By Alex J. Pollock Wednesday, July 13, 2011
Filed under: Economic Policy, Government & Politics
Over the last four decades, the U.S. government has engaged in a financial experiment, or adventure, of exploding agency debt relative to Treasury securities.
The huge debt of Fannie Mae, Freddie Mac, other government-sponsored enterprises, and other off-budget government agencies (â??agency debtâ?) fully relies on the credit of the United States. This means it by definition exposes the taxpayers to losses, but it is not accounted for as government debt. As the Federal Reserve carefully notes in its â??Flow of Fundsâ? report, non-budget agency and GSE debt is not â??considered officially to be part of the total debt of the federal government.â?
Not â??considered officially,â? as they sayâ??but what is it really? It puts the federal budget at risk, subjecting it to uncertainty of defaults. It has proven its ability to generate huge losses for the taxpayers. In other words, it is really, if not â??officially,â? government debt.
The vast majority of agency debt goes to subsidized housing finance though Fannie Mae, Freddie Mac, the Federal Home Loan Banks, and the FHA/Ginnie Mae combination. It represents off-balance sheet financing and hidden leverage for the government. Fannie and Freddie in particular have reasonably been characterized as â??government SIVs,â? (â??structured investment vehiclesâ?) which failed, just as did the SIVs of Citibank and other bailed out banks.
Agency vs. Treasury Debt
In 1970, Treasury debt held by the public (â??Treasury debtâ?) was $290 billion. Agency debt was small by comparison: it totaled only $44 billion.
But by 2006, at the height of the housing bubble, while Treasury debt was up to $4.9 trillion, agency debt had inflated to $6.5 trillion. Treasury debt had increased 17 times during these years, but agency debt had multiplied 148 times!
At the end of 2010, Treasury debt was $9.4 trillion, and agency debt was $7.5 trillion.
Graph 1 shows the remarkable history of agency vs. Treasury debt.
Graph 1
In 1970, agency debt represented only 15 percent of Treasuries. By 2006, this had inflated to 133 percent. At that point, there was a lot more agency debt than Treasury debtâ??more government debt off-balance sheet than on it. To move debt off balance sheet was a preference shared by many financial actors during the bubble, but the government had been consistently doing so for almost 40 years. At the end of 2010, agencies were 81 percent of Treasuries, still a notably high level.
Graph 2 displays the trend of agency debt as a rapidly increasing percentage of Treasury debt. The percentage is falling at the end because of the huge increases in Treasury debt we all know about, reflecting among other things the problem that some of the off-balance sheet debt had to be taken back on balance sheet in the crisisâ??just as happened with private SIVs.
Graph 2
If we add these two types of debt together, we can get a total of â??effective government debtâ? (debt dependent on government credit) held by the public. Graph 3 compares this â??effective government debtâ? with Treasuriesâ??an instructive comparison.
Graph 3
Observe that effective government debt now totals nearly $17 trillion.
Increasing the Cost of Treasury Debt
The expansion of agency debt not only imposes risk and realized losses on taxpayersâ??we recall the $160 billion that the U.S. Treasury has been forced to put into Fannie and Freddie to prevent their financial collapseâ??it also increases the cost of Treasuryâ??s direct financing by creating a huge pool of alternate government-backed securities to compete with Treasury securities, and thus increases the interest cost to taxpayers.
So although agencies are not â??officially government debt,â? they undoubtedly increase the required interest rates on Treasury securities, in my judgment, and thus increase the federal deficit. The greater the amount of agency securities available as potential substitutes for Treasuries, the greater this effect must be. As a manager of a major institutional investor told me recently, â??We view Fannie and Freddie MBS as Treasuries with a higher yieldâ??so now we own very few Treasuries.â?
It is difficult to put an exact number on the counterfactual question of how much this increased cost has been. However, a quantitative suggestion is implied by a recent Federal Reserve analysis (Joseph Gagnon et al., â??Large-Scale Asset Purchases by the Federal Reserve: Did They Work?â?â??FRBNY Economic Policy Review, May 2011). The authors conclude that by taking $1.7 trillion in securities out of the market by Federal Reserve purchases, of which more than $1 trillion were purchases of agency securities, the interest rates on ten-year Treasuries were reduced by â??somewhere between 30 and 100 basis points.â?
Suppose we run this logic in reverse: if the supply of effective government debt is increased by trillions of dollars of agency debt, perhaps that increases the cost of long-term Treasuries by at least a like amount.
This result depends on the idea that investors will substitute agency debt for Treasuries and thus reduce the demand for Treasuries from what it would have been. We can observe a striking example of this substitution in the aggregate balance sheet of the commercial banks.
In 1970, commercial banks owned $63 billion in Treasuries and $14 billion in agency securities. Their Treasury holdings were more than four times their agency holdings. By 2006, at the peak of the bubble, all commercial banks owned only $95 billion in Treasuries, which was dwarfed by their $1.14 trillion in agencies. They then had 12 times the investment in agencies as in Treasuries.
At the end of 2010, the corresponding totals are $299 billion of Treasuries and $1.35 trillion in agencies. This long-term trend of agencies versus Treasuries in banking is shown in Graph 4.
Graph 4
Expressed as a percentage of banking assets, investment in Treasuries falls from 12 percent to less than 1 percent, then recovers to only 2 percent under current circumstances. Meanwhile investments in agencies inflates from less than 3 percent of banking assets to over 10 percent, then ended last year at 9.4 percent. This substitution is shown in Graph 5.
Graph 5
Why Was the Agency Debt Inflation Possible?
How was it possible for agency debt, and the corresponding taxpayer exposure, to grow so much for so long?
Well, bond salesmen, peddling trillions of dollars of Fannie, Freddie, and other agency securities to investors all over the world, told them something like this: â??You canâ??t go wrong buying these, because they are really a U.S. government credit, but they pay you a higher yield! So you get more profit with no credit risk.â?
In contrast, a senior member of the Financial Services Committee memorably opined that Fannie and Freddie had â??no explicit guarantee . . . no implicit guarantee . . . no wink and nod guarantee.â? Official voices liked to point out that the offering memoranda for GSE debt said right there in boldface type that these securities were not guaranteed by the United States.
Nonetheless, what the bond salesmen said was right, as events have conclusively demonstrated. A good sense of the resulting situation is described by then-Secretary of the Treasury Henry Paulson in his memoir of the financial crisis:
Foreign investors held more than $1 trillion of the debt issued or guaranteed by the GSEs . . . To them, if we let Fannie or Freddie fail and their investments got wiped out, that would be no different from expropriation. They had bought these securities in the belief that the GSEs were backed by the U.S. government. They wanted to know if the U.S. would stand behind this implicit guaranteeâ??and what this would imply for other U.S. obligations, such as Treasury bonds.
Note how clear in this description is investorsâ?? belief that agency debt is simply government debt.
Risk vs. Uncertainty
Of course, using the credit of the United States to make, guarantee, insure, or finance mortgage loans through any of the agencies which do so entails credit losses. This in itself is not the most difficult problem: if we knew what the losses would be, or knew what they would be within a narrow range, the losses could be easily priced and budgeted for.
For entities subject to the Federal Credit Reform Act, the expected (best guess estimates) of losses must be reflected as costs in the federal budget. This requirement was without doubt a major improvement over previous practice, but it does not address the fact that we simply do not know what the losses will be. As the Congressional Budget Office points out, the FHA, for example, has often had to make big increases in its estimates of credit losses.
Even more troublesome is the fact that the expansion of leverage created by the very programs in question tends to make the losses worse. This certainly happened with Fannie and Freddieâ??s hyper-leveraging of mortgage credit risk.
Huge increases in loss estimates characterized the failure of Fannie and Freddie. The limits of the most expert knowledge of the future extent of losses is highlighted by this statement of the then-Director of the Office of Federal Housing Enterprise Oversight: â??Let me be clearâ??both [Fannie and Freddie] have prudent cushions above the OFHEO-directed capital requirements.â? This March 2008 statement was indeed clear, but wrong; only six months later both agencies collapsed.
Two months before the collapse, in July 2008, then-Chairman of the Senate Banking Committee pronounced: â??Whatâ??s important are factsâ??and the facts are that Fannie and Freddie are in sound situation.â?
The same month, the Chairman of the Federal Reserve assured us that Fannie and Freddie â??are in no danger of failing.â?
As the Congressional Budget Office says: â??The expected cost of defaults â?¦ do not account for the uncertainty about how costly such defaults ultimately will beâ? [emphasis added]. Absolutely right.
Make Treasury Responsible for Managing the Governmentâ??s Credit
Managing the issuance of Treasury securities under the circumstances of the last decade deals with only about half, and sometimes less than half, of the effective government debt.
Read Full Article »