One of the approaches to determine or predict curency exchange rates is that involving the analysis of interest rate differentials (the Interest Rate Approach). This involves a number of different principles and we shall go through them briefly and in turn. The first principle involves the basic interest rate parity theory, which is that:

An exchange rate’s forward % premium/discount = its interest rate differential

Thus, for instance, the traditional forward discount on the dollar–yen exchange rate should equal the interest rate differential between the two currencies. This is seen as the equilibrium reflecting the relationship between the exchange rate and interest rates. Because forwards are a traded instrument and thus subject to supply and demand, the forward premium or discount can vary briefly from this equilibrium, but should always revert to norm. After all, if for argument’s sake the forward premium/discount for some reason did not equal the interest rate differential between the two currencies an arbitrageur could in theory make risk-free profits by borrowing in one currency, investing in the securities of the other currency and simultaneously opening a forward contract in the exchange rate for the same period as the initial loan. This is called covered interest rate arbitrage.

The theory of interest rate parity is a guiding principle for several economic and financial models. Under this theory, it is assumed that the expected (interest rate) returns of a currency should be equalized through speculation in another country once converted back to the first currency. This may sound like gibberish, but basically this is an interest rate version of PPP— and like PPP its results are decidedly mixed. Indeed, there can be significant violations of the interest rate parity theory for substantial periods of time without the immediate reversal that covered interest rate arbitrage might suggest. Not too surprisingly, this is a dismal predictor of exchange rates.

Indeed, before we go further into the theory, it is important to point out a practical flaw in the theory involving incentive, which is undoubtedly a key contributing factor to its poor predictive track record—the theory supposes an automatically causal relationship between interest rates and the exchange rate, yet in practice most currency market practitioners trade currencies with directional rather than interest rate considerations in mind. Even this statement is a generalization. On a simple numerical basis, the majority of currency market practitioners are made up of interbank dealers, thus it is important and necessary to look at their motivation for trading. Spot traders for the most part care not one whit about a currency’s interest rate, in part because they hold positions for too short a time for it to matter, in part because they are seeking to predict direction—and thus make capital gains on their position, not primarily to make interest income. Forward traders are a different breed entirely and more akin to money market or interest rate traders. Indeed, theway they hedge out their forward exposure frequently involves an array of interest rate-related instruments. Eventually, interest rate parity violations will be reversed, but there is little incentive to do so in the immediate term if you don’t care about the interest rate in the first place.

Returning to the theory for now, interest rate parity theory states that the difference between a spot and forward exchange rate expressed as a percentage should equal the interest rate differential between the two currencies. Yet, we know from the PPP principle that exchange rates and inflation rates are linked. Can we not link these also with interest rates? Indeed we can, thanks to the seminal work of the economist Irving Fisher. Thus, according to what has become known as the “Fisher effect:”

The difference in interest rates = the difference in expected inflation rates

Thus, we have gone from the difference between the spot and the forward exchange rate equating to the interest rate differential through the interest rate parity theory, which in turn equates to the difference in expected inflation rates through the Fisher effect. Yet, PPP tells us that absolute or relative price growth levels can be used to forecast future exchange rates.

Thus, through PPP we can extrapolate this one stage further to suggest that:

The difference in expected inflation rates = the expected exchange rate change

Bringing all these together, we get: (1) The difference in spot and forward rates = the difference in interest rates (Interest rate parity theory)

(2) The difference in interest rates = the difference in expected inflation rates (Fisher effect)

(3) The difference in expected inflation rates = the expected change in spot exchange rate (Purchasing Power Parity)

Logically from this, one may conclude that the difference between the spot and forward rates expressed as a percentage should equal the expected change in the spot exchange rate. This is known as the expectations theory of exchange rates.

Finally, there is the theory that: (4) The difference in interest rates = the expected change in the spot exchange rate (International Fisher effect)

On the face of it, the ideas presented above seem logical and follow a clear and persuasive train of thought. There is only one small problem—this clear train of thought rarely works in practice. More specifically, the difference in interest rates or expected inflation rates may be equal to the theoretical construct of the “expected change in the spot exchange rate”, but in practice it is of the future exchange rate. In line with this, the forward rate is also a very poor predictor of the future exchange rate, a fact that economists have labelled “forward rate bias” or the “forward premium puzzle”. As Bansal and Dahlquist confirmed in their exhaustive study, in contrast to the theory, empirical evidence suggests that in fact current interest rate differentials and future spot exchange rates are frequently negatively correlated. This is particularly the case within the developed economies, though the picture is somewhat more mixed within emerging market economies.

Over the long term, the interest rate parity theory is seen to work as enough market participants can be found to “discover” the opportunities available for covered interest rate arbitrage between currencies and interest rates, thus in the process eliminating such disparities. However, there are much longer lags than the theory might suggest is possible. Here again, the issue of incentive must be a focus. As noted earlier, it should behove the theorists to know that the majority of currency market practitioners are currency interbank dealers and moreover that the main incentive for these to trade is directional gain rather than interest income. Currency markets do focus on interest rate differentials for extended periods of time, but equally they focus on other factors, in many cases completely disregarding interest rates.

Real Interest Rate Differentials and Exchange Rates

Currency strategists do however use models comparing the real interest rate differential with either the nominal or the real exchange rate between two countries. The logic behind this relates to both the international Fisher effect and to PPP, where on the one hand the difference in interest rates should, if not be exactly equal to an expected change in the spot exchange rate, at least be an important driver of it, and on the other hand where nominal interest rate differentials are adjusted for inflation (i.e. domestic price growth) and thus relate to the exchange rate through the law of one price.

The link or correlation between real interest rate differentials and the exchange rate appears to have grown exactly in line with the gradual move since the end of the Bretton Woods exchange rate system to liberalize capital flows globally. As barriers to capital movement have fallen, so the overall importance of capital flow has grown exponentially relative to that of trade flow. Exchange rate models that focused solely on the current account no longer seemed appropriate in such a world, those that focused on capital flows seemed increasingly so. As capital flows have gained in importance, so their importance within overall currency market flows has grown and thus the correlation between the two increased. Thus, currency strategists across the market continue to track this relationship between real interest rate differentials and nominal exchange rates as one of many useful and important indicators of currency over- or undervaluation. Figure 1 compares the Euro–dollar exchange rate against the 10–year bond yield differential from 1996 through to the end of January, 2002.

The traditional exchange rate models, which are based on some form of equilibrium value, do offer an important and useful long-term guide towards exchange rate prediction. Indeed, without these long-term signposts, currency strategists might be quite lost in seeking to predict exchange rates past one year out. As a result, in terms of their usefulness to specific types of currency market practitioner, corporations, strategic “real money” investors or “macro” hedge funds with a multi-month or even multi-year perspective would probably find them most valuable as an analytical tool. On the other hand, these traditional exchange rate models are unlikely to be of more than passing interest or use to short-term speculators or interbank dealers whose time frame is far shorter.

The apparent weaknesses of traditional exchange rate models, I would suggest, adds to this case that such an integrated currency strategy framework be adopted. It has to be said that to date, when faced with the unsatisfying results that the traditional exchange rate models have produced as far as predicting exchange rates is concerned, the economic community has for the most part either ignored these inconvenient results or declared that it is impossible to forecast short-term exchange rate moves as they are determined by the so-called “random walk” theory. Occasionally, there has been a paper, illuminating in both its honesty as well as its intellectual acumen, which has “fessed up” to both the failure of these models as predictive tools and a lack of understanding as to why that may be the case. The majority of the time, however, the reaction to the obvious question has been denial or the random walk excuse. According to the latter, since traditional exchange rate models do not appear able to predict short-term exchange rate moves, it must follow logically that such short-term exchange rate moves cannot in fact be predicted at all and must therefore follow a “random walk” path, suggesting an equal probability of appreciating or depreciating over time. Fortunately, however, recent developments in technical and capital flow analysis have achieved significantly better results in predicting exchange rates than the random walk would imply. Thus, the correct approach to analysing and predicting exchange rates would seem to be to use market-based approaches such as technical and flow analysis for short-term exchange rate moves and the traditional exchange rate models for medium- to long-term predictions. This is certainly not the whole story in trying to create an integrated framework for analysing currencies, but it forms a good start in our understanding of how we should approach exchange rates.

Building on this, going forward, it seems logical to assume that traditional exchange rate models should be modified to suit the modern structure of currency market flows. More specifically, trade flows, which form the premise behind the PPP, Balance of Payments and External Balance Approaches, were once seen as the main driver of currency market overall flow. However, nowadays, they make up only around 1–2% of the USD1.2 trillion in daily volume going through the currency market. Hence, as the overall importance of trade to total market flow has declined, so to a degree has the relevance of those exchange rate models that rely solely on shifts in trade flow patterns. Meanwhile, just as the pre-eminence of trade flows has declined, so the importance of portfolio flows has grown exponentially as barriers to capital have been lifted over the past two decades. The Portfolio Balance Approach is clearly an attempt to focus on asset markets and specifically the bond market as a driver of exchange rates, yet this model remains unsatisfactory as a predictor of exchange rates for the reasons given. In order to try to get to a better answer of exchange rate movement over the short term, we have to define the main flow drivers of exchange rates:

1. “Speculative” flow (without an underlying attached asset)
2. Equity flow
3. Fixed income flow
4. Direct investment flow
5. Trade flow

By far, speculative flow is the main driver of exchange rates over the short term. It is not sufficient to suggest that speculative flows follow a “random walk” for the simple reason that both technical and flow analysis have discovered consistent patterns in short-term exchange rates which should not exist under random walk theory. Within asset market flow, equity and fixed income flows continue to do battle for pre-eminence. For instance, from 1998 to mid-2000, net inflows to the US equity markets were a key driver of dollar strength. Equally, as the US equity market began to falter, the resulting equity outflows from the US market weighed on the US dollar. Eventually, however, these flows were more than made up for by fixed income inflows to the US fixed income markets as the Federal Reserve continued to cut interest rates to support the economy. Direct investment is also an increasingly important driver of exchange rates, both in the developed economies and in the emerging markets, as barriers to inward investment have also fallen away. In 2001, the top five performing currencies in the world against the USD were the Mexican peso, Polish zloty, Czech koruna, Hungarian forint and Peruvian sol, all of which benefited from substantial direct investment inflows which had a significant impact on their exchange rates.

The importance of all of these flow types continues to fluctuate in line with market trends. What is clear however, is that until there is a specific exchange rate model which focuses on the main flow dynamic of the currency market, namely speculative flow, it is unlikely that exchange rate models in general will be able to improve upon their current accuracy to any significant degree. In the next chapter, this is in fact what we will try and do—to build a simple exchange rate model which focuses on speculative flow. In addition, we also examine how to use “currency economics”, or the bits of economic theory that are relevant to the currency market, in a practical manner for currency forecasting, trading and investing.

January, 2003