Research

Traditional optimal commodity tax analysis, dating back to Ramsey (1927), prescribes that to maximize welfare one should impose higher taxes on goods with lower demand elasticities. Yet policy makers do not stress minimizing efficiency costs as a desideratum. In this note we revisit the commodity tax problem, and show that the attractiveness of the Ramsey inverse-elasticity prescription can itself be inverted if the tax system is chosen -- or at least strongly influenced -- by taxpayers who are overly confident of their ability, relative to others, to substitute away from taxed goods.

In this paper, I develop a theory of the jointly optimal choice of nonlinear income taxes and commodity taxes which accounts for the presence of tourists. Standard results are augmented in two key ways. First, if commodity taxes are uniform, I show that the Pareto efficient rate is the one that maximizes revenue from tourists. Impacts of a uniform commodity tax on residents are irrelevant, as they can always be compensated through income tax changes. Second, I show that the presence of tourists in the tax base overturns the classic Atkinson-Stiglitz result justifying uniform commodity taxation. Efficient rate differentiation is characterized by a Corlett–Hague-style rule which requires placing lower taxes on goods which tourists can more easily substitute for expenditures outside of the destination economy. Notably, the impact of commodity taxes on residents do not influence the pattern of efficient differentiation nor of the overall level of commodity taxation: rather, resident consumption behavior only influencing the magnitude of efficient rate differentiation. By contrast, the endogeneity of tourist arrivals may also influence the efficient degree of rate differentiation: taxes should also be lower on those goods whose consumption most strongly predicts a larger extensive margin response. That is to say, rate differentiation may be efficient if it can achieve a tagging objective, targeting the greatest tax burden to those tourists who are relatively more price sensitive at the extensive margin or whose visits generate the relatively more tax revenue. My results can rationalize some commonly observed tax policies in tourist destinations, such as VAT refunds to tourist on large purchases, or reduced VAT rates on hotel accommodation. On the other hand, these results do not support other common practices, such as levying higher rates on goods that are disproportionately consumed by tourists.

When should tax policy be used to influence political donation behavior? In a model of electoral politics where campaign spending is financed by citizen donations, inequality of political influence favoring the "donor class" can arise. Adopting the normative stance that such inequality is undesirable, characterizations of optimal linear and nonlinear taxation of political donations are presented. Sufficient statistics for optimal policy include not only donation demand elasticities, but also the marginal efficacy of campaign spending, and the effect of taxes on the sensitivity of donations to candidate policy platforms. The results help to rationalize some observed policies. For example, taxing donations reduces campaign spending, driving up the marginal return to campaign spending and potentially increasing political inequality. Nonlinear subsidies targeting small donors - such as those seen in Canada - can decrease the relative influence of large donors without decreasing campaign spending.

This paper revisits classic results in optimal income taxation by incorporating non-filers, whose undermining the efficacy of income tax-based redistribution. Under Atkinson-Stiglitz preference assumption, the addition of non-filers rationalizes commodity subsidies as an alternative approach to achieving redistribution. The imperfection of this approach may also rationalize differential commodity taxation/subsidization. Moreover, when filing is an endogenous choice—affected by demogrant incentives—the optimal income tax follows a modified ABC rule, with social marginal utility adjusted to reflect imperfect take-up. These findings reveal that accounting for non-filers can increase or decrease optimal redistribution.

We ask how externalities should be taxed when redistribution is costly. In our model, the government raises revenue using distortionary income and commodity taxes. If more or less productive people have identical tastes for the externality-generating activity, the government optimally imposes a ``Pigouvian'' tax equal to the marginal damage from the externality. This is true regardless of whether the tax is regressive. However, if regressivity partly reflects different preferences of people with different incomes, the tax optimally deviates from the Pigouvian benchmark because this helps redistribute income efficiently. The overall tax may be higher or lower, and may even reverse sign relative to the externality. We derive sufficient statistics for optimal policy, and use them to study carbon taxation in the United States. Throughout most of the income distribution, our empirical results imply an optimal carbon tax below marginal damage, but this reverses for very high-earning households.

This paper presents a simple extension of probabilistic voting theory that accounts for various potential sources of inequality in political influence. In this model, the need for political parties to secure political donations leads to a distortion of policy choices in favor of the marginal donor, even without an explicit quid pro quo. All that is required for political inequality to emerge due to donation incentives is that parties anticipate the donation responses of the "donor class" to policy announcements. However, money is not the only source of political inequality in this model: parties also cater to the preferences of the marginal donor of volunteer labor, and to the marginal donor of social capital. Critically, in the case of each of these factors of electoral production, it is not the largest donor who wields the greatest influence, but rather the one whose donation of a productive factor is most responsive to policy declarations. The model in this paper helps to clarify various potential sources of inequality, and fundamental challenges to measuring political inequality. It may also prove useful in predicting how political inequality responds to shocks to the electoral production function due to changes such as online campaigning, targeted advertising, and the potential future automation of campaign tasks.

A rigorous & intuitive guide to bunching method, complete with visual illustrations.

Is the coexistence of political equality and economic inequality stable? I consider this question using a simple dynamic model of democratic redistribution which captures the idea that economic and political inequality may be mutually reinforcing. Two candidates iteratively compete in elections fought over income tax policy. They engage in campaign spending financed by citizen donations. This campaign finance mechanism creates a feedback loop. In each period, the citizens with higher levels of after-tax income have higher relative political influence, which in turn results in more favorable tax treatment of these citizens in the next period. Long run convergence to plutocracy can occur for arbitrarily small levels of initial economic inequality. However, the opposite is also possible: a society which is initially extremely unequal may be destined for egalitarianism. The long run outcome may also exhibit strong sensitivity to initial conditions.

Abstract It is well-known that static labor supply theory predicts that the density of taxable income will feature bunching (a mass point) at a convex kink point in a tax schedule. In this paper, I highlight another prediction of the theory: the density of taxable income should feature a discontinuity at a "dent point" in the tax schedule (a point where the curvature of the tax schedule is discontinuous). As in the case of bunching at kink points, this property of the distribution of taxable income can—in principle—be used to learn about the behavioral response to taxation. In particular, the size of the density discontinuity nonparametrically identifies the local average ETI at the dent point. This result is especially intriguing given that recent work has show that the bunching mass cannot be used to achieve nonparametric identification. However, the applicability of this identification strategy is limited by the relative scarcity of dents points in real world tax or price schedules. However, I show that a variant of the same identification strategy can be used to identify the ETI in a population of "ironing" agents: taxpayers who make decisions as if they believed their marginal tax rate were equal to their average tax rate. I consider possible applications of this version of the strategy to tax analysis and utilities pricing.

We develop an optimal tax framework that combines two recent extensions of tax analysis: a tax-systems emphasis on non-rate policy instruments, and a recognition of the role of behavioral biases. Although the implications of taxpayers' biases for optimal tax rates have received considerable attention, a complete analysis of this aspect of optimal tax theory must account for the fact that such biases are often endogenous to the non-rate aspects of a tax system. We first generalize and extend the analysis of optimal tax systems to incorporate endogenous behavioral biases. We then develop a novel and important application of this issue, showing how misperception of the tax rate affects the optimal breadth of the tax base.

We investigate the impact of the US drone program in Pakistan on insurgent violence. Using details about US-Pakistan counterterrorism cooperation and geocoded violence data, we show that the program was associated with monthly reductions of around nine to thirteen insurgent attacks and fifty-one to eighty-six casualties in the area affected by the program. This change was sizable, as in the year before the program, the affected area experienced around twenty-one attacks and one hundred casualties per month. Additional quantitative and qualitative evidence suggests that this drop is attributable to the drone program. However, the damage caused in strikes during the program cannot fully account for the reduction. Instead, anticipatory effects induced by the program played a prominent role in subduing violence. These effects stemmed from the insurgents’ perception of the risk of being targeted in drone strikes; their efforts to avoid targeting severely compromised their movement and communication abilities, in addition to eroding within-group trust. These findings contrast with prominent perspectives on air-power, counterinsurgency, and US counterterrorism, suggesting select drone deployments can be an effective tool of counterinsurgency and counterterrorism.