A Dynamic Competitive Forecasting Model Incorporating Dyadic Decision Making
提出一个考虑医生和患者二元决策的模型，用于预测新药销售，发现纳入二元决策能提升预测效果，且多数情况下患者用药后效用高于用药前。
New products are often launched sequentially, by different firms, and the purchasing decisions are sometimes made by dyads. This paper proposes a new model that explicitly considers dyadic decision making in drug prescription and allows assessment of the relative influence that physicians and patients have in making decisions concerning new as well as existing ethical drugs. Modeling sequentially launched competing products in a category allows for parsing out effects that are hard to differentiate in models designed to capture only a single product's dynamics. The proposed model is applied to prescription drug data sets in the pharmaceutical industry, and it also explicitly captures both physicians' and patients' pretrial and posttrial utilities of each drug in the therapeutic category. Based on the model's fit and out-of-sample forecasting performance, we find that, in many cases, the incorporation of the dyadic decision making leads to better performance vis-à-vis models where such decision making is not explicitly considered. We also find that in many cases the posttrial utility of a drug is greater than its corresponding pretrial utility, lending partial empirical support to the prevailing industry practice of spending on various activities (e.g., sampling to physicians) needed to get potential patients to try a new drug. The proposed model enables managers to predict in advance the sales of sequentially launched new drugs and plan the new product launch and strategy accordingly. The model is also applicable to other product categories involving more than a single decision maker, including business-to-business products (e.g., office equipment) as well as to products targeting children (e.g., toys).