The economics profession's record on forecasting exchange rate movements has not exactly been good. Many have given up and now generally accept the notion that exchange rates follow a 'random walk'. This means that the best guess you have about where a certain exchange rate will be tomorrow is where it is today. But every now and again something happens to restore some faith in the textbooks. And this could be one such time.

With the exception of 1991, the US has run a persistent current account deficit since 1982. The cumulative deficit now totals over $5 trillion (£2.5 trillion). Over the past five years, the size of the deficit has grown in US dollar terms, in real terms and as a share of GDP. By the end of last year, it had reached 7% of GDP - almost certainly a record for the US and a highly unusual position for the world's most capital-abundant country to find itself in. As a country's current account deficit over any period of time is equivalent to the change in the value of its net claims on the rest of the world, it is equal to the change in its net foreign assets. Hence, a country with a positive net export balance must be acquiring foreign assets of equal value. Since 1982, the US net foreign asset position has swung from a surplus of 7% of GDP to a deficit of 50%.

Presented with these facts, orthodox economic theory says the dollar must fall. And lo and behold, it has. The trade-weighted dollar, which accounts for the US pattern of trade, has lost around a quarter of its value in both nominal terms, and real, inflation-adjusted, terms. Its decline is similar in scope to the one that occurred between 1985 and 1987, between the Plaza and Louvre Accords, which were designed respectively to bring the dollar's value down and to stop it falling any further.

: IMF, Fatham

So far so good. But there remains a puzzle behind the dollar's recent decline. During the late 1980s, the dollar's decline eventually shaved 3.5 percentage points of GDP from the current account deficit by 1990.

So why is it not yet having an impact on the trade data?

It is true that the deficit narrowed sharply in the final quarter of 2006, but this was largely an oil story. The non-oil deficit was more or less unchanged. There are a number of possible answers to this puzzle.

The first is the economist's old friend, long and variable lags. There is probably some truth in this, but it is a bit of a stretch to argue that this can explain the lack of improvement since 2002. The current account usually responds to real exchange rate changes with a lag of two to three years.

Another possible explanation is that the dollar has fallen against the wrong currencies. Currencies like the Chinese yuan and the Japanese yen have got off more or less scot-free. In fact, the dollar has risen against both in real terms since 2002. Meanwhile currencies like the euro, and latterly sterling, have borne the burden of the dollar's slide. The trade-weighted euro exchange rate has risen by over 20% in nominal terms since 2002. Allowing for relative inflation, it has risen by 23%, taking it to its highest point since 1998, the year before the euro was launched.

However, the EU accounts for only 12% of the total US deficit. By contrast, Asia and the Pacific region accounts for nearly 60% of the total US current account deficit, nearly two-thirds of which is with China. In other words, the deficit happens to be widest with countries against which the dollar has not fallen, or has not been allowed to.

A final alternative is that the historical relationship between the real exchange rate and the current account has changed. The standard model assumes that the US earns the same rate of return on its foreign assets as foreigners earn on their US assets. But the US has earned a higher rate of return on its foreign assets than it has paid out on its liabilities every single quarter since records began in 1960.

It is hard to think of any genuinely convincing economic arguments for why this should be the case. Some have argued that this reflects an innate superiority of US investors, or more plausibly their greater tolerance for risk, which naturally brings its own reward. Alternatively it may simply reflect the dollar's role as reserve currency. Whatever the explanation, if we assume that this difference is sustainable and will not be arbitraged away, it dramatically changes our results. Now we find that in order to maintain the net external asset position at -50% of GDP, the real trade-weighted dollar needs to rise by nearly 2% from its current level. Then again, it might just follow a random walk.

Danny Gabay is director of Fathom Financial Consulting and was previously UK economist for JP Morgan Chase and the Bank of England.www.fathom-consulting.com.