Statistical manna from heaven
The BEA’s current account data contained two surprises. One that I expected. Another that I did not.
The “expected” surprise was that the 2005 and 2006 data looks to have been revised to show much (and I mean MUCH) higher central bank purchases of US assets. The “unexpected” surprise — at least to me — was that the data on income payments was also revised down quite significantly and a result of the large change in the income balance, both the 2005 and 2006 current account deficit were revised down. The latest data puts the 2006 deficit at close to $810b, well below the $855 in the last BEA release.
The q1 income balance was consistent with the revised 2006 income balance. The ongoing surplus in the income balance offset a portion of the trade deficit, contrary to my expectations. As a result, the overall q1 current account deficit ($193b) was somewhat lower than I expected.
My guess is that both changes reflect, at least in part, the data in the most recent “survey” of foreign portfolio investment in the US. The June 2005 to June 2006 survey showed two things. One was a much larger increase in “official” holdings of US debt than had been reported in the ongoing “TIC data.” and the other was a much smaller overall increase in foreign holdings of US debt than reported in the TIC data.
Daniel Gros of the Center for European Policy Studies has argued that the fact that a certain amount of US debt sold to foreigners “disappears” every year after the survey — he calls this “disappearing into a black hole” — is evidence that the US current account deficit is underreported. Others argue that this “statistical manna from heaven” (See John Kitchen and Bill Cline) shows that the US current account deficit can be sustained for far longer than pessimists think – as ongoing trade deficits simply haven’t produced either a significant deterioration in the United States net international investment position (the broadest measure of the US external position — one that includes equity investments as well as debt) or the United States income balance.
That said, the revisions to the income balance seem to come far more from more interest income on US lending than from lower payments on US borrowing — and above all from higher dividend income (counting reinvested earnings) on US direct investment abroad and lower dividend payments (counting reinvested earnings) on foreign investment in the US. And the changes in the “income” of US direct investment abroad and foreign direct investment in the US do not fall out of the survey.
There is little doubt that the data on “official purchases” were revised quite significantly. The last release for 2006 showed “official purchases” of $200b in 2005 and $300b in 2006. (data here) The revised data puts official purchases at $270b in 2005 and a rather impressive $440b in 2006. That is a big change, but one that is very consistent with my world view. I have consistently argued that the US data tends to understate ongoing central bank inflows to the US.
It is worth noting that the $440b in official financing provided over half the net inflow needed to cover the revised $810b US deficit. The $440, incidentally, likely will be revised up — the q3 and q4 data for 2006 don’t yet reflect the additional official purchases that tend to show up in the survey data.
In q1, central banks were even more generous, providing about $150b in financing ($147.8b) – an annualized pace of close to $600b. That provided about 3/4s of the financing needed to sustain the US external deficit in q1. And if anything, the current data likely still understates official inflows. The revisions to the initial data releases on “official inflows” have consistently revised the data up, and the various measures of global reserve growth that I track suggest that central banks had a ton of cash to lend to the US in q1.
Central bank inflows inq1 incidentally are far stronger now they were than back in 2004, when Japanese financing of the US was very widely discussed in the US media.
Incidentally, US companies invested way more abroad in q1 ($75) and foreign companies invested in the US ($25). That implies that the US needed to sell about $250b in securities to the rest of the world to cover the US current account deficit (a bit under $200b) and the capital outflow that stems from US direct investment abroad.
That is the bad news.
The good news is that the income balance was also revised, in a way that shifted the 2006 income balance from a deficit to a surplus. That reduced the 2006 current account deficit. It also kept the income balance in surplus in q1. Lower payments on foreign direct investment in the US helped — foreign investors continue to get a very raw deal on their investment in the US, at least if you believe the payments reported in the US data.
But the q1 income surplus is generally consistent with the revised data. That is where the big changes came. The amount of income the US received on its investment abroad was revised up by about $30b, from $620b to $650b (with about 1/2 the gain coming from higher returns on direct investment and 1/2 the gain from other source), and the amount the US paid on investment in the US (and borrowing from abroad) fell by about $25b, from $630b to $605b. Most of the fall came from lower returns on direct investment, not from smaller debt payments.
The revisions to the data certainly were not consistent with my expectations. And to be honest, the gap between the return on the reported return on US FDI abroad (decent) and the reported return on foreign direct investment in the US (terrible) strikes me as a bit too big to be plausible. But if the data consistently goes against your expectations, it is also worth looking a bit more closely at your assumptions, and trying to understand the data a bit better. I hope to do so over the next couple of weeks.
To be honest, I think I understand the downward revisions in US interest payments on US debt held abroad — a reflection of the debt that seems to disappear in the Survey — far better than I understand the revisions to the data on the returns of US FDI abroad and the returns on foreign FDI in the US. i remain confident of one thing though. The striking fact is not high returns on US FDI abroad, but rather the very low returns on foreign direct investment in the US.
19 Responses to “Statistical manna from heaven”
Guest • June 16th, 2007 at 4:31 am
Brad – overview commentary on China currency situation from John Mauldin latest, including specific points on April TIC data and the redirection of flows out of Treasuries (which I imagine you’ll post on separately)
bsetser • June 16th, 2007 at 8:39 am
i’ll post on the TIC data on monday. The Chinese shift toward agencies has been ongoing for sometime. in q1, there was more central bank demand for agencies than treasuries — and in 06, the revised data puts agency and treasury demand at both around $200b. i don’t think a shift from treasuries to agencies alone would be enough to explain the recent market moves — there is a treasury v agency arbitrage that other players would likely do.
plus it is clear that the TIC data has understated chinese purchases significantly for the past serveral months. $10b in net purchases is too low v. $37-38b in valuation adjusted reserve growth. I am a bit less in the “be careful what you wish for school” since i rather doubt a gradual RMB appreciation will slow Chinese reserve growth in the short-run, and so long as china wants the appreciation to be gradual is has to keep on buying $ (at least that is what I think). it can shift to the short-end — but it will still want to get interest. And it is hard for me to see how China gets out of the reserve accumulation game/ stops seeing its trade and current account surplus rise without letting the RMB appreciate.
df • June 16th, 2007 at 9:48 am
heavan ?
Guest • June 16th, 2007 at 11:58 am
and “Central bank inflows in q1 incidentally were stronger than bask in 2004…” although spell check wouldn’t have caught that. But given how very late/early this was posted, perhaps not a whole lot of functional proofreaders at that time…
bsetser • June 16th, 2007 at 4:39 pm
remind me again not to post late at night. the spell check doesn’t extent to the title field, which is something of a problem for me, especially if i am rushing. my apologies.
Guest • June 16th, 2007 at 6:23 pm
Good grief – what a sorry lot to be complaining about such trivalities.
They’re lucky to have anything this reliable and good to read on a regular basis.
Keep posting late at night, and spelling minutiae be damned – for those of us who appreciate the substance and your efforts.
Guest • June 16th, 2007 at 6:26 pm
See how I spelled trivialities?
And I’m careful!
See how easy it is, other guests?
Guest • June 17th, 2007 at 9:12 am
the Treasury’s explanation of the ways TNC, and presumably I-bank transactions are categorized in the TIC data is also not clear to me. I’m also starting to wonder how transactions by entities such as the World Bank and IMF may be categorized, assuming they must also hold, buy and sell some of the securities monitored by the TIC data.
Guest • June 17th, 2007 at 11:46 am
as the comment immediately preceeding 2007-06-17 09:12:36 was deleted, twice, the 2007-06-17 09:12:36 entry may not make sense.
Guest • June 17th, 2007 at 12:25 pm
2007-06-16 18:26:09 – as we all know errors and inaccuracies are tough to avoid at the best of times, are you suggesting we ignore them (how does Brad feel about this?) even if such ‘trivialities’ may distort meaning or (presume falsely) indicate a broader disregard for readers or sloppiness in statistical details.
I don’t see anyone ‘attacking’ Brad’s spelling or suggesting that he change his posting habits.
some guidelines as to the ways readers are expected to react to errors, as well as any explanation of why some comments disappear, would be appreciated.
Guest • June 17th, 2007 at 12:58 pm
Having authored 2007-06-16 18:26:09, perhaps I put our host in a slightly awkward position in order to respond gracefully to the headache of such a petty conversation. So I apologize to him first. But I’ll also presume to recommend my own guidelines for those who desperately need such lessons – don’t do it unless the error makes the content not understandable. And if you must do it, be polite and respectful about it. In this case, the meaning was not at all distorted, and our host was the one who alluded to changing his posting habits (which you would know if you actually read and absorbed his comment). Now please drive on and let the man do the productive and enjoyable part of his job.
Julien • June 17th, 2007 at 1:50 pm
B.Setser you might check the xls sheet for full Net direct investment.
- DI outflows : -75.4 (from -$66b Q4)
- DI inflows : +23.4 (from +$45.5b Q4)
So Net Direct Investment tumbles at -51.9 from -$20.5b, it’s fiveteen quarters low (Qtrsa).
is the money leaving this country ?
Guest • June 17th, 2007 at 1:55 pm
if you can let Brad respond and “please drive on and let the man do the productive and enjoyable part of his job.” With all due respect, whether or not you understand the concepts, more lectures from you about politeness and discipline aren’t helping.
Emmanuel • June 17th, 2007 at 5:08 pm
To me, it does the BEA little favor to adjust their figures willy-nilly as it destroys the continuity of figures with past data series. In particular, these revisions that make the US income balance positive from Q4 2005 onwards are mind-boggling and inconsistent with simple calculations of how much interest the US needs to pay on its accumulated debts.
As an aside, there was a whole slew of articles (I have a summary) in BusinessWeek recently touting how labor productivity increases in the US are likely misattributed because they allegedly come from offshoring, not increased labor productivity. This resulted in so-called “phantom GDP”, questioning BEA accounting again.
It seems to me that the BEA needs to do a better job of explaining these statistical sleights of hand to a broader audience. With more and more folks paying attention to SNA and BOP accounting which were previously esoteric topics, the BEA needs to justify the changes it makes better as well as address criticisms of its methods.
Do you believe in magic?
bsetser • June 17th, 2007 at 6:10 pm
I too would like to learn what led the BEA to revise its income series. the revisions to the official financing series come from the survey — i now understand that bit fairly well. but the revisions to the FDI income series are a bit more puzzling — tho as usual the real puzzle is the low level of income reported on foreign direct investment in the us (notably low levels of reinvested earnings). it also seems like the survey discovered additional us holdings of foreign debt (the int. income on us assets abroad was revised up) without additional foreign holdings of US debt. I suspect some of this will be clearer once the survey of current business comes out with the FDI data, and once the NIIP data is released (the niip clearly will not deteriorate in 06, incidentally).
bsetser • June 17th, 2007 at 11:58 pm
why do posts disappear …
hmmm — probably b/c they contain words like c_sinos and g_ming … words that set off our spam filter. P_ker does too.
I would like to see the conversation return to more substantive issues. Guest — i will say that when i first read your corrections, I was a little annoyed (Tho appreciate pointing out the error in the title so it could be corrected quickly, as errors in the title become errors last google stores forever). My proof-reading does often leave a little to be desired — but i was scrambling to get something up on the road, and wanted to post on an interesting data release before it was too old, and was hoping for a bit of leeway …
Guest • June 18th, 2007 at 5:54 am
Perhaps RGE’s filters can also be instructed to proofread, and delete trivialities and obscenities (abbreviated or not)
The link was posted in an attempt to break the chill. And as the whole post had degenerated into trivialities, and as everyone seems to recognize that the data is terrible, and given the great deal of guessing that seems to go into its interpretation and what it may indicate – and as there also seems to be a great deal of speculation and concern about: 1) the ways PE and hedgefunds; 2) the U.S. residential housing downturn may affect global capital flows, as you know the post was, in part, a question about whether recent significant transactions in the g****g industry, which is not unsubstantial, may be worth some attention.
“Penn National G****g Inc., the owner of 18 c*****s and horse-****** tracks, agreed to be taken private by buyout firms Fortress Investment Group LLC and Centerbridge Partners LP for $6.1 billion. Penn National’s revenue quadrupled in the past five years as more U.S. states legalized g*****g, adding s**t machines and table g****s to boost their tax base…” http://www.bloomberg.com/apps/news?pid=20601087&sid=advKIWgQqDNs&refer=home
It’s the ‘little’ things…
Estragon • June 18th, 2007 at 10:29 am
I wonder how much, if any, of the differential in FDI returns arises from corporate tax differentials and transfer pricing issues.
A. P. Simkin • June 21st, 2007 at 4:27 am
I suspect that part of the reason for the revision of the net income figure has to do with the inclusion of the unrepatriated earnings both for U.S. companies and foreign companies in the U.S. I also suspect that foreign companies in the U.S. use a lot of domestically borrowed capital and deduct the interest from their reported earnings. And, as the last guest pointed out, there is the matter of transfer pricing.
Dr. S. I take it you noticed that in April residents of China were net sellers of long-term treasuries, and big buyers of short-term treasuries. Evidently, we can’t tell if this continued in May. Assume it did. My guess is that the motivation was not financial, but rather was Peking’s way of sending a message to Washington: we’ve got each other by the you know whats. Your comments, please.
