Inequality in leisure across gender and income groups over time: Evidence from individual RUT subsidy data
Economists and social scientists have long been interested in how home and market production are organized. Leisure and home production are central to understanding a wide range of issues related inequality in living conditions, in addition to differences in income and wealth. How do individuals distribute their time between work and leisure and what role do differences in these play for inequality in living conditions?
This project uses new data to answer these classic questions based on a tax change - the so-called RUT deduction - that was introduced on July 1, 2007. The reform entailed a 50% subsidy of the labor cost for domestic services. One consequence of the reform is that there is now a unique dataset which contains third-party reported consumption of housing services, which otherwise is generally measured using surveys that have issues with selection and measurement error.
Our project will use the RUT data to analyze how demand for RUT services depends on the characteristics of individuals. By buying household services, people can increase their available time. We then use a demand analysis to estimate different individuals valuation of their leisure time. By analyzing the variation in people s reactions to RUT, we can also estimate elasticities for labor supply for different types of individuals. Finally, we study how inequality measures change when one takes into account differences in leisure and home work and the distribution effects of the RUT reform.
Final report
Project Purpose and Development
This project aimed to investigate how individuals allocate their time between leisure, household work, and market labor and how this allocation influences economic inequality and living conditions. We utilized a unique natural experiment—the introduction of the Swedish tax subsidy (RUT-avdrag)—to examine both short- and long-run behavioral responses to substantial relative price changes in domestic services. As we began to analyze the data, however, our preliminary findings caused us to shift the focus of our study. In particular, we became interested in the question: How strongly do people respond to tax changes? Since taxes are typically levied on market transactions, the answer depends crucially on the substitution possibilities between market and non-market activities. Understanding this substitution margin is central for evaluating a wide range of policies, from minimum wage laws to income taxes and childcare subsidies. Thus, we could leverage our findings on RUT-avdrag to shed light on these other topical policy proposals. Policy analysis requires estimates of long-run elasticities that incorporate equilibrium responses. However, most empirical work relies on short-run partial equilibrium (PE) responses, often with an empirical design featuring a "missing intercept" problem (Wolf, 2023), precluding the identification of the full general equilibrium (GE) effects. Thus, our primary objective then turned to developing an empirical strategy to quantify these short- and long-run elasticities to deepen understanding of market and home production interactions.
Brief Description of Implementation
How big is the difference between short-run PE responses and long-run GE responses? This project made progress on this question by studying a large, long-lasting, national level, tax reform in Sweden to examine both short-run and long-run behavioral responses to changes in the relative price of household services. In July 2007, Sweden introduced the RUT tax credit, which provided a 50 percent subsidy for household services, including cleaning, with an annual cap of 50,000 SEK. The project capitalized on detailed Swedish registry data encompassing the entire adult population, integrated with comprehensive records of domestic service purchases under the RUT reform. Using the full population administrative data that tracks individual consumption of subsidized services through social security-number linked tax information, we document that long-run responses are dramatically larger than short-run responses. The one year elasticity is around 0.8, whereas at ten years the elasticity is about 6. The elasticity continues to rise until the end of our sample: in 2010 about 5% of Swedish adults had used the RUT deduction, but there was a gradual and nearly constant rise to 40% by 2020.
How should we conceptualize this phenomenon? Exploiting the richness of the data, we first show that the slow adoption is fundamentally about understanding extensive margin decisions. There are large heterogeneities in how much different RUT users spend, but all of the increase in RUT usage over time is accounted for by a higher share of the population using it, with stable average spending conditional on use. From this, we conclude that a model of the RUT usage should feature an extensive margin. Moreover, we show that this extensive margin decision looks like an adoption decision. Before using RUT, there is a very low probability that you will start using it. However, once you have used RUT, there is a high probability that you will keep using it. We further show that this slow adoption is not driven by selection over time in who adopts. The usage for adopters remains constant over time. This is despite the usage being quite different in the cross section. To model this we use a technology adoption framework. We estimate a Hidden Markov Model (HMM) to characterize consumer adoption choices and persistence over time. The implementation of the HMM combined empirical microeconomic analysis, extensive data collection, and advanced econometric modeling.
Significant Results and Discussion
Our analysis reveals several key findings. First, the adjustment process is dominated by the extensive margin, with conditional usage remaining stable over time. Second, we find no evidence that early adopters are "high-value" users; while income predicts usage, both usage levels and adoption patterns remain consistent over time. Third, the data strongly suggest that taking advantage of the RUT credit represents an adoption decision, akin to the diffusion of new technologies.
We demonstrate that a three-state hidden Markov model successfully captures the key features of the adoption process. The model reveals highly persistent states with slow transition rates, implying that the journey to steady state is extraordinarily long, with a half-life of approximately 6.5 years. Information frictions cannot explain this slow adjustment, as the reform was well-publicized. Furthermore, we show that supply-side constraints are also unlikely to be behind the results, given that service providers rapidly entered the market. Rather, the patterns align with gradual changes in household production technology and social norms surrounding outsourcing household work.
Our findings carry important implications for policy analysis and welfare calculations. The stark difference between short-run and long-run responses suggests that standard empirical approaches focusing on short-run elasticities may significantly underestimate the eventual impact of permanent policy changes affecting the margin between home and market production. While we concentrate on a specific reform, our insights likely have broader applicability to policies that require extensive margin adjustments or adoption decisions, such as childcare subsidies or earned income tax credits.
New Research Questions
Several new research questions emerged, notably:
• What specific informational, normative, or market frictions underpin the observed slow adoption?
• How do behavioral patterns differ when comparing domestic service subsidies to other labor market interventions?
• What broader implications might similar adoption dynamics hold for other policy contexts, such as childcare subsidies or Earned Income Tax Credits?
Dissemination and Collaboration
Research results were actively disseminated through presentations at major international conferences, including the Society for Economic Dynamics Annual Meeting 2024 in Barcelona, Spain, and through invited seminars at leading research institutions worldwide. The research team collaborated internationally, notably with researchers at the University of Minnesota and the University of Edinburgh. Our presentations have sparked interdisciplinary discussions around policy design and economic inequality.
List of Publications
Working Papers:
• Boppart, T., Gottfries, A., Malmberg, H., & Mitman, K. "How Long is the Long Run? Theory and Evidence from a Swedish Tax Reform"
Conference Presentations:
• Society for Economic Dynamics Annual Meeting, 2024.
Links the personal websites of the authors are available below:
• Kurt Mitman (http://www.kurtmitman.com)
• Timo Boppart (https://www.econ.uzh.ch/en/people/faculty/boppart.html)
• Axel Gottfries (https://sites.google.com/site/axelgottfries/)
• Hannes Malmberg (https://sites.google.com/site/hannesmalmbergweb/)
This project fully complies with RJ's open access requirements.
This project aimed to investigate how individuals allocate their time between leisure, household work, and market labor and how this allocation influences economic inequality and living conditions. We utilized a unique natural experiment—the introduction of the Swedish tax subsidy (RUT-avdrag)—to examine both short- and long-run behavioral responses to substantial relative price changes in domestic services. As we began to analyze the data, however, our preliminary findings caused us to shift the focus of our study. In particular, we became interested in the question: How strongly do people respond to tax changes? Since taxes are typically levied on market transactions, the answer depends crucially on the substitution possibilities between market and non-market activities. Understanding this substitution margin is central for evaluating a wide range of policies, from minimum wage laws to income taxes and childcare subsidies. Thus, we could leverage our findings on RUT-avdrag to shed light on these other topical policy proposals. Policy analysis requires estimates of long-run elasticities that incorporate equilibrium responses. However, most empirical work relies on short-run partial equilibrium (PE) responses, often with an empirical design featuring a "missing intercept" problem (Wolf, 2023), precluding the identification of the full general equilibrium (GE) effects. Thus, our primary objective then turned to developing an empirical strategy to quantify these short- and long-run elasticities to deepen understanding of market and home production interactions.
Brief Description of Implementation
How big is the difference between short-run PE responses and long-run GE responses? This project made progress on this question by studying a large, long-lasting, national level, tax reform in Sweden to examine both short-run and long-run behavioral responses to changes in the relative price of household services. In July 2007, Sweden introduced the RUT tax credit, which provided a 50 percent subsidy for household services, including cleaning, with an annual cap of 50,000 SEK. The project capitalized on detailed Swedish registry data encompassing the entire adult population, integrated with comprehensive records of domestic service purchases under the RUT reform. Using the full population administrative data that tracks individual consumption of subsidized services through social security-number linked tax information, we document that long-run responses are dramatically larger than short-run responses. The one year elasticity is around 0.8, whereas at ten years the elasticity is about 6. The elasticity continues to rise until the end of our sample: in 2010 about 5% of Swedish adults had used the RUT deduction, but there was a gradual and nearly constant rise to 40% by 2020.
How should we conceptualize this phenomenon? Exploiting the richness of the data, we first show that the slow adoption is fundamentally about understanding extensive margin decisions. There are large heterogeneities in how much different RUT users spend, but all of the increase in RUT usage over time is accounted for by a higher share of the population using it, with stable average spending conditional on use. From this, we conclude that a model of the RUT usage should feature an extensive margin. Moreover, we show that this extensive margin decision looks like an adoption decision. Before using RUT, there is a very low probability that you will start using it. However, once you have used RUT, there is a high probability that you will keep using it. We further show that this slow adoption is not driven by selection over time in who adopts. The usage for adopters remains constant over time. This is despite the usage being quite different in the cross section. To model this we use a technology adoption framework. We estimate a Hidden Markov Model (HMM) to characterize consumer adoption choices and persistence over time. The implementation of the HMM combined empirical microeconomic analysis, extensive data collection, and advanced econometric modeling.
Significant Results and Discussion
Our analysis reveals several key findings. First, the adjustment process is dominated by the extensive margin, with conditional usage remaining stable over time. Second, we find no evidence that early adopters are "high-value" users; while income predicts usage, both usage levels and adoption patterns remain consistent over time. Third, the data strongly suggest that taking advantage of the RUT credit represents an adoption decision, akin to the diffusion of new technologies.
We demonstrate that a three-state hidden Markov model successfully captures the key features of the adoption process. The model reveals highly persistent states with slow transition rates, implying that the journey to steady state is extraordinarily long, with a half-life of approximately 6.5 years. Information frictions cannot explain this slow adjustment, as the reform was well-publicized. Furthermore, we show that supply-side constraints are also unlikely to be behind the results, given that service providers rapidly entered the market. Rather, the patterns align with gradual changes in household production technology and social norms surrounding outsourcing household work.
Our findings carry important implications for policy analysis and welfare calculations. The stark difference between short-run and long-run responses suggests that standard empirical approaches focusing on short-run elasticities may significantly underestimate the eventual impact of permanent policy changes affecting the margin between home and market production. While we concentrate on a specific reform, our insights likely have broader applicability to policies that require extensive margin adjustments or adoption decisions, such as childcare subsidies or earned income tax credits.
New Research Questions
Several new research questions emerged, notably:
• What specific informational, normative, or market frictions underpin the observed slow adoption?
• How do behavioral patterns differ when comparing domestic service subsidies to other labor market interventions?
• What broader implications might similar adoption dynamics hold for other policy contexts, such as childcare subsidies or Earned Income Tax Credits?
Dissemination and Collaboration
Research results were actively disseminated through presentations at major international conferences, including the Society for Economic Dynamics Annual Meeting 2024 in Barcelona, Spain, and through invited seminars at leading research institutions worldwide. The research team collaborated internationally, notably with researchers at the University of Minnesota and the University of Edinburgh. Our presentations have sparked interdisciplinary discussions around policy design and economic inequality.
List of Publications
Working Papers:
• Boppart, T., Gottfries, A., Malmberg, H., & Mitman, K. "How Long is the Long Run? Theory and Evidence from a Swedish Tax Reform"
Conference Presentations:
• Society for Economic Dynamics Annual Meeting, 2024.
Links the personal websites of the authors are available below:
• Kurt Mitman (http://www.kurtmitman.com)
• Timo Boppart (https://www.econ.uzh.ch/en/people/faculty/boppart.html)
• Axel Gottfries (https://sites.google.com/site/axelgottfries/)
• Hannes Malmberg (https://sites.google.com/site/hannesmalmbergweb/)
This project fully complies with RJ's open access requirements.