Merger waves and the structure of merger and acquisition time-series
用两状态马尔可夫转换模型分析并购时间序列，发现并购活动在高低水平间切换，比传统ARIMA模型拟合更好，并给出浪潮的具体日期。
What is the best characterization of mergers and acquisitions time-series? The traditional response is that mergers occur in 'waves'. I estimate a two-state, Markov switching-regime model which should capture wave structure if it is present in the data. Linear and nonlinear diagnostics tests suggest that the switching regime model fits the data well, and better than ARIMA models. Said differently, the underlying pattern in the M&A data can be characterized by dichotomous shifts between high and low levels of activity. In addition, objective inferences about the precise dates for these waves are available through a nonlinear filter.