Public Debt Management in Brazil
Alessandro Missale and Francesco Giavazzi*
December 8, 2003
Abstract
This paper derives the optimal composition of the Brazilian public debt by looking at
the relative impact of the risk and cost of alternative debt instruments on the probability
of missing the stabilization target. This allows to price risk against the expected cost of
debt service and thus to find the optimal combination along the trade off between cost
and risk minimization. The optimal debt structure is a function of the expected return
differentials between debt instruments, of the conditional variance of debt returns and
of their covariances with output growth, inflation, exchange-rate depreciation and the
Selic rate. We estimate the relevant covariances by: i) exploiting the daily survey of
expectations; ii) simulating a small structural model of the Brazilian economy under
different shocks; iii) estimating the unanticipated components of the relevant variables
with forecasting regressions. The empirical evidence strongly supports the funding
strategy of Brazilian Treasury in 2003 of relying heavily on fixed-rate LTN bonds. It
also supports its recent decision to revitalize the market for price-indexed bonds with
the new NTN-B program of IPCA indexation. Though decreasing, the exposure to
exchange rate risk appears too large suggesting that more efforts should be made to
reduce funding in foreign currencies.
JEL Classification: E63, H63.
Corresponding author: Alessandro Missale; Dip. di Economia e Politica Aziendale; Via
Mercalli 23; 20122 Milano; Italy. Tel. +39-02-50321512 Fax +39-02-50321505. E-mail:
[email protected]
tWe thank Fernando Blanco and Santiago Herrera who provided valuable information and data on the
Brazilian Economy. We are grateful to Alfonso Bevilaqua and Paulo Levy for insightful discussions on the
effect of GDP growth and inflation on the primary surplus. We thank Carlo Favero for many comments and
suggestions. We also benefited from comments of participants in the seminars at the Brazilian Ministry of
Finance, the Cental Bank of Brazil and IPEA. Marco Aiolfi and Andrea Civelli provided excellent research
assistance. Financial support from the World Bank is gratefully acknowledged. The authors are associated
with Universita di Milano; Universita Bocconi, IGIER, CEPR and NBER, respectively.