Monetary Policy Choice in Developing Countries: A Multinomial Probit Model

The Asian crisis of the 1990s brought into question the adequacy of soft pegs for developing countries. The alternatives suggested by the literature are fixed exchange rates and inflation targeting. This study uses data for 142 developing countries whose monetary policies are classified into three mutually exclusive categories: fixed exchange rates (or hard pegs), inflation targeters, and the reference group (soft pegs). Since the dependent variable is unordered and categorical, a multinomial probit model of monetary policy choice is estimated, in which each country chooses from the three monetary policies the one that provides the highest utility. The multinomial probit model is a choice probability model that allows the three choices to be evaluated simultaneously: based on independent variables that represent country characteristics such as openness to trade, vulnerability to external shocks, fiscal dominance and central bank characteristics, the model reports two sets of results - the log of the ratio of the probability for a country to choose inflation targeting relative to soft pegs, and the log of the ratio of the probability for a country to choose fixed exchange rates relative to the reference choice. The study finds that relative to soft pegs, inflation targeters are less open to trade, are more developed, have more diversified exports and have more independent central banks (as reported by the presence of both lower turnover rates for Central Bank governors and tighter restrictions on government borrowing to finance the budget deficit). Countries with fixed exchange rates are more developed and their central banks have less actual independence than those in the reference group. After accounting for the above country characteristics, the results suggest that fiscal dominance (proxied by variables such as government consumption expenditure, central government debt, share of government purchases and cash surplus/deficit, all expressed as percentage of real GDP) does not play a role in the choice of monetary policy. Average inflation rates were also added to the previous estimations, and the inflation coefficients did not attain statistical significance in any of the specifications suggesting that inflation levels do not influence the choice of monetary policy.