Using variation in bank scope generated by the stepwise repeal of the Glass–Steagall Act in the US, we show that the deregulation of universal banks allowed them to finance firms with 14% higher volatility. This increase in risk is compensated by lasting improvements in firms’ total factor productivity of 3%. Using bank scope-expanding mergers to identify shocks to universal banks’ private information about borrower firms, we provide evidence that informational economies of scope across loans and non-loan products account for the firm-level real effects of universal banking.

How do bank networks facilitate information flows that shape market outcomes? Using international banks’ advisory activities in corporate takeovers as their source of private information, we show in supervisory data that banks with closer ties to the target, but not the acquirer, advisor trade profitably in the target’s stock prior to the deal announcement. This trading behavior is associated with a higher premium paid without compromising deal success. As connected banks’ incentives are aligned only with target shareholders’ interests, which are represented by the target advisor, our evidence suggests that economic incentives determine which banks share information and with whom.

Cash- and stock-financed takeover bids induce strikingly different target revaluations. We exploit detailed data on unsuccessful takeover bids between 1980 and 2008, and we show that targets of cash offers are revalued on average by +15% after deal failure, whereas stock targets return to their pre-announcement levels. The differences in revaluation do not revert over longer horizons. We find no evidence that future takeover activities or operational changes explain these differences. While the targets of failed cash and stock offers are both more likely to be acquired over the following eight years than matched control firms, no differences exist between cash and stock targets, either in the timing or in the value of future offers. Similarly, we cannot detect differential operational policies following the failed bid. Our results are most consistent with cash bids revealing prior undervaluation of the target. We reconcile our findings with the opposite conclusion in earlier literature (Bradley, Desai, and Kim, 1983) by identifying a look-ahead bias built into their sample construction.