Labor clauses in trade agreements: Hidden protectionism?

We explore the impact of the introduction and design of labor clauses (LCs) in preferential trade agreements (PTAs) on bilateral trade flows over the period 1990–2014. While it is not a priori clear if the inclusion of LCs in PTAs will decrease or increase bilateral trade, we expect the direction of trade to matter, that is, we expect to observe the (negative or positive) impact of LCs in the South-North trade configuration. We also expect, in that configuration, stronger LCs to yield stronger (negative or positive) effects on bilateral trade flows. Using a novel dataset on the content of labor provisions in PTAs, we find in line with our first expectation that while the introduction of LCs has on average no impact on bilateral trade flows, it increases exports of low and middle-income countries with weaker labor standards in North–South trade agreements. Consistent with our second expectation, this positive impact is mostly driven by LCs with institutionalized cooperation provisions. In contrast, LCs with strong enforcement mechanisms do not have a statistically significant impact on exports of developing countries in North–South PTAs. The results are inconsistent with the ideas that LCs are set for protectionist reasons or have protectionist effects, casting doubt on the logic for the reluctance of many developing countries to include LCs in their trade agreements. Supplementary Information The online version contains supplementary material (Online Appendix with annexes 1-4) available at 10.1007/s11558-021-09423-3.


Annex 1. Control variables
The United Nations Statistical Division (COMTRADE database) provides the original trade data from customs declarations. CEPII uses COMTRADE to create BACI that reconciles differences in the declarations between exporters and importers. 1 This harmonization procedure enables to extend considerably the number of countries for which trade data are available, as compared to the original COMTRADE dataset, but only from 1995 onwards.
is the value of imports of country i from country j in year t (in thousands of US dollars).
The geodesic distance is calculated following the great circle formula, which uses latitudes and longitudes of the most important cities/agglomerations (in terms of population); see Mayer and Zignago (2011) for more details. Other gravity variables include dummy variables indicating whether the two countries are contiguous, share a common language, have had a common colonizer after 1945 or have ever had a colonial link. The common language dummy is set to one if a language is spoken by at least 9 percent of the population in both countries.
Trying to give a precise definition of a colonial relationship is obviously a difficult task.
Colonization is here a fairly general term that we use to describe a relationship between two countries, independently of their level of development, in which one has governed the other over a long period of time and contributed to the current state of its institutions. All the gravity variables are from CEPII's BACI database.
The measure of depth is a dummy variable based on an additive index that combines seven key provisions that can be included in PTAs (see Dür, Baccini and Elsig 2014). The first provision captures whether the agreement foresees that all tariffs (with limited exceptions) should be reduced to zero (that is, whether the aim is to create a full free trade area). The other six provisions capture cooperation that goes beyond tariff reductions, in areas such as services trade, investments, standards, public procurement, competition and intellectual property rights. For each of these areas, Dür, Baccini and Elsig (2014) code whether the agreement contains any substantive provisions. We use an updated variable of their measure that was kindly provided to us by Andreas Dür. The dummy we use equals unity if the depth index is larger or equal to its median, namely 3. As shown in table below, PTAs with LCs are significantly "deeper" than PTAs without LCs. "Depth_index" refers to the index (in the range of 0-7) of Dür, Baccini and Elsig (2014) and "depth_dummy" refers to our dummy.
In annex 4 we use as additional control variable the pre-agreement tariff. More precisely, the pre-agreement MFN tariff is the latest MFN tariff available for a country before the years before the implementation of the PTA (from t-1 to t-5). We extract from the World Integrated Trade Solution (WITS -UNCTAD TRAINS database) the average MFN tariff only for on manufacture sectors, weighted by bilateral manufacture trade flows.
In fine, we use a sample of 64,297 observations, including 5,605 country-pairs and 437 PTAs over 1995-2014 (PTAs are reported in annex 4 Table A4). Second, if an agreement is amended without any changes in its LC, this is not taken into account (only the first date is used to define the dummy); if an agreement is amended with a change in its LCs, this is taken into account. For instance, if a subset of countries signs a first agreement with no LC and, several years after, the same subset of countries sign a new agreement with LC, we consider that the first agreement is amended, the LC dummy is set to unity for all country pairs within this PTA.
Evolution in LCs over 1995-2014 within a subset of countries is important in our study as this is the way the LCs impact is identified in the gravity equation with dyad fixed effects.
Over the 437 PTAs considered in this study, only 9 reports action in their LCs, and only 6 over 1995-2014.