In an influential paper Angeletos (2002) argues that, even in the absence of state contingent debt, governments can achieve a complete market outcome through issuing bonds of different maturities. The key insight is that fluctuations in the yield curve are exploited through holding or selling bonds of different maturities. This framework can therefore be used to provide insights into optimal debt management. Angeletos further claims that even if the conditions for this result (namely that there exist more maturities than there are states of nature) do not hold exactly governments can still use debt management to get close to the first best. Nosbusch (2006) provides some simulation based support for this claim. In this paper we review the ability of this complete market approach to provide a plausible model of debt management. In the first section we outline some properties of OECD debt management since 1970. We find that the composition of government debt shows substantial stability over time; governments issue positive amounts of debt instruments and that these are always a relatively small proportion of GDP. We then consider, both analytically and numerically, the implications of models of debt management where maturity structure is used to achieve complete market outcomes. In particular, we extend the existing literature in two significiant directions - by adding capital accumulation and ruling out the assumption that governments buyback all their debt every period and then restructure. Buera and Nicolini (2004) have already documented how in a basic model the complete market approach to debt management requires governments to hold debt positions which are huge positive and negative multiples of total output. We show how introducing capital accumulation and forcing authorities to hold debt to maturity exacerbates this problem even further. It also creates additional problems which reveal that the implications of the model for debt management are wildly at odds with the observed practive of OECD countries. Not only does the model recommend extreme positions but optimal positions show great sensitivity to small changes in the model and recommended portfolio shares are not a smooth function of maturity. Further allowing for capital accumulation also overturns the strong predictions of these models to issue long term debt and buy short term bonds. We conclude by offering some insights as to why the complete market approach to debt management has such strong counterfactual predictions and suggest areas of future research to provide a more plausible model of debt management.
