instability. In the case of Model 1 and persistent shocks in only 0.4% of cases with complete markets
do we Hnd evidence that debt is explosive or ’’unsustainable” e.g estimates of b > 1. By contrast
under incomplete markets this occurs in 36.4% of cases. When we add capital accumulation to
the model then the relevant proportions are 0.8% for complete markets and 38.7% for incomplete
markets. It should be stressed that these Hndings that debt is on an ’unsustainable” path under
incomplete markets are misleading. By construction debt is sustainable and is indeed following
an optimal path. It is simply that under incomplete markets debt is used as a bu≡er and during
these periods displays such persistent and long lasting swings that simple AR models suggest data
is following a non-stationary path9.
From these simulation results we draw the following conclusions:
• Tax smoothing does not produce complete debt stability even in the case of complete markets.
Therefore evidence of substantial fluctuations in debt does not rule out the existence of
complete markets.
• Complete markets produces less volatility in debt compared to incomplete markets for a
given economic structure. Therefore changes in debt issuance practice that lead to lower debt
fluctuations reflect better Hscal insurance and a move towards complete markets.
• Debt management can have an important role in helping to achieve Hscal rules involving debt
targets (speciHed levels for α∕(1 — 6)). The better is debt management e.g. the closer policy
is to complete markets, then the quicker debt is mean reverting and the less persistent are
swings in debt.
• Under incomplete markets debt displays large and long term fluctuations which may ap-
pear ’unsustainable” and debt based Hscal rules are likely to prevent optimal policies being
followed.
These results suggest that regardless of whether the aim of debt management is to smooth
taxes or stabilize the level of debt our performance indicators will rank different portfolios in
9It should also be stressed that Bohn (2005) shows how the non-stationarity of debt alone is not sufficient to
infer that debt is unsustainable. The intertemporal budget constraint places a restriction on the relative orders of
integration of debt and the primary deHcit rather than imposes that debt is stationary.
14