The value effects of Mergers and Acquisitions in the US financial market.

To conclude the VAR analysis, it has been reported the variance decomposition as output of the VAR estimation: Variance decomposition determines how much of the forecast error variance of each variable can be explained by exogenous shocks generated by the other variables.

Thus, in the specific case of the Merger between Bank of America and Merrill Lynch, the variance decomposition obtained after the estimation of the VAR(5) model is reported in figure 3.9. Observing the table and the graph, the 99% of the forecast error variance of BOA is explained by itself; otherwise, the 70% of the forecast error variance of MER is explained by itself, while the 30% is explained by BOA.

As remarked before, the VAR analysis is used to identificate the possible relationship between two or more time series, and through the evaluation of the Impulse Response Function, it's possible to analyze the short-run effects due to the shocks caused by one of the variables inclused in the model.

Anyway, in order to obtain a future vision about the relationship between two or more series, Engle and Granger [8] in 1987 pointed out that a linear combination of two or more non-stationary series may be stationary, defining in this way a stationary linear combination that is called cointegrating equation which may be interpreted as a long-run equilibrium relationship among the variables.

Therefore, to identificate the possible long-run relationship between two series,has been used the Johansen Cointegration Test[25] proposed by Soren Johansen between 1988 and 1995[26]; this test has been applyed on the price time series (because non-stationary) of BOA and MER in the period 2005/2008. The Johansen's test has a number of desirable properties, including the fact that all test variables are treated as endogenous variables.