Firm learning in a selection market
研究了企业在创建新市场时如何通过适应性学习掌握需求和成本信息，利用加州ACA交易所数据发现学习模型比完全信息假设更符合实际，且完全信息会损害纳税人利益。
Abstract Creating new markets is a prevalent approach for implementing large social programs. Assuming firms have full information about the relevant parameters upon market inception is commonplace in the literature. In contrast, we develop an adaptive learning model with selection to study how firms' knowledge of demand and cost affects the market equilibrium. We estimate alternative learning models with data from the California ACA exchange and assess their external validity using novel data on firms' predicted costs from insurer rate filings. The learning models provide statistically significant improvements in fit relative to the standard model that assumes firms have full information. Most of the improvement results from allowing firms to learn about the relationship between demand and cost. Firms with full information can increase profit, but at taxpayers' expense. Regulation that prohibits firms from using certain consumer information to set premiums makes them react more to the information they can use.