We characterize a …rm’s pro…t-maximizing turnover policy in an environment where managerial productivity changes stochastically over time and is the managers’private information. Our key positive result shows that the productivity level that the …rm requires for retention declines with the managers’tenure in the …rm. Our key normative result shows that, compared to what is e ¢ cient, the pro…t-maximizing policy either induces excessive retention (i.e., ine ¢ ciently low turnover) at all tenure levels, or excessive …ring at the early stages of the relationship followed by excessive retention after su ¢ ciently long tenure. JEL classi…cation: D82

How do discount rates affect agents ’ decisions and valuations? As economists, we are routinely confronted with this question: whether they concern savings and refinancing decisions, the exercise of options, or experimentation with reforms or new technologies, discount rates are at the heart of economic analysis. Given its importance, one might expect there to be

Executive summary. My research explores, challenges, and extends our understanding of auctions and other strategic interactions between buyers and sellers. Auctions have been studied intensively for decades, but my work has pushed the envelope in several new and interesting ways, notably: (i) to lay the foundation for theoretical and empirical studies of “multi-unit auctions ” (funded by 2003-2006 NSF grant), (ii) to endogenize the rules of the auction (“credible sales mechanisms”), (iii) to endogenize what bidders believe about each other’s private information (“secrecy in auctions ” and “strategic ignorance”), and (iv) to endogenize what the seller knows about bidders ’ private information (“disclosure in auctions”). In another vein, funded by 2008-2011 NSF grant, I have developed the theory of stochastic games with re-matching (“dynamic partnerships”). This research sheds light on how partnerships form, thrive, struggle, and fail in changing environments, with implications for labor markets and strategic learning in relationships, among other things. A notable and unusual aspect of my research record is that I have been able to publish successfully in so many styles, from pure theory (see #1,#6,#10 on my ranked list) to applied theory (see #3,#4,#7,#9), econometric theory (see #5,#8) and econometric applications (see #2). This multi-pronged research approach reflects my eagerness to learn and deploy whatever tools are needed to deeply and completely understand my subject. Outline. The remainder of this personal statement is organized into nine parts:

In an environment where managers ’ ability to generate pro…ts changes stochastically over time, we characterize the optimal turnover policy for a …rm hiring a succession of managers. The focus is on the interaction between the compensation scheme used to incentivize the managers’e¤ort and the retention policy used to screen the managers’types and induce the desired turnover. Our key positive result shows that the …rm’s optimal retention policy becomes gradually more permissive over time: the productivity level that the …rm requires to an incumbent manager to be retained declines with the manager’s tenure in the …rm. This result helps explain the practice of rewarding managers that are highly productive at the early probatory stages (and hence generate, on average, high pro…ts) by o¤ering them job stability once their tenure in the …rm becomes long enough. Our key normative result shows that, when compared to what is e ¢ cient, the …rm’s pro…t-maximizing contract either induces excessive retention (i.e., ine ¢-ciently low turnover) throughout the entire relationship, or excessive …ring at the early stages followed by excessive retention in the long run. JEL classi…cation: D82

This paper studies relational contracts with partially persistent states, where the distribution of the state depends on the previous state. The optimal contracts have properties similar to those of stationary con-tracts in Levin (2003), but stationary contracts are no longer optimal. With endogenous states, the dynamic enforcement constraint and the second-best contract are different from the i.i.d. case. The paper then applies the results to study implications for markets where the principal and the agent can be matched with new partners. (JEL C73, D86, L14) 1

in any medium now known or hereafter created.