Journal Issue

Q&A: Seven Questions on the Implications of Global Supply Chains for Real Effective Exchange Rates

International Monetary Fund. Research Dept.
Published Date:
March 2013
  • ShareShare
Show Summary Details

Rudolfs Bems

The real effective exchange rate (REER)—the most commonly used measure of competitiveness—requires a conceptual update to reflect the rise of global supply chains. This article summarizes recent research that develops a value-added REER, measuring competitiveness for value-added exports.

Question 1: What do we know about global supply chains?

The term “global supply chains” broadly refers to the shift of production to a multi-stage arrangement that can stretch across countries. In recent decades we have seen a proliferation in global supply chains, especially after 1990. A common way to measure global supply chains is to estimate re-exported imports, which have gradually increased to account for around one-fourth of global gross exports. With these networks, one can split gross exports of a country into two parts: (i) re-exported imports and (ii) exports of value added. Proliferation of global supply chains leads to an increasing “round-tripping” of goods across borders, as countries import intermediate inputs, and export new products after additional value is added to the input. As a result, the share of value-added exports in gross exports falls, while the share of re-exported imports rises.

Question 2: What are real effective exchange rates (REER) measuring?

The question that motivates the construction of REER indices is “How is demand for a country’s output affected by changes in prices of output relative to competitors?” The framework that underlies REERs (see Armington, 1969; McGuirk, 1987) postulates that the answer depends on three factors. First, the degree of openness—if a country is closed then changes in prices relative to competitors do not affect demand for output. Second, an elasticity of substitution, which captures the sensitivity of demand to changes in relative prices. Finally, it depends on the REER, which summarizes relative price developments weighted by trade patterns.

For our purpose, the key simplifying assumption of the conventional REER index is that countries produce goods entirely at home and compete with each other in various markets. Because there are no intermediate inputs, there is no distinction between a price of a country’s value-added and gross output. The two are identical. Similarly, there is no distinction between exports in value-added and gross terms. The two are, again, assumed to be identical.

Question 3: Does the rise of global supply chains warrant a rethinking of REERs?

Accounting for global supply chains can alter our interpretation of the state of the international economy. Estimates show that global supply chains can change bilateral trade flows. For example, in value-added terms, China’s trade surplus with respect to the United States is roughly 25 percent to 40 percent smaller, because headline gross-trade based surplus includes value added from other countries (Johnson and Noguera, 2012a). Also, in value-added terms the United States, not Germany, is France’s largest trade partner. A more subtle point is that sectoral composition of exports can change: in value-added terms one half of U.S. exports are services. In gross terms services account for one third of exports.

Turning more specifically to the framework that motivates REERs, all three previously mentioned ingredients can in principle be affected by the rise of global supply chains. Most importantly, in the presence of global supply chains (and intermediate inputs more generally) there is a multitude of output prices and trade weights because quantities and prices in value-added terms and gross terms are distinct. Consequently, a question arises as to what are the most appropriate prices, weights, and formula for the REER.

Question 4: How to modify the REER that account for global supply chains?

In a recent paper, Bems and Johnson (2012) account for global supply changes by generalizing the framework that motivates conventional REERs. In essence, they introduce demand for intermediate inputs into the framework, in addition to the final demand. They re-derive the REER index with the same underlying motive, although the more general framework allows for a more pointed question: “How is demand for a country’s value added affected by changes in prices of value added relative to competitors?” The answer is a new value-added index: Value-Added REER or VAREER.

There are two key advantages to using this particular generalization of the conventional framework. First, it collapses to the Armington demand system that motivates conventional REERs, when intermediate inputs are assigned zero weights in production. Second, the new framework has the same make/use structure as input-output tables, so all the model parameters (except elasticities) can be easily mapped into data.

Question 5: What are the findings conceptually and empirically?

Conceptually, Bems and Johnson (2012) find that the gist of the REER remains valid—it summarizes relative price developments weighted by trade patterns. The key new insight concerns the types of data required to build the index. To measure relative prices and trade links, the VAREER uses different data from the conventional REER:

  • GDP deflators to measure changes in relative prices. Intuitively, they are the most direct summary measure for factor (capital and labor) costs.

  • Bilateral trade in value added to construct country trade weights. What matters for a country’s competitiveness is demand for its value added, rather than for its gross output. Intuitively, global supply chains redefine a country’s competitors. There is less competition between countries that share a supply chain because each can affect the price of the final product and hence the competitiveness of any other country in the supply chain. VAREER reflects these considerations by assigning smaller weights to countries that share a supply chain.

Empirically, Bems and Johnson (2012) construct the VAREER index for 42 countries over the 1970–2009 period and compare it to the conventional CPI-based real effective exchange rate. As one would expect, the two indices move in the same direction. Year-to-year differences are small, but over time sizable deviations can accumulate in some cases. For example, while the conventional REER for China exhibits no trend over the 1990–2009 period, VAREER shows a 20 percent appreciation over the last decade. There are also significant differences between VAREER and CPI-based REER for Eurozone countries, in the post-1995 period. Among these countries, the VAREER moves more strongly in directions that are consistent with the widening of current account imbalances prior to the onset of the crisis.

The authors decompose the differences between VAREER and conventional REER and show that the bulk of deviations between the two indices stems from the shift in prices from CPI to GDP deflators. Although there can be sizable changes in trade weights, these do not correlate systematically with changes in relative prices and, hence, at least historically, have not had a significant impact on differences between the two price indices.

Question 6: Is the new index feasible to build for a large set of countries?

Yes, it is relatively easy to implement the new VAREER index. Because the gist of the REER formula does not change, one simply needs to substitute weights and prices. The price data—GDP deflators—are available at quarterly frequencies for a large set of countries. The weights—based on bilateral trade in value-added—have a more limited coverage, but their availability has increased rapidly in recent years. Several datasets are now available, including from Timmer (2012) and Johnson and Noguera (2012). Furthermore, because weights contribute little to the deviations between the two indices, weights based in bilateral gross trade flows provide a good proxy for the VAREER index. Using this shortcut, the VAREER can be computed for all countries that report GDP deflators and bilateral trade flows.

Question 7: How synchronized are national recessions around episodes of global recessions? And how do national cycles interact with the global cycle during these periods?

Work on the effect of global supply chains on price competitiveness needs to revisit the other two factors that have been, justifiably or not, neglected by the conventional framework: the degree of openness and elasticity of substitution (see Question 2 above). First, global supply chains reinterpret the traditional measures of openness. Empirically, countries are more open in value-added terms (i.e., value-added trade as a share of GDP) than in gross terms (i.e., gross trade as a share of gross output). Furthermore, for the majority of economies openness has been increasing over time. Changes in openness, in turn, lead to reinterpretation of the macroeconomic impact of a given change in a price index. The impact is larger when the economy is more open. Results in Bems and Johnson (2012) suggest that macroeconomic implications of this channel can be significant.

Second, accounting for global supply chains and intermediate inputs more generally requires a reinterpretation of price elasticities of demand. Available empirical estimates of relevant macro and micro elasticities are all based on expenditure data and, therefore, estimate the effect of changes in the relative price of final goods on demand for goods. What is missing are comparable elasticity estimates for value added. It is conceivable that some of the deviations between the VAREER and the conventional REER index stem from differences in underlying elasticities rather than price indices. This is an empirical question that requires an answer (see Bems, 2012, for further discussion).

Finally, in the context of the new VAREER1 index, it would be beneficial to gain a better understanding of deviations between prices of a country’s value added and gross output. In particular, to what extent are such deviations driven by domestic intermediate inputs as opposed to imported intermediate inputs. If the source of deviations is domestic, then it is domestic intermediate inputs, such as


    ArmingtonPaul S.1969. “A Theory of Demand for Products Distinguished by Place of Production,IMF Staff Papers Vol. 16 No. 1 pp. 159178.

    BayoumiTamimSarmaJayanthi and JaewooLee.2006. “New Rates from New Weights,IMF Staff Papers Vol. 53 No. 2 pp. 272305.

    BemsRudolfs.2012. “Intermediate Inputs, External Rebalancing, and Relative Price Adjustment,” (unpublished; Washington: International Monetary Fund).

    BemsRudolfs and RobertJohnson2012Value-Added Exchange Rates,NBER Working Paper No. 18498 (Cambridge, Massachusetts: National Bureau of Economic Research).

    JohnsonRobert C. and GuillermoNoguera.2012a. “Accounting for Intermediates: Production Sharing and Trade in Value Added,Journal of International Economics Vol. 82 No. 2 pp. 224236.

    JohnsonRobert C. and GuillermoNoguera.2012b. “Fragmentation and Trade in Value Added Over Four Decades.NBER Working Paper No. 18186 (Cambridge, Massachusetts: National Bureau of Economic Research).

    McGuirkAnne.1987. “Measuring Price Competitiveness for Industrial Country Trade in Manufactures.IMF Working Paper 87/34 (Washington: International Monetary Fund).

    TimmerM.P. (2012ed) “The World Input-Output Database (WIOD): Contents, Sources and Methods,WIOD Working Paper 10.

The full dataset with VAREERs is available from the authors’ homepages: or

Other Resources Citing This Publication