cost minimization (at the cost of increasing the exposure to roll-over risk.)
Secondly, it is likely that the amount of price-indexed bonds that the market is willing
to absorb at current interest rates is limited. If the government placed increasing amounts
of such debt its interest rate would rise. The extent of indexation may also be limited by
reasons of political opportunity: inflation indexation of interest income may give rise to
pressures for extending indexation to other types of income. Moreover, it is often argued
that indexation reduces the cost of inflation and thus the incentives for anti-inflationary
fiscal and monetary policy. Fixed-rate debt may also enhance the effectiveness of monetary
policy in controlling aggregate demand (see Falcetti and Missale 2002). Finally, issuance of
fixed-rate conventional bonds can be motivated by the objective of developing a domestic
market for fixed-rate bonds.
It is worth examining under which conditions substituting fixed-rate 1-year bonds for
dollar denominated bonds and for Selic-rate bonds is optimal, while taking the shares of
the other types of debt constant at the current level. Table 12 shows (for various pairs of
the expected debt reduction and the probability that debt stabilization fails) the interest
rate differential between 1-year fixed-rate bonds and dollar denominated bonds, FP,below
which it is optimal to issue fixed-rate bonds in exchange for dollar denominated bonds.
Since the current exposure to exchange rate risk (after swaps), considering the net external
debt, is currently as high as 40%, substituting fixed-rate bonds for foreign currency debt
would be optimal even for a very high perceived probability that debt stabilization may fail.
For instance, with the current 4.3% expected return differential the exposure to exchange
rate risk would be optimal only if the perceived probability of failure were as high as 40%.
Table 13 shows the interest rate differential between 1-year fixed-rate bonds and Selic-
indexed bonds, TP, below which issuing fixed-rate bonds in exchange for Selic-indexed
bonds is optimal. With an expected debt reduction equal to 2% of GDP, and 6% probability
that the debt ratio would not stabilize, fixed-rate bonds should replace Selic-indexed bonds
as soon as the term premium falls below 2.8%. However, if the probability of failure were
lower, say 3%, then fixed-rate bonds should be issued even if the term premium were as
high as 4.6%. Although, these numbers should be regarded as just indicative, they show
the large scope for improvement in the composition of the Brazilian debt.
6. Policy conclusions
In this paper we have presented a framework for the choice of debt instruments that is
relevant for countries where fiscal vulnerability makes debt stabilization the main goal of
debt management.
The optimal debt composition has been estimated by looking at the relative impact
of the risk and cost of various debt instruments on the probability that the government
might miss the stabilization target, which we have defined as a pre-assigned level of the
debt-to-GDP ratio.
The empirical evidence suggests that a large share of the Brazilian debt should be
indexed to the price level. Price-indexed bonds appear to consistently provide a good hedge
against all types of shocks, although their role is limited in the case of EMBI shocks. Price
indexation should be preferred to Selic-rate indexation, and the share of dollar denominated
(and indexed) bonds should be drastically reduced. These policy prescriptions are robust
18