SOUTH AFRICA Selected Issues

This Selected Issues paper for South Africa presents a quantitative analysis of inflation dynamics in the country. The conduct of monetary policy has been complicated by a variety of unanticipated events that have had important effects on inflation. Exposed to exchange rate and other shocks, the model confirms that a delayed policy response to inflation shocks leads to persistently higher inflation rates and, subsequently, to a sharp real contraction of the economy.

Abstract

This Selected Issues paper for South Africa presents a quantitative analysis of inflation dynamics in the country. The conduct of monetary policy has been complicated by a variety of unanticipated events that have had important effects on inflation. Exposed to exchange rate and other shocks, the model confirms that a delayed policy response to inflation shocks leads to persistently higher inflation rates and, subsequently, to a sharp real contraction of the economy.

II. Public Debt in South Africa: A Risk Analysis1

A. Introduction

1. Analyses of public debt sustainability commonly rely on medium-term projections of the debt-to-GDP ratio given macroeconomic forecasts and fiscal policy assumptions. While such projections per se do not allow to determine the sustainability of a particular public debt position, the expected debt path nevertheless provides some indication as to whether the underlying policies can be sustained under plausible macroeconomic conditions without endangering government solvency. Specifically, a projected decline in the debt ratio will generally be interpreted as a signal that government policies do not jeopardize sustainability, whereas a positive trend or even stabilization at a high level may motivate concerns about sustainability.

2. Uncertainty about future macroeconomic conditions and fiscal policy inevitably weakens the diagnostic based on such analyses. An assessment of the risks affecting the “baseline” projection of public debt could thus help form a more nuanced and more credible assessment of long-term sustainability. One simple way to appraise those risks is to estimate alternative debt paths that would prevail under less favorable circumstances than in the baseline. These alternative scenarios, or “bound tests,” typically envisage pessimistic macroeconomic forecasts (low growth, high interest rates,…), fiscal policy slippages and exogenous debt shocks—such as valuation effects resulting from exchange rate movements or the recognition of off-budget obligations, including loan guarantees. Although bound tests give a good sense of the sensitivity of the sustainability assessment to adverse developments, it is difficult to quantify the risks to the baseline scenario.

3. To measure those risks properly, a complete probability distribution of the debt ratio would be needed for each year of projection. One way to estimate such distributions is to build a large sample of bound tests corresponding to different constellations of shocks likely to affect the debt dynamics. The sample can be generated by means of stochastic simulations where both the size and co-movements of random shocks are calibrated to fit the statistical properties of relevant historical data, including correlations among economic variables, as well as their respective dynamics (see e.g., Garcia and Rigobon, 2004). Particular attention also needs to be paid to the systematic response of fiscal policy to these shocks and to public debt developments themselves (accounting for the government’s explicit concern for solvency, see Bohn, 1998), a response that can be appraised with estimated fiscal policy reaction functions. One important dimension of this approach is that every individual scenario produced in the context of the simulation exercise makes a more efficient use of the data needed to carry out standard debt sustainability analysis (DSA). This chapter uses the simulation algorithm developed in Celasun, Debrun, and Ostry (2005), which explicitly accounts for endogenous fiscal policy.2

4. From a policy perspective, more complete information on the debt risk profile should help improve the design of medium-term fiscal policy plans.

  • First, awareness of the risks to public debt would promote greater caution in the conduct of fiscal policy. In particular, this could imply a lesser reliance on debt to finance new productive expenditure programs, thereby reducing the likelihood that the debt dynamics spins out of control due to extraneous macroeconomic factors and ultimately forces cutbacks in those valuable programs. More generally, governments with low credibility and operating in a volatile economy could better grasp ex-ante the potential costs of policies implying higher public debt ratios, while governments with greater credibility and facing relatively stable economic conditions could avoid taking excessive comfort in a favorable baseline outlook for public debt.

  • Second, an explicit quantification of risks to the debt dynamics could help in the design of fiscal consolidation strategies. Governments could indeed evaluate the relative merits of alternative adjustment plans in regard of the corresponding probabilities to bring the public debt below a certain target.

5. The risk analysis in this chapter confirms South Africa’s sound public debt position. Upside and downside risks are well-balanced under the baseline policy scenario, whereas the upside risks associated with a moderate fiscal expansion sustained over the medium term, and with an adverse shock on the public debt stock (akin to the recognition of contingent liabilities) appear to remain manageable. The analysis underscores the responsiveness of the primary balance to the public debt as a key determinant of that relatively benign outlook.

6. This chapter is structured as follows: Section B briefly illustrates the outcome of the traditional approach to debt sustainability; Section C describes the interaction between fiscal policy and debt dynamics, suggesting that policy behavior is important for a proper assessment of the debt risk profile; Section D describes the stochastic simulation method and applies it to South Africa; policy implications are discussed in Section E.

B. Debt Sustainability Analysis and Risk Assessment

7. To evaluate the risks to the debt dynamics, the commonly used debt sustainability analysis subjects the baseline projection to a series of deterministic and isolated shocks (“bound tests”) likely to deteriorate the outlook. These typically include lower GDP growth, higher interest rates, a weaker primary balance, a depreciation of the exchange rate, and the recognition of off-budget obligations. The historical variance of the underlying series generally determines the magnitude of the simulated disturbances, but co-movements among them are ignored. Also, and most importantly, fiscal policy is assumed not to respond to the simulated economic developments.

8. Calibrating bound tests to reflect economic and policy patterns observed in a particular economy is challenging. One possibility is to devise a small number of standardized scenarios—where isolated shocks are expressed as a fraction or a multiple of historical standard deviations of the variables—such that both the shock itself and the resulting debt path appear plausible in probabilistic terms (IMF, 2003a). By nature, this approach lends itself very well to the construction of standardized bound tests applicable to many different countries, and requires only a fairly parsimonious dataset. One inevitable downside, however, is that the underlying scenarios hardly ever follow these common patterns in the economy where shocks typically trigger persistent responses from other variables—including those that matter for debt dynamics. Moreover, the small number of scenarios prevent quantifying the risks to public debt. And from a purely presentational perspective, there is also a chance that observers might judge the plausibility of these scenarios on the sole basis of their simplified core assumptions—e.g., a growth slowdown without repercussions on interest rates or fiscal policy—rather than on the intrinsic likelihood of the resulting debt path.

9. A legitimate question is thus to ask whether a diagnostic based on a few highly stylized scenarios is sufficiently robust to more realistic constellations of shocks. If a joint distribution of economic disturbances can be estimated for the country under review, stochastic simulation reflecting actual co-movements of shocks in the economy can produce a large sample of more realistic bound tests from which frequency distributions of debt can be derived. These frequency distributions provide a quantitative assessment of the risks to baseline debt projections, that may ultimately help refine fiscal policy recommendations.

10. Another important issue is the extent to which the sustainability diagnostic is sensitive to the assumed fiscal policy behavior. Commonly used DSA scenarios assume that fiscal policy is invariant to the stylized shocks. However, recent attempts to characterize systematic patterns in the conduct of fiscal policy have established that the primary balance systematically responds to variations in public debt and to business cycle developments, among other factors.3 The estimation of fiscal policy reaction functions provides a quantitative characterization of the average relationship between economic developments and policy scenarios (see below). Integrating systematic features of the policy process into the analysis should clearly improve the reliability of the risk analysis.

11. Overall, the stochastic simulation approach emerges as a potentially interesting complement to deterministic bound-testing. The main features of the two methods are summarized below (Table II.1).

Table II.1.

—Debt Sustainability Analysis and Risk Assessment

article image

12. The outcome of the IMF’s deterministic bound-testing approach is illustrated in Figure II.1 for South Africa over the 2005–2010 time horizon. The IMF template provides debt paths corresponding to several standardized scenarios: the baseline (reflecting macroeconomic projections and policy assumptions); small but permanent, adverse shocks (half a standard deviation) to real GDP growth, the real interest rate and the primary balance; a combination of these three shocks (this time assuming a quarter of a standard deviation); and two large temporary disturbances, namely a 30 percent real depreciation and a shock to the debt stock (mimicking the recognition of contingent liabilities) equivalent to 10 percent of GDP.

Figure II.1.
Figure II.1.

South Africa: IMF Standard Debt Sustainability Analysis—2004–2010

Citation: IMF Staff Country Reports 2005, 345; 10.5089/9781451966763.002.A002

Sources: South African National Treasury and IMF staff estimates.

13. The selected bound tests suggest fairly low risks to public debt sustainability over the medium term. Although all permanent shocks entail a growing debt ratio, government debt remains below 40 percent of GDP in 2010. Only a one-off realization of contingent liabilities equivalent to 10 percent of GDP in the first year of the projection yields a significant increase in the public debt ratio. However, the size of the shock in that particular scenario may be deemed unlikely in the case of South Africa. This would indeed presume that virtually all identified contingent liabilities of the national government materialize in one year (or equivalently that national government would take over the debt of all state-owned enterprises, or any convex combination of both).

14. Before presenting the results of the stochastic simulation approach, the next section discusses the importance of fiscal policy behavior in shaping the risks to debt dynamics.

C. Debt Dynamics and the Conduct of Fiscal Policy

15. A fiscal reaction function can illustrate the contribution of fiscal policy to debt sustainability. This section shows how the systematic policy response to variations in public debt affects debt dynamics and, thereby, the risks to long-run sustainability. Estimates of explicit reaction functions for a panel of emerging market economies (including South Africa), and for South Africa itself, are also presented and discussed.

16. The linkage between fiscal policy and debt dynamics is described by the following identity:

Δdt(r(dt1)gt)dt1+pt(dt1,Z)+xt(1)

where dt is the debt-to-GDP ratio at time t ; r (d ) represents the real interest rate (which itself depends on the level of public debt, with ∂r/∂d > 0 ); g symbolizes the real growth rate; x is an exogenous shock to debt; and Δ denotes the first difference operator. The primary surplus pt is a function of the lagged public debt and other non-debt determinants (gathered in the vector Z ) such as business cycle conditions and relevant commodity prices (see below).

17. Looking at descriptive evidence for South Africa, changes in the public debt have indeed been related to the level of the primary surplus (see Figure II.2). In particular, high primary surpluses helped bring down public indebtedness while periods of rising public debt were associated with very low primary surplus and primary deficits. Overall, fiscal policy behavior has been a key determinant of public debt dynamics, and this relationship does not appear to have been weakened by significant stock-flow (“below-the-line”) adjustments.

Figure II.2.
Figure II.2.

South Africa: Changes in the Public Debt Ratio and the Primary Surplus, 1990–2005

(in percent of GDP)

Citation: IMF Staff Country Reports 2005, 345; 10.5089/9781451966763.002.A002

Sources: South African National Treasury and IMF staff calculations.

18. Fiscal policy itself seems to have been responsive to developments in the debt-to-GDP ratio (Figure II.3). The strong, positive relationship between the lagged public debt ratio and the primary surplus is consistent with the idea that debt stabilization has been a key ingredient of the fiscal policy strategy since 1994.4 It is also particularly interesting to note that starting in 2002, the primary surplus has declined significantly, suggesting that the government has taken advantage of more favorable debt dynamics (helped by higher growth and lower interest rates) to increase spending and offer tax relief. That result now needs to be corroborated econometrically. Econometric methods will also make sure that the unconditional correlation is not spurious.

Figure II.3.
Figure II.3.

South Africa: Primary Surplus and Lagged Public Debt, 1990–2005

(in percent of GDP)

Citation: IMF Staff Country Reports 2005, 345; 10.5089/9781451966763.002.A002

Sources: South African National Treasury and IMF staff calculations.

19. In standard general equilibrium frameworks, such a positive response of the primary balance to an increase in the debt ratio (i.e., ∂pt/∂dt-1 > 0) is sufficient to ensure long-run solvency (Bohn, 1998).5 If that condition is fulfilled, then the absence of drift in the elements of Z implies that the debt ratio is always going to converge towards some finite value (mean-reversion).6

20. In a partial equilibrium context, phase diagrams neatly illustrate the importance of the primary surplus’ response to dt-1. Along the separation line DD in Figure II.4, the primary surplus is such that the debt ratio is constant. The greater the difference between the real interest rate and real GDP growth, the steeper DD—the convexity of DD captures the impact of rising public debt ratio on interest rates. Any primary surplus located above (below) DD is consistent with a declining (rising) debt ratio. The line AA (left panel) depicts a fiscal reaction function with a weak response of the primary surplus to the debt. In that case, the debt ratio is either on an explosive path or converges towards zero; and shocks (vertical shifts in AA) can easily jeopardize sustainability. In contrast, the line BB (right panel) describes an aggressive response of the primary surplus to variations in the debt ratio; and the debt ratio will always converge to some finite debt level (denoted by d*), irrespective of the shocks affecting the budget.

Figure II.4.
Figure II.4.

Debt Dynamics and the Response of the Primary Surplus to the Debt

Citation: IMF Staff Country Reports 2005, 345; 10.5089/9781451966763.002.A002

Source: IMF staff.

21. The specification of the fiscal reaction function is fairly standard7 and takes the following general form:

pi,t=αi+pdi,t1+βDi(di,t1d˜)+Σk=1KγkZk,i,t+ϵi,t(2)

where i is a country-specific subscript, α,ρ,β and the γ’s are parameters to be estimated, and εi,t is an error term. Notice that αt is a country-specific intercept (fixed effect). In line with IMF (2003b), equation (2) also allows for a “spline” parameter that captures a possible break in the relationship between debt and the primary surplus at a debt level d˜ ;Di is a dummy variable equal to 1 when di,t1 >d¯ and to zero otherwise.

22. The short time-series available for fiscal variable in most countries (some of them with no more than 15 years of annual data) motivated attempts to estimate fiscal reaction functions with panel data techniques.8 One obvious caveat is that linear panel estimation presupposes identical fiscal behavior across countries. As suggested in IMF (2003b), allowing for non-linear relationships between the primary balance and its determinants may help alleviate the problem. Panel estimates presented in Table II.2 cover a broad sample of 33 emerging market economies9 over the period 1990–2004. These estimates allow for two types of “non-linearities”: a “kinked” response to the public debt (assuming that responsiveness to debt changes beyond a given threshold), and an asymmetry in the cyclical response between “good times” (positive output gap) and “bad times” (negative output gap). Celasun, Debrun and Ostry (2005) discuss in detail technical issues related to the specification and estimation of fiscal reaction functions, including the risk of a small upward bias in the estimated value for ρ.

Table II.2.

Selected Emerging Market Economies: Fiscal Policy Reaction Functions (1990–2004)

Dependent variable: Primary surplus of NFPS in percent of GDP or GNP

article image

Panel two-stage EGLS, instrumenting the output gap; country fixed effects included.

Panel IV, instrumenting the output gap; country fixed effects included.

Robust standard errors in brackets.

23. In terms of coverage, the dataset generally refers to the widest possible definition of the public sector as the government is likely to provide at least implicit guarantees to decentralized entities and key public enterprises. For most countries, debt and primary balance data cover the nonfinancial public sector (NFPS) or the general government. For data availability reasons, central government data had to be used for Côte d’Ivoire, Indonesia, Morocco, South Africa (“national government”), and Ukraine.

24. The estimated reaction functions yield interesting results regarding debt sustainability in emerging market economies (Table II.2). On average, the primary surplus exhibits a positive and significant response to the debt ratio, in line with long term debt sustainability. However, that sensitivity sharply declines once debt exceeds 50 percent of GDP, a threshold often considered in the literature as the upper limit of the “comfort zone” for an average emerging market economy in terms of debt accumulation.10 Beyond that point, the risk of embarking on an unstable debt path is indeed non trivial as illustrated in Figure II.5.

Figure II.5.
Figure II.5.

Primary Surplus and Lagged Debt in Emerging Market Economies

Citation: IMF Staff Country Reports 2005, 345; 10.5089/9781451966763.002.A002

25. The estimated reaction function also allows to quantify policy-related risks.11 For instance, fiscal policy reacts to random cyclical developments in a way that may amplify their impact on the debt ratio. The estimated coefficients of the output gap indeed point to a mildly counter-cyclical behavior, as the primary balance deteriorates (improves) in the event of an adverse (positive) GDP shock. Notice that in good times (GDP above trend level), the coefficient is not significantly different from zero, suggesting that pro-cyclical discretionary actions (for example, additional spending financed with revenue windfalls) undermine automatic stabilizers.12 Also, oil exporters’ budgets appear quite sensitive to oil price developments, with oil-related revenue windfalls or shortfalls being partly reflected in the primary balance. (The impact of non-oil commodity prices was found to be non-significant among commodity exporters). Interestingly, election years13 tend to be associated with lower primary balances, supporting the idea of a political business cycle. Hence, political instability and the risk of frequent elections may entail more frequent negative shocks to the primary surplus. Other possible sources of risks captured by the estimated reaction function include discretionary fiscal surprises (reflected in the residuals), and the uncertainty about the true magnitude of the responses to these events (captured by the standard error of estimated coefficients). The results in Table II.2 thus allow to study the effect of plausible changes in fiscal behavior (assuming that such changes have no effect on the estimated distribution of disturbances).

26. Finally, other structural and institutional factors potentially affecting the capacity to generate primary surpluses can be assessed. These factors include the overall quality of economic and political institutions (government stability, low corruption, high bureaucratic quality, efficient law enforcement), adherence to an IMF-supported program, and default/restructuring episodes. Everything else being equal, countries with better institutions seem to need lower surpluses, presumably because they enjoy greater credibility and correspondingly lower financing costs. As expected, countries implementing a program supported by the IMF generate higher primary surpluses on average, whereas the tight financing conditions faced by countries renegotiating their debt obligations (or in default) force them to run higher surpluses.

27. South Africa appears to behave quite differently from the rest of the sample as regards countercyclicality (more pronounced) and responsiveness to debt (almost 3 times stronger). This is evident from the last column of Table II.2 which provides South-Africa specific estimates for these two key coefficients, while keeping all other coefficients uniform across the panel (including for South Africa). The magnitude of these differences prompted us to base the stochastic simulation exercise on a simpler, country-specific reaction function estimated over the last 15 years (Table II.3).

Table II.3.

South Africa: Fiscal Reaction Function (1990–2004)

Dependent variable: Primary surplus of general government in percent of GDP

article image
Robust standard errors in brackets.

28. The estimated coefficients on debt and the output gap are close to those identified with panel estimation, which controls for other potential determinants of the primary surplus. They have the advantage to come from a model providing the best fit to South African budgetary data.

D. Assessing the Risks to Debt Dynamics

29. This section applies a stochastic simulation algorithm to derive frequency distributions of South African government debt over the next 5 years.14 As in the standard DSA framework, each individual simulation yields public debt projections using equation (1). However, unlike the standardized, deterministic bound tests, shocks stem from a random draw of disturbances affecting real GDP growth, the real interest rate (long-term government bond deflated by GDP prices), the real effective exchange rate, and the primary fiscal balance. The joint distribution of all non-fiscal shocks is calibrated using the variance-covariance matrix of a VAR model estimated with quarterly data over the period 1980–2004. The fiscal shock is assumed to be orthogonal to the other variables and is consistent with the standard error of regression (OLS) reported in Table II.3. New macroeconomic disturbances occur every quarter and feed into VAR forecasts of the 3 non-fiscal variables. The persistence of shocks is therefore fully accounted for. After annualization of macroeconomic variables, debt and primary balance projections are determined recursively using equation (1). Annual frequency distributions for public debt are calculated on a sample of 1000 stochastic simulations.

30. While this approach makes a more efficient use of the data entering standard DSA, two significant limitations remain. First, the disconnect between low frequency (annual) fiscal data and higher frequency macroeconomic data prevents the simulation algorithm to incorporate the feedback effect of fiscal variables (deficit and debt) on growth, interest rate and the real exchange rate.15 The uneasy mix of annual and quarterly data present in the simulation framework is a sacrifice to the fact that most countries only publish reliable and detailed budgetary data at annual (or at best semi-annual) frequency. Besides, quarterly fiscal data reflect a number of developments unrelated to policy behavior per se, and with only little bearing on medium-term sustainability concerns. Still it would be ideal from a statistical point of view to incorporate the fiscal reaction function into the VAR model whenever sufficiently long and reliable quarterly fiscal series are available (see Penalver and Thwaites, 2004).

31. Second, the risk analysis only concerns debt projections and cannot appraise the solvency risk per se. Barnhill and Kopits (2003) directly estimate the probability of government insolvency by deriving probability distributions for the government net worth. However, this “Value-at-Risk” approach is much more demanding in terms of data as it requires reliable estimates of the government balance sheet.

32. The following simulations will thus ignore potentially important aspects, such as the growth effect of higher public investment, or the reaction of interest rates to rising deficits and debts. One might conjecture that a strong response of interest rates to public debt developments is likely to widen confidence intervals around projected debt paths, whereas strong growth effects of high-quality fiscal expansions would presumably contain the upside risks associated with a deterioration of the primary balance. These are important considerations to bear in mind when interpreting the results.

33. In the remainder of this section, particular attention is paid to the public debt risk posed by three hypothetical fiscal developments in South Africa. First, it is assumed that a discretionary increase in public expenditure by 1 percentage point of GDP is sustained over 5 years and entirely financed with new debt over the first 3 years (scenario 1). After 3 years, the primary surplus is progressively allowed to reflect increases in the public debt, and an adjustment takes place (in line with the reaction function). This presumes that the expenditure boost will ultimately be funded with new revenues and expenditure reallocations. Second, a reduction in the sensitivity of the primary surplus to the public debt (by one standard deviation of the estimated coefficient—see Table II.3) is introduced to examine the implications of a reduced focus on debt sustainability in overall policy behavior. Third, a contingent liability shock identical to the IMF standard bound test (10 percent of GDP) is imposed, mainly to show the importance of introducing an endogenous policy response.

34. First, a baseline scenario is constructed, based on VAR projections of the macroeconomic variables and the reaction function reported in Table II.3. A series of shocks to interest rates, growth and the exchange rate affect the economy over a time horizon of 5 years. The risks to the debt dynamics are best summarized by a fan chart (Figure II.6). Different colors delineate deciles in the distributions of debt ratios, with the zone in dark grey representing a 20 percent confidence interval around the median projection and the overall cone, a confidence interval of 80 percent. Comparing Figure II.6 with the outcome of simple bound tests (reported in Figure II.1) suggests that the debt paths associated with most bound tests fall well within the 80 percent confidence interval. The fan chart also nicely illustrates the significant uncertainty typically surrounding public debt projections, giving a better idea of the overall risks to public debt in comparison to Figure II.1 (especially as regards the balance between upside and downside risks).

Figure II.6.
Figure II.6.

South Africa: Public Debt Risk Profile, 2005–2009 (baseline)

Citation: IMF Staff Country Reports 2005, 345; 10.5089/9781451966763.002.A002

35. Scenario 1 indicates that financing new expenditure plans with debt while retaining the same reaction function does not substantially deteriorate the overall debt risk profile. The probability to observe the national government debt above 40 percent of GDP by 2009 rises from 0.25 in the baseline to 0.38 (see Table II.4) while the median debt projection for the same year increases from 34.6 percent of GDP under the baseline to 37.7 percent. The fan chart (Figure II.7) also illustrates that the balance of risks is now tilted to the upside. The fiscal adjustment implied by the estimated reaction function is visible near the end of the forecasting horizon, with the debt ratio reverting to a steady state level of about 35 percent of GDP.

Table II.4.

South Africa. Public Debt Risk Profile under Alternative Scenarios: Key Statistics

article image
Source: IMF staff calculations.
Figure II.7.
Figure II.7.

South Africa: Public Debt Risk Profile, 2005–2009

(scenario 1: higher discretionary spending financed with new borrowing)

Citation: IMF Staff Country Reports 2005, 345; 10.5089/9781451966763.002.A002

36. A reduction in the responsiveness of the primary balance to the public debt entails a non-negligible worsening of the public debt risk profile. In that scenario, a lesser concern for debt stabilization (by one standard deviation or 0.062) entails an initial reduction in the primary balance by 2.2 percentage points of GDP (that is 0.062 times 35.8 percent, the debt-to-GDP ratio at the end of 2004). The primary balance is then allowed to adjust upwards as the debt ratio initially rises, in line with the estimated reaction function. The probability to observe a public debt ratio above 40 percent of GDP by 2009 nevertheless increases to 0.58 (up from 0.25 in the baseline). The fan chart confirms that upside risks now clearly dominate the outlook (Figure II.8), and that the change in fiscal behavior raises the steady state debt ratio to about 40 percent of GDP.

Figure II.8.
Figure II.8.

South Africa: Public Debt Risk Profile, 2005–2009

(scenario 2: lower responsiveness of the primary surplus to the debt)

Citation: IMF Staff Country Reports 2005, 345; 10.5089/9781451966763.002.A002

37. The importance of fiscal policy behavior is further illustrated by analyzing the impact a large contingent liability shock (10 percent of GDP), or, alternatively, a full takeover of SOEs’ debt. Strikingly, the probability of the debt ratio exceeding 40 percent of GDP by 2009 is below that in the scenarios 1 and 2 (0.37 against 0.25 in the baseline) despite the size of the initial shock. Again, the fan chart shows the impact of the automatic fiscal adjustment imposed by the fiscal reaction function (Figure II.9). This comes out as a startling illustration of how the responsiveness of the primary surplus to the public debt is essential in shaping the overall risk profile of a country’s public debt. The contrast with the bottom panel of Figure II.1, which shows the resulting debt path under the invariant-policy assumption is also quite remarkable.

Figure II.9.
Figure II.9.

South Africa: Public Debt Risk Profile, 2005–2009

(scenario 3: contingent liability shock)

Citation: IMF Staff Country Reports 2005, 345; 10.5089/9781451966763.002.A002

E. Conclusions and Policy Implications

38. This chapter complements the standard debt sustainability analysis—based on medium-term projections of the debt-to-GDP ratio under a limited number of scenarios—by offering a complete assessment of the risks surrounding the baseline debt projection. By means of stochastic simulations, frequency distributions of public debt are obtained over the entire forecasting horizon, allowing for a probabilistic analysis of debt dynamics. An important dimension of the study is to allow for fiscal policy to endogenously react to macroeconomic shocks while preserving the flexibility to look into the consequences of specific policy initiatives (or changes in fiscal behavior) for the public debt risk profile.

39. Overall, the analysis supports the relatively benign assessment resulting from the commonly used DSA, but brings useful nuances to the policy advice. Specifically, the chapter suggests that a protracted fiscal expansion accompanying new spending initiatives should only entail a relatively limited worsening of the debt risk profile. However, the analysis indicates that this benign outlook hinges critically on the continuation of the prudent fiscal policy behavior observed over the last 10 years. For instance, it appears that a lesser attention to public debt stabilization than in the recent past could significantly increase upside risks. Balancing upside and downside risks thus points to a strategy in which new expenditure programs would require additional revenues and expenditure reallocations to be phased in over the medium term. A focus on growth-promoting initiatives would also help contain upside risks although the algorithm used in this chapter could not quantify the gains from such focus in terms of reduced risks to the public debt.

F. References

  • Balassone, Fabrizio, and Maura Francese, 2004, “Cyclical Asymmetry in Fiscal Policy, Debt Accumulation and the Treaty of Maastricht”, Temi di Discussione No 531, (Rome: Banca d’Italia).

    • Search Google Scholar
    • Export Citation
  • Barnhill, Theodore, and George Kopits, 2003, “Assessing Fiscal Sustainability under Uncertainty”, IMF Working Paper No 03/79, (Washington, DC: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Bohn, Henning, 1998, “The Behavior of U.S. Public Debt and Deficits”, Quarterly Journal of Economics, 113: August pp.949 -63.

  • Celasun, Oya, Xavier Debrun, and Jonathan D. Ostry, 2005, “Primary Surplus Behavior and Risks to Fiscal Sustainability in Emerging Market Countries: A ‘Fan-Chart’ Approach”, mimeo, (Washington, DC: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Fatàs, Antonio, and Illian Mihov, 2003, “On Constraining Fiscal Policy Discretion in EMU”, Oxford Review of Economic Policy, 19 (1) : pp. 1 -28.

    • Search Google Scholar
    • Export Citation
  • Favero, Carlo, 2002, “How Do European Monetary and Fiscal Authorities Behave?”, CEPR Discussion Paper, No 3426, (London: Center for Economic Policy Research).

    • Search Google Scholar
    • Export Citation
  • Galí, Jordi and Roberto Perotti, 2003, “Fiscal Policy and Monetary Integration in Europe”, Economic Policy, October pp. 37:535 -572.

    • Search Google Scholar
    • Export Citation
  • Garcia, Márcio, and Roberto Rigobon, 2004, “A Risk Management Approach to Emerging Market’s Sovereign Debt Sustainability with an Application to Brazilian Data”, NBER Working Paper No 10336, (Cambridge, MA: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2003a, “Sustainability Assessment—Review of Application and Methodological Refinements”, www.imf.org, (Washington, DC: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2003b, “Public Debt in Emerging Markets: Is it too High?”, Chapter III in World Economic Outlook, September, (Washington, DC: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2004, “Has Fiscal Behavior Changed Under the European Economic and Monetary Union?”, in Chapter II of World Economic Outlook, September, (Washington, DC: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Nowak, Michael, and Luca Antonio Ricci (eds.), Post-Apartheid South Africa: The First Ten Years, forthcoming (Washington, DC: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Penalver, Adrian, and Gregory Thwaites, 2004, “Analysing Sovereign Debt Sustainability: A New probabilistic Approach”, mimeo, (London: Bank of England).

    • Search Google Scholar
    • Export Citation
1

Prepared by Xavier Debrun (FAD).

2

Marcos Souto (Research Department) assisted in the preparation of the simulation algorithm. Related algorithms have been developed by Penalver and Thwaites (2004) and Garcia and Rigobon (2004).

4

For detailed information on fiscal policy developments in South Africa since the end of apartheid, see Nowak and Ricci (forthcoming).

5

Of course, that relationship needs to hold only on average, not every single year.

6

For a more comprehensive discussion, see IMF (2003b). Mean-reversion cannot rule out that the debt-to-GDP ratio would stabilize at an implausibly high level. One should also bear in mind that long-run solvency per se does not imply a “default-proof” fiscal behavior since default can also be triggered by liquidity constraints, or by strategic considerations.

7

See Favero (2002), Gali and Perotti (2003) and Fatàs and Mihov (2003) for detailed discussions of the issues related to the specification and estimation of fiscal policy equations.

9

Argentina, Brazil, Bulgaria, Chile, China, Colombia, Costa Rica, Côte d’Ivoire, Croatia, Ecuador, Hungary, India, Indonesia, Israel, Jordan, Korea, Lebanon, Malaysia, Mexico, Morocco, Nigeria, Pakistan, Panama, Peru, Philippines, Poland, Russia, South Africa, Thailand, Turkey, Ukraine, Uruguay, and Venezuela.

10

See IMF (2003b) for a fuller discussion. Cross-country variation in public sector coverage within the panel creates some uncertainty as to whether this ceiling applies to the NFPS or the central government. Since most of the data used in the estimations covered the NFPS, the latter might be the most appropriate concept. In South Africa, adding municipal and non-financial state-owned enterprise (SOE) debt to national government debt yields a debt-to-GDP ratio close to the 50 percent threshold. However, as shown in the last column of Table II.2, the national government exhibits a much stronger debt-stabilizing response than other countries in the panel, in line with a stable, low-debt equilibrium (Figure II.4). As both municipal and SOE’s debt has exhibited a downward trend (in terms of GDP) comparable to (or more pronounced than) national government debt, it is likely that South African NFPS behavior is not materially different from that of the national government.

11

In Figure II.5, the realization of upside (downside) risks to public debt entails a downward (upward) shift in the AA schedule.

12

If left unchecked, that tendency may lead to a growing deficit bias, raising concerns about long term sustainability (IMF, 2004, and Balassone and Francese, 2004).

13

The “election year” dummy variable identifies presidential elections. The effect of legislative elections was found not to be significant.

14

Technical aspects are described Celasun, Debrun, and Ostry (2005).

15

Notice that in a comparable exercise for Brazil, Penalver and Thwaites (2004) used quarterly fiscal data and could formally test the null hypothesis of orthogonality between fiscal and non fiscal disturbances. They failed to reject the hypothesis.

South Africa: Selected Issues
Author: International Monetary Fund
  • View in gallery

    South Africa: IMF Standard Debt Sustainability Analysis—2004–2010

  • View in gallery

    South Africa: Changes in the Public Debt Ratio and the Primary Surplus, 1990–2005

    (in percent of GDP)

  • View in gallery

    South Africa: Primary Surplus and Lagged Public Debt, 1990–2005

    (in percent of GDP)

  • View in gallery

    Debt Dynamics and the Response of the Primary Surplus to the Debt

  • View in gallery

    Primary Surplus and Lagged Debt in Emerging Market Economies

  • View in gallery

    South Africa: Public Debt Risk Profile, 2005–2009 (baseline)

  • View in gallery

    South Africa: Public Debt Risk Profile, 2005–2009

    (scenario 1: higher discretionary spending financed with new borrowing)

  • View in gallery

    South Africa: Public Debt Risk Profile, 2005–2009

    (scenario 2: lower responsiveness of the primary surplus to the debt)

  • View in gallery

    South Africa: Public Debt Risk Profile, 2005–2009

    (scenario 3: contingent liability shock)