Mark Whitehouse highlighted the deterioration in the US income balance in yesterday’s Wall Street Journal. The US paid more in interest and dividends on its external borrowing (and foreign equity investment in the US) than it received on its external lending (and its equity investment abroad).
It many ways, though, the deterioration in the US income balance has just begun. In this post, I’ll argue that that US lending abroad has a shorter-term structure than US external borrowing. Most US lending and US borrowing is denominated in dollars. Consequently, an increase in US short-term (policy) interest rates tends to increase the return on US lending abroad faster than it increases the interest rate the US pays on its borrowing.
In the first half of 2006, this effect turned out to be quite significant – slowing the pace of deterioration in the US income balance. US lending abroad is around $3.9 trillion (end 2005 data), so a gap on US borrowing and lending rates can have a significant impact on the overall income balance.
But as US policy rates stabilize and the interest rate on US borrowing catches up with the interest rate on US lending, this effect will disappear. The result: higher net income payments, even if the gap between the returns on US equity investment abroad (decent) and the returns on foreign equity investment in the US (embarassingly low) continues …
The gory details (and graphs) follow.
In absolute terms, interest payments on US external debt have grown faster than the interest the US receives on its external lending. It should: The US external borrowing exceeds US external lending by about $2.5 trillion (mid 2006 estimate). Rising net interest payments (and a stable surplus on direct investment) explain why the US income balance is now negative, as following chart shows.
However, the pace of deterioration in the interest balance seems to have slowed in the first part of 2006. Why? Largely because the interest income on the United States $4.7 trillion in external lending has been rising almost as fast as interest payments on the United States $7.25 trillion in external borrowing (mid 2006 estimates).
Why is it growing fast? The most likely explanation is that US lending abroad is very short-term, so it reprices more quickly. The implied interest rate on US external lending fell faster when US interest rates were falling. And it is now rising faster when US interest rates are rising.
Alas, with the Fed on hold, the interest rate the US gets on its external lending won’t continue to grow. And barring a miracle, the average cost of US external borrowing will continue to climb. The net result: a further $60b deterioration in the income balance.
This phenomenon can be discussed in slightly different terms. The debate over dark matter has died down, but I wouldn’t have noticed the swing in the implied interest rate the US gets on US external debt if Hausmann and Sturzenegger had not raised the topic (an updated version of their argument can be found here).
Hausmann and Sturzenegger postulated that the US borrowed cheaply from the rest of the world to lend money out at higher rates. Think of it as borrowing cheaply from the People’s Bank of China to buy high yielding Brazilian debt. The argued that the United States ability to provide a safe store of value allowed the US to earn an “insurance” premium. The US offered the world’s investors safety, and then put the funds that came in to work.
It was an interesting idea, but it didn’t turn out to be true. The amount of US lending to high-yielding emerging economies is very small relative to total US lending to the world.
Most US external lending seems to be short-term, dollar-denominated floating rate loans to borrowers located in offshore financial centers, whether London or the Caribbean.
Consequently, when US short-term interest rates were low, the interest rate the US earned on its external lending was below the interest rate the US paid on its external borrowing.
That is the opposite of what Hausmann and Sturzenegger postulated. The US was paying an insurance premium, not getting it.
But that changed, big time, this year. The implied interest rate on US lending has increased from around 2.5% in 2004 to around 5% in the first half of 2006. The implied interest rate on US borrowing has increased by far less, rising from around 3% to a bit over 4%. As a result, the US is now earning a bit of an insurance premium.
But it isn’t really an insurance premium, at least not in the sense of Hausmann and Sturzenegger thought of it. It really looks more like a fixed to floating swap, with the US paying something closer to a fixed rate on its longer-term borrowing, and the US getting a floating rate on its short-term lending. That swap was out of the money when US rates were low in 2002, 2003 and 2004. And it is now “in the money.”
The following chart (an updated version of the charts in my earlier post) tries to illustrate the various sources of what Hausmann and Sturzenegger label “dark matter” – namely the imaginary assets that keep the US income balance from being as negative as it should be, given that the US is a substantial net debtor. Some dark matter comes from the fact that US FDI abroad pays higher dividends than foreign direct investment in the US. Some comes from higher reinvested earnings on US direct investment abroad than on foreign direct investment in the US (though this appears to be of declining importance). Some comes from the fact that US equity investment abroad exceeds foreign equity investment in the US, and the US ends up making money issuing debt to buy equity. Some comes from the fact that some US external liabilities don’t pay any interest – lots of dollars circulate abroad. Some comes from the fact that Hausmann and Sturzenegger assume that all US debt should pay a 5% interest rate, and when US rates are lower than 5%, their methodology allows the US to book the difference between its borrowing rate of 4% and the 5% it should be paying as an external asset. And some comes from the difference between what the US pays on its external borrowing and what it gets on its external lending – the “insurance” or “swap” component.
Assuming I did the math correctly, the total stock of dark matter in my graph should match the stock calculated by Hausmann and Sturzenegger.
What stands out in this graph, apart from the fact that the stock of dark matter seems to be shrinking?
Let me highlight four things.
First, the US doesn’t really borrow all that much to invest in foreign equities, despite what Jim O’Neill of Goldman --who I greatly respect and generally agree with -- and many others argue. The gap between total US equity investment abroad and total foreign equity investment abroad at the end of 2005 was around $2.25 trillion. That gap was basically zero in 2001-02, but the strong relative performance of foreign stock markets led to a surge in the market value of US equity investment abroad.
However, most US equity investment abroad, however, is balanced by foreign equity investment in the US, not by US borrowing from the rest of the world. What might be called the equity for equity swap on the US balance sheet ($5.9 trillion at the end of 2005) is far larger than the debt for equity swap ($2.25 trillion).
The equity for equity swap also generates the majority of US dark matter, largely because of very low reported returns on foreign equity investment in the US. The dark matter from the equity for equity swap is the sum of the light blue (FDI dividends) and purple (reinvested earnings) bars; the dark matter from the debt for equity swap is represented by the tan line (borrowing to invest).
Second, from 2000 to 2004, the difference between the (large) reinvested earnings on US firms abroad and the (small) reinvested earnings of foreign firms operating in the US accounted for the majority of US dark matter. Look at the purple bar. That changed in 2005. US firms stopped “reinvesting” their earnings and instead made huge dividend payments to take advantage of the homeland investment act. And so far in 2006, the net gains on reinvested earnings are far smaller than they were from 2000-2004 – largely because foreign firms in the US are now reporting somewhat larger reinvested earnings. The difference in reinvested earnings generated $500b in dark matter in the first half of 2006 v $2000b in dark matter during the 00-04 period.
Incidentally, the surge in the market value of US FDI abroad in 2005 has had the effect of converting dark assets to “real’ assets, as it reduced the difference between implied rates of return on US investment abroad and foreign investment in the US.
Third, the pink bar shows the “dark matter” from the US net debt position. This reflects the fact that Hausmann and Sturzenegger assumed that the US should be paying 5% on its external debt, and – as explained earlier – if the actual interest rate was lower than 5%, there methodology effectively counted the difference between what the US should be paying and what it really was paying on its net debt (net debt is the difference between total US lending and US borrowing) as a “dark asset.”
Not surprisingly, the value of this dark asset (the pink bar) soared as US interest rates fell, reaching almost a $1 trillion in 2003 and 2004. It is now shrinking, though, as one would expect.
Fourth the bright red bar represents the gains (and losses) on the portion of the US external balance sheet that comes from borrowing from the world to lend to the world (Insurance services in Hausmann/ Sturzenegger land, or a fixed for floating swap in Setser land). That red bar was negative from 2001 through 2004.
Why – the rapid fall in interest rate on US external lending fell pushed the interest rate on US external ending below the interest rate on US external borrowing.
It was quite positive in the first half of 2006 (all numbers have been annualized) – generating almost a trillion ($800b) in “dark matter.”
Let me convert dark matter into plain English.
The fact that US lending repriced faster than US borrowing saved the US about $20b on its interest bill in the first half of 2006. If that gap persists for the full year, the total savings in 2006 will be around $40b .
I suspect, however, that this advantage will disappear over the next year and half, as US external borrowing reprices.
I also would assume that the average interest rate on US borrowing will climb toward 5%, eliminating another $400b in dark matter (i.e. pushing US annual borrowing costs up by about $20b … )
If I am right, even if the US went cold turkey, stopped borrowing and stopped running a current account deficit, the United States net interest bill will rise by another $60b as the interest rate on US external borrowing rises to 5% and the gap between what the US gets on its lending and what it pays on its borrowing shrinks.
Suppose the $60b gain that the US gets from a borrowing rate below 5% disappears by q2 of 2007. And suppose the US current account deficit over the next four quarters is close $900b.
The $900b in additional debt would imply another $45b in net interest payments (annualized). The repricing of the United States existing debt would add another $60b.
That would turn the $16b or so negative income deficit the US has now ($4b in q2, annualized) into a $120b deficit.
Now, I don’t know the term structure of US external debt, so the repricing process may take a bit longer. But – barring a Roubini-esque recession that leads to a big cut in US short-term rates and eventually a big fall in long-term rates – it is coming. The only real question is how fast. This is following graph shows what happens if it takes two years rather than one for the $60b loss from repricing to play itself out. And in that context, the US income balance deteriorates from say $30b this year to $190b in 2008.
$60b from “repricing” and $100b or so from all the additional the US takes on to finance its external deficits.
For that matter, if a Roubini-eseque recession leads to a big fall in policy rates, income on US lending should fall faster than income on US borrowing .. it wouldn’t help as much as you might think.
One last note: I would like to try to publish my work on the disaggregated sources of dark matter somewhere. Any ideas for a good target journal?
The disaggregated sources of dark matter are calculated by matching US external assets against comparable US external liabilities using the data in the US NIIP. Gross debts exceed gross US lending. Some of the difference is matched against US equity investment abroad. The residual is the United States net debt. Income payments are calculated from the BEA.
The implied interest rate on US borrowing and lending is calculated by subtracting estimated dividend payments on portfolio equity from the sum of government and other payments in the income balance. Since data on 2006 dividend payments isn’t available, I have assumed that portfolio equity dividends remained at their 2005 levels. The stock of debt is calculated by adding debt creating flows from the first half to the totals in the NIIP data. I annualized both the income payments and the flows to estimate 2006 interest rates in a way comparable with previous years.