The US economy is outperforming the eurozone. Official figures next week are likely to show that industrial production in the US is growing nicely, while that in the eurozone has fallen since the winter.
Common sense says this should strengthen the US dollar. Faster growth means more chance of higher interest rates – the Fed could raise them next week – so traders should buy the dollar in anticipation of higher income. And more US economic activity means more demand for US dollars, and of course higher demand should mean higher prices.
If this seems obvious, it shouldn’t be. The facts tell a different story. Since 1999 there has actually been no correlation between differences in annual growth between the US and eurozone (as measured by industrial production) and changes in the €/$ rate.
Yes, sometimes the dollar has fallen when US growth has been weaker than the eurozone’s – such as in 2006-07 – and has strengthened when US growth has outstripped the eurozone’s, as happened in 2008-09 and 2011-12. Just as often, though, the €/$ rate has moved in the opposite direction from the growth differential. For example, the dollar fell in 2002-04 while the US outperformed the eurozone but rose in 2010 and 2015 when US growth was weaker than the eurozone’s.
Common sense, then, is wrong. Why?
One reason is simply that traders sometimes anticipate economic growth – at least outside of recessionary periods – and so price it into currencies before it actually happens.
Partly for this reason, what we have here is an example of something first pointed out back in 1983 by Ken Rogoff and Richard Meese, and which has remained mostly true since. It’s that conventional economic models do a lousy job of explaining and predicting exchange rate moves. Common sense and economic theory says that things such as interest rates, growth and money demand should drive exchange rates. But, in fact, this is only sometimes the case. Occasionally, there are relationships between some economic fundamentals and exchange rate moves, but these are often only temporary.
This doesn’t necessarily mean that exchange rates are always unpredictable. Northwestern University’s Sergio Rebelo and colleagues have shown that they often overshoot, with the result that the real exchange rate (that is, the exchange rate adjusted for differences in consumer prices) mean-reverts over longish periods: a high real exchange rate leads to a fall in the nominal rate, and a low real rate to a rising nominal rate.
This is simple enough. But things get simpler. Because relative prices in developed economies don’t vary very much, this implies that the level of the nominal exchange rate predicts its subsequent changes, at least over medium-term horizons. Sure enough, since 1999 there has been a correlation of minus 0.33 between the €/$ rate and changes in it over the following 12 months. This might not seem very high. But it’s much better than the correlation between growth rates and the change in the exchange rate.
Sadly, though, the rate is now close to its long-run average, which means this fact doesn’t tell us where the dollar is heading. Very often, we just cannot predict exchange rates. Except in rare cases, retail investors should not try to do so.
