27th CEIES Seminar

ISSN 1725-1338 Ζ
Q
LU
«ί
ο
O
27th CEIES seminar

Producing business statistics
Cost and efficiency
Recent developments and experiences
Dresden, 14 and 15 October 2004
THEME
* *
A<* General
EUROPEAN and regional
COMMISSION eurostat • s - » statistics
Our mission is to provide the European Union
with a high-quality statistical information service
Eurostat publications cover nine themes:
53 General and regional statistics External trade
gjg Economy and finance

£, Transport
8ÎÂÔ Population and social conditions 36 Environment and energy
Industry, trade and services Science and technology
Agriculture and fisheries
' Eurostat publications are divided into collections adapted to the needs of all users: ' CONTENT FORMAT
Press releases: Eurostat publishes more than 200 press releases every year in as Paper
connection with the release of new data. The press releases include the key figures f- HTML
on the EU, the Member States and their partners. They are provided free of charge. PDF
They are also available on the Internet as soon as they are published:
www.europa.eu.int/comm/eurostat/
Paper
Statistics in focus: This collection is published regularly by Eurostat and provides
up-to-date summaries of the main results of statistical surveys, studies and analyses.
Ini m PDF
It covers all themes and consists of about 4 to 12 pages per issue. Eurostat issues around ¡it Φ
200 Statistics in focus per year. Available as single copies or on subscription.
P o c k e t b o o k s : This collection provides the main indicators on a theme or Paper

sub-theme in an easy-to-carry booklet of up to 150 pages. Available as single
Ini PDF
copies or on subscription. m
P a n o r a m a of t h e E u r o p e a n U n i o n : This collection highlights the features Paper

IM :::::
and trends of a multi-sectoral theme or a sectoral sub-theme. The publications CD-ROM
consist of around 200 pages and include analyses as well as tables and graphs. ==ΐ © PDF
Available as single copies or on subscription.
M e t h o d s a n d n o m e n c l a t u r e s : Intended for specialists who want to consult PDF

=
the methodologies and nomenclatures used for a theme, a sub-theme or a sector.
Available as single copies.
D e t a i l e d t a b l e s are intended for specialists. They present part, or all, of the ..... Paper
statistical data compiled on a theme, sub-theme or sector. Available as single
Hil CD-ROM
copies or on subscription. B PDF
Catalogues provide rapid access to concise information on Eurostat's products a¡ Paper

and services. They are free of charge and include the Mini-guide, a selection of CD-ROM
Eurostat products and services, and Statistical references, the newsletter on
Eurostat's latest products and services.
Graphs Tables as Comments Maps

ISSN 1725-1338 Ζ
O
iii.
o
o
27th CEIES seminar

Producing business statistics
Cost and efficiency
Recent developments and experiences
Dresden, 14 and 15 October 2004
THEME
* * EUROPEAN =¿L Λ* General

and regional
COMMISSION eurostat statistics
Immediate access to harmonised statistical data
www.europa.eu.int/comm/eurostat/
Eurostat o n t h e I n t e r n e t j
Our website is updated daily. Visit it today and get:
• direct and free access to all Eurostat PDF publications;

• direct and free access to our databases, presenting the latest and most complete statistical
information available on the European Union, the EU Member States, the euro zone, the
European Economic Area and other European partner countries;
• alert me — customisable e-mail alerts that inform you immediately when new data or publi
cations on your preferred topics become available;
• specialised access to short-term economic data;
• complete information on all Eurostat products and services.
European Statistical Data Support J
Eurostat has set up with the members of the 'European statistical system' a network of support centres
which will exist in nearly all Member States as well as in some EFTA countries. Their mission is to pro
vide help and guidance to Internet users of European statistical data.
Complete details concerning this support network can be found on our Internet site.
M e d i a S u p p o r t Eurostat J
Journalists can contact the media support service:

Tel.(352)4301-33408
Fax (352) 4301-35349
E-mail: eurostat-mediasupport@cec.eu.int
Europe Direct is a service to help you find answers to your questions about the European Union
Freephone number (*):
00 8 0 0 6 7 8 9 10 11
(*) Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed.
A great deal of additional information on the European Union is available on the Internet.
It can be accessed through the Europa server (http://europa.eu.int).
Cataloguing data can be found at the end of this publication.
Luxembourg: Office for Official Publications of the European Communities, 2005
ISBN 92-894-8298-2
ISSN 1725-1338
© European Communities, 2005
Printed in Luxembourg
PRINTED ON WHITE CHLORINE-FREE PAPER

EUROSTAT
L-2920 Luxembourg Tel. (352)43 01-1 —www.europa.eu.int/comm/eurostat/
eurostat
Eurostat is the Statistical Office of the European Communities. Its task is to gather and analyse
figures from the different European statistical offices in order to provide comparable and
harmonised data for the European Union to use in the definition, implementation and analysis of
Community policies. Its statistical products and services are also of great value to Europe's
business community, professional organisations, academics, librarians, NGOs, the media and
citizens.
To ensure that the vast quantity of accessible data is made widely available and to help each user
make proper use of the information, Eurostat has set up a publications and services programme.
This programme makes a clear distinction between general and specialist users and particular
collections have been developed for these different groups. The collections Press releases,
Statistics in focus, Panorama of the European Union, Pocketbooks and Catalogues are aimed at
general users. They give immediate key information through analyses, tables, graphs and maps.
The collections Detailed tables and Methods and nomenclatures suit the needs of the specialist
who is prepared to spend more time analysing and using very detailed information and tables.
As part of the new dissemination policy, Eurostat has developed its website. All Eurostat
publications are downloadable free of charge in PDF format from the website. Furthermore,
Eurostat's databases are freely available there, as are tables with the most frequently used and
demanded short- and long-term indicators.
Eurostat has set up with the members of the 'European statistical system' a network of support
centres which will exist in nearly all Member States as well as in some EFTA countries. Their
mission is to provide help and guidance to Internet users of European statistical data. Complete
details concerning this support network can be found on our Internet site.
Eurostat
1st day
14 October 2004
REGISTRATION
OPENING SESSION
Welcome to the participants
B. Graf Puckler, Chairman of the CEIES Subcommittee on Economic and Monetary Statistics
Ms I. Schneider-Böttcher, President of the Statistisches Landesamt des Freistaates Sachsen, Germany
Mr G. Kopsch, Director, Federal Statistical Office of Germany
KEYNOTE SPEECHES
Reasons for being more efficient (from CEIES point of view)
Mr U. Heilemann, Vice-Chairman of CEIES
Recent developments and point of view of Eurostat
Mr M. Glande, European Commission, Director, Eurostat
1. COST AND EFFICIENCY

CHAIR: M R J. RICHTER, WIRTSCHAFTSKAMMER ÖSTERREICH, AUSTRIA
1.1 TECHNIQUES TO MEASURE COST AND EFFICIENCY

Mr B. Gorzig, German Institute for Economic Research
1.2 PERCEPTION BY THE BUSINESS WORLD

Mr A. Tortopidis, Federation of Greek Industries, Greece
Mr T. Donohoe, Irish Business and Employers Confederation
LUNCH BREAK
1.2 (Continued) PERCEPTION BY THE BUSINESS WORLD
Mr R. Amler, Siemens AG, Germany
OPEN DISCUSSION
2. MEASURES TO IMPROVE EFFICIENCY

CHAIR: M R I. MACLEAN, STATISTICS USER COUNCIL, UNITED KINGDOM
2.1 GENERAL STRATEGY OF NSI's

Mr T. Mesenbourg, US Census Bureau
Mr M. Cloutier, Statistics Canada
OPEN DISCUSSION
END OF FIRST DAY
27th CEIES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience
2nd day
15 October 2004
2.2 DATA REPORTING ORGANISATION

Mr J. Ritzen, Statistics Netherlands
2.3 SURVEY ORGANISATION

Mr M. Lagerström, Statistics Sweden
2.4 USE OF ADMINISTRATIVE DATA

Mr S. Egmose, Statistics Denmark
2.5 IMPROVEMENT OF THE BENEFIT OF OFFICIAL STATISTICS

Mr U. Scheinost, Zentralverband der Elektrotechnik und Elektronikindustrie, Germany
Mr Ρ Boegh Nielsen, European Commission, Directorate General Entreprise
LUNCH BREAK
OPEN DISCUSSION
Summing up
B. Graf Pückler, Chairman of the CEIES Subcommittee on Economic and Monetary Statistics
Reaction from Eurostat

Mr M. Glande, Director
Closing remarks
B, Graf Pückler, Chairman of the CEIES Subcommittee on Economic and Monetary Statistics
END OF THE SEMINAR

* * * * * * * * * *
27th CELES seminar: Producing business statistics Costs and efficiency Recent developments and experience
m
Background and aim of the seminar
The European Commission has recognized the problem that excessive regulation places burdens on enterprises.
It highlighted, in a contribution to the special European Council in Lisbon, that "a number of areas are
underperforming. These require urgent attention to improve the Union's medium term economic outlook.
Regulatory costs are far too high placing unnecessary and costly burdens on European enterprises. The costs
of red tape, partly at European level, but mostly due to national and regional overregulation, have been
estimated by the OECD to be between 3 and 5 percent of the EU's GDP. A new initiative is needed to improve
the regulatory environment at all levels."'
From its earliest beginning, the European Union has needed harmonized statistical data from the National
Statistical Institutes (NSI). The creation of European Economic and Monetary Union, the establishment of the
European Central Bank and the accession of new Member States, have increased acceptance of the need for
new supranational statistics. This need has been expressed by the government sector, the business community
and by the financial markets. As increased data requirements from the Council and the Commission may have
led to cost increases not only in the enterprises, but also in the national statistical offices, the question of the
efficiency of official statistics has to be raised both regarding their production process and their significance
for the users.
Years ago - in 1995 - the CEIES Subcommittee on Business Statistics dealt with the subject of "Statistical
Burdens on Businesses". The main conclusions of this study were:
1 ) «The load of statistical surveys on businesses is small by comparison with the costs of complying more
generally with administrative requirements of government. The burdens are nevertheless perceived to be
considerable and thus merit serious attention.
2) The burdens on businesses, which depend on the costs and uses of the statistics, vary enormously and are
influenced by a range of factors"2
For these reasons, the CEIES Subcommittee on Monetary and Economic Statistics has decided to put this
subject on its agenda again. It has decided to organise a similar seminar which will, inter alia, analyse to what
extent the 1995 conclusions are still valid and which measures still have to be taken to balance costs and
benefits. Two key questions are:
• are we producing the right statistics and are we doing it in the most efficient and cost effective way?"
• are we taking the right steps to ensuring that firms understand both the reasons for collection and the benefit
to them of the outputs?
The aims of the seminar are:
• to discuss the problems of measuring efficiency in official statistics
• to get a more concrete idea of the perceived burden on the respondents
• to discuss the preconditions (technological, methodological, etc.) regarding a more general frame for reducing
the burden for the respondents
• to highlight how the perceived value of the output of official statistics could be raised
• to look for possibilities to enhance the benefits of statistical data for management decisions
• to present concepts for improving efficiency in the light of decreasing resources of Statistical Institutes.
What is C E I E S ?
CEIES stands for Comité consultatif européen de l'information statistique dans les domaines économique et
social; in English: 'The European Advisory Committee on Statistical Information in the Economic and Social
Spheres'. Its task is to assist the Council and the Commission in the co-ordination of the objectives of the
Community's statistical information policy, taking into account user requirements and the costs borne by the
information producers.
1
'Hie Lisbon European Council - An Agenda of Economic and Social Renewal for Europe
Contribution of the European Commission to the Special European Council in Lisbon, 23-24th March 2000; DOC/00/7, p. 14
;
5th Meeting of the European Advisory Committee on Statistical Information in the Economic and Social Spheres (CEIES)
Report of the CEIES - Subcommittee " Business Statistics" "Statistical Burdens on businesses", p. 1
The committee was set up by Council Decision 91/116/EEC of 25 February 1991. The original decision was
amended by Council Decision 97/25 5/EC of 19 April 1997 taking into account the accession of Austria, Finland
and Sweden,
CEIES is chaired by the Commissioner responsible for statistics. The vice-chairman is Mr Ulrich Heilemann
from Germany. CEIES is composed of two private members per Member State, three members from the
European Commission, the Chairman of the Committee on Monetary, Financial and Balance of Payments
Statistics (CMFB) and the Presidents or Directors-general of the National Statistical Institutes of the Member
States.
The European Advisory Committee on Statistical Information in the

Economic and Social Spheres Secretariat: Eurostat, Unit A-2
http://europa.eu.int/comm/eurostat
e-mail : estat-ceies@cec.eu.int
Organisers: The CEIES subcommittee on Economic and Monetary Statistics: B. Graf Pückler, chairman, Ms F
Ruane, Mr J. Le Duigou, Mr M. Brachmond, Ms S. Schnorr-Backer, Mr I. Maclean, chairman of the CEIES
Dissemination Statistics subcommittee andfrom Eurostat, Mr J. Debruyn
CEIES Secretariat: Ms A. Näslund-Fogelberg, Ms N. Lauwerijs, Ms I. Schön
CONTENTS
The proceedings are a collection of papers prepared by the speakers in advance of the seminar. They do not
include the open discussions nor the discussions in the round table/panel session.
The papers presented and published herein only represent the views of their authors and do not necessarily
reflect an official position of their institutions or organisations.
WELCOME TO THE PARTICIPANTS

B. Graf Pückler, Chairman of the CEIES Subcommittee on Economic and Monetary Statistics 10
Ms I. Schneider-Böttcher, President of the Statistisches Landesamt des Freistaates Sachsen, Germany . 11
Mr. G. Kopsch. Federal Statistical Office of Germany 13
KEYNOTE SPEECHES
1. REASONS FOR BEING MORE EFFICIENT (FROM CEIES POINT OF VIEW)

Mr. U. Heilemann, Vice-Chairman of CEIES 17
2. RECENT DEVELOPMENTS AND POINT OF VIEW OF EUROSTAT

Mr. M. Claude, European Commission, Director, Eurostal 21
COST AND EFFICIENCY

Mr. B. Görzig, German Institute for Economic Research 28
Mr. A. Tortopidis, Federation of Greek Industries 36
Mr. T. Donohoe, Irish Business and Employers Confederation 39
Mr. R. Amler, Siemens AG, Germany 47
MEASURES TO IMPROVE EFFICIENCY

Mr. T. Mesenbourg, US Census Bureau 58
Mr. M. Cloutier, Statistics Canada 64
Mr. .1. Ritzen, Statistics Netherlands 77
Mr. M. Lagerström, Statistics Sweden 85
Mr. S. Egmose, Statistics Denmark 93
Mr. U. Scheinost, Zentralverband der Elektrotechnik und Elektronikindustrie, Germany 100
Mr. P. Boegh Nielsen, Eumpean Commission, Directorate General Enterprise 107
Mr. F. Pfrommer, German Statistical Office 114
Summing Up and Opinion

B. Graf Pückler, Chairman of the CEIES Subcommittee on Economic and Monetary Statistics 121
Reaction from Eurostat

Mr. M. Claude, Director 128
List of Participants 131
27th CEIES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience 9
BOTHO, GRAF PÜCKLER
Chairman of the CEIES Subcommittee
on Economic and Monetary Statistics
Welcome to this seminar, and I trust you had a pleasant journey to Dresden. I would like to extend a particularly
warm welcome to the President of the Statistical Office of the Free State of Saxony, Mrs Schneider-Böttcher,
the Vice-President of the CEIES, Professor Dr Heilemann, and Mr Kopsch, Director of the Federal Statistical
Office, who will speak to us on behalf of its President, Johann Hahlen.
Before we start the seminar proper, I would just like to say a few words on why we are devoting our time to this
topic.
The introduction of the single market in the early 90s and of Economic and Monetary Union at the end of the
90s and, last but not least, the Lisbon summit in 2000, led to a considerable increase in the data required by the
European Commission and by the Member States themselves. The increased demand for data placed a
particularly heavy burden on industry.
The CEIES Business Statistics Subcommittee began to address this issue in the mid-90s and concluded that
businesses felt the burden to be particularly acute, even though, comparatively speaking, it was less than that
imposed by the administrative cost of legal regulations and the like. The committee came up with a number of
recommendations as to how the burden could be eased.
In the past, the national statistical offices have done a great deal to minimize the data burden placed on
businesses as a result of additional data requirements. The continued production of business statistics by the
statistical offices is thus crucial. This is a topic to which we will be paying close attention here, and we hope
that, at the end of the seminar, we will be able to draw a number of conclusions that will help with policy-
making and benefit NSIs and businesses alike.
Before I hand over the floor to Mrs Schneider-Böttcher, a few remarks of a practical nature:
1. The conference languages are English, French and German.
2. The City of Dresden has kindly 'invited the participants of this seminar to a reception this evening at seven
thirty. If you wish to attend, which I very much hope you do, I would kindly ask you to let us know in advance,
as this will make it easier for our hosts to plan the evening.
After introductory remarks by Mrs Schneider-Böttcher and Mr Kopsch, we will be hearing keynote addresses
from Professor Heilemann and Mr Glaude from Eurostat. These should certainly give us plenty to think about!
After the coffee break, we will be continuing with "Costs and efficiency ", with Mr Richter from the Austrian
Chamber of Industry in the chair. I would kindly request all speakers not to overrun their allotted times. I'm a
football fan, so I've brought along three cards - green, yellow and red. As in a football match, red follows
yellow. The chairman of each topic section may show speakers a yellow card if they look likely to overrun.
Please avoid getting a red card!
Thank you.
10 27th CEIES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience
IRENE SCHNEIDER-BÖTTCHER
President of the Statistisches Landesamt des Freistaates Sachsen, Germany
The Prime Minister of the Free State of Saxony, Dr George Milbradt, has taken on the sponsorship of this
seminaron the costs and efficiency of producing business statistics which we are holding here in Dresden. The
Prime Minister would have liked to welcome you personally, but sadly, this was not possible for scheduling
reasons. He asked me to greet you on his behalf, and to express his hope that the seminar will be very successful
and that you will find your stay in Saxony and the beautiful city of Dresden enjoyable.
Prime Minister Milbradt is very knowledgeable about official statistics and he highly values the contribution
they make in policy terms. The Prime Minister is also particularly keen to reduce the administrative burden on
businesses, and official statistics are clearly part of this burden. I am pleased to see that you will be focusing
on this topic in this seminar.
In the age of globalisation, the knowledge-based economy and the information society, data and information
on the economic and structural development of global markets and society are increasingly important.
Decision-makers in industry, politics and government need relevant, up-to-date and reliable information if their
plans are to have a viable basis. This applies particularly to statistical information concerning companies and
business.
However, the perception of statistical data differs greatly according to the type of enterprise. While large
companies that are active in international markets need statistical information to analyse their markets and
develop suitable marketing strategies, this is less important for smaller companies. For the most part, small and
medium-sized enterprises with fewer than 500 employees, which data from the Institute for Small Business
Research show make up approximately 99.7% of all enterprises in Germany, are only active regionally and/or
nationally. They see official statistics as a costly, bureaucratic and time-consuming burden rather than a useful
tool for making informed business decisions.
So it is not surprising that official statistics are seen as part and parcel of the intense debate about streamlining
bureaucracy going on in Germany, as elsewhere. As I mentioned previously, I therefore welcome the fact that
this seminar will be dealing not only with the statistical burden placed on businesses but will also be
concentrating on what statistical information means to small and medium-sized enterprises and how it is used.
Ultimately, information on regional economic structures and development is needed for fostering economic
development, for example here in the Free State of Saxony.
One major problem is the discrepancy between the data requirements of the European Commission and other
European institutions, such as the European Central Bank, on one side, and trade and industry on the other.
While the Commission mainly needs highly aggregated data, companies at European level find branch-based
data most relevant. What Eurostat publishes thus take on a special meaning in terms of how official statistics
are used.
The danger at present is that the statistical data being produced at European level are increasingly geared to the
needs of the Commission to the detriment of trade and industry. The European Commission needs to be made
more aware that European statistics have to satisfy industry's need for information as much as its own. Only if
business managers feel that the statistical information offered serves a purpose will they be willing to provide
quality "raw material" - i.e. the requisite data from company accounts - with due care.
As already mentioned, there is no doubt that there is a need for official statistics in the Information Society.
Nevertheless, I would join the Prime Minister in calling on all of you taking part in this seminar and on all
decision-makers to do your utmost to keep the statistical burden placed on enterprises by official statistics as
light as possible. No one disputes that companies are overburdened with bureaucratic red tape, and the burden
is just getting worse. Bureaucracy affects and influences crucial business decisions. The fact that it stifles the
entrepreneurial spirit, the main source of a dynamic economy, is particularly serious.
For this reason, the effort being made to modernise the production of statistics should be welcomed, as should
on-line data collection of company statistics, the anticipated automatic collection of data from company
accounts, use of data from the business register, and greater use of administrative records. These measures will
eventually free companies from the burden of reporting obligatory information. The burden aspect must play a
central role in the efforts currently being made to reform company statistics. In future, official statistics should
not focus only on the needs of large companies - regional, branch-based information should also be readily
available to small and medium-sized enterprises. Enterprises will also have to be encouraged to take greater
interest in official statistics, this in turn fostering greater understanding of the data that needs to be collected.
With this in mind, I hope that you will find the talks and discussions you will have over the next two days
stimulating, and wish you an enjoyable stay in our beautiful capital of Dresden.
12 27tli CEIES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience
GUNTER KOPSCH
Director, Federal Statistical Office, Germany
It is my pleasure to welcome you here today in Dresden to the CEIES Seminar on Producing Business Statistics
- Costs and Efficiency. I say this not as the host of the event, who is of course the previous speaker, Mrs
Schneider-Böttcher of the Statistical Office of Saxony, which has its head office very close to Dresden, noras
the organiser which is the CEIES with Graf Pückler as the Chairman of the Subcommittee on Economic and
Monetary Statistics, but - since this is an international event in the field of statistics being held on German soil
and thus involving the Federal Statistical Office - it is my honour as its representative to deliver this welcoming
address.
The topic of this seminar is that of costs and efficiency in the production of business statistics or, to be more
precise, an attempt to find answers to the question of how we can improve the efficiency of our production
processes and reduce costs. Rising to this challenge has, for some time now, been crucial to the survival of
official statistics and will continue to be so.
The production of statistics generates costs in two respects. On the one hand, there are those borne by the
statistical offices, but we must also appreciate the burden on, and thus the costs for, the enterprises surveyed,
without whose information business statistics cannot be compiled. Official statistics come in for public
criticism on account of this information burden. On top ofthat, discussions in Germany on the programme for
official statistics put considerable emphasis on reducing the burden on respondents, who have in the meantime
mobilised a powerful lobby, to which the statistical offices must respond. Reducing this burden of providing
information is therefore a substantial consideration in terms of the Federal Statistical Office's strategic
objectives.
Public discussions focus less on the dwindling resources of the statistical offices. With each passing year the
number of people working for the statistical offices is falling. While official statistics may in great demand,
there is no strong lobby to support them in budget negotiations.
The juggling act required of statistical offices in Germany (and elsewhere too no doubt), involves performing
ever increasing tasks with constantly dwindling resources, or - to seek an analogy - it is the art of using less
and less wool to knit a longer and longer jumper. However, in order to ensure that the looser stitches do not let
in the cold, we need better quality wool and, above all, better knitting techniques. The statistical offices in
Germany are taking up these challenges by introducing a variety of innovations at all stages in the production
of statistics. Pride of place amongst the Federal Statistical Office's strategic objectives for 2004 - 2008 goes to
the standardisation and modernisation of work processes and use of the latest technology alongside the
aforementioned goal of easing the burden on respondents.
Given these two dimensions to our efforts to reduce costs, the most valuable innovations are those which both
ease the burden of response and streamline working practices in the offices to make them more cost-efficient.
When this has been achieved, the increase in efficiency is particularly striking, and there are many examples
of such success stories in Germany. These have been brought together under the master plan for the reform of
official statistics which was adopted by the statistical offices in 2003:
• I would like to present some of our activities in the area of optimising data collection to illustrate how
increased efficiency can benefit both respondents and the statistical offices:
For all online statistics which it itself collects, the Federal Statistical Office offers the possibility of online
reporting. The regional statistical offices have also already made it possible for a significant proportion of
the data required for decentralised statistics to be reported online. There are obvious advantages to this: it is
easier for businesses to submit data, and the statistical offices can then process these on the same medium
right through to the production of the end product. We have however set our sights even higher. We are
working with leading software companies on the automatic extraction of data from company accounts. The
aim is to use appropriate software packages to automatically obtain the information needed to produce
statistics - i.e. without imposing any additional workload on businesses - and to relay this directly to the
statistical office- i.e. the data is more up-to-date than in traditional methods of collection.
The tests we are currently conducting on the use of administrative data for short-term statistics will show to
what extent reductions can be made in primary surveys whilst maintaining - or even extending - the data base
for official statistics. At present, we are concentrating here on the use of turnover and employment data from
the tax authorities and the employment agency.
The introduction of a single business code, as already exists in many other EU countries, would make data
exchanges between business and administration - as well as between administrative authorities - much easier.
The world of statistics is just one of the many beneficiaries here and we strongly support this work being
carried out by the German administrative authorities.
Whilst we are cooperating with external partners on the above-mentioned activities in order to increase our
work efficiency, there are other innovations which affect working practices within the German statistical
system:
• At the Federal Statistical Office and in several of the regional offices, we have used comprehensive process
analyses to analyse and, where necessary, re-organise the range of working practices which have developed
over the course of time. This has generally led to double-digit gains in efficiency. The monitoring instrument
used is the cost and management accounting system which was introduced some years ago and enables us
to determine the resources used for each area of activity and each product.
• The division of labour between the Federal and regional statistical offices should be promoted through
optimised cooperation. It is common knowledge that statistics in Germany are organised on a federal basis
with a federal statistical office and 16 regional offices in the Bundesländer. There has long been cooperation
in the field of IT development on the basis of "one for all". In future, individual statistical offices should act
as processing centres which carry out statistical production work on behalf of all or some of the other offices.
There is certainly room to further expand this division of labour on a national scale, along the lines of our
plans for the European trend of Centres of Excellence. The basic idea is, however, the same. A specific task
is performed by one or more offices on behalf of all. In Europe, this is more than likely to be development
work. I would like to take the opportunity offered by this European Forum to argue for the introduction of
Centres of Excellence, as this topic will be examined by the next SPC meeting in November.
• Another measure is the standardisation of production processes when compiling statistics. As with the
concept of using platforms in the car industry, standardised sub-processes can also be carried out using multi-
statistical IT procedures to compile different sets of statistics. Our priorities in Germany are to devise uniform
questionnaires, provide opportunities for online reporting, establish uniform identifiers to control processes,
compile uniform or central databases for the management of surveys, develop standard procedures for
checking incoming data and for reminders, to standardise manuals, marking, plausibility checks and the
evaluation of data.
We are convinced that with such measures we are on the right road towards mapping out the future of official
statistics in Germany by letting the statistical offices develop their own initiatives.
On the topic of the "burden on businesses", I would like to add that we are working together with a renowned
research institute on producing a comprehensive and meaningful measurement of the reporting burden on
respondents. At the Federal Statistical Office, we intend to use this measurement to develop a permanent
information system on the trends in this burden on industry.
One of the key tasks of statistical offices - and one which, as I said at the start, may even be vital for their
survival - is to analyse the cost of statistical production and the efficiency of our work processes. In order,
however, to conduct a thorough evaluation of our work, we also need to look at the information we are offering.
At the end of the day, it is both the range and, above all, the quality of our products and how they are rated by
our users which let us know whether the work we are doing is good. Cheap statistics and no statistics at all can
both turn out to be very expensive, if the inaccuracy or absence ofinformation results in wrong decisions being
taken. We must keep our sights fixed on ensuring that our statistical information is of the highest quality, at not
just national but also at European level.
The current discussions on the deficits notified by Greece within the Economic and Monetary Union underline
the political and economic importance of reliable and meaningful data. The quality and reliability of our
statistics cannot, however, be guaranteed unless Eurostat and the statistical offices in the Member States are
able to work on an objective, impartial and professionally independent basis and are given sufficient resources.
We feel that some radical improvements can still be made to the general framework required for this purpose
both at European level and in the 25 Member States:
- For example, the discussions on establishing a constitution for Europe did not lead to the Article on statistics
in the current EU Treaty being extended so as to reinforce the legal basis for the professional independence
of the statistical offices in the EU and entrust the European statistical system with the objective, impartial and
scientifically-verifiable production and dissemination of statistics. Work has, nonetheless, started recently on
amending the European statistics regulation, and I am confident that we will manage this time to strengthen
the legal basis for the independence of European statistics.
In order to produce specialised high-quality statistics, Eurostat and the national statistical offices must be given
the necessary human and material resources, and clear priorities must be set in the Statistical
Programme. This point was also emphasised by the Economic and Financial Affairs (ECOFIN) Council at its
meeting of 2 June 2004, as was the importance of the independence of official statistics:
"High-quality statistics are fundamental for European policies. The Council considers that integrity,
independence and accountability of data compilers, and the transparency of the compilation methods,
underpinned by the appropriate institutional arrangements, are crucial to ensure such high quality
statistics The capacity of the statistical authorities is fraught with both large national and increasing
European requirements for statistics. To preserve and improve the quality and reliability of statistics, it is
important to review priorities of the statistics and to reduce (legal) requirements for areas which are now
considered to be of less importance."
In order to improve prioritisation, a Eurostat Task Force chaired by Denmark has drawn up a list of measures
which was endorsed in May of this year by the SPC and, I feel, can lead to progress. The following proposals
in particular should be implemented as quickly as possible:
• Eurostat and the Member States produce cost calculations for statistical projects (beginning with new projects
from 2005 onwards),
• Eurostat draws up an inventory of European requirements and legal bases,
• a time limit for statistical legislation,
• a thorough sector-by-sector review of the Statistical Programme, beginning with economic statistics from
2006 onwards
and first and foremost
• greater participation by the Member States and European users - not just the European Commission, but also
the European Central Bank and the CEIES.
This brings me to the third and last topic of this address: a plea for a strong forum of users in the European
statistical system.
We at the Federal Statistical Office see strengthening the CEIES as an important objective in the discussions
on the further development of the European statistical system, with one of the aims of the reforms also being
to make the CEIES work more effectively.
These considerations become all the more pressing with the accession of the 10 new Member States. A Eurostat
Task Force is currently drawing up proposals on specifying the tasks of the CEIES and amending the number
and composition of its members. One of the alternatives envisaged is to slash the number of members from the
current 58 to 12. This would be done by reducing the number of so-called "private users" from business, science
and society from the current two representatives per Member State to a total number of 2 or 4 members. A
compromise proposal being discussed at present suggests cutting the total number of members to 19, rather
than 12, and giving the "private users" one more voice than institutional users from European and national
institutions.
There is no doubt that the enlargement of the EU to 25 Member States means that the structure of the CEIES
needs to be changed. The Federal Statistical Office, however, sees no justification for the drastic reduction
being discussed at the moment; nor can this be reconciled with our shared vision of a stronger CEIES.
The CEIES is similar to the Statistical Advisory Committee in Germany in that it has an important role to play
as a go-between in balancing the interests of European official statistics and those of European users from
politics, business, science and society. The Federal Statistical Office is therefore arguing the case for a viable
CEIES which at the same time can adequately represent the users of the European statistical system. No
fundamental changes should therefore be made to the current structure of the CEIES. One possible option is
that in future there should be just one private user per Member State sitting on the CEIES and the number of
statistical offices represented could be reduced to 5 or 6.
The idea of slashing the number of members on the CEIES is also incompatible with the large number of tasks
it has to carry out, especially since these are increasing rather than decreasing. One of these tasks is the
organisation of seminars, such as that being held here in Dresden over the next two days, which in the past
have stimulated important developments in official statistics in Europe. Such seminars cannot, however, be
organised and run unless the necessary human and financial resources are made available. All the more reason
why appropriate human resources should still be available to the CEIES and thought should be given to
providing it with better funding.
Ladies and gentlemen, I look forward to interesting talks, animated debate and some good fringe discussions
over the next two days. Thank you very much for your attention.
m
REASONS FOR BEING MORE EFFICIENT

( F R O M C E I E S P O I N T O F V I E W )
Producing Business Statistics - Costs and Efficiency

Recent Developments and Experiences
Some Comments on the Topic1
by Ullrich Heilemann2
I. Welcome speech as Vice-chairman of the CEIES

1. Happy about the large number of participants and looking forward to the excellence of the contributions,
judging from a first reading
2. Delighted that the seminar is being held in Saxony, in Dresden (Leipzig would of course have been even nicer...)
3. An important historical location for us, since Saxony - along with Prussia and Bavaria - is the birthplace of
German statistics and its first Director, Ernst Engel, is, without any fear of contradiction, regarded as the
greatest German statistician of the 19th century ...who secured his place at the forefront of economics with
his basic scientific findings - Engel's law.
II. Wider connections between the location and the topic

1. There are two main roots to official statistics in Germany
a. the first dimension, as in all western European countries, is information on the country, which had been
developed for government and administrative practice since the emergence of nation states in the 16th/17th
centuries and
b. the second, which arose to a certain extent in the course of industrialisation, was the need to procure
information on
1 ) industry, population, production and consumption
2) poverty and risk
3) «morals», in the sense of delinquency, births inside marriage, etc.
4) disablement (health, illness, age).
c. A decisive factor in the current context is that interest in many of these areas - such as industry - developed
not just at State level, where there was even a certain lack of enthusiasm, but rather among traders and
businessmen.
In order to manage their activities effectively they required information on
1) the labour potential
2) health and other risks
' Keywords on the papers presented at the 27th CEIES-Seminars entitled "Producing business statistics - costs and efficiency - recent developments and
experiences" held on 14 and 15 October 2004 in Dresden.
;
Prof. Dr. Ullrich Heilemann. University of Leipzig, Institute for empirical economic research, Vice-chairman of the CEIES.
27lh CEIES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience 17
3) costs, prices, etc.
and they naturally sought a comparison - with no ifs and buts.
d.The State (Saxony, Prussia, etc.) did not feel able at first to do very much about this, for cost and other
reasons, perhaps because
1) it did not appreciate the significance of these issues,
2) there was an initial lack of budgetary resources,
3) the State - clearly more so than businesses - shied away from publicising the situation for internal and
external reasons.
Sound familiar?
e. The initial response was therefore to establish statistical offices, which tackled these statistical issues with
financial support from the State.
The financial outlay was modest....
As far as I know, however, the 'burden' of statistics was less of an issue back then, either because
1) there was less bureaucracy as a whole
2) businesses considered that the advantages far outweighed any individual burdens
3) there was also a financial aspect: the cost of the three-yearly population census in Saxony in 1848
amounted to 1 800 thalers, whilst the 1846 census of industry had swallowed up 2 000 thalers, at a time
when the budget available to the Statistischer Verein (the statistical office at that time) was only 2 000
thalers ...
III. Cost and efficiency of statistical production - a recurring topic for the CEIES
A. The cost and efficiency of statistical production is - explicitly and implicitly - a key issue for the CEIES
and should be viewed in connection with the statistical programme. It is thus a recurring topic - as of course
it is for EUROSTAT and the national statistical offices too - but that is only of indirect interest to us here!
1. Directly ... more on this presently..
2. Indirectly, because whenever statistical programmes are discussed it is easy enough to find good reasons
for doing "more", "better" and "faster", but when it comes to doing "less", ...
3. Market-failure and non-market-failure (Heilemann 1998)
Β. The cost and efficiency of statistical production is - leaving producers out of the equation at any rate - an
issue which has received little and sporadic empirical attention.
1. "The Economist" investigated the matter in depth in 1993 and 1995 and discovered a high positive
correlation between "statistical quality" and "expenditure on statistics" with Canada (US$ 8.2 per capita
per annum) leading the field and Belgium (USS 3.6) at the bottom.
2. The CEIES looked at this question as far back as 1995 - i.e. shortly after it was founded (1991) - , taking
a study by Machin (1995) as the basis, and put the problem firmly into context. For the EU Member States:
a. 5% of the total costs were caused by State rules
b. SMEs spent 24 hours of their annual working time on statistics - i.e. 0.1% of the annual total...
c. there were major fluctuations from business to business
d. there was a massive difference between actual and perceived burdens
3. Input orientation of the analysis ...
4. The time when the marginal return from expenditure on official statistics seemed to justify virtually every
amount - the Recession - (Arrow 1966 [ 1957]), is long since gone.
1 ) In order to improve the situation, more needs to be done to make it clear to users what all can be
achieved with statistics -Anmiller paper (?). Academic teaching has sidelined the role and importance
of descriptive statistics very much in favour of methods ...
2) The trade offs between the different criteria - quality v. timeliness - need to be explained, and above all
what they cost!
3) The problem of the distribution between producers and users needs to be made clear.
4) Last but not least: when doubts are cast on the benefits of State action in general, this naturally has
repercussions on statistics. The 19th century was obviously smarter- see above - in focussing on the
key role of statistics for private sector activity.
C. Cost overview for individual statistics (Heilemann 1998)
1. Concept of margins for statistics - Radermacher 2004
2. Problem: Evaluation of benefits? - how much do businesses spend and on which information from private
data producers?
3. Cost statistics (Heilemann 1998, EUROSTAT) as a first step:
a. CPI: Norway: 9, Sweden: 14, Netherlands: 24 staff...
b. Trade integration: Sweden: 14, Netherlands: 48
D. Citizens and respondents must be able to see the benefits of statistics
1. The problems in recent years are well-known. Think here of
a. the calculation of the deficit for Greece ...
b. perceived inflation in relation to the introduction of the euro
c. Italy, England (sic), USA etc. - Economist article
2. Important: - national and European " statistics" are hauled over the coals for everything — even for wrong
economic forecasts etc. - so we are all in the same boat.
IV. Conclusions
A. This is not an attempt to anticipate other papers or the conclusions of the CEIES!
Β. Analysing the costs and efficiency of statistics is important and is an ongoing task as in any activity.
C. All experience - and particularly that of recent years - has however shown that there may still be
considerable cost and efficiency potential in this area. To put it bluntly: cut-off points also have their limits
and a price!
D. I feel that in order to ensure and develop the supply of high-quality statistics in the statistical era - as
famously proclaimed by Goethe in 1786 when he visited the Bolzano Trade Fair during his Italian journey
- these need to be legitimised first and foremost by the role they play
1. in meeting the demands made by the State
2. in meeting the needs of the economy and businesses
3. for civil society.
E. There is no question that users and respondents have a decisive role to play on both the supply and demand
sides in order to achieve this aim - just as they did in the early days of official statistics in the 19th century.
There is also no question that it is currently much more difficult, for a variety of reasons, to do this at
European level than at national level where there are well-established, and tried and tested structures
governing supply, demand and dissemination.
This is precisely the task which the CEIES has to perform.
Events such as today's seminar make a major contribution to this w o r k - given their content and purpose -
but they are also a forum for the users and producers of European statistics to exchange information on their
expectations, demands, inputs and opinions - be these in the domains of
the State,
the economy
or civil society.
Bibliography
Amler, R. (2004), Measuring the efficiency of statistical surveys. Paper presented to the 27th CEIES Seminar
"Producing Business Statistics - Costs & Efficiency - Recent Developments And Experiences", 14 and 15
October 2004 in Dresden.
Arrow, K. J. (1966 [1957], Statistik und Wirtschaftspolitik [Statistics and Economic Policy]. In: G. Gäfgen
(Ed.), Grundlagen der Wirtschaftspolitik. (Neue Wissenschaftliche Bibliothek, 11.) Köln, p. ....... (Reprint
from Econometrica, 25 (1957), pp. 523-530).
Economist (1993), The good statistics guide. September 11"1, p. 63
Economist (1995), The insatiable in pursuit of the unquantifiable, March 4th, p. 81-82
Heilemann, U. (1998), Die amtliche Statistik-Dienstleister in der Informations-gesellschaft [Official statistics
- providing a service in the information society]. In: Statistisches Bundesamt im Auftrag der Statistischen
Ämter des Bundes und der Länder (Ed.), 50 Jahre Bundesrepublik Deutschland: Amtliche Statistik - ein
konstitutives Element des demokratischen Staates. Wiesbaden, pp. 87-107.
Machin (1995), Statistical burdens on businesses. Paper presented to the 5th CEIES meeting, Brussels, 29
September 1995.
Schmidt, D. (2004), Staat und Statistik [The State and statistics] - Über Statistische Bureaus und statistisches
Wissen im neunzehnten Jahrhundert, am Beispiel des Königreichs Sachsen. Leipzig Dissertation.
Radermacher, W (2004), Demand, quality, burden: optimisation to balance interests. Paper presented to the
90th DGINS Conference, Paris, 13 and 14 September 2004.
20 27th CEIES seminai·: Producing business statistics - Costs and efficiency - Recent developments and experience
m
RECENT DEVELOPMENTS AND EUROSTAT'S STANDPOINT
Michel G laude
Director Internal-market, employment and social-affairs statistics
European Commission, Eurostat
1. Introduction
Before I get down to business I would like to thank the organisers for including the vital question of cost-
effectiveness in the production of economic statistics on the agenda for this seminar, since one of the tasks of
the CEIES is to help the Council and the Commission (in this case, Eurostat) to take greater account of users'
needs and the costs incurred by the suppliers ofinformation when drawing up Community statistical policy. It
is also important that the frequently-raised question of the statistical burden on businesses should be regularly
discussed in view of developments in demand and collection technologies.
The June 1995 conclusions of the Subcommittee on Business Statistics on this subject of the statistical burden
are very valuable here. We will discuss them later, but we can already say that, overall, while certain conditions
may have changed over the intervening period of almost ten years, many if not all of the recommendations are
still valid today: better justification of requests, better assessment of costs, improved coordination of
methodologies, regular review of programmes, keeping abreast of technological developments (electronic data
collection etc.).
Does this simply mean that we have not made enough effort to apply these recommendations?
I suggest, therefore, that we resume this investigation, taking as our starting point, as always, the needs and
requests for statistics expressed by the various parties involved. We will see in particular the great impact that
the "EMU action plan" of 2000 has had on the system of short-term economic statistics. Then, on the basis of
a more thorough analysis of the system for producing Community statistics (the European Statistical System )
and the interplay of the various operators, we will see how specific advances can be made at the various levels
of the system.
More particularly for the National Statistical Institutes, it will be vital to foster the exchange ofinformation and
good practice: collection via the Internet, the NSI commitment agreement [charte d'engagement] on good
relations with businesses etc.
At European level, various activities for improving the efficiency of the system will be proposed:
- at the methodological level, with the opportunity offered by the dissemination of the new accounting
standards and the implementation of business registers incorporating the group dimension;
- at the level of the statistical programme, by taking account of "negative priorities", designing European
samples (in connection with the "First for Europe" principle) and developing new "light" surveys.
- at the level of access to data, with the project for dissemination of data free of charge on the Internet.
2. Rapidly growing demand

It is a commonplace to say that quantitative information has become a vital ingredient for all the economic
operators in our developed societies, particularly as the speed at which information travels has been increased
thanks to the new technologies. At all levels, from managing directors to the mayors of major cities to the
European Council, the decision-makers want data that are reliable, up-to-date, consistent, intelligible,
comparable and suited to their needs. Let us, however, try to establish more precisely what the various needs
are.
2.1 At the macroeconomic level, the authorities responsible for economic and monetary policy need to keep a
very close eye on short-term economic developments and inflationary tendencies. In particular, under the
action plan associated with Economic and Monetary Union and on the basis of a comparison with the data
available in the United States, a whole series of short-term indicators were defined in 1998 (STS Regulation).
These monthly or quarterly indicators of activities, orders, staff, staff costs and production costs etc. are
collected at the 2-digit level of NACE. Even after five years of implementation, however, not all the data for
some Member States are available by the deadlines and this was some cause for concern at the ECOFIN
meeting of 2 June. Furthermore, the Council is currently examining new requests. This statistical burden, of
which the NSIs and businesses are very much aware, should, however, be considered in the light of the
importance of the use made of these data in connection with economic and monetary policy, which very much
affects the lives of all economic operators in the Union, and businesses in particular.
2.2 As regards the sectoral and structural polices of the Union, the demand for data has come to focus on the
annual monitoring of a series of indicators. These include the structural indicators accompanying the
implementation of the Lisbon process (March 2000, with the aim of making the European Union, by 2010, the
most competitive and dynamic knowledge-based economy in the world, capable of growth and sustainable
development with more high-quality jobs and greater social cohesion), and the indicators associated with the
e-Europe 2005 action plan. It should be noted that the importance of these indicators in terms of policy has
increased particularly as a result of the "open coordination method", which makes it possible to encourage
national social policy to move toward common objectives while observing the principle of subsidiarity.
However, many other types of sectoral data collected annually under the SBS Regulation are frequently used
for specific studies at both European and national level.
2.3 Requests by businesses, as relayed by the professional federations primarily concern "knowledge of their
market". This implies effective harmonisation and an adequate level of detail in the classification of activities,
and short return periods.' From this viewpoint, despite a positive verdict on the NACE 2007 revisions,
businesses have nevertheless expressed concern about the simplification of the PRODCOM statistics so
strongly urged by the NSIs. In addition to a better picture of sales and output, businesses also need the labour-
market indicators (working hours, costs, productivity etc.). Finally, in his last contribution to the 90th DGINS
conference (Paris, 13 and 14 September), Daniel Dewavrin, Chairman of the French Association of Industrial
Federations, added to this list the need for investment indicators - for example, on direct foreign investment.
2.4 Finally, the academic world is very keen to have microeconomic data in order to carry out the economic
studies needed for its activities since, in addition to the analysis of macroeconomic indicators we have
mentioned, the definition and assessment of public policies (such as the effects of R&D expenditure or
reductions in working hours on business activities) depend very much on studies by economists using complex
econometric analyses requiring individual data on businesses. Their needs are sometimes substantial, covering
several fields (accounting data, data on activity, employment, investment etc.) and relating to several years
(panel data). In addition to the problem of availability- often requiring file-matching - there is the problem of
the confidentiality of these data.
2.5 The question of priorities when considering these many and varied requests is absolutely crucial in view
of the limited resources available for meeting them.
In their contribution to the last DGINS conference in Paris, entitled "Demand, quality, burden: optimisation to
balance interests", W. Radermacher, J. Weisbrod and D. Asef of the German Federal Statistical Office discussed
this question on the basis of an appropriate diagram produced by S. Linacre'.
This diagram, which illustrates the relationship, where short-term statistics are concerned, between the
expertise and influence (on the work programme and methods of the statistical institute) of the various
' Many requests by businesses are at a very detailed geographical level (location of an establishment, for example). In such cases, as with other requests
by local decision-makers (impact of a business shutting down, creation of a free zone etc.), the NSIs are responsible for dealing with the requests.
1
Linacre, Susan (2001): "Understanding users and managing quality in a statistical agency". Q2001 Stockholm.
requesters, clearly shows that the most competent and the most influential are the Ministries (and the
Commission) and the central banks (including the ECB). At the other extreme (little expertise or influence) are
the general public and students. On the other hand, while academic staff have considerable competence but little
influence, professional associations and the media occupy the middle ground on both scales. It would also be
possible to add a time dimension to this diagram showing how these positions have shifted in recent years. With
the introduction of the euro the influence of the European Institutions has undoubtedly increased.
Ministries - j p τ τ
Influence
f' Elected pol.
' „representatives y■ i Banks }
ι Associations ..-'"Supranational "

institutions h
international '"
institutions
Public )
/ ■ " '
Institutes
Students
III Universities.: IV
Expertise
Is it possible to bring abut developments in this type of situation by defining and implementing different
priorities, and should this be done?
Where shortterm statistics are concerned, the completion of the action plan in connection with Economic and
Monetary Union will continue to be one of Eurostat's chief priorities, in view of the current situation in Europe
(particularly in the light of the statistics available in the USA) and the opinions expressed by the Council of
Finance Ministers.
This should not, however, prevent the European Statistical Office from devoting considerable efforts and
resources to meeting other categories of requests as effectively as possible.
3. The supply side: the European Statistical System (ESS)

After this first look at needs and priorities, let us now attempt a better analysis of the system for producing
European statistics its structure, the constraints and the roles of the various operators so as to provide a
firmer foundation for the ways forward that we will propose in the final section.
3.1 The European statistical system is highly decentralised, virtually all basic statistics being produced at
national or infranational level, and then aggregated at European level by Eurostat.
Coordination and harmonised definition of the concepts and classifications used by the Member States are
therefore vital, particularly as the diversity of the national statistical systems means that harmonisation must be
by product ("output harmonisation") rather than by collection method ("input harmonisation"). With this
approach based on statistical products rather than the production processes, great importance is also attached
to checking the quality of the statistics produced (reliability, timeliness, comparability etc.) on the basis of
"quality reports" by the NSIs, the dissemination of good practice among the Member States and quality control
at Eurostat by the various people responsible for the processing of the data.
27th CEIES seminar: Producing business statistics Costs and efficiency Recent developments and experience 23
m
3.2 Statistical harmonisation has by no means been achieved, and we should bear in mind that it is not always
the most recent arrivals that provide the worst harmonised statistics. In particular, the new Member States that
joined the Union on 1 May have made considerable efforts to incorporate the acquis communautaire. These
efforts have been successful and I would like to congratulate them publicly.
Moreover, concepts and definitions develop as a function of economic and social conditions. Thus the new
International Accounting Standards (IAS) interfere with statistics and disturb their comparability overtime and
space.
3.3 In the ESS, the tasks and responsibilities of all those involved are clear. Businesses are involved at both
ends of the production chain- as suppliers of the basic information (by replying to surveys and administrative
questionnaires) and as users of the resultant statistics. The national statistical systems produce the large part of
the national and local statistics in various institutional set-ups and provide elements that permit aggregation at
European level. In order to do this they already draw extensively on administrative data in addition to collecting
data from businesses. Finally, Eurostat coordinates and compiles the national data in connection with
Community decisions that take the institutional form of Decisions and Regulations (Framework Regulations of
the Council and the European Parliament and implementing Regulations adopted at Commission level) but also
draw on the "Gentlemen's agreements" concluded at the level of the Statistical Programme Committee.
This division of labour does not prevent the ESS from being prone to various tensions that are typically
encountered where the distribution and allocation of inherently scarce resources are concerned (budget, staff
etc.). The NSIs are therefore often confronted with major cuts in the their budgets and sometimes complain
about the impact that the Community work programme has on their own work programme.
3.4 It is mainly businesses, however, that regularly complain about a «statistical burden", which they regard as
excessive and detrimental to their competitiveness. Before discussing the precise extent of this burden, I would
like to make two remarks:
- Aren't businesses being a little inconsistent by asking to be regularly supplied with more economic data
at a fine level of detail that will enable them to make better decisions, while constantly complaining
about the burden of the collection required for producing these data?
Maybe the inconsistency should be made more specific in tenns of sectors and size classes, since the statistical
burden and the need for relevant data vary depending on the types of business concerned (small or large,
operating on a local or international market etc.). In addition, the people in businesses who are responsible for
providing data (and complain) are often not the ones who need figures. I hope that our discussions at this
seminar will enable us to clarify this debate.
- Could it not also be argued that since the activities of businesses affect the outside world so much (in terms
of employment and the environment etc.) that it is reasonable that they should provide the authorities with
sufficient information to enable them to draw up effective public policies in all fields and at all levels (from
local to international)?
3.5 Now let us return to the central issue of the costs connected with the statistical burden. This seminar should
enable us to clarify this question, since it has been examined in several countries in recent years. Without going into
too much detail, I can tell you from my previous experience in France that a distinction should be made between:
- collection ofinformation from administrative sources in general (i.e. including statistics) and a fair number
of surveys of various origins that swamp businesses without the latter often being able to distinguish between
compulsory surveys and optional ones;
and between:
- statistical requests and others within the field of collection from administrative sources so that the parties
responsible can be properly identified.
4. Coordinating efforts at all levels

However, even if the statistical burden is at the end of the day only a small part of the requests for information
addressed to businesses, this does not mean that it does not need to be justified on the one hand and
administered efficiently on the other. Quite the reverse!
There have already been efforts in this direction, but considerable progress still needs to be made at all levels
of responsibility for the ESS - particularly by improving the coordination of action at national (or regional) and
European level.
4.1 Priorities at national level

Efforts must focus primarily on those statistical services at national level that are in direct contact with
businesses:
- in order to reduce as faras possible duplicate collection ofinformation. We still sometimes see a new survey
being launched when administrative data on the same subject could be used. This increased use of data
already collected for purposes other than statistics should also be accompanied in the longer term by work on
developing these administrative data with a view to integrating statistical requirements more effectively when
legislation is recast or amended. A more ambitious aim that would make it possible to optimise the
administrative and statistical burden on businesses could involve improved coordination of all these
obligations (for example, by standardising the classifications and definitions, such as the concept of
workforce). Even if the NSIs are well placed - as specialists on the information systems - to contribute to this
coordination, the implementation of this kind of project goes far beyond the scope of the statistical set-up and
depends very much on the national contexts;
- i n order to justify requests for information more convincingly without relying exclusively on legal
obligations but trying to demonstrate the usefulness of collecting the data, occasionally referring to previous
results;
- in order to provide businesses with better information on collections planned for the future so that they can
form a picture of the burden they will have to deal with and adapt themselves accordingly. In a nutshell, there
needs to be a certain transparency in the lasting relations between partners;
- in order to exploit the new technologies for collecting and transmitting information so as to reduce the
workload on businesses and the associated costs. I will not go into these technological developments (such
as electronic transmission of data, collection via the Internet etc.) as various contributions to this seminar are
to be devoted to them, but it would appear that an intelligent application of technological progress would
make it possible to reduce the collection burden while improving the quality of the data collected.
- in order to improve feedback to businesses on the information collected, both in a general way, by sending
them extracts of the publications and studies produced, and in a more personalised fashion, by showing them
where their business stands in relation to others in the same sector or region. Even though this personalised
feedback using automatic data-processing software is prone to errors, businesses often welcome it.
All these ways forward (to which could be added optimised sampling so as to prevent the same businesses being
consulted for different surveys) should be exploited in the more general context of a Charter of good relations
between the statistical service and businesses (or a code of good practice), which could also serve as a basis for
organising institutional communication.
All these initiatives concern the national or infranational level, but coordination and exchange of good
practice could also usefully be organised between the NSIs with the active involvement of Eurostat.
4.2 Priorities at European level

Let us now consider the specific activities that need to be taken up or pursued at European level with a view to
reducing the burden and increasing the efficiency of data collection.
4.2.1 First, methodology

- Given the tasks of the European Statistical Office, harmonisation of definitions, concepts and classifications
is the top priority. Apart from being vital for the quality of the data produced, this harmonisation also provides
a guarantee of efficiency in production, particularly for multinational businesses, which would clearly have
difficulty in replying in each country in terms of specific classifications and concepts. Our efforts under this
heading will be concentrated on two activities in addition to the introduction by 2007 of the new version of
NACE/CPA and the development of data collection on branches of foreign firms in Europe and of European
firms in the rest of the world.
- producing registers of groups harmonised at European level. A proposal for a Regulation on this question is
currently being discussed at the level of the Statistical Programme Committee (SPC); this would make it
possible for confidential information to be exchanged between countries. This proposed Regulation would
enable administrative sources to be mobilised more effectively and statistical surveys to be better coordinated,
and would provide essential information on the demography of the groups.
It could also serve as a basis for direct collection from the groups of information on all their European activities,
thus simplifying their task of replying to surveys and improving the quality of collection. Responsibility for
such matters could lie with a "Centre of excellence" for a given Member State (in liaison with other NSIs) with
a view to better division of labour in the ESS.
- Monitoring the impact of the phased adoption of the new international accounting standards (IAS/IFRS),
which has already been felt by listed companies and is likely to spread rapidly to the others. The burden of
replying to surveys could be reduced in two ways - in terms of the level of detail required and through the
general use of a new system for data transmission (XBRL). A close eye must, however, be kept on the
implications for the quality of data, and Eurostat is considering concerted action at Commission level.
4.2.2 The second way of reducing the statistical burden is by revamping Community requests.
New requirements are emerging as a result of new Council policies, and these necessitate the collection of new
harmonised data. We have already mentioned the case of short-term statistics but the same is true in the field
of services or the new technologies. In this case, old priorities are becoming things of the past and the
legislation governing the corresponding collection needs to be reviewed.
The subject of "negative priorities" has recently arisen in the ESS. This is not an easy matter, however, (for
example, in the case of agricultural statistics, which are being left out of account) for a variety of reasons: there
is never complete agreement on what reductions are necessary and requests for eliminating items are often
largely attributable to very specific circumstances. A new "responsibility culture" needs to be developed
collectively. Work on simplification is well under way in the case of business data, for which the first proposals
from the Member States were discussed by a working party before being submitted to the user Directorates-
General of the Commission and finally submitted to the SPC.
- In the case of PRODCOM, allthe quarterly requests are to be discontinued (more than 1 000 series) and the
classification will be simplified (elimination of a hundred or so elementary headings). New proposals on,
for example, the monthly steel-production figures are currently being examined. This simplification is also
being carried out in conjunction with improving the quality of the remaining data.
- In the case of structural statistics, proposals are being drawn up with a view to striking a new balance in
industrial statistics, which have a long tradition of collection, and placing more emphasis on services, and
for other simplifications (elimination of optional variables, data collected at the level of kind of activity,
energy consumption etc.).
- In the case of short-term statistics, it is not so much a question of reducing requests, as mentioned, but of
an approach ("First for Europe") in which the criteria of statistical quality complement the strict
juridical/institutional approach. Two important innovations have recently been introduced:
•Applying a different treatment for the "small countries" representing less than 1% of the European
aggregate in terms of the level of detail required and the periodicity for supplying the results.
• A system of sampling at European level has been proposed for certain variables. By abandoning, in some
cases, the idea that representativeness is necessary at national level in favour of the European Union and
the Eurozone, the sample of respondents can be substantially reduced, thus reducing the statistical burden
on businesses and the NSIs.
- On the other hand, the negative priorities should make it possible to introduce new "light surveys"
corresponding to new economic concerns.
26 27 th CEIES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience
m
In order to obtain a better picture of the implications of the "information society", an information system has
gradually been set up ("e-action plan") comprising in particular surveys covering households as well as
businesses so as to produce a whole series of monitoring indicators. The same has been done in order to monitor
the effects of innovation and R&D on the growth and development of businesses with the introduction of the
Community Innovation Survey (CIS). More generally, Eurostat wants to develop this type of light survey of
businesses on emergent topics. In 2003 this was done for "relations between businesses" and a project on
"knowledge management" is under discussion for 2005.
4.2.3 Finally I would like to conclude this presentation with one last example of a way forward that concerns
the dissemination of data and their use by the various partners, including in particular businesses.
As from 1 October, Eurostat has made all its non-confidential data available free of charge at its website,
including in particular the "New Cronos" database, which comprises 4 500 multidimensional tables (detailed
classification, time series) plus the detailed external-trade and Prodcom data. The metadata needed for efficient
use and a user-friendly access system have also been reviewed. We hope that, among the many potential users
of the European statistics, businesses will be able to take advantage of this in order to target their strategic
choices and development more appropriately.
The dissemination of economic data on businesses always raises the crucial question of confidentiality. There
are strict rules on this point (see Regulation 831/2002), but one may wonder whether these rules could not be
relaxed a little in the case of data that are sufficiently old or made public through other channels. Finally, could
not file-matching (at the individual level that researchers are so keen to have) be better organised with strict
respect for confidentiality? All this will have to be further clarified and discussed with all the parties involved,
however, (including the Committee on Statistical Confidentiality).
Thank you for your attention. I look forward to your comments and questions. I will pay great attention to the
various presentations and the ensuing discussions and, at the end of the seminar, will try to comment on the
various points made, particularly as regards Eurostat.
27th CEIES seminar: Producing business statisücs - Costs and efficiency - Recent developments and experience 27
COST AND EFFICIENCY
MEASURING THE EFFICIENCY OF STATISTICAL SURVEYS
Bernd Gôrzig
Technical University, Berlin
Summary
Efficiency means the relationship between the output and input of a statistical survey. To measure this, it is of
paramount importance that the output and input indicators produced are factually correct and comprehensive.
The output indicators must be chosen so as to give a satisfactory description of the aim of the survey. The input
indicators must, above all, be complete.
There is no generally accepted measurement procedure to determine the efficiency of statistical surveys. The
DEA method used for comparable issues has not yet been applied to measuring the efficiency of statistical
surveys. Just like all other measurements of efficiency, this procedure does, however, require a clear and
comprehensive description of the objectives.
1. Preliminary remarks
Efficiency currently ranks as one of the key demands made of statistical surveys. For example:
• The President of the German Federal Statistical Office links the demand for efficiency in the official statistical
survey programme with focusing on users and minimising the burden on respondents (Hahlen, 2002).
•The Japanese «Statistics Council» calls for a reduction in the burden on respondents, the use of new
technologies, and simplification and efficiency in surveys (Statistics Council, 1995).
• The US Bureau of Labor Statistics (BLS) talks of using legislation to raise efficiency, reduce the burden on
respondents and improve the quality of surveys (BLS, 2003).
One could see these statements as just a general declaration of intent, using placatory words with the right
associations. What public body is likely to pursue its objective of carrying out the tasks assigned to it by society
in an inefficient manner? What these positions do, however, indicate is that efficiency is regarded as an
important objective and that this objective is generally mentioned in association with other objectives.
It does, however, remain to be seen what is meant here by the concept of efficiency. There is also sometimes
uncertainty as to the level at which this objective of efficiency is supposed to be achieved. Is the whole
statistical system to be organised efficiently? Or are only some individual surveys to be made efficient?
The question of how to measure efficiency cannot therefore be separated from the objectives which are to be
achieved by measuring efficiency. This report cannot take on the task of defining these objectives. As is so often
the case with societal issues, these are dependent upon political decisions. An attempt will, however, be made
to clarify certain aspects associated with the problem of measuring efficiency, and thus to indicate ways of
finding a solution.
2. What is efficiency?
There are different linguistic uses attached to the concept of efficiency. In technical terms, it expresses a
relationship between the use of resources and the result achieved. The idea of efficient action assumes that there
m
is a process by which a specific effect, the output, can be achieved through varying the level of the input. In this
sense, ISO standard 9000:2000 gives an initially neutral definition of efficiency as the «relationship between
the result achieved and the resources used». (ISO 2000)
Confusion arises when the term efficiency is also used in other contexts to denote the calculation of an optimum
outcome. The demand for efficient action in this instance describes all measures which lead to the attainment
of a specific objective, the output, using as few resources as possible. In this sense, everything which diverges
from an efficient outcome, is inefficient and thus wasteful.
When efficiency is described in terms of setting an objective and minimising the resources required to achieve
it, then we speak of the efficiency of expenditure. If the resources used can be expressed in value terms, the
concept of cost efficiency is commonly used. This paper will use the general term input efficiency.
i i
Output
-*«ί.
_ II III
-*«% _J l_J υ
Input efficiency
-««Si I 1" 111
.
-afe . . . 1l·III
- * ►
Input
Figure 1: Input efficiency

In order to increase the input efficiency of statistical collection, the measures proposed are primarily those
which rationalise statistical surveys. Survey costs need to be lowered by means of various technical measures
such as
• standardisation,
• the harmonisation of surveys,
• the shared use of data, particularly administrative data
• improved survey methods, through the use of modern I CT techniques,
• the greater use of on-line surveys,
• the use of sampling techniques.
This will result in greater efficiency without making any changes to the output.
m
The relationship between output and input can, however, also be efficient, in the sense that a maximum output
is achieved with a given volume of input. In this case, the terms used are efficiency of yield or the efficiency
of effect. The term used here will be output efficiency.
<
åi Λ å k h
'. ; ■
Output *o - |Mlu
c r—f
TJ
C
í—^
Φ
=ft
O
φ
3
O Ώ
["¿'t E'-fl Γ"
Q □ E
►
Input
Figure 2: Output efficiency
To increase output efficiency, the measures proposed include:
• the better exploitation of existing surveys,
• the development of new uses, particularly the greater evaluation of microdata,
• the extension of existing surveys to new reporting fields and items to be declared,
• the introduction of new surveys.
From a formal viewpoint, there is no difference between these two definitions of efficiency. On the other hand,
when assessing the efficiency of statistical surveys, differentiating between a maximising and minimising
objective is of some relevance. It is a matter of- political - objective-setting whether efficiency is achieved by
improving the level of performance from existing resources or seeking to attain a given objective using the
minimum amount of resources.
It is therefore important to read declarations of intent concerning efficiency very carefully. They do not
automatically express a commitment to cost-cutting, unless this objective is explicitly stated. Nor does the
measurement of efficiency as the relationship between output and input say, in itself, to what extent input is to
be minimised or output maximised.
From a scientific viewpoint, it is not possible in many cases to obtain an unambiguous measurement of
efficiency unless clear objectives have been set. If there are different processes available to describe the
respective combinations of input and output, then all that can initially be established is to what extent these
processes are more advantageous as regards input or output efficiency.
Microeconomic production theory has developed an instrument in the form of the frontier production function
with which to determine all those combinations of input and output which can be classified as equally efficient
in terms of optimisation calculation. (Fried/Lovell/Schmidt, 1993) This is the so-called efficiency frontier.
30 27th CEIES seminai·: Producing business statistics - Costs and efficiency - Recent developments and experience
¿L
Increased output
efficiency
Output
O*
Input
Figure 3: Efficient and inefficient processes
3. Absolute and relative efficiency

It is doubtful just how possible it is in practice to determine an absolute level of efficiency for a survey. In
most cases it will only be possible to determine an achievable increase in efficiency undergiven conditions.
In this respect, most statements about efficiency are relative, i.e. the aim is not to determine an absolute level
of efficiency. In fact, when statements are made about the efficiency of a statistical survey, the data on the
respective level of efficiency are linked to a reference value.
This reference value can originate from different sources. A preceding period is very often selected for
comparison to illustrate an increase or reduction in efficiency. It is common practice to prove that efficient
action has been taken by showing that cost savings have been made possible. Reference is also often made to
improvements in quality and additional uses for surveys.
"Benchmark analyses" make use of a comparable survey as the reference value. Comparative analyses with
other surveys are used to show to what extent efficiency can be increased. At present, such comparisons are
made primarily between European and US surveys, especially with respect to the timeliness parameter, which
is important for observing the economic cycle. (Statistical Board, 2002)
4. A simple measurement of efficiency

The corresponding definition forms a useful basis for a ratio to measure efficiency. The ratio should provide
quantitative information on the extent of the relationship between output and input. If the ISO definition of
efficiency is used, then one can imagine the initial measurement to be very simple:
„r~ . Output
Efficiency—j^.
The higher this figure is, then the more efficient the process described. Sometimes this measurement is standardised
by setting the highest attainable efficiency at 1. In this case one can also talk about the degree of efficiency.
As long as output, and input, can each be described by a single parameter, this sort of ratio is well suited to
describing efficiency. Output and input can also be of differing dimensions in this calculation. They can,
depending on the issue, be defined by quantities, specific ratios or values.
In practice, however, it is not so easy to measure efficiency. In particular, the specific form of the ratios to be
used for output and input requires careful analysis.
4.1. T he objectives of α survey

The output ratios should describe the aim of the survey. As already explained, societal objectives can seldom be
reduced to a single target parameter. The description of output ratios can turn out to be just as difficult. The aims
can clash with each other and there is also often a hierarchy of objectives. A fairly loosely defined onedimensional
objective is made up of less vaguely formulated partial objectives. These can also be at odds with each other.
27th CEIES seminar: Producing business statistics Costs and efficiency Recent developments and experience 3J_
If the demand for an optimum survey which meets all requirements is broken down into individual components,
the following list of objectives may, for example, be identified:
• User friendly,
• Burden reducing, and
• Efficient.
In practice, it is likely to be difficult to attain all these objectives. A user expects to get as much information as
possible from a userfriendly survey. This would lead to a significant expansion of the survey programme. A survey
which minimises the burden on respondents, on the other hand, is more likely to require cutbacks in the survey
programme. In order nonetheless to pursue these two objectives simultaneously, the third objective of efficiency or,
to be more precise, increased efficiency, may be brought into play. Increasing output efficiency implies that user
friendliness has to be improved without increasing the burden on respondents. Improving input efficiency could
reduce the burden on respondents without detracting from userfriendliness. Unless additional targets are set,
however, it is not possible to resolve the issue of how both aims are to be achieved at the same time.
Optimal survey
Output efficiency J) (Γ Input efficiency
User- ,_„. Burden-

,. .. Efficient ,
friendly reducing
Figure 4: Hierarchy of objectives and conflicting objectives

The targets have to be identified depending on the level at which efficiency is measured. This is generally easier
when the efficiency of a specific survey is to be measured rather than that of statistics as a whole. In no way do
efficient individual surveys also guarantee the efficiency of the entire statistical system. The efficiency of the
statistical system as a whole also depends upon how efficiently the individual surveys are linked to each other.
However, even if only specific surveys are to be measured for efficiency, it is not easy to clearly define the
target, i.e. the output. It is often the case that one single survey is used for several objectives. It should also be
taken into account that survey results can be put to different uses, due to the fact that the statistical surveys are
used by users with different interests.
It should, for example, be pointed out that a large number of the current statistical requirements in the EU
Member States are defined by corresponding EU Regulations. The Federal and Länder Committee on reducing
bureaucracy in German statistics highlighted in its report that changes in the survey programme had to be
viewed in the context of demands made by the national accounts (Federal and Länder Committee, 2004). It also
has to be taken into account that surveys are then processed in very different ways in official statistics. Even
within the field of official statistics it is therefore possible to identify users with differing priorities.
Alongside these requirements emanating from official statistics, there are however many other demands made
specifically of business surveys. In particular, these have to provide information for businesses and business
associations. With respect to the demand for userfriendliness, the field of research should also not be
overlooked, whether statistical surveys are intended to serve independent research objectives or are directly
connected with the provision of economic policy advice through research.
In addition to these multiple objectives, it should also be bome in mind that almost every statistical survey
consists of a large number of individual indicators. In production theory terms, this is known as coproduction.
If the efficiency of each output indicator were to be analysed separately, then it would need to be indicated
which component of input was to be ascribed to the respective individual output indicator. On the other hand,
the overall efficiency of a survey can generally be increased by the simultaneous collection of characteristics,
because fixed expenditure is also incurred in every survey, regardless of the number of output indicators.
32 27th CEIES seminar: Producing business statistics Costs and efficiency Recent developments and experience
4.2 Efficiency and effectiveness
It is also necessary to be clear about the objective of the survey in order to ascertain whether the issue at stake
is really one of efficiency. Efficiency relates to the relationship between expenditure and income. If, on the
other hand, the extent to which objectives are achieved is to be determined, then this is a question of
effectiveness. Particularly when the targets set are unclear, it becomes difficult to separate the measurement of
efficiency from the issue of effectiveness.
If, for example, the aim is to cover all economic activities in a country's economy, then this will clearly not be
achieved if the survey relates only to industry. This survey might in itself even stand out on account of its
exceptionally good expenditure-income ratio. For a whole range of economic statistics, it may suffice to
conduct surveys only on industry. It is, however, essential for a large number of other objectives - and
particularly those concerning economic policy - to also include in the survey those parts of the economy which
do not belong to the industrial sector. In this respect, even though a survey on industry may well be carried out
efficiently, it may not be particularly beneficial.
This distinction is, above all, also important for interpreting results. Unfortunately, it does often happen that
partial surveys are also used to draw conclusions for the whole economy for the simple reason that there are no
others. In this respect, declarations of intent on raising efficiency are only of any value when accompanied by
a list of the objectives to be achieved.
4.3 Exhaustive/less of the input indicators

In general, it should be less problematic to establish input indicators to determine the efficiency of a statistical
survey than to establish output indicators. Care should be taken however to ensure that the input indicators are
exhaustive. If an output can be determined with several inputs used to varying degrees, then a distinction must
be made between partial efficiency and overall efficiency. If the aim is to determine the overall economic
efficiency of a survey (Federal and Länder Committee, 2004), then it is necessary to cover all the necessary
inputs. This must be done regardless of the economic unit involved.
It would be easy to raise the partial, commercial efficiency of a statistical office if its own input was reduced
to be replaced by inputs produced by other units. This could, for example, be done by passing some of the input
to be provided on to the parties obliged to report information. If this results in an improved division of labour,
then it is also conceivable that the overall efficiency would be improved by shifting the burden in this way. This
can, however, only happen if the additional objective of minimising the burden is not included among the output
indicators and does not therefore come into the efficiency measurement.
If administrative data are used in the place of their own surveys, the partial efficiency of a statistical office can
also be increased, although it is not possible to obtain administrative data without incurring some costs. As with
a statistical survey, there is some expenditure involved - both for the parties obliged to report information and
for the body collecting the administrative data.
The reason usually given for using administrative data is to avoid the unnecessary duplication of surveys. This,
in itself, would certainly also seem to improve overall efficiency. It cannot, however, be assumed that this results
in an automatic increase in the overall efficiency of collecting information. The advantages of the synergies
produced by the fact that the administrative data are used for several purposes are offset by the disadvantages
of the data not generally being produced in accordance with statistical rules. This can, for example, mean that
the output does not meet the quality criteria required of a statistical survey. Nor is there any a priori guarantee
that the administrative data have been collected efficiently.
5. Extending the measurement of efficiency

Once the respective inputs and outputs for a statistical survey have been clearly defined, then the above-
mentioned measurement of efficiency could be used even in the case of multiple inputs and outputs, if these
could be added up. Given, however, that output and input indicators often have different characteristics and are
of differing dimensions, a simple addition is only possible in a few cases.
This is possible when a weighting can be found for the respective indicators. In this case, the efficiency of a
statistical survey could be described by the following equation:
27tli CEIES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience 33
Fff · _ O u t P u t _ Weighted sum of outputs ,
Input Weighted sum of inputs '
Xvi*xi Xt Output/; l/. Weight of output ƒ
Xuj*yj' yi Input/· u Weight of input j
j ' '
One of the main tasks in the quantitative assessment of an efficiency figure is therefore to determine the
weights for the input and output indicators used. Prices could, for example, be used as weights, chiefly for those
input indicators for which monetary expenditure is incurred.
This method should not, however, be confused with the standard calculations often used in administrations and
businesses. If the value of the output is, for example, determined by the value of the expenditure, any further
measurement of efficiency is superfluous. The only meaningful evaluation of output can be one which is
independent of the inputs, e.g. through the price obtained on the markets.
In the case of outsourced expenditure, such as certain services used by businesses, the expenses paid to these
companies could provide some information on the weighing. In many cases, technical conversion factors may
also be available. Sometimes experts might be asked to carry out the weightings. This should even be
compulsory in the case of output indicators, when fixed targets are set. In this case, the weighting reflects the
externally-defined prioritisation of objectives.
6. Formalised procedures
In recent years, increasingly formalised procedures have come to the fore for analysing the efficiency of
processes. One procedure which is regularly discussed in comparable contexts is the DEA (Data Envelope
Analysis, 2004) method. This is an instrument to measure the technical efficiency of organisational units. It is
used for analysis in a range of very different contexts, such as, for example, measuring the efficiency of non
profit organisations, public authorities and business units not directly involved in the marketplace, as well as
for the evaluation of research.
It is a nonparametric procedure, in which the respective units or processes are compared to each other and
related to the best process in each instance. The procedure is based on a linear programming approach. Use of
the procedure requires the presence of a set of comparable processes with different input and output indicators.
Max E0 Σ« J
* ■>ν
for each process τ.
'Χ,
subject to 1
=s 1 for each process x,
"Vi
with weights u r v. a Ö.
Efficiency frontier
Output
Increased
output efñcienc
Ρ
Inefficient processes
/increased input 'I j ^ Q
-.iennv
efficiency ·
Input
Figure 5: Frontier production function
In terms of its general approach, there would seem to be no reason why this procedure could not also be used for
determining the efficiency of statistical surveys. No examples of this could, however, be found in the relevant
literature. Nor can it be assumed that this procedure can lead to a clear and universally accepted solution. Given
the multitude of different variants to the model, use of the approach does not just require an informed choice of
input and output indicators, but also selection of the appropriate model. Some authors have also questioned the
general applicability of this approach (LiebDoczy, E., G. Shuttleworth, 2002). The DEA approach would seem
to hold a number of advantages in comparison to the regression approach also sketched out in the literature, the
most important of which is the relatively low demands made regarding the number of comparable processes. It
also highlights the possibility of being able to establish a separate weighting scheme for each process.
Bibliography
BLS, 2003, Compensation and Working Conditions Online, "Confidentiality Information Protection and
Statistical Efficiency Act of 2002", by Jeffrey L. Schildkraut.
BundLänderAusschuss Statistik (Federal and Länder Committee on Statistics), 2004, Bericht des Bund
LänderAusschusses Statistik an die Wirtschaftsministerkonferenz der Länder zur weiteren Möglichkeit des
Bürokratieabbaus aus dem Bereich der Wirtschaftsstatistik (Report to the trade and industry ministers of the
Länder on further possibilities for reducing bureaucracy in the field of economic statistics), Bonn.
Charnes, Α.; Cooper, WW; Rhodes, E., 1978, Measuring the efficiency of decisionmaking units, in: European
Journal of Operational Research, Vol 2, p. 429444.
DEA, 2004, http://www.deazone.com
Fried, H., K. Lovell, S. Schmidt, 1993, "The measurement of productive efficiency", New York.
Hahlen, J., 2002, Press conference held on the occasion of the submission of the report from the Statistischer
Beirat (Statistical Board) to the Federal government on 31 July 2002 in Berlin.
ISO, 2000, http://www.iso.org
LiebDoczy, E., G. Shuttleworth, 2002, Sinn und Unsinn des Benchmarking (Sense and Nonsense in
Benchmarking), NERAEconomic Consulting.
Statistics Council, 1995, New Strategies for Government Services for the Coming Decade. Report to the
Director General of MCA.
Statistischer Beirat (Statistical Board), 2002, Empfehlungen zur Weiterentwicklung der amtlichen Statistik
(Recommendations for the further development of official statistics), Report to the F ederal government,
Wiesbaden.
COST AND EFFICIENCY
PRODUCING BUSINESS STATISTICS - COST AND EFFICIENCY
PERCEPTION BY THE BUSINESS WORLD
Antonios Tortopidis
Coordinator, Research and Analysis, SEV Federation of Greek Industries
Businesses are the single source ofinformation for the production of business statistics. They devote time and
money in order to contribute accurate information. The questions raised are:
• First, what is the relevant cost?
• Second, can it be reduced?
• Third, can there be a benefit?
The cost of providing business statistics

There are direct and indirect costs in providing information. Often, what counts is not only the absolute value
of the costs. The perception of the costs may also define a positive or negative attitude on the part of the
contributor of the information.
Direct costs
Providing information to public statistical authorities usually takes the form of an obligation in the sense that
there can be repercussions in the case of non-compliance.
Every year, an average medium sized manufacturing company has the obligation to answer approximately ten
questionnaires sent by statistical bodies for strictly statistical purposes. Of these,
• one is repeated every month,
• two are quarterly questionnaires,
• seven are annual.
That means that, on average, it has to fill in close to 2.2 statistical questionnaires per month.
On the assumption that the responding company has three discretely located facilities, an analysis of these
questionnaires leads to an estimate that it has to complete a total of 158 pages of questionnaires per year, with
approximately 50 entries per page On average and a total of 7,871 entries per year.
The time required to fill in these entries is estimated at 0.35 person-years per year. This is a considerable but
not prohibitive cost item. Businesses generally comply with the required data provision.
Indirect costs
Is data provision for statistical puiposes risk free?
Not entirely. For an experienced analyst, in certain cases, individualised business information can be
approximated by using branch/regional business statistics, especially when dealing with branches/regions
characterised by a limited number of companies.
m
On the other hand, business statistics are sometimes required before they are finalised or before they are made
available to the public at large. In such instances, especially for companies listed in the stock exchange, there
is always the risk of being accused for providing information possibly misleading or favouring interested
stakeholders.
Undertaking such risks constitutes an element of indirect costs.
'The perception of costs
Enterprises provide business information fora number of public entities, other than statistical bodies. Such
entities include:
• Stock market authorities
• Internal revenue and VAT authorities
• Labour and social security bodies
• Various Ministries
• Central banks
• Local and regional government authorities
Responding to such obligations adds to their costs. Providing business statistics is only part of the burden for
complying with regulatory red tape. As a result, the perceived cost is higher than the actual.
Furthermore, enterprises often have to provide statistics in periods when their relevant business units are under
pressure. In such cases both the actual (due to overtime required) and the perceived costs are higher.
Reducing the cost

Avoiding duplication
Providing business statistics is based on stand alone or integrated information systems. If the required
information changes frequently, additional time and costs for adapting these systems are necessary.
Basic business information is provided in formal statements that enterprises have to submit to government units
periodically. There is no need to resubmit such information for statistical purposes.
Making questionnaires clear and corresponding lo business accounts
Responding to statistical questionnaires takes more time when questions are not straightforward or
standardised. At minimum, standard forms may be used. Any extra detail should be asked only if it is
sufficiently justified.
Avoiding frequent changes
Are there business benefits?

At their discretion, businesses provide information to bodies, such as:
• Commercial banks
• Business Information data banks
• Universities and their students
• Public and private research centres and companies
• Private researchers
• Regional, national or international business organisations/ associations
•The public at large
They do so because they can derive a benefit, direct or indirect, immediate or distant, quantitative or qualitative.
m
They cannot always determine such benefits from the information they provide to official statistical
questionnaires. The design of these questionnaires usually reflects needs of government bodies and neglects the
needs of the enterprises. Furthermore, the results from data processing of such questionnaires are often
produced with time delays that make them obsolete for business needs.
Improving the relations between enterprises and statistical departments

Arising from the considerations above, here are some ideas to improve the co-operation with business:
• Care to formulate clear and standardised questions.
• Avoid frequent changes.
• Avoid requiring data in periods of pressure for the enterprises, mainly the first three months of the year.
Allow a reasonable time to comply.
• Make contacts and requirements more business friendly - explain clearly the purpose of data collection
- avoid asking for data of marginal value for policy formulation.
• Distinguish business statistics requirements from other regulatory red tape. Make it look more like a
business relation.
• Ensure compatibility of data required with data readily available in business accounting systems and
standards.
• Provide responding businesses with privileged access to information on the progress of data processing
and the intermediate, provisional and final results available for their use.
• Cooperate with business organisations in order to enhance the usefulness for businesses of the results
from the processing of the information provided.
• Produce meaningful and timely results
38 27th CEEES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience
COST AND EFFICIENCY
PRODUCING BUSINESS STATISTICS - COSTS & EFFICIENCY
PERCEPTION OF THE BUSINESS WORLD
Tony Donohoe
Head of Research and Information,
Irish Business and Employers Confederation
Summary
• European-based businesses face much higher costs through regulation than those in the United States or in
the Far East. However worthy reams of red tape may seem in theory, in practice they are undermining
competitiveness and costing jobs. It is in this context that the issue of the statistical burdens on business must
be considered.
•The burden varies considerably between companies, depending on the nature of the business as well as the
size of the enterprise. It also varies between countries. In small economies the NSIs have a relatively limited
business sample and are forced to survey the same companies. Small firms also bear a disproportionate
burden.
• Consultative procedures that establish effective user/producer dialogues are now pre-requisites for NSIs to
ensure that they are producing the statistics required by users and disseminating them in a format that
enhances their usability.
• Statistical offices need to adopt a 'whole system' approach, involving all areas of the public sector where
administrative records are maintained and from which official statistics can be generated.
• Apparent or actual duplication of requests for the same information is a particular source of irritation for
business. Therefore the use of integrated data sets is the key to adding value to the statistics that can be derived
from administrative records.
• Companies will have very different perceptions of the relevance of statistics. Large companies that are
competing in international markets are more likely to be interested in trends. They are also more likely to
appreciate the importance of statistics for the development of economic and social policy.
• Most official statistics are not produced to answer the specific requirements of business. This is not their
function. The primary role of the NSIs and Eurostat is to provide data so that national governments and
European institutions can define, implement and analyse policies. The statistics used by business are in many
ways a by-product of this process. However the uses of official statistics are becoming more diverse and there
needs to be a reorientation of official statistical systems to meet these broader needs.
• A number of important gaps, particularly around the inter-related areas of the new economy and globalisation,
have developed within the last decade. First steps in this process are under way in the NSIs and Eurostat but
the sheer pace of change means that there is an increasing risk that official statistics may 'miss' the
information economy.
•The high commitment to accuracy and the rigorous standards maintained by the NSIs and Eurostat is
acknowledged and respected by the business community. Business users tend to be less satisfied with the
timeliness of statistics.
• Unfortunately for statistical offices, the 'bar' continues to be raised in terms of customer expectations. Users
now expect greater flexibility and to have interactive tools to produce their own views of the underlying data.
• Our US competitors have access to a huge wealth of easily accessible, relevant and timely business statistics,
freely available at their finger-tips - and in a format that the user does not have to be a statistician or economist
to assimilate. European businesses require similar services.
PRODUCING BUSINESS STATISTICS - COSTS & EFFICIENCY

PERCEPTION OF THE BUSINESS WORLD
Men are disturbed not by things, but by the view which they take of them.
Epictetus. (55 AD - 135 AD)
Introduction
In 1995, the CEIS sub-committee on business statistics concluded:
The load of statistical surveys on business is small by comparison with the costs of complying more genemlly
with administrative requirements of government. The burdens are nevertheless perceived to be serious'
The committee also proposed a comprehensive list of measures to address tins issue. They included use of
alternative administrative data, fair distribution of the reporting burden, dialogue with data providers,
harmonisation of classifications, use of electronic data collection methods and advice to companies on uses of
statistics
This paper sets how to assess, almost a decade later, business continues to perceive the burden and benefits of
official statistics. As the paper's title and the quotation at the top of the page emphasise, whether businesses see
a survey as a burden or not will be a matter of perception. However it is a perception that is based on reality. It
would be incorrect to assume that companies view official surveys - particularly those that are reinforced by
statutory orders - any differently than the other regulatory burdens to which they are subject. European-based
businesses face much higher costs through regulation than those in the United States or in the Far East. However
worthy reams of red tape may seem in theory, in practice they are undermining competitiveness and costing
jobs. It is in this context that the issue of the statistical burdens on business must be considered.
Notwithstanding the above, the reporting burden is considerably less onerous if the benefits can be identified.
Therefore, the second part of the paper provides a business perspective on official statistics in terms of their:
• Relevance
• Accuracy
• Timeliness
• Accessibility
Reporting burden
A considerable amount of research has already been undertaken to measure the cost to business of responding
to government requests for information. Judging by the widely varying estimates, this would not appear to be
an exact science. However, it is clear that the burden varies considerably between companies, depending on the
nature of the business as well as the size of the enterprise. It also varies between countries. In small economies
like Ireland, for example, the Irish Central Statistics Office (CSO) has a relatively limited business sample and
is forced to continuously survey the same companies.
' Statistical burdens on businesses Report of the CEIS Business Statistics Sub-Committee June 1995
m
There appears to be a consensus that small firms bear a disproportionate burden. For businesses with less than
20 employees, the compliance costs borne are least 35% higher than for firms with more than 500 staff. This
figure is regarded as an absolute minimum. In the majority of cases the costs borne by small firms are relatively
much higher. Nor can the National Statistical Institutes (NSIs) afford to be sanguine about their contribution
to this burden. While statistical enquiries only account for a fraction of the total administrative burden, almost
five percent (and 17.2% in the 50+ employees category) of respondents to a UK survey3 identified this as the
area of red tape that had caused the biggest increase in the previous year in compliance time. NSIs in a number
of European countries have also reported evidence of survey fatigue and downturns in response rates.
A strategy to alleviate the statistical burden and improve response rates was well articulated by CEIS in 1995.
However, the implementation of certain aspects of this strategy has been slow in many European countries. One
obvious example is the relatively slow adoption of IT to optimise the burden on individual respondents. At a
minimum, all NSI's should have data management systems that would allow the collection of data from a
variety of electronic sources and integrated survey management systems that would identify the surveys each
company has been included in.
Businesses perceive the statistical burden to be heavier when they have no interest in the results or do not
understand the purposes of the survey. There are basic 'housekeeping' measures that will improve co-operation
such as ensuring the survey is directed at the appropriate person, giving a full explanation of the survey's
purpose, and providing confidentiality assurances. However NSIs could go further than this by automatically
providing results to respondents (again, technology may have a role to play here).
Consultative procedures that establish effective user/producer dialogues are now pre-requisites for NSIs to
ensure that they are producing the statistics required by users and disseminating them in a format that enhances
their usability. This happened in Ireland during preparations for the new National Employment Survey (see
below) when the Irish Central Statistics Office (CSO) conducted intensive consultations with businesses in the
development of the survey. From a business perspective (and hopefully from the CSO's perspective), this was
an extremely worthwhile process.
Statistical offices should remain highly aware of the data demands that they place on their respondents,
particularly smaller businesses. The statistical reporting burden on business may be quite modest in the context
of the overall information demands made by public bodies but it attracts considerable criticism as the
information required is often relatively complex and must be specially extracted.
NSIs can take many steps to minimise the reporting burden:
• Use statutory inquiries only for censuses or when high response is essential;
• Accept best estimates if exact figures are not readily available;
• Whenever possible, inquiries should be based on samples, so that a maximum number businesses are
excluded from the need to respond;
• Reporting thresholds (e.g. enterprises employing more than 10 persons) should be used in many inquiries,
thus exempting small enterprises from reporting;
• Field representatives should be used to help businesses in the completion of returns.
Administrative records
While inherently important in itself, the issue of response burden has consequences for survey response rates
and hence on analysis and interpretation of survey results. As well as spreading response burden as fairly as
possible, it is important that NSIs are in a position to demonstrate that the information collected is used
efficiently and effectively. In this regard, it is vital that the statistical infrastructure is developed by reference
not just to statistical surveys, but also to the large amount of administrative data that is in existence now or that
will exist in the future.
' Regulatory Burdens of Small Business - Chittenden, Kauser and Poutziouris. Manchester Business School, October 2001
1
Small Business Research Tnist. (2000) 'NatWest SBRT Quarterly Survey of Small Businesses in Britain', Vol. 16. No. 1. SBRT.
•uro·»
Again, this point was highlighted by the CEIS in 1995 and has appeared in numerous NSI strategy documents
since then. However, outside Scandinavian countries, it appears to remain an aspiration. This is not to under-
estimate or over-simplify the obstacles and drawbacks associated with this approach. However, the arguments
in its favour are compelling. The CSO has started to embark on this route with the adoption of a 'whole system'
approach, involving all areas of the public sector where administrative records are maintained and from which
official statistics can be generated.
The approach4 adopted here involves several key changes, some of which build on what has been evolving in
the statistical arena for the past decade. For example:
• All surveys funded by government departments and agencies will from now on be planned in partnership with
the CSO, should be integrated, where relevant, with CSO surveys and be treated as part of official statistics;
• Data from administrative sources will be developed in a very conscious manner as part of a total data system;
and
• Data users and producers must take joint responsibility for ensuring that the data collected are those which
meet important information needs; and serious and substantial analyses of these important data are
undertaken so that society gains fully from the endeavour.
Irish business has welcomed this as an extremely positive development and we are already seeing its potential.
In recent months, for example, an inter-department group has already identified close to 200 potentially
important data sources for social 'statistics within the government departments and agencies it reviewed. The
potential for using data collected by taxation authorities, in particularly, is enormous. Obviously the wider use
of administrative datasets for statistical purposes will require full safeguards in respect of privacy.
Integration of data sets

Apparent or actual duplication of requests for the same information is a particular source of irritation for
business. Therefore the use of integrated data sets is the key to adding value to the statistics that can be derived
from administrative records.
A practical example of how integration could work is provided by the CSO's recent New Earnings Survey
(NES), which is an integrated survey of employers and employees. Data is collected directly from both. In the
survey currently under progress, about 7,000 private sector businesses as well as almost all public sector
organisations have been included. These organisations provide information at the level of the organisation as
well as earnings, hours worked and occupational and other details for a sample of employees. Each sampled
employee is then sent an individualised questionnaire, which asks for characteristics such as age, sex, family
status, working patterns etc.
A primary focus of the NES is employee earnings, but the survey has been designed to have the potential to
cater for other topics - both of a quantitative and qualitative nature. It also has the potential to act as a link
between the many business surveys already conducted and household surveys at the micro and macro levels are
envisaged. At the micro employee level, the collection of the Personal Public Service Number (PPSN) which is
issued to all Irish residents will facilitate direct linkage to other household surveys which may in the future
collect the PPSN as well, of course, as a linkage to administrative sources containing the PPSN.
At the macro employee level, the use of common classifications in the NES and household surveys will provide
a valuable method of analysing the situation of groups and sub-groups of the population. For businesses, direct
linkages with other CSO surveys will be possible. For instance, economic variables relevant to business
performance would be available alongside other organisational characteristics and practices.
Relevance
The question of relevance is at the heart of any discussion about the statistical burden. Companies will have
very different perceptions of the relevance of statistics. Large companies that are competing in international
markets are more likely to be interested in trends. They are also more likely to appreciate the importance of
4
Strategy for Statistics 2003-2008, National Statistics Board. Ireland, July 2003.
statistics for the development of economic and social policy. Small firms operating in local markets are less
likely to the benefit for their businesses.
There are as many business reasons for interpreting official statistics as there are published statistics but these
are just a few reasons why businesses use them:
• Measure companies and their products.
• Measure business costs.
• Analyse the market potential for their products.
• Judge if the time is right to proceed with a new capital investment, launch a take-over, or move into a new
market.
• Get the best returns on their investment.
• Get a better understanding of how an economy is performing.
• Judge the government's economic policies.
• Obtain a feel for an unfamiliar economy.
• Compare several countries.
• Make a forecast.
Most official statistics are not produced to answer the specific requirements of business. This is not their
function. The primary role of the NSIs and Eurostat is to provide data so that national governments and
European institutions can define, implement and analyse policies. The statistics used by business are in many
ways a by-product of this process.
The business community recognises, more than most, that the demands of evidence-based policy-making are
creating new pressures for data to allow policy-makers to analyse and understand the complex and fast-
changing environment. However the uses of official statistics are becoming more diverse and there needs to be
a reorientation of official statistical systems to meet these broader needs.
Because Ireland is exceptionally open in terms of international trade and has a very high level of foreign direct
investment (FDI) in the economy, the international context is especially important for us. Ireland is virtually
unique in Europe in the scale of the importance of "new economy" activities in both manufacturing and
internationally traded services and in having a strongly dualist manufacturing sector. So the increased
fragmentation of global production impacts in a magnified way in the case of Ireland.
A number of important gaps, particularly around the inter-related areas of the new economy and globalisation,
have developed within the last decade. First steps in this process are under way in the NSIs and Eurostat but the
sheer pace of change means that there is an increasing risk that official statistics may 'miss' the information
economy.
New economy: The world economy is becoming significantly more ICT intensive with major effects on other
technology fields and the speeding up of economic development processes. This has led to (i) a rapid expansion
in products and services available to consumers and business and (ii) changes in the way consumers behave and
interact. These issues were highlighted at the 2002 London Conference of the International Association for
Official Statistics. Delegates warned5:
• Statistical needs exist in significant areas of ICT related capital formation, such as accounting for software,
and the wider category of intangible assets, which clearly fall within any definition of the 'knowledge based
economy'.
• ICT has also changed the economic characteristics of assets as inputs to productive processes. For example,
the behaviour of knowledge as an asset, which can gain value in use, is very different from that of physical
assets.
1
Official Statistics and the New Economy; Report of the 2002 IAOS conference, UK Office for National Statistics
• New dimensions of measurement are also needed in assessing the effects of expansion in communication, as
the 'networked society' allows more interactions - both social and economic - between individuals.
Globalisation: the term 'globalisation' is used to describe a range of changes in the way the international
economy works. It includes the behaviours of firms which operate across national boundaries, the increasingly
ability of consumers to access international supplies, and the decreasing importance of geography in the
choices firms make about where to carry out specific parts of their operations and what operations they do
themselves. This raises a number of key issues6 including
• Understanding the 'disintegration of business operations, including splitting of value chains across borders;
• Limitations in data derived from single country views of larger, multinational operations;
• International movement of intangibles, and of investment by household and firms.
Services sectors: Services output and prices have traditionally been a weak link in national accounts, yet
economic activity continues to be more heavily concentrated in the services sector. For example, services now
account for over 67% of all employment in the EU compared to 60% ten years ago. Manufacturing employment
has declined from 33% to 28%. The relative importance of the two sectors is not reflected in official statistics,
which continue to have a strong manufacturing bias.
Consumer trends: there is a strong producer bias in official statistics. While houschold'consumer expenditure
surveys are more expensive and resource-intensive to conduct, official statistics should reflect our consumer-
driven economies. Business has an obvious interest in this data, but there is also a more general requirement to
understand and track the changing consumer behaviour.
Geographic breakdown: since the changes in most economies are not evenly spread across the country, national
averages are no longer enough to understand the economic changes in different areas of a country. As a
consequence, there are new demands for detailed geographic breakdown of the full range of economic data.
Accuracy
The high commitment to accuracy and the rigorous standards maintained by the NSIs and Eurostat is
acknowledged and respected by the business community. The dismantling of trade barriers leading to the free
movement of goods and the size of the informal economy in some countries conspire to make measurement
increasingly difficult. However, most of us who have had dealings with our NSI would acknowledge the
technical expertise and integrity of the staff.
Timeliness
Business users tend to be less satisfied with the timeliness of statistics. Obviously, this reflects the classic
conflict facing a statistical office in attempting to simultaneously satisfy two dimensions of statistical quality
- speed and accuracy. However, the time lag - particularly around one-off or less regular household based
surveys - limits their value.
Accessibility
Improved dissemination of official statistics through the web is the single most important development for
business users within the last decade. At a minimum, most European statistical publications are made available
on the NSI at the time of release - sometimes the first release is via the web. These are often backed up with
email information services.
The impact was already apparent in a user survey conducted by the Irish CSO two years ago7. About 70% of
respondents indicated that there had been an improvement in the quality of CSO products. In a similar survey
conducted in 1997, only 46% of the responses indicated an improvement. The role of the web in this
improvement can be seen in table 1 :
6
Globalisation: new needs for statistical measurement. Economic Trends No 598 September 2003
' Survey of CSO Users 2002. The National Statistics Board, Ireland.
m
Table 1: Reasons for improvement in CSO services 207 replies
Website 53
Staff; Customer focus 35
Timeliness 26
Automation; computerisation; e-mail 24
Better dissemination or presentation 20
Wider range of statistics 15
Quality of publications 7
Easier to contact 5
More flexible, adaptive or responsive 5
More open or approachable 5
Availability on disc 3
Data quality 3
Management 3
Ability to do special rims 1
More resources 1
Yearbook 1
Unfortunately for statistical offices, the 'bar' continues to be 'raised' in terms of customer expectations. Users
now expect greater flexibility and to have interactive tools to produce their own views of the underlying data.
The number of websites offering access to databases is steadily increasing. The amount of information
obtainable and the facility for queries vary significantly across different countries. Some NSIs provide only
basic indicators with a limited number of options on how to create a query, while others have all statistical data
available with advance query options.
A problem that foreign users frequently encounter when making queries from these databases is that the user
interface is only available in the local language. If translation is considered too expensive, then a short online
dictionary of statistical terms from the local language into English would help.
The relationship between NSIs and business users has been evolving over time and extends far beyond
collecting and disseminating statistics. For many years, NSI customers were confined to policy makers,
researchers and academics.
This trend has stimulated new demands. For example, statistical offices are generally much more familiar with
the strengths and weaknesses of their statistics than users or other intermediary analytical agencies. This
expertise should be used to inform and guide users to ensure that they can correctly identify and interpret the
fundamental messages in the statistics. Presenting the results of surveys in a format that is readily
comprehensive to the 'average user is one of the more challenging tasks facing NSIs today.
Overall quality
It is important that NSIs continue to gather the views and preferences of their users. In its 2002 survey, the Irish
CSO asked respondents to rate the quality of its products on a scale of 1 to 7 (l='ExtremeIy Good'),
7='Unsatisfactory').
Table 2 shows that users were most satisfied with the accuracy of CSO's statistics (1.9) and least satisfied with
their timeliness (2.9). It shows that users, in general, felt that the CSO had made considerable progress since
1997 in meeting this challenge, as the satisfaction level with timeliness improved from 4.2 to 2.9 while at the
same time there was a slight improvement in the rating for accuracy.
Progress was made on all the characteristics except relevance where a slight disimprovement was registered. In
the context of relevance, it was suggested that the CSO should broaden its range of statistics to include more
on the new economy and on social issues.
Table 2: Average rating of level of satisfaction with the quality of CSO products
User category Level Timeliness Accuracy Relevance Style Overall Req'ment

of detail of presenting cost fulfilled
Consultants; Market Research 1.7 2.3 1.9 2.0 2.0 2.2 2.0
Third Level; Research 2.5 3.3 2.0 2.4 2.2 2.3 2.4
Financial; Stockbrokers 2.1 3.0 1.6 2.1 2.3 1.8 2.4
Government Departments 1.6 2.5 1.6 2.3 2.7 2.2 2.3
Media; Political; Business 2.0 2.9 1.7 2.9 2.6 2.5 2.3
Local Administration 2.1 2.4 2.0 2.0 2.6 2.> 2.4
Representative Bodies 3.2 4.0 2.6 2.8 2.8 2.4 3.4
State Sponsored Bodies 2.5 3.1 2.1 2.2 2.0 2.0 2.7
All organisations 2002 2.2 2.9 1.9 2.3 2.4 2.2 2.5
All organisations 1997 2.6 4.2 2.0 2.2 2.8 2.6 3.1
EU data
Increasing international co-operation, market liberalisation and globalisation have created a demand for
statistics that allow businesses to benchmark national performance against other countries and other economic
regions. The EU as a major economic force also wishes to benchmark its performance on various measures. A
key requirement for national benchmarking is that the EU can produce quality and timely statistics that match
those of the US.
Unfortunately, while there has been considerable progress over the last decade, we still have a considerable way
to go. EU statistical databases tend to suffer from significant data gaps and time lags, and are difficult to use.
Networks of information intermediaries and fee-based access are not the answer. Our US competitors have
access to a huge wealth of easily accessible, relevant and timely business statistics, freely available at their
finger-tips - and in a format that the user does not have to be a statistician or economist to assimilate. European
businesses require similar services.
The new internet dissemination policy of Eurostat based on free of charge access for all users is a welcome
development. The statistical domains contained in its databases New Cronos and Comext are due to be made
available to the broad public from this month. Hopefully the tools for accessing and analysing this data will
usable by the general business visitor.
Conclusion
The economic, social and environmental phenomena to be measured by statistical offices are becoming more
complex due to more diverse lifestyles, greater deregulation, the advent of the information society and
increased globalisation. At the same time, the requirements of business users continue to become more detailed
and sophisticated. In order to meet these requirements, statistical offices need to take a 'whole system'
approach, involving all areas of the public sector where administrative records are maintained and from which
official statistics can be generated. This is a formidable task. There are important issues around the structuring
and linking of administrative data. NSIs will need to play a central role in relation to the use of statistical
classifications and data standards in order to make this happen.
m
COST AND EFFICIENCY

PERCEPTION BY THE BUSINESS WORLD
Robert Am 1er
SIEMENS AG
- The spoken word applies
Statistical declarations by Siemens Automation and Drives (A&D)

First of all I want to give you some information on Siemens A&D, the world market leader in the field of
industrial automation and drives technology. Products and systems - like switches and circuit breakers,
different motors and industrial controls including software and communication technology for machines as
well as for plant automation for process and manufacturing industries — are produced by A&D at 65 large
production locations worldwide. With approx. 50,000 employees A&D realized a sales volume of more than
8.4 billion EUR and a profitability of 9,6% of sales which is higher than the profit range of most of A&D's
competitors.
For such a large enterprise it is not easy at all to say how many statistical reports and declarations were issued.
The following slide gives an overview on the main statistical declarations in value and units for numerous
product lines.
Automation and D rives
Robert Amler Cost and efficiency on the business side
Statistical declarations by Siemens A&D (examples)
27th CEIES Declaration by Category

Seminar "j; Branch offices (Germany) pjoduction/sales monthly
Producing ^· Plants (Germany) production/salesfem ployees monthly

Business Statistirs
ν - Internationa! offices and plants sales monthly
Cost and Efficiency
\s" Special issues, e.g.
*s Capital expenditure yearly
¿ Environmental declarations yearly
xt Wages and salaries yearly
*s M arket shares for antitrust sales/market volume yearly
commissions
k Censusof wages and salaries 5 years period
xs Software sales erratic
Dresden, 1 y\/e estimate internal cost for declaration business data to statistical
Oct. 14-15.2004 I ~ „ . . .~ ....
offices to about 2 millions of Euro
SIEMENS i:\ad* \1_Akia-d_rBj \Pr Sseri ¡farm Wcrtr ^ e M r ø n íScreíge iCBES ASDST1 AtguáZXtM S
kk
As can be seen we declare our orders, sales volume, numbers of employees, wages and salaries, environmental
statistics as well as market share information to antitrust authorities and last but not least specific statistical
declarations and information for organizations and associations. Depending on the category this reports are
issued monthly, yearly or in perennial cycles. As an internationally active company we issue such declarations
in Germany, in European countries, in the Americas there especially in the US — and as well in our Asian
facilities and locations. If you roughly estimate the costs for our statistical declarations they amount to a total
of approx. 2 million Euro a year. But costs for statistical declarations are not reported maybe a gap in the field
of business statistics?
Achievements of business statistics

From my point of view statistics and business statistics are a method for modeling nearly everything by using
figures. So a short version of the "world" is shown that means statistics do not show everything. But the
illustration in figures allows analyzing, falsifying or verifying specific circumstances in a quite universal way.
When reading business magazines you will realize that most of the written texts are an interpretation of
historical, present and future figures, demonstrating that they are interpreted, compared with the past etc.
Automation and Drives
W h a t would economy be without business data?
27th CEIES
Economic {»'ess
Seminar
"Highest oil price since . . . ";
Produang ". . . xy company back to profit";
Business Statistics ".. . machinery orders 5% up ...";
Cost and Efficiency
" . . . labour cost per hour in Switzerland 30% above US
ps- What do we know about Nigeria and what about Europe about what
can you even get knowledge in Nigeria and Europe?
less much too much
Dresden,
Oct 14 -15,2004 Who will define and decide V/hat' is too much?
SIEMENS lAadsfrllJSfctcrdnai VPrSsrtDticnen Vcrlr £ùe>_Mappen \5cn¡*9e \COE5 A4DST1 AIJSM¡*2X3334 Side
Do business statistics cause additional efforts?

Surely, they cause additional effort if you don't use the statistics yourself.
Definitely not, if you use business statistics deriving from existing data of the company's business reporting
system for
• recognizing deviations in longer time periods
• benchmarking your own competition situation
• preparing decisions with continuative analyses
• recognizing interdependencies
• etc.
With other words: If there were no business statistics, what would we know about a country like Nigeria or
what do we know about Nigeria at all in comparison to Western Europe?
With this simple slide I would like to spotlight the central question of our seminar: "Do we need all this business
statistics, is it worth the effort?" At first glance business data or statistics are of no advantage. But if they are
used to serve a certain purposes then advantage can be taken out of reporting of business statistics. It is assumed
that this advantage is judged differently from person to person. With the following slide I would like to explain
again the interdependency between (normally useless) data and information.
Automation ar*d Drives

i&O
Data and information
27th CEIES
Seminar information is
Producing phenomena of the real and for the purpose of
τ
Business Statistics reflection on
abstract world completing a task
Cost and Efficiency
formation relation subject/object relation applicai ion retati»!
pragmatica!
syntactical and semantical dimension
dimension
objective level subjective (value) level
Dresden,
Oct. 14 -15,2004
SIEMENS l:\*yc\1./Hcnlj»j tR-Sseriitknen Vcrtr SjeJAsppeTi iStrtsfcje \CBES A&D ST i A i r e l a 2C04 Side
According to the previous slide only the user of data and statistics can decide whether to discontinue business
statistics or not. But as long as statistics hardly cost anything the decision is easy. Only when a high price has
to be paid costs and profit will be compared and measured. But I do not want to further stress this daring
statement. We definitely know a lot of apparently expensive things which would be discontinued especially in
times of crisis or when research institutes are short of money etc. But it is necessary to consider that once
discontinued data and information could hardly be revitalized and resumed.
Users of business statistics at Siemens A&D

Many of the above mentioned declarations of business statistics are used by different organization units within
the company. A brief overview is shown in the following slide.
50 27th CEIES seminai" Producing business statistics - Costs and efficiency - Recent developments and experience
m
Automation arïd D rives
User of business statistics
27th CEIES
Seminar
identity new market requirements,.
Producing
Business Statistics
Cost and Efficiency benchmark own position, costs, efficiency,
Production: productivity measures, quality information, motivation

of employees,...
Management internal and external presentations, basis for decision

making: investments, new products, capital
expenditure,...
Dresden, Data about internal business are interesting, comparable data about
Oct 14 -15,2004
external business are exiting
SIEMENS iAa&M_AMcRl_nai VPrSartaticnm Voir S)e Mappen \ScnsSge \CHES A&D ST 1 ΡϋφΛΖΚΖΧη aide
Researchers and developers use statistics to be able to react to the requirements of customers. Salesmen use
statistics to gather information on competitors and the own position. In production statistics are used to
compare costs. Managers use statistics to decide about capital expenditure or for internal and external
presentations. All these presentations and comparisons would not be possible if the company itself did not have
the data and if others did not declare their business in statistics. It has to be mentioned that these analyses or
presentations are prepared by group departments. These departments play an important role in generating new,
intelligent and significant analyses as well as in creating values out of statistics instead of only allocating data.
Advantage of business statistics for Siemens Automation and Drives explained with two
examples
Forecasts for orders and sales

As input for forecasts Siemens A&D uses business test data on one hand, e.g. results from surveys by ifo-
Institut. On the other hand a lot of data time series of branches and countries referring to orders, sales, capacity
usage and so on are analyzed. As a result the following illustration generates.
«¡rami
Automation una D rives

Robert Amler Costand efficiency on the business side
Cyclicality of A&D Business — J uly 2004
(Smoothed business climate industry by A&D Cà sa.esstaic&re and development of orders)
Gtrmany
i
S.
Veston Europe (Wo Germany)
—V* y ΊΒΝ^-
S -o
e c: USA
■S ,a
a
ϊι
»fepsn
TI
C» «■ Business Climate
11
e ■"Orders I
FY 1995 1997 2001 2003 2005

>»5 ;33î ;SÎ» ;coi :ot
The business climate in spite of the slow down in Germany

further indicates an upward trend. Due to the strong Euro only
a moderate upswing is anticipated.
l:\siot\1JMtni_nai iPl-Ssritfcrw) Vcttf SjeJÄHxn \Scn*(ft \CBSS A&D ST 1 Ai^staiO» Side
The slide shows the direction of our orders during the next months. The thick grey line indicates the business
climate, the dark blue line indicates the development of our orders (the dates shown refer to the end of Siemens
fiscal year ending September 30). In this discussion we cannot go further into the method how to extract the
leading indicator out of the numerous data (a publication dealing with this method is available from the author).
Regarding the advantages of business statistics I would like to state the following.
The picture shows that in the past orders of Siemens Automation and Drives closely followed the business
climate indicator. We assume that this trend will continue which means that A&D's orders will increase during
the next half year. The percentage of the forecast value is shown in the next slide. The absolute figures for the
forecast value and the growth rate were eliminated due to stock exchange rules.

Using external/internal business statistics for forecasts
27th CEIES
Seminar
Producing
Business Statistics
Cost and Efficiency
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
Moving 12 months average
n
3
t»
^*
t %***
CL
UI r l r ""jf
ie
SE W^^
Dresden,
Oct. 14 -15,2004
1995 1996 1997 1S98 1999 2000 2001 2002 2003 2004 2005
SIEMENS IAatg<\1_Ak!m!_nai \FVSMrtEticren Vatr ^B_Mappen \Soni*ge \CBES AID STI Αχμ*23,20Ο4 ST* S
The forecast figure is very important information referring to asset management for instance regarding
personnel capacity and material procurement. In order to give an idea of the relevance of such analyses and
forecasts generated in my department it has to be mentioned that we prepare analyses and forecasts for
hundreds of aggregates out of our company worth 8.4 billion E U R - for nearly every business unit and division,
for Germany as well as for many European countries, for the US and for China. But we as well do this for
specific product lines as forecasts by value and units. This is done mostly in a quarterly cycle. The growing
demand for our forecasts by colleagues in our plants and sales divisions in my point of view demonstrates the
usefulness of dealing with business statistics. Surely it is not possible to quantify the amount of money business
statistics are worth. But the following slide points out that this method also contributes to A&D's good
profitability in phases of downturns.
l:\eytM_rtdcnl_nai WSserfrticneri W a t t l e / A p p a l \Scrrigî VCHES A&DST1 iuiElT.roi 3de S
Although similar slides could be shown for our competitors we would like to abandon this idea for obvious
reasons. But we can show the development of A&D's profit margin in comparison to our competitor's.
Automation arai D iivo

Profit Margin Development
A&DandPeers -Group(entirecompanies} DQ3_2004
16%
01 I t i l l I I I | l ! I 1 t I I 1 I I 1 1 I I I I
Q 1 2 . 3 4: 1 2 - 3 4 1 2 3 4 12 3 4 . 1 2 3 4 1 2 3 4 1 2 3
FY 1999 i 2000 : 2001 2002 . 2003 2004 2005
A&D performs at a very high group profit margin level above the
target range and above the average of the Peers -Group
h ¡< ! v« ar i<a:i Ρ"Ίο n fn*n c rr -rprt
!:\n*jK U.Aktora rrau VPrStisnMionen WortrSçjajWappen \Sor»tige \CI~ES AKDST1 ΑΐφΒ!23.2<»4 SUM 10
Especially in the minima and maxima of the cyclicality we succeed to stabilize our profit situation.
As a result of my demonstrations I do believe that every economically thinking man can answer the question
regarding the advantage of business statistics: A profit margin of additionally half a percentage point means
approx. 40 million EUR profit for us. As for the costs it has to be said that we accomplish these forecasts and
analyses with four excellently trained employees and few contact persons in our sales, manufacturing and
regional organizations. It is up to you to judge the costs. The data basis we established for this purpose is shown
in the following slide:
Automali on and D rives

Basic data
27th CEIES MÍ External time series

Seminar Economic cycle time series:
Countries/ regions Number of industries with indicators Total
Producing USA 100 industries/branches/aggregates - 1 0 indicators 1Λ00
Business Statistics Germany 70 industries/branches/aggregates ~ 20 indicators 1400
Cost and Efficiency Italy 60 industries/branches/aggregates - 6 indicators 350
Japan 200 ¡ndustries/branches /aggregates - 5 indicators 1.000
China 400 industries/branches/aggregates ~ 4 indicators 1.600
20 more European countries 15-30 industries/branches/aggregates - 5 - 2 0 indicators 6.000

40 Asia /Pacific/other 1-20 m d usines/branches /aggregates : ~ 5 indicators 650
countries . , : . . ■ ; . :
Othertime series (product/subject specific etc): κ J::.:-K.;J >:>:-:-Kf$;;.l·:.:
Construction Automotive Chemical Tool machines Mechanical engineering 5.000

To be developed: Food and drug
I basic time series

Number of internal basic time series
Dresden.
Oct. 14 -15,2004 Internal : External =1:17
SIEMENS l:\agv\iJUaaj*u ',RSserf*crien Vatt ífceJvfeHxn vSerelge \CglES A&OST1 te&si23,7m Side
I want to point out the fact that we use at least ten to twenty times as much external statistical time series than
internal ones. Our statistical data base for big countries and capital goods branches is well filled; the financial
burden is irrelevantly low.
Let me state my second example:
Information on regional markets

As a global player we invest a lot of money in extending our business in China. As well known the Chinese
growth rates increased enormously in the last couple of years you could have hardly gone wrong there.
Meanwhile we reached a point where we have roughly marked the main points. In order to guarantee further
growth you have to know exactly in which other provinces to expand, which other customer groups to serve
with the right kind of product etc. Because we are a business enterprise we have to act according to economic
aspects, that means to reach high profit with few input factors.
Therefore we gathered all statistical Chinese data of each province for certain branches and for the indicators
employees, shipments and number of enterprises. We had the Chinese data translated to English and allocate
them in a data base. The three indicators for the 30 Chinese provinces and 233 branches respectively sub
branches add up to approx. 21,000 data sets. In an intermediate step just to enhance clarity 32 branch units
were created for the business plan of Automation and Drives. On one hand with a quick analysis of a province
ora selection of provinces according to the branches relevant to us we are able to develop a business plan, which
gives us the necessary information on the number of sales personnel etc., tells us the necessary number of
locations in the individual provinces, and on the kind of products in demand. On the other hand this is an
m
opportunity for a quick analysis of the market size of A&D-relevant products in the individual provinces in
order to optimize our market presence. Let me introduce two examples beyond the business plan of Siemens
Automation and Drives.
Imagine you would have to deliver machines to manufacturers of cigarettes or cigars - where do you think most
customers will be found? Our data base can answer this question quickly.
ST' Robert Amler Cost and efficiency on the business side

Information about regional markets (example 2}
CH INA: Manufacturing of cigarettes and cigars by provinces
Number of Companies üacräcj
Shipments b y Province
by Province Jh ¡In b i l l i o n RMB)
Hefcngiang
InrcrMcngcéa
Tferfn
Hebe
i:% Shaw
ShsnaxQ
Henen
Siasral
Gettai
Mlncpda
Xrpng
Stenet ia
.Êangsu
mm: Zhsjiang
Anhi
HU»
Hlrø!
Guengtfcng
Gi.*nçf<;
Hären
Fijian
Charging
Schien
Giishai
YUTSI
Girci**
MatøcVl/McRLpau VPrSsertätkren Vatr ^eMappat \Screågs \CHES A&DST1 AuguS23,3J04 Side
As you can see in the slide major and minor cigarette manufacturers are located throughout China. Sales
departments have to be positioned throughout the country. Especially in the Eastern region around Shanghai
and Jiangsu good business could be made and further emphasis has to be placed on Yunnan.
The second example deals with the manufacturers of bicycles - equally common as cigarettes in China. The
same question can be asked in this context: Does the sales departments have to be positioned throughout the
country in order to be able to do business with bicycle manufacturers? Our data base gives first hints quickly:
Automation αικί Drives

Information about regional markets (example 1 )
27th CEIES
Seminar Ν umber of Companies Shipments by Province
by Province (in billion RMB)
Producing
Business Statistics
Cost and Efficiency
Dresden
Oct 14 -15.2004
SIEMENS l:\a&\-\Jtttro_nBj \PrS«rtalimen Voir StpJAfpBn \Scn*je \CBES A&D ST 1 Αιχυ.ΙΖιΊΤΛ Side
Bicycle production is restricted to few provinces in the Eastern and North Eastern region as well as to the
Southern province Guangdong.
As we know from experience the illustration of branches concerning our product range caused surprise - even
sales managers with long China experience did not know all the details.
The investment in necessary business statistics is modest - this applies for the previous examples as well -
compared to the advantage that can be capitalized on the declarations released by companies. Alternatively you
can have your analyses made by consultants. But without a statistical data base not even consultants will be
successful.
Particularly taking into consideration the view of a global player I have made my beliefs clear: Statistics are
good and helpful - it depends on what you make of it. And the costs are definitely tolerable.
GENERAL STRATEGY OF N S I S
U.S. Census Bureau Initiatives to Reduce

Business Reporting Burden
Thomas L. Mesenbourg and Ronald H. Lee

U.S. Census Bureau
INTRODUCTION
The 1980 and 1995 Paperwork Reduction Acts (PRA) provide the framework for ensuring that all of the Federal
Government's information collections satisfy a programmatic need, that the information has practical utility,
and that the burden on the public of these activities is curtailed. The Bush Administration, while recognizing
the public benefits ofinformation collections, is committed to reducing needless paperwork burdens and other
violations of the Paperwork Reduction Act.
In FY 2003, the U.S. public devoted an estimated 8.2 billion "burden hours" and $320 billion responding to all
Government requests for information. Roughly 63 percent of this paperwork burden was imposed on businesses.
The Federal Government has over 8,000 separate information collection requests authorized by OMB.
Government places a heavy and expensive reporting and recordkeeping burden on all businesses, but relatively
more attention has been focused on small businesses. Additionally, small businesses bear a disproportionate
share of the total burden. According to a 2001 report, "The Impact of Regulatory Costs on Small Firms" by W.
Mark Crain and Thomas D. Hopkins,1 for firms employing fewer than 20 employees the annual reporting burden
is $6,975 per employee—nearly 60 percent more than that for firms with more than 500 employees, at $4,463.
While the Internal Revenue Service (1RS) accounted for over 80% of the total business burden, the Census
Bureaus share was not small. Annually, the Census Bureau's 100 plus current economic surveys impose almost
2 million hours of burden on businesses and state and local governments. Every five years the Economic Census,
the Nation's single largest collection of economic data, imposes an additional 5 million hours of burden.
The Census Bureau is the largest statistical agency of the Federal Government. Best known for the decennial
census of population and housing, the Census Bureau also conducts surveys and censuses that measure
changing individual and household demographics and the economic condition of the Nation. The Census
Bureau's Economic Programs Directorate is responsible for statistical programs that measure and profile U.S.
businesses and government organizations. This includes conducting an Economic Census and a Census of
Governments every five years, collecting data for years ending in "2" and "7"; over 100 separate surveys taken
monthly, quarterly, and annually, including 13 principal economic indicators; voluminous merchandise export
and import statistics produced monthly; numerous reimbursable surveys undertaken for other Federal agencies;
and numerous research and technical studies.
Relevant, accurate, and timely statistics require the continued Cooperation of our data providers and suppliers.
Continued support of the business community cannot be taken for granted. Recent modest downturns in
response rates for both our surveys and censuses support this assertion. Economic conditions appear to be the
single best predictor of response rates shifts, but corporate downsizing, perceptions of increasing government
reporting burden, and growing privacy concerns all may impact the willingness of businesses to cooperate.
' W. Mark Crain & Thomas D. Hopkins, "The Impact of Regulatory Costs on Small Firms," Report to the Small Business Administration (2001).
Background
The reporting burden imposed on businesses by the Census Bureau is significant. It is especially significant to
small businesses. In a 1995 Hopkins and Diversified Research study2 that asked small businesses how burdens
might be reduced, 95% recommended simplifying reporting and recordkeeping. For the Census Bureau to be
successful it must proactively and aggressively seek innovative ways to reduce reporting burden.
This very philosophy is embedded in the Economic Directorate's strategic plan. Recently released, the 2004-
2008 plan describes the Census Bureau's economic statistics priorities and work plans for the next five years
(www.census.gov/econplan ). The plan contains six goals and thirty-one associated objectives. Six of those
objectives along with associated performance measures and targets relate to reducing burden by simplifying
and facilitating reporting.
To accomplish these objectives, we must work closely and collaboratively with data providers, other agencies
supplying us with information, accounting groups, and organizations representing businesses and
governments. Most importantly, we need to put ourselves in the shoes of the business and government entities
we are surveying. We must shift from our traditional program or survey-centric approach to communicating
with and collecting data from data providers, to a company-centric approach. To achieve these goals, we must
better understand our data providers - their record keeping practices, organizational structure, and data
availability. We also must seek more information about how well they understand our concepts, questions, and
instructions, identify key reporting problems and concerns, and seek their ideas about how we can facilitate and
simplify reporting.
Our overall strategy has three primary components: (1) re-engineering of economic statistics programs, (2)
easing reporting by automating data collection, and (3) expanding communication with the business
community.
Re-engineering Economic Statistics Programs

In 1993, the Statistics 2000 Task Force, chaired by the Census Bureau and containing representatives from the
Bureau of Economic Analysis (BEA) and the Bureau of Labor Statistics (BLS), and the National Agriculture
Statistics Service (NASS) identified 34 specific opportunities to reduce the reporting burden on businesses;
eliminate or reduce duplicate data requests; facilitate and simplify business reporting; and improve the
efficiency of Federal statistics programs.
The single most important recommendation called for legislation to permit broader sharing of data among
Federal statistical agencies for exclusively statistical purposes. Nearly a decade later, this recommendation
became a reality with the enactment of the Confidential Information Protection and Statistical Efficiency Act
of 2002 (CIPSEA), part of the Ε-Government Act of 2002. The new law protected the confidentiality of
propriety information companies supply to the federal government on surveys and permitted business related
information to be shared among a small set of statistical agencies.
To date CIPSEA has had little impact upon Census Bureau processes, program, and budget because changes to
tax law and regulation that would permit the sharing of federal tax information (FTI) has not been enacted.
While CIPSEA authorizes the Census Bureau, the Bureau of Economic Analysis (BEA) and the Bureau of
Labor Statistics (BLS) to exchange business records for statistical purposes only, much of the Census Bureau
data that could be exchanged either contains federal tax information or is derived from FTI . Tax law and
regulations promulgated by the Treasury Department and administered by the Internal Revenue Service (1RS)
specify what agencies may get access to FTI, the specific items of FTI that may be accessed, and the uses of
the data that justify access to FTI. Until tax law and regulation are modified to accommodate CIPSEA, data
sharing activities among the three statistical agencies will be limited.
The Census Bureau and BEA have entered into one modest data sharing project, which can take place because
no data containing FTI or derived from FTI will be exchanged. The project entails comparison of data on
research and development (R&D) activities as reported on BEA's foreign direct investment surveys (incoming
and outgoing) with data on R&D activities as reported on the industrial research and development survey that
' Thomas D. Hopkins and Diversified Research, "A Survey of Regulatory Burdens." The study was prepared at the direction of the Small Business Ad
ministration's Office of Advocacy in 1995.
the Census Bureau conducts for the National Science Foundation. The project is exploratory in the sense that
it examines data from 1997 and 1999 to determine whether there are sufficient data matches from the two
samples to generate useful information and justify establishing an ongoing data sharing program. Both agencies
believe the program may bring to light inadequacies in their respective survey samples. If this proves to be the
case, then data sharing would be illustrating one of its purported benefits, namely helping the statistical
agencies improve the completeness and efficiency of their samples.
Another Statistics 2000 recommendation was to create subcompany organizational links on the Census
Bureau business register. Identifying links among all the organizational units of a company allows Census to
match requests for financial and similar information to current company organizational structures and record
keeping methods, making it easier for organizations to respond to census and survey requests. Aggregated
reporting units, constructed around North American Industry Classifications System (NAICS) industries, are
widely used in retail, wholesale, and service current surveys and for banks, telecommunication, utilities and a
couple of other "networked" industries in the 2002 Economic Census.
In a number of other instances Census has already re-engineered its data collection methods so that economic
information can be improved or continued while reporting burden on the business community is reduced or
avoided. Examples include:
•The U.S./Canada Data Exchange has relieved U.S. exporters of 3.8 million hours of reporting burden each
year since 1990. Under an historic agreement, the U.S. and Canada exchange import data monthly in lieu of
collecting and compiling documents on cross-border export trade; Canadian imports from the U.S. are used
to measure U.S. exports to Canada and U.S. imports from Canada measure Canadian exports to the U.S.
Because of this agreement, Shipper's Export Declarations no longer need to be completed and filed for
shipments to Canada by U.S. exporters.
• Use of administrative record information from the Internal Revenue Service (1RS) to relieve small
businesses of the reporting burden associated with completing mail questionnaires. Examples include:
- In the Economic Census the use of administrative information in lieu of direct reporting for over 2 million
small employer firms and 16.5 million businesses with no paid employees, thus eliminates millions of
hours of reporting burden, reducing processing costs, and increasing processing efficiency while extending
coverage of all U.S. businesses.
- We are also studying the feasibility of using tax return data from small businesses in lieu of data collection
for a number of our surveys. One such survey is our Business Expenses Survey.
Innovative research and methods have also helped to reduce reporting burden and facilitate reporting. For
example:
• Minimize the sample re-selection overlap - In order to reduce burden on small and medium sized
businesses, we attempt to minimize the number of small and medium-sized firms that are selected in
consecutive samples. We do this only when it does not increase nonsampling error into our estimates.
• For a number of our service sector programs, we were able to avoid re-selecting about 80% of the businesses that
were in our prior surveys. In addition, small businesses participating in the Quarterly Financial Report program
now are ineligible for re-selection for a 10-year period following completion of their 2-year reporting period.
• Minimize sample sizes needed to achieve desired levels of reliability - We select new samples for each of
our indicator surveys of the retail, wholesale, and service industries at approximate five years intervals. This
allows for some relief of reporting burden for units in the survey. We scientifically design each sample to
produce estimates of a desired reliability at specified industry levels with the fewest units possible. Although
small and medium-sized firms are more numerous in the population than large firms, their economic data tend
to be much less variable. Stratified random sampling allows us to take advantage of this relationship so that
small and medium-sized firms are sampled at a much lower rate than larger firms, thus reducing the overall
burden on the business population. We use a subsample of the annual surveys to conduct the monthly or
quarterly surveys rather than select a different sample.
• Use of the Census of Manufactures as a sample frame for current manufacturing surveys, rather than the
Annual Survey of Manufactures (ASM), spread reporting burden more equitably across all manufacturers and
reduced the number of ASM companies also included in other current surveys.
• Expand coverage of the Census Bureau's business register to include annual receipts data for businesses with
and without paid employees. This has allowed further burden reductions through more efficient sample
designs and new economic data publications without imposing additional burden.
• Ensuring data requests reflect corporate and government accounting conventions, concepts, and
definitions are responsive to our data suppliers' accounting and recordkeeping practices. This is a new
initiative and represents a significant change in mindset. Accounting concepts and conventions are going to
be key determinants of data collectability. To assist us, we plan to contract with a leading accounting firm so
we can acquire the expertise needed to systematically review our report forms and instructions as well as
being better informed about changing tax laws and regulations.
• Increasing the use of cognitive research techniques to evaluate existing and new economic data program
questionnaires for recommending changes in content, structure, and methods that would make timely and
accurate responses less burdensome.
Cognitive testing has been a major contributor to the development and testing of the questionnaire collecting data
for the new economic indicator in the Services Sector, the Quarterly Services Survey. Testing plans included
exploratory interviews with trade association representatives and business respondents, identified their
measurement and determined business respondents' abilities to provide requested data. Similar testing was
conducted for our newly expanded Annual Capital Expenditures Survey and Services Annual Survey. In all cases,
the cognitive work has liad a major influence on data content, questions, and instructions. The Census Bureau has
recently adopting a pretesting policy that requires all new questions be tested prior to full data collection.
A variation on cognitive pre-testing is usability testing, which examines respondents' interactions with the
graphical user interface of electronic instruments. Usability principles are user-centered, so results of testing
ease the interaction between the user and the instrument, reducing respondent burden in surveys. Usability
testing has been conducted for the electronic instruments collecting the Quarterly Financial Report (QFR) and
the monthly manufacturing indicator.
Easing Reporting Through Increased Automation
Economic Census:
The Census Bureau has aggressively expanded electronic reporting capabilities in a number of programs and
worked closely with businesses to ensure that their needs are being met. Our experience has demonstrated that
implementing an electronic reporting capability, if done effectively, demands substantial Census Bureau resources
and significant changes to existing processing systems. With the 2002 Economic Census the Census Bureau set a
major milestone in collecting data from companies. More than 3.5 million businesses were offered the opportunity
to file electronically via the Internet. Almost 475,000 establishments filed their reports electronically. This
represents about 14% of the total reports filed. In fact, 3 out of 4 responses by our Nation's largest companies were
submitted electronically. It is also almost a 60% increase over the 1997 Economic Census. While participation was
impressive, we believe those numbers can be significantly increased for the 2007 Economic Census.
A business plan that evaluates the 2002 efforts and recommends strategies for improving response is already in
place. The goal for the 2007 Economic Census is 900,000 establishments or almost twice the number reported
electronically for 2002. Under this plan, we will offer an electronic reporting option to an additional 1.5 million
businesses that receive a short report form. The strategies focus on improving response across the spectrum of
all companies. One strategy is to improve awareness through a systematic marketing campaign. During the
2002 census only 4,000 single-units filed electronically. Our research showed that these small companies filed
electronically only because they "stumbled" across the instrument - there was no direct marketing of the
instrument to them. Another strategy is to increase participation of the medium and large companies. This will
be accomplished through a major redesign of the instrument aimed at making it more appealing (i.e., less
onerous looking), increasing its functionality by making the tools more intuitive and less error prone, and
improving the instrument's overall download and upload performance. We are also considering having staff
visit the very largest of companies to assist them with their electronic reporting needs.
Current Surveys:
So far nine of our current surveys offer electronic reporting. Respondents in the annual Company Organization
Survey (COS) have embraced electronic reporting. In 1999, 400,000 establishments reported electronically,
450,000 in 2000, and 500,000 in years 2001-2003. In all five years the response rate for electronic reporters
exceeded the overall response rate by 2-4%. More importantly, the electronic reporters report earlier than the
paper form filers. Since electronic reporting offers the most benefits to large companies that are requested to
fill out multiple requests, the acceleration in reporting is important in permitting us to release our annual
economic statistics in a more-timely manner. Feedback from companies using electronic reporting indicates
that they also experience cost savings. One of the largest oil companies stated that electronically filing the COS
form cut the preparation time by 25%.
In our Automated Export System, exporters have reported that the average time to complete an electronic
Shipper's Export Declaration is 3 minutes compared to 11 minutes for a paper form. Electronic reporting is
easier to use and through import functionality offers data collection efficiencies not available with paper-based
forms. Also, electronic reporting provides a higher quality report due to the embedded edits, which translates
into fewer time-consuming follow-ups. In our export program, electronic documents have a 4% error rate, while
paper forms have a 50% error rate. Notably innovative is the Internet filing option the Census Bureau
introduced in 1999. This free, off-the-shelf option permits the on-line interactive or batch filing of export
information and now has more than 4,700 companies participating. We also plan to expand and enhance the
functionality and support for the AES to meet the requirements of the export trade community and the
legislative requirements for mandatory use of AES.
Electronic reporting capability also has been expanded to include the Annual Survey of Manufacturers, the
Quarterly Summary of State and Local Tax Revenues, the Quarterly Survey of the Finances of Public Employee
Retirement Systems, and the Annual Survey of Government Employment. Electronic reporting is also now
being offered to three of our principal economic indicator surveys, foreign trade statistics program, the
Quarterly Financial Report (QFR) program, the Manufacturers' Shipments, Inventories, and Orders Survey
(M3), and the Quarterly Services Survey (QSS). We also implemented a computerized self-interviewing system
that pennits businesses to enter information using Touchtone Data Entry (TDE) systems for selected surveys.
These systems provide reporting flexibility by allowing companies to report any time of day or night. For
example, currently more than 1,200 companies use TDE monthly to report in the Manufacturers' Shipments,
Inventories, and Orders Survey (M3). This technology is also offered in the monthly Building Permits program
and the Monthly Advance Retail Trade Survey (MARTS). We have found that in indicator programs such as the
M3, that converting existing TDE or FAX reporters to electronic filing, is challenging. The existing options are
convenient, while electronic filing does not provide significant value-added. In the case of a new survey, it is
much easier to sell electronic filing. For example, our new quarterly Services indicator started collecting data
in April 2004 and already has a 25% electronic filing rate, a significantly greater rate than experienced in the
monthly manufacturing survey.
Further expansion of electronic reporting to our current surveys is more problematic and budget dependent.
With more than 100 different surveys, the degree of customization and the amount of infrastructure required to
support such an effort is considerable. We currently have a FY2005 budget initiative that would pennit
businesses to file electronically in any one of almost 100 cunent economic surveys. If funded, the expansion
will take place over several years. If not funded, we will move more slowly.
Expanding Communication With Businesses

The quality and the success of the Census Bureau's economic statistics programs require the cooperation and
participation of the business community. Several initiatives aimed at proactively seeking business support and
increasing business data use are described below. We welcome additional ideas.
An integral step in survey and census planning is systematic contact with trade associations and other business
groups to review and comment on current survey and census content. For example, over 3,000 groups were
asked to comment on the 2002 Economic Census report forms, and the comments received were instrumental
in making the forms more effective in content and format. For the 2007 Economic Census we are going begin
our content review process with a systematic review of all census program components and content with our
most important customers, the Bureau of Economic Analysis, the Federal Reserve Board, and the Bureau of
Labor Statistics. We will be asking these agencies to prioritize programs and related content. After reviewing
the census, we will conduct a similar review of our current economic statistics programs.
The Census Bureau established the 2002 Economic Census Account Manager program to support and provide
resources for the 1,000 largest companies to complete their census forms. These businesses account for nearly
500,000 business locations and over a third of U.S. employment. More than 140 Census Bureau statisticians
and economists have been assigned as Account Managers to work with these companies.
The Account Manager program is an extension of the pilot Customer Relationship Management (CRM)
program launched in 2001. CRMs work closely with data providers and survey programs to identify and resolve
resource issues; provide annual reporting calendars to summarize the Census Bureau infonnation requests
being sent to the company; and facilitate the survey reporting process, leading to more accurate and timely data
for Census Bureau customers and stakeholders. CRMs are internal advocates for data providers concerned
about reporting burden.
A 24/7 Internet/Help Desk facility was also established for the 2002 Economic Census. During the conduct of
the Census the site received more than 9 million hits, over 300,000 visits, and close to 100,000 requests for
extensions, re-mails, and additional forms were made by using the site's functionality.
In July 2004 a Small Business Ombudsman position was established and filled. The position will coordinate
activities across the Economic Directorate aimed at finding ways to facilitate reporting and minimizing small
business reporting burden. Being proactive and providing small businesses with a voice at the Census Bureau
will help us identify additional opportunities for simplifying reporting and minimizing reporting burden.
Next steps are to further expand and improve communication with the business community by:
• Increasing contacts with organizations that represent small businesses.
• Informing small businesses that the Census Bureau has a small business ombudsman office, whose objectives
are to serve their needs and minimize reporting burden.
• Extending the successful Account Manager Program and the 24/7 Internet Help Site used in the 2002 censuses
by adapting them for use in economic surveys.
•Adding a Small Business component of the Economic Directorate's "1-800» hotline and "24/7 internet
assistance" that provides small businesses with "one-stop-shopping" for "plain language" technical assistance
and information regarding Census Bureau programs and requirements.
CONCLUSION
One of the Census Bureau's most daunting challenges is to continue to meet the information needs of private
and public sector customers while minimizing the reporting burden imposed on the business community. This
paper described some of our accomplishments and future plans. To achieve these objectives, we need to create
new alliances and partnerships with the business community; we must reach out to seek new ideas and
approaches; and we must listen and act on what we hear.
As a first step, we welcome your comments and suggestions about the ideas presented in this paper as well as
any other opportunities to reduce burden or improve the usefulness of our programs. Please send your
comments to: Tom Mesenbourg, Assistant Director for Economic Programs, U.S. Bureau of the Census, Room
3045-3, Washington, D.C. 20233. His fax number is 301-457-3767 and his Internet address is
tmesenbo@census.gov.
General Strategy of NSI's

The Unified Enterprise Survey
Michel Cloutier
Statistics Canada
Executive Summary
This paper presents an overview of Statistics Canada's Unified Enterprise Survey (UES) as an example of an
approach for developing efficient surveys. The paper will discuss information and communication technologies
issues related to the UES systems and data warehouse, UES strategies for reducing response burden and overall
concepts for improving the efficiency of business surveys within National Statistical Institutes. In addition new
standards such as the Chart of Accounts and XBRL will be discussed. The paper also presents some metrics
regarding the UES program (concrete outcomes), as well as the challenges and lessons learned from this
project.
The Unified Enterprise Survey

In the late 1990's, Statistics Canada decided to redesign its entire framework for conducting business surveys.
The goal was to integrate all business surveys, into a single master survey program called the Unified
Enterprise Survey (UES). The UES was designed to collect more industry and commodity detail at the
provincial level than was previously possible and while avoiding, as much as possible, any overlap between
different survey questionnaires.
The UES was created to achieve a number of objectives as part of the Project to Improve Provincial Economic
Statistics (PIPES). Overall the goal was to improve business survey data in terms of consistency, coherence,
breadth and depth. In summary, these objectives or UES principles include:
• Frame/Sampling and Coverage: use of a single unduplicated frame, a common sample design, enterprise
centric approach ensuring full unduplicated coverage of all establishments, focus on large complex
enterprises for profiling and survey samples, use of tax data for smaller simple enterprises, expanded
coverage of business surveys to include all sectors of the economy, sample sizes appropriate for the production
of provincial estimates by industry
• Content and Collection: use of common concepts, terminology and classifications (standards), elimination of
duplicate data requests using an enterprise centric strategy and questionnaires with a common look and feel,
Key Provider Manager (KPM) program for most important businesses, electronic data collection and/or
integrated data collection options
• Processing and Post-Collection: use of common generic processing systems and methods (common edit and
imputation methodologies), centralized UES warehouse for processing and analytical purposes, allocation of
head office expenditures and enterprise level survey responses to accurately measure value added by industry
using the North American Industrial Classification System (NAICS) and province
• Outputs: timely, detailed and reliable economic data allowing the production of provincial input output tables,
while imposing minimal burden on respondents, elimination of incoherence between the establishment data
and enterprise data, facilitate more analysis through the centralized UES warehouse.
An integrated (unified) approach to business surveys has been found to be essential in helping us to understand
and resolve response burden and efficiency issues. The strategic direction for the UES is as follows:
• continue to centralize operations and survey infrastructure (including a centralized data warehouse)
• increase the number of surveys and clients using the existing UES infrastructure and leveraging UES services
to create further efficiencies thereby permitting greater emphasis on analysis in subject matter areas;
• fill data gaps through administrative and survey data;
• maximise the use of tax data to reduce response burden for small and medium sized businesses;
• employ a more integrated approach to deal with large businesses using the Key Provider Manager (KPM)
program as a model.
• use of the Chart of Accounts and XBRL to standardize concepts and facilitate response
The strategy for the UES has been to centralize core UES-related services associated with content harmonization
and preparation of mail out materials, collection monitoring and budget management, respondent relations
(complaint resolution), processing of collected data, the development and maintenance of the centralized UES
data warehouse and associated tools, and the monitoring and assurance of overall UES data quality. Through this
process the best ideas and practices from all UES survey managers are incorpo rated in the standard UES systems
used by all participating surveys. The use of these generalized systems across all UES surveys reduces
development and maintenance costs for the agency and allows us to maintain a constantly improving and flexible
processing system for the UES. These functions are essential to meeting UES objectives.
The UES Data Warehouse is also key to gaining efficiencies in the survey process. The following aspects of the
warehouse are discussed: major components, underlying software, user interfaces, interaction with other
systems, integration of survey, administrative and other data sources and the number of surveys currently
included. The potential for expanded survey coverage, increased analysis through the use of data marts and
better data coherence will also be discussed. Finally the role of the warehouse in understanding and managing
respondent burden is presented.
The benefits of this approach include:
Reduced development costs: By developing a single system, efficiencies are leveraged both in terms of system
hardware (reduced number of servers) and system development (single group of developers).
Equity: All survey managers use the same tools and processes. There is no longera tendency for one group to
have a better set of tools than another.
Knowledge Transfer: Since the same set of tools are used for a wide variety of surveys, survey managers who
transfer from one subject matter area to another need only worry about learning a new subject matter since the
tools and methodologies that they are exposed to do not change.
Development of best practices: By developing and maintaining a centralized processing system, the best ideas
and practices of the various survey managers can be incorporated into the processing system and these ideas
and practices can in turn be shared across all surveys.
1) The Unified Enterprise Survey

In the late 1990's Statistics Canada decided to redesign its entire framework for conducting business surveys.
The goal was to integrate all business surveys, into a single master survey program called the Unified
Enterprise Survey (UES). The UES was designed to collect more industry and commodity detail at the
provincial level than was previously possible and while avoiding, as much as possible, any overlap between
different survey questionnaires.
The UES was created to achieve a number of objectives as part of the Project to Improve Provincial Economic
Statistics (PIPES). Overall the goal was to improve business survey data in terms of consistency, coherence,
breadth and depth. In summary, these objectives or UES principles include:
• Frame/Sampling and Coverage: use of a single unduplicated frame, a common sample design, enterprise
centric approach ensuring full unduplicated coverage of all establishments, focus on large complex
m
enterprises for profiling and survey samples, use of tax data for smaller simple enterprises, expanded
coverage of business surveys to include all sectors of the economy, sample sizes appropriate for the production
of provincial estimates by industry
• Content and Collection: use of common concepts, terminology and classifications (standards), elimination of
duplicate data requests using an enterprise centric strategy and questionnaires with a common look and feel,
Key Provider Manager (KPM) program for most important businesses, electronic data collection and/or
integrated data collection options
• Processing and Post-Collection: use of common generic processing systems and methods (common edit and
imputation methodologies), centralized UES warehouse for processing and analytical purposes, allocation of
head office expenditures and enterprise level survey responses to accurately measure value added by industry
using the North American Industrial Classification System (NAICS) and province
• Outputs: timely, detailed and reliable economic data allowing the production of provincial input output tables,
while imposing minimal burden on respondents, elimination of incoherence between the establishment data
and enterprise data, facilitate more analysis through the centralized UES warehouse.
Statistics Canada's business surveys are the responsibility of different program areas, such as distributive
trades, services, transportation and manufacturing, dispersed in over a dozen subject matter divisions.
Enterprise Statistics Division (ESD) was therefore created as a central area charged with managing and co-
coordinating all UES activities including the UES Warehouse. It is a highly integrated and interdependent
approach to conducting business surveys which involves many service area partners. ESD coordinates the work
of these partner divisions, including the business register, centralized collection services, operations research
and development, methodology services, tax data and standards.
The UES strategy emphasizes a centralized approach where many operational activities are concentrated in
ESD, allowing subject matter areas to focus on data validation, analysis and subject matter specific research.
The UES has allowed Statistics Canada to achieve a much more robust annual statistics program compared with
what existed before PIPES. Nevertheless, the constantly changing environment of business surveys including
the new ways businesses are organizing themselves and operating in the global economy (new business
models), financial pressures within the agency and increased reliance and response burden on large businesses,
dictates a need for further improvements in our survey programs. An integrated (unified) approach to business
surveys is more and more essential in helping us to understand and resolve these issues.
Overview of ESD within the Business and Trade Statistics Field

The UES program is managed in a matrix environment through a streamlined number of strategic and
operational committees, with partner divisions providing methodological, frame, standards, collection and
application development services. ESD's role is to collaboratively orchestrate this group of subject matter
clients and partners, and to plan, manage and process the UES so as to meet operational and strategic objectives.
ESD is also responsible for providing core UES-related services associated with content harmonization and
preparation of mail out materials, collection monitoring and budget management, respondent relations
(complaint resolution), processing of collected data, the development and maintenance of the centralized UES
data warehouse and associated tools, and the monitoring and assurance of overall UES data quality. Through
ESD the best ideas and practices from all UES survey managers are incorporated in the standard UES systems
used by all participating divisions. The use of these generalized systems across all UES surveys reduces
development and maintenance costs for the agency and allows us to maintain a constantly improving and flexible
processing system for the UES. These functions are essential to meeting UES objectives and are not replicated
in other divisions. Beyond the UES, ESD provides corporate respondent relations services across all STC
business surveys, including survey complaint resolution and the KPM Program for large, complex businesses.
2) Results of the UES (program outcomes)

Over the last three years a great deal has been accomplished to improve the UES. The timeliness target of releasing
UES survey results fifteen months after the reference year-end were achieved for most UES surveys as of reference
year 2001, shaving as much as nine months off the previous nonns. A systematic multi-year plan was put in place
and implemented to replace survey data with tax data for simple businesses across all UES surveys, thus
significantly reducing response burden. Collection costs have been reduced significantly (almost 20%) through the
aforementioned initiative and the use of a "score function" to prioritize delinquent and edit follow-up activities in
the centralized collection areas. Over the last few years, ESD has implemented a fully redesigned systems
infrastructure which integrates tax data in the UES process and ensures that all tax data are available in a systematic
way for business surveys. The processing environment has been re-engineered using BANFF (Statistics Canada's
new SAS based generalized E&I system) to allow us to produce estimates much earlier in the survey cycle (while
collection is still ongoing), thus ensuring a more top down approach to data validation and enabling more analysis.
Response burden reductions in the last few years have been significant (with further reductions planned) due
to greater use of tax and shorter questionnaires based on the Chart of Accounts (COA), which brings our data
requests more in line with standard accounting and business reporting practices. The KPM program has
matured and now covers most of the largest businesses in Canada. Pilot work on coherence analysis of
establishment versus enterprise data has set the foundation for introducing such data confrontation into the
ongoing UES production process for reference year 2003.
Through ESD, we have also achieved most, if not all of the basic objectives for the UES. These objectives include
the use of a common survey frame, integrated sampling and estimation, common concepts and definitions,
common look and feel for questionnaires and mail out materials, common/generic collection and processing
systems with lower maintenance and development costs and a common database through which we can leverage
our data holdings. Centralizing the UES management and core process functions in ESD has meant timely
delivery of change initiatives (timeliness improvements, tax replacement, system and database enhancements)
in a cost effective manner. Continuous improvement of these processes to optimize all aspects of the UES is also
at the heart of the ESD mandate. Details are provided in the following sections for these achievements.
2.1 Timeliness
The length of time between the end of the reference period and the appearance of the estimates for publication
was problematic in the start-up years of the UES surveys. The pilot year for new surveys and the following
transitioning of existing surveys into the UES covered the reference years 1997 to 2000. All components within
the survey process contributed to the problem; namely questionnaire and collection applications development,
data collection, survey operations processing and analysis. However, the principles of the UES did clearly
identify timeliness as a priority given its importance to the relevance of the data to the user community.
In June 2001, a UES task force was instituted to deal with the timeliness issue with ESD taking up the
leadership and coordination role. A plan was devised for improving the timeliness of all survey processes within
the UES. A target was set as "15 months after the reference period" to be applied to reference year 2001. For
all future reference years, the 15-month target would be maintained and where possible more improvement to
a 12 month target would be sought. In summary, major improvements were achieved as of reference year 2001
with most surveys releasing within 15 to 17 months. Further improvements in timeliness were achieved for
2002 and 2003 reference year UES surveys with some surveys achieving releases within twelve months.
2.2 Use of Administrative Data in business surveys
Significant progress has been achieved in making greater use of tax data. In the Unified Enterprise Survey
(UES), survey data is being replaced with tax data whenever possible, both as planned tax replacement and for
survey non-response ( 14 000 tax records from a sample of 55 000 in reference year 2002). More importantly,
this initiative is allowing us to reduce response burden for smaller enterprises. Over 50% of simple businesses
previously in sample will no longer be receiving a UES questionnaire. We have also implemented a fully
redesigned systems infrastructure which integrates tax data in the UES process and ensures that all tax data are
available in a systematic way for business surveys.
Research is ongoing to evaluate the best ways to fully utilize the available tax data and increase the rate of tax
replacement wherever possible, to further reduce response burden. A number of simulations have been done
with different rates of tax replacement for simple enterprises. These simulations have clearly shown that we will
need either a small sample of simple businesses or characteristics surveys to model the non-financial
information. A characteristics survey pilot project is currently under way using a simplified one-page
questionnaire which requests percentages (rather than dollars) that could be applied to tax data to generate the
eurotta!
type of detail required. The lessons learned from this project and the simulations will allow us to expand our
use of tax data and reduce response burden, while maintaining data quality and timeliness.
The benefits of increased use of tax data include a major reduction in respondent burden, a more reliable flow
of data, potential for increasing the coverage of the survey programs, and a reduction in data collection costs.
Over the last few years, STC has successfully developed a complete infrastructure that integrates tax data into
the survey process. With the infrastructure now in place, we are able to offer the tax replacement process to all
surveys within the UES. This centralization of the infrastructure has led to both a quick implementation of the
strategy and wide usage. If each survey program had been left to its own devices to develop and implement a
tax use strategy, the process would have been more expensive and adoption much slower.
The UES tax strategy consists of two separate initiatives. The first is the use of tax data to estimate for the
smallest businesses within the population, those cumulatively representing the lowest percentage of total
revenue. For UES 19982001, a "RoyceMaranda exclusion threshold" of 5% was used. For reference year 2002,
the threshold was increased to 10% reducing the UES sample by some 4,000 units. Table 1 shows the share of
the takenone population relative to the total population for the entire UES Program from 19972001. It is evident
from this table that data for the majority of businesses in Canada are obtained from administrative records.
Table 1 - Take-None Population sv . Total Population of UES Program
Reference Number of . Take-None Total Revenue Total P opulation Total UES

Year Industries Population Take-None Revenue
Surveyed Population 'SBiilion
:
'SBiilion
1998 18 181,270 7.2 400,823 485.9
1999 20 411,985 31.3 940,647 891.4
2000 40 301,421 60.1 843,970 1,505.8
2001 42 331,326 73.1 857,968 1,558.8
2002 42 448,293 134.7 864,757 1,726.5
The second initiative involves obtaining data for a subset of businesses administrative records rather than
through the survey process. This initiative incorporates both a 'preidentified' component and replacement for
non response. The preidentified component involves preselecting units within the UES sample and obtaining
their administrative data rather than burdening them with a questionnaire. The replacement for nonresponse
component involves using administrative data to replace and impute questionnaire responses for delinquents.
Table 2 shows the number of sampled units for which administrative data were used in lieu of survey response.
Table 2- UES Tax Replacement Progre

ss in RY2002
Industry Number of Simple Pilot Year Pre-identified Tax Non Response Tax Total Tax
Single Enterprises PreidentifiedTax Replacement Replacement Replacement
in the sample Replacement Targets Units Received Units Received Units Received
Number Percentage
:::V::;:' ■ ! '. - , ' of Units of Simple
αν
Single
Sample
Distributive 11,821 1,072 588 2,209 2,797 24%
Trades
Service 17,854 4,813 3,194 3,920 7,114 40%
Industries
Manufacturing 10,780 835 367 2,661 3,028 28%
Other 967 0 0 167 167 11%
Total 41,422 6,720 4,149 8,957 13,106 32%
Nole: When lhe reduction in sample due to the 10% Royce/Maranda threshold are included, the effective tax replacement rate tor the UES is approxi
mately 40% in RY 2002. For RY 2003 this rate has increased to 50%.
68 27th CEIES seminai" Producing business statistics Costs and efficiency Recent developments and experience
m
Ideally we would like to use administrative data for all simple units within the UES sample. The reality is that
administrative records do not contain all the information required by users of the UES data. Non-financial
information is required and must be obtained directly from at least a small sub-sample of respondents to allow
us to model questionnaire data for simple businesses. Research is currently being undertaken within the UES
to determine the best way to collect and process non-financial information and marry this with the financial
information obtained from tax sources. The goal of this research is to enable the UES to maximize the use of
administrative data while maintaining a high level of data quality.
2.3 Centralized Processing Systems and Databases

A principal goal of the UES was to create a suite of centralized tools that could be used to both process and
analyze data. The rationale for the centralization of these services includes:
• Reduced development costs as we move away from stovepipe, stand-alone systems and approaches.
• The development of best practices that can be embedded in the processing tools and shared across the bureau.
• Reduction in the amount of knowledge transfer required as individuals move from one subject matter area to
another.
• Centralization of the databases, moving from a distributed model to corporate data management.
• A single point of access to and centralized security of all UES-related data rather than multiple access points
and multiple / differing security levels.
• Rationalization of hardware to minimize the number of servers.
In the early stages of the UES, many of these systems were developed independently of one another. This was
not so much by choice as it was a result of the requirement to get the system up and running in time to meet the
UES production schedule. One of the major accomplishments of ESD over the last two years has been
incorporating the UES processing system and databases into an integrated suite of tools that have a common
look and feel and shared methodology. This current suite of tools includes:
• USTART: The Unified Survey Tracking Analysis and Retrieval Tool is a SAS-based, web-deployed analytical
tool that provides analysts and survey managers access to all the data contained within the UES Data
Warehouse. USTART offers the analyst data retrieval, querying and report building-functionality. In addition
it provides analysts access to all the UES metadata and scanned UES images. Currently there are over 512
USTART users in 24 STC divisions.
• FRIC: The Facility to Review Impute and Correct is a visual-basic correction and analysis tool that allows
analysts to produce "flash estimates" at the Industry/Geography (NAICS/GEO) dimension and to drill
through to the micro data to examine top contributors and outliers. FRIC allows the analyst to correct capture,
processing or response errors at the micro level. FRIC includes a scries of edits to ensure consistency
following micro data changes by analysts. There are currently over 175 FRIC users across seven STC
divisions.
• Edit and Imputation: A SAS-based, web-deployed batch E&I system using STC generalized systems. Over
the last year, ESD has converted its E&I system from GEIS to BANFF (the SDD/BSMD re-designed
generalized imputation system). The UES edit and imputation system utilizes generalized functions such as
error localization, deterministic, donor, proration and estimation. Every survey that is part of the UES passes
through this system.
• Allocation/Estimation: A SAS-based, web-deployed system used to allocate survey information from the
collection entity to the statistical establishment or location. In addition BSMD has constructed a survey
estimation module that ESD has incorporated within the processing system. The estimation module calculates
the final survey weights, CV's, response rates and imputation fractions. All UES surveys pass through this
system.
• I O M S - The Integrated Questionnaire Metadata System is a Visual Basic application that stores the majority
of the UES metadata (cell descriptors, formats, and code sets). It is also the tool that survey managers use to
enter their data capture edits and output specifications. The tool is evolving toward becoming the warehouse
forali UES related metadata. Currently there are 162 IQMS users across seven STC divisions.
• KPMIS - The Key Provider Manager Information System is a Visual Basic application that allows survey
managers to monitor the interactions between the key provider managers and Canada's largest businesses. The
system allows managers to review reporting arrangements, response rates, KPM comments and other relevant
information. There are currently 160 users of the KPMIS in seventeen divisions.
2.4 Respondent Burden Reduction

When the Unified Enterprise Survey was conceived under the PIPES initiative, it was fully recognized that in
the short to medium term, improvements in coverage and quality would result in significant response burden
increases. Following the pilot survey for reference year (RY) 1997, industry surveys were added annually
culminating in the inclusion of the Annual Survey of Manufacturers (ASM) for reference year 2000.
Questionnaire content, driven chiefly by SNA requirements, was ambitious. While there was an established
principle of ensuring that the variables requested were consistent with record keeping practices of the business
community, in reality there was little time to fully test the consistency of our needs with the capability of
businesses to report. There was no GAAP-consistent accounting framework on which subject matter divisions
could shape their survey questions or through which the SNA could express its data needs. The use of tax data,
a fundamental PIPES principle, was not yet feasible for the most part as standardized electronic tax data was
just becoming available and deemed to be unreliable and too late to be of real use. Under the "Royce-Maranda"
guidelines however, the smallest firms representing the bottom five percent of revenues were eliminated from
sample and domain estimated. For the largest most complex businesses, the KPM program was established to
manage relationships and facilitate data reporting. This was accomplished by negotiating reporting
arrangements, and prioritizing surveys in concert with subject matter clients.
Reduction in questionnaire content and the Chart of Accounts

Since the introduction of the ASM three years ago, a number of significant steps have been taken to reduce the
respondent burden associated with the UES. Faced with increased collection difficulties and poor response to
some questions, starting in RY2001 divisions slashed the size of their questionnaire typically from 12 to 14
pages to 4 or 5. This represented a combination of fewer variables requested (see chart 1) and moving detailed
instructions from the form to accompanying "guides". While the latter strategy likely did not decrease the
actual time required to complete the forms, it was felt that the survey questionnaire would at least appear to be
far less intimidating, thereby eliciting a better response. For reference year2002, a number of initiatives came
on stream to more aggressively tackle the burden issue. Divisions began incorporating the Chart of Accounts
(COA) standard that had been approved in principle by the Methods and Standard's Committee, and which was
designed to be GAAP-compliant and consistent with SNA/Subject Matter needs. While there is no empirical
evidence to quantify the impact of this change for respondents, it is felt that this has made reporting easier. It
also opens the way down stream to more powerful burden-reduction technologies such as XBRL.
Chart 1 -Average Number of Variables Requested per UES Question n aire (Excludin g ASM)
240
m 2000
220
Β 2001
200 □ 2002
180
160
140
120
100
80
Aqua DTD HO lOFD SiD Transpo. UES
Aqua - Aquaculture Survey; DTD - Distributive Trades Division; HO - Head Office Survey; IOFD - Industrial
Organisation and Finance Division; SID - Service Industries Division; Transpo -Transportation Surveys; UES
- Unified Enterprise Surveys.
m
Between the survey years of 2000 and 2003 expected response burden as measured by hours required to
complete surveys, will have decreased by almost 25% through a combination of tax replacement and reductions
in survey questionnaires (see Table 3). This significant reduction will provide relief to a large number of small
businesses previously receiving UES questionnaires.
Table 3 - Potenti al Response Burden Hours for the UES

Survey Grouping by Industry 2000 2001 2002 2003
Services 22,478 26,026 18,010 24,750
Trade 43,253 38,852 27,696 21,500
ASM 49,044 51,221 53,704 40,125

Other 2,270 2,701 1,939 2,000
UES Totals 117.045 118,800 101,349 88,375
Reduction i n sample si ze
RY2002 also saw the first use of tax data to replace survey data at the micro level. Approximately 6,800 simple
businesses were eliminated from survey collection, reducing overall burden by some 13,000 hours. For
UES2003, an additional 8,000 businesses (16,000 hours) will be replaced (see Chart 2). Further, the percentage
of small businesses eliminated from sample and estimated through use of tax' using the Royce-Maranda
threshold, was doubled to ten percent (4,600 units and 9,000 hours).
Chart 2 - Questionnaires Mailed by Reference Year
r
65,000
60,000 - 1 m 2000
H 2001
55,000 -
: : : : : : : : □ 2002
50,000
45,000 1
40,000
35,000 - _ .■ ■ . ■■■
.'.:) , ;■',:■': . . . . . ,'..'■
30,000 -
1"otal UES
Increased KPM coverage
The KPM program has been effective in improving response, timeliness and quality of survey data. A great deal
of this success has been the result of lessening the burden imposed on KPM enterprises. This has been achieved
through streamlined reporting arrangements and the development and extensive use of electronic spreadsheets
customized to match their companies' accounting records and aligned to the business survey questionnaires.
Subject-matter divisions continue to refer enterprises to the KPM Program for problem resolution. To meet this
growing demand, two KPMs were introduced in the Eastern Region in 2001-02 and the working arrangement
has proven extremely successful. We are partnering with Transportation Division to introduce "subject matter
KPMs". The objectives are to extend the coverage of KPM treatment without additional resources and to give
subject matter managers experience in dealing with large, complex businesses. As envisioned, the subject
matter KPM would have responsibility for a small number of companies and would focus on maintaining rather
than building the relationships. The KPM Program would continue to provide overall coordination and support.
1
Λ completion lime of 2 hours per questionnaire is used based on the median time reported by UES respondents.
Chart 3 - KPM Enterprise Coverage -1997 to 2003
200 478-
163 163 163 T67™
150 JL31
104
BFmmF
91 97 96
100
50
0
1997 1998 1999 2000 2001 2002 2003
Enterprises M Top Industry Contributors
Future Strategies
The UES plans to follow a much more aggressive tax replacement approach by replacing the financial data now
collected from all simple businesses with data derived from tax records. Testing will be done to determine the
feasibility of replacing complex businesses that are in a single industry. Characteristics surveys of a small
sample of tax-replaced businesses would be required to obtain non-financial information that is not available
from tax sources. The planned strategy is the development of an enhanced respondent relations program aimed
at larger businesses not currently receiving full KPM treatment. Another key piece of the vision involves
expanding the current KPM role to include coordination and transmittal of 'best practices' related to all
respondent relations activities.
Importance of our relationships with large businesses

Recognizing the increased and growing importance of large businesses, the KPM Program continues to
encourage subject-matter divisions to treat respondents as their 'customers', to foster a more enterprise-centric
view of these large businesses and to communicate corporate intelligence to agency staff. Over the next year
we will be working towards a better integration of our respondent relations and KPM programs to find new
innovative ways of dealing with more large businesses. We will work towards increasing the number of
companies covered and the knowledge we have about these important respondents, so we can increase response
rates and data quality. At the same time we will explore the possibility of consolidation between Business
Register profilers and KPMs as part of the latest Strategic Streamlining Initiatives.
Other respondent relations activities in Statistics Canada

With increased globalization and overall competitiveness, Statistics Canada's task of producing good quality
and timely business information has become increasingly difficult. The business community has become more
resistant to traditional survey collection methods and is less willing to devote the same resources it once did to
completing survey questionnaires. Collection costs have increased while response rates have been difficult to
maintain. At the same time, clients of our data have demanded quicker turnaround and better quality
information. It is no secret that sensitivity to the needs and expectations of businesses and the development of
positive relationships with respondents are critical elements to the ongoing success of any business survey
program. However, developing a strong respondent relations program must also factor in fiscal realities,
operational constraints, the needs of the client community, and technological evolution. Below is a list of the
business respondent relations-related activities currently underway at Statistics Canada:
• Use of administrative data in preference to surveys
• Measurement of respondent burden
• Electronic reporting
• Communication with respondents
• Responding to business survey complaints
m
2.5 The UES Warehouse
Characteristics
The UES Warehouse is essentially a series of related SAS data files. It is packaged together in a secure area
and is managed by Enterprise Statistics Division.
The UES Warehouse includes a number of major components:
• UES Metadata repository: provides descriptions of data stored in the warehouse
• UES processing metadata: coded instructions and tables required to process the UES using generic post
collection systems and to manage the data warehouse
• UES Central Data Store (CDS): data and processing flags for all UES surveys
• UES Data Marts: an area where UES data is repackaged to make it more useable for analysis using various
On Line Analytical Processing (OLAP) tools (ex: coherence data mart).
• UES Information Mart: includes information related to UES data such as scanned questionnaire images and
KPM reports on enterprises, comments, MIS reports
• External Data: these are data sets which originate outside ESD but have been integrated in the UES framework
(ex: business register data or tax data). The external data can be replicated (copied) from the original source
and in some cases further modified or it can be accessed directly from its place of origin through a distributed
model which avoids unnecessary data duplication
• UES Archives: repository of inactive data
The benefit of the warehouse to the subject matter, SNA and various other data clients at Statistics Canada are:
• More consistent use of terms, concepts, and classifications across the wide range of UES industries.
• More coherent data between divisional program areas and other data sources.
• More comparable data over time within and between program areas.
Roles and advantages

The UES Warehouse plays a central role in many aspects of the UES. It is the central repository for all UES
data and metadata and therefore a structured environment for maintaining and verifying content and standards.
It provides a platform from which all processing and analysis activities can be conducted (dual purpose). It
allows for the storage of multiple views and slices of data, as surveys go through each processing step. It
facilitates easy access to a wide variety of survey and administrative data for analysts across Statistics Canada
thus fostering more data sharing and cross cutting analysis. It reduces costs by eliminating the need for separate
processing and retrieval systems and databases.
The UES Warehouse facilitates many different kinds of analysis. On such example is research which allows us
to start understanding and reducing incoherence in the data collected from business enterprises. This was one
of the important objectives of the UES design. For example, variables that can be measured only at the level of
the enterprise, such as corporate profits or balance sheet elements, may not be coherent with variables that are
normally measured at the level of the establishment, such as production and sales. Through the coherence
analysis data mart (see further details below), it helps us to resolve data problems related to the creation and
death of establishments, duplication in coverage between various industry surveys and misunderstandings by
respondents in completing our questionnaires
Scope of data holdings

The UES warehouse includes all files resulting from any UES process. They can be broadly categorized as:
• front-end files from the Business Register (survey universe files, including the post-stratified version, the
sample control file, and the survey interface file),
• processing phases of the survey data and related files including tax comparison and tax replacement files, and
the questionnaire, establishment and location reference files which together elaborate the inter-relationships
of the collection entity level data coverage and the statistical entity level data which is the output of the
processing steps leading to tabulated and publishable estimates,
• other supporting and related files such as questionnaire images and metadata tables containing standard
classifications, and other code sets and data descriptions needed for the processing system and other tools,
• external files of related data which include data from the enterprise level surveys, from previous historical
iterations of the same or similar surveys or related sub-annual data for comparison and confrontation
purposes,
• cross-products or merged and/or rolled-up data from these foregoing files which contain useful process
related, or higher level and alternative aggregations of the micro-data, or statistical and economic information
derived from the micro-data intended for further analysis, reporting to the SNA and publication processes.
For the year 2002 the UES Warehouse included 42 surveys distributed over 56 separate data sources (types of
questionnaires). There were just over 41,000 simple enterprises in the UES sample for 2002. In addition the
warehouse included enterprise, establishment and location data for approximately 14,000 more complex
enterprises. In total, for year 2002 there were over 55,000 collection entities in the UES warehouse covering
97,220 establishments and 62,455 locations. UES surveys account for almost two thirds of Canadian GDP
(about $630 billion). In addition the warehouse requires 1428 cell transfonnations to convert data capture files
to the UES standards. There are almost 47,000 cells in the warehouse with almost 6300 variables and an
additional 2400 derived variables (financial and characteristics data from survey as well as comparable tax
data). Validation tables include almost 31,000 valid values for these variables. The data warehouse contains
information for the 6 full years of the UES (1997 to 2002) representing over 100 million records of data and
1.1 terabytes of storage space stored across 4 separate servers and 4 simultaneously maintained environments
(Production, Simulation, Test and Development).
Managing Respondent Burden

The role of the UES Warehouse in understanding and managing respondent burden can be viewed from a
number of different angles. The influence of the warehouse is in some cases very explicit and in other cases
more subtle. The UES has allowed for the harmonization of survey content across most of our annual business
surveys. The Chart of Accounts (COA) is now being used to further standardize the concepts used across all
UES questionnaires. In addition all available tax data is being mapped to the COA for use in replacing survey
data wherever possible. The UES strategy from the start has been to maximize the use of tax data to reduce
response burden. In as much as the UES warehouse has allowed us to get a better understanding of our data
holdings and the relationship between survey and tax data, it has allowed us to rationalize our data needs and
reach our response management objectives. As of the RY 2004 UES survey will be using tax data instead of
survey data from respondents for about 26,000 simple enterprises (59% of the simple enterprises compared to
the 2001 UES).
The UES has also reduced response burden through the use of a single unduplicated frame, common concepts,
classifications and definitions and a standard look and feel for all UES questionnaires, although the benefit to
respondents in sample is less obvious. In addition as we do more analysis of the data in the warehouse it should
allow us to identify possible overlaps in our data requests and eliminate incoherence between enterprise and
establishment data. The UES strategy is one of constant improvement under one integrated framework. More
importantly the UES Warehouse allows us to easily implement corporate initiatives and improvements, such as
our tax replacement strategy, simultaneously across a large number of surveys, since they are all integrated
under one framework. In the long term this should allow us to improve the survey process to the benefit of the
respondents, the internal users and outside clients.
We have perhaps been less successful in minimizing response burden on large complex enterprises. Ideally we
should be able to eliminate duplicate data request through the UES. In the case of some of enterprises covered
by the KPM program this has been the possible through the use of integrated questionnaires and contacts
tailored to the enterprise. In practice most large enterprises still receive a large number of separate
questionnaires (both UES and other surveys). We have not yet progressed to the stage where integrated data
requests from Statistics Canada eliminate all duplication in our questionnaires. Also we have not yet been able
to compile a complete inventory of all surveys in the field by enterprise which is something we really need to
effectively manage response burden.
A new avenue which might have great potential for further reducing response burden is XBRL. We are currently
doing some preliminary tests to see if an application of the Extensible Business Reporting Language (XBRL)
can help us to achieve this goal. Statistics Canada is working closely with selected businesses, software
providers and the Canadian Institute of Chartered Accountants to implement an XBRL taxonomy which is
mapped to our Statistics Canada COA. This may finally allow us to received unduplicated data from business
and efficiently allocate it across all of our surveys and industries.
3) Strategic Priorities, Plans and Program Options

Although much has been accomplished over the last few years, we also recognize that much remains to be done.
Current fiscal realities dictate that further efficiencies be unearthed. Extending tax replacement to all (or
almost all) simple businesses, and where possible to complex businesses, offers promise but will require
research and the use of "characteristics surveys" to collect non-financial data not available through tax.
Extending the UES to incorporate all annual business surveys in the Business and Trade Statistics Field offers
the potential for additional savings through processing efficiencies. But more importantly, centralizing these
surveys in the UES framework with its common concepts and content, survey frame, and access tools will
greatly increase the analytical potential of the UES data warehouse and free up subject matter staff to mine it.
To date, the UES data warehouse has been constructed and used primarily as a production tool. But the potential
exists to extend the content to incorporate variables from non-UES surveys (e.g. Customs / export and
investment data) that use the BR as their frame, and to use tax data to populate a virtual universe of businesses.
ESD will be consulting with industry analysts within and outside the Agency to assess demand for specific
analytical data marts to address current issues and to leverage our investment in PIPES via a much more
comprehensive data warehouse for analytical purposes.
The role of the KPM Program is also expected to evolve under the direction of the KPM Steering Committee,
with a pilot project for subject matter KPMs that will manage relationships with key industry-specific
enterprises. The KPMs are also becoming important providers of strategic information to analysts regarding
complex businesses (e.g. Auto industry and Multi National Enterprises (MNE) Project). The replacement of
simple units is increasing our dependence on larger businesses in an environment of more respondent resistance
to collection. This means we must find economical ways to improve our relations with respondents and to make
reporting easier for them through a variety of means including electronic data reporting (EDR) and more
focused content.
The greatest ongoing threat to the business statistics program is the increased reluctance of respondents to
provide data through our traditional survey questionnaires. We must change the way in which we interact with
these businesses. Respondents are now less willing to complete the large number of separate survey requests
we send them each year. They expect from us a more coordinated, integrated and unduplicated approach to data
collection, with the option of responding electronically in a user friendly environment. We need to continue to
respond to this challenge by understanding our total data needs and dealing with companies in an integrated
way (from the companies' perspective), instead of one survey at a time. Our challenge is to transform our
surveys, our collection activities and our respondent relations program to interact with companies in a way that
makes sense both for them and us. This will require a complete survey inventory of all our data requests by
company, focused on the larger businesses. We will need to coordinate all our contacts with enterprises, so that
we ask for data items only once. We must also continue to review our content and make maximum use of tax
data, while constantly improving our systems and processes. ESD and the UES have provided us with the
correct framework for doing all of this. We now need to expand this enterprise centric approach, so that we can
better manage response burden, take full advantage of our data holdings and truly interact with businesses in a
fully integrated way for all Statistics Canada surveys.
Historically, financial data have been collected from businesses via a series of regular monthly, quarterly and
annual surveys managed by multiple divisions within the bureau. With the development of a common financial
accounting standard through XBRL International ana XBRL Canada, the potential exists to re-engineer the way
the Agency collects information from large complex enterprises. ESD is currently coordinating a small pilot
project with several partner divisions and some businesses, to assess feasibility and benefits / costs of this type
of reporting. Should this exercise prove promising, plans are to extend the pilot to a number of other companies
now working with XBRL Canada. Even more cost-effective options could open up as the major suppliers of
■ura·»
business accounting software / services fully embrace XBRL and the new Canadian taxonomy into their
accounting packages. This should greatly aid companies to generate detailed "instance documents" (i.e.
standardized reports aimed at meeting particular client's needs) consistent with Statistics Canada's financial
data requirements.
4) Conclusion
The UES Warehouse has been at the center of Statistics Canada's new framework for business surveys. It has
been a challenge to implement given the very distributed nature of survey taking activities in the organization.
It has however helped us achieve a number of improvements in terms of consistency, coherence, timeliness and
overall data quality. It is also beginning to pave the way for more data analysis and a macro orientation in the
way we process (top down and drill down data validation), collect and view our data holdings, thus reducing
costs of all kinds and unnecessaiy editing of micro data.
The UES is still in a period of significant transformation. The full integration of ASM and other services
surveys, the increased use of tax data, the need to transform our respondent relations program to focus on larger
businesses, the addition of characteristics surveys, the increase in the scope of the UES Data Warehouse, the
integration of coherence analysis in the production flow and the efforts to constantly improve timeliness and
quality all contribute to a constantly changing environment. In this context the UES has not yet reached full
maturity and will continue to require further adjustments and improvements. In addition, all of the objectives
of the UES have not been fully achieved. We have made important progress towards many of these objectives
(central frame and sampling, common systems, central data store, better timeliness, coordinated respondent
relations, KPM, consistent content, etc.), but still have some areas where we need to do more (details available
for the SNA, coverage of the UES, more use of tax data, enterprisecentric approach to data collection, better
understanding of large enterprises and their data). As we strive to meet these challenges, the UES will continue
to adapt its organization and services as needed, to respond to the priorities and requirements of the
department's business survey program.
Ackn owledgements :
Much of the information in this paper was compiled using various UES documentation papers and internal
Statistics Canada documents. Thanks are due to Jim Tebrake, Wayne Baxter, Les Graham, Dennis Leblanc, Ed
Bunko, Monique Gaudreau, Tony Dupuis, Daniel Scott, Carey Olineck for their contribution through these
papers. The views expressed in this paper are those of the author and do not necessarily reflect those of Statistics
Canada.
Statistical process redesign in changing circumstances

Standardised protocols and electronic data extraction
in reporting for statistics
Jean Ritzen
Statistics Netherlands
Summary
The circumstances in which official statistics are compiled have long been subject to radical change. The
requirement for statistics is changing and still constantly increasing; hence the need for constant critical
assessment of the program. There are also increasingly vocal complaints from society in general, and above all
from businesses, about the administrative burdens involved in meeting obligations. This also undennines the
willingness of companies to provide reports for statistical purposes, even though statistics account for only a
tiny fraction of their total workload. These developments, combined with heavy pressure on the funds available,
mean that we need to find ways of compiling statistics at lower cost and with less of a burden on businesses.
Good use of ICT makes a substantial contribution here. Standardisation of content and procedures is also of
prime importance.
The social context

The following developments are relevant to the statistical work at Statistics Netherlands (CBS):
• The demand for statistics is changing and constantly increasing. This means that there is a constant need to
revise priorities. New statistical requests can be met only if funds are made available, by cutting parts of other
programmes, or by boosting the efficiency of the processes used.
The output of Statistics Netherlands has been ranked recently after a review of the statistical program. Some
existing statistics are considered posteriorities, so these will be dropped in favour of other existing products
or of new products which are to be considered as priorities. In this review many key users have been asked to
give their comments and suggestions.
• There has long been talk of reducing the budgets made available for statistics by central government. From
2006 the overall budget of Statistics Netherlands will be some 10% below its 2000 level. This percentage is
expected to rise still further when more stringent government measures come into effect.
m
The diagram shows the development' of staff size 1980 - 2004 with an extrapolation to 2010.
Staff size development 1980 -2010

(Excluding EU -extra fundings and interviewers)
fte
□ S taff size
1980 1985 1990 1995 2000 2004 2010
Businesses feel that they are carrying a heavy administrative burden from the obligations placed on them.
Statistics are one of these obligations. Even though statistics account for only a small fraction of the total,
estimated at far less than 0.5%, they are perceived as posing a heavier burden than obligations from other
areas. The target for the whole of government is to lighten the administrative load by 25% in the period 2003-
2006. Statistics Netherlands has drawn up plans for meeting this target, regardless of the fact that the
administrative burden emanating from Statistics Netherlands, expressed in real terms (hours), fell by 60%
between 1994 and 2003. It is unlikely that the perceived burden followed the same trend.
The burden can be reduced more drastically by cuts in the statistical program, new or different data-collection
methods, standardisation of concepts and procedures, and the use of ICT. In practice it will be achieved by a
combination of these options.
The diagram shows the development in burden reduction. The burden is continously measured in all
questionnaires of CBS asking how much time it takes to complete the questionnaires, including activities like
looking things up and logistics.
Administrative burden from statistics 1995 -2010

(0.2% of total national burden)
□ response
burden in
million euro
(price level 2002)
1995 2000 2004 2010
' Statistics Netherlands received some additional funds in the period 2000-2003 lo meel some new oulput requirements from regulations
m
The strategy report "Statistics that count" of May 2001 outlines the strategy of Statistics Netherlands for
meeting the challenges in changing circumstances2.
The new status of Statistics Netherlands

The status of Statistics Netherlands has changed from January 2004, from an organisational subdivision of the
Ministry of Economic Affairs into an Autonomous Agency. The background to this is a new 'CBS Law'. An
important article of this law requires statistical data to be collected at the smallest possible administrative cost to
information providers, with business surveys to be conducted only where the data are not available elsewhere,
e.g. from registers. The new CBS Law also gives the Statistics Netherlands access to all registers held by
government that contain information useful for compiling statistics. A substantial increase in the use of register
information is the crux of Statistics Netherlands policy. We refer to this as secondary data collection. F rom the
point of view of statistical quality, Statistics Netherlands will hopefully get more of a say in the content and
organisation of secondary data sources (registers). It can achieve this by fulfilling and expanding its role as the
"Bureau of Standards (BOS)". Infrastructure tools for processing information from business registers have been
developed within Statistics Netherlands. This development was made possible by a sharply increased use of ICT.
Increased efficiency and ICT

The severe budget cuts at Statistics Netherlands since 2000 have been almost entirely offset by efficiencyboosting
measures. These went hand in hand with a major reorganisation, initiated in 2000, which made the phases of the
statistical process, not the statistical product, central in determining the organisational structure. Business
statistics as well as social and regional statistics are now based on the following phases, in this order: registers
maintenance, data collection, analysis and publication. These phases are now reflected in the organisation chart,
together with support and development units. To make these changes progressively and with the desired upturn in
efficiency, there was heavy investment in ICT for example in the form of the "IMPECT" program for economic
statistics. Here, similar process phases were standardised and placed in the same "production strata". F urther
standardisation and automation are still possible. The transition between the various phases and from data
collection to analysis requires special attention. It is not only automation aspects that are important here, but also
and especially the fact that colleagues have to interpret the data differently in the different phases.
The diagram below shows the current organisational principle of Statistics Netherlands
From PRODUCT into PROCESS

One Process Statistics
Γ
2Ö- ^
1
Output I
1
1
1
Analysis .
1
Data
collection |
1
Registers !
1
1
See www.cbs.nl/organisaIion/Corporate information/Strategy report/
A renewal of the data collection strategy
A second important development as a consequence of ICT-developments is to aim to conduct many direct
surveys of businesses electronically. For a number of data-collection exercises, the percentage of reports
received by email is already up by around 50%. This is efficient not merely for Statistics Netherlands, but also
for companies themselves, since time-consuming logistic work such as form-filling is no longer necessary.
Data collection remains to serve required statistical output, which means that an optimal data collection
strategy has to be set up for the required statistical output. Optimal means optimal efficiency in that the strategy
must result in the required output at minimum costs and imposing minimal administrative burden. The
following modes for data collection are available:
•The use of data already existent and available in registrations;
• Electronically data collection;
• Estimation of the data using models (e.g. time series or indirect calculation);
• Data collection by telephone interview;
• Postal data collection using paper forms;
• Data collection by face to face interviewing (by field officers).
These modes are ranked by increasing burden, so the last-mentioned modes must be minimised in favour of the
first mentioned. Electronic data collection includes a range of variants like touch tone, e-mail, web-surveying
and automatic reporting from information systems at enterprises. A highly professional data collection
organisation is needed to optimise the data collection process. At Statistics Netherlands a Data Collection
Expertise Program has been set up for this. This program is staffed with a number of high level professionals
and works closely with other existing facilities at CBS, e.g. with the questions lab.
An example may clarify the new approach. Monthly statistics on turnover are based on a sample from the
business register. Timeliness is especially important for the first aggregate results. First results should show
valid figures on the percentage change relating to the previous month and become available within a few
weeks after the reference month. Later on the VAT-system of the fiscal authorities will yield information
on turnover, but this will come too late for the first results. In such cases we could compile a flash estimate
from data reported by an a-select relatively small number of enterprises, preferably reported electronically.
For more definitive and more disaggregated results, the information from the VAT-systems will probably
be available. This approach is called the 'multi-wave approach', in which we can wait for data becoming
available depending on the timeliness requirements. Time series estimates can also contribute to first
results.
The use of data from administrative registers requires tools to translate these data into statistical concepts,
especially into statistical units. At Statistics Netherlands this translation is done in the Baseline database which
is directly connected with the statistical business register.
An advantage of the mixed mode approach is that the enterprise situation can be taken in consideration. In the
optimal data collection strategy, data reporting by enterprises can be done in a way that fits best with enterprise
preference. This leads at least to the perception of a reduced burden and poses an interesting challenge in
communication with respondents.
The new data collection strategy is being introduced. Experiences in data collection for short-term statistics,
about 40,000 questionnaires a month in total, already show interesting results.
m
STS statistics; highly efficient in 2004
Sample
Dispatch, — Dispatch and

reminder and I reminder via
return via paper forms
Internet
·· Reminder by
L
———-1 telephone
processing info
via OCR
Aut. check and control

Manual checking
Results
!
The combination of measures in economic statistics production processes, so for SBS and STS statistics, has
resulted in a considerable reduction of staff involved. This is expressed in the next diagram with an
extrapolation to 2008.
Staff capacity for SBS and STS

350 -f
300 -'
250'
|f «j
2ÜÜ ■'
□ Annual
150' figures
100 -' ι
50 '
0> H ■pm *
1996 2000 2004 2008
It is expected that the 2008 estimate will be realised from the implementation of the described data collection
strategy. Much is expected of more sophisticated electronic data collection and the use of data in administrative
registrations.
ICT and further alleviating the administrative burden
As described earlier in this paper, Statistics Netherlands has paid much attention recently to response burden
reduction, enhancing efficiency in processes, quality of output, and timeliness. In 2003 politics set a target for
the reduction of the administrative burden in the Netherlands by 25% before 2006, to which ICT solutions must
contribute substantially.
27th CEIES seminar: Producing business statistics Costs and efficiency Recent developments and experience
One aim of the Dutch government is the multiple use of information acquired only once. This requires
standardised concepts and procedures and the use of ICT. The national ICTAL program (ICT and administrative
burden reduction) was launched in 2002. ICTAL's aim is to develop and operate standardised infrastructures
that enable businesses to report to the government through standardised procedures, and laying solid
foundations for the multiple use ofinformation provided only once.
Statistics Netherlands is involved in this program, both in developing the infrastructure tools and the report
flows. Three national infrastructure tools are being developed within the ICTAL program:
• The official business information portal (BL): a central point providing businesses with answers to questions
on obligations and legislation in various fields. Statistics Netherlands will take part in this, starting with
information on the new CBS Law. At a later stage it will be possible to complete and send off electronic forms
from this portal. In the first operational version, businesses can find answers to frequently asked questions on
duties from the tax administration and chambers of commerce.
•The Official Business Register (BBR): this register will be the basis for a single, authentic identification of
businesses with some important characteristics. This register is of great importance to Statistics Netherlands,
because it will facilitate the necessaiy link to available government data registers. Statistics Netherlands has
developed its own tools to transform units and definitions into statistical concepts.
• Official Transaction Portal (OTP): this will be the official electronic government address for reports from
businesses. The OTP is therefore.responsible for onward routing to parties requesting government data
officially, among which Statistics Netherlands. The OTP is developing standards for data protection and
transmission.
Within the ICTAL program, Statistics Netherlands and several industry boards are conducting a monthly
turnover statistics project. On the basis of definitions and specifications, the data are derived from available
information systems at enterprises and are directed according to OTP protocols to the OTP, which sends them
on to the end consumer. This is illustrated in the following diagram:
individual
report
handling
σ> </·>
c φ
o u ·
<r» φ eφ
*-*
ω o ο>
U o «3
om- Statistics
put
A field trial is being conducted together with the Dutch Retail Board (HBD) and the Dutch Board for the Hotel
and Catering Industry (BH&C), using the NACE groups Retail Trade and Hotels and Restaurants. Use is being
made of companies' point-of-sale information systems, from which a report is compiled automatically.
Participating software developers made a special functionality for this compilation in software packages for the
reporting enterprises. At the end of the reporting period the report can be compiled directly right away, although
the enterprise may have to make some corrections. Once the report is compiled, OTP-software picks it up,
makes the necessary encryptions for security, and sends the report via the Internet to the OTP. From here the
report will be sent to the end users of the data according agreed safe protocols. Enterprises will get an electronic
message confirming that the report was received.
Negotiations with the software developers were difficult because the government does not reimburse
development costs. They must participate at their own risk.
The procedure being developed has many characteristics of secondary data collection (use of data from
administrative registers) although it is mainly primary data collection. It is the definition of the unit that causes
the difference. This is not a statistical unit, but a unit that needs a translation according to the translation
procedures for information received from administrative registers. Therefore a unique identifier for the unit
must be available in the future at the BBR. For now, the identifying number of the Chamber of Commerce is
used, this is also stored in the statistical business register and can be used for translation into statistical concepts.
This procedure may well have wider applications in the future. First and foremost are the financial data
necessary for statistics based on the STS and SBS Regulations concerned. For annual financial data, a
taxonomy is being developed in collaboration with accountancy organisations and some leading data clients,
including Statistics Netherlands and the tax authorities. This consists of specifying variables and their
definitions. XBRL was chosen as the reporting standard; the taxonomy is to be part of this.
As well as specifications of variables, the identification of units and their conversion to statistical concepts require
close attention. The aim is to derive as many reports as possible, in this case for Statistics Netherlands via software
applications, from available information systems at businesses and then to transmit these via the infrastructure tools.
Businesses themselves remain responsible for reporting. It is undesirable and unworkable for Statistics
Netherlands to become responsible for hauling in the requested business data. This is why the actual reporting
is the responsibility of businesses, even though Statistics Netherlands does its utmost to facilitate this task and
to contribute to optimising data collection tools.
The developments will have far-reaching consequences for the way in which statistics are compiled in the future
and for the associated costs.
Other developments in the Netherlands

The ICTAL developments described above will have a nation-wide impact. But other developments will also
greatly affect the compilation of statistics. One is the redesign of the Dutch social security systems into one
standardised system in which all relevant information is stored and continuously updated. This is an integrated
policy administration for social security for all individuals involved. The information can be used for many
statistics, both social an economic. This means there will be one information source instead of the many we
have now, which is a major improvement of efficiency and burden reduction. Statistics Netherlands participates
in the development as the Bureau of Standards for definitions and classifications.
For Statistics Netherlands it is a challenge to find out which data systems are available at the many official
agencies, and how data stored there can be used for statistics. This would serve as a replacement for data
collection by Statistics Netherlands itself. A point of departure can be a list of the 200 registrations which
impose the most administrative burden on enterprises. This list has been compiled within the ICTAL program.
Some statistical data collection activities are part of this list, e.g. foreign trade and monthly statistics.
Incidentally, these are also in the picture for burden reduction at Statistics Netherlands.
Concluding remarks
In the last decade Statistics Netherlands experienced many developments and changes as a result of external
and internal causes. The relationship with our environment is completely different, as is the internal
27tli CEIES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience 83
Organisation. ICT developments greatly influenced these changes. This process of change is likely to continue.
Demands for statistical information will also continue to change at an increasing pace. The pressure to be as
efficient as possible, and to lower the burden on the enterprises, will increase. This requires more complex
methodologies. Staff size is expected to fall, staff composition is being changed because in all stages of the
statistical process we now need highly educated and highly skilled professionals to meet the challenges.
Information system for the Swedish Accommodation Statistics
How to use the Information Technology in a more efficient manner in order to meet the
demands of our users and respondents -
The Swedish experience.
Martin Lagerström & Sara Frankl

Statistics Sweden
Executive Summary
Demands on increased quality regarding timeliness, relevancy, accessibility, usability and ease of supplying
data in the Swedish Accommodation Statistics have been put forward by interested parties, respondents and
users of the statistics, e.g. some of the major hotel chains and The Swedish Hotel and Restaurant Association,
SHR. Discussions on how to better meet these demands on increased timeliness, relevancy, accessibility,
usability and ease of supplying data was initiated in the summer of 2002.
In the fall of 2002 a project group with representatives from Statistics Sweden, The Swedish Tourist Authority
and SHR was put together and a project plan was formed on how to best meet the demands put forward. A
reference group consisting of representatives from some of the major hotel chains was also formed. The
mission statement of the project is to develop an integrated IT-system that meets the different requirements,
needs and prerequisites of both respondents and users. The three main objectives of the project are to increase
the value of the information to our users, to minimize the response burden and to increase the flexibility of the
system.
It was agreed on that the best way to meet the demands was to, as far as can be done, automatize the processes
of supplying, collecting and disseminating data. In general this requires a more efficient use of the Information
Technology both at the input and output side. In other words, a more flexible IT-solution is needed to integrate
the input with the output side. On the input side we stipulated three major strategies - textfile, websurvey and
paperform - in order to fulfill the mission statement and the three main objectives of the project. Our strategies
at the output side is to develop different interactive webbased solutions.
This paper presents our experiences in this work process - how we have chosen to use Information Technology
to improve the quality of the Swedish Accommodation Statistics. The main focus will be on the efforts made
to implement an automatized process of data collection using the respondents' booking systems.
Our conclusions of the work up to date are that a lot of proposals made by the users, respondents and reference
group have been fulfilled.
Introduction
Statistics Sweden has produced Accommodation Statistics since 1978, from 1995 on behalf of the Swedish
Tourist Authority, which since 1995 is the authority responsible for official statistics on Tourism in Sweden.
The Swedish Accommodation Statistics is subjected to the law of the Swedish official statistics which among
other things means that anyone supplying accommodation (of a certain extent) commercially, is obligated to
supply accommodation data to the producer of the Swedish Accommodation Statistics, i.e. Statistics Sweden.
27lh CEIES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience 85
Demands on increased quality in the Swedish Accommodation Statistics have been put forward by interested
parties, respondents and users of the statistics, e.g. some of the major hotel chains and The Swedish Hotel and
Restaurant Association, SHR. Discussions on how to better meet these demands on increased timeliness,
relevancy, accessibility, usability and ease of supplying data was initiated in the summer of 2002.
In the fall of 2002 a project group with representatives from Statistics Sweden, The Swedish Tourist Authority
and SHR was put together and a project plan was formed on how to best meet the demands put forward. A
reference group consisting of representatives from some of the major hotel chains was also formed. It was
agreed on that the best way to meet the demands was to, as far as can be done, automatize the processes of
supplying, collecting and disseminating data.
This paper presents our experiences in this work process how we have chosen to use information technology
to improve the quality of the Swedish Accommodation Statistics. The main focus will be on the efforts made
to implement an automatized process of data collection using the respondents' booking systems.
The Swedish Accommodation statistics
Collection process at present

The Swedish Accommodation Statistics is a monthly survey that consists of approximately 2 700 Hotels and
similar establishments; 1700 hotels, 300 holiday villages and 700 youth hostels. Infonnation is collected from
all hotels and similar establishments with at least five rooms or nine beds. Today there are four full time
employees that are working with the statistics. The information is collected by paper form or print outs from
booking systems and all forms are manually registered. There is no fine for non respondents.
: ; ' . : ■ . . . . . . . . . . . ■ ; : . - ; ; ; : ; y ; : ; ■ ;: ; : .
Information system for trie

Swedish Accommodation Statistics
Accommodation Statistics production process

c
V
"g 4. Reminder 2 17th ¿. sending out questionnaires
(questionnaire per fax or 20th each month for
U phone) _^^^k^«^ measuring the next month
5. Start of production 25th
ΙΛ
3. Reminder 1 10th
(postal)
«to
jg 1| Wt 6. Production & sending out

js ^^. Jm subscriptions 2th
<u
m 7. Publication of the statistics 4th
¡je
2. Last date of sending in the
questionnaire 5 t h
(Λ
Figure 1: Today's collection process.

Figure 1 illustrates the present production process of the Swedish Accommodation Statistics. The 20"' each
month questionnaires are sent out to 2 700 hotels and similar establishments with at least five rooms or nine
beds. This population and threshold level means, according to our estimates, that approximately 98 percent of
the population is covered. The last date of sending in the questionnaires is the 5th of the following month.
Thereafter, two reminders are sent to the respondents that have not answered within the original time limit. The
production starts the 25u' each month and the results are published four to five weeks after the reference month.
Dissemination
Data is disseminated 4-5 weeks after the measured month. The statistics is published at regional level on
Statistics Sweden's homepage (www.scb.se). Feedback on a number of variables, e.g. capacity, occupied room
and beds, average revenue per room and overnight stays, is submitted each month on paper to those respondents
who send in data on time. Data is also disseminated through subscriptions and single orders of statistics on
regional and local level or by markeWcompetitive profiles (sets of minimum five units).
Background
Demands and requirements - the challenge.

Users, as well as respondents, of the Accommodation Statistics has over the years put forward demands on the
collection and dissemination processes and on the content of the statistics. The demands can be summarised in
three major aims:
• to shorten the time between collection and dissemination of data
• to increase the user value for the Accommodation industry as well as for the Tourist industry in general
• to minimize response burden
All the above should also be done with kept or increased quality. Moreover the market situation in Sweden is
rather competitive in this field. There are e.g. private companies that offer instant dissemination of local level
data via the Internet.
Strategy
Statistics Sweden's strategy to meet this challenge was to start a project with the aim of meeting these demands
and requirements. A project group was formed in the autumn of 2002 with Statistics Sweden, the Swedish
Tourist Authority and the Swedish Hotel and Restaurant Association. A reference group was also formed with
the major hotel chains. On the first meeting with the reference group we set out to map needs and proposals.
The output of this analysis can be organised into the following three main sections: collection of data,
dissemination of accommodation statistics and marketing, please see below.
Collection of data
• Develop a function in the establishments booking system for automatized transfer of data to Statistics
Sweden's database
• Develop a electronic form on the Internet
Dissemination of accommodation statistics

• Develop a function that gives access to last months statistics when you have sent in this months data
• Develop a more relevant feedback to establishments that send in data on time (with e.g. prices and average
revenue per room)
• Develop a better looking feedback with better graphics and more manageable diagrams
• Develop quick feedback based on a selection of larger establishments and/or hotel chains
Marketing
• Work out a systematic plan of communication to inform about, and actively market, processed data on a
continuous basis
• Actively form opinion through The Swedish Hotel and Restaurants Association, The Swedish Tourist
Authority, Statistics Sweden and hotel chains
tumia
Working process
A project plan was formed and accepted by all parties in autumn 2002. The general financial terms of the
project is that the cost of the product development is divided equally between the three parties, the Swedish
Tourist Authority, the Swedish Hotels and Restaurant Association and Statistics Sweden. The project plan
consists of a long term objective (mission statement/vision), project objectives, organisation, financial term etc.
An abbreviated form of this project plan is presented below.
Project plan
Mission statement
• An integrated IT-system that meets the different requirements, needs and prerequisites of both respondents
and users
ftifarmnrfnn system farttm

α
Vision of tomorrow'
:#ί:ί::..
interactive web
based solutions
integrated
systems
"Lightly" interactive
feedback to respondents
Statistics packages
»■-'
- ~ ^spppj
«5 if-1
3Γ ' ι i-T" I "ifJf
ír~ s r
l·
-li i ι iinrømìli
-lini iltiìiilil
Figure. 2: Mission statement (vision of tomorrow).

Figure 2 illustrates the close connection'integration between the input and output side for this project. On the
input side three major strategies — text file, web survey and paper form are needed in order to fulfil the
mission statement and the objectives of the project. On the output side these input solutions should also be
compatible with developing interactive webbased solutions as e.g. web portal, lightly interactive feedback to
the respondents and statistical packages.
The main objectives

The three main objectives of the project are presented below:
Increasing the value of the information to the users
• Increasing usability and relevance of the feedback to our users.
cSS 27th CEIES seminar: Producing business statistics Costs and efficiency Recent developments and experience
m
• Shorten the time between the collection and dissemination of data with 50%.
Minimizing the response burden
• Make it easier for the respondents to distribute data to Statistics Sweden.
• Integrate the quality check of data with the respondents distribution of data for fewer "correction" calls.
• Motivate our respondents with better feedback
Increasing the flexibility of the system
• To a greater extent adapting the collection and feedback of data to the different requirements, needs and
prerequisites of different kinds of respondents and statistics users
Long term effects:

The project will also generate the following spin-off effects in the long term perspective:
• Increased credibility: A consequence of the users getting more involved in the process of continuous quality
work
• A mutual understanding of needs and requirements: An effect of the communication between the producers,
users and respondents - a prerequisite of successful development and quality work
• Statistics Sweden's understanding of the users and respondents priorities has already increased up to date and
will increase more as a result of the next steps in the project - with a few simple changes and improvements
the usability of the data has improved - Quality improvements
• The users insight into the production process has already increased their understanding of the problems
involved in shortening the time between collection and dissemination of data and we think this insight will
increase more in the future.
• International cooperation: Many countries face the same problems and we now have a great opportunity to
share experiences of similar projects
Strategy for the ongoing work

The project group agree that the best strategy to develop these input and output solutions efficiently is a step
by step approach, that is to say to:
• Split up the project into smaller projects with separate mission statements, objectives, financial terms and
strategies, minimizing the risk of developing huge, expensive solutions, not in line with the users demands,
needs and prerequisites
• An important condition: the respondents, users, software companies etc. must be involved in the process from
the beginning
• Qualitative and quantitative market research
Please see figure 2 in annex foran overall overview of the different subprojects.
The project so far

In October 2004 the following steps have been accomplished:
• A more relevant feedback to the respondents and users has been developed (step 1). The objective of step 1
was to increase the user value at an early stage.
• A prototype for automatic delivery of the required data directly from the hotels booking system to Statistics
Sweden's database for Accommodation Statistics has been developed in close cooperation with one of the
major hotel chains, (step 2). The major objective of this second step - to develop a prototype - was that the
work carried out would serve as an important input for further work with the general text file solution. The
output of this work was a proposal for a file design which could be used for this purpose. Thereafter we held
an information meeting to inform about the project etc. The proposal for file design was then circulated
among the major software suppliers of booking systems for comments and followed up with individual
meetings between Statistics Sweden and the software company.
Ά final file design for supplying data by text file was agreed upon, tested and launched (step 3). More
information was given to respondents and users. Figure 3 below gives an overview of this process.
1
Market research (qualitative and quantitative) was conducted in order to get an in-depth understanding of the
needs and prerequisites of different respondents and users regarding IT, type of booking system in use etc.
(step 4). The purpose of the study was to gain valuable knowledge to enable us to plan the continuous work
in accordance with the strategies set up earlier.
1
In September 2004 we started to develop a web questionnaire (step 5).
information system for Ute

Suppiying data by text f He

c An application in the hotel's booking system
α/ Hotel booking system
"O creates a text file (ASCII)
β>
5 • The text file is sent to Statistics Sweden by
«Λ
FTP or by e-mail
• The process can be totally automatized or
ν>
involve a few manual elements
c
■ra
£ Accommodation Statistics' database

19
C
The data is imported into the f*?~ :
Accommodation Statistics" database
-χ
Quality checks are performed as the data
19
is imported (logical checks)
It's possible (and desirable) to build in
some quality tests in the application in
the booking system
Figure 3: Supplying data by text file.

The figure shows the process of supplying data automatically from the hotels booking systems. An application
has been developed in the hotel's booking system, which creates a text file. This text file is sent to Statistics
Sweden by File Transfer Protocol (FTP) or by e-mail. The process is now totally automatized only involving a
few manual elements compared to the earlier process. The data is then imported into the Accommodation
Statistics database where quality checks are performed at the same time the data is imported. It is possible (and
also desirable) to build in some quality tests in the application in the booking system.
The present situation regarding the text file solution is that two software suppliers have developed the
application and two more will start working with it during fall. Recently more software suppliers have promised
to develop the application very soon.
Experiences from working with the text file

Our experiences from working with the text file so far are summarized below:
• Smaller suppliers of software seems more interested in developing the application; competitive advantage?
m
•The major software suppliers have a list of development work and a new way of supplying data to the
accommodation statistics does not have the highest priority
Strategies for the continuing work with the text file.

Our strategies for the continuing work with the text file are presented below:
• Work on the major suppliers while working with the minor ones
• The hotels demand is essential, so a lot of marketing will be done onwards!
• Once a certain volume is reached, even the major suppliers will have to develop the application
Conclusions
Conclusions of the work up to date are that a lot of the proposals made by the reference group have already
been fulfilled, see figure 4.
Regarding the collection of data almost everything has been accomplished, that is to say a function has been
developed, tested and launched for automatically transferring data from the establishments booking system to
Statistics Sweden and the project of developing a Form on the Internet has just started and will be finished in
October 2005.
Regarding the marketing proposal everything has been done and more will and must be done.
Regarding the proposals made with respect to dissemination of data a more relevant feedback to establishments
has been developed, which is more userfriendly with better and more manageable diagrams. The work that
remains to do with respect to dissemination is to develop a function that gives access to last months statistics
and quick feedback based on a selection of larger establishment.
D Proposais from the reference group

'■, \;\
c Collection of data
ο φ · F unction in the establishments booking system for automatized transfer of data to
«Λ Statistics Sweden's database
Form on the Internet
1 Dissemination of accommodation statistics
Function that gives access till last months statistics then you have sent in this
C months data
•t9
L·· • Relevant feedback to establishments that send in data on time (p rises and RevPar)
• Better looking feedback with better graphics and more manageable diagrams
c
g · Quick feedback based on a selection of larger establishments and/or hotel chains
l/>
Marketing
.ïï G · Inform about and actively market processed data
# · · Actively form opinion through The Swedish Hotel and Restaurants Association,
The Swedish Tourist Authority, Statistics Sweden and hotel chain s
Figure 4: Proposals from the reference group.
T h e next steps
Textfile
The next steps in the project regarding the text file consist mainly of marketing and communication of the
advantages of delivering information by text file. According to our estimations based on the software
companies that already have developed this solution it only takes 11 hours in total for the software company to
develop this text file solution, which implies that it should not be any major obstacles with respect to the
continuing process regarding this.
Web questionnaire - step 5

The next major step needed in order to fulfil the mission statement is to develop a web questionnaire and
solutions for interactive web based dissemination of data.
The purpose of the subproject 5 (web questionnaire) is to:
• Minimize the response burden
• Make the production more efficient
The project group has recently agreed upon the financial terms and timetable regarding the subproject 5. The
project started in September 2004 and the questionnaire will be launched in October 2005. Experiences and
conclusions from the qualitative and quantitative pilot study will be used for developing a user-friendly Web
portal. Another important input to this work is the Web portal that Statistics Norway launched (please see
separate paper).
Interactive web based solutions - step 6

The purpose of step 6 is to develop a more user friendly presentations of data. This project has only been
discussed at the present moment and no formal contract regarding this has been finalized yet.
Annex
Projectplan Information System

702233-8
s— Statistics Sweden, TD och SHR

Promet 1000
-* Develop new feedback Subobjective 1
Project 2000
Textflle — * ■ Subobjective 4
Project 2100
—►
Prototype Subobjective 2
Projekt 3000
-* Marketresearch —► Subobjective 3
Projekt 4000
-* inaire
Web questionnaire —► Subobjective 6
:"■■'.■■.". ■ : ■■■■■ .■■:■:■:■ : . -'■:' ■:"■■ .-: . ' ■ ' :■':.'■■ ■ ■■: - : ■■. " ■ ■ ". .. ' . : ■ :■- : - ■■'■■' : ■- :■ ■ . :■■ .-:-:--■--;■■■■-;---■, - - ■ ■ -
Projekt 5000
-* Webdissemin.
Projekt 5100
-*' Feedback Subobjective 5
P 5200
-* Webportal Subobjective 7
Projekt 6000
-* Marketing & Information
*
Figure 6: Overview of the different subprojects.
MEASURES TO IMPRO VE EFFICIENCY
Use of Administrative Data
Business statistics based on admin istrative data ■

The Dan ish case
Sven Egmose
Statistics Denmark
Executive summary
Statistics Denmark is the central statistical office of Denmark, and the principal responsibilities and rights of
Statistics Denmark are laid down in the Act on Statistics Denmark, which was adopted by Parliament in 1966
with later amendments. The Act on Statistics Denmark grants access to the basic data necessary to produce the
statistics, and in that respect, the act states that public authorities must supply the information they possess
when requested to do so by Statistics Denmark.
This provision is the basis on which Statistics Denmark has build up a statistical programme which to a very
high degree is based on data from administrative sources. This means that Statistics Denmark has been able to
build a rather comprehensive statistical system at a rather low cost and without posing a totally overwhelming
burden on persons and enterprises.
The provision in the act is very important but in addition to that Statistics Denmark has build up a good and
trustful cooperation with the administrative bodies which collect administrative data. Regarding business
statistics, this concerns in particular the Tax and Customs authorities with whom it has even been possible to
make an agreement that purely statistical information is collected in connection with one of their collection of
administrative data.
In particular, the social statistics are based on data from administrative registers, but also some very important
business statistics are based on data from administrative registers. For the business sector, it is possible to make
up statistics on main economic features and employment by using data from administrative registers, but data
for more specific features such as statistics on products used in the production or products sold, no
administrative data exist. For such statistics, Statistics Denmark also depends on surveys.
In the paper, some important structural statistics based on administrative data are described, and also some
important short term statistics are described. For the short term statistics based on administrative data,
timeliness in particular is a problem. The enterprises have some respite to inform the administrative system and
that makes it very difficult to produce very quick statistics based on administrative data. This is a very severe
problem. It might mean that the statistical institutes have to bother the enterprises with statistical questionnaires
on subjects which the enterprise also reports to administrative bodies. This would be a very unfavourable
situation which should be avoided. Maybe the new technical opportunities for enterprises to report
electronically will present a solution to the problem of timeliness for statistics. In the paper, this is referred to
in the description of the statistics on turnover.
Introduction
Statistics Denmark is the central statistical office of Denmark and responsible for co-ordination of all official
statistics concerning Denmark and the Danish society. The principal responsibilities of Statistics Denmark are
laid out in the Act on Statistics, which was adopted by Parliament in 1966 with later amendments. Statistics
Denmark's principal task is to collect process and publish statistical information on social and economic
conditions. This task should be carried out using efficient and scientific methods. In addition, Statistics
Denmark may take on paid statistical activities. Statistics Denmark will also participate in international
statistical activities in order to increase the usefulness of the statistical information. And, finally, Statistics
Denmark may undertake statistical analyses and forecasts.
Statistics Denmark has been established as the central authority for Danish statistics. This means among other
things that:
• Other public authorities and institutions must co-ordinate their efforts with Statistics Denmark if they intend
to collect and prepare statistical information.
• Statistics Denmark shall be in charge of, or assist in, the organisation and utilisation of public authority's
central registers which can be used for statistical purposes.
The Act on Statistics Denmark grants access to the basic data necessary to produce the statistics. The Act states
that public authorities must supply the information they possess when requested to do so by Statistics Denmark.
The private sector is also obliged to supply certain information for statistical purposes. The justification for
Statistics Denmark's access to administrative register information is the reduction in the burden of businesses
brought about by utilising information already collected by other public authorities. In accordance with this,
Statistics Denmark has striven to develop a statistical system which to a very large extent is based on the
administrative registers of other public authorities. And Statistics Denmark is one of the statistical institutes in
the world which depends most heavily on administrative registers, in particular, concerning statistics on
persons, but also to a very large extent concerning business statistics.
Social Statistics
The introduction of the administrative central register on persons in 1968 has had an enormous influence on
social statistics in Denmark. The personal number which is the identifier in this register is widely used in all
relations and a lot of additional registers on persons has been established in Denmark also using this personal
number as the identifier. The environment in Denmark is therefore such that administrative registers for persons
on very many different items exist in Denmark. And as Statistics Denmark according to the Law has access to
all registers within the public sector there is no limitation on the access. This has, in agreement with the
intentions of the Law, been used in an unambiguous strategy to base the social statistics on data from
administrative registers.
Business Statistics
For business statistics, the picture is not quite as clear-cut as for social statistics. The reason for this, is partly
that the legislation on a unique identifier for business entities (legal units) came much later than was the case
for persons. But even more important is that there is no administrative need to have as comprehensive
administrative data on business units as there is to have data on persons.
The unique identifier for business units, the CVR-number, was determined by the law of 1999 on a central
administrative business register, which is to be used by all the public sector. From that point of time all
communication between the public sector and the business units shall utilize the CVR-number. New registers
on business topics created within the public sector will contain the identifier, the CVR-number, and already
existing administrative registers have obtained the CVR-number as an item in the registers.
The central administrative business register in itself is the most important administrative register for business
statistics. This register contains basic information on all business units in Denmark for the two business unit
types, legal units and the LKAUs belonging to the legal units. The register is updated in common by the
Companies House, the Tax and Customs Authorities, the Factories Inspectorate and Statistics Denmark.
Statistics Denmark runs its own statistical business register. This statistical register contains a full copy of the
administrative central business register and in addition it contains supplementary unit types, the enterprise
group (for the most important units), the enterprise, the LKAU belonging to the enterprises and the
administrative units which among other things are used for paying value added tax. Because of the close
connection to the administrative central business register the quality of the statistical business register is good.
That is extremely important for business statistics as almost all business statistics are based on information
from the statistical business register. That is the case both for the statistics which are categorized as based on
administrative registers and for statistics based on surveys. For both of these types of statistics that register
element exist that information from the business register is indispensable for the statistics. For surveys, the
business register gives both the frame for grossing up and the background data such as activity code, and for
statistics based on data from administrative registers, the business register gives the population and the
background data for the statistics.
Apart from the administrative central business register, the different administrative registers on business units
cover overall subjects such as general economic information. More specific data on the goods and services
produced or the process of production is not covered in administrative registers in Denmark. Consequently, the
business sector statistics which is based on administrative registers in Denmark is the general statistics on
turnover, accounts data and employment while more specific statistics on the production split up in different
goods and services and on the goods and services used for the production are based on surveys.
Nevertheless, the statistics based on administrative registers is a very significant part of the Danish business
statistics. The comprehensive and consistent use of administrative registers as a basis also for business statistics
is the major tool to restrict the statistical response burden posed on business units. For the statistical institute
the use of administrative data for statistics also is a rather cheap solution compared to a statistics based on
surveys. Surveys are very costly and by using administrative data the data are for free for the statistical institute.
The data are checked by administrative bodies for administrative purposes, which mean that the quality of the
data is rather high. But one shall not ignore that a range of statistical checks are necessary before administrative
data are useful for statistics. So even if the data itself are for free for the statistical institute, a rather substantial
work has to be done within the statistical institute before the statistics based on administrative data can be
published.
A main feature for statistics based on administrative data is that these statistics in most cases have full coverage.
The whole population will most often be covered by the administrative requirements in case and consequently
will the basis for the statistics be data for the whole population. This gives associations to structural statistics,
and in some respects the statistics based on administrative sources really are structural statistics. But for some
of the register based statistics they are at the same time short term statistics. In the following a description will
be given of five very important register based business statistics, two of them short term statistics and three
entirely structural statistics.
Short Term Statistics

Turnover and on employment are two very basic data for business statistics, and for both of these features the
main short term statistics in Denmark are based on data from administrative registers.
The Statistics on Turnover

The register based statistics on turnover is based on the business units value added statements submitted to the
Tax and Customs authorities combined with information in the statistical business register. The VAT-declaration
contains information on VAT-amounts payable and receivable as well as VAT-exempt export. These figures can
be directly converted to turnover, purchases (purchases for intermediate consumption as well as purchases of
investment goods) and export since only one VAT-rate exists in Denmark.
The duty to settle VAT which covers almost all activities is in Denmark restricted to businesses which have a
yearly turnover of 50,000 DKK (corresponding to 6,700 Euro). The smallest business units settle VAT on a half
yearly basis, business units with a yearly turnover of 1,000,000- 15,000,000 DKK settle quarterly, while
business units with a yearly turnover of more than 15 mill DKK (2 mill Euro) settle every month.
The largest units are by far the most important for this statistics, and as these units settle every month the
statistics is made up and published monthly. Those who settle monthly have to send their VAT declaration to the
Tax and Customs Authorities within 25 days after the end of the month. The settlements are registered by the
Tax and Customs Authorities and a copy of the file is given to Statistics Denmark. Up till recently Statistics
Denmark received this file around 14 days after the final day for declarations which made it possible to make
the necessary statistical checks and publish the statistics within 2 months after the relevant month.
This is a rather rapid total-coverage statistics but it is not rapid enough to meet the requirements for timeliness
set up by Eurostat and the Central Banks. These very hard requirements on timeliness is an obstacle to a wider
use of register-based statistics for short term statistics and it will be both costly for the statistical institute and
pose an additional burden on the enterprises if it will be necessary to supplement the register-based statistics
with survey statistics in order to meet the requirements on timeliness.
Maybe the new technical solutions to reporting can bridge the troubles. Nowadays more and more of the
enterprises make their VAT-declarations in an electronic way, which will bring forward the point of time when
the Tax and Customs Authorities will have a file with all VAT-reports registered. That opens new opportunities
for the statistics, and Statistics Denmark has already made a new agreement with the Tax and Customs
Authorities to get access to their file with registered VAT-reports only a few days after the deadline for VAT-
reports from the enterprises. The coverage is rather high at this early point of time, and in the latest months
Statistics Denmark has managed to make up the statistics on turnover and publish it within 1_ months after the
relevant month. In order to avoid bothering both the enterprises and ourselves with additional surveys Statistics
Denmark will make a great effort to bring the processing period further down, so that the register based short
term statistics on turnover can meet the requirements on timeliness.
As the data received from the Tax and Customs authorities are the individual declarations where each unit is
identified by an identification number which also exist in the statistical business register, the data can be linked
together with the structural data from the business register (e.g. activity code, location and ownership code).
The accounting unit will in most cases be identical with the legal unit and in cases where this is not the situation,
Statistics Denmark splits up or pools the accounting units to the corresponding legal units before the statistics
is made up.
Due to the fact that the VAT-statistics cover most business units and therefore also industries for which
information is not normally collected, great importance is attached to these statistics both as an important
source of assessing economic trends and as a source of information concerning economic variables for a range
of industries, especially in the services sector.
Besides being published as separate statistics, the VAT-data on turnover is stored in the business register and
used as an important stratification variable.
The Quarterly Statistics on Employment

The quarterly ATP statistics of the amount of work performed is compiled on the basis ofinformation from
employers on quarterly payments of contributions to the Labour Market Supplementary Pension Scheme
(ATP). The employer identification number is the same as the identification number used by the Tax and
Customs authorities. All employees aged 16-66 years and who are employed by an employer for at least 9 hours
a week are included in the schemes. The size of the contribution depends on the number of hours worked:
- Full contribution is paid for persons with more than 27 hours employment a week;
- 2/3 offrili contribution is paid for persons employed between 18 and 27 hours;
- and 1/3 of full contribution is paid for persons employed between 9 and 18 hours a week.
Structural information like industry code and type of ownership is taken from the statistical business register.
The statistics on the volume of employment is compiled by dividing the total contributions paid by each
employer with the average quarterly contributions for a full time employee. The resulting figures therefore can
be said to express the number of full time employment required to perform the work actually performed by full
time employees and part time employees, some of whom are employed only part of the quarter.
This statistics is a very important short-time indicator for employment with preliminary figures published
roughly 7 weeks after the relevant quarter. Final results are published 4-5 months after the relevant quarter.
The results from the ATP statistics are stored in the business register and are an important stratification variable.
m
Structural business statistics
The Accounts Statistics

The account statistics shall fulfil two main purposes. The first is to give an economic description of the Danish
businesses, useful for business economic analyses, for business political decisions and for evaluations of the
pursued industrial policy. The second is to be input into the compilation of the national accounts. In both areas
there is both an urgent national need and important EU-demands. The EU-demands concerning the first
purpose can be found in the regulation on structural business statistics.
To fulfil these demands the information collected on questionnaires has to be rather substantial. The
questionnaire on enterprise level comprises questions on profit and loss items (including financial items and
extraordinary items), on balance sheet items and on investment items. Besides this there are supplementary
questions on financial leasing and on goods resold without transformation. In total there are about 100
questions in the questionnaire. For enterprises with important activities in separate kind-of-activity units a
supplementary and simplified questionnaire has to be filled out for each kind-of-activity unit. The simplified
questionnaire contains questions on turnover, purchases, stocks, investments and goods resold without
transformation.
The basic data for the account statistics is (1) the filled in questionnaires sent to a stratified sample of
enterprises, (2) information from the statistical business register maintained by Statistics Denmark and (3)
information from the Tax- and Customs authorities in the form of standardised account information.
From the business register all enterprises active in the relevant calendar year is extracted with information on
activity, ownership form, number of full time employees and turnover according to VAT-settlements. On the
basis of this information the enterprises are stratified and following that a stratified sample is drawn. All
enterprises with at least 50 full time employees are included in the sample, 50 percent of the enterprises with
from 20 to 49 full time employees, 20 percent of the enterprises with from 10 to 19 full time employees and
finally 10 percent of the enterprises with from 5 to 9 full time employees. Questionnaires are sent out and the
enterprises can either return a filled in questionnaire or send a copy of their accounts from which Statistics
Denmark shall be able to extract the requested information. The last possibility is preferred by roughly 50
percent of the respondents. The sample consisted in 2002 of 9,200 enterprises.
From the Tax- and Customs authorities Statistics Denmark receives a copy of the standardised accounting
information which both companies and unincorporated enterprises since 1986 are obliged to report to theTax-
and Customs authorities on a yearly basis. For each enterprise about 20 main accounting items are reported
from the profit and loss account and the balance sheet. All market activities are in principle covered by the
obligation to report, but a number of exceptions exist, e.g. enterprises with a yearly turnover under 0.5 Mio.
DKK, enterprises registered on the official stock exchange and partnerships. In 2002, the number of enterprises
reporting to the Tax- and Customs authorities and situated in the relevant activity groups (and not included in
the above mentioned random sample) was approximately 68,200.
The total population of enterprises active in the calendar year and with the relevant activities is known from the
business register. Therefore, the number of enterprises in the residual group can be calculated to approximately
105,500 enterprises. These enterprises are nearly all relatively small, e.g. 2/3 of them have no employees.
The principle behind the following calculations is to construct a total set of accounting information on the most
detailed level (the questionnaire level) for every active enterprise. This is a three step procedure. The first step
is for every stratum to calculate distribution figures based on the information in the questionnaire sample. The
second step is for every enterprise in the Tax- and Customs file to use the relevant strata distribution figure to
calculate the sub-items under each main item. The third step is to calculate accounting figures per full time
employed for each stratum on the basis of the above mentioned information and thereafter calculate detailed
accounting figures for each enterprise in the residual group where only activity, ownership and employment is
known. For the very small units with less than half a man year of employment, the third-step-calculation is
based on information on VAT-turnover instead offrili time employment.
The enterprises in the «questionnaire» group covered in 2002 roughly 5 percent of the total number of
enterprises but covered 68 percent of total turnover. The corresponding figures for the Tax- and Customs
authorities group were 37 percent and 19 percent, and the figures for the residual group were 58 percent and
13 percent.
On the basis of this complete statistical register with detailed account information for each enterprise some
further calculations are performed with the purpose of calculating a limited set of accounts data for each local
kind of activity unit corresponding to the information collected by the questionnaire on the kind-of-activity
level. The detailed procedure will not be described here but the basic principle is to calculate standard
accounting figures per full time employed for each activity/ownership strata on the basis of accounting figures
from enterprises placed in that strata and whose local kind of activity units all have the same activity. These
standard figures are then used to split the information on enterprise level down to local kind of activity level
for enterprises with local kind of activity units in different activities.
The general register based employment statistics

The general register based employment statistics is a yearly statistics containing information on:
• number of local kind of activity units and their employment ultimo November
• full time employment and wages during the year
It is important to note ( 1 ) that number of local kind of activity units ultimo November includes units where only
the proprietor is working, (2) that employment ultimo November includes both self employed and employees
and (3) that employment ultimo November can be described by the variables socio-economic status, sex, age
and education (4) that all local kind of activity units and all employment is covered by this statistics.
The data sources for these statistics are extracts from several administrative and statistical registers among
which the business register. The primary source for the statistics is extracts of the infonnation sheets submitted
yearly by employers to the Tax- and Customs authorities concerning wages and pension payments. On this
information sheet the Tax- and Customs authorities have agreed to put a field for "code for workplace" which
is only used for statistical purposes.
This agreement with the Tax- and Customs authorities is the crucial precondition for the register based
employment statistics as every employer with more than one workplace has to fill in this field with the relevant
"code for workplace" for each of the employees.
Ifall information sheets were filled out correctly by the employers, the extract of the information sheets would
produce a list of all workplaces (local kind of activity units) and assign all jobs to the correct workplaces right
away. But of course this is not the case. Therefore a rather burdensome task has to be done by Statistics
Denmark to correct and secure the quality of the information. This task is called the work place project.
Tax and Customs authorities send out the uncompleted information sheets to all employers in
November/December each year. These forms are accompanied by workplace lists for enterprises that have
several workplaces. The workplace lists are produced by Statistics Denmark. They contain a list of the
workplaces which Statistics Denmark has recorded for the enterprise in question as they appear in Statistics
Denmark's statistical business register. The lists contain the name, address and activity code of each workplace.
Finally, the workplace lists contain the code for the workplace, which must be indicated on the information
sheet for each individual employee. The employer must update the workplace list with appropriate information
on the individual workplace, should there be errors or inconsistencies in the list. The employer must also
complete the workplace list indicating new workplaces and others that have closed so that finally the list
contains complete information on all workplaces which have been active during the course of the year. Besides
the current updating of the business register information on local kind of activity units, a complete update of
the information in the business register on the local kind of activity units is therefore done once a year. This
procedure is followed for private employers and for institutions belonging to the state.
For employers within counties and municipalities the information sheet is not used as a basis for the workplace
project, since information on workplaces etc. is already recorded in the computerised wage and salary payment
systems of local government. The workplace project therefore uses information drawn directly from these
systems which however first must be supplemented and systematised. This is a laborious process.
m
Besides the information from the workplace project on employers and their employees, information on the self-
employed can be extracted from the business register so that the total number of workplaces and total
employment on these workplaces can be defined.
The integrated enterprise statistics

This statistics is an integration of three of the aforementioned statistical products:
• The VAT-statistics
• The register based general employment statistics
• The accounts statistics
The integrated enterprise statistics combines a central of economic, employment and accounting information
on the same statistical unit level - the enterprise. The unit level in the VAT-statistics is the legal unit, the unit
level in the employment statistics is the local kind of activity unit and the unit level in the general accounts
statistics is the enterprise.
The population of the integrated enterprise statistics is defined by the statistical business register, and the
register for the statistics is made up by attaching information from the three above mentioned statistics to the
units of the statistical business register. Both for the VAT-statistics and for the accounts statistics the situation
is so that not all units are covered by the statistics. Some activities are exempted from VAT, and the accounts
statistics only covers the activities demanded by the SBS Regulation. For the variables from those statistics the
variables can only be stated for the business units covered by the statistics. For the rest of the population these
variables must remain unknown in the integrated enterprise statistics. On the other hand, for the register based
employment statistics all activities are covered and consequently employment data can be attached to all the
population of the integrated enterprise statistics.
This statistics is made by integrating different unit types and by this integration of data from different sources
of course a large number of data confrontation problems will show up. These problems are handled by the staff
of the integrated enterprise statistics by gathering supplementary information from different sources, and if
necessary some corrections to the data from the original statistics are made.
The data content is the following variables: Number of enterprises, turnover, profit or loss on ordinary
activities, export, number of full time employees, employed end of November, wages, fixed assets, net capital,
balance and investments.
m

IMPROVEMENT OF THE BENEFIT OF OFFICIAL STATISTICS
Major aspects of a harmonized system of official business statistics'
Ulrich Scheinost
ZVEI Frankfurt am Main
Major aspects of a harmonized system of official business statistics
1) Introduction
One brief preliminary remark to start with:
I understand "harmonised" to mean not just the comparability of data from different national sources but also
the comparability and linkability of data from different survey areas (e.g. industry - services), as this reflects
the purpose of the data - i.e. to portray and analyse the real economic situation.
From a German point of view at least, the cost side plays the dominant role in the current political discussions
on costs and benefits. These discussions are chiefly conducted using terms such as rationalisation, the
streamlining of statistics and reducing the burden on respondents. The content and value of the information, on
the other hand, are relegated to the background. This rather contradicts the proud claim made at the same time
that our society can be characterised as a "knowledge-based information society" and the need for individual
and general economic decision-making processes to be grounded on a solid statistical basis. This means that
the provision of statistical information involves an infrastructural dimension which demands attention and has
to be allocated sufficient resources. Given such a distorted viewpoint, how are the available data to be
evaluated'?
2) Information fields and sources
Businesses are primarily interested in information which enables them to better evaluate their market
opportunities, i.e. their attractiveness to their (current and potential) markets, as well as their (relative)
competitive position in these markets, in order then to be able to devise market strategies. (Markets are
therefore to be evaluated not just with respect to sales but also with regard to production possibilities). Five
main fields ofinformation were identified:
a) Information which is external to the business and describes the market or location (the market "demand"
profile and market attractiveness), such as
• the number of players in the market (suppliers, consumers, mediators)
• (per capita) purchasing power in the respective markets, market volume
' Author's note: the observations are based on business statistics in Germany
' Author's note: the observations are based on business statistics
• market growth, conditioned by the trend in demand, changes in legislation (recycling of electrical
appliances, environmental provisions, shop opening hours, etc.), innovations/new technologies (medicines,
Internet, Diesel - soot filters, etc.), changing values (environmental protection, globalisation, etc.)
• characteristics of the consumers (age, sex, education, profession, place of residence, Internet connection,
attitudes, etc.) for market segmentation
• consumers' demands (willingness to pay, payment conditions, location, speed of delivery, demand for
information, etc.)
• market structures (depiction of relations between market players with regard to product, financial and
information flows)
• production conditions in different locations: factor costs, taxes, suppliers, human capital, access to
knowledge (research, technology transfer) etc.
• prices in the sales markets and procurement markets
b) Information which is external to the business and describes the competition (the "suitability" profiles of
competitors) such as:
• competitive conditions: on the supply and demand side (market power, market concentration)
• market entry conditions: regulations, customs etc.
• number of competitors
• services offered by competitors and market strength (market shares, level of recognition)
• suitability profile of competitors vis-à-vis the demand profile of purchasers
• resources (staff, fixed assets, capital, core competencies, knowledge, Intellectual Property Rights)
c) in-house information to compare a business's own suitability profile with the market demand profile and the
suitability profiles of competitors and to determine the (relative) market position, as above
d) in-house operating ratios for the planning, execution and control phases, in which mainly nominal
parameters are of interest
• turnover
• costs: wages, energy, financing, taxes, intermediate consumption, etc.
• storage
• yield
• new orders received (by region)
• productivity
It is immediately apparent, above all in forecast data for planning purposes, that official statistics cannot be
the only supplier of information for businesses, but that market and economic research institutes,
associations, notary statistics, etc. also come into play.
Whilst the primary objective of official statistics is to record actual changes in living standards, they must
nonetheless attempt to portray real trends. Price trends are vital for monetary policy and calculating national
economic output, but also play a role in wage formation (productivity trends and price components).
e) Innovation, technological development, technology transfer
In addition to the infonnation on existing markets targeted by a company portfolio, infonnation on newly-
emerging markets is also required to further develop this portfolio.
New markets can also emerge or competitive positions change through technological developments. In order
to describe these, certain information (already mentioned in part above) is required on
• patent developments
27th CEtES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience 101
• licensing
• business start-ups
• new product and process developments
• State aid
• the development of market volumes in new markets
Different sources and deductions

Data of interest to businesses emanate from the businesses themselves (in-house data), from indirect sources
(patents, foreign trade flows), and from other statistics describing the market in categories a, b and d.
In principle, the only data which should be collected from businesses are those covered by their normal
information system. Given that the markets - and thus also competitors - are international, any improvements
to the statistical information base for businesses must also always be harmonised at international level.
Official statistics can always only supply aggregates, which need to be further processed for specific purposes.
Some of the key problems are:
• The institutional demarcation into businesses/firms and the market breakdown by product/service unci region
mean that when assessing competitiveness, for example, the product-specific competitive conditions cannot
be directly deduced.
• The (cross-border) links among businesses with transfer prices result in the distortion of key parameters such
as productivity and profitability and of the flows of services between economies. Given that this problem is
unlikely to be solved, it is often more important that the analysis focus on changes than on the absolute level.
Timeliness and trends are thus more important than accuracy.
• The varying quality of the data (compatibility, accuracy, timeliness, etc.) from different countries or regions.
From this viewpoint, the overriding criterion for businesses when evaluating data is that no wrong strategic
decisions (e.g. stronger business expansion in Asia instead of North America) be made.
• Official statistics can never be more than the point of departure for more thorough analyses describing the
market and competitive position.
3) Legal bases of the surveys - costs and benefits of statistics

There are many causes for justified and unjustified complaints and these partly reflect a dilemma genuinely
faced by official statistics, or the public perception of this dilemma.
Official economic statistics are compiled in Germany on the basis of laws. These specifically regulate all the
details of the respective surveys by reporting population, characteristics and periodicity for the individual
sectors of the economy (industry, agriculture, trade, insurance, Horeca), with the survey units generally being
allocated by economic activity. There are no provisions, in principle, for cross-sectoral surveys, and in this
respect limits are already placed on comparative analyses of the different sectors.
The regional (Länder) statistical offices are generally responsible for carrying out the surveys, whilst the
Federal Statistical Office {Statistisches Bundesamt) plays a steering role for the methodology, survey bases and
compilation of the results. This organisational structure is unwieldy and inflexible when it comes to changing
individual surveys and/or overall concepts.
With global efforts being made to harmonise statistics (e.g. OECD) and the EU Regulations on carrying out
uniform surveys, it has become even harder to adapt survey systems to technological and economic
developments even if this is felt to be necessary at national level. This problem begins with the implementation
bases (e.g. the definitions of nomenclatures, characteristics and methods) und carries right through to (partly
cultural) differences in opinion on the need for, and accuracy of, statistics.
The dialogue conducted among providers, producers and users on the problems with statistical surveys takes
place at many levels and is not always very transparent, especially since the different groups are often arguing
from very different viewpoints. For example, industrial enterprises subject to statistical reporting requirements
m
can contain both providers and users, who generally belong to different areas of responsibility and therefore
obviously present their differing views. Leaving aside this conflict which also exists within businesses and the
tendency to dismiss statistics as an administrative activity and thus primarily as a cost factor, it must be
acknowledged that "Industry" is not just one of the most heavily-solicited data providers but is also a very
demanding user of data. The ratio does not seem to be very balanced, in terms of taking the legitimate interests
of other «major users», such as the European Central Bank or other State authorities, into account. It is essential
to meet the demands for more detailed data below the level of the highly-aggregated sectoral data which
apparently suffice for general economic analyses. This would be beneficial in conscientiously fulfilling
reporting obligations and would thus improve the relevance and quality of the figures as a whole.
4) Differing coverage of economic sectors
Statistical Data Basis

Sections C to F NACE Rev. 1 Sections G to Ρ
Goods-Producing Industry = 33% GDP, 257 Classes 25 Services = 66% GDP

Production Statistics (Quantity and Value) 1 ) With relatively good data sources
Production by more than 6,000 items quarterly - Public Administration
External trade by 'item' and country monthly - Defense
- Social Insurance
Short Term Data monthly Problem: price deflation
Turnover - national
New orders from abroad 2) With sufficient data, e.g. from annual enterprise surveys
Employees - white and blue collar - Trade
Wages and salaries - Hotels and restaurants
Production index - Transport
Price index - Financial intermediation
- Insurance and pension funding
Structural Data Annual Problems: intermediate consumption, investments
Investments
Cost structure 3) With insufficient sources - 20% GDP
(Source: e.g. VAT Statistics]
Problems: - Real estate activities
Trade related goods _ - Leasing
Services _ - Computer and related activities
Externalisation of Functions _ - Telecommunications
- Research and development
- Other business activities
The differentiation of the statistical information on the individual economic sectors is clear from the (very
simplified) overview. It can be said that the relatively well-developed short- and long-term statistical surveys in the
industrial sector - even though these are suffering from constant erosion - are not matched by a similar survey
stmcture of the same standard for the services sectors which are of significantly greater importance for GDP
growth. Compared to the mass of statistics for agriculture, we find a genuine dearth of data in the fastest-growing
areas of product-related services. It is therefore not possible to portray either the intra-sectoral structural change
in industry or the changes between sectors. Even when it is available, statistical information is not comparable either
by type of characteristics or by scope and periodicity, even when it relates to the same market. Unfortunately, the
new Law on Statistics on Services does little to change this situation, as the existing gaps and shortcomings are not
fully remedied. Nor should the diagram lend support to the fallacy that the data for industry are in themselves
cohesive and complete. This is not the case, as here too there are many regrettable conceptual inconsistencies.
It should be stated at this point that absolutely identical economic circumstances, in accordance with the legal
reporting obligations - or rather on account of the legal regulations - lead to totally different statistical
coverage. The scope and temporal structure of notification processes are covered in different ways. One
particularly serious problem is that of the erosion of the statistical database for industry caused by differences
in coverage in the case of so-called "out-sourcing". Because of apparently structural discontinuities in the time
of the event, considerable damage is done to the time comparisons which are so popular with economic
analysts, showing, for example, a reduction in employment figures which did not take place. Similar
misinterpretations arise from the fact that finns which were established from the outset as service enterprises
are not covered at all and cannot be included in the description of the market. This leads to distorting effects on
downstream measurements and indicators for calculating sectoral figures which are also of considerable overall
economic importance, such as productivity in wage negotiations.
5) European statistics and globalisation of the economy

With the completion of the European internal market, the domestic market became less strategically important
for enterprises. This naturally also led to a shift in businesses' demands for market research: national data are
of less and less interest and only attain the desired measure of relevance within the context of European figures.
Even this European perspective, whilst it does remain the first step, is increasingly found wanting in this age
of creeping globalisation.
There has been a lot of activity in European statistics in these fields over the last decade. The EU has defined
minimum requirements and introduced harmonised nomenclatures. This involved drawing up provisions for the
national statistics which were designed to ensure that the Community has a cohesive and uniform statistical
system. It is a laborious and ongoing process which had already started back in the 1970s.
In the meantime, whilst many European figures are available in accordance with the new arrangements, by no
means all of them meet the requirements. One of the reasons for this is that the databases run by Eurostat are
still far from complete because not all.countries submit comprehensive and reasonably up-to-date data.
The harmonisation of statistics in the EU has tended rather to exacerbate many of the above-mentioned
problems. Above all, however, the goal of comparability within the EU has not been attained, as substantial
differences remain in the individual countries between the methodology and characteristics used and with
respect to timeliness. This paper cannot delve further into the serious problem of inconsistencies in the
implementation of European harmonisation measures by the national statistical authorities, and must restrict
itself here to this cursory reference. But harmonisation alone is not enough. In addition to the comparability of
data in an international context, the relevance of their content must also be guaranteed, which brings us straight
to the issue of «nomenclatures».
6) Significance of nomenclatures
Nomenclatures are a vital prerequisite for the relevance of the data collected. Unfortunately, they contain many
design flaws, as in the endeavour to accelerate the harmonisation process some key points were not properly
discussed and a lot of unsatisfactory compromises were made.
From an economic point of view, the list of breakdowns takes pole position as this is the place where decisions
are made as to what extent the data can be used for detailed analyses. Politicians, who are mainly interested in
overall figures, may view this differently. It is, however, important to bear in mind that - as one of the main
suppliers of data - the economic sector also has the right to ask for a usable output. Besides, the practicality of
a nomenclature plays a decisive role in the quality of the data collected.
The written version of this paper will contain a detailed examination of this issue, which must be omitted here
due to time constraints.
7) Quality of statistical data

The next point deals with the deterioration in quality which has crept in with the transition to new EU statistics.
This starts with the fact that impracticable nomenclatures inevitably lead to uncertainties in reporting and also
reduce the reporting body's willingness to offer reliable cooperation. For the national statistical offices, much
EU legislation effectively meant more work but with no funding. This work is not therefore done and the figures
continue to be collected and processed using a wide range of different methods. EUROSTAT, the last link in
the chain, is happy if the national offices manage to transmit figures more or less on time. As a result, the user
is continually coming across figures in the European data pool which are manifestly wrong, without however
being given the means to make any sort of plausible corrections.
Once again, there is not time to discuss any examples here, but these can also be found in the written version.
Official statistical circles are also well aware of the problems with the quality of the available European
harmonised data, as proven by the many attempts to highlight and combat the causes. These problems are,
however, ones which can be tackled by practical measures - and not by the theoretical models which have been
all the rage in recent years.
8) Product-related services alter the concept of production and the process of creating value added
Product-related services are playing an increasingly important role in the information society, and they are seen
to be of particular importance for the competitiveness of the capital goods industries. Despite its high
productivity, German industry will never be able to win any price-cutting competition with low labour cost
countries. In order to avoid price competition, German industry has been advised on many occasions to make
the changeover from manufacturing products to providing systems. This requires a full range of material goods
and product-related services. Such services cover the pre-sales aspect as well as the lifetime of the product, and
continue beyond the end of its economic life. In addition to the traditional services of training and maintenance,
this category also includes innovative engineering services, operator models, financial services, product
upgrades and collection and disposal. The increasing importance of software used to nin appliances and
systems has long been recognised.
In order to reflect the new core competencies at an internal level also, functions which were previously separate
are merged more and more frequently within businesses. Electronic sensors and actuators are linked to each
other by communications networks. Precision control devices are the key building blocks for the factories of
today and tomorrow. This fusion is not, however, limited to the traditional production process, as internal
information and communications systems are also being integrated with production.
These technical and organisational moves towards more services are also, however, driven by consumer
demand. The increasing performance levels and complexity of products and systems lead to a growing need for
advice and service. Demanding clients are not just looking for products, but tailor-made solutions which
include the requisite services. The ability to meet consumers' demands in the area as well is becoming a
decisive competitive factor alongside cost and technical aspects. The range of services which a modern
manufacturer offers - and must offer - nowadays extends from engineering via systems integration to the
provision of the necessary infrastructure and the day-to-day operation and maintenance of systems.
Product-related services play a key role in the success of businesses. The more complete the range of
complementary services offered by the manufacturer, the greater the efficiency of the manufactured products.
The spotlight is fixed not just on the technical merits of the product, but also on the services which open up
possibilities for the customer.
Product-related services have therefore radically changed business procedures, the process of creating value
added, the content of the product and the nature of investment and lie behind the «tertiarisation" of the
production of physical goods. This new orientation is not, however, adequately reflected in official statistics
which mainly give an institutional record of economic transactions. Industrial enterprises obtain the services
they require not just from specialised service providers belonging to the tertiary sector. The notion of core
competences determines which services enterprises provide for themselves and which ones they purchase from
other enterprises. The capital goods industry is the very sector in which the aforementioned tertiarisation of
enterprises is in fact taking place. The statistics do not, however, show the dramatic nature of the changes taking
place within industry in favour of services, which are the exact same as those provided and marketed outside
of industry and which fall under the heading of «other services". The new law on statistics in services is, at any
rate, incapable of doing justice to the demands for meaningful basic data in an information society, as it is based
on an old-fashioned «pigeon-holing" mentality, whereby services can only provided by those enterprises which,
by dint of their economic characteristics, have been assigned to this sector. Nor do the plans for annual reviews
of survey characteristics and - above all - the nomenclatures allow us to hope that reasonably meaningful data
for analysis will be available any time in the near future.
9) Possible solution
How then are statistics to deal with all these problems - and this is just a selected sample?
Everyone knows that new statistical ground needs to be broken:
27th CEfES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience 105
«As we approach the start of the new parliamentary term, there is a also need to develop concepts with respect
to the necessary changes in individual statistical provisions, which meet the demand for information and
efficiency by placing the emphasis on systemic relationships and the intensive use ofinformation which has
already been obtained." This can be read in a letter written on 20 January 1987 by the then President of the
Statistisches Bundesamt, Mr Holder.
Since the last "genuine" reform in 1976 with the changes in the reporting system for industry, there have been
decades of discussions on adjustments and further development. In recent times, the topics for consideration
have included re-calculation of the production index as well as the following issues:
- coverage of the "New Economy"
- coverage of biotechnology
- coverage of services in industry
These are all developments which have been taking place for some considerable time now in our economic
environment, but the analyses undertaken to date seem more a reflection of helplessness than a full command
of the situation.
Statistics do, in principle, have problems in providing a quick and relevant picture of real changes. On the one
hand, there is a justifiable interest in the continuity of data, on the other hand making any changes to our
complicated statistical apparatus is a long drawn-out affair. This becomes really difficult when the response to
changing circumstances also entails spending more money.
This institutional inertia in statistics is now coming up against an economy which is changing faster and more
radically than ever before. The main impetus for this has come from information and communications
technology, which has also made such profound changes to all processes in the "old economy" that this
distinction between "new" and "old" has been obsolete from the outset. The importance of services is a
particularly good example of this.
We need to have the courage to take a good hard look at our statistics in general conceptual terms and to devote
our resources not just to individual, more or less successful, repair measures. This obviously applies above all
at European level. This review process should also embrace academic statisticians, who to all appearances have
been very much involved in the past with analytical methods and have had less to do with the "nitty-gritty" of
the statistical data base. This has led to a wide discrepancy between the possibilities for data analysis and the
low information value of the available figures, which are not appropriate for these procedures. As a result,
theoretically rigorous models can, on the whole, only be tested using simulation data and are not operational.
There is a causal link between a further problematic development and the unsatisfactory coverage of our
statistics. The larger the extent of the shady or grey areas, the greater the recourse to ad hoc surveys conducted
by all types of bodies to close (or allegedly close) these data gaps. Even the best of these, however, can only
produce results which make no significant improvements in the consistency of the whole system- even though
this is in the economic self-interest of the bodies involved. The result is that in addition to the considerable
resources devoted to official statistics, there is further substantial expenditure on this statistical «sub-culture».
In expert circles, e.g. the German Statistical Society or the Federal Statistical Board, the aforementioned
problems as such are, essentially, not even controversial topics for discussion among the experts from the fields
of official statistics, science and business. It does, however, turn out to be particularly difficult to try and instil
the systemic mentality required to change the legal framework for compiling statistics. It is only with the aid
of a new overall concept that we will be able to improve the existing survey systems by filling in the data gaps
and making them more meaningful. The burden on respondents would then be counterbalanced by a
commensurate benefit and this recalibration would also change the whole basis of the discussions.
Improvement of official statistics:

Proposal for improving the relevance of European business statistics
Peter Boegh Nielsen

DG Enterprise, European Commission
The views expressed are ¡mrely those of the author and may not in any circumstances be regarded as stating an
official position of the European Commission.
I. Introduction
Statistical information has in recent years gained a more prominent position in political decision making. The
European Council meeting in March 2000 in Lisbon "invites the Commission to draw up an annual synthesis
report on progress on the basis of structural indicators to be agreed relating to employment, innovation,
economic reform and social cohesion ". Another example is the e-Europe Action Plan 2005 that includes a
comprehensive list of benchmarking indicators. In order to ensure that political decision-making is carried out
based on a reliable and sound basis, official statistics need to be capable of meeting the growing demand for
supply of statistical information.
The main missions of official statistics are to provide and disseminate reliable, timely and relevant statistical
information as a basis for making the democracy and society work. From a user perspective, the EU business
statistics can be considered fulfilling the criterion of reliability, but question marks has to be put to the criteria
of timeliness and relevance of the European business statistics. This paper addresses only the issue of relevance,
as a crucial element in the justification of official statistics is that the statistics produced are relevant, i.e.
reflecting the economy and society in a comprehensive and meaningful way.
The statistical production process is fundamentally based on the approach of measuring a range of phenomena
in the society and possible changes in these phenomena overtime. This approach has a potential built-in conflict
with user needs being often of urgent character focusing on changing elements due to shifting political
priorities and agenda. In order to overcome this dilemma, a number of core policy needs of a certain
"sustainable" character have to be identified by the main users and communicated to the statistical society.
Regarding EU business statistics, a useful benchmark is the needs embedded in the so-called Lisbon strategy.
The ambitious and ultimate goal of the Lisbon strategy is that the EU shall "become the most competitive and
dynamic knowledge-based economy in the world, capable of sustainable economic growth with more and better
jobs and greater social cohesion."1 This goal encompasses a range of different policy areas such as economic,
social or enterprise policy. This paper will focus only on the needs related to enterprise policy and especially
address the problem of the inadequate sectoral coverage of business statistics. The current business statistics do
not fully reflect the scale of the services sector and its economic importance which is a serious deficit in the
European Statistical System, as the statistical coverage of the services sector is crucial for the monitoring of
the progress of achieving the Lisbon goals.
1
European Council: Presidency conclusions, Lisbon European Council 23 and 24 March 2000.
The Lisbon European Council highlighted the role of services in the economy and their potential for growth
and employment creation. The importance of the services sector is justified by its sheer weight in the economy
(around 70% of EU 15 GDP) and the increasing consumption of services by manufacturing industry, affecting
the cost, price and quality of manufactured goods. Consequently, the business statistics cannot be classified as
fulfilling the requirement of relevancy, as long as the statistical coverage of the services sector is so incomplete
as it is the case with the current EU business statistics.
II. Services sector statistics: current situation

During the late eighties and the nineties national and international statistical institutes began developing
statistics on the services sector, due to their growing economic importance and the derived needs for statistical
information from different groups of users2.
As a consequence, several EU Regulations covering the services sector were adopted, i.e. on statistical business
registers, structural and short-term business statistics. Following the implementation of these Regulations, the
European Statistical System has been considerably improved. As an illustration of the improvement: before the
adoption of the Structural Business Statistics Regulation December 19963 harmonised structural business
statistics were only available for extraction, manufacturing, energy and water supply and construction. Thus,
the implementation of the SBS Regulation also covering business-related services and other services sectors
must be seen as a very important achievement.
Nevertheless, serious deficiencies in our understanding of the structure of the services sector and the factors
influencing the growth of services enterprises remain. The available statistical material does not reflect
appropriately the dominant position of services in the economy and also fails to show the extent of the
interdependence between services and the other sectors.
There are several reasons for this situation:
• Services are difficult to define and thus to measure. The main characteristics of the services sector are that
the output can generally be characterised as being of intangible nature. However, the introduction of
information technology has changed this situation in the sense that services nowadays can be imbedded in
physical products and thus delivered and stored via physical means. Normally, a service contains a
considerable element of personal contact between the producer and the client - a social inter-action. This
social interaction puts focus on the so-called "soft technology", being the qualifications of the employees in
terms of level of knowledge, capability of presenting to and communicating with the customer, etc.
• The development of new technologies, services products, business models and types of enterprise relations is
fast and changing rapidly; consequently the need for inclusion of new indicators or variables puts pressure on
the existing data collection to an extent not previously experienced.
• Furthermore, many statistical offices were faced with severe budget cuts in the nineties, hampering the
enlargement of the statistical programmes. Also the political demand for reduction of the respondent burden has
made new statistical surveys more difficult in many countries. The improvement of services statistics has to be
seen in a global context covering all statistical information to be supplied by enterprises: indicating possible
reductions in other types of business statistics or innovative approaches utilising existing data in new ways.
III. Commission Communication on Business-related services

The European Commission finds that the time has come to make a serious effort to put the economic
importance and potential of services on the EU policy agenda as a key component of its revisited
competitiveness strategy. There is a need for the Commission to enter into a dialogue with business operators,
professional organisations, unions, policy makers, economists, researchers and the media. They are, in the
opinion of the Commission, not sufficiently aware of the role of, the functioning and the potential impact that
services possess for the development of the EL economy.
!
For an overview of the conceptual development and practical measurement issues in the field of services statistics, see UN-Economic and Social Coun-
cil. Report of the Australian Bureau of Statistics on statistics of services (doc. E/CN.3/2003/12).
' Council Regulation (EC, EURATOM) N° 58/97.
There are several reasons for this lack of awareness. Firstly, the services sector comprises a large number of
different activities, which are difficult to classify and to treat under one heading. Secondly, official statistics at
national and EU level do not properly reflect the importance of services and their interaction with the rest of
the economy. Thirdly, and as a consequence of the lack of data, economic research on services is weak
compared to the primary and secondary sectors. Fourthly, services are not well organised on the business
market, mainly because of their diversity and a perceived lack of common interests of business operators.
It is against this background that the European Commission adopted the Communication on The
Competitiveness of business-related services and their contribution to the performance of European
enterprises." The Communication contains statistical and economic analysis of the services sector forming the
basis for policy actions to be implemented through a wide range of specific actions inside a coherent EU policy
framework.
Box 1. Definition of Business-related sendees in accordance with the EU statistical activity nomenclature, NACE.
Business Services (NACE 70-74) This group can be divided into two groups: 1) Knowledge-Intensive
Business Services, which are professional services, such as IT-consulting, management consulting, R&D
services, and advertising. 2) Operational services consisting of services such as industrial cleaning, security
services and secretarial services.
Distributive Trade (NACE 50-52) This group consists of enterprises facilitating the distribution of goods
and services to other sectors of the economy and to final consumers.
Network Services (NACE 40-41,60-64) This composite group consists of electricity, gas and water supply,
transport and communication services.
Financial Services (NACE 65-67) This group consists of enterprises offering inter-mediation of financial
services such as banks and insurance companies.
The Communication identifies the following key policy issues:

1. Market integration and competition in business-related services markets is not vigorous enough to ensure
and strengthen their competitiveness;
2. The inputs necessary for the production (labour qualifications, integration of ICT and capital) are lacking in
quality and quantity;
3. The outputs from the business-related services enterprises are not sufficiently transparent (standards), valued
(reporting on intangible assets) or documented (quality);
4. The provision and use of business-related services is limited in less developed regions and candidate
countries, mainly affecting SMEs and convergence processes;
5. Knowledge about the sector and the markets is scarce, hampering the decision making of enterprises and
policy makers.
The Communication states that "Improvement of the knowledge and statistical coverage of the business-related
services is an important action in order to better understand the realforces behind competitiveness. This is an
essential instrument for giving guidance to decision making by business operators, policymakers and other
stakeholders and for monitoring of progress in the implementation of policy areas of action described in this
Communication.
The importance given to the need for developing sufficient knowledge about the services sector is further
underlined by the fact that only key policy area being addressed in more details is the issue of lack of statistical
information. The Communication presented a relative detailed strategy for the development of services
statistics, briefly described below.
' COM(2003) 747 final
IV. Strategy for development of services sector statistics
In order to furnish the policy makers with needed and relevant statistical information and indicators on
business-related services as a tool in shaping Community enterprise policy, a huge range of topics has to be
addressed. As the business-related services are very heterogeneous, ranging from highly specialised,
knowledge intensive activities such as computer services and engineering services to low skilled jobs in
industrial cleaning, growing demands not only for a larger, but also for a more detailed statistical coverage are
frequently brought forward by users from governments, branch organisations, researchers, etc.
In order to or organise the expressed user needs in a constructive and forward looking manner, allowing for
future implementations of the different parts of the strategy plan, a set of building blocks for the development
of new statistics on services has been identified. The building blocks are the following:
• Structure of the services sector including issues such as enterprise dynamics, entrepreneurship, international
trade in services, (exports and imports, foreign direct investment)). Exports and imports of manufactured
goods are covered by detailed statistics, but information about the international trade in services at "product
level" is missing. Research on globalisation issues (e.g., the impact of migration of jobs to Asia, or
competitiveness issues) is high on the political agenda as well as a better knowledge on the processes
surrounding the business survival and success of service companies.
• Labour input (employment qualifications, skills, education, hours worked). To understand the job creation
processes in the services sector, reliable statistics on the qualifications of the employees such as level of
education, work experience, other skills, gender and age are essential. Research on training processes and
educational levels is also needed, in particular focusing on those skills shortages hampering the growth
possibilities of services activities.
• Technology and knowledge input (ICT usage, e-business, e-commerce, research and development, innovation
and intangibles). Innovation in the services sector is generally brought about by investment in acquisition of
new skills, new organisational structures and new ways of co-operation with customers and suppliers. These
types of innovation are fundamentally different from innovation in physical products and require new methods
of measurement, concepts and further research;
• Enterprise interactions (outsourcing, purchases of services, networking, strategic alliances, franchising,
capital chains). Existing statistics on business-related services have focused on the supply side. To understand
the development of the services sector and the competitiveness of all sectors of the economy, statistics on the
demand for, and use of, services by manufacturing enterprises and other enterprises or the public sector are
essential. From the research point of view, new concepts and approaches need to be developed to approach
new changes on forms of service provision and entrepreneurial behaviours.
• Services outputs (products, prices, productivity). The concepts of productivity work well for manufacturing
enterprises, but entail considerable uncertainty when it comes to measuring productivity in the service
sectors. The understanding of productivity in services requires new conceptual and theoretical developments
and research able to explore the factors enhancing productivity in services. The interrelations between
productivity, quality and innovations are particular outstanding for further research projects.
Figure I. Building blocks of statistics on the services sector.
TECHNOLOGICAL DEVELOHWENT AN» INfMOVATlON
.QTBER.SECICHI8 OF THE ECONOMY
V. Follow-up to the Communication

As a follow-up to the Communication the Commission has established a European Forum on Business-Related
Services, composed of the Community institutions, Member State representatives, professional organisations,
workers representations, research institutions and other enterprise-related stakeholders, with the objective of
ensuring a continuous dialogue with all involved parties on how to improve the competitiveness of business-
related services. As a first step in this direction, the Forum will assist the Commission in preparing an Action
Plan for the implementation of the key policy issues identified in the Communication.
In order to come up with proposals for targeted actions, the Forum has established working groups addressing
the different key policy issues identified in the Communication. One working group addresses the issue of
knowledge and statistics with the following twofold and mutually supportive mandate:
1. To contribute to improving knowledge about and research on services on those topics requiring more
analytical or empirical works. The working group will identify current research deficits and put forward
proposals to overcome these.
2. To develop elements to be incorporated in a long-term strategy for improving the statistical coverage of the
services sector and to present specific proposals for future statistical pilot projects on business-related services
in order to overcome the serious deficiencies in our understanding of the structure of the business-related
services sector and the factors influencing the growth of services enterprises.
The working group has not finalised its work at this point in time but some preliminary outcome of the current
discussions can shortly be described, using the following main headings:
• Promotion of integration and closer co-operation between research and official statistics
• Development of future services statistics
• Development of tools for lowering the respondent burden
• Increased use of existing data and micro data analysis
• Continuous revision of existing data collections
In this paper, only the 4 last headings shall be further described.
V.I Development of future services statistics

From the discussions crystallised a number of main topics for the development of future services statistics, also
taking the policy issues and the derived statistical needs of the Lisbon strategy into consideration. As the needs
identified in the meetings of the working group to a large extent are identical with the elements in the strategy
presented above, the contents of the main proposals will only be listed.
/ . Knowledge-based economy
The knowledge-based economy is characterised by the crucial importance of interaction and relations, being
between enterprises, across sectors, between employees or between suppliers and clients. It is a huge challenge
for statistical offices to develop a methodology to identify and survey these types of more soft and intangible
phenomena. Under this heading, issues such as enterprise networking, innovation in services (non-
technological innovation), human skills and knowledge management will be categorised.
2. Internationalisation of services
Industrial restructuring has been part of the economic development in the EU in the last decades, especially
influencing the manufacturing sector, and leading to a concern about a process of deindustrialisation in
Europe.5 A more recent trend, having caught high policy attention, is the apparent increasing global sourcing
of services. The concern has been based on anecdotal evidence from newspapers, as no official statistical data
covering the issue giving a global picture are currently available. Under this heading, issues as international
trade in services (detailed services product data), global sourcing of services (measurement problems, outwards
FATS) and Internal Market for services (barriers, types of services purchased across border, location of
provider) will be categorised.
.?. Competitiveness of the European services sector

Competitiveness is at the core of Community enterprise policy and the focal point of the Lisbon strategy. The
main factors enhancing competitiveness are the use of ICT by enterprises and integration into the business
processes, innovation and human capital - and the interaction of these factors. Indicators for competitiveness
have to be developed. The cmrent main indicator is productivity. With the growing importance of services and
their reported poorer productivity performance compared to manufacturing and the US services sector, the
issue of productivity in services is important for the Lisbon process.6
As mentioned above, reliable measurement of services productivity is a challenge for national statistical
institutes as we are confronted with major problems both measuring inputs (e.g. labour) and output (e.g. prices,
quality and changes in quality).
V.II Is it possible to lower the respondent burden?

On the one hand, policy shaping is hampered by the lack of statistical coverage of the services sector, on the.
other hand the administrative burden on enterprises is of concern for policy makers both at the EU and national
level.
The respondent burden must be considered a major obstacle for developing business statistics as many
statistical offices are facing obligations for considerable cuts in respondent burden to be realised in the coming
years. In reply, measures are taken by NSIs to extract the most recent ICT technologies to deliver the data
electronically from companies, even directly from business accounts (XBRL).
Another tool for replying to the user needs is to utilise existing statistical or administrative data by data linking
in new and innovative ways, the recent workshop "Services sector statistics - Future needs and possible
answers" organised by Eurostat and DG Enterprise gave examples of such uses to provide rapid information
'· COM (2004) 274 final

' O'Mahony, van Ark (ed.): EU productivity and competitiveness: An industry perspective. European Commission 2003
on topics of emerging policy interests.7 Data linking should not be restricted to only enterprise statistics but due
to the importance of human skills in understanding the performance of services, it is crucial to investigate the
possibilities of linking Labour Force Survey data with enterprise data
A further innovative tool is the use of firm level data to supplement economic analysis based on aggregated
data with new knowledge without launching new surveys.
As mentioned, NSIs are under pressure from national governments to reduce the respondent burden put on
enterprises and on the other hand obliged by EU regulations to collect a majority of the data currently in their
work programmes. There is an urgent need for revising the existing EU requirements in order to set negative
priorities and to develop procedures for continuous revision of statistical regulations and directives in order to
reflect future changes in the economy adequately.
VI. Conclusion
From the discussions in the working group and the recent workshop, the conclusion can be drawn that by taking
the different tools in terms of 1) lowering the existing respondent burden by introducing digital and
standardised reporting and revising existing data collections by introducing negative priorities solutions, 2)
utilising existing data registers in new and innovative ways by data linking and 3) supplement existing macro
analysis with firm level analysis into consideration, there is still room for manoeuvre, i.e. for introducing new
statistical data collections regarding the services sector in order to put the coverage of manufacturing and
services on equal footing.
For policy shaping purposes, official business statistics is considered the main source of statistical information
as official statistics is regarded being of high quality. But the quality of the official business statistics is
threatened, if the cunent imbalanced coverage of the economic sectors are not overcome and the flexibility and
capability of official statistics to reply to emerging policy issues not improved. On the long run, the situation
might force policy makers to base themselves on other statistical sources, if the relevancy of the EU business
statistics is not improved considerably. The on-going activities of the European Commission related to the
development of services statistics is an important action to prevent the deterioration of the quality and relevance
- and thus the utility - of official business statistics.
7 Tony Clayton: Offshore service purchases and other issues: can we tackle new questions by linking existing data sources? and Dorte Koch: Factors of
business success - results from Danish registers
27lli CEIES seminar: Producing business statistics - Costs and efficiency - Recent developments and experience 113
•urani

GENERAL STRATEGY OF N S I ' S
«Concepts and Tools for EDR

in the federal environment of Germany»
Fritz Pfrommer
Federal Statistical Office of Gennany
1. Introduction
Like many statistical organisations, the Federal Statistical Office of Germany and the Statistical Offices of the
German states have given top priority to projects which will reduce costs in statistics and improve the overall
efficiency of the statistical system. To reduce the burden upon the respondents has been on the political agenda
for many years. The use of the newest technology for collecting data und the statistical production system has
been a constant effort since the introduction of computers into statistics. There were many initiatives to increase
the use of administrative data bases and to get the data directly from the ERP-systems of the enterprises by
using electronic data reporting (EDR). For many years this has been done in an administrative environment,
which has not changed in its basic structure since the end of the sixties of the last century. But now new efforts
are being made to reform the federal division of work and to introduce new forms of cooperation and
participation between the German statistical offices. Mayor initiatives have taken place to restructure the
statistical administration corresponding to basic statistical processes and to install new forms of cost control
and product oriented budgets.
From a technical point of view there are three main fields of work:
• Standardisation of data and production
• Development of IT-Tools and of introduction of new IT-Infrastructure
• Use of administrative and business data
In all three fields the Federal statistical office has been active for many years. Therefore we have been able to
go forward on a solid base of experience. The objective of this presentation is to inform on the experience with
new IT- tools and new IT-infrastructure to support this situation. Naturally it will concentrate on the first two
topics.
2. Background
The work in statistics is part of the general development in government and society. As in all other countries
German government authorities are reacting to the combined challenge of global economy and information age
society. This challenge is a big chance for official statistics. There are several actions and programs in this
context:
2.1 E-Government
Since introducing the action plan «Innovation and Jobs in the Information Society of the 21st Century» of
September 1999 the Gennan govermnent has put forward a master plan aimed at making Germany a European
leader in the information age. The Federal Government of Germany has included in this action a general
m
program for Modernisation of Government and Administration. In the framework of this program the electronic
data reporting projects of the Federal Statistical Office have to be seen. They are considered an important part
in a general effort to improve the relation between government and citizen and to enhance the performance of
administration and economy in a global market. The Federal Statistical Office has to report regularly to the
Conference of Economic Ministers of German States on the enhancement on EDR. All government services
provided by the Federal Government will be accessible via Internet in 2005. By using internet technologies
Germany intends to participate in the global trend towards e-government und e-business.
2.2 E-Stutistics
As essential contribution to the overall objectives of e-government will be the use of modern information and
communication technologies for official statistics. It is natural to call this e-statistics. Nowadays, "modern
communications technologies" is a synonym for the Internet, so e-statistics is also very narrowly connected to
this technology. Attractive though the idea of collecting data using this technology may be, it constitutes a
challenge for traditional aspects of working with statistics. This concerns not only the use of complex
technologies that are constantly in flux, but also and especially survey design, new working methods, the
security of data transmission and the protection of confidential information.
Internet-based e-statistics has to be used to achieve both economic and qualitative targets. For statistical offices,
e-statistics provides opportunities to streamline the surveying process and simplify workflow. Thanks to basic
verification, data should already be of better quality at the input stage. This is expected to lead to more efficient
use of resources and a reduction of processing costs. In addition to modemising data collection, e-statistics also
provides an opportunity to promote the disemination of statistical data on the Internet.
2.3 Master Plan Statistics

In 2002 the federal and state courts of auditors presented a report on the production processes in the federal
statistical system. They questioned the cost efficiency and adequacy of the technology, which is used. To cope
with these questions the heads of the German statistical offices introduced a general action called "Masterplan
Statistics". This action is aimed at the improvement of statistical production and information collection. Two
essential support factors for the achievement of this strategic goal were identified:
• Information technology and statistical methology
• Development of Tools and concepts for organisational management:
- standardisation of business processes
- quality management
- budget and cost control
- accounting
An important factor was seen in the introduction of competition in the division of work between Gennan
statistical offices. Competition will improve quality and efficiency of the federal structure and create some kind
of market. Equally important will be a strategy to concentrate on core statistical competences. On the federal
and state level statistical offices have to stay close to the citizen and customer. Elements of competition and of
best practices will be combined to a kind of "optimised cooperation", which in future will lead to the
improvement of quality and the reduction of costs. That means an environment will be created which might be
called "competition federalism". In the future scenario German official statistics will consist of a network of
competent statistical institutes concentrating on core businesses like survey processing and data analysis. All
other processes basically will be delegated to "Centres of excellence", which are organized according to
principles of organisational management. High adaptability and high efficiency in the spending of resources
will make this system a reference for modern government administration.
3.The Experience with Electronic Data Reporting in Germany
3.1 Intrastat
Traditionally in Germany Foreign Trade Statistics is centrally processed in the responsibility of the Federal
Statistical Office. An important part of the staff and the budget was spent for foreign trade. For a long time
«nomi
foreign trade figures were produced by using one copy of the customs declaration. This copy was sent by the
custom authorities to the statistical office. There processing started with data entry, data editing up to table
production and publication of the figures. This situation changed when the European Union introduced the
common domestic market and customs declarations were abolished. A completely new statistical ITsystem was
installed the German Intrastat processing system (ASI). During the construction of this new the F ederal
Statistical Office got the chance to an overall modernisation of the processing of statistical returns. Many new
technical instmments were introduced by the federal statistical office:
• Scanning of paper declarations
• Optical Code recognition
• Integrated Paperless Survey Processing
• Automated Register for generation of reminders
• Propagation of data deliverance in electronic form
• Extensive use of Interactive Data Entry programs (IDES)
3.2 W3Stat-for Intrastat

Even though data transmission was included in the original concept, at the beginning only electronic media
could be used for EDI. Already in 1998 the Federal Statistical Office decided to implement an EDRsoIution
totally based on the Internet. It was called W3stat and was developed in cooperation with a German Software
Company Werum Inc. Lüneburg. The budget was provided by the Federal Statistical Office and by the Edicom
program of the European Union.
The W3stat service enabled companies to submit Intrastat declarations via the Internet. W3stat offered
providers of statistical information (PSIs) the following options for transmitting their returns to the F ederal
Statistical Office:
• online declarations with direct browserlevel input of data onto forms,
• transmission of externally prepared declaration files in all permissible formats,
• transmission of data generated by an interactive data entry software (IDES).
Since mid1999, a production version has been available. This initial version of the W3stat webserver already
featured the following functions:
• online documentation,
• guest access enabling users to try out the system,
• help functions for users (F AQ and support),
• user registration,
• downloading of IDES,
• online forms for direct submission of returns,
• transmission of IDES data,
• data encryption and security.
The use and development of EDR for Intrastat and foreign trade was a big success. In July 2004, there were
more then 26 000 registered respondents, which mean that 57 percent of all enterprises which have to provide
declarations (PSI), are registered. Over 50 percent of the data for Intrastat is sent via the W3StatSystem.
3.3 W3Statfor other centrally processed statistics

As the federal Statistical office was taking part in the ΕGovernment action of the Federal government, it had
the obligation to provide internet access for all statistics which are processed by the federal office. F or this
purpose the w3statsystem was redesigned, so that it could be used for several statistics. At the beginning of
2005 all 23 statistics in the responsibility of the Federal Statistical Office will have Online EDRfacilities. In
August 2004 there were more then 10000 providers of statistical information for these survey, which were
registered and actually sending data via Internet.
3.4 Statistik-Online
The statistical offices of Germany started in 2001 to test the Internet as a survey medium for statistics which
were processed by the state statistical offices. For this purpose, they set up a working group to develop a
concept for a single organisational approach and to lay the technical foundations for an implementation at all
the statistical offices. Within the group, the North-Rhine Westphalian Authority for Data Processing and
Statistics (Landesamt für Datenverarbeitung und Statistik Nordrhein-Westfalen - LDS NRW) carried out a
design study to develop up-to-date survey forms and sounded out respondents to gauge the degree of
acceptance. During this study forms had to be provided to the participants. In view of the limited human
resources and a short deadline a solution based on technical concepts that could be implemented quickly but
which would presumably have to be modified or even completely redeveloped for the nationwide statistical
application was chosen. With this limited solution the first online survey for North-Rhine Westphalia took place
in December 2001. The resulting system was called "Statistik-online" and is still in full production. In August
2004 more than 30 statistical applications for the state North-Rhine Westphalia were online and application
hosting for several other states is provided.
3.5 Other EDR applications in Germany

In several German states there are also application specific solutions (e. g. Bavaria for salary survey). The state
statistical office of Saxony uses its own software for Internet based surveys. This software was developed in
cooperation with a local Software company and is also implemented according to the common concepts of the
German statistical offices.
4. Standardisation of Data Formats

Currently, most data collection procedures at the Federal Statistical Office rely on proprietary, survey-specific
document and file formats for transmitting and collecting raw data. Sometimes - as in the case of the Intrastat
even multiple formats are used simultaneously for historical or technical reasons. Metadata, if collected at all,
are provided in disparate formats. All this increases development and maintenance costs and makes it difficult
or even impossible to share resources and services among surveys. Also data quality is immediately linked to
and dependent on the availability and quality of metadata throughout and at the end of the statistical production
process. All main categories of metadata have to be provided: statistical, organisational and technical metadata.
In a remarkably short period of time, the Extensible Markup Language (XML) has established itself as a core
technology. Its platform-independence, suitability for generic processing and inherent simplicity help integrate
information and systems and promise stability, secure investments and a long-term perspective for data in
rapidly changing IT and business environments.
Considering these facts the German statistical offices decided to develop common standard document formats
based on the Extensible Markup Language (XML). They were called DatML. DatML is a whole family of
document subtypes supporting the whole of the statistical production progress, starting with data collection
including data editing, dissemination and data archiving. The use of these format guarantees a consistent flow
of metadata through the whole of the statistical production progress.
Each of the formats is especially adapted to a specific purpose. There was a real challenge in modelling the
business requirements and information into these XML-based document types.
Dat/ML consists of following subformats:
• DatML/RAW - generic data type for raw data transmission
• DatML/RES - acknowledgement of transmission (complementary to DatML/RAW)
• DatML/ASK - description of electronic forms
• DatML/EDT - rules for checks and imputation in data editing
• DatML/SDF - survey description (enables generation of DatML/RAW messaged)
•urani
• DatML/MAP - mapping between different data formats ( e. g. generation of forms using DatML/RAW and
DatML/ASK )
• DatML/SSP - defines common XML elements for all Dat/ML formats (e.g. date, time, address)
• DatML/SET - document type for data archives
For Electronic data reporting the message family consisting of DatML/RAW, DatML/RES, DatML/ASK,
DatML/EDT and DatML/SDF is a sound basis to build generic tools for the statistical production process.
It is a strategic objective that DatML/RAW will be used for all electronic data transfer to statistical offices and
all electronic services will be built on it. So all raw data will be transferred into this format, regardless of
whether it is supplied in paper form, on electronic media or on-line. It can then be stored in one data base. From
this starting point the statistical production workflow can lead to the next stages also based an Dat/ML
standards.
5. The new standardized IDEV data collection software

For the working group, which was set up to develop a concept for a single organisational approach it was clear
that they had to lay the technical foundations for a unique software implementation at all the statistical offices.
They did a study on the functionality of the different solutions in Germany and designed a concept for a
common solution for the whole of German statistics.
The following basic properties were considered essential for the design of a common software tool:
Scalability: The system has to provide services for big and small surveys.
Common Interface: It has to provide a common interface for the statistical system distributing returns to the
different state statistical offices.
Economy: It should be based on open standards and use open software, so that it can easily be installed in
different locations and environments.
As the federal Statistical office had to redesign its own W3Stat-Software, it offered include all the functionality
in its new solution and to provide software, which could be used for the whole of the German statistics. It was
called IDEV (Internet Datenerhebung für den Verbund) and was finished at the beginning of 2004. Since the
beginning of 2004 it is installed at the federal Statistical office and several others offices. Extensive tests and
a number of functional adaptations have been done, so that the migration of the production to the new system
can start at the beginning 2005. As the statistical offices cannot tolerate an interruption of the service the
migration has to be done under production conditions.
Relying on the experiences with W3Stat and Statistik-Online the new software could be developed with all
basic functions considered essential for providing EDR facilities to providers of statistical information (PSI) in
Germany.
These are the following:
5.1 Editing System for data editing facilities

An important advantage of modern EDR via Internet is that one can hopefully improve data quality at the input
side. Electronic questionnaires are much better suited for supporting PSI to provide correct statistical
information. In German statistics formal check and editing rules were traditionally provided at survey design
and implemented in survey specific software. It is one of the main objectives of the federal statistical office to
be able to use fonnal descriptions to generate software modules. Naturally this should also be possible for
checks during data transmission by the web. Therefore the IDEV package includes functions for rule
generation. It will be possible, to define all checks and imputation rules and use all or part of this rules either
as client-side checks, as server-side checks or use them only for in house data editing. This gives the subject
matter statistician full flexibility to decide at which point of the data flow checks are done. Also at this point all
error messages for the different level of users are defined. The generated modules are included in standard
DatML/EDT documents.
5.2 Design and Editing of electronic forms
Even though at first sight it might be desirable to keep Online-Forms similar to paper forms, it would mean that
many interactive facilities could not be used. Therefore the working group decided to develop electronic
questionnaires separately from paper forms. There are many software products on the market which can
produce and process electronic forms. After looking at this products the working group agreed, that it is not
advisable to use one of these product. On one side these products tend to be very expensive when used with the
amount of data common for statistics. On the other side these products cannot be fully integrated into the
statistical workflow. Essential for this integration is support of the corresponding Dat/ML format family. In
developing IDEV by ourselves we were able to support these formats. The software is designed to be able to
design questionnaires according to the standards common in German statistics. The rule and checks defined for
data editing can be included as DatML/EDT documents and the output is produced in standardized XML-
format (DatML/ASK).
5.3 Administration data base for Surveys, PSIs and declarations

The IDEV administration database administers surveys and users of the service. It stores all the necessary
information relating to registration of PSIs and is a means of identifying all data provided by the users of the
system. PSIs can access the database and change entries such as address and contact person, where necessary.
For the purpose of processing registration inquiries, there can be a link to other registers, and a letter of
confirmation is drawn up assigning a user name and password to an approved PSI. After the Webserver has
received a valid PSI data entry (anonymous user name), protocol data such as the time of filing are entered in
the administration database.
In this database, the processing status of a declaration can be tracked from computerised storage of the protocol
data right the way through to data input into the processing system. If the PSI so wishes, an e-mail can be
automatically generated confirming that the return is being processed.
Another important function of the administration database is to keep track on the different versions of forms
and editing rules which are provided in standard DatML/ASK and DatML/EDT formats. Forms and
verification rules often change between reporting periods. Validation rules often refer on data from other
periods. The corresponding entries in the database have to guarantee that every data entry is done with the
correct form and correct validation rules.
5.4 IDEV-Web-Server
There is a special Webserver providing the IDEV- Service. From this server, files are passed through the
firewall to an internal data distribution Server, where the declarations are accepted and then decrypted. The data
provided by the respondent is converted to DatMl/RAW and later distributed to statistical applications. The
input protocols on the data submitted are made available in the administration database.
The webserver is providing the following functions:
PSI registration and authentication

It is essential that Statistical Offices know for certain the identity of the person or the enterprise submitting
data. Every information provider therefore has to be registered before he or she can use EDR. The sender of a
declaration can then be authenticated. User names and passwords are being employed as a means of identifying
users. Depending on the statistic either all respondents are sent a letter containing a user name with password
or respondents can register online via the Internet using a special registration form. Usemame and password
have to be entered for each session and are the sole means of identifying the PSI up to the point at which the
data are accepted into the processing system.
Processing of Online returns

The mechanism for filing returns on-line is specially suited for PSIs who have to make only a few entries per
reporting period. Forms are generated from DatML/ASK and DatML/EDT documents and provided for input.
Input data can be stored in different formats locally or on the server. This stored data can be used to keep
information from one period to the next.
Transmission of declaration files
If a PSI prepares a return locally, files thus prepared can be transmitted to the Statistical Office via IDEV Even
though for the future DatML/RAW will be the preferred data format, all other proprietary formats which are
permissible for returns to Statistical Offices are also supported, as are standard data compression program
formats (such as .zip). It is planned to convert all data to DatMl/RAW. A file dialogue facility enables the user
to select the declaration file on his or her local computer. File transmission can subsequently be activated.
5.5 Common online Raw Data entry (Core)

Electronic questionnaires and data uploading on the Web may be very fine for small enterprises or for
enterprise which have a low level of business process integration. Typically big companies or big service
providers have a situation, where they have to provide Statistical information in different fonns for many
statistical surveys, for themselves or their clients. This is a situation where in Germany typically they have also
to provide data to several state statistical offices. To cope with such a situation there is functionality within the
IDEV-Software to offer a unique data entry point for such PSIs. At this entry point, it will be possible to transfer
all kind of statistical information for all statistical surveys and for all recipients in German statistics. This is
quite similar to the way German tax declarations are sent to the finance administration. But this functionality
can only be used by PSIs which are able to provide their data in Dat/ML format and after a suitable XML-
definition (DatML/SSP) has been made accessible on the common statistics portal. To facilitate the generation
of XML-data they can use the repository on this portal. The federal Statistical there offers for free a library of
modules (Core.connect), which can be integrated into business software or a program to generate valid
messages locally (Core .reporter).
6. Summary and Conclusions

In German statistics we have achieved a high level of technology and experience on Electronic data reporting.
We can be very optimistic on the future progress concerning all the different components of e-statistics.
Nowadays the mayor tools are available. We make fast progress in the level of standardisation. We consider this
is the only way to do statistics in the information age. Of course there is much field work to be done. But also
there is big chance to collect good and tasty fruit.
SUMMING UP
Botho, Graf Pückler

Chairman of the CEIES Subcommittee
1. The need for statistical information on the dynamics of structural economic and social change has increased
dramatically in recent years. The causes are many. Introduction of the internal market in the early 1990s,
realisation of European economic and monetary union and institutionalisation of the European Central Bank
at the end of the 1990s and, last but not least, the decisions taken by the Lisbon summit in 2000 have been
decisive reasons for the increased need for information. This need has also been increased by globalisation
and by the move to a more knowledge-based society. Official statistics are among the most important sources
for meeting this need for information.
Production of statistical information uses up considerable resources, not only in national statistics agencies
but also among the companies surveyed. Given the tightness of public budgets and sharp cost management
in companies, the efficiency with which corporate statistics are prepared is becoming an increasingly
important element.
2. The former CEIES subcommittee "Business Statistics" looked at these questions as long as nine years ago.
It was therefore time to address this theme again in the framework of a seminar and to ascertain what results
could be achieved in terms of improved efficiency.
3. It is unarguable that decision-makers in politics, administration and business need relevant, up-to-date and
reliable statistical information in order to place their actions on a sound basis. A first finding of this seminar
was that perception of the need for and use of available statistical data is very different in the political and
business worlds. Whereas politicians tend to need highly aggregated data for their decisions, companies
require disaggregated data so that they can analyse their markets and develop business strategies. In this
regard, it was pointed out critically that the profile of statistical data on offer at EU level does not correspond
to the business community's needs.
4. Another important finding was that the need for statistical information increases in line with company size
and the degree of globalisation. Conversely, small and medium-sized enterprises (SMEs) which operate only
on regional and/or local markets overwhelmingly tend to have little or no interest in economic data, since
they believe themselves to be close to their market.
It is therefore not surprising that representatives of both national statistics agencies and the business
community reported that statistical returns are regarded by many SMEs as a bureaucratic burden. Research,
such as that carried out in the USA (Hopkins and Diversified Research Study), shows that the bureaucratic
load on SMEs is relatively heavier than that on large companies, where the smaller relative load can be
attributed to statistics-linked activities. This confirmed an observation already made by the Business
Statistics Subcommittee. Furthermore, a striking piece of evidence for this is the declining response rate
among SMEs.
5. The heterogeneity of the business world is also demonstrated by the fact that large companies with global
operations often want even more detailed statistics whereas SMEs in particular call simultaneously for more
limited statistics and hence a reduction in the bureaucratic load. In order to do justice to both these requests,
the efficiency with which statistical data are collected and processed needs to be increased.
6. Yet the question is: how can this efficiency be measured, and can it also be used for an international
benchmarking exercise? In this context the data envelope analysis method (DEA method) was proposed. This
is an instrument for measuring the technical efficiency of organisational units, for instance non-profit
organisations, public administrations or business units which do not operate directly on the market, and for
evaluating the findings. It was urged that consideration be given to carrying out work to this end at
supranational level, whose outcome would be an internationally comparable indicator which makes it
possible to benchmark the efficiency of producing business statistics.
7. Small and medium-sized enterprises in particular feel put upon by the demand for statistical data. According
to an estimate produced by the Federation of Greek Industries, a medium-sized enterprise has to earmark
0.35 man-years a year in order to respond to all the statistical returns in which this business is involved. This
represents a considerable cost.
8. The burden however is not measured in financial cost alone, but also in the perceived burden. This depends
on objective and subjective factors, for instance the current sensitivity of the person who has to prepare the
data. However, if other administrative work has to be done, a distinction is no longer made between general
administrative and statistical tasks. The time available for dealing with statistical returns in companies is also
relevant. After all, in times of hectic business activity, companies often have little time to spare, which means
that extra hours have to be worked so that surveys can be carefully completed. If the requested data first have
to be generated by the company's accounting department and this may require modifications to business
software, or if other departments of the business are involved, this increases the burden considerably. For that
reason, the conceptual design of surveys should be changed as little as possible and should also be carried
out essentially only when companies are not under pressure. With an eye to the aspect of data quality, they
should be given a reasonable timeframe for responses.
Lastly, the frequency with which companies have to take part in statistical surveys is a considerable load
factor. This frequency depends on the size of the sample, the size of the company and the size of the country
or region.
9. Many proposals were tabled for reducing the level of discontent about official statistics among respondent
companies. Avoidance of double collection of the same information, stronger orientation on the business
accounting model, improvement of information in national statistics agencies about intra-company data
production and administration, verification of survey forms to ensure that they are comprehensible,
improvement ofinformation on introduction of an integrated statistics system which also incorporates the
public domain in its data files, and identification of what data business needs. It is important that relations
between national statistics agencies and companies are placed on a footing that is usual in the business world.
This means that the conduct of these agencies vis-à-vis companies should be marked not by authoritarianism
but rather by a sense of partnership, where the emphasis should be on the benefit of information for
companies. This applies with regard to providing assistance for completing surveys and also with regard to
the willingness to provide clarification and the way information is provided to users. This is an area where
national statistics agencies should unreservedly demonstrate their service function vis-à-vis users. For
instance, this might entail companies taking part in a survey also gaining preferential treatment in the form
of advance information (executive results). Establishment of a business relations department to improve
relations with the business world would be very helpful.
10. How and with what success statistical data can be used is shown in a report by a company that is strongly
integrated in the world economy as a global player. It demonstrates the usefulness of official statistics for
forecasting corporate success indicators for identification and penetration of markets (using China as an
example). The finding that emerges is that this company has been better able to come to grips with the ups
and downs of the economic cycle than other large competitors. What is determinant for the usefulness of
official statistics for companies is that the data offered should be suitable for describing markets and market
conditions, and should make it possible for companies to produce in-house success and/or planning
indicators which allow benchmarking based on sectoral characteristics. Seen from the vantage point of a
globally active company, the usefulness of official statistics is all the greater if their data can be integrated
efficiently in the company's planning and control operations. However, this view of matters is not shared
by many small and medium-sized enterprises. The latter assess the cost aspect to be higher, and the benefit
correspondingly lower.
11. The efficiency and effectiveness of official statistics at national and supranational level depends to a
considerable extent on the quality of the findings. Despite minimum requirements and common
nomenclatures, the quality of statistical information at EU level in the industrial field is given a critical
assessment. Comparability of data is still not given. Harmonisation of statistics is inadequate. It is of
decisive importance that the content is meaningful and relevant. Both these attributes must be ensured.
12. What measures have statistics agencies taken to increase the efficiency of statistics production, reduce the
burden on companies and improve their willingness to provide information? We heard reports from the
USA, Canada, Germany, Denmark, the Netherlands and Sweden. All these countries saw themselves
confronted with the problem of having to manage growing demands with shrinking resources. They have
made considerable efforts to increase efficiency in the production of business statistics. These countries
have followed different routes. Prominence is given to the following initiatives: review of the survey
programme, standardisation of production processes, large investments in modern information and
communication technologies, introduction of modern survey and transmission methods, a n d - in particular
in the USA - more intensive communication with companies in order to secure better information about
what data companies hold and its availability. In the case of the Netherlands, the reform even went as far as
to give the agency a new status which allows it have an influence on the content of secondary sources in its
capacity as a standards agency (BOS).
In addition, many measures have been taken to facilitate data collection and thereby to reduce the burden
on companies. In all countries, part of this is growing use of administrative data, an area in which some
countries have made more progress than others. In some cases, primary surveys may only be carried out if
the data are not available in registers. Such shifts of emphasis in data production make a considerable
contribution to reducing the burden on business.
13. A particularly important factor for increasing efficiency in producing business statistics is the introduction
of modern information and communication technologies. All representatives of national statistics agencies
referred to the increased efforts that are being made in precisely this area. This is because these technologies
not only make it possible to improve internal data collection, processing and management but also increase
the efficiency of data transmission from the companies surveyed to statistics offices. This is not merely a
matter of transmitting a questionnaire by e-mail to the data collection points but also means that companies
can extract data directly from their accounting records. The introduction of EU-wide accounting rules with
standards and definitions that are binding across the Community seems to be helpful in this connection.
14. There were references in the debate to various measures for achieving efficiency improvements at
supranational level. These include the planned establishment of so-called centres of excellence as well as
the best practices method. It was pointed out that national and supranational statistics systems must at all
costs be prevented from drifting apart. Bearing in mind the scarcity of budgetary and human resources, it
is unacceptable to have two systems running in parallel.
15. Although efficiency in the production of official business statistics has increased a great deal objectively
since 1995 (much more is produced, more rapidly and without quality losses), this is not the perception of
the companies in question.
16. By way of conclusion, the findings can be summarised as follows:
• Benchmarking and best practices should become a permanent component in official statistics;
• The experience of several national statistics agencies shows that considerable efficiency gains can be
achieved through greater use of administrative data. Nevertheless, this approach also is limited^, above all
in terms of the need for up-to-the-minute data and the need for meaningful data;
• Use of the most modern technologies and survey techniques should be pushed forward in order to increase
efficiency gains. But they are only fully effective if they are also integrated with corporate information
management systems;
• Business statistics can only be produced efficiently in cooperation with the business world. Hence,
relations between official statistics agencies and businesses must be placed on a more businesslike,
partnership-type basis. Timely flanking measures designed to secure acceptance among companies are
therefore necessary;
•The official statistics system must be subjected to an ongoing improvement process. This involves
dropping statistics that have become obsolete as well as including new elements. This applies at both the
national and the supranational level. Under no circumstances must the European system and the national
systems be allowed to develop in different directions, creating a duplication of the load. This is
unacceptable bearing in mind ever scarcer resources.
All the stated objectives of the seminar were addressed and concrete proposals were tabled on how to increase
the efficiency of statistics production, reduce the burden on companies and improve assessment by companies
of the value of statistical information. The seminar also clearly highlighted that national statistics agencies have
made considerable efforts since 1995 to improve the efficiency of business statistics production and to improve
the offer of statistical information. However, it also became clear that there is still scope for further efficiency
improvements. The Subcommittee therefore calls on the Commission and national statistics agencies to step up
their efforts to improve official statistics.
OPINION
Botho, Graf Pückler

Chairman of the CEIES subcommittee
1. The information value and quality of official statistics, especially of business statistics, depend mainly on
the willingness of respondents to answer the questions of official statistics as truthfully and rapidly as
possible. Here we have the dilemma that the profile of statistical data offer - at the national level and even
more at EU level - does not correspond to the profile of businesses demand. This is why the CEIES dealt
with that issue already in 1995. Nearly 10 years later, the findings obtained at the time have not only been
confirmed now; the situation has become even worse, both from the point of view of businesses and of
official statistics. The introduction of the single market in the early 1990s, the economic and monetary union
in the late 1990s and the Lisbon summit have crucially contributed to the substantial rise in statistical
requirements on the part of the Commission. However, they are accompanied by a range of data offered to
businesses that is by far not adequate. For example, the Commission and the European Central Bank need
rather highly aggregated data of the member states, whereas the businesses wish to receive market-oriented,
i.e. detailed data. What is more, businesses are not a homogeneous group; they are quite heterogeneous not
only in terms of size but also in terms of market orientation. The larger and more globally oriented an
enterprise is, the larger will be its demand for detailed data. At the same time, there is an increasing
dissatisfaction with the government to be observed among businesses and in the society, so that activities to
be performed for the government are perceived as a burden. That effect is even stronger where those having
to perform such activities do not directly benefit from them.
2. Especially for small and medium-sized enterprises, the burden caused by demanding statistical data is
particularly heavy. However, the cost burden is not all that counts - what is also important is the burden
perceived. The latter depends not only on the actual time to be spent on completing the questionnaire but also
on how often a business is included in statistical surveys. That frequency depends, among other things, on
the economic branch and the size of the enterprise, the periodicity of the survey, the size of the Land or
region, etc.
The negative attitude is directly proportional to the complexity of the survey and to special computations that
may be required to transmit the data demanded. In order to reduce the dissatisfaction with statistics among
the questioned enterprises, many proposals have been developed, referring mainly to the following aspects
of official statistics:
(1) General approach
• Asking several times for the same information must definitely be avoided.
• The statistical offices should apply a holistic approach, including a graded programme of characteristics
and a combination of primary and secondary surveys that lives up to problems and demand.
• Surveys based on laws should be conducted only where a high response rate is required.
• Accepting best estimates where exact figures are not available in time or would involve unjustifiably
high costs.
• Gains in timeliness must not adversely affect quality where poor quality increases the risk of wrong
decisions being taken.
m
(2) Methodical-technical aspects
• Introducing cut-off limits to disburden small businesses.
• Making maximum use of administrative data.
• Close consultation should take place between users, data providers, and data producers, so that the
results of statistics meet user requirements.
(3) Actions enhancing acceptance
• Official statistics must abandon its nature of a public administrative institution and move towards a
service business. Staff must be trained accordingly. The goal must be co-operation in the spirit of
partnership rather than in a spirit of state authority. It is recommended to set up a «business relations»
department in the NSI that would be responsible for external relations with businesses. One of its tasks
might be to provide executive results to enterprises participating in the relevant survey.
• The statistics and their results must be relevant not only for politics but also for businesses; the results
of official statistics that are relevant for businesses must not be obtained as a side-product. To this end
research should be undertaken to establish what are the decisions taken by business that would benefit
from improvements in the detail ,timeliness and frequency of official statistics. Special attention should
be placed on comparability, trade , production , sales and stocks are for example all collected by
different surveys.
• More than in the past, official statistics has to point out to businesses what data are available, what
benefits they involve, and how businesses may use them for their purposes. What is crucial here is to
demonstrate to the businesses that goals can often be reached in various ways. Office staff should be
trained in their service function and be ready to give advice to businesses.
3. NSIs are in a difficult position, too. Increasing requirements are contrasted by decreasing resources. It would
be helpful in this context to model an efficiency indicator which - such as the Data Envelope Analysis
Method (DEA) - is suited to measure the efficiency of statistics production and which can be used for
international benchmarking. This would force the offices to take appropriate action.
Introducing new survey concepts and methods, standardising production processes, using new technologies,
using more register data - these are permanent tasks.
At the European level, too, statistics must be re-engineered. So-called Centres of Excellence should be set
up, i.e. one or several countries are entrusted with finding a solution to a specific problem, which will then
be taken over by the other countries. Also, the method of best practices should be introduced.
The following three-track strategy suggests where re-engineering should be started:
• Re-designing the programme of business statistics
• Facilitating response through automated data collection, and
• Extending communication with businesses.
In this context, a variety of measures have been proposed to facilitate data collection and to reduce the burden
on businesses, such as using more administrative data, automating data collection and transmission. Also, co-
operation with businesses must be intensified and shaped more closely in order to understand their
organisational structures and internal recording practices and to identify data availabilities; such knowledge of
what is feasible will then directly be used in shaping the questionnaires. Questionnaires must use clear
language, their content must be comprehensible, and they must be compatible with the terms and definitions
used within the business accounting system. Changing the catalogue of variables at short intervals should be
avoided because this can cause costly software adjustments in the businesses. Early involvement of the business
associations at the national and supranational levels is suited to increase the benefit for businesses and thus the
acceptance of statistics.
A holistic approach to business statistics would be desirable. As far as possible, existing and new surveys of
business statistics should be combined - at least in conceptual terms - to form one standard business survey.
The goal here should be to collect and represent the data in a more detailed breakdown in terms of region,
economie branch, and product. At the same time, overlaps between the individual surveys must be avoided.
Also, core services should be centralised, a data warehouse should be set up, etc.
In addition to all those measures aiming mainly at the cost side, it is also necessary to emphasise more strongly
the benefit statistics provide for businesses. Businesses need a multitude ofinformation to exploit their market
potential and to develop appropriate international market strategies. Serious gaps in the statistical system - such
as in the service sector, and especially regarding the product-related services and the intersectoral structural
change in the industry - should be closed. On a cross-country scale, especially for the EU area, the
comparability of results is of prime importance; it may be further improved if differences between applied
methods and variables and in timeliness are further harmonised at the European level.
Conclusion:
The efficiency of official statistics has improved since the first seminar was held in 1995. More data are
produced more rapidly and with better quality, even though businesses involved do not perceive it in this way.
Benchmarking and best practices should become a permanent element of official statistics. The experience
acquired by various national statistical institutes has shown that increased use of administrative data may lead
to considerable gains in efficiency. However, there are limits especially in terms of timeliness and partly in
terms ofinformation value. Also, using highly modern technologies and survey techniques has produced large
gains in efficiency. However, they will develop their full effect only if they are integrated in the business'
internal information management system. Such measures must always be accompanied by appropriate actions
attempting to obtain acceptance from businesses. This must be a permanent task. It goes without saying that the
system of official statistics must be subject to a continuous process of further development. It includes deleting
obsolete statistics and integrating new surveys. This applies both to the national and the supranational level. It
must definitely be avoided that the European system and the national systems drift apart from each other, as
this would involve double burdens which cannot be borne because of decreasing resources.
REACTION FROM EUROSTAT
Michel Glaude
Director, Eurostat
First ofall I should like to thank you for all these very interesting comments and contributions which have made
it possible to consider thoroughly this topic concerning the needs and costs of producing business statistics, for
users as well as producers.
Several matters have been raised, and among them I was particularly struck by the issues concerning the
usefulness of business statistics, a question which was put right at the outset of our work by Mr Heinemann.
Complex issue of the usefulness of statistics

This is a complex question. To be sure, developed societies need and will increasingly need quantitative data
in order to take strategic decisions, but don't let that allow us to fall into a technocratic trap since it is
perfectly clear that sometimes the figures are used merely to justify decisions which have been taken for
other reasons. A degree of modesty concerning the usefulness of this or that set of statistics is therefore
called for, and we never know how the data we put out will be used in public debate. Take, for example, the
recent case of the tremendous impact of the OECD data on the level of schooling of young people of 15
(PISA programme).
This should not however prevent us - indeed, quite the contrary - from continuing our efforts to get a better
understanding of this vital question. There are two types of ongoing action which to my mind need to be
intensified.
Paying more attention to users

The fact is that we need to move beyond the first stage of simply listing the needs and wishes of users to
get their opinion on priorities which would be regularly reviewed - every year - as statistical programmes
develop.
At Eurostat, this regular dialogue will be undertaken with the Commission's thematic directorates-general
(hearings), with the Member States (discussion of the annual work programme) and as part of CEIES.
At the national statistical institutes, discussion will be organised according to the economic, social and
institutional structure of each country.
We also need to have methods of analysing and establishing priorities in order to incorporate the various
needs which are expressed. The talks and opinions we have shared with out American and Canadian
colleagues during this seminar have given us some examples of the criteria to be used. We have to
remember, too, that the budget cuts we sometimes have to cope with also have a salutary effect of «forcing»
us to determine priorities.
Lastly, I feel that the specific nature of the ESS is a help in detecting the priorities to be accorded to the various
types of European statistics. On the one hand, listening to users and especially to firms is pared down (by a
factor of 25) and is thus more solidly founded. On the other hand, the requests from the Council - and the
Commission - are pitted against the resources which the NSIs need to call on to meet their needs in a context
of often fruitful critical discussion.
m
Better use of data
As several speakers mentioned, making better use of data is a simple and effective means of increasing the
efficiency of statistical compilation. It is also, as I see it, an essential condition for fostering useful discussion
with users. You can have an informed opinion about a product only if it has been used enough.
In this connection, I was very interested by the frequent comments during the discussions about the difficulties
of communication between firms and statisticians on the use of data. This matter of «language» was mentioned
in particular in the case of SMEs, where it was apparent that while the average burden was not so great (since
it was spread over many firms), the perceived burden was very great (when a firm is surveyed, and more
generally because small firms cannot see the point of the statistics that are compiled).
There is obviously scope here for some progress which goes beyond just offering free access to data, although
that is still an essential first step. In this regard, Eurostat is ready to work with volunteer NSIs in connection
with a «common dissemination platform» and more generally as part of the introduction of «centres of
excellence», given that initiatives in this field need to come from the NSIs.
Lastly, let us not forget what still needs to be done to allow researchers to access confidential microdata via
«safe centres».
Cutting costs
Let us consider now the second way of improving efficiency in producing business statistics, i.e. making the
most of input and cutting costs.
By first assessing costs

We were given several examples of how to achieve this aim. They show that any strategy if it is at all serious
must be based on a better assessment of costs and especially of the costs incurred by firms.
Some Member States already do this, apparently using similar methods («standard cost models»). Would it not
be possible for us to make systematic use of this kind of approach and extend it to all the Member States, thus
making some very interesting comparisons possible?
All grist to the mill

We also heard about some other ways of cutting costs.
With regard to access to administrative data, Eurostat can help some Member States by including this
requirement on a more systematic basis in draft regulations. A national approach must still be preferred,
however, especially in the medium tenn, in order to include statistical requirements at an initial stage during
any review of the arrangements for collecting administrative data which is needed as a result of technological
progress.
The work of harmonising concepts and nomenclatures will be continued. It should be pointed out that already
at NUTS-2 level of NACE 2007 departments now have 87 headings against 65 before, as part of a better balance
between statistics on industry and services.
During the seminar we have also heard a lot about the possible progress, but also about the limits, of new tools
for collecting and transmitting data. Technology is not the answer to everything, of course, but Eurostat and the
Commission need to help in popularising tools such as XBRL.
While the question of sharing the statistical burden among firms was often mentioned, the question of sharing
among countries was not. Nevertheless, with the introduction of European samples which I mentioned in my
opening paper, this is another option for reducing the overall burden imposed by European requirements which,
in my opinion, needs to be pursued.
Finally, let us not forget that Eurostat has promised and will continue to review systematically its work
programme and the related regulations as part of the «negative priorities» in order to be able to cope with new
demands while managing to keep, or even reduce, the cunent workload.
Three final comments
Thinking ahead, with a proactive attitude

The question of statistical burden is an old one, and it will stay on the agenda for a long time for various reasons:
the significance of the business lobby and the careful attention they get from the politicians, inadequate efforts
on the administrative and statistical side, etc.
To avoid the risk of any kneejerk political reaction (abolition of funds at national level or regulations at
Community level, etc), the question needs to be carefully handled (with measures and indicators) and we need
to have a proactive attitude to important improvements.
Avoid pitting European against national requirements

There are three reasons why we must not make «Brussels» or «Luxembourg» the statistical scapegoats for all
the dissatisfaction felt by firms:
- first of all, because it is of course the Member States which decide on the various European requirements in
the Council or at the European Parliament;
- secondly, because so-called «European» statistics are an essential complement to «national» statistics for
many decision-makers, and especially for firms;
- lastly, and above all, because it is the same collection system which makes it possible to satisfy all these needs.
We therefore need to construct the ESS in a complementary way and in a spirit of collaboration, while
recognising what this work of construction means at the various levels. The short-term statistics provide a good
example, with a European sample which possible national expansions can usefully complement.
Exchanging good practices

In view of the interest which was aroused during the seminar by the various presentations of national
information systems for business statistics, I feel that it is vital to continue these exchanges of good practices.
I congratulate CEIES for contributing to this, but this is not the context in which it should be continued. There
are various forums at international level - ISI, I AOS, UN City Groups, etc - but the involvement of Eurostat at
DGINS level or with the regular meetings of the business statistics directors (BSDG) is a fact. Let's make an
effort to continue along this path.
PARTICIPANTS LIST
AMLER Robert BENGTSSON Gunnel BLESKESTAD Björn

Siemens AG A&D ST 1 Deputy Director- Head of Division
Gleiwitler Str. 555 Economic Statistics Statistics Norway
D-90475 Nuernberg Statistics Sweden Boks 8131 Dep.
Robert.amler@siemens.com Klostergatan 23 N-0033 Oslo
S-Örebrö bbl@ssb.no
Gunnel.bengtsson@scb.se
BØEGH NIELSEN Peter CARDOSO Anabela CLOUTIER Michel

DG Enterprise Economist Director,
Rue de la science 27 Banco de Portugal Enterprise Statistics Division
B-1040 Bruxelles Av. Alreirante Reis, 71 Statistics Canada
Peter-Boegh.Nielsen@cec.eu.int P-l 150 Lisboa Main Building, Room 1005-A
adcardoso@bportugal.pt Tunney's Pasture
Ottawa, Ontario Kl A 0T6
Canada
cloumic@statcan.ca
DEBRUYN Johan DEFAYS Daniel DJERF Olle

European Commission European Commission Economist
Eurostat Eurostat Statistics Sweden
BECH B2/409 BECH B2/380 Box 24300
Jean Monnet Building Jean Monnet Building S-10451 Stockholm
L-2920 LUXEMBOURG L-2920 LUXEMBOURG 011e.djerf@swipnet.se
Johan .debruyn@ceeu. int Daniel.defays@cec.eu. int
DONOHOETony DRAPAL Stanislav EGMOSE Sven

Head of Research and Vice President Head of Division
Information Services Czech Statistical Office Statistics Denmark
Irish Business and Employers' Napadesatem 81 Sejrøgade 11
Confederation (IBEC) CZ-Prague 10 DK-Copenhagen2100
84/86 Lower Baggot Street drapal@gw.czso.cz seg@dst.dk
IRL-Dublin 2
Tony.donohoe@ibec.ie
EPLER Margit GEORGIOU George CLAUDE Michel

Bundeskammer fur Arbeiter Principal Statistics Officer Director
und Angestellte - Statistical Service European Commission
Statistische Abteilung Michalakis Kazaolis Str. Eurostat
Prinz Eugen Strasse, 20-22 1444-Nicosia BECH B2/380
A- 1040 WIEN CYPRUS Jean Momiet Building
Margit.EPLER@akwien.or.at ggeorgiou@cystat. mof.gov.cy L-2920 LUXEMBOURG
Michel.Glaude@cec.eu.int
GÖRZIG Bernd GRÖNDAHL Ola HEILEMANN Ullrich

German Institute for Economic Adjoint scientifique Universität Leipzig
Research (DIW) Swiss Federal Statistical Office Wirtschaftsw issenschaftliche
Koenigin Luise Strasse Espace de l'Europe 10 Fakultät
D-14191 Berlin CH-2010 Neuchatel Institut für Empirische
bgoerzig@diw-berlin. de 01a.groendahI@fbs.admin.ch Wirtschaftsforschung - IEW
Marschnerstr. 31
D-04109 Leipzig
heilemann@wifa.uni-leipzig.de
•uréraí
KALENGA M'PALA Roger KAUMANNS Sven KOPSCH Günter

Chargé de Mission Assistant Head of Section Director
Politique Scientifique Fédérale Statistisches Bundesamt Statistisches Bundesamt
8, rue de la Science Graurheindorfer Strasse 198 Stresemannring 11
B-1000 Bruxelles D-53117 Bonn D-65189 Wiesbaden
kale@belspo.be Sven.kaumanns@destatis.de Guenter.kopsch@destatis.de
KOTULICS Tamas LAGERSTRÖM Martin LAMEL Joachim

Director of Business Statistics Head Travellers & Tourism Untere Oden 28
Hungarian Central Statistical Statistics Sweden A-3400 Klosterneuburg
Office Box 24 300 Joachim. lamel@aon.at
Keleti Károly u. 5-7 S-10451 Stockholm
HU-1024 Budapest Martin.lagerstrom@scb.se
Tamas.kotulics@office.ksh.hu
LARSEN Richard B. LAUWERIJS Nicole MACLEAN Ian

Director European Commission Statistics Users' Council
Confederation of Danish Industries ESTAT Lancaster House - More Lane
DK-Copenhagen V Unit A-2/CEIES Secretariat UK-Esher, Surrey KT108AP
rbl@di.dk BECHA4/123 Ian@worldtradestats. com
L-2920 LUXEMBOURG
Nicole.lauwerijs@cec.eu.int
MANCINI Andrea MESENBOURG Thomas MISÄNS Kaspars

Director of Business Statistics Assistant Director for Vice-president
ISTAT Economic Programs Central Statistical Bureau of
Via Tuscolana, 1788 U.S. Bureau of the Census Latvia
Roma 4700 Silver Hill Road Lãcplêsa Street 1
ITALY Room 2069/3 Riga
anmancin@istat. it Washington, D.C. 20233 U.S.A. LATVIA
tmesenbo@census.gov kaspars.misans@csb.gov.lv
NASLUND-FOGELBERG OSCHISCHNIG ULRIKE PAPAJ Karol

Annika Director of Statistics Department Head of the General
European Commission Austrian Chamber of Commerce Methodology Unit
ESTAT Wiedner Hauptstrasse 63 Statistical Office of the
Unit A-2/CEIES Secretariat A-1045 Wien Slovak Republic
BECHA4/124 Ulrike.oschischnig@wko.at Mileticova 3
Bâtiment Jean Monnet SK-Bratislava
L-2920 Luxembourg Karol.papaj@statistics.sk
aniiika.naslund-fogelberg@cec.eu.int
PAPO Maura PENNECK Stephen PFROMMER Fritz

Head of Unit on Business Statistics Director - Statistical Output Group Statistisches Bundesamt
ISTAT Office for National Statistics Stresemannring 11
Via Tuscolana, 1788 Room Dl/16, 1 Drummond Gate D-65189 Wiesbaden
I-Roma UK- London SW1V 2 QQ Fritz.pfrommer@destatis.de
ITALY Stephen.penneck@ons.gov.uk
Papo@istat.it
POLONI Paolo PÜCKLER Botho Graf RICE Melissa

Economist-Statistician Bundesvereinigung der Researcher
Eurotheum Deutschen Arbeitgeberverbände UK Statistics Commission
Neue Mainzer Landstr. 66 Breite Strasse 29 10 Great George ST
D-Frankfurt am Main D- 10178 BERLIN UK-London S WIP 3 AE
Paolo.poloni@ecb.int B.Graf_PueckIer@bda-online.de mrice@statscom.org.uk
RICHTER Josef RITTER Gert RITZEN Jean
Federal Economic Chamber Regierungsdirektor Program Manager
Windschutzstr. 14 Bundesministerium Statistics Netherlands
A-l 140 Vienna f. Wirtschaft u. Arbeit Klosterweg 1
Josef. richter@aon.at Villemombler Str. 76 NL-6412CN Heerlen
D-53123 Bonn jrtn@cbs.nl
Gert.ritter@bmwa.bund.de
ROBIN Yves SCHEINOST Ulrich SCHNEIDER-BÖTTCHER Irene

Chef de Service Abteilungsleiter Präsidentin
Service des Etudes et des Zentralverband Elektrotechnik u. Statistisches Landesamt des
Statistiques Industrielles Elektronikindustrie (ZVEI) Freistaates Sachesen
10, rue Auguste Blanqui Postfach 70 13 61 Macherstr. 63
F-93186 Montreuil CEDEX D-60591 Frankfurt am Main D-01917Kamenz
Yves.robin@industrie.gouv.fr GERMANY praesident@statistik.sachsen.de
scheinost@zvei.org
SCHNORR-BÄCKER Susanne SCHÖN Irina STAEGLIN Reiner

Statistisches Bundesamt European Commission Chairman of the German
Gustav Stresemann Ring 11 ESTAT Statistical Society
D-65180 Wiesbaden Unit A-2/CEIES Secretariat Free University and DIW Berlin
Susanne.schnorr-baecker@destatis.de BECH A4/128 Koenigin-Luise-Str. 5
Bâtiment Jean Monnet D-14195 Berlin
L-2920 Luxembourg rstaeglin@diw.de
Irina.schoen@cec. eu. int
STALMASKOVÁ Marianna STURM Roland TALON Guillaume

Coordinator of SBS and STS Head of Section Advisor,
Statistical Office of the Slovak DESTATIS Current Account Division
Republik Gustav-Stresemann-Ring 11 Banque de France
Mileti_ova 3 D-65189 Wiesbaden BDF-51-1562 SETEL
82467 Bratislava Roland.sturm@destatis.de F-75049 Paris Cedex 01
SLOVAK REPUBLIC Guillaume.talon@banque-france.fr
Marianna.Stalmaskova@statistics.sk
TORTOPID1S Antonios VIS Agathe ZAC H Sabine

Head - Research and Analysis Unit Advisor Head of Service Statistics
Federation of Greek Industries Eurochambres Statistics Austria
5 Xenofontos Street Avenue des Arts 19 A/D Guglgasse 13
GR-Athens Brussels A-1230 Wien
lazari@fgi.org.gr BELGIUM Sabine.zach@statistik.gv.at
vis@eurochambres.be
ZALETEL PLANTAN Mojca

Senior Adviser
Statistical Office of the Republic
of Slovenia
VozarskiPOT 12
Ljubljana
SLOVENIJA
Mojca.zaletel@gov.si
European Commission
27th CEIES seminar — Producing business statistics — Costs and efficiency — Recent developments and experiences
Luxembourg: Office for Official Publications of the European Communities
2005 — 133 pp. — 21 χ 29.7 cm
Theme: General statistics and regional

Collection: Eurostat news
ISBN 92-894-8298-2
ISSN 1725-1338
SALES AND SUBSCRIPTIONS
Publications for sale produced by the Office for Official Publications of the European
Communities are available from our sales agents throughout the world.
How do I set about obtaining a publication?
Once you have obtained the list of sales agents, contact the sales agent of your choice and
place your order.
How do I obtain the list of sales agents?
• Go to the Publications Office website http://publications.eu.int/
• Or apply for a paper copy by fax (352) 2929-42758
ω
ι
■ο
ψ
ό
■&·
ι
ο
ο
ω
m
ζ
ι
ο
ISBN 92-894-8298-2
Publications Office
Publications.eu.int
789289"482981

27th CEIES Seminar - Producing Business Statistics Cost and Efficiency Recent Developments and Experiences

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

27th CEIES Seminar - Producing Business Statistics Cost and Efficiency Recent Developments and Experiences

Caricato da

Copyright:

Formati disponibili

ISSN 1725-1338 Ζ

Dresden, 14 and 15 October 2004

Eurostat publications cover nine themes:

53 General and regional statistics External trade

gjg Economy and finance

8ÎÂÔ Population and social conditions 36 Environment and energy

Industry, trade and services Science and technology

Agriculture and fisheries

P o c k e t b o o k s : This collection provides the main indicators on a theme or Paper

P a n o r a m a of t h e E u r o p e a n U n i o n : This collection highlights the features Paper

M e t h o d s a n d n o m e n c l a t u r e s : Intended for specialists who want to consult PDF

Catalogues provide rapid access to concise information on Eurostat's products a¡ Paper

Graphs Tables as Comments Maps

27th CEIES seminar

Dresden, 14 and 15 October 2004

* * EUROPEAN =¿L Λ* General

Our website is updated daily. Visit it today and get:

• direct and free access to all Eurostat PDF publications;

European Statistical Data Support J

Journalists can contact the media support service:

Cataloguing data can be found at the end of this publication.

Luxembourg: Office for Official Publications of the European Communities, 2005

© European Communities, 2005

PRINTED ON WHITE CHLORINE-FREE PAPER

1. COST AND EFFICIENCY

1.1 TECHNIQUES TO MEASURE COST AND EFFICIENCY

1.2 PERCEPTION BY THE BUSINESS WORLD

1.2 (Continued) PERCEPTION BY THE BUSINESS WORLD

Mr R. Amler, Siemens AG, Germany

2. MEASURES TO IMPROVE EFFICIENCY

2.1 GENERAL STRATEGY OF NSI's

END OF FIRST DAY

2.2 DATA REPORTING ORGANISATION

2.3 SURVEY ORGANISATION

2.4 USE OF ADMINISTRATIVE DATA

2.5 IMPROVEMENT OF THE BENEFIT OF OFFICIAL STATISTICS

Reaction from Eurostat

END OF THE SEMINAR

The European Advisory Committee on Statistical Information in the

WELCOME TO THE PARTICIPANTS

1. REASONS FOR BEING MORE EFFICIENT (FROM CEIES POINT OF VIEW)

2. RECENT DEVELOPMENTS AND POINT OF VIEW OF EUROSTAT

COST AND EFFICIENCY

MEASURES TO IMPROVE EFFICIENCY

Summing Up and Opinion

Reaction from Eurostat

List of Participants 131

REASONS FOR BEING MORE EFFICIENT

Producing Business Statistics - Costs and Efficiency

I. Welcome speech as Vice-chairman of the CEIES

II. Wider connections between the location and the topic

This is precisely the task which the CEIES has to perform.

RECENT DEVELOPMENTS AND EUROSTAT'S STANDPOINT

2. Rapidly growing demand

ι Associations ..-'"Supranational "

3. The supply side: the European Statistical System (ESS)

4. Coordinating efforts at all levels

4.1 Priorities at national level

4.2 Priorities at European level

4.2.1 First, methodology

Figure 1: Input efficiency

3. Absolute and relative efficiency