The more the dollar falls, the more dollars you need to buy …

That, more or less, is the conclusion of “portfolio balance” models of global capital flows.    

The dollar tanks.   Its share of your portfolio falls.   So sell euros (or RMB) and buy dollars to keep the dollar share of your portfolio up.  Or, if your assets are growing, direct a disproportionate share of the increase toward the dollar.

US stock markets under-perform foreign stock markets.  Better buy more US stocks to keep the US share of your portfolio up.

The worse the US does (financially speaking), the more financing it gets from the rest of the world.  A new IMF paper by Guy Meredith includes a graphic illustration of this.   The US has run huge current account deficits for the past few years.   In dollar terms, the rest of the world’s claims on the US are growing.   But because the dollar has depreciated and US markets have underperformed, the share of US assets in foreign portfolios has been constant at around 40% of the rest of the world's GDP over the past few years … see Meredith's Figure 8 on p. 11.  

I thought one of the core insights of economics is that incentives matter.   But doesn’t this pattern kind of create perverse incentives for the US?   The worse US financial assets do, the more financing the US gets --  at least so long as foreigners care more about maintaining a constant dollar share of their portfolio than little things like the actual return on dollar assets .... 

This sounds, I realize, like criticism of the Meredith's paper.  It isn’t really.  A portfolio balance model does a better job of explaining ongoing inflows to the US than other models over the past few years.   It fits the data.   

And one of the strengths of Meredith's paper is that the paper tries to match the model with actual stylized facts in the data.  For example, Meredith correctly note that the “US deficits reflect fast growth in the US” story doesn’t hold up.  US growth hasn’t been faster than world growth – the US share of world GDP isn’t rising.    He notes that much of the financing for the US deficit has come from central banks -- though I wish he used the same sketpical eye that he applies to the real exchange rate data (which he argues understates the real appreciation of the dollar because of differences in the way inflation is calculated) to the US official inflow data (which also likely understates official accumulation of US assets).   And he note that observed return (looking at the income line on the US balance of payments) on foreign investment and lending to the US has been very low – and well below the observed return on US investment and lending abroad. 

Throw in valuation losses from the dollar’s fall and low interest/ earnings on US investment and certainly those financing the US haven’t been doing so to make money.  Even if you are among the lucky few who haven’t lost money lending to the US/ investing in US markets (measured in your own currency), you have generally make less money than you would have investing elsewhere. 

That after all is why the US net international investment position hasn’t deteriorated by much (or even at all) recently: US investment abroad has done far better in dollar terms than foreign investment in the US.  

And it is why the euro/ dollar may matter for global adjustment, even if trade between the US and Europe is relatively limited.   Most US foreign investment is in Europe, so moves in the euro dollar deliver big valuation gains for the US.  I haven’t read the latest Philip Lane/ Gian-Maria Milesi-Ferretti paper on this (only Menzie Chinn's summary), but I would bet – based on their earlier research – that this dynamic drive their most recent findings. 

Of course, a portfolio balance model cannot explain the surge in inflows to the US in the late 1990s, when US markets outperformed foreign models.   Remember, in these models, outperformance is a reason to get the hell out and move into underperforming places.  Otherwise the share of underperforming markets in your portfolio falls. 

Most authors get around this problem by assuming that foreign appetite for dollar assets rose at some point.    If you think that this has continued, then there are large potential flows waiting around to finance even larger US deficits.   Foreigners have been desperately trying to raise the US share of their portfolios, but falls in the dollar (and strong performance of equity markets abroad) keep getting in the way … 

My actual criticism of this paper doesn’t center on the portfolio balance assumption.  It centers on the low returns on foreign investment in the US will continue assumption.  I think the low returns in the 2000-2005 period reflect very low US interest rates and a likely undercounting of the profits on foreign direct investment in the US (transfer pricing and the like, along with likely undercounting of reinvested earnings).

That seems to be changing.   The average interest rate on US borrowing rose (my calculations) to about 4.3% in 2006 – that is up from around 3% a few years ago, though still below the 6% rate observed back in 2000.   The US income balance didn’t deteriorate more in 2006 because the interest rate on US lending rose to about 5.5% (from under 2.5% in 2003).    The BEA seems to have made a real attempt to do a better job of counting reinvested earnings on foreign investment in the US – the difference between US returns on its direct investment abroad and foreign returns on their direct investment in the US seems to be shrinking.  (This reflects both an absolute rise in foreign returns and a rise in the valuation of US investment abroad that hasn’t been matched by a rise in the actual income stream).  I’ll have a lot more on this soon; I am in the process of updating my earlier analysis of the income balance.

But I’ll grant those (Cavallo and Tille, Kitchen, Meredith) who argue that the US current account position is rather more sustainable than people like me think one thing: 

If you assume, based on a portfolio balance model, that big falls in the dollar will generate both big valuation gains on US investment abroad and ongoing inflows into the US as foreigners struggle to keep the dollar share of their portfolio constant in the face of a sliding dollar;

If you assume, based on historical norms, that some of the inflows that the BEA measures in the TIC data will disappear from the survey data, reducing the measured US debt stock in the net international investment position (this is Daniel Gros’s explanation for “other valuation changes” otherwise known as statistical manna from heaven); 

And if you assume that the US will continue to attract large net inflows even though US investment abroad continues to generate more interest/ dividend income than foreign investment in the US;

Then the US external position looks pretty good for quite some time so long as the US trade deficit doesn't get a lot bigger.   The NIIP doesn’t deteriorate all that much (though a lot depends on the scale of the dollar's depreciation).   The income balance also doesn’t deteriorate -- at least not in an explosive way. 

The only question I have is whether these assumptions make sense over time.

But I grant that they do fit the post-2002 data …