Value-Added Real Effective Exchange Rates: Testing for Countries with High and Low Vertical Specialization in Trade

VI. Conclusion

The rise in vertical specialization as a phenomenon in world trade indicates that REERs constructed to reflect value-added trade and value-added prices might be more suitable for explaining changes in a country's export volume if its export structure is characterized by high levels of vertical specialization. Bems and Johnson (2012) construct such an exchange rate, and observe substantial differences between VAREERs and conventional REERs, determined primarily by the replacement of consumer prices with GDP deflator as a proxy for value-added prices, and not by the modification of bilateral trade weights to reflect value added trade.⁵⁸

This paper tests the suitability of the replacement of consumer prices with GDP deflator by constructing conventional and value-added REERs for two countries with different levels of vertical specialization: Germany and Belgium. I estimate an error-correction model, as prescribed by Engle and Granger (1987), to compare the relative explanatory strength of the value-added and conventional real effective rates for Germany and Belgium. The results of the ECM indicate that, in the long-run, the VAREER surpasses the ability of the conventional REER to explain export demand for Belgium, while the opposite is true for Germany, as expected.

The ECM also shows, however, that the value-added and conventional REERs have negligible short-run explanatory power. It was suggested that part of this may be due to the inability of the ECM to take into account optimal lags for each variable. To investigate this possibility, I regress an ARDL model using first-differences to estimate Belgian and German exports with lagged exports, lags of OECD real GDP, and lags of the two different REERs. The results of the ARDL regressions indicate that the value-added and conventional REERs are significant for both countries in the short-run, and the relative size of the F-statistics of the different rates correspond with the predictions of the theory; specifically, the joint significance of the VAREER is slightly greater than that of the conventional REER for Belgium, while the VAREER significance is slightly lower than that of the conventional REER for Germany.

The ARDL results, though, also suggest that there may be differences between the two countries not properly controlled for in the initial regressions. As a further robustness check, I estimate a fixed effects panel regression of the VAREER regressed over exports with the country as the panel variable. The results of the panel regression cloud the picture provided by the ECM and ARDL model. The panel joint significance tests suggest that the VAREER might not be able to claim statistical significance as a predictor of exports for either country in the short-run. Furthermore, the relative magnitude of the F-statistics in the panel slightly contradicts the relationship predicted by the theory as well as the findings of the ARDL model, since the joint significance of the German VAREER exceeds that of the Belgian VAREER in the panel. The panel results, however, are consistent with the findings of the ECM, which also indicated negligible short-run effects in the export equations for both countries.

To address the failure to find conclusive results in the shortrun, there are typically three reasons why the quantitative predictions of a theory are not visible in the data: (i) the effect is too small, (ii) the theory is wrong, or (iii) the methodology is imperfect/not very powerful. Since the long-run results in the ECM strongly corroborate the theory put forward by Bems and Johnson (2012), it seems unlikely they have made a theoretical error. However, there may be merit in explanations (i) and (iii).

With regards to the first explanation, one potential issue that may explain the failure of the panel and short-run ECM estimates to distinguish between the German and Belgian rates in a manner consistent with the theory is that the level of vertical specialization in each country likely varied at different rates during the period studied. This would certainly have a substantial effect on the findings of the short-run models. Furthermore, the scarcity of more recent data on vertical specialization makes it difficult to completely ascertain the relative size differences in the level of vertical specialization between Belgium and Germany over the entire period.

With regards to the third explanation for the failure to find conclusive short-run results, the lack of an error-correction model that uses optimal lags in the short-run may account for the statistical insignificance of the first differences of the REERs. To fix this, future work with an ARDL error-correction model, which contains short-run variables consistent with the long-run relationship and chooses optimal lags for the shortrun, may give a more accurate picture of the short-run effect of REERs on export demand.

In sum, the results of the Engle-Granger ECM are promising for the long-run validity of the claims of Bems and Johnson (2012). However, since the panel regression seems to call into question the short-run results of the ARDL model, the VAREER may have negligible utility as a policy tool for assessing competitiveness, since the short-run relationship holds greater weight in policy deliberations. Further research that looks into the longand short-run suitability of the VAREER in other contexts would be valuable in helping to clarify these issues. Specifically, expansion of the work done in this paper using ARDL and panel error-correction models that account for optimal lags in the short-run could provide a more accurate picture of the value-added rate's suitability over both periods. Furthermore, as more data on vertical specialization become available, extending these techniques to a broader range of countries to get a more comprehensive view of the differences between the value-added and conventional REERs would represent another important contribution. Additional research that more directly tests for a connection between the suitability of the VAREER and rising vertical specialization by incorporating continuous measures of vertical specialization in place of using countries with different levels of vertical specialization as an indirect test of the relationship would also be of value.

References

1. Bayoumi, Tamim, Jaewoo Lee, and Sarma Jayanthi. 2006. "New Rates from New Weights." International Monetary Fund, Staff Papers 53(2): 272-305.
2. Bems, Rudolfs, and Robert C. Johnson. 2012. "Value Added Exchange Rates: Measuring Competitiveness with Vertical Specialization in Trade." Working paper, IMF Research Department.
3. Bems, Rudolfs, Robert C. Johnson, and Kei-Mu Yi. 2011. "Vertical Linkages and the Collapse of Global Trade." American Economic Review 101(3):308-317.
4. Breda, Emanuele, Rita Cappariello, and Roberta Zizza. 2008. "Vertical Specialization in Europe: Evidence from the Import Content of Exports." Bank of Italy Temi di Discussione 682: 1-21.
5. Chowdhury, Abdur. 1993. "Does Exchange Rate Volatility Depress Trade Flows? Evidence from Error-Correction Models." Review of Economics and Statistics 75(4): 700706.
6. Eurostat. 2002-2013. "Gross Domestic Product at Market Prices."
7. Engle, Robert F. and C.W.J. Granger. 1987. "CoIntegration and Error Correction: Representation, Estimation, and Testing." Econometrica 55(2): 251-276.
8. Global Insight. 1980-2012. "GDP, Real, US Dollars." Key Indicators.
9. Hummels, David, Jun Ishii, and Kei-Mu Yi. 2001. "The Nature and Growth of Vertical Specialization in World Trade." Journal of International Economics 54: 75-96.
10. International Monetary Fund. 1980-2012. "Exchange Rates: Market Rate Period Average." International Financial Statistics. Accessed with IHS Global Insight.
11. International Monetary Fund. 1980-2012. "Exports To: World." Direction of Trade Statistics. Accessed with IHS Global Insight.
12. International Monetary Fund. 1980-2012. "National Accounts: GDP Deflator." International Financial Statistics. Accessed with IHS Global Insight.
13. International Monetary Fund. 1980-2012. "Prices, Production, Labor: Consumer Prices." International Financial Statistics. Accessed with IHS Global Insight.
14. International Monetary Fund. 1980-2012. "Imports From: World." Direction of Trade Statistics. Accessed with IHS Global Insight.
15. Khan, Mohsin S. 1974. "Import and Export Demand in Developing Countries." International Monetary Fund, Staff Papers 21(3): 678-693.
16. MacKinnon, James G. 1990, 2010. "Critical Values for Cointegration Tests." Queen's Economics Department Working Paper, No.1227.
17. Organization for Economic Cooperation and Development. 1980-2012. "Consumer Price Index." Main Economic Indicators. Accessed with IHS Global Insight.
18. Schaffer, Mark E. 2010. "egranger: Engle-Granger (EG) and Augmented Engle-Granger (AEG) Cointegration Tests and 2-step ECM Estimation." http://ideas.repec.org/c/boc/bocode/s457210.html
19. Stock, James H., and Mark W. Watson. 2011. Introduction to Econometrics. Third ed. Boston, MA: AddisonWesley.
20. Yi, Kei-Mu. 2003. "Can Vertical Specialization Explain the Growth of World Trade?" Journal of Political Economy 111(1): 52-102.
21. United Nations. 1980-2012. "UN COMTRADE." SITC Revision 1. Accessed with World Bank World Integrated Trade Solution.

Endnotes

1. Special thanks to my faculty advisor, Dr. Iqbal Zaidi, and my graduate student advisor, Olivier Darmouni, for their invaluable guidance during this project.
2. Hummels, Ishii, and Yi 2001, 76.
3. Yi 2003, 91.
4. Hummels, Ishii, and Yi 2001, 77.
5. Bems, Johnson, and Yi 2011, 316-317.
6. Bems and Johnson 2012, 2.
7. Bems and Johnson 2012, 2.
8. Bems and Johnson 2012, 3.
9. Bems and Johnson 2012, 1-2.
10. Hummels, Ishii, and Yi 2001, 84; Breda, Cappariello, and Zizza 2008, Tab. 1-2.
11. Hummels, Ishii, and Yi 2001, 77.
12. Hummels, Ishii, and Yi 2001, 83-5. Recognizing that changes in the price of imported oil could change their measure of each country's level of vertical specialization, the authors found it necessary to calculate the vertical specialization share of exports twice for each country and excluded energy trade in the second calculation.
13. Hummels, Ishii, and Yi 2001, 83.
14. Breda, Cappariello, and Zizza 2008, Tab. 2.
15. Bems and Johnson 2012, 2-3.
16. Bems and Johnson 2012, 2.
17. Bems and Johnson 2012, 18.
18. Bems and Johnson 2012, 15.
19. Bems and Johnson 2012, 18.
20. Bems and Johnson 2012, 17-24
21. Bems and Johnson 2012, 24.
22. Bems and Johnson 2012, 17.
23. Bems and Johnson 2012, 24.
24. Bems and Johnson 2012, 16.
25. Bayoumi, Lee, and Jayanthi 2006, 272.
26. Hummels, Ishii, and Yi 2001, 84; Breda, Cappariello, and Zizza 2008, 6.
27. Breda, Cappariello, and Zizza 2008, Tab. 1-2.
28. Author's calculation based on data gathered from the IMF's Direction of Trade Statistics.
29. Breda, Cappariello, and Zizza 2008, Tab. 1-2.
30. Bems and Johnson 2012, 2.
31. Khan 1974, 682.
32. Engle and Granger 1987, 253.
33. Engle and Granger 1987, 254.
34. Engle and Granger 1987, 264-267.
35. Engle and Granger 1987, 275.
36. Stock and Watson 2011, 553.
37. Stock and Watson 2011, 553.
38. Stock and Watson 2011, 553.
39. Stock and Watson 2011, 552.
40. Chowdhury 1993, 701.
41. Engle and Granger 1987, 275.
42. Schaffer 2010.
43. Stock and Watson 2011, 552.
44. Engle and Granger 1987, 265.
45. Engle and Granger 1987, 252.
46. Chowdhury 1993, 703; Engle and Granger 1987, 262.
47. Stock and Watson 2011, 538.
48. Stock and Watson 2011, 535.
49. Stock and Watson 2011, 538.
50. Stock and Watson 2011, 545.
51. Stock and Watson 2011, 544.
52. Stock and Watson 2011, 356.
53. Global Insight. 1980-2012. "GDP, Real, US Dollars." Key Indicators.
54. Bayoumi, Lee, and Jayanthi 2006, 279-280.
55. Bayoumi, Lee, and Jayanthi 2006, 286.
56. Bems and Johnson 2012, 24.
57. Schaffer 2010; Critical values reported by "egranger" come from MacKinnon (1990, 2010).
58. Bems and Johnson 2012, 24.