Working Papers
I study price setting behavior by sales agents of an electrical wholesale company following a change in their compensation contract. Originally, agents received a fixed share of the revenues from their sales. Under the new scheme, agents’ commission rates increase with the price-cost margin of the sale. This contract creates a multitasking conflict among products historically sold with different margins. Employees of the firm decide how much effort to allocate to the sale of each good and can modify prices by offering discounts. A simple multitasking principal-agent model predicts that, even when incentives increase for all products, if efforts across goods are substitutes for the agent, the new scheme will increase price and effort of the more-compensated goods at the expense of price and effort of the less-compensated goods. The reform was enacted at different times in different stores, enabling measurement of its impact by difference-in-differences. Commissions on 95% of goods increase but workers do not raise prices on all these products. Despite the stronger financial incentives, the price of 18% of goods decreases and increases for the rest, suggesting workers reallocate effort among products. These changes are exacerbated when products are bundled together, but exist for sales of a single item, suggesting additional effects beyond gaming.
- The Determinants of Teachers’ Occupational Choice (with Kevin Lang)
We examine the appropriate fit of a dynamic optimization model with an application to teacher earning structure. To do so, we adjust the specification of the model by varying the unobserved heterogeneity in individuals’ preferences and earnings. We find that even a model with no unobserved heterogeneity fits well within sample. Testing formally selects models with substantial unobserved heterogeneity, suggesting that relying on in-sample fit is problematic. In the application to teacher earnings, we show that a reform that adjusts teacher compensation to mimic the return to skills and riskiness of the non-teaching sector would be expensive and challenging; overall compensation in teaching would increase, but the majority of current teachers would be made worse off. Importantly, these conclusions are sensitive to the degree of heterogeneity allowed in the model.
I explore the importance of employee-customer relationships as an incentive and price discriminating tool. The model assumes that customers differ in their valuations and on their probability of returning (q). The distribution of valuations and q are known, and in each interaction the sales agent exerts effort to learn the customer’s valuation. The agent earns a commission based on the client’s payment and has full pricing flexibility. The two main insights are, first that when the valuation is unknown, effort is increasing in q. Second, that a commission raise increases the learning speed, and under certain conditions the learning speed on customers with higher q’s increases more. Average prices should be increasing in effort. Using administrative data from an electrical wholesale company, I show that the data supports the theoretical insights.