Check out Menzie Chinn’s graph over at Econbrowser

In his most recent post, Menzie Chinn plots actual yields on ten-year benchmark Treasury bonds v. the rates predicted by an econometric model based on a series of relatively standard variables (the inflation rate, output gap and projected debt-to-GDP ratio two years out)

The result: starting in 2002 -- and certainly by 2003 -- US interest rates are well below those predicted by the model.    The model expected rates to be above 6% from say 2005 on, and actual rates haven't been anything close to that high.   While Menzie (with Dr. Frankel) finds a bigger gap than some others, the basic result is fairly common -- starting in 2003 or so, traditional models start predicting higher US rates than have been observed.

What else happened about the same time that a large gap emerged between the model's output and the market's output?   Central bank reserve growth went from $100-200b a year to something more like $700b a year, if not more. 

Coincidence?   I don't think so.

Proving it?  Rather hard.

Why?  In part, data difficulties.  From 2002-2004, recorded inflows into the US (and strong recorded demand for Treasuries) matched growth in global reserves quite closely.   But in 2005 those series started to diverge -- largely, I think, because the US data does a far better job of capturing official purchases from Japan than official purchases from the Middle East.    In 2005, overall demand for Treasuries remained robust, but recorded official demand fell sharply.  I think that reflects the undercounting of official purchases from the Middle East --  and some undercounting of Chinese purchases as well (especially from June 2005 on; at the data from the first half of the year has been revised to reflect the results of the Treasury survey -- which showed much larger Chinese purchases than in the TIC data). 

What of 2006?  Overall global reserve growth was quite strong.  And I suspect demand for dollars also picked up relative to 2005.  Recorded official demand for Treasuries rose from its 2005 level - but overall foreign demand for Treasuries clearly fell.   Central banks increasingly shifted toward agencies and even mortgage-backed securities (hopefully not ones too exposed to the subprime market).   And they also increased their offshore dollar deposits -- and bank lending financed by these deposits indirectly supported a lot of different asset prices.

Going forward, at least one large player looks likely to directly support almost all asset classes -- I suspect Chinese official institutions will have roughly $400b to invest in 2007 and they intend to spread the wealth around, so to speak.

All of this is to say that even though the emergence of the gap between the interest rates that Chinn and Frankel forecast and actual rates correlates closely with the emergence of large-scale dollar reserve growth, actually matching the gap to the official US data is difficult.  Especially after 2004, when reserve growth shifted away from Japan and toward the emerging world.    

A final note on RGE's new policy on comments.  As Dr. Roubini made clear last week, commenting on all RGE blogs now requires registration.  Registration requires a valid email address, but once you have registered, it is possible to post comments on the blog anonymously (as has been case). 

Those who register to comment are automatically also registered for a two-week free trial of RGE's subscription service and are added to RGE's mailing list.  After the free trial ends, those who registered will remain able to comment on the blogs even in the absence of a RGE subscription.  It is also possible to take your name off the RGE mailing list (there is an unsubscribe option at the bottom of RGE's twice a week email) and remain registered to comment on the blog.   

Those who indicate as part of the registration process that they are signing up for blog comments only will not be added to the mailing list.  For technical reasons though they are effectively given a trial subscription to the site.

I realize that this processs is a bit cumbersome and the notion of an expired free trial (which still allows blog commenting) is a bit confusing.  It does though offer those who register (temporary) access to my more detailed work on reserves (often done in collaboration with Christian Menegatti).  And if you do opt to take a look, do tell me what you think I have right and what I have wrong.