South Africa: Selected Economic Issues

This Selected Economic Issues paper examines economic developments in South Africa during 1993–94. After a cumulative fall of 3.5 percent between 1989 and 1992, GDP at market prices grew by 1.1 percent in 1993. The major contribution to growth came from the turnaround in the inventory cycle, with positive investment in inventories recorded for the first time since 1989. Private consumption expenditure remained subdued in 1993, rising by only 0.5 percent; by contrast, public consumption grew by 1.8 percent in 1993.

Abstract

This Selected Economic Issues paper examines economic developments in South Africa during 1993–94. After a cumulative fall of 3.5 percent between 1989 and 1992, GDP at market prices grew by 1.1 percent in 1993. The major contribution to growth came from the turnaround in the inventory cycle, with positive investment in inventories recorded for the first time since 1989. Private consumption expenditure remained subdued in 1993, rising by only 0.5 percent; by contrast, public consumption grew by 1.8 percent in 1993.

III. The Labor Market

1. Introduction

Unemployment is a central economic problem in South Africa. In 1993, 44 percent of the labor force was unable to find work in the formal sector and the official unemployment rate for October was 29 percent, or 3.6 million persons. 1/ Of the 3.6 million unemployed in 1993, half were between 15 and 29 years old, and three fifths had been without work for more than a year. Few of the unemployed have access to unemployment insurance, and there is no publicly provided unemployment assistance.

The White Paper on Reconstruction and Development states that “The fundamental goal of the RDP is an employment-creating, labor-absorbing economy which will ultimately lead to full employment.” A key element in achieving this goal is understanding why unemployment is so high now.

This chapter discusses the key features of the labor market, including its institutional structures, in order to shed some light on this question. It relates the data and the institutional structures to a number of common propositions about the roots of South African unemployment. It thus builds upon the analysis of unemployment contained in the 1993 Selected Economic Issues Paper (SM/93/255), pp. 29-37.

2. Employment and unemployment

There are two key measures of unemployment: the proportion of the labor force without employment in the formal sector--the “formal employment gap” (FEG)--and the “official measure” of unemployment. The two measures reflect different definitions of employment and the labor force. The FEG is simply the labor force less formal sector employment, expressed as a proportion of the labor force--it thus takes no account of employment in the “informal sector”. 2/ The official unemployment rate includes employment in the informal sector also. The FEG adopts the “strict” definition of the labor force; this excludes persons who declare a desire to work, but who make no job search efforts--the so-called “discouraged workers.” The official measure uses the “expanded” definition of the labor force, which includes those who declare a desire to work but who engage in no job-search efforts. Estimates of the FEG are available from 1960 onward, while the official rate of unemployment was measured for the first time in October 1993, in the October Household Survey (OHS).

Chart 13 shows the population, labor force, formal employment and FEG since 1960. It shows that the FEG began to rise in the mid 1970s, but accelerated from the early 1980s. Chart 14 highlights the developments in formal sector employment underlying the FEG. It shows that employment growth outside government slowed in the 1970s, and stopped in the early 1980s. This coincided with stagnation in employment in the secondary sector and in unskilled employment. 1/ However, employment in government, the tertiary sector (which includes government), and higher skilled occupations continued to grow until the late 1980s.

CHART 13
CHART 13

SOUTH AFRICA LABOR FORCE AND FORMAL SECTOR EMPLOYMENT, 1960–94

Citation: IMF Staff Country Reports 1995, 021; 10.5089/9781451840919.002.A003

Sources: South African Labor Statistics 1994; Bureau for Market Research; Statistical Service.1/ According to the “strict” definition which excludes “discouraged workers”.2/ Standardized employment in the formal sector.3/ Labor force divided by the population.4/ Percentage of the labor force outside the formal sector.
CHART 14
CHART 14

SOUTH AFRICA FORMAL SECTOR EMPLOYMENT, 1960–93

Citation: IMF Staff Country Reports 1995, 021; 10.5089/9781451840919.002.A003

Sources: South African Labor Statistics 1994; Institute for Futures Research.1/ Central, provincial, and local government.2/ Transport, finance, communications, government, and domestic service.3/ Agriculture and mining.4/ Manufacturing, electricity, and construction.

Chart 15 shows that during the 1980s, when the FEG was rising, the real wages of black people were also rising in most parts of the manufacturing sector, including those with the lowest average income levels. 2/ The average real wage of employed black people rose faster than that of white people throughout the 1980s. In the same period, vacancy rates for unskilled labor remained low, with vacancy rates for skilled labor being substantially higher.

CHART 15
CHART 15

SOUTH AFRICA AVERAGE BLACK WAGES RELATIVE TO SUBSISTENCE, 1984–92

MANUFACTURING 1/

Citation: IMF Staff Country Reports 1995, 021; 10.5089/9781451840919.002.A003

Sources: Central Statistical Service; and Institute for Planning Research, University of Port Elizabeth.1/ Subsistence is measured as HSL-6 each March.2/ Sector legends are as follows:A: Clothing B: Leather. C: Footwear. D: Wood products. E: Textiles. F: Non metal minerals.G: Furniture. I: Food. J: Plastic. K: Metal goods. L: Other manufacturing. N: Electrical.O: Beverages. P: Professional. Q: Paper goods. R: Machinery. S: Iron and Steel. U: Rubber.

Some of the rise in FEG since the 1960s--and possibly a sizable part--has been absorbed into the informal sector. Although historical data are limited, 3/ they do suggest a rapid expansion of the informal sector during the 1980s. According to the OHS (which excludes the TBVC States), 2.4 million black people were employed on their own account in the informal sector in 1993, compared with total black employment of 5 million. This source also suggests that wage rates in the informal sector are substantially lower than those in the formal sector. It implies that the monthly contribution to GDP per person working in the informal sector is some R 680, compared with average monthly wages of R 1500 for black people in formal sector manufacturing in the third quarter of 1993, and R 950 a month for black people employed in formal sector construction in the same period.

The OHS conducted in October 1993 measured the official rate of unemployment. 1/ These data are some of the first to provide a clear indication of the characteristics of the unemployed. Among the most notable characteristics are the following:

  • * Half are under 30 years old;

  • * Three fifths have never held a previous job;

  • * Some two thirds have been seeking work for over a year;

  • * The overwhelming bulk have below standard 10 education;

  • * Half do not engage in job search activities--they are the so-called “discouraged workers.”

The OHS does not report unemployment rates by region. However, data from the Development Bank, based upon the OHS, suggest that while there are considerable regional variations in the rate of unemployment--from 50 percent in the Northern Transvaal to 17 percent in the Western Cape--it is high in all regions. Data on unemployment rates by age are not published in the OHS. However, staff calculations using the labor force data published elsewhere 2/ suggest that the OHS found unemployment rates among those under 20 years old of over 50 percent, nearly double the average unemployment rate for all age groups. 3/

3. Why are South African unemployment rates so high?

Unemployment in South Africa has certain marked features, including the relentless rise in the FEG in the 1980s alongside evidence suggesting that real wage rates were rising, and the prominence of the long-term and young unemployed in official unemployment in 1993. These help to identify its roots, but there are gaps in the official data which complicate efforts to account for its increase precisely.

The severity of unemployment--by either measure--and the survey evidence that many unemployed have been without work for long periods despite a desire to work would indicate that the bulk of unemployment at present is neither voluntary nor frictional. The fact that the FEG has risen steadily for 15 years to its current level indicates that little of the current stock of FEG is cyclical--the bulk is structural. Both the OHS evidence that the unemployed are predominantly unskilled, and the evidence that unskilled employment has stagnated since the late 1970s indicate that the roots of the problem are to be found in the structural features of the market for unskilled labor.

Moving beyond these conclusions, however, to identify exactly what underlies the high structural unemployment of unskilled labor is not straightforward. The involuntary nature of the unemployment and the evidence of rising real wage rates over the 1980s naturally directs attention to impediments to downward real wage adjustment. The fact that the FEG has been high and rising for so long narrows that focus down to accounts of wage rigidity that operate in the medium- to long-term, and away from accounts of short-term wage rigidity such as those sometimes thought to underlie business cycles.

The published labor market data, however, are often insufficiently disaggregated to distinguish unambiguously between alternative accounts of impediments to real wage adjustment over the medium term. Furthermore, the wage setting institutions and the behavior of the main participants within them defy simple generalizations. In light of these constraints, the following sections discuss three impediments to real wage adjustment over the medium term that commentators have suggested might be important in South Africa: the notion that wages for unskilled labor in the formal sector are so low that they cannot fall any further, the growing strength of trade unions, and the mechanics of the wage determination system.

4. A floor on unskilled wages in the formal sector

One aspect of the substantial economic inequalities in South Africa is that wages for unskilled labor in the formal sector are generally low. It has been suggested that unskilled wage rates may be so low that employers have been reluctant to reduce them further, despite the incentive to do so provided by rising unemployment of unskilled labor. One version of this argument traces employers’ reluctance to reduce real wages to workers’ nutritional requirements; if wages fall below a level necessary to provide adequate nutrition to workers, the physical productivity of labor declines. 1/ If the real wages of unskilled labor were generally at such a floor in South Africa, then this might account for the failure of real wages to adjust despite the rising FEG in the 1980s.

One approach to assessing the potential role of this factor in South Africa would be to compare unskilled wage rates directly with a measure of “minimum subsistence income.” However, such an exercise requires that “subsistence” can be measured in absolute terms, and it is not clear that this can be done satisfactorily. 1/ An alternative approach is to assess the potential role of this factor indirectly, examining wage behavior (as opposed to the wage levels). If wage rates for unskilled labor had reached a nutritional floor, they should remain unchanged relative to subsistence thereafter, neither rising (given high unemployment) nor falling, and they should closely reflect regional differences in the costs of subsistence.

The wage data for the manufacturing sector shown in Chart 15 show average wages for black people deflated by the Household Subsistence Level for a household of six low-income persons (HSL-6). For most sectors, these series were rising during the 1980s. This may reflect, in part, some progression of black people up the skill ladder during this period, so these data do not give unambiguous information about the evolution of wage rates for unskilled labor.

The upper panel in Chart 16, however, shows wages for unskilled labor in selected industries deflated by the same subsistence basket. These industries are chosen because they report wage rates by skill level to the Central Statistical Service. Unskilled wage rates in the metal and engineering industries, as well as in civil engineering were rising strongly relative to subsistence in the 1980s. Even in the building industry, wage rates deflated by HSL-6 rose by over 10 percent between 1980 and 1984, despite the rising FEG since the mid-1970s. While the data for domestic service appears to show little change relative to subsistence, both its components--for full- and part-time domestic servants--move substantially relative to HSL-6 in the 1980s, falling for full-time and rising for part-time domestic servants. 2/ What direct evidence there is from official sources on unskilled wage rates contradicts the nutritional floor account of downward wage inflexibility underlying the rising FEG.

CHART 16
CHART 16

SOUTH AFRICA AVERAGE WAGES FOR UNSKILLED LABOR, 1980–92

Relative to Subsistence 1/

Citation: IMF Staff Country Reports 1995, 021; 10.5089/9781451840919.002.A003

Sources: South African Labor Statistics 1994; Barker (1992); Institute for Planning Research, University of Port Eliz1/ Wages deflated by the household subsistence level for a household of six low income persons.

The lower panel in Chart 16 shows the minimum wages for unskilled labor as required by Wage Boards, all deflated by HSL-6. Wage Boards are statutory bodies which set minimum wage rates for unskilled labor. In 1991, the 27 Wage Board rulings then in force applied to between 500,000 and 700,000 workers. The chart shows the conservative stance taken by such Boards, since in most cases, minimum wages fell relative to HSL-6 during the 1980s, sometimes substantially so. This evidence rules out the possibility that the evidence of rising wages for unskilled labor relative to HSL-6 during the 1980s merely reflects the impact of minimum wage requirements as set by the Wage Boards.

The Peromnes surveys of manufacturing firms, cited by Hofmeyr, 1/ supplement the data on unskilled wage rates published by the Central Statistical Service. These show that between 1985 and 1990, the average annual rate of growth of black male wages deflated by the low income consumer prices index 2/ was 3.2 percent for unskilled labor, the highest rate of increase for any skill level. This pattern broadly matches that for female unskilled wage rates. These data therefore suggest that the evidence of rising unskilled wage rates provided by the official published data are not unrepresentative. The combined evidence of rising wage rates for unskilled labor throughout the 1980s is difficult to reconcile with the nutritional wage hypothesis.

The relation between wages and subsistence costs by region is also difficult to reconcile with the nutritional wage floor hypothesis. Chart 17 shows HSL-6 for all centers surveyed, between September 1989 and September 1993, with centers ordered according to their relative cost of subsistence in 1989. These data show that the regional variations in the cost of subsistence are marked, and, with some exceptions, are relatively stable over time.

CHART 17
CHART 17

SOUTH AFRICA HSL-6 IN REGIONAL CENTERS, 1989–93

in current rand. 1/

Citation: IMF Staff Country Reports 1995, 021; 10.5089/9781451840919.002.A003

Source: Institute for Planning Research University of Port Elizabeth.1/ Centers are ordered according to the cost of household subsistence (HSL-6) in September 1989.1: Boksburg. 2: Jo’burg. 3: Germiston. 4: Kimberly. 5: Benoni. 6: Krg’dorp.7: Cape Town. 8: Mossel bay. 9: George. 10: Springs. 11: Vaol Triangle. 12: Pretoria.13: Brakpan. 14: Durban. 15: E. London. 16: Brits. 17: Uitenhage. 18: Os’town.19: Pt. Elizabeth. 20: Bloem’tn. 21: K. William’s Town. 22: P’maritzburg. 23: Peddie.

The absence of data on unskilled wage rates by region made a direct examination of the relationship between average unskilled wages and regional subsistence costs impossible. Accordingly, the following exercise examined data on average wages for all workers in the manufacturing sector--using data for 1988, the latest year for which the necessary regional wage data are available. For each subsector within manufacturing, the average wage relative to the HSL-6 was calculated for each regional center. The weighted average of these ratios in each subsector was then calculated, weighted by black employment in each subsector in each region. This indicates the subsector average of wages relative to subsistence costs, and was defined as 100. Finally, the percentage deviation of each regional center from this weighted average was calculated for each subsector, giving a measure of the regional dispersion of average wages relative to HSL-6. 1/

The results of this exercise are shown in Chart 18. The bars indicate the highest and lowest deviations of average wages from the subsector average of wages relative to HSL-6. The stars indicate the regional centers which account for more than 5 percent of subsector employment. The sectors have been ordered, so that those sectors with the lowest average wage relative to HSL-6 appear on the left, increasing toward the right. The weighted average of each subsector’s average wage relative to HSL-6 is noted along the horizontal axis--for sub sector A (clothing), average wages are 90 percent of HSL-6. This ordering is an attempt to show those sectors where the unskilled workers dominate on the left hand side of the chart. 2/

CHART 18
CHART 18

SOUTH AFRICA WAGE DISPERSION AROUND SUBSISTENCE, IN 1988 IN MANUFACTURING 1/

Citation: IMF Staff Country Reports 1995, 021; 10.5089/9781451840919.002.A003

Sources: South African Labor Statistics 1994; Institute for Planning Research, UPE.1/ Subsistence is measured as the average of HSL-6 in March and September.2/ Sector legends are as follows:A: Clothing B: Leather. C: Footwear. D: Wood products. E: Textiles. F: Non metal minerals.G: furniture. H: Pottery. I: Food. J: Plastic. K: Metal goods. L: Other manufacturing.M: Vehicle Darts. N: Electrical. 0: Bsveraaes. P: Professional. 0: Porier ooods. R: Machinerv.

The data suggest high rates of regional variation--of around 40 percentage points around each sub sector’s weighted average--even just focusing on those regions that account for more than 5 percent of subsector employment. There is some suggestion in the data of a diminution of dispersion in sectors with lower average wages, but sectors A-C--clothing, leather products, and footwear--employed only 23,000 black people in 1988, compared with black employment in all manufacturing of some 471,000 in the same year, and have plants in only a few regional centers. Hence, this suggestion of more compressed regional dispersion in the lower income subsectors does not appear to be significant.

Clearly, this data on regional wages is less than ideal: it includes skilled labor, so part of the high variation may reflect the dispersion of skilled wages within each subsector. Nevertheless, if there is a nutritional floor on the wages of unskilled workers, then it is not apparent from these data, with the possible exception of the very lowest income sectors.

The weight of the evidence from wage behavior is against the nutritional floor hypothesis account of wage inflexibility in the formal sector during the 1980s. This conclusion is buttressed by the evidence that wages in the informal sector are substantially below those in the formal sector, that young people are disproportionately represented among the unemployed (despite their lesser familial responsibilities), and that unskilled employment has stagnated since the late 1970s. All these cast doubt on this account of downward wage inflexibility underlying the rising FEG.

5. The rise of trade unions

An alternative account the failure of real wages to adjust to the rising FEG in the 1980s emphasizes the growth of trade unions. Efforts to unionize the black labor force go back to the 1920s. But these efforts did not make a substantial impact on labor market behavior until the 1970s, when union growth was spurred following by the 1972-3 strike wave in Natal and in the Cape. The strikes stimulated the establishment of several new trade unions, the direct antecedents of the modern unions in South Africa.

Following the emergence of black union activity in the 1970s, the Industrial Conciliation Amendment Act of 1979 radically changed official treatment of black trade unionism: it recognized black people as employees, thereby incorporating them into the statutory wage negotiating structures (discussed below), it withdrew restrictions on mixed-race unions, it repealed most job reservation by racial group, and it made a number of other notable changes including the establishment of the Industrial Court.

The rise in the FEG is correlated with the growing economic strength and legal standing of black trade unions. Membership in registered trade unions 1/ rose from 677,000 in 1977 to 1,879,000 in 1987, with the Manpower Commission estimating that a further 240,000 members belonged to 88 unregistered unions in 1987, bringing total union membership to around 2.1 million or over half of those employed in the formal sector outside government. 2/ By 1993, total registered union membership had risen further, to 2.5 million persons. The largest of the union federations, COSATU, represents almost entirely unskilled and semi skilled workers; its membership stood at some 1.2 million persons in the late 1980s. The number of registered strikes rose from 101 in 1979 to 1,148 in 1987.

The link posited between growing union legal and economic strength and the FEG data has one key merit: it provides an account of the timing of the acceleration in the FEG in the late 1970s, and of its continued rise during the 1980s. But it does not stand alone as an account of the rising unemployment: some countries have had large trade unions and low unemployment rates, such as the former West Germany and Austria, which makes clear that union behavior is at least as important as the degree of unionization in determining the link between unionization and unemployment. Furthermore, the fact remains that, union membership of some 2.5 million people in 1993 compares with labor force estimates of some 15 million people. If union behavior lies at the heart of the rising FEG, then it is highly leveraged, even recognizing the much higher degree of unionization of the employed labor force. This has prompted suggestions that the economic power of unions in the 1980s has not rested solely on their improved legal standing and membership, but has been buttressed by other factors. Two such factors that have been suggested are the monopoly power of the producers with which the unions were negotiating, and the international support unions received as part of the sanctions campaigns against the former regime.

It has been suggested that employers enjoying monopoly power in goods markets made concessions to union claims more readily than employers in competitive goods markets, because concessions by monopolists merely represent transfers of part of the economic rent; they do not threaten the viability of the enterprise. However, the role of this factor in the rising FEG cannot be assessed. Notwithstanding the prominent monopoly producers that can be readily identified, the extent of monopoly power in goods markets is not well established, 1/ and the available data do not permit a decomposition of wage data by the monopoly power of employers.

Multinational firms operating in South Africa became increasingly subject to international pressures for political reform in South Africa during the 1980s. In seeking to follow guidelines such as the Sullivan principles, some made substantial concessions to union pressures for recognition and for wage increases, and there appears to have been a “demonstration effect” on locally owned companies. This may, in part, account for the particularly substantial real wage gains secured for unskilled labor in the metal and engineering industries shown in Chart 16. Nevertheless, the contribution of these factors to rising FEG rates throughout the 1980s is unclear.

While unionization appears to account for the timing of the rise of the FEG, and for its relentless rise in the 1980s, the leverage that union power obtained is difficult to account for. However, a third set of factors that may have accounted for the leverage that union power enjoyed in the labor market concerns the institutions of wage determination. These issues are discussed below.

6. The structure of pay bargaining

There are a variety of structures governing pay determination. These vary across sectors and firms. In the cases of the mining sector and Government, wage determination predominantly involves centralized direct negotiations between employers and union representatives. Agriculture and domestic service are largely not unionized, while the informal sector is totally ununionized. Wage Boards--Statutory bodies reporting to the Ministry of Labor--have powers to determine minimum wage rates in sectors in which there is no union representation. However, other sectors--notably manufacturing--which are highly unionized, often use what are known as Industrial Councils (ICs) to determine wages, conditions, and disputes settlement.

ICs are formed at the voluntary initiative of both employer and employee representatives. Each constitutes a negotiating forum with established disputes procedures. ICs set minimum wage rates for each skill category in each industry. These rates obtain legal standing and may be enforced in the courts, though some exemptions are granted to specific firms on appeal. There were 86 IC agreements in force in 1993. 1/ 2/

IC agreements set floors on wages for all firms covered by the IC. ICs rarely determine actual wage rates, or percentage increases in wage rates; these are determined in plant level bargaining that occurs subsequent to IC agreements. 3/ Wages can be substantially above the IC minima: in the iron and steel industries, for example, the current wage rates for unskilled workers are commonly 15-20 percent above the IC wage floors for their skill levels, while for artisans, the actual wage rates are commonly 50-60 percent above the IC wage floors for artisans. There is, however, wide dispersion in behavior in this regard between sectors and firms.

In the period after 1979, the black unions tended to focus their efforts on plant level negotiation, and in the process, the gap between IC wage minima and actual wages tended to widen. IC wage agreements came to be perceived as the starting point in wage negotiations, rather than as the principal substance of those negotiations. This accounts for the evidence that the pay for unskilled labor has risen more rapidly than the IC minima during the 1980s. This is most apparent in the metal and engineering sector, where the IC rates for unskilled labor in iron and steel and in the motor industry remained broadly unchanged relative to HSL-6 while average wages rose strongly. In the building industry, the IC floors on wages for unskilled labor fell relative to HSL-6, while average rates of pay rose. Accordingly, plant level negotiation, rather than IC agreements, seems to underlie the evidence of rising wage rates for unskilled labor in the face of the rates during the 1980s.

Despite the apparent importance of plant level negotiations in wage determination in firms that use ICs, there are three mechanisms within the IC structures which may leverage union power, and strengthen their hand in plant level negotiations. First, wages and conditions negotiated at IC level are enforced not only in those firms which are directly represented on the IC, but they can also be enforced by law in firms that are not represented. This extension is determined on a case-by-case basis by the Ministry for Labor at the request of the IC. It is approved if the parties on the IC are viewed as being “sufficiently representative” of the sector as a whole. 1/ Second, the floors that ICs set on wages are applied uniformly throughout the country; they take no account of regional disparities in the costs of subsistence. This, compounded by plant level wage negotiations, may account for the wide dispersion of wages relative to subsistence shown in Chart 18. Not only might this discourage greater regional decentralization of industrial activity, but it may reinforce the bargaining position that unions are able to achieve in the “core” producing regions of each industry. Finally, neither IC nor Wage Board wage floors take any account of age--an unskilled worker who is 16 years old has the same IC and Wage Board wage floor as another of 60. This may leverage union power by reducing the power of unemployed unskilled youth to bid themselves into work. In a population so much of which is relatively young, and where such a large share of the unemployed are young, this mechanism may be important in leveraging union power.

7. Participation, youth unemployment, and rural-urban migration

The suggestion that real wages became increasingly inflexible downwards in the formal sector during the 1980s occurred in a context where participation rates were rising, where youth unemployment was increasing, and where controls on rural-urban migration were substantially eased. This section discusses the role of each factor in raising the FEG.

Increases in the participation rate (Chart 13)--a result of falling participation rates for males and rising participation rates for females--and growing social tensions were both notable features of labor market developments in recent years. Downward inflexibility of unskilled wages meant that the FEG bore the brunt of the burden of adjusting to both developments; those unable to secure work in the formal sector were forced into the informal sector, or into open unemployment. However, the rise in the participation rate was not substantial, and in that sense it accounts for a relatively small part of the rise in the FEG.

As the FEG rose, there is some evidence that young people were disproportionately affected. A tracer survey of two cohorts of standard-10 school leavers from schools in the Johannesburg and Pretoria areas--the classes of 1984 and 1988--conducted in 1992 1/ found that after leaving school, unemployment rates--defined in the same way as the official unemployment rate--are high, but they fall steadily over the subsequent years. In 1992, the unemployment rates for the class of 1984 were 17.8 percent and 24.4 percent for males and females, respectively, compared with 43.1 percent of and 65 percent for the class of 1988 (see Table 11). When the researchers compared the class of 1984 with that of 1988 for a given number of years in the labor force, the latter experienced considerably higher rates of unemployment. 2/

Table 11.

South Africa: Activity of 1984 and 1988 Standard 10 School Leavers in 1992

article image
Source: P. Bennell and M. Monyekolo, 1992.

The results from the OHS are consistent with the tracer survey data: clearly, youth unemployment rates are substantially higher than those for other age groups. The pattern of youth employment--high rates immediately after leaving school but falling steadily over time--may account for the otherwise puzzling dominance of long-term unemployment alongside evidence that few of the unemployed have held a previous job; many of the long-term unemployed may be relatively young persons.

Conventional models of rural-urban migration 3/ predict that the increases in formal sector real wages rates raise the incentive for rural residents to migrate to urban areas, by raising the expected income from migrating. As a consequence, urban unemployment rates rise as recent migrants seek work. It is sometimes suggested that this may account, in part, for the rising FEG. However, this is unlikely. As a purely statistical point, the definition of the labor force in the FEG includes all persons engaged in subsistence agriculture, but subsistence agriculture is excluded from the formal sector employment data. Hence, migration from subsistence agriculture cannot account for the rising FEG. Even the relaxation of influx control laws in the mid-1980s, which made migration to urban areas substantially easier, occurred after the FEG began to rise. Finally, empirical surveys of migrant behavior in Africa and elsewhere suggest that in most cases, migrants only move to urban areas after first having obtained some assurance of employment there, usually through informal information networks. 1/

8. Synthesis and conclusions

While the current unemployment of skilled labor may have a substantial cyclical element, the unemployment of unskilled labor is overwhelmingly structural. This conclusion is underlined by the prolonged stagnation of unskilled employment. The roots of this problem appear to lie in the growth of union leverage for employed insiders in plant level wage negotiations. Its consequences have been particularly injurious for the young, for those forced into informal employment, and for those of the unemployed with limited or no support from their families.

Nevertheless, there is much that is unclear about unskilled unemployment. The relative roles of the IC structures and union behavior cannot be unambiguously disentangled using the official data in those sectors and firms where ICs apply, and wage and employment behavior in sectors and firms outside IC structures is an area where further work is necessary. Both problems limit the conclusions that can be drawn about how to address the unemployment problem. But the discussion does suggest some guidelines that the new government might follow in its approach to unemployment:

  • * There is a clear need for upgraded training, both in the education system and for adults. Firms are most likely to fund programs of firm-specific training, and these might also be the quickest to yield benefits. But training is typically slow to produce results discernible at the level of the general economy.

  • * Consideration should be given to differentiating wage minima by age as part of the attempt to address youth unemployment. The Netherlands provides a model for this: minimum wage legislation there provides a steeply inclined schedule rising with age, and the full minimum wage is only payable to those over 22 years old. The ratio of youth to total unemployment in the Netherlands is among the lowest of the industrial countries. 2/ Such a schedule could be incorporated both into the Industrial Council agreements, into Wage Board rulings, and into public sector wage scales.

  • * Industrial Council agreements should explicitly recognize regional differences in the cost of living in the wage floors they set as part of an effort to achieve a better reflection of regional living costs in wages. As the first stage in the wage determination process, IC agreements constitute a useful forum where this initiative could be taken.

  • * The mandatory extension of Industrial Council agreements to firms that are not represented on the councils--commonly smaller firms--should be terminated. If firms that are not party to IC negotiations wish to adopt IC wages and conditions, they should be able to do so, but on a voluntary basis.

  • * Despite the evident caution exercised by the Wage Boards in setting minimum wages in the past, it is clear that their wage rates are binding in many cases. Given the severity of the present unemployment problem, the conservative stance of the Wage Boards should, at the very least, be maintained.

But these specific steps should not obscure the need for a greater responsiveness on the part of unions at plant level to the needs of unemployed unskilled labor. The impact of the measures noted above in addressing the unemployment problem would be substantially enhanced if plant level negotiations actively incorporated them. Clearly, the call made by President Mandela at the COSATU conference in 1994 for wage agreements to reflect the need to stimulate job growth is well placed.

APPENDIX I: The Household Subsistence Level

This measure has been calculated by the Institute for Planning Research at the University of Port Elizabeth on a consistent basis since 1972.

This data set shows the monthly cost of a basket of basic commodities for households of different sizes and compositions in up to 23 different mainly urban centers, and it is calculated in March and September of each year.

The Household Subsistence Level for a household of six low-income persons is used here because that is roughly the average size of low income households for the period being studied. More recent data suggest that average household size for low income groups has fallen to five persons. This reduces the household subsistence level by over 10 percent below that for a household with six persons--the exact decline depending on whether one removes an adult or a child from the household composition used for constructing the subsistence level.

From September 1994, the consumption bundle used for the calculation was changed--the first time this has happened--on the basis of new nutrition scales from the Ministry of Health. This reduced the level of the low income subsistence index further. (See Fact Paper 98 for details on this change to the basket.) This change, along with that suggesting a diminution in average household size, illustrates the problems with defining subsistence levels in absolute terms.

The nutritional component of the basket used up to September 1994 (and therefore in all exercises in this chapter) was defined by the Ministry of Health, based on Food and Nutrition Board, NRC USA 1974. The researchers producing the index indicate that since the basket is based on the needs of U.S. residents, the requirements of South Africans may be somewhat overstated by the index.

The clothing component is taken from Suttner (1966), with adjustments for children on a sliding scale according to age. Other items in the basket include fuel, lighting, cleaning materials, heating, rent, and commuting transport costs.

The basket excludes, notably, medical and education expenses, holidays and entertainment, insurance, depreciation of household assets, and noncommuting transport costs.

Prices are surveyed in supermarkets. Special discounts are ignored, and where goods are sampled several times in one area, the mid-price charged for the good is used.

References

  • Banerjee B.,Information flow, expectations, and job search: rural to urban migration in India.Journal of Development Economics (1984) Vol 15 pp. 239257.

    • Search Google Scholar
    • Export Citation
  • Bennell P. with Monyokolo M.,The Lost generation? A tracer survey of African Secondary school leavers in the PWV area.University of Witwatersrand October 1992.

    • Search Google Scholar
    • Export Citation
  • Bliss C.J. and Stern N.H.,Productivity, Wages and Nutrition, 1: The Theory.Journal of Development Economics Vol. 5. (1978) pp. 331362.

    • Search Google Scholar
    • Export Citation
  • Epstein A.L.,Urbanisation and social change in Africa” in G. Breese (ed) “The City in Newly Developing Countries: Readings on Urbanism and Urbanization” Prentice Hall. NJ 1969.

    • Search Google Scholar
    • Export Citation
  • Harris A. and Todaro M.,A model of labor migration and urban unemployment in less developed countriesin the American Economic Review March 1969 (vol 59) pp. 139148.

    • Search Google Scholar
    • Export Citation
  • Hofmeyr J.,African wage movements in the 1980sSouth African Journal of Economics, 61, 4, 1993.

  • Moghadam R.,Why is unemployment in France so high?IMF Working Paper Series WP/94/58 1994.

  • Nickell S., Vainiomaki J., and Wadwhani S.,Wages and Product Market Power.Economica (November 1994.)

  • Suttner S.,The cost of living in Soweto 1966South African Institute for Race Relations, Johannesburg 1966.

  • “White Paper on Reconstruction and Development” Cape Town 15 November 1994 WPJ/1994.

  • Fact Paper 98The Household Subsistence Level in the Major Urban Centers of the Republic of South AfricaUniversity of Port Elizabeth September 1994.

    • Search Google Scholar
    • Export Citation
  • SALDRUSouth Africans Rich and Poor: Baseline Household StatisticsSALDRU School of Economics, University of Cape Town. August 1994.

    • Search Google Scholar
    • Export Citation
1/

This excludes the TBVC States--Transkei, Bophuthatswana, Venda and Ciskei--which were treated as independent states when the rate of unemployment was measured in October 1993.

2/

The formal sector is defined as firms that are enumerated by the employer surveys of the Central Statistical Service.

1/

This classification of skills is based on the Manpower surveys. These define unskilled employment as that which requires at most a short in-house training period (2 weeks) and primary school level literacy and numeracy. Many persons employed in these occupations exceed these educational requirements. The data show that the share of unskilled employment in total employment was a little over 60 percent until the 1970s, when it began falling slowly and steadily to a little under 60 percent in 1980. Thereafter, it fell more rapidly to 50 percent in 1990.

2/

In Chart 15, wages are deflated by an index of subsistence known as the Household Subsistence Level, which is described in Appendix 1. It moves closely with the low income consumer price index.

3/

Unofficial estimates of the trends in employment and value added in the informal sector have been based on various measures, including the discrepancy between the output and expenditure estimates of GDP and on the behavior of notes and coin. Many official impediments to informal sector activities were relaxed during the 1980s and anecdotal evidence is suggestive of a mushrooming of activity consequent upon this relaxation. The OHS contains the current official estimate of employment and value added in the sector.

1/

Even though the OHS in 1993 suffers from the defect that the TBVC states are excluded, its findings on unemployment rates are close to those of the SALDRU survey which included the TBVC states. This was a large household survey undertaken by a consortium of foreign donor and private domestic research organizations in 1993. See SALDRU 1994.

2/

South African Labor Statistics 1994.

3/

Unemployment rates are 47 percent for those between 20 and 24 years of age, 32 percent for those between 25 and 34, 20 percent for people between 35 and 54, and 15 percent for those between 55 and 64 years old.

1/

A nutritional floor on wages for a household depends inter alia on household size and composition, on the number of income earners, on the incidence of income in kind, on own production by the household, on the pattern of consumption by the household, on the pattern of intra and inter household transfers, and on the mean and variance of all these characteristics amongst the unskilled labor force. Furthermore, the relevant concept is not the “actual” nutritional floor, as might be defined by nutritionists, but what employers perceive to be the floor.

2/

However, the concept of wages relative to subsistence is more troublesome for domestic workers than for others, since domestic workers receive payments in kind to cover part of their subsistence needs.

2/

The low income CPI is based on the consumption patterns of low income households. Its movements are usually close to those of the HSL-6.

1/

Consider the following illustrative example. In a subsector, average wages in three regions are 100, 150 and 200 rand a month, while HSL-6 in all three regions is 150. If employment in all three regions is the same, the weighted average of the three ratios of average wages to HSL-6 in each region would be 1. The percentage deviation of each region’s ratio from the weighted average for the whole sector would be -33, 0, and + 33, respectively. On Chart 18, this would be shown as 67, 100, and 133.

2/

Space constraints prevent the display of the results for all subsectors in manufacturing in Chart 18. The sectors not shown are those with relatively high wage rates, but they show the same degree of wage dispersion as the sub sectors shown.

1/

Until 1979, no black unions were registered, though many existed.

2/

SALDRU’s estimates of union membership are somewhat higher for the beginning of 1988; they estimated it at 2.6 million.

1/

The evidence on monopoly power in goods markets is discussed in Chapter VI. See Nickell et al (1994) for an analysis of these issues in the United Kingdom.

1/

ICs were first established by the Industrial Conciliation Act of 1924. From 1988, teachers or lecturers employed in institutions fully or partly state funded were brought under the ambit of the IC system.

2/

The Industrial Council (IC) structure into which the black unions were incorporated in 1979, along with other labor legislation, provides a number of legal restrictions on unions. Strikes (and lockouts) are illegal if implemented before the established disputes settlement process has run its course, if they relate to matters already settled in a current IC agreement, or if the case is submitted to voluntary arbitration. The legislation also requires that strikes must be preceded by a secret ballot in which a majority of the trade union members supported the action. (Similar rules apply to lockouts.) Ballots cannot be held until a 30-day cooling off period has elapsed. Strikes are always illegal in essential services--power, light, water, sanitation, passenger transport, and fire extinguishing services “within the area of a local authority”.

3/

However, the iron and steel industry is moving to establish a direct link between the percentage increases agreed for IC wage floors for each skill category, and the percentage increases in actual wages that are paid to employees in those skill categories.

1/

A variety of measures have been used to determine whether or not an IC is sufficiently representative, most notably the share of employees that employers on the IC employ, and the number of employers as a proportion of all employers in an area or sector--to assess the representativeness of employers--and the share of employees represented by the trade unions on the IC--to measure the representativeness of the unions. When extension has been granted, it applies not only to existing firms in the industry, but also to any subsequent entrants while the IC agreement is in force.

2/

It should be noted that students that have attained standard 10 education represent an educational elite among black people.

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

    SOUTH AFRICA LABOR FORCE AND FORMAL SECTOR EMPLOYMENT, 1960–94

  • View in gallery

    SOUTH AFRICA FORMAL SECTOR EMPLOYMENT, 1960–93

  • View in gallery

    SOUTH AFRICA AVERAGE BLACK WAGES RELATIVE TO SUBSISTENCE, 1984–92

    MANUFACTURING 1/

  • View in gallery

    SOUTH AFRICA AVERAGE WAGES FOR UNSKILLED LABOR, 1980–92

    Relative to Subsistence 1/

  • View in gallery

    SOUTH AFRICA HSL-6 IN REGIONAL CENTERS, 1989–93

    in current rand. 1/

  • View in gallery

    SOUTH AFRICA WAGE DISPERSION AROUND SUBSISTENCE, IN 1988 IN MANUFACTURING 1/