We examine changes in risk following US bank mergers in the period 1981–2014. Short-run increases in acquirer risk following mergers occur only in the first few mergers undertaken by the same acquirer, and only in systematic risk. The equity volatility of acquirers does not increase. Using a new approach to measure the long-run effect we find that these results persist, consistent with banks maintaining a constant level of total equity risk in the long run. Constant acquirer risk means that all diversification benefits of the mergers are dissipated. We measure the loss of diversification associated with mergers and find it to be 40% of the risk level in 1981. Almost all of this occurred prior to 2004. In addition, there has been a large increase in correlations between the largest banks, much of which has come from sources other than mergers. The results are inconsistent with these mergers being motivated by the ‘too big to fail’ put. They suggest that if one wanted to reduce the risk of the banking system by demerging major banks one would have to reach back to the structure that existed before 2004. Simply reversing recent mergers would not have much effect on stock market measures of risk.

The literature on international equity holdings distinguishes between home bias (overweighting of home stocks) and foreign bias (relative underweighting for more “distant” countries). The two biases can be integrated into one distance-based model. We define pure home bias as the excess of home bias relative to this model, and find pure home bias only in emerging markets. Countries with high tax rates and low credit standing have higher pure home bias, and more development comes with lower distance aversion. Methodologically, the choice of portfolio bias measure matters. We find the best measure to be a covariance-based measure relative to the world average.