Working Papers

Informative Milestones in Experimentation
Job Market Paper 

I study a continuous-time moral hazard problem with private learning about a project of unknown quality. There is ex ante symmetric information and full commitment for the principal. The project generates a profit if two consecutive stages are completed. The amount of experimentation required to complete the first stage (milestone) is informative but not conclusive about the quality of the project. The informativeness of the milestone yields an incentive to privately shirk in the first stage. This increases the principal's pessimism in the second stage and thereby induces more favorable second-stage contract terms for the agent. In the optimal contract, the reward for a first-stage success is decreasing in its arrival time to prevent effort delays. The composition of the reward changes with the success time: early successes are rewarded with long second-stage deadlines and no bonus payments in the first stage, while later successes are rewarded with first-stage bonus payments and less continuation value from second-stage experimentation. Allowing for agent replacement between stages, I show that the principal wants to replace the agent in the second stage if the success arrives late.

We study competitive awarding procedures of short haul railway passenger services in Germany from 1995 to 2011 by means of a newly collected data set. In particular, we use regression techniques to investigate the determinants of the number of bidders, the identity of the winning bidder and the subsidy level. We find that there are more bidders when the contract duration is high and the revenue risk low. The dominant operator is more likely to win contracts if it is the incumbent, the network is large, the contract duration is high, when used rolling stock is admitted and when there are few other bidders.

Work in Progress

The Effect of Asymmetric Revenue Uncertainty on Procurement Outcomes - Evidence from German Railway Passenger Services (joint with Stefan Weiergräber)

We provide a framework that allows to disentangle the effects of asymmetries in the private value distribution from the effects of asymmetric information about the common value component in an auction. We combine structural econometric methods with a detailed contract-level data set of the market for short-haul railway passenger services (SRPS) in Germany to study the effects of asymmetries in the cost distribution between the incumbent and the entrants from the effects of asymmetric information about the revenue component of a SRPS contract. More specifically, data on gross auctions, in which the firm does not face revenue risk, allow us to back out the cost distribution for each firm. Moreover, data on net auctions, in which firms bear the revenue risk, enable us to quantify the effect of revenue uncertainty on bidding strategies. Our preliminary results indicate that bidding behavior is indeed structurally different in gross and net auctions. In particular, we find that the incumbent is slightly more cost-efficient on most lines and holds more precise information on the future ticket revenues. Finally, in the absence of asymmetric information on ticket revenues, entrants would have bid much more aggressively than in the status quo.

Consumer Rating Dynamics (joint with André Stenzel)

We propose a framework to assess dynamic pricing incentives of firms in online markets with rating systems where product quality is private information. Consumers use a heuristic: Inference is characterized by the pair of quality and past consumers' tastes such that purchasing decisions are individually rational based on the current price and are consistent with the current aggregate rating. This gives rise to price serving as a selection device: It determines the set of purchasing consumers, and thereby the induced review and future aggregate rating and profits. We show that the firms' tradeoff between flow and future profits may give rise to either over- or underpricing relative to the myopic optimum, depending on the sensitivity of consumers' utility and review functions to quality, taste, price, and expectation. Finally, we use the model to provide numerical evidence that rating systems with limited memory, as recently suggested by e.g., may have adverse effects on the accuracy of consumers' inference, as well as consumer welfare.

Collaborating under Asymmetric Information (joint with Sinem Hidir)

In a dynamic moral hazard model, we analyze the interaction of learning about a team member's productivity and freeriding. In many team settings it is realistic to assume that players are uncertain about other team members' productivity. We consider the case in which one player's productivity is commonly known while the other player is privately informed about her productivity. The team has to obtain a single breakthrough to realize the benefits of the project which the players share. The uninformed player only wants to remain in the partnership if she is sufficiently optimistic that the other player's productivity is high. Her outside option is to work on the project alone which increases her reward in case of a success but also the expected cost of obtaining it because she has to complete the project by herself. The informed player always wants to remain in the relationship. As the absence of a breakthrough is bad news about the productivity of the collaborator, the uninformed player becomes increasingly pessimistic over time. As a consequence, she will increase her effort until she quits the collaboration or a breakthrough is obtained. This increases the freeriding incentive of the informed player and the high-productivity type of the informed player will work less in equilibrium than under complete information.