Research

Abstract We derive the optimal joint-income tax schedule for couples, addressing the distinction between interpersonal and inter-household inequality. Households are comprised of two spouses with possibly unequal access to the family's economic resources. Individual-oriented utilitarianism typically leads to a misalignment between the households' and the government's objectives, which Apps and Reese (1988) have termed dissonance. We solve the multidimensional Mirrlees' (1971) program by assuming identical iso-elastic preferences that collapse the two-dimensional heterogeneity in productivity into a single one that preserves single-crossing. The optimal tax formula adds a Pigouvian term to the traditional ABC formulae to correct for the presence of dissonance. In our numeric exercise, the Pigouvian term is positive, leading to a system with higher marginal tax rates relative to the non-dissonance benchmark. When generalizing the specification, whether marginal taxes should be increased or reduced as a consequence of dissonance hinges on whether the less powerful spouse's marginal contribution to household earnings is less or greater than her marginal entitlement to the household's consumption.

Abstract: Notional Defined Contributions (NDC) systems mimic the incentive structure of fully funded social security while preserving the Pay-as-you-go nature of most current systems. We study size-preserving social reforms that replace the current US system with NDCs characterized by alternative contribution rules and deficit/GDP ratios.If one retains the current mandatory age-independent contribution rules we find a change to an NDC to reduce welfare: the sacrifices in distributive and insurance properties are not compensated by the efficiency gains. NDCs are, however, flexible enough to allow for alternative contribution rules that increase welfare while preserving actuarial fairness. Contributions ought to be age-dependent and concentrated later in a worker's career. The incentive structure induces an increase in capital accumulation that results, through general equilibrium effects, in welfare gains larger for low-productivity workers, despite the increase in income inequality as captured by the Gini coefficient.

Abstract: This paper explores optimal distributive policies using a Collective approach to household behavior. This approach allows individual preferences for each spouse, which is crucial when examining policies targeted to women, like Mexico's Prospera or Brazil's Bolsa Familia programs. We assume the spouses' decisions follow a Nash-bargaining procedure with internal threat points. We show that the taxation principle does not apply, meaning that the optimal tax schedule is dominated by the optimal mechanism. This is because a single tax schedule cannot optimally influence threat points and induce households to choose desired allocations simultaneously. By permitting couples to opt for joint or individual tax filing, we significantly increase the set of implementable allocations. The central role of threat points motivates an extension of the model in which marriage market negotiations partially determine threat points after marriage. This extension endogenizes the distribution of couples and formalizes how general equilibrium considerations and social norms affect bargaining power within a couple. The extended model generalizes previous approaches to the optimal taxation of couples. Finally, we parametrize and calibrate this model to empirically evaluate the relevance of our theoretical findings. The ability to influence threat points has large effects on equilibrium allocations and the evaluation of optimal policies.

Abstract: When searching for employment, workers take different job characteristics into account. We study an environment where unemployed workers search for jobs with different effort requirements in a labor market characterized by directed search. In equilibrium, firms are more likely to post vacancies for high-effort jobs, as these are more profitable. Hence, low-effort jobs are hard to come across. The non-observability of employment contracts influences the optimal unemployment insurance (UI) program, leading to distortionary taxation. Calibrating the model to the U.S. economy, we find that non-observability of employment contracts results in faster declining UI benefits, steeper taxes upon re-employment, distortionary taxation, and a 10.5\% costlier program than an observable contract scenario providing equal welfare.

This paper has previously circulated with the title "Who should bear the risk of economic growth?"

Abstract: This paper investigates how risk should be shared between workers who require incentives and retirees who do not. We analyze the impact of incentives on workers' ability to bear risk and examine how the timing of incentives through entitlements should vary depending on the state of the economy. Our findings suggest that perfect risk sharing is optimal when the utility from consumption is logarithmic or when aggregate productivity growth is independent and identically distributed. Our numerical analysis suggests that the deviation from perfect risk sharing is small. We relate the quantitative findings to the failure of a consumption-based stochastic discount factor (SDF) to price economic growth, which is reminiscent of the equity premium puzzle. Once we augment our model with shocks that generate volatile enough SDFs, numerical deviations from perfect risk sharing are substantial.

Abstract: What is the cost of informality? On the one hand, an informal sector creates a restriction on the set of policies that can be implemented. On the other hand, its existence offers an alternative for those for whom the benefits of formal relations do not compensate for the costs. Based on Mirrlees (1971), we propose an optimal tax formula that accounts for the existence of informality. This allows us to adopt an inverse-optimum procedure to recover the social objective that rationalizes the current tax system and use it to evaluate the welfare consequences of eliminating the informal sector. Using survey data from Brazil that encompasses formal and informal workers' wages, we calibrate the model to recover the main parameters that underlie the formalization decision, i.e., the joint distribution of productivity and the formalization costs. We find welfare gains of 6.3% which can be decomposed into a 2% gain from a direct increase in tax revenues and a 4.3% gain from re-optimizing the tax system. 

Abstract: Tax perturbation methods rely on the sufficiency of local conditions to describe the behavior of agents in a taxation context. For this to be the case agents' choices must be `well behaved'. We show that this requirement may be too restrictive when the Spence-Mirrlees' condition (SMC) does not hold, once budget sets are endogenous in an optimal tax problem. In a quasi-linear version ofMirrlees' (1971) model without the SMC, we show that the optimal tax system promotes income segregation when agents are allowed to be indifferent between two different bundles. This type of indifference is generically ruled out under the tax perturbation method because of the sufficiency of local incentive constraints. Moreover, when this condition is not imposed, aggregate choices are still required to be well-behaved, which we show may not be the case at the optimum when global constraints are explicitly taken into account. 

Abstract: We revisit classic optimal tax theory while recognizing that most people live in multi-person households. We show how optimal tax formulae must be adjusted and highlight the usefulness of non-rate policies along the lines of Moore and Slemrod (2021). We show, among other things, that the presence of household 'internalities' makes lump-sum taxes seldom optimal, and that separability of preferences no longer suffices for Atkinson and Stiglitz's (1976) uniform tax result.We explore the `behavioral' nature of collective households comprised of fully rational agents to draw a parallel to Farhi and Gabaix's (2019) seminal work.

Abstract: We revisit the efficiency-equity trade-off of optimal tax theory by emphasizing the consequences of increased progressivity on \textit{growth}. We use an endogenous growth framework that considers both the decision to become a researcher and the effort established entrepreneurs make to improve their products. We find that the optimal level or progressivity is lower than the current one but that welfare gains are moderate -- CEV = 0.5%. However, if one disregards the growth impact one would prescribe a substantially higher level of progressivity at significant welfare cost -- CEV = -9%.