A growing literature integrates debt management into models of optimal fiscal policy. One promising theory argues the composition of government debt should be chosen so that fluctuations in its market value offsets changes in expected future deficits. This complete market approach to debt management is valid even when governments only issue non-contingent bonds. Because bond returns are highly correlated it is known this approach implies asset positions which are large multiples of GDP. We show, analytically and numerically, across a wide range of model specifications (habits, productivity shocks, capital accumulation, persistent shocks, etc) that this is only one of the weaknesses of this approach. We find evidence of large fluctuations in positions, enormous changes in portfolios for minor changes in maturities issued and no presumption it is always optimal to issue long term debt and invest in short term assets. We show these extreme, volatile and unstable features are undesirable from a practical perspective for two reasons. Firstly the fragility of the optimal portfolio to small changes in model specification means it is frequently better for fear of model misspecification to follow a balanced budget rather than issue the optimal debt structure. Secondly we show for even miniscule levels of transaction costs governments would prefer a balanced budget rather than the large and volatile positions the complete market approach recommends. We conclude it is difficult to insulate fiscal policy from shocks using the complete markets approach. Due to the yield curve's limited variability maturities are a poor way to substitute for state contingent debt. As a result the recommendations of this approach conflict with a number of features we believe are integral to bond market incompleteness e.g. allowing for transaction costs, liquidity effects, robustness etc. Our belief is that market imperfections need to be explicitly introduced into the model and incorporated into the portfolio problem. Failure to do so means that the complete market approach applied in an incomplete market setting can be seriously misleading.
