A key stylised fact in international macroeconomics is that the real exchange rate is highly volatile and exhibits, at best, a slow rate of convergence towards a constant long-run equilibrium. Fluctuations in the real exchange rate may be brought about by deviations from the law of one price of traded goods or differential movements in the relative price of non-traded to traded goods in different countries. Traditionally, traded goods have been assumed to obey the law of one price, leaving changes in the relative price of nontraded to traded goods as the only source of real exchange rate fluctuations.
However, a vast empirical literature has rejected the law of one price and found evidence of large fluctuations in the relative prices of traded goods across countries. Moreover, in a very influential paper, Engel (1999) claims that the relative price of non-traded to traded goods accounts for essentially none of the real exchange rate fluctuations. Specifically, he finds that over 90 percent of the fluctuations in the real exchange rate can be attributed to fluctuations in the relative price of traded goods for several OECD countries relative to the US. Chari, Kehoe and McGrattan (2002) find similar results. This evidence has been interpreted to imply that it is not important to distinguish between tradable and nontradable goods to understand the cyclical real exchange rate fluctuations (Burstein Eichenbaum and Rebelo, 2005) and has motivated a tremendous increase in research on models where traded goods prices account for all the movements in the real exchange rate.
Burstein, Eichenbaum and Rebelo (2005) question the results obtained by Engel (1999) on the grounds that he uses consumer price indices to measure traded goods prices. The traded goods included in the consumer price index (CPI) are highly contaminated by non-traded components such as wholesale, distribution and retail services. In addition, goods specified as traded in the CPI may not actually be subject to trade. This may create a bias towards finding a greater importance of the traded goods sector in explaining real exchange rate fluctuations. To circumvent these possible problems BER use export price indices (EPI) and import price indices (IPI) to construct the relative price of traded goods. Using this measure they find that the nontraded component accounts for about half of the fluctuations in the real exchange rate, suggesting that the distinction between non-traded and traded goods is important forunderstanding real exchange rate fluctuations.
In this thesis I argue that the approach taken by BER is likely to overestimate the importance of non-traded goods in explaining real exchange rate fluctuations. By using total import and export price indices they include several components not included in the CPI, e.g. raw materials such as raw oil and gas, and investment goods. If these components have a higher tendency to obey the law of one price than consumer goods, their inclusion in the price index of traded goods will lead to an overestimation of the importance of non-traded goods in explaining real exchange rate fluctuations.
The contributions in this thesis are twofold: First, I decompose the fluctuations in the Norwegian-US real exchange rate using the methods proposed by Engel (1999) and BER. A main motivation for this analysis is to see whether Norway as a raw material based economy is different from the countries studied by those authors. Second, I decompose the real exchange rate using a new measure of traded goods prices based on import and export prices of consumer goods. This decomposition has been made possible by the fact that Statistics Norway recently published export and import price indices categorised by Broad Economic Categories (BEC). The advantage of using this measure compared to the measure used by BER is that it is possible to exclude all goods but consumer goods from the export and import price indices. Hence, I ensure that the goods used to calculate traded goods prices are similar to the ones included in the consumer price indices used to calculate the real exchange rate.
As expected, the share of real exchange rate fluctuations attributed to the relative price of non-traded to traded goods depends critically on the price measure used to calculate the relative price on traded goods. Not only does the distinction between retail and ‘at the dock’ prices matter, but also the composition of goods in the export and import price indices. Using retail prices, an upper bound of the importance of the relative price of non-traded to traded goods is found to be about 3 percent. By contrast, the upper bound is 65 percent when using aggregate export and import price indices. These results are similar to the results obtained by Engel (1999) and BER. However, the exclusion of all but consumer goods from the export and import price indices significantly lower the importance of the relative price of non-traded to traded goods, and the upper bound falls to 31 percent. Moreover, using a new real exchange rate decomposition I am able to explain most of the discrepancy between the results obtained using retail prices and ‘at the dock’ prices of consumer goods: The local distribution costs of traded goods are found to account for about 25 percent of the real exchange rate fluctuations. This suggests that it is important to distinguish between retail prices of traded goods and pure ‘at the dock’ prices of traded goods when accounting for real exchange rate fluctuations.