Economic development and the role of currency undervaluation.

• Author: Bhalla, Surjit S.
• Position: Report

This article is concerned with the determinants of economic growth, and, in particular, with the role of policy in directing the pattern of growth in developing economies. Over the years, economists and policymakers have focused primarily on fiscal and exchange rate policy. While the role of fiscal deficits is well understood, the same agreement does not hold with regard to exchange rate policy.

Part of the reason for the controversial nature of exchange rate policy is that it comes in various styles--floating, dirty float, managed, dirty managed, pegged, and fixed. Besides floating, the success or failure of a particular exchange rate policy appears to be contingent not on the nature of the regime but rather on the direction of the misalignment of the currency. Bad exchange rate policy, in the form of an overvalued exchange rate, has been much analyzed and the results are well known. A persistently overvalued exchange rate leads to factor misallocations, loss in efficiency, higher inflation, and lower GDP growth. This conclusion is widely accepted, but the parallel theoretical and analytically equivalent conclusion--that exchange rate undervaluation is helpful to growth--is not. Moreover, even if accepted in theory, in practice the discussion gets involved with definitional and measurement issues. How do we define the real exchange rate? How does one measure it? Most important, how does one measure equilibrium? It is the latter that allows misalignment to be measured.

Most of the empirical results to date do not support the conclusion that exchange rate undervaluation is helpful for growth. Easterly (2005) concludes, on the basis of an updated Dollar (1997) measure of currency undervaluation, that a policy of undervaluation does not matter as a determinant of economic development. Acemoglu et. al (2003) and the IMF (2005) reach the same conclusion, and find the impact to be even weaker if institutions are introduced into the growth model. Lately, however, some evidence of the positive growth effects of currency undervaluation is beginning to surface. In various articles (Bhalla 2002, 2005, 2008a, 2008b), I document this effect, as do Johnson, Ostry, and Subramaniam (2007, hereafter JOS) and Rodrik (2007).

The plan of this article is as follows. In the next two sections, I discuss the definition of the real exchange rate and the different methods of measurement. I then examine the empirical basis for the notion that the "real exchange rate is endogenous (RERIE)." Phrased differently, the RERIE argument is the same as saying that the "impossible trinity" holds--namely, that it is impossible to simultaneously target the nominal exchange rate, have freely floating capital, and have an independent monetary policy. Next, I explain the mechanism through which exchange rate undervaluation is likely to work--by increasing the profitability of investment, leading to higher growth and savings and, therefore, to the operation of a virtuous cycle of investment-growth-savings. Empirical results are then presented for the effect of currency undervaluation (and overvaluation) in standard growth models. This section documents the large empirical role of two variables related to currency undervaluation: the valuation in the initial year, and the average change in undervaluation over the period examined. The latter variable is found to be particularly significant, and it helps explain the fast growth episodes of several economies. In the final section, I offer some policy conclusions.

The Real Exchange Rate: Definition and Measurement

There are two closely related definitions of the red exchange rate (RER). The two definitions represent the external sector (primary definition) and the internal sector (secondary definition). The "external" definition of the RER is the ratio of the overall price levels between two economics, (1) while the "internal" definition is the domestic economy ratio of the price of nontradable goods ([P.sub.N]) to the price of tradable goods ([P.sub.T]). For both definitions, a rise in the RER constitutes an appreciation.

The Primary Definition

The most widely used ratio of price levels is that published as "Penn World Tables 6.1" or PWT. (2) These data use the periodic ICP (International Comparison of Price) surveys of different countries to compute an intertemporal price series for each country. The ratio of price levels is presented in a common purchasing power parity (PPP) currency, with the U.S. price level defined to be 100 in each year. This ratio is identically equal to the ratio of the exchange rates between each country (i) and the United States (usa); it is the ratio of the current PPP exchange rate. (3)

Thus,

(1) RER = ratio of country price levels = [P.sub.i] / [P.sub.usa]

(2) RER = ratio of exchange rates = exchange rate with respect to PPP dollar / exchange rate with respect to U.S. dollar = [E.sub.i] / [E.sub.usa]

(3) RER = [P.sub.i] / [P.sub.usa] = [[E.sub.i] / [E.sub.usa].

In two articles, Balassa (1964) and Samuelson (1964), hereafter B-S, show that the RER is positively associated with the level of per capita income. Balassa estimates the RER to be a function of real per capita income for 12 OECD economies. A strong cross-sectional relationship was observed for the year 1960 and Balassa theorized that this cross-sectional relationship between overall price levels and income was a consequence of a time-series relationship within each country. In particular, that the price level was higher in richer countries, because the process of becoming rich involved a faster productivity growth in the manufacturing tradable goods sector. Faster growth in the tradable sector means slower income and productivity growth in the nontradable services sector. Likely, higher relative productivity growth will lead to lower relative prices for tradables, ceteris paribus. Higher CDP growth over a sustained period of time leads to a higher income level (as in the United States), and is associated with a higher overall price level (price of goods the same in the different economies but price of services higher in the richer economy). Thus, both definitions of the RER (ratio of overall price levels between economies or ratio of [P.sub.N] to [P.sub.T] within an economy) will increase with economic growth or an increase in per capita income.

The Secondary Definition

The association between the pattern of growth in relative price levels and the assumed cause, slower productivity growth in services, has led to a parallel indirect definition of the real exchange rate--namely, the ratio of prices of nontradables to tradables.

According to this definition,

(4) [RER.sub.N/T] = [P.sub.N] / [P.sub.T].

Implicit in this definition of the RER is the assumption that over time, changes in tradable goods prices are broadly equal in country i and the United States. Thus, changes in overall price levels can occur only via changes in the prices of nontradables in the two economies. This price change is higher in the faster-growing economy. Over time, among comparable economies, countries that are richer will be revealed to have a higher [RER.sub.NT], the same result as that obtained for the first inter-country definition of the RER--that is, [P.sub.i] / [P.sub.usa].

Measurement of the Equilibrium Real Exchange Rate

Does reality conform to the B-S predicted pattern of the RER increasing with per capita income? It is hard to test for the relationship between [P.sub.N] / [P.sub.T] and income because data for the decomposition of P between [P.sub.N] and [P.sub.T] are not readily available. Nevertheless, several attempts have been made, and the most comprehensive evaluation for developed economies (Engel 1999) finds that more than 80 percent of the variation in the RER is due to variation in tradable goods prices. This result is completely contrary to the prediction that variations in [RER.sub.N/T] are due to variations in [P.sub.N]. Besides data availability, this is another argument for using the easily available ratio of price levels to measure the RER.

There have been several attempts to capture the evolution of the equilibrium real exchange rate ([RER.sup.*]) and its dependence on real per capita income (Y). Broadly, three different methods can be outlined. First is the direct method of estimating the relationship between [RER.sup.*] and Y; a functional form relates RER and Y, and the fitted value measures [RER.sup.*]. The indirect method postulates that [RER.sup.*] be estimated in terms of its determinants--for example, terms of trade, share of government expenditure, and net foreign capital inflows (see Razin and Collins 1997).

The third approach was pioneered by Williamson (1994), who argued in favor of estimating the Fundamental Equilibrium Exchange Rate for each country. The equilibrium exchange rate was defined to be the rate that would achieve a zero external balance, but was later modified to be one that would achieve a "target" level of the current account deficit. The FEER has had several variants for example, BEER, where B is behavioral, and PEER, where P is permanent (see IMF 2007).

Once the functional dependence of [RER.sup.*] on per capita income has been estimated, it is then straightforward to derive the undervaluation (UV) of a currency at any point in time:

(5) [RER.sup.*] = fitted function of Y, other determinants, and

(6) UV = 100 log (RER/[RER.sup.*]).

When UV is negative, RER is below its equilibrium value [RER.sup.*], and the real exchange rate is deemed to be undervalued; or equivalently; the nominal exchange rate would need to appreciate to bring about equilibrium. When UV is positive, the real exchange rate is considered overvalued, and the nominal exchange rate would need to depreciate to bring about equilibrium.

The most common approach to estimating [RER.sup.*] is via the log-log equation (7):

(7) rer = a + by + u,

where rer and y are the log values of RER and Y, and u is the error...