Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
TABLE OF CONTENTS
PAGE
Acknowledgements (i)
Table of Contents (ii)
List of Tables (iv)
1. INTRODUCTION
PAGE
LIST OF TABLES
(1) The terms "family" and "household" are used loosely. The
unit of concern is generally the income unit for social
security purposes.
Employment
status Works At home Works At home Works At home
Employment
status Works At home Works At home Works At home
Wife 20.70 10.00 17.45 9.75 17.15 8.10
1 12.1 4. 9 17.0
2 13.1 6.7 20.0
3 14.5 8.0 22.5
4 15.7 9.3 25.0
5 16. 9 10.6 27.5
6 18.2 11.8 30.0
7 19.4 12.6 32.0
8 20.0 14.0 34.0
9 21.2 14.8 36.0
10 21.8 16.2 38.0
11 22.4 17.6 40.0
12 + 24.2 19.8 44.0
Note: The figures in brackets are the relativities derived if a couple with a head not working
is taken as the base. m = male, f = female.
SOURCE: Henderson, 1975, pp. 354-355.
14
Supplementary Piachaud
Benefit relativities *
relativities
Sole parent,
two children
(2 and 5 years) 1.00 1.11
Sole parent,
three children
(2, 5 and 8 years) 1.21 1.44
Sole parent,
four children
(2, 5, 8 and 11 years) 1.47 1.79
Pension Levering
Relativities Relativities
Sole parent,
one child (2 years) 0.77 0.78
one child (5 years) 0.77 0.81
Sole parent,
two children (2 and 5 years) 0.90 0. 93
Sole parent,
three children (2,5 and 8 years) 1.03 1.12
Sole parent,
four children (2,5,8 and 11 years) 1.17 1.33
Couple, no children 1.00 1.00
Couple, one child (2 years) 1.11 1.12
one child (5 years) 1.11 1.15
Couple, two children (2 and 5 years) 1.24 1.28
Couple, three children
(2,5 and 8 years) 1. 38 1.47
Couple, four children
(2, 5, 8 and 11 years) 1.52 1.67
* The "pension relativities" include family allowances; they
would have increased slightly because of actions taken in the
1983-84 and 1984-85 Budgets. "Lovering relativities" for sole
parents take as the base payment the standard rate of pension
plus mother's/guardian's allowance.
SOURCE: Lovering, 1983, p.62 and author's own calculations.
* Includes income units whose principal source of income was from business
or farm, partnership, superannuation, Interest, rent or dividend and other.
(A) Figures In brackets are equivalence scales Implicit in the level of median
and mean net incomes.
SOURCE: Australian Bureau of Statistics, Income Distribution, Australia
1978-79: Income Units (Catalogue No. 6523.0)
24
This point of view is not new. In 1776 Adam Smith (10) stated
that "by necessaries I understand, not only the commodities which
are indispensably necessary for the support of life, but whatever
the custom of the country renders it indecent for creditable
people, even of the lowest order, to be without" (quoted in
Townsend, 1979, p. 32n). The specific concept of relative
deprivation is derived from the work of sociologists - Stouffer,
Merton and Runciman -who used it to denote feelings of
deprivation relative to others. Townsend expanded the concept to
refer to conditions, not only feelings, and distinguished such
objective deprivation from conventionally acknowledged or
normative deprivation and individual or group subjective
deprivation. All are important in considering people's relative
welfare, poverty or needs.
In contrast with the relative income approach, Townsend's concept
of relative deprivation involves the detailed analysis of
people's styles of living and their resources. Style of living
is the set of customs and activities which people are expected to
share or in which they are expected to join. Conformity is not
rigidly prescribed. People engage in the same kind of activities
rather than the same specific activities, just as they select
from a fairly limited and familiar range of foodstuffs or other
commodities. Different but overlapping sets of activities are
expected of people of different age and sex and family
membership. Communities differ according to geographical
situation, composition and the kind of resources that are readily
available to them. The style of living of a society consists
more of elements which are heterogeneous, but ordered and
interrelated, rather than rigidly homogeneous. Any attempt to
define this style and represent it in some form of operational
index or poverty line, so that the conformity of a population can
be measured statistically, is bound to be rough and ready. It
might be added that the problem is perhaps analogous to that of
measurement of intellectual ability, for example, or educational
achievement where instruments of some value can be developed
although it may not be clear exactly what is being measured.
Townsend argues that broad measures of resources should replace
simple income measures in the study of inequality and poverty and
also "style of living", as discussed above, should replace
"consumption" (or more narrowly still, "nutritional
requirements") in determining what levels in the ranking of
resources should be regarded as constituting deprivation.
Townsend's measure of resources included all forms of cash
income, capital assets, the value of employment benefits and of
public social services, as well as various forms of private
income in kind. Townsend tentatively hypothesised that, "as
resources for any individual or family are diminished, there is a
point at which there occurs a sudden withdrawal from
participation in the customs and activities sanctioned by the
culture. The point at which withdrawal 'escalates'
disproportionately to falling resources could be defined as the
poverty line".(p.57)
Characteristic % of population
with characteristic
Table 2.9: Townsend equivalence scales for selected household types - U n i t e d Kingdom, 1968-69
Note: The figures in brackets are the relativities derived if a couple with the head not working
is taken as the base. For each person with an appreciable disablement, add 0.17 (0.20), with
a severe disablement, add 0.33 (0.40), and with a very severe disablement, add 0.50 (0.60).
SOURCE: Townsend, 1979, pp. 268-269.
33
It has been shown that the two extreme views - that poverty is
either a condition bordering on starvation or any deviation from
absolute equality - are of very limited usefulness in an analysis
of social conditions. If, therefore, it is accepted that there
is a continuum of standards of living between starvation and
equality, and that poverty can be defined below a point or band
somewhere along that continuum, then the case for Townsend's
general conception is compelling. That is, poverty can or should
be thought of as the economic threshold for participation in the
normal life of a given society.
There is, however, the second range of issues for consideration.
If poverty is conceptualised in terms of participation and non-
participation in the normal activities of society, then is the
idea credible that there is a set of equivalent income levels or
poverty lines below which people cannot participate in social
activities and above which people can? Views about this differ.
For instance Tulloch notes:
"The principle of validity requires that the definition of
poverty demonstrates a difference in kind as opposed to
degree, between the poor and the rest. Poverty is
'something different', more than just another dimension of
inequality. Like disease it must appear to be a 'real
phenomenon', its attribute to exist in concrete reality -
and not in the eyes of the beholder" (1980, p. 16).
Tulloch argues that the distinction between the poor and the rest
of society cannot be sustained and that it seems difficult,
if not impossible, to give priority to a particular definition of
poverty on factual grounds -
"The 'real phenomenon' view of poverty is difficult to
support. It is not the 'facts' which determine the value
position, but the other way round. Poverty is a label which
the researcher, for moral, emotive, and political reasons,
applies to a particular level of inequality. It is
essentially a value judgement, expressing preferences about
social change and the shape and nature of society" (1980),
pp. 17-19)
A similar view is reached in the SWPS Report on Poverty
Measurement which concluded that none of the methods proposed
for measuring poverty are entirely satisfactory, and "in
particular, we do not feel that any of the criteria which we have
discussed would allow us to state with any confidence whether the
money amounts of pensions and benefits were adequate or not" (p.
60) .
1 0.60 0.69
2 1.00 1.00
3 1. 20 1. 28
4 1.40 1. 52
5 1.60 1.70
6 - 1.86
7 - 2.04
Table 3.2 shows the basic scales estimated using this method.
47
1 person 0. 59
2 persons 1.00
3 persons 1. 18
4 persons 1.35
5 persons 1.53
6 or more persons 1.60
This table is directly comparable with Table 3.1 and shows that
the results for Australia and Canada were very similar, with
those for Australia generally showing greater economies of scale
for households larger than two persons.
3 0 1.17
3 1 1.92
3 2 1.72
4 0 1.74
4 1 2.55
4 2 2.25
Household Composition:
Adults Children Food Only All "necessities"
2 0 1.00 1.00
2 1 1.27 1.14
2 2 1.43 1.34
2 3 1. 91 1.71
2 4 1.88 1.59
2 5 2.05 1.63
SOURCE : SWPS, 1981, p. 105.
BASKET OF COMMODITIES
Low Med ium High Low Medium High Low Med ium High
Income Income Income Income Income Income Income Income Income
The most important point to note about the Engel model is that it
is simply a method of analysing the available survey data. It
does not provide a theory relating the scales to consumer
behaviour. In effect, it is assumed that households of different
54
(13) While Engel could have estimated equivalence scales from the
data available, he preferred to base scales on studies of
physiological "needs". See Muellbauer, 1980, pp. 153-154.
55
where xij is expenditure on the ith good by the jth family, mij
is the effect on expenditure on commodity i of the composition of
the jth type of family, yj is the income of the jth type of
family and moj is the overall effect on consumption of all
commodities or the composition of the jth type of family. Thus,
this equation is the same as that shown by (3.3.2), except that
it allows for the differential effect of household composition on
consumption of different types of commodity. The mijS are the
commodity specific equivalence scales and the moj is the overall
household income equivalence scale, in which we are interested.
This method assumes that for each type of family it is possible
to determine the relationship between income and consumption of
each commodity that goes to make up total expenditures. It is
assumed that these expenditures on all commodity groups must add
up to total family income; that is, an income change will always
be absorbed by additional expenditures on all commodities.
Commodities are generally treated as fairly broad aggregates such
as food or clothing rather than specific items such as butter or
men's shirts (although there is no inherent reason why a more
detailed analysis could not be carried out). From equation
(3.4.1) it is possible to derive the relationship between the
specific commodity scales and the overall income scale. In fact,
the income scale is the weighted average of the specific
commodity scales, with the weights being approximately
proportional to the proportions of expenditures on the
commodities. Using the commodity scales and the income scale,
Prais and Houthakker could express actual household income and
expenditure in terms of equivalent household income and
expenditure, and thus for each commodity fit all household types
to a single equivalent Engel curve.
The results shown in Tables 3.8 and 3.9 are very similar, the
major difference being that Podder did not present scales varying
by the age of children. Podder's finding that a single adult
requires only 49 per cent of the income of a married couple is
implausible. He suggests that this may be because the latter
family has to buy proportionately more food to entertain guests
(p.186). It may be, however, that the result reflects the
effects of age differences in the composition of these household
types - a greater proportion of single person households may be
retired.
In any case, it must be emphasised that Podder's methodology does
not provide a satisfactory solution to the identification
problem. While Podder claims to be using the Prais-Houthakker
methodology, in effect he only applied the proportional approach,
and the resultant scales must therefore be considered to be
biased upwards.
Since the Prais-Houthakker study there have been many other
attempts to apply their methodology. The identification problem
has continued to be the major stumbling block, and different
solutions have been proposed by different researchers. Forsyth
(1960) concluded that scales could not be estimated from
cross-sectional data on expenditures alone. The issue has also
been discussed by Barten (1964) and Cramer (1969) and by Singh
and Nagar (1973), who suggested an alternative iterative
procedure which was later adapted by McClements (1977).
McClements analysed data from the 1971 and 1972 Family
Expenditure Surveys and produced the very detailed results shown
59
in Table 3.10. The most notable feature of these scales is the
extent to which the costs of children increase with age - from 9
per cent of the costs of a married couple for a child under 1
year of age to 36 per cent for a 16 to 18 year old.
A.
Household type All Households Retirement
Pensioners
Male Female Male Female
Number of children
Number of no 1 2 3 4 5 6
Adults children child children children children children children
SOURCE: 1-3, Nicholson, 1976, pp. 10-11; 4, Henderson, 1950, p.274; 5, calculated from Rothbarth, 1943, p. 129; 6,
Garganas, 1977, p.106 - (i) calculated assuming first and second children are under 4 years of age, (ii)
calculated assuming all children are between 5 and 14 years.
64
91 = 91 (P1, P2, Y) )
) or qj = q j (p 1 , P2 Y), j = l,2..(3 . 6. 2 )
q2 q2 (P1, P2, Y) )
These functions indicate the amount demanded of each of the goods
at alternative combinations of prices and income. Holding the
price of the second good and income constant gives the demand
curve for the first good
=
P1 92 El(Y) = P1, 91 (P1 P2, Y) ....(3.6.5)
The Engel curves, which we have met before, indicate the effect
of a change in income on the expenditure on each good at fixed
prices. In econometric studies of consumer behaviour, the major
interest is in the elasticities of demand, which can be defined
using the demand function, the demand curve, or the Engel curve.
The own price elasticity of demand for a good can be defined as
the percentage change in the quantity demanded given a 1 per cent
change in the price of this good; the income elasticity of demand
for a good can be defined as the percentage change in the
quantity demanded given a 1 per cent change in income; and, the
cross price elasticities of demand can be defined as the effect
of a change in the price of one good on the demand for the other
good(s). These elasticities are therefore convenient summaries
of the responsiveness of the quantity of a good demanded to the
factors influencing it.
68
Table 3.14: Equivalence scales with age differences using the Barten model -
United Kingdom, 1975
Household Type
Weekly Expenditure Couple, Couple, Couple, Couple, Couple, Couple,
Level for reference no one one two two two
household children child child children children children
(age 0-4) (age 5-16) (0-4,5-16) (both aged (both aged
0-4) 5-16)
20 pounds 1.00 1.16 1.30 1.42 1.27 1.57
30 pounds 1.00 1.12 1.25 1.34 1.19 1.48
40 pounds 1.00 1.09 1.22 1.28 1.13 1.42
50 pounds 1.00 1.07 1.20 1.24 1.09 1.37
70 pounds 1.00 1.04 1.17 1.17 1.02 1.31
100 pounds 1.00 1.00 1.13 1.11 0.96 1.24
Household Type
Couple , Couple, Couple, Couple, Couple, Couple, Couple,
Single no one two three four five six
Income level adult children child children children children children children
Less than $4,880 0.60 1.00 1.21 1.38 1.48 1.54 1.59 1.64
$6,000 0.61 1.00 1.21 1.38 1.48 1.52 1.57 1.61
$8,000 0.61 1.00 1.21 1.37 1.47 1.51 1.55 1.57
$10,000 0.61 1.00 1.20 1.37 1.47 1.50 1.54 1.55
$12,000 0.61 1.00 1.20 1.37 1.47 1.50 1.53 1.54
$14,000 0.61 1.00 1.20 1.37 1.46 1.50 1.52 1.53
$16,000 0.61 1.00 1.20 1.37 1.46 1.49 1.52 1.52
A. Overall scales
Head Works
Wife
Wife not Head Not
Household Type Works * Wo rking * Working **
A. Basic scales
Household Type Equivalence Scale
Single Mult
male 0.90
female 0.74
Sole parent, one child 1.09
Couple, no children 1.00
(Head aged 35-54)
Couple, one child o 1.24
Couple, two children 1.28
Couple, three children 1.40
Household Type
Single adult
male under 35 0. 96
female under 35 0.79
male 35-54 0. 90
female 35-54 0.74
male 55-64 0.66
female 55-64 0.50
male 65 plus 0.51
female 65 plus 0.34
Couple, no children
head under 35 1.06
head 35-54 1.00
head 55-64 0.76
head 65 plus 0.62
C. Effect of age of children
Couple Couple Couple Couple
under 35 35-54 55-64 65 -
No children 1.00 1.00 1.00 1.00
Child under 6 1.22 1.24 1.31 1.38
Child 6 - 1 1 1.31 1.32 1.42 1.50
Child 12 - 17 1.40 1.43 1.56 1.69
Child 18 plus 1.43 1.46 1.60 1.74
SOURCE: Van der Gaag and Smolensky, 1980, Table 3,
79
Despite these general issues, the main problems with the Barten
model arise in its specific estimations. Muellbauer's (1977)
estimation showed that the original model gives implausibly low
scales for children, apparently because of excessive
substitution. The choice of a different cost function improves
the plausibility of the scales, but raises the question of the
dependence of the results on the choice of the cost function.
Overall, there have been very few estimations of the original
Barten proposal and the model can be regarded as still being
tested.
1 0.81 0.83
2 1.00 1.00
3 1.13 1.12
4 1.24 1.21
5 1. 32 1.28
6 1.40 1.35
7 1.47 1.41
A.
B.
Age of adults Relative Allowance
Female Male
25 27 1.00
25 40 1.10
50 52 1.34
55 57 1.36
The attitudinal approach has been developed only over the past
fifteen years, and, since the methodology has been outside the
mainstream of earlier research on either poverty or consumer
behaviour, it lacks a critical literature through which it would
be possible to assess more comprehensively its weaknesses and
strengths. Given the limitations of earlier methods of deriving
86
On the broader issue, there are good reasons for considering that
individuals' welfare evaluations are not necessarily good guides
to their needs. In his study of poverty in the United Kingdom,
Townsend (1979) sought information on subjective feelings of
deprivation as well as objective deprivation. This was not used
to derive the poverty standard, but to analyse whether people's
feelings of deprivation were consistent with objective measures.
While Townsend found that the majority of people did reflect in
their attitudes to their own living standards the resources which
87
This in turn implies that the family head (usually the husband)
will make the same evaluation of a lower or higher income level
as all other members of the family. If someone other than the
head undertakes most of the household budgeting and purchases
most of the groceries and children's clothing, and pays most of
the recurring bills, then it is likely that the family head's
evaluation of income will be based on ignorance. It would seem
very likely that this is the case. This and the unequal sharing
of resources within the family unit must be regarded as a problem
for all equivalence scale research, but for the attitudinal
approach they appear far more significant. (See Edwards (1981)
and the discussion on page 99-100 of this paper).
In conclusion, it would appear that the limitations of the
attitudinal research discussed here outweigh its attractions.
While the equivalence scales shown in Table 4.1 and 4.2 are
generally reasonable, there are problems with the increasing
allowances for adults as they grow older and the flat
relativities for children as they age. The theoretical problems
are even more daunting.
89
5 CONCLUSION
Economies of scale
The concept of "economies of scale" can be distinguished from
that of the "economies of married life", as discussed
previously. The term "economies of scale" refers to the effects
of joint consumption of some goods and services. For example,
the addition of a child involves extra food consumption if the
previous standard of living of a household is to be maintained,
but due to sharing of housing services, little extra spending on
housing is required. Joint consumption occurs where two or more
individuals in a household can share a good or service without
reducing the satisfaction derived by any one person. Joint
consumption arises because units of a good or service may be
indivisible. It is not possible to have half a kitchen, half a
television set or half a refrigerator. Because these items come
in single units, consumption cannot be attributed entirely to one
individual or another. If one person left the household,
consumption of indivisibles would remain unchanged. (22) At the
other end of the sharing range, there are goods like food and
clothing which are divisible and may be able to be ascribed to
individuals. But even here there may be an element of joint
consumption; for example, in the cooking and preparation of food,
costs may not increase in the same proportion as the costs of
food items themselves. In practice most goods will fall betwen
the two extremes of pure joint consumption and entirely
independent consumption. Most goods will be partly but not
entirely attributable to individuals.
The new theory of consumer behaviour
From these examples it is apparent that economic analysis has
concentrated on the material aspects of a family's standard of
living. The simple version of utility theory outlined previously
ignores conflict and co-operation within households, essentially
treating each household as if it had only one member, and
focussing on the effects of changes in money income and money
prices on the allocation of income among market goods. A major
problem with this approach is that to the extent that income and
prices do not explain consumption behaviour, the explanation must
rest with changes in taste. For example, if a household derives
utility from heating fuel then, in the simple theory, its tastes
who leaves school late in a financial year may have a very low
measured annual income, but because they have been part of a
wider income unit their overall standard of living may not be
accurately represented by their apparent annual income (Ingles,
1981, p. 3.). Current income, when measured, may also not be a
good indicator of living standards because it is variable,
particularly among the poor. (See Cox, 1976, pp. 423-442.) In
addition, in the analysis of trends in living standards, any
income measure may be inaccurate to the extent that it cannot
take account of changes in relative economic status because of
inflation. If the price of food, for example, rises faster than
the prices of other commodities, then the living standards of the
poor may fall, both relatively and in "real" terms, even if their
incomes rise with the overall level of inflation, because food is
a greater component of their total expenditures. This issue has
been extensively discussed in the United Kingdom, in relation to
the development of a retail price index for pensioners
(McClements, 1978, pp. 146-166; Diamond, 1978, pp. 330-335, 361-
365, 441-457). There is very limited Australian literature on
the differential effects of relative price changes, apart from
Kakwani (1980, pp. 188-190). In summary, it should be clear that
there are many issues of timing that may affect the accuracy of
measurement of a family's standard of living. It is also
apparent that many of these issues might best be approached
through a longitudinal survey of living standards.
For sole parents with one child, the central range is 0.86 to
0.98. The Henderson scales are in the middle of this range, as
are the ABS/SWPS (ELES) scales. The Podder scale is low. The
ABS/SWPS (Canadian) relativities are just high of the range. The
Kakwani scales are low. Many of the administrative scales are
within the range, but Australia's is well below it. The
geometric mean is 0.90.
For sole parents with two children, the central range of values
is 1.00 to 1.20. This too is a wide range. The Henderson scales
are in the middle of this range. The ABS/SWPS (Canadian) scales
fall in the range but the detailed ELES scales are well below it;
the Kakwani and Podder scales just below the bottom of the
range. A majority of administrative scales fall in the range,
but Australia's is significantly below it. The geometric mean is
1.10.
For sole parents with three children, the central range is 1.32
to 1.38. The Henderson scales are over the top of this range.
The ABS/SWPS (ELES) scales are very low, but the ABS/SWPS
(Canadian) scales are generally in range. Kakwani and Podder
both produce relatively low scales. None of the administrative
scales fall in this range (those of the UK and the FRG are
higher); the Australian pension scales are low. The geometric
mean is 1.31.
For couples with one child, the central range is 1.16 to 1.24.
The Podder and Kakwani scales are near the top of the range. The
ABS/SWPS (Canadian) and the Henderson scales fall near the bottom
of this range. The ,ELES scales are under the range, as are the
Australian pension relativities, and a number of other
administrative scales. The geometric mean is 1.20.
For couples with two children, the central (wide) range of values
is 1.30 to 1.48.The Henderson and the Kakwani scales are in the
middle of this range. The Canadian scales are near the bottom of
the range, with those derived from the ELES method being very
low. The Podder scales are near the top of the range. The
Australian pension scales are well below the range, but not so
low as the ABS/SWPS (ELES) scales. The geometric mean is 1.38.
For couples with three children, the central range of values is
1.48 to 1.70. The Henderson scales are high. The two ABS/SWPS
scales and the Kakwani scale are generally towards the bottom of
the range, while the Podder scale is close to that of Henderson.
The Australian pension relativities are well below the range. The
geometric mean is 1.59.
105
Family type
Method of Single Single with Single with Single with Couple with Couple with Couple with Date Source
derivation adult one c h i l d two children three children Couple one c h i l d two children three children
Administrative
1. Australia 0.60 0.77 0.89 1.04 1.00 1.12 1.25 1.39 1984 Aust.
2. New Zeal and 0.60 1.00 1.09 1.17 1.00 1.08 1.17 1.25 1982 N.Z.
3. United Kingdom 0.62 0.83 1.14 1.52 1.00 1.21 1.42 1.74 1982 U.K.
4. Canada 0.63 0.90 1.02 1.14 1.00 1.12 1.24 1.35 1981 Can.
5. Fed. Rep. Germany 0.56 0.90 1.44 2.22 1.00 1.34 1.88 2.66 1980 F.R.G.
Budgetary
Henderson
6. - Head working 0.76 0.91 1.14 .44 .00 1.15 .37 1.68 1954 U.S.
7. - Head not working 0.68 0.86 1.11 .45 .00 1.17 .42 1.76 1954 U.S.
8. Plachaud 0.62 0.84 1.11 .44 .00 1.23 .50 1.82 1979-80 U.K.
9. Lovering 0.60 0.78 0.93 1.12 .00 1.12 .28 1.47 1983 Aust.
Rowntree
10. - Head working 0.81 (1.01) (1.11) ( .18) .00 1.20 .30 1.37 1936 U.K.
11. - Head not working 0.63 (0.90) (1.03) ( .09) .00 1.27 .40 1.46 1936 U.K.
12. Orshansky 0.69 (0.85) (1.17) (1.41) .00 1.16 1.48 1.72 1965 U.S.
13. Beveridge 0.59 (0.83) (1.07) (1.31) .00 1.24 1.48 1.72 1942 U.K.
Proportional
Family type
Method of Single Single with Single with Single with Couple with Couple with Couple with Date Source
derivation adult one c h i l d two children three children Couple one c h i l d two children three children
30. - high income 0.86 1.00 1.24 1.24 1.00 1.24 1.24 1.34 1968-69 Israel
31. - Podder 0.49 (0.74) (0.97) (1.17) 1.00 1.25 1.48 1.68 1966-68 Aust.
32. - (Prals/Houthakker) 0.53 (0.72) (0.99) (1.30) 1.00 1.19 1.45 1.77 1937-39 U.K.
33. - Bureau van
Statistiek 0.53 (0.95) (1.24) (1.48) 1.00 1.42 1.71 1.95 1917 Neth.
34. - Nlcholson (food) 0.54 (0.85) (0.99) (1.19) 1.00 1.31 1.55 1.75 1965 U.K.
Consumption Theory
McCIements
35. - Young children 0.53 (0.62) (0.80) (1.01) 1.00 1.09 1.27 1.48 1971-72 U.K.
36. - Older children 0.53 (0.76) (1.01) (1.28) 1.00 1.23 1.48 1.75 1971-72 U.K.
37. Jensen 0.60 0.92 1.20 1.46 1.00 1.27 1.53 1.77 -
38. Nlcholson - - - - 1.00 1.25 1.40 - 1949 U.K.
39. Nicholson 0.64 (0.77) (0.89) (1.00) 1.00 1.13 1.25 1.35 1976 U.K.
40. Cramer - - - - 1.00 1.12 - - 1953-54 U.K.
41. A.M. Henderson - - - - 1.00 1.17 1.25 - 1937-38 U.K.
42. Rothbarth - - - - 1.00 1.15 1.20 1943 U.K.
Garganas
43. - Young children - - - - 1.00 1.15 1.21 1.33 1971 U.K.
44. - Older children - - - - 1.00 1.29 1.52 1.64 1971 U.K.
Barten (Muellauer)
45. - overall - - - 1.00 1.15 1.30 1.44 1975 U.K.
-
46. - low income - - - - 1.00 1. 16 1.42 - 1975 U.K.
47. - high income - - - - 1.00 1.00 1.11 - 1975 U.K.
48. Lazear & Michael 0.94 (1.15) (1.33) (1.53) 1.00 1.21 1.39 1.59 1960-61 U.S.
Kakwani
49. - low income 0.60 (0.81) (0.98) (1.08) 1.00 1.21 1.38 1.48 1966-68 Aust.
50. - high income 0.61 (0.81) (0.98) (1.07) 1.00 1.20 1.37 1.46 1966-68 Aust.
ABS/SWPS
51. - basic scales 0.67 0.92 1.07 1.05 1.00 1.11 1.20 1.31 1974-75 Aust.
52. - head words 0.71 (0.82) (0.86) (1.04) 1.00 1.11 1.17 1.33 1974-75 Aust.
53. - head not working 0.59 0.87 0.85 1.17 1.00 1.13 1.20 1.38 1974-75 Aust.
54. Van der Gaag and
Smolensky 0.74 1.09 1.00 1.24 1.28 1.40 1960-61 U.S.
Attitudinal
55. Goedhart et al 0.81 1.00 1.13 1.24 1.00 1.13 1.24 1.32 1975 Neth.
Kapteyn and Van Praag
56. - Young f a m i l y - - - - 1.00 1.19 1.28 1.36 1971 Neth.
57. - Older family — — — 1.00 1.14 1.21 1.27 1971 Neth.
Other
Townsend
58. - head working 0.71 0.88 1.04 1.29 1.00 1.17 1.33 1.58 1968-69 U.K.
59. - head not working 0.65 0.85 1.06 1.35 1.00 1.20 1.41 1.71 1968-69 U.K.
60. - Root N 0.71 1.00 1.22 1.41 1.00 1.22 1.41 1.58
Table 5.3 shows one further set of scales, those proposed by the
O.E.C.D. for "countries which have not established their own
equivalence scales" (O.E.C.D., 1982,p.37). These scales were
developed as part of the O.E.C.D. social indicators program, and
appear to be purely arbitrary. They are derived by setting the
value for a single adult as 1.0, that for second and subsequent
adults as 0.7, and that for each child as 0.5. The results shown
in Table 5.3 were then derived by re-calculating with two adults
as the base. It can be seen that these results vary quite
significantly from those shown by the geometric means or by the
central values discussed on pages 102 to 106. This variation
arises because of the high relative allowances for children and
because the O.E.C.D. results do not allow for economies of
scale. It is, of course, important to note that both the results
in Tables 5.2 and 5.3 also average out the differences that could
be expected to arise because of the effects of labour force
participation, age of children or household head, and household
income. It is to some of these details that the paper now
turns.
109
Overall Australian
Family Geometric Geometric
Composition Mean Mean
Sole parent,
one child 0.90 0.88
Sole parent,
two children 1.10 1.01
Sole parent,
three children 1.31 1.17
Couple,
no children 1.00 1.00
Couple,
one child 1.20 1.16
Couple,
two children 1.38 1.30
Couple,
three children 1.59 1.48
SOURCE: Calculated from Table 5.1. The results for the overall
geometric mean are calculated from all scales except the
administrative and root N results. The other results
are calculated from Australian research (not including
the Henderson relativities). The geometric mean is
defined as the nth root of the product of n numbers, in
this case the relevant equivalence scale values.
Family Composition
Adults/Children Equivalence scale value
1 0 0. 59
1 1 0.88
1 2 1.18
1 3 1.47
2 0 1.00
2 1 1.29
2 2 1.59
2 3 1.88
1. McClements, Age 0-4 18 2. McClements, Age 0-1 9 3. Piachaud, Age 4. Lovering, Age
United Kingdom 5-10 21 United Kingdom, 2-4 18 United Kingdom, 2 23 Austral la, 2 12
1971-72 11-22 26 1971-72 5-7 21 1979 5 27 1983 5 15
13-15 32 8- 10 23 8 32 8 17
16-18 38 11-12 25 11 35 11 22
Average 23 13-15 27 Average 29 Average 16.5
16-18 36
Average 22.7
5. Townsend, Age 0-10 17 6. Muellbauer, Age 0-4 12 7. Garganas, Age 0-4 13 8. Van der Gaag Age
United Kingdom, 11-15 25 United Kingdom 5-16 25 United Kingdom, 5-14 29 and Smolensky, 0-6 24
1968-69 Average 21 1975 Average 18.5 1971 Average 21 United States 6-11 32
1960-61 12-17 43
18 plus 46
Average 36.3
9. Supplementary Age 0-10 21 10. EEC Expenditure Age 0-2 11 11. Department of Age 0-4 15 12. Henderson Age 0-5 15
Benefit Rate 11-15 31 Survey 3-4 17 Health, Education 5-15 37 United States
, 6-15 22
Scale, 16-17 38 5-6 22 and Welfare, 16-18 52 1954 16 plus us 31
United Kingdom, Average 30 7-8 28 United States, Average 34.7 Average 22.3
November 1983 9- 10 33 1976
11-12 39
13-14 44
Average 27.7
13. Beveridge Report Age 0-4 16 14. The Amsterdam Age 0-14 27
United Kingdom, 5-9 22 Scale 15-17 (male) 52
1942 10-13 25 15-17 (female) 47
14-15 28 18plus (male) 53
Average 23 18plus (female) 47
Average (male) 44
Average (female) 40.3
SOURCES: Items 1, 2, 6, 7, 10, 11 are taken from Isherwood (1978) pp. 4-5;
Item 9 is taken from Social Work Today, 22 November 1983;
Item 13 comes from McClements (1978), p. 114; Items 3, 4, 5, 8 and 12 are
derived in other parts of this paper; Item 14 comes from Deaton and MuelIbauer (1980) p. 193.
112
Table 5.5: The costs of working - Percentage increases in income required to
maintain equivalent standard of living when household head works
1. Rowntree, United Kingdom, 1936 2. Henderson, United States, 1954
Household Type Per Cent Household Type Per Cent
Single man 14.1 Single man 18.8
Single woman 21.8 Single woman 25.0
Couple, no children 14.9 Couple, no children 13.6
Couple, one child 9.1 Couple, one child 11.7
Couple, two children 6.6 Couple, two children 9.6
Sole parent, one child 19.7
3. Townsend, United Kingdom, 1968 4. ABS/SWPS (Proportional) Australia 1974-75
Household Type Per Cent Household Type Per Cent
Single person 31.5 Single person 21.1
Couple, no children 20.5 Couple, no children 7.5
Couple, one child 17.0
Couple, two children 13.7
Sole parent, one child 23.9
5. ABS/SWPS (ELES) - Australia, 1974-75
Household Type Per Cent
Single person 39.2
Couple, no children 14.9
Couple, one child 13.3
Couple, two children 12.5
Sole parent, one child 14.5
NOTE: These figures have been calculated by subtracting the equivalence scale value
for a household with the head not working from the value for the corresponding
household with the head working, and expressing this result as a percentage of
the head not working scale. The exception is the result for sole parents
given in 5, in which the value for the head not working is subtracted from an
"overall" scale result.
This pattern reflects the fact that the costs of working enter
these results as fixed costs. For example, say a single man and
a married man both need an extra $50 a week to compensate for
working costs, this $50 is a greater proportion of the single
man's "standard costs" than of the married man's. Again, for a
couple without children, the base will be lower than that for a
couple with children, and hence the pattern of declining costs.
Therefore, this table should not be taken to show that the costs
of a working family decline if they have a child or extra
children. The comparison only applies to identical family types,
in work and out of work.
It may be possible to derive some further indirect information on
the costs of working, if it were assumed that the age of the
household head was a proxy for the status of working. This would
seem reasonable when comparing those over 65 years of age, who
could reasonably be assumed to be retired, with those who were of
working age. However, usually it is not then possible to control
for the work status of the spouse in such comparisons and the
results would" heed to be treated with even greater caution.
One-earner and two-earner families
Table 5.6 shows the information available on the differences
between the costs of families with working and dependent
spouses. These costs have been calculated by subtracting the
estimates of the costs of couples with and without children and
where the head works and the spouse does not from estimates for
corresponding couples where both spouses work, and then
expressing this difference as a percentage of the costs for the
couple where the head only works. As such, these figures can be
thought of as the additional compensation required by different
family types where the "head" already works, when the spouse goes
out to work.
Table 5.6 shows considerable variability between scales, and no
consistent regularity in the pattern of results emerges. Some of
the results can be considered to be more reasonable than others.
For example, Townsend's estimates of the costs faced by a working
spouse are likely to represent only those relevant to very low
income earners. Since the Townsend equivalence scales are
derived from the relative deprivation poverty standard, it is
likely that working spouses in this situation were both uncommon
and, where existing, likely to be working very few hours. It
would therefore seem reasonable to conclude that Townsend's
scales underestimate the costs faced by working spouses.
Similarly, the Henderson scale estimates are likely to be
inappropriate as a basis for comparing the average situation of
couples where the head works and the spouse does not, since they
are derived from poverty lines and are therefore not
representative of the community average situation.
114
Household Type
1. Henderson,
United States, 1954 21.0 18.3 16.1
2. Townsend, United
Kingdom, 1968 6.0 5.1 5.3
3. ABS/SWPS (Proportional)
A u s t r a l i a , 1974-75 2.0 31.6 14.3
4. ABS/SWPS (ELES )
A u s t r a l i a , 1974-75 9.0 18.0 24.8
The two Australian results derived by the ABS and SWPS are
clearly inconsistent, since the proportional estimates rise
massively and then decline, while the ELES estimates rise
smoothly, if significantly. The ELES results, a priori , appear
more reasonable. The presence of children implies higher direct
costs due to childcare, but if these childcare costs then double
in terms of direct expenses, because the number of children
double, they should decline somewhat relative to overall costs.
The Lazear and Michael (1980) estimates were derived specifically
in order to determine the costs of a working spouse, but their
plausibility depends on an a priori judgement whether more or
less than an additional 30 per cent is required when a spouse
goes out to work.
first child adds 11 per cent, the second 6 per cent, the third 16
per cent, the fourth 3 per cent and the fifth 17 per cent. It is
difficult to conceive of the reasons why this should be so.
not be. These scales are, of course, the Henderson scales, which
can be judged to lack both theoretical and empirical validity,
even if they have a considerable amount of public support.
Manning (1982, pp.4-5) argues that the Henderson equivalence
scales should continue to be used for the time being until
something better is developed to replace them. He notes that
"the expectations created by long-established relativities are a
force to be reckoned with in the politics of social security, and
while a proposed alternative equivalence scale may serve to
indicate groups which should receive priority in any increase in
rates, few reformers are willing to tie their whole campaign to a
particular scale" (p.5). "Long-establishment" is not a
particularly convincing argument for anything, particularly not
for the use of a set of equivalence scales. The weaknesses of
the Henderson scale were well recognised at the time of the
Poverty Inquiry. The First Main Report stated:
"Relativities between income units of different size and
composition must be reconsidered when they are available,
and this is not just an exercise in arithmetic. Studies
elsewhere have shown that actual expenditure data are
necessary but not sufficient to determine needs. They must
be supplemented by nutritional studies, studies of housing
and other reseach" (1975, p. 300).
While there are therefore many good reasons for rejecting nearly
all available equivalence scales, the problem remains that
equivalence scales are unavoidable in many important areas of
social research and social policy, and some sort of choice must
therefore be made. This raises a specific conflict for the
author of this paper - after spending considerable time in
criticising all available equivalence scales, it seems necessary
to opt for one of the discredited approaches. In this context, I
would make a very tentative choice based on the full range of
research. For indicative purposes (for example, in terms of
setting goals for social security rates) I would choose the
"central range" of equivalence scale values discussed on pages
102 to 106. For the purposes of Australian poverty measurement,
where the scales used most necessarily be precise, I would choose
the geometric mean of the Australian research, shown in Table 5.2
on page 108.
These judgements can clearly be subject to substantial
criticism. Rather than choosing one discredited approach, I
would opt for the average of all of the discredited approaches!
(27) This should not be taken to suggest that successive mistaken
(27) i.e. "running with the pack". See Manning, 1982 p.12.
131
Sources: Australian Budget Statements and Department of Social Security Annual Reports
for relevant years; and Development Division, Department of Social Security, Research
Paper No. 20, Developments in Social Security, June 1983.
Equivalences are based on the net value of assistance for a pensioner couple with
no dependent children who are not paying rent.
136 A.2
EQUIVALENCES IMPLICIT IN AUSTRALIAN TAXATION SYSTEM - 1976-77 and 1983-84
Notes: Derived from personal calculations. The equivalences for those at 50 per cent, 100 per cent and 200 per
cent AWE are those implicit in the after-tax incomes of families whose pre-tax income are at these
respective levels. It is assumed that for a couple where both work, the income is plit 60:40. Family
allowances are treated as a tax rebate.
137 A.3
Equivalence Ratios Implicit in International Family Allowance Programs in 1981
(a) Equivalences are based on the level of assistance for the first child in the family who qualifies for a payment.
(b) Universal programs are most often funded from general revenue.
(c) Some countries pay an age loading which varies with the age of the child assisted, and is added to the basic
payment. The value of these loadings relative to the value of the payments for the first child are represented
in diagram ( ), which should be read in conjunction with this table.
(d) Employment related programs are funded from employee and employer contributions, and are often limited to certain
industrial sectors.
SOURCES; SOCIAL SECURITY PROGRAMS THROUGHOUT THE WORLD 1981, U.S. Department of Health and Human Services, Research
Report 58, SSA 13-11805, July 1982. Quebec, BASIC ASSISTANCE REGULATIONS, January 15, 1981.
A.4
138
Age Loadings in International Family Allowance Programs in 1981
(a) Loadings are calculated as a proportion of the basic level of assistance for the
first qualified child of the family.
BIBLIOGRAPHY
Habib, Jack and Tawil, Yossi, EQUIVALENCE SCALES FOR FAMILY SIZE:
FINDING FROM ISRAELI DATA, The National Insurance Institute,
Bureau of Research and Planning, Jerusalem, March 1974.
142
Habib, Jack, Kohn, M., and Lerman, Robert, SOME ISSUES IN THE
DEFINITION AND MEASUREMENT OF POVERTY IN ISRAEL, Brookdale
Institute of Gerontology and Adult Human Development in Israel,
American Joint Committee, Jerusalem, 1975.
Harding, Ann, An introduction to the social wage, SOCIAL SECURITY
JOURNAL, December 1982, 13-21.
Harper, R.J.A. Survey of Living Conditions in Melbourne - 1966,
THE ECONOMIC RECORD, 43(102), June 1967, 262-288.
Henderson, A.M., The Cost of a Family, REVIEW OF ECONOMIC
STUDIES, Vol 17, 1949a, 127-148.
Henderson, A.M., The Cost of Children, Part I POPULATION STUDIES,
3(2) 1949b, 130-150.
Henderson, A.M., The Cost of Children, Parts II and III,
POPULATION STUDIES, 4(1) 1950, 267-298.
Henderson, R.F., COMMISSION OF INQUIRY INTO POVERTY, FIRST MAIN
REPORT, Australian Government Publishing Service, Canberra, 1975.
Lazear, E.P. and Michael, R.T., Family Size and the Distribution
of Real Per Capita Income, THE AMERICAN ECONOMIC REVIEW, 70 (1),
March 1980, 91-107.
Love, R., and Oja, G., Low Income in Canada, REVIEW OF INCOME AND
WEALTH, 23, No. 1, 1977, 39-61.
Piachaud, David, ROUND ABOUT FIFTY HOURS A WEEK : THE TIME COSTS
OF CHILDREN, Poverty Pamphlet 64, Child Poverty Action Group,
October 1984.
Pollak, R.A. and Wales, T.J., Equity: The Individual versus the
Family. Welfare Comparisons and Equivalence Scales, AMERICAN
ECONOMIC REVIEW, 69 (2) 1979, 216-221.
Powell, A.A., A Complete System of Consumer Demand Equations for
the Australian Economy Fitted by a Model of Additive Preferences,
ECONOMETRICA, 34 (3) 1966.
Powell, A.A., Estimation of Lluch's Extended Liner Expenditure
System from Cross Sectional Data, AUSTRALIAN JOURNAL OF
STATISTICS, 15, 1973.
Praag, Bernard van, INDIVIDUAL WELFARE FUNCTIONS AND CONSUMER
BEHAVIOUR, North-Holland, Amsterdam, 1968.
Praag, Bernard van, The Welfare Function of Income in Belgium: An
Empirical Investigation, EUROPEAN ECONOMIC REVIEW, 2, 1971,
337-369.
Praag, Bernard van, Goedhart, Theo and Kapteyn, Arie, The Poverty
Line - A Pilot Survey in Europe, REVIEW OF ECONOMICS AND
STATISTICS, 62 (3) 1980, 461-465.
Prais, S.J. and Houthakker, H.S., THE ANALYSIS OF FAMILY BUDGETS,
Cambridge University Press, Cambridge, 1955.
Rathbone, Eleanor R., FAMILY ALLOWANCES, (first issued as THE
DISINHERITED FAMILY, 1924) 4th Edition, George Allen and Unwin,
London, 1949.
Rathbone, Eleanor R., THE ETHICS AND ECONOMICS OF FAMILY
ENDOWMENT, The Beckly Social Security Lecture 1927, Epworth
press, 1927.
Raymond, Judy and Whiteford, Peter, SOME IMPLICATIONS OF THE
INTERACTION OF THE PERSONAL INCOME TAXATION AND SOCIAL SECURITY
SYSTEMS, Research Paper No 22, Development Division, Department
of Social Security, March 1984.
Rein, Martin, Problems in the Definition and Measurement of
Poverty, in Ferman, Louis A. et al (eds), POVERTY IN AMERICA,
University of Michigan Press, Ann Arbor, 1968, 116-133.
Rothbarth, E., Note on a method of determining equivalent income
for families of different composition, Appendix IV in Madge, C.,
WAR-TIME PATTERNS OF SAVING AND SPENDING, University Press,
Cambridge, 1943, 123-130.
Rothschild, M. and Stiglitz, J.E., Some Further Results on the
Measurement of Inequality, JOURNAL OF ECONOMIC THEORY, 6 (2)
1973, 188-204.
Rowntree, B.S., POVERTY: A STUDY OF TOWN LIFE, MacMillan,
London, 1901.
147
Rowntree, B.S., POVERTY AND PROGRESS, Longmans, Green and Co.,
London, 2nd edition, 1942.
Sarma, J.T.R., Effects of Household Structure, Size and Household
Income on Expenditure Patterns, POPULATION AND EMPLOYMENT WORKING
PAPER 31, ILO, Geneva, 1976.
Saunders, C., Measures of Total Household Consumption, THE REVIEW
OF INCOME AND WEALTH, 26, 1980, 351-366.
Saunders, Peter (ed), THE POVERTY LINE: METHODOLOGY AND
MEASUREMENT, Social Welfare Research Centre, Sydney, October
1980a.
Saunders, Peter, What's Wrong with the Poverty Line? THE
AUSTRALIAN QUARTERLY, Summer 1980b, 388-397.
Sen, A.K., Poverty, an Ordinal Approach to Measurement,
ECONOMETRICA, 44 (2) 1976.
Sen, A.K., Poverty, Inequality and Unemployment: Some Conceptual
Issues in Measurement, ECONOMIC AND POLITICAL WEEKLY, 8, 1973.
Seneca, J.J. and Taussig, M.K., Family Equivalence Scales and
Personal Income Tax Exemptions for Children, REVIEW OF ECONOMICS
AND STATISTICS, 53 (August) 1971, 253-262.
Singh, B. and Nagar, A.L., Determination of Consumer Unit Scales,
ECONOMETRICA, 41 (2) 1973, 347-356.
Smith, Philippa, LIVING ON THE EDGE : A STUDY OF 90 LOW INCOME
FAMILIES, Australian Council of Social Service, Sydney, 1982.
Social Welfare Policy Secretariat, REPORT ON POVERTY MEASUREMENT,
Australian Government Publishing Service, Canberra, 1981.
Stanton, David I., Approaches to the Measurement of Poverty and
their Possible Application to Australia, DSS, mimeo, June 1972.
Stanton, David I., Determining the Poverty Line, SOCIAL SECURITY
QUARTERLY, Spring 1973, 18-32.
Stanton, David I., The Henderson Poverty Line - a Critique,
SOCIAL SECURITY JOURNAL, December 1980, 14-24.
Stone, R., Linear Expenditure Systems and Demand Analysis: An
Application to the Pattern of British Demand, ECONOMIC JOURNAL,
64, 1954.
Sydenstricker, E. and King, W.I., The Measurement of the
Relative Economic Status of Families, QUARTERLY PUBLICATION OF
THE AMERICAN STATISTICAL ASSOCIATION, Vol.17, 842-857, 1921.
Townsend, P., POVERTY IN THE UNITED KINGDOM, Penguin,
Harmondsworth, 1979.
148