Abstract: Our joint commentary identifies two risk measures—capital ratios and duration—which may be helpful in assessing outcomes of the Fed’s exit strategy. We have avoided a fully prepared technical paper and admit that the issues are complex for many of our readers. To assist readers, we have divided this paper into two parts. The first part describes the Fed’s balance sheet in terms that are more commonly understood because they are applied to banks. Of course, we know that the Fed is not a bank. We choose the bank as a reference since it is the basis now for many federal programs. We demonstrate that if the Fed were a bank it would be capital constrained. In the second and more technical part, we use McCauley duration to measure the risk building in the Fed’s new balance sheet construction. We estimate that a 1% upward parallel shift in the yield curve could render the Fed technically insolvent, if it had to mark to market as a bank is required to do.
If the Fed were a bank and had to live up to the same capital requirements as the institutions it regulates, would it be adequately capitalized? This may seem like an absurd question. After all, a central bank can’t go bankrupt. This is especially true for the Federal Reserve, whose balance sheet historically has been quite simple compared to even that of a moderately sized bank.
Until recently, the Fed’s assets consisted primarily of US Treasury debt, typically with relatively short maturities. There was no credit risk and the Fed usually didn’t engage in the buying and selling of those assets. Rather, it bought and held them to maturity. This strategy provided the justification for the Fed not to mark those assets to market as interest rates changed.
Open market operations were usually conducted using repurchase agreements (RP) and reverse repurchase agreements (RRP), where securities were temporarily sold (bought) with an agreement to buy (sell) them back to the holder at an agreed-upon price. These RP transactions were normally done for overnight or for three-day maturities. They served to decrease or (increase) the outstanding supply of Federal Funds, which are the cash reserves banks trade among themselves. Most importantly, they determined the overnight Fed Funds rate, which was the primary instrument of Fed policy. They also linked the Treasury securities market to the Federal Funds market, because Treasury securities were used in the RP transactions.
Until the recent crisis, the Fed’s liability structure consisted mainly of currency held by the public (Federal Reserve notes), deposits held by member banks in the form of excess and required reserves, and deposits of the US Treasury. Currency accounted for about 90% of the Fed’s outstanding liabilities. Individual notes roll over as they wear out, but they are simply replaced with new ones. The total currency outstanding volumes fluctuated slightly with changes in the demand for cash. Cash demanded by the public includes foreign demand and what was needed in automated teller machines (ATMs).
Currency is effectively a permanent source of funding for the Fed, and its maturity is nearly infinite. Until just recently, none of the reserve deposits paid interest. Of course, cash pays no interest. So the currency component of the Fed’s balance sheet has a zero cost of carry.
Today, the Fed holds a wide range of loans and other assets acquired as part of its efforts to combat the financial crisis. The financing of those assets has also changed. As a result, the Fed’s balance sheet is not so simple and is now exposed to a significant amount of both credit and interest-rate risk.
The Fed just released a consolidated balance sheet as of May 27, as part of a new monthly report on “Credit and Liquidity Programs.”1 The report lists consolidated Federal Reserve System assets at $ 2.082 trillion and total capital at $45 billion. This implies a capital-to-assets ratio of 2.16%., which would also be the Fed’s tangible equity capital ratio. Keep in mind that the Fed’s target tangible equity capital ratio for banks is 4%.
Of course, the Fed’s tangible equity capital ratio isn’t its risk-based capital ratio. If the Fed were a bank, the government securities portion would require a 0% risk weight. That means no capital would have to be held against those assets. As a bank, the Fed’s mortgage-related holdings would require a 50% risk weight under current standards. If one considers all the other lending of the Fed to be equivalent to assets that require a 100% risk weight, the Fed, as a bank, would have a Tier 1 capital ratio of approximately 4.9%.
Now, we know that the Fed is committed to expanding its mortgage portfolio to about $ 1.2 trillion from its current size of $428 billion. This would increase the Fed’s risk-based assets by another $386 billion and reduce its Tier 1 capital ratio to 3.6%, assuming a 50% risk weight. The Fed has virtually no options to increase its capital account, except to retain more of its earnings rather than return those funds to the Treasury. The Fed cannot engage in a stock offering.
Its current capital base provides precious little margin to absorb any additional losses on its AIG- and Bear Stearns-related portfolios or to deal with substantial write downs in asset values. Furthermore, there is virtually no cushion to absorb the capital losses that might have to accompany any upward movement in interest rates. An increase in inflation expectations may require the Fed to raise policy rates needed to fight inflation. This could trigger losses on long-term, lower-rate assets as the Fed’s attempts to exit from its quantitative easing strategy.
Finally, comparing the published financial statements in the Fed’s H.1 report with its new monthly disclosures on its lending programs suggests that the Fed has not yet recognized the embedded losses in the Maiden Lane portfolios in its capital account. Currently, those portfolios are under water by $13 billion. Loss recognition would reduce the Fed’s estimated capital-to-total-assets ratio to 1.5% and reduce its estimated Tier 1 capital ratio to 3.5%. Should the Fed’s mortgage related assets expand to their targeted level and Maiden Lane losses be recognized, this combination would reduce its Tier 1 capital ratio to 2.5%.
It is hard to believe that in today’s environment mortgage loans and mortgage-related assets would be regarded as half as risky as traditional bank lending. If mortgage loans were considered on par with commercial and industrial loans, then the Fed’s Tier 1 capital ratio would be about 3.9% without recognition of these losses and about 2.8% after recognizing the embedded Maiden Lane losses. In short, by its own standards, the Fed is arguably not adequately capitalized and would be perilously close to being turned over to the FDIC – if the Fed were a bank.
This view of the Fed’s capital adequacy is obviously somewhat tongue –in cheek, but the issues raised are important for two reasons. First, possible asset sales have featured prominently in the exit strategy that the Fed has put forward as a means to reduce the outstanding liquidity it had injected into markets previously. Secondly, the Fed has accumulated a much larger portfolio of longer-term Treasuries and mortgage-related and other longer-term assets whose liquidity may be problematic and whose changes in market value may be substantial if longer-term interest rates increase.
(Note to readers. What follows is a more technical discussion of the Fed’s balance sheet and risk exposure.)
The Fed’s report provides some information on the maturity of those assets. Over half of its outstanding loans and other asset holdings have a maturity of over one year, and over 32% have a maturity of five years or more. This means that the durations of the Fed’s assets have increased substantially during the crisis period and are likely to increase even more as its mortgage-related holdings continue to expand. The increase in duration plus expansion of its mortgage-related assets to potentially $1.2 trillion means that the Fed’s interest-rate and credit risk exposure has increased substantially.
This would not necessarily be a problem because of the unusual nature of the Fed’s liability structure, which until recently consisted mainly of approximately infinite-duration non-interest-bearing liabilities (namely currency). However, this too has changed as the Fed has responded to the financial crisis. In particular, the Fed’s liabilities are now nearly half overnight reserve deposits held by member banks and half in Federal Reserve Notes. Reserve deposits have a near-zero duration, while Federal Reserve Notes have nearly an infinite duration. Thus the Fed’s liability structure is a “barbell” shape whose average portfolio duration has shortened significantly at the same time the duration of its assets has increased. Assuming the mortgage-related portion of the portfolio continues to expand as promised and is financed through the creation of bank reserves, the trend towards a widening asset/liability mismatch will persist. If the Fed were to expand its mortgage holdings to $1.2 trillion and finance it with short-term reserves, then short-term liabilities would account for over 60% of its liabilities.
At some point in the process, the risk of even small changes in interest rates will wipe out the Fed’s capital if it marks assets to market. For example, if one assumed an average asset duration of five years and that duration of its liabilities shrank as low as eight years, then back-of-the-envelope calculations using McCauley duration suggest that a 1% increase in interest rates would reduce the Fed’s capital by 38% if assets were marked to market. 2 If the Fed were a bank, this would be sufficient to trigger liquidation by the FDIC.
The first point of this illustration is simply that asset sales or other policies to exit from the Fed’s quantitative easing strategy have a risk if interest rates move by even a relatively small amount. This poses significant risks to the Fed’s capital structure and the public’s perception of its solvency.
The second point is that the duration calculation has suddenly become an important test of the Fed’s ability to achieve an exit strategy. The Fed will remain technically solvent as long as the duration of its liabilities is substantially greater than the duration of its assets. As a practical matter, because of their prepayment feature, mortgage instruments complicate the calculation of their duration, which increases as interest rates rise, other factors being equal. This means that policies designed to fight inflation by raising interest rates will also narrow the gap between the duration of the Fed’s assets and liabilities and increase the risk that its capital will fall below acceptable levels. An examination of the Fed’s balance sheet suggests that the Fed is getting close to failing this test.
Third, because of its interest-rate risk exposure and its likely impact on Fed exit strategies and credibility, we expect that there will be great interest in the Fed’s balance sheet duration. If the Fed truly seeks transparency and discloses its internal calculations of duration, it will also have to estimate how the duration of the mortgage portfolio changes as interest rates change and what impact this will have on its capital structure. So far we have not seen any Fed computations of duration made public.
One final note is needed to be technically correct. If the Fed were a bank and if the Fed were marking its assets and liabilities to market, it would also have to mark its gold hoard to market. Such a change would add about $150 billion to the carried book value of the Fed’s gold. That addition could also be reflected in the Fed’s capital account. It would also open the Fed to scrutiny as the gold price fluctuated. And it would certainly enhance the market’s view of precious metals.
The authors are not advocating a revaluation of gold. In our view, gold trades just like any other commodity. It has no place in a fiat-currency world of monetary policy making. Any one who wishes to use it as a hedge or a store of value may do so on their own. Furthermore, basing US monetary policy on revaluing gold on the Fed’s balance sheet borders on lunacy. Remember, the largest hoards of gold are found in the ground in Russia and South Africa.
1 See “Federal Reserve System Monthly Report on Credit and Liquidity Programs and the Balance Sheet,” Board of Governors of the Federal Reserve System, June 2009.
2 The method used here is based upon Bierwag, G.O. and George G. Kaufman, “Duration Gap for Financial Institutions,” Financial Analysts Journal 41, March/April 1985, pp 68-71.