The Impact of Industry Classifications on Financial Research

Preprint
Publisher Website
Google Scholar
Abstract
Although industrial classifications are an important part of financial research, few researchers explicitly recognize the data base they use in classifying industrial structure. There are substantial differences, however, in primary SIC codes for firms on Compustat and CRSP. Over 36% of the classifications disagree at the two-digit level and nearly 80% disagree at the four-digit level. Thus, research results and significance levels based on industry comparisons could differ depending on the source of SIC codes, irrespective of the underlying economics of the problem being analyzed. This research analyzes the impact of industrial classification on financial research. Using approximately 10,000 firms jointly covered by Compustat and CRSP, we simulate a typical research experiment. Specifically, we: 1) examine the extent of agreement between commonly used industrial classifications, 2) examine how SIC codes change over time, and 3) test the specification and power of alternate industrial classifications in typical financial research. Our results document dramatic differences in industrial classifications and analyze the impact of these differences upon statistical inference in finance.