> bonds is overcounted; reported in the WSJ 4/16 "US Global Debt
> May Be Smaller, Fed Study Finds." Hence the Current Account
> debt is less than figured.
>
> The report is at
> http://www.federalreserve.gov/pubs/ifdp/2002/722/ifdp722.pdf
>
> Any significance? Anybody?
(A long post, but it's a crucial subject.)
Their main point, as far as I can tell (the devil is always in the details) is that the statistics on cross-border flows of capital are a bit unreliable due to the existence of offshore financial centers like Hong Kong and the Caymans, and that the transactions data (what credit instruments get bought or sold per month) doesn't match with survey results of asset holdings. They run monthly transactions data through a rather arbitrary set of valuation adjustments, which has the happy end result of rendering the US slightly less bankrupt than the Federal Reserve statistics say it is, i.e. the official data supposedly undercounts US ownership of foreign stocks and somehow overcounts foreign purchases of US bonds, to an unknown degree.
There are two problems with this: monthly transaction data, unless it's accurate to a whole bunch of decimal points, which it never is in the wide, wide world of multinational economics, is never comparable with annual benchmark surveys. Example: in 1998, the Japanese yen hit 140 to the dollar, all sorts of buying and selling occured at that level, and then went to 110 to the dollar -- all this resulted in all sorts of zany monthly flows into and out of offshore accounts, not all of which are well-reported. Throw in bankruptcies, write-offs, and accounting changes, and incompatible national accounting regulations (e.g. what's an asset, what's a liability, market value vs. purchased value, depreciation, you name it, all are a little different)... you get the idea.
Second, their use of benchmark surveys of custodial assets, held by large banks or other financial institutions, doesn't capture churn -- money flying between institutions -- nor does it capture the underground economy, not to mention the complications of securitized assets, which combine revenue streams from all sorts of different instruments. I suspect what they're really capturing is nothing more dramatic than the fact that the US financial system is relatively well documented, whereas asset holdings abroad are not as well documented or even documentable, for various reasons (i.e. US multinationals who like to park their money overseas, and invest in Singapore rather than Detroit; rentiers stashing millions in the Caymans; developmental states who keep a tight lid on domestic finance, etc.). This may explain why their system accurately predicts foreign purchases of US stocks and US purchases of overseas bonds, and why certain instruments in the Eurozone seem to tally with the benchmarks: these are more transparent financial systems anyway.
This isn't to minimize the genuine problems with the data by any means; more and better reporting would be nice. It'll be interesting to see what Eurostat's take on this is (they've started to report net international investment figures for the eurozone).
-- Dennis