by James D. Hamilton, Econbrowser.com
Originally published at Econbrowser 17 March 2013.
The Federal Reserve has increased its assets from $900 billion in 2007 to over $3,150 billion and still climbing today. On the liabilities side of the Fed’s balance sheet, reserve balances held by banks have gone from $10 B in 2007 to $1,750 B and climbing today. My expectation had always been that this would be a temporary situation, with a return to historical norms when economic conditions improved. Recently, three different teams have independently studied what that transition back to normal might look like. One study was carried out by Robert Hall and Ricardo Reis (professors at Stanford and Columbia, respectively) and another by Seth Carpenter, Jane Ihrig, Elizabeth Klee, Alexander Boote, and Daniel Quinn (all economists at the Federal Reserve Board). I participated in a third study with David Greenlaw at Morgan Stanley, Peter Hooper at Deutsche Bank, and Frederic Mishkin at Columbia. Our analysis used a similar methodology to that conducted by the Fed staff, and we reached similar conclusions to theirs, while Hall and Reis took a broader and more theoretical perspective. Here I describe the methods and findings of our study.
In our baseline calculations we assumed that the Fed continues to buy long-term assets at its current pace through the end of 2013. We further assumed that beginning in January, the Fed would buy new assets only to the extent necessary to replace maturing holdings so as to keep total assets steady through 2014, and then begin to sell MBS late in 2015 in order to arrive at zero MBS holdings after 2019. These and other assumptions are summarized in the table below, with the implied path for Fed assets displayed in the following graph.
Reserves would return to more normal historical levels from one of two forces. Some of the reserves will end up as currency held by the public, represented by the green region in the graph below. Selling off the Fed’s assets (or allowing them to mature without replacing) would be the other main force bringing excess reserves back down.
We assumed that interest rates would rise over the next several years as predicted in Blue Chip consensus forecasts, as summarized in the figure below.
The relevance of these assumptions is that if the Fed is buying assets when long-term rates are low, and selling them when rates have risen, the Fed will make a capital loss on its purchases, as indicated by the red region on the graph below. The Fed will also have to be paying more in interest on reserves as rates rise (the blue region below).
According to our baseline calculations, during 2017-2019 these expenditures for the Fed will be greater than income coming in from interest on retained assets (the green region in the graph above). This can be seen more clearly in the graph below, which plots Fed inflows minus outflows for each year.
Is this a problem that the Fed will be operating at a loss? One perspective is to note that for the past several years the Fed has been making unusually large profits which it returned to the Treasury, and we are anticipating that this will continue to be the case for the next several years. One calculation of interest is to look at the cumulative profit or loss by the Fed since the financial crisis began. This cumulative net sum remains positive in our baseline simulation (the black line in the figure below) – the Fed’s profits in its good years exceed its losses in its bad years.
On the other hand, there is still an interesting logistical question of how the cash shortfall will be covered. JP Koning argues that recapitalization of the Fed by the Treasury in such circumstances is not a sure thing. The Fed’s Plan A appears to be that they will simply create new reserves as necessary to cover any losses incurred. In terms of the accounting, the creation of new reserves (a Fed liability) coincides with the acquisition of an asset recorded in the Fed’s new “deferred asset account“. The theory is that this is simply an account that gets credited, and represents a claim on future Fed earnings which normally would have been returned to the Treasury but which instead will be retained by the Fed in order to rebuild the Fed’s working capital. That is, the deferred asset account is an IOU from the Treasury to the Fed, counted by the Fed as one of its assets.
Accounting aside, there is an interesting question how that would play out in terms of politics and public perceptions of the Fed. I could certainly imagine politicians grandstanding over the fact that the Fed will be paying out billions to big Wall Street banks at the same time that the Fed is losing billions of dollars for taxpayers. And how would this affect expectations of inflation in an environment when on the fiscal side interest expenses and debt continue to grow while the Fed’s only resource for covering its losses is to print more money?
The Fed could spare itself some of this awkwardness by simply refusing to sell any of its assets, as a result of which it would never have to realize the loss. According to our market model of mortgage prepayment, the Fed could allow its MBS simply to mature on their own, in which case we would be back to normal levels of reserves by 2020 instead of 2018 (see the red hashed region in Figure 2 above). This would allow the Fed to continue to report a positive profit, as represented by the dashed blue line in Figure 7 below.
On the other hand, if interest rates should rise more quickly than the Blue Chip consensus, the Fed’s losses would be larger than anticipated under our baseline assumptions. In our paper we evaluate a number of alternative possibilities. To highlight the most dramatic of these, if the Fed continues expanding its balance sheet during 2014 rather than hold it constant as assumed in our baseline scenario, and if interest rates are 200 basis points higher starting in 2016 than assumed in our baseline, we calculate that the Fed’s deferred asset account could reach as high as $370 B.
Now, some might argue that this is exactly the purpose of the Fed’s large-scale asset purchases, namely, to take maturity risk off of holders of Treasury bonds (and thereby reduce the term premium on long-term bonds), and absorb that risk on the Fed’s own balance sheet. Absorbing risk means that in some states of the world you take a loss. Nor should any of our calculations be taken to suggest that the Fed’s efforts so far to provide support for a weak economy have been a mistake – the benefits in terms of lower unemployment could greatly exceed the loss, even if those benefits were valued solely for their implications for receipts and expenditures of the U.S. Treasury.
Notwithstanding, before we made these calculations, my assumption had been that the Fed would eventually sell most of the assets it has recently required. However, I’m now guessing that the Fed may never be comfortable doing that. And if you’re never going to sell, it’s a reason to be a little more cautious about how much more you plan to buy, particularly if, as I believe, the incremental benefit of additional purchases at this point is small.