Empirical Economics
https://doi.org/10.1007/s00181-020-01955-8
Willingness to pay to ensure a continuous water supply
with minimum restrictions
Clevo Wilson1 · Wasantha Athukorala2 · Benno Torgler1 · Robert Gifford3 ·
Maria A. Garcia-Valiñas4 · Shunsuke Managi1,5
Received: 14 October 2016 / Accepted: 27 July 2020
© Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract
This study provides a quantitative assessment of the willingness to pay to avoid
water use restrictions taking into account psychological, attitudinal and behavioural
influences. We analyse determinants of households’ willingness to pay to ensure a
continuous water supply in Brisbane, Australia. The results show that in addition
to socio-economic variables, attitudinal and behavioural factors—including values,
norms, and beliefs—influence residents’ valuation. They also underscore the importance of accounting for socio-economic variables and pertinent psychological and
behavioural aspects when implementing policies to manage and conserve urban water.
Keywords Water use restrictions · Willingness to pay · Residential sector ·
Behavioural change
1 Introduction
Water shortages lead to inconvenience for residents, the imposition of an economic cost
on the community, and a loss of a producer surplus for the water authority (Dandy 1992;
Hensher et al. 2005; March et al. 2012). In such cases, residents look for alternative
ways to avoid or comply with the restrictions. In general, the willingness to pay
(WTP) to avoid water restrictions (WTPAR) is driven by such factors as the costs
B
Shunsuke Managi
managi@doc.kyushu-u.ac.jp
1
QUT Business School, Queensland University of Technology, Brisbane, Australia
2
Department of Economics, University of Peradeniya, Kandy, Sri Lanka
3
School of Psychology, University of Victoria, Victoria, BC, Canada
4
School of Economics and Business, University of Oviedo, Asturias, Spain
5
Urban Institute & Departments of Urban and Environmental Engineering, Kyushu University, 744,
Motooka Nishi-ku, Fukuoka, Japan
123
C. Wilson et al.
of mitigating action, the value that residents attach to maintaining their gardens, and
inconvenience factors associated with water rationing (Hensher et al. 2006). Hence,
WTPAR represents the value that citizens attach to maintaining a continuous water
supply.
One of the costs attached to water restriction is the cost of creating substitute
supplies such as rainwater tanks (purchase and installation), which may be linked to
some residents’ WTPAR (Brennan et al. 2007). Other costs include those for waterefficient landscaping, replacing plants that die during a period of drought, and finding
alternatives for car washing and yard clean-up (Blamey et al. 1999). The WTPAR
will not only vary across a community according to residents’ preferences, but also
because the restrictions will impact different citizens to varying degrees. For example,
residents who value their well-maintained gardens as a visual amenity and a means
of relaxation may pay a higher price to avoid restrictions. In other words, the loss of
utility from loss of gardening would be reflected in the WTPAR estimates.
Although previous studies apply various methods to assess consumers’ valuation
of water use restrictions (Dandy 1992; Griffin and Mjelde 2000; Gordon et al. 2001;
Hensher et al. 2006), none undertake empirical analyses which take into account both
socio-economic and behavioural/psychological factors influencing WTPAR. They do,
however, provide evidence that water consumers dislike constraints on water usage
and are willing to pay higher bills to avoid them (Griffin and Mjelde 2000; Gordon
et al. 2001; Hensher et al. 2005). Nevertheless, a number of previous studies (not only
limited to urban water and energy use) also suggest that the explanatory power of socioeconomic variables is relatively limited in the context of residents’ environmental
behaviour (Dietz et al. 1998; McFarlane and Boxall 2003; Halkos and Matsiori 2018).
The concept of financial sustainability is one of the important aspects for sustainable
operation of a public water system. Financial sustainability provides the monetary
resources necessary to enable a system to operate and distribute water effectively while
allowing for expansion to serve additional customers. In this context, the existing rate
structure or tariff structure plays a key role. In general, the existing tariff structure
should have the ability to generate the revenue necessary to cover the system’s fixed
and operating expenses and contain an incentive to conserve water. At the same time,
the rate structure needs to be “fair” to different classes and levels of users. In order
to design such a system, it is important to know users’ WTP under different rating
schemes where users facing multiple options. This issue is also investigated by this
study which is important for policy makers who need to understand peoples’ WTP for
different rate structures.
This study further contributes to these findings by indicating that a large majority of
residents are willing to pay to ensure a continuous water supply with minimum restrictions. An important contribution to the literature by our study is therefore to employ, in
addition to socio-economic variables, psychological, attitudinal and behavioural variables to estimate the factors that influence residents’ WTPAR. By doing so, we aim
to make an important contribution to the development and implementation of policies
designed to more efficiently conserve urban water resources.
The remainder of the paper unfolds as follows. Section 2 reviews previous research
on water use restrictions and describes the study’s context. Section 3 outlines the
economic model for investigating WTPAR determinants and the corresponding esti-
123
Willingness to pay to ensure a continuous water supply…
mation procedure. Section 4 reports the empirical results, and Sect. 5 provides the
paper’s conclusions and policy recommendations.
2 Literature review
The early research on WTPAR is to be found in Russell et al.’s (1970) study of a 1960’s
drought in New England. This study identifies small observable losses to water users
of between US$5 and US$13 per capita. Later studies by Dworkin (1973) and Kates
(1979) confirmed this finding for the USA. They found that the out-of-pocket expenses
associated with drought-induced shortages for about 2.5 million urban residents per
year were modest—at between US$3 and US$5 per capita in 1973. In contrast, research
on the 1976–77 droughts in Northern California estimated quite large costs attributable
to drought-induced water shortages (Meral 1979; Nelson 1979).
To measure actual WTPAR, Howe et al. (1994) conducted a contingent valuation
study of residents in the three US towns of Boulder, Aurora and Longmont, which
were faced with restricting their outdoor water use to 3 h every third day for the
months of July, August and September. These respondents indicated that they would
accept payment of between US$4.53 and US$13.99 per month, on average, as security
against a decrease in supply. In a similar study, Griffin and Mjelde (2000) e-mailed
a questionnaire to 4856 households in seven cities in Texas, two of which had previously experienced water supply shortfalls. A closed-ended question asked participants
about their valuation of a hypothetical current shortfall in terms of restrictions of varied strength and duration. Not only did the participants prefer to pay rather than be
restricted in their water usage but their WTPAR increased as the duration of the restrictions increased. Specifically, when facing a 20% shortage in water supplying capacity,
they expressed a WTP an average of US$27.46, US$29.86 and US$32.27 to avoid
restrictions covering periods of 14, 21 and 28 days, respectively.
Such a WTPAR is also evident in Koss and Khawaja’s (2001) study on contingent householder valuation of water security in 10 Californian water districts. These
respondents indicated that they were willing to pay, on average, between US$11.67
and US$16.92 per month to avoid water use restrictions dependant on the frequency
and severity of restrictions. In Australia, Gordon et al. (2001) found that residents
would be willing to pay AUS$10.00 on average to avoid restrictions when facing
a 10% supply shortage. Two surveys of residential water tariffs were conducted by
the OECD in 1999 and in 2008. Linear volumetric tariffs were found to be the most
common form of water tariffs in OECD countries which were used by 90 out of 184
utilities. Similarly, Hensher et al. (2005), using a choice experiment methodology to
assess consumers’ WTPAR in Canberra, hypothesise that WTP was conditional, not
only on the restriction’s duration, but also on the type of activities restricted and day
of the week.
On average, in 2003, they estimated that consumers were willing to pay as much as
AUS$239.00 annually to avoid virtually any chance of restrictions on sprinkler use,
lawn watering and any other outdoor water uses. Willis et al. (2005) used a stated
choice experiment model to estimate benefits to water company customers in the
UK by valuing 14 distinct attributes of water and waste water service provision. One
123
C. Wilson et al.
of those attributes was the frequency of water restrictions. The study surveyed 1000
households and 500 businesses and found that, on average, Yorkshire households were
willing to pay £3.20 per year to avoid water restriction for a 3-month period. Another
choice experiment by Tapsuwan et al. (2007) showed that those households willing to
move away from the status quo (a scenario in which they would have to endure severe
water restrictions) would be willing to pay 22% more on their water bill to be able
to use their sprinklers up to 3 days a week. They would also, on average, be willing
to pay around 50% more in water charges to finance a new supply source rather than
enduring severe water restrictions.
Martin-Ortega et al. (2011) assessed the non-market value of allocating enough
water to the environment to ensure environmental services are sustained when water
is scarce. This study also investigated the non-market value of guaranteeing water
supply for secondary household uses. The results show that the population derives
significant benefits not only from the direct use of water, but also from non-use values
related to ecological uses which, however, has a considerably lower impact on the
consumer surplus (Martin-Ortega et al. 2011). Cooper et al. (2011) used a contingent
valuation study to investigate consumers’ WTP to avoid urban water restrictions in the
Australian states of New South Wales and Victoria. It also investigated the influence
of cognitive and exogenous dimensions on the utility gain associated with avoiding
water restrictions.
A number of studies employ the applied choice experiment method or the contingent valuation method to identify households’ valuation of energy restrictions such
as electricity. Reichl et al. (2013) examined the costs of power outages to Australian
households using contingent valuation method. Woo et al. (2014) assessed residential
preferences for the reliability of electricity in Hong Kong. Hensher et al. (2014) investigated the household’s WTP to avoid specific restrictions on service supply quality in
terms of residential electricity supply quality and reliability. Morrissey et al. (2018)
examined residents’ WTP for avoiding a power outage while Carlsson and Martinsson
(2008) examined Swedish households’ marginal WTP for reducing unplanned power
outages. Kim et al. (2019) analysed residential consumers’ (WTP) to avoid power outages using 1000 households’ data in South Korea. The results show that the mean of
households’ monthly WTP amounts to USD 1.41. Converting it into an annual value
and then expanding the value to the country indicate that the annual national value
amounts to USD 335.3 million.
Although the reasons for the wide variations observed in WTPAR are unclear,
such differences are probably linked partly to the methods used and partly to differences in the specific trade-offs being targeted in particular countries. In particular,
behaviourally, different responses can be anticipated dependant on the historical context of water availability and expectations of shortages. This makes it difficult to
draw any general policy conclusions. It also needs to be pointed out that existing studies focus on estimating the amount of WTPAR and not its determinants
which are important inputs for policy decision-making. Moreover, many studies apply
the choice experiment method, the relevance of which for policy making is limited
because the corresponding models ignore important socio-economic, psychological
and behavioural aspects. This study aims to fill this research gap by incorporating
all three types of variables and determining how these factors influence WTPAR. In
123
Willingness to pay to ensure a continuous water supply…
addition, this study is designed to reveal a community’s preference between the two
main types of pricing regimes—a flat rate and an increasing block rate system—and
to reveal the extent to which WTP levels differ for both regimes.
3 Method and data
The data for this study were collected through a survey conducted among residents
living in suburbs within the Brisbane City Council region. The objective was to assess
their water use and attitudes to water use management and conservation. Commencing
in 2009, this survey employed a multi-stage sampling procedure to produce a random
sample of Brisbane residents. In the first stage, we ranked the Brisbane suburbs by
taking the percentage of households belonging to the highest income group and then
selected every second suburb to yield a sample containing 90 of the 189 suburbs under
Brisbane. In the second stage, using proportional-to-population size sampling and
including only owner-occupied homes, we selected every third household from a list
of all households in the selected 90 suburbs. Only those that owned houses and paid
water rates were included. This sampling resulted in a total of 37,341 households being
sent letters asking their consent to participate in the three-year study. Of these, 3475
households agreed and in 2010 were sent a detailed questionnaire. The initial survey
response rate was 2141, but because some respondents did not answer all questions
on variables used in the study, they were dropped from the final data set, leaving 1202
completed cross-sectional observations. It is evident that the non-response issue is
one of the major problems in a survey such as this which occurs when the potential
participant is unwilling to participate or impossible to contact. Non-response can result
in a reduction in precision of the study and may bias results. In general, increasing
response rates will necessarily reduce non-response bias. In order to reduce the nonresponse rate, we sent the reminders for the respondents which helped us to increase
the respond rate during the survey.
The survey questionnaire had nine main sections. Section 1 covered general information on such household water conservation measures as domestic fixtures, domestic
appliances, garden and lawn maintenance, pool, rainwater tanks, grey water use, household water consumption during the 2008–2010 period and water conservation habits
and strategies. Section 2 sought details of future water saving strategies. Section 3 dealt
with water demand strategies that could be used by water management authorities such
as restricting the supply of water, water pricing, provision of incentives, buying back
surplus water, education, moral persuasion, promotion of low-consumption technologies and increasing the supply of water to residents. Section 4 collected information on
households’ attitudes towards water conservation. Section 5 asked about households’
awareness and knowledge of water pricing and particularly about their WTP to avoid
water restrictions. Section 6 included items on barriers to water conservation, while
Sects. 7 and 8 addressed environmental and social attitudes and other general questions.
The final Sect. 9 collected socio-economic data, as well as household demographics
such as age, gender, education level, household size and income.
In order to measure the WTP to ensure a continuous water supply with minimum restrictions, we employ both OLS and GLS regressions. In this study, based on
123
C. Wilson et al.
Amemiya (1977), we employed a three-step feasible generalised least squares (FGLS)
model. First, we perform a preliminary estimation to obtain cross section specific
residual vectors. Then, we use these residuals to form estimates of the cross-specific
variances. The estimates of the variances are then used in a weighted least squares
procedure to form the feasible GLS estimates. The general specification of the OLS
model is as follows:
′
Yi X i δ + εi
(1)
where Y is the households’ WTPAR,1 X represents a vector of the explanatory variables defined in Appendix Table 6, β is a vector of parameters, and ε is a mean-zero
disturbance term that captures idiosyncratic factors that contribute to different WTPAR
levels for households. We assume that the variance of εi is given by:
σεi2 X i θ
(2)
We estimated δ and θ using a three-step FGLS procedure following Amemiya
(1977). First, we estimate Eq. (1) using OLS procedure and predicted the residuals
from Eq. (1) to estimate the following equation using OLS:
∧
(εi )2 X i θ + µi
(3)
The prediction values from this auxiliary regression can be used to transform the
equation, and this transformed equation is estimated using OLS to obtain an asymp∧
totically efficient FGLS estimate, θ which is a consistent estimate of σεi2 (Günther
and Harttgen 2009), the variance of the idiosyncratic component of WTPAR. The
∧
estimates of σ are then used to transform Eq. (1) for our estimation which yields a
εi
consistent and asymptotically efficient estimate of εi (see, for example, Wooldridge
2001).
In our analysis, the dependent variable is derived from the following survey question:
(1) Assume that there are two pricing structures available for domestic water use as
listed below. These pricing structures can be used to reduce water restrictions. In
this case, the amount you will have to pay for water consumption will be higher.
With this in mind, which policy would you prefer?
(A) You would pay a flat price for each unit that you consume
(B) A tiered (block) rate system. That is, you would pay different prices for
different blocks of units of water you consume
Under the flat rate system indicated in (A), a consumer would pay the same price
for each unit consumed, and the average price would equal the marginal price. The
1 The dependent variable Y indicating households’ WTPAR is a censored variable. For example, the
dependent variable for households whose WTPAR equals zero is coded 0. Given such households total less
than 5% of respondents, we excluded the Tobit model from our analysis.
123
Willingness to pay to ensure a continuous water supply…
block tariff referred to in (B), on the other hand, is a volumetric scheme under which
consumers pay different amounts for different consumption levels. This means that the
rate per unit of water increases as the volume of consumption increases. Consumers
thus face a lower rate through the first block of consumption then a higher price up to
the limit of the second block and so on up till the highest consumption block. At the
highest level, consumers can use as much water as they wish, but for each additional
water unit consumed, they pay the highest price in the rate structure.
Respondents selecting (A) are asked to answer sub-question (i), while those choosing (B) are directed to sub-questions (ii), (iii), and (iv):
(i) If you select option (A) how much would you be willing to pay per kilolitre (KL)
of water to ensure continuous water supply with minimum restrictions?
(ii) If you select option (B), how much would you be willing to pay per KL of
water used in tier (block) 1 to ensure continuous water supply with minimum
restrictions? (Currently, up to 62 KL in tier 1 is charged at $0.62 per KL)
(iii) If you select option (B), how much would you be willing to pay per KL of
water used in tier (block) 2 to ensure continuous water supply with minimum
restrictions? (Currently, up to 14 KL in tier 2 is charged at $0.66 per KL)
(iv) If you select option (B), how much would you be willing to pay per KL of
water used in tier (block) 3 to ensure continuous water supply with minimum
restrictions? (Currently, more than 76 KL in tier 3 is charged at $1.17 per KL)
In general, our data analysis reveals three possible way of responding to (i): selecting
the flat price system, choosing the tiered (block) rate system, or selecting neither A
nor B but satisfactorily answering all other survey questions. In the latter case, which
applies to approximately 5% of the respondents, we assume a WTPAR of zero and no
WTP. Sub-questions (ii), (iii), and (iv), on the other hand, reflect a three-tier system
ordered according to an increasing block rate, with each including information on
both consumption limits and current charges, so the respondent can make a rational
judgement. By adding the values for all three tiers together and dividing by three, we
estimate a simple average that represents the respondents’ WTP per KL of water with
no restriction under the block rate system.
Four additional sub-questions—(v) through (viii)—ask respondents to select one
payment schedule from eight bid options that range from “less than $0.50 per 1KL”
to “$4.01 to $5.00 per 1KL”. As these ranges are specific, we use the middle value as
their WTPAR. For example, respondents choosing the first option (less than $0.50 per
1 KL) have a WTPAR value of AUS$0.25, while those selecting the second ($0.51 to
$0.75 per 1 KL) have a WTPAR value of AUS$0.63.
We also ask households whether they would pay a fixed fee that would enable
Brisbane to operate and maintain water services, as well as a fixed state government
bulk water charge, in the form of a quarterly payment, paid for every KL used. We base
these rates on the standard charges in Brisbane between 1 July, 2009, and 30 June, 2010,
which consisted of a local water access charge of AUS$ 38.81 and a state government
bulk water charge of AUS$ 1.22 per 1KL. We derive the dependent variable from
respondents’ WTPAR per KL and use standard socio-economic variables to capture
the factors that influence it. The remaining independent variables, which are listed in
Appendix Table 6 with their expected signs, are either psychological (e.g. PET, EFFE,
123
C. Wilson et al.
REDU and WP) or behavioural/attitudinal (e.g. MENB, PARE, ATI_1, ATI_2,WC_A,
RELI and TRUS).
Because the OLS results for the factors influencing WTPAR are subject to a certain
degree of unequal variance and inter-observational correlation, we apply GLS to obtain
our final estimates. The general model specification can be written as:
Z i∗ β0 + β1 AGE + β2 EDU_1 + β3 EDU_2 + β4 HHS + β5 INC
+ β6 BORN + β7 MENB + β8 PARE + β9 AWARE + β10 ATI_1
+ β11 ATTI_2 + β12 EFFE + β13 PET + β14 REDU + β15 POOL
+ β16 RWT + β17 GAR + β18 WC + β19 WC_ A + β20 WP + β21 RELI
+ β22 TRUS + Ui
where Z i∗ is the dependent variable representing WTPAR, and all other independent
variables are as defined in Appendix Table 6. The significant variables in the model are
likely to provide important insights into the factors that policy makers need to take into
account. Variables other than those for age, income, household size and level of water
consumption are represented by dummies. We hypothesise that most are likely to be
positively correlated with WTPAR. For example, we can expect that households scoring higher on the important explanatory variables of income, household size, average
water consumption and education level are likely to pay more to ensure a continuous
water supply with minimum restrictions. We also include having a swimming pool
and/or rainwater tank(s), as well as garden size, as additional determinants of WTPAR.
To gauge the importance of the above variables, we incorporate psychological variables such as ATTI_1 and ATTI_2 and behavioural variables which include MENB,
PARE, REST, RELI and TRUS (see Table 7).
4 Results and discussion
The average age of respondents from the 1202 households surveyed is 55 with a
lower limit of 20 and an upper limit of 83 years. Male respondents account for 48%
and females for 52%. The average household size is 2.8, with a maximum and minimum of 8 and 1 individual(s), respectively. Respondent education levels are high:
approximately 52% have a degree or postgraduate qualification. The average income
per suburb is approximately AUS$56,503, and around 79% of the respondents were
born in Australia. Interestingly, approximately 67% seemed unaware of the Brisbane’s
water pricing structure. Household water consumption, which averaged around 36 KL
a quarter, showed a high level of variation with a minimum of 1 KL and a maximum
of 725 KL per quarter. Another interesting observation is that approximately 67% of
the respondents have rainwater tanks.
For the survey item measuring WTPAR, we offer a choice of two price schemes:
a flat price per unit consumed and a tiered rate system that prices by blocks of water
consumed. Around 29% of respondents selected the flat price system (A), compared
to 66% that preferred the block rate system (B). Only 5% would not pay to ensure
123
Willingness to pay to ensure a continuous water supply…
Table 1 Estimates of average WTPAR under the two rate systems
Category
Average WTP (per KL)
Fixed charges (per quarter)
Flat price system
0.760
33.042
Tiered (block) rate system
0.858
33.717
Both categories
0.809
33.379
Only 1152 respondents selected one of these options. Of these, 351 (30%) opted for the flat price system,
while 801(70%) chose the tiered rate system
a continuous water supply with minimum restrictions. Hence, a large proportion of
respondents selected either (A) or (B).
Before estimating the proposed models, we needed to understand the nature of
households’ WTPAR. So we calculated the average WTPAR under the two different
price systems. The estimated average WTPAR per additional KL of water and the average WTPAR of any other fixed charges under different pricing schemes are reported in
Table 1. Interestingly, the average WTPAR per additional KL of water is higher among
those who opted for a tiered rate system than among those who selected the flat price
option. It is also higher than the average prices charged by the local government in
Brisbane (or Urban Utilities) under blocks 1 and 2, which at the time of study, were
0.62 and 0.66 per KL, respectively.
Table 2 compares the estimated WTPAR averages with actual water charges under
the two rate systems. In the first two quarters of 2006, Brisbane employed a flat
pricing structure in which each KL of water cost a household AUS$0.89 plus a water
access charge of AUS$27.50 per quarter. Beginning in the third quarter 2006, Brisbane
implemented a tiered pricing water tariff structure with marginal prices of AUS$0.91,
$0.95 and $1.20, respectively, for the first 50 KL, 51–75 KL and more than 75 KL
consumed.
This structure also includes an AUS$0.05 water surcharge for every KL used and
an AUS$28.25 water access charge per quarter. Rate revisions over the last few years,
however, have significantly decreased the water rates in the increasing block rate
system. For example, by 2010, water rates were AUS$0.62, $0.66 and $1.17 for the
first 62 KL, 63–77 KL and over 77 KL consumed, respectively. Water access charges,
on the other hand, increased to AUS$38.81 per quarter, while the state government
bulk water charge rose to AUS$1.22 for every KL used.
As the table clearly shows, whereas the actual price under the flat rate system is
AUS$0.99 per KL, the average WTPAR per KL of households which selected the flat
rate system is AUS$0.76. These households would also be willing to pay AUS$0.63
in state government bulk water charges. The implication of these observations is that
households in this category would be willing to pay AUS$1.39 per KL of water in
a situation without water use restrictions. Under the increasing block rate system,
the Brisbane City Council has been charging AUS$0.91, $0.95 and $1.20 for up to
62 KL, 62–77 KL and above 78 KL, respectively. But the survey results indicate
that households would be willing to pay an additional AUS$0.56, $0.78 and $1.23,
respectively, for the same blocks. Specifically, the additional WTPAR for the water
access charge is AUS$33.02, while the additional WTPAR for the state government
123
C. Wilson et al.
Table 2 A comparison of average WTPAR under the two rate systems
Category
Existing chargesa
Study results
Rate (per KL) Access feeb Bulk water
charges (per
KL)
Average WTP Access feeb Bulk water
(per KL)
charges (per
KL)
Flat price systemc 0.89
27.50
–
0.76
32.41
0.63
38.81
1.22
0.56
33.02
0.69
Tiered rate
Tier 1 (up to 62
KL)
0.62
Tier 2 (63–77 KL) 0.66
0.78
Tier 3 (above 78) 1.17
1.23
Tier rate
(average)
0.86
0.81
a This tier rate was introduced on 1 July, 2009, and remained in place until 30 June, 2010
b The water access fee is a fixed charge payment for each quarter, whereas the state government bulk water
charges apply for each KL of water
c The flat rate figures are for 2005/6. Translated into 2010 prices, they are equivalent to a water rate of
AUS$0.99 per KL and an access fee of AUS$30.89
bulk water charge is AUS$0.69. It is clear that the WTP figures in this study are not
comparable with others general studies in the literature (Willis et al. 2005; Cooper
et al. 2011) as we analysed the WTP under the increasing block rate system while
most other studies were based on lump sum payments for avoiding water restrictions.
Because the Brisbane City Council is the largest council in Australia, both socioeconomic and weather variables are likely to vary between its suburbs. It is therefore
also important to assess inter-suburb differences under the different average WTPAR
categories outlined in Appendix Tables 8 and 9. We therefore calculate the average
WTPAR per KL of water and the fixed charges, including those that enable the water
authority to operate and maintain water services, as well as the state government bulk
water charge. According to our results, the maximum WTPAR is AUS$4.05 per KL,
with a minimum of zero, while the maximum WTPAR for fixed changes per quarter
is AUS$59.72, also with a minimum of zero.
We also estimate the number of households which fall into each of the two WTPAR
categories—those selecting the flat system and those choosing the tiered system (see
Tables 4, 6, respectively). As Table 3 shows, approximately 5% of respondents would
be unwilling to pay to ensure a continuous water supply, implying that the opportunity
cost of water restriction to those households is zero. However, approximately 75, 16
and 3%, agreed they would pay AUS$ 0.01–1.00, 1.01–2.00 and 2.01–3.00 per KL,
respectively, to maintain a continuous water supply. Around 1.5% would be willing
to pay over AUS$3.00 per KL. A similar division of the fixed charge category (see
Table 4) shows that approximately 20% of respondents would not want to pay an
additional amount as a fixed charge although around 50% would agree to pay a fixed
charge of over AUS$40 per quarter, which is higher than the rate at the time of survey
In the next study stage, we regress households’ WTPAR against a range of
the independent variables listed in Annex Table 6. The estimation methods used
123
Willingness to pay to ensure a continuous water supply…
Table 3 Number of households
under different WTPAR ranges
(unit charges)
WTPAR ranges (AUS $)
Number of households
Percentage
0.00–0.00
61
5.07
0.01–1.00
902
75.04
1. 01–2.00
187
15.56
2. 01–3.00
34
2.83
In calculating the average
WTPAR, we do not differentiate
between the flat and tiered rate
systems
3. 01–4.00
10
0.83
4. 01–5.00
8
0.67
1202
100
Table 4 WTPAR for households
in fixed charge category
WTP ranges (AUS$)
In calculating the average
WTPAR, we do not differentiate
between the flat and tiered rate
systems
Total
Number of households
Percentage
0.00–0.00
239
1.00–20.00
43
3.57
21.00–40.00
307
25.52
41.00–60.00
565
46.97
61.00–80.00
40
3.33
9
0.75
1203
100
More than 80.00
Total
19.87
include weighted least squares (which account for heteroskedasticity), generalised
least squares, ordinary least squares with heteroskedastic-consistent errors and truncated regressions (to test the robustness of our results). Table 5 lists the variables
that appear to have a statistically significant effect on WTPAR in both the OLS and
GLS estimations. Each slope coefficient in both regression models is a partial one that
measures the changes in the estimated unit change in a given regressor’s value while
holding other regressors constant. We thus interpret them as the marginal impact of
the right-side variable on the dependent variable.
As the table shows, most variables are significant at acceptable margins, and their
coefficients have the expected signs. Although it is not appropriate to directly compare
OLS and GLS results, it is possible to compare the signs on the coefficients and their
significance levels. Such a comparison reveals that although the estimated coefficients
of many variables are significantly different from zero at standard significance levels in
both sets of results, in the GLS results, most variables also have significantly improving
R2 values. This suggests that the GLS provides better results than the OLS.
In particular, the estimations show that age (AGE) is a significant variable in both
models, although its negative sign indicates that the WTPAR of younger residents is
higher than that of older residents. This implies that WTPAR declines with increasing
age. The second most significant result is for education (EDU), a higher level of which
was expected to be associated with a higher WTPAR. To allow such a comparison, we
used two dummies to divide households into those whose education level is either less
or more than year 9. Both education variables are significant and have the expected
sign.
123
C. Wilson et al.
Table 5 Estimation of WTPAR determinants
Variables
OLS: coefficients
GLS: coefficients
Constant
− 1.461 (3.234)
3.858 (2.789)*
AGE
− 0.013 (0.004)****
− 0.008 (0.003)***
EDU_1
0.347 (0.170)***
0.641 (0.171)****
EDU_2
0.659 (0.161)****
0.807 (0.170)****
HHS
0.072 (0.049)*
0.103 (0.038)****
INC
0.250 (0.301)
− 0.152 (0.258)
BORN
− 0.079 (0.143)
− 0.522 (0.091)****
MENB
0.358 (0.174)***
0.258 (0.142)**
PARE
0.246 (0.096)***
0.091 (0.079)
AWAR
0.578 (0.097)****
0.518 (0.078)****
ATI_1
0.117 (0.124)
0.020 (0.102)
ATI_2
0.223 (0.165)*
− 0.204 (0.136)*
EFFE
0.050 (0.133)
0.097 (0.108)
PET
− 0.311 (0.106)****
0.021 (0.090)
REDU
− 0.275 (0.208)*
− 0.657 (0.142)****
POOL
0.030 (0.114)
− 0.333 (0.091)****
RWT
− 0.050 (0.099)
− 0.196 (0.081)***
GAR
0.274 (0.120)*
0.307 (0.101)*
WC
− 0.003 (0.001)**
0.000 (0.001)
WC_A
− 0.372 (0.116)****
− 0.174 (0.100)**
WP
0.554 (0.089)****
0.286 (0.085)****
RELI
− 0.299 (0.094)****
− 0.460 (0.079)****
TRUS
0.132 (0.099)*
0.011 (0.087)
Number of observations
1202
1157
LR chi2 (22, 1179)
8.03
19.07a
p value
0.00
0.00
R2
0.144
0.271
SE are shown in brackets
*, **, ***, and **** denote significance at the 20%, 10%, 5% and 1% levels, respectively. OLS estimations
are shown with robust standard errors
a F(22, 1134)
Other interesting results include those for household size and income, being born
in Australia, and active membership in an environmental organisation. The significant
results for household size (HHS) clearly indicate that the larger the household, the
higher the WTPAR. The results for household income (INC), in contrast (proxied
by average income for the suburb) are not significant in either model, refuting the
hypothesis that higher income households are more likely to pay more to ensure a
continuous water supply. In our assessment of whether the WTPAR of residents born
in Australia (BORN) differs from that of those born overseas, the only significant
result is the GLS coefficient, which implies that the former is relatively lower than the
123
Willingness to pay to ensure a continuous water supply…
latter. The results for active membership in an environmental organisation (MENB),
in contrast, clearly indicate that these respondents have a higher WTPAR than others,
meaning they are more likely to pay more to ensure a continuous water supply.
In addition to the above variables, we use the PARE dummy to test whether
households with highly water/conservation conscious parents are different from those
without. The OLS result for this variable is significant, implying that parental behaviour
can indeed influence their children’s water valuation decision. Of the two variables
designed to assess whether residents’ attitudes can influence their WTPAR decision
(ATI_1 and ATI_2), interestingly, only the second attitude variable is significant in
either model. This indicates that the WTPAR of residents trying to reduce future water
consumption is higher than that of others. Interestingly, a further important variable
influencing attitudes is religiosity, which is significant in both models. The negative
sign of its coefficient, this implies that the WTPAR among those who define themselves
as religious is less than among those who do not.
In terms of household environs, although ownership of a pool (POOL) or rainwater
tank (RWT) is an important indicator of WTPAR, only the GLS methods produces
significant coefficients for both these variables, both with negative signs. These results
indicate that households which have rainwater tanks are less likely than other households to pay a higher amount for a continuous water supply. This in turn would suggest
that such ownership makes them more secure from water restrictions and/or the lower
WTPAR reflects the fact that they have already made an investment in avoiding restrictions.
The garden variable (GAR), on the other hand, is significant in both models, implying that households with large gardens are more willing than those with smaller gardens
to pay for a continuous water supply. This finding, which was expected, reflects the
enjoyment derived from gardening and the correspondingly higher WTP to avoid
restrictions. The coefficients for the two-water consumption-related variables (WC
and WC_A), in contrast, are only highly significant in the OLS model and negatively
so. This finding implies that the WTPAR of households which consume more water
is relatively lower than that of those which consume less. It may be assumed this is
because residents who do not attempt to conserve water do not want to pay more to
avoid water restrictions.
We gauge household awareness of the existing system of water charges (AWAR)
in the expectation that households with a clear knowledge of existing price structures
will be more likely to pay more to ensure a continuous water supply. This variable is
indeed significant in both models and has the expected sign.
The WP variable tests whether residents’ attitudes to pricing policies affect their
WTPAR. That is, such attitudes relate to water pricing being an important economic
instrument for improving water use efficiency, enhancing social equity and securing
financial sustainability of water utilities and operators. This variable is significant and
has the expected sign in both models indicating that residents who appreciate the
long-term beneficial effects of water pricing have a higher propensity for a WTPAR.
Finally, because econometrics has shown that multicollinearity is a common problem among certain explanatory variables (Greene 2000)—which makes the estimation
of accurate and stable regression coefficients difficult, we derived the collinearity
matrix of the variables used in our models. The condition index of 0.38 indicates a low
123
C. Wilson et al.
level of collinearity (Greene 2000), a level which is unlikely to harm our estimations
given the relatively high t values.
5 Conclusions
As urban cities globally face more frequent residential water supply shortages, an
increasing number of research studies have sought to better understand the policy
instruments that water managers can employ to balance available water supply with
demand. Water shortage issues in the agricultural sector are also of concern and there
is much debate as to how agricultural water, too, can be better managed (see, for
example, Mallawaarachchi et al. 2020). In the case of residential water, one strategy is
to manipulate water demand by encouraging the use of alternative water sources such
as recycled water, rainwater tanks and water saving devices (e.g. low-flow toilets) and
by restricting specific activities that consume excessive quantities of water. To throw
a light on which strategies might be most effective, this study focuses on the factors
that influence the willingness to pay to avoid water restrictions (WTPAR) using survey
data from residents of 90 suburbs in Brisbane, Australia.
When water authorities implement bans or place restrictions on water usage in
outdoor activities such as watering lawns and gardens, washing cars, filling swimming
pools and using sprinklers for recreational purposes, responses vary widely across the
community. It is therefore problematic to evaluate WTPAR based on the types of
anecdotal information presented in the literature, especially given that few previous
studies include measures for behavioural and physiological variables that could also
have an impact. Our study results show that WTPAR is significantly influenced by
a number of socio-economic, behavioural and physiological variables. In particular,
they include behavioural variables (e.g. parent’s behaviour), educational variables (e.g.
awareness about the current system of charges), and attitudes/beliefs (e.g. religiosity).
Similarly, respondents with large families (i.e., large household size) and those with
more education appear to receive significantly differing levels of utility from water
restriction avoidance. WTPAR can also be affected not only by personal circumstances
(e.g. a love of gardening or having a lawn), but also by exogenous factors such as the
severity and duration of water restrictions imposed within the respondent’s city.
The results of this study have numerous policy implications. Most importantly, they
lend strong encouragement for policies which are designed to use a pricing mechanism
to avoid water rationing. Moreover, the wide range of motivations which are shown to
shape residents’ WTPAR provide highly relevant guidelines for government authorities
in developing the specifics of such water supply management policies. For example,
given that respondents who are aware of current water pricing policies and those who
are environmentally active both have a higher WTPAR, the implementation of polices
designed to educate rate payers on these issues would appear to be a logical strategy.
That is, better appreciation of issues surrounding water management, current pricing
policies including environmental concerns, could clearly help raise a community’s
overall WTPAR and therefore receptivity to policies which are designed to avoid
water restrictions.
123
Willingness to pay to ensure a continuous water supply…
The particularly wide mix of significant socio-economic, psychological, attitudinal and behavioural influences could provide useful guidelines for the content and
targeting of marketing and advertising used in educational campaigns. For example,
advertising content may well need to reflect the fact that activities—such as gardening—are more important indicators of a resident’s WTPAR than income levels.
Equally, the higher WTPAR of younger people provides pertinent demographic data
on which to base a communications policy. Useful policy indicators are also provided
by the strong preference shown by residents for an increasing block rate system of
water pricing.
It is also evident from this study that given use of a price mechanism to avoid water
restrictions will generate extra revenue, there is scope for developing parallel water
infrastructure investment policies which will meet changes to water demand generated by such a policy. Thus, the WTPAR values for the preferred increasing block rate
system of charging, provides a valuable indicator of to what extent a politically acceptable pricing regime can fund required changes to the water storage and distribution
infrastructure.
In general, no matter how carefully a sample is selected, some potential participants
do not respond within the time provided (that is, the potential respondent is unwilling
to participate or is impossible to contact). Those who do respond to a mail survey
may therefore be different in their demographic characteristics from those who do
not respond to the survey continuously. It is thus clear that a non-response error can
reduce the precision of the study and may bias results. As we do not compare the
characteristics of respondents and non-respondents, it is identified as one of the main
limitations of this study.
Appendix
See Tables 6, 7, 8 and 9.
123
C. Wilson et al.
Table 6 Independent variables used in the regression models
Code
Definition
Expected sign
AGE
Age (number of years)
±
EDU_1
Education (dummy): 1 if completed grade 10 to 12, 0 otherwise
+
EDU_2
Education (dummy): 1 if completed a degree or postgraduate qualification, 0
otherwise
HHS
Household size (number)
+
INC
Income: average suburb income (AUS$)
+
BORN
Dummy: 1 if born in Australia, 0 otherwise
±
MENB
Dummy: 1 if an active member of environmental organisation, 0 otherwise
+
PARE
Dummy: 1 if parents are very water/conservation conscious, 0 otherwise
+
AWAR
Dummy: 1 if aware about current water charging system, 0 otherwise
+
ATI_1
Dummy: 1 if agreed or strongly agreed that, to save money, they will be more
conscious in the future about the amount of water used, 0 otherwise
+
ATI_2
Dummy: 1 if agreed or strongly agreed that they will make sure their water
usage is reduced as much as possible in the future, 0 otherwise
EFFE
Dummy: 1 if answered affirmatively that water restrictions are an effective
water conservation strategy, 0 otherwise
+
PET
Dummy: 1 if answered affirmatively that water patrols are effective as part of
the strategy to enforce water restrictions on households, 0 otherwise
+
REDU
1 if answered affirmatively that they have reduced their water usage as a
result of the water restrictions, 0 otherwise
+
POOL
1 if reported having a pool on their property, 0 otherwise
+
RWT
1 if reported having a rainwater tank or tanks, 0 otherwise
+
GAR
1 if approximate garden size is 51 m2 or above, 0 otherwise
+
WC
Water consumption level (KL)
+
WC_A
1 if their water consumption in the most recent quarter was above the suburb
average, 0 otherwise
WP
1 if agreed that although higher water prices are usually regarded
unfavourably by the general public, relatively cheap water will not be an
incentive to save water, 0 otherwise
+
RELI
1 if reported being a religious person independent of attendance at a church,
temple or place of worship, 0 otherwise
+
TRUS
1 if agreed that most people can be trusted; 0 if agreed one cannot be too
careful in dealing with people
+
Data for the variables were obtained from the survey except for average income for the suburb, which was
based on 2009 estimations by the Australian Tax Office
123
Willingness to pay to ensure a continuous water supply…
Table 7 Descriptive statistics
Variables
Mean
Maximum
Minimum
SD
WTP
0.81
4.50
0
0.62
AGE
55.15
83
20
13.60
EDU_1
0.19
1
0
0.39
EDU_2
0.76
1
0
0.42
HHS
2.87
8
1
1.34
INC
57,273
101,257
40,670
10,588
BORN
0.78
1
0
0.41
MENB
0.13
1
0
0.34
PARE
0.46
1
0
0.49
AWAR
0.66
1
0
0.47
ATI_1
0.76
1
0
0.42
ATI_2
0.89
1
0
0.31
EFFE
0.80
1
0
0.39
PET
0.19
1
0
0.39
REDU
0.90
1
0
0.29
POOL
0.29
2
0
0.45
RWT
0.67
2
0
0.47
GAR
0.85
1
0
0.35
WC
36.52
725
1
36.57
WC_A
0.25
1
0
0.43
WP
0.54
1
0
0.49
RELI
0.46
1
0
0.49
TRUS
0.72
1
0
0.44
Table 8 Average WTPAR and number of suburbs
Average WTPAR (KL)
Less than 0.1
Number of suburbs
Percentage
4
4.44
0.10–0.59
15
16.67
0.60–1.09
62
68.89
1.10–1.59
7
7.78
1.6 0–2.00
2
2.22
Because respondent number differs in different suburbs, we first calculate the average WTPAR for the
suburb and then estimate the number of suburbs in each category
123
C. Wilson et al.
Table 9 Fixed charges and number of suburbs
Fixed charges (per quarter)
Number of suburbs
Percentage
Less than 10
6
6.67
10.19
6
6.67
20–29
15
16.67
30–39
48
53.33
40–50
15
16.67
Because respondent number differs in different suburbs, we first calculate the average WTPAR for the
suburb and then estimate the number of suburbs in each category
References
Amemiya T (1977) The maximum likelihood estimator and the non-linear three-stage least squares estimator
in the general non-linear simultaneous equation model. Econometrica 45(4):955–968
Blamey R, Gorden J, Chapman R (1999) Choice modelling: assessing the environmental values of water
supply options. Aust J Agric Resour Econ 43(3):337–357
Brennan D, Tapsuwan S, Ingram G (2007) The welfare costs of urban outdoor water restrictions. Aust J
Agric Resour Econ 51:243–261
Carlsson F, Martinsson P (2008) Does it matter when a power outage occurs?-A choice experiment study
on the willingness to pay to avoid power outages. Energy Econ 30:1232–1245
Cooper B, Burton M, Crase L (2011) Urban water restrictions: attitudes and avoidance. Water Resour Manag
47(12):1–13
Dandy G (1992) Assessing the economic cost of restrictions on outdoor water use. Water Resour Res
28(7):1759–1766
Dietz T, Stern PC, Guagnano GA (1998) Social structural and social psychological bases on environmental
concern. Environ Behav 30:450–471
Dworkin J (1973) Global trends in natural disasters 1947–1973. Natural Hazards Research Working Paper
No. 26, Institute of Behavioral Science, University of Colorado, Boulder, Colorado
Gordon J, Chapman R, Blamey R (2001) Assessing the options for the Canberra water supply: an application
of choice modelling. In: Bennett J, Blamey R (eds) The choice modelling approach to environmental
evaluation. Edward Elgar, Cheltenham
Greene WH (2000) Econometric analysis, 4th edn. Prentice Hall, Upper Saddle River
Griffin RC, Mjelde JW (2000) Valuing water supply reliability. Am J Agric Econ 82:414–426
Günther I, Harttgen K (2009) Estimating household’s vulnerability to idiosyncratic and covariate shocks: a
novel method applied in Madagascar. World Dev 37(7):1222–1234
Halkos G, Matsiori S (2018) Environmental attitudes and preferences for coastal zone improvements. Econ
Anal Policy 58:153–166
Hensher D, Shore N, Train K (2005) Households’ willingness to pay for water service attributes. Environ
Resour Econ 32(4):509–531
Hensher D, Shore N, Train K (2006) Water supply security and willingness to pay to avoid drought restrictions. Econ Rec 82(256):56–66
Hensher DA, Shore N, Train K (2014) Willingness to pay for residential electricity supply quality and
reliability. Appl Energy 115:280–292
Howe C, Smith M, Bennett L, Brendecke C, Flack J, Hamm R, Mann R, Rozaklis L, Wunderlich K (1994)
The value of water supply reliability in urban water systems. J Environ Econ Manag 26(1):19–30
Kates RW (1979) The Australian experience: Summary and prospect. In: Heathcote RL, Thom BG (eds)
Natural hazards in Australia. Australian Academy of Science, Canberra, pp 511–520
Kim JH, Lim KK, Yoo SH (2019) Evaluating residential consumers’ willingness to pay to avoid power
outages in South Korea. Sustainability 11(5):1258
Koss P, Khawaja MS (2001) The value of water supply reliability in California: a contingent valuation study.
Water Policy 3:165–174
123
Willingness to pay to ensure a continuous water supply…
Mallawaarachchi T, Auricht C, Loch A, Adamson D, Quiggin Q (2020) Water allocation in Australia’s
Murray-Darling Basin: managing change under heightened uncertainty. Econ Anal Policy 66:345–369
March H, Therond O, Leenhardt D (2012) Water futures: reviewing water-scenario analyses through an
original interpretative framework. Ecol Econ 82:126–137
Martin-Ortega J, Giannoccaro G, Berbel J (2011) Environmental and resource costs under water scarcity
conditions: an estimation in the context of the european water framework directive. Water Resour
Manag 25(6):1615–1633
McFarlane BL, Boxall PC (2003) The role of social psychological and social structural variables in environmental activism: an example of the forest sector. J Environ Psychol 23(1):79–87
Meral GH (1979) Local drought-induced conservation: California experiences. In: Proceedings of the
conference on water conservation: needs and implementing strategies, American Society of Civil
Engineers, New York
Morrissey K, Plater A, Dean M (2018) The cost of electric power outages in the residential sector: a
willingness to pay approach. Appl Energy 212:141–150
Nelson JO (1979) Northern California rationing lessons. In: Proceedings of the conference on water conservation: needs and implementing strategies, American Society of Civil Engineers, New York
Reichl J, Schmidthaler M, Schneider F (2013) The value of supply security: the costs of power outages to
Austrian households, firms and the public sector. Energy Econ 36:256–261
Russell CS, Arey DG, Kates RW (1970) Drought and water supply: implications of the massachusetts
experience for municipal planning. Johns Hopkins Press for Resources for the Future Inc, Baltimore
Tapsuwan S, Brennan D, Ingram G, Michael B (2007) Household willingness to pay to avoid drought water
restrictions: a case study of Perth, Western Australia. http://works.bepress.com/sorada_tapsuwan/3.
Accessed 28 Mar 2013
Willis KG, Scarpa R, Acutt M (2005) Assessing water company customer preferences and willingness to
pay for service improvements: a stated choice analysis. Water Resour Res 41:W02019. https://doi.org/
10.1029/2004WR003277
Woo CK, Ho T, Shiu A, Cheng YS, Horowitz I, Wang J (2014) Residential outage cost estimation: Hong
Kong. Energy Policy 72:204–210
Wooldridge JM (2001) Econometric analysis of cross section and panel data. The MIT Press, Cambridge
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.
123