ESG Ratings and Real Estate Key Metrics: A Case Study

Abstract

This study examines whether and through which channels ESG ratings influence key metrics in the real estate industry. Focusing on Switzerland as a case study and concentrating on commercial real estate investors and their income properties, we utilize unique datasets and employ an OLS post-LASSO estimation procedure to identify and quantify the associations between ESG ratings and four key metrics: appraisal-based and transaction-based discount rates, rental incomes, and vacancy rates. Our results demonstrate that ESG ratings maintain a significant association with all four key metrics even after undergoing robustness checks. When dissecting the total ESG rating into its components, the environmental rating stands out as the most significant. While largely dependent on the specific metric being analyzed, the association of social and governance ratings tends to be less pronounced. Delving deeper into individual ESG rating levels, our findings suggest potential signaling effects, as properties with higher ESG ratings demonstrate heightened sensitivity to both types of discount rates and vacancy rates. Overall, our findings deepen the understanding of the association between ESG ratings and real estate markets, illuminating the intersection of sustainability and financial relevance.

1. Introduction

In 2004, the United Nations [1] introduced Environmental, Social, and Governance (ESG) ratings, reshaping the dialogue around asset sustainability. Unlike earlier frameworks like Corporate Social Responsibility (CSR) [2], ESG quantifies risks and opportunities across three key dimensions, emerging as an essential tool for assessing non-financial performance and enhancing risk management in the modern financial landscape [3]. Real estate, distinguished by its tangible and immovable nature, occupies a unique position within this paradigm. Delving deeper, the sector contends with fundamental questions regarding the essence of ESG ratings, recognizing their adaptation toward a comprehensive approach to sustainability.

Fueled by growing intra-industry interest in the association between ESG ratings and financial performance, the real estate sector has embraced this topic. Recent literature offers preliminary insights into the connection between ESG ratings and Real Estate Investment Trusts (REITs), marking a pivotal step toward understanding how sustainability metrics intersect with property investments. At the object-level, however, the association between ESG ratings and key performance metrics remains largely unexplored, prompting critical questions about their applicability and significance. Central to this inquiry is the underlying mechanism by which sustainability attributes translate into the financial performance of real estate assets, alongside the appropriate methodologies for evaluating these potential associations. The ambiguity surrounding ESG ratings is further highlighted by the complex and diverse regulatory landscape governing sustainability at both the international and national levels. While the European Union has taken a major step with the EU Taxonomy, establishing a regulated framework for measuring sustainability attributes, ESG rating methodologies remain predominantly in the hands of independent agencies, with no direct oversight from governmental or international regulatory bodies.

Amid the dynamic landscape of sustainability matters, this study examines whether and through which channels ESG ratings influence key metrics in the real estate industry. Focusing on Switzerland as a case study, we leverage the country’s regulatory enforcement of sustainability standards and concentrate on commercial real estate investors and their income properties. We utilize unique datasets comprising approximately 6300 expert-based DCF valuations and around 850 real market transactions in Switzerland spanning from 2019 to 2022. From these datasets, we analyze the association between ESG ratings and four key metrics derived from a conceptual framework designed for applicability across different countries: appraisal-based and transaction-based discount rates, rental income, and vacancy rates. The evaluation of these metrics is conducted through an ESG rating, approved by Global Real Estate Sustainability Benchmark (GRESB) and acknowledged for its objectivity and meticulous quantification consistently applied across Swiss locations. Using an OLS post-LASSO estimation procedure, we seek to identify and quantify the association between ESG ratings and key real estate metrics, employing a flexible and adaptive approach to explore the extent to which sustainability ratings intersect with real estate markets.

Our contribution can be summarized as follows. First, by disentangling total ESG ratings into their individual sub-ratings, we offer a detailed examination of the fundamental characteristics and complexities of such ratings, thereby exploring the intersection between broad sustainability coverage and targeted financial relevance. Second, leveraging appraisal and transaction processes, we enhance the understanding of the association between ESG ratings and four key metrics for income properties, contributing to the identification of influencing channels. Third, we offer new insights into a relatively unexplored type of ESG rating, advancing the field by critically assessing its relevance and practical application within the real estate industry.

The remainder of this study is structured as follows. Section 2 provides an overview of the institutional background and regulatory framework of ESG ratings, followed by a review of relevant literature in Section 3. Section 4 details the materials and methods employed in this study, including the development of a conceptual framework to identify influencing channels, and a description of the data and methodologies used. In Section 5, we present the results, with a discussion in Section 6. Finally, the study concludes in Section 7.

2. Institutional Background and Regulation of ESG Ratings

On a global scale, the emergence of ESG ratings is linked to economic crises, extreme weather events, or even wars. A prominent example is the economic crisis in 2008, which affected the public and private sectors and put investors’ investment decisions, social impacts, and respective governance practices to the fore [4]. ESG ratings also gained prominence through a key initiative by the European Commission, particularly for companies in the EU. In July 2018, the Commission established the Technical Expert Group (TEG) on sustainable finance, composed of experts from academia, industry, and finance, to develop a unified classification system for sustainable economic activities [5]. The resulting EU Taxonomy aims to create a standardized and transparent framework for determining the environmental sustainability of economic activities, guided by six key environmental objectives: (1) climate change mitigation, (2) climate change adaptation, (3) sustainable use and protection of water and marine resources, (4) transition to a circular economy, (5) pollution prevention and control, and (6) protection and restoration of biodiversity and ecosystems. The EU Taxonomy focuses on industries classified under the Nomenclature of Economic Activities (NACE) system, targeting sectors responsible for approximately 93.5% of the EU’s greenhouse gas emissions. Firms that comply with the Taxonomy’s technical screening criteria are required to disclose their proportion of Taxonomy-aligned economic performance (e.g., turnover, capital expenditure), and financial intermediaries investing in these companies can report the percentage of their investments aligned with the Taxonomy. The Taxonomy was enacted as law by the end of 2020, with compliance obligations for market participants starting in December 2021 [6].

Similar to the EU Taxonomy, the ESG rating encompasses a wide range of sustainability attributes to evaluate the long-term risk and opportunities associated with financial investments. However, and in contrast to the EU Taxonomy, ESG ratings are assessments provided by independent agencies. The intersection between ESG ratings and the EU Taxonomy has, therefore, been further explored in the literature. As previously discussed by Dumrose et al. (2022) [7], ESG ratings displays similarities to the EU Taxonomy, arising from the fact that it is primarily framed around environmental sustainability, with its current iteration concentrating exclusively on six environmental objectives. Yet, some studies approached a more granular understanding about the intersection between ESG ratings and the EU Taxonomy, contributing to the consolidation of the two concepts [8]. In particular, Sica et al. (2023) [9] compared ESG rating providers and found that environmental and social indicators align closely with both rating agencies and scientific literature, forming consistent clusters. In contrast, governance indicators remain distinct, suggesting that agencies provide unique, independent contributions in this area. Additionally, the authors found that environmental indicators align closely with the EU Taxonomy, while social and governance indicators are influenced by the European Commission’s early guidelines but lack a fully established reference framework.

Several initiatives have been implemented to enhance the transparency, accuracy, and consistency of ESG ratings. For example, the European Securities and Markets Authority (ESMA) has issued guidelines for ESG rating providers, detailing best practices for ESG rating methodologies and disclosure requirements [10]. Additionally, organizations, such as the Global Reporting Initiative (GRI), Sustainability Accounting Standards Board (SASB), and Task Force on Climate-related Financial Disclosures (TCFD), have developed frameworks for companies to report their ESG performance [11]. Overall, the determination of specific submatrices within an ESG framework hinges on the principle of materiality, with the alignment of these submatrices to financial materiality being a pivotal requirement to substantiate their relevance in evaluations conducted by investors and stakeholders in the real estate industry. Of course, the alignment necessitates ongoing adjustments to address emergent global challenges effectively [12], which also contributes to a multitude of indicators across ESG ratings furnished by different providers. As pointed out by Berg, Kölbel, and Rigobon (2022) [13], different indicators across ESG ratings also imply that they may not inherently be comparable. In fact, differences between ESG ratings may arise through different definitions of ESG ratings across providers, such as the inclusion of alternative sub-ratings and different weighting approaches [14]. Furthermore, the difficulty in comparing ESG ratings is compounded by the lack of transparency in methodologies, data sources, and weighting systems used by different providers [15].

Regulation of Sustainability Matters in Switzerland

Although Switzerland is not part of the EU and therefore not bound to the EU Taxonomy, a number of significant initiatives and regulations pertaining to sustainability have been undertaken. Early instances include the implementation of the federal energy law in 1999, emphasizing energy efficiency and renewable energies, as well as the introduction of the federal CO2 law, aiming to diminish greenhouse gas emissions in accordance with international reduction commitments [16]. In 2013, the Swiss Sustainable Construction Standard (SNBS) was introduced, which evaluates the sustainability of buildings based on criteria, such as energy, materials, ecology, and quality. In 2018, the Swiss parliament approved a revision of the Energy Law, which includes measures to promote energy efficiency in buildings. The revised law mandates that building owners disclose their buildings’ energy performance when selling or renting them and introduces financial incentives for energy-efficient renovations [17].

A pivotal action was undertaken in 2019 when the Swiss parliament approved the Swiss Climate Policy 2050, a strategy designed to achieve carbon neutrality by 2050 through the reduction of greenhouse gas emissions in Switzerland. The policy includes measures to promote energy-efficient buildings, such as stricter building codes, increased subsidies for energy-efficient renovations, and incentives for building owners to invest in renewable energy sources [18]. Recently, the Federal Council adopted the executive order on climate reporting for large Swiss companies, which came into force on 1 January 2024. Public companies, banks, and insurance companies that employ at least 500 people and have a balance sheet total of at least 20 million Swiss francs or revenues of more than 40 million Swiss francs are required to report publicly on climate issues [19].

Compliance with regulations and initiatives enacted by the Swiss Parliament and Federal Council is monitored by the Swiss Financial Market Supervisory Authority (FINMA), which defines guidelines that obligate financial institutions to consider ESG factors in their investment processes and to transparently communicate their ESG policies and practices to stakeholders. Similarly, pension funds in Switzerland are mandated by the federal act on Occupational Retirements, Survivors’ and Disability Pensions (BVG) to take ESG considerations into account when making investments. Pension funds are required to provide regular updates on the ESG performance of their portfolios and to furnish detailed information on ESG aspects of their investments. Therefore, regulations in Switzerland serve to force financial institutions and pension funds to consider ESG factors in their investment processes and to promote transparency in their ESG policies and practices [20]. Furthermore, organizations, such as Swiss Sustainable Finance (SSF), seek to provide a platform for aligning with international standards, without being bound to the EU’s framework.

As part of the financial industry, the real estate industry in Switzerland is directly affected by the enacted regulatory framework and initiatives. However, the real estate industry possesses distinct characteristics, which differentiate it from other industries. Accounting for about one quarter of total greenhouse gas emissions [21], largely due to heating and cooling systems, sustainability aspects have become paramount in the real estate industry. Furthermore, as tangible assets bound to a specific location, real estate properties may also be affected by a wide range of locational characteristics, such as the socioeconomic structure, environmental hazards, or climate change risks. In addition, real estate investments are often of a long-term nature and their consequences on their location and environment can persist beyond the investment duration.

3. Review of Literature

ESG-related studies gained cross-industry interest, and a significant portion of the literature, particularly in the real estate industry, has focused on the connection between sustainability ratings and REITs. Focusing on REITs in the United States (US) between 2000 and 2011, Eichholtz, Kok, and Yonder (2012) [22] identified a positive association between the share of Leadership in Energy and Environmental Design (LEED) and Energy Star-certified buildings and operating performance measures, such as the return on assets, return on equity, and ratio of operational to total revenue. Similarly, Fuerst (2015) [23] analyzed the association between sustainability benchmarking ratings provided by the GRESB and the financial performance of REITs. Using a dataset of 442 detailed GRESB ratings for REITs in the US, Asia, and Europe between 2011 and 2014, the study concluded that investing in sustainability enhances operational performance while reducing risk exposure and stock volatility.

ESG ratings have emerged as a key area of focus in real estate literature among the various sustainability metrics available. Brounen and Marcato (2018) [24], for instance, analyzed the performance of US REITs from 2011 to 2018, evaluating the relevance of ESG ratings from GRESB, Thomson Reuters, and KLD MSCI within an asset pricing framework. Their results indicated performance premiums associated with Reuters and KLD ratings, while GRESB ratings were correlated with lower REIT returns. Moreover, the study revealed that sub-ratings for social and governance factors positively influenced returns, whereas environmental ratings were linked to diminished returns. In a later study, Aroul, Sebherwal, and Villupuram (2022) [25] corroborated the positive relationship between ESG ratings and REIT performance. Examining publicly traded US REITs from 2019 to 2020, they demonstrated that REITs with higher ESG ratings operated more efficiently and exhibited enhanced operational performance.

A second strand of the literature centered on an object-level analysis, examining the association between sustainability ratings and various key metrics. A prevalent approach in this literature involves investigating real estate prices using hedonic pricing methods, thereby exploring the underlying pricing mechanisms [26,27,28]. The range of analyzed key metrics and their association with sustainability ratings is notably multifaceted, encompassing factors, such as occupancy rates [29,30], rental rates [31], and capitalization rates (cap rates) [32]. These studies commonly focus on eco-labels, such as LEED certification, Energy Star-certified buildings, or Energy Performance Certificates (EPCs). Furthermore, research on the connection between sustainability ratings and real estate key metrics has focused on investigating potential signaling effects, suggesting that a given association may not necessarily be linear [33].

Given the critical role of ESG ratings in complementing standard financial metrics, it is somewhat surprising that the prevalent literature offers very few results about the association between ESG ratings and real estate key metrics on an object-level. This also counts for Switzerland, where research, including Feige, McAllister, and Wallbaum (2013) [34] or Kempf and Syz (2022) [35], has concentrated on the association between rental prices and sustainability attributes or MINERGIE-certified residential properties. A somewhat broader approach was taken by Marty and Meins (2017) [36], who assessed potential rent premiums in Switzerland based on Economic Sustainable Indicators (ESIs). They found that above-average premiums are paid for factors related to “location and mobility”, such as good public transportation connections or short distances to the city center or recreational areas, and particularly for factors related to “health and comfort” (ESI represents a sustainability rating, encompassing object- and location-based indicators across five following groups: flexibility and polyvalence, energy and water dependency, location and mobility, safety and security, health and comfort). Similarly, Meins et al. (2010) [37] examined the feasibility and practicality of ESI indicators in Switzerland to improve the transparency of the Discounted Cash Flow (DCF) method for assessing property risk, particularly in determining the discount rate.

Overall, the exploration of whether and how ESG ratings may be associated with key metrics for income properties is a frontier largely unexplored, leaving a significant gap in the current understanding of sustainable finance. Leveraging comprehensive appraisal and transaction data, we meticulously analyze the association between ESG ratings and four key metrics. Moreover, by deconstructing the overall ESG ratings into their individual sub-ratings, we shed light on the intricate interplay between broad sustainability coverage and specific financial implications, providing a more granular understanding of how ESG factors may drive financial performance in real estate markets.

4. Materials and Methods

In the pursuit to identify and quantify the channels through which ESG ratings may influence real estate key metrics, we first introduce a conceptual framework to pinpoint these influencing channels (Section 4.1). We then present the estimation approach designed to assess the association between ESG ratings and real estate key metrics, highlighting its advantages and potential limitations (Section 4.2). Finally, we discuss the dataset used for the empirical analysis, detailing the ESG ratings and key real estate metrics considered, along with alternative variables utilized in the empirical investigation (Section 4.3).

4.1. Conceptual Framework to Identify Influencing Channels

Concentrating on commercial real estate investors and their income properties, our approach to identify adequate real estate key metrics leverages the evaluation of these assets. Income properties are primarily evaluated through the income approach, where market values are typically assessed by discounting future cash flows to the present. In Switzerland, the Discounted Cash Flow (DCF) method is frequently applied to evaluate the market value of income properties:

$$P{V}_{i,t}={\displaystyle \sum }_{t=1}^n{\displaystyle \frac{C{F}_{i,t}}{{\left(1+d_{i }\right)}^t}}+{\displaystyle \frac{T{V}_{i,n}}{{\left(1+d_{i }\right)}^n}}$$

(1)

where $P{V}_{i,t}$ is the resulting present value, or market value more generally, of a real estate object $i$ at time $t$, $C{F}_{i,t}$ are object-specific cash flows, and the discount rate is represented by $d_i$ [38]. The second term in Equation (1) represents the terminal value, which estimates the future sale price or residual value of the income property at the end of the holding period. Consequently, the market value of an income property is essentially driven by cash flows and the discount rate, which forms the cornerstone in the related literature exploring the connection between eco-labels. Specifically, cash flows and discount rates encompass a variety of components acquired through appraisal and transaction processes, offering a robust foundation for critically assessing their association with ESG ratings.

To objectively evaluate the market value of an income property, appraisers require detailed information about actual incomes and costs, as summarized in object-specific cash flows in the DCF framework. Among these, actual rental incomes and vacancy rates are particularly suitable for analyzing their association with ESG ratings, given their direct linkage to real estate markets and their reflection of market participants’ immediate responses to location-specific sustainability criteria. Rental incomes, in this regard, may be associated with ESG ratings through real estate owners or investors in this regard, passing on the quality of a given location on local rental prices, which is capitalized by agents’ willingness to pay through hedonic mechanisms [34]. Alternatively, ESG ratings may also affect object-specific cash flows via a demand-side effect induced through the increased quality of a given location, potentially influencing vacancy rates [39]. In addition to these potential effects arising from object-specific cash flows, an association with ESG ratings may also be mediated through the discount rate [32]. The discount rate reflects the probability with which the expected rental income will be realized and corresponds to the return that a given investor can expect under normal market conditions. The discount rate is generally composed of a risk-free rate ($R{F}_t$), capturing the compensation for lost liquidity, a compensation for inflation ($I_t$), and an object-specific risk premium for risk-taking ($R{P}_{i,t}$) [40]:

$$d_{i }=R{F}_t+I_t+R{P}_{i,t }$$

(2)

The risk premium stands out as particularly intriguing, acknowledging that the real estate literature commonly agrees on its association with object- and location-specific characteristics [41,42]. If ESG ratings are perceived to be associated with lower risks, higher future cash flows, and improved sustainability, they can lead to lower discount rates. Conversely, if ESG ratings are perceived to be associated with higher risks, lower future cash flows and reduced sustainability, they can increase discount rates [43].

Throughout this study, we will analyze appraisal-based ($d_i^{apr}$) and transaction-based ($d_i^{trx}$) discount rates. Whereas appraisal-based discount rates are determined by appraisers using the DCF method in real estate valuations, transaction-based discount rates are implicitly derived from market transactions and reflect investors’ perception of risks at a given point in time. While appraisal-based discount rates facilitate the examination of whether appraisers consider ESG-related indicators when determining discount rates, the analysis of transaction-based discount rates enables us to infer whether investors prioritize these indicators when making investment decisions, eventually through a mark-up in the object-specific risk premium.

4.2. Method

In order to identify and quantify the connection between ESG ratings and real estate key metrics, our estimation procedure combines Ordinary Least Squares (OLS) with Least Absolute Shrinkage and Selection Operator (LASSO). Specifically, we employ an OLS post-LASSO estimation methodology [44], wherein we use LASSO for variable selection and subsequently employ OLS to quantify the connection among ESG ratings, their corresponding sub-ratings, and predefined key metrics. The employment of the LASSO estimation technique in our analysis serves a dual purpose. Firstly, it facilitates a meticulous evaluation of these established drivers, and secondly, it systematically examines the association of ESG ratings, including their sub-ratings, with the designated outcome variables. This hybrid approach integrates a priori variable selection based on intuition and theory with a data-driven model, offering a robust framework for investigating the association between ESG ratings and the predefined outcome variables.

Our baseline estimation equation is summarized as follows:

$$Y_{i,t}={\alpha }_0+{\gamma }_i·ES{G}_{i,t}+{\delta }_i·{\psi }_{i,t}+{\vartheta }_{i,t}$$

(3)

where $Y_{i,t}$ denotes the key metrics encompassing appraisal-based and transaction-based discount rates, along with rental incomes and vacancy rates. $ES{G}_{i,t}$ reflects either the total ESG rating or sub-ratings for the pillars $E_{i,t}$, $S_{i,t}$, and $G_{i,t}$. ${\psi }_{i,t}$ summarizes control variables, and ${\vartheta }_{i,t}$ is an error term.

The LASSO method extends OLS estimation by augmenting the model with a penalty term on the sum of the absolute coefficients. This penalty term facilitates variable selection by setting many potential explanatory variables to have zero coefficients [45]. In particular, applying a linear LASSO solves the following absolute value function:

$$\widehat{\beta }=\underset{\beta }{argmin}\left({\displaystyle \frac{1}{n}}{\displaystyle \sum }_{i=1}^n{\left(Y_i-X_i{\beta }^{\prime }\right)}^2+\lambda {\displaystyle \sum }_{j=1}^p\left|{\beta }_j\right|\right)$$

(4)

where $\beta =\left[{\alpha }_0,{\gamma }_1, {\gamma }_2,\dots ,{\delta }_1, {\delta }_2,\dots \right]$ and $X_i$ represents $p$ potential covariates in the dataset of $n$ observations and $Y_i$ is the outcome variable. In Equation (5), $\lambda$ represents the LASSO penalty term ($\lambda >0$), which belongs to the tuning parameters in our estimation procedure [46]. All variables have been standardized according to the $z$-score normalization. The optimal value of the regularization parameter $\lambda$ was determined using 10-fold cross-validation, extracted from the fitted LASSO model. In fact, applying $k$ = 10 in $k$-fold cross-validation can be seen as common practice in the empirical literature [47]. To enhance the robustness of our findings, we conduct a thorough examination (Section 5.4) using alternative variable selection methods. This evaluation assesses the consistency of selected variables across various cross-validation settings, contributing to the reliability of our results.

Upon identifying the pertinent regressors selected through the LASSO estimation procedure ($l$), we incorporated them into an OLS:

$$Y_{i,t}={\alpha }_0+{\gamma }_i·ES{G}_{i,t}^l+{\delta }_i·{\psi }_{i,t}^l+{\vartheta }_{i,t}$$

(5)

A list of all variables incorporated in our estimation procedure is attached in Appendix A.1 (

Table A1. Variable classification.

Variable Name	Description	Classification
ESG rating	Total location-based ESG rating (Min = 1, Max = 5)	Continuous
E-rating	Location-based rating for the environment (E) (Min = 1, Max = 5)	Continuous
S-rating	Location-based rating for the society (S) (Min = 1, Max = 5)	Continuous
G-rating	Location-based rating for the governance (G) (Min = 1, Max = 5)	Continuous
$d_i^{apr}$	Appraisal-based discount rate	Continuous
$d_i^{trx}$	Transaction-based discount rate	Continuous
Rental_I_sqm	Effective rental income in Swiss Francs (CHF) derived from accounting figures divided by the total usable area	Continuous
Vacancy_R	Vacancy rate derived from accounting figures	Continuous
Age	Difference between the construction year of a given object and the valuation date	Continuous
LogSqm	Logarithm of the usable area, measured in square meters (sqm)	Continuous
Micro_L	Rating for the location of a given object within a given municipality (Min = 1, Max = 5)	Continuous
Macro_L	Locational rating characterizing the municipality of a given object municipality (Min = 1, Max = 5)	Continuous
Object_R	Standardized rating characterizing the overall quality of given object (Min = 1, Max = 5)	Continuous
Standard_R	Standardized rating characterizing the standard of a given object (Min = 1, Max = 5)	Continuous
Condtion_R	Standardized rating characterizing the condition of a given object (Min = 1, Max = 5)	Continuous
IR_20	Yields on governmental bonds with 20-year maturity	Continuous
Inflation	Inflation according to the consumer price index	Continuous
YD_19	Year dummy for the year 2019	Dummy
YD_20	Year dummy for the year 2020	Dummy
YD_21	Year dummy for the year 2021	Dummy
YD_22	Year dummy for the year 2022	Dummy
Type_D1	Real estate type dummy for residential real estate properties	Dummy
Type_D2	Real estate type dummy for mixed usages (residential, industrial, or office purposes)	Dummy
Type_D3	Real estate type dummy for industrial real estate properties	Dummy
Type_D4	Real estate type dummy for special usages	Dummy
CT_D1	Regional type dummy for canton Zurich	Dummy
CT_D2	Regional type dummy for canton Berne	Dummy
CT_D3	Regional type dummy for canton Lucerne	Dummy
CT_D4	Regional type dummy for canton Uri	Dummy
CT_D5	Regional type dummy for canton Schwyz	Dummy
CT_D6	Regional type dummy for canton Obwalden	Dummy
CT_D7	Regional type dummy for canton Nidwalden	Dummy
CT_D8	Regional type dummy for canton Glarus	Dummy
CT_D9	Regional type dummy for canton Zug	Dummy
CT_D10	Regional type dummy for canton Fribourg	Dummy
CT_D11	Regional type dummy for canton Solothurn	Dummy
CT_D12	Regional type dummy for canton Basel-City	Dummy
CT_D13	Regional type dummy for canton Basel-Country	Dummy
CT_D14	Regional type dummy for canton Schaffhausen	Dummy
CT_D15	Regional type dummy for canton Appenzell Ausserrhoden	Dummy
CT_D16	Regional type dummy for canton Appenzell Innerrhoden	Dummy
CT_D17	Regional type dummy for canton St. Gallen	Dummy
CT_D18	Regional type dummy for canton Graubünden	Dummy
CT_D19	Regional type dummy for canton Aargau	Dummy
CT_D20	Regional type dummy for canton Thurgau	Dummy
CT_D21	Regional type dummy for canton Ticino	Dummy
CT_D22	Regional type dummy for canton Vaud	Dummy
CT_D23	Regional type dummy for canton Valais	Dummy
CT_D24	Regional type dummy for canton Neuchâtel	Dummy
CT_D25	Regional type dummy for canton Geneva	Dummy
CT_D26	Regional type dummy for canton Jura	Dummy

Notes: the table specifies the variables (names, descriptions, and classification) for expert-based DCF valuations and real market transactions between 2019 and 2022 applied in the OLS post-LASSO regression approach, as discussed in Section 6. Variables are specified according to Table 2 and are defined identically in both datasets used for the empirical analysis.

). While our estimation approach, guided by intuition and theory, employs a comprehensive OLS post-LASSO regression model with cross-validation to explore the associations between ESG ratings and key metrics of income properties, it is crucial to acknowledge the inherent challenges in attributing causal relationships in this study. In fact, the country-wide obligation for commercial real estate investors to report such ratings, along with the distinctives of ESG ratings as outlined in Section 4.3, forms essential components within the framework for assessing their connection to income properties’ key metrics. However, with the emergence of sustainability criteria and ESG ratings, they face opposition from traditionally heterogeneous perceptions among market participants in real estate markets. For instance, there might be a scenario where an association between vacancy rates and rental incomes with ESG ratings only emerges for particular “green” tenants who either choose to locate themselves in sustainable areas or are willing to pay a rent premium. Additionally, appraisers and investors may implicitly incorporate factors contained in ESG ratings, predating the formal conceptualization and regulation of these ratings. This underlines the importance to control for a multitude of empirically grounded variables that collectively influence the key metrics under consideration, all of which are assessed through LASSO, to mitigate omitted variable bias and achieve conditional exogeneity.

4.3. Data

In the context of this study, the relevance of sustainability criteria influencing Swiss real estate markets can be traced back to the year 2019 when the Swiss parliament approved the Swiss Climate Policy 2050. Subsequently, sustainability criteria gained in importance among commercial real estate investors, and its increase in significance further accelerated with the Federal Council’s recent adoption of the executive order on climate reporting for Swiss companies, effective from the start of 2024. Building on this, the empirical part of this study analyzes the time frame between 2019 and 2022 using two unique datasets—a dataset capturing expert-based valuations and a dataset containing real market transactions. The geographical distribution of the real estate objects analyzed is displayed in Figure 1.

Both datasets provide detailed information on the real estate key metrics identified in Section 4.1, along with additional valuable object-specific data that are incorporated into the empirical analysis. To assess the association between ESG ratings and appraisal-based discount rates, rental incomes, and vacancy rates, we use a dataset capturing approximately 6300 expert-based DCF valuations of commercial investors’ income properties. The effectiveness of the DCF methodology stems from the combination of appraisers’ knowledge and detailed, high-quality data, including rent rolls, which provide comprehensive information on actual rental incomes and vacancy rates. The second dataset contains roughly 850 income properties that have been sold in open-market transactions. For each income property from the second dataset, an expert-based DCF valuation exists, together with transaction data providing information about transaction-based discount rates.

The datasets have been enriched with an ESG rating, as outlined in

Table 1. ESG rating description.

ESG Sub-Rating	Criteria	Indicators
Environment indicators (E)	Climate change and risks	Heat days in 2020 (RCP45), difference in heat days between 2060 and 2020 (RCP45), cooling degree days in 2020 (RCP45), difference in cooling degree days between 2060 and 2020 (RCP45).
	Greenery and sealing	Proportion of shrubs, proportion and diversity of greening, surface sealing.
	Mobility	Public transport quality class, car sharing locations, future public transport infrastructure, public e-car charging stations.
	Resource utilization	Population density.
Social indicators (S)	Health and well-being	Road noise during the day, road noise at night, railway noise during the day, railway noise at night, aircraft noise, long-term air pollution index, radon.
	Safety and natural hazards	Hail, flood, fall, landslide, storm, debris flow, avalanche, earthquake, surface runoff.
	Socioeconomic structure	Diversity of household sizes, diversity of age structure, diversity of socio-economic milieus, diversity of income, price range distribution.
	Building stock	Diversity of building categories, mix of dwelling sizes, diversity of building ages.
	Recreational areas	Recreational and green areas, public meeting places.
Governance indicators (G)	Real estate market	Supply ratio, vacancy rates.
	Renewable energy	Energy city label, utilization of solar potentials.
	Spatial planning	Building permits, conversion shares in building permits, densification potential.

Notes: the table specifies each sub-rating, respective criteria, and indicators entering the ESG rating. More detailed information, including the sources and weightings of each indicator contributing to the ESG rating, is provided in Appendix A.2 (Table A2). Source: Wüest Partner AG.

. Similar to most ESG ratings in the real estate industry, the ESG ratings considered consist of multiple indicators grouped under the three core pillars, emphasizing the need for transparency in their disclosure [48]. Environmental indicators encompass a wide array of measures related to climate change and associated risks, as well as multifaceted indicators of greenery, land sealing, and mobility aspects. Social indicators address aspects directly linked to society, including health and well-being, safety and natural hazards, socioeconomic structure, building stock, and recreational areas. In contrast, governance indicators focus on aspects related to subordinate developments, such as current trends in local real estate markets, initiatives toward renewable energy, and spatial planning efforts. Further information about the ESG rating considered is provided in Appendix A.2 (

Table A2. Detailed ESG rating description.

ESG Sub-Rating	Criteria (Weight)	Indicator	Radius	Source
Environment Indicators (E)	Climate change and risks (12)	Heat days in 2020 (RCP45)	Point	National Centre for Climate Services
		Difference in heat days between 2060 and 2020 (RCP45)	Point	National Centre for Climate Services
		Cooling degree days in 2020 (RCP45)	Point	National Centre for Climate Services
		Difference in cooling degree days between 2060 and 2020 (RCP45)	Point	National Centre for Climate Services
	Greenery and sealing (12.5)	Proportion greening	50 m	LFI
		Diversity greening	50 m	LFI
		Surface sealing	50 m	Copernicus
	Mobility (12.5)	Public transport quality class	Point	ARE
		Car sharing locations	2000 m	BAV, BAKOM, Wüest Partner
		Future public transport infrastructure	Municipality	Wüest Partner
		Public e-car charging stations	4000 m	BFE
	Resource utilization (10.5)	Population density	300 m	BFS
		Employment density	300 m	BFS
Social Indicators (S)	Health and well-being (12.5)	Road noise during the day	Point	BFS
		Road noise at night	Point	BFS
		Railway noise during the day	Point	BFS
		Railway noise at night	Point	BFS
		Aircraft noise	Point	Wüest Partner
		Long-term air pollution index	Point	Meteotest
		Radon	Point	BAG, Wüest Partner
	Safety and natural hazards (12.5)	Mudslide	Point (50 m buffer)	BAFU, Geotest AG, Wüest Partner
		Hail	Point	MeteoSchweiz, Wüest Partner
		Flood	Point (50 m buffer)	BAFU, Geotest AG, Wüest Partner
		Fall	Point (50 m buffer)	BAFU, Geotest AG, Wüest Partner
		Landslide	Point (50 m buffer)	BAFU, Wüest Partner
		Storm	Point	BAFU, Wüest Partner
		Debris flow	Point (50 m buffer)	BAFU, Geotest AG, Wüest Partner
		Avalanche	Point (50 m buffer)	BAFU, Geotest AG, Wüest Partner
		Earthquake	Point	BAFU, Geotest AG, Wüest Partner
		Surface runoff	Point (50 m buffer)	BAFU, Geotest AG, Wüest Partner
	Socioeconomic structure (9)	Diversity of household sizes	300 m	BFS STATPOP, Wüest Partner
		Diversity of age structure	300 m	BFS STATPOP, Wüest Partner
		Diversity of socio-economic milieus	300 m	Microm, Wüest Partner
		Diversity of income	300 m	Wüest Partner
		Price range distribution	Municipality	Wüest Partner
	Building stock (7.5)	Diversity of building categories	300 m	GWR, Wüest Partner
		Mix of dwelling sizes	300 m	GWR, Wüest Partner
		Diversity of building ages	300 m	GWR, Wüest Partner
	Recreational areas (6)	Recreational and green areas	Municipality	BFS
		Public meeting places, e.g., fire pits and playgrounds	1000 m	BAV, BAKOM, Wüest Partner
Governance Indicators (G)	Real estate market (2)	Supply ratio	Municipality	Wüest Partner
		Vacancy rates	Municipality	Wüest Partner
	Renewable energy (1.5)	Energy city label	Municipality	Energiestadtlabel
		Utilization of solar potentials	Municipality	Energyreporter geoimpact AG
	Spatial planning (1.5)	Building permits, densification potential	Municipality	Wüest Partner
		Densification potential population	Municipality	Wüest Partner
		Densification potential employees	Municipality	Wüest Partner
		Conversion shares in building permits	Municipality	Wüest Partner

Notes: the table summarizes each sub-rating of the total ESG rating, respective criteria, and indicators. Weightings for each sub-rating are tailored to regional characteristics, with a relatively higher emphasis on environmental and social indicators compared to governance. This approach reflects regional diversity, where urban areas consistently achieve higher governance ratings. The column “radius” characterizes the spital scale at which the respective indicator is measured. “Point” refers to the process of verifying the property’s coordinates against the corresponding raster value. “Point (50 m buffer)” entails pre-buffering the raster, selecting the point where the property’s coordinates intersect with the raster. Distances are measured in meters (m), while spatial political ranges (“Municipality”) represent the range within which the respective indicator is recorded. RCP45 represents the scenario RCP 4.5, which predicts a global warming of 2.6 degrees Celsius by the year 2060. Abbreviations used: BFS = Federal Statistical Office, LFI = National Forest Inventory, ARE = Federal Office for Spatial Development, BAV = Federal Office of Transport, BAKOM = Federal Office of Communications, BFE = Federal Office of Energy, BAFU = Federal Office for the Environment, GWR = Federal Register of Buildings and Dwellings. Source: Wüest Partner AG.

).

The considered ESG rating aims to provide a comprehensive assessment of sustainability for any given location in Switzerland, incorporating distinctive data beyond the conventional characteristics typically used to describe a location. It includes indicators based on expert classifications, such as raster data for natural hazards and air pollutants, as well as benchmark-based indicators. From these indicators, an overall ESG rating, along with sub-ratings for the E, S, and G pillars, is calculated on a scale from 1 to 5, with 5 representing the highest value. These ratings are subsequently compared and reclassified using precalculated ESG ratings for 2 million Swiss buildings, yielding the total ESG rating and sub-ratings for each pillar.

As a data-driven assessment, this ESG rating enhances objectivity by resisting influence from stakeholders involved in the appraisal or transaction process, unlike object-specific ESG ratings, which may be more prone to such biases. This objectivity improves comparability across different locations and has led to the rating’s approval by GRESB. Although this study does not delve into the inclusion of specific indicators, the ESG rating provides a robust foundation for evaluating its association with key real estate metrics, incorporating a diverse set of factors commonly associated with sustainability in the real estate industry.

When compared to the EU Taxonomy, the ESG ratings reveal distinct approaches to environmental, social, and governance indicators. The EU Taxonomy emphasizes a broad regulatory framework focused on climate change mitigation, energy efficiency, and biodiversity protection. In contrast, the ESG rating underlying this study offers a more detailed assessment of specific climate metrics, urban infrastructure, and community health. In terms of environmental indicators, both frameworks prioritize climate change, but the ESG rating uniquely addresses local conditions, such as heat days and urban greenery. For social factors, the ESG rating provides deeper insights into health, well-being, and socioeconomic diversity, while the EU Taxonomy offers broader social safeguards. Regarding governance, the EU Taxonomy focuses primarily on regulatory compliance, whereas the ESG rating emphasizes real estate market health, renewable energy initiatives, and urban planning efforts. Overall, while both frameworks are aligned with sustainability goals, the EU Taxonomy adopts a broader, regulatory approach, whereas the ESG rating is more granular and specifically adapted to the real estate sector.

In our estimation procedure to assess the association between ESG ratings and key real estate metrics, we incorporate various control variables, acknowledging that these outcome variables are typically influenced by a range of additional factors.

Table 2. Variable description.

Variable Name and Description	Source
ESG rating Total location-based ESG rating (Min = 1, Max = 5)	Wüest Partner AG
E-rating Location-based rating for the environment (E) (Min = 1, Max = 5)	Wüest Partner AG
S-Rating Location-based rating for the society (S) (Min = 1, Max = 5)	Wüest Partner AG
G-Rating Location-based rating for the governance (G) (Min = 1, Max = 5)	Wüest Partner AG
$d_i^{apr}$ Appraisal-based discount rate	Wüest Partner AG
$d_i^{trx}$ Transaction-based discount rate	Wüest Partner AG
Rental_I_sqm Effective rental income in Swiss Francs (CHF) derived from rent rolls divided by the total usable area	Wüest Partner AG
Vacancy_R Vacancy rate derived from rent rolls	Wüest Partner AG
Age Difference between the construction year of a given object and the valuation date	Wüest Partner AG
LogSqm Logarithm of the usable area, measured in square meters (sqm)	Wüest Partner AG
Micro_L Rating for the location of a given object within a given municipality (Min = 1, Max = 5)	Wüest Partner AG
Macro_L Locational rating characterizing the municipality of a given object (Min = 1, Max = 5)	Wüest Partner AG
Object_R Standardized rating characterizing the overall quality of a given object (Min = 1, Max = 5)	Wüest Partner AG
Standard_R Standardized rating characterizing the standard of a given object (Min = 1, Max = 5)	Wüest Partner AG
Condtion_R Standardized rating characterizing the condition of a given object (Min = 1, Max = 5)	Wüest Partner AG
IR_20 Quarterly yields on governmental bonds with 20-year maturity	Swiss National Bank
Inflation Yearly inflation according to the consumer price index	Federal Statistical Office

Notes: variable names, descriptions, and sources for the subsequent empirical analysis. In both datasets used, the variables are defined identically.

summarizes the relevant variables, while summary statistics for both datasets are attached in Appendix A.3 (

Table A3. Summary statistics—valuation dataset.

Variable	Min.	Max.	Mean	Median	Std. Dev.
$d_i^{apr}$	1.50	6.00	2.99	2.95	0.45
Rental_I_sqm	30.48	788.36	230.70	216.49	86.67
Vacancy_R	0.00	100.00	4.33	1.45	8.55
Macro_L	1.18	5.00	4.37	4.53	0.62
Micro_L	1.00	5.00	3.58	3.50	0.62
Quality_R	1.18	5.00	3.42	3.40	0.49
Standard_R	1.30	5.00	3.36	3.30	0.53
Condition_R	1.00	5.00	3.50	3.40	0.75
Age	1.00	822.00	52.21	48.00	42.64
Sqm	130.00	97425.00	4209.61	2418.75	5996.55
IR_20	−0.41	1.20	0.15	−0.06	0.52
Inflation	−0.70	2.80	1.16	0.60	1.38
ESG rating	1.00	5.00	3.89	4.10	0.91
E-rating	1.00	5.00	3.61	3.80	1.04
S-rating	1.10	5.00	3.66	3.70	0.90
G-rating	1.00	5.00	3.70	4.00	1.00

Notes: summary statistics reflect variations across expert-based DCF valuations between 2019 and 2022. The dataset is used in the empirical investigation to analyze the connection between ESG ratings and appraisal-based discount rates, rental incomes, and vacancy rates. Variables are defined according to Table 2.

and

Table A4. Summary statistics—transaction dataset.

Variable	Min.	Max.	Mean	Median	Std. Dev.
$d_i^{trx}$	1.36	6.85	2.74	2.71	0.61
Macro_L	1.20	5.00	4.20	4.40	0.82
Micro_L	1.80	5.00	3.68	3.60	0.63
Quality_R	2.30	4.88	3.39	3.32	0.47
Age	0.00	599.00	58.88	51.00	47.90
Sqm	77.00	73383.00	3060.90	1731.30	5054.85
IR_20	−0.41	1.20	0.23	0.06	0.57
Inflation	−0.70	2.80	1.13	0.60	1.48
ESG rating	1.00	5.00	3.85	4.20	1.04
E-rating	1.00	5.00	3.48	3.6	1.15
S-rating	1.30	5.00	3.68	3.80	0.92
G-rating	1.00	5.00	3.55	3.70	1.11

Notes: summary statistics reflect variations across real market transactions between 2019 and 2022. The dataset is used in the empirical investigation to analyze the connection between ESG ratings and transaction-based discount rates. Variables are defined according to Table 2.

).

Considering first appraisal-based and transaction-based discount rates. As outlined in Equation (2), from a theoretical standpoint, discount rates encompass a risk-free rate, an inflation compensation, and a risk premium. To approximate the risk-free rate, we incorporate yields from government bonds with a 20-year maturity. We have opted for this extended maturity period under the presumption that longer government bond maturities and the associated higher interest rates more accurately reflect investors’ return expectations in a low-interest-rate environment. We consider the yearly inflation rate to represent the compensation for inflation. Including the yearly inflation rate corresponds with the accessible data available to appraisers and investors, offering a pragmatic approach to assess the direct effects of inflation on future cash flows.

Besides these macroeconomic variables affecting both appraisal-based and transaction-based discount rates, we approximate the risk premium using an array of locational and object-specific variables. To account for locational characteristics, we include the macro- and micro-location rating. Whereas the macro-location rating is a standardized assessment that reflects the quality of a municipality as a whole, the micro-location rating specifies the precise location of an income property within a particular municipality. The macro-location rating classifies the municipality, capturing fundamental elements, such as the municipality type, tax burden, and price developments. Both the macro- and micro-location rating are tied to the specific type of a given income property, encompassing condensed information about locational characteristics that traditionally affect key metrics. To control for the overall quality of a given object, we incorporate the object-specific quality rating, which is a standardized rating defined by any appraiser during the assessment of a given object. Besides these standardized ratings, we control for the object-specific age and the size, which are variables that traditionally drive discount rates [44].

In our empirical investigation to analyze the role of ESG ratings on rental incomes, we control for variables that are generally applied in hedonic estimation procedures [49]. We incorporate both the macro-location and micro-location rating to consider local quality criteria that influence rental prices. Because rental prices are primarily influenced by distinct quality criteria, we consider the overall condition and the standard of a given object. Considering the standard and condition of an object provides a more detailed and granular understanding of the rental object’s attributes, allowing for accurate price differentiation between objects. We additionally incorporate the specific age and size of a given object to more effectively account for variations in rental incomes across different locations.

Concerning vacancy rates, we incorporate the macro-location and micro-location rating to control for local characteristics that influence local vacancy rates. We also consider the object-specific quality rating to accommodate differing tenant preferences among locations, which can impact vacancy rates. Controlling for the object-specific age and size follows a similar approach.

Across all key metrics under consideration, we include dummy variables controlling for seasonality, regionality, and real estate type-specific differentials. Seasonality is addressed with the inclusion of year dummy variables, accounting for significant events between 2019 and 2022, such as the emergence of the COVID-19 pandemic in 2020 and changes in monetary policy, including increasing target interest rates in 2022. Real-estate-type dummy variables control for variations in key metrics inherently tied to specific property categories due to structural circumstances. Examples include the persistent rent control system in Switzerland, allowing property owners or tenants to adjust rental incomes according to the reference index or the Consumer Price Index (CPI) [50], the historically high share of Swiss tenants resulting in low vacancy rates in the residential real estate segment, and the risk premium asked by commercial investors for this particular segment (the reference index is a weighted average of mortgage interest rates in Switzerland). Regional dummy variables, including those for the 26 cantons, accommodate Switzerland’s spatial political structure. This incorporation addresses the pronounced heterogeneity within Switzerland’s regulatory landscape, arising from the Swiss federalist system, which grants each canton the authority to establish distinct locally binding regulations [51]. As a result, the key metrics considered exhibit significant fluctuations across regions.

5. Results

The results are organized according to the key metrics identified in the study. Section 5.1 presents the findings related to appraisal-based and transaction-based discount rates, while Section 5.2 focuses on rental incomes and vacancy rates. In Section 5.3, we extend the empirical analysis by exploring the association between key metrics across different rating levels. Section 5.4 provides a robustness check through the assessment of alternative model specifications.

5.1. Appraisal-Based and Transaction-Based Discount Rates

Considering first the findings presented in

Table 3. ESG ratings and discount dates.

Dependent Variable:	$\mathbf{Log} ({\mathit{d}}_{\mathit{i}}^{\mathit{a}\mathit{p}\mathit{r}}$)	Log (${\mathit{d}}_{\mathit{i}}^{\mathit{a}\mathit{p}\mathit{r}}$)	Log (${\mathit{d}}_{\mathit{i}}^{\mathit{t}\mathit{r}\mathit{x}}$)	$\mathbf{Log} ({\mathit{d}}_{\mathit{i}}^{\mathit{t}\mathit{r}\mathit{x}}$)
Model:	1	2	3	4
Constant	1.558 ***	1.553 ***	1.333 ***	1.318 ***
	(0.018)	(0.019)	(0.070)	(0.072)
Macro_L	−0.056 ***	−0.050 ***	−0.031 ***	−0.014
	(0.003)	(0.003)	(0.011)	(0.013)
Micro_L	−0.032 ***	−0.031 ***	−0.042 ***	−0.038 ***
	(0.002)	(0.002)	(0.010)	(0.010)
Quality_R	−0.041 ***	−0.042 ***
	(0.003)	(0.003)
Age	−0.0004 ***	−0.0003 ***	−0.001 ***	−0.001 ***
	(0.00003)	(0.00003)	(0.0001)	(0.0001)
LogSqm	0.022 ***	0.021 ***	0.016 **	0.012 *
	(0.001)	(0.001)	(0.007)	(0.007)
IR_20	−0.045 ***	−0.045 ***	−0.039 ***	−0.040 ***
	(0.004)	(0.004)	(0.013)	(0.013)
Inflation	0.002	0.002
	(0.002)	(0.002)
ESG rating	−0.032 ***		−0.042 ***
	(0.002)		(0.007)
E-rating		−0.024 ***		−0.053 ***
		(0.002)		(0.007)
S-rating		−0.004 **
		(0.002)
G-rating		−0.011 ***		−0.006
		(0.002)		(0.007)
Regional dummies	Yes	Yes	Yes	Yes
Year dummies	Yes	Yes	Yes	Yes
Type dummies	Yes	Yes	Yes	Yes
Observations	6166	6166	836	836
Adjusted $R^2$	0.593	0.598	0.435	0.45
Residual Std. Error	0.094 (df = 6127)	0.093 (df = 6130)	0.159 (df = 812)	0.156 (df = 799)

Notes: the results displayed are derived from an OLS post-LASSO estimation procedure, in which we employ LASSO for variable selection and OLS for quantification. In the LASSO variable selection procedure, the optimal value of the regularization parameter $\lambda$ was determined using 10-fold cross-validation. The optimal value of λ is 0.0020 in Model 1, 0.0012 in Model 2, 0.0103 in Model 3, and 0.0010 in Model 4. The year, type, and regional dummies are included in each model. The time frame under consideration refers to the years from 2019 to 2022. Variables are defined according to Table 2. Signif. codes: ‘***’ 0.001, ‘**’ 0.01, ‘*’ 0.05, ‘.’ 0.1

, which summarizes the results from the OLS post-LASSO estimation procedure for appraisal-based and transaction-based discount rates, in Model 1, the total ESG rating is selected by LASSO and significantly contributes to the equation for appraisal-based discount rates, suggesting that sustainability characteristics, on an aggregated level, are positively associated with appraisal-based discount rates. The results indicate that a one-unit increase in the total ESG rating is associated with an approximate 3.2% decrease in appraisal-based discount rates. To achieve a nuanced understanding of the association between ESG ratings and appraisal-based discount rates, Model 2 applies our OLS post-LASSO estimation procedure to the sub-ratings of the total ESG rating, specifically focusing on the environment (E-rating), society (S-rating), and governance (G-rating). Apparently, all sub-ratings have been selected by LASSO and display a significant inverse connection with appraisal-based discount rates. Among those ratings, appraisal-based discount rates appear to be particularly sensitive to changes in the E-rating, whereas the S-rating and G-rating exhibit comparatively smaller sensitivities. These results suggest that environmental factors may have a more significant influence on appraisal-based discount rates, while social and governance factors likely play a subordinated role.

The relevance of environmental factors in explaining appraisal-based discount rates is supported by the significant association with the micro- and the macro-location rating. In Models 1 and 2, both locational quality criteria enter significantly into the estimation equations. Similarly, the object-specific quality rating displays a notable sensitivity to appraisal-based discount rates, suggesting that a considerable share in the variation of appraisal-based discount rates hinges on the object quality itself. The age is characterized by a counterintuitive estimate, as one would expect the risk premium to increase with the age of a given real estate object. Explanations may be found in the DCF method, wherein depreciation is explicitly accounted for over time, stable income that is independent of object characteristics is considered, and routine maintenance and renovations are already factored into the object-specific quality rating. Conversely, appraisal-based discount rates appear to increase with the size of an object, indicating an increasing risk premium associated with relatively sizable real estate objects.

The risk-free rate enters the estimation equations in Models 1 and 2 with a counterintuitive sign. Throughout the period under consideration, interest rates predominantly remained in the negative territory, particularly in 2021, before experiencing a subsequent increase in 2022. Average discount rates, in contrast, consistently decreased during the same time frame. Due to the conventional time lag in the responsiveness of discount rates to changes in the interest rate environment, the influence of interest rates on discount rates may have become attenuated during the period under investigation. The same holds true for inflation, which has a minor and insignificant effect on appraisal-based discount rates.

In Models 3 and 4, we examine the association between transaction-based discount rates and ESG ratings. Compared to appraisal-based discount rates, transaction-based discount rates reflect investors’ perceptions, providing information on whether ESG ratings may be reflected in the risk premium of a given real estate object. In Model 3, the total ESG rating emerges as a relevant predictor for transaction-based discount rates, having been selected by LASSO, and revealing an inverse association between the two variables.

Notably, transaction-based discount rates exhibit a more pronounced sensitivity to variations in the total ESG rating compared to appraisal-based discount rates. A one-unit increase in the ESG rating is associated with a decrease in transaction-based discount rates of approximately 4.2%. Furthermore, in Model 4, we include ESG sub-ratings in our OLS post-LASSO estimation, revealing that the predominant association of the total ESG rating observed in Model 3 primarily stems from the E-rating. The relevance of the S-rating is not retained by LASSO, resulting in its attenuation as a regressor in the model, while the G-rating does not significantly enter the estimation equation. This observation highlights a significant limitation of relying solely on associations resulting from aggregated ratings, which may obscure important variations within the data and potentially serve as a foundation for aggregation errors.

Overall, the association of ESG ratings with transaction-based discount rates mirrors the pattern observed with appraisal-based discount rates, with environmental ratings serving as the primary driver of the aggregated association detected. In contrast to appraisal-based discount rates, transaction-based discount rates exhibit a lower sensitivity to variations in the macro-location rating. Nonetheless, it is noteworthy that the micro-location rating displays an increased degree of responsiveness in relation to transaction-based discount rates, suggesting that investors place a heightened emphasis on the precise geographical location of a particular real estate object compared to the overall rating of a given municipality. Different weightings between appraisers and investors also become apparent considering the object-specific quality rating or inflation, which have been excluded by LASSO from the estimation equation for transaction-based discount rates.

5.2. Rental Incomes and Vacancy Rates

Table 4. ESG ratings, rental incomes and vacancy rates.

Dependent Variable:	Log (Rental_I_sqm)	Log (Rental_I_sqm)	Vacancy_R	Vacancy_R
Model:	5	6	7	8
Constant	4.441 ***	4.374 ***	8.859 ***	8.966 ***
	(0.061)	(0.061)	(1.655)	(1.703)
Macro_L	0.075 ***	0.069 ***	−0.746 **	−0.694 **
	(0.011)	(0.011)	(0.295)	(0.277)
Micro_L	0.094 ***	0.089***	−0.521 ***	−0.481 ***
	(0.007)	(0.007)	(0.183)	(0.184)
Standard_R	0.090 ***	0.091 ***
	(0.010)	(0.010)
Condition_R	0.076 ***	0.081 ***
	(0.007)	(0.007)
Quality_R			0.547 **	0.518 **
			(0.237)	(0.237)
Age	−0.0004 ***		−0.003	−0.003
	(0.0001)		(0.003)	(0.003)
LogSqm	−0.069 ***	−0.063 ***	0.366 ***	0.341 ***
	(0.005)	(0.005)	(0.126)	(0.126)
ESG rating	0.063 ***		−0.809 ***
	(0.007)		(0.185)
E-rating		0.043 ***		−0.797 ***
		(0.006)		(0.146)
S-rating		0.009 *		−0.149
		(0.006)		(0.149)
G-rating		0.017 ***
		(0.006)
Regional dummies	Yes	Yes	Yes	Yes
Year dummies	Yes	Yes	Yes	Yes
Type dummies	Yes	Yes	Yes	Yes
Observations	6261	6261	6166	6166
$\mathrm{Adjusted} R^2$	0.256	0.256	0.058	0.060
Residual Std. Error	0.317 (df = 6228)	0.317 (df = 6225)	8.284 (df = 6131)	8.277 (df = 6127)

Notes: the results displayed are derived from an OLS post-LASSO estimation procedure, in which we employ LASSO for variable selection and OLS for quantification. In the LASSO variable selection procedure, the optimal value of the regularization parameter $\lambda$ was determined using 10-fold cross-validation. The optimal value of λ is 0.3901 in Model 5, 0.3238 in Model 6, 0.0483 in Model 7, and 0.0412 in Model 8. The year, type, and regional dummies are included in each model. Variables are defined according to Table 2. Signif. codes: ‘***’ 0.001, ‘**’ 0.01, ‘*’ 0.05, ‘.’ 0.1.

summarizes the results of applying our empirical strategy to analyze the association among ESG ratings, rental incomes, and vacancy rates. Examining first the results for rental incomes, as displayed in Models 5 and 6, Model 5 reveals that the total ESG rating is directly associated with rental incomes. Notably, a one-unit increase in the ESG rating is linked to an average increase of approximately 6.3% in rental incomes. In Model 6, we further decompose the association of the total ESG rating into its individual sub-ratings. Similar to appraisal-based discount rates (Table 3), all sub-ratings are selected by LASSO. A substantial portion of the association between the total ESG rating and rental incomes is primarily mediated through the E-rating. Rental incomes display a relatively small sensitivity to changes in the G-rating, which becomes even smaller for the S-rating. While the finding that rental income associations are primarily mediated through environmental factors is well-documented in prior studies [36,37], we hypothesize that the reduced sensitivity of social and governance aspects may result from varying perceptions of these attributes among market participants.

The responsiveness of rental incomes to various influencing factors becomes evident through the remaining LASSO-selected variables in Models 5 and 6. While the significant association with both micro- and macro-location ratings unveils rental incomes’ sensitivity to locational characteristics, tenants’ willingness to pay for unique object-specific attributes is illustrated by the direct association with standard and condition ratings. Interestingly, the object-specific age is not a significant variable in the estimation equation, while rental incomes inversely correlate with the object-specific size. This suggests that newer properties do not necessarily command higher rents, potentially due to overriding factors, such as the location or amenities, and indicates a decreasing rent premium for larger properties.

In Models 7 and 8, we analyze the association between ESG ratings and vacancy rates. Our OLS post-LASSO estimation procedure selects the total ESG rating as a relevant variable for object-specific vacancy rates. A one-unit increase in the total ESG rating is associated with an approximate 0.8 percentage point decrease in average vacancy rates. In Model 8, we break down the total ESG rating and asses the relevance of ESG sub-ratings to vacancy rates. We find that the E-rating remains the most influential predictor among all ESG sub-ratings, whereas the S-rating exhibits an insignificant estimate, and the G-rating has not been selected by LASSO. We conjecture that this observation may be attributed to the underlying mechanisms governing real estate markets, where location attractiveness—largely influenced by the quality-related factors encompassed in the E-rating—exerts a positive demand-side effect, thereby influencing vacancy rates.

The significance of locational quality criteria also becomes evident in the estimation equations for vacancy rates. An increase in both the micro and macro-location ratings is associated with a decrease in vacancy rates. Furthermore, our results suggest nuanced interplays among object-specific variables in explaining vacancy rates. Surprisingly, the object-specific quality rating enters into the estimation equations for vacancy rates with a positive sign, which suggests that even high-quality objects may experience vacancies. Conversely, our results reveal that vacancy rates are not consistently affected by the object-specific age but tend to increase with the size of a given real estate object.

5.3. Exploring ESG Rating Levels

Our prior analyses have shed light on the connection between ESG ratings and the pertinent metrics, operating under the assumption of a linear association. Yet, beyond this linear perspective, ESG ratings may implicitly signal value, implying that their connection with outcome variables could differ across varying rating levels [33]. To further explore potential signaling effects of both high and low ESG ratings, Figure 2 summarizes the results from our established estimation procedure, illustrating the association between various ESG rating levels (e.g., below 3, between 3 and 4, and above 4) and the previously discussed real estate key metrics.

In Panel A, the analysis reveals a pronounced sensitivity between ESG ratings and appraisal-based discount rates, with the most substantial association observed in the upper bound, particularly for ESG ratings exceeding 4. Intermediate ratings (between 3 and 4) do not exhibit a significant association with appraisal-based discount rates, though there is a slight increase in significance for ESG ratings in the lower bound (below 3). A similar pattern emerges in Panel B for transaction-based discount rates, highlighting that the primary association is particularly pronounced in properties with ESG ratings above 4. Since the existing Swiss regulatory framework imposes obligations on commercial real estate investors, the inclusion of risk discounts in discount rates is likely indicative of a future-proofing effect for sustainable investments, as previously reported by Kempf and Syz (2022) [35], which may reveal an implicit risk discount for strategic investment decisions towards sustainable locations.

Applying the same approach to rental incomes (Panel C) suggests that particularly low levels of ESG ratings are associated with increasing rental incomes. We speculate that this might suggest that rental incomes in the upper range of the ESG rating already account for locational sustainability characteristics. As a result, the association between higher ESG ratings and rental incomes appears marginal, limiting the potential for additional mark-ups on rental prices, especially given the constraints of Switzerland’s rent-control system.

Panel D highlights a notable sensitivity of vacancy rates to higher ESG ratings (above 4), providing evidence of a higher signaling effect for heightened ESG ratings. We posit that this increased sensitivity is likely attributed to properties with higher ESG ratings attracting greater demand, potentially leading to a reduction in vacancy rates. Thus, tenants sorting themselves around sustainable locations may be inherently linked to the fundamental mechanisms composing real estate markets, where local quality characteristics are transmitted through object-specific cash flows via a demand-side effect.

5.4. Assessing Model Specifications

In order to further test the obtained results, we apply additional selection methods to assess the validity of LASSO-selected variables, as displayed in

Table 5. Assessing variable selection.

Dependent Variable	$\mathbf{Log} ({\mathit{d}}_{\mathit{i}}^{\mathit{a}\mathit{p}\mathit{r}}$)				$\mathbf{Log} ({\mathit{d}}_{\mathit{i}}^{\mathit{t}\mathit{r}\mathit{x}}$)				Log (Rental_I_sqm)				Vacancy_R
Model	1		2		3		4		5		6		7		8
Method	EN	AL	EN	AL	EN	AL	EN	AL	EN	AL	EN	AL	EN	AL	EN	AL
Macro_L	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
Micro_L	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
Quality_R	Y	Y	Y	Y	N	N	N	N					Y	Y	Y	Y
Standard_R									Y	Y	Y	Y
Condition_R									Y	Y	Y	Y
Age	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	N	Y	Y	Y	Y
LogSqm	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
IR_20	Y	Y	Y	Y	Y	Y	Y	Y
Inflation	Y	Y	Y	Y	N	N	N	N
ESG rating	Y	Y			Y	Y			Y	Y			Y	Y
E-rating			Y	Y			Y	Y			Y	Y			Y	Y
S-rating			Y	Y			N	N			Y	Y			Y	Y
G-rating			Y	Y			Y	Y			Y	Y			N	N

Notes: selected variables (Y = Yes, N = No) by Elastic Net (EN) and Adaptive Lasso (AL) using 10-fold cross-validation to determine the optimal regularization parameter $\lambda$. With the exception of applied selection methods, selected variables are the results of the same estimation procedure as displayed in Table 3 (Models 1 and 2 for appraisal-based discount rates, Models 3 and 4 for transaction-based discount rates) and in Table 4 (Models 5 and 6 for rental incomes, Models 7 and 8 for vacancy rates). The variables are defined according to Table 2.

. We first perform a robustness check using Elastic Net regression, which is a regularization technique that combines the penalties of both Ridge regression and LASSO. Specifically, Elastic Net penalizes the absolute size of regression coefficients (LASSO penalty) and the squared size of coefficients (Ridge penalty). This combination allows Elastic Net to select variables like LASSO while also handling multicollinearity more effectively, akin to Ridge regression [52].

Next, we employ the Adaptive LASSO as an additional variable selection method. The Adaptive LASSO extends the conventional LASSO methodology by incorporating data-driven weights for each predictor variable during the regularization process. In contrast to standard LASSO regression, which uniformly penalizes all predictors, the Adaptive LASSO assigns unique weights to each predictor based on their estimated coefficients derived from an initial model. This adaptive approach employs cross-validation to iteratively fit the model, enabling the determination of adaptive weights by varying the regularization parameter λ [53]. This proves especially advantageous in scenarios with collinearity or varying importance among predictors.

While LASSO’s stronger regularization generally leads to a more stringent variable selection process—resulting in fewer non-zero coefficients and a sparser model—both the Elastic Net and Adaptive LASSO methods offer distinct advantages in managing collinearity and highlighting variable importance. As illustrated in Table 5, Elastic Net and Adaptive LASSO produce comparable variable selection outcomes. Despite differences in the selection methods, the qualitative consistency of the results reaffirms the reliability of the selected variables.

6. Discussion

The question of whether and through which channels ESG ratings may be associated with key metrics in the real estate industry is intrinsically linked to the fundamental dynamics governing real estate markets. The integration of sustainability concerns into financial decision-making processes, particularly in real estate, requires an understanding of both regulatory frameworks and market forces that convert non-financial attributes, such as environmental, social, and governance considerations, into measurable financial outcomes. This study, although focused on Switzerland, offers insights with broad relevance, as the underlying principles transcend national boundaries and can inform global investigations into the intersection of sustainability ratings and financial performance across various markets.

In real estate markets, the transformation of sustainability attributes into financial materiality is critically dependent on regulatory policies and market-driven mechanisms. Recent developments in Switzerland serve as a valuable case study to demonstrate how sustainability factors, such as ESG ratings, are progressively embedded into financial practices. By exploring Switzerland’s regulatory landscape, where sustainability considerations are becoming central to investment strategies, this study highlights how ESG ratings can impact key income-related factors, such as rental rates, through hedonic pricing models, and demand-side dynamics, as evidenced by variations in vacancy rates. Additionally, the study explores potential associations with financial materiality, mediated through discount rates in both appraisals and transactions. These processes reflect broader global trends in which regulatory efforts increasingly position sustainability at the heart of financial and investment decision-making.

The study’s conceptual framework, detailed in Section 4.1, offers a robust and adaptable theoretical foundation that accommodates the complex interactions between ESG ratings and financial metrics. This framework is not limited to the Swiss market but can be reutilized for diverse regulatory environments and real estate markets, allowing for a more comprehensive understanding of how ESG ratings influence financial performance internationally. By capturing these interactions, the framework establishes a foundation for future research, enabling scholars and practitioners alike to explore the intricate ways in which sustainability is integrated into the financial fabric of real estate and beyond, under varying regulatory and market conditions.

An important feature of this study is its methodological approach, which integrates theoretical insights with a data-driven model. The combination of variable selection methods, such as the LASSO technique, with subsequent OLS estimation, proves effective in addressing the inherent complexity and ambiguity found in ESG ratings. This methodology allows us to navigate the uncertainties surrounding sustainability ratings with their inconsistent regulatory frameworks, while still delivering reliable preliminary results about their relevance and practical application within the real estate industry. This adaptability is crucial as sustainability ratings remain fluid across different industries and regions, with varying degrees of regulation and standardization.

However, a critical limitation of this research—and indeed, any study attempting to correlate ESG ratings with financial metrics—is the availability of granular data. In addition to ESG ratings and real estate metrics, a range of object-level control variables must be accessible to mitigate omitted variable bias and ensure conditional exogeneity. Without access to detailed, high-quality data on these variables, any analysis risks oversimplifying the association between sustainability ratings and real estate key metrics.

7. Conclusions

In an era increasingly dedicated to sustainability, our study has endeavored to assess the intricate association between sustainability and real estate assets. Drawing insights from the Swiss regulatory framework, we leveraged appraisal and transaction processes to identify key metrics potentially associated with ESG ratings, including appraisal-based and transaction-based discount rates, rental incomes, and vacancy rates. Guided by comprehensive datasets featuring 6300 expert-based valuations and around 850 real market transactions, we applied an OLS post-LASSO estimation procedure to identify and quantify the association between ESG ratings and real estate key metrics.

At an aggregated level, our results indicate that the total ESG rating is significantly associated with the key metrics considered. Our results argue in favor of a robust connection between ESG ratings and all key metrics considered. Specifically, we observed a sensitivity of 3.2% in appraisal-based discount rates and 4.2% in transaction-based discount rates in response to a one-unit change in the total ESG rating. We also identified a direct connection between ESG ratings and rental incomes. A one-unit increase in the total ESG rating has the potential to increase average rental incomes by approximately 6.3%. Similarly, the total ESG rating is significantly connected to object-specific vacancy rates, resulting in a reduction of 0.8 percentage points in response to a one-unit increase in the total ESG rating.

Dissecting the total ESG rating into its sub-ratings reveals that the overall association is primarily mediated through the environmental sub-rating. Among all key metrics considered, this sub-rating demonstrates the highest sensitivity and establishes the most significant associations. Notably, the relevance of the social and governance sub-ratings varies substantially across the key metrics under consideration. While LASSO identifies all ESG sub-ratings in the context of appraisal-based discount rates, indicating that appraisers consider a diverse range of factors when evaluating a real estate object, transaction-based discount rates are predominantly influenced by the environmental sub-rating. Conversely, in the case of rental incomes, robust associations are mainly evident for the environmental and the governance sub-rating, while for vacancy rates, the environmental sub-rating is the sole sub-rating selected by LASSO and characterized by a significant association. These results question the intersection between broad sustainability coverage and targeted financial relevance, suggesting that both perspectives do not necessarily align and vary depending on the key metric being considered.

In our further exploration of potential signaling effects at different levels of total ESG ratings, we observed that high ESG ratings, particularly those exceeding 4, exhibit an increased sensitivity in the context of appraisal-based and transaction-based discount rates, as well as in vacancy rates. However, for rental incomes, we did not observe a consistent association between higher ESG ratings and increased rental levels. The detection of potential signaling effects for specific key metrics highlights that an association with sustainability attributes may not be linear, further underscoring the varied perceptions of sustainability attributes among market participants.

Several parallels can be drawn when comparing our results to the existing literature, especially regarding the sign and magnitude of previously documented findings. Among the few empirical findings about the connection between sustainability ratings and discount rates, our results validate their inherent inverse connection to the risk premium, as reported by McGrath (2013) [32]. Furthermore, our results align with a broader body of evidence on rent premiums, supporting findings from both international studies [31] and national investigations [34,35,36]. While there is a limited benchmark for vacancy rates, our study advances this area by reporting a consistent inverse association between ESG ratings and vacancy rates, mirroring the observations of Fuerst and McAllister (2009) [29] and Holtermans and Kok (2017) [30].

Building on our findings and anticipating upcoming regulatory efforts to incorporate and report sustainability ratings, we advocate for the standardization of ESG rating. Specifically, we emphasize the development of transparent frameworks together with unified methodologies for measuring financial relevance alongside adequate sustainability indicators to foster transparency and the effective management of sustainability issues. Thus, we encourage policymakers to develop comprehensive guidelines on the scope of sustainability ratings and their practical application in object valuations and investment decisions. Consequently, future research should further explore the fundamental mechanisms through which sustainability ratings may be associated with real estate markets, thereby deepening our understanding of this critical issue.