The Employment Consequences of Anti-Dumping Tariffs: Lessons from Brazil


With the promise of "bringing jobs back", import tariffs are advocated to increase national employment. Despite its relevance to policy and prominence in the political debate, the aggregate effect of tariffs is still a source of debate among economists. On the one hand, tariffs shift demand from overseas to the home market. Therefore, through this shift in demand, the national producer and sectors upstream to it could increase production and employment. But, on the other hand, downstream firms, the ones using the taxed good as input, have a higher cost, which could lead to lower employment. Therefore, the final effect of tariffs on employment will depend on the employment elasticity of the national producer, upstream firms, and downstream firms. In a recent paper (de Souza and Li (2020)), we estimate how tariffs propagate through the economy and quantify their aggregate effect on employment.

Our first step is to understand how firms adjust to an import tariff. But that is not so simple! Tariffs are endogenous. They are usually introduced in declining sectors with large international competition after strong lobbying. As a consequence, we cannot simply tease out the effect of tariffs from, say, the effect of strong lobbying.

To disentangle the effect of tariffs from other trends affecting the labor market, we exploit WTO anti-dumping (AD) regulation. According to WTO rules, the decision to impose a dumping tariff depends on the price of the taxed good internationally. Therefore, AD tariffs should not depend on labor market trends, lobbying, or any other endogenous object, but only on some pre-determined characteristic!

Using micro-data for Brazil, we implement differences-in-differences comparing sectors with AD tariff considered but rejected (control group) against sectors that had an AD tariff (treatment group). The identifying assumption of parallel trends is supported by institutions and the data. According to WTO institutions, these two sectors should differ only on the level of international prices of the taxed good but not on its trends. Indeed, we show these two sectors were on parallel trends on ALL the variables we consider until the tariff was imposed.

The fact that WTO rules require AD to be immune to political pressures and uncorrelated with sectoral shocks does not mean that local governments follow their rules. It could well be the case that AD rules are used by influential firms to benefit themselves. Validating our identification, we show that AD tariffs in Brazil do not correlate with lobbying, campaign contribution, procurement contracts, subsidies, tax breaks, and other policies implemented in the period. We show that, using only international prices and a linear model, we can predict AD tariffs with an R-squared of above 0.9. We also implement a series of placebo tests showing that our effects cannot be explained by sectoral shocks or trend breaks. Moreover, sectors that are beneficiaries of AD tariffs are not more likely to make campaign contributions, to receive other protectionist tariffs, or to receive subsidy from the government. The data strongly supports the idea that AD tariffs in Brazil are imposed according to WTO regulations, which supports our identifying assumption.

We find that tariffs increase employment in the sector protected by the AD tariff. But, on the other hand, they decrease employment of firms using the taxed good as inputs, i.e., the downstream firms showed in figure 1b. Surprisingly, firms that sell goods to the protected sector do not increase employment (1c). Therefore, the aggregate effect of tariffs will depend on how they propagate through the input-output connection of firms.

To move from these micro elasticities to macro estimates, we use a model with inputoutput connections and endogenous labor supply decisions. Using the micro elasticities to identify key parameters of the model, we found that by setting higher tariffs on downstream sectors, the government can increase employment by as much as 2.8%. But that comes with a large welfare cost. Consumption would fall by 15.9%.


Using a novel identification strategy and a model, we found that, indeed, tariffs can "bring jobs back", but, that comes with a large welfare cost.


de Souza, G. and H. Li (2020): \The Employment Consequences of Anti-Dumping Tariffs: Lessons from Brazil," Available at SSRN 4099664

Share this