Uncovered Interest Parity in Practice
There are two basic approaches to exchange rate determination (both of which are often combined in models): the goods market approach and the asset market approach.45 While in the goods market approach the idea is that exchange rates are determined basically through the trade of real assets, the asset market approach points to the importance of capital flows. The main feature of the first is purchasing power parity (PPP) and the notion of the second is uncovered interest parity (UIP). The concept of PPP states that the exchange rate equates the national price levels of two countries in the sense that the purchasing power of a unit of currency is the same in both countries.46 On the other hand UIP states that the change in the exchange rate over time equals the interest rate differential between two countries. The following chapter is designed to outline the theoretical concept and methods for testing the hypothesis of UIP.
KeywordsExchange Rate Interest Rate Central Bank Sharpe Ratio Foreign Exchange Market
Unable to display preview. Download preview PDF.
- 45.The following draws on Isard (1995), Lewis (1995) and Meredith and Chirm (1998).Google Scholar
- 46.See Sarno and Taylor (2001b). The equalisation of two price levels is termed absolute PPP. In contrast, relative PPP holds when the change in the exchange rate is equal to the difference of the inflation rates between two countries.Google Scholar
- 47.Consider et to be the US$/euro exchange rate: To arbitrage, a US investor can borrow one US$, buy one euro, invest in euro-denominated bonds paying interest of 1 + i* and sell the return forward at rate tt. If the forward discount is smaller than the interest rate differential, the US investor locks in a riskless profit. Thus, covered interest parity must hold. However, if transaction costs are introduced, small deviations of covered interest parity are possible.Google Scholar
- 48.An equal derivation of (3) is the following: (1+it)/(1+it’=Et[et+1]/et subtracting 1 from both sides yields [(1+i t)/(1+i t’)]−[(1+i t’)/(1+i t’)]=[(E t[e t+1]/e t)−(e t/e t)] and hence (i t−i t’)/(1+t’)=(E t[e t+1]−e t)/e t. For small values of i’ this is approximately equal to it−it’≈ln(E[et+1])−ln(et).Google Scholar
- 49.One way to analyse this is by using survey data. Studies using such data (see Froot and Frankel, 1989) find that risk premiums only account for a small proportion of the bias in the interest differential. Expectational errors seem to be much more important. 50 This corresponds with the procedure used among others by Flood and Rose (1996, 2001).Google Scholar
- 51.All data are obtained from Thomson Financial Datastream (Mnemonics: BBAUDSP, BBNZDSP, BBGBPSP, BBCADSP, BBSEKSP, AUST90D, NZTBL3MGoogle Scholar
- 52.Surveys are provided among others by Hodrick (1987), Froot and Thaler (1990), Lewis (1995), Flood and Rose (1996, 2001).Google Scholar
- 53.An interesting result has been detected by Flood and Rose (1996). They find that the forward discount puzzle might depend upon the choice of exchange rate regime. Using data from the European Monetary System they show that UIP performs much better in this system of fixed but adjustable exchange rates than under floating arrangements. However, since our focus is solely on countries that have no official exchange rate target and are classified as independent floaters their result has no relevance for our case.Google Scholar
- 54.From interviews with proprietary traders he reports that they require a Sharpe ratio (i.e. risk-adjusted returns) that is larger than 0,4 (which is the Sharpe ratio of a buy-and-hold strategy in U.S. equities). For the major floating exchange rates this would require that β in equation (5) would need to be either smaller than-1 or larger than +3-well outside a standard deviation band of ±2σ around 1 that a statistician would require to accept the null hypothesis of uncovered interest parity.Google Scholar
- 35.This result is also found by Flood and Rose (1996) who study the performance of uncovered interest parity within fixed exchange rate regimes and conclude that the “[…] deviation from uncovered interest parity, known as the ‘forward discount puzzle’ does not appear to characterise our fixed exchange rate data set.” (p. 751).Google Scholar
- 56.Interestingly, Svensson also refers to foreign exchange intervention in his comment (though he does not interpret them as an additional policy tool for central banks). He asks: “Most macro studies find little or no effect of sterilised intervention. If they are correct, why do central banks keep doing it?” (Svensson, 1996: 300). He suggests that the micro approach to the exchange rate might help to answer this question.Google Scholar