Mar Reguant discussion of: Policy responses to energy price shocks

Mar Reguant discussion of: Policy responses to energy price shocks Mar Reguant IAE-CSIC and Northwestern In this paper, the authors undertake a large microdata collection effort to quantify the energy shock households faced in England and Wales during the energy crisis, examining the incidence across demographics and building characteristics. Understanding the equity impacts of the energy crisis and transition is crucial. The UK context is particularly relevant, as many households struggled with energy poverty during the crisis due to a combination of severe price increases and poor building insulation. While energy use is lower for low-income households, it represents a disproportionate burden on their income. As in other countries, this crisis hit low-income households the hardest in relative terms. The Annual Fuel Poverty Statistics report estimates that while fuel poverty only increased slightly in 2022, reaching 13.4% of households in England, the monetary gap required to relieve households from energy poverty increased by 37% (Department for Energy Security and Net Zero & Department for Business, Energy & Industrial Strategy, 2023). To assess the impacts of the energy crisis, the authors develop a measurement framework combining a variety of detailed data sources to provide baseline energy consumption before the crisis. They then compute the shock households would have faced if they had maintained their usual consumption, offering a reliable estimate of the size of the ‘shock’ faced by households. The authors also construct counterfactuals to project energy shocks under various price scenarios, including changes in the United Kingdom’s energy price cap as seen in the relief policy. The authors find that wealthier regions experienced larger shocks in absolute terms, likely due to the characteristics of Economic Policy October 2024 pp. 761–763 Printed in Great Britain © CEPR, CESifo, Sciences Po, 2024. This is an Open Access article distributed under the terms of the Creative Commons AttributionNonCommercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact [email protected] Downloaded from https://academic.oup.com/economicpolicy/article/39/120/761/7816096 by Consejo Superior de Investigaciones Cientificas (CSIC) user on 12 November 2024 the properties in these areas (larger, older) and higher consumption of high-income households on average. To reach these conclusions, the study primarily relies on the Energy Performance Certificate (EPC) database, which contains over 22 million certificates providing estimates of energy expenditure for around 15 million unique properties. This represents more than 50% of the residential building stock in England and Wales. Each EPC includes model-based estimates of energy consumption for space heating, hot water generation, and lighting, derived from the building's physical characteristics, a thermodynamic modelling approach, and assumptions about occupancy. To refine these model-based engineering estimates, the study incorporates two additional data sources. First, the authors match properties to microdata on energy and gas consumption from the National Energy Efficiency Data-Framework (NEED), based on property characteristics reported in both EPC and NEED, as well as the region and consumption deciles (ranked). Second, they merge these properties with anonymized individual-level meter reading data from the Department of Business, Energy, and Industrial Strategy (BEIS), which is as granular as the postcode level (covering 15 houses). The resulting dataset is averaged, and the analysis is conducted at the Middle Layer Super Output Area (MSOA) level (6,791 areas in England) to align with available socio-economic data. My main suggestion is that the paper should preserve the data variation as much as possible, rather than averaging it out across MSOAs. Even if some socio-economic variables only vary at that level, it would still be valuable to understand how the distribution of impacts differs within areas. The paper employs impressive data collection methods and takes great care in matching these sources, but ultimately the data’s potential is somewhat limited by its aggregation at such a high level. Given the paper's focus on counterfactual energy shocks and income, I believe it would have been preferable to avoid linking the data sources at such a high level of aggregation. Although the authors keep the unit of observation at the MSOA level, I would have kept it at the postcode level, which allows for matching the energy certificate and BEIS consumption data—a rich and valuable data source. For the income variables, I would have used at least the MSOA experimental statistics provided by the Office for National Statistics (ONS), rather than relying on coarse matching with the NEED microdata. Additionally, if accessible, the Lower layer Super Output Area (LSOA) income data could be used. The issue of geographical aggregation can significantly mute our understanding of the distributional impact of the crisis, as that MSOAs mask substantial within-region heterogeneity. This effect is well-explained by Banzhaf et al. (2019), who review the problem in the context of the geographical distribution of pollution and environmental justice. As they highlight, if pollution impacts households heterogeneously within an area, but areas appear similar between them, a more aggregate analysis may substantially diminish the identified racial disparities. In the context of the energy crisis, if there is significant heterogeneity within an area, e.g., along the income vector or building characteristics, a comparison between areas will underestimate these differences. 762 MAR REGUANT Downloaded from https://academic.oup.com/economicpolicy/article/39/120/761/7816096 by Consejo Superior de Investigaciones Cientificas (CSIC) user on 12 November 2024 For future research, new income experimental data have been released at the LSOA level for Wales and England. There are 33,755 LSOAs in England and 1,917 in Wales, substantially increasing the resolution of the income variable and leveraging this valuable data source more directly. This is also the level at which fuel poverty statistics are reported, providing greater detail to understand the impact of the crisis. Additionally, one could consider using methods from Cahana et al. (2022) or other ecological literature (see King, 1997) to further refine the income estimates. The approach in Cahana et al. (2022) proposes to extend the percentile-ranking method in Borenstein (2012) to a less extreme sorting approach, based on co-variates. Another avenue for future research could involve analysing realized energy consumption at the household level in later years to document household responses to the energy shock, which could offer insights into how households adapt to changes in energy prices. As a minor note, it would be helpful to further elaborate on how the best subset approach can be used to compare across counterfactuals. Since different covariates are selected depending on the counterfactual experiment, I am not entirely confident that the coefficients between regressions can be compared in a ceteris paribus sense. As an extension, one might consider how to perform best subset selection in the presence of multiple regressions, perhaps by using machine learning selection methods that account for this structure (e.g. group lasso and other variants). More simply, one could present the same specification for the two counterfactuals. REFERENCES Banzhaf, S., L. Ma and C. Timmins (2019). ‘Environmental justice: the economics of race, place, and pollution’, Journal of Economic Perspectives, 33, 185–208. Borenstein, S. (2012). ‘The redistributional impact of nonlinear electricity pricing’, American Economic Journal: Economic Policy, 4, 56–90. Cahana, M., N. Fabra, M. Reguant, and J. Wang (2022). ‘DP17200 The Distributional Impacts of Real-Time Pricing’, CEPR Discussion Paper No. 17200. CEPR Press, Paris & London. https://cepr.org/publications/dp17200. Department for Energy Security and Net Zero & Department for Business, Energy & Industrial Strategy (2023). Annual Fuel Poverty Statistics in England, 2023 (2022 data). National Statistics, 28 February 2023, (https://assets.publishing.service.gov.uk/media/63fcdcaa8fa8f527fe30db41/ annual-fuel-poverty-statistics-lilee-report-2023-2022-data.pdf) King, G. (1997). A Solution to the Ecological Inference Problem: Reconstructing Individual Behavior from Aggregate Data, Princeton University Press, Princeton. DISCUSSION 763 Downloaded from https://academic.oup.com/economicpolicy/article/39/120/761/7816096 by Consejo Superior de Investigaciones Cientificas (CSIC) user on 12 November 2024