We analyze holdings of public bonds by over 20,000 banks in 191 countries, and the role of these bonds in 20 sovereign defaults over 1998-2012. Banks hold many public bonds (on average 9% of their assets), particularly in less financially-developed countries. During sovereign defaults, banks increase their exposure to public bonds, especially large banks and when expected bond returns are high. At the bank level, bondholdings correlate negatively with subsequent lending during sovereign defaults. This correlation is mostly due to bonds acquired in pre-default years. These findings shed light on alternative theories of the sovereign default-banking crisis nexus.

From a historical point of view, the main activity of investment banks is what today we call security underwriting. Investment banks buy securities, such as bonds and stocks, from an issuer and then sell them to the ?nal investors. In the eighteenth century, the main securities were bonds issued by governments. The way these bonds were priced and placed is extraordinarily similar to the system that inve- ment banks still use nowadays. When a government wanted to issue new bonds, it negotiated with a few prominent “middlemen” (today we would call them investment bankers). The middlemen agreed to take a fraction of the bonds: they accepted to do so only after having canvassed a list of people they could rely upon. The people on the list were the ?nal investors. The middlemen negotiated with the government even after the issuance. Indeed, in those days governments often changed unilaterally the bond conditions and being on the list of an important middleman could make the difference. On the other hand, middlemen with larger lists were considered to be in a better bargaining position. This game was repeated over time, and hence, reputation mattered. For the middlemen, being trusted by both the investors on the list and by the issuing governments was crucial.