A significant depreciation of the dollar, predicted by economists for years, was finally realised during the last twelve months. Yet short-term exchange rate movements are notoriously difficult to forecast. How can we resolve the tension between economists being ignorant in the short run and right in the long run?
A simple (and snarky) answer would be that any prediction about exchange rates will eventually come true if you wait long enough – if the euro-dollar exchange rate is a random walk, then sometime in the future there will be a period during which the dollar declines. Economists who predict a drop in the dollar without specifying a timeframe can always wait a bit longer to be inevitably right.
But that explanation is too cynical. There are good reasons for the difference in our ability to forecast exchange rates in the short run and the long run. They lie in unobserved variables, expectations, and the role of economic fundamentals. We know our economic destination, even if the exact route of the journey is impossible to specify.
The short-run random walk
It is almost impossible to predict exchange rates in the short run. Forecasts delivered by macroeconomic exchange-rate models are generally found to be less accurate than the random walk, i.e. forecasts based on the idea that the exchange rate tomorrow will be just the same as it is today. This famous result of the early 1980s has been persuasively persistent.1
For example, consider the euro-dollar exchange rate’s behaviour in response to US monetary decisions since the beginning of the subprime crisis. On 18 September 2007, the Fed announced its first interest-rate cut – by 50 basis points, from 5.25 to 4.75%. During the days preceding the FOMC meeting, the dollar had depreciated sharply against the euro because markets already expected a monetary easing and hence a fall in the return to dollar assets. After the meeting, the dollar continued to depreciate, but more smoothly (Figure 1). Several rate cuts followed during the fall and winter, cumulating 175 basis points.
Figure 1. The euro-dollar exchange rate in September 2007
On 18 March 2008, the FOMC again cut the Fed Funds rate, from 3 to 2 ¼ percent.2 Markets anticipated this move, so the dollar depreciated prior to the meeting. However, market analysts did not expect such a large cut (75 basis points). Based on the standard portfolio-choice theory of exchange-rate determination, the dollar should have depreciated after the meeting because of the lower remuneration on dollar-denominated assets. In fact, the dollar appreciated, due to market expectations that the vigorous policy decision would likely successfully dismiss the risk of a deep economic crisis (Figure 2).
Figure 2. The euro-dollar exchange rate in March 2008
Econometric techniques are ill prepared to tackle such diverse reactions to the same event. In the standard exchange rate model, a 75 point cut in the Federal Funds rate should trigger a larger depreciation of the dollar than a 50 point cut. But the exchange rate depends upon unobserved market expectations, which are difficult to capture in these models, as illustrated above.
The long-run destination
The prolonged weakness of the dollar has long been anticipated. In the late 1990s, most economists were already worried about the US current-account deficit, although some claimed that, for various reasons, a large deficit was sustainable in the long run. At least, market economists, surveyed by Consensus Forecasts in London, increasingly expected that the dollar would depreciate against other currencies, including the euro. In fact, the weakness of the euro in its early years came as a surprise.
How was the dollar’s weakness forecast long before its arrival when exchange rates are unpredictable? The reason is that, paradoxically, it is less difficult to predict long-run than short-run evolutions, because long-run trends are cleaned up from noise and bubbles that are especially difficult to model, and because no exact timing is asked for when looking at “long run” forecasts.
What is the “long run”?
The “long run” can itself be decomposed into three successive time horizons. In the “very long run”, the real exchange rate (the relative price of the same basket of goods across two countries) is expected to come back to a constant level, due to the convergence of the prices of both traded goods (due to international arbitrage) and non-traded goods (due to productivity equalisation all over the world). In the jargon, this is called purchasing power parity, and it acts as a very long run attractor.
In the “long run”, the net foreign asset (NFA) position of a country is expected to stabilise in percentage of GDP at some equilibrium level that can be related to the demographic patterns and the level of development of countries. The stability of the NFA-to-GDP ratio puts some constraint on the trade account. The trade account must be positive if the NFA ratio is negative, to compensate for net interest payments to the rest of the world and keep the NFA ratio constant. Since a positive trade account requires a weak currency, a negative NFA ratio is associated with a weak currency in the “long run”.
Note that a convergence of demographic structures and development levels across countries translates into a convergence in NFA-to-GDP ratios in the “very long run”. Since all NFA positions should sum to zero worldwide, this means that NFA ratios as well as trade accounts are balanced in the very long run, which is consistent with purchasing power parity.
In the “medium-long run”, NFA positions need to adjust to their equilibrium levels through current account adjustments as well as valuation effects (i.e. the potential large wealth transfers across countries due to asset and currency price fluctuations). With relatively large adjustments needed and limited sensitivity of trade flows to relative prices, large exchange rate adjustments are expected to take place in the “medium-long run”. When the NFA-to-GDP ratio has converged to its equilibrium value, the current account no longer needs to be different from zero and the “long run” value is then reached.
Where is the euro-dollar exchange rate headed?
What does this analysis mean for the euro, the dollar, and their exchange rate? In recent work (Bénassy-Quéré, Béreau and Mignon (2008)), we argue that the euro and the dollar both seem overvalued against trade-weighted basket of currencies based on any of the three definitions of the long run. With relatively optimistic assumptions concerning valuation effects and the reaction of trade flows to exchange rates, the euro is found to be overvalued in bilateral terms against the dollar: the “medium-long run” equilibrium rate is around 1.30-1.40, while today’s exchange rate is around 1.60. With more pessimistic assumptions, the euro still needs to appreciate against the dollar. However, in the long and very-long run, the equilibrium value of the euro is found much lower – around 1.10. Thus, our work suggests that the euro may have peaked and be due to fall.
Still, the dollar appears overvalued against a number of other currencies, especially in Asia. The development of international securities denominated Asian currencies is a pre-condition for international markets to stop considering the euro as the only alternative to a weakened dollar. By triggering faster appreciation of Asian currencies, this may speed up NFA adjustments and reduce the cost of the rise and fall for the Euro area.
Bénassy-Quéré, A., Béreau, S. and V. Mignon (2008), “Equilibrium Exchange Rates: a Guidebook for the Euro-Dollar Rate”, CEPII working paper 2008-02, March.
Cheung, Y.W., Chinn M.D. and A.C. Garcia-Pascual (2005), “Empirical exchange rate models in the nineties: are any fit to survive?” Journal of International Money and Finance, 24, 1150-1175.
Meese, R.A. and K. Rogoff (1983), “Empirical models of the seventies : do they fit out of sample?” Journal of International Economics 14, 3-243.
1 See Meese, R.A. and K. Rogoff (1983), Cheung, Y.W., Chinn M.D. and A.C. Garcia-Pascual (2005).
2 The last rate cut of this round took place on April 30 (25 basis points, to 2 percent).