News Archive
TRENDS IN PRIVATE ANTITRUST LITIGATION: 1980-2004
Andrew E. Abere, Ph.D.
Senior Economist, Princeton Economics Group, Inc.
Introduction
November 2005 marked the 20th anniversary of the Georgetown project conference on private antitrust litigation. The project, commenced in 1983, was a two and a half year effort to compile a new database on private antitrust litigation designed to inform the antitrust policy debate. The project was motivated by mounting concern about the nature, volume, and consequences of private antitrust litigation, on which there had been little systematic study. At the time, more than 1,000 private antitrust cases were filed annually in Federal district courts. The anniversary of the conference motivated us to collect and analyze data on the number of private antitrust cases filed since the time of the Georgetown project in the early 1980's.
Trends Since 1980
In the early 1980's, more than 1,000 private antitrust cases were filed annually in Federal district courts. In Figure 1, we present data on the number of private antitrust cases filed in Federal district courts for the period 1980 to 2004. From 1980 to 1983, for example, the average number of private cases filed per year was 1,245. It is interesting to note, though, that there was a downward trend during this period that continued through the rest of the decade, when the number of cases filed reached a low of 452 in 1990.
The number of private cases has trended upward since 1990, although there was a slight dip in the late 1990's. The number of cases filed just at the end of the 1990's markedly increases, rising, from 608 in 1999 to 811 in 2000. This may have been due to the numerous private antitrust cases brought against Microsoft around that time.
Despite the upward trend since 1990, the level of private antitrust activity since the end of the 1990's has been significantly less than that of the early 1980's when the Georgetown project was undertaken. For the five years from 1980 to 1984, the average number of private cases filed per year was 1,216 versus 757 for the years 2000 to 2004.
We were curious to know whether the number of private antitrust cases was related to the level of Federal enforcement. Often, a private case against a defendant may be a "follow-on" case to a successful government action against the same defendant. In Figure 2, we present data on the number of U.S. government antitrust cases (criminal and civil) filed in Federal district courts for the period 1980 to 2004. The number of U.S. government filings in any given year is a fraction of the number of private filings in the same year, such that private filings never account for less than eighty percent of all antitrust filings in a year.
There is a modest positive correlation between the number of private cases filed and the number of U.S. government cases filed. We present a graph of the two plotted against each other in Figure 3. The correlation coefficient between the two is 0.40. This result is statistically significant at the five percent level.
We also considered whether there may be a lag between the level of Federal enforcement and private litigation, in that private litigation may be generated by earlier, rather than contemporaneous, government enforcement. There is no statistically significant correlation between the number of private cases filed annually and the number of U.S. government cases filed in the prior year. There is however, a modest and statistically significant positive correlation between the number of private cases filed annually and the number of private cases filed two years earlier. The correlation between the two is virtually the same as the correlation between the contemporaneous measures.
Thus, the level of private enforcement does not seem to be strongly related to the level of government enforcement. These results should, however, be interpreted with caution for a number of reasons. For one, not all government cases may be the same. One government case may spawn only a few "follow-on" private cases, while another may spark tens or hundreds. In addition, the number of government cases filed may be an imperfect measure of enforcement activity. Government action does not necessarily involve the filing of a case (e.g., the use of "fix-it-first" remedies in the merger and acquisition area). Finally, the government (specifically, the Department of Justice) has the ability to bring criminal cases as well as civil cases, and thus its enforcement activities may not be as sensitive to changes in factors that affect private cases (e.g., changes in the standard for summary judgment).
Conclusion
Private antitrust litigation, as measured by the number of cases filed in Federal district courts, has increased in the last several years. The number of cases, though, remains below the level of the early 1980's, when there was concern raised about the volume of private antitrust litigation. The level of private antitrust litigation does not seem to be strongly related to the level of U.S. government cases, but the number of government cases may be an imperfect measure of enforcement.
1 A volume of edited papers and comments from the conference was published in 1988. See White, ed., Private Antitrust Litigation: New Evidence, New Learning (Cambridge: MIT Press, 1988).
2 Data are from the Administrative Office of the United States Courts. Data from 1980 to 1991 are for the year ended June 30 and data from 1992 to 2004 are for the year ended September 30.
3 See Thomas M. Kauper and Edward A. Snyder, "Private Antitrust Cases That Follow on Government Cases," in White, ed., Private Antitrust Litigation: New Evidence, New Learning (Cambridge: MIT Press, 1988).
4 U.S. government cases include both civil and criminal filings, and include cases where the U.S. government was a plaintiff or a defendant.
5 The correlation coefficient is a measure of the linear association between two variables. The value of the coefficient ranges from -1.0 (which indicates two series move perfectly in opposite directions) to +1.0 (which indicates two series move perfectly in the same direction). A value of zero indicates no relationship.
6 By this we mean that the probability that we obtained the observed result in our sample when there is in fact no relationship between the two is less than five percent.
7 The correlation coefficient between the two is 0.30 but this is not statistically significant.