South Africa faces unemployment and deindustrialisation. Manufacturing tariff cuts in the 1990s were expected to reallocate workers into services. Our research finds the opposite: municipalities with larger tariff cuts experienced weaker employment growth in manufacturing and services. Wages did not adjust – the shock was absorbed through job losses. Negative spillovers from manufacturing constrained services from absorbing displaced workers; they either exited the labour force or migrated to other municipalities.

Introduction

South Africa’s unemployment rate has hovered above 30% for much of the past decade. Amid this jobs crisis, policymakers have pinned hopes on the services sector to absorb workers displaced from manufacturing. In 2006, manufacturing accounted for 16.4% of total employment in formal non-agricultural businesses in 2006; by 2015, this share had fallen to 12.5%. This has raised concerns about premature deindustrialisation and its impact on future growth and job creation.

South Africa is currently renegotiating trade agreements and debating the future of its industrial policy. Understanding how tariff reforms affect employment across sectors is essential for policymakers. Our evidence shows that, under the current tariff policy, the services sector is unable to absorb displaced manufacturing workers. This study demonstrates that the effects of tariff reductions extend beyond manufacturing. Because manufacturing and services are closely interconnected, adverse shocks to manufacturing can also weaken employment growth in services, limiting structural transformation.

The findings are particularly relevant as South Africa seeks to revitalise manufacturing while expanding employment opportunities in services. Current debates on industrial policy, localisation, and trade agreements often treat manufacturing and services as disconnected sectors. Our findings suggest that this distinction underestimates the broader employment consequences of trade shocks.

Background

During the 1990s and early 2000s, South Africa dramatically reduced import tariffs as part of its reintegration into the global economy. Lower tariffs exposed domestic firms to greater international competition. While this improved efficiency and lowered prices for consumers, it also placed pressure on previously protected industries. Manufacturing was one of the sectors most affected. Average tariff exposure fell from 16.6% in 1996 to 8.6% in 2011, with much of the decline occurring in the 1990s, as shown in Figure 1.

Interestingly, female employment-weighted tariffs were higher and fell more sharply than those for males (24% to 11% compared with 15% to 7%). This difference can be attributed to the relatively sharp reductions in tariffs on industries such as clothing, textiles, and footwear, in which female employment is concentrated. This suggests that trade liberalisation may have a gendered impact, with more pronounced effects on women than men.

Figure 1: Evolution of average municipality-level tariff exposure

Notes: Simple average manufacturing tariff exposure measure across municipalities, calculated using 1996 employment shares of all workers, male workers and female workers within manufacturing in each municipality using the 10% sample of the 1996 Population Census and tariff data updated from Edwards (2005).

The post-apartheid period also corresponds with substantial changes in the sectoral composition of employment. Table 1 presents employment levels and shares across major sectors. Services account for the bulk of employment, with their share rising from 69.7% in 1996 to 82% in 2011. Manufacturing, in contrast, has experienced a decline in its share of total employment, falling from 14% to 9.8%, despite an increase in the total number of manufacturing workers. This structural shift towards services is consistent with several factors, including deindustrialisation arising from relatively rapid productivity growth in manufacturing, increased international competition following trade liberalisation, and the servicification of manufacturing – the growing use of services such as logistics, finance, and information technology in manufacturing production. Our study focuses on the role of trade liberalisation in driving this trend, while accounting for spillover effects, including those linked to servicification.

Data and empirical approach

To understand the effect of manufacturing tariffs on the services sector, we examine wage and employment outcomes at the local level. This level of analysis is motivated by the fact that services are largely non-traded and depend on local demand from manufacturing firms and households. If manufacturing contracts, demand for services inputs – such as transport, finance, maintenance, and business services – declines, as does household spending and investment in local infrastructure. These spillover effects are most evident at the municipal level, where the linkages are strongest.

Our empirical analysis draws on employment data for 234 municipalities from the South African Population Censuses for 1996, 2001, and 2011, combined with industry-level nominal tariff data from Edwards (2005), which we update using published tariff schedules. This provides a rich dataset covering a period of substantial trade liberalisation.

Municipalities differed considerably in their pre-liberalisation industrial composition. Municipalities with a high share of employment in protected industries faced larger tariff cuts, while those with more diversified or less protected manufacturing bases faced smaller changes. This variation in tariff exposure across municipalities allows us to compare employment effects of tariff reductions while controlling for pre-liberalisation municipality differences. These include the skill composition of the working-age population, the unemployment rate, the manufacturing employment share, and measures of infrastructure. We also control for time-varying factors, such as the size of the working-age population, the labour force, migration rate, and period fixed effects to capture general trends in labour market outcomes common to all municipalities.

What does the evidence show?

The analysis produced several important findings that challenge conventional views of how local labour markets adjust to trade liberalisation.

Trade liberalisation had little effect on wages. Firms responded to increased import competition by reducing employment rather than reducing wages. This may reflect South Africa’s relatively rigid wage-setting institutions, where collective bargaining and sectoral wage agreements limit downward wage adjustments.

Employment grew more slowly in both manufacturing and services. Municipalities exposed to larger tariff reductions experienced slower employment growth in both manufacturing and services. This finding is surprising because economists typically expect workers leaving manufacturing to find employment in the growing services sector.

Negative spillover effects from manufacturing to services. Manufacturing and services are closely interconnected through production, income, and local investment linkages. When manufacturing contracted in response to import competition, demand for services also weakened. Rather than absorbing displaced workers, the services sector also came under pressure as there was less money in circulation to pay for them.

The adjustment process differed for women and men. Women were more heavily exposed to tariff reductions because they were concentrated in industries such as clothing, textiles, and footwear, where tariffs fell the most. At the same time, the adverse spillover effects on services employment were stronger for men, highlighting that trade reforms affect workers differently depending on where they are employed, underscoring the importance of incorporating gender into trade adjustment policies.

Displaced workers also did not move to other sectors. We found little evidence that displaced manufacturing workers found employment in the primary sector. Instead, workers appear to have left the labour force altogether or migrated to other regions in search of work.

Overall, the findings suggest that labour market rigidities, sector-specific skills, and limited labour mobility constrained worker adjustment.

Why didn’t the services sector absorb displaced workers?

Lower demand for services by manufacturing firms. Manufacturing firms rely on transport, financial services, maintenance, security firms and many other business services. When manufacturing output contracts, demand for these services also falls, weakening employment growth in both sectors.

Household spending declined. Workers who lose manufacturing jobs have less income to spend in local shops, restaurants, and other service businesses. Lower household spending reduces demand for local services, creating second-order effects of employment losses beyond manufacturing itself.

Infrastructure investment slowed. Slower economic activity reduces investment in infrastructure and local development. With fewer new projects and weaker business confidence, opportunities for employment growth in services become even more limited.

Policy implications

The evidence points to several key policy implications.

Improve coordination between trade, industrial, and labour market policies. Trade policy is often developed separately from labour market and industrial policy. This separation underestimates the broader employment consequences of trade reforms. Greater coordination across government departments would help ensure that trade liberalisation is accompanied by complementary measures that support firms, workers, and local economies during adjustment.

Invest in skills and regional support. Retraining programmes can help displaced workers move into new occupations, but they must be aligned with areas where employment opportunities are likely to emerge. Training initiatives should be designed alongside industrial and regional development strategies, rather than operating in isolation. Municipalities heavily exposed to trade shocks require targeted support, including infrastructure investment and incentives to attract new investment, to diversify their economic base and create new employment opportunities.

Develop gender-specific policies. Women and men experienced different labour-market adjustment paths following trade liberalisation because they were concentrated in different industries and occupations. Policies supporting displaced workers should recognise these differences when designing retraining programmes, employment services, and income support. A more gender-responsive approach would improve the effectiveness of adjustment assistance and promote more inclusive labour market outcomes.

Conclusion

Trade liberalisation remains an important tool for improving productivity and long-term growth. However, our findings show that its labour market consequences are more complex than is often assumed. Manufacturing and services are deeply interconnected, meaning that policies aimed at supporting structural transformation cannot focus on individual sectors in isolation. Trade policy, industrial policy, and labour market policy need to work together if South Africa is to achieve more inclusive growth.

International experience suggests that such policy coordination is both feasible and beneficial. For example, Denmark’s flexicurity model illustrates how flexible labour markets can be combined with active labour market policies, retraining, job-search assistance, and income support to facilitate workers’ adjustment to economic restructuring. More broadly, evidence from OECD countries suggests that complementary labour market policies can facilitate workers’ adjustment to trade-induced structural change by reducing adjustment costs and supporting labour reallocation.

While South Africa’s institutional context differs substantially from that of Denmark and other OECD countries, these experiences highlight the importance of coordinating trade, industrial, and labour market policies to support structural transformation and promote more inclusive growth.

Written by Refilwe Lepelle and Lawrence Edwards, Econ3x3