The Determinants of the Merger Arbitrage Spread: Panel Data Approach Evidence from the UK.

Abstract

This thesis investigates merger arbitrage, an important hedge fund strategy. It addresses a significant gap in the literature by taking a panel data approach to ascertain the statistically significant determinants of the merger arbitrage spread (“spread”) for mergers and acquisitions (“M&A”) transactions involving UK listed companies. It also contributes to the extant literature concerning the risk return trade‐off and level of efficiency of the UK equity market. A spread is created when one listed company, the bidder, announces its intention to acquire another listed company, the target, and the consideration offered to the shareholders of the target contains an equity element. It is calculated as the percentage difference between the value of the bidder’s offer on a per share basis and the share price of the target and varies throughout the offer period which ends when the deal either completes or fails. The collected data, a sample of 36 recent M&A transactions involving UK listed companies, demonstrate that spread for completed deals converges to parity over time as deal completion uncertainty is resolved. Conversely, there is no convergence of the spread for deals which fail to complete. These results are as expected in an efficient equity market with the presence of active arbitrageurs. The data also shows that the transactions take different lengths of time to complete or fail. Accordingly, and to create a balanced panel data‐set so that a range of panel data estimators could be employed, the offer period of each deal was innovatively converted from natural time into percentiles. This standardisation of time was possible because of the high frequency (minute‐by‐minute) nature of the collected data and it was found, via a number of robustness checks, not to significantly affect the conclusions drawn from the panel data regressions. The first stage of the analysis was to econometrically estimate the determinants of the spread for a typical completed and failed deal using the static fixed and random effects models with ordinary least squares and generalised least squares estimators, respectively. Whilst the former produced plausible results when the autocorrelation and heteroscedasticity in the residuals were corrected for, it was found that a number of the variables were non‐stationary and the results were, therefore, potentially spurious. Evidence of an underlying cointegrating relationship between the spread and its hypothesised determinants was found for the completed deals and, accordingly, the next stage of the analysis was to estimate the determinants of the spread using a dynamic autoregressive distributed lag model with the pooled mean group estimator of Pesaran et al. (1999). The panel regression found that the statistically significant determinants of the spread were, as hypothesised, the premium payable by the bidder for the target, the recommendation of the deal by the target’s board of directors to its shareholders, the presence of an alternate bidder for the target, the receipt of regulatory clearance and the approval by the target’s shareholders. Conversely, the proportion of the consideration offered by the bidder which is in cash was found to be significant but positively related to the spread rather than negatively as hypothesised. This suggests that the proportion of cash consideration contributes to deal completion risk rather than supporting the notion that shareholders prefer the certainty of cash over other forms of consideration. For the failed deals, there was no evidence of cointegration and thus estimation of the panel was not possible. These findings lend support to the conjecture that the UK equity market is informationally efficient and also strongly suggest that merger arbitrageurs in UK deals are rewarded for bearing unsystematic risk, primarily deal completion risk. They should also be of use to merger arbitrageurs for enhancing their trading profitability, in particular by exploiting the underlying relationship and short‐run error correction mechanism identified between the spread and the tested determinants. Furthermore, the innovative methodological approach taken may stimulate further research into the determinants of the spread in other jurisdictions or may be applied to the investigation of other financial datasets.