When Less is More: Control Limit Strategies in a Diagnosis Crowdsourcing Platform
Jingxuan Geng, Guangwen Kong, Marco Shaojun Qin
Under 2nd round review at Management Science
We analyze a telemedicine platform where patients obtain multiple diagnoses from doctors for a single fee. Using commission-based pricing leads to downward price distortion to curb doctor over-participation. Introducing a control limit (capping diagnoses per inquiry) enables the platform to appropriately allocate doctors across patients. Counterintuitively, patients’ delay sensitivity—their preference for earlier diagnoses—reduces late-arriving doctor participation, acting as a natural control limit. Consequently, profit gains from control limits diminish as delay sensitivity increases. When doctors differ in quality, adverse selection occurs, i.e., low-quality doctors crowd out high-quality ones when service differentials are small. Control limits counteract this by increasing the high-quality doctors’ reward probability. Extensions such as general utility function, endogenous commission rate, endogenous effort, price differentiation, and stochastic delay time confirm the robustness of the main result. We test the predictions of our model leveraging a natural experiment, and found consistent results.