Abstract
The first Chapter of this Dissertation analyzes the general equilibrium welfare effects of implementing a job-search assistance program for unemployed jobseekers. While beneficial to participants, empirical evidence suggests these programs impose costs on non-participants through a crowding out effect on hiring. In the model, agents are heterogeneous with respect to assets, human capital, and the duration of their current unemployment spell. I consider a policy in which all agents out of work one year or more are automatically enrolled in a government-funded assistance program, which increases their job-finding chance by about 48 percent, and I analyze the welfare gains and losses associated with the creation of this policy for all agents. I then decompose these welfare effects into a partial equilibrium channel (individual-level job-search efficiency) and general equilibrium channels (labor market composition, firm hiring behavior, and price and tax effects). I find that the program is welfare-improving by about 0.22 percent. However, general equilibrium effects are negative and non-negligible in magnitude, accounting for a welfare loss of about 0.11 percent. A decline to labor market tightness, which reduces job-finding rates, accounts for most of this welfare loss. Price effects represent a welfare gain of about 0.1 percent, and tax effects are a welfare loss of 0.1 percent. Next, I show that high-skill and asset-poor agents benefit the most from the policy. Finally, I compare the search assistance policy to an unemployment subsidy with the same cost, and show the former delivers a higher welfare gain than the latter.
Chapter 2 studies the optimal scheme of search assistance in an economy populated by heterogeneous agents. By improving the job-finding chances of one agent, all other agents are made worse off, all else equal. This Chapter introduces a planner who takes this fundamental tradeoff into account when choosing the optimal search assistance intensity for all types of agents in the economy in order to maximize the sum of all agents' utility. The model has a two-period horizon and is simple enough to allow an analytical characterization of the planner's choice. For example, I show that the planner's choice of assistance intensity is an increasing function of agents' skill level, as the surplus from re-employment is increasing in skills. I solve three different (but closely related) versions of the model; one in which the planner solves directly for the optimal assistance scheme for all types, and two in which the planner chooses the assistance scheme based on functional rules; these last two allow me to characterize results for a large state space using only a small number of choice variables. After presenting results for these models, I solve an alternative specification that removes all employed agents from the planner's objective function, as these agents pay most of the cost of the program but receive none of the benefit. Now, the planner provides a much higher level of assistance to unemployed agents.
Chapter 3 adds endogenous search effort to the analysis, to study how the creation (and expansion) of a search assistance program affects agents' job-search behavior, and the implications of this behavior for welfare gains and losses. Empirical research has shown that agents' job-search effort is countercyclical; that is, worse labor markets are associated with higher effort. I extend this logic to the context of a job-search assistance program. I define an infinite-horizon model of heterogeneous agents in which the government introduces a search assistance policy that makes jobseekers more likely to find re-employment. I employ two separate matching functions and compare their results. The first function induces agents to reduce their search effort when assistance intensity increases, and the second leads them to increase their effort. For each of these, I analyze the welfare gains from the assistance policy when allowing for this endogenous effort, and compare this to a world where search effort is fixed at a baseline value. I show that the variable-effort specification yields a higher welfare gain for the first matching function, but welfare gains are essentially equal with the second function.
