Optimal labor income taxation with perks

Incentive problems arise in many economic relationships, between workers and firms as well as between those agents and the fiscal authority. If it is well known, in fact, that labor contracts tying wages to performance may mitigate the efficiency costs from unobservable effort, it is also an empirical fact that real word contracts, and the incentives provided wherein, are more frequently based on some sort of non-monetary compensation – fringe benefits, or perks, as they are called. As a result, the fiscal authority is more often called to decide on the eligibility of those perquisite goods and, therefore, on their taxability. In spite of their diffusion, however, there is still little consensus among researchers on the reasons why perks should be provided, and what their effect on welfare might be. The standard explanation for perks usually relies on some form of agency problem. However, monitoring concerns are certainly less relevant when dealing with a fiscal authority, that, for the most part, may monitor the provision of perks better than the payment of cash. Motivated by the concerns about how fringe benefits may restrict the revenue collection problem and affect the redistribution of resources and incentives, this research aims at investigating the relationship between the provision of perks and i) the progressivity/regressivity of the optimal transfer-tax system, and ii) the top-income marginal tax rate. To the best of our knowledge, this thesis constitutes the first theoretical attempt to deal with the issue at hand. In fact, all existing papers that analytically study the moral hazard problem with perks, either focus on their effect on the effort responsiveness to monetary incentives, or they study the cash-perks substitutability in the optimal labor contract, absent any consideration for the insurance/incentive purposes of income taxation. To some extent, we also contribute to the literature on optimal taxation, by emphasizing the connection between the agents' responses to the fiscal system and the structure of the labor market. Methodologically, we investigate our research agenda by following two different approaches. In Chapter 2, we develop a static version of a standard stochastic Ramsey's problem with a representative agent and an utilitarian, resource-seeking government who takes the post of the principal and owner of the only firm in the economy. When designing the optimal rules for cash payment and perk provision, the government, who is constrained by his budget balance, takes into account the agent's unobservable reactions to the tax system, and set optimal taxes so as to provide either insurance and right incentives. In Chapter 3, however, we take an alternative perspective, and study a decentralized economy wherein an independent, self-interest firm retains full control of the provision of perks and wage payments, given the fiscal policy announced by the government. Our analysis aims at highlighting the existence of a trade-off between the opposite interests of the government and the employer, that will serve as a rationale for the characterization of the equilibrium marginal tax rate that we derive from a Nash non-cooperative interaction game between the fiscal authority and the agents in the labor market. Our analysis suggests that, whenever perks are efficiently provided, the government trades off progressivity for perk provision. Our main conclusion, in fact, is that the (centralized) second-best efficient taxation scheme with perks is more regressive compared to its perk-less equivalent. Notably, we also find that, whenever society's preferences for public expenditure and agents' risk aversion are sufficiently low, a regressive top-income marginal tax is also consistent with a positive provision of perks. In a numerical exercise, for example, we show that a marginal tax rate of 30% allows for either i) a positive provision of perks that accounts for 1.6% of the gross income (3.2% of the agent's wage), and ii) a per-capita public expenditure which is 7% of the top-brackets taxpayers' gross income (15% of their earnings). An equilibrium allocation without perks results, instead, for a 45% marginal tax rate. However, in spite of the ability of our theoretical framework to capture the relationship between risk aversion and efficient provision of perks, through their effect on the optimal trade-off between the insurance and incentive motives for taxation, our model has its weakness, as to regard its sensitiveness to the stochastic dominance properties of the probability distribution and the level of income in the economy. If, in the former case, deviations from the benchmark distribution are qualitatively important but quantitatively small, in the latter case, the quantitative implications of different income levels are quite important. A more precise calibration of the models and a better characterization of the results are left to future works.