To a twenty five year old entering the labor market, the landscape must feel full of uncertainties. Will they land a good job relatively quickly or will they initially bounce from one job to another in search of a good match? What opportunities for on-the-job training and other forms of skill accumulation will they find? How well will they take advantage of these opportunities? Just how good are they? How high will they rise?
Will they advance steadily within a firm or industry, or be laid off and have to reinvent themselves elsewhere? For all these reasons, young workers must find it challenging to predict how much they will be making at, say, age fifty. More generally, an ample body of empirical research has documented that individuals face significant uncertainty in their lifetime earnings (see for example Storesletten, Telmer, and Yaron (2004) for a recent study).
This considerable uncertainty in lifetime earnings generates an important role for social insurance. In advanced countries, social insurance is to a large extent provided through the tax system as a tool for social insurance, taxing both labor and capital income, and financing various forms of transfers. What does economic theory say about this approach and how can it help guide its design? The goal of this proposal is to develop simple and robust insights to construct tax systems that do a good job of insuring idiosyncratic risks over the life cycle.
Early writings by Milton Friedman argued that taxes should play an important role in the provision of social insurance. In Friedman (1962), he proposed a "negative income tax" where individuals with low enough income would owe a negative tax, and thus collect transfers. In particular, he advocated a simple system where a constant marginal tax on income is combined with a lump sum rebate -- resulting in a linear tax function with a negative intercept (Sheshinski 1972). For a given distribution of before-tax income, such a tax reduces the dispersion in after-tax income, thus achieving a more equitable outcome.
Mirrlees' seminal work (Mirrlees 1971) refined this idea by allowing for fully non-linear taxation of income. Since then, with a few notable exceptions, optimal tax theory has mostly worked with a static model (e.g. Diamond 1998; Saez 2001). How are the lessons from the static models relevant for the dynamic problem faced by actual workers? Or better, how can we extend optimal tax models to dynamic settings to take into account the uncertainty and dynamics faced by workers? This question has recently spurred interest in studying more realistic dynamic settings (see Golosov, Tsyvinski, and Werning 2006, and the references therein). Because these models present a number of technical difficulties, relatively little is known how taxes should optimally be designed.
In Farhi and Werning (2010), we characterize theoretically and quantitatively the optimal design of taxes in such a dynamic setting. We show an important role for taxes to depend on age. Because the optimal tax system requires sophisticated tax instruments, we also consider simpler tax systems. In particular, we find that simple generalizations of Milton Fridman's "negative income tax" perform almost as well as the optimal tax system. It combines a lump-sum rebate with linear taxes on labor that increase with age until retirement.
In our proposed research, we plan to complete this work and also research the following. First, we would like to considerably extend the scope of our numerical simulations. These simulations are demanding, requiring the use of modern parallel computing methods. We plan to devise programs and algorithms to allow a wider set of scenarios, regarding worker preferences and behavior as well as the process for uncertainty they face. Second, we would like to enrich the study to consider human capital accumulation, on the job and at school. This makes the age of entry into the labor market a choice, instead of a given. Third, we would also like to allow for a decision to retire and analyze its effects on the optimal tax and social security system. Finally, we would like to extend our analysis to incorporate general equilibrium considerations. The treatment that we provide in Farhi and Werning (2010) is essentially in partial equilibrium at a given interest rate. As we have shown in previous work (Farhi and Werning 2008, 2009), taking into account the effects on interest rates may be important in evaluating tax systems.
