VII Lessons from an Empirical Model of Fiscal Policy in Emerging Markets

Donal McGettigan, Xavier Debrun, Mark Griffiths, Reza Moghadam, Ernesto Ramirez Rigo, Martin Schindler, Mats Josefsson, Cheng Lim, Christian Keller, Christoph Klingen, Chris Lane, Oya Celasun, Ashoka Mody, Dewitt Marston, and Mathew Verghis
Published Date:
September 2005
  • ShareShare
Show Summary Details
Xavier Debrun

The degree and sustainability of “fiscal effort” in Turkey can be measured using a simple empirical model of fiscal policy behavior that integrates the role played by economic, political, and fiscal institutional factors.1 The model provides estimates of feasible primary surplus targets that reflect the state of the economy, the stock of public debt, the quality of economic governance, past adjustment efforts, and the institutional capacity to generate surpluses.

The analysis suggests that the targeted primary surplus of 6.5 percent of GNP in 2004 could be achieved with the same degree of effort as in 2003. However, over the medium run, fiscal reforms (whose positive effect on the surplus-generating capacity is estimated to be between 1.5 and 2.5 percent of GNP) seem critical to sustain high primary surpluses. The analysis also indicates that greater political stability and further improvements in economic governance would lower the necessary level of primary surpluses, as borrowing costs would fall in response to stronger policy credibility. Broad-based structural reforms thus appear as an important avenue to progressively alleviate the burden of high primary surpluses while maintaining ambitious debt reduction goals.

Debt Sustainability and the Primary Surplus

From an accounting perspective, the necessary primary balance consistent with debt sustainability depends on growth, interest rates, and the stock of liabilities. In normal circumstances, if the objective of the medium-term fiscal strategy is to reduce the debt-to-GDP ratio, b, the government will have to target a primary surplus P¯t such that P¯t>bt(rg), where r is the real interest rate and g the real growth rate. The more ambitious the debt reduction objective, the higher the primary surplus target.

However, in practice, the appropriate primary surplus target goes beyond accounting identities and needs to reflect what the authorities can credibly achieve. For example, a key factor is the strength of fiscal institutions, notably in terms of their capacity to raise revenue and control expenditure. Another factor affecting the credibility of fiscal targets is the success of previous fiscal adjustment episodes. A situation in which repeated fiscal slippages have put at risk the perception of a government’s ability to honor its commitment would be particularly challenging. On the one hand, keeping debt dynamics under control may require historically high primary surplus targets. On the other, excessively high targets may be self-defeating, as creditors price in a high likelihood of failure of the adjustment strategy. Hence, when debt sustainability is at risk, it is essential to anchor the inevitably high primary surplus targets in a coherent reform strategy that lifts the capacity to sustain these surpluses in the short run while reducing the need to maintain them for too long.

The Empirical Model

The cross-country empirical model used in this chapter follows Bohn (1998), but is extended to quantify the effect of institutions on fiscal behavior. The model accounts for cross-country heterogeneity with country fixed effects and nonlinearities in the relationship between the dependent variable and explanatory variables. It is written as:

where pi,t is the primary balance in country i at time t; bi,t+1 is the public debt in country i at the end of period t–1; αi is a country-specific intercept (fixed effect); εi,t is an error term; Z denotes the institutional quality indicator; and Xj’s are cyclical variables explaining changes in the primary balance unrelated to the long-run solvency requirement.2 Finally, equation (1) also allows for a “spline” parameter that captures a possible break in the relationship between debt and the primary surplus at a debt level b*; Di is a dummy variable equal to 1 when bi,t+1 > b* and to zero otherwise.

Key Results

The model is estimated using a panel of 23 emerging market economies—including Turkey—over the period from 1990–2002.3 Fiscal policy is described by the primary balance of the general government (or the closest available entity). The quality of economic governance is captured by an indicator averaging indices of government stability, bureaucratic quality, and perceived corruption from the International Country Risk Guide (Political Risk Services Group, 2003). Xj’s are proxied by the output gap and inflation (for commodity exporters, deviations of oil and nonoil commodity prices from their Hodrick-Prescott [HP] trend are also included).

The estimated coefficients of equation (1) confirm the significant role of institutional variables (Table 7.1). However, the overall effect on the primary surplus is ambiguous, as the estimates reveal the following three channels through which institutional factors work:

  • Better institutions are associated with lower surpluses, suggesting a reduced need to generate large primary surpluses, presumably because the authorities’ commitment to honor obligations is more credible. Hence, good institutions—as reflected by political stability, low levels of corruption, and a high-quality bureaucracy—would tend to allow countries to borrow at a lower cost and with longer maturities, thereby reducing the primary surplus consistent with solvency at any given level of debt.
  • Stronger institutions are associated with a greater countercyclicality of fiscal policy, which indicates better control of expenditures during booms, when revenue windfalls and easier financial conditions make spending increases difficult to resist. Procyclical fiscal policies are frequently observed in emerging market economies (see Talvi and Végh, 2000; and Tornell and Lane, 1999), and can easily lead to a deficit bias if procyclical loosening is less than offset by procyclical tightening. This is generally the case. Good institutions limit that risk and contribute to healthier primary balances.
  • Better institutions are associated with a greater concern for solvency, as the primary surplus responds more strongly to stabilize the public debt ratio. However, this result is imprecisely estimated and should therefore be treated with caution.
Table 7.1.Panel Estimates of the Fiscal Policy Equation(Dependent variable: primary balance in percentage of GDP)
Output gap−0.081−1.1140.260
Public debt (t–1)0.1288.9110.000
Spline (50 percent)−0.129−8.2540.000
Institutional quality−0.543−4.7650.000
Debt and institutions0.0021.0050.316
Output gap and institutions0.0543.2420.001
Oil prices0.0682.4340.016
Nonoil commodities prices0.0672.1710.031
Default-restructuring dummy0.4772.6830.008
Weighted R-squared0.662
Unweighted R-squared0.504
Number of observations244
Note: Feasible generalized least square estimates with robust standard errors. Country fixed effects not reported.
Note: Feasible generalized least square estimates with robust standard errors. Country fixed effects not reported.

The country fixed effects provide additional information on surplus-generating capacity. Although they capture any country-specific feature systematically affecting fiscal performance, they also encapsulate key features of the fiscal regime relevant for surplus-generating potential, such as weak expenditure control and low revenue-raising capacity. Figure 7.1 depicts a positive relationship between the government revenue-to-GNP ratio and the estimated fixed effects.4 The relatively low revenue-to-GNP ratios characteristic of many emerging market economies partly explain why, on average and irrespective of the role of other factors (such as institutions), these economies find it difficult to generate the high primary surpluses required to stabilize high levels of public debt (see Figure 7.2). As Table 7.1 confirms, the debt-stabilizing response of the primary surplus weakens considerably (and in fact disappears) once public debt exceeds 50 percent of GNP.5

Figure 7.1.Fixed Effects and Total Government Revenues

(In percent of GDP, average over 1990–2002)

Figure 7.2.Revenue Ratios and Effective Tax Rates in Selected Emerging Market Economies

Source: IMF staff calculations.

The Case of Turkey

For Turkey, the model is used to generate predicted values for the primary surplus. These can then be compared to actual levels, and a measure of “fiscal effort” can be derived as the difference between actual and predicted primary surpluses (that is, the residuals of the regression). Table 7.2 shows significant adjustment efforts starting in 2000 and continuing through 2004, when the effort needed to reach the 6.5 percent target is estimated to remain high, at 4.2 percent of GNP. By definition, positive “residuals” cannot be observed for long unless a structural break in fiscal behavior occurs. In practice, this means that sustaining the high primary surpluses consistent with a rapid reduction in the debt-to-GNP ratio makes structural reforms an essential part of the consolidation strategy.

Table 7.2.Alternative Measures of Feasible Primary Surpluses(Nonfinancial public sector, in percent of GDP)
19911992199319941995199619971998199920002001200220032004Average 1993–20033
Actual primary balance:1 (a)n.a.n.a.−5.6−0.22.7−1.2−2.10.5−
Predicted primary balance: (b)−0.8−1.40.0−1.50.4−−
Relative adjustment effort:2
Measures of feasible surplus:
(b) + maximum effort over 1993–20034.
(b) + average positive effort 1993–20031.
(b) with “ambitious” fiscal reforms41.
(b) + average positive effort 1993–2003, and “ambitious” fiscal reforms4.
(b) + average positive effort 1993–2003, and fiscal reforms53.
Memo items: Total revenue22.824.125.325.623.724.627.129.632.139.438.834.334.830.5
Average fixed effect in primary surplus equation1.8
Turkey fixed effect in primary surplus equation2.8
Expected Turkey fixed effect on the basis of revenues1.3

Targeted surplus for 2004 as in the “program baseline” of the fiscal strategy note.

The average is based only on positive efforts.

The average fiscal effort is calculated based on the positive numbers only.

The potential effect of “ambitious” fiscal reforms on the surplus-generating capacity is estimated as (minus) the country’s fixed effect in the primary surplus model.

The primary surplus effect of fiscal reforms is estimated as the difference between the country fixed effect and the predicted fixed effect on the basis of revenues (see Figure 7.2).

Targeted surplus for 2004 as in the “program baseline” of the fiscal strategy note.

The average is based only on positive efforts.

The average fiscal effort is calculated based on the positive numbers only.

The potential effect of “ambitious” fiscal reforms on the surplus-generating capacity is estimated as (minus) the country’s fixed effect in the primary surplus model.

The primary surplus effect of fiscal reforms is estimated as the difference between the country fixed effect and the predicted fixed effect on the basis of revenues (see Figure 7.2).

Turkey’s low country fixed effect in relation to revenues suggests that the surplus-generating potential can be improved by enhancing expenditure management. In Figure 7.1, the expected increase can be measured by the vertical distance separating Turkey from the regression line, which is 1.3 percentage points of GNP. There is thus merit in paying particular attention to expenditure management in the fiscal reform plan. In that respect, it is worth noting that the low country fixed effect is reflected one-for-one in the average fiscal effort reported in Table 7.2. Hence, fiscal reforms resulting in a greater fixed effect would reduce the measured effort, bringing the current consolidation strategy in line with “normal,” and therefore more easily sustained, fiscal behavior.

On the revenue side, there also seems to be potential for improvement. In comparison with other emerging market economies, Turkey’s total revenue ratio is not high. Besides, the direct tax-to-GNP ratio is strikingly low (Figure 7.2), whereas the indirect effective tax rate is above average for emerging market economies. This lends support to the view that tax administration and tax policy reform, along with a rebalancing in the composition of revenues, could prove fruitful.

The results bring to the fore the issue of the highest feasible primary surplus given Turkey’s economic situation and institutions. Table 7.2 considers various possibilities, based on the continuation of past consolidation efforts (that is, ignoring negative values of the effort variable in the sample) and on the plausible benefits of fiscal reforms. Depending on the methodology used, the feasible range is estimated to be between 3.5 percent and just over 6 percent of GNP, on average. Clearly, more ambitious reforms and greater effort could, of course, deliver higher surpluses. However, structural improvements in expenditure control and revenue-raising capacity would likely be needed to avoid adjustment fatigue, given the degree of recent fiscal effort.

Further insights into risks to the sustainability of fiscal efforts can be gained from simple descriptive statistics of fiscal adjustments undertaken in the sample. An interesting first question is how persistent have fiscal adjustments been. Persistent fiscal adjustment is defined here by two criteria: (i) a large initial adjustment (that is, an initial increase of 1 percent of GNP in the primary balance, which turns out to be the threshold for the 75th percentile of all adjustments in the sample); and (ii) an average primary balance over the three subsequent years that is at least 1.5 percent higher than at the end of the initial year.

Although 60 cases satisfy the large-initial-adjustment criterion, only 10 of them pass the persistence criterion. For these persistent adjustment cases, Figure 7.3 shows in the top panel the typical path of primary balance three years before and up to five years after the initial adjustment year. The median adjustment over six years (from t–1 to t + 5) turns out to be quite large (on the order of 6 percentage points of GNP). The bottom panel of Figure 7.3 compares the adjustment path with the median, and indicates that the adjustment path in Turkey is somewhat higher than the median path of persistent adjustments, especially at the end of the period (2003–04).

Figure 7.3.Successful Fiscal Adjustments in Emerging Market Economies

But how does Turkey’s adjustment compare against all other attempts (failed as well as successful) in the sample? Figure 7.4 shows the probability distribution of cumulative fiscal adjustment after four, five, and six years. Although not without precedent, Turkey’s fiscal adjustment would be unusually successful if maintained, when compared to all other cases of intensive adjustment.

Figure 7.4.Probability Distribution of Fiscal Adjustment in Emerging Market Economies


The empirical evidence presented in this chapter indicates that, by any measure, the recent fiscal adjustment in Turkey has been remarkable in size and longevity, at least when compared to a sample of 34 emerging market economies over the period from 1990–2002. If maintained, it would easily qualify as one of the few successful (persistent) adjustment episodes in that sample.

The challenge now is to sustain Turkey’s strong performance. More specifically, the challenge is to reduce the debt ratio at least to below 50 percent, the threshold identified by the econometric analysis in this chapter as an upper limit to what a typical emerging market economy can sustain without endangering long-run sustainability. The empirical evidence points to a risk of adjustment fatigue, as large fiscal efforts can hardly be continued without a permanent change in fiscal behavior. It therefore appears critical to anchor the medium-term fiscal consolidation strategy in a broad structural reform package that would at the same time boost the surplus-generating potential of the country and progressively reduce the need for exceptionally high surpluses. From that perspective, the analysis points to expenditure and tax policy reforms as two clear areas of priority.

1The model draws on and extends work published in the September 2003 edition of the International Monetary Fund’s World Economic Outlook. The current variant of the model was reestimated using the most recent fiscal data for Turkey.
2See Favero (2002), Gali and Perotti (2003), and Fatás and Mihov (2003) for detailed discussions of the issues related to the specification for fiscal policy equations.
3The total sample contains 34 countries. See IMF (2003) for details.
4Figure 7.1 excludes the transition economies, namely Poland, Hungary, Ukraine, and Russia, which are more in line with advanced economies than typical emerging market economies.
5The 50 percent threshold maximizes the model’s goodness of fit. As shown in IMF (2003), such a break in the debt-stabilizing response of fiscal policy sharply contrasts with the behavior of fiscal authorities in advanced economies, which tends to raise the debt-stabilizing response of the primary surplus once the public debt ratio exceeds 80 percent of GNP.

    Other Resources Citing This Publication