The last two decades have seen the proliferation of regional trade agreements in developing countries. In some cases preferential liberalisation has been accompanied by domestic unilateral trade reform. However, despite substantial progress towards trade reform, a major constraint for further liberalisation in developing economies is the perception of potential large adjustment costs. Uncertainty regarding adjustment costs and reallocation of resources implies more reluctance to sign new trade agreements and especially to unilaterally liberalise trade restrictions. As a result, the main objective of this review is to synthesise the empirical evidence regarding two main elements of adjustment to tariff reductions in developing countries: the impact on employment and the impact on tax revenue.
This review is a synthesis a quantitative studies. We include quantitative evaluations in developing countries that controls for other factors affecting employment and tax revenue. We also include Computable General Equilibrium (CGE) studies, and compare simulation results to the econometric evidence. In addition, given the large number of specifications within papers and the fact that different authors employ different reduced form equations, we combine several methodologies to synthesise the results. First, we use a two stage methodology. In the first stage, we compute a single estimate by study. Then we cluster studies according to the use of the same variable for employment/tax and trade policy. In this second stage, we perform meta-analysis between studies at each cluster. The synthesis results for all the clusters are then combined to derive conclusions and policy implications. The second methodology uses the same cluster methodology but using all the available observations by study and controlling for covariates that may impact the impact of trade policy on employment and tax revenue using meta-regression. Furthermore, we analyse what factors may impact the statistical significance of the coefficients.
In total from a sample of 3,487 searched studies, we include 109 studies, 67 econometric and 42 CGE. The ex post econometric evidence on revenue effects is available in 21 studies. The remaining econometric studies, 45 examine employment effects. There are no econometric studies that look at both of these variables. In contrast 8 CGE studies included in this review look at both employment and revenue impacts of trade liberalisation. The rest of the CGE studies either exclusively look at employment effects, 22 or at revenue effects, 11.
The results from the synthesis indicate that:
- Studies that use tariffs to measure trade policy find an ambiguous effect on employment. When we select studies based on using tariffs as trade policy variable and also controlling for potential endogeneity, we find that in general trade reductions tend to be correlated with reductions in employment. However, studies differ in the size of this effect.
- Studies that use trade flows to proxy trade liberalisation or use dummies to capture trade reform episodes (around 67% of specifications) tend to find positive effects on employment, on tax revenue and on labour reallocation. This result is quite robust, but it should be considered of lower quality since trade flows are a poor proxy of tariff reductions in some cases, and, therefore causality is unclear.
- Regarding employment reallocation we find that most studies show significant reallocation effects, especially regarding increased employment in export sectors following liberalisation. Regarding import competing sectors the evidence is mixed. The body of evidence includes studies that find employment destruction, studies that find negligible effects and studies that find employment growth.
- Ex ante CGE simulation studies that allow for trade reform impacts on the level of aggregate employment commonly predict a moderately positive net job creation effect (36 of 36 included studies). On average, a 1% increase in the volume of trade due to trade reform raises aggregate employment in the reforming country by 0.34% (31 studies).
- Ex ante CGE simulation studies with fixed aggregate employment commonly predict job destruction in import-competing and non-traded-goods sectors and job creation in export-oriented sectors after a trade reform. For each percent of additional trade volume generated by a trade liberalisation scheme, 0.12 percent of the labour force relocates to a new job in a different sector within the liberalising country (8 studies).
- The majority of CGE simulation studies that address the fiscal effects of trade reforms involving tariff reductions report negative total tax revenue impacts or the need for increases in other tax rates in order to compensate for lost tariff revenue (24 of 28 included studies).
- While CGE studies allow a cleaner isolation of trade policy impacts from other influences compared with econometric studies, the results need to be interpreted with caution. Regarding tax revenue, CGE results allow us to look at the isolated impact on tax revenue from reducing tariffs selectively. This is very informative since econometric evidence is likely to pick the impact of simultaneous interventions affecting tax revenue. However, the assumption of a frictionless reallocation of labour and other factors across sectors is an oversimplification not always supported by the econometric evidence.
In general our interpretation of the findings of the synthesis is that as expected these are country and trade policy specific. However in line with the preferred econometric evidence overall employment is likely to mildly decrease in the short-run following liberalisation, although this depends on the extent of the trade policy shock. These results are in contrast with the CGE findings, which by design incorporate projections of the medium-run economy-wide knock-on effects suggested by economic theory. In addition, the evidence points towards an expansion of employment in the export sector, but with an unclear prediction regarding the size and the sign of changes in import competing sectors.
Regarding tax revenue, tariff reductions are likely to reduce trade tax revenue in the short-run. In this regard, CGE simulations predict that only in the case of very high import demand and substitution elasticities, the generated increase in imports would be sufficient to compensate for the tariff cut. Thus, the likely outcome following liberalisation or implementation of a trade agreement is one of lower trade tax revenue, other things constant. The econometric evidence clearly points towards a positive impact of larger shares of trade to GDP on total tax revenue. Therefore, the impact in the medium-run could be positive, especially if complementary tax policies and increases in customs tax collection efficiency are implemented, although the effects of these last two channels cannot be properly quantified from the reviewed studies.
EPPI-Centre, Social Science Research Unit, Institute of Education, University of London, London, UK, ISBN: 978-1-907345-12-8, 99 pp.