The U.S. Treasury has upped its estimate of China’s holdings of U.S. securities last year, to $1.61 trillion as of June 30th, of which $1.48 trillion was long-term debt, typically Treasury bills with an original term-to-maturity of more than one year. The change is a function of the way the U.S. Treasury collects the statistics as much as anything, of which more later.
The preliminary data, as this revision is known, suggests that any thoughts that Beijing is divesting itself of U.S. financial assets are wide of the mark. Final data is due to be published towards the end of April.
China accounts for one seventh of the foreign holdings of U.S. securities and more than one fifth of the foreign holdings of America’s long-term debt, on the basis of these latest numbers. It has long been believed that the U.S. Treasury undercounts China’s holdings of U.S. securities. Chinese purchases made through international financial centers, notably London, and held by a custodian there, show up under the third country’s ledger, the U.K. in this case. This custody bias, as it is known, is a methodological problem that affects more purchasing countries and financial centers than just China and London. The successive rounds of estimates are intended to minimize it and thus give a truer picture of who really owns what.
The numbers are further distorted by the U.S. Treasury’s way of getting round the time lag in reporting purchases of long-term securities by adding monthly net transactions to the number established at the previous annual report. Most of the trading of long-term securities takes place in international financial centers and netting out transactions between them doesn’t necessarily reflect the nationality of the ultimate buyers and sellers.
A final wrinkle in the numbers is that short-term bill holdings are reported to the U.S. Treasury at face value but the numbers for long-term securities are collected at market value and non-marketable securities at current value. Thus the table of Major Foreign Holders of U.S. Securities is a hybrid of market and face values. (More on the methodology from a 2006 paper that is as good and long now as it was then, here; and from a shorter though no more contemporary note, here.)