Extended Cumulation, Convergence of Rules of Origin, and Regional Value Chains
TO DEVELOP REGIONAL SUPPLY CHAINS, COUNTRIES
NOT ONLY NEED ACCESS TO EACH OTHER’S MARKET, BUT ALSO SHARED ACCESS TO THIRD MARKETS
agreement with the importing partner (so-called nonoriginating inputs), even if Costa Rica has a PTA with the country where the input comes from.
The story, though, can only be fully understood with the introduction of another country, say Nicaragua. As mentioned above, Nicaragua has an agreement with Costa Rica that allows for extended cumulation with Mexico. This situation im- plies the following: other things being equal, when Costa Rica imports inputs to be used in goods for domestic final consumption, it has more incentives to trade with Nicaragua than with Argentina because it has a trade agreement with the former but not with the latter. This is similar to the situation between Peru and Argentina. Again, this is the bilateral trade impact that is typically captured in the literature. But the second effect now implies that when it comes to importing inputs that will subsequently be used in exports to a country like Mexico, Costa Rica has an additional incentive to import from Nicaragua at the expense of Peru (or Argentina) because it can cumulate materials from Nicaragua into ex- ports to Mexico without losing its preferences there. The extended cumulation with Mexico gives an additional value chain incentive for sourcing inputs from Nicaragua.
The gravity model is used to estimate these two effects arising from sharing (a) a trade agreement and (b) RoOs that allow for extended cumulation. In the context of the road map discussed earlier, this can be seen as an initial, par- tial equilibrium, approximation of the RVC gains of a more ambitious move toward a regionwide FTA. More specifically, the estimated impact from the gravity model will be the sum of two impacts. First, an increase in value-add- ed from country j to country i that takes place if they sign a trade agreement. Second, an increase in the value-added from j to i that arises if i is allowed to accumulate origin from j because it is now part of its trade network and thus they all share the same set of rules of origin, making cumulation across them possible.
Before presenting the results, it is important to note that the rules of origin in the gravity equation were modeled based on a very simplified assumption, which is explained below. There are at least two reasons for this. First, the dataset on val- ue-added is available at a very aggregate level, which makes it difficult to match the data on rules of origin, which are normally written at very disaggregated product levels. Second, to obtain robust estimates, the model was run using data not only for LAC but for the whole world, across three different time periods (see box 5.3). Obtaining detailed data on RoOs for all world PTAs at different time periods would be a monumental task, one that goes beyond the scope of
this study. Therefore, a very simple proxy was used to capture the potential dis- incentives for country i to import from j because the rules of origin in PTAs with third partners prevent i from doing so: the number of agreements that i has signed with third countries and which j is not party to.84 This simplification
comes at some cost. Because this is a crude proxy of the extent and the level of RoO restrictiveness, the results from the estimations should be viewed with some caution. The attention should focus more on the direction of the effects and the relative magnitudes rather than on the absolute size of the impacts.85
Figure 5.4 shows the results for this exercise. In all the cases, the variable of in- terest is the change in the value-added from country j (the exporting country) that is eventually used by country i (the importing country) in its own exports. The first column shows the average across all source countries and destinations
FIGURE 5.4 CHANGE iN ExPORTS OF vALUE-ADDED FROM LAC
COUNTRiES THAT ARE SUBSEQUENTLY USED AS iNPUTS iN OTHER LAC COUNTRiES’ ExPORTS
0 2 6 4 12 10 8 Per cent ag e
All countries Central America
& Mexico Andean Southern Cone PTA RoOs
Source: IDB estimates.
84 An alternative proxy was also used: the total market potential that country i has access to through the
agreements that it has signed with third countries and in which country j does not participate. See box 5.3.
85 In this type of analysis there are always two econometric challenges: i) the explanatory variable does not
pick up the effects of other factors or policies (say, export processing zones) and ii) the causality goes from the explanatory variable to the dependent variable and not the other way around. The estimation presented in Box 5.3 addresses the first challenge by including a full set of fixed effects that control, among other things, for time-variant importer and exporter factors. The estimation of the model using lagged values of the explanatory variable seeks to address the second challenge.
FIGURE 5.5 CHANGE iN ExPORTS OF vALUE-ADDED FROM CENTRAL AMERiCA AND MExiCO THAT ARE SUBSEQUENTLY USED AS iNPUTS iN SELECTED REGiONS’ ExPORTS 0 2 10 8 6 4 16 14 12 Per cent ag e
Central America & Mexico Andean Southern Cone PTA RoOs
Source:í îïðñ òóôõ òðñ ö
in LAC. PTA measures the increase of value-added that arises from sharing a trade agreement, and RoO measures the increase of value-added that arises from sharing the same set of RoOs. The results, shown in the first column, point to an average increase in value-added of around 9%. That is, for the average LAC country, exports of intermediate goods that are subsequently used as inputs in other LAC countries’ exports increase by 9%, with equal contributions from the PTA and RoO effects.
The rest of the columns show the results when the sources of value-added (that is, the exporting countries) are grouped by subregion. The second column pres- ents the average for Central America and Mexico when they export interme- diate goods that are used as inputs in the exports of other LAC countries. The third and fourth columns do the same for the Andean and the Southern Cone countries, respectively.86 The results indicate the existence of some heterogene-
ity across subregions, with the countries in the Central America/Mexico region exhibiting slightly larger average effects.
86 Central America/Mexico includes: Costa Rica, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, and
Panama. The Andean countries includes Bolivia, Colombia, Ecuador, Peru and Venezuela. The Southern Cone includes Argentina, Brazil, Chile, Paraguay, and Uruguay.
CONVERGENCE OF RoOs IN THE CONTEXT OF A LAC-FTA IS LIKELY TO INCREASE THE USE OF REGIONAL INPUTS IN LAC EXPORTS BY 9%
FIGURE 5.6 CHANGE iN ExPORTS OF vALUE-ADDED FROM ANDEAN COUNTRiES THAT ARE SUBSEQUENTLY USED AS iNPUTS iN SELECTED REGiONS’ ExPORTS 0 2 10 8 6 4 12 Per cent ag e
Central America & Mexico Andean Southern Cone PTA RoOs
Source: IDB estimates.
FIGURE 5.7 CHANGE iN ExPORTS OF vALUE-ADDED FROM
SOUTHERN CONE COUNTRiES THAT ARE SUBSEQUENTLY USED AS iNPUTS iN SELECTED REGiONS’ ExPORTS 0 2 10 8 6 4 12 Per cent ag e
Central America & Mexico Andean Southern Cone PTA RoOs
Figures 5.5 to 5.7 present more disaggregated results by exporting and import- ing regions, and thus provide more insight on what is going on behind the av- erages. For instance, figure 5.5 shows the results when the countries of Central America and Mexico export intermediate goods that are used as inputs in each of the other subregions’ exports. Note, for example, that the potential increase in exports of value-added to countries in the same region is very modest. This is because most of these countries already have trade agreements and rules of origin that allow cumulation among them; therefore, a regionwide FTA does not add much to what they already have. Note, however, that this is not the case with exports of value-added to the Andean or Southern Cone countries. Since there are currently relatively fewer trade agreements between Central America/ Mexico and countries in these other regions, the proposed LAC-FTA has a much larger impact on the potential formation of supply chains across these re- gions. Figures 5.6 and 5.7 show a similar situation for the Andean and Southern Cone countries.
Overall, the results suggest the move toward convergence and, eventually, a re- gionwide FTA might have significative and positive impacts on the formation of regional supply chains. The largest impacts appear to be in the formation of supply chains across subregions, a result that arises from the fact that there is more room for integration across subregions than within them.