Salem Witch Trials

The need to understand the mechanics of exchange rates and their developments has generated a

vast theoretical and empirical literature. The flexible price monetary model, which subsequently

gave way to the overshooting or sticky-price model, the equilibrium and liquidity models as well as

the portfolio balance approach have characterised three decades of research, from the 1960s to the

1980s. More recently, since the publication of Obstfeld and Rogoff's (1995) seminal "redux" paper,

the new open-economy macroeconomics has attempted to explain exchange rate developments in

the context of dynamic general equilibrium models that incorporate imperfect competition and

nominal rigidities. Empirically, these theoretical developments have fared poorly at explaining

exchange rate dynamics, at least over relatively short horizons, and several exchange rate puzzles

have been highlighted.

The increasing role played by international financial markets in developed economies

constitutes a persuasive argument to explore possible relationships between returns on risky assets

and exchange rates dynamics. Recently, a new strand of research has investigated the

interconnections between equity and bond returns, on one side, and exchange rate dynamics, on the

other side, with promising results (see Brandt et al., 2001; Pavlova and Rigobon, 2003; Hau and

Rey, 2004 and 2005).

In this paper, by employing the Lucas' (1982) consumption economy model, we introduce a

new framework explaining exchange rate dynamics. We propose an arbitrage relationship between

expected exchange rate changes and differentials in expected equity returns of two economies.

Specifically, if expected returns on a certain equity market are higher than those obtainable from

another market, the currency associated with the market that offers higher returns is expected to

depreciate. A resident in the market which offers higher expected returns suffers a loss when

investing abroad, and therefore she has to be compensated by the expected capital gain that occurs

when the foreign currency appreciates. This ensures that no sure opportunities for unbounded

profits exist and, therefore, the equilibrium is re-established. Due to the similarity with the

Uncovered Interest Parity (UIP) condition, the equilibrium hypothesis proposed and tested here is

baptised Uncovered Equity Return Parity condition (URP). There is, however, a key difference

between the two arbitrage relations. In the UIP return differentials are known ex ante, since they are

computed on risk-free assets, while in the URP are not.

Risk-averse agents investing in risky assets denominated in a foreign currency usually

require a market and a foreign exchange risk premium, which can be time varying. We begin our

study assuming that investors are risk-neutral, which implies that the URP does not include risk

5

ECB

Working Paper Series No. 529

September 2005

premia. Next, we relax the risk-neutrality assumption and we enrich the URP by employing

additional financial variables, which are related to the business cycle. We use differentials in

corporate earnings' growth rates, short-term interest rate changes, and annual inflation rates, as well

as net equity flows. In line with previous studies (see, for instance, Fama and French, 1989; Chen,

1991; and Ferson and Harvey, 1991b), these variables can be thought of as proxies for the risk

premia. The URP with risk premia turns out to explain a large fraction of the variability of the

European currencies, particularly of the euro, the British pound and the Swiss franc.

We also test the forecasting ability of the URP with time-varying risk premia and find that it

beats the naïve random walk model with drift at two- and three-month ahead forecasts and always

predicts about two thirds of directional changes for the EUR/USD exchange rate. The forecasting

performances for the Swiss Franc are better in terms of mean square errors, but not in terms of the

sign tests. The specification for the British pound does not beat the naïve random walk model,

despite it predicts correctly more than 60% of its directional changes.

Similar specifications fail in explaining the evolution of the Japanese Yen and the Canadian

dollar against the US dollar, possibly due to the systematic intervention policy of these countries in

the foreign exchange market.

6

ECB

Working Paper Series No. 529

September 2005

"My experience is that exchange markets have become so efficient that

virtually all relevant information is embedded almost instantaneously in

exchange rates to the point that anticipating movements in major currencies is

rarely possible..... To my knowledge, no model projecting directional

movements