Microeconometric Analysis with Scanner Data

我们使用扫描仪数据进行微观计量分析，以实证调查消费者和零售商的决策和行为。我们的项目包括两个子项目：1。零售商的定价决策与食品品牌在其门店中的竞争消费者对音乐cd的购买决策在第一个项目中，我们发表了一篇综述文章和一份研究报告，研究了88家超市的贝叶斯分层模型。扫描一本书中的数据。在后者中，我们表明，虽然不稳定的价格弹性估计是通过使用每个商店数据单独的普通最小二乘估计产生的，但稳定的估计是通过同时使用所有商店数据的贝叶斯分层估计获得的。我们准备了一篇投稿期刊的论文原件。在第二个项目中，我们使用日本一家音乐CD店的销售数据进行了实证分析，该数据发表在经过评审的会议论文集上。我们的数据的显著特征是它包含每个客户的ID号，这使我们能够使用他/她的购买历史来检查每个客户对音乐类型的偏好。我们使用这个扫描面板数据来估计每个客户对音乐cd的需求。对于估计，我们采用贝叶斯层次模型。利用估计的顾客偏好，我们提出了零售商结合优惠券的分类变化，以提高其盈利能力，并保持其在顾客中的声誉。在这两个项目中，我们使用了马尔可夫链蒙特卡罗（MCMC）方法进行估计。

We conducted microeconometric analyses using scanner data in order to empirically investigate decisions and behavior of consumers and retailers. Our project includes two sub-projects:1. Retailers' pricing decision and competition among food brands in their stores2. Consumers' purchase decision on music CD.1. In the first project, we published a review article and a research note on a Bayesian hierarchical model using 88 supermarkets .scanner data in a book. In the latter, we showed that, while unstable price elasticity estimates were produced by separate ordinary least square estimation using each store data, stable estimates were obtained by a Bayesian hierarchical estimation using all stores data simultaneously. We prepare an original paper for submitting to a journal.2. In the second project, we conducted an empirical analysis using sales data from a music CD store in Japan, which is published in refereed conference proceedings. The distinguishing feature of our data is that it contains each customer's ID number, which enables us to examine each customer's preference for music genre using his/her purchase history. We used this scanner panel data to estimate each customer's demand for music CDs. For estimation, we adopted a Bayesian hierarchical model. Utilizing estimated customers' preference, we proposed the retailer's assortment change combined with couponing both to improve its profitability and to maintain its reputation with customers.In Both projects, we used a Markov Chain Monte Carlo (MCMC) method for estimation.