رژیم نرخ ارز و عملکرد درآمد در کشورهای جنوب صحرای آفریقا

It has been argued that the institutions of the CFA Franc zone may have reduced inflation but that they also induced misalignment of the real exchange rate and that this is the explanation for their dismal revenue performance. This paper uses a panel of 22 countries in Sub-Saharan Africa to estimate revenue performance over the period from 1980 to 1996. It finds that the poor cumulative relative revenue performance of the Franc zone countries is mainly attributable to differences in environmental and structural factors, and to their different responses to changes in the equilibrium real exchange rate, but that the misalignment of the real exchange rate also played a part.

During the 1980s and early 1990s, the countries of the CFA Franc zone experienced a steady decline in the ratio of tax to GDP, from almost 15% of GDP in the early 1980s to only 11.5% in the early 1990s. By contrast, over the same period the tax yield in countries outside the zone remained remarkably stable at around 15.5%. Following the devaluation of the CFA Franc in January 1994, the decline in the tax ratio halted and there is evidence of a slight recovery by 1996. This pattern of relative decline in the CFA zone is repeated for each of the major components of total tax revenue, namely income taxes, domestic indirect taxes and taxes on international trade (see Fig. 1 and Table 1).1A natural question is whether these differences can be explained solely in terms of a different evolution of the determinants of revenue productivity or whether the two groups of countries differ in a more fundamental sense. A substantial literature has explored how the institutional characteristics of the zone may have contributed to a distinctive macroeconomic response to external events. This literature argues that the mechanism guaranteeing the convertibility of the CFA Franc, the overdraft facility provided by the French Treasury, relieves the domestic economy of the obligation to adjust to external disequilibria, at least in the short to medium term. It has been suggested that whilst this structure successfully delivered low inflation and high growth in the early 1980s, the inflexibility it imparted to the domestic price system retarded adjustment to deteriorating external circumstances in the late 1980s and 1990s. Thus, domestic price inflexibility contributed to the emergence of large and persistent misalignment of the real exchange rate relative to its equilibrium level in the CFA zone, long after other economies of Sub-Saharan Africa had established more flexible exchange rate arrangements (see, for example, De Melo and Devarajan, 1991). Nashashibi and Bazzoni (1994) develop this argument, conjecturing that: a major factor in the deterioration of fiscal performance in the fixed-rate [i.e., CFA] countries during the second half of the 1980s was that the real exchange rate increasingly diverged from its equilibrium path. Conversely, the variable rate countries were able to improve their fiscal performance because their real exchange rate was converging towards its equilibrium path [pp. 118–119]. In this paper, we subject this conjecture to econometric scrutiny by conducting an empirical investigation of revenue performance in Sub-Saharan Africa for the period from 1980 to 1996. The paper follows in the tradition of Heller (1975) and Leuthold (1991) and corresponds quite closely to a recent paper by Ghura (1998), which examines the impact of macroeconomic policy choices and of measures of corruption on total revenue productivity for the period 1985–1996. Ghura's work partially supports the Nashashibi and Bazzoni view by finding significant differences between CFA and non-CFA countries—reflected in a significant group-specific intercept shift—but she finds no significant role for changes in the real exchange rate or its fundamentals. Although we use a similar data set, we are able to examine more closely the role of the real exchange rate by adopting a less restrictive econometric specification and by disaggregating total tax revenue into its principal components which, from a theoretical perspective, may be expected to respond differently to the real exchange rate and its determinants.2 This approach allows us to address two key issues not addressed in earlier work. First, since our sample spans the period from 1980 to 1996, we can explicitly test a time-invariant common model of revenue performance against an alternative model in which the transmission from external and policy factors to revenue performance differs between the two regimes and over time. Of particular interest, of course, is whether the CFA devaluation of 1994 has altered revenue productivity in the zone. Second, by estimating an explicit model for the evolution of the equilibrium real exchange rate, we can distinguish between the response of tax revenue to the misalignment of the real exchange rate as opposed to equilibrium movements in it. Our empirical strategy is not without limitations. As with most studies of revenue performance, we lack reliable information on the evolving structure of the tax system, hence our analysis is restricted to crude tax yields. However, while it is not possible directly to control for changes in the tax regime, panel data methods do allow us to control for unobservable time-invariant country-specific characteristics determining revenue performance, which will include amongst other features the general stance of the tax system (see also Leuthold, 1991). A second limitation is that we are unable to distinguish between the effects of the different institutional elements of the CFA zone, for example, the effect of the fixed nominal exchange rate and the effect of guaranteed convertibility. At best, therefore, we are able to comment on the net effect on revenue productivity of the CFA institutions as a whole rather than directly on the role of the nominal exchange rate regime itself (although our focus on the real exchange rate and the inflation rate means that we do control for some of the implications of the regime). To set the stage, we start with a brief presentation of the data including some preliminary estimates of the buoyancy of the tax system. This is followed in Section 3 with a simple model of revenue performance, which identifies the role of the real exchange rate and its determinants (the so-called fundamentals). A complication for the empirical analysis is that these ‘fundamentals’ may have a direct impact on revenue productivity as well as an indirect one via their effect on the exchange rate. The model thus provides a framework within which these effects can be disentangled in the empirical application. Section 4 presents the results from the estimated revenue and real exchange rate equations while Section 5 concludes with a summary of the results

A natural question is whether these differences can be explained solely in terms of a different evolution of the determinants of revenue productivity or whether the two groups of countries differ in a more fundamental sense. A substantial literature has explored how the institutional characteristics of the zone may have contributed to a distinctive macroeconomic response to external events. This literature argues that the mechanism guaranteeing the convertibility of the CFA Franc, the overdraft facility provided by the French Treasury, relieves the domestic economy of the obligation to adjust to external disequilibria, at least in the short to medium term. It has been suggested that whilst this structure successfully delivered low inflation and high growth in the early 1980s, the inflexibility it imparted to the domestic price system retarded adjustment to deteriorating external circumstances in the late 1980s and 1990s. Thus, domestic price inflexibility contributed to the emergence of large and persistent misalignment of the real exchange rate relative to its equilibrium level in the CFA zone, long after other economies of Sub-Saharan Africa had established more flexible exchange rate arrangements (see, for example, De Melo and Devarajan, 1991). Nashashibi and Bazzoni (1994) develop this argument, conjecturing that: a major factor in the deterioration of fiscal performance in the fixed-rate [i.e., CFA] countries during the second half of the 1980s was that the real exchange rate increasingly diverged from its equilibrium path. Conversely, the variable rate countries were able to improve their fiscal performance because their real exchange rate was converging towards its equilibrium path [pp. 118–119]. In this paper, we subject this conjecture to econometric scrutiny by conducting an empirical investigation of revenue performance in Sub-Saharan Africa for the period from 1980 to 1996. The paper follows in the tradition of Heller (1975) and Leuthold (1991) and corresponds quite closely to a recent paper by Ghura (1998), which examines the impact of macroeconomic policy choices and of measures of corruption on total revenue productivity for the period 1985–1996. Ghura's work partially supports the Nashashibi and Bazzoni view by finding significant differences between CFA and non-CFA countries—reflected in a significant group-specific intercept shift—but she finds no significant role for changes in the real exchange rate or its fundamentals. Although we use a similar data set, we are able to examine more closely the role of the real exchange rate by adopting a less restrictive econometric specification and by disaggregating total tax revenue into its principal components which, from a theoretical perspective, may be expected to respond differently to the real exchange rate and its determinants.2 This approach allows us to address two key issues not addressed in earlier work. First, since our sample spans the period from 1980 to 1996, we can explicitly test a time-invariant common model of revenue performance against an alternative model in which the transmission from external and policy factors to revenue performance differs between the two regimes and over time. Of particular interest, of course, is whether the CFA devaluation of 1994 has altered revenue productivity in the zone. Second, by estimating an explicit model for the evolution of the equilibrium real exchange rate, we can distinguish between the response of tax revenue to the misalignment of the real exchange rate as opposed to equilibrium movements in it. Our empirical strategy is not without limitations. As with most studies of revenue performance, we lack reliable information on the evolving structure of the tax system, hence our analysis is restricted to crude tax yields. However, while it is not possible directly to control for changes in the tax regime, panel data methods do allow us to control for unobservable time-invariant country-specific characteristics determining revenue performance, which will include amongst other features the general stance of the tax system (see also Leuthold, 1991). A second limitation is that we are unable to distinguish between the effects of the different institutional elements of the CFA zone, for example, the effect of the fixed nominal exchange rate and the effect of guaranteed convertibility. At best, therefore, we are able to comment on the net effect on revenue productivity of the CFA institutions as a whole rather than directly on the role of the nominal exchange rate regime itself (although our focus on the real exchange rate and the inflation rate means that we do control for some of the implications of the regime). To set the stage, we start with a brief presentation of the data including some preliminary estimates of the buoyancy of the tax system. This is followed in Section 3 with a simple model of revenue performance, which identifies the role of the real exchange rate and its determinants (the so-called fundamentals). A complication for the empirical analysis is that these ‘fundamentals’ may have a direct impact on revenue productivity as well as an indirect one via their effect on the exchange rate. The model thus provides a framework within which these effects can be disentangled in the empirical application. Section 4 presents the results from the estimated revenue and real exchange rate equations while Section 5 concludes with a summary of the resultsCounterfactual 1 implicitly assumes that all the elements of the vector X are exogenous to the policy process. Certainly, the vector contains structural characteristics, such as the composition of production, the terms of trade, the level of per capita income and the equilibrium real exchange rate, which are independent of the conduct of macroeconomic policy, at least in the short-run. However, X also contains factors more closely under the control of the authorities including the rate of inflation and the degree of misalignment of the real exchange rate. In Counterfactual 2, we therefore isolate the contribution of the difference in the environment made by these so-called “policy factors” from the structural factors. We define the counterfactual as: equation(19) View the MathML source where the subscript P denotes the policy components of X. Counterfactual 2 thus measures the contribution of the policy stance to total tax revenue performance. Roughly speaking, when the CF2 lies below the actual trajectory of CFA tax revenue the actual policy stance of the CFA countries worked in their own favour, and vice versa when CF2 lies above actual revenue. Thus, in the early 1980s, the lower inflation and lower real exchange rate misalignment in the CFA zone implied total revenue was higher (to the tune of 0.75% of GDP) than it would have been if these variables were at the average values of the non-CFA countries. From the mid-1980s until the CFA devaluation in 1994, even though inflation was still lower within the CFA than outside, the persistent misalignment of the real exchange rate in the CFA zone at a time when non-CFA countries were rapidly eliminating the most egregious misalignment worked to the CFA's disadvantage, lowering by 0.75% of GDP the actual revenue yield relative to the counterfactual. With the devaluation of 1994 and the fall in inflation in the non-CFAs, the difference between the two groups becomes negligible, although because of still lower average inflation there is a modest gain in favour of the CFA countries. So far, we have maintained the assumption that the behavioural response of the different country groups to their environment, broadly defined, is given. As we noted in Section 1, it was suggested that the institutional features of the CFA Franc zone contributed to the inflexibility of the regime in response to (common) external changes, most importantly in the equilibrium real exchange rate. To conclude our analysis of the results, we therefore consider a third counterfactual in which we assume that the CFA countries responded to movements in the equilibrium real exchange rate in the same fashion as occurred in the non-CFA countries. Letting the subscript R denote the equilibrium real exchange rate, this counterfactual is then defined as: equation(20) In this case, there is a significant increase in the counterfactual tax yield, contributing in excess of an additional one percentage point of GDP by the end of the period. In fact, Counterfactual 3 generates a tax yield that is higher than that observed for the non-CFA countries by the very end of the period.15 Table 8 repeats this analysis for the individual components of the total tax yield where for convenience we have taken sub-period averages over the full sample. Each row of the table measures the marginal contribution to revenue in the CFA countries as a percentage of GDP under the three counterfactuals, where a positive (negative) value indicates the increase (decrease) relative to the actual outturn that would have been enjoyed by the average CFA economy.Although these counterfactuals are necessarily crude—not least since we make strong assumptions about the exogeneity of the independent variables including real GDP—three main conclusions emerge from Fig. 5 and Table 8. The first is that differences in structure taken crudely account for a large proportion of the difference in revenue performance, given the structure of taxes in the two regions. By the mid-1990s, differences in structure accounted for approximately 63% of the total revenue gap. The second important conclusion is that our results bear out Nashashibi and Bazzoni's (1994) conjecture concerning the behaviour of CFA revenue performance in the late 1980s, which motivated this paper. Although there were initial revenue gains from low inflation, these were progressively offset by increasing real exchange rate misalignment, which reached a peak in the early 1990s before the devaluation of the CFA Franc. Indeed, over the middle decade of our sample, 1983–1992, our second counterfactual suggests that the CFA revenue yield would have been more or less stationary in the absence of differential misalignment and inflation instead of actually falling by around 1.5% of GDP. This effect is seen to work most strongly through the effect on trade and indirect taxes which, combined, would have been about 0.5% of GDP higher in the late 1980s had the CFA countries been able to limit the growing misalignment.16 Over the entire sample period, however, although these policy factors had a powerful effect on the time path of revenue they had no significant impact on the cumulative decline in the total revenue yield. Finally, these results suggest that the differential response to movements in the equilibrium real exchange rate does matter for revenue, adding up to around an additional 1.5% of GDP to revenue performance. The capacity of the arrangements embodied in the CFA Franc zone to deliver consistently reduced inflation relative to other countries in Sub-Saharan Africa has featured prominently in discussions of their relative merits for many years. More recently, there has been much interest in their possible role in inducing cumulative distortions, especially misalignment of the real exchange rate. The data presented earlier in this paper demonstrate the continuing truth of the proposition concerning inflation. They also underline two other important differences in performance between the two groups of countries, where by contrast the non-CFA countries have outperformed the CFA ones: they have generated growing as opposed to falling per capita incomes, and they have sustained the share of tax revenues in GDP as opposed to seeing this fall. We have had nothing to say about the first of these outcomes, but have focussed on the second. It has previously been conjectured that the CFA arrangements not only resulted in exchange rate misalignment but that this misalignment had itself caused the revenue decline. In this paper, we have derived estimates of equilibrium real exchange rates and hence of the extent of misalignment, which broadly support the first contention. Specifically, the non-CFA countries were able during the 1980s to rectify their initial very high real overvaluation, whereas this increased, albeit from a low initial level, for the CFA countries, up until the devaluation of 1994. Perhaps more importantly, allowing for country specific effects, we found that we could pool across the two types of nominal exchange rate regimes: the way in which the fundamentals determine the equilibrium exchange rate is common to both groups of countries, as theory would suggest. The CFA institutions would then have their differential impact via different inflation, different misalignment and other mechanisms, but not via a shift in the relation between the fundamentals and the equilibrium real exchange rate. We then proceeded to investigate the second contention concerning the link between the real exchange rate and revenue performance, first in a simple theoretical model, and then econometrically. We examined to what extent the differential- and relatively poor-revenue performance of the CFA countries could be attributed to three different groups of factors. The first comprised differences in structure and in the economic environment experienced by the two groups of countries (and hence inter alia in their equilibrium real exchange rates), the second comprised differences in policy factors associated with the exchange rate regime, such as inflation and misalignment, and the third concerned differences in their revenue responses to changes in the equilibrium real exchange rate. Our conclusion was that structural and environmental factors were very important, explaining well over half the discrepancy, and that during the early 1990s the failure to eliminate exchange rate misalignment contributed a large proportion of this shortfall. Finally, we suggested that differential behavioural responses to the same movements in the equilibrium real exchange rate were also important, accounting for 15–30% of the discrepancy: these presumably reflect other institutional differences between the CFA countries and the rest, over and above those specific to nominal exchange rate management. There are of course a multitude of reasons for wishing to minimize both inflation and exchange rate misalignment, and it does seem to be true that CFA-type arrangements involve a trade-off between them. From the narrow but important perspective of revenue productivity, these results suggest that reduced inflation and increased misalignment at the magnitudes associated with the CFA zone can have significant temporary revenue consequences. However, the poor cumulative revenue performance of the CFA countries over the entire period appears to result mainly from a different evolution of environmental and structural features and also from the different (institutionally determined) ways in which revenue responds to this evolution. The exchange rate regime itself does not seem to have been of central importance to this cumulative story.