The Distribution of Energy Efficiency and Regional Inequality.

AuthorSinghal, Puja

    Residential heating accounts for one-fifth of the final energy consumed in Germany (BMWi, 2019) and continues to be primarily powered with fossil fuels. (1) The initial objective for energy efficiency policies in leading industrialized nations was to reduce the foreign dependence on fossil fuels (i.e. energy security). Now, energy efficiency measures play an increasing role in achieving greenhouse gas emission reductions globally. However, the design of policy instruments such as building standards, subsidized retrofit programs, and mandatory energy performance certificates tend to lack information on the socioeconomic distribution of energy efficiency technology and the energy savings achieved. (2)

    Here we take a long-term perspective and assess the net outcomes of major policy efforts (energy standards for new construction and renovations, financial subsidies for home retrofits) that have already taken place in Germany. The main goal of the paper is to measure and describe how energy efficiency outcomes are distributed in the existing building stock. By focusing on the existing building stock, with an average year of construction of 1967, we circumvent the problem of measuring energy efficiency outcomes over a short-time frame as highlighted by Kotchen (2017). We investigate simultaneously both the geographic and socioeconomic distribution to shed light on regional inequality in energy efficiency, in particular the East-West divide in Germany. (3) This spatial disaggregation allows us to gain some insight into the mechanisms driving estimated differences in energy efficiency.

    We propose a new methodology to estimate the energy performance of buildings. We combine annual heating bills for a large sample of the existing building stock in Germany with daily temperature data from weather stations to estimate the causal response of heat energy demand to fluctuations in the total annual sum of heating degree days (HDD). Using this building-level response of heat energy consumption to temperature as an overall measure of energy efficiency, we investigate some of the main drivers of the heterogeneous response to temperature shocks. To this end, we employ both fixed effects regressions and causal forests to explore all observable dimensions in a systematic manner. We thus demonstrate how our method makes it possible to accurately assess building-level performance (as an alternative to energy performance certificates), which can then be used to identify buildings most in need for energy efficiency measures.

    Housing-related energy policies may have particularly strong distributional effects and thus may have the additional objective of economic equity, such that the costs and benefits of environmental policy are allocated progressively across income groups or regions (Bento, 2013). Evidence suggests that investments towards energy efficiency vary significantly by income groups and thus the improvements in energy intensity are likely distributed unequally as well. Borenstein and Davis (2016) and Jacobsen (2019) show that financial incentives such as income tax credits for home weatherization, hybrid or electric cars, solar systems, and energy-efficient appliances are predominantly received by higherincome households in the United States. McCoy and Kotsch (2021) investigate the returns to energy efficiency investments by income group and find that the poorer households experience lower energy savings both in the short- and long-run in the United Kingdom. Similarly, Bruegge et al. (2019) find that building codes in California did lead to energy savings for the lower income groups, but this was due to decrease in square footage rather than improvements in building energy efficiency attributes.

    We add to the literature on the distributional impacts of energy policies in the residential building sector and come to a number of conclusions. Overall, we do find that regions with the lowest unemployment rates are home to a disproportionately higher share of energy-efficient buildings. However, the zip codes associated with the highest unemployment rates (regions in the East), also benefit from the use of both carbon-intensive heating fuel type and energy-efficient buildings, owing to larger buildings, renovation efforts, and efficient construction that took place post-reunification in East Germany between 1990 and 2001. For instance, we document that the older building stock (built before 1975) in the East of Germany is significantly more energy efficient than the comparable group in the West. At the same time, buildings in East Germany are on average larger than those in West Germany, and this is further associated with lower annual energy demand per square meter of living space in response to an increase in heating degree days. A closer look reveals that the geographic distribution of energy efficiency outcomes is well-explained by regional differences in the intensity of the winter heating season across Germany.

    In this study, we further argue that temperature change is a dominant factor in explaining the observed decline in heat energy use by the German housing sector in the last decade, and discuss potential implications for energy efficiency investment incentives. Germany experienced an unwavering increase in average temperatures in the past decade from 2010 to 2019. (4) Unsurprisingly, the heating sector is particularly affected by global warming because the demand for energy to heat homes falls, ceteris paribus. (5) We quantify the fall in energy demand for space heating due to climate change (8 percent from 2014 to 2018), and thus the potential weakening economic incentives for homeowners to invest in thermal-efficiency of the existing building stock.

    The next section discusses the unique data used in this paper. Section 3 reports on the underlying trends in heat energy demand and heating degree days. Section 4 explains the identification strategy and estimates the average response of energy demand to temperature. Section 5 considers the heterogeneous responses to temperature, especially the regional distribution of building energy efficiency. Section 6 offers a forest-based machine learning method to estimate building-level marginal effects to further understand the sources of the heterogeneity in the response of heat demand to temperature. Section 7 concludes.

  2. DATA

    The analysis in this paper is based on data from three sources: (1) data on building-level heating bills and energy performance certificates from a leading energy-metering company, (2) weather station data from the German Weather Service (Deutscher Wetterdienst), and (3) socio-de-mographic data from RWI-GEO-GRID (RWI and microm 2020).

    2.1 Heating Bills

    The primary data used come from a large panel of building-level heating bills for 420,573 residential buildings (3,215,800 bills) in Germany, with 12-month billing cycles that start during January 2008 to June 2018. 12-month billing means that all heating bills are for either 365 or 366 days, but the billing start and end dates vary.

    The billing dataset contains information on the actual (metered) units of energy consumed for space heating and water heating, along with yearly costs incurred. The billing data also contains important building characteristics that help determine the energy requirements of buildings: living space (in square meters), building size (in number of apartments), location by zip code, and main heating fuel type.

    The main dependent variable is the annual quantity of heating energy consumed per square meter of heated living space in a building. We calculated this variable in several steps: first, building-specific consumption values are limited to the amounts of energy used for heating space (excluding warm water). Second, the metered consumption value is multiplied by the net calorific value corresponding to the building's fuel type, giving us the absolute heat energy consumption in kilowatt-hours (kWh) for a building during the annual billing period. Third, we divide total kilowatt-hours consumed by the amount of heated living space in the building. The units are therefore, kilowatt-hours per square meter of heated living space per year (kWh/[m.sup.2]a). We calculate the average price per kWh of heat energy billed by dividing total energy costs reported on each bill by the annual units of kWh consumed.

    To create the estimation sample, we only consider heating bills from buildings that use either natural gas, heating oil, or district heating as the main fuel type, which is 98 percent of all buildings observed. We trim the sample further by removing the top and bottom 1% tails from the distribution of heat energy consumption, used as the main dependent variable: such that consumption is above 30 kWh/[m.sup.2]a and below 400 kWh/[m.sup.2]a of heated living space. See Figure A3 for the distribution of the heat energy consumption variable, trimmed above 600 kWh/[m.sup.2]a.

    Finally, we only consider those buildings observed at least two times in the (unbalanced) panel. After these steps, the full sample consists of 384,223 buildings with a total of 3,030,063 observed heating bills. On average we observe a building 9 times, minimum number of 2 times, and a maximum number of 11 times.

    2.2 Energy Performance Certifcates

    We also observe energy performance certificates (EPC), for about 40 percent of the full sample of buildings, issued from 2008 to 2019. We do not observe whether an EPC is consumption-based or an engineering estimate of the energy requirements of a building, however. (6) EPC data allow us to observe important measures of the thermal efficiency attributes of the buildings, including the energy performance score, construction year of the building, year (or retrofit year) of the heating system, roof, loft ceiling, exterior wall, windows, and basement ceiling. For energy performance certificates issued from 2014 to 2019 (about 20% of the sample), we...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT