Rethinking the Exchange Rate Impact on Trade in a World with Global Value Chains

ABSTRACT

Global value chains (GVCs) are a prominent feature of global production and trading systems. Using the OECD-WTO database on trade in value added, this paper examines the exchange rate elasticities of GVC-related exports and imports and compares them with elasticities for trade in traditonal goods. We find that a real depreciation raises both the foreign and domestic value-added content of GVC-related exports. The size of these elasticities is found to be smaller when the import content of GVC exports is larger. Among the key policy implications of these results is that exchange rate changes by small contributors of value added have little effect on their own production or the production of their supply chain partners. On the other hand, large contributors to the value-added of the final product create spillovers to their smaller supply chain partners, obviating traditional beggar-thy-neighbor concerns.

KEYWORDS:

• Global value chains
• exchange rate elasticity
• trade

Acknowledgements

We would like to thank the editor, in addition to Tam Bayoumi, Daniel Leigh, participants at the Center on Asia and Globalization (CAG) conference on finance, trade and investment in Singapore as well as several IMF seminars in Washington, DC, for helpful comments. Any remaining errors and omissions are ours.

Notes

1 Because intermediate products traverse international borders multiple times, this leads to the well-documented result that GVCs tend to swell the amount of international trade.

2 The FVA share in the total VA of exports is also relevant to the concept of effective protection, whereby a tariff is levied on the entire product, but the benefits accrue only to the DVA of the country that imposes the tariff. Thus, the effective protection rate (EPR) exceeds the nominal tariff rate whenever the FVA share is larger than zero, and the EPR escalates as the FVA share rises (Corden, 1966). When production is fragmented across countries, the concept of cumulative effective protection, which takes account of tariffs imposed along the entire production chain, is relevant (Rouzet & Miroudot, 2013).

3 Aggregation bias that arises from using gross exports and gross imports may help account for the wide variation in real exchange rate elasticities obtained in empirical studies using different country samples and time periods, as documented in Tokarick (2010), and Auboin and Ruta (2011).

4 However, over the long run, persistent country-specific REER changes may push out more expensive suppliers from the GVC in favor of lower-cost countries.

5 Note that (1–S_i) would express this share in terms of country i’s own DVA contribution. The FVA formulation is easier to work with in the empirical estimations but, being a simple linear transformation, generates equivalent empirical estimates.

6 The FVA share in a country's gross GVC exports is a proxy for the share of other countries’ VA contributions in the total value added of the final product. This proxy is used because the OECD-TiVA data do not report trade along individual supply chains. Using instead the FVA share in a country's gross GVC exports tends to understate other countries’ VA contribution to the final product for countries upstream in the production chain. On the other hand, exclusion of the final stage of GVC products from GVC trade data because it is not subsequently re-exported for further processing (and which TiVA records under other trade) tends to overstate other countries’ VA contribution to the final product in the case of countries downstream in the production chain. It is assumed that these measurement issues neutralize one another, so that the FVA share is a good proxy for all other countries’ VA contribution to the final product. Other empirical studies of the role of imported intermediates are subject to this same data limitation.

7 See Fally (2011).

8 As a robustness check, we also use the DVA REER series constructed by Bems and Johnson (2012). This series differs from the one defined above mainly in that it uses the GDP deflator rather than CPI to derive the bilateral real exchange rates. The Bems and Johnson VA REER series is directly obtained from the following website: http://dl.dropbox.com/u/7512224/webdata4VAREERpaper_WIOD.zip.

9 Moreover, and as expected, the FVA contribution in gross GVC exports is found to not significantly affect the response of non-GVC trade to REER changes. Coefficient values and their significance are little affected by omitting this variable from regressions 3 and 4.

10 Also, the GVC-related export and import elasticities are fairly similar in magnitude, suggesting limited substitutability in production.

11 This is because: (i) the estimated export elasticities corresponding to FVA shares of 50–80% lie within the 90% confidence interval spanning zero, suggesting that the elasticities are not statistically distinguishable from zero. For import elasticities, the corresponding FVA share range is 38–62%, but above this range, a positive elasticity cannot be rejected. And (ii) the maximum FVA contribution to GVC–related gross exports for any country in the dataset is less than 80%, with the average FVA share around 50–60%. Thus, most countries tend to operate in the range where GVC elasticities are about zero.