Vertical specialisation and gains from trade

Abstract

This refers to the sequential manner in which most goods and services are produced, where a given product uses intermediate products as inputs, which are themselves produced with other intermediate inputs, and so forth. 1 The nature of this production structure has important implications for productivity and the gains from international trade. As has long been acknowledged by trade economists (e.g. Krugman & Venables, 1995), trade in intermediate inputs provides a magnification effect that raises welfare significantly more than trade in final goods alone. This effect is typically captured by assuming a 'roundabout' production structure, which delivers a sizable welfare impact (see Costinot & Rodríguez-Clare, 2014). 2 Although the roundabout assumption is convenient in a number of ways, it abstracts from an additional channel whereby multi-stage production might facilitate gains from trade (GFT), namely through specialisation across production stages or 'vertical specialisation' (VS). Consider, for example, smartphone production. Some popular models are assembled as final products in China, using intermediate inputs like the LCD screen, gyroscope and flash memory sourced from countries like Japan, Switzerland and Korea. In this example and others, the final stage of production (e.g. assembly) is technologically distinct from upstream production stages (e.g. gyroscope production). Vertical specialisation also features prominently in maquiladora trade for Latin American countries and export-processing zones for China and other Asian countries.1 Multi-stage production is also commonly referred to as sequential production.2 Roundabout production assumes that any given good can be used either for final consumption or as an input into production.

On the Geography of Global Value Chains

“…On the other hand, our model nests “state‐of‐the‐art” Ricardian models featuring roundabout production. Specifically, letting , the gains from trade are given exactly by the formula in () but depend on the aggregate domestic expenditure on industry k final goods given by and the aggregate domestic expenditure on industry k intermediate inputs by each industry given by , as in the recent work by Alexander (). Further restricting the model to for all delivers exactly the influential roundabout model of Caliendo and Parro (), which includes multiple industries and input‐output linkages but which features a single aggregate domestic expenditure share for each industry.…”

“…For computational reasons, we will mostly focus on an application in which the various industries in a WIOT are collapsed into a single sector. Nevertheless, following the lead of Alexander (), we will allow for asymmetries in the technology to produce final goods and inputs. It is undeniable that an estimated multi‐industry version of our model could deliver estimates with significantly different implications for the counterfactuals studied in Section , so readers are entitled to treat the results in the next two sections as a proof of concept, rather than as a definitive quantitative exercise.…”

“…Second, production depends on the Cobb–Douglas input expenditure shares and , which are allowed to vary across the production of intermediate‐input and final‐good varieties, and which are stage‐specific but common across countries. Third, labor productivity depends on the country‐specific state‐of‐technology levels and , which are also allowed to differ across intermediate input and final‐good varieties (as in Alexander ()). The parameter θ governing the strength of comparative advantage is instead assumed common across stages and countries.…”

“…We find that the gains from trade (i.e., the income losses from reverting to autarky) emanating from our model are, on average, 60% larger than those obtained from a version of our model without multi‐stage production. This variant of our model is a generalization of the Eaton and Kortum () model—akin to the work of Alexander ()—estimated to match all the entries of the WIOT. Similarly, we find that the real income losses from a hypothetical U.S.‐China trade war are, for those two countries, almost 50% higher with sequential production than without it, though such a war may well benefit third countries.…”

“…On the quantitative side, and as mentioned above, our work builds on and expands on previous work by Yi (, ), Fally and Hillberry (), and Johnson and Moxnes (). Other authors, including Caliendo and Parro (), Alexander (), Antràs and Chor (), and Baqaee and Farhi (), have developed quantitative frameworks with Input–Output linkages across countries, but in models with a roundabout production structure without an explicit sequentiality of production. The connection between our framework and these previous contributions has been further explored in de Gortari (), who blended several strands of this literature by generalizing the formulas on value‐added content and downstreamness within the context of a multi‐sector Armington model with sequential production.…”