This Selected Issues paper reviews the status and potential impacts of HIV/AIDS in South Africa. It draws on demographic projections and economic studies, and summarizes the official policy response and actions taken by the nongovernmental and business communities. It presents some stylized facts on property price developments in South Africa, and reviews briefly the link between asset price developments and economic activity. It also looks at the ability of policymakers to identify asset price booms.

Abstract

This Selected Issues paper reviews the status and potential impacts of HIV/AIDS in South Africa. It draws on demographic projections and economic studies, and summarizes the official policy response and actions taken by the nongovernmental and business communities. It presents some stylized facts on property price developments in South Africa, and reviews briefly the link between asset price developments and economic activity. It also looks at the ability of policymakers to identify asset price booms.

V. What Determines Fluctuations in South African Manufacturing Output?1

A. Introduction

1. Over recent decades, South African manufacturing output has exhibited a strong pro-cyclical pattern over the business cycle. Though its relative importance in total output declined slightly (Figure 1), the manufacturing sector remains an important sector in the South African economy, contributing about 18 percent to total GDP in constant prices in 2003. Since the mid-1990s, the long-run trend in total output and manufacturing output growth picked up significantly due to broad economic and structural reforms following the end of apartheid and successful stabilization efforts. However, manufacturing output remained very volatile and annual growth rates of quarterly manufacturing output fluctuated strongly.2 These sharp fluctuations raise the question of what are the underlying sources of variation in manufacturing output.

2. This section finds that deviations of manufacturing output from trend are strongly driven by domestic and global demand conditions. Relative price factors, as captured by real exchange rate movements do not seem to have had a large impact on fluctuations of manufacturing output during the 1980s. However, more recently, real exchange rate changes do appear to have had a significant impact on manufacturing output, reflecting the greater trade openness and increased exposure to international competition. The structure of the section is as follows; Subsection B provides a description of the evolution of manufacturing output, domestic and foreign demand, and CPI and unit labor cost (ULC) based real exchange rate measures and their trend and cyclical components, and discusses their main descriptive statistics. Subsection C presents the econometric methodology, estimation results, and concludes.

B. Empirical Evidence

3. While manufacturing output grew by about 1.9 percent on average during 1980-2003, year-on-year quarterly growth rates fluctuated widely from negative 9 to 11 percent. Growth in manufacturing output picked up in the mid-1990s, reflecting the transition to a higher growth path of total output (Figure V.1). Following the end of the apartheid regime, sustained higher growth was achieved through wide-ranging economic and structural reforms and successful macroeconomic stabilization. High growth in total factor productivity (TFP), which in turn seems to have been driven by greater trade openness and private-sector participation in the economy, has been the main cause of the recent growth performance in South Africa. Jonsson and Subramanian (2000) estimate that trade liberalization may have accounted for up to 90 percent of actual TFP growth during the 1990s. Any model that aims to explain the dynamics of manufacturing over time would need to take into account these factors, in addition to other input factors of manufacturing production and relative cost-and-demand factors. In this section, the focus is on short- to medium-run dynamics of manufacturing output and, therefore, the analysis is confined to identifying those variable that have a strong impact on these dynamics.

Figure V.1.
Figure V.1.

The Evolution of Manufacturing Output in South Africa During 1978-2003

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A005

Source: South African Reserve Bank.

4. The trend and cyclical components of manufacturing output are determined by applying a Hodrick-Prescott (HP) filter, a commonly used method for identifying cyclical components of macroeconomic time series (Figure V.2). The cyclical component represents variation above business cycle frequency. King and Plosser (1993) provide a detailed discussion of the HP-filter and its properties. Applying a HP-filter (with the main parameter lambda=1600) to quarterly data is similar to first differencing the data and then applying a band pass filter that eliminates all variation with a periodicity greater than eight years. According to this procedure, over the past ten years, manufacturing output growth exceeded its trend during 1995Q1 to 1998Q2 and during 2000Q3 to 2003Q1, with the exception of 2001Q3, before slowing significantly during the remainder of 2003.

Figure V.2.
Figure V.2.

The Trend and Cyclical Components of Manufacturing Output

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A005

Source: South African Reserve Bank, and staff calculations.Note: Variables are in logarithms. The cyclical component is the difference between actual data and trend.

5. Following the literature, the set of variables that can potentially explain fluctuations in manufacturing output includes foreign and domestic demand and various exchange rate measures.3 Most of South Africa’s exports of manufactured goods are sold in Europe and other advanced economies. Global demand conditions or foreign demand for South African manufactured goods is, therefore, approximated as a weighted average of total domestic demand in South Africa’s main advanced trading partner countries. Total domestic demand is measured as the sum of total consumption and investment in South Africa. Real exchange rate measures based on consumer price indices (CPI) and on unit labor costs (ULC) are employed as possible indicators of short-run changes in South Africa’s international competitiveness position. The weights used to calculate real exchange rates and foreign demand correspond to each country’s share in South Africa’s total exports to its main trading partners.

6. Compared to real exchange rates based on consumer price indices, measures based on unit labor cost better indicate developments in relative production costs of internationally traded, manufactured goods.4 A rise in unit labor costs in the South African manufacturing sector, relative to its main trading partners, makes production in South Africa more expensive than in those countries and may thus force companies to reduce production, if they loose the market share, or to exit markets. In contrast, an increase in the CPI-based real exchange rate could reflect a rise in the relative price of nontraded goods with little impact on manufacturing production costs if the input of nontradables into manufacturing production is low.

7. Manufacturing output is strongly correlated with foreign and domestic demand (Table V.1). The correlation with the highly volatile ULC-based real exchange rate is negative and relatively low, while the correlation with the CPI-based real exchange rate is virtually zero.5 The difference between the CPI- and ULC-based exchange rates is relatively small (Figure 3) and, accordingly, the correlation of the cyclical components of these measures turns out very high in both samples. This fact could imply that much of the short-run variability in both real exchange rates is actually driven by changes in the nominal exchange rate. The positive—but low—correlation between the domestic and foreign demand components indicates that South Africa’s business cycle is only weakly synchronized with those of its main trading partners.

Table V.1.

Standard Deviations and Cross-Correlations of Cyclical Components

article image
Sources: South African Reserve Bank and International Financial Statistics (IMF); and staff calculations.Note: Cyclical components are calculated as the difference between variables (in logarithms) and their HP-trends, 1980Q1-2003Q4.
Figure V.3.
Figure V.3.

Estimates of the Cyclical Component of Manufacturing, 1980-2003

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A005

Sources: South African Reserve Bank; and Fund staff calculations and estimates.

C. Estimation Procedure and Results

8. Simple econometric models find a statistically significant impact of foreign and domestic demand and relative unit labor costs on manufacturing output for the entire sample (see, Table V.2). The impact of ULC-based real exchange rate is relatively small, and in a model (Model 4 in Table V.2) in which the ULC-based real exchange rate is replaced with the CPI-based measure, the coefficient is statistically insignificant. However, statistical tests cannot reject the hypothesis that a structural break occurred in the beginning of the 1990s.

9. In a model that splits the ULC-based real exchange rate into two variables that assume zero values for the periods 1980Q1-1989Q4 and 1990Q1-2003Q4, respectively, the ULC-based exchange rate only turns out to be significant for the latter period, and the estimated coefficient is much larger compared to the coefficient that was estimated for the entire sample.6

Table V.2.

Determinants of Fluctuations in Manufacturing Output, 1980-2003

article image
Note: The dependent variable is the cyclical component of manufacturing. Standard errors are in parentheses. Model 1 and 2 are estimated using OLS. However, the residuals of these regressions are serially correlated. The models are, therefore, reestimated including an autoregressive first-order process for the original residual. Marquardt’s nonlinear least squares algorithm is used to estimate Models 3, 4, 5 and 6. In Model 5, ULC-REER(1) is set equal to ULC-REER for the period 1980Q1-1989Q4 and set equal to zero for the remainder of the sample, while ULC-REER(2) is set equal to zero for the former period and set equal to ULC-REER for the latter period. In Model 6, the same procedure has been applied to CPI-REER.

10. In this model (Model 5 in Table V.2), a one percent increase of the ULC-based real exchange rate now triggers a 0.06 percent fall in manufacturing output below trend, compared to -0.02 during the 1980s, while a one-percent rise of foreign or domestic demand above trend causes a contemporaneous increase in manufacturing output of 1.1 or 0.3 percent above its trend value. The coefficient of the first-order autoregressive term for the residual, which is 0.75, indicates that the response of manufacturing to shocks is relatively sluggish. Even though the real exchange rate measures are much more volatile than foreign and domestic demand, the above coefficients imply that much variability in manufacturing output is caused by changes in foreign and domestic demand.

11. To better quantify the individual impact of the determinants on fluctuations in manufacturing output, measures (modified R-squared) are calculated assuming each determinant were the only variable contributing to variations. While domestic demand would account for about 38 percent of fluctuations in manufacturing output and foreign demand for about 11 percent, the ULC-based real exchange rate would only account for about 7 percent. Figure 3 shows the isolated impact of movements in the demand and real exchange variables on deviations of manufacturing output from trend, actual, and fitted values of the cyclical component of manufacturing output.

12. The sharp appreciation of the real exchange rate indicators during 2002 and 2003 following the rand crisis in the fourth quarter of 2001 may have depressed manufacturing output by up to 3 percent. While the real exchange rate measures were about 25 percent below trend in the first quarter of 2002, they were about 20 percent above trend in the fourth quarter of 2003 (see Figure V.4).7 These deviations drove manufacturing output about 1½ percent above trend and then later 1½ percent below trend.

Figure V.4.
Figure V.4.

Trends and Cyclical Components of the Main Determinants of Fluctuations in Manufacturing Output

Citation: IMF Staff Country Reports 2004, 379; 10.5089/9781451841022.002.A005

Sources: South African Reserve Bank and International Financial Statistics (IMF); and staff calculations. Note: All Variables are in logarithms. The cyclical components measure deviations from trend in percent.

13. This section found a strong impact of domestic and foreign demand on the cyclical component of manufacturing output. Real exchange rate measures, including relative unit labor costs, are estimated to have a smaller impact. However, there is evidence that the impact of the real exchange rate on manufacturing output has increased since the beginning of the 1990s. The sharp depreciation at the end of 2001 is estimated to have pushed manufacturing output above trend by up to 1½ percent in early 2002. Since then, the subsequent appreciation of the ULC-based real exchange rate may have depressed manufacturing output by up to 3 percent. Finally, the highly aggregated nature of the data requires a cautious interpretation of these results. On the individual firm level, output adjustment can take place by changing firm-individual output or by enter and exit decisions of firms. To achieve a more coherent picture of the change in manufacturing output, firm level data would offer a much richer picture and would also allow for better analysis of the impact of demand and exchange rate changes on profits and employment, pointing the way for further research

Variables: Definitions and Source

The dataset consists of quarterly data from 1980 to 2003 for South Africa and its main trading partners. Data series for unit labor costs in manufacturing, manufacturing value added, gross domestic product, total consumption and total investment are from the South African Reserve Bank, Quarterly Bulletin. Data series for the CPI-based real effective exchange rate and unit labor costs in manufacturing, as well as total domestic demand, of South Africa’s main trading partners is from IFS.

Domestic demand is defined as the sum of total investment and total consumption.

Foreign demand is defined as a weighted average of real total domestic demand in South Africa’s main advanced trading partner countries

The ULC-based real effective exchange rate is defined as unit labor costs (manufacturing) in South Africa relative to a geometrically weighted average of unit labor costs (manufacturing) in its main trading partners’ countries all measured in U.S. dollars.

The main trading partners include Germany (0.22), United States (0.16), Japan (0.15), United Kingdom (0.12), Italy (0.09), France (0.07), Belgium (0.04), Netherlands (0.04), Switzerland (0.03), Canada (0.02), Spain (0.02), Australia (0.02) and Sweden (0.01). Relative weights used to calculate the ULC-based real exchange rate are in parenthesis.

South Africa: Tax Summary as of June 20041

(All amounts in South African rand)

article image
article image
article image
article image
article image
article image
article image
article image
article image
article image
article image
article image

References

  • Cogley, Timothey and James Nason, 1995, “Effects of the Hodrick-Prescott Filter on Trend and Difference Stationary Time Series: Implications for Business Cycle Research”, Journal of Economic Dynamics and Control, Vol. 19: 25378.

    • Search Google Scholar
    • Export Citation
  • Jaeger, Albert, 1994, “Mechanical Detrending by Hodrick-Prescott Filtering: A Note”, Empirical Economics, Vol. 19: 493500.

  • Jonsson, Gunnar and Arvind Subramanian, 2000, “Dynamic Gains from Trade: Evidence from South Africa”, IMF Working Paper, 00/45 (Washington; International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • King, Robert and Sergio Rebelo, 1993, “Low Frequency Filtering and Real Business Cycles”, Journal of Economic Dynamics and Control, Vol. 17: 20731.

    • Search Google Scholar
    • Export Citation
  • Lipschitz, Leslie and Donogh McDonald, 1991, “Real Exchange Rates and Competitiveness: A Clarification of Concepts, and Some Measurements for Europe”, IMF Working Paper, 91/25 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Marsh, Ian and Stephen Tokarick, 1994, “Competitiveness Indicators: A Theoretical and Empirical Assessment”, IMF Working Paper, 94/29 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
1

Prepared by Thomas Harjes.

2

In this section, manufacturing output refers to the real value added of the manufacturing sector in the national accounts.

3

For the remainder of this section, domestic and foreign demand, manufacturing output and real exchange rates refer to their cyclical components unless otherwise noted.

4

Unit labor costs in South Africa and its main advanced trading partners reflect costs in the respective manufacturing sectors. See Marsh and Tokarick (1994), and Lipschitz and McDonald (1991) for a comprehensive discussion of various exchange rate measures as indicators of competitiveness.

5

Jaeger (1994), and Cogley and Nason (1995) have shown that, under certain conditions, HP-filtering can induce spurious correlations, especially if the time series are nonstationary. However, differencing the data first produces the same qualitative results.

6

Similar experiments with the domestic and foreign demand variables did not indicate any further structural changes.

7

It is important to note that the above-mentioned trend is purely defined in a statistical sense by the trending method (i.e. the HP-filter) and does not reveal anything about real exchange rate equilibrium values in regard to economic fundamentals.

1

Updated by M. Horton, Fiscal Affairs Department, June 2004. For further information, see http://www.sars.gov.za or http://www.treasury.gov.za.

2

The worldwide basis for income taxation was introduced from January 1, 2001.

3

The capital gains tax became effective on October 1, 2001.

4

The average rate for 2002/03 was R 1.40 per R 100 of earnings.

5

A tax of 0.25 percent of the purchase value of marketable securities (Act No. 32 of 1948) was repealed on December 22, 2003.

6

Fuel excise rates are from April 7, 2004.

7

In the 2004/05 budget, the stamp duty on mortgages for home purchase was proposed for repeal, with effect from March 1, 2004.

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

    The Evolution of Manufacturing Output in South Africa During 1978-2003

  • View in gallery

    The Trend and Cyclical Components of Manufacturing Output

  • View in gallery

    Estimates of the Cyclical Component of Manufacturing, 1980-2003

  • View in gallery

    Trends and Cyclical Components of the Main Determinants of Fluctuations in Manufacturing Output