VoxEU Column Industrial organisation Labour Markets

How privatisation impacts workers: Evidence from Brazil

In the early days of his administration, Brazilian President Jair Bolsonaro announced plans to privatise several of the country's largest state-owned enterprises and airports. Fearing such a move would lower both wages and employment, labour unions organised in opposition to Bolsonaro’s plans. This column looks anew at evidence testing whether privatisation offers more than merely an immediate infusion of revenue, and finds that while increases in efficiency might contribute to Brazil’s overall economic growth, privatisation could also expose the country’s most vulnerable workers to significant risk of decreased wages.

The privatisation of state-owned enterprises (SOEs) remains a popular policy tool for debt-burdened governments looking to raise large amounts of revenue quickly. Currently, Brazil is on the precipice of a new privatisation wave. After taking office in October 2018, Brazilian President Jair Bolsonaro announced plans to sell several state-owned assets, including multiple airports and portions of some of the country’s largest companies, including the utility giant Eletrobras and the petroleum company Petrobras. Supporters of Bolsonaro’s plans argue that privatisation will raise desperately needed revenue for the government and improve the efficiency of the privatised firms. Opponents of the plan include workers and trade unionists, who have organised multiple protests in the fear that privatisation will lower wages and employment. 

There is limited empirical evidence to show how privatisation impacts workers. Most of the academic literature on privatisation focuses on how it impacts firm profits and efficiency (Megginson and Netter 2001, Porta and Silanes 1999). However, the source of these gains is not always clear, and one possibility is that firms increase profits by lowering wages and employment. Therefore, while privatisation may benefit firm owners, it could also lead to declines in worker welfare, as union members fear. 

In a recent paper (Arnold 2018), I aim to contribute evidence to this debate by studying Brazil’s privatisation programme in the 1990s, which raised roughly $87 billion in revenue, making it one of the world’s largest privatisation programmes. Brazil makes an ideal case study for privatisation, given both the current policy debates and the available administrative data. My study uses matched employer–employee data, which contains earnings and demographic information for the country’s entire formal sector. To estimate the impact of privatisation on workers, I compare earnings for workers in SOEs that become privatised to otherwise similar workers both before and after privatisation. Given the rich administrative data, I am able to construct a valid control group for the privatised workers by matching each privatised worker to a worker of similar age employed in the same occupation and industry. 

Figure 1 plots the average log monthly wage (in real terms) for workers in the privatised SOEs alongside workers in the matched control group. As can be seen in Figure 1, in the years prior to privatisation, earnings are similar for both the privatised and control workers. After privatisation, earnings for the privatised workers fall dramatically. Ten years after privatisation, the privatised workers earn roughly 26% less than the control workers.  

Figure 1 On average, workers in privatised SOEs lose significant earnings following privatisation

This indicates that, on average, workers in privatised SOEs lose significant earnings following privatisation. But it does not show how those losses are distributed across the workforce. It could be that politically connected managers are able to inflate their wages in SOEs relative to private-sector firms. To explore this possibility, Figure 2 plots the impact of privatisation on log monthly earnings for three different groups: workers without a high-school degree, workers with a high-school degree, and workers with a college degree. As can be seen in the figure, earnings losses are by far the largest for workers without a high-school degree, suggesting that the losses are not solely attributable to large rents for politically connected managers. For these workers, earnings fall by about -50 log points relative to the matched control group, in contrast to a -20 log-point decline for other groups. Therefore, privatisation leads to persistent earnings losses for all groups, but the losses are greatest for the least economically advantaged group. 

Figure 2 Earnings losses are by far the largest for workers without a high-school degree

These direct effects of privatisation are large and economically meaningful. Even so, they will underestimate the total impact of privatisation on workers in the presence of general equilibrium spillovers, which might arise for three reasons. First, changes in product-market competition due to privatisation may impact labour demand. Second, large declines in employment at SOEs will increase the supply of workers to private-sector firms, thereby reducing wages in the private sector. Lastly, large reductions in wages offered by SOEs may result in less wage-competition between SOEs and private-sector firms, resulting in further wage decreases at firms not directly impacted by privatisation.

To test for spillovers, I construct a measure of exposure to privatisation that traces how the effects of privatisation spread to those private-sector firms connected to privatised SOEs by labour mobility. For example, before privatisation a large fraction of electrical technicians were employed in SOEs, leaving electrical-technician jobs at private-sector firms highly exposed to privatisation. I find that exposure to privatisation is associated with a significant decline in wages. Specifically, a one standard deviation in exposure to privatisation is associated with about a 1.6% decrease in the market wage. 

To put this number in perspective, I compute a simple summary calculation which suggests that privatisation decreased the aggregate wage in the formal sector of the Brazilian labour market by about 4.4%. Roughly one quarter of this effect can be attributed to the direct effect on privatised workers. That is, taking into consideration only the impact of privatisation on workers employed at privatised SOEs, the aggregate wage in Brazil would have declined by about 1.1% following privatisation. Taking into account spillovers increases this number to 4.4%. Incorporating spillovers is therefore crucial for understanding the total impact of privatisation on workers in this context.  

These results suggest that privatisation can have a sizeable negative impact on workers. However, even with this negative impact, privatisation may still be welfare-improving to an economy overall. Increases in efficiency will contribute to economic growth and could outweigh the negative impact on exposed workers. On the other hand, if privatisation increases profits primarily through reduced wages and employment, then the perceived ‘efficiency benefits’ of privatisation might actually be due more to decreased worker welfare than to increased output and lower prices for consumers. 


Arnold, D (2018), “The Impact of Privatisation of State-Owned Enterprises on Workers”, Industrial Relations Section Working Paper 625.

Megginson, W L and J M Netter (2001), “From State to Market: A Survey of Empirical Studies on Privatisation”, Journal of Economics Literature 39(2): 321-389.

Porta, R L and F Lopez de Silanes (1999) “The Benefits of Privatisation: Evidence from Mexico”, The Quarterly Journal of Economics 114(3): 1193-1242.

6,614 Reads