This paper attempts to shed light on the causes of the procyclical behavior of fiscal policy in Latin America. This empirical regularity was brought to the forefront of recent academic and policy discussions by Gavin and others and Gavin and Perotti, and it is now a widely accepted characterization of fiscal policy in the region.1 The paper's contribution is twofold: it attempts a rigorous documentation of the procyclical nature of fiscal policy in Latin America, and it seeks to establish a causal link between limited creditworthiness—that is, the lack of access to credit during economic downturns—and procyclical fiscal policy.
The authors first compute the structural primary balance for nine countries in the region for the period 1981-2004: they use a Hodrick-Prescott filter to cyclically adjust revenues and then compute the structural primary balance as the difference between the cyclically adjusted revenues and actual expenditures. They find that the structural primary balance is negatively correlated with the output gap, implying procyclical fiscal policy. This result is not surprising, since it confirms the results found in the literature.2 The authors claim that "the empirical approaches for rigorously testing and explaining the issue are scant, despite the conventional wisdom that fiscal policy is procyclical in Latin America." Their structural balance calculation, however, does not necessarily improve on the currently used methodologies. The literature to date shows that the key difference between developed and developing countries is in the behavior of government expenditures, not revenues, with procyclical spending in developing countries and countercyclical spending in developed ones. Since the authors' methodology cleans out the cyclical component of revenues but not of expenditures, their methodology is simply capturing the positive correlation between spending and the cyclical component of GDP, which is already well documented. [End Page 185]
The paper's primary innovation is the attempt to find a causal relation between limited creditworthiness and procyclical fiscal policy. The authors claim that the markets' perception of the sustainability of Latin American economies worsens in economic downturns. This reduces access to credit for these countries, which forces a fiscal adjustment and thus leads to procyclical fiscal policy. To test this hypothesis, the authors construct a measure of fiscal sustainability at each point in time. They find that the structural primary balance improves when sustainability worsens. Since sustainability worsens in bad times, this would explain why Latin American governments adjust during recessions.
The authors define the current threshold balance (CTB) as the primary balance that renders the ratio of public debt to GDP stable. The ideal measure of this threshold is the primary balance at which the present value of future primary balances is sufficient to pay off the stock of net debt. Given data limitations, however, the authors estimate the current threshold balance as
where ρ is the ratio of interest payments to debt, g is the GDP growth rate, and D is the debt stock.
They find a positive correlation between changes in the estimated current threshold balance and their measure of the structural primary balance. This implies that governments adjust the discretionary component of fiscal policy when the perception of insolvency increases. Since this happens during downturns, discretionary fiscal policy is procyclical. One problem with this interpretation, however, is that it is not obvious that forward-looking agents would believe the debt position of a country to be less sustainable during a cyclical downturn. Moreover, the estimate of the current threshold balance automatically increases during downturns, so the methodology does not clearly identify a channel from perception of creditworthiness to fiscal adjustment.
Another issue is related to sample selection. If the hypothesis is that limited creditworthiness is key for procyclical fiscal policy, then it would be interesting to include countries with few financial problems in the sample. Data might also be a problem. The IMF's Government Finance Statistics database used in the paper does not include subnational spending, which is very relevant for federal countries such as Argentina, Brazil, Mexico, and Venezuela. Including subnational data probably would not alter the sign of the results, since the [End Page 186] available evidence indicates that subnational spending is also procyclical, but it...