Energy efficiency investments in the context of split incentives among French households

Energy Policy 87 (2015) 465–479 Contents lists available at ScienceDirect Energy Policy journal homepage: www.elsevier.com/locate/enpol Energy efﬁc...

Download PDF

700KB Sizes 3 Downloads 67 Views

Report

PDF Reader
Full Text

Energy Policy 87 (2015) 465–479

Contents lists available at ScienceDirect

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

Energy efﬁciency investments in the context of split incentives among French households Dorothée Charlier Université Montpellier 1, Site Richer, Rue Raymond Dugrand, 34960 Montpellier Cedex 2, France

H I G H L I G H T S

I provide empirical evidence of underinvestment due to split incentives. I investigate the inﬂuence of tax credit and energy burden on EE expenditures. Results show that tax credits are ineffective in a context of split incentives. Mandatory measures such as minimum standards seem to be appropriate. Financial support from a third party ﬁnancer can be also a solution.

art ic l e i nf o

a b s t r a c t

Article history: Received 14 April 2015 Received in revised form 3 September 2015 Accepted 4 September 2015 Available online 3 October 2015

The residential sector offers considerable potential for reducing energy use and greenhouse gas (GHG) emissions, particularly through energy-efﬁcient renovations. The objective of this study is twofold. First, I aim to provide initial empirical evidence of the extent to which split incentives between landlords and tenants may lead to underinvestment. Second, I investigate the inﬂuence of tax credits and energy burdens on energy efﬁciency expenditures. Given the complexity of studying the decision to invest in energy-saving renovations, I use a bivariate Tobit model to compare decisions about energy-efﬁcient works and repair works, even when the renovation expenditures seem quite similar. The analysis shows that tenants are doubly penalized: they have high energy expenditures due to energy-inefﬁcient building characteristics, and because they are poorer than homeowners, they are unable to invest in energysaving systems. The results also conﬁrm that tax credits are ineffective in the split incentives context. In terms of public policy, the government should focus on low-income tenants, and mandatory measures such as minimum standards seem appropriate. Financial support from a third-party ﬁnancer also might be a solution. & 2015 Elsevier Ltd. All rights reserved.

Keywords: Energy efﬁciency Split incentives Energy burden Tax credit Public policy

1. Introduction The residential sector offers considerable potential for reducing energy use and greenhouse gas (GHG) emissions, particularly through energy-efﬁcient renovations. Research has often claimed that differing incentives between tenants and landlords of residential housing units lead to inefﬁcient uses of energy (Blumstein, 1980; Brown, 2001; Fisher and Rothkopf, 1989; Jaffe and Stavins, 1994a, 1994b; Sutherland, 1991). Split incentives1 are an E-mail address: [email protected] Split incentives arise when participants in an economic exchange do not share the same goal. When the owner and the occupier of a housing unit are different people, a split in incentives occurs. Whereas landlords want to minimize the purchase cost of energy systems (heating and hot water) and have no return on this investment, tenants want to minimize their energy bill. Therefore, neither party wants to invest in energy-efﬁcient systems. Landlords are not inclined to make investments in energy efﬁciency because tenants are the ones receiving the dividends. 1

http://dx.doi.org/10.1016/j.enpol.2015.09.005 0301-4215/& 2015 Elsevier Ltd. All rights reserved.

important barrier to reducing energy consumption in the residential sector (International Energy Agency, 2007). In 2012, residential buildings made up just over 26.65%2 of ﬁnal energy consumption in the European Union (29%3 in France). Considering that 70.3%4 of all occupied housing units are rental units (36.3% in France), the amount of energy consumption affected by these misaligned incentives might be substantial. However, empirical evidence of the extent of the split incentive issue remains rather limited. Elucidating the split incentives problem has been a challenge for economists. Several studies address the magnitude of the problem and energy efﬁciency issues more broadly (Hassett and 2 3 4

Bertoldi et al. (2012). Ministère de l'Ecologie du Développement Durable et de l'Energie (2012). Eurostat (2012).

466

D. Charlier / Energy Policy 87 (2015) 465–479

Metcalf, 1995; Murtishaw and Sathaye, 2006; van Soest and Bulte, 2001). In general, researchers agree that tenants are reluctant to invest (Arnott et al., 1983; Davis, 2010; Levinson and Niemann, 2004; Rehdanz, 2007). However, empirical literature on the role of split incentives in the decision to invest in energy efﬁciency is sparse. Davis (2010) compares energy-saving system patterns between owner-occupiers and tenants using household-level data. Controlling for household characteristics such as income, tenants are signiﬁcantly less likely to use energy-saving systems. A landlord, free of a dwelling’s energy bill charges, is less likely to invest in energy-saving systems. Gillingham et al. (2012) ﬁnd evidence of split incentives when the occupant does not pay for heating or cooling; households paying for heating are 16% more likely to change their heating system. Gillingham et al. (2012) also show that owner-occupied dwellings are 20% more likely to have insulated attics or ceilings and 13% more likely to have insulated exterior walls. However, these ﬁndings are not directly applicable to France, where all tenants have to pay their energy bills, yet 92% of energy-saving renovations are done in owner-occupied housing units (Sofres-ADEME, 2009). In their analysis, Gillingham et al. neither consider energy burdens (or energy-to-income ratios) as a possible determinant of underinvestment nor measure the impact of public policy on energy efﬁciency expenditures according to occupancy status. This study aims to offer some of the ﬁrst empirical evidence of the extent to which split incentives between landlords and tenants may lead to underinvestment. The emphasis is on analyzing the inﬂuence of tax credits on energy efﬁciency expenditures. I also provide evidence that underinvestment could be related to income (or fuel-poverty5) issues and occupancy status, by investigating whether low-income households with high energy expenditures (i.e., fuel-poor households) invest in energy efﬁciency systems. Shedding light on these determinants can contribute to the improvement of public policies. Many research ﬁelds have addressed occupancy status and housing tenure. Occupancy status is a factor in many household decisions. For example, some authors investigate the effects of housing occupancy on employment and unemployment durations (Battu et al., 2008). Others show that the tenure choice and mobility decisions are correlated (Ioannides, 1987). Home ownership is viewed as one of the crowning achievements in a person’s life cycle. Moreover, home ownership and the capital gains it generates are the primary means of wealth creation for households.6 5 In this article, I assess fuel poverty by measuring the energy burden. The energy burden can be broadly deﬁned as the burden put on a household’s welfare due to the cost of energy expenditures. It is commonly measured as the ratio of energy expenditures to household income (Hills, 2011, 2012; Palmer, 2008). A household bears an energy burden when this energy–income ratio is greater than 10%; that is, the household devotes more than 10% of its income to energy expenditures. 6 Five factors explain how home ownership is a means of wealth creation (see Herbert et al. (2012)). First, the widespread use of amortizing mortgages to ﬁnance the acquisition of the dwelling results in forced savings because a portion of the ﬁnancing cost each month goes toward principal reduction. Second, dwellings are generally assumed to experience some degree of real appreciation over time, reﬂecting increased overall demand for housing due to growth in both population and incomes against a backdrop of a ﬁxed supply of land located near centers of economic activity. Third, a homebuyer with a modest down payment gets the beneﬁt of increases in the overall asset value despite his or her small equity stake. Although a situation of negative leverage can result if the increase in home values is lower than the cost of ﬁnancing (so that the ﬁnancing costs exceed the increase in the asset value), this risk diminishes over time as the value of the home compounds while the debt payment is ﬁxed. However, the latter two arguments are no longer valid in view of the current ﬁnancial crisis. Fourth, income tax beneﬁts from ownership can also be substantial. The ability to deduct mortgage interest and property taxes is the most apparent of these beneﬁts. Fifth, ownership provides a hedge against inﬂation in rents over time (Todd and Souleles, 2005). However, all these arguments are suspended on the time perspective. Indeed, a very long time

Many governments encourage and try to facilitate home ownership through public policies. Occupancy status can also provide insights into a household’s investment decisions. Tenants and landlords have speciﬁc determinants (e.g., income, age, capital access) that explain why their investment decisions may be different. Thus, in terms of energy efﬁciency, it seems pertinent to explore the link between occupancy status and energy efﬁciency investments. Incentives for agents differ depending on whether the housing unit is occupied by a tenant or a homeowner. Consequently, the split incentive issue is particularly relevant to the issue of energy efﬁciency. According to the 2006 Enquête Logement database, 62% of homeowners reporting cold problems in their housing units replaced their equipment, whereas only 32% of tenants experiencing such problems did so. On average, 75% of households that decided to make energy-savings investments were homeowners. Renters are often poorer than homeowners, so the former therefore devote a larger share of their income to energy expenditures, which constitutes the so-called energy burden. According to Boardman (2010), a “household is in fuel poverty if it needs to spend more than 10% of its income on fuel to maintain a satisfactory heating regime and all other energy services.” This deﬁnition also applies to the energy-to-income ratio (De Quero and Lapostolet, 2009) and provides the indicator used by the European Union to measure energy burdens. Moreover, it is often said that low-income households are obliged to “choose” low-cost housing units with many energy efﬁciency problems (e.g., bad insulation, dampness, poor heating systems). These households live in the least energyefﬁcient dwellings and emit more GHG emissions. Such poor quality housing affects social health, incurs cumulative costs, and accelerates housing degradation because of the lack of renovations. Unfortunately, economists have not studied the split incentive issue from an energy burden viewpoint. In addition, the hypothesis that tenants produce more GHG emissions than landlords has yet to be conﬁrmed. To verify these hypotheses, I sought to obtain information about GHG emissions and energy consumption. Data on energy consumption (in kW h/m2/year), energy savings (in euros), and GHG emissions savings (in kg.CO2) were obtained through the PROMODUL software. Thus, PROMODUL served as a tool to provide the data used to feed the model. This approach is one of the original features of this study and constitutes a value added to the literature. The recent rise in energy prices and further expected rises will make it increasingly difﬁcult for tenants to pay their bills (Baxter, 1998). Between 2005 and 2008, 4.2 million principal residences in France received tax credits, equivalent to a total public cost of €7.8 billion. The corresponding cost between 2009 and 2010 was €4.2 billion. Considering the €10 billion devoted to the French Energy Transition bill (June 18, 2014) and the maintaining of tax credits, evaluating the effect of the tax credit scheme on energyefﬁciency investment decisions in the residential sector should be a topmost priority. Not only have few studies examined the effect of tax credits, but the results that exist diverge (Hasset and Metcalf, 1995; Mauroux, 2012; Nauleau, 2014; Pon and Alberini, 2012). Moreover, previous analyses have mainly focused on homeowners and have not considered the split incentive context. Thus, the second objective of my investigation is to determine the effectiveness of tax credits, especially in the context of split incentives. This article presents an empirical analysis of expenditures on different types of energy-saving investments (energy efﬁciency and repair works) according to occupancy status. I found the (footnote continued) horizon is required to produce beneﬁcial effects of home ownership as a means of wealth creation.

D. Charlier / Energy Policy 87 (2015) 465–479

amount spent on each kind of renovation to be similar. However, whereas 25% of households undertook repair works in 2006, only 4% decided to make energy-saving investments. The decision to invest in energy-saving renovation was complicated to analyze for two reasons. First, approximately 71% of households reported no expenditures on renovations, so estimating a linear regression would have induced computational complexities. I therefore applied a Tobit regression (Amemiya, 1973; Heckman, 1979; Tobin, 1958) with left-censored dependent variables (at a zero level). Second, interdependence between the two expenditure types is possible. Thus, I opted to use an econometric model that could take into account both censoring and interdependence: the bivariate Tobit model (Amemiya, 1974; Maddala, 1983). This model extends the single regression model by means of a censored normal dependent variable. I ﬁnd that energy efﬁciency underinvestment is closely linked to energy burden issues, as well as occupancy status. The results also conﬁrm that tax credits are ineffective in inﬂuencing the investment decision in the context of split incentives. In terms of public policy, the government should pay attention to low-income tenants. The remainder of this article is structured as follows. In Section 2.1, I describe the data and variables used in this study. Section 2.2 introduces the econometric models considered. In Section 3, I present the results of my analysis, then discuss them in detail in Section 4. In Section 5, I describe policy recommendations and offer ﬁnal conclusions.

2. Methods 2.1. Data, variables, and descriptive statistics 2.1.1. Data In this study, I used the 2006 Enquête Logement7 (a disaggregated household-level survey) and the Travaux database from INSEE. These are the latest versions publicly available. In these databases, information about households (age, occupancy status, income), dwellings (period of construction, climate zone, surface area), and energy efﬁciency policies targeted at households (tax credits, zero-rate bank loans, subsidies) is available. Merging the two surveys provided information for 22,228 households. Energy efﬁciency renovations are distinguished from repair works. In line with the approach adopted by the Observatoire Permanent de l'amélioration Energétique du logement (OPEN), I consider eight types of energy-saving renovations: double glazing, roof insulation, wall insulation, ﬂoor insulation, mechanical ventilation, new heating system, new hot water system, and chimney. Then repair works include all other types of works (e.g., refurbishment such as painting or expansions). In France, according to Article 7f of the Act of July 6, 1989, a tenant cannot “transform the rented premises and equipment without the written agreement of the landlord.” Incidentally, lease contracts stipulate that tenants are only 7 The choice of the year 2006 is more a constraint than a voluntary decision. First, this is the latest version available of the Enquête Logement. Second, to properly test the effects of the tax credit scheme, it would have been preferable to include the year before the entry into force of the scheme; this would have allowed for an effective comparison of the pre-treatment and post-treatment situations. Thus, applying the econometric analysis to a panel (or pseudo-panel) ranging from at least 2004 to 2007 would be preferable, but the only database available over the period is the ADEME-Sofres Maîtrise de l’Energie surveys from 2001 and 2008, which are conducted by a private ofﬁce. Because that investigation is a property of ADEME, only its staff can use the database. However, given that these data are not available, focusing on 2006 underlines several noteworthy points. First, the problem of split incentives is emphasized. Second, the EPEE (European Fuel Poverty and Energy Efﬁciency) council and the ONPE (Observatoire National de la Précarité Energetique) have based their current analysis on the 2006 Enquête Logement.

467

obliged to maintain the overall quality of the housing.They have to pay for the maintenance ofdoors, windows and light-blocking devices, ceilings, interior walls, ﬂooring and other ﬂoor coverings, electricaland heating installations (e.g., wall sockets, timers, boilers), and ﬁttings (e.g., lime scale removal); they are also required to replace switches or sockets, valveseals, and so on. In contrast, it is the landlord's responsibility to make the necessary repairs to ensure that the rented accommodation is kept in good condition, such as ensuring that electrical systems and plumbing comply with current standards; replacing boilers,water heaters, and hot water tanks; and conducting repairsrelated toconstructiondefectsor faults.However, the landlord has no obligation to undertake renovations that are not deemed necessary. Unfortunately, information on GHG emissions, energy consumption, and energy savings relating to energy efﬁciency renovations was not available in the 2006 Enquête Logement database, even though they are highly pertinent variables in the investment decision. Prior literature has shown that energy savings should be considered a driver of investment (Banﬁ et al., 2008; Grösche and Vance, 2009; Nair et al., 2010). The costs of renovation and the expected gains also emerge as key variables. To verify these hypotheses, I needed information about GHG emissions, energy consumption, and energy savings. Energy consumption (in kW h/m2/year), energy savings (in euros), and GHG emissions savings (in kg.CO2) were obtained with the PROMODUL software by estimating theoretical energy consumption, GHG emissions, and energy expenditures for each dwelling category. PROMODUL can estimate theoretical energy consumption, GHG emissions, and energy expenditures for each category of dwelling using the 3CL method.8 The PROMODUL software is available for residential buildings (multi-family and single-family homes) and allows the user to simulate a diagnosis of energetic performance for each dwelling according to its characteristics (e.g., quality of insulation, type of heating system, localization, type of dwelling). To assess energy expenditures precisely, the housing stock is categorized into different types on the basis of type of dwelling, climate zones (for a map of the French climate zones, see Fig. A-2 in the Appendix), period of construction, type of glazing, type of roof insulation, type of ventilation system, and type of main fuel. The choice of these categories reﬂected my effort to merge the databases (for a summary of the categories, see Table A-1 in Appendix A). The energy consumption, GHG emissions, energy expenditures, energy savings, and GHG emissions savings associated with a renovation were calculated for each type of renovation. Then, for each type of housing unit, I assess the energy expenditures, GHG emissions, and energy consumption after each type of renovation. The energy savings from a speciﬁc investment are calculated as the difference between pre-renovation and postrenovation energy expenditures. Similarly, energy consumption savings were determined as the difference between pre-renovation and post-renovation energy consumption. The same approach is used to compute GHG emissions savings. This procedure is repeated for each category—that is, a total of 2160 times. Appendix A and Table A-2 present an example of the simulation. I merged these new variables with the 2006 Enquête Logement database. The PROMODUL simulation of energy consumption and GHG emissions was applied for 2160 representative dwellings according to the type of dwelling, the quality of insulation, the localization, the type of fuel, and the period of construction. The same information is available in the 2006 Enquête Logement database. Thus, for each dwelling, according to its characteristics, I can determine theoretical energy consumption and theoretical GHG emissions by 8 This computation method, which was described in a French decree in September 2006, is used to establish the dwelling's Energetic Performance Diagnosis.

468

D. Charlier / Energy Policy 87 (2015) 465–479

Table 1 Expenditures (in euros) on repair works and energy-efﬁciency renovations according to occupancy status. Variables

Household characteristics No qualiﬁcations Infbac Bac Bac þ2 Supbac þ 2 Quintile1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 With mobility desires Without mobility desires Building characteristics Before 1974 1974–1981 1982–1989 1990–2001 After 2002 Climate zone 1 Climate zone 2 Climate zone 3 Climate zone 4 Individual housing unit Collective building with individual heating system with collective heating system Electricity Gas Oil Mean

All households

Owner-occupied dwellings

Renter-occupied dwellings

Energy-efﬁciency renovation

Repair work

Energy-efﬁciency renovation

Repair work

Energy-efﬁciency renovation

Repair work

5544 6572 5821 3409 8350 7072 6662 6053 6009 5377 6108 6282

6380 5829 7712 5619 6377 6966 5807 5787 6476 6178 5653 6461

6164 6203 5164 3528 8848 7667 6091 6446 5635 5412 6315 6195

6201 5770 6779 5721 5903 6814 5007 5681 6002 5972 5699 5963

3878 7836 7503 3090 6508 4877 7831 5039 7061 5156 3756 6474

6544 6210 8687 5497 7106 7061 6413 5918 7196 6922 5592 7075

6186 6245 7721 5174 6353 6518 5740 6587 6560 6102 6364 6496 5901 6063 6163 6611 6232

6340 5041 5420 6087 8483 6008 5964 7518 5912 5884 6565 6640 6249 6303 6290 6019 6228

6216 5257 8682 4950 6313 6605 5275 7318 7107 5453 6977 7225 6088 6722 5984 5950 6237

5976 5024 5131 6027 7354 5220 5962 6835 5896 5513 6261 6034 7305 6274 5786 5483 5886

6080 9538 5204 5062 6631 6271 7534 4170 5422 7816 4131 3787 5265 4042 6725 8594 6215

6776 5063 5793 6161 9547 6997 5966 8321 5934 6365 6936 7405 5134 6344 6871 6680 6658

merging information with my PROMODUL simulation. The ﬁnal sample contained 16,111 households. A list of the variables used in the empirical analysis appears in Table A-3 in Appendix A. 2.1.2. Variables and descriptive statistics 2.1.2.1. Energy-saving renovations, home energy rating, and occupancy status. The annual disposable income of French households in 2010 was €43,700 for homeowners, €27,000 for tenants living in private housing, and €22,000 for tenants living in public housing (Commissariat Général du Développement Durable, 2012). Low-income households seem to experience many energy-efﬁciency problems such as bad insulation, dampness, and poor heating systems. The energylabel distribution according to occupancy status (see Fig. A-1 in Appendix A) shows that, in 2006, tenants lived in the most poorly insulated dwellings, while landlords lived in the best-insulated (see Table A-4 in Appendix A). Moreover, considering this distribution according to income (see Table A-5 in Appendix A), the wealthiest households lived in the most energy-efﬁcient and climate-efﬁcient homes. According to the data set, 9.51% of households with an energy income ratio greater than 10% were tenants, and 6.15% were landlords. This is understandable, in that renters are generally poorer than homeowners and live in the most poorly insulated dwellings. Therefore, the distribution of households in terms of dwelling energy efﬁciency is a function of occupancy status. Another noteworthy point is that though tenants who invested in energy efﬁciency renovations faced high energy costs, on the whole, very few tenants actually invested in energy efﬁciency renovations. According to Table 1, similar amounts were invested in repair and energyefﬁciency works. In 2006, only 4.25% of households undertook energy-saving renovations versus 14% for repair works. On average, households spent € 6232 on energysaving works and € 6228 on repair works. Thus, at ﬁrst glance, it appears that similar investments were made in repair and energy efﬁciency works. If a person were to undertake both types of

renovations together, some information9 would be lost. The magnitude of the investment (i.e., proportion of households that invested) is very different between these two renovation types. Therefore, I chose to study the decision to invest in energy-saving systems and repair works separately to determine the reasons for the low rate of energy efﬁciency renovation. A not able result pertains to the importance of occupancy status: 75% of households that decided to make energysavings investments were homeowners, whereas only 56% of repair works were undertaken in owner-occupied dwellings. Two observations can be made about energy efﬁciency investments with respect to occupancy status. First, energy efﬁciency renovations were mainly undertaken in owner-occupied dwellings, regardless of the energy label (Table 2). Although a signiﬁcant proportion of tenants lived in dwellings with low energy label categories (F and G), few of them invested in energy efﬁciency renovations. The opposite is true for homeowners who lived in these categories of dwellings. Second, tenants who invested in energy efﬁciency renovations faced high energy costs, especially those living in the least energyefﬁcient dwellings (i.e., categories F and G). In general, renters were poorer than homeowners and devoted the highest portion of their income to energy expenditures (see Table 3). The ratio of energy expenditures to income among renters was more than 10%, especially among those living in dwelling categories E, F, and G, which indicates that these households were living with an energy burden (or in fuel poverty). In the database, 20% of the households can be considered fuel poor, of which 60% were tenants. Moreover, among poor households, only 22% invested in energy efﬁciency renovations, even though

9 The proportion of repair works is greater than that of energy-efﬁciency works, and some speciﬁc aspects relating to energy-efﬁciency investments could be neglected.

D. Charlier / Energy Policy 87 (2015) 465–479

Table 2 Energy-label distribution of households according to occupancy status. Energy label

A B C D E F G Total

Table 3 Energy expenditures, energy–income ratio, theoretical energy savings and theoretical GHG emissions savings according to energy label and occupancy status.

Households investing in EE renovations (in %)

All households (in %)

Homeowner

Tenant

Mean

Homeowner Tenant Mean

0.00 8.70 31.53 24.37 17.60 12 5.80 100

0.00 3.57 32.15 29.76 20.24 8.33 5.95 100

0.00 7.45 31.68 25.69 18.25 11.09 5.84 100

0.01 8.75 29.23 23.83 12 11.54 6.65 100

Energy Label

0.00 7.25 25.53 27.08 20.17 12.62 7.35 100

469

0.01 8.06 27.52 25.33 20.07 12.04 6.97 100

such renovations were particularly proﬁtable for these households. On average, energy expenditures were higher in dwellings heated by oil, especially those occupied by tenants. Households faced an average bill of €1290 for fuel, €1098 for electricity, and €1024 for gas. This bill reached €1363 for oil in renter-occupied dwellings, versus €1222 in owner-occupied dwellings. There also were more energyefﬁciency investments among dwellings, especially renter-occupied dwellings, if the main fuel was oil. These ﬁndings suggest that tenants are doubly penalized. On the one hand, they have high energy expenditures as a result of building characteristics related to energy efﬁciency; on the other hand, they are poorer than homeowners (see Table A-4 in the Appendix) and, consequently, are unable to invest in energy-saving systems. There is also evidence of underinvestment in collective buildings with collective heating systems. In such buildings, apartments are heated by a collective system, and the energy bills, including costs related to excess energy consumption, are divided among all the building residents. Twenty percent of principal residences in France are included in this category. In such buildings, energy efﬁciency renovations are voted on at owners’ meetings. Forty-four percent of collective buildings with a collective heating system are tenant occupied, which can reinforce the problem of underinvestment in energy-saving systems. A ﬁnal observation pertains to mobility desires. The proportion of households investing in energy efﬁciency systems was very low, which highlights the inﬂuence of the return on investment: Households with mobility desires may not stay long enough in a dwelling to make a potential investment proﬁtable. Thus, the proﬁtability of investment and the potential energy savings should be taken into account in the decision to invest. 2.1.2.2. Energy-efﬁciency expenditures, energy costs, and energy savings. According to prior literature, it is crucial to take energy costs and savings from reduced energy usage into account (Banﬁ et al., 2008; Grösche and Vance, 2009; Nair et al., 2010). In my sample, energy savings were roughlyequivalentamong owner-occupied and renter-occupied dwellings. However, GHG emissions savings were greater for homeowners who invested inmeasures leading to betterenergy conservation (see Table 3). A more detailed analysis reveals that the renovations done most often involved double glazing and heating systems changes (Table 4). These renovations are not necessarily the most beneﬁcial in terms of energy savings and GHG emissions savings. However, double-glazing insulation can provide other advantages, such as noise reduction. Moreover, heating systems change can be an obligation after a breakdown. In France, landlords are responsible for replacing heating and hot water systems, which may explain the high occurrence of this renovation compared with other types of renovations in tenant-occupied dwellings. Unfortunately, GHG emissions savings were higher in category D than in category G, which is the most polluting category.

Homeowners

Tenants

Mean

Ratio of energy expenditure to income (energy–income ratio) in % A 0.44 0.44 B 2.46 4.17 3.17 C 4.72 6.84 5.35 D 7.79 10.05 8.53 E 6.91 11.13 8.59 F 7.98 11.99 9.6 G 7.67 12.95 9.52 Average energy expenditures for households who invested in energy-efﬁciency systems (in euros) A 0 0 0 B 544.57778 391 526.5098 C 823.76205 925.22884 849.01185 D 1230.8333 1412.96 1282.5739 E 1237.1978 1362.7941 1271.36 F 1473.4848 2002.3067 1570.8993 G 1348.1871 1052.9 1274.3653 Theoretical energy savings in euros according to occupancy status (in euros) A 0 0 0 B 38.4 29.5 37.4 C 41.9 39.8 41.4 D 36.7 30.8 35 E 32 37.7 33.6 F 61.4 61.9 61.5 G 50.7 53.8 51.5 Theoretical GHG emissions savings in kg.CO2 according to occupancy status A 0 0 0 B 20.8 10.1 19.5 C 19.29 14.45 18.1 D 25.58 13.03 22 E 21.86 19.75 21.3 F 21.53 23.14 21.8 G 19.02 9.72 16.7

2.1.2.3. Energy efﬁciency expenditure and public policies. Government intervention through public policy may be a particularly effective solution to the split incentive problem. In France, current environmental policies target the residential sector to better exploit its energy-saving potential. In recent years, several measures have been introduced, including tax credits, subsidies, and zero-rate bank loans, to encourage households to undertake energy-efﬁcient renovations (see Table 5). The tax credit scheme allows taxpayers to deduct part of the renovation expense from their income taxes. The deduction rate depends on the equipment (e.g., double glazing, heating system), and the proportion of expenses deducted depends on the number of people in the household. All households are eligible for this policy. A zero-rate bank loan is available to homeowners who carry out several renovations or acquire a given level of energy efﬁciency. It can be combined not only with tax credits and subsidies but also with bonuses. Only owner-occupiers, landlords, and households that have access to credit are eligible for these incentives. Homeowners who perform insulation renovations or install efﬁcient heating systems can also receive a subsidy depending on their household income. This subsidy is only available for owneroccupiers and landlords. Furthermore, none of these measures actually focuses on low-income households living in the least insulated dwellings. These households face difﬁculty gaining access to bank credit and zero-rate bank loans and have no money to undertake renovations. In 2006, 78.8% of households that undertook energy-saving renovations did not receive any public policy incentives (13.43% claimed tax credits, 4% received subsidies, and 3.8% were granted zero-rate bank loans). Only 7.7% of tenants who invested in energy-saving systems claimed tax credits, relative to 15.8% of homeowners (see Table 6). In general, very few tenants beneﬁted from any of the aforementioned public policy schemes.

470

D. Charlier / Energy Policy 87 (2015) 465–479

Table 4 Energy savings and GHG emissions savings in renovated dwellings according to the type of renovation. Double glazing

Wall

Roof

Floor

2.92 3.09 2.38

Chimney

MV

Hot water

Heating system

Distribution All households Homeowners Tenants

42.63 43.71 39.29

11.39 12.38 8.33

9.93 10.25 8.93

Average energy savings from renovation

41.3

47.7

48.7

27.8

45.9

50.6

42.2

33.2

Average GHG emissions savings from renovation

24.3

25.4

27.4

25.6

3.1

30.7

25.9

8.5

Energy-efﬁciency expenditure (in euros by square meter)

93.7

101.9

78.4

19.9

94.3

11.1

220.5

91.9

The subsidy program seems to be the most effective in terms of energy savings but not in terms of GHG emissions savings. However, this result must be interpreted carefully, because of the lack of information and the problem of theoretical GHG emissions estimations.10 A not able observation about zero-rate bank loans can be made: the expenditures and the number of renovations associated with this credit are very high compared with the other policies, but the effectiveness in terms of GHG emissions savings and energy savings is questionable. This result is particularly true for homeowners. It is possible that zero-rate bank loans are effective for tenants but not for homeowners. However, the effectiveness of this measure has not been assessed in prior literature. Finally, the tax credit scheme seems to target homeowners. This measure does not seem very appropriate for inducing investment among low-income households. Moreover, though the number of renovations increased with tax credit claims, savings hardly rose. I carried out a t-test to compare the actual difference between the two means in relation to the variation in the data (expressed as the standard deviation of the difference between the means) and found that the results were not signiﬁcant. Neither the tax credit scheme nor the subsidy program inﬂuenced energy savings. Only the zero-rate bank loan policy had an effect that was strongly correlated with credit access and, therefore, the number of renovations and the expenditures involved. These results suggest that ﬁscal measures, especially tax credit schemes, may not be appropriate to induce investment, especially in cases in which the dwelling is renter occupied.

10 Today, the energy diagnosis performance analysis is the only way to gain information about theoretical energy consumption and theoretical GHG emissions before and after energy efﬁciency renovations. However, it is important to note that there are some problems with the use of these labels. Two elements can explain the gap between theoretical energy consumption and effective energy consumption. First, the main reason given is the existence of the “rebound effect.” The rebound effect refers to a behavioral response to the introduction of new technologies that increase the efﬁciency of a heating system. These responses tend to offset the beneﬁts of the new technology. For example, the households can choose to live at a more comfortable temperature (for an illustration of the French case, see Alibe (2012), Khazzoom (1980) and Sorrell (2007, 2009)). Second, the quality of diagnoses made by professionals who produce the building’s energy performance rating must be called into question. Information is not completely available, and assumptions must be made about the quality of wall insulation, for example. Moreover, there is no control on the professionals who prescribe and install equipment, and they may not have the skills necessary to perform the diagnosis. Thus, we consider that the gap between effective energy consumption and theoretical energy consumption is the result of interactions between the professional actors and the households’ energy usage behavior. Unfortunately, in the absence of available information about households’ behavior in the database, the energy diagnosis performance analysis is currently the only available way to learn about potential energy savings and GHG emissions savings that result from an energy efﬁciency renovation. To my knowledge, there is no other way to assess the potential energy and GHG emissions savings that result from energy-efﬁciency renovations.

5.11 5.22 4.76

1.46 1.74 0.6

3.07 3.87 0.6

23.50 19.73 35. 12

2.2. Empirical analysis of the decision to invest in energy efﬁciency systems according to occupancy status Analyzing the decision to invest in energy-saving renovations proved quite complicated for two reasons. First, approximately 71% of the households in the sample reported no expenditures on renovations, so estimating a linear regression would have induced computational complexities. I therefore applied a Tobit regression (Amemiya, 1973; Heckman, 1979; Tobin, 1958) with left-censored dependent variables (at a zero level). Assuming that households could underconsume energy goods (i.e., homeowners constrained by the expenditure function), the issue of censoring had to be considered. Second, interdependence was possible between the two expenditure types. A bivariate Tobit model is an econometric model that can take into account both censoring and interdependence (Amemiya, 1974; Maddala, 1983). It extends the single regression model with a censored normal dependent variable. In this model, one may deﬁne yit for each household i as

yi1 = α′i1 + μi1 if yi1 = 0

yi2 = α′i2 + μi2 yi2 = 0

α′i1 + μi1 > 0,

if α′i1 + μi1 ≤ 0 if

(1)

α′i2 + μi2 > 0,

if α′i2 + μi2 ≤ 0

where i ¼1,2,…,n, yi1 and yi2 are the dependent variables, xi is a vector of independent variables, and αi1 and αi2 are the corresponding parameter vectors of unknown coefﬁcients. The error terms (mi1 and mi2) are independent of xi. These disturbances have a joint normal distribution with variances of σ12 and σ22 , where μi1, μi2 : N(0,0, σ12, σ22,ρ12) and the covariance is given by s21,2 ¼ ρ σ12, σ22. Multivariate Tobit estimates of two-equation Tobit models rely on the maximum simulated likelihood.

3. Results Before choosing to work with the multivariate Tobit framework, I compared the results of univariate Tobit with those of a type II Tobit model. In the former model, I made a distinction between the decision to invest and the amount spent, assuming these two decisions to be independent. However, a type II Tobit model requires an exclusion variable to avoid collinearity problems. In other words, the selection equation needs an exogenous variable that is excluded from the outcome equation. Unfortunately, I could not ﬁnd such exclusion variable in the database. Thus, the multivariate Tobit approach seemed to be the best model for my analysis. I proceed in two stages. First, I determine the amount spent according to potential energy savings. Second, I determine the amount

D. Charlier / Energy Policy 87 (2015) 465–479

471

Table 5 Public policies. Measure

Description

Tax credit

A part of the expenses on energy-saving renovations can be 15% for double glazing deducted from the household income tax (or refunded if the household pays no income tax). 25% for roof and wall insulation This concerns only a range of speciﬁc renovations and the expenses deducted are limited to a certain amount, depending on the household characteristics. 25% for modernization of heating system 40% for adoption of renewable energy Bonus 1350 euros

Bonus

Rate and/or amount

Households eligible for the policy All households

Middle-income households, according to their income and the number of persons in the dwelling. Only owner-occupiers. Lump sum Landlords who invest can require tenants to repay a lump 10€ for 1 room, 15€ for 2–3 rooms and Landlords sum. 20€ for 4 or more rooms Zero-rate bank No interest on the amount borrowed. It concerns homeowers Minimum of two energy-efﬁciency re- Only owner-occupiers and landlords. Similar novations or minimum level of energy conditions of allocation to other types of loan who perform several renovations or make a signiﬁcant encredit. efﬁciency. ergy-saving investment. The loan amount depends on the renovation. Subsidy A subsidy for homeowners, depending on household income. 35% of renovation expense Only owner-occupiers and landlords NB: To receive such ﬁnancial aid, a household has to hire a company to carry out renovations. Households who perform renovations on their own are not eligible for subsidies, VAT reductions, tax credits or zero-rate bank loans.

Table 6 Public policies and energy-saving investments according to occupancy status.

All households Number of households Average energy-efﬁciency expenditures in euros Number of energy-efﬁciency renovations Theoretical GHG emissions savings in kg.CO2 Theoretical energy savings in euros Owner-occupied dwellings Number of households Average energy-efﬁciency expenditures in euros Number of energy-efﬁciency renovations Theoretical GHG emissions savings in kg.CO2 Theoretical energy savings in euros Renter-occupied dwellings Number of households Average energy-efﬁciency expenditures in euros Number of energy-efﬁciency renovations Theoretical GHG emissions savings in kg.CO2 Theoretical energy savings in euros

No incentives received

Tax credit

Subsidy

Zero-rate bank loan

539 5286 0.7 19.1 40.9

92 5375 1.8 22.5 45.8

28 5670 0.2 27.6 41.6

26 13,883 7.7 25.2 20.5

392 5761 0.9 20.7 41.9

79 5400 2.1 25.3 44.7

21 6306 0.1 27.0 45.3

25 13,629 8 25.4 19.7

147 3960 0.5 14.6 38.2

13 5227 1.8 16.1 52.9

1 3943 0.2 29.4 30.4

7 19,730 7.7 21.7 41

spent according to the energy–income ratio. Tables 7 and 8 present the results. I used a likelihood ratio test to examine the statistical signiﬁcance of the model according to a null hypothesis whereby all the slope coefﬁcients are zero. The χ² statistics for the estimations indicated the rejection of this null hypothesis. To test for interdependence between the two expenditure types, I incorporated a ttest likelihood ratio to constrain the correlation coefﬁcients of the error terms (ρEE,RW) to zero in the two expenditure equations of the univariate model. The t-value for the estimates of ρEE,RW was signiﬁcant at the 1% level, meaning that each null hypothesis could be rejected. I studied the null hypothesis of the independence of renovation work expenditures by means of a log-likelihood ratio test in which the restricted model forces the off-diagonal elements of the covariance matrix to equal zero. Because the resulting χ2 statistics were statistically signiﬁcant, I rejected this null hypothesis. All these test results reinforced my decision to analyze the data within a multivariate Tobit framework. Moreover, the correlation coefﬁcient was positive, implying that the two types of expenditures were positively correlated.

4. Discussion Among the different kinds of renovations, expenditures on repair works were highest in the newest housing units. It is reasonable to assume that energy-efﬁcient improvement expenditures were not necessary in these housing units, which had to comply with certain thermal regulations and labels in place at the time of construction. For example, the “low-energy building” label was introduced in 2005 to households with energy consumption up to 50 kW hpe/m2/year. Most repair works were undertaken in owner-occupied dwellings. Expenditures were higher in dwellings where gas and oil were the main fuel than in dwellings that predominantly used electricity. In renteroccupied dwellings, energy efﬁciency expenditures were higher when oil was the main fuel than in dwellings that mainly used electricity. Therefore, the decision to invest could be linked to energy costs. This result is reinforced by the effect of energy savings observed among the households. The estimated energy savings are positive and statistically signiﬁcant (i.e., households with high energy expenditures before renovation and low expected expenditures after renovation were more willing to invest in energy-efﬁcient renovations), which is in line with the ﬁndings of Banﬁ et al. (2008), Grösche and Vance (2009), and Nair

472

D. Charlier / Energy Policy 87 (2015) 465–479

Table 7 Results of the bivariate Tobit model with energy savings. All households

Household characteristics Quintile 1 Quintile 2 Quintile 3 Quintile 4 Homeowners Age Agenhomeowners Building characteristics Bef1974

Owner-occupied dwellings

Renter-occupied dwellings

EE

RW

EE

RW

EE

RW

0.176 (0.818) 0.923 (0.820) 0.467 (0.780) 0.645 (0.789) 11.31nnn (1.109) 0.0982nnn (0.0197) 0.160nnn (0.0205)

1.693nnn (0.492) 2.583nnn (0.495) 0.390 (0.461) 0.783n (0.461) 2.048nnn (0.795) 0.00187 (0.0128) 0.0258n (0.0153)

2.791nnn (1.079) 1.480 (0.957) 0.629 (0.870) 1.069 (0.872)

3.141nnn (0.670) 3.720nnn (0.654) 0.997n (0.553) 1.328nn (0.539)

1.408 (1.516) 1.192 (1.582) 0.857 (1.627) 0.348 (1.672)

0.525 (0.853) 0.492 (0.865) 1.344 (0.870) 0.873 (0.894)

0.269nnn (0.0260) 0.263nnn (0.0191)

0.0539nn (0.0263) 0.0669nnn (0.0232)

0.000825 (0.0269)

0.0143 (0.0148)

2.226nnn (0.563) 1.596nn (0.675) 2.372nnn (0.718) 2.101nnn (0.645) 0.0704nnn (0.0128) 0.000139nnn (5.05e 05) 0.148 (0.477) 0.245 (0.461) 1.197nn (0.523) 0.546 (0.367) 0.601 (0.414)

0.390 (1.238) 0.233 (1.464) 0.796 (1.464) 0.323 (1.383) 0.136nnn (0.0255) 0.000227nn (9.36e–05) 1.459 (0.962) 0.725 (0.928) 1.390 (1.098) 1.567n (0.860) 1.557n (0.928)

2.194nnn (0.748) 2.093nn (0.915) 2.512nnn (0.934) 1.902nn (0.841) 0.0640nnn (0.0142) 0.000118nn (5.40e–05) 0.297 (0.611) 0.199 (0.593) 1.068 (0.677) 0.467 (0.471) 0.463 (0.536)

1.524 (1.561) 1.435 (1.948) 1.515 (2.000) 0.155 (1.885) 0.137nnn (0.0292) 0.000280nnn (0.000108) 2.682n (1.493) 1.535 (1.459) 4.183nnn (1.561) 1.613 (1.333) 2.833nn (1.322)

2.402nnn (0.829) 0.979 (0.981) 2.170nn (1.102) 2.429nn (0.980) 0.0876nnn (0.0173) 0.000206nnn (6.76e–05) 0.227 (0.762) 0.525 (0.732) 1.182 (0.822) 0.638 (0.582) 0.738 (0.650)

3.426nnn (0.188) 0.210nnn (0.0229)

4.944nnn (0.259) 0.252nnn (0.0276) 0.0508nnn (0.00715) 3.228 (2.486) 39.99nnn (2.461) 0.348nnn (0.0354) 8673

3.259nnn (0.206) 0.187nnn (0.0232)

7.357nnn (0.600) 0.573nnn (0.0923) 0.0626nnn (0.0110) 72.70nnn (5.746) 49.52nnn (3.188) 0.257nnn (0.0465)

3.782nnn (0.438) 0.267nnn (0.0596)

0.564 (1.017) 1974–1981 0.383 (1.216) 1982–1989 0.201 (1.235) 1990–2001 0.0161 (1.148) Surface 0.134nnn (0.0237) Surface2 0.000230nnn (8.76e 05) Climate1 1.980nn (0.820) Climate2 1.205 (0.799) Climate3 2.469nnn (0.908) Gas 1.505nn (0.734) Oil 1.922nn (0.768) Renovation characteristics and public policy NB 5.379nnn (0.264) NB2 0.300nnn (0.0309) Energy savings 0.0554nnn (0.00610) Tax credit 2.472 (2.404) Constant 50.61nnn (2.157) Rho 0.327nnn (0.0286) Observations 16,111

18.75nnn (1.195)

15.93nnn (1.400)

21.28nnn (1.726)

7438

NB: Robust standard errors are given in brackets. Standard errors are shown in parentheses. n

Signiﬁcant at 10%. Signiﬁcant at 5%. Signiﬁcant at 1%.

nn

nnn

et al. (2010). A €1 increase in potential energy savings led to a 5.54% increase in energy efﬁciency expenditures. This evidence indicates that households are sensitive to potential energy savings. Results for income quintiles 1 and 2 are negative and statistically signiﬁcant at the 1% level for repair works but not for energy efﬁciency works. For example, a household in the ﬁrst quintile spent less on repair works than a household in the ﬁfth quintile. Analyzing the regression by category (renter-occupied vs. owner-occupied households), the income quintile has an impact when the dwelling is owner occupied. This result is not able in terms of public policy. In owneroccupied dwellings, measures to help low-income households could be appropriate. However, in tenant-occupied dwellings, income has no

effect: The lack of investment is mainly related to occupancy status. My analysis demonstrates that, in general, when a housing unit is owner occupied, it signiﬁcantly and positively affects energy-efﬁcient expenditures and repair expenditures. Therefore, occupancy status is a determinant of energy-saving investment. There is a signiﬁcant difference in renovation expenditures between households in rented dwellings and those in owner-occupied ones. These ﬁndings are consistent with those obtained by Arnott et al. (1983), Rehdanz (2007), and Davis (2010). One explanation for why tenants might not invest in such expenditures is that their expected length of occupancy may not be sufﬁcient to render any energy-saving investment proﬁtable. I also estimated the model according to the type of

D. Charlier / Energy Policy 87 (2015) 465–479

473

Table 8 Results of the bivariate Tobit model with energy–income ratio. All households IE Household characteristics and fuel poverty Energy income ratio 0.149nnn (0.0532) Energy income ratio2 0.00131n (0.000760) Homeowners 13.92nnn (1.312) Age 0.110nnn (0.0221) Agenhomeowners 0.196nnn (0.0245) Building characteristics Bef1974 0.202 (1.087) 1974–1981 0.611 (1.299) 1982–1989 1.212 (1.337) 1990–2001 1.215 (1.249) Public policy Tax credit 2.335 (2.534) Constant 37.14nnn (1.612) Rho 0.458nnn (0.0262) Observations 16,111

Owner-occupied dwellings

Renter-occupied dwellings

IR

IE

IR

IE

IR

0.0150 (0.0319) 0.000233 (0.000443) 3.325nnn (0.812) 0.00169 (0.0131) 0.0405nnn (0.0157)

0.0789 (0.0725) 0.00189 (0.00116)

0.0215 (0.0456) 5.68e 05 (0.000657)

0.202nn (0.0786) 0.00136 (0.00101)

0.0406 (0.0454) 0.000400 (0.000609)

0.320nnn (0.0295) 0.331nnn (0.0233)

0.0672nn (0.0266) 0.0915nnn (0.0236)

0.00250 (0.0300)

0.0139 (0.0153)

2.833nnn (0.593) 1.632nn (0.716) 2.571nnn (0.758) 2.737nnn (0.684)

0.739 (1.412) 0.659 (1.658) 0.158 (1.686) 1.387 (1.588)

2.779nnn (0.811) 2.302nn (0.981) 2.760nnn (1.009) 2.518nnn (0.920)

1.299 (1.646) 0.250 (2.076) 2.199 (2.136) 1.723 (2.002)

3.050nnn (0.869) 0.878 (1.045) 2.435nn (1.155) 3.149nnn (1.029)

1.376 (2.638) 25.08nnn (1.869) 0.470nnn (0.0329) 8673

12.98nnn (0.887)

61.14nnn (1.720) 34.57nnn (2.312) 0.433nnn (0.0435) 7438

9.748nnn (1.077)

13.09nnn (1.164)

NB: Robust standard errors are given in brackets. Standard errors are shown in parentheses. n

Signiﬁcant at 10%. Signiﬁcant at 5%. nnn Signiﬁcant at 1%. nn

Table 9 Public policy recommendations. Proposals Mandatory measures to retroﬁt buildings and to improve energy efﬁciency

Third-party investment with energy performance contract and “warm rents”

Beneﬁts

Limitations

Direct or indirect rebound effect (rising en-

Individualization of heating systems and direct metering in collective buildings with collective heating systems

heating system (collective or individual) but did not observe any signiﬁcant differences between renter-occupied and owner-occupied dwellings (see Table B-1 in Appendix B). Nonetheless, I ﬁnd that very few households living in collective buildings with collective heating systems made investments. The only noticeable difference arises due to the impact of occupancy status. The amounts spent on energy efﬁciency renovations were higher in owner-occupied dwellings, but the degree of inﬂuence of occupancy status differed depending on the type of heating system. In owner-occupied dwellings, energy efﬁciency expenditures were higher for individual heating systems. Table 8 shows the results of my analysis related to the energy-toincome ratio. To avoid collinearity problems, I removed many variables

Bonus on the housing market value Bonus on the housing rental value Energy savings Less dependence on rising energy prices Bonus on the housing market value Bonus on the housing rental value Energy savings Less dependence on rising energy prices Relief landlords/tenants of the burden of debt Avoid a rent increase in a short term Provide information on the energy cost Easier for EE decision making

ergy consumption) Expenditures and maintenance cost Indirect costs or disturbance costs Possible rent increase Direct or indirect rebound effect (rising energy consumption) Negotiation between the tenant and the landlord Indirect costs or disturbance costs

Expenditures cost Indirect costs or disturbance costs

describing the building stock. The coefﬁcient of the energy–income ratio is positive and statistically signiﬁcant, but the square of the energy–income ratio is negative and statistically signiﬁcant. This suggests that expenditures initially increase with the energy–income ratio and then decline after a peak. This result shows that households with low income and high energy bills (i.e., fuel-poor households) are not able to ﬁnance energy efﬁciency expenditures. Finally, the results conﬁrm that the tax credit scheme had no effect on the decision to invest. I do not interpret the negative effect of the tax credit among renter-occupied dwellings because of issues pertaining to the relevance of the results given the low number of households that received such credits (only 13 in the database).

474

D. Charlier / Energy Policy 87 (2015) 465–479

5. Conclusion and policy implications The residential sector offers considerable potential for reducing energy use and GHG emissions, particularly through energy-efﬁcient renovations. With residential buildings making up more than 26.65% of ﬁnal energy consumption in the European Union in 2012 and 70.3% of all occupied housing units being rental units, the amount of energy consumption affected by misaligned incentives might reasonably be expected to be substantial. However, empirical evidence related to the extent of split incentive issues is rather limited. Using the 2006 Enquête Logement database, this study aimed to provide some of the ﬁrst empirical evidence of the extent to which split incentives between landlords and tenants might lead to underinvestment, with a particular focus on the role of tax credits and energy burden in energy efﬁciency expenditures. In this sample, 75% of households that decided to make energy-saving investments in 2006 were homeowners. The econometric results show that ﬁnancial measures, especially the tax credit scheme, are ineffective in inducing energy-efﬁcient investments in the context of split incentives. Between 2005 and 2008, this measure cost the French government €7.8 billion. A key measure of the French Energy Transition bill related to energy-efﬁcient renovations is the maintenance of the 30% rate tax credit. I recommend a strong assessment of this measure to avoid free-ridership and inefﬁciency on the part of the public action. I have shown that tenants are doubly penalized: They have high energy expenditures because they live in less energy-efﬁcient housing units and are poorer than homeowners. Thus, they are not able to invest in energy-saving systems. In terms of public policy, the government should address the problem of split incentives and also focus on fuel-poverty issues. Whereas landlords want to minimize the purchase cost of energy systems (heating and hot water) and have no return on this investment, tenants want to minimize their energy bill. Therefore, neither party wants to invest in energy-efﬁcient systems. Landlords are not inclined to make investments in energy efﬁciency because tenants are the ones receiving the dividends. If a landlord decides to invest, a possible consequence is larger tenancy deposits or the transition of investment costs to tenants in the form of higher rental rates. This solution is not viable for lowincome tenants. However, it is possible to develop ﬁnancing schemes adapted to the split incentives context. Energy efﬁciency renovations have a speciﬁc economic proﬁle because, in most cases, the return on investment is ensured through energy savings (non-expenses) and not through an increase in revenues (unlike renewable energy projects, which generate a positive cash ﬂow). Tenants have a strong mobility desire, and it might not be possible for them to make EE investments proﬁtable. This partially explains the difﬁculty for tenants to consider EE investments. A recent proposal is energy efﬁciency ﬁnancing, whereby landlords (borrowers) can obtain ﬁnancial support from a third-party ﬁnancer such as an international bank. The ﬁnancial support occurs in a manner that allows the borrower to repay the lender from the energy savings. For example, a third party can ﬁnance the energy efﬁciency investment and capture the energy savings over the occupancy period. Third-party investment solutions are appealing because they relieve landlords of the burden of debt and avoid a rent increase in the short run. The third-party investor has a claim on the future energy savings and may take the risk of not achieving the expected savings (mainly due to the tenant’s mobility desire). To ensure the repayment of energy savings, an energy performance contract (EPC) can be developed (Bullier and Milin, 2013). Such EPCs have been implemented in the industry for many years and, to a lesser extent, in buildings. An EPC11 is a contractual arrangement in which an energy service company (ESCO) designs and implements an energy retroﬁt 11 For more complete descriptions of EPCs, see Bertoldi et al. (2007), Eurocontract (2008), and http://iet.jrc.ec.europa.eu/energyefﬁciency/european-energyservice-companies/ﬁnancing-options.

with a guaranteed level of energy savings. A speciﬁc status for EPC fees in the rental sector could be implemented (Bullier and Milin, 2011). The energy savings would be used to repay the ESCO’s initial investments (though an EPC can also be ﬁnanced directly by the landlord). The landlord or tenant may beneﬁt from part of the energy savings. After all investments are repaid, the contract ends, and the homeowner and/or tenant beneﬁts from all energy savings. Alternatively, to ensure the repayment of energy savings, “warm rent” might be a solution. “Warm rent” is a system in which energy costs are included in the rent, which enables the landlord to recoup the savings to pay the third-party investor (Bullier and Milin, 2013). This system already exists in countries such as Sweden and could be introduced in other countries (Högberg, 2014). In this case, the landlord is responsible for utility contracts, and the rent includes costs for running utilities (heating and water). The rent negotiations include utility costs based on previous consumption, which are split between the landlord and tenant. In practice, the landlord limits consumption; however, the tenant may have no incentive to limit energy consumption. Thus, the third-party investment with EPC and “warm rent” could be a solution for the rental sector. Low-income tenants who live in less well-insulated dwellings and who are vulnerable to rising energy prices do not have the cash to pay for investments. Soft loans or zero-rate bank loans can make the investment more attractive, though the tenant still bears the costs and risks. A large share of low-income tenants cannot or do not want to increase their debt by investing in an EE renovation. For those tenants, solutions based on a third-party investment may be more appealing. With the French Energy Transition bill, the creation of a guarantee fund for energy renovations would help ﬁnance energy efﬁciency renovations for low-income households. Another possible solution in terms of public policy would be to implement mandatory measures such as minimum standards to improve bad energy label categories. These categories could be improved through minimum standard regulatory measures applicable to landlords. I share the view of Bird and Hernández (2012) that public policy should be aimed at enforcing mandatory minimum standards among landlords to support low-income renters. The obligation to retroﬁt tenant-occupied dwellings was discussed during the Grenelle de l’environnement (Pelletier, 2008, p. 86). For example, one possibility would be to ensure that, with every change in dwelling occupancy, homeowners whose dwellings fall below a certain energy label or energy consumption threshold must undertake measures to upgrade it. According to an assessment by Giraudet et al. (2011), this measure would be particularly effective in the landlord–tenant dilemma. Similar regulatory improvements have been introduced in the Netherlands with some success (Scholten, 2013). Finally, I have mentioned the underinvestment in collective buildings with collective heating systems. In such buildings, apartments have a common heating system for which the energy bills are divided among all the residents, who also share the cost of excess energy consumption. Households have an imperfect knowledge of their energy consumption. In this particular case, a direct metering would be the most accurate and reliable system for measuring the amount of energy that individual households use to heat their homes. Households that are more sensitive to their energy consumption might try to change their behavior. The French Energy Transition bill provides for the deployment of smart meters for gas and electricity to better inform households about their energy use. Moreover, in such buildings, energy efﬁciency renovations are voted on at owners' meetings. Forty-four percent of collective buildings with a collective heating system are tenant occupied, and this can reinforce the problem of underinvestment in energy-saving systems. Yet when the building is mainly tenant occupied, measures are never determined by voting. In this case, in terms of energy efﬁciency, individualization of heating systems enables households to choose their equipment. Table 9 summarizes policy recommendations based on the results of my analysis.

D. Charlier / Energy Policy 87 (2015) 465–479

Acknowledgment

Appendix A

I would especially like to thank the ANR-14-CE05-0008-02, the French National Research Agency, by funding this research.

A. Data and descriptive statistics

475

Appendix A-2 Example simulation using PROMODUL software See Fig. A-1,Fig. A-2,Table A-1,Table A-2,Table A-3,Table A-4, Table A-5.

Fig. A-1. Dwelling energy label in kW hef/m2/year and climate label in kg.CO2.

Fig. A-2. The French climate zone.

Table A-1 Housing stock categories.

Type of fuel Climate zone Periods of construction Glazing Ventilation Roof insulation Type of heating

Individual housing units

Collective buildings

Electricity, gas, oil 4 climates zones (1 is the coldest) 5 periods (before 1974, from 1975, to 1981, from 1982 to 1989, from 1990 to 2001, after 2002) Double glazing or simple glazing Mechanical ventilation or not Good, intermediate, bad

Electricity, gas, oil 4 climates zones (1 is the coldest) 5 periods (before 1974, from 1975, to 1981, from 1982 to 1989, from 1990 to 2001, after 2002) Double glazing or simple glazing Mechanical ventilation or not Good, intermediate, bad Individual: only for one dwelling, or Collective: common for the building 1440 2160

Total number of categories 720 TOTAL

476

D. Charlier / Energy Policy 87 (2015) 465–479

Table A-2 Example of simulation.

Without renovation Improvement Insulation Works (IIW) Double glazing Wall insulation Roof insulation Floor insulation Equipment Replacement Works (ERW) Mechanical ventilation New heating system New hot water system Chimney a

Energy (Kw h/m2/ year)

GHG emissions (kg.CO2)

Expenditures by m2 and year (euros)

747

48

33.8

703 661 622 667

45 42 38 42

32.3a 30.7 29.1 30.9

645

41

30.9

713

46

32.6

740

47

33.6

686

37

31.2

After a double glazing renovation, the average energy expenditure is 32.3 euros per square meter, which amounts to energy savings of 1.5 euros per square meter.

Table A-3 List of variables used in empirical approach. Variable Dependent variables Expenditures in EE Expenditures in RW Independent variables Socioeconomic characteristics of households Income quintile Energy–income ratio Energy income ratio2 Occupancy status

Name

Deﬁnition

Units

EE RW

The amount spent on energy-efﬁciency works The amount spent on repair works.

In euros In euros

Quintile Energy–income ratio

Binary variable for each income quintile (5 quintiles) Continuous variable. The ratio of income to energy expenditures in % Energy–income ratio2 Continuous variable. Square of energy expenditures income ratio. Homeowners Binary variable introduced for homeowners Tenants Binary variable introduced for tenants Age Continuous variable of age WDesireMob When the household has mobility desires WoutDesireMob When the household has no mobility desires Binary variable introduced for each degree level (5 modalities) Ref Households with no qualiﬁcations Infbac Households with qualiﬁcations lower than baccalaureate level Bac Households with baccalaureate level Bac þ2 Households with two years of post-baccalaureate qualiﬁcations

Age With mobility desires Without mobility desires Degree level No qualiﬁcations Lower than baccalaureate Baccalaureate Two years of higher studies after baccalaureate More than two years of higher studies after Supbacþ 2 baccalaureate Characteristics of buildings Period of construction Before 1974 Bef1974 1974–1981 1974–1981 1982–1989 1982–1989 1990–2001 1990–2001 After 2002 Before 2002-Ref Surface area Surface Square of surface area Surface2 Climate zone Climate zone 1 Climate1 Climate zone 2 Climate2 Climate zone 3 Climate3 Climate zone 4 Climate4 -ref Individual housing unit Indhousing Collective building Collbuilding Individual heating system Indheating Collective heating system Collheating Main Fuel Gas Oil Electricity Characteristics of renovation Energy savings Number of renovations

Public policy Tax credit

Energy savings NB NB2

Tax credit

0/1 Continuous (between 0 and 100) Continuous 0/1 0/1 0/1 0/1 0/1 0/1 0/1 0/1

Households with more than two years of post-baccalaureate qualiﬁcations

0/1

Binary variables are introduced for each period of construction Dwelling constructed before 1974 Dwelling constructed between 1974 and 1981 Dwelling constructed between 1982 and 1989 Dwelling constructed between 1990 and 2001 Dwelling constructed after 2002 Average surface area per dwelling in 2006 Square of average surface area per dwelling in 2006 Binary variable for each climate zone (4 zones)–see Fig. A-2 Households in climate zone 1 Households in climate zone 2 Households in climate zone 3 Households in climate zone 4 Households in an individual housing unit Households in a collective building When the dwelling has an individual heating system When the collective building has a collective heating system

0/1 0/1 0/1 0/1 0/1 0/1 in m2 in m4 0/1 0/1 0/1 0/1 0/1 0/1 0/1 0/1 0/1

When the main fuel used for heating and hot water system is gas When the main fuel used for heating and hot water system is oil When the main fuel used for heating and hot water system is electricity

0/1 0/1 0/1

Theoretical energy expenditures before renovation minus theoretical energy expenditures after renovation Number of renovations in 2006 Square of number of renovations in 2006

In euros

Binary variables are introduced when a household receives a tax credit

continuous continuous

0/1

D. Charlier / Energy Policy 87 (2015) 465–479

477

Table A-4 Labela distribution (%) according to occupancy status. Label

Energy label Total

A B C D E F G Total

Climate label Homeowner

Tenant

Total

Homeowner

Tenant

0.01 8.06 27.52 25.33 20.07 12.04 6.97

0.01 8.75 29.23 23.83 12 11.54 6.65

0 7.25 25.53 27.08 20.17 12.62 7.35

0.70 6.37 21.52 30.82 19.11 14.41 7.06

0.96 7.31 22.51 30.8 19.17 12.93 6.33

0.40 5.28 20.37 30.86 19.04 16.13 7.92

100.00

100.00

100.00

100.00

100.00

100.00

a

The energy efﬁciency of a dwelling is rated according to energy labels. Energy labels were introduced in 1995 and are aimed at providing consumers with information about the energy efﬁciency of a dwelling. A very efﬁcient dwelling is classiﬁed as A for energy consumption ( o 50 kW hfe/m2/year), while a very inefﬁcient dwelling is classiﬁed as G ( 4450 kW hfe /m2/year). A very efﬁcient dwelling is classiﬁed as A for GHG emissions ( o 50 kW hfe/m2/year), while a very inefﬁcient dwelling is classiﬁed as G ( 4450 kW hfe/m2/year). A ﬁgure is provided in Appendix A (Fig. A-1). In France, the average energy consumption is 195 kW hfe/m2/year. Most households have a D energy rating.

Table A-5 Average household income according to occupancy status and energy label. Label

Energy label

Climate label

Total

Landlord

Tenant

Total

Landlord

Tenant

A B C D E F G

35,655 34,964 31,547 28,198 25,207 16,105 20,148

38,018 26,083 34,662 31,786 28,420 19,017 25,645

26,127 32,788 26,930 24,357 22,581 14,416 15,497

30,694 30,043 28,295 28,263 28,043 26,336 29,292

32,018 33,560 32,178 32,110 31,665 30,231 33,400

27,034 24,368 23,293 23,785 22,696 25,463 15,497

Mean

28,153

31,984

23,687

28,153

31,984

23,687

Energy expenditures were simulated using PROMODUL software. I computed estimations of theoretical energy consumption, GHG emissions, and energy expenditures for each dwelling category using the 3CL method. This computation method was described in a French decree in September 2006. Thus, PROMODUL served as a tool to provide data for feeding the model. To approximate energy expenditures more precisely, I split the housing stock into different categories according to dwelling type (individual or collective), climate zone (four zones), period of construction (ﬁve periods), glazing (double or not), roof insulation (good, intermediate, or bad), ventilation system (mechanical ventilation or not), and main fuel used (electricity, gas, or oil). I chose these categories with a view to merge the databases. The following assumptions were made: – None of the dwellings have verandas or southern exposures. – The accommodation is on one level for individual housing units and an intermediate level for collective buildings. – The same type of fuel is used for heating and hot water. – Only the best renovation solution is chosen.

were calculated (in euros) for each type of renovation. Then, for each type of housing unit and each type of renovation, post-renovation energy expenditures were calculated. The energy savings from a speciﬁc investment were computed as the difference between pre-renovation and post-renovation energy expenditures (measured in euros). This procedure was repeated for each category; that is, a total of 2160 times. For an individual housing unit (1) using electricity as a main fuel, (2) constructed before 1974, (3) with an average surface area of 110 square meters, (4) located in the ﬁrst climate zone, (5) with poor roof insulation, and (6) without double glazing or a mechanical ventilation system, I obtained an average theoretical energy consumption of 747 kW h/m2/year, average GHG emissions of 48 kg.CO2, and an energy-consumption expenditure of €33.80 per year and per square meter. Energy consumption, GHG emissions, and energy expenditures were then calculated (in euros) separately for each type of renovation.

Appendix B I selected the characteristics of each dwelling on the basis of assumptions and information available in the 2006 Enquête Logement database. Energy consumption, GHG emissions, energy expenditures, and energy savings associated with a renovation

B. Results See Table B-1

478

D. Charlier / Energy Policy 87 (2015) 465–479

Table B-1 Results of the bivariate Tobit model with energy savings according to the type of heating system. All households EE Household characteristics Quintile 1 0.176 (0.818) Quintile 2 0.923 (0.820) Quintile 3 0.467 (0.780) Quintile 4 0.645 (0.789) Homeowners 11.31nnn (1.109) Age 0.0982nnn (0.0197) AgenHomeowners 0.160nnn (0.0205) Building characteristics Bef1974 0.564 (1.017) 1974–1981 0.383 (1.216) 1982–1989 0.201 (1.235) 1990–2001 0.0161 (1.148) surface 0.134nnn (0.0237) surface2 0.000230nnn (8.76e 05) Climate1 1.980nn (0.820) Climate2 1.205 (0.799) Climate3 2.469nnn (0.908) Gas 0.564 (1.017) Oil 0.383 (1.216) Renovation characteristics and public policy NB 5.379nnn (0.264) NB2 0.300nnn (0.0309) Energy savings 1.505nn (0.734) Tax credit 1.922nn (0.768) Constant 0.0554nnn (0.00610) Rho 0.327nnn (0.0286) Observations 16,111

Dwellings with individual heating systems

Dwellings with collective heating systems

RW

EE

RW

EE

RW

1.693nnn (0.492) 2.583nnn (0.495) 0.390 (0.461) 0.783n (0.461) 2.048nnn (0.795) 0.00187 (0.0128) 0.0258n (0.0153)

0.0591 (0.913) 0.585 (0.905) 0.115 (0.863) 0.145 (0.872) 12.21nnn (1.179) 0.0977nnn (0.0213) 0.176nnn (0.0218)

1.970nnn (0.552) 2.473nnn (0.550) 0.562 (0.517) 0.745 (0.516) 2.663nnn (0.863) 0.00789 (0.0141) 0.0396nn (0.0166)

1.276 (1.823) 1.922 (1.808) 2.595 (1.772) 2.551 (1.801) 5.150n (2.667) 0.0757 (0.0491) 0.0576 (0.0509)

0.685 (1.070) 2.936nn (1.140) 0.288 (1.018) 0.740 (1.027) 0.863 (2.001) 0.0480 (0.0305) 0.0405 (0.0388)

2.226nnn (0.563) 1.596nn (0.675) 2.372nnn (0.718) 2.101nnn (0.645) 0.0704nnn (0.0128) 0.000139nnn (5.05e 05) 0.148 (0.477) 0.245 (0.461) 1.197nn (0.523) 2.226nnn (0.563) 1.596nnn (0.675)

0.161 (1.100) 0.454 (1.315) 0.184 (1.358) 0.0776 (1.242) 0.127nnn (0.0232) 0.000207nnn (8.49e 05) 2.206nn (0.907) 1.525n (0.891) 2.613nnn (1.000) 0.161 (1.100) 0.454 (1.315)

2.246nnn (0.629) 1.832nn (0.756) 2.261nnn (0.803) 2.023nnn (0.718) 0.0665nnn (0.0134) 0.000126nnn (5.23e 05) 0.175 (0.529) 0.378 (0.516) 1.184nn (0.579) 2.246nnn (0.629) 1.832nn (0.756)

1.545 (2.480) 1.010 (3.035) 0.923 (2.947) 0.902 (2.920) 0.188nnn (0.0390) 0.000434nnn (0.000142) 2.016 (1.878) 0.172 (1.727) 2.766 (2.087) 1.545 (2.480) 1.010 (3.035)

2.350n (1.278) 0.914 (1.498) 2.937n (1.600) 2.645n (1.481) 0.0885nnn (0.0256) 0.000202nn (0.000101) 0.0332 (1.103) 0.114 (1.027) 1.144 (1.204) 2.350n (1.278) 0.914 (1.498)

3.426nnn (0.188) 0.210nnn (0.0229) 0.546 (0.367) 0.601 (0.414)

5.205nnn (0.287) 0.284nnn (0.0329) 1.117 (0.785) 2.174nnn (0.817) 0.0493nnn (0.00647) 0.333nnn (0.0316) 12,898

3.347nnn (0.204) 0.200nnn (0.0242) 0.490 (0.410) 0.682 (0.462)

5.872nnn (0.519) 0.346nnn (0.0667) 5.997nn (2.513) 3.404 (2.650) 0.120nnn (0.0253) 0.412nnn (0.0605) 3213

3.845nnn (0.391) 0.273nnn (0.0560) 0.830 (0.818) 0.463 (0.926)

NB: robust standard errors are given in brackets. Standard errors are shown in parentheses. n

Signiﬁcant at 10%. Signiﬁcant at 5%. Signiﬁcant at 1%.

nn

nnn

References Alibe, B., 2012. Modélisation des consommations d’énergie du secteur résidentiel français à long terme, CIRED. cole des Hautes Etudes en Sciences Sociales. Amemiya, T., 1973. Regression analysis when the dependent variable is truncated normal. Econometrica 41, 997–1016. Amemiya, T., 1974. Multivariate regression and simultaneous equation models when the dependent variables are truncated normal. Econometrica 42, 999–1012. Arnott, R., Davidson, R., Pines, D., 1983. Housing quality, maintenance and rehabilitation. Rev. Econ. Stud. 50, 467–494. Banﬁ, S., Farsi, M., Filippini, M., Jakob, M., 2008. Willingness to pay for energysaving measures in residential buildings. Energy Econ. 30, 503–516. Battu, H., Ma, A., Phimister, E., 2008. Housing tenure, job mobility and

unemployment in the UK. Econ. J. 118, 311–328. Baxter, L.W., 1998. Electricity policies for low-income households. Energy Policy 26, 247–256. Bertoldi, P., Hirl, B., Labanca, N., 2012. Energy Efﬁciency Status Report 2012: Electricity Consumption and Efﬁciency Trends in the EU-27. JRC Scientiﬁc and Policy Report, European Commission Report EUR 25405 EN. Bird, S., Hernández, D., 2012. Policy options for the split incentive: increasing energy efﬁciency for low-income renters. Energy Policy 48, 506–514. Blumstein, C., 1980. Program evaluation and incentives for administrators of energy-efﬁciency programs: can evaluation solve the principal/agent problem? Energy Policy 38, 6232–6239. Boardman, B., 2010. Fixing Fuel Poverty: Challenges and Solutions. Earthscan, London. Brown, M.A., 2001. Market failures and barriers as a basis for clean energy policies. Energy Policy 29, 1197–1207.

D. Charlier / Energy Policy 87 (2015) 465–479

Bullier, A., Milin, C., 2011. Energy refurbishment of social housing using energy performance contract. ECEEE 2011 Summer Study Proceedings: Energy Efﬁciency First: The Foundation of a Low-Carbon Society, pp. 1049–1060. Bullier, A., Milin, C., 2013. Alternative Financing Schemes for Energy Efﬁciency in Buildings. Panels of the ECEEE 2013 Summer Study, pp. 795–805. Commissariat Général du Développement Durable, 2012. Les conditions d’occupation des logements au 1er janvier 2011, Chiffres & Statistiques, pp. Service de l'Observation et des Statisques, Ministère de l'Ecologie, du Développement Durable et de l'Energie. Davis, L.W., 2010. Evaluating the Slow Adoption of Energy Efﬁcient Investments: Are Renters Less Likely to Have Energy Efﬁcient Appliances? NBER Working Paper No. 16114. De Quero, A., Lapostolet, B., 2009. Groupe de Travail Précarité Energétique. Plan Bâtiment Grenelle. Eurocontract, 2008. Comprehensive Refurbishment of Buildings with Energy Performance Contracting. Graz Energy Agency, Berlin Energy Agency and Austrian Energy Agency, 2008. Eurostat housing statistics, 2012. Housing Statistics. 〈http://ec.europa.eu/eurostat/ statistics-explained/index.php/Housing_statistics〉 (Accessed 25.08.15). Fisher, A.C., Rothkopf, M.H., 1989. Market failure and energy policy: a rationale for selective conservation. Energy Policy 17, 397–406. Gillingham, K., Harding, M., Rapson, D., 2012. Split incentives in residential energy consumption. Energy J. 33, 37–62. Giraudet, L.-G., Guivarch, C., Quirion, P., 2011. Comparing and combining energy saving policies: will proposed residential sector policies meet French ofﬁcial targets? Energy J. 32, 213–242. Grösche, P., Vance, C., 2009. Willingness to pay for energy conservation and freeridership on subsidization: evidence from Germany. Energy J. 30, 135–153. Hassett, K.A., Metcalf, G.E., 1995. Energy tax credits and residential conservation investment: evidence from panel data. J. Public Econ. 57, 201–217. Heckman, J.J., 1979. Sample selection bias as a speciﬁcation error. Econometrica 47, 153–161. Herbert, C.E., McCue, D.T., Moyano, R.S., 2012. Is Homeownership Still an Effective Means of Building Wealth for Low-income and Minority Households? (Was It Ever?). Joint Center for Housing Studies. Harvard University, Cambridge, MA. Hills, J., 2011. Fuel Poverty: The problem and its measurement Interim Report of the Fuel Poverty Review. Centre for Analysis and Social Exclusion. Case Report 69. Hills, J., 2012. Getting the measure of fuel poverty, Hills Fuel Poverty Review. Centre for Analysis and Social Exclusion. Case report 72. Högberg, L., 2014. Inclusive Rent and Its Impact on Energy Efﬁciency Investments. European Commission, Joint Research Center, Overcoming the Split Incentive Barrier in the Building Sector Workshop 2014. International Energy Agency, 2007. Mind the Gap: Quantifying Principal Agent Problems in Energy Efﬁciency. 〈https://www.iea.org/publications/free publications/publication/mind_the_gap.pdf〉 (Accessed 23.08.15). Ioannides, Y.M., 1987. Residential mobility and housing tenure choice. Reg. Sci Urban Econ. 17, 265–287. Jaffe, A.B., Stavins, R.N., 1994a. The energy-efﬁciency gap: what does it mean? Energy Policy 22, 804–810. Jaffe, A.B., Stavins, R.N., 1994b. The energy paradox and the diffusion of

479

conservation technology. Resour. Energy Econ. 16, 91–122. Bertoldi, P., Boga-Kiss, B., Rezessy, S., 2007. Latest Development of ESCO’s Across Europe: A European Update. Joint Research Center (JRC), 2007. Khazzoom, J.D., 1980. Economic implications of mandated efﬁciency in standards for household appliances. Energy J. 1, 21–40. Levinson, A., Niemann, S., 2004. Energy Use by apartment tenants when landlords pay for utilities. Resour. Energy Econ. 26, 51–75. Maddala, G.S., 1983. Limited-Dependent and Qualitative Variables in Econometrics. Cambridge University Press, New York. Mauroux, A., 2012. Le credit d’impôt dédié au développement durable: une évaluation économétrique. Document de travail INSEE. Ministère de l'Ecologie du Développement Durable et de l'Energie, 2012. Indicateurs de Développement Durable Nationaux: Consommation d’énergie des secteurs résidentiel et tertiaire. 〈http://www.statistiques.developpement-dur able.gouv.fr/indicateurs-indices/f/1932/1339/consommation-denergie-sec teurs-residentiel-tertiaire.html〉 (Accessed 25.08.15). Murtishaw, S., Sathaye, J., 2006. Quantifying the Effect of the Principal-Agent Problem on US Residential Energy Use. Environmental Energy Technologies Division. University of California, Berkeley. Nair, G., Gustavsson, L., Mahapatra, K., 2010. Factors inﬂuencing energy efﬁciency investments in existing Swedish residential buildings. Energy Policy 38, 2956–2963. Nauleau, M.-L., 2014. Free-riding on tax credits for home insulation in France: an econometric assessment using panel data. Energy Econ. 46, 78–92. Pelletier, P., 2008. Rapport au Ministre d'Etat, ministre de l'Ecologie, du Développement et de l'Aménagement durables. Comité opérationnel ⟪rénovation des bâtiments existant⟫. Pon, S., Alberini, A., 2012. What Are the Effects of Energy Efﬁciency Incentives? Evidence from the US Consumer Expenditure Survey. Working paper presented at the International Association for Energy Economics, 9–12 September. Rehdanz, K., 2007. Determinants of residential space heating expenditures in Germany. Energy Econ. 29, 167–182. Scholten, N.P.M., 2013. Environmental Performance Regulation in The Netherlands. 4th International Conference Civil Engineering 13, Proceedings Part 1. Sofres-ADEME, 2009. Maîtrise de l'énergie: Attitudes et comportements des particuliers. Note de synthèse. Sorrell, S., 2007. The Rebound Effect: An Assessment of Evidence for Economy-Wide Energy Savings from Improved Energy Efﬁciency. UK Energy Research Centre report. Sorrell, S., 2009. Jevons’ Paradox revisited: the evidence for backﬁre from improved energy efﬁciency. Energy Policy 37, 1456–1469. Sutherland, R.J., 1991. Market barriers to energy-efﬁciency investments. Energy J. 12, 15–34. Tobin, J., 1958. Estimation of relationships for limited dependent variables. Econometrica 26, 24–36. Todd, S., Souleles, N.S., 2005. Owner-occupied housing as a hedge against rent risk. Q. J. Econ. 120, 763–789. van Soest, D.P., Bulte, E.H., 2001. Does the energy-efﬁciency paradox exist? Technological progress and uncertainty. Environ. Resour. Econ. 18, 101–112.

Energy efficiency investments in the context of split incentives among French households

Energy efficiency investments in the context of split incentives among French households

Recommend Documents