Working Papers

Peer Feedback in Clinical High-Volume Activities: How Ranking Designs and Ability Impact Physician Effort
(joint with Y. Ndiaye, C. Waibel and D. Wiesen), accepted at Management Science

Peer feedback has strongly been advocated by medical professional societies as a key non-monetary strategy to improve the quality of care. Currently, if at all, clinical leader often provide feedback on key quality indicators in a rather pragmatic way. Systematic evidence on how to design peer feedback is scarce. We study how physicians' medical service provision is causally affected by peer feedback. Using a behavioral experiment, which mimics high-volume activity in a clinical setting, we systematically analyze how different non-anonymous ranking designs that vary in their granularity affect physician effort provision. In particular, we analyze how the heterogeneity in the physicians' abilities explains the responses to the different ranking designs. From a theoretical model of status concerns, we derive behavioral predictions. We report three main findings. First, peer feedback in form of rankings enhances physician performance on the aggregate compared to a non-ranking baseline. Overall, the effects tend to be stronger the more granular the ranking is. Second, the ranking design affects physician performance, and the impact of the ranking design depends on the physicians’ abilities. For an individual physician, increased granularity particularly motivates within her ability range - and may demotivate outside. Finally, our results indicate that small ranks at the tails of a ranking are salient. For clinical leaders, our results suggest that the use of rankings improves physician effort and imply that the choice of a ranking design should account for the composition of the physicians' abilities.

Public Assignment of Scarce Resources under Income Effects

The use of money is restricted for certain goods like school places or human organs. Yet, proponents of selling the goods emphasize efficiency benefits. Without income effects an approach that sells the good for a market clearing price indeed has superior efficiency and welfare properties. We show that under income effects this may no longer hold. We explicitly model income heterogeneity in a model of non-linear preferences and study characteristics of different assignment approaches under utilitarian objectives. We show that for positive income effects a fair lottery can be ex-post efficient despite heterogeneity in the willingness to pay for the good. It furthermore may even Pareto-dominate an approach that offers superior access to the good in return for the payment of a price that is redistributed to the population. If poorer agents benefit more from the goods than richer ones do, a fair lottery can be even optimal though the rich are willing to pay more for the good. In general, smoothing access to the goods by involving randomization on the cost of redistribution is particularly attractive from a utilitarian view if the preference space is narrow or benefits from consuming the goods are large. Repugnance towards classical price mechanisms thus can be supported by efficiency and welfare arguments if income effects play a role. 

Constraints on Matching Markets Based on Moral Concerns (joint with A. Wambach)

Monetary transfers are banned or restricted on many markets. These restrictions are often motivated by moral concerns. However, it is not obvious whether the observed restrictions on monetary transfers are the appropriate market design answer to these concerns. Rather than exogenously imposing constraints on monetary transfers for the assignment of indivisible goods, we introduce discrimination-freeness as a desideratum based on egalitarian objectives. Discrimination-freeness requires that the objects are allocated independent of wealth endowments. A key assumption in our model are positive income effects. We show that whenever income types are dispersed enough money cannot be used to Pareto-improve ordinal and money-free assignments without violating discrimination-freeness. Furthermore, if a discrimination-free assignment of objects and money is implementable then the respective object assignment is also implementable without money. Once money can be used outside a market designer's control, further restrictions than only money-freeness are required to address discrimination concerns. 

The Streetlight Effect: Focusing on Unimportant Attributes (joint with W. Mimra)

We model competition in the provision of a multi-attribute good, for which a consumer observes attribute quality imprecisely before purchase. Attributes are not equally important for the consumer, high quality in one attribute is more important for ex-post utility than high quality in the other attribute. Quality production is stochastic and providers can shift resources to increase expected quality in some attribute. We say that there is consumer focusing on an attribute if the consumer's best response prescribes to choose according to a high quality signal in this attribute, independent of signals in the other attribute. We show that there may be focusing on the unimportant attribute if the relative signal precision in this attribute is high. Under focusing on the unimportant attribute, providers invest in quality in this attribute in the unique symmetric equilibrium and any equilibrium is inefficient. Increasing signal precision, wile improving selection based on realized quality, may lower welfare by shifting quality investments. We discuss the implications of different regulations for quality reporting.