This volume comprises eight contributions on impact evaluation in trade and trade policy. It is a unique bundling of a series of studies dealing with this rather novel approach in empirical trade research.
The technique of impact evaluation originally was developed in medical science to evaluate the effect of a treatment on patients. In the meantime, the social sciences have also adopted this methodology primarily in fields such as health, education, and labor economics. One of the purposes of this current World Bank volume is to assess whether impact evaluation has a future as a methodology in trade-related matters. The balance seems to be a positive one. In an introductory chapter, Cadot et al. make a number of convincing arguments as to why this methodology offers potential to be used in other areas of economics, in particular in the evaluation of trade facilitation and export promotion policies. Impact evaluation can be used on firm-level data or at a more aggregate level such as on country-level data. The procedure followed when applying this technique is the same independent of the level of aggregation and consists of three steps. First, researchers have to determine who features in the group of the ‘treated’ firms. Second, a group of ‘untreated’ firms have to be selected. Third, an econometric technique is used to evaluate the outcome variable and to see whether it has evolved differently for the ‘treated’ versus the ‘untreated’.
The difficulty lies in the construction of a group of ‘untreated’ firms. Typically, the literature has used various methods of ‘matching’. A matching method is directed at identifying firms that are similar to the untreated firms, but did not receive the treatment. This ‘similarity’ may involve a number of different criteria and clearly depends on the type of research question that is undertaken. For example, to evaluate the impact of import protection on domestic firms' productivity, Konings and Vandenbussche (Reference Konings and Vandenbussche2008) compare the protected firms to a group of unprotected firms but with similar characteristics as the protected ones, such as the sector they operate in, the size of the firm, their export status, their productivity in the period before protection, their exposure to foreign competition, etc. Alternatively, some studies construct a random selection of ‘untreated’ firms as opposed to more complex econometric ‘matching’ techniques and then compare treated firms to the average untreated firm. In studies of a more experimental nature, one often assigns treatment randomly, which is a way to avoid selection effects. A true impact evaluation study not only wants to compare productivity of the protected firms over time, but at the same time compare it to the evolution of productivity of unprotected firms, which is what is called a ‘difference-in-difference’ approach. This ‘double difference’ effect is important for a correct interpretation of the facts. A ‘single difference’ approach would only evaluate the productivity increases of protected firms over time. Suppose, with a single difference approach, results show that firm-productivity rises after protection. But when we compare this rise in productivity to unprotected firms through double differencing, we would find that productivity rises far less for the protected firms than for the unprotected firms. This simple example shows that double differencing in any impact evaluation study is important to arrive at the right conclusions. In the example above, based on the single difference results, our conclusion may have been that protection is good for firm productivity, whereas based on the double differencing we would arrive at the opposite conclusion.
Most of the studies listed in this World Bank volume used a variation of the double differencing methodology described above. Christian Volpe uses the impact evaluation method on trade programs in Latin America and finds a positive impact. Especially small firms benefited, and trade support affected the extensive margin (the number of products exported) more than the intensive margin (exports of existing products). Gourdon et al. evaluate a World Bank-financed export promotion programme for Tunisia and find it to have a relatively large increase in beneficiaries of the program not just in terms of their exports but also in terms of employment. Atkin and Khandelwal describe a project to assist micro-entreprises in Egypt to enter into export markets. Their impact evaluation consists of comparing enterprises that benefit from assistance to a random sample of firms that did not and to compare their export performance. In Chapter 5, Sequeira discusses a transport infrastructure program in Mozambique and evaluates the rehabilitation of an old railway as a reduction in transport cost and how this will affect the performance of firms. The ‘treatment’ in this case is the distance between a firm and the new railway. Datt and Yang, in chapter 7, engage in a natural experiment in the Philippines. At some point there was a change in the law which made it harder for firms to avoid pre-shipment inspections. The law consisted of reducing the minimum value threshold under which shipments were exempted from inspections which previously had led firms to slice shipments to avoid duties. Datt and Yang evaluated duty collection before and after the law changed but did not find a significant increase of duty collection after the change which the authors explain by the fact that firms switched to another loophole in the Philipines' law which made avoiding duties possible. Cantens et al. in chapter 7 evaluate a recent reform in Cameroon. The study consists of comparing the performance of customs inspectors by giving them different performance contracts. But this study suffered from a small sample bias which made a reliable impact evaluation study impossible. For this reason, the study was reduced to simply evaluate the before and after performance of inspectors which is like the single difference approach explained above.
And finally there is also a chapter by Ferro et al. dealing with an aid program in services and their impact on downstream exports by manufacturing sectors in about 106 countries. The results are positive and show that especially the assistance to the energy and the banking sector has led to a positive impact on downstream manufacturing exports.
The outcome of these impact evaluation studies led the World Bank Chief Economist Justin Yifu Lin to conclude that there seems to be a role for policy to enhance trade competitiveness, especially when policy can address market failures. Cadot, Fernandes, Gourdon, and Mattoo have succeeded in bringing together academic work that aims to objectively evaluate the success or failure of various trade policy interventions. A valid concern that will need to be addressed in the future is no doubt whether the type of unilateral interventions studied in the chapters here ultimately also result in multi-lateral gains. In other words, whether export promotion by one country occurs at the expense of other countries' exports. While trade policy evaluation studies are very important to study policy effectiveness, the idea of ‘survival of the fittest’ in export markets should not become the ‘survival of the fattest’ at the expense of the fittest. By that I mean that government intervention should not result in favoring unproductive firms through a support system of subsidies in one country at the expense of productive firms in another country where firms receive little or no government support.
