Heterogeneous Effects of Economic Integration Agreements

By Scott L. Baier (Clemson University), Jeffrey H. Bergstrand (University of Notre Dame), and Matthew W. Clance (University of Pretoria)

It is now widely accepted that economic integration agreements (EIAs) and other trade-policy liberalizations contribute to nations’ economic growth and development. EIAs have proliferated among North-North (N-N), North-South (N-S), and South-South (S-S) country-pairs. While such agreements inevitably alter distributions of income within countries, for the most part EIAs are believed to raise economic welfare. A major recent advance in the international trade literature — in the wake of and building upon theoretical developments associated with firm heterogeneity and export fixed costs — is the development of the “new quantitative trade models.”[1] These models provide calculations of general equilibrium trade and welfare effects of trade liberalizations using exogenous (variable-cost) “trade elasticities” estimated from structural gravity equations combined with aggregate bilateral trade data. Moreover, estimates of welfare effects of EIAs can be computed once one has partial treatment effects from a properly specified gravity equation with EIA dummy variables and an exogenous trade-elasticity (parameter) value.[2]

However, an important unresolved and hardly explored issue is whether — and by what factors — trade elasticities with respect to trade-policy changes vary across time and space, that is, are sensitive to “particular settings”; this is particularly important in contrasting trade elasticities for N-N, N-S, and S-S EIAs. In a recent study, we address three particular questions related to this issue.[3] First, how are trade elasticities — fixed-cost-trade-policy trade elasticities as well as variable-cost ones — theoretically related to levels of fixed and variable trade-cost variables, which vary dramatically between N-N, N-S, and S-S pairs? Second, is there convincing empirical evidence supporting these theoretical interactions? Third, how important quantitatively is the heterogeneity in partial equilibrium trade impacts in determining the general equilibrium welfare impacts of trade-policy liberalizations?

To address these questions, we provided three contributions. First, we extended a standard Melitz model of trade to show theoretically how extensive-margin, intensive-margin, and trade elasticities are endogenous to the levels of theoretical bilateral variable and fixed, policy and non-policy trade costs — even with CES preferences and with an untruncated Pareto productivity distribution.[4] Among several theoretical results, we note three. While the intensive-margin elasticity of tariff rates is sensitive only to the relative levels of variable policy and non-policy trade costs, the extensive-margin elasticity is sensitive also to the relative importance of bilateral endogenous export fixed costs (via network effects) in total bilateral export fixed costs. While the intensive-margin elasticity of policy export fixed costs is zero, the extensive-margin elasticity of policy export fixed costs is sensitive to the relative importance of bilateral endogenous export fixed costs in total bilateral export fixed costs as well as the relative importance of exogenous policy export fixed costs to exogenous non-policy export fixed costs. The theoretical comparative statics provide numerous predictions about how proxies for (time-invariant exogenous) natural variable trade costs and policy and non-policy export fixed costs influence the expected partial effects of EIAs on intensive margins, extensive margins, and bilateral trade.

Second, we evaluated empirically our theoretical hypotheses. We provided empirical evidence confirming our theory and demonstrated the heterogeneity of EIAs’ trade effects depending upon country-pairs’ geographic, cultural, institutional, and development characteristics. Extending earlier work, this is the first study to show evidence that extensive-margin, intensive-margin, and trade-flow EIA elasticities are indeed sensitive to levels of (observable) bilateral variable and fixed, policy and non-policy trade costs in a manner consistent with theoretical comparative statics.[5] Trade elasticities with respect to trade-policy changes do vary across “particular settings.” Geographic, cultural, institutional, and development country-pair characteristics all significantly influence the extensive margin elasticity, whereas primarily geographic variables (distance and adjacency) influence the intensive margin elasticity, consistent with our theory.

Finally, our framework allows us to put to ex ante use the partial effects of EIAs. By explaining the heterogeneity of EIAs’ effects according to theoretically-motivated factors, one can use the heterogeneous partial (treatment) effects for ex ante predictions and we demonstrate empirically that the partial effect of an EIA tends to be much larger for a pair of developing economies. Moreover, in the context of the new quantitative trade models, we demonstrate empirically using two approaches how sensitive quantitatively general equilibrium welfare effects of EIA liberalizations are to the bilaterally heterogeneous (partial) trade elasticities. In one approach, we calculate the general equilibrium welfare effects for importers of 1,358 bilateral EIA liberalizations among N-N, N-S, and S-S country-pairs. Consistent with theory, we show that 98-99 percent of the variation in these 1,358 welfare changes can be explained by the variation in two statistics: the estimated pair-specific bilateral EIA partial (treatment) effect and the share of the importer’s national expenditures on exports from the EIA partner. In the other approach, we show that the probability of two countries having an EIA — which in the context of a theoretical model is related to the net welfare gain from such EIA — is highly correlated with the heterogeneous EIA coefficients and the trade shares.[6] Our results suggest that a 10 percent lower average per capita income of a country-pair is associated with a 60 percent higher partial (trade) effect of an EIA. We close our study by demonstrating the relevance of our findings to the current trade-policy debate, analyzing the partial effect of “Brexit” from the European Union (EU), as well the potential effects of two EU members that are developing economies exiting the EU.

References

Arkolakis, C., A. Costinot, A. Rodriguez-Clare, (2012); “New Trade Models, Same Old Gains?American Economic Review, 102 (1), 94-130.

Baier, S., and J. Bergstrand, (2004); “Economic Determinants of Free Trade Agreements.Journal of International Economics, 64 (1), 29-63.

Baier, S., J. Bergstrand, and M. Clance, (2018); “Heterogeneous Effects of Economic Integration Agreements.Journal of Development Economics, 135, 587-608.

Baier, S., J. Bergstrand, and M. Feng, (2014); “Economic Integration Agreements and the Margins of International Trade.Journal of International Economics, 93 (2), 339-350.

Costinot, A., and A. Rodriguez-Clare, (2014); “Trade Theory with Numbers.” In Handbook of International Economics, Volume 4, edited by G. Gopinath, E. Helpman, and K. Rogoff. Elsevier Science: Amsterdam.

Head, K., and T. Mayer, (2014); “Gravity Equations: Workhorse, Toolkit, and Cookbook.” In Handbook of International Economics, Volume 4, edited by G. Gopinath, E. Helpman, and K. Rogoff. Elsevier Science: Amsterdam.

Melitz, M., and S. Redding, (2015); “New Trade Models, New Welfare Implications,” American Economic Review, 105 (3), 1105-1146.

Novy, D., (2013); “International Trade without CES: Estimating Translog Gravity,” Journal of International Economics, 89 (2), 271-282.

Endnotes

[1] See Arkolakis, Costinot, and Rodriguez-Clare (2012), Head and Mayer (2014), and Costinot and Rodriguez-Clare (2014).

[2] See Head and Mayer (2014).

[3] See Baier, Bergstrand, and Clance (2018).

[4] Novy (2013) generated endogenous trade elasticities by assuming transcendental logarithmic preferences and Melitz and Redding (2015) generated endogenous trade elasticities by assuming a truncated Pareto productivity distribution.

[5] See Baier, Bergstrand, and Feng (2014) and Head and Mayer (2014) for earlier work.

[6] See Baier and Bergstrand (2004) for underpinnings on this methodology.