By contrasting the issues in

By contrasting the issues in this literature with the ones of structural breaks, Luporini (1999) contributes to the debate by investigating the performance of the Brazilian fiscal policy by dividing the original sample into sub-periods, according to the occurrence of changes in the debt/GDP ratio. In so doing, the sample she uses is divided into two sub-periods due to the impact of the oil shock of 1979 on the Brazilian public accounts. Although the results of the tests indicate the sustainability of debt in the BTL-104 1966–1996, it is found that from 1981 on this debt would follow an unsustainable trajectory. Imposing ex-ante the structural breaks in a time series, as done by Luporini (1999), constitutes a shortcoming in this procedure for lacking acceptable economic arguments. Indeed, by taking a glance at Fig. 1 in section, it is impossible to identify an exact number of these breaking changes. So, instead of being a deterministic event, it seems more theoretically plausible to accept the breaking dates as a stochastic variable. It is also worth pointing out another potential shortcoming that might be implicit in the modeling of the papers cited previously for using Brazilian historical series of public finances. This may occur because the longer are the series on these data, the more inaccurate are the informations on them, due to the lack of incorporating the so-called skeletons and/or investments of state-owned enterprises under the heading of public expenditure. Goldfajn (2002) corroborates this argument by performing simulations with different long-term economic scenarios for the Brazilian economy. He infers about the sustainability of the debt/GDP over the next decade, and concludes that even for conservative scenarios on the GDP growth rate and the fiscal surplus, the results point to debt sustainability. By the same token, Mendonça et al. (2008) estimate a fiscal reaction function via Markov-Switching model for Brazil in the post-Real Plan (1994) period. They find a regime change after 2005 and a loss of response by the Brazilian government for generating primary surplus after that year. Another study of Mendonça et al. (2008) applies the same estimation technique that allows regime change and takes into account the importance of equity adjustments in the bulk of the public debt. They confirm the sustainability of the public debt in the medium term. This article will rely upon a recent proposal of Bohn (2006), which presents a critique on the traditional techniques of stationarity and cointegration tests, besides stating that the government solvency is achieved if the debt is stationary from any finite number of differentiations. In other words, the ROI remains satisfied if revenues and expenditures are stationary in differences for any arbitrary order without any cointegration prerequisite. Thus, there is a broad class of stochastic processes that violate the conditions of stationarity and traditional cointegrations for testing the sustainability of the fiscal, although the ROI remains satisfied. A testable implication of Nuclear lamina argument is that the common practice of judging whether a government is solvent or not from the unit root and cointegration tests fails. Besides, Bohn (2006) suggests that the procedures for testing sustainability through the estimation of the government reaction function provide solid results for understanding deficit trajectories. Thus, the methodology proposed in the following section adds the possibility of structural breaks in the parameters of the government reaction function. Contrary to the arguments of Mendonça et al. (2008), the methodology scheme to be employed will allow multiple regimes in accordance with a finite number of possible structural breaks according to the coefficients of the estimated function.
Methodology
Empirical application The limitations of traditional unit root tests and cointegration techniques in the analysis of public debt sustainability, especially in the presence of structural breaks in the time series, this article relies on a combination of tax smoothing model of Barro (1979) with a procedure that allows multiple endogenous structural breaks in the coefficients of the government reaction function.