Project topics and materials
Get Complete Project Material File(s) Now! »
Housing benefits and monetary incentives to work: Simulations for France
Paper published in French and English in Economie & Statistique / Economics & Statistics. Motivation. Housing benefits schemes aim at helping low-income households cover their housing expenditures. In France, this is primarily achieved through monetary transfers to tenants that are increasing with the rent (benefits-rent linkage) and decreasing with households’ earnings (means-testing). Earlier work shows that the linkage with the rent causes 50 to 80% of housing benefits to be captured by homeowners through rents increases (Fack, 2006). These results have prompted several reform proposals aiming at alleviating this linkage, effectively turning the scheme into a means-tested transfer programme. This raises questions about the articulation of housing benefits with other means-tested transfers in order to design an effective redistributive policy that preserves work incentives and does not generate a poverty trap.
Paper. Monetary incentives to work can be decomposed between incentives to work more when in-work (intensive margin) and incentives to start working when out-of-work (extensive margin). This paper characterizes incentives to work at both margins through, respectively, effective marginal tax rates and effective participation tax rates. They are estimated using the 2011 Enquˆete Revenus Fiscaux et Sociaux (ERFS, Insee) for childless singles with the TAXIPP microsimulation model, and decomposed by tax and transfer instruments. Meanstesting implies that a 1-euro increase in gross labor earnings reduces housing benefits by 27 cents on average. Combined with reductions in other means-tested transfers (30 cents) and the payment of social contributions (21 cents) this translates into effective marginal tax rates close to 80%. Means-testing also induces a reduction in housing benefits upon taking a job which acts as a participation tax. Its magnitude depends on whether individuals receive unemployment benefits when out-of-work. Unemployment benefits increase overall partici-pation tax rates by providing higher replacement earnings but decrease the participation tax linked to housing benefits by reducing the amounts of housing benefits received. Contribution. The paper contributes to the literature on monetary incentives to work in France and underlines the adverse incentive effect of housing benefits means-testing. While (Sicsic, 2018) focuses on the historical evolution of aggregate estimates of effective marginal and participation tax rates for different household compositions, this paper provides indi-vidual estimates shedding light on heterogeneity of incentives to work for childless singles. It further investigates how incentives to work are affected by (a) whether individuals receive unemployment benefits when out-of-work, (b) whether employer contributions are actually paid by workers or firms, (c) whether social insurance contributions are treated as taxes or as forced savings that guarantee future consumption. The analysis highlights the com-plexity of the French tax-benefit system and raises novel questions about the articulation of means-tested transfers and unemployment benefits.
Inattention and the taxation bias (with J. Boccanfuso)
Paper awarded with the ITAX PhD Student Award at the 2019 IIPF Conference.
Motivation. Tax-benefit systems are complex. Accordingly, a growing body of evidence documents that agents exhibit substantial inattention to, and misperceptions about, tax policies (Bernheim & Taubinsky, 2018; Stantcheva, 2020). These need not entail negative effects. For instance, Rees-Jones and Taubinsky (2020) document that agents underestimate personal income tax rates in the US, and that this reduces the efficiency cost of taxation because agents act as if taxes were lower. In light of this evidence, a burgeoning normative literature analyzes the design of optimal tax policy accounting for these features (Farhi & Gabaix, 2020). Yet, surprisingly little work analyzes how these features affect the conduct of actual tax policy, a question we tackle in this paper.
Paper. This paper shows that agents’ inattention to taxes interacts with policymaking and leads governments to implement inefficiently high tax rates: this is the taxation bias. Intuitively, inattentive agents partially ignore tax reforms which induces governments to increase tax rates. Yet, these reforms induce equilibrium adjustments in agents’ tax percep-tions which imply that these tax increases are too large ex-post, reflecting a commitment problem in the choice of tax policy. Deriving a sufficient statistics formula for the taxation bias, we take it to the data and estimate the magnitude of the taxation bias for the US econ-omy. Since inattention decreases with earnings, the taxation bias is more pronounced at low earnings levels and affects the progressivity of the tax system. Moreover, the efficiency gains from potential tax underestimation may be dominated by the efficiency costs of the taxation bias when attention is limited. Overall, the paper sheds a new light on the implications of tax inattention and misperceptions.
Contribution. The paper contributes to the expanding behavioral public finance littera-ture by showing that governments may no longer select the optimal tax policy in the presence of inattention to taxes. It thus provides a theoretical framework to analyze the policy adjust-ments that have been empirically observed after changes in the salience of taxes (Finkelstein, 2009; Cabral & Hoxby, 2012). The fact that inattention induces a commitment problem in the choice of tax policy builds an unexpected bridge to earlier works on the inconsistency of optimal plans (Kydland & Prescott, 1977) and on the existence of an inflation bias in monetary policy (Barro & Gordon, 1983). This novel finding supports the idea that inatten-tion to taxes may entail substantial costs, and contributes to our understanding of complex tax-benefit systems.
Make work pay or make search pay? Redistributive taxation and unemployment insurance
Motivation. Modern welfare states operate large-scale redistribution and social insurance programmes, which are usually studied independently. For instance, the optimal taxation literature (Mirrlees, 1971; Saez, 2001) analyzes an equity-efficiency trade-off in which gov-ernments value redistribution but must preserve work incentives (make work pay). Similarly, the optimal unemployment insurance literature (Baily, 1978; Chetty, 2008) analyzes a trade-off in which governments value insurance provision to the unemployed but must preserve job search incentives (make search pay). Yet, redistribution and unemployment insurance are hard to separate in practice, which begs the question of whether these problems interact and if so, how these interactions affect the optimal design of tax-benefit systems.
Paper. This paper analyzes the optimal design of redistributive taxation and unemploy-ment insurance. First, the two problems interact because agents’ work and search decisions are affected by both taxes when employed and unemployment benefits when unemployed. Sufficient statistics formulas for optimal taxes and optimal benefits thus depend on both work and search elasticities, as well as on the redistributive tastes of the government and on unemployment. Second, the efficient allocation of resources implies a balance between agents’ consumption when employed and when unemployed. The implied efficiency condition shows it is efficient to redistribute through both redistributive taxation and unemployment benefits. Third, this efficiency condition also reveals an important fiscal externality: pro-viding generous unemployment benefits is not only costly to unemployment insurance but also costly to redistribution because of the foregone taxes people would pay if they were employed. This externality happens to be strongly increasing with income, calling for an unemployment insurance system with decreasing replacement rates. A calibration to the US economy shows actual net replacement rates decrease with earnings but suggests some scope for Pareto-improvements. These findings highlight that redistributive taxation and unemployment insurance interact in important ways and that these interactions matter for the design of public policies.
Contribution. The paper broadly contributes to the public economics literature by con-necting two large and yet disconnected strands of research on optimal redistribution and optimal unemployment insurance. It more specifically contributes to the optimal income tax literature that considers unemployment (e.g. Hungerb¨uhler, Lehmann, Parmentier, & Van der Linden, 2006; Kroft, Kucko, Lehmann, & Schmieder, 2020) by introducing un-employment benefits that depend on earnings when employed, a crucial feature of actual unemployment insurance systems which plays a prominent role in the analysis. While the optimal unemployment insurance literature usually focuses on linear unemployment benefits and constant replacement rates, this paper applies methods of the optimal taxation litera-ture to characterize optimal nonlinear unemployment benefits and shows that replacement rates should be decreasing in income. Moreover, the analysis underlines that in settings with earnings heterogeneity, the provision of unemployment insurance is only possible (incentive compatible) in the presence of eligibility conditions such as minimal work durations. Concep-tually, the paper shows that problems of optimal redistribution and optimal unemployment insurance cannot be separated.
Sufficient statistics for nonlinear tax systems with pref-erence heterogeneity (with B. Lockwood and D. Taubinsky)
Motivation. An influential result on the optimal design of tax-benefit systems due to Atkinson and Stiglitz (1976) states that redistribution should operate through progressive income taxes rather than through differential taxation of consumption categories. This result relies on the assumption that tastes are homogeneous in the population, and does not generalize when tastes are correlated with earnings ability (Mirrlees, 1976). A prominent justification for the taxation of capital income, savings, bequests, and certain commodities is thus that it is efficient to tax consumption categories that are preferred by those with higher earnings ability. Yet, much of the existing literature on the topic focuses on linear taxation whereas actual taxes on savings, or on capital gains, can be nonlinear or feature earnings-dependent linear tax rates. Moreover, existing work on preference heterogeneity is mostly qualitative, or restricted to special assumptions about agents’ utility functions. Our understanding of optimal tax systems in the presence of preference heterogeneity thus remains limited.
Paper. This paper provides sufficient statistics characterizations for optimal nonlinear tax systems in the presence of correlated preference heterogeneity. Our results encompass un-restricted tax systems that implement the optimal allocation of resources in the economy, as well as separable tax systems that combine a nonlinear earnings tax with a nonlinear capital income tax, or with an earnings-dependent linear capital income tax. Under some surprisingly general assumptions, we show that such separable tax systems also implement the optimal allocation. Across all tax systems, the key sufficient statistic for preference het-erogeneity is the difference between the cross-sectional variation of consumption with income and the causal effect of income on consumption. Our formulas for optimal differential com-modity taxation produce empirically-implementable generalizations of the Atkinson-Stiglitz theorem, and take a familiar form that resembles the sufficient statistics formula for the optimal nonlinear earnings tax (e.g. Saez, 2001). Contribution. The paper contributes to our understanding of optimal tax systems in the presence of preference heterogeneity. First, it provides empirically implementable character-izations of optimal tax systems that do not rely on the estimation of structural preference parameters. The key sufficient statistic for preference heterogeneity can be directly estimated from administrative or survey data using quasi-experimental variations like tax reforms or wage changes. Second, we derive conditions under which the type of separable tax systems implemented in practice are optimal. This result on the optimal structure of tax systems shows that separability can be without loss of generality. Overall, the analysis paves the way for an empirical assessment of optimal tax systems given the amount of preference heterogeneity measured in the data.
Table of contents :
Remerciements – Acknowledgements
R´esum´e – Summary
Introduction g´en´erale
General introduction
Chapter 1 – Housing benefits and monetary incentives to work: Simulations for France
Chapter 2 – Inattention and the taxation bias (with J. Boccanfuso)
Chapter 3 – Make work pay or make search pay? Redistributive taxation and unemployment insurance
Chapter 4 – Sufficient statistics for nonlinear tax systems with preference heterogeneity (with B. Lockwood and D. Taubinsky)
References
