Selected Issues

Abstract

Selected Issues

The Growth-Inflation Tradeoff of Monetary Policy in South Africa1

The combination of low growth and stubbornly high inflation expectations for a protracted period has complicated monetary policy decisions. Staff analysis contributes to the ongoing growth-inflation tradeoff discussion in South Africa, concluding that there is limited growth tradeoff of monetary policy efforts to anchor inflation expectations at a lower level at present.

A. Introduction

Growth and Inflation

1. Economic growth declined in South Africa and started to undershoot its peers in the past decade, highlighting the role of unaddressed structural constraints. As discussed in Annex I. South Africa’s macroeconomic performance appears to have witnessed an important inflection point between the 2000s and the 2010s, which broadly coincides with the global financial crisis in 2008–09. Real GDP growth (year on year) halved after the Global Financial Crisis, from 4.3 percent during 2000–07 to 2 percent during 2010–17 and eked out just 0.8 percent in 2018. Monetary and fiscal policies have lent significant growth support, but structural rigidities have kept business confidence low, depressing private investment and growth. Demand for credit has weakened, while banks have refrained from expanding loans, in the low growth environment.

2. The SARB’s target measure of inflation, the monthly consumer price index (CPI), has been affected by a range of factors. The inflation index includes all the major metropolitan areas and the largest towns in each of the nine provinces. Prices of imported goods, hence exchange rate performance, play an essential role in the inflation process in an open emerging economy such as South Africa. The rand is a relatively volatile currency, reflecting the extent of domestic and external shocks, and the SARB’s policy of no foreign exchange intervention. The exchange rate pass through to inflation has declined (Kabundi and Mbelu, 2018), owing partly to greater central bank credibility (Kabundi and Mlachila, forthcoming). Administered price inflation has generally been above overall CPI inflation, and in recent years has on numerous occasions underpinned elevated wage demands on occasions in the context of a rigid labor market, despite unemployment being persistently high.

Impact of Monetary Policy

3. The central bank may influence some of the supply factors of inflation through the exchange rate and credibility channels. Gumata et al. (2013) highlight the importance of the exchange rate and expectations channels in South Africa during the two decades since 1990.

  • The exchange rate represents a relatively important component for domestic food and petrol prices.2 Accommodative monetary policy may weaken the exchange rate and lift non-core and core inflation.

  • Central bank credibility may have a role in setting the extent of the positive feedback loop between inflation and inflation expectations, particularly that of price and wage setters (South African Reserve Bank, 2016). Inflation expectations of businesses and trade unions closely follow inflation outturns with a lag, which can be attributed to the expectation trap hypothesis, where trade unions determine wages with little consideration of the central bank objective, and with the view that the central bank will accommodate their demand for higher wages to support economic activity.

4. In this context, the impact of monetary policy on growth may have diminished from pre- to post-2008, but it is less clear whether its impact on inflation has weakened as well. When economic activity is suppressed due to structural rigidities, subdued business confidence, low investment, and risk aversion on the part of lenders, monetary policy could affect growth little by marginally changing the cost of funds. This would also imply a weaker monetary policy transmission through demand to inflation. However, monetary policy could affect inflation expectations to a greater extent through the exchange rate and credibility channels as discussed above.

5. This note helps understand how the growth-inflation tradeoff of monetary policy has evolved by (i) describing the conventional Phillips curve for South Africa, and, (ii) estimating econometrically the impact of real policy rate changes on domestic demand growth and inflation expectations.

B. The Growth-Inflation Phillips Curve

6. The slope of the growth-inflation Phillips curve has been generally negative in South Africa (Figure 1). During the 1970s, the relationship between inflation and real GDP growth was around zero, that is, inflation changed little systematically when real GDP growth changed. The fact that the rand was pegged may have partly been responsible for the observed disconnect. In the 1980s, the slope started to turn negative, that is, lower (higher) inflation was more systematically associated with higher (lower) real GDP growth. The negative slope steepened markedly in the 1990s as higher domestic demand growth was associated with significantly lower inflation. The slope flattened back in the 2000s and 2010s, as periods of lower demand growth were associated with more muted increases in inflation. The fact that inflation generally remained in single digits may have contributed to the flattening of the curve (see Annex I).

Figure 1.
Figure 1.

South Africa: Relationship Between Real GDP Growth and Inflation

(Coefficient of inflation on GDP growth)

Citation: IMF Staff Country Reports 2020, 034; 10.5089/9781513528519.002.A001

Sources: Haver and IMF staff calculations.

7. The observation that higher GDP growth and lower inflation go hand in hand can be interpreted in several ways. One view is that growth mainly stems from productivity improvements, which is disinflationary at the same time. Therefore, structural reform implementation is important to boost productivity and growth. It also facilitates the SARB’s job to anchor inflation and inflation expectations at a lower level. Another view is that inflation dampens growth in the long run, thus the SARB should continue to maintain inflation and inflation expectations low and stable (Ndou and Gumata, 2017).

C. Econometric Analysis

Baseline Model and Results

8. The impact of monetary policy action on domestic demand growth and inflation expectations is estimated using four key macroeconomic variables (Figure 2). These variables are 1) real domestic demand growth to capture the strength of economic activity; 2) a survey-based indicator of inflation expectations one year ahead (average for financial analysts, businesses, and trade unions) to reflect the fact that monetary policy tends to affect inflation 12–18 months ahead; 3) real policy interest rate (deflated by inflation expectations) to measure monetary policy action to the extent that the monetary policy stance could ease after a rate hike if inflation (expectations) rose more; and 4) nominal effective exchange rate to capture its essential role in the inflation process in an open emerging economy such as South Africa.

Figure 2.
Figure 2.

South Africa: Macroeconomic Indicators Used for Regression Analysis

Citation: IMF Staff Country Reports 2020, 034; 10.5089/9781513528519.002.A001

Sources: BER South Africa, Haver, Stats SA, and IMF staff calculations.

9. Using the four variables, a quarterly VAR model is estimated for 2000Q3–18Q2. The variables are detrended, except for inflation expectations, to reduce the chance of obtaining spurious results. The variables are stacked in the order of domestic demand growth, inflation expectations, real policy rate, and NEER growth. The model is estimated for a period through 2008 (2008Q2) and after 2010 (2010Q3) to assess changes in the impact of monetary policy action. Three lags are used to estimate the VAR. Generalized impulses are reported.

10. Results suggest that a rise in the real policy rate tends to reduce inflation expectations throughout the two sample periods, but its impact on domestic demand growth weakens from the 2000s to 2010s (Figure 3, and Appendix II for the responses of other variables). A 1 percentage point increase in the real policy rate is associated with a 2 percentage point decline in domestic demand growth during the 2000s but does not generate a systematic impact during the 2010s. By contrast, monetary policy action systematically reduces inflation expectations in both sample periods, but the impact diminishes from 1–1.5 percentage points during the 2000s to 0.5 percentage point during the 2010s. These results appear consistent with those by Ramcharan (2008), who, using data up to the middle of the 2000s, finds that a 1 percentage point increase in the nominal policy rate is associated with a 0.8 percentage point decline in GDP growth two quarters later (similar to findings in the literature of the US economy).

Figure 3.
Figure 3.

South Africa: Response to One Percentage Point Increase in Real Policy Rate

(Percentage points)

Citation: IMF Staff Country Reports 2020, 034; 10.5089/9781513528519.002.A001

Sources: Haver and IMF staff estimates.Note: Time scale in quarters. Broken lines are the 95 percent confidence bands.

Robustness Checks

11. Additional models are estimated to address the following questions and confirm some of the baseline results. The first question is whether there are threshold effects. One view is that monetary policy may have a stronger effect when economic activity is stronger, and interest rate and inflation expectations are higher. This could be driving the baseline findings, as the 2000s (2010s) can be characterized by high (low) growth/interest rate/inflation regime. Another question is the role of core inflation. The baseline results suggest that the monetary policy transmission through demand growth to core inflation is likely muted.

12. Additional findings confirm the prior hypothesis. To address the first question, the baseline model is re-estimated after splitting the whole sample period (excluding quarters around 2008–09) into “high” and “low” real policy rate regimes. A threshold of 1.5 percent splits the whole sample broadly in half. The results show that both demand growth and inflation expectations respond to monetary policy action during the “high” real rate regime, but only inflation expectations respond in the “low” real policy rate regime. To address the second question, the baseline model is estimated after replacing (headline) inflation expectations with (realized) core inflation (detrended). The model is estimated for the 2010s only as the sample for the 2000s is short when core inflation is used. Results show that core inflation declines with a lag after monetary policy tightening, but the relation is statistically insignificant.

D. Conclusion

13. The findings in this note suggest that the SARB should continue its efforts of anchoring inflation and inflation expectations at a lower level because monetary policy lends limited support to growth dampened by structural issues. During the 2010s, domestic demand growth responded little to monetary policy action. The environment of weak growth, low interest rates, and relatively moderate inflation (expectations) could have muted monetary policy transmission. Ultimately, the constraints to economic growth need to be removed. Meanwhile, inflation expectations continue to respond to monetary policy action albeit to a lesser extent. Monetary policy transmission through demand has weakened––demand growth does not systematically respond to monetary policy action nor does core inflation––but the exchange rate and credibility channels appear to remain operational.

Appendix I. Scatter Plot

Figure AP1.1.
Figure AP1.1.

South Africa: Real GDP Growth and Inflation

(Percent)

Citation: IMF Staff Country Reports 2020, 034; 10.5089/9781513528519.002.A001

Appendix II. Detailed Results

1. During 2000Q3–2008Q2, inflation expectations increased in response to stronger domestic demand and a lower real policy rate. The real policy rate increased in response to stronger domestic demand and higher inflation expectations. It declined in response to a weaker currency, probably through higher inflation.

2. During 2010Q3–2018Q2, the real policy rate declined in response to higher domestic demand and inflation expectations, potentially as the response of the nominal policy rate lagged a rise in inflation. It declined in response to a weaker currency.

Figure AP2.1.
Figure AP2.1.

South Africa: Response to Generalized One Standard Deviation Shocks

(2003Q3–2008Q2)

Citation: IMF Staff Country Reports 2020, 034; 10.5089/9781513528519.002.A001

Figure AP2.2.
Figure AP2.2.

South Africa: Response to Generalized One Standard Deviation Shocks

(2010Q3–2018Q2)

Citation: IMF Staff Country Reports 2020, 034; 10.5089/9781513528519.002.A001

References

  • Gumata, N., A. Kabundi, and E. Ndou. 2013. “Important Channels of Transmitting Monetary Policy Shock in South Africa”, ERSA Working Paper.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund. 2017. South Africa Staff Report, Annex III, “Exchange Rate Pass-Through to Core Inflation in South Africa”, 17/189.

    • Search Google Scholar
    • Export Citation
  • Kabundi, A. and A. Mbelu. 2018. “Has the Exchange Rate Pass-Through Changed in South Africa,” South African Journal of Economics, 86, 3, September, 33960.

    • Search Google Scholar
    • Export Citation
  • Kabundi, A., and M. Mlachila. forthcoming. “The Role of Monetary Policy Credibility in Explaining the Decline in Exchange Rate Pass-Through in South Africa,” Economic Modelling.

    • Search Google Scholar
    • Export Citation
  • Ndou, E. and N. Gumata. 2017. “Inflation Dynamics in South Africa”, Palgrave Macmillan.

  • Ramcharan, R. 2008. “The Impact of Interest Rates on Real Activity in South Africa”, South Africa: Selected Issues, 09/276.

  • South African Reserve Bank. 2016. “Inflation Mechanisms, Expectations and Monetary Policy”, BIS Papers No 89, 34548.

1

Prepared by Ken Miyajima, reviewed by Ana Lucía Coronel.

2

The exchange rate pass through to these prices is relatively immediate and large. That to core inflation is estimated to be smaller, at around 10–15 percent (IMF, 2017).

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

    South Africa: Relationship Between Real GDP Growth and Inflation

    (Coefficient of inflation on GDP growth)

  • View in gallery

    South Africa: Macroeconomic Indicators Used for Regression Analysis

  • View in gallery

    South Africa: Response to One Percentage Point Increase in Real Policy Rate

    (Percentage points)

  • View in gallery

    South Africa: Real GDP Growth and Inflation

    (Percent)

  • View in gallery

    South Africa: Response to Generalized One Standard Deviation Shocks

    (2003Q3–2008Q2)

  • View in gallery

    South Africa: Response to Generalized One Standard Deviation Shocks

    (2010Q3–2018Q2)