VoxEU Column International trade Poverty and Income Inequality

Trade and inequality: From theory to estimation

What is the effect of trade on inequality? This column presents a unique study examining wage inequality in Brazil after liberalisation. Starting from a closed economy, the column finds that wage inequality will initially rise as only some firms take advantage of the new opportunities. But as trade costs continue to fall and more firms start to trade, wage inequality peaks and begins to fall back.

Until recently, research on the labour market effects of international trade has been heavily influenced by traditional theories such as the Heckscher-Ohlin and Specific Factors models. Those theories provide predictions about relative wages across skill groups or across occupations and sectors. In contrast to predictions of those theories, empirical studies find increased wage inequality in both developed and developing countries, growing residual wage dispersion among workers with similar observed characteristics, and increased wage dispersion across plants and firms within sectors. In a large part due to this disconnect, previous studies have concluded that the contribution of international trade to growing wage inequality is modest at best (see for example the survey by Goldberg and Pavcnik 2007).
We argue that these apparently discordant empirical findings are in fact consistent with a trade-based explanation for wage inequality, but one rooted in recent models of firm heterogeneity and trade. In Helpman et al. (2012) we begin by documenting a number of stylised facts about the level and growth of wage inequality in Brazil that provide support for these recent theories.

  • First, much of overall wage inequality occurs within sectors and occupations rather than between sectors and occupations. We document this fact with a decomposition of the variance of the log wage into the variance within sector-occupations and the variance between sector-occupations. We find that the within sector-occupation component of wage inequality accounts for over two-thirds of both the level and growth of wage inequality in Brazil between 1986 and 1995, as illustrated for the growth of wage inequality in Figure 1.
  • Second, a large share of the wage inequality within sectors and occupations is driven by wage inequality between rather than within firms.
  • Third, both prior findings are robust to controlling for observed worker characteristics. This robustness suggests that wage inequality between firms within sector-occupations is largely residual wage inequality.

We arrive at the latter two findings after separating sources of wage inequality behind the within sector-occupation component.1 The bulk of the change in within sector-occupation inequality is driven by between-firm wage inequality during our sample period. In Figure 2, we plot the change in wage inequality within sector-occupations and its components relative to the base year of 1986. Between-firm wage dispersion dominates the evolution of wage inequality within sector-occupations and drives an inverted U-shaped pattern in wage inequality within sector-occupations (which in turn drives the inverted U-shaped pattern of overall wage inequality in Figure 1 before). This is the component of wage inequality that we aim to capture in our structural model.

Figure 1. Changes in log wage inequality and its components

Source: RAIS 1986-1998 (administrative linked employer-employee records), manufacturing workers with positive wage (last-held top-paid job of year) at firms with at least five employees.

Note: Decomposition of total log wage for workers into between and within components. 1986 is used as the base year, with each series expressed as the difference from its 1986 value

Figure 2. Changes in log wage inequality within sector-occupations and its components

Source: RAIS 1986-1998 (administrative linked employer-employee records), manufacturing workers with positive wage (last-held top-paid job of year) at firms with at least five employees.

Note: Decomposition of log wage inequality within sector-occupations (employment-weighted) based on the Mincer regression of log worker wages on observable worker characteristics as well as a time varying firm effect and the subsequent variance decomposition of the total variance of the log wage into the variance of the worker observables prediction term, the variance of the firm effect, twice the covariance between the worker observables prediction term and the firm effect, and the variance of the Mincer residuals. Reported changes relative to the base year of 1986.

Motivated by these and related findings, we estimate a structural econometric model of firm export status, employment and wages, which is derived from our previous work in Helpman et al. (2010). This model incorporates three sources of firm heterogeneity: productivity, export-market entry costs, and worker screening costs, each of which is central to matching the data. We show that the estimated model is empirically successful in accounting for the observed distributions of wages and employment for parameter values consistent with theoretical restrictions. Exporters pay higher wages than non-exporters for two reasons in the model.

  • First, high-revenue firms pay higher wages. In the model, high-revenue firms choose to be more selective in the labour market, to hire workers with higher match quality, and to pay these workers higher wages as a result. At the same time, the presence of fixed exporting costs implies that high-revenue firms are more likely to enter export markets. This is a selection effect.
  • Second, entering the export market increases firm revenue, some of which is shared with workers in the form of higher wages. That is a foreign market access effect. We find that firm trade participation – through a combination of the firm selection and the foreign market access effects – is important for the model's explanatory power. We show that counterfactual changes in variable trade costs can have sizeable effects on wage inequality through this mechanism of wage dispersion between firms.

A key prediction of our approach is an inverted U-shape relationship between wage inequality and trade openness. Starting from the closed economy, reductions in trade costs necessarily increase wage inequality. But as trade costs continue to fall, wage inequality reaches a peak and starts to fall back. The intuition for this result stems from the discrete jump in a firm's wages that occurs when a firm enters the export market. When some firms export, while others do not, then the additional revenue generated at exporters is shared with only the workers at exporters so that export-market access contributes to wage inequality. When no firm exports, a small reduction in trade costs increases wage inequality, because it induces some firms to export and raises the wages paid by these exporting firms relative to domestic firms. At the other extreme, when all firms export, a small increase in trade costs raises wage inequality, because it induces some firms to cease exporting and reduces the wages paid by those domestic firms relative to exporting firms.

Figure 3 illustrates this prediction for the estimated model using Brazilian data. At the origin, trade costs are prohibitively high and result in a foreign market access measure of zero (on the x-axis in Figure 3). As trade costs fall, the foreign market access measure increases and so does log wage inequality initially. Log wage inequality would reach its peak at a level of Brazil's foreign market access that is outside the observed range of foreign market access during our sample period 1986-1998. Subsequent further trade liberalisations in Brazil may therefore still lead to an increase in inequality up to the peak, from which point on inequality would start to decline with further trade liberalisation.

Figure 3. Counterfactual reduction in variable trade costs and log wage inequality


Note: Counterfactual variation in the variance of log worker wages in response to a reduction in variable trade costs from their autarky level (at close-to-zero log market access) to complete trade openness (when almost all workers are employed by exporting firms). The variance of log worker wages is plotted against the log foreign market access measure, which increases as variable trade costs fall. The distribution of wages across workers weights firm wages by firm employment. Dashed lines correspond to the estimated values of log foreign market access in 1986 and 1990 under the preferred specification (for details see Helpman et al. 2012). The red square indicates the prediction of the model given the baseline 1990 estimated parameter vector.

This prediction is relevant for policymakers concerned about the effect of trade liberalisation on wage inequality. Once an economy is open to trade, our approach suggests that further trade liberalisation can either raise or reduce wage inequality, depending on the initial level of trade openness and whether it is above or below the value at which the peak of wage inequality occurs.

Our model is representative of a wider class of theories in which inherently more capable firms embrace new economic opportunities more frequently than other firms. In our model, globalisation and easier foreign market access present such an opportunity. Other examples of new economic opportunities include the arrival of novel production or communication technologies that facilitate management practices or novel product designs that open consumer markets. In our model, the more capable firms do not only adopt the new opportunity but also simultaneously select workers who are more able to perform the newly created jobs. For this wider class of theories, our analysis suggests that, at times of fast arising new opportunities and accelerated economic growth, economies may experience spikes in inequality but, as the new opportunities become accessible to a larger number of firms and workers, inequality can be expected to fall again.

References

Goldberg, Pinelopi K and Nina Pavcnik (2007), “Distributional Effects of Globalization in Developing Countries”, Journal of Economic Literature, 45(1):39-82.

Helpman, Elhanan, Oleg Itskhoki, and Stephen Redding (2010), “Inequality and Unemployment in a Global Economy”, Econometrica, 78(4):1239-1283.

Helpman, Elhanan, Oleg Itskhoki, Marc-Andreas Muendler, and Stephen Redding (2012), “Trade and Inequality: From Theory to Estimation”, NBER Working Paper 17991.


1Concretely, we split within sector-occupation inequality into four terms: the variance contribution of observable worker characteristics, the between-firm component (from firm fixed effects by year), the covariance between observable worker characteristics with the firm component, and the within-firm residual variance.

3,045 Reads