Advertisement

Energy Efficiency

, Volume 11, Issue 8, pp 2033–2056 | Cite as

Rental tenants’ willingness-to-pay for improved energy efficiency and payback periods for landlords

  • Matthew Collins
  • John CurtisEmail author
Original Article
  • 192 Downloads

Abstract

Throughout the developed world, residential buildings in the rental sector exhibit lower levels of energy efficiency than the owner-occupied building stock. This study estimates Irish rental tenants’ willingness-to-pay for energy efficiency improvements. A double-bounded dichotomous choice contingent valuation method is used to examine how much renters are willing to pay in their monthly rent for improved energy efficiency, measured via energy performance certificates. Using an administrative dataset from a grant scheme for residential energy efficiency retrofits, we examine the upfront cost to landlords of engaging in energy efficiency retrofits and calculate associated payback periods. Tenants in Ireland are willing to pay an average of €38 for a one-grade improvement along a 15-point energy performance certificate scale. Providing additional information about energy performance certificates and the potential impact on energy costs reduced mean willingness-to-pay, implying that in the absence of information tenants overvalued energy efficiency labels. Based on tenants’ willingness-to-pay, investment payback periods for attic and cavity wall insulation are relatively short but prohibitively long for external wall insulation and solar heating retrofits.

Keywords

Willingness-to-pay Rental markets Double-bounded contingent valuation Energy performance certificates 

Introduction

The European Union has set a target of 20% reduction in energy use by 2020 and further reductions thereafter (European Parliament and the Council of the European Union 2012), while the Paris Agreement has emphasised the need to reach peak greenhouse gas emissions as soon as possible (United Nations 2015). One sixth of emissions in the European Union are estimated to occur in residential buildings (European Commission 2011a), while space and water heating account for 67 and 14% of residential energy consumption, respectively (European Commission 2011b). Similarly, 22.5% of energy consumption in the USA occurs in residential buildings (Department of Energy 2012). Two thirds of buildings in the European Union were built before the introduction of energy performance standards, with an average of only 1% of these buildings being renovated each year (European Commission 2016). As such, improving the energy efficiency of the residential building stock provides a significant opportunity to contribute to limiting global warming to below 2 C.

The rental housing market accounts for a large share of residential buildings throughout the world. In the European Union (EU), 29.9% of residential buildings are occupied by tenants, rising to 33.1% in the Euro area and as high as 35.2 and 47.6% in the UK and Germany, respectively (Eurostat 2015). In the USA, 34.6% of occupied housing units are occupied by rental tenants (U.S. Census Bureau 2015). The rental sector, however, is generally less energy efficient than that of owner-occupied homes. In the USA, 70% of rental properties are considered either ‘well insulated’ or ‘adequately insulated’, relative to 84% of owner-occupied properties, while 13% of rental properties possess water heaters more than 20 years old, relative to just 8% of owner-occupied homes (EIA 2013). Similar patterns exist in the EU. In England, for example, private rental accommodation had lower levels of cavity or loft insulation than owner-occupied homes (DfCLG 2013). In Ireland, where the energy performance certificate (EPC) database is publicly available, the distribution of energy labels for rental and other properties can be examined in detail. EPCs in Ireland are termed building energy rating (BER) and Fig. 1 compares the distribution of private rental dwelling BERs with BERs for non-rental properties. Both distributions are heavily skewed toward less efficient grades. While there is slightly greater proportion non-rental properties in the least efficient grades, ‘F’ and ‘G’, overall, the proportion of properties with low-energy efficiency BER assessments, i.e. C3–E2, is relatively higher for rental compared to non-rental properties.
Fig. 1

Proportional distribution of Irish BERs

There is clearly potential to improve the energy efficiency of the rental stock and, as such, it is of interest to understand whether landlords require incentives to do so. As will be discussed in ‘Relevant literature’, improved energy efficiency, in the form of improved EPCs, has been shown to attract a price premium in many rental markets, which in turn provides an incentive to landlords to improve their properties’ energy efficiency. In a market with increasing rental prices, for example, due to supply constraints, both landlords’ and tenants’ decisions may be affected. In such a market, landlords’ rental income increases irrespective of housing quality and some landlords may be less inclined to invest in energy efficiency upgrades. Upward pressure on rental prices will diminish tenants’ willingness-to-pay for improved energy efficiency, as higher rent costs will reduce their ability to pay. Additionally, some housing attributes, including energy efficiency, may have a lower priority compared to attributes such as size or location meaning that willingness-to-pay for energy efficiency may also be affected by rental market conditions. For example, the Irish rental market in recent years has faced a number of market stresses. In the aftermath of the recession, economic recovery has been confined mainly to its cities, leading to a growth in housing demand in these locations. At the same time, rental housing supply contracted, with a 12% reduction in the number of homes available for rent nationwide in 2016 (Lyons 2016). Tighter mortgage regulation in Ireland has forced households from the home purchase to the rental market, while a deficit in social housing has led to a movement of households into the rental market who would otherwise receive social housing. This has lead to an increase in rental prices. Average annual rental growth rates for 2016 were 7.8% (RTB 2016), compared to an average increase in average weekly earnings of 0.95% (CSO 2016). These market forces will necessarily impact on tenants’ ability to pay for energy efficiency but this is an issue that we do not specifically address in this paper. We examine tenants’ willingness-to-pay for energy efficiency using cross-sectional data, which is at one point in time within the rental market. However, within the context of the existing stressed rental market, estimates of tenants’ willingness-to-pay are likely to be conservatively low compared to the situation that might occur in more ‘normal’ rental market where tenant’s budget constraints may be less binding.

This study aims to use survey data to estimate the willingness-to-pay of rental tenants for improved energy efficiency, measured using energy performance certificates, and how this varies across the characteristics of households. We use Ireland as a case study, where high rental demand, particularly in cities, combined with a decreasing supply has created significant upward pressure on rents. We then use a dataset of energy efficiency retrofits to gauge the costs incurred and energy efficiency improvements available through engaging in various retrofitting measures. These are used to calculate guideline payback periods for investments in energy efficiency by landlords.

The remainder of the paper is organised as follows. Section ‘Relevant literature’ outlines related literature. Section ‘Data and methodology’ outlines the data used and methods of analysis. Section ‘Results and discussion’ presents and discusses the results of the analysis. Section ‘Conclusion and policy implications’ then concludes.

Relevant literature

The literature on willingness-to-pay for household energy efficiency is dominated by stated preference techniques. Discrete choice experiments are used to model the preferences of home owners with regard to energy efficiency measures under choice models such as those developed by McFadden (1984). Modelled trade-offs in utility between costs and energy efficiency improvements are then used to estimate the willingness of home owners to pay either for overall measured energy efficiency improvements (Achtnicht 2011) or for specific energy efficiency measures (Cameron 1985; Jaccard and Dennis 2006; Banfi et al. 2008; Kwak et al. 2010). Others have used discrete choice methods to examine revealed preferences, imposing ex-post discrete choice sets comprised of observed retrofits among home owners and other retrofit options foregone for these observed choices (Grösche and Vance 2009).

A wide literature exists examining the observed price premiums for energy efficiency in the purchase sector. Carroll et al. (2016) provided a review of literature on this relationship. There is an expansive literature with regard to energy efficiency price premiums for the sale of homes in various countries, including the USA (Bloom et al. 2011), across EU countries (DG Energy 2013) and more specifically in Germany (Cajias and Piazolo 2013), the Netherlands (Brounen and Kok 2011), England (Fuerst et al. 2015), Wales (Fuerst et al. 2016) and Ireland (Hyland et al. 2013). In the rental sector, however, there exists a narrower but emerging literature on rental price premiums for energy efficiency. In a review of EPCs in selected European countries, DG Energy (2013) found rental price premiums for an increase of one EPC letter grade or equivalent of 4.4% in Austria, 1.5–3.2% in different areas of Belgium and 1.4% in Ireland. Hyland et al. (2013) also found an average increase in rental costs of €5 per month for each EPC grade in Ireland.

Looking specifically toward willingness-to-pay for rental accommodation, discrete choice modelling of stated preference data again dominates the literature. For example, Farsi (2010) undertook a discrete choice experiment and used random effects regressions of various functional forms to analyse risk premia and willingness-to-pay for energy efficiency in rental apartments in Switzerland, finding a willingness-to-pay for various retrofit measures of between 0 and 11.3% of monthly rent. Phillips (2012) also used a discrete choice experiment to elicit willingness-to-pay for energy efficiency ratings of home owners, renters and landlords. Using a nested logit, Phillips found a median willingness-to-pay of $3.23 per week for Home Energy STAR certification in the USA. Galassi and Madlener (2017) used a discrete choice experiment, asking survey respondents to imagine they lived in a cold apartment and presenting a series of discrete choices between retrofitting measures for that cold apartment. Using a mixed effects logit, they found that rental tenants expected a greater disutility from the costs of retrofitting than home owners. As such, renters were less likely to choose greater energy efficiency improvements as these would be more expensive. In Ireland, Carroll et al. (2016) conducted a survey of rental tenants, providing respondents with a discrete choice between differing apartments. Carroll et al. (2016) found, using a mixed logit, that tenants were willing to pay more for EPC improvements than they would be expected to save on energy costs, with willingness-to-pay falling when moving from less efficient to more efficient grades.

We add to the literature on household willingness-to-pay for energy efficiency in the rental sector in three ways. Firstly, we understand that not all households in the sector are willing to pay for improved energy efficiency and attempt to determine whether significant predictors of whether a household is in fact willing to pay for such improvements exist. Secondly, in addition to complementing the literature on the magnitude of the willingness-to-pay, we examine whether the introduction of information regarding EPCs affects either of these issues. Thirdly, we examine the necessity of state aid, in the form of grant aid and/or financing options as a means of improving the energy efficiency of the rental building stock.

Data and methodology

Data collection

To explore rental tenants’ willingness-to-pay for improved energy efficiency data was collected as part of a wider survey of energy-related decision-makers in Ireland. An online survey was designed in three iterations. Firstly, the survey was developed and pre-tested by colleagues, most of which possessed post-graduate degrees in economics or other social sciences. This led to the exclusion or modification of several items. This was followed by a pilot survey to test to a small sample of respondents recruited by a market research firm. The final survey was launched using the panel from an international online consumer panel company with approximately 54,000 panellists across Ireland. This panel is demographically representative of gender, age, region and principal economic status in Ireland. The survey was conducted in November 2016 and elicited 2430 responses. The sample of surveyed households was targeted to be representative of the national population according to the age of the head of household, their principal economic status and gender. Based on a comparison with the Central Statistics Office’s Quarterly National Household Survey (QNHS) Q4 2016, this was largely achieved. The focus of this analysis is a subset of the surveyed sample; respondents living in rental accommodation, renting either from a private landlord, from a local authority or from a voluntary or co-operative housing body. Of the survey sample, 866 respondents lived in rental accommodation. Screening questions were used to first ensure respondents have responsibility for energy-related decisions in the household. Additional screening questions were included in the middle and at the end of the survey to ensure respondent attention and accuracy (Sills and Song 2002; Podsakoff et al. 2003; Bertsch et al. 2017). Respondents with survey response times below the 1st percentile or above the 99th percentile were also excluded. Our final sample comprises 436 households living in rental accommodation. The proportions in different types of rental accommodation differs somewhat from the 2016 Census of Population where 66% of rental accommodation is privately owned, compared to 76% in our sample; 30% is owned by local authorities compared to 20% in our sample, and the balance being voluntary bodies. Average survey response time to the entire survey by the 436 respondents in the sample was 20 min

Survey design

The survey included several modules related to energy efficiency and background characteristics, eliciting information on the characteristics of respondents’ dwellings and their rental tenure. The central element of the questionnaire relevant to this paper is a series of questions within a dichotomous choice contingent valuation methodology (CVM) framework used to estimate tenants’ willingness-to-pay. Each survey respondent is asked two practically identical CVM questions, which are sandwiched between an ‘information’ treatment that explains BERs, including what a BER rating measures, and how much a grade improvement along the BER scale can affect energy costs. A comparison of the two sets of responses enables us to examine how knowledge of energy efficiency, as measured by the BER scale, can affect tenants’ willingness-to-pay for energy efficiency. The good that tenants are asked to value in the CVM questions is a 1-point improvement in energy efficiency on the 15-point BER scale. The CVM question is based on each tenant’s current rental property and an affirmative response to the CVM question secures the tenant an exactly similar property with the same attributes (e.g. size, rooms, location and facilities) except that the BER rating is 1 point better. This design of the CVM question means that we have clearly isolated energy efficiency for valuation and that respondents have full information on the property’s other characteristics, as they currently live in the property.

The CVM component of the survey begins by asking respondents the following:

‘Consider asituation where you are approached with the following proposition regarding your accommodation.

Your landlord offers you asimilar property to your current accommodation, which is the same size, has the same number of rooms, same location in terms of proximity to shops, transport, neighbours, work, college etc. The only difference is that the new accommodation is more energy efficient and therefore has alower combined cost for heating, lighting and ventilation. Energy efficiency is measured on the BER scale with 15 grades from A1 to G, with A1 being the most efficient.

Taking into account your own circumstances, would you be willing to pay more in your monthly rent if the new accommodation was one grade better on the BER scale (e.g. D1 instead of D2 or B3 instead of C1)?’

Respondents answering ‘yes’ to that question then faced adouble-bounded dichotomous choice willingness-to-pay bidding scenario as follows:

‘Would you be willing to pay €XX in increased monthly rent if your landlord engaged in renovations that would improve the BER of the property by ONE GRADE (e.g. from E1 to D2, or from C3 to C2)?’

Respondents were randomly assigned starting bids of €20, €30 or €40. Positive responses led to subsequent bid of €10 greater, while negative responses were followed by a second bid €10 less than the original bid. These bid prices were chosen based on the estimated savings associated with improvements in a dwelling’s BER, as published by the Sustainable Energy Authority of Ireland (SEAI 2014). The bid values were broadly spread across the central distribution of willingness-to-pay from the pilot survey responses.

Respondents that answered ‘no’ to the initial question, i.e. were unwilling to pay for an improvement in energy efficiency, were asked the reasons for their response before facing an ‘information’ treatment. ‘Yes’ respondents also faced the same ‘information’ treatment. The information treatment was designed to succinctly explain the BER rating scheme and what it measured, as well as provide estimates of typical energy cost savings associated with a1-point BER improvement. The information treatment concluded with ascreening question similar to the initial question as follows:

‘A building energy rating (BER) is an indication of the energy performance of ahome. BER is the calculated energy use for space and hot water heating, ventilation and lighting based on standard consumption, in kilowatt hours, for every square metre of adwelling. ABER is similar to the energy label for ahousehold electrical item like your fridge. The label has ascale of A–G. A-rated homes are the most energy efficient and tend to have the lowest energy bills.

In atypical apartment, aone-grade improvement from D1 to C3 could save the occupant approximately €200 in energy costs each year. For alarge, detached house, the same improvement could save €800, based on standard occupancy. The reductions in costs associated with improvements in BER vary depending on the size of ahome and how efficient it is prior to having energy efficient renovations undertaken.

In light of this information, taking into account your own circumstances, would you be willing to pay more in your monthly rent if the new accommodation was one grade better on the BER scale’

Respondents answering in the affirmative faced a further double-bounded dichotomous choice willingness-to-pay bid scenario similar to that described above.

Those who were an unwillingness-to-pay for an improved BER following the treatment were screened to a question regarding reasons for this unwillingness. In addition to an open-ended response, respondents were asked to state their agreement with a number of potential explanations. The distribution of responses to this question is provided in Fig. 2. As can be seen, the most common reason provided for this unwillingness was that respondents did not feel they could afford to pay any more in their monthly rent, with approximately 85% of those unwilling to pay agreeing either somewhat or strongly with the statement. This was followed by those who believed any energy cost savings would be offset by increased rent. The third-most ‘agreed with’ statement was that respondents’ accommodation was already suitably energy efficient. Further reasons cited were that respondents did not want to provide more income to their landlord, while approximately 30% of those unwilling to pay for improved energy efficiency expressed a lack of trust in BERs as a reliable indicator of energy efficiency. We therefore categorise responses to this question as either those of respondents possessing a willingness-to-pay (WTP) of zero or as protest responses. Those who agreed with the former three statements were deemed to possess a WTP of zero. Responses which did not express agreement with these three statements but did agree with one or both of the latter two statements were deemed protest responses. This is because those lacking trust in BERs may be otherwise willing to pay for greater energy efficiency, while those not willing to provide their landlord with greater income may be willing to do so with a more preferable landlord. Responses agreeing with one or both of the latter two statements who also agreed with one of the former three statements were categorised as possessing a WTP of zero. This is because these three statements dominate the latter two. We provide an example of a respondent who does not want to provide their landlord with more income but who also cannot afford more rent. In the case that the respondent was in a similar rental situation but not averse to providing the landlord more revenue, they would remain unable to afford higher rent costs. Further respondents were also classed as protest responses based on open-ended responses, such as those misunderstanding of the question and citing an unwillingness to go through the process of moving home, a misunderstanding of how BER is calculated, and complaints about landlords being unwilling to refurbish their accommodation. Of 436 respondents, removing protest responses left a remaining sample of 415 respondents.
Fig. 2

Reasons given for expressing an unwillingness to pay for improved energy efficiency

Table 1 provides summary statistics of the responses to the double-bounded dichotomous choice questions. After the ‘information’ treatment, a greater number of respondents expressed a willingness-to-pay for improved energy efficiency, although the proportion of those responding negatively to the first bid also increased. Without information, 156, or 38% of respondents, expressed a willingness-to-pay more in their monthly rent for an improved BER. This proportion increased to 55.2%, or 229 respondents, following the provision of information. Figure 3 provides the proportional distribution of respondents present in each bounded willingness-to-pay bracket based on answers provided to the dichotomous choice questions. The introduction of information led to changes in the proportion of households in the €0–30 and €40–50 categories but otherwise has not altered the shape of the distribution. Ordinarily, one would expect the plots in Fig. 3 to be bell-shaped, with low proportions of respondents willing to pay the highest bid amounts. The absence of such responses from the plots indicates that our bid vector, with a maximum value of €50, was insufficient to broadly span the full distribution of WTP. The implication for the analysis is that estimates of WTP are likely to be biased downwards.
Table 1

Responses to double-bounded dichotomous choice questions

Without information

Non-zero willingness-to-pay

No

Yes

Respondents

259

156

(Proportion)

(62.4)

(37.6)

First bid

 

No

Yes

Respondents

 

25

131

(Proportion)

 

(6)

(31.6)

Second bid

 

No

Yes

No

Yes

Respondents

 

14

11

53

78

(Proportion)

 

(3.4)

(2.7)

(12.8)

(18.8)

With information

Non-zero willingness-to-pay

No

Yes

Respondents

186

229

(Proportion)

(44.8)

(55.2)

First bid

 

No

Yes

Respondents

 

53

176

(Proportion)

 

(12.8)

(42.4)

Second bid

 

No

Yes

No

Yes

Respondents

 

29

24

64

112

(Proportion)

 

(7)

(5.8)

(15.4)

(27)

Fig. 3

Bounded willingness-to-pay of respondents to dichotomous choice contingent valuation survey

Other information collected included a range of socio-demographic characteristics of respondents, including their working status, location, age, the number of and age of occupants in the household and whether they are in receipt of specific housing supports1. With regard to the characteristics of their rental tenure, respondents are asked the length and type of their tenure, in addition to the cost of rent. Respondents are also asked whether they know the BER of their accommodation and if so, into which letter-grade category does their accommodation fall. Unknown BERs are estimated according to Curtis et al. (2015). Also collected are questions on energy-related knowledge and on pro-environmental and energy-related behaviours. The answers to these questions are collated to create knowledge and behaviour indices, both of which are standardised about zero. Details of questions asked and the calculation of these behaviour and knowledge indices are provided in Appendices A and B, respectively. Descriptive statistics are provided in Table 2.
Table 2

Descriptive statistics

 

Observations

Proportion

 

Observations

Proportion

Tenure

  

Tenure length

  

Rent from a private landlord

317

0.76

Less than 1 year

76

0.18

Rent from a local authority

85

0.20

1–3 years

136

0.33

Rent from a voluntary/co-operative housing body

13

0.03

3–5 years

83

0.20

 

415

 

5–10 years

61

0.15

Type of accommodation

  

10 + years

59

0.14

Student/shared

13

0.03

 

415

 

Other

402

0.97

Working status

  
 

415

 

Working full-time

167

0.40

Receipt of subsidies

  

Working part-time

74

0.18

Yes

152

0.37

Working the home/carer

46

0.11

No

263

0.63

Unemployed

40

0.10

 

415

 

Retired

29

0.07

BER

  

Student

37

0.09

ABC

161

0.39

Unable to work due to sickness/disability

22

0.05

DEFG

254

0.61

 

415

 
 

415

 

Region

  

Respondent knew BER

  

Dublin City

109

0.26

Yes

137

0.33

Other cities

64

0.15

No

278

0.67

Greater Dublin Area (ex. D)

61

0.15

 

415

 

Rest of Ireland

181

0.44

    

415

 
 

Observations

Mean

Std. dev.

Min

Max

Occupants under 18

415

0.75

1.06

0

5

Occupants aged 19 - 64

415

1.93

0.96

0

7

Occupants aged 65 +

415

0.10

0.37

0

2

Age

415

37.37

12.74

18

78

Energy-related behaviour (scale 0-1)

415

0.62

0.14

0.15

0.91

Energy-related knowledge (scale 0 - 14)

415

5.59

1.75

1

11

Rent per person (€)

4081

320.50

253.84

6

2500

1 Reduced sample size due to non-completion or incorrect completion of question and removal of outiers

Methodology

Possessing a non-zero willingness-to-pay

Assuming non-negative willingness-to-pay for improved energy efficiency and given that only a proportion of respondents expressed a non-zero willingness-to-pay, we are interested in examining whether there are any significant predictors of willingness-to-pay. We therefore specify a selection model of the likelihood of possessing a non-zero willingness-to-pay for improved energy efficiency as follows:
$$ \text{Pr}(\text{WTP}~>~0)~=~\text{Pr}(y_{1i}= 1)=\alpha z_{i}+u_{i1} $$
(1)
where \(y_{1i}\) takes a value of one if respondent i possesses a non-zero willingness-to-pay, \(\alpha \) is a vector of parameters, \(z_{i}\) is a vector of explanatory variables and \(u_{1}\) is the error term. We estimate this model using a standard probit regression. To test for sample selection bias in the outcome model, i.e. the willingness-to-pay equation, we include an explanatory variable which affects the likelihood of possessing a non-zero willingness-to-pay but which does not affect the outcome. We use the region variable for this purpose. The region variable was constructed on the basis of responses to a question about how rental market pressures affected respondents’ choice of rental accommodation. Responses by county were aggregated into regions of market stress. The regions of greatest market pressure are Dublin City and the Greater Dublin Area (GDA), which comprises Dublin County and the counties of Meath, Kildare and Wicklow. Approximately 55% of respondents in these areas indicated that market pressure exerted a major influence on, or took precedence in their accommodation decision, as shown in Fig. 4. Other larger cities, i.e. Cork, Limerick, Galway and Waterford, had moderate market stress levels, followed by a regional category for the rest of Ireland.
Fig. 4

Reported influence of ‘rental market pressures’ on respondents’ choice of rental accommodation, by region

Contingent valuation

We employ the well-established interval-data, double-bounded dichotomous choice contingent valuation model to estimate tenant’s willingness-to-pay for energy efficiency (Alberini 1995; Cameron and Quiggin 1994; Haab 1998; Hanemann et al. 1991). Conditional upon possessing a non-zero willingness-to-pay, the true willingness-to-pay for improved energy efficiency, \(y^{*}\), of respondent i can be expressed as follows:
$$ y^{*}_{2i} = \beta x_{i} + u_{i2} $$
(2)
where \(\beta \) represents a vector of parameters, \(x_{i}\) represents a vector of determinants of willingness-to-pay and the error, \(u_{2}\), is assumed to possess a normal distribution with mean 0 and standard deviation \(\sigma \). This equation forms the basis for estimating the valuation function of each respondent. In order to estimate a valuation function depicting the monetary value of improved energy efficiency, we categorise each respondent possessing a non-zero willingness-to-pay into four binary outcomes. These are based on responses to bid offers presented in the contingent valuation questions and described as follows, with \(B_{1i}\) and \(B_{2i}\) representing the first and second bid presented to respondent i, respectively:
$$\begin{array}{@{}rcl@{}} I^{\text{YY}}_{i} &=& 1(B_{1i}=`yes',B_{2i}=`yes') \\ I^{\text{YN}}_{i} &=& 1(B_{1i}=`yes',B_{2i}=`no') \\ I^{\text{NY}}_{i} &=& 1(B_{1i}=`no',B_{2i}=`yes')\\ I^{\text{NN}}_{i} &=& 1(B_{1i}=`no',B_{2i}=`no') \end{array} $$
(3)
where \(1(.)\) is an indicator function taking a value of one when the argument is true and zero otherwise. Taking this into account, the log-likelihood function of the double-bounded dichotomous choice contingent valuation model is as follows:
$$\begin{array}{@{}rcl@{}} \ln L\!&=&\! {\Sigma}^{\mathrm{N}}_{i = 1} \left\{I^{\text{YY}}_{i}\ln\left[1\,-\,\phi\left( \frac{B_{2i}\,-\,\beta x_{i}}{\sigma}\right)\right]\right.\\ &&\!\left.+I^{\text{YN}}_{i}\ln\left[\phi\left( \frac{B_{2i}\,-\,\beta x_{i}}{\sigma}\right)-\phi\left( \frac{B_{1i}\,-\,\beta x_{i}}{\sigma}\right)\right]\right.\\ &&\!\left.+I^{\text{NY}}_{i}\ln\left[\phi\left( \frac{B_{1i}\,-\,\beta x_{i}}{\sigma}\right)-\phi\left( \frac{B_{2i}\,-\,\beta x_{i}}{\sigma}\right)\right]\right.\\ &&\!\left.+I^{\text{NN}}_{i}\ln\left[\phi\left( \frac{B_{1i}\,-\,\beta x_{i}}{\sigma}\right)\right]\right\} \end{array} $$
(4)
where \(\phi (.)\) is again the standard normal cumulative distribution function. The model is estimated with a user-written Stata®; module, ‘DoubleB’.2

Estimating such a model yields an estimate of the willingness-to-pay of rental tenants for improved BERs. However, our dataset is comprised of a sample of respondents who expressed an explicit unwillingness-to-pay for improved energy efficiency and a sample of respondents who are willing to pay varying amounts for a one-grade BER improvement. As we can only analyse our model based on those who expressed a willingness-to-pay of greater than zero, it is possible that our estimates are subject to selection bias, i.e. estimated levels of willingness-to-pay would likely be greater than those of the population as those possessing a willingness-to-pay of zero are not included in the analysis. In order to test for selection bias, we follow Heckman (1979) in including the inverse Mills ratio, calculated using the estimation of results of the probit model discussed in ‘Possess-ing a non-zero willingness-to-pay’, as an explanatory variable.

If the underlying distribution of willingness to pay changes between the first and second bids, the interval data model above is misspecified and a more general bivariate probit may be preferable (Cameron and Quiggin 1994). Originally, Cameron and Quiggin (1994) suggested that the misspecification ‘vastly’ distorts the estimates. Following correction by Haab (1998) and confirmation by Cameron and Quiggin (1998), the magnitude of distortion is not as serious as initially feared. Alberini (1995) examined the tradeoff between the bias and efficiency associated with the interval-data model compared to a bivariate probit and concluded that ‘the bias of the estimate of the mean/median WTP, if any, is negligible’ and that internal-data estimates (as specified by the model above) are robust to even low values of the coefficient of correlation in the bivariate model. Haab and McConnell (2002, p. 117) provide a further practical reason for using the interval-data model rather than the bivariate probit. In the bivariate probit model, if the parameter estimates differ between the initial bid-price and the follow-up, the researcher is faced with a choice of which to use to calculate WTP. Furthermore, Haab and McConnell (2002, p. 123) note that the interval-data modelling approach yields the greatest increase in efficiency associated with the introduction of the second bid in the contingent value survey plus has the least ambiguity about the recovered preferences compared to other modelling approaches such as the bivariate probit.

Results and discussion

Willingness to pay for energy efficiency

Likelihood of possessing a non-zero willingness-to-pay

We are first interested in identifying characteristics of rental tenants who may be more likely to possess a willingness-to-pay (WTP) of greater than zero. These are tenants which provide an incentive for landlords to improve the energy efficiency of rental properties. In case rental tenants possess a non-zero WTP, landlords could improve energy efficiency and extract greater surplus without reducing consumer welfare by investing in improved energy efficiency. We estimate the selection equation discussed in ‘Possessing a non-zero willingness-to-pay’, results of which are presented in Table 3.
Table 3

Likelihood of possessing a non-zero willingness-to-pay

 

Without information (1)

With information (2)

Tenure (ref = rent from a private landlord)

    

Rent from a local authority

0.295

(0.202)

0.0600

(0.199)

Rent from a voluntary/co-operative housing body

\(-\) 0.452

(0.459)

\(-\) 0.202

(0.388)

Student/shared accommodation

0.0540

(0.377)

0.0575

(0.399)

In receipt of subsidy

\(-\) 0.105

(0.179)

\(-\) 0.330*

(0.180)

Tenure length (ref = less than 1 year)

    

1–3 years

0.517**

(0.203)

0.473**

(0.195)

3–5 years

0.794***

(0.226)

0.702***

(0.222)

5–10 years

0.620**

(0.242)

0.911***

(0.240)

10 + years

0.166

(0.274)

0.601**

(0.269)

Rent per person (€)

\(-\) 0.000464

(0.000330)

\(-\) 0.000584*

(0.000307)

Region (ref = Dublin city)

    

Greater Dublin Area (ex. Dublin city)

\(-\) 0.258

(0.216)

\(-\) 0.266

(0.210)

Other cities

\(-\) 0.177

(0.221)

\(-\) 0.0989

(0.220)

Rest of Ireland

\(-\) 0.313*

(0.178)

\(-\) 0.192

(0.174)

Working status (ref = working full-time)

    

Working part-time

0.363*

(0.197)

0.103

(0.193)

Working in the home/ carer

0.443*

(0.249)

0.297

(0.251)

Unemployed

0.437

(0.270)

0.198

(0.280)

Retired

0.622

(0.464)

0.559

(0.496)

Student

0.641**

(0.253)

0.559**

(0.250)

Unable to work due to illness/disability

0.478

(0.332)

0.469

(0.339)

Age

\(-\) 0.0159*

(0.00832)

\(-\) 0.0255***

(0.00817)

Occupants 18 or under

\(-\) 0.0726

(0.0735)

\(-\) 0.0844

(0.0711)

Occupants aged 19–64

\(-\) 0.0122

(0.0858)

\(-\) 0.0769

(0.0840)

Occupants aged 65 +

0.0780

(0.266)

0.175

(0.286)

BER = DEFG

\(-\) 0.0162

(0.169)

0.243

(0.163)

Knew BER

\(-\) 0.160

(0.177)

0.0893

(0.172)

Behaviour (z)

0.143**

(0.0721)

0.123*

(0.0690)

Knowledge (z)

0.00104

(0.0704)

0.0500

(0.0672)

Constant

\(-\) 0.00562

(0.480)

0.858*

(0.458)

Observations

408

408

Pseudo R2

0.0776

0.0696

Robust standard errors in parentheses (***p < 0.01, **p < 0.05, *p < 0.1). Reduced sample size due to non-complete responses to cost of rent and removal of outliers

Findings can be categorised into four principal areas. These are that tenure length, socio-demographics, region and individual characteristics matter. The results show that the length of time a household has lived in their accommodation has an effect on the likelihood that decision-makers are willing to pay for improved energy efficiency. Those living in their accommodation for between 1 to 3 years, 3 to 5 years and 5 to 10 years are all more likely to possess a non-zero willingness-to-pay than short-term tenancies of less than 1 year and long-term tenancies of greater than 10 years. The differences across these three categories, however, are not statistically different to one another. Other rental characteristics were not found to be significant predictors prior to the treatment.

With regard to market stress, respondents in the Greater Dublin Area and other cities were found to be no more or less likely than those in Dublin City to possess a non-zero WTP. Those in the rest of Ireland, however, are found to be less likely to possess a non-zero WTP. While the exact cause of this is unclear, it is possible that tenants in a tighter rental market, i.e. the cities and commuter belt, are forced to or feel obliged to choose accommodation which does not meet their initial preferences, while those in low-pressure areas are more likely to find accommodation with which they are satisfied.

Age and working status are found to be significant predictors. As people age, they are less likely to possess a non-zero WTP. When age is specified as a categorical variable, this result is largely applicable to the oldest age categories. Respondents working part-time, students and those working the home and carers were all found to be more likely than those working full-time to be willing to pay more in their monthly rent for improved energy efficiency. Those with higher scores in self-reported pro-environmental behaviour were also found to be more likely to possess a non-zero WTP, although those with greater levels of energy-related knowledge were not. The structure of the household, with regard to the number of occupants of varying age categories, was not found to be statistically significant. The BER of properties did not play a significant role in prediction, nor did the indicator variable of whether a respondent knew their dwelling’s BER. When the the imputed BER variable, i.e. ‘building energy rating = DEFG’ is dropped from the regression, there is no substantial change in the estimated parameter. This result may be a reflection of the high growth rates in rental prices, which is shrouding out the importance of other property attributes, such as BER rating.

The introduction of information does not lead to any substantial difference in the model estimates on the likelihood of being willing to pay for improved energy efficiency. The one exception, albeit only marginally, is for respondents with higher rents. The estimated coefficient is now negative and statistically significant, though only at the 10% level. After receiving the ‘information treatment’ those paying higher rents also became less likely to possess a non-zero WTP.

As discussed in ‘Survey design’, a number of options were presented to home owners who expressed an unwillingness-to-pay for improved energy efficiency. The reasons most cited were that respondents could not afford to pay higher rents, that energy cost savings would be offset by increases in rent and that their accommodation was already suitably energy efficient. That 85% of those expressing an unwillingness-to-pay for energy efficiency cited an inability to pay higher rent could be seen as worrying from a standard of living perspective, although rent costs per occupant was not found to be a significant predictor of possessing a non-zero WTP in the no-information condition. Similarly, the model was also estimated using other variations of rental costs, including total rent, rent as a proportion of the mid-point of respondents’ income categories and rent per occupant as a proportion of the mid-point of respondents’ income categories. None of these were found to possess a statistically significant relationship with the likelihood of possessing a non-zero willingness-to-pay. Overall, there is little indication from the estimates in Table 3 that there is a sample selection issue relating to the respondents that completed the CVM questions, i.e. those with non-zero willingness-to-pay. The pseudo R2 for both models is relatively low, while the McFadden’s adjusted R2 is actually negative. For completeness, we continue to include the inverse Mills ratio in the CVM model.

Conditional willingness-to-pay

We next estimated the willingness-to-pay for improved energy efficiency, conditional on possessing a non-zero WTP. Results of the WTP estimation are presented in Table 4. Estimates from our model indicate a mean WTP in our sample of €41.72 with a standard deviation of 8.71. As discussed previously, we include the inverse Mills ratio of the selection model as an explanatory variable in the outcome equation, i.e. the willingness-to-pay model. It is not statistically significant either with or with the ‘information’ treatment. As such, we conclude that sample selection is not an issue in the outcome equation and therefore that the selection and outcome equations are independent.
Table 4

Conditional willingness-to-pay for improved energy efficiency

 

Without information (3)

With information (4)

Tenure (ref = rent from a private landlord)

Rent from a local authority

\(-\) 15.54***

(5.773)

\(-\) 9.138**

(4.016)

Rent from a voluntary/co-operative housing body

\(-\) 46.16***

(15.92)

\(-\) 1.584

(6.890)

Student/shared accommodation

\(-\) 6.521

(9.910)

\(-\) 6.063

(6.672)

In receipt of subsidy

\(-\) 0.474

(4.399)

7.784

(5.685)

Tenure length (ref = less than 1 year)

1–3 years

\(-\)2.152

(8.320)

\(-\)6.648

(8.765)

3–5 years

\(-\)7.422

(11.88)

\(-\)12.60

(12.15)

5–10 years

\(-\)0.399

(10.24)

\(-\)13.32

(13.97)

10 + years

\(-\)0.0119

(8.411)

\(-\)12.68

(11.53)

Rent per person (€)

\(-\) 0.00936

(0.0101)

0.0154

(0.0102)

Working status (ref = working full-time)

Working part-time

\(-\) 2.439

(6.282)

\(-\) 1.315

(4.499)

Working in the home/ carer

\(-\) 6.329

(7.621)

\(-\) 2.526

(6.628)

Unemployed

\(-\) 3.068

(7.872)

0.356

(5.483)

Retired

\(-\) 4.504

(11.78)

\(-\) 1.152

(13.09)

Student

\(-\) 0.781

(9.701)

\(-\) 11.06

(9.020)

Unable to work due to illness/disability

\(-\) 12.56

(7.881)

\(-\) 11.08

(9.200)

Age

0.171

(0.268)

0.276

(0.395)

Occupants 18 or under

1.697

(1.867)

3.726**

(1.847)

Occupants aged 19–64

\(-\) 2.822

(2.285)

0.642

(2.079)

Occupants aged 65 +

\(-\) 3.812

(5.004)

\(-\) 5.490

(6.684)

BER = DEFG

\(-\) 11.36***

(3.610)

\(-\) 9.209*

(4.703)

Knew BER

1.258

(4.288)

\(-\) 6.644*

(3.610)

Behaviour (z)

\(-\) 0.951

(2.350)

0.110

(2.234)

Knowledge (z)

0.690

(1.567)

\(-\) 0.473

(1.523)

Inverse Mills ratio

\(-\) 11.59

(17.45)

\(-\)23.54

(23.79)

Constant

70.93***

(20.82)

56.86***

(15.11)

Sigma (σ)

14.97***

(1.537)

16.74***

(1.433)

Log likelihood

− 170.43

− 263.28

Observations

152

225

Robust standard errors in parentheses (***p < 0.01, **p < 0.05, *p < 0.1). Reduced sample size due to non-complete responses to cost of rent. Willingness-to-pay calculated from above as follows: WTP= β0 + ΣβiXi

We identify significant predictors of household WTP for improved energy efficiency. We find the type of rental tenure to be significant, with those renting from a local authority found to possess a WTP of €15.54 less than those renting from private landlords, increasing to €46.16 for those renting from voluntary or co-operative housing bodies. This is possibly due to budget constraints. These effects diminish with the introduction of information, with the difference in WTP between those renting from private landlords and those renting from local authorities falling to €9.14, while the difference between those renting privately and those renting from housing bodies losing significance. The introduction of information led to an increase in WTP of those with members of the household under the age of 18, with WTP rising by €3.72 for every additional minor.

WTP of respondents living in energy-inefficient homes is €11.36 less than those living in ‘A’-, ‘B’- or ‘C’-rated homes, a difference that falls to €9.21 post ‘information’ treatment. This indicates a longer payback period for investments in poorly rated properties. However, there is generally diminishing marginal returns in efficiency gains as BER rating increase. The improvement in BER rating associated with any retrofit measure is likely to be greater for lower rated properties, i.e. BER improvement is greater for an F-rated than C-rated property. Therefore, the investment cost necessary to achieve a 1-point improvement in BER rating in properties at the lower end of the BER scale is likely to be lower than those higher up the scale. So while the estimates nominally suggest that WTP is lower and therefore the payback period is longer among occupants of lower rated properties compared to others this result may be negated due to diminishing marginal returns on investments in energy efficiency.

Figure 5 shows the comparison of the distributions of conditional willingness-to-pay before and after the treatment. The pane on the left of the figure shows the overall distribution of conditional WTP for all respondents possessing a non-zero WTP either before or after the information treatment, while the pane on the right compares these distributions including only those who expressed a non-zero WTP both before and after the treatment. Both panes present similar patterns, with the introduction of information leading to a convergence in WTP toward a greater peak in the distribution, with a reduction in the size of the tails. It is likely that the information provided gave respondents a reference case for cost reductions available from improving energy efficiency and thus a benchmark level of increased rent respondents would be willing to pay for improved energy efficiency.
Fig. 5

Distribution of predicted willingness-to-pay before and after treatment

As shown in Table 5, the point estimate of conditional WTP for the sample fell by almost €9, although this change is not statistically significant. However, the introduction of information led to an increase in the number of respondents expressing a willingness-to-pay for improved energy efficiency. As shown in Table 1, the number of such respondents possessing a non-zero WTP rose from 156 to 226, an increase of almost 45%. Including those expressing an unwillingness-to-pay for improved energy efficiency, the mean unconditional WTP of the sample increased from €17.45 to €20.77 with the information treatment.3
Table 5

Point estimates of willingness-to-pay for sub-groups of the sample

 

Pre information treatment

Post information treatment

 

WTP point estimate

Standard error

WTP point estimate

Standard error

 

(€)

 

(€)

 

Full sample

46.84

(7.67)

37.66

(2.49)

BER = ABC

53.94

(7.47)

43.51

(2.75)

BER = DEFG

42.58

(8.09)

34.30

(3.67)

Private landlord

51.44

(8.47)

39.64

(2.58)

Non-private landlord

34.41

(6.75)

31.22

(3.90)

Occupants under 18

48.39

(8.53)

41.29

(2.49)

No occupants under 18

45.59

(7.19)

34.96

(3.22)

BER = DEFG, private landlord, no occupants under 18

45.93

(8.34)

33.59

(4.41)

BER = DEFG, non-private landlord, occupants under 183

31.69

(7.89)

31.49

(4.47)

1 Point estimates calculated using estimated parameters of models (3) and (4), holding all characteristics other than those of interest at their mean sample value

2 Calculated using the delta-method (Oehlert 1992)

3 Mean number of occupants under the age of 18 of households comprising one or more occupant under the age of 18

Our estimates of conditional WTP differ to those found by Carroll et al. (2016), who found a progressively increasing WTP when moving from more to less efficient grades. At the lowest grades, where a change in alphanumeric grade corresponds to a change in letter grade, i.e. from ‘G’ to ‘F’ and from ‘F’ to ‘E’, Carroll et al. find a WTP of €82 and €61, respectively. While we estimate WTP for improvements of one alphanumeric grade, e.g. ‘D1’ to ‘C3’, Carroll et al. (2016) consider changes in letter grade. WTP at other grades are therefore not directly comparable. For example, an estimated WTP of €39 to improve from an E to a D may exceed that of certain sub-groups in our population in case this improvement was from a ‘E1’ to a ‘D2’, whereas this would not be the case for an improvement from ‘E2’ to ‘D1’ or from ‘E1’ to ‘D1’. Our estimates are also higher, in turn, than the rental premium associated with energy efficiency in the market estimated by Hyland et al. (2013), who found an average premium of 0.5% of rental costs for each alphanumeric grade. With an average rental price of dwellings of €1005, this equates to a premium of €5 per grade, much lower than the conditional WTP estimated in this study. Given that only a sub-sample of tenants possess a non-zero WTP, similarly, landlords might not all see value in improved BERs with regard to setting prices for rental properties.

Payback period of investment

In order to examine the returns available to landlords from engaging in energy efficiency investments, we calculate a guideline payback period of certain retrofit measures. This is achieved by analysing an administrative dataset comprising the costs of retrofits and measured energy efficiency improvements. The Sustainable Energy Authority of Ireland (SEAI) provides grants to home owners, including landlords, for a range of retrofit measures via the Better Energy Homes scheme. The grant-aided retrofit measures include attic insulation, three types of wall insulation (cavity wall insulation, internal dry-lining or external wall insulation), three types of heating system upgrade (high efficiency gas or oil boiler with heating controls or heating controls upgrade only) and/or solar heating. The dataset includes information on the retrofit measures for which grant aid was applied, the total cost of each measure and the overall energy efficiency improvement as a result of each retrofit.4

We estimate the payback period of investment using the average total cost of a variety of measures and discounted future income based on increasing rental costs at the average conditional willingness-to-pay of two sub-groups, these being those living in A-, B- and C-rated homes and those living in D-, E-, F- and G-rated homes. The analysis is completed using the total cost of retrofitting plus the post information treatment estimates. In addition, we examine how the receipt of grant aid via the Better Energy Homes scheme impacts these payback periods, reducing costs by the applicable level of grant aid currently available under the Better Energy Homes scheme. We choose an annual discount rate of 10%, which is representative of the interest rate offered on personal loans by leading Irish banks for amounts similar to the costs of retrofitting. We also estimated payback periods using a discount rate of 15% but payback periods for measures which were deemed to be affordable to landlords were not sensitive to this change.

Table 6 presents the estimated payback periods of some of the more popular retrofit combinations observed in the Better Energy Homes scheme, conditional on tenants possessing a non-zero willingness-to-pay for improved energy efficiency. The payback period is measured as the number of months required for a landlord to recover their investment. Strikingly, payback periods for external wall insulation and solar heating are virtually infinite. For the purposes of this analysis, we capped payback periods at 300 months, equivalent to 25 years, as investments with greater payback periods are unlikely to be seen as worthwhile. Without grant aid, only apartments or mid-terrace houses possessing a BER between ‘D’ and ‘G’ possess a payback period for external wall insulation of fewer than 10 years, although the provision of grant aid does reduce the payback period below 10 years for all homes rated ‘D’ or worse and A-, B- or C-rated apartments and mid-terrace houses. This is quite striking, given that the majority of rental properties for permanent residence are located in cities, where solid walls are much more common than cavity walls. With high rental prices, landlords are unlikely to be willing to engage in investments possessing a long-term return structure as that may require foregoing rental income while renovations are undertaken. It may also be seen as undesirable to engage in such an investment when rental costs are rising as renovations need not be prioritised as a means of increasing income. In devising policy aimed at improving the energy efficiency of the rental market, the results of this analysis indicate that grant aid may be required for these measures with very long payback periods. This is because without grant aid, these investments would not be profitable for landlords. Low-interest financing for landlords would be unlikely to lead to an increase in installations of external wall insulation or solar heating as these payback periods are not very sensitive to changes in the discount rate.
Table 6

Payback periods of specific energy efficiency retrofit investments for landlords, in months

BER of property:

No Grant Aid

Better Energy Homes grant aid

 

ABC

DEFG

ABC

DEFG

Attic and cavity wall insulation

42

24

25

15

External wall insulation - apartment/mid-terrace house

\(>\)300

103

120

55

External wall insulation - end-of-terrace/semi-detached house

\(>\)300

297

\(>\)300

108

External wall insulation - detached house

\(>\)300

> 300

181

84

High-efficiency boiler with heating controls

49

40

38

31

Heating controls only

40

24

22

14

Solar thermal

\(>\)300

> 300

253

\(>\)300

Attic and cavity wall insulation, high-efficiency boiler with heating controls

81

45

50

30

Attic and cavity wall insulation, heating controls

39

32

18

15

Mean costs and BER grades improved:

 

Costs (€)

BER Grades Improved

 

Total

With Current Grant

ABC

DEFG

Attic and cavity insulation

1537

937

1

2

External wall insulation - apartment/mid-terrace

5285

3035

1

2

External wall insulation - end-of-terrace/semi-D

8337

4937

1

2

External wall insulation - detached house

8656

4156

1

2

Boiler with heating controls

3494

2794

2

3

Heating controls only

1470

870

1

2

Solar thermal

5923

4723

1

1

Attic, cavity, boiler

5157

3557

2

4

Attic, cavity, HC

2880

1380

2

3

 

ABC

DEFG

  

Average WTP (€)

43.51

34.3

  

1 Payback period calculated with annual discount rate of 10%

2 Mean WTP for each sub-group are those presented in table 5 and may vary by the average grade number of grades associated with BER improvements for each retrofit measure

3 The amount of grant aid awarded for external wall insulation varies by dwelling archetype. As such, costs and improvements associated with these archetypes, while WTP is not found to vary across archetypes. Information on the level of grant aid available for retrofit works is available at http://www.seai.ie/Grants/Better_energy_homes/About_the_Scheme/

Each of the other retrofit combinations examined possesses quite short payback periods. Even without grant aid, each of attic and cavity wall insulation, heating system upgrades and combinations thereof possesses payback periods of approximately 4 years or less in homes considered energy inefficient. For homes in the three most efficient letter grades, payback periods do not exceed 7 years, even in the absence of grant aid. On this evidence, grant aid may not seem necessary to induce retrofit activity in rental properties, provided tenants are in fact willing to pay for improved energy efficiency. Were policy-makers to consider state aid schemes to promote retrofitting activities in the rental sector, it may be prudent to introduce either low-interest financing or a nominal level of grant aid for these measures, which may act as a nudging mechanism to raise the awareness of the availability of funding for these works. For those measures discussed above, however, where payback periods are very long or virtually infinite, a more comprehensive system combining direct aid and financing options may be required to encourage investment.

Robustness checks

As discussed in ‘Survey design’, not all respondents knew the BER of their accommodation and, as such, unknown ratings were estimated according to Curtis et al. (2015). It is possible that these are not entirely accurate. As a robustness check, we estimate mean WTP before and after the treatment for both the sample as a whole and including only those respondents with knowledge of the BER of their accommodation. This is done by re-estimating models (3) and (4) for both the full and reduced samples. We compare the models’ point estimates of mean WTP and associated confidence intervals, which are presented in Table 7. While those who know their BER have a much larger estimated WTP pre information treatment, the confidence intervals are large, perhaps owing to the small sample size. Regardless, differences in estimates are not statistically significant.
Table 7

Mean willingness-to-pay for improved energy efficiency for sub-groups of the sample

 

Respondents

Mean WTP (€)

95% confidence intervals

Log likelihood

Before treatment

All respondents

46.84

31.8068

61.8732

\(-\) 170.42976

 

All respondents who knew BER

73

31.79

114.22

\(-\) 33.180335

After treatment

All respondents

37.66

21.8036

53.5164

\(-\) 263.27974

 

All respondents who knew BER

39.99

35.45

44.53

\(-\) 80.50003

1 Point estimates calculated using estimated parameters of models (3) and (4), holding all characteristics other than those of interest at their mean sample value

2 Confidence intervals calculated using delta-method standard errors (Oehlert 1992)

Conclusion and policy implications

Residential retrofits have been identified by policy-makers in Ireland as an opportunity for policy-makers to help meet obligated reductions in energy consumption. The majority of retrofitting work completed through the market-based Better Energy Homes scheme have occurred in owner-occupied homes. Due to upward pressure on rents in Ireland over a number of years, there is less incentive for landlords to improve the energy efficiency of rental properties. We examine, using stated preference data, tenants’ willingness-to-pay for improved energy efficiency. Using an administrative dataset of grant-aided retrofits, we examine whether tenants’ willingness-to-pay is sufficient to encourage landlords to invest in energy efficiency retrofit works.

Conditional upon possessing a non-zero willingness-to-pay, we find that tenants in Ireland are willing to pay an average of €46.84 for each one-grade improvement in their accommodation’s BER. This falls to €37.66 on the introduction of improved information regarding BERs, though this difference is not statistically significant. This reduction in willingness to pay implies that in the absence of information tenants overvalue energy efficiency labels. We find short payback periods for attic and cavity wall insulation and virtually infinite payback periods for external wall insulation and solar thermal panelling. This has significant welfare implications, as investing in those measures with short payback periods can lead to welfare improvements for both landlords and tenants.

Various policy implications can be taken from the findings of this research. Information has been shown to increase the likelihood that tenants are willing to pay increased rent for energy efficiency, with those possessing a non-zero willingness-to-pay also more likely to value non-monetary benefits of engaging in retrofit works, such as improved comfort and health. While informing tenants of these benefits may induce a willingness-to-pay for improved energy efficiency, certain retrofit measures will not provide a return on investment for landlords without support in the form of grant aid. This is true for external wall insulation and solar heating.

Footnotes

  1. 1.

    Subsidies chosen are those which provide eligibility for the Sustainable Energy Authority of Ireland’s Better Energy Warmer Homes scheme, which provides grant aid for households subject to fuel poverty.

  2. 2.
  3. 3.

    Excluding protest responses, mean WTP is calculated as a weighted average of the point estimate for the sample of respondents possessing a willingness-to-pay and 0 for those unwilling to pay for improved energy efficiency

  4. 4.

    This is calculated as the difference between the estimated BER of the property prior to retrofitting and a registered BER assessment following retrofit works. For a more detailed discussion of this process, please see Collins and Curtis (2016)

Notes

Acknowledgments

We acknowledge the Sustainable Energy Authority of Ireland for access to the anonymous dataset of Better Energy Homes scheme applications. This research has been financially supported by the Sustainable Energy Authority of Ireland and ESRI’s Energy Policy Research Centre.

References

  1. Achtnicht, M. (2011). Do environmental benefits matter? Evidence from a choice experiment among house owners in Germany. Ecological Economics, 70(11), 2191–2200.  https://doi.org/10.1016/j.ecolecon.2011.06.026.CrossRefGoogle Scholar
  2. Alberini, A. (1995). Efficiency vs bias of willingness-to-pay estimates: bivariate and interval-data models. Journal of Environmental Economics and Management, 29(2), 169–180.  https://doi.org/10.1006/jeem.1995.1039.CrossRefGoogle Scholar
  3. Banfi, S., Farsi, M., Filippini, M., Jakob, M. (2008). Willingness to pay for energy-saving measures in residential buildings. Energy Economics, 30(2), 503–516.  https://doi.org/10.1016/j.eneco.2006.06.001.CrossRefGoogle Scholar
  4. Bertsch, V., Hyland, M., Mahony, M. (2017). What drives people’s opinions of electricity infrastructure? Empirical evidence from Ireland. Energy Policy, 106, 472–497.  https://doi.org/10.1016/j.enpol.2017.04.008.CrossRefGoogle Scholar
  5. Bloom, B., Nobe, M., Nobe, M. (2011). Valuing green home designs: a study of ENERGY STAR\(^{{\circledR }}\). Journal of Sustainable Real Estate, 3(1), 109–126.  https://doi.org/10.5555/jsre.3.1.a611518768931237.Google Scholar
  6. Brounen, D., & Kok, N. (2011). On the economics of energy labels in the housing market. Journal of Environmental Economics and Management, 62(2), 166–179.  https://doi.org/10.1016/j.jeem.2010.11.006.CrossRefGoogle Scholar
  7. Cajias, M., & Piazolo, D. (2013). Green performs better: energy efficiency and financial return on buildings. Journal of Corporate Real Estate, 15(1), 53–72.  https://doi.org/10.1108/JCRE-12-2012-0031.CrossRefGoogle Scholar
  8. Cameron, T.A. (1985). A nested logit model of energy conservation activity by owners of existing single family dwellings. The review of Economics and Statistics, 205–211.  https://doi.org/10.2307/1924719.
  9. Cameron, T.A., & Quiggin, J. (1994). Estimation using contingent valuation data from a “dichotomous choice with follow-up” questionnaire. Journal of environmental economicsand management, 27(3), 218–234.  https://doi.org/10.1006/jeem.1994.1035.CrossRefGoogle Scholar
  10. Cameron, T.A., & Quiggin, J. (1998). Estimation using contingent valuation data from a “dichotomous choice with follow-up” questionnaire: reply. Journal of EnvironmentalEconomics and Management, 35(2), 195–199.  https://doi.org/10.1006/jeem.1998.1026.CrossRefGoogle Scholar
  11. Carroll, J., Aravena, C., Denny, E. (2016). Low energy efficiency in rental properties: asymmetric information or low willingness-to-pay? Energy Policy, 96, 617–629.  https://doi.org/10.1016/j.enpol.2016.06.019.CrossRefGoogle Scholar
  12. Collins, M., & Curtis, J. (2016). Value for money in energy efficiency retrofits in ireland: grant provider and grant recipients. ESRI Working Paper Series, 542. http://www.esri.ie/pubs/WP542.pdf.
  13. CSO. (2016). Average earnings, hours worked, employment and labour costs by economic sector NACE Rev 2, type of employee quarter and statistic. Technical report, CentralStatistics Office.Google Scholar
  14. Curtis, J., Devitt, N., Whelan, A. (2015). Using census and administrative records to identify the location and occupancy type of energy inefficient residential properties. Sustainable Cities and Society, 18, 56–65.  https://doi.org/10.1016/j.scs.2015.06.001.CrossRefGoogle Scholar
  15. Department of Energy. (2012). Buildings share of U.S. primary energy consumption (percent). Buildings energy data book: Table 1.1.1. Department of Energy, United States of America.Google Scholar
  16. DfCLG. (2013). English housing survey headline report. Department for Communities and Local Government.Google Scholar
  17. DG Energy. (2013). Energy performance certificates in buildings and their impact on transaction prices and rents in selected EU countries. European Commission. https://ec.europa.eu/energy/sites/ener/files/documents/20130619-energy_performance_certificates_in_buildings.pdf.
  18. EIA. (2013). 2009 residential energy consumption survey data. U.S. Energy Information Administration.Google Scholar
  19. European Commission. (2011a). Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions: a roadmap for moving to a competitive low carbon economy in 2050. COM(2011) 112 final. European Union.Google Scholar
  20. European Commission. (2011b). Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions: energy efficiency plan 2011. COM(2011) 109 final. European Union.Google Scholar
  21. European Commission. (2016). Putting energy efficiency first: consuming better, getting cleaner. European Commission MEMO/16/3986.Google Scholar
  22. European Parliament and the Council of the European Union. (2012). Directive 2012/27/EU of the European Parliament and of the Council of 25October 2012 on energy efficiency, amending directives 2009/125/EC and 2010/30/EU and repealing directives 2004/8/EC and 2006/32/EC. European Union.Google Scholar
  23. Farsi, M. (2010). Risk aversion and willingness to pay for energy efficient systems in rental apartments. Energy Policy, 38(6), 3078–3088.  https://doi.org/10.1016/j.enpol.2010.01.048.CrossRefGoogle Scholar
  24. Fuerst, F., McAllister, P., Nanda, A., Wyatt, P. (2015). Does energy efficiency matter to home-buyers? An investigation of EPC ratings and transaction prices in England. Energy Economics, 48, 145–156.  https://doi.org/10.1016/j.eneco.2014.12.012.CrossRefGoogle Scholar
  25. Fuerst, F., McAllister, P., Nanda, A., Wyatt, P. (2016). Energy performance ratings and house prices in Wales: an empirical study. Energy Policy, 92, 20–33.  https://doi.org/10.1016/j.enpol.2016.01.024.CrossRefGoogle Scholar
  26. Galassi, V., & Madlener, R. (2017). The role of environmental concern and comfort expectations in energy retrofit decisions. Ecological Economics, 141, 53–65.  https://doi.org/10.1016/j.ecolecon.2017.05.021.CrossRefGoogle Scholar
  27. Grösche, P., & Vance, C. (2009). Willingness to pay for energy conservation and free-ridership on subsidization: evidence from Germany. The Energy Journal, 135–153. http://www.jstor.org/stable/41323237.
  28. Haab, T.C. (1998). Estimation using contingent valuation data from a “dichotomous choice with follow-up” questionnaire: a comment. Journal of Environmental Economics and Management, 35(2), 190–194.  https://doi.org/10.1006/jeem.1998.1025.CrossRefGoogle Scholar
  29. Haab, T.C., & McConnell, K.E. (2002). Valuing environmental and natural resources: the econometrics of non-market valuation. Cheltenham: Edward Elgar Publishing.CrossRefGoogle Scholar
  30. Hanemann, M., Loomis, J., Kanninen, B. (1991). Statistical efficiency of double-bounded dichotomous choice contingent valuation. American Journal of AgriculturalEconomics, 73(4), 1255–1263. http://www.jstor.org/stable/1242453.CrossRefGoogle Scholar
  31. Heckman, J.J. (1979). Sample selection bias as a specification error. Econometrica, 47(1), 153–161. http://www.jstor.org/stable/1912352.MathSciNetCrossRefGoogle Scholar
  32. Hyland, M., Lyons, R.C., Lyons, S. (2013). The value of domestic building energy efficiency-evidence from Ireland. Energy Economics, 40, 943–952.  https://doi.org/10.1016/j.eneco.2013.07.020.CrossRefGoogle Scholar
  33. Jaccard, M., & Dennis, M. (2006). Estimating home energy decision parameters for a hybrid energy-economy policy model. Environmental Modeling & Assessment, 11(2), 91–100.  https://doi.org/10.1007/s10666-005-9036-0.CrossRefGoogle Scholar
  34. Kwak, S.-Y., Yoo, S.-H., Kwak, S.-J. (2010). Valuing energy-saving measures in residential buildings: a choice experiment study. Energy Policy, 38(1), 673–677.  https://doi.org/10.1016/j.enpol.2009.09.022.CrossRefGoogle Scholar
  35. Lyons, R. (2016). The Daft.ie Rental Report. Daft.ie. https://www.daft.ie/report/2016-q4-rental-daft-report.pdf [accessed 15 May 2017].
  36. McFadden, D.L. (1984). Econometric analysis of qualitative response models. In Arrow, K.J., & Intriligator, M.D. (Eds.) Handbook of econometrics.  https://doi.org/10.1016/S1573-4412(84)02001-8, (Vol. 2 pp. 1395–1457). North-Holland.
  37. Oehlert, G.W. (1992). A note on the delta method. The American Statistician, 46(1), 27–29. http://www.jstor.org/stable/2684406.MathSciNetGoogle Scholar
  38. Phillips, Y. (2012). Landlords versus tenants: information asymmetry and mismatched preferences for home energy efficiency. Energy Policy, 45, 112–121.  https://doi.org/10.1016/j.enpol.2012.01.067.CrossRefGoogle Scholar
  39. Podsakoff, P.M., MacKenzie, S.B., Lee, J.-Y., Podsakoff, N.P. (2003). Common method biases in behavioral research: a critical review of the literature and recommended remedies. Journal of Applied Psychology, 88(5), 879–903.  https://doi.org/10.1037/0021-9010.88.5.879.CrossRefGoogle Scholar
  40. RTB. (2016). The RTB rent index. Residential Tenancies Board. http://www.rtb.ie/docs/default-source/default-document-library/rtb-rent-index-2017-29-march.pdf [accessed 15 May 2017].
  41. SEAI. (2014). A guide to building energy ratings for homeowners. Technical report, Sustainable Energy Authority of Ireland.Google Scholar
  42. Sills, S.J., & Song, C. (2002). Innovations in survey research: an application of web-based surveys. Social Science Computer Review, 20(1), 22–30.  https://doi.org/10.1177/089443930202000103.CrossRefGoogle Scholar
  43. U.S. Census Bureau. (2015). 2011-2015 American community survey 5-year estimates. B25033. United States Census Bureau.Google Scholar

Copyright information

© Springer Science+Business Media B.V., part of Springer Nature 2018

Authors and Affiliations

  1. 1.The Economic and Social Research InstituteDublinIreland
  2. 2.Sustainable Energy Authority of IrelandDublinIreland
  3. 3.Trinity College DublinDublinIreland

Personalised recommendations