Information about Asia and the Pacific Asia y el Pacífico
Journal Issue
Share
Article

Measuring Competitiveness

Author(s):
Tamim Bayoumi, Mika Saito, and Jarkko Turunen
Published Date:
May 2013
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details

I. Introduction

1. In the past few years, two alternatives to the Fund’s standard formula for calculating real effective exchange rates (REER) have been proposed.2 Both alternatives aim to address the implications of emerging global supply chains for assessing international price competitiveness. Bems and Johnson (2012) propose an alternative method based on trade in tasks and value-added trade data. Thorbecke (2011), as applied in Unteroberdoerster, Mohommad, and Vichyanond (2011), propose a so-called integrated effective exchange rate (IEER) that separately accounts for trade in intermediate inputs. Results from these alternative approaches can diverge from each other depending on the country. The main purpose of this paper is to propose a measure of competitiveness based on trade in goods as in the standard REER, but with the price of goods reflecting the cost structure of production with global supply chains. This is a new approach but shares the same intuition as was seen in IEER.

2. Bems and Johnson (2012) follow the original work by Armington (1969) and McGuirk (1987) closely, but make two important improvements. First, they switch from a world where goods are traded to a world where “tasks” are traded. With the growth of global supply chains, they argue that any task can be purchased either as intermediate inputs to production or as part of final goods consumed. They show that under certain conditions, the Fund’s standard REER formula holds but the weights and the price index need to reflect value added or tasks. They compute REER indices using weights reflecting value-added trade patterns and prices for value added. They compute alternative REER indices for 42 countries from 1970 onwards and find that differences between standard REERs and their measures are driven by differences in prices, not weights. For example, they find that the Chinese REER measured in value-added terms has appreciated more since the early 2000s than the standard REER would suggest.

3. Thorbecke (2011) argues that in the context of global supply chains, analyzing the impact of exchange rate changes on trade needs to account for changes in the prices of imported intermediate inputs that are embodied in final goods exports. Based on this work, Unteroberdoerster, Mohommad, and Vichyanond (2011) compute an IEER which accounts for changes in the exchange rates of suppliers vis-à-vis the final export market currencies for a number of East Asian economies. They find that taking value-added in intermediate inputs into account, the Chinese IEER has appreciated less than the standard REER since 2008.

4. We argue that because of production sharing through global supply chains, changes in relative prices of goods have become less sensitive to domestic factor price movements, especially for emerging market economies. As a result, production sharing has helped some emerging market economies, such as China, to maintain competitiveness when faced with increasing prices of domestic production factors. To demonstrate this point, we construct a new index, REER in Goods, with prices of those goods reflecting “global supply chain” production structure. That is, goods are no longer produced using domestic production factors only, but include production factors from abroad. More specifically, by expressing the price of goods as a function of price of production factors embedded in goods, we show that the Fund’s standard REER formula needs to be expressed in terms of prices of factors (proxied by the GDP deflator3) and needs an additional term. This additional term captures the role of outsourcing or foreign value added in easing appreciation pressure coming from, for example, a relative increase in the cost of domestic factors of production or an appreciation of the nominal exchange rate. We compare this new index with the standard REER index and an index that reflects trade in tasks (REER in Tasks).

5. We find that there are significant differences between the standard REER and REER in Goods. For example, reflecting a more pronounced increase in the relative prices of production factors (compared to the price of final consumption goods) for China, the REER in Goods suggests more appreciation over time than the standard REER. This result is similar to Bems and Johnson (2012). However, for many emerging market economies, including China, an appreciation owing to an increase in the price of domestic production factors is moderated by the use of foreign production factors embedded in goods. Our results therefore confirm that in a world with production sharing, a loss in competitiveness due to an increase in relative factor costs does not necessarily get translated into a loss in competitiveness of goods. This is the moderation effect of production sharing discussed in Thorbecke (2011) and Unteroberdoerster, Mohommad, and Vichyanond (2011).

II. Model

6. Original work by Armington (1969) and McGuirk (1987) is based on a model of consumer demand. For example, consumer in country j maximizes his or her utility subject to a budget constraint and chooses consumption demand of goods produced in countries 1 to n, C1j, … , Cnj. They specify a utility function with the CES functional form. Prices P1, … , Pn are therefore prices that the consumer faces; which justifies the use of CPI.4

7. Based on the Armington model, the Fund’s current formula (Bayoumi, Lee, and Jayanthi, 2005) in calculating the REER of country j is

where Rj and Rk are respective exchange rates and wjk is the trade weight that captures import competition (i.e., competition in country j), export competition (i.e., competition in trading partner country k) and the third market competition (i.e., competition between j and k in all other markets).

8. Bems and Johnson (2012) introduce global supply chains to this setting and also move away from a goods world to a value-added world. More specifically, they model intermediate input demand as well as consumption demand, by assuming a production function with a global supply chain structure (similar to the CES production function described below). For example, producer in country j minimizes his or her cost of production subject to gross production technology and chooses intermediate input demand of inputs produced in countries 1 to n, X1j, … , Xnj as well as factors of production, capital (Kj) and labor (Lj).

With resource constraints satisfied in both goods and factor markets, they solve for equilibrium level of gross output Xj. They then show that when the same functional form as in the original Armington is assumed to describe consumer preferences and production technology, the Fund’s standard REER formula remains the same but the weights need to be calculated to reflect value-added (or task) trade patterns, and Pj value-added prices (proxied by the GDP deflator).

9. Our main goal is to show how production sharing can help a country’s goods competitiveness when relative prices of goods do not rise as fast as the relative costs of domestic factors.5 We therefore derive the price equation for (gross) goods that captures the global supply chain structure and replace the consumer price within the original Armington structure.6 What simplifies greatly in deriving the REER formula with global supply chains is that the total cost function, production technology and the price equation, which can be derived by specifying technology, can all be expressed in terms of value added of all countries in the world. After all, goods must be a product of someone’s value added. For example, the two expressions of total cost below are equivalent, where the first expression can be read off of any input-output tables, while the second needs to be calculated as in Koopman and others (2010):

where qj is price of value added, Dj is the domestic production factor, and Mij is the foreign production factors (from country i ≠ j) used in country j. Equivalently, production function Xj = F(Lj, Kj, X1j, … , Xnj) can be expressed in terms of value added, Xj = G(Dj, M1j … , Mnj). The benefit of using these alternative expressions is the ability to express the price of goods in terms of the price of value added of all countries. This simplifies the REER formula greatly.

10. We find that the REER formula in the world with production sharing needs to be expressed in terms of the GDP deflator but also needs an additional term. The additional term is expressed as a weighted average of relative prices of foreign value added used in production (relative to that of trading partners). It captures the role of outsourcing or foreign value added in easing an appreciation pressure coming from, for example, a relative increase in the cost of domestic production factors or an appreciation of the nominal exchange rate. The IEER shares the same intuition. The rest of this section gives the derivation of the REER formula in more detail.

11. Suppose that a producer minimizes the following cost function for gross output Xj

Note that production technology takes the two-level constant elasticity of substitution (CES) functional form, which separates the production of domestic value-added and foreign value added used in domestic production. σj is the elasticity of substitution between domestic value-added and aggregate foreign value-added imports and σjM is the elasticity of substitution among foreign value-added imports from different countries. If σj = σjM, we are back to the standard Armington world (i.e., one-level CES). δj is the domestic value-added cost share in total cost, and φij is the foreign country i’s value-added cost share in total cost of foreign value added.

12. Solving for the first order conditions, we can derive the cost function as a function of output and prices:

13. Under perfect (or monopolistic) competition, the price of output Pj equals average cost and therefore the price of goods Pj can be expressed as a function of the price of production factors qj:

Note that this implies that the CES aggregator of GDP deflators, which measure prices of production factors or value added of different countries, is the best available proxy to capture the cost of production and hence the price of goods when there is production sharing.

14. One approach to computing the price index for goods when there is production sharing is to estimate values for σj and σjM, compute the price of goods produced in country j, Pj, and replace the CPI with this price index in computing the standard REER. There are pros and cons in doing so. One obvious advantage is that the price index computed this way would reveal the cost structure with global supply chains more accurately than an alternative approach described below, since the elasticity of substitution between domestic and foreign inputs is known to differ from that among foreign inputs (e.g., Saito, 2004). One disadvantage is that there is no consensus over the wide range of elasticities found in the literature.

15. An alternative is to assume that σj and σjM approach 1, and hence production technology is described using the Cobb-Douglas functional form:

16. The attraction of this approach is that key parameters are measureable: δj is the domestic value-added cost share in total cost, and φij is the foreign country i’s value-added cost share in total cost of foreign value added. In this case, the price of output is:

where the constant term, cj=(δjδj(1δj)(1δj)Πjiφijφij). The price of goods is expressed as a weighted average of prices of production factors or value added of different countries. Note that if the price of imported value added from each country i is expressed in local currency, it needs to be converted to country j’s currency where Ri and Rj are respective exchange rates:

Note that when there is no production sharing (i.e., δj = 1) and no intermediate inputs are traded, the price of goods in country j equals the price of domestic production factors (or value added) Pj = qj. Equivalently, the price of goods with no production sharing equals the price of final goods captured in the GDP deflator.

17. The REER of goods produced in county j under this setting is as follows:

18. The first product sum captures the role of outsourcing or foreign value added in easing an appreciation pressure coming from a relative increase in the cost of domestic production factors or an appreciation of the nominal exchange rate. For example, when there is no production sharing (i.e., , δk = δj = 1), no intermediate inputs are traded and thus the competitiveness in prices of goods is that in prices of final goods or equivalently that in factor costs:

With production sharing, however, relative prices of goods do not necessarily appreciate at the same rate as relative factor costs or as the rate of nominal appreciation. For simplicity, let us suppose that trading partners do not outsource (i.e., δk=1). In such a case, an increase in relative prices of goods due to an increase in relative cost of domestic production factors or a nominal exchange rate appreciation can be moderated by a decrease in relative costs of foreign production factors:

The moderation effects (the first product sum) are larger, the larger the foreign value-added cost share 1 – δj is. This expression captures the intuition that underlies the IEER.7

III. Empirical Applications

A. Alternative REER Indices

19. In this section, we present empirical examples of three different measures of competitiveness. Main sources of data are: the OECD Input-Output Database, the OECD Bilateral Trade Database, the United Nations Comtrade, the World Economic Outlook and the International Financial Statistics. In order to be able to compare with the standard REER index, we use the same formula as in Bayoumi, Lee, and Jayanthi (2005) as a starting point for all three measures. One difference from the standard REERs is the number of countries included: we use only 42 OECD countries and emerging markets where input-output tables are available.8

  • Standard REER: The standard measure of competitiveness currently used at the IMF is computed as follows:REERjStandard=Πkj(PjRjPkRk)wjk,where Pj and Pk are measured by the consumer price index (CPI) and wjk are the weights computed using gross trade data taken from the United Nations Comtrade.
  • REER in Goods: The REER proposed in this paper is computed as follows:REERjGoods=Πkj(cjΠij(qjRjqiRi)φij(1δj)ckΠik(qkRkqiRi)φik(1δk))wjkΠkj(qjRjqkRk)wjk,where qj and qk are proxied by the GDP deflator, and φij is the foreign country i’s value-added cost share, and δj is the domestic value-added cost share in total cost. To obtain foreign value-added shares, we supplement the OECD Input-Output Database with the OECD Bilateral Trade Database. By making the proportionality assumption as in Koopman and others (2010), we first decompose gross exports of country j into domestic value added and intermediate inputs by source. Intermediate inputs from country i (including domestic country j) are then decomposed into domestic value added of country i and intermediate inputs by source (which are then assumed to contain no further foreign value added). The reason for using export demand rather than total final demand to compute domestic value added shares is to capture the cost structure of tradables rather than that of goods more generally.
  • REER in Tasks: The first order approximation of the Value-Added Real Effective Exchange Rate (VAREER) proposed in Bems and Johnson (2012) is computed as follows:REERjTasks=Πkj(qjRjqkRk)vjk,where qj and qk are proxied by the GDP deflator and vjk are the weights computed using value-added trade data. For the empirical application in this paper, the same formula as above is used but trade and output data are adjusted using value-added data.9 For example, to obtain value-added exports, we multiply gross exports used to compute REERjStandard by the domestic value-added share in exports as in Hummels, Ishii, and Yi (2001) based on the OECD Input-Output Database. Adjusted home sales are computed using value-added exports and value-added output (obtained by multiplying gross output by the domestic value-added share in output). These adjustments are used in the construction of new weights for manufacturing and non-tourism services only, whereas other components of the REERs (weights for commodities and tourism) are computed as described in Bayoumi, Lee, and Jayanthi (2005). GDP deflator data are mainly quarterly and taken from the WEO or the IFS. Lower frequency source data have been intrapolated to generate monthly data.

B. Standard REER versus REER in Tasks

20. We first compare the standard REER that is based on trade in goods, but not adjusted for imported intermediates, and REER in Tasks, our approximation of the Bems and Johnson (2012) VAREER index. There are significant empirical differences between the two REERs that also vary across countries. This is illustrated for selected countries in Figure 1. For China and Japan, the REER in Tasks suggests more appreciation over time than the standard REER. For China the difference emerges from the early 2000s onwards, reflecting a gradual increase in domestic production costs, whereas for Japan the difference emerges in the early 1990s narrowing somewhat in the early 2000s. In cumulative terms the REER in Tasks suggest an additional close to 30 percent appreciation for both countries. For the United States, the REER in Tasks suggests a gradual improvement in competitiveness over time (about 15 percent in cumulative terms) compared to the standard REER. For Germany, the compression in domestic labor costs contributed to a decline in the REER in Tasks from the mid 1990s onwards. Overall, these differences are similar to those reported in Bems and Johnson (2012).10

Figure 1.Standard REER versus REER in Tasks

Source: Fund staff estimates.

21. The difference between the two indices comprises those in weights (i.e., gross versus value-added trade weights) and prices (CPI versus GDP deflator). As in Bems and Johnson (2012), most of the difference comes from using different price indices, not weights. For prices, this reflects the observation that when “tasks” (rather than goods) are traded across the world, the relative price of tasks (measured by the GDP deflator) should be used to measure competitiveness. Using the GDP deflator rather than the CPI to compute REERs shows a more significant deterioration in Chinese and Japanese competitiveness compared to the United States and Germany. The view that prices matter is not new and using different price indices has been regularly used to evaluate different aspects of price competitiveness.11 However, Bems and Johnson (2012) argue that trade in tasks provides a theoretical justification for using value-added prices when assessing competitiveness.

Gross Trade versus Value-Added Trade Weights

22. Differences in gross trade versus value-added trade based weights are small. Comparing weights for the latest time period (mid-2005) shows that the average decline/increase is about one-tenth of a percentage point. The largest decline is just below 3 percentage points and the largest increase is about 5 percentage points. Given the significant growth in global supply chains over time, the result that value-added weights do not matter seems counterintuitive. Bems and Johnson (2012) argue that the pattern of revisions to weights is not correlated with the pattern of cross-country differences in rates of change of exchange rates or prices, and therefore the net combined effect of all the revisions to country weights tends to not be very far from zero. The alternative explanation is that the reduced role of domestic value added in exports affects exports to all destinations equally, leaving the relative importance of trading partners unaffected. A simple example, where trade weights are based on export shares, can illustrate this point.12

23. Suppose that total gross exports of country j, Ej are the sum of exports to all foreign countries k:

Suppose also that trade weights are computed using these gross exports data, wjk = Ejk/Ej for k ≠ j. Given that gross exports are a product of value-added from all countries, country j’s bilateral exports to country k can be rewritten as the sum of domestic value-added Vjjk and the aggregate of foreign value-added Σi≠j Vijk embedded in bilateral exports to country k:

24. In a world with global supply chains with heavy reliance on outsourcing, country j’s foreign value added can carry much larger weight than in the past. For example, foreign content share in exports has increased on average from under 0.2 in early 1970s to 0.33 in mid-2000s (Riad and others, 2011). This change is revealed in the fact that the second term (i.e., the aggregate of foreign value-added Σi≠j Vijk) in each bilateral exports (in each bracket) has become relatively large, compared to the first term (i.e., the domestic value-added Vjjk) over time. Because the reduced role of domestic value added in exports affects exports to all destinations more or less equally, the relative importance of trading partner k remains relatively unaffected. It is only relative since some differences remain because of differences in export product composition across different destinations.

25. Though small in magnitude, compositional differences between gross and value-added trade weights tend to be in favor of advanced economies. The weight of advanced economies generally increases, whereas the weight of emerging market economies and newly industrialized Asian countries declines (Figure 2). This moderates the finding in Bayoumi, Lee, and Jayanthi (2005) suggesting a significant increase in trade weights towards emerging market economies. The increase in the weight of advanced economies reflects an increase in the weights for the United States and Japan, whereas the weights for the euro area countries tend to decline. The increase in the U.S. weight is broad-based across markets, whereas the increase in the weight of Japan is largest among partners in the region (in particular Korea, Thailand and Taiwan). While the weight of emerging economies as a whole tends to decline, the weight of China increases for some trading partners, mainly within the region and the United States. In terms of the REERs, weights do matter for some countries that are more involved in (regional) supply chains. For Japan, for example, the value-added weights alone (excluding any change in the price measure used) result in somewhat more appreciation of the REER over time.

Figure 2:Differences in Value-added Versus Gross Trade Weights

Source: Fund staff estimates.

Note: 20 largest trading partners.

26. These results suggest that the overall position in supply chain matters for weights: upstream countries with higher domestic value-added content in exports tend to gain, whereas downstream countries with more processing (less domestic value-added in exports) tend to lose weight. This reflects the fact that upstream countries’ value-added exports to China (for example) are now correctly allocated to weights for upstream countries. Changes in weights over time also reflect an increased role of supply chains. Results for other periods show similar small differences in weights on average. Overall, however, differences in weights are somewhat larger and more widespread in the 2000s and mid-2000s than in the mid-1990s, likely reflecting an increase in the importance of global and regional supply chains.

27. In summary, over the last 30 years or so, domestic value added in a country’s gross exports has declined, on average from about 80 percent to about 66 percent of total exports. Because of the reduced role of domestic value added in exports affecting exports to all destinations equally, however, the relative importance of export destination in value-added trade weights look very similar to those in gross trade weights. However, within the small differences observed in gross versus value-added trade weights, upstream countries with higher domestic value-added content in exports tend to gain, whereas downstream countries with more processing (less domestic value-added content in exports) tend to lose weight.

C. REER in Goods versus REER in Tasks

28. We turn to a comparison of the REER in Tasks, our approximation of the Bems and Johnson (2012) VAREER index and our REER in Goods measure next. The main conceptual difference between Bems and Johnson (2012) and the REER in Goods measure proposed in this paper is whether competitiveness is over tasks or goods. When one is interested in measuring a country’s competitiveness where “tasks” are traded, it is sufficient to compute REER in Tasks. If however, one is interested in goods’ competitiveness, it can be useful to derive explicitly the price of goods that fully reflects the cost structure of production. Section II showed that REER in Goods needs to be expressed in terms of the price of value-added and also needs an additional term. This term is a weighted average of relative prices of foreign value-added used in production (relative to those of trading partners), and therefore can capture potential loss (or gain) in competitiveness due to a rise (or a fall) in the cost of importing value-added from the rest of the world.

29. The differences between REER in Tasks and REER in Goods are generally smaller, but matter in some cases (Figure 4). Empirical differences between REER in Tasks and REER in Goods mostly come from the additional term, which captures the role of outsourcing in easing the appreciation pressure on the relative prices of goods. For the United States, Japan and Germany the two series are nearly identical. However, for China, REER in Goods suggests lower appreciation than REER in Tasks. This implies that a relative increase in domestic factor costs in China has not been translated into an equivalent increase in the relative price of Chinese goods. In cumulative terms since 1990, REER in Tasks reveals an appreciation of factor costs by 34 percent, a substantial difference from what is revealed in the standard REER, an appreciation of relative goods prices of 7 percent. REER in Goods indicates an appreciation of relative prices of goods by about 14 percent. The direction of adjustment for the last few years is consistent with results shown in Unteroberdoerster, Mohommad, and Vichyanond (2011).

Figure 4.REER in Goods versus REER in Tasks

Source: Fund staff estimates.

30. Emerging market economies in Europe (e.g., Estonia, Hungary, Romania and Slovak Republic) and those in Asia (e.g., China and Vietnam) saw largest cumulative differences between REER in Tasks and REER in Goods (Figure 5). The differences reflect an important role of global supply chains for these countries, reflected in a large share of foreign value added in their exports, as well as an increase in domestic factor prices over this time period. For these countries, participation in global supply chains has helped offset the impact of an increase in domestic costs, maintaining their competitiveness. While magnitudes are smaller, the reverse is true for many advanced economies that are well integrated in Global supply chains: their competitiveness has been eroded by the increasing cost of imported production factors from emerging market economies.13 Overall, both appreciations and depreciations tend to be mitigated when the use of outsourcing is taken into account, except for a number of countries where losses in competitiveness owing to a relative increase in domestic costs are made worse by an increase in the cost of imported intermediate inputs (e.g., Greece, the Netherlands, Portugal, and Spain).

Figure 5:Change in REER: Tasks versus Goods

(cumulative percentage change, 2000-11)

Sources: Fund staff estimates.

31. Incorporating global supply chains seems to improve the fit of trade equations too. A simple net export equation is estimated using the standard REER and two alternative REERs. Because of the presence of time-series properties in data, the dynamic panel data analysis is applied. The net export equation is assumed to be a function of trading partner’s output, its own output, and real effective exchange rates, measured either in terms of the standard REER or the two alternatives. Given that net export is a value-added concept, REER in Tasks and REER in Goods should (in theory) be equivalent in fitting net export data. The results show that REER in Tasks and REER in Goods are both improvements in terms of fitting net export data, reflected in lower sigmas (the standard error of the regression) when the new indices rather than the standard REER are used as regressors.

Table 1:Dynamic Panel Data Analysis(all variables in first differences, 1990-2011)
Dependent Variable:Ln (net exports)t
StandardGoodsTasks
constant0.0010.0030.002
Ln (net exports)t-10.766**0.713**0.725**
Ln (net exports)t-20.048**0.041*0.045**
Ln (output_partner)t0.149*0.152*0.147*
Ln (output_partner)t-1-0.228*-0.225*-0.210
Ln (output_partner)t-20.0180.013-0.004
Ln (output_own)t-0.259**-0.241**-0.246**
Ln (output_own)t-10.289**0.254**0.265**
Ln (output_own)t-2-0.021-0.004-0.008
Ln (reer)t-0.057**
Ln (reer)t-10.024**
Ln (reer)t-20.005
Ln (reer in goods)t-0.078**
Ln (reer in goods)t-10.036**
Ln (reer in goods)t-2-0.002
Ln (reer in tasks)t-0.061**
Ln (reer in tasks)t-10.029**
Ln (reer in tasks)t-20.003
sigma0.01810.01750.0176
No of countries424242
No of observations785785785
Sources: Fund staff estimates.

The One-Step GMM estimator (Arellano and Bond, 1991) used. ** and * denotes statistical significance at the 1- and 5-percent levels, respectively.

Sources: Fund staff estimates.

The One-Step GMM estimator (Arellano and Bond, 1991) used. ** and * denotes statistical significance at the 1- and 5-percent levels, respectively.

IV. Concluding Remarks

32. With an increasing role of global supply chains, the need to revisit the standard formula in computing REERs has intensified. Two alternative approaches, both trying to incorporate the role of global supply chains, have reached different empirical findings for some countries. Bems and Johnson (2012), by measuring competitiveness in “tasks,” have found that China’s competitiveness, for example, has appreciated more than what is revealed in the standard REER.

33. We argue that with production sharing, changes in relative prices of goods have become less sensitive to changes in relative factor prices. To demonstrate this point, we construct a new index, REER in Goods, with prices of those goods reflecting “global supply chain” production structure. We compare this new index with the standard REER index and an index that reflects trade in tasks (REER in Tasks). Using the new index, we find that thanks to their participation on global supply chains, many emerging market economies have maintained goods competitiveness while losing competitiveness due to a relative increases in domestic factor costs or an appreciation of nominal exchange rate.

References

    Armington, Paul S.,1969, “A Theory of Demand for Products Distinguished by Place of Production,” IMF Staff Papers, Vol. 6, pp. 159–78.

    Bayoumi, Tamim, RichardHarmsen, and JarkkoTurunen,2011, “Euro Area Export Performance and Competitiveness,” IMF Working paper 11/140 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    Bayoumi, Tamim, JaewooLee, and SharmaJayanthi,2005, “New Rates from New Weights,” IMF Working Paper 05/99 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    Bems, Rudolfs, and RobertC. Johnson,2012, “Value-Added Exchange Rates,” NBER Working Paper 18498 (Cambridge, MA: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    Hummels, David, JunIshii, and Kei-MuYi,2001, “The National and Growth of Vertical Specialization in World Trade,” Journal of International Economics, Vol. 54, pp. 75–96.

    • Crossref
    • Search Google Scholar
    • Export Citation

    Koopman, Robert, WilliamPowers, ZhiWang and Shang-JinWei,2010, “Give Credit Where Credit Is Due: Tracing Value Added in Global Production Chains,” NBER Working Paper 16426 (Cambridge, MA: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    Lipschitz, Leslie and DonoghMcDonald,1992, “Real Exchange Rates and Competitiveness: A Clarification of Concepts and Some Measurements for Europe,” Empirica, Vol. 19, No. 1, pp. 37–69.

    • Crossref
    • Search Google Scholar
    • Export Citation

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

    • Search Google Scholar
    • Export Citation

    Riad, Nagwa, LucaErrico, ChristianHenn, ChristianSaborowski, MikaSaito, and JarkkoTurunen,2011, Changing Patterns of Global Trade (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    Saito, Mika,2004, “Armington Elasticities in Intermediate Inputs Trade: a Problem in Using Multilateral Trade Data,” Canadian Journal of Economics, Vol. 37, No. 4, pp. 1097–117.

    • Crossref
    • Search Google Scholar
    • Export Citation

    Thorbecke, Willem,2011, “Investigating the Effect of Exchange Rate Changes on China’s Processed Exports,” Journal of the Japanese and International Economies, Vol. 25, No. 2, pp. 33–46.

    • Crossref
    • Search Google Scholar
    • Export Citation

    Unteroberdoerster, Olaf, AdilMohommad, and JadeVichyanond,2011, “Implications of Asia’s Regional Supply Chain for Rebalancing Growth”,Chapter 3, Regional Economic Outlook: Asia and Pacific,April (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
1

We would like to thank Rudolfs Bems, Olaf Unteroberdoerster, Steve Phillips, and participants in the SPR seminar series for helpful comments, and Emmanuel Hife and Xingwei Hu for excellent research assistance.

2

For the current method used at the Fund, see Bayoumi, Lee, and Jayanthi (2005).

3

The GDP deflator is an imperfect proxy for the price of production factors since it also captures changes in profits. However, in the absence of comprehensive and comparable data on factor costs, we argue that the GDP deflator is the best available proxy for factor costs.

4

Assuming perfect competition, it can also justify the use of unit labor costs (assuming further that labor is the only factor of production) or GDP deflator as the price of final goods.

5

This reflects a gain in competitiveness stemming from participating in global supply chains. Symmetrically, some countries may face a loss in competitiveness when relative prices of goods rise faster than relative costs of domestic factors. Note that this gain/loss owing to participating in global supply chains refers to changes in relative prices over time, not to the level of relative prices. Outsourcing of production, typically from advanced economies to emerging markets, has often been driven by a lower level of labor costs, resulting in a gain in goods’ competitiveness from participating in global supply chains. The gain addressed in this paper is distinct from this since the REER is about relative changes and not about levels.

6

Goods include both final goods and intermediate inputs. The consumer price index, which is used in to compute the REER at the Fund is an imperfect proxy of the price of goods since it only covers the price of final goods, and only partially.

7

One caveat of this approach is an implicit assumption that aggregate supply is fixed at ex-ante equilibrium. Armington derives the demand for a product in a particular market using a two-step procedure: (i) the demand for the good in general is derived by maximizing a utility function; and (ii) the demand for a good produced by an individual country is determined by minimizing the cost of purchasing the amount of the good derived in the first step. We assume that the intermediate input demand is determined by cost minimization for a given level of aggregate supply Xj and the demand for consumer goods by utility maximization subject to a given level of factor income (which is a also function of aggregate supply). Keeping the standard trade weights in our analysis implies that the aggregate supply Xj remains fixed at ex-ante equilibrium. Bems and Johnson (2012) instead solve for an equilibrium level of output where aggregate supply equals aggregate demand. The formula for trade weights is different, reflecting differences in goods versus value-added trade patterns.

8

The standard weights used to compute the REER, the so-called “INS weights,” include only partner countries with a weight above 1 percent in trade and therefore the number of partner countries in INS weights varies both over time and across countries. A comparison of our gross trade based weights for the mid 2000s with the latest available INS weights suggest that the larger coverage of trading partners results in minor changes in the REER indices.

9

Value-added trade weights computed here are different from those computed in Bems and Johnson (2012) for two main reasons. First, Bems and Johson (2012) solve analytically for value-added trade weights and compute weights according to the new formula. Our value-added trade weights are only a first order approximation of their value-added trade weights. Second, their value-added trade weights change every year while ours only three times during the sample period (mid 1990s, early 2000s, and mid 2000s) as in the standard trade weights currently used at the Fund. There may also be differences in the underlying input-output data used to compute value-added trade.

10

The differences are similar but smaller since value added trade weights in this paper are only a first order approximation of the value added trade weights computed in Bems and Johnson (2012).

11

See Lipschitz and McDonald (1992) and Bayoumi, Harmsen, and Turunen (2011) for a comparison of REERs with different deflators for European countries.

12

Note that the INS weights are not a simple export trade weights.

13

Note that this apparent loss owing to outsourcing refers to changes in relative prices over time, not to the level of relative prices. Owing to large remaining level differences in costs, it is likely that outsourcing continues to contribute positively to competitiveness in advanced economies.

Other Resources Citing This Publication