Publications

This paper examines the conventional wisdom, expressed in McAfee and McMillan's (1987) widely cited survey paper on auctions, that links increased variance of bidder values to increased information rent. We find that although the conventional wisdom does indeed hold in their (1986) model of a linear contract auction, this relationship is an artifact of that particular model and cannot be generalized. Using Samuelson's (1987) model, which is similar but allows for unobservable costs, we show that increased variance does not always imply increased information rent. Finally, we give the appropriate measure of dispersion (different from variance) that provides the link between the bidder value distribution and information rent.

Research papers

Many activities such as smoking and gambling have long been subject to government regulation. Congress has also considered passing laws to encourage healthy eating and some corporations subsidize exercising. Each of these activities can be addictive and is more enjoyable when done with others. These features, addiction and social interactions, make regulation potentially important but difficult to design. This paper develops and estimates the first model of demand for a good that is both addictive and affected by social interactions. A key benefit of jointly modeling these two phenomena is that addiction introduces dynamics that allow the researcher to identify social interactions under relatively weak assumptions. The model also illustrates that the type of externalities that arise depends greatly on the precise manner in which social interactions affect individual utility. Finally, I apply my proposed identification methodology to data on cigarette consumption from the National Health Interview Survey and find strong evidence for the presence of both addiction and social interactions. Accounting for social interactions together with addiction triples the estimated effect of anti-smoking laws on consumption compared to a model that ignores them.

While conducting empirical work, researchers sometimes observe changes in outcomes before adoption of a new treatment program. The conventional diagnosis is that treatment is endogenous. Observing changes in outcomes prior to treatment is also consistent, however, with the anticipation effects that arise naturally out of many theoretical models. This paper illustrates the importance of properly distinguishing endogeneity from anticipation effects by showing that incorrectly equating pre-period trends with endogeneity can lead to biased estimates. The paper then provides a framework for comparing the different methods for estimating anticipation effects and proposes a new set of instrumental variables that can address the problem that subjects' expectations are unobservable. Finally, this paper examines a specific set of tort reforms that are unlikely to be endogenous to physician supply but likely to have been anticipated by physicians. Accounting for anticipation effects increases the estimated effect of these reforms by a factor of four compared to a model that equates pre-period changes with endogeneity.

[Under review.]

[Under review.]

© Copyright Julian Reif. All rights reserved
Steve's free web templates