Cross-Section Regularities in Returns on Investment in Common Stocks

Abstract

Fisher and Lorie and the University of Chicago Center for Research in Security Prices (sponsored by Merrill Lynch, Pierce, Fenner, and Smith) continue to perform a major service by collecting and presenting data on stock prices in Some Studies of Variability of Returns on Investments in Common Stocks.' Their basic data can be stated in the form of a rectangular matrix of annual wealth ratios by company and by year. There are an average of around 900 companies and 40 years so that there are an aggregate of 35,407 individual company, annual wealth ratios. They develop and report on many distributions of wealth ratios and say: results are presented primarily in tables which, we hope, will provide reference material for specialists in the field. Since we have spent considerable time examining the material in the tables we have made a few comments. However, most analysis will be left to the reader (p. 100). I would like to take advantage of this invitation to discuss some regularities shown in their tables involving the division of total variance between variance over time and other variance and the effects of crosssection diversification brought about by building portfolios with different numbers of stocks. The large mass of data should permit adequate testing. Unless otherwise specified, all standard deviations of distributions of terminal wealth used in this paper are taken from Fisher and Lorie's tables.