CRSP for Teaching

Abstract

Most finance professors, at least in the North America, are quite familiar with the CRSP database, which contains historical trading data for all American listed stocks. However, few use it in their classrooms for three reasons: it is difficult to retrieve data; time consuming to explain and does not have recently available data, i.e. for yesterday or last month. In this paper, I present an effective way to solve those shortcomings. Two words could be used to describe my method: “trivial” and “fast”. “Trivial” means that the method is so simple that it needs just a one-page long instruction. “Fast” means that loading a data set takes just a few seconds. For example, it takes just 5 seconds to load CRSP monthly data (which contains 32,676 stocks over 92 years). Another advantage of my approach is that we could retrieve IBM’s data from 1926 up to yesterday by combining CRSP with Yahoo!Finance. Over the last 5 years, more than 50 schools have launched Business Analytics/Data Analytics programs. For those programs, CRSP is an ideal database they should use for machine learning, data visualization, cluster analysis, portfolio optimization and bankruptcy prediction.