A macroeconomic model of U.S. Merger activity

Abstract

Studies by Nelson [The Corporate Merger, 1966], Weston [The Role of Mergers, 1961], Beckenstein [Antitrust Bulletin, 1979], Clark, et al. [Quarterly Review ofEconomics and Business, 1988] and others have sought to determine the relationship between macroeconomic variables and U. S. merger activity. Using regression analysis, most studies have found that stock prices are positively correlated to merger activity; however, there is less unanimity concerning the relationship of the business cycle and interest rates to merger activity. Only one study (Beckenstein) tested the effect of the government's antitrust attitude on mergers, and this study concluded that the government's attitude was an insignificant determinant of merger activity. Past regression analyses have suffered from either autocorrelation (due to the exclusion of significant variables) or multicollinearity. This study has sought to better specify and test the relationship between macroeconomic variables and merger activity. According to the expectations theory of merger activity, the level of merger activity should be positively related to expectations of economic growth, as evidenced by the business cycle and stock prices. Another theory suggests that merger activity is determined by capital market conditions. In general, one would expect that interest rates would be negatively correlated with merger activity. In this study, past merger activity, stock prices, interest rates, and the unemployment rate were used to predict the annual number ofU. S. mergers from 1919-79. The annual series of U. S. manufacturing and mining mergers with a value of over $1 million was obtained from Nelson [Merger Movements, 1959] and the FTC. This series was regressed on stock prices, the yield on Aaa corporate bonds, and the unemployment rate (which was a proxy for the business cycle). A preliminary Box-Jenkins univariate analysis indicated that current merger activity is significantly influenced by merger activity in the preceding two periods. Consequently, the explanatory variables included the dependent variable lagged one and two periods, and a variable representing a deterministic trend. The results indicate that stock prices are positively related to merger activity, while interest rates and unemployment are negatively related to mergers. The adjusted R2's on various specifications were in excess of.80. Multicollinearity was not a serious problem since the Belsley, Kuh, and Welsch condition numbers were less than 30. This specification was a better fit than the specifications used in past studies. A further specification added an explanatory variable which represented the government's antitrust position. The FFC's budget as a percentage of the government's budget was found to be a significantly negative variable when yield and unemployment were not included; however, when either or both of those variables were introduced, multicollinearity became a significant problem, and the FFC variable became insignificant. Other proxies of the government's antitrust position were tested, but none of these variables proved significant. First differences and log-log regressions were also tested, but the OLS specifications with lagged dependent variables was the superior model.