By Pierre-Richard Agénor
A dynamic stochastic general equilibrium model is used to study the optimal fiscal response to commodity price shocks in a small open low-income country. The model accounts for imperfect access to world capital markets and a variety of externalities associated with public infrastructure, including utility benefits, a direct complementarity effect with private investment, and reduced distribution costs. However, public capital is also subject to congestion and absorption constraints, with the latter affecting the efficiency of infrastructure investment. The model is parameterized and used to examine the transmission process of a temporary resource price shock under a benchmark case (cash transfers) and alternative fiscal rules, involving either higher public spending or accumulation in a sovereign fund. The optimal allocation rule between spending today and asset accumulation is determined so as to minimize a social loss function defined in terms of the volatility, relative to the benchmark case, of private consumption and either the nonresource primary fiscal balance or a more general index of macroeconomic stability, which accounts for the volatility of the real exchange rate. Sensitivity analysis is conducted with respect to various structural parameters and model specification.
The topic of this paper must be the most burning one right now, with various oil-exporting developing economies going through a very hard times following the spectacular drop in oil prices. The paper looks at the angle of a windfall in oil revenues, but the model is symmetric and one can thus draw opposite conclusions for the opposite shock. But is this right? This maybe a poster case for when log-linearization around a steady-state does not apply. The shocks are too big to be only small(ish) deviations from the steady-state that are OK with linearization, and it seems to me that there is a definitive non-linearity once you consider that a country may have to default on sovereign debt. Of course, looking at this kind of situation may not be what the author had in mind when drawing the model and its solution method.