The global financial system seems to have a black hole at its centre. Over the last two decades, US residents have sold a total of about $5,500bn worth of IOUs to foreigners, yet the officially recorded net investment position of the US has deteriorated only by a little more than half of this amount ($2,800bn). The US capital market seems to have acted like a black hole for investors from the rest of world in which $2,700bn vanished from sight - or at least from the official statistics.

How can $2,700bn disappear?

It is often argued that the US can simply make large capital gains on its gross positions because its assets are denominated in foreign currency and its liabilities in dollars. However, the available data indicate that over the last two decades this factor has netted the US at most $300bn-$400bn. This still leaves a loss of well over $2,000bn to be explained.

The explanation comes in two tranches of about $1,000bn each.

I.
The first source of accounting revenues for the US derives from an anomaly in the item "reinvested earnings" on foreign direct investment in the US balance of payments. This item improves the current account by about $100bn a year because foreign companies systematically report abnormally low profits for their US operations to avoid US corporate income taxes. If one assumes that foreign companies earn the same rate of return on their direct investment in the US as on their portfolio investment in equity, the US current account would deteriorate by about $100bn. Properly measured, the country's current account deficit would thus be about 1 per cent of gross domestic product larger than officially reported.

The underreporting of the current account deficit implies that US indebtedness is also underestimated. Over the past two decades the cumulative correction for the anomaly in "reinvested earnings" would lead to a higher US net debtor position of about $1,000bn.

II.
A second source of gains comes from very large residuals - labelled "other changes" by the Bureau of Economic Analysis in its statistics on the evolution of the net US international investment position - the total of which also reaches about $1,000bn over the past two decades.

The US international investment position today should in principle be equal to the sum of past current account balances (mostly deficits). However, this is not the case by far, even taking into account the balancing item "errors and omissions". The quite detailed data available for the period 1989-2004 show that the exchange rate and valuation adjustments mentioned above have netted the US $300m-$400m, still leaving a discrepancy of around $1,000bn.

The discrepancy arises for a simple reason: the current account data are based on actual flows of payments recorded in the balance of payments. By contrast, the data on the US international investment position are based on surveys of depository institutions, which year after year tend to lose sight of US assets held by foreigners, especially portfolio investment and real estate.

Could the discrepancy be due to inaccurate statistics on the balance of payments? This is unlikely because the financial flows are just the mirror image of the current account which can be accurately measured given that it consists mostly of business-to-business transactions. With the improvement in current account statistics, the global current account balance discrepancy has now almost totally disappeared. If the current account figures constitute a more reliable source (except for "reinvested earnings"), it is likely that the true US net external debtor position is around $4,000bn (about 40 per cent of GDP) rather than the $2,500bn reported officially for end-2004. Taking into account the current account deficit of about $800bn for 2005 would bring the [true] net current US debtor position to more than $4,500bn. (The official US net international investment position as of the end of 2005, to be published soon, is likely to again include a significant write-down of foreign assets in the US, so the official data are likely to show a net indebtedness below $3,000bn.)

[So this tranche is $4,500bn - $3,000bn or $1,500bn.]

Conclusion.

A closer look at the data thus suggests that both the current account deficit and the net debtor position of the US are even worse than officially reported. This can only mean that the need for a substantial depreciation of the dollar and/or a period of sub-par growth is even bigger than generally accepted.

The writer is director of the Centre for European Policy Studies in Brussels and chairman of the board of San Paolo IMI Asset Management. For more details see CEPS working documents nos 242 and 243 (http://shop.ceps.be)