Aggregate Merger Activity: New Evidence on the Wave Hypothesis

Abstract

The conventional opinion is that aggregate merger activity occurs in waves. An early proponent of this view was Nelson [13; 14] who concluded that merger behavior could be described by bursts of high activity followed by long periods of low activity. Numerous other authors have advanced similar interpretations of the data, including Scherer and Ross [16], Blair, Lane and Schary [1] and Golbe and White [7]. In fact, some have argued that merger waves remain an important but as yet not totally resolved puzzle (Brealey and Myers [2], Weston, Chung and Hoag [21]). Figure 1 presents a plot of the annual number of U.S. mergers over the period 1895-1994. It is not difficult to see from the figure how the popular belief about merger waves arose. The view that mergers come in waves has been the subject of considerable interest since Nelson's work first appeared.' However, the formal statistical evidence regarding the wave hypothesis is mixed. Shughart and Tollison [17] examine several time series of merger activity and generally do not reject the null hypotheses that each of the series examined follows either a random walk or an AR(1) process with first order autocorrelation close but not equal to 1.0. They argue that such processes are not consistent with the wave hypothesis. Golbe and White [7] on the other hand fit a sine wave based model to a similar aggregate merger series and conclude that there is a close correspondence between the data and their specification. In this paper we examine the issue of waves in merger activity through formal tests of the hypothesis that aggregate merger activity follows a stationary regime switching process. Specifically, we argue that shifts in an unobservable state variable ignite shifts in the characteristics of the process driving mergers. The appeal of this specification is threefold. First, it provides an intuitively plausible model which generates waves in merger activity. Second, as we show later, it provides an explanation for the results in Shughart and Tollison while at the same time providing an interpretation of the data which adds content to the findings of Golbe and White, thus reconciling the evidence within an intuitive framework. Third, the model also provides a foundation for further considerations of the factors which might ignite merger waves.2