Consequences of Riding Takeover Waves: Australian Evidence

Abstract

This paper uses Australian data to analyze takeover bid premiums and longterm abnormal returns for mergers that occur during wave and non-wave periods. Findings reveal that bid premiums are slightly lower in wave periods, and bidding firms earn normal post-takeover returns (relative to a portfolio of firms matched on size and survival) if their bids were made in non-wave periods. However, bidders who announced their takeover bids during wave periods exhibit significant underperformance. For mergers that took place within waves, there is no difference in bid premiums nor is there a difference in the long-run returns of bidders involved during the first half and second half of the waves. We find that none of prominent theories of merger waves (managerial, misvaluation, and neoclassical) can fully account for Australian takeover waves and their effects. Instead, our results suggest that BanalEstanol et al.'s screening theory of merger activity, by combining the misvaluation and neoclassical theories, may provide a better explanation.

Aggregate Australian Takeovers: A Review of Markov Regime Switching Models

“…Second, the underlying assumption of the two‐state Markov switching regime is that a higher mean and higher variance would be observed in a wave state than in a normal state, while the Poisson distribution assumes that the mean and variance of a Markov state are equal, which makes it potentially too restrictive when characterizing wave behavior. As a result, Duong and Izan () propose to utilize the Gaussian distribution and extend the AR(1) processes using more general ARMA modeling. They use Kim's () methodology, which combines the State‐Space model with Hamilton's () Markov switching model in order to cover more generalized time series.…”

“…In another, merger time series can be captured by a linear process (e.g., Shughart and Tollison ; Barkoulas et al. ), or by a nonlinear two‐state Markov switching model (e.g., Town ; Gartner and Halbheer ; Duong and Izan ). The latter approach is preferable in this analysis since the existence of merger waves is quite obvious in recent literature (e.g., Rhodes‐Kropf and Viswanathan ; Harford ; Banal‐Estanol et al.…”

“…It is believed that the two‐state Markov switching model alone does not optimize wave locality and may not be sufficient to describe transient behavior. Consequently, Duong and Izan () propose a model that uses the procedure described in Kim () to analyze the time‐series behavior of merger activity and verify the existence and extent of merger waves as Hamilton's () approach was formulated for AR processes and it cannot be easily modified to accommodate ARMA processes. The two‐state Markov switching‐regime model for ARMA(1,1) process of Duong and Izan () can be represented by the equation: …”

“…Because data for some macroeconomic variables are only available for periods in the late 1970s, this analysis only examines the influences of macroeconomic and financial market variables lagged by one‐quarter on the current level of takeover activity during the period 1980–2004. Figure shows the probability of being in a wave state when modeling annual time series of the number of Australian takeover bids by all three models: Gaussian AR(1) (Hamilton ), Poisson AR(1) (Kendig ), and State‐Space ARMA(1,1) (Duong and Izan ). As quarterly data series are used in the regression, it is assumed that the probability of four quarters in a given year remains the same.…”

“…There are currently two Markov switching models specifically applied to takeover data in the Australian market. They are Poisson Markov switching AR(1) model (Kendig ), and State‐Space Markov switching AR moving average (ARMA) (1,1) model (Duong and Izan ). As both models are derived from the Gaussian Markov switching AR(1) model of Hamilton (), it is necessary to include this model in the comparison although it is primarily used to characterize US takeovers (Town ; Linn and Zhu ).…”