It's the job of watchdogs to bark, and according to reports the International Monetary Fund will use its new economic forecasts this week to warn about the threat posed to growth and stability by global economic imbalances. This is hardly a new concern, with the United States having had a large and growing current account deficit since the mid-1990s, but the IMF thinks high energy prices are exacerbating the pressures and "heightening the risk of a sudden disorderly adjustment". It calculates that higher oil prices directly explain half the worsening of the US deficit in the latest three years, while the recycling of petrodollars is indirectly making matters worse by reducing world interest rates and thus prompting American consumers to carry on spending like there's no tomorrow.

Earlier episodes of high oil prices set ominous precedents, of course. The imbalances which built up in the late 1970s paved the way for the Third World debt crisis of the early Eighties. The lesser oil price hike in [1989-90] unwound in a bond market bubble which burst in 1994. It clearly has been difficult in the past for world financial markets to absorb and recycle such large sums in a smooth and orderly way. But are we doomed to a repetition of history?

This is the majority view among economists and commentators. But two Harvard economists, Ricardo Hausman and Federico Sturze-negger, have challenged - and enormously irritated - the doomsters with the intriguing suggestion that the current account deficit isn't nearly as large as the official figures suggest. Their starting point is the observation that the US has had a deficit in every year but one since 1983, and while it has been deteriorating sharply, it's hard to characterise a 20-year phenomenon as unsustainable. They also note that consistent and rising trade deficits imply that America's net holdings of foreign assets should have been falling (as the figures indicate) and therefore America's earnings on its overseas assets should have been falling too. But in fact the figures show those earnings have been fairly constant.

Figures from the Bureau of Economic Analysis show net foreign assets were $365bn in 1980, delivering a net return of $30bn. Between 1980 and 2004 the current account deficit accumulated to $4.5 trillion dollars. Net foreign assets declined to minus $4.1 trillion. Simple arithmetic suggests that net foreign earnings should have been negative too (say minus $210bn at a 5 per cent rate of return) but in fact the figure for 2004 was still plus $30bn. The two economists write: "Isn't there something misleading about calling a country which makes money on its financial position the world's largest debtor?"

Balance of payments statistics are fraught with accounting difficulties and judgements. For one thing, the global figures show the world to be in an overall deficit, so unless there's unnoticed trade with Mars something is obviously being omitted from the statistics somewhere. The current account is equal (as an accounting identity) to the trade balance plus the interest earned on net foreign assets, which in turn should equal the change in the stock of net foreign assets. The obvious ways to make the official US figures satisfy these accounting identities are to assume either that there have been huge capital gains on the US stock of overseas assets, or that US investors overseas have consistently earned higher rates of return than foreign investors in the United States.

Hausman and Sturze-negger take a different tack, arguing that America has a large amount of intangible and unmeasured overseas assets - the "dark matter" (by analogy with the unseen dark matter in space which astronomers have to postulate to explain gravity). They suppose that US overseas assets can be valued on the basis of their returns, just as properties are valued on the basis of their rentals, or companies on the basis of their earnings. Assume for the purposes of illustration that the overseas rate of return to US investors is an average 5 per cent. This implies a price-earnings ratio of 20. Multiplying the recorded $30bn of foreign earnings in 2004 by 20 implies the US is a net creditor to the tune of $600bn, in contrast to the official figures showing it to be a net debtor to the tune of $2.5 trillion. The difference of $3.1 trillion between the two is the dark matter. Calculating the annual change in net foreign assets on the same basis for the past 20 years gives an alternative measure of the current account deficit which has been roughly in balance all that time. The first chart shows the contrast between the dark matter current account and the official measure.

Britain happens to be in the same position. While official figures show the current account to be in a large deficit in recent years, the dark matter figures indicate a large surplus. For 2004 they are minus $42bn and plus $173bn respectively.

So what is dark matter? And is it sustainable - if real, does it eliminate the need to worry about global imbalances? It isn't difficult to think of intangible assets which are not well measured by accountants. There is also quite a range of evidence indicating that US-owned firms are more productive than their local equivalents. The reason is widely thought to be their intangible know-how and better management techniques, in addition to other intangibles such as brand value. It seems quite likely therefore that the value of US direct investment overseas is understated to some extent in the statistics.

However, although there is likely to be some intangible value, it is also likely to be somewhat fragile. To the extent that the value of a business rests on brand, reputation and the expertise of its managers, it can evaporate overnight; we have seen prominent examples in recent times of large companies becoming almost worthless. So even believers in dark matter can't afford to be completely relaxed about the US current account.

The dark matter approach has also unleashed a storm of controversy among economists. In a recent paper, the Federal Reserve Board of New York admits there is a "puzzle" in the fact that US companies earn a much higher rate of return on foreign direct investment overseas than do foreign firms operating in the US. Nevertheless, it argues that the deterioration in the US net asset position has taken a long time to feed through to earnings from foreign direct investment, but is about to do so big time. This is one of the main contentions of the dark matter critics.

So is the IMF right to sound its new warning about global imbalances? Surely the truth lies somewhere between the radical dark matter vision and the gloom-laden warnings of imminent catastrophe. Hausman and Sturzenegger are right to say a 20-year "unsustainable" deficit needs explanation, and this is likely to lie in the productivity of US companies which invest overseas. But the pace at which the US current account deficit is deteriorating, exacerbated by energy prices, is probably not sustainable.

Diane Coyle is the founder of Enlightenment Economics. Stephen King is away