ARTICLE IN PRESS

Energy Policy ] (]]]]) ]]]]]]

Contents lists available at ScienceDirect

Energy Policy
journal homepage: www.elsevier.com/locate/enpol

Economic efciency of solar hot water policy in New Zealand

Kenneth Gillingham 
Stanford University, Department of Management Science & Engineering, and University of Auckland Energy Centre, P.O. Box 16336, Stanford, CA 94309, USA

a r t i c l e in fo

abstract

Article history:
Received 5 August 2008
Accepted 7 January 2009

New Zealand has recently followed the path of several other countries in promoting solar hot water
(SHW) systems in the effort to reduce greenhouse gas emissions, yet the economic efciency of largescale policies to encourage SHW remains a pressing question for policymakers. This paper develops an
economic framework to examine policies to promote SHW in New Zealand, including the current
information, training, and subsidy policy. The economic framework points to environmental, energy
security, and average-cost electricity retail pricing market failures as motivation for SHW policy, with
the global climate change externality the most important of these. The results indicate that domestic
SHW systems are close to being nancially attractive from a consumer perspective, but a more
substantial subsidy policy would be necessary for SHW to appeal to a wider audience. Such a policy is
far more likely to have positive net benets than a policy of mandating SHW on all homes or all new
homes in New Zealand, and could be justied on economic efciency grounds under reasonable
assumptions. However, this result reverses under an economy-wide carbon trading system that
internalizes the environmental externality.
& 2009 Elsevier Ltd. All rights reserved.

Keywords:
Solar
Greenhouse gases
Electricity

1. Introduction
With the threat of global climate change a high priority for
policymakers around the world, policies to promote renewable
energy technologies have come into fashion. Some of the most
common policies focus on solar energy technologies, such as solar
photovoltaics (PV) and solar hot water (SHW) heaters. Germany,
Japan, and California have implemented large-scale subsidy
policies to promote solar PV. Similarly, there are signicant
subsidy policies for SHW heaters in Germany, Austria, Sweden, the
Netherlands, and France, and mandatory installation policies in
Spain and Israel (Roulleau and Lloyd, 2008). Some of these policies
began decades ago during the oil crises in the 1970s for energy
security reasons, but both the number and extent of the policies
have gained steam in recent years as part of broader efforts to
reduce greenhouse gas emissions.
New Zealand solar policy follows a similar pattern. From 1978
to 1982 New Zealand experimented with a SHW subsidy policy for
energy security reasons. The NZ$500 subsidy policy was discontinued largely due to low take-up and poor system performance.
With concerns about global climate change growing, New Zealand
revived the solar subsidy program in 2002 with a NZ$300 subsidy
towards the interest on a loan to nance the SHW installation,
along with training and information policies. In 2006, the subsidy

 Tel.: 650 353 6578; fax: 650 350 4330.

E-mail address: kgilling@stanford.edu

was increased to NZ$500 that could be taken directly as a grant or

used towards the interest on a loan. One notable aspect of this new
policy is that this subsidy will only be granted for purchases of
systems that meet a cost-effectiveness threshold set by the Energy
Efciency and Conservation Authority (EECA) (EECA, 2007).
New Zealands policy has been met with some initial success in
increasing diffusion of SHW systems, with SHW annual sales
increasing from approximately 1000 in 2002 to 3500 in 2006
(EECA, 2006a). However, in the broader picture, SHW reduces
both peak and total electricity demand by such an insignicant
amount that the Ministry of Economic Development (MED) does
not even examine SHW explicitly in its Energy Data File (MED,
2007a). Moreover, it remains unclear whether the 2006 policy
changes will be successful in fostering a sustainable SHW market.
There are several open questions pertaining to New Zealand
SHW policy that have broader implications for SHW policy
throughout the world. First, what does it mean for a SHW policy
to be economic efciency-improving? In other words, under what
conditions would a SHW policy improve social welfare by
reducing market failures? Second, is SHW in New Zealand
currently nancially attractive and if New Zealand is serious
about promoting SHW, how large of a subsidy policy might be
needed to ensure SHW is nancially attractive? Third, what are
the implications of large-scale SHW policies for electricity use and
greenhouse gas emissions? Finally, would these policies be
economic efciency-improving?
Previous publications that address the economics of SHW
policy in New Zealand (e.g., EECA (2006a), EHMS (2006), EECA

0301-4215/$ - see front matter & 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.enpol.2009.01.005

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005

ARTICLE IN PRESS
2

K. Gillingham / Energy Policy ] (]]]]) ]]]]]]

(2001), Sumner (2004), and McChesney (2005)) have provided

useful technical overviews of the nancial attractiveness of
individual systems or examined the maximum technical potential
for solar in New Zealand. This paper aims to address the questions
posed above by presenting an economic framework for examining
SHW policy based on market failures, and then using this
framework as guidance for an economic analysis of several
large-scale SHW policies.
The rest of this paper is organized as follows. Section 2
provides this framework for examining the economic efciency of
SHW markets. Section 3 contains an analysis of the nancial
attractiveness of typical SHW systems to New Zealand consumers.
Section 4 examines the implications of several larger-scale SHW
policies than the current EECA policies. Section 5 concludes.

2. Economic efciency in SHW markets

In many respects, solar technologies are no different than any
edgling technology. For any technology, improvements in the
technology based on research and development or learning-bydoing can lead to improved performance and lower costs with
additional research effort or cumulative installations. Lower costs
imply increased sales and possibly greater consumer benets. Yet
we do not subsidize the market for every new gadget. So, why is
solar different?
Clearly, there is the environmental externality, which has been
the primary motivation for solar policy around the world. But
there are several other market failures that may provide
additional motivation, some of which are unique to solar in
New Zealand.1 The following sections discuss market failures that
could be argued to exist in the New Zealand SHW market in order
to provide an economic basis for SHW policy analysis.
2.1. Environmental externalities
The environmental externalities avoided by installing SHW are
the primary motivation for SHW policy throughout the world. In
New Zealand, over 70% of hot water heaters use electricity, with
most of the remainder using natural gas (Isaacs et al., 2006). Thus,
from the perspective of the SHW market, there is a positive
externality from the installation of each additional SHW system
due to reduced use of fossil fuels in the generation of electricity or
combustion of natural gas. Quantifying the externality provides
guidance for policy intervention.
The magnitude of the global climate change externality is
signicantly more controversial and uncertain than whether or
not the externality exists. Gillingham et al. (2004, 2006) perform a
literature review of several economic studies on the value of
carbon dioxide external damages, and nd a mean value of US$30
per tonne of carbon (approximately NZ$43) and a median value of
US$26 per tonne of carbon (approximately NZ$37).2 However,
these studies are far from denitive, for estimating the value of
external damages implicitly requires ethical judgments about the
value of human life and biodiversity. Thus, it is not uncommon to
see an estimate of US$100 per tonne of carbon (NZ$142) or above
for the value of the external costs. With such uncertainty, the
economic analysis in the later sections of this paper is neutral on
magnitude of the external costs by providing insight into the
1
Absent market failures, economic theory suggests that the competitive
market without intervention maximizes social welfare, and thus the justication
for any policy intervention would have to be made on equity grounds (i.e., for
ethical reasons due to distributional consequences).
2
All dollar values in this paper are denominated in 2007 New Zealand dollars
unless otherwise specied.

implications of policies under different assumptions of the

external costs.
There is also an externality due to emissions of sulfur dioxide
(SO2), nitrogen oxides (NOX), particulate matter (e.g., PM10, PM2.5),
and mercury from fossil-fuel-based electricity generation, all of
which are known to have health consequences. New Zealand
benets from a relatively small amount of fossil-based electricity
generation and a copious amount of wind, so emissions of these
pollutants are often quickly spread over a large area and out
to sea. Thus these costs are not likely to be important in
New Zealand, but may be important in other countries. As a
benchmark, Gillingham et al. (2004), combine the results of
several studies in the United States to nd that the external costs
from these other pollutants are roughly half of the external cost of
carbon dioxide. Assuming the ratio of carbon to the other
pollutants is roughly constant among all thermal generators,
and taking the NZ$43 mean value of external damages from
carbon dioxide, this implies that air pollutants impose an
additional external cost of around US$22 (NZ$31) per tonne of
carbon.

2.2. Energy security externality

The case for an energy security externality has been argued
from a few angles. One is that importing fuel leaves the economy
vulnerable to the macroeconomic effects of price shocks, a factor
that is not reected in the price consumers pay for fuel. Another is
that for countries like the United States, there is a national
security and international diplomacy external cost. The idea
behind this external cost is that consumers purchasing fuel do not
take into account the cost of military and diplomatic expenditures
to ensure a reliable supply of fuel (Bohi and Toman, 1996). This
argument is mostly used in the context of imported oil.
Neither of these cases appears to apply very strongly to the
New Zealand electricity market at the current time. Sixty-two
percent of electricity generation is supplied by hydropower and
geothermal electricity generation. The only imported feedstock for
electricity generation is Indonesian coal for use at the Huntly
thermal plant (12% of electricity generation in 2006).3 All of the
natural gas currently used in New Zealand is supplied by domestic
elds (21% of electricity generation in 2006). This may change in
the future as New Zealand natural gas reserves become depleted
and pressure builds for the construction of LNG facilities, but
currently it appears that the energy security externalities in the
electricity market are small.

2.3. Average-cost electricity pricing market failure

As in most countries, New Zealand residential electricity
consumers are charged a price per kWh by the retailer that
reects the average cost of electricity and not only does not vary
during the day, but often does not vary even by the season.
However, demand uctuates throughout the day, and supply in
New Zealand uctuates seasonally due to constrained supply of
hydropower during dry winters. As noted in Joskow and Tirole
(2007), price signals may fail to appropriately reect the scarcity
of electricity on the market under average-cost pricing. Assuming
a competitive market, the average-cost electricity price is set to
smooth out the uctuations in wholesale spot prices, but ensure
the retailer covers its costs. Hence, the use of electricity by
consumers during peak times has an external cost, due to the
3
Roughly 50% of Huntly coal generation is from New Zealand coal and 50% is
from Indonesia (Genesis Energy, 2007).

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005

ARTICLE IN PRESS
K. Gillingham / Energy Policy ] (]]]]) ]]]]]]

effect that consumers consumption has on the average electricity

rates paid by all ratepayers.
If solar PV or SHW systems generate more electricity at times
of higher spot prices, but consumers receive electricity bill savings
based on the average rate, then consumers will have an incentive
to under-invest in solar systems. Borenstein (2005) notes that
higher spot prices tend to occur during the sunniest times of the
day in California. Borenstein then nds a correlation between
hourly wholesale spot prices for electricity in California and
hourly solar PV system electricity generation, and uses this result
to estimate that the long-run value of solar PV electricity
generation to all ratepayers is 2948% greater than the value to
the solar PV consumer who faces a average-cost electricity price.
In New Zealand, the hourly electricity savings from SHW
systems are highly dependent on the household pattern of hot
water use. A typical household has major draws of hot water
around 8 am and again around 6 pm, which correspond to, and
most likely contribute to the morning and evening peaks (Thomas
and Lloyd, 2006). For SHW systems with a large well-insulated
cylinder, enough hot water may be available from the sun to last
into the night, and a timer system can ensure that the back-up
turns on before the morning peak. Thus, it is not clear whether
there is any correlation between hourly electricity spot prices and
hourly electricity savings from SHW systems.
On the other hand, in New Zealand there is longer term
seasonal correlation between wholesale spot prices and solar
radiation, and thus a seasonal correlation between wholesale spot
prices and SHW electricity generation. Wholesale spot prices tend
to be highest during winters when there has not been much
rainfall in the preceding few months because of a lack of water to
run the hydropower stations that provide 55% of New Zealands
electricity (MED, 2007a). It only follows that years with dry
winters have more sun on average. Aggregated data on electricity
spot prices from the Electricity Commission (EC (2007)) and solar
insolation from the National Institute for Water and Atmospheric
Research (NIWA (2007) suggest that there is a positive correlation
between average annual spot prices and solar insolation
(Table 1).4 The Pearson correlation coefcient using the data in
Table 1 is 0.42, which gives a rough indication of the relationship
between spot prices and insolation (with only 11 years of data, the
correlation coefcient is signicant in a one-tailed t-test at a 90%
condence level).
To illustrate more concretely how this correlation can lead to a
market failure in competitive electricity markets, consider a
simple two period stylized model based on Borenstein (2005),
Poletti (2006), and Joskow and Tirole (2006, 2007). Assume two
equal length time periods, periods 1 and 2. Period 1 represents the
peak period, so the wholesale electricity spot price in period 1 (p1)
is greater than the spot price in period 2 (p2). Due to the positive
correlation between wholesale electricity spot prices and solar
radiation, the output from a solar system in period 1 (s1) is greater
than the output from a solar system in period 2 (s2). There are N
consumers and the percentage of consumers that purchase
solar systems is given by yA[0,1]. The output from a solar system
st is assumed to be the same for all consumers who purchase a
solar system.
All consumers are assumed to be charged an average-cost
electricity tariff p.
The individual demand for electricity for a
non-solar consumer in time period t is a function of the electricity
t
tariff and is given by d p.
Assuming no net metering, the demand

4
The spot prices are weighted averages based on electricity consumption over
all GXPs and months of the year. The solar insolation data are averaged over all
monitoring stations and months of the year. Since both GXPs and monitoring
stations are more likely to be closer to populated areas, these provide a rough
indication of the spot prices and solar insolation relevant to consumers.

Table 1
Daily solar radiation and nominal spot prices 19972007.
Year

Radiation (kWh/m2)

Final price (NZ$/

MWh)

1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007a

3.90
3.85
3.92
3.91
3.80
3.89
3.97
3.81
3.97
3.94
3.46

$46.1
$39.0
$40.3
$41.3
$86.8
$46.5
$91.2
$35.9
$75.6
$79.2
$33.9

Data were only available up to February 2007.

for electricity at time t for a consumer who purchases a solar

t
system is d p
 st . Thus the aggregate demand for electricity
during period t, Dt p,
is
t

Dt p;
y N yd p
 st N1  yd p.

(1)

Under perfect competition in retail electricity markets, the

long-run retailer prot function Pp;
y equals zero, so
the average-cost rate must be set to cover the cost of purchasing
the electricity on the spot market.5 Hence, the at rate must be
chosen to satisfy
1
2
Pp;
y p  p1 D p;
y p  p2 D p;
y 0.

(2)

Eq. (1) can be substituted into Eq. (2) and the implicit function
theorem used to analyze how the at rate price changes with the
percentage of people who buy solar systems:
@p
Ns1 p1  p
s2 p2  p


.
@Pp;
@y
y
@p

(3)

Since s14s2 and p14p2, we have that

s1 p1  p4s

2 p2  p,
implies
so the numerator in Eq. (3) is positive. Now @Pp;
y=@p40

that if retailers were not constrained by competition and could raise

the price they charge, then retailer prots would increase, which is a
standard assumption. Hence, Eq. (3) implies that
@p
o0.
@y
Thus, if more consumers add solar systems, then all consumers
face a lower average-cost retail price, so there is an external benet
to non-solar consumers from a solar purchase implying that solar
systems are underprovided relative to the social optimum.
Of course, this market failure depends on both spot market and
average-cost prices being set competitively. If there is market power
at either the generator or retailer end, SHW consumers may be
overcompensated in electricity bill savings relative to the social
optimum, in which case SHW systems may be overprovided due to
this market failure. Further research into the presence and nature of
market power in the New Zealand electricity market may provide a
quantitative sense of the extent of this market failure.6
5
This abstracts from non-energy xed per day charges to pay for transmission,
distribution, and any sunk costs.
6
A cursory analysis using aggregated data and assuming competitive markets
indicates that the undervaluation is only a few percent.

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005

ARTICLE IN PRESS
4

K. Gillingham / Energy Policy ] (]]]]) ]]]]]]

2.4. Learning-by-doing appropriability externality

Each SHW installation may also have a dynamic positive
externality if the cost of future installations declines with each
additional installation today (learning-by-doing) and this cost
decline is not fully appropriated by the solar installer (i.e., there
are spillover benets to other solar installers). Of course, this
learning effect is likely present with most edgling technologies,
so a case can be made for SHW policy based on learning only if
learning in SHW technology is at least to some degree nonappropriable and SHW technology has above average learning
characteristics. Learning has been examined in detail in the
context of solar PV (van Benthem et al., 2008; Jamasb, 2007;
Nemet, 2006), yet little evidence exists on learning for SHW. In
New Zealand, average SHW system prices have risen in the past
few years (Table 2) due to other factors, such as increased
enforcement of building consent requirements and increased
labor costs, obfuscating any learning effects.
Thus, while learning currently does not appear to be a
signicant force in the New Zealand SHW market, future learning
is certainly possible and even likely if installations continue to
increase. Whether or not this learning in SHW is nonappropriable
is a topic for future research.
2.5. Informational market failures
If consumers do not have complete and accurate information on
the costs and benets of a SHW system, there may be an
informational market failure whereby consumers over- or underinvest in SHW systems. For example, if consumers decide not to
purchase SHW systems based on a misconception that they
perform poorly and provide little electricity savings, then consumer decision-making would be biased. This may provide
motivation for government intervention through product testing
and informational campaigns. Similarly, if there are principalagent issues or asymmetric information, consumer decisionmaking may again be biased, and there is some evidence it is
biased towards underinvestment in energy-saving technologies
(Sanstad et al., 2006). Of course, informational market failures may
exist in any market, not just the SHW market. Identifying the
nature of the informational market failures in the SHW market and
the cost-effectiveness of policies to correct these market failures
are important questions, but are out of the scope of this paper.
2.6. Summary and discussion
This discussion points to the key importance of the environmental externality in motivating SHW policy, with energy
security, and learning-by-doing externalities providing weaker
motivation. The average-cost pricing market failure may be
important, but further research is needed in this area.
A few effects of SHW purchases are notably not described above.
One is the avoided costs from the construction of new electricity
generation capacity. If spot prices send the correct price signal to
potential investors in new electricity generation capacity, then there
is no market failure. A second is avoided transmission and
distribution (T&D) costs. This is more complicated since T&D costs
are not included in the average-cost rate per kWh, but are charged
to electricity customers per day. Whether there is an external effect
due to T&D costs depends on what rate plan the consumer is on, and
whether this rate plan changes with the installation of a SHW
system. On average, there may not be any effect.
A third is due to the ripple control load-management system
whereby electricity distribution companies (lines companies) can
reduce peak load by sending a signal to electric hot water heaters

Table 2
Range of SHW prices over time (in nominal NZ dollars).
Year

Lower estimate

Upper estimate

Source

2001
2004
2006

NZ$3500
NZ$4000
NZ$4000

NZ$5500
NZ$7500
NZ$8000

EECA (2001)
EECA (2004)
EECA (2006a)

to turn them off. To the extent that electric hot water heaters are
replaced by SHW systems, this may imply additional system costs
due to the loss of a load-management technique.7 Combining
MED (2007b) estimates of the number of residential customers
each lines company serves with Vector Limited (2007) and
Eastland Infrastructure (2007) ripple control capacity estimates
yields an extrapolated total ripple control capacity in New Zealand
of between 450 and 600 MW, a signicant amount. However, this
capacity is somewhat unreliable and degrading, and SHW systems
could easily be installed with ripple control, possibly providing at
least some of the ripple control capacity. Thus, it is not likely there
is a signicant external effect due to ripple control.
This framework of market failures in the SHW market can be
applied to policy analysis. An economically efcient solar policy
would subsidize SHW (and provide information programs) in
order to maximize the net benets of the program. The policy will
have positive net benets when the government cost of the
program (plus any deadweight loss from raising the revenue) is
less than the benets of the program from correcting each of the
externalities described above. Since the environmental externality
appears to be the most important, this implies that if the
environmental externality is already internalized, then the
motivation for a SHW policy may be weak. This is important,
given the Climate Change (Emissions Trading and Renewable
Preferences) Bill 187-1, which would establish an economy-wide
greenhouse gas emissions trading scheme, and is still actively in
the policy debate (NZ Parliament, 2008).

3. Financial attractiveness of SHW in New Zealand

Underlying any economic analysis of SHW policy is a model of
the nancial attractiveness of the technology. The focus in this
paper is on residential SHW systems, which comprise nearly 97%
of the SHW market (Burleigh Evatt, 2006). In any SHW market, the
nancial attractiveness of a SHW system is a function of the
installation cost of the system, the hot water output of the system,
the life of the system, and the avoided cost of the alternate hot
water heater energy source. Of course, the installation cost of
SHW systems varies considerably depending on the size of the
system, type of system, the installer, and the type of home.
This analysis assumes that a SHW system replaces an electric
hot water system, and has an electric rather than natural gas backup element. This seems reasonable since 70% of hot water systems
are electric and, while natural gas systems have come more into
favor lately, the imminent depletion of the largest New Zealand
natural gas elds within the next 10 years is likely to dampen the
recent increase in natural gas systems. This analysis also divides
up the residential SHW market into three segments: family homes
and non-family homes, following Statistics New Zealand (2001)
in dening a family household as a household containing two
or more people usually living together with at least one couple
and/or parentchild relationship, with or without other people.
7
This is a concern because many SHW systems are not installed with ripple
control, and even the ones that are would not turn off when the sun is shining and
they are not drawing any electricity.

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005

ARTICLE IN PRESS
K. Gillingham / Energy Policy ] (]]]]) ]]]]]]

Table 3
Financial attractiveness analysis assumptions.

Collector size (m2)

System efciency (%)
Annual retail electricity variable cost
increase (%)
System lifetime
Mean kWh used for hot water per day (kWh)
Total system cost
Cost of electric system
Marginal system cost (before incentives)

Family
system

Non-family
system

3.5
65
4.2

2.5
65
4.2

20 yrs
10
NZ$7000
NZ$2000
NZ$5000

20 yrs
6
NZ$5500
NZ$1600
NZ$3900

uses for hot water varies greatly by consumer behavior, EECA

estimates that 610 kWh per day are required for home hot water
use (EECA, 2004).
The system costs are based on Table 2 and discussions with
industry. This analysis assumes a SHW system cost of NZ$7000 for
a family household and a system cost of NZ$5500 for a non-family
household. These estimates are compared to the cost of an electric
hot water heating system, which is assumed to be NZ$2000 for a
family household and NZ$1600 for a non-family household. Thus
the marginal system cost is NZ$5000 for SHW systems installed
on a family residence and NZ$3900 for a SHW system installed on
a non-family residence.

3.1. Numerical results

The baseline best estimates of the key parameters for family and
non-family systems are given in Table 3.
While collectors can range in size between 2.5 and 8 m2, they
are more typically sized between 2.5 and 4 m2, which is
sufciently large so that the SHW system provides nearly all of
the hot water for the house in the summer, and at least
signicantly contributes in the winter (EECA, 2006b). One square
meter of collector typically serves about 4070 l of cylinder
volume, depending on the location of the cylinder and the type of
system (EECA, 2004).
The average system efciency assumption of 65% conceals the
heterogeneity in SHW adsorptive surfaces and system designs, but
very roughly captures the general efciency of systems at
converting sunlight to heat. The collector efciency is a function
of the temperature of the heat transfer uid (e.g., water or a
waterglycol mixture) the hotter the uid, the lower the
efciency. If the uid is cold, collectors can be 8090% efcient,
but if the uid is at or near boiling, the efciency drops to close to
0% (EECA, 2001). In a properly designed system, the uid going
through the collector (i.e., the water coming from the bottom of
the tank or the heat transfer uid) will be little more than
lukewarm much of the time. But if no hot water has been used for
several hours, as may be the case late in the afternoon, the uid is
already hot and the efciency of the system signicantly drops.
Thus, consumers who use a moderate amount of hot water
throughout the day when the sun is shining are most likely to
achieve the highest efciency from their system (and may not
have the back-up heating element turn on at all). Pumped systems
also have a slightly lower efciency since they require some
electricity to run the pump. This analysis averages out all of these
factors to yield a reasonable estimate of 65% efciency.
The average SHW system efciency has recently been a subject
of controversy in New Zealand, with some evidence of rapid
degradation of the most commonly sold systems due to poor
quality systems and improper installation, leading to considerably
reduced performance in a matter of a few years (Kane et al., 2007).
This analysis assumes quality control and proper installation
so that the SHW system efciency does not degrade appreciably
over time.8
The 4.2% annual increase in electricity variable cost is
calculated based on historical data from MED (2007b) and is
used to calculate the price of the electricity that solar offsets in
the future. The electricity used for hot water per day sets the
maximum amount of electricity savings possible from a SHW
system in any given day. Unlike a photovoltaic system, a SHW
system cannot sell back to the grid and save more electricity than
the household uses for hot water. While the kWh a household

8
An examination of the sensitivity of the modeling results to this assumption
reveals it is not very sensitive within a reasonable range of efciencies (4585%).

NIWA (2007) data on solar insolation and the parameters in

Table 3 can be used to compute the average annual electricity bill
savings from SHW systems for family and non-family households
by region in New Zealand (Table 4).
The electricity bill savings calculations reveal a surprisingly
small difference in savings across regions. The greatest savings for
systems on family households are in Gisborne (NZ$521 per year),
where both the solar insolation and electricity rates are high.
Interestingly, SHW system production in Gisborne is higher than
in some places that get more total sun in New Zealand (e.g.,
Nelson and Tasman). This is because more sun comes in the
winter months in places like Gisborne than in Nelson and Tasman.
The extra sun in the summer months in Nelson and Tasman does
not translate into more SHW system production since production
per day is limited by the amount of water people use.
Nevertheless, one clear result from examining electricity bill
savings is that the difference between regions is much smaller
than the difference between SHW systems installed in family and
non-family households. How much hot water a household uses is
the most important determinant of the electricity bill savings
from a SHW system households that use more hot water can
save much more on their electricity bill from installing a SHW
system, regardless of where they are in New Zealand. This result
also holds when the cost of the system is factored in, even though
larger systems are more expensive. Table 5 gives two of the
standard metrics used to assess the nancial attractiveness of an
investment (internal rate of return (IRR) and payback period) for
SHW systems without a subsidy.
The results indicate that if consumers of SHW systems on
family homes have a discount rate of less than approximately 10%,
then the SHW system is a good investment. Similarly, for SHW
systems on non-family homes, a consumer must have a discount
rate of less than 8%. Of course, in reality, there are numerous
investments available to consumers, so a SHW system would have
to have a rate of return exceeding that of other investments with a
similar risk prole for a SHW system to be the best investment.
This is not likely to be the case in New Zealand now, when the
nearly risk-free investment in a 12-month term deposit at the
National Bank offers an 8.5% nominal interest rate (National Bank,
2007).
For SHW systems on family homes, the payback period is
around 9 years, and on non-family homes it is around 10 years.
The payback period ignores the time value of money and thus IRR
is a much preferred metric. Nonetheless, most consumers would
balk at a payback period of 10 years, for the average length of time
that New Zealanders stay in any one home is only around 710
years (Deppe, 2007).
Thus, with the current SHW technology and cost, SHW systems
appear to be a marginal investment that might make the most
sense for environmentally minded consumers who use more hot

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005

ARTICLE IN PRESS
6

K. Gillingham / Energy Policy ] (]]]]) ]]]]]]

Table 4
Annual electricity bill savings calculations for typical SHW systems (in NZ$).
Region

Family

Non-family

Savings (kWh)

Electricity rate
(c/kWh)

Electricity bill
savings

Savings (kWh)

Electricity rate
(c/kWh)

Electricity bill
savings

Northland
Auckland
Waikato
Bay of Plenty
Gisborne
Hawkes Bay
Taranaki
Manawatu-Wanganui

2967
2923
2682
2918
2881
2857
2885
2805

15.7
14.6
14.9
15.6
18.1
15.7
15.6
16.0

$465
$426
$399
$455
$521
$450
$451
$449

1900
1879
1755
1875
1844
1833
1852
1798

15.7
14.6
14.9
15.6
18.1
15.7
15.6
16.0

$298
$274
$261
$292
$333
$289
$289
$288

Wellington
Tasman
Nelson
Marlborough
West Coast
Canterbury
Otago
Southland

2733
2862
2862
2872
2651
2649
2625
2498

15.2
14.4
15.0
16.2
18.7
14.3
15.0
15.2

$414
$412
$428
$465
$495
$378
$395
$380

1754
1836
1835
1839
1715
1710
1691
1613

15.2
14.4
15.0
16.2
18.7
14.3
15.0
15.2

$266
$264
$275
$298
$320
$244
$254
$245

Table 5
Financial attractiveness of SHW systems (no subsidy).
Region

Family

Table 6
Financial attractiveness of SHW systems ($1500 subsidy).
Non-family

Region

IRR (%)

Payback
period (yrs)

IRR (%)

Payback
period (yrs)

Northland
Auckland
Waikato
Bay of Plenty
Gisborne
Hawkes Bay
Taranaki
Manawatu-Wanganui
Wellington
Tasman

11
10
9
10
12
10
10
10
9
9

9.1
9.7
10.3
9.2
8.2
9.3
9.3
9.3
10.0
10.0

8
7
7
8
10
8
8
8
7
7

10.7
11.4
11.8
10.8
9.7
10.9
10.9
10.9
11.7
11.7

Nelson
Marlborough
West Coast
Canterbury
Otago
Southland

10
11
12
8
9
8

9.7
9.1
8.6
10.7
10.4
10.7

7
8
9
6
7
6

11.4
10.7
10.0
12.5
12.1
12.4

water than the typical customer. Adding a government nancial

incentive can make a difference. With an incentive of NZ$500, as
in the current incentive scheme for systems that meet the capital
threshold, the payback period decreases by about a year and the
IRR increases by about 2%.
Adding the more signicant incentive of NZ$1500 makes much
more of a difference, as shown in Table 6. NZ$1500 is chosen as a
good benchmark because it represents how much the upper
estimate of SHW system prices increased between 2001 and 2007,
as shown in Table 2.
With a nancial incentive policy of NZ$1500, the IRR for SHW
systems on family homes becomes a quite attractive 1318%, and
the payback period of 68 years is at least in the range of the
length of time that people live in a home. The estimates improve
even more for SHW systems on non-family homes, since the
NZ$1500 incentive applies for any system, regardless of its size.
The IRR for SHW on non-family homes ranges from 12% to 17% and
the payback period from 6.4 to 8.4 years. While still not as

Family

Non-family

IRR (%)

Payback
period (yrs)

IRR (%)

Payback
period (yrs)

Northland
Auckland
Waikato
Bay of Plenty
Gisborne
Hawkes Bay
Taranaki
Manawatu-Wanganui
Wellington

16
15
14
16
18
15
15
15
14

6.7
7.2
7.6
6.8
6.0
6.9
6.9
6.9
7.4

15
14
13
15
17
14
14
14
13

7.1
7.6
7.9
7.2
6.4
7.3
7.3
7.3
7.8

Tasman
Nelson
Marlborough
West Coast
Canterbury
Otago
Southland

14
15
16
17
13
13
13

7.4
7.2
6.7
6.3
8.0
7.7
7.9

13
14
15
16
12
13
12

7.8
7.6
7.1
6.6
8.4
8.1
8.4

attractive as family homes, non-family homes still appear

relatively attractive under such an incentive policy. With these
estimates for systems on family and non-family homes, it seems
reasonable that SHW sales would rapidly increase from the
current relatively limited scale.

4. Economic analysis of large-scale SHW policies

While it is clear that a NZ$1500 subsidy would signicantly
improve the nancial attractiveness of SHW, it is much less clear
whether this is a sensible policy and how it would compare to
policies that mandate SHW installations. This section examines
the cost and implications for energy use and greenhouse gas
emissions of two mandate policies and a 10 year large-scale
subsidy policy designed to promote a sustainable SHW market.
Each of these policies is compared to a no policy baseline case
where the SHW market grows at a modest 5% per year, implying

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005

ARTICLE IN PRESS
K. Gillingham / Energy Policy ] (]]]]) ]]]]]]

annual SHW sales increasing from 3500 today to 56,800 by 2075.

The economic framework developed in Section 2 provides a lens
for examining the economic efciency of each of these policies.

Table 7
Installations, electricity savings and cost of SHW with all homes mandate.
Region

Installations
(thousands)

4.1. SHW mandate policies

Mandate policies can be thought of either as large-scale
policies where the government buys massive quantities of SHW
systems and has them installed, or a policy where the government
requires homebuilders or homeowners to install only SHW
systems.9 The implications for energy use, greenhouse gas
emissions, policy costs and policy benets remain the same in
either case.
4.1.1. Mandated SHW on all homes
The rst mandate policy installs SHW systems on all gridconnected homes in New Zealand over 5 years, regardless of how
old the water heater is. This policy gives a sense of the upper
bound on the electricity savings possible. For comparison
purposes, this analysis computes all energy savings as electricity
savings, rather than dividing up natural gas savings from
electricity savings.
This analysis uses owner-occupied homes in New Zealand from
the 2006 New Zealand Census (2006) for the total number of
homes, and estimates from Statistics New Zealand (2001) to
divide total homes into family and non-family homes. It again
assumes a NZ$7000 cost for a family home and a NZ$5500 cost for
a non-family home.10 However, here the analysis assumes that the
average electricity savings in each home is 80% of the current
estimate used in Section 3. Currently, SHW systems are more
likely to be installed in homes that have a north-facing roof and
relatively high hot water use. Installing SHW systems on all
homes would imply installing SHW on many homes that use less
hot water and have relatively shaded roofs. These factors may be
somewhat offset by the improved installation techniques that
almost certainly would be associated with such a massive
installation program. 80% may be a somewhat optimistic estimate,
but lower values of electricity savings would only serve to
reinforce the result described below.
Table 7 shows that a policy to install SHW on all homes would
be quite expensive, at NZ$1.76 billion per year11 over 5 years for
electricity (or equivalent natural gas) savings of 3223 GWh per
year or 11.6 PJ per year (8.8% of 2006 electricity consumption)
(MED, 2007a). Combining the total cost with the total electricity
savings over the lifetime of the systems implies a cost of
NZ$0.14 per kWh displaced, more expensive than the cost of
other competing sources of electricity.12
The carbon emissions reductions depend on what the marginal
generator is when the electricity savings occur. The marginal
electricity generator during the day in New Zealand is usually the
9
Note the distributional consequences of the policy would be different if it is
borne by taxpayers, borne by homeowners, or shared by homebuilders and
homebuyers (based on the elasticity of demand).
10
There are two offsetting factors that may inuence the cost. The rst is the
economies of scale of installing SHW systems on such a large scale. The second is
that by installing SHW systems on all homes, many homes that are not well suited
for SHW systems will also have systems installed. This would include homes that
have roofs that need reinforcing to put the system up there, homes that are less
accessible, and homes that are shaded by trees that would need to be trimmed to
allow light through for the collector. I assume these factors balance.
11
Note any administrative costs are not explicitly included in the government
costs in any of the analyses in this paper, but these costs will likely be small
relative to the system costs.
12
This estimate of $0.14 per kWh would be slightly higher if the kWh from
SHW systems that would have been installed anyway in the absence of the policy
(free riders) were included. Given current SHW sales, this number is small relative
to the kWh savings from this policy.

Northland
Auckland
Waikato
Bay of Plenty
Gisborne
Hawkes Bay
Taranaki
Manawatu-Wanganui
Wellington

Tasman
Nelson
Marlborough
West Coast
Canterbury
Otago
Southland
TOTAL

Electricity
savings
(GWh/year)

Annual cost
(million NZ$)

56
439
141
97
16
56
40
85
169

129
1006
274
222
37
126
94
190
365

$66
$520
$167
$114
$19
$66
$48
$101
$201

17
17
17
13
203
76
36
1478

41
40
40
26
412
153
69
3223

$21
$21
$20
$15
$246
$92
$37
$1755

coal-red power plant at Huntly, which has a carbon intensity of

0.101 tonnes of carbon per GJ or 0.364 tonnes of carbon per MWh.
Much, but not all, of the electricity savings would likely come
from less coal electricity generation. The weighted average carbon
intensity for all electricity generation is signicantly less:
0.027 tonnes of carbon per GJ or 0.096 tonnes of carbon per
MWh. This is even less than the carbon intensity of natural gas,
which is 0.055 tonnes of carbon per GJ or 0.198 tonnes of carbon
per MWh (EC, 2007). Combining the carbon intensity of the
electricity displaced (70% of the systems) with the carbon
intensity of the natural gas displaced (30% of the systems) yields
0.035 tonnes of carbon per GJ for the average electricity production and 0.872 tonnes of carbon per GJ for coal electricity
production. The analysis presents estimates using both the
average carbon intensity and the Huntly coal carbon intensity,
for the actual carbon intensity of electricity reduced by a SHW
system will highly likely fall between these estimates.
The difference in the carbon dioxide savings from the all
homes mandate and the no policy case is 18,210 tonnes of carbon
(coal carbon intensity) or 7331 tonnes of carbon (average carbon
intensity) over the 20 year lifespan of the SHW systems. Dividing
the cost of the policy by these savings implies a cost of carbon of
NZ$434 per tonne of carbon (coal) or NZ$1077 per tonne of carbon
(average). However, this simple calculation ignores the time value
of money.
A more useful analysis examines the smallest magnitude of the
environmental externality sufcient for the policy to have positive
discounted net benets.13 The policy will have positive net
benets if and only if the environmental externality is larger than
this value (Table 8).
The estimates in Table 8 are above the reasonable range of
NZ$43NZ$142 even under the more optimistic scenario of a 5%
discount rate and the electricity displacing only coal electricity
generation. Thus policymakers would have to believe that the
environmental consequences of global climate change are extremely large in order for the policy of SHW systems on all homes
to be economic efciency-improving. Similar results hold if the
policy installed SHW systems on only all family homes, which

13
This can also be thought of as an aggregate value of the environmental,
energy security, and at rate pricing market failures.

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005

ARTICLE IN PRESS
8

K. Gillingham / Energy Policy ] (]]]]) ]]]]]]

Table 8
Minimum externality for SHW to be economic efciency-improving for the all
homes mandate.

Table 9
Installations, electricity savings and cost of SHW with new homes mandate.
Region

Carbon intensity

Average NZ electricity system

Huntly coal

Externality (NZ$ per tonne of carbon)

5% discount rate

10% discount rate

1647
663

2301
926

account for approximately 70% of all homes (Statistics New

Zealand, 2001).
4.1.2. Mandated SHW on all new homes
The second mandate policy requires SHW systems on all new
grid-connected homes, which includes both the volume built and
custom built new home markets (each about half of the total new
home market) (EECA, 2006a). The policy examined is implemented for a single year, but could be easily extrapolated to a
multiyear policy.14 Since new homes can be designed with SHW
systems in mind, some of the costs of the SHW system, such as the
building consent and installation, can be included in the cost of
the building consent and construction of the entire home. Thus,
this analysis assumes a slightly lower average SHW system cost of
NZ$6000. Similarly, new homes can be designed to have a northfacing roof amenable to a SHW system, so the same electricity
savings can be used as in Section 3. The number of SHW system
installations is calculated as the number of new homes per year
(23,000) distributed to the regions of New Zealand by the
percentage of population in 2006 in each region (EECA, 2006a;
Statistics New Zealand, 2006). The results are shown in Table 9.
Table 9 indicates that a policy to install SHW on all new homes
for 1 year would cost NZ$138 million and would displace
73.7 GWh per year or 0.27 PJ per year (0.2% of 2006 electricity
consumption) if all the new homes had installed electric hot water
systems instead. Combining the cost of the policy with the total
electricity savings over the lifetime of the systems implies a cost
of NZ$0.09 per kWh displaced, which may not be much more
expensive than some thermal electricity generation, especially as
natural gas prices rise.15 The amount of energy savings from the
policy is relatively small though for such a large-scale policy.
Following the same methodology as in the mandate on all
homes, the difference between the carbon dioxide emissions
savings under the policy and in the no policy case is 4712
thousand tonnes of carbon (coal) or 166 thousand tonnes of
carbon (average). Simply dividing the cost of the policy by these
savings yields a cost of carbon of NZ$978 per tonne of carbon
(coal) or NZ$394 per tonne of carbon (average). This again ignores
the time value of money. Table 10 accounts for discounting by
presenting the smallest value of the environmental externality
sufcient for the policy to have positive net benets.
While the estimates in Table 10 are more promising than for
the mandating SHW in all homes policy (Table 8), they are still
above the reasonable range of NZ$43NZ$142 even under the
more optimistic scenario of a 5% discount rate and the electricity
displacing only coal electricity generation. Again, policymakers
would have to believe that the environmental consequences of
global climate change or the other externalities are extremely
14
If the policy was in place for several years, one could imagine costs dropping
due to economies of scale and learning.
15
This estimate of $0.09 per kWh would be slightly higher if consumers who
would have purchased SHW anyway (free riders) are explicitly accounted for. As
before, this number is likely to be small relative to the electricity savings from the
policy.

Northland
Auckland
Waikato
Bay of Plenty
Gisborne
Hawkes Bay
Taranaki
ManawatuWanganui
Wellington

Tasman
Nelson
Marlborough
West Coast
Canterbury
Otago
Southland
TOTAL

Installations
(thousands)

Electricity savings
(GWh/year)

Policy cost
(millions NZ$)

0.8
9.2
2.3
1.7
0.1
0.5
0.3
0.6

2.7
30.9
6.6
5.7
0.2
1.7
1.0
1.9

$4.8
$55.3
$13.9
$10.2
$0.3
$3.0
$1.7
$3.5

2.4

7.5

$14.3

0.3
0.2
0.3
0.1
3.1
0.9
0.1
23.0

1.1
0.7
1.1
0.4
9.4
2.7
0.3
73.7

$1.9
$1.3
$1.9
$0.8
$18.9
$5.6
$0.7
$138.0

Table 10
Minimum externality for SHW to be economic efciency-improving for the all new
homes mandate.
Carbon intensity

Average NZ electricity system

Huntly coal

Externality size (NZ$ per tonne of carbon)

5% discount rate

10% discount rate

1132
456

1581
637

large in order for the policy of SHW systems on all new homes to
be economic efciency-improving.

4.2. Subsidy policy

In contrast to a mandate, a subsidy policy does not guarantee a
particular environmental benet, but rather allows the market to
determine the nal allocation. In a stable market the optimal
subsidy policy under perfect information is simply to set the
subsidy to the magnitude of the externality. However, in an
emerging market like the SHW market, a subsidy policy is
complicated by diffusion of the new technology into the market,
and the inuence that a subsidy policy may have on this diffusion.
Along with an information policy, a subsidy policy that makes
SHW nancially attractive to more consumers can lead to more
rapid diffusion, which has the potential to bring down costs,
through learning-by-doing, with the potential to eventually lead
to a sustainable market.
Of course, modeling any new technology, including SHW
systems, is difcult due to the complex interplay of the new
technology prices, competing technology prices, lack of information in the market, and learning effects. A common assumption
that captures all of these factors in one simple framework is to
model the diffusion of a new technology as a dynamic system that
is represented by a logistic or S-shaped curve. An underlying
premise of this logistic model is that different categories of people
adopt at different times: rst the venturesome innovators and
early adopters, later the majority, and nally the laggards. The
price must decrease and characteristics of the technology must

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005

ARTICLE IN PRESS
K. Gillingham / Energy Policy ] (]]]]) ]]]]]]

SHW Industry Sales

Table 11
Ordinary least squares regression for SHW adoption model.

4000
3500
Units sold per year

New volume built

New custom built

Retrot

4.85 (0.18)
0.33 (0.04)

3.48 (0.18)
0.36 (0.04)

5.69 (0.18)
0.32 (0.04)

3000
ln(U0)
t

2500
2000
1500

Standard errors in parentheses.

 Indicates signicant to the 1% condence level.

1000
500
0
2001

2002

2003

2004

2005

2006

2007

70,000

Financial Year ending in

Historical Data

Fitted Data (Policy)

No Policy

60,000

dXt
a  bXt,
dt

(4)

where a is the diffusion rate and b is the crowding effect. The

diffusion rate captures the base rate at which the technology will
be adopted and the crowding effect captures how the rate of
adoptions slows as the market becomes saturated. In fact, when
the growth of sales is set to zero (set the left-hand side in Eq. (4)
to zero), then X(t) a/b (Kingsland, 1995).
The solution to this differential equation can be seen most
easily by examining the ratio of total sales to potential remaining
sales before saturation U(t), dened as
Xt
.
Ut
X max  Xt
This ratio is easy to interpret. When U(t) 1, exactly one half of
the potential adopters have adopted the technology. When U(t) is
very small, most of the market has not adopted the technology;
when U(t) is very large, most of the market has adopted.
The solution to Eq. (4) can then be written as
Ut

X0
eat .
X max  X 0

(5)

Noting the denition of U(t), Eq. (5) can be rewritten as

at

Ut U 0 e .
Taking the logarithm of both sides of this equation yields the
linear equation
lnUt lnU 0 at.

(6)

Assuming a mean zero stochastic error term, Eq. (6) can be

estimated by ordinary least squares to give the best linear

50,000
40,000
30,000
20,000
10,000

29

27

20

25

20

23

20

21

20

19

20

17

20

15

20

13

20

11

20

09

20

07

20

05

20

03

20

20

01

0
20

improve as the technology is adopted by later groups of people

(Rogers, 2003).
Since 2001 and until 2006, the SHW market has followed the
beginning of a classic logistic curve (Fig. 1). However, the increase
in system costs (Table 2) has recently caused new installations to
level off. With a NZ$1500 subsidy SHW systems would be
nancially attractive to a wider audience and the logistic diffusion
model may just be a reasonable model of technology adoption.
Along with the logistic diffusion model, system costs will likely
decrease, especially if the subsidy is implemented in concert with
an information and training program, as is currently the case with
EECA policy. This analysis models the decrease in system cost as a
simple learning-by-doing process. It is assumed that the government sets the nancial incentive so the marginal cost of the
system after the incentive remains constant. In this manner, the
incentive is phased out after 10 years, by 2018.
The logistic dynamic system models the growth of SHW sales
X(t) by the following differential equation:

Annual SHW Sales

Fig. 1. Solar hot water industry sales. Source: EECA (2006a) and Deppe (2007).

financial year ending in

Fig. 2. Installations over time.

predictor. This analysis uses historical SHW sales data (Fig. 1)

for new volume built homes, new custom built homes, and new
retrot homes (EECA, 2006b). Table 11 displays the estimation
results.
The estimation results are not easily interpretable, but using
the tted values yields number of SHW sales (X(t)) in each year for
each market following the classic S-shaped logistic diffusion
curve. Fig. 2 illustrates the total SHW sales under the subsidy and
the no policy baseline.
Such diffusion of SHW systems could only occur with
decreases in prices of SHW systems, as well as improvements in
quality and better publicity. The cost of SHW systems is modeled
by the standard learning-by-doing relation:
CK aK b ,
where C is the cost of SHW systems, K is the cumulative installed
SHW systems, a is a normalization parameter to match the base
year cost, and b is the learning elasticity. This structure implies
that a doubling of cumulative installations will reduce system
costs by a factor of 2b, which is also known as the progress rate
(Loschel, 2002). There is little evidence for a progress rate for SHW
systems, but a common assumption for emerging renewable
energy technologies is in the range of 0.80.9, suggesting that a
baseline value of 0.85 is reasonable (Jamasb, 2007; van Benthem
et al., 2008).

4.2.1. Results
Fig. 3 indicates the path over time of cumulative installations,
marginal system costs, and the nancial incentive. Prior to 2008 is
historical data. The current cost (NZ$4670) is a weighted average
of the marginal installation cost without an incentive for systems
on family homes and systems on non-family homes, and the same
weighted average cost is used for all years. This time path of
subsidies and cumulative installations can be used to calculate the
cost of the policy (Table 12).

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005

ARTICLE IN PRESS

$5

450

$5

400

$4

350

$4

300

$3

250

$3
200

$2

150

$2
$1

100

$1

50

$0

SHW systems (Thousands)

K. Gillingham / Energy Policy ] (]]]]) ]]]]]]

SHW system cost (Thousands)

10

financial year ending in

system cost

incentive

cum. installations

Fig. 3. Marginal system costs and incentives in nancial incentive policy.

Table 12
Incentive per system and cost of policy.
Year

Financial
incentive ($)

Table 13
SHW systems and electricity savings.
SHW system
sales

Cost of policy
(millions $)

Historical data
2001
2002
2003
2004
2005
2006
2007

300
500a

500
1000
1500
1750
2800
3500
3500

1.0

Policy case
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018

1500
1374
1227
1066
898
729
565
407
258
119

5581
7241
9236
11,590
14,326
17,466
21,017
24,948
29,179
33,575
37,956

8.4
9.9
11.3
12.4
12.9
12.7
11.9
10.1
7.5
4.0
0.0

Only for systems that meet the capital threshold.

Since any improvements in the SHW efciency over time are

likely to be offset by decreases in the number of most productive
new sites for SHW systems, this analysis assumes that
the electricity savings per system remains the same over time.
Table 13 presents the total electricity savings from the policy by
year until 2025, and then every 5 years until 2075.
By 2075 the no policy SHW system sales are beginning to catch
up with the sales with the subsidy. The cumulative difference in
electricity savings between the policy and no policy is 310 PJ from
2008 to 2075, with a mean difference of 4.6 PJ/year and a
maximum difference of 7.1 PJ/year by 2047. While not insignificant, these results mirror those of the mandate policies, which
show that SHW can only play at most a reasonably small role in
the future energy system.
Following the same methodology as in the mandate policies,
the mean carbon dioxide emissions savings until 2075 due to the

Year Incentive policy

No policy

Difference

Cumulative Electricity Cumulative Electricity Cumulative Electricity

systems
savings
systems
savings
systems
savings
(thousands) (PJ/year)
(thousands) (PJ/year)
(thousands) (PJ/year)
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2030
2035
2040
2045
2050
2055
2060
2065
2070
2075

18.9
20.4
22.5
25.2
26.6
28.6
30.5
34.3
39.4
46.2
54.9
65.8
79.1
95.1
114.1
136.1
161.2
189.2
219.7
252.4
286.6
321.7
357.2
392.7
427.7
588.3
718.4
820.3
899.4
960.6
1008.0
1044.7
1073.1
1095.0
1112.0

0.20
0.21
0.24
0.26
0.28
0.30
0.32
0.36
0.41
0.48
0.57
0.69
0.83
0.99
1.19
1.42
1.68
1.98
2.30
2.64
2.99
3.36
3.73
4.10
4.47
6.15
7.51
8.57
9.40
10.04
10.53
10.92
11.21
11.44
11.62

18.9
20.4
22.5
25.2
26.6
28.6
30.5
32.5
34.5
36.6
38.8
41.1
43.5
46.0
48.6
51.4
54.2
57.2
60.3
63.6
67.0
70.5
74.3
78.2
82.3
105.9
135.8
173.9
222.3
284.0
362.8
463.2
591.3
722.2
823.6

0.20
0.21
0.24
0.26
0.28
0.30
0.32
0.34
0.36
0.38
0.41
0.43
0.45
0.48
0.51
0.54
0.57
0.60
0.63
0.66
0.70
0.74
0.78
0.82
0.86
1.11
1.42
1.82
2.32
2.97
3.79
4.84
6.18
7.55
8.61

1.8
4.9
9.6
16.1
24.7
35.6
49.1
65.4
84.7
106.9
132.0
159.4
188.8
219.6
251.2
283.0
314.5
345.5
482.4
582.6
646.5
677.1
676.6
645.3
581.5
481.8
372.8
288.5

0.02
0.05
0.10
0.17
0.26
0.37
0.51
0.68
0.89
1.12
1.38
1.67
1.97
2.29
2.62
2.96
3.29
3.61
5.04
6.09
6.75
7.07
7.07
6.74
6.08
5.03
3.90
3.01

policy (i.e., difference between the policy and no policy) are 398
thousand tonnes of carbon per year (coal) or 241 thousand tonnes
per year (average). The maximum carbon dioxide emissions
savings due to the policy in any given year occurs in 2048 with

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005

ARTICLE IN PRESS
K. Gillingham / Energy Policy ] (]]]]) ]]]]]]

Table 14
Minimum externality for SHW to be economic efciency-improving for the largescale subsidy policy.

Table 15
Minimum externality with higher initial SHW system cost (NZ$1000).
Carbon intensity

Carbon intensity

Average NZ electricity system

Huntly coal

Externality size in $
per tonne of carbon
5% discount rate

10% discount rate

29
16

86
44

a savings of 715 thousand tonnes of carbon per year (coal) or 443

thousand tonnes per year (average).
But is this massive subsidy policy economic efciencyimproving? Ignoring the time value of money, a simple division
of the total cost of the policy by the total carbon dioxide savings
(up to 2075) due to the policy yields a value of NZ$3.79 per tonne
of carbon (coal) or NZ$6.25 per tonne of carbon (average).16
However, the costs are in the near future while most of the
emission reductions are farther in the future, so ignoring the time
value of money produces a result strongly biased downwards.
Table 14 accounts for discounting by presenting the smallest value
of the environmental externality sufcient for the policy to have
positive discounted net benets with the result that the policy
can be economic efciency-improving.
These results in Table 14 suggest that if the subsidy policy
(possibly coupled with information and installer training policies)
is successful in causing the market diffusion modeled by the
logistic curve, then the policy would have positive net benets
under reasonable assumptions of the size of the externality.
4.3. Sensitivity analysis
A thorough sensitivity analysis indicates that the primary
results appear to be relatively robust to varying all of the major
parameters. A substantial increase in the assumed initial SHW
system cost (by NZ$1000) requires a higher assumed value for the
environmental externality in order for the policy to have positive
discounted net benets (Table 15).
The subsidy policy results are, of course, most sensitive to the
assumption of the logistic model of SHW technology diffusion.
Alternative assumptions of the diffusion of new technologies
typically follow a similar S-shaped curve, but all have a few
caveats. Most importantly, modeling the diffusion of a new
technology with an S-shaped curve assumes that there are no
competing technologies that also are diffusing in the market and
that the technology continues to improve during the diffusion to
become appealing to a wider range of consumers (Rogers, 2003).
These appear to be reasonable assumptions in the case of SHW
systems, but relaxing them would render the subsidy policy much
less economically efcient, for its benets rely at least partly on
the policy instigating diffusion of SHW systems.
5. Conclusions
With the increased interest in renewable energy policies to
meet long-run greenhouse gas emission reduction goals, understanding the implications and economic merits of policies to
promote SHW is a critical task for policymakers. This paper
provides guidance to policymakers looking to understand the
economic motivation for SHW policy, and applies this economic

11

Externality size in $ per tonne of carbon

5% discount rate

10% discount rate

SHW system on all homes policy

Average NZ electricity system
Huntly coal

1894 (1427)
763 (575)

2647 (1994)
1065 (803)

SHW systems on new homes policy

Average NZ electricity system
Huntly coal

1320 (1132)
532 (456)

1845 (1581)
743 (637)

84 (29)
46 (16)

221 (86)
112 (44)

Financial incentive policy

Average NZ electricity system
Huntly coal

Parentheses indicate the values with baseline SHW system costs.

framework to examine large-scale policies to promote SHW. Both

the economic framework and the policy modeling are most
relevant to New Zealand policy, but they provide a methodology
for applications around the world.
The investigation into the economic efciency of SHW policy
points to the critical importance of the global climate change
externality underpinning the rationale for policy. This is particularly salient to policy in New Zealand, for a successful economywide greenhouse gas cap-and-trade program would internalize
this externality, removing much of the motivation for SHW policy.
Energy security, learning-by-doing, and average-cost pricing
market failures provide much weaker motivation than the
environmental externality.
The examination of large-scale SHW policies indicates that
even under the most extreme case mandated SHW systems on
all homes SHW will still only reduce demand by 9% of the 2006
New Zealand electricity consumption. Moreover, policies mandating SHW systems on all homes and all new homes do not appear
to be justiable on economic efciency grounds.
However, SHW systems are currently marginally nancially
attractive with an IRR of 10% for a typical family home. The IRR
increases to a very attractive 18% for family homes with a
substantial subsidy of NZ$1500, which was chosen to match the
increase in the upper bound of the range of prices between 2001
and 2006. The modeling results indicate that a large-scale subsidy
program in concert with informational policies that lead to a
sustainable SHW market could be economic efciency-improving
under reasonable assumptions of the magnitude of the environmental externality. This result is quite robust to most major
parameters, but is dependent on the assumed form of diffusion.
These results imply that if New Zealand policymakers do not
implement a cap-and-trade policy, then a large-scale subsidy and
informational policy to promote SHW may be worth considering.
Future research may shed light on the importance of the averagecost pricing market failure, which could provide motivation for solar
policy absent a climate change externality. On the other hand,
shifting to real-time-pricing or time-of-use pricing may internalize
this market failure, and provide greater benets for load-management and reducing peak loads. Finally, broadening the scope of
SHW policies to commercial and industrial buildings may increase
the total potential for energy savings from SHW systems.

Acknowledgments
16

Since the no policy and policy cases are rapidly converging after 2075,
extending the time series further does not signicantly lower the cost per tonne of
carbon.

The author would like to thank Bart van Campen, John Panzar,
Will Oxley, Rodney Deppe, Joseph Mayhew, Jonathan Leaver, and

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005

ARTICLE IN PRESS
12

K. Gillingham / Energy Policy ] (]]]]) ]]]]]]

Jim Sweeney for useful discussions. All remaining errors are the
sole responsibility of the author. The author would also like to
thank Fulbright New Zealand for providing the fellowship making
this research possible.
References
Bohi, D., Toman, M., 1996. Economics of Energy Security. Kluwer Academic
Publishers, Norwell, MA.
Borenstein, S., 2005. Valuing the Time-Varying Electricity Production of Solar
Photovoltaic Cells. CSEM WP 142. University of California Energy Institute,
Berkeley, CA.
Burleigh Evatt, 2006. Solar Water Heating Strategy Options for EECA. Burleigh
Evatt.
Deppe, R., 2007. Personal Communication on the Solar Hot Water Industry, July 25.
Eastland Infrastructure, 2007. Personal Communication with Henrike Seifert of
Eastland Infrastructure, June 15, Gisborne, New Zealand.
EC, 2007. Centralised Dataset. New Zealand Electricity Commission, Wellington,
New Zealand.
EECA, 2001. Solar Energy Use and Potential in New Zealand. Energy Efciency and
Conservation Authority, Wellington, New Zealand.
EECA, 2004. Solar Water Heating Fact Sheet 3. Energy Efciency and Conservation
Authority, Wellington, New Zealand.
EECA, 2006a. Increasing the Uptake of Solar Water Heating: A Government
Discussion Document. Energy Efciency and Conservation Authority, Wellington, New Zealand.
EECA, 2006b. Solar Water Heating Guidebook: A Technical Guide for Building
Industry Professionals. Energy Efciency and Conservation Authority,
Wellington, New Zealand.
EECA, 2007. Solar Water Heating Programme Outline. Energy Efciency and
Conservation Authority, Wellington, New Zealand.
EHMS, 2006. Microgeneration Potential in New Zealand: A Study of Small-scale
Energy Potential Generation, Prepared for Parlimentary Commissioner for the
Environment. East Harbour Management Services, Wellington, New Zealand.
Genesis Energy, 2007. Huntly Power Station. Genesis Energy, Auckland,
New Zealand.
Gillingham, K., Newell, R., Palmer, K., 2004. Retrospective Examination of Demand
Side Energy Efciency Policies. Resources for the Future Discussion Paper 0419REV, Washington, DC.
Gillingham, K., Newell, R., Palmer, K., 2006. Energy efciency policies: a
retrospective examination. Annual Review of Environment and Resources 31,
161192.
Isaacs, N., Camilleri, M., French, L., Pollard, A., Saville-Smith, K., Fraser, R., Rossouw,
P., Jowett, J., 2006. Energy Use in New Zealand Households: Report on the 10
Year Analysis for the Household Energy End-use Project (HEEP). BRANZ,
Porirua, New Zealand.

Jamasb, T., 2007. Technical change theory and learning curves: patterns of progress
in electricity generation technologies. Energy Journal 28, 5171.
Joskow, P., Tirole, J., 2006. Retail electricity competition. Rand Journal of Economics
37, 799815.
Joskow, P., Tirole, J., 2007. Reliability and competitive electricity markets. Rand
Journal of Economics 38, 6084.
Kane, C., Pollard, A., Zhao, J., 2007. An Inspection of Solar Water Heater
Installations: Final Draft. BRANZ Limited, Porirua City, New Zealand.
Kingsland, S., 1995. Modeling Nature. University of Chicago Press, Chicago, IL.
Loschel, A., 2002. Technological change in economic models of environmental
policy: a survey. Ecological Economics 43, 105126.
McChesney, I., 2005. Developing the market for larger scale solar water heating
applications in New Zealand. East Harbour Management Services, Ltd. and
Enercon Ltd.
MED, 2007a. New Zealand Energy Data File. Ministry of Economic Development,
Wellington, New Zealand.
MED, 2007b. Schedule of Domestic Electricity Prices: Updated 15 February 2007.
Ministry of Economic Development, Wellington, New Zealand.
National Bank, 2007. National Bank: Term Deposits. National Bank of New Zealand,
Wellington, New Zealand.
Nemet, G., 2006. Beyond the learning-curve: factors inuencing cost reductions in
photovoltaics. Energy Policy 34, 32183232.
NIWA, 2007. Clio Database. National Institute for Water and Atmospheric
Research.
NZ Parliament, 2008. Climate Change (Emissions Trading and Renewable
Preference) Bill 187-1, Wellington, New Zealand.
Poletti, S., 2006. Ph.D. Dissertation, Economics Department. University of Auckland, Auckland, New Zealand.
Rogers, E., 2003. Diffusion of Innovations. Free Press, New York.
Roulleau, T., Lloyd, C.R., 2008. International policy issues regarding solar water
heating, with a focus on New Zealand. Energy Policy 36, 18431857.
Sanstad, A., Hanemann, M., Auffhammer, M., 2006. End-use Energy Efciency in a
Post-Carbon California Economy: Policy Issues and Research Frontiers,
Managing Greenhouse Gas Emissions in California. The California Climate
Change Center at UC-Berkeley, Berkeley, CA.
Statistics New Zealand, 2001. National Family and Household Projections:
2001(base)-2021 Update. Statistics New Zealand, Wellington, New Zealand.
Statistics New Zealand, 2006. 2006 New Zealand Census. Statistics New Zealand,
Wellington, New Zealand.
Sumner, I., 2004. Solar Water Heater Performance Evaluation. URS New Zealand
Ltd., Auckland, New Zealand.
Thomas, S.E., Lloyd, C.R., 2006. Experimental and Simulated Performance of
Commercially Available Solar and Heat-pump Water Heaters in New Zealand.
University of Otago, Dunedin, New Zealand.
van Benthem, A., Gillingham, K., Sweeney, J., 2008. Learning-by-doing and the
optimal solar policy in California. Energy Journal 29, 131151.
Vector Limited, 2007. Personal Communication with Will Oxley of Vector Limited,
July 31, Auckland, New Zealand.

Please cite this article as: Gillingham, K., Economic efciency of solar hot water policy in New Zealand. Energy Policy (2009),
doi:10.1016/j.enpol.2009.01.005