Accounting for U.S. Regional Real Exchange Rates
Chen, Lein-Lein, Choi, Seungmook, Devereux, John, Journal of Money, Credit & Banking
THE DISTINCTION BETWEEN tradables and nontradables is at the core of open economy macroeconomics. (1) Engel (1999), however, has recently questioned the empirical relevance of this distinction. Using various real exchange rate measures, he finds that changes in the relative price of nontradeables explain little of U.S. real exchange rate movements at short or medium time horizons for fixed or floating exchange rate regimes. As noted by Obstfeld and Rogoff (2000) and Obstfeld (2001), these findings are devastating for traditional tradables/nontradables models.
Engel's results are plausible for floating exchange rates where changes in nominal exchange rates tend to overwhelm price level movements. We would, however, expect changes in the relative price of nontraded goods to play a larger role where exchange rates are fixed. Mendoza (2000) provides some early support for this position. Using data for the Mexican/U.S. real exchange rate, he finds that changes in the prices of nontradables explain 70% of real exchange rate movements during periods of fixed rates or managed floating.
In this paper, we examine the relationship between the relative price of nontraded goods and the real exchange rate with data from four U.S. regions: the Northeast, Midwest, South and West. In terms of size and economic structure, these regions are comparable to large developed countries. They allow us therefore to study real exchange rate movements in economies with permanently fixed rates as well as high levels of factor mobility and goods market integration. Furthermore the U.S. regional data are superior to data used internationally in they are collected for identical sets of goods and services, the weights are similar across regions for broad aggregates and the same methods are used to introduce new goods and to adjust for quality changes. Thus, many of the data difficulties faced internationally are not present for U.S. regions. (2)
We show that changes in the relative price of nontradables account for a large portion of regional real exchange rate movements over medium and longer run horizons. Indeed, they explain 80% of real exchange rate changes at horizons above two years. In addition, we find that the dominance of nontradables is because departures from purchasing power parity (PPP) for tradables are short lived at the regional level.
The final portion of the paper compares the U.S. regional and international evidence on the relative importance of nontradables. We argue that differences between the regional and the international results arise because traded good markets are better integrated across U.S. regions. In addition, we show that the findings for international data depend on the share of nontradables in expenditure. Using plausible expenditure shares, we find that nontradables can account for 50% of U.S. real exchange rate changes with Germany, France, and Japan for the Bretton Woods system.
We proceed as follows. Section 1 outlines real exchange rate accounting. Section 2 applies real exchange rate accounting to regional data while Section 3 extends the results to city data. Section 4 discusses the relative importance of nontradables for Canada, France, Germany, Italy, and Japan during fixed rates between 1962 and 1972. Section 5 summarizes.
1. REAL EXCHANGE RATE ACCOUNTING
This section introduces real exchange rate accounting drawing on Engel (1999). The next section applies the approach to regional data.
Assume that the overall price level for the i'th region is given by Equation (1), where [P.sub.i] is the log of price level and [p.sub.i.sup.T] and [p.sub.i.sup.N] are traded and nontraded prices, respectively, in logs and [alpha] is the share of the nontradables in expenditure.
[p.sub.i] = (1 - [alpha])[p.sup.T.sub.i] + [alpha][p.sup.N.sub.i]. (1)
We further assume that the share of nontradables is the same for all regions. Following Engel (1999), we express the real exchange rate between the i'th and j'th regions, denoted by [q. …