Optimal taxation when people do not maximize well-being

Abstract

I derive the optimal nonlinear income tax when individuals do not necessarily maximize their own well-being. This generates a corrective argument for taxation: optimal marginal taxes are higher (lower) if individuals work too much (too little) from a well-being point of view. I allow for multi-dimensional heterogeneity and derive the optimal tax schedule in terms of measurable sufficient statistics. One of these statistics measures the degree to which individuals fail to optimize their labor supply. I empirically estimate this by using British life satisfaction data as a measure of well-being. I find that low-income workers tend to work 'too little' and high-income workers 'too much,' providing a motive for lower marginal tax rates at the bottom and higher marginal tax rates at the top of the income distribution

Optimal Taxation with Behavioral Agents

“…This could provide a behavioral rationale for the EITC (Earned Income Tax Credit) program. In parallel and independent work, Gerritsen (2016) and Lockwood (2017) derive a modified Saez formula in the context of decision vs. experienced utility model. Lockwood (2017) provides an empirical analysis documenting significant present-bias among EITC recipients, showing that a calibrated version of the model goes a long way towards rationalizing the negative marginal tax rates associated with the EITC program.…”

“…Liebman and Zeckhauser (2004) study a Mirrlees framework when agent misperceive the marginal tax rate for the average tax rate. Two recent, independent papers by Gerritsen (2016) and Allcott, Lockwood and Taubinsky (2019) study a Mirrlees problem in a decision vs. experienced utility model. Our behavioral Mirrlees framework is general enough to encompass, at a formal level, these models as well as many other relying on alternative behavioral biases.…”

Behavioral Optimal Taxation: The Case of Aspirations

“…The abundance of evidence of nonrational behavior has lead scholars to call for or engage in the development of behavioral models for welfare evaluations (Bernheim and Rangel, 2005;McCaffery and Baron, 2006;Kőszegi and Rabin, 2008;Riedl, 2010;Mullainathan et al, 2011;Chetty, 2015). However, such models are still extremely rare in the field of optimal taxation (some notable exceptions are O' Donoghue and Rabin, 2006;Chetty et al, 2009;Farhi and Gabaix, 2015;Gerritsen, 2016). The question whether the results obtained in a rich literature of optimal taxation theory carry over to a world populated by actual humans rather than rational agents is clearly important.…”

Measuring “Schmeduling”

“…2 See, e.g., Gerritsen (2016), Blomquist and Micheletto (2006), Kanbur et al (2008), Kanbur et al (2006), as well as Bernheim and Taubinsky (2018) for a review.…”