123II  E DISTRIBUTIONAL CONSEQUENCES OF TRADEE DISTRIBUTIONAL CONSEQUENCES OF TRADECountries liberalize trade because they expect gains for their economy. Previous sections have provided detailed descriptions of the different mechanisms that allow countries to reap such gains from trade and have shown that the gains are likely to be significant. Why is it then that countries sometimes hesitate to reduce trade barriers and why is it that outright opposition to liberalization can sometimes be observed? This section provides some answers to these questions by focusing on the distribution of the gains from trade within countries.1 Not all individuals within an economy necessarily become better off with trade liberalization and this section will pay particular attention to those individuals that may lose from trade liberalization, either temporarily or permanently. The last sub-section analyzes how to ensure that the most vulnerable individuals in an economy, i.e. the poor, are among those gaining from liberalization. 1. TRADE AND INEQUALITY (a) What do trade models say about the distributional changes resulting from trade liberalization?Trade liberalization provides new commercial opportunities for companies that are able to export and provides consumers, through imports, with access to cheaper and different goods. Those imports, however, may be in competition with local production and the relevant local producers may suffer from the new competitive pressure. New export opportunities and the increased competition from imports will lead to the expansion of some activities and the reduction of others and – as is often the case with changes resulting from policy reform – some individuals may gain and others may lose in this process. Since individuals do not necessarily know in advance whether they will be among the losers or winners, they may fear liberalization because of the uncertainty it brings. Others will focus on possible difficulties in the short term. For instance, they may be afraid of having to change jobs, even though they are likely to become better-off in the long term. Regarding the long-term distributional consequences of trade reform, an important question is whether the relatively well-off or the not so well-off gain from trade liberalization, i.e. whether trade liberalization is likely to increase or decrease inequality in societies. Economists today consider the answer to this question to be highly situation-specific, and economic thinking on this question has undergone certain changes over time. The classical link between trade and income inequality is based on the Stolper-Samuelson Theorem developed in a traditional trade model (Heckscher-Ohlin) that assumed full employment. In this model, trade flows are determined by comparative advantage and the latter, in turn, depends on each country’s resources.2As developing countries are typically well endowed with low-skilled labour relative to developed countries, the former were expected to start exporting low-skill labour-intensive goods to the industrialized world. Relative demand for low-skill workers would increase in developing countries and decrease in industrialized countries and the theorem predicted that inequality between high-skill and low-skill workers would probably increase in industrialized countries as a consequence of trade with developing countries.3Along the same lines, inequality would be expected to decline in developing countries. A similar argument could be made with respect to the gains of capital compared with labour. If industrialized countries are considered to be relatively rich in capital, capital-labour inequality would increase in industrialized countries as a result of trade and decrease in developing countries. The Stolper-Samuelson Theorem thus predicted that trade would lead to changes in rewards that were factor specific. Certain factors were expected to gain, independent of whether they were employed in exporting or importing sectors, or companies, while others were expected to lose, again independent of their employment. The theorem applies to trade among rather different countries – for example, industrialized versus developing countries – and predicts that relative rewards move in opposite directions as a consequence of trade. Traditional theory is less useful for predicting the distributional effects of trade among similar countries. This is a potentially important question since industrialized countries trade more with other industrialized countries than with developing countries. The predictions of traditional theory also124WORLD TRADE REPORT 2008appear to be in conflict with the evidence from firm-level data indicating that companies differ significantly within sectors, that only a subset of companies within a given sector exports and that those companies tend to pay higher wages than non-exporting companies (Bernard and Jensen, 1999). More recent contributions to the economic literature have analyzed how trade among similar countries, i.e. among industrialized countries, may affect factor prices. Matsuyama (2007) argues that the act of engaging in international trade may require the services of skilled labour, meaning labour with expertise in areas such as international business, language skills and maritime insurance. As a result, increases in trade can lead to a worldwide increase in the relative price of skilled labour. Epifani and Gancia (2006) argue instead that trade can benefit skilled workers because they can better take advantage of larger markets. They show that skilled workers, in any country, tend to constitute a minority of the labour force and tend to be employed in sectors with high plant-level fixed costs that produce highly differentiated goods that are gross substitutes for less skill-intensive products. In such a situation, trade will lead to a rise of the relative output of sectors characterized by economies of scale, i.e. the skill-intensive sectors. As a result, the relative demand for skilled workers goes up.Another set of models, in which fixed costs also play a role, allow for differences between firms and a so-called continuous distribution of skills among workers (Manasse and Turrini, 2001; Yeaple, 2005). In these models there is no clear line of separation between “high skill” or “low skill” workers, but rather a large variety of workers with different skill levels. In both models, the highest-skilled workers will end up by