Do private rental tenants pay for energy efficiency?: The dynamics of green premiums and brown discounts

ABSTRACT

The rental market, notably within the UK, is facing increased scrutiny with the tightening of regulation in relation to energy performance and the increases in energy costs and rental unaffordability. Whilst a sizeable volume of research has examined the pricing effects of EPCs and house prices, scrutiny of this relationship within the private rental sector remains more embryonic. This study, using 2,914 transactions for Northern Ireland, extends the traditional analysis beyond the conditional mean estimate by examining the quantiles of the relationship between EPCs and rental prices. The findings provide evidence of a rental premium of 0.2% for a one-point improvement in energy efficiency. In terms of EPC ratings, we find premiums for B- and C-rated dwellings of 8.2% and 2.4% and also discount effects for E, F and G-rated properties ranging between 3.9%-5.5%. The quantile findings exhibited a parabolic effect across the price distribution demonstrating the lowest and highest priced properties to comprise higher price premiums of 13% and 16% for B-rated dwellings, however more pronounced discount effects for F- and G-rated properties at the lowest and highest deciles. The findings provide evidence to help alleviate the split-incentive problem between landlords and tenants within the UK.

KEYWORDS:

• Energy performance certificates
• quantile regression
• hedonic pricing model
• rent prices
• energy efficiency

Introduction

Climate change and energy consumption have become two of the most well studied, researched and debated topics in recent times. In this context, a variety of treaties and policies have been implemented globally to tackle the ongoing and damaging effects of greenhouse gases. The Brundtland Report (1985) published by the World Commission on Environment and Development identified seven thematic sustainable development goals relating to the political, social, economic, technological, and ecological systems which foster international sustainability and problems arising from disharmonious development (Brundtland, 1985). In 1990, at the Second World Climate Conference, countries recognised that the principle of equity and differentiated responsibility of countries should form the basis of a global response to climate change (Egypt, 1989). The Rio summit in 1992 followed suit by creating a blueprint for international action on development and environmental issues, with the Kyoto Protocol (1997) the first legally binding international agreement that set targets for greenhouse gas emission reductions (Kim et al., 2020).

Following the Kyoto Protocol, the European Union (EU) reduced its greenhouse gas emissions by approximately 8% (by 2015), leading to a reduction of 23% relative to 1990 levels (Rayer & Jordan, 2016). Despite these reported reductions, the debate surrounding carbon abatement have intensified due to the increased frequency of extreme weather events and climate change, indicating that the existing agreements were not adequate. In December 2015, the Paris Agreement on Climate change was a landmark legally binding international treaty which saw 196 parties at COP 21 agreeing to limit global warming to below two degrees Celsius compared to pre-industrial levels. These new targets have spawned various legislative instruments and produced workable directives to combat the growing effects and damage caused by emissions and over consumption within the EU. For example, The European Green Deal which has strategic targets to ‘cut greenhouse gas emissions by at least 55% by 2030’ (European Commission, 2021).

The real estate sector remains a key contributor to climate change and adaptation given that it is responsible for 40% of the global carbon dioxide emissions with the sector also responsible for approximately 22% of total energy consumption (Gynther et al., 2015; Loga et al., 2016), with approximately 75% of the EU building stock considered energy inefficient (European Commission, 2021). Consequently, the existing housing stock is of fundamental importance for carbon abatement targets, with approximately 80% of existing housing stock within Europe expected to require renovation or retrofit to achieve the 2°C targets by 2050 (Energy Savings Trust, 2021).

To combat the challenges of climate change, a number of EU directives have come into force since the early 2000’s. In 2002, The EU formed the Energy Performance of Buildings Directive (EPBD) to help reduce energy usage within the building sector. In an initiative to increase awareness regarding carbon emissions and reduction, the EU directive introduced Energy Performance Certification (EPC’s) in 2005. These energy labelling certifications were first introduced to the United Kingdom (UK) in 2008 through the EPBD, meaning that any building newly constructed, sold or let to a new tenant required an EPC certificate. This action also sets a minimum energy standard requirement comprising an EPC rating of E or above in order for the property to legally transact. This step was envisaged to encourage property owners and landlords to improve the energy efficiency, reduce greenhouse emissions, and encourage the retrofitting or redevelopment of energy inefficient buildings. More recently, in an attempt to further heighten energy awareness, current legislative reform requires all newly rented properties to have an EPC rating of C or above from 2025 onwards (Ferentinos et al., 2021).

Following the implementation of this mandatory certification, an array of research studies have examined whether there is a capitalisation effect between property value and energy labelling (Brounen & Kok, 2011; Cerin et al., 2014; Fuerst et al., 2013). Within Northern Ireland (NI), several investigations have examined the nature and significance of EPCs and house prices (P. Davis et al., 2017; P. T. Davis et al., 2015; McCord et al., 2020; McCord, et al., 2020; McCord, et al., 2020). Yet despite this body of evidence, research investigating the Private Rental Sector (PRS) has remained limited, despite the sector containing some of the least energy efficient housing (Ambrose, 2015). It is estimated that private rented housing constitutes approximately 30% of the overall EU housing stock (Chegut et al., 2020), and it is less energy efficient than owner-occupied buildings (Burfurd et al., 2012; Lang et al., 2022), due to under-regulation and institutional barriers. Indeed, seminal research such as that of Laquatra (1987), highlighted that institutional barriers within the rental housing sector have contributed to a lack of thermal integrity in structures, which in many cases are inhabited by those who can least afford energy price increases (Wrigley & Crawford, 2017). Notably, as Wrigley and Crawford (2017) contend, rental properties in particular are unique in that they suffer from a range of barriers and market failures, including misinformation, split incentives in relation to who pays for energy upgrades, and an uneven power dynamic between renters and landlords.

Within Northern Ireland, rising rents and energy costs are placing increased financial stress on tenants, with existing research showing rents to have increased by 53 percentage points on the rental index since 2016, and by 35 percentage points since 2021.¹ Further governmental research has also revealed that approximately 84% of tenants within the PRS in NI currently spend more than 25% of their income on rental housing costs.² Pertinently, as Ambrose (2015) identifies, the choices that occupants make regarding energy are constrained by the material characteristics of a property, something only the landlord can alter, and that initiatives which convince landlords of the benefits of improving energy efficiency remain elusive.

Consequently, this study firstly examines a sample of PRS dwellings in Northern Ireland using hedonic regression analysis to establish whether a rental premium is found for energy efficient properties, when controlling for several rental determinants. The findings suggest that a capitalisation effect is evident for more energy efficient dwellings with discounts observed for rental properties with poorer energy efficiency. Secondly, when further examining the existence of a premium effect across the price distribution using quantile regression, the findings reveal a parabolic effect across the quantiles suggesting that price premiums and discounts are not symmetrical. This finding indicates differences in how tenants ‘value’ energy efficiency, suggesting that tenants residing in the lowest and highest priced properties capitalise energy efficiency into their housing costs more so than the conditional mean estimates derived from stand hedonic modelling suggests, and importantly that socio-economic standing may impact upon behaviour.

Therefore, the contribution of this paper is twofold. It provides empirical evidence and analysis of the value of energy efficiency in the private rented sector in Northern Ireland, using a nuanced dataset, revealing evidence of both capitalisation and discount effects in relation to energy performance comparable to other European markets. Moreover, to the best of the authors knowledge, this is one of only few studies which examines a rental price capitalisation effect across the price distribution using quantile regression. Whilst most of the existing analysis has generally tended to use traditional hedonic specifications which estimates the conditional mean of the response variable providing a single coefficient estimate, we employ quantile regression to further examine the nature of the relationship. This approach is an extension of linear regression and used when the conditions of linear regression are not met. The application of quantile regression is therefore used as the conditional quantile functions of energy performance are of interest, and to further unearth whether energy performance is valued differently for those renters living in lower and higher priced dwellings.

Following this introduction, the paper is divided into five sections. Section two provides an overview of the existing literature focusing on the rental market and energy performance and pricing, the role of information asymmetry and the willingness to pay for energy efficiency. Section three details the data and methodology used in the study, with Section four presenting the empirical findings. Section five provides a discussion on the modelling outputs, with Section six presenting conclusions.

Literature review

Numerous research papers internationally have investigated the relationship between energy performance within both the residential and commercial property sectors, revealing mixed findings, notably relating to the magnitude of the premium effects (Cajias & Piazolo, 2013; P. T. Davis et al., 2015; Fuerst et al., 2016; Gabe et al., 2014; Marmolejo-Duarte & Chen, 2019; Murphy, 2014), their spatial composition (Galvin, 2023; McCord, Davis, et al., 2020; Taltavull et al., 2017) and their overall level of statistical significance. Despite a voluminous body of research in this area, the literature in relation to the PRS is not as extensive or conclusive, although this evidence base is beginning to become more prominent over time due to the emerging importance of rental markets globally. The ‘Willingness to pay’ (WTP), information asymmetry, awareness and disclosure are key themes which have emerged within the energy efficiency debate, with findings also varying due to several factors such as socio-economic profile, energy costs, and the potential savings that can be realised from retrofit energy-efficient upgrades (Bird & Hernández, 2012).

Rental price premiums?

Studies investigating the nexus between energy efficiency and rental pricing have begun to emerge recently and generally have identified both premium and discounts at varying degrees. Existing studies have found sizeable positive pricing effects ranging between circa 5% and 14% (de Ayala et al., 2016; Hyland et al., 2013; Im et al., 2017; Liu et al., 2018), whereas other research has shown the size effect to be more negligible such as Cajias et al. (2019) who revealed premiums ranging from 1.4% to 0.2% for A+ to C-rated properties, and discounts of 0.1% and 0.5% for F, G and H-rated dwellings. The findings across a number of existing studies have suggested that the magnitude and differences in the size of the premium or discounts to be a consequence of micro-economic market structure due to geographical variation, spatial unevenness, demand and supply imbalance and during periods of economic recession (Cajias et al., 2019; Galvin, 2023; Petrov & Ryan, 2021), limited retrofitting due to insufficient payback period for landlords (Galvin, 2023) and inefficient heating technology attributable to landlord and renter behaviour as well as the wider institutional setting (Taruttis & Weber, 2022). Indeed, in the work of both Petrov and Ryan (2021) and Hyland et al. (2013) they found that rental properties are, on average, 10% less efficient than owner-occupied properties and that larger premiums are observable for owner-occupied dwellings when compared with rental units.

Willingness to pay

The role of energy efficiency and the WTP for improved energy performance has been subject to a number of studies across rental markets globally, with general findings identifying challenges relating to awareness, demographic profile, and cost savings. Empirically, the research undertaken by Collins and Curtis (2018) showed that tenants in Ireland are willing to pay, on average, €38 for a one-grade improvement along a 15-point EPC scale. However, other findings from the study undertaken by Carroll et al. (2016), again for Ireland, revealed that tenant’s willingness-to-pay did reduce when moving from less efficient to more efficient EPC grades, concluding that higher energy efficiency levels only appear achievable if a renter’s WTP exceeds the landlord’s investment costs. In a similar study, Fuerst et al. (2020) also highlighted that the value of future energy cost savings exceeds tenants’ implicit WTP by a factor of 2.5 which the authors suggest is invariably a consequence of the market power of tenants, uncertainty in the rental relationship and the ‘landlord – tenant dilemma’ complicated by the split-incentive problem. Further research has examined the WTP and found it to be related to socio-demographic profiling, with older tenants more likely to possess a non-zero WTP than younger tenants, and that older tenants are less aware of energy certification (Collins & Curtis, 2017; Marmolejo et al., 2020).

Information asymmetry between landlord and tenants

This ‘landlord-tenant problem’ in relation to information on energy efficiency within the residential rental market remains an important issue. The role of asymmetric information, and more specifically, whether energy cost information asymmetries exist between landlords and tenants has been subject to a number of studies which have shown a misalignment between capital costs and utility cost savings and disclosure (Bian & Fabra, 2020; Carroll et al., 2016; Myers, 2020; Phillips, 2012). In a paper questioning the efficiency of the market for considering ‘energy efficiency’, Sieger and Weber (2023) concentrating on the apartment sector, explored the relationship between ‘a total-cost-of-renting perspective’ based on rational behaviour that monthly rent should reflect differences in expected monthly heating costs. Applying 844,229 apartment listings between 2014 and 2020, they established a small premium to exist for more energy efficient apartments, suggesting that if the energy performance score decreases by 10 kWh/m²a, the monthly basic rent increases, on average, by roughly €0.01 per square metre. Their analysis indicates that expected energy cost savings exceed the premium by a factor of three to seven, and that discounts of up to 9.2% are observable if apartments use inefficient heating technologies. This was also the subject of research undertaken by Franke and Nadler (2019). Studying the role and influence of energy efficiency on tenants and owners, they found that owners give energy performance a stronger consideration than tenants, which the authors attribute to purchase decisions being more long-term oriented and cost intensive. Alternatively, research exploring landlords’ attitudes to energy efficiency have demonstrated that this is likely shaped by a lack of knowledge (Ambrose, 2015; Hope & Booth, 2014), as well as the principal-agent problem (Ambrose, 2015), compounded by the absence of direct financial incentives, cultural factors, and low-cost retrofit activity (Lang et al., 2022).

This existing body of research demonstrates that the relationship between rent prices and energy performance is complex and the extent to any capitalisation effect is premised on awareness, information on cost savings, affordability, demographics, socio-economic profiles and the ‘landlord – tenant dilemma’ complicated by the split-incentive problem. Indeed, these issues imply that choice decisions and trade-offs result in pricing differentials with varying levels of energy efficiency (McCord, Lo, et al., 2020), determined by segmentation as the energy performance relationship differs according to the type of housing and particular housing segments (Cerin et al., 2014) the deterioration of rental affordability (Wrigley & Crawford, 2017), and the financial barriers for tenants to cover the cost of energy efficiency improvements. Accordingly, this paper is positioned in this debate and seeks to add to the literature base by identifying the extent to which energy labelling is associated with energy performance within the rental market. In doing so, we suggest that tenants are willing to pay more for energy efficiency, and that this willingness to pay fluctuates across the rent price distribution due to the relationship between rental costs, income, energy costs and by property type.

Data and methodology

This study uses 2,914 rental transactions obtained from the Ulster University Rental Price Index (UURPI) for the period Q2, 2022 to Q4, 2022 for Northern Ireland. The UURPI is an established rental market index developed in 2014 which is based on a robust sample of achieved ‘Let agreed’ rental transactions obtained from estate and letting agents on a monthly basis. The data is therefore the price agreed within the open market based on a list and then rent agreed price and is verified and validated using robust data checks and testing procedures. Given that all rental properties in the UK cannot be advertised or rented without having an energy performance certification, the sample is based on all the transactional evidence available, thus mitigating any potential selection bias. The data includes several physical attributes of properties alongside neighbourhood and location identifiers as observed in Table 1.

Table 1. Property variables and descriptions.

Variable	Description	Type
Housing attributes
Rental price	Monthly rent price when ‘Let agreed’	C
In(Price)	Log of monthly rent price	C
Size	Floor area in m^2	C
Property Type	Type of property (1 if terrace; 0 otherwise)	B
Property Age	Year the property is built (in decades) (1 if 1900–1910; 0 otherwise)	B
Bedrooms	Number of bedrooms (1 if Beds1; 0 otherwise)	B
Reception rooms	Number of reception rooms (1 if Receps1; 0 otherwise)	B
Grade	Condition of the property (1 if Grade B; 0 otherwise)	B
Furnished	Whether the property is furnished (1 if part furnished; 0 otherwise)	B
Floor	Floor property is located on [Apartments only] (1 if 1^st Floor; 0 otherwise)	B
EPC rating	Assessed energy efficiency rating in bands (A-G) (1 if B; 0 otherwise)	B
EPC score	Assessed energy efficiency score (0–100)	C
Neighbourhood and location variables
Postcode	The postcode where the property is located to represent a granular submarket area (e.g. 1 if BT13; 0 otherwise or 1 if BT06; 0 otherwise)^1	B
Rural	If the property is located within a settlement development limit (1 if rural; 0 if urban)^2.	B
Settlement Band	Settlement bands where the property is located^3.	B
Multiple Deprivation	Decile of Multiple deprivation where a property is located (1 if Decile1; 0 otherwise	B

C = continuous; B = binary. ¹The Postcode unit used is at the two-digit level. There are 80 postcodes areas. ² Settlement Development Limits are a statistical classification and delineation of settlements in Northern Ireland defined by the Department for the Economy’s Planning Service. ³ Settlements are grouped with other settlements of similar characteristics. There are 8 Bands (A-H) based on Census population only.

The neighbourhood and locational variables were attributed by layering and undertaking spatial joins from databases from official government data using Geographical Information Software (GIS). Where applicable, the variables were transformed into binary state for the hedonic modelling. To measure the effect of EPCs on rental pricing, the study applies both the EPC rating (bands) and the assessed EPC score. As observed in the study of McCord, Davis, et al. (2020), the EPC score is applied due to it being a continuous variable and thereby intuitively providing a more deterministic relationship that allows for model estimates of the EPC score to be assessed as a unitary effect.

In line with existing research into energy efficiency and EPC performance, the data comprises some limitations related to missing determinants of energy efficient features. In this study, one omitted variable relates to the heating type of the rental properties which was not available. It should be acknowledged however that this characteristic should be implicitly priced into the assessed value and EPC scores through their original valuations and energy performance inspections which renters (ought to) take account of when selecting a rental property.

Descriptive analysis

A summary of the descriptive statistics for the data is presented in Table 2. The sample mean rental price is £802 per calendar month, with a standard deviation figure of £330. The average floor size is 93 m², with the average EPC score of 61.7 or rating category D, and in line with the wider UK average residential EPC band rated D.³ In terms of EPC ratings within the sample, 8.1% are B-rated with 30.7% C-rated and 32.4% D-rated. Just under 20% of the sample is E-rated (18.1%) with 7.7% F-rated and 3.0% rated G.

Table 2. Descriptive statistics.

Variables	Minimum	Maximum	Mean	Std. Dev
Rental Price	250	3,600	802.13	330.459
In(Price)	4.61	8.19	6.62	0.33824
Size (m^2)	14.80	510.00	93.31	40.389
Bedrooms	1	6	2.53	1.017
Reception rooms	1	4	1.19	0.455
EPC Score	16	93	61.73	15.065
Classification type	Private built	89.9%	Public built	11.9%
Settlement type	Rural	12.5%	Urban	87.5%
Furnish type	Furnished	60.1%	Unfurnished	23.0%
	Part furnished	11.8%	Lux Furnished	5.1%
EPC Band	Band A	0.1%	Band B	8.1%
	Band C	30.7%	Band D	32.4%
	Band E	18.1%	Band F	7.7%
	Band G	3.0%

Methods

Hedonic modelling is typically applied common approach applied within property analysis to ascertain the marginal effects of property attributes. Typically, the functional relationship between the price $\mathit{P}$ of a heterogeneous good $\mathit{i}$ and its quality characteristics represented by a vector ${\mathit{x}}_i$;

(1) $P_i=\mathit{f}\left({\mathit{x}}_i;\mathit{\beta }\right)+u_i$(1)

Where $P_i$ represents a property with a price P, ${\mathit{x}}_i\mathit{\hspace{0.17em}}$ the structural attributes, $\mathit{\beta }$ relates to the vector of coefficients which are estimated for the characteristics, with $u_i$ representing the error term.

We adopt a semi-logarithmic hedonic specification in order to generate coefficients that can be interpreted as percentage differentials as follows:

(2) $\mathit{In}\left(P_i\right)=\mathit{\alpha }+{\sum }_{j=1}^J{\beta }_j{z}_{\mathit{ji}}+u_i$(2)

where the natural logarithm of the rent of the ith property is a function of the J characteristics assumed to influence price, and ɑ, β the coefficients estimated, and u the normally distributed error term.

Overall, the regression model specifications include the type and size (in sqm) of the property, the year it was built, year the number of bedrooms, reception rooms and the floor on which an apartment is on, whether it was furnished, its grade, deprivation decile and ‘settlement band’, as well as the two-digit postcode and whether it is in a rural or urban area.

Quantile regression

Previous studies have investigated the nature of traditional least squares regression with general criticisms notably around violation of the independent observations assumption, error-bias, and that it only provides the conditional mean of the distribution based on squared residual errors being minimised (Mosteller & Tukey, 1977). Quantile regression introduced by Koenker and Bassett (1978), however extends classical least squares regression to an ensemble of models for conditional quantile functions which more fully represents the conditional distribution (Koenker & Bassett, 1978) by minimising an asymmetrically weighted sum of the residual errors (Koenker & Bassett, 1978; Koenker & Hallock, 2001), and are therefore insensitive to heteroscedastic errors and dependent variable outliers (Buchinsky, 1991) or when the error term is non-normal (Buchinsky, 1998) – a notable feature of property data and markets.

For a random variable Y, the $\mathit{\tau th}$ quantile is defined by a value y such that the probability of finding a smaller y is less than or equal to $\mathit{\tau }$, and the probability of finding a larger y is less than or equal to 1 - $\mathit{\tau }.$ Similarly, the $\mathit{\tau th}$ quantile regression function, $\mathbf{B}(\mathit{\tau }$), corresponds to a linear or quadratic function fit through the data such that approximately $\mathit{\tau }$ proportion of the observations are less than $\mathbf{B}(\mathit{\tau }$) and 1 – $\mathit{\tau }$ proportion of the observations are greater than $\mathbf{B}(\mathit{\tau }$). Estimates b$(\mathit{\tau }$) of $\mathbf{B}(\mathit{\tau }$) are obtained by minimising the absolute values of the residuals where positive residuals are given weights equal to $\mathit{\tau }$ and negative residuals are given weights equal to 1-$\mathit{\tau }$ (for full mathematical descriptions of the algorithms see Koenker & Bassett, 1978). The basic quantile regression can be written as:

(3) $y_i=x_i^{\prime }{\beta }_{\theta }+u_{\mathit{\theta i}}\mathit{with}\mathit{Quan}{t}_{\theta }\left(y_i|x_i\right)=x_i{\beta }_{\theta }$(3)

where $x_i^{\prime }$ denotes a vector of regressors, β_θ represents the vector of parameters to be estimated, and u_θi is a vector of residuals. Quant_θ(y_i|x_i) represents the θ^th conditional quantile of y_i given $x_i^{\prime }$. The θ^th regression quantile solves the following problem:

(4) $\underset{\beta }{min}=\sum _i\mathit{\theta }\left|y_i-x_i\mathit{\beta }\right|+\sum _i\left(1-\mathit{\theta }\right)\left|y_i-x_i\mathit{\beta }\right|\underset{\beta }{=min}\sum _i{\rho }_{\mathit{\theta }}{u}_{\mathit{\theta it}},{\theta }_{\epsilon }\left(0,1\right)$(4)

where ${\rho }_{\mathit{\theta }}$ is known as the ‘check function’ and defined as:

${\rho }_{\theta }\left(\mathit{\epsilon }\right)=\mathit{\theta \epsilon }\mathrm{if}\mathit{\epsilon }\ge 0$

$\left(\mathit{\theta }-1\right)\mathit{\epsilon }\mathrm{if}\mathit{\epsilon }<0$

Equation (4)(4) $\underset{\beta }{min}=\sum _i\mathit{\theta }\left|y_i-x_i\mathit{\beta }\right|+\sum _i\left(1-\mathit{\theta }\right)\left|y_i-x_i\mathit{\beta }\right|\underset{\beta }{=min}\sum _i{\rho }_{\mathit{\theta }}{u}_{\mathit{\theta it}},{\theta }_{\epsilon }\left(0,1\right)$(4) is then solved by the linear programming technique. The median regression, which is a special case of the quantile regression, is obtained by setting u ($\mathit{\tau }$) = 0.5. Other quantile of the conditional distribution can be obtained via variation of u ($\mathit{\tau }$). To convey a sense for the relationship of selected explanatory variables across the entire conditional rental price distribution, the results are reported across quantile deciles ranging from the 10^th decile ($\mathit{\tau }$ = 0.1) to the 90^th decile ($\mathit{\tau }$ = 0.9).

Empirical findings

The empirical analysis is conducted on two base OLS models which contain the EPC coefficients as ratings (bands) in Model 1, and the EPC Score (Model 2). A further interactive model controlling for Type*Age applied to add robustness to the findings (Models 3–4), to allow the relationships between property attributes to be more non-linear. We further specify property type regression models (Models 5–12) to measure the EPC rating and EPC score pricing effects for each segment of the rental market. Finally, ten quantile regression models (Models 13–21) which represent the quantiles ($\mathit{\tau })$ ranging from 0.1–0.9 are developed based on EPC ratings, and then applying the EPC score variable (Models 22–30). Examination of the base OLS model coefficients exhibit general conformance with exceptions in terms of magnitude and direction.

In terms of energy performance, when considering the base EPC rating model (Model 1), the findings show that EPC bands comprise varying degrees of positive and negative effects on rental prices. The findings show EPC B-rated properties to display an 8.2% premium effect, statistically significant at the 5% level. This effect diminishes to 2.4% (p < .01) for properties classified with an EPC rating of C (Table 3). Rental housing with an EPC classification E, displays a negative effect of 3.9%, with Bands F exhibiting a 5.5% discount significant at the 1% level of significance (Figure 1). In terms of the EPC score (Model 3), the findings indicate that a unitary increase in EPC score, ceteris paribus, culminates in a 0.19% increase in rental prices, significant at the 5% level. Thus, a ten-point increase in the EPC score translates to a 1.9% rental premium. Further examining the interaction between property type and age (Model 2), the EPC rating model exhibits largely consistent results with Model 1, demonstrating an overall pricing effect for a B-rated rental dwelling to be 8.2%, with a C-rated property showing a 1.8% premium. Rental properties which are E, F and G-rated exhibit discounts of 4.0%, 5.6% and 3.8% (Figure 1), although it must be acknowledged that G-rated properties display limited statistical significance which may be due to the small sample size.

Figure 1. Rental market EPC rating price premium and discount effects.

Table 3. OLS EPC rating and score regression models.

	Model 1 (Rating)	Model 2 (Score)	Model 3 (Rating Int)	Model 4 (Score Int)	Model 5 (Apt)	Model 6 (Det)	Model 7 (Semi)	Model 8 (Terr)	Model 9 (Apt)	Model 10 (Det)	Model 11 (Semi)	Model 12 (Terr)
(Constant)	6.69453***	6.2639***	6.6845***	6.5654***	6.7461***	7.1152***	6.8508***	6.4866***	6.6875***	6.8025***	6.7243***	6.3656***
EPC B	0.08216**		0.08241**		0.1010**	0.0519	0.0735**	0.0240*
EPC C	0.02420***		0.01783***		Holdout	0.0310*	Holdout	0.0237**
EPC D	Holdout		Holdout		−0.031**	Holdout	0.0124	Holdout
EPC E	−0.03695***		−0.04040***		−0.035**	−0.01581*	−0.0300**	−0.0404***
EPC F	−0.05319***		−0.05616***		−0.135***	−0.02047*	−0.0257**	−0.0594***
EPC G	−0.02999		−0.03779*		−0.065	−0.03432	−0.1310	−0.0123
EPC Score		0.00191***		0.001851***					0.0029***	0.0001	0.0021**	0.0019***
Size (m^2)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Type dummies	Yes	Yes	NA	NA	N/A	N/A	N/A	N/A	N/A	N/A	N/A	N/A
Age dummies	Yes	Yes	NA	NA	N/A	N/A	N/A	N/A	N/A	N/A	N/A	N/A
Type*Age dummies			Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Bed dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Recep dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Grade dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Floor dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Furnished dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Set. Band dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Rural dummy	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
MDM dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R^2	0.717	0.71613	0.726	0.724	0.785	0.878	0.853	0.788	0.780	0.878	0.854	0.787
Adj. R^2	0.695	0.69517	0.699	0.699	0.680	0.690	0.785	0.757	0.680	0.707	0.790	0.758
F	32.630***	33.584***	27.361***	27.973***	7.495***	4.662***	12.455***	25.597***	7.776***	5.138***	13.279***	26.433***
N	2,914	2,914	2,914	2,914	1,096	244	415	1,159	1,096	244	415	1,159

‘*’, ‘**’ and ‘***’ indicate statistical significance at the 10%, 5% and 1% level, respectively. Parsimonious model presented. Full model including time and space dummies and Std. errors, t-statistics and p-values available upon request. Huber Sandwich Standard Errors & Covariance. Sparsity method: Kernel (Epanechnikov) using residuals. Bandwidth method: Hall-Sheather.

The type-based models further highlight the potential differential effects between EPCs and the capitalisation effect (Table 3). For B-rated rental dwellings, apartments exhibit the most pronounced premium effect of 10.1%, significant at the 5% level, compared to a C-rated apartment, followed by the semi-detached sector which displays a statistically significant premium of 7.4% (β = 0.0735, p < .05) for a B-rated rental property. For the detached segment, the findings show 5.2% and 3.1% premiums for B- and C-rated houses, however these a not statistically significant at the conventional level. Terrace rental units exhibit the lowest premium effect across the property type models of 2.4% for both B and C-ratings, significant at the 5% level, relative to a D-rated terrace rental property. Apartments further reveal increased price discounts for poorer energy performance. D-rated and E-rated apartments show a discount of 3.1% and 3.5% (β = 0.0310; β = 0.035, p < .05) with apartments with an F-rating more sizeable at 13.5% (β = 0.1351, p < .01). Terrace rental stock exhibits price discounts for E- and F-rated properties of 4.0% and 5.9% (β = 0.0310; β = 0.035, p < .01) which are both statistically significant at the 1% level, with the semi-detached sector showing discounts of 1.2% for D-ratings, 3.0% and 2.6% for E- and F-rated properties. Interestingly, the detached sector exhibits lower price discounts for poorer EPC labelling, observing a 1.6% decrease for E-rated dwellings and 2.0% for properties with an EPC rating of F.

Examination of the EPC scores across the various market sectors also revealed some differential effects in the size of the EPC coefficients. The apartment sector revealed the largest premium effect of 0.29% for a unitary increase in EPC score, with semi-detached and terrace housing both exhibiting a 0.21% and 0.19% effect respectively, significant at the 5% and 1% level. The detached sector however revealed a 0.01% effect which is not statistically significant, suggesting that renters of larger detached housing do not capitalise energy efficiency into their decision making, a finding in line with the EPC ratings which exhibited reduced premium and discount effects for detached housing.

Quantile regression findings

For the quantile analysis, the size coefficient estimates show a varying degree of the marginal percentage effects across the price distribution and indicates that the size of rental properties is valued differently across the quantiles and that conditional quantiles are not identical (Table 4). The size coefficient increases across the quantile range from 0.04% in the lowest quantile ($\mathit{\tau }=0.1$) to 0.22% in the highest quantile ($\mathit{\tau }$ = 0.9) demonstrating that lower priced rental properties comprise a lower pricing effect relative to higher priced properties. In relation to the condition of the rental properties, the quantile analysis also shows that lower priced rental properties to have a higher statistically significant effect relative to higher priced rental properties.

Table 4. Quantile EPC rating regression results.

	Model 13 (tau = 0.1)	Model 14 (tau = 0.2)	Model 15 (tau = 0.3)	Model 16 (tau = 0.4)	Model 17 (tau = 0.5)	Model 18 (tau = 0.6)	Model 19 (tau = 0.7)	Model 20 (tau = 0.8)	Model 21 (tau = 0.9)
(Constant)	6.089481***	6.343022***	6.286529***	6.35886***	6.93294***	7.142382***	7.20565***	7.335191***	7.087429***
EPC B	0.131966***	0.0854**	0.073791**	0.075634	0.080532**	0.082056*	0.139769***	0.082846**	0.159521**
EPC C	0.031011**	0.03593***	0.023964**	0.024219*	0.030372**	0.037145***	0.055344***	0.057539***	0.055481**
EPC E	−0.04696**	−0.05411***	−0.034217**	−0.029115**	−0.031871**	−0.03105***	−0.03436***	−0.03163**	−0.02787**
EPC F	−0.06028**	−0.03656**	−0.038417**	−0.02004	−0.035541**	−0.03652**	−0.04922***	−0.06589***	−0.0576**
EPC G	−0.08951	−0.08482	−0.036584	0.017314	−0.009756	−0.01945	−0.06213**	−0.07295	−0.12084
Size (m^2)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Type dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Age dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Bed dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Recep dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Grade dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Floor dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Furnished dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Set. Band dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Rural dummy	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
MDM dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R^2 (Pseudo)^1	0.322232	0.343189	0.368387	0.400168	0.431511	0.45957	0.484259	0.507374	0.51715
Adj. R^2	0.303555	0.325089	0.350981	0.383638	0.415845	0.444678	0.470047	0.493799	0.503844
Quan. dep. var	6.309918	6.388561	6.437752	6.476972	6.55108	6.620073	6.745236	6.856462	7.130899
Sparsity	0.751393	0.451563	0.392087	0.364276	0.356455	0.3531	0.390747	0.503103	0.740636
N	2,914	2,914	2,914	2,914	2,914	2,914	2,914	2,914	2,914

Examination of the EPC rating quantile coefficients within Models 13–21, as observed in Table 4, clearly demonstrates that energy performance is not constant or significant across the entirety of the rental pricing structure. Indeed, across the rent price distribution all quantiles show a positive premium effect for rental properties with EPC ratings of B and C, with discount effects observable across the quantiles for those properties with EPC ratings E, F and G. Notably, whilst the conditional mean estimate models showed that EPC band B exhibited an 8.2% premium, examination of the quantile coefficients reveals a more varied and parabolic pricing picture. Rental prices within the lowest quantile ($\mathit{\tau }=0.1$) display a higher premium effect of 13.2% (β = 0.132, p < .01) relative to the lower and mid-range price distribution (2^nd up to the 6^th quantiles) which show relatively consistent pricing effects ranging between 7.5% and 8.2% (Figure 2), with the higher end of the price distribution ($\mathit{\tau }=0.9$), showing the largest premium of 16.0% (β = 0.1595, p < .05). This finding infers that rental properties comprising the lowest and highest values with an EPC rating B show the largest premium effects – suggesting that EPC B-rated properties are valued differently across the quantiles and that conditional quantiles are not identical.

Figure 2. EPC Rental market price premium and discount effects.

For EPC C-rated rental properties, the quantile analysis shows higher premiums of 5.5% and 5.8% at the higher end of the price distribution ($\mathit{\tau }=0.7-0.9$) relative to the remaining quantiles (Figure 3). Like the EPC B-rated rental units, there does appear to be a slightly higher capitalisation effect for the lowest rental price quantiles ($\mathit{\tau }=0.1\mathit{and}0.2$) relative to the mid-point price range. In contrast, rental properties with lower energy performance reveal brown discounts across the quantiles. The results reveal that properties with an EPC rating of E have a relatively uniform discount effect of between 2.8% and 3.4% for the majority of the quantiles ($\mathit{\tau }=0.3-0.8$), with the highest quantile ($0.9$) showing the lowest discount effect (β = 0.0278, p < .05). Notably, the findings exhibit the lowest quantiles ($\mathit{\tau }=0.1\mathrm{and}0.2$) to comprise the largest discount effects of 4.7% and 5.4% significant at the 5% and 1% level for E-rated rental properties (Figure 2). When considering F-rated properties the size and nature of the discount effect generally increases across the quantiles, however it is notable that the increases are more pronounced for rental properties within the higher quantiles of the rental price distribution ($\mathit{\tau }=0.7-0.9$) ranging between 4.9% and 6.6%, and for the lowest quantile ($\mathit{\tau }=0.1:\mathit{\beta }=$-0.06028**, p < .05), – again resembling a parabolic effect across the price distribution. Similarly, G-rated properties exhibit further discount effects at the higher and lower ends of the price distribution with no real sizeable effects evident within the mid-range of the price distribution ($\mathit{\tau }=0.4-0.6$), although the results are not statistically significant at the conventional level.

Figure 3. Rental market quantile EPC-ratings price premiums and discounts.

When further considering the EPC score variable, the conditional mean estimate showed that a unitary increase in EPC score translates to a 0.19% increase in rental pricing, significant at the 5% level. Examination of the quantile distribution (Models 22–30) also shows differential effects across the quantiles. As observed in Table 5, the EPC score coefficients somewhat confirm the findings of the Banded EPC models (Models 13–21) insofar that there is a parabolic effect ranging across the quantile range. As observed in Figure 4, for the lowest quantile ($\mathit{\tau }=0.1$), a one-point increase in EPC scores exhibits a 0.252% increase in value (β = 0.002518, p < .01), or for every ten-point increase in EPC score, there is a premium statistically significant effect of 2.52%. The size of this effect reduces at the fifth quantile to 0.18%, before increasing to 0.21% at the highest quantile ($\mathit{\tau }=0.9$). This suggests that the premium is greatest in the lowest two deciles and highest decile and remaining largely stable across the remaining deciles, albeit displaying a lower capitalisation effect.

Table 5. Quantile EPC score regression results.

	Model 22	Model 23	Model 24	Model 25	Model 26	Model 27	Model 28	Model 29	Model 30
	(tau = 0.1)	(tau = 0.2)	(tau = 0.3)	(tau = 0.4)	(tau = 0.5)	(tau = 0.6)	(tau = 0.7)	(tau = 0.8)	(tau = 0.9)
(Constant)	5.977593***	5.526232***	5.807485***	6.217973***	6.099502***	6.230759***	6.239025***	6.583708***	7.303557***
EPC Score	0.002518***	0.002138***	0.001786**	0.001879**	0.001778***	0.001807**	0.001891***	0.001993***	0.00206***
Type dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Size	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Age dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Bed dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Recep dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Grade dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Floor dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Furnished dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Set. Band dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Rural dummy	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
MDM dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R^2 (Pseudo)^1	0.402995	0.420526	0.445895	0.473427	0.501149	0.52256	0.543364	0.560415	0.56885
Adj. R^2	0.363499	0.382189	0.409237	0.43859	0.468146	0.490973	0.513154	0.531333	0.540326
Quan. dep. var	6.309918	6.388561	6.437752	6.476972	6.55108	6.620073	6.745236	6.856462	7.130899
Sparsity	0.473959	0.324275	0.281785	0.263062	0.261926	0.263455	0.294549	0.350458	0.494445
N	2,914	2,914	2,914	2,914	2,914	2,914	2,914	2,914	2,914

‘*’, ‘**’ and ‘***’ indicate statistical significance at the 10%, 5% and 1% level respectively. Parsimonious model presented. Full model including Std. errors, t-statistics and p-values, Location and time dummies available upon request. Huber SandwichStandard Errors & Covariance. Sparsity method: Kernel (Epanechnikov) using residuals. Bandwidth method: Hall-Sheather.

Figure 4. Rental market EPC score OLS and quantile percentage effects.

Discussion

Energy efficiency has emerged as a critical component of sustainable living, enhanced liveability and housing quality and comfort, lower utility and housing costs and for the reduction of greenhouse gas emissions. In the context of the private rental system, the promotion of energy efficient housing has significant and far-reaching benefits in meeting the net-zero housing challenge and the much-needed cost savings for tenants on utility bills, improving affordability and alleviating energy and fuel poverty.

Within the UK, the regulations introduced in April 2018, specified that landlords are required to have valid Energy Performance Certificates for their properties, and cannot grant new tenancies or renew existing tenancies for properties with an EPC rating below an E (unless an exemption is granted). From April 2020, this requirement was further extended to cover all existing tenancies and impending legal requirements to try and improve CO₂ emissions and achieve net zero carbon targets set by the government, mean that from 2025 all newly rented properties will be required to have an EPC rating of C or higher, with existing tenancies seeing the regulations come into effect in 2028. This study has revealed that approximately 10% of the rental units within the sample are below an EPC rating of E, and therefore below the Minimum Energy Efficiency Standards (MEES) for private rented properties. When considering the new regulations coming into effect within the next two years, this statistic increases to an alarming 61% of the rental stock within the region falling below the new energy efficiency regulation standards to be introduced.

Similar to the findings of Cajias et al. (2019), the findings emerging from this study clearly demonstrate the presence of a measurable rental premium for energy performance for B- and C-rated rentals, with ‘brown’ discounts evident for E-, F-, and G-rated rental housing. This indicates that energy efficient rental properties may have a competitive advantage in the market whereby landlords who invest in energy efficient measures can attract premiums ranging between 2.4% and 8.2%, on average, if increasing the energy efficiency from a D-rating to a C- or B-rating. This provides some evidence to help alleviate the split-incentive problem between landlords and tenants and this demonstrable link between rents and energy efficiency levels, informs landlords of the monetary or economic incentive for investing in energy efficiency. Indeed, as an example, applying the average rent within the market (£802), our estimates reveal that Landlord’s would see rent premiums range between £231 and £789 per annum, with discounts ranging between £295 and £510 per annum. Assuming the average cost of a small solar photovoltaic (PV) system installation is £4,500,⁴ the payback period would equate to just over five and a half years if obtaining the 8.2% premium for a B-rated property.

When dissecting the analysis by property type, the results showed apartments commanded both higher premiums and discounts compared to the other market segments for superior and poorer energy performance classification. This finding is parallel with previous research undertaken by McCord, Davis, et al. (2020) who observed differing property types to comprise very distinct and complex relationships in terms of price and EPC banding, with apartments demonstrating increased likelihood to be more energy efficient and detached dwellings the least efficient for commanding premiums. Further, Cerin et al. (2014) also pointed out that the relationship between energy performance and pricing is determined by housing segmentation as the energy performance relationship differs according to the type of housing and thus particular housing segments need policy targeting and support.

When further empirically examining the nature of the price and EPC relationship, the findings show that for B-rated properties, and to a lesser extent, C-rated properties, a parabolic effect can be observed across the price distribution from the lowest to the highest quantiles. This is also notable for the level of discounts attributable to E, F and G-rated rental properties. These findings clearly suggest that the premium is greatest in the lowest two deciles and is largely stable for mid-range deciles before increasing at the top decile, inferring that tenants residing in the lowest and highest priced properties capitalise energy efficiency into their housing costs – showing a willingness to pay for energy efficiency more so than tenants residing in the mid-priced range rental accommodation. This analysis is in line with the study of Fuerst et al. (2020) which indicated that renters residing within lower priced rental units generally tend to be on lower incomes and are particularly vulnerable to energy pricing changes, and therefore are more likely to want energy efficient housing to save on energy costs. In contrast, tenants with higher incomes and higher energy bills are likely to have a greater willingness to pay for energy efficiency.

The findings are suggestive that the higher and premiums and discounts for properties within the highest quartile may be a consequence of both the trade-off between a combination of upfront costs, operational savings, environmental considerations, and the perceived long-term value associated with living in an energy-efficient home. Higher priced properties often incorporate advanced energy-efficient technologies and features, thus the upfront cost of integrating these technologies can contribute to attracting environmentally conscious tenants and consequently higher rents. Moreover, whilst renters pay a premium upfront, energy-efficient properties typically result in lower monthly utility bills and provide a higher quality of living through improved indoor air quality, and enhanced comfort contributing to a healthier and more comfortable living environment. In contrast, discounts may emerge in scenarios where older period properties with original architectural features may take precedence over incorporating modern energy-efficient technologies. Also, for those on higher incomes renting expensive properties, the proportion of their income spent on utilities might be relatively low compared to the overall cost of living. In such cases, tenants may be less concerned about energy efficiency. Further, in rental market areas or sectors with heightened competition, renters may have fewer alternatives or choices, and more willing to accept lower energy efficiency. A number of these trade-offs remain symptomatic of behaviour and motivations in relation to ‘energy desire’ versus energy choice/needs.

Our results are interesting in that they clearly show that an overall analysis of the conditional mean estimate would miss the nuance at play across the quantiles in valuing energy performance. These differences are partly extent of a ‘common’ effect, such as the premium for a B-rated property, but not in all cases, with G-rated properties in some instances showing the largest discount, whilst in others a lower discount than the higher EPC rated E and F properties, a somewhat confounding result taken to its extreme in fourth quantile, where G rated properties receive a premium. In terms of the B-rated properties, in the NI and indeed UK generally, a B-rated property is a performance outlier, almost certainly heavily marketed as such, with expected running cost savings clearly articulated, and likely to include more advanced equipment and/or specific design – a ‘stand out’ property. Alternatively, G-rated properties may typically lack key components such as double glazing – in many ways falling into the ‘refurbishment’ category in market parlance, again justifying a discount. Equally however, they may be large period dwellings which carry a certain market ‘cache’ capable of muting, or even reversing the discounting effect. The similarity of B and G-rated rental dwellings is that they are extremes, at either end of the efficiency ‘spectrum’, and are easier for the market participants to comprehend and demonstrate awareness. The intermediate EPC grades are perhaps harder to distinguish from a WTP perspective, with scores driven increasingly by basic inescapable factors of construction (on the one hand) and more marginal, changeable aspects (such as additional loft lagging) on the other.

Moreover, given that not only rental properties with higher rents exhibit enhanced energy efficiency, and that premiums exist for the lowest priced rental units, there is to a degree some consciousness or awareness which exists towards energy performance certification, and this appears to have perhaps incentivised some landlords to make energy-efficient improvements and that tenants equally value the cost saving benefits. In contrast, the brown discounting evident for properties with lower EPC ratings clearly shows the need continued policy intervention and incentivisation within the sector to upgrade the energy profile of the stock. Pertinently, with our analysis showing reduced premiums, and indeed discounts within the mid-range of the rental price distribution, this is suggestive of limited capitalisation effects, consciousness and awareness for a large majority of tenants and landlords, highlighting the need for collaboration and innovative solutions to overcome the barriers to improving energy efficiency within the PRS.

This is therefore clearly positioned relative to the obvious challenges and barriers in relation to the costs associated towards the transition to a low-carbon housing economy within the rental sector. As identified by Bahareh et al. (2018) there remain several economic, technical and social barriers towards green retrofitting and mitigating carbon intensity within the housing sector and the extent to which intervention measures can be capitalised in terms of added economic value, liveability, health and wellbeing. Therefore, to combat the misalignment of incentives requires strategies which embrace, and cut across, government programmes, finance, education, and practice which are necessary for fostering awareness and change. Indeed, Government or utility programmes can offer financial support to both landlords and tenants through schemes such as grants, low-interest loans and tax credits for landlords who invest in energy upgrades and through approaches such as rebates or discounted energy rates for tenants which directly allows them to benefit from energy-related measures and ultimately participate in energy saving initiatives.

Given the substantial upfront costs, nuanced leasing agreements also have the potential to help both the landlord and tenant share the cost-benefit of energy improvements. These negotiated lease arrangements could include for example agreement to marginally increase the rent for investment in energy efficient appliances to offset costs over time. Notably, green leases as pointed out by Fuerst et al. (2020) include a cost-benefit sharing mechanism providing improved (sub)metering and measurement of a tenant’s energy consumption as well as a clause that allows landlords to pass a part of the costs of energy upgrades to the tenant. Equally, enabling tax ‘offsetting’ of energy efficiency improvements would increase landlord’s willingness to make improvements as found in existing studies (De T’Serclaes, 2007; Petrov & Ryan, 2021; Phillips, 2012; Wrigley & Crawford, 2017) as this tax restructuring would permit energy performance management to be offset against rental income could help make investment more attractive for landlords incentivising retrofit activity. These tax exemptions for energy efficiency improvements will also help with the split incentive dilemma as investment is more attractive for landlords, while reducing energy bills for renters. Consequently, enhancing energy performance necessitates a ‘win-win’ proposition within the UK. Best practices from other jurisdictions such as the Netherlands and Germany, for example, the ‘Energiesparcontracting’ scheme offers energy-efficient renovations for rental dwellings with the initial costs covered by an energy service company, with the upfront capital repaid through the achieved energy savings over time – benefitting both landlord and tenant.

Whilst the UK leasing sector has grown, it is still minor in comparison with the lifelong leasing norm in Germany. For example, considerable policy and practice reform is still required to improve living standards in UK residential leasing. Particularly as the emphasis is perhaps still more on providing warm, dry and affordable homes, which has the potential to dovetail with the energy efficiency agenda in a rare virtuous cycle. Lessons and experience from other markets are unlikely to flow naturally to the UK context and would require specific policy action and considerable engagement from what is still a very heterogenous and small-scale market. The more recent engagement of large institutional players, via the Build-to-Rent expansion is perhaps the best possibility to introduce and prove new leasing formats, but this is very much at an early stage and generally dominated by new, multi-family properties that are already at the upper end of performance in energy terms.

Conclusion

Energy efficiency within the private rental market holds immense potential to create positive outcomes for both landlords and tenants and improve the housing stock. Over the past decade, there has been an increasing policy focus targeting energy performance within the rental sector with the introduction of mandatory certification within EU Directives intended to incentivise and foster consumer behaviour and awareness. Numerous studies have examined the role of energy efficiency within the rental market from a variety of perspectives and, to some extent, are in agreement that this desired impact is not being realised, remaining partial from a behavioural, assessment and pricing standpoint.

Whilst a number of studies have shown rental premiums and the capitalisation of energy efficiency, this study has taken a nuanced approach by examining the nature of the EPC and price relationship using the quantile process to measure the EPC premium effect across the entirety of the rental price distribution. Our findings whilst showing that premiums exist, reveal that conditional quantiles are not symmetrical and that the capitalisation effect is far from complete, uniform and priced in by both tenants and landlords. We show that there are significant discounts for rental properties which are more prominent at either end of the price distribution. Especially where the market is pricing energy efficiency, and where behaviour or sentiment is accounting for a brown discount.

Importantly, moving beyond the conditional mean price estimate provides some important policy messages with the mid-range of the pricing segment of the market seemingly ‘idle’. Thus, these findings suggest that whilst consciousness and sentiment appear evident for some market participants, there remains a large segment of the market which does not price in energy efficiency. In essence, we see a misalignment which is undoubtedly due to the split-incentive dilemma whereby the cost to improve the energy efficiency requires initial (upfront) investment which can be a significant barrier for landlords, especially those with limited financial resources, and where tenants cannot afford to undertake the necessary improvements or are reluctant to as they will not benefit long-term. Further research is required to build upon the findings of this study, and from a Landlord perspective, examine typical refurbishment costs relative to the magnitude of the premiums for determining the economic viability of such implementation.

From a practice and policy perspective, our findings suggest that implementing effective strategies are needed to promote energy efficiency and pave the way for a sustainable future. These strategies include financial incentives, education campaigns, regulation, and collaboration. Most importantly, holistic strategies require more empirical depth of research to support energy efficiency in the rental market sector, which in turn will provide both short- and long-term benefits to both tenant and landlord (De T’Serclaes, 2007; Wrigley & Crawford, 2017). This research begins to meet this gap by providing useful findings as to how both performance premiums and brown discounts are distributed within the private rented sector.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Michael McCord

Michael McCord is a Reader in Real Estate valuation, finance and market analysis, and an established academic, lecturing and researching into the fields of real estate econometrics, property finance and investment. Michael has a sizeable portfolio of esteemed commissioned research projects, peer reviewed journal publications, book chapters and international conference papers, receiving a number of international best paper and reviewer awards. Michael is currently a consultant for the World Bank with reference to mass appraisal exercises and revenue estimation models and manages a number of price performance indices for measuring housing market performance. He has been involved in an extensive range of projects valuing the public sector asset register and the undertaking of statistical analysis for regional and local governments to examine the potential of policy changes. He has been the recipient of a number of prestigious international research grants and fellowships for entities such as European Bank for Reconstruction and Development, the Foundation for Ecological Security and the Omidyar Network, Lincoln Institute of Land Policy and the European and Social Research Council: Administrative Data Research Centre Northern Ireland. Dr McCord has acted as PI and Co-Investigator on seven EU Framework 7 and H2020 projects relating to urban security, the Impact of Extreme Weather on Critical Infrastructures” and the decarbonisation of the built environment.

John McCord

John McCord is Course Director for the LLM in Clinical Legal Education and Ulster Law Clinic, he has a sizeable portfolio of esteemed commissioned research projects, peer reviewed journal publications, book chapters and international conference papers in law, socio-legal and property related disciplines. This includes property tax, compulsory purchase and energy efficiency. John has been successful in securing a number of EU FP7 and Horizon 2020 research projects concerning urban security, resilience, disaster recovery and critical infrastructure protection; crime & terrorism and community policing.

Peadar Davis

Peadar Davis is a Chartered Surveyor and Senior Lecturer in Property Appraisal and Management. He was previously employed in private practice as a property valuer and asset manager specialises in providing research, consultancy and training solutions for developing/transitional jurisdictions and building bespoke databases and analysis for enhancing property tax systems. Recent property related work includes: advising the World Bank, FAO UN and the Ugandan Government on Government property valuation process modernisation and the introduction of CAMA; advising the Ethiopian Government on appropriate valuation training reform; advising the Dubai Land Authority on the creation of property market indices. He is heavily involved in EU research concerning urban security, resilience, disaster recovery and critical infrastructure protection. Peadar is the Co-Editor for the Journal of Financial Management of Property and Construction.

Martin Haran

Martin Haran is Professor of Real Estate and Urban Studies at Ulster University. He holds a First Class Hons Degree in Business and Finance and a PhD in Urban Regeneration, Development and Investment. He has an extensive research portfolio comprising over £4.5 m in secured funding from a diverse range of prestigious bodies including the European Commission, HMRC, the World Bank, the World Built Environment Forum and the European Regeneration and Development Bank. He has served as PI on more than 50 externally funded research projects. He recently led a series of work packages within the multi-award winning H2020 Carbon Risk Real Estate Monitor (CRREM) project which centred on the decarbonisation of the built environment and the promotion of environmental and social value within cities. Professor Haran has also led a series of WBEF and RICS commissioned research projects to inform urban infrastructure investment and financing models to ensure that large scale infrastructure projects serve more effectively as a catalyst for area based regeneration and renewal. Professor Haran has also collaborated extensively with the International Federation of Red Cross (IFRC) and ACAPs to champion the need for more effective harnessing of local skills and competencies to bolster community resilience in response to climate change elated events.

Graham Squires

Graham Squires is a Professor of Property Studies (Real Estate and Urban Economics) at Lincoln University, New Zealand. His research interests and publications have been in Development, Planning and Housing through the lens of Economics and Finance. In addition to journal and research report outputs, he has authored six books in the discipline areas of Economics, Property, Planning, Urban and Environmental Studies, Real Estate Development, and Construction Procurement. He has worked on a number of esteemed commissioned research projects with grant funders including The World Bank (WB), The United Nations (UN), National Government Departments, National Research Councils, and The Fulbright Commission (US-UK) – Visiting Researcher at The University of California, Berkeley. Graham serves as the Co-Editor In Chief of the journal Property Management (PM).

Notes

1. Ulster University Rental Performance Index Report H2 2023, Issue 22. Available at: https://www.ulster.ac.uk/research/topic/built-environment/research-property-planning/housing-market-reports/rental-index

2. Department for Communities, Intermediate Rent research: https://www.nihe.gov.uk/getattachment/727d6567-2df9-4e68-8aaa-f6ae3b221abe/Intermediate-Rent-Report-NI.pdf

3. See:https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/729052/EPB_Cert_Statistics_Release_-_Qtr_2_2018_final.pdf

4. Cost of PV installations derived from the https://solarnrgni.com/knowledge-hub/solar-panels-in-northern-ireland-cost-all-the-details-2023/