Skip Navigation

Assessing the Public Costs and Benefits of Fannie Mae and Freddie Mac May 1996

The housing government-sponsored enterprises are able to claim that they impose no cost on taxpayers only because the subsidies they receive--so far, at least--are in noncash benefits rather than cash. Actual costs of the GSEs to the government, however, are no less real than if the Federal National Mortgage Association and the Federal Home Loan Mortgage Corporation were granted free use of federal buildings, land, and personnel. If sold competitively, the benefits the federal government provides to the housing GSEs would command billions of dollars. That forgone annual income is the cost of the housing GSEs to taxpayers and the government. When the housing GSEs do indeed acknowledge receiving benefits, they suggest that the services they provide to the government and to the public are of at least equal value.
 

What Does GSE Status Mean?

To refer to Fannie Mae and Freddie Mac as government sponsored gives only a hint of the extent of the relationship between government and these privately owned corporations. A more complete description lists all of the features of federal law and regulation that constitute GSE status. The list usually begins by noting that federal (rather than state) law charters the enterprises and it continues with a variety of other features (see Box 1).
 

The totality of the defining characteristics of GSE status, however, constitutes a benefit not expressed or referred to in any provision of law. That benefit consists of a strong implication from the federal government to investors that GSE obligations are safe from the risk of default. That assurance is conveyed by frequent parallel treatment of GSEs and federal entities in law and the contrasting treatment of GSEs with fully private entities. As shown in Box 1, numerous instances occur in which federal law treats the risk of GSE securities as no greater than that of risk-free U.S. Treasury securities. Issues of Fannie Mae and Freddie Mac share several characteristics of Treasury debt: they are approved by the Treasury, paid by and eligible for purchase by the Federal Reserve, equivalent to federal debt under the Securities Exchange Act of 1934, and suitable collateral for protecting the government's own money. Of equal importance, in common with Treasury securities, GSE securities are exempt from disclosures of risk required by the Securities and Exchange Commission (SEC) and from restrictions on holdings by entities whose liabilities the federal government explicitly guarantees. Those provisions endow GSE securities with the appearance of being significantly safer than the intrinsic credit quality of the GSE would ordinarily warrant.

Short of placing an explicit guarantee on the securities of the housing GSEs, the law could hardly be more clear: the government's financial interests in the safety of Fannie Mae and Freddie Mac ensure that their obligations are safe from the risk of default. Although the federal guarantee is only implicit, the financial markets grant "agency status" to the securities issued by GSEs on the basis of those statutory provisions, as though the enterprises were part of the federal government.

That federal seal of approval is immensely valuable to the housing GSEs. It provides Fannie Mae and Freddie Mac with savings through a reduced cost of funding that is not available to any fully private firm. On debt issued by the two agencies to fund their portfolio holdings of mortgage and mortgage-backed assets, interest rates are lower than on obligations issued by fully private firms whose financial condition is stronger than that of the housing GSEs. On mortgage-backed securities issued by the GSEs, agency status permits Fannie Mae and Freddie Mac to sell securities regarded as being of the highest quality by simply declaring the securities "guaranteed," without incurring the expense of putting in place the credit enhancements--such as subordinated securities, third-party insurance, or reserve guarantee funds--that are required of fully private intermediaries if their MBSs are to be rated Aaa or Aa.⁽¹⁾
 

The Credit Enhancement of Government-Sponsored Enterprises

The agency status of Fannie Mae's and Freddie Mac's obligations transforms the market's view of the credit quality of the housing GSEs and vaults their securities from a rating of A or Aa based on their intrinsic financial condition to super Aaa because the risk of default is seen to be lower than on even the highest-rated fully private securities.⁽²⁾ That transformation suggests that the financial equivalent of GSE status is the difference between the financial strength of an A- or Aa-rated firm and that of super Aaa.

The ability of a firm to meet its financial obligations is measured primarily by the excess of the value of its assets over its liabilities or by its equity position. That measure is a key one because it indicates the margin by which the resources the firm commands exceed those it owes. A firm with liabilities equal to 97 percent of its assets has only a 3 percent cushion of value to protect its creditors, whereas a firm with liabilities equal to 94 percent of assets offers an equity buffer that is twice as large. For fully private firms, a higher ratio of equity to liabilities means--all other factors being equal--a higher credit rating and a lower cost of borrowing. Equity, therefore, can mean the difference between a lower credit rating and a higher one.

Granting GSE status can be thought of as equivalent to giving an asset and an equal increase in the equity of the enterprise, which significantly enhances its credit quality. But such an infusion of assets and equity is remarkably different from what a fully private firm obtains from selling additional stock to the investing public.

It differs in a number of ways. First, the infusion is large. After the taxpayer equity is added, a GSE has more capital than any fully private firm would pay to acquire or find cost-effective. Second, the government's commitment as a provider of equity is potentially open ended. Assume that a GSE grows, takes on more risk, or reduces the amount of equity that private shareholders have invested (for example, by repurchasing its own stock). If so, the government is effectively committed to providing additional amounts of capital to ensure that the enterprise retains its super Aaa rating. In that sense, taxpayers do not enjoy the "limited liability" granted to private investors. In financial terms, that open-ended commitment means that the government has given the GSE an option to sell additional amounts of stock to the government. The safety and soundness regulation of a GSE can be interpreted as an attempt to limit the government's transfer of equity to the enterprise. Third, the equity does not require the user to compensate the provider (in the case of GSEs, the provider is the government). Instead, all after-tax earnings and increases in the value of the enterprise, including those attributed to the government's equity, accrue to the private shareholders. And fourth, the value of the equity conveyed to the GSE does not appear in the accounts of the government or the enterprise.
 

Annual Costs of the GSEs

Each year taxpayers provide the GSEs with the benefits of enhanced credit standing. Even though taxpayers have not laid out any money for those benefits, enhancing the GSEs' credit standing is costly to taxpayers because--instead of providing it free--the government could sell the right to share its credit standing. With the receipts from such a sale, the government could increase spending for any public purpose or reduce taxes. The annual cost to taxpayers thus depends on how much the use of the government's credit standing is worth to the GSEs and their potential competitors. That value can be measured in several ways. One is to look at the change in the market value of a GSE caused by the transfer of government equity to (or withdrawal of it from) the GSE each year. Another is to measure the annual flow of returns that taxpayers could have received on their equity if placed with terms of comparable risk.

Measuring the equity provided by the government has the advantage that it recognizes the full cost when the capital is transferred. By contrast, the flow of forgone returns smooths the cost of the equity transfer over many years. In fact, as explained in Box 2, the amount and direction of the transfer of equity can vary sharply over time.
 

If the government was starting a new GSE today, the transfer of equity would probably be the preferred measure of annual cost because it recognizes the full cost of the capital when it is transferred to the GSE. But the government is not creating a new GSE. The sponsored enterprises have already received unrecognized infusions of taxpayer equity. For those past transfers, the current annual cost is the forgone taxpayer return. For new transfers of equity, the current annual cost is the full amount of the equity transferred in the current period. An earlier study of the annual cost of the housing GSEs, therefore, defined the government's cost as the sum of the annual change in the value of the government's equity plus the forgone return on equity provided in earlier periods.⁽³⁾ That study estimated that in 1979, for example, the government provided $3.7 billion in equity to Fannie Mae and gave up $300 million in earnings on pre-1979 equity, thus providing $4 billion in value.

Estimated Federal Equity

Although the value of the government's equity position is not recognized on the books of either the government or the GSEs, shareholders in the financial markets recognize and value it. Hence, estimates of the value of credit enhancement to the shareholders can be obtained by calculating the difference between the market value and the balance-sheet value of the enterprises.

The accounting "fair value" of a firm is the market value of on-book assets minus the market value of on-book liabilities. That value is reported in the annual reports of the housing GSEs beginning in 1992. The contrasting market value of the firm is the number of common shares outstanding multiplied by the stock price on the day for which the accounting fair value is calculated (see Table 4 for both of those values for Fannie Mae and Freddie Mac along with their difference). The market value of both Fannie Mae and Freddie Mac is a multiple of the value that would be expected from their on-book assets and liabilities alone. The difference is especially large for Fannie Mae, whose market value is more than three times the "fair market" accounting value. If all of the difference is the result of the unbooked value of federal credit enhancement, taxpayer equity in the two GSEs is now nearly $30 billion, with over 75 percent of that going to Fannie Mae.
 

However, the precise extent to which federal credit enhancement contributes to the excess of market value over accounting value for Fannie Mae and Freddie Mac is not known. But it is known to have considerable value. For example, it permits the GSEs to borrow in the capital markets at lower interest rates than are available to commercial banks, even though the GSEs have less than half the capital required for banks.⁽⁴⁾ Nevertheless, other unbooked assets and liabilities affect the difference between market and accounting values.⁽⁵⁾ Most non-GSE firms have a market value that exceeds their "book value" (which does not adjust booked items for market value as the fair value measure does). Only a few fully private firms with high ratios of market to book value are comparable in size to the GSEs. For example, a search of a commercial database containing 1,613 financial services, real estate, and insurance firms found 298 firms with a market-value-to-book-value ratio of 2 to 1 or more on April 12, 1996. Most of those firms were small, however. Only 30 also had a market value of $4 billion or more. Finally, only 12 of the 30--including Fannie Mae and Freddie Mac--had market-to-book-value ratios greater than 2 to 1 for the previous three years. For Fannie Mae and Freddie Mac, the market-to-book-value ratio now exceeds 3 to 1.

The change in the value of the credit enhancement determines the flow of taxpayer equity between the government and the GSE, and is one component of annual taxpayer cost. If the value of other unbooked assets and liabilities changes slowly, then most changes in the difference between market value and accounting value are likely to stem from changes in the market value of federal credit enhancement. The difference between market and balance-sheet value for both GSEs combined rose $600 million from 1992 to 1993, then fell $4.3 billion in the next year before surging $17.5 billion in 1995. Even after adjusting for the robust increase in the value of the entire stock market in 1995, the market is indicating a sharp rise in the net value of unbooked items for Fannie Mae and Freddie Mac in the last year.

The value of a portfolio of all traded stocks increased about 35 percent in 1995. Previous studies have found that a 1 percent change in the overall market is accompanied by a 1.3 percent change in the same direction in Fannie Mae and Freddie Mac stock.⁽⁶⁾ Given that relationship, the housing GSEs would have been expected to go up about 45 percent in 1995. In fact, the value of Fannie Mae stock rose 70 percent and Freddie Mac stock 65 percent.

The difference between market and accounting value in 1995 is probably the largest ever for the housing enterprises. The highest estimated difference in value before that was $11.3 billion for Fannie Mae, which occurred in 1981 when the market value of the GSE was $500 million whereas its balance-sheet value was negative $10.8 billion.⁽⁷⁾ That market value indicates that although Fannie Mae was insolvent on the basis of the fair market value of booked assets and liabilities, the value of an unrecognized asset--the federal government's credit enhancement is an obvious candidate--made the firm solvent. Over the 1978-1985 period, the value of the credit enhancement averaged 9 percent of the face value of Fannie Mae's mortgage portfolio. If that value is extrapolated to 1995, the value of the government's equity position would be about $32 billion, or $3 billion more than the difference between the market and accounting value of $29.3 billion.

Forgone Return on Equity

The second component of the annual cost to the taxpayers is the return that they would have received from other firms for the use of equity capital. (This component applies only to the taxpayers' equity balance at the beginning of the period for which cost is being calculated--that is, on December 31, 1992.) That return is approximated for firms whose risk is comparable to that of Fannie Mae and Freddie Mac by the average annual rate of return earned by GSE shareholders from dividends and stock appreciation over the period from year-end 1992 to year-end 1995. Taxpayer equity at the end of 1992--measured as the difference between market value and fair market value--was $15.5 billion. The weighted average return to shareholders of the housing GSEs was 20.6 percent for the 1993-1995 period. Thus, taxpayers would have received an average of more than $3 billion annually in returns on their pre-1993 equity over the three years.

The total annual cost to taxpayers--the sum of the change in government equity (measured by changes in the difference between market and accounting value) and the forgone return--has ranged from negative $1.1 billion in 1994 to $20.7 billion in 1995 (see Table 5). The three-year average annual cost is $7.8 billion by that measure.
 

That measure is not biased upward by the assumption that the entire difference between market and accounting value results from the government's credit enhancement. Although other unbooked assets push market value above accounting value, unbooked liabilities have the opposite effect. In fact, without the infusion of federal equity, the housing GSEs could have a market value that is less than accounting value by an excess of unbooked liabilities over unbooked assets. Note also that the cost measure attributes all of the 1994 decline in the net value of unbooked items to a fall in government equity even though changes in the value of other unbooked items could have caused the decline.

Thus, the amount of the federal credit enhancement might be larger than the difference between market and book value. Indeed, an extrapolation from an earlier estimate suggests that the federal equity position is $3 billion more than the difference. If so, $7.8 billion is an underestimate of the average annual cost to the government. Nor is there much consolation in finding that a handful of large, fully private financial firms have a ratio of market to book value in excess of 3 to 1. GSE status is so integral to the financial performance of Fannie Mae and Freddie Mac that no one could assert with any confidence that those enterprises would be among the elite of financial firms if they were fully private.
 

Estimating the Annual Funding Savings to the GSEs

Measuring the cost of GSEs to the government is a difficult task in part because the government gives the GSEs nonpriced benefits rather than cash payments. It is useful, therefore, to develop estimates from more than one perspective if possible. Recent estimates of the reduced funding cost that sponsored status confers on the GSEs provide an independent measure of the government's cost. The logical link between the funding cost savings to the GSEs and the cost to the government is that those savings are what the GSEs and their potential competitors would pay for those benefits in a competitive sale.

To measure the estimated cost savings realized by Fannie Mae and Freddie Mac as a result of their GSE status requires using as a benchmark the costs they would face as fully private entities. Creating such a benchmark requires making a number of assumptions:

• That without GSE status, securities issued by the two enterprises would have identical credit ratings of Aa (one or both might be rated A, unless management was willing to take steps to shore up the credit quality of the enterprises);
  
  
• That the mix of callable and noncallable debt and MBSs issued by Fannie Mae and Freddie Mac is unaffected by GSE status; and
  
  
• That the per-dollar funding subsidy for callable and noncallable debt is constant over the 1991-1995 period (that assumption is likely to be violated because the subsidy on debt varies with the riskiness of the enterprise, shareholder capital, and market conditions).

Those assumptions permit the funding subsidy on debt issues to be developed from the estimates for the 1991-1994 period, in which savings were 105 basis points on callable debt and 46 basis points on noncallable debt.⁽⁸⁾ For all debt funding, the weighted-average savings were about 70 basis points (see Table 6 for the calculated subsidy values for debt securities for both Fannie Mae and Freddie Mac).⁽⁹⁾ During the 1991-1995 period, that component of the taxpayer subsidy to the housing GSEs rose from $0.9 billion to $2.6 billion per year.
 

For mortgage-backed securities, the cost savings from GSE status is 40 basis points based on a variety of estimates that place the yield on MBSs issued by GSEs at 25 to 60 basis points lower than the yield on the highest-rated fully private issues.⁽¹⁰⁾ That estimate is approximate because of differences in structure between private MBSs and those issued by GSEs. For example, the two types of securities differ in the method of allocating credit risk between issuer and investor. In addition, as is also true for the debt subsidy rate, those spreads change with market conditions and--more recently--with the growing acceptance of private MBSs.⁽¹¹⁾

The estimated value of GSE status for cost savings on mortgage-backed securities is $3.8 billion in 1995, a $1.2 billion increase from 1991. Total cost savings on debt and MBSs combined from the GSE status now totals $6.5 billion per year.
 

Benefits, Subsidies, and Value Added to the GSEs

GSE status clearly confers substantial funding benefits on Fannie Mae and Freddie Mac. Some observers have claimed, however, that not all of those benefits result from federal risk-bearing. Some of the benefits could stem from other features of government-sponsored status (for alternative ways of splitting up those federal benefits, see Box 3).
 

The housing enterprises have argued, however, that a portion of the benefit is the result of the superior liquidity that agency status gives to securities issued by the enterprises. In the view of the housing GSEs, the liquidity premium from GSE status is an added value in excess of the direct cost to taxpayers.⁽¹²⁾ Accordingly, Fannie Mae and Freddie Mac believe that the cost of GSEs to the government should only include the expected outlays associated with the insolvency of the enterprise, which they believe is less than the cost savings in raising funds.

If the benefits provided to the housing enterprises had no alternative uses, that narrow definition of cost would be sensible. But in fact potential competitors of the GSEs--not to mention Fannie Mae and Freddie Mac themselves--would willingly pay for the right to issue huge volumes of highly liquid, near-Treasury-quality securities. Awarding that right exclusively to the enterprises without charge has a cost. The cost is not having the financial resources that firms would pay to the government for that right. Under Fannie Mae and Freddie Mac's definition of costs, any value that the government might create through its guarantees has no cost to the government because that added value belongs to the housing GSEs.

The financial activities of Fannie Mae and Freddie Mac may also add value to the cost of inputs, but that value must be carefully delineated by source: government and the housing enterprises. The government's contribution of added value belongs to the government and society at large. When that value is transferred to the GSEs, it is a subsidy to the recipient and a cost to the government. By contrast, value added by the government-sponsored enterprises through their own efforts is not a subsidy and would still accrue to the firms if they were fully private. It is the government's contributed value that Fannie Mae and Freddie Mac are battling for in the privatization debate.

Consider a parallel case: the government's cost of giving away a $1 bill. The direct cost to the government of the paper and ink required to produce a bill is less than 5 cents. But the recipient is willing to provide the government with goods and services worth a full dollar in exchange for the bill. In the terms used by Fannie Mae and Freddie Mac, the government has "created value" of more than 95 cents. The housing GSEs would then claim that the government's cost of giving away a $1 bill is less than 5 cents because the rest is value added by the government's action. That restricted measure of cost understates the value that would be sacrificed by giving away dollar bills and hides a transfer of public resources to private entities. The free grant of GSE status to Fannie Mae and Freddie Mac has precisely the same effects.
 

Assessing the Estimates

In light of the number and variety of assumptions required to estimate the subsidies to the housing GSEs, estimates need to be corroborated whenever possible. Thus, the funding benefit to the GSEs, though smaller by about $2 billion per year on average over the 1993-1995 period, is roughly consistent with the estimated cost of government equity. As stated above, the per-dollar subsidy on debt is significantly greater than the per-dollar subsidy on MBSs. That finding is consistent with the higher profitability of the portfolio noted earlier and management's decision to increase net mortgage portfolios--financed by debt--much faster than outstanding MBSs. During 1991 through 1995, the ratio of portfolio holdings to outstanding MBSs for both Fannie Mae and Freddie Mac increased from 21 percent to 37 percent. The estimated subsidy on callable debt is substantially larger than that for noncallable debt, which is consistent with the rapid increase in the use of callable debt by the GSEs. During the 1991-1994 period, Freddie Mac increased the percentage of long-term debt with features for call, or other downward rate adjustments, from 38 percent to 88 percent. For Fannie Mae, that percentage rose from 22 percent to 55 percent. For 1995, the percentages were 73 percent for Freddie Mac and 48 percent for Fannie Mae.

The debt estimates used in this study are consistent with the low end of the range of a previous estimate (70 to 154 basis points) and with a direct comparison of the Lehman Brothers Agency Bond Index with the Lehman Brothers Aa Bond Index for 1991 through 1994 (101 basis points).⁽¹³⁾ They are also consistent with independent estimates of the amount of subsidy passed through to home buyers and the earnings performance of the housing GSEs, as well as with the current structure of the secondary market for conforming mortgages and the Housing and Community Development Act of 1992. Each of those consistencies is discussed below.

Consistency with the Subsidy Pass-Through and Earnings

With the cost advantage in funding that is not available to any fully private intermediary, the housing GSEs are able to offer higher prices for mortgages than their competitors. Those higher prices bid for mortgages push down interest rates paid by home buyers and forestall entry by potential competitors. Fannie Mae and Freddie Mac thereby pass through some of the funding subsidy to home buyers.⁽¹⁴⁾

A recent study estimates that the GSEs reduced interest rates on average by about 35 basis points during the 1989-1993 period.⁽¹⁵⁾ That amount is the estimated difference between interest rates on "jumbo" mortgages (those above $207,000 currently) and conforming mortgages ($207,000 or less) that Freddie Mac and Fannie Mae are eligible to purchase. The rationale for using the difference between jumbo and conforming rates as the measure of the subsidy pass-through is that the housing GSEs cannot participate in the jumbo market--which is served by fully private, competitive firms. The difference in rates between the two markets, therefore, approximates the amount by which GSEs are passing through the benefits of GSE status in lower mortgage interest rates.

According to Unicon Research, conforming rates were 15 to 60 basis points lower than jumbo rates for 30-year fixed-rate mortgages originated over the 1989-1993 period (for a representative sample of their estimates, see Table 7). In estimating that pass-through, the analysts controlled for ratios of mortgage loan to home value, loan size, and new versus existing homes. Unicon's results provide a wider range of estimates than had been found previously. However, the estimated spread diminishes over the period, with the smallest differences reported in the latest year.⁽¹⁶⁾ The decline in the estimated spread between jumbo and conforming mortgage rates coincides with the growth of activity by private conduits in the jumbo mortgage market and the increasing liquidity of that market. Those developments would tend to reduce the rate in the jumbo mortgage market and diminish the spread between jumbo and conforming mortgages.
 

The Unicon Research study also finds a discontinuity or "stacking" in the distribution of mortgages at the conforming ceiling. Specifically, the analysts find that the probability that a mortgage will be at or below the conforming limit is much greater than the probability that the mortgage will be just above the conforming limit. That finding confirms that a difference in interest rates exists and that borrowers adjust the size of their loan to take advantage of the lower rates available on conforming mortgages. Stacking appears to diminish with the fall in the estimated spread between jumbo and conforming mortgages.

The interest rate savings accruing to the housing GSEs of $6.5 billion in 1995 is consistent with the estimated 35 basis-point subsidy pass-through and with the two agencies' earnings performance. As shown in Table 8, 35 basis points on the $1.2 trillion in mortgages funded in 1995 would have consumed $4.4 billion.⁽¹⁷⁾ That amount leaves a difference of $2.1 billion as retained by the housing GSEs, which could raise their return on book equity from 11 percent a year to 20 percent.
 

Some observers have suggested that Fannie Mae and Freddie Mac should be credited with reducing interest rates on all conforming, conventional, fixed-rate mortgages and not just those that they buy or securitize. Clearly, the willingness of the housing GSEs to buy conforming mortgages at posted prices has reduced rates on some mortgages held by others. That rate reduction, however, does not change the conclusion that Fannie Mae and Freddie Mac are retaining $2 billion in benefits that they could use to reduce mortgage interest rates further if they chose to do so.

Fannie Mae has chided some analysts for failing to back up the claim that the housing GSEs overcharge home buyers with estimates of the extent of the overcharge--that is, the amount by which interest rates on conforming mortgages are higher than warranted by Fannie Mae's costs of funding.⁽¹⁸⁾ As Table 8 indicates, the GSEs use $4.4 billion to reduce mortgage interest rates by 35 basis points. If they passed through the entire estimated subsidy to home buyers, interest rates on conforming mortgages would be about 50 basis points lower than rates on jumbo mortgages. By retaining a portion of the subsidy, the housing GSEs are charging home-buying families 15 basis points more than would be possible with complete pass-through. The enterprises have claimed, however, that the benefit to home buyers and to the GSEs is not a cost to taxpayers (see Box 4).
 

The estimated retained subsidy accounts for over 40 percent of the GSEs' pretax net income in 1995. Moreover, the significance of the retained subsidy to the earnings of the GSEs appears to be increasing over time (see Table 9). The housing GSEs are subject to federal income taxes. On an after-tax basis, the retained subsidy in 1995 is $1.5 billion but accounts for the same share of after-tax income as pretax income. (The cost estimates shown in Table 9 are used as the estimated cost of the housing GSEs for the remainder of this report.)
 

Consistency with Market Structure

Estimates of the gross and passed-through subsidy are also roughly consistent with the observed features of the market for securitizing conforming mortgages. Specifically, because the government has given the subsidy exclusively to the housing GSEs, they have been able to maintain a duopoly for decades, despite sustained high earnings and the active presence of potential competitors--using the same technology--in the parallel market for jumbo mortgages. More precisely, either a large subsidy to Fannie Mae and Freddie Mac has kept other firms out of the conforming market, or some unidentified competitive advantage of the GSEs has done so. But if the explanation is some factor other than the federal subsidy, Fannie Mae and Freddie Mac would not lose that advantage with privatization.

Other studies have found that the housing GSEs appear to be "tacitly colluding duopolists in the conforming segment of the conduit market."⁽¹⁹⁾ If one recognizes a mutual dependence of interests by the GSEs, along with the nature of the subsidy provided, and that Fannie Mae and Freddie Mac are profit seekers, those findings are consistent with estimates that some but not all of the subsidy is passed through to home buyers.

It is uncertain that the housing GSEs succeed in maximizing profits. But it appears that they have limited their expansion short of the level of lending that would be observed in a purely competitive market in which the subsidy was available to all competitors and all of the subsidy was passed through to borrowers.

Consistency with the 1992 Act

The Congress seemed aware of the nature, extensive size, and disposition of the GSE subsidy when it adopted the Housing and Community Development Act of 1992. Specifically, in light of the subsidy estimates, one can understand the rationale behind creating the Office of Federal Housing Enterprise Oversight (OFHEO) to regulate the levels of shareholder capital in the two enterprises. The new authority of the Secretary of Housing and Urban Development to establish affordable housing goals for Fannie Mae and Freddie Mac is also understandable in light of those subsidy estimates.⁽²⁰⁾

Establishing OFHEO, with its heavy emphasis on ensuring soundness by enforcing an adequate level of shareholder equity, appears intended to limit the ability of management to increase the risk exposure of taxpayers. OFHEO's mandate is to determine--by simulating the effects on the housing GSEs of two extreme stress scenarios--the levels of market shareholder equity that would be sufficient to absorb all losses so that even in those two cases taxpayers would not be called on to cover losses.

The experience of bank regulators in attempting to promote the safety and soundness of federally insured institutions with risk-based capital standards is not entirely encouraging.⁽²¹⁾ Risk takes so many forms that it may be impossible for a central authority to measure and monitor risk with sufficient precision to protect the taxpayers' interest. OFHEO believes, however, that its capital standards will represent a significant advance over the standards used by bank regulators.

The willingness of the Congress to impose presumably costly but affordable housing goals on Fannie Mae and Freddie Mac is also consistent with Congressional understanding that the GSEs are retaining some subsidy as surplus. The language of the statute speaks of "activities relating to mortgages on housing for low- and moderate-income families involving a reasonable economic return that may be less than the return earned on other activities" [emphasis added].⁽²²⁾
 

Sensitivity Analysis

Measuring the funding cost savings to Fannie Mae and Freddie Mac from GSE status is hampered by a scarcity of comparable securities issued by fully private entities. That problem, however, is not just a matter of bad luck. It is an inevitable consequence of sponsored enterprises. Only the implicit federal guarantee permits an intermediary to issue MBSs without the credit enhancements required of private issuers. Sponsored status also enables GSEs to issue callable debt that is regarded as free of the risk of default. That quality of credit makes those securities much easier to sell than callable debt offered by private firms. The lack of comparable private issues is part and parcel of the advantages of GSE status to the enterprises.

One cannot address satisfactorily the shortage of comparable agency and private securities by measuring spreads on the small number of comparable issues over a long period, since yield spreads between agency and private issues vary substantially over time. To take an extreme example, the funding cost advantage of Fannie Mae was probably in the neighborhood of 1,500 to 2,000 basis points in the early 1980s when the GSE was economically insolvent. Its funding advantage is now much less than it was then, but small changes in market and enterprise conditions can also significantly affect the value of GSE status.

In commenting on a draft of this report, both Fannie Mae and Freddie Mac noted the limited number of comparable GSE and private securities. One of Fannie Mae's solutions was to ask dealers what they thought the spreads might be on comparable securities if they existed. Freddie Mac suggested using only noncallable debt in the comparisons, even though over 40 percent of Freddie Mac's debt maturing in 1997 and beyond is callable. The Congressional Budget Office prefers statistical analysis of the limited data. The estimated subsidies for debt that this report uses are based on an analysis that controls for many of the factors that affect spreads. It uses data for the shortest, most recent period of estimate (1991 through 1994). At least nine matched pairs of bonds were available for comparison in every month in the estimation period.

Nonetheless, a limited sample size means that the estimates of the cost advantage of funding vary widely around the true value. The small sample, however, does not mean that the estimates are biased. They may be too low; they may be too high. To test for the significance of bias, the funding cost subsidies are recalculated under four alternative sets of assumptions (see Table 10).
 

First, to allow for the possibility that both the MBS subsidy and pass-through rates are now lower than they were during the estimation period, those rates are reduced to 30 basis points and 25 basis points, respectively. The net effect of those two adjustments is to raise the retained subsidy as a share of net income from 42 percent in the base case for 1995 to 48 percent.

Second, to address the claim advanced by the GSEs that they pass through 100 percent of the funding subsidy on MBS issues, the MBS and pass-through rates are both set at 25 basis points. The retained subsidy declines to 38.6 percent of 1995 net income. That is a significant reduction from 42 percent, but the net income of the GSEs still depends heavily on the retained subsidy from debt-financed portfolio lending.

Third, estimates of the debt subsidies are reduced by 10 basis points for Fannie Mae and Freddie Mac. That calculation reduces the retained subsidy as a share of net income by about 4 percentage points from the second alternative, but the retained subsidy still makes up more than one-third of the net income of the housing GSEs. Fourth, it is necessary to assume that the GSEs pass through 50 basis points in lower mortgage interest rates in order to reduce the estimate of the retained subsidy to near zero in 1995.
 

Summing Up

Government-sponsored enterprises are costly to the government and taxpayers in that they receive a benefit for which others would pay a substantial sum. In addition, the GSEs are retaining a substantial share of that benefit for management and shareholders rather than passing it through to home buyers. That conclusion holds for a variety of estimating methods and assumptions. If one measures the costs of the GSEs as the annual change in the value of the government's credit enhancement and the annual return on previously invested capital, then the average annual cost to the government--though highly volatile--averaged $7.8 billion per year during the 1993-1995 period. If measured as the estimated funding cost advantage that the GSEs receive, the benefit is currently worth $6.5 billion annually.

The estimate of funding cost savings excludes the direct savings to the GSEs from the special provisions of law that confer agency status, such as the exemptions from state and local income taxes and from SEC registration fees. The income tax exemption alone provides a net benefit of more than $250 million at current income levels (see Table 11). In addition, the General Accounting Office has estimated that the exemption by the Securities and Exchange Commission is worth more than $100 million annually.⁽²³⁾ Thus, the $6.5 billion annual value of the subsidy to the housing GSEs is a significant understatement of the cost to the government and taxpayers.
 

1. Private credit-rating agencies evaluate the risk of default on debt and mortgage-backed securities and assign letter grades to each issue. The highest rating is Aaa, the next highest is Aa, and so on down to C. For a discussion of the rating process, see Congressional Budget Office, Controlling the Risks of Government-Sponsored Enterprises (April 1991), pp. 50-52.

2. Super Aaa is not a grade assigned by the credit-rating agencies. It is used here to indicate that the market regards GSE securities as lower in risk than those issued by the highest-rated fully private firms. Aaa-rated issues have a small but positive expected default rate. See Congressional Budget Office, Controlling the Risks of Government-Sponsored Enterprises, p. 65.

3. Edward J. Kane and Chester Foster, "Valuing and Eliminating Subsidies Associated with Conjectural Government Guarantees of FNMA Liabilities" (research report prepared for the Department of Housing and Urban Development, August 30, 1985).

4. Panos Kostas, "Government-Sponsored Enterprises: Their Role as Conduits of Credit and as Competitors of Banking Institutions," Banking Review, Federal Deposit Insurance Corporation, vol. 8, no. 2 (Spring 1995), pp. 13-24.

5. An example of an unbooked asset is the relationships the housing GSEs have established with lenders in the primary mortgage market. An example of an unbooked liability is the difference between the accounting provision for credit losses and the actual expected credit losses.

6. Vassillis Lekkas and Robert Van Order, "Taking Stock: Applying Financial Theory to Freddie Mac and Fannie Mae Stock," Secondary Mortgage Markets (Fall 1991), pp. 24-27.

7. Kane and Foster, "Valuing and Eliminating Subsidies."

8. A basis point is one-hundredth of a percentage point.

9. These calculations use the same funding savings rate for all debt regardless of maturity because comparable estimates are not available for maturities of one year or less. If the savings on debt of less than a year was only one-half (or one-fourth) of the savings rate on longer-term debt, the estimated cost savings on debt in 1995 for both GSEs would be about $600 million (or $920 million) lower than shown in Table 6. See footnote 17, however.

10. Eric I. Hemel, "GSEs and Mortgage Finance," U.S. Investment Research (New York: Morgan Stanley, November 23, 1994), p. 3; John Goodman and Wayne Passmore, Market Power and the Pricing of Mortgage Securitization, Finance and Economics Discussion Series No. 187 (Federal Reserve Board, March 1992); Brent Ambrose and Arthur Warga, "Implications of Privatization: The Costs to FNMA and FHLMC," HUD Studies (May 1996), p. 194. See also Salomon Brothers, "The Mortgage Securities Market--First Quarter 1991 Review and Current Outlook," (April 15, 1991); and Michael Lea, Housing and the Capital Markets, Working Paper No. 8 (Cambridge, Mass.: MIT Housing Policy Project, 1988).

11. James Rothberg, Frank Nothaft, and Stuart Gabriel, "On the Determinant of Yield Spreads Between Mortgage Pass-Throughs and Treasury Securities," Journal of Real Estate Finance and Economics, vol. 2 (1989), pp. 301-315.

12. Benjamin Hermalin and Dwight Jaffee dismiss the claim that investor liquidity preference is a significant cause of lower interest rates on GSE securities. See Hermalin and Jaffee, "The Privatization of Fannie Mae and Freddie Mac: Implications for Mortgage Industry Structure," HUD Studies (May 1996), pp. 263, 273.

13. Kenneth Thygerson, "Federal Mortgage Credit Agencies and the Decline in Thrift Charter Value" (paper presented at the annual meeting of the American Real Estate and Urban Economics Association, December 1990). One reviewer of the commissioned papers (Lawrence White) uses a 55 to 60 basis-point spread on GSE and fully private debt issues, but then adds basis points for the GSE exemption from state and local income taxes, exemption from SEC registration fees, relief from the necessity to purchase pool insurance for MBS issues, and so forth. Also see Brent Ambrose and Arthur Warga, "Reply to Shilling," HUD Studies (May 1996), p. 223.

14. Given the nature of the subsidy and the market, this partial pass-through is consistent with maximizing profits. See Hermalin and Jaffee, "The Privatization of Fannie Mae and Freddie Mac," pp. 282-289.

15. Robert Cotterman and James Pearce, "The Effects of FNMA and FHLMC on Conventional Fixed-Rate Mortgage Yields," HUD Studies (May 1996), pp. 97-168.

16. Previous estimates of the pass-through were 30 basis points in the 1986-1987 period and 10 to 23 basis points in 1987. See Patric Hendershott and James Shilling, "The Impact of the Agencies on Conventional Fixed-Rate Mortgage Yields," Journal of Real Estate Finance and Economics, vol. 2 (1989), pp. 101-115; and ICF Inc., Effects of the Conforming Loan Limit on Mortgage Markets (prepared for the Department of Housing and Urban Development, March 1990).

17. This estimate applies the same 35 basis-point subsidy pass-through rate to all mortgages purchased by the GSEs, including adjustable-rate, multifamily, intermediate-term, second, and government-insured home loans, because no comparable estimates of the pass-through on those mortgages are available. If the pass-through rate on those loans is only one-half (or one-fourth) of the pass-through on conventional, fixed-rate, conforming mortgages, the estimated pass-through for 1995 would be $900 million (or $1.3 billion) lower than shown in Table 8.

18. The Federal National Mortgage Association's review of Benjamin E. Hermalin and Dwight M. Jaffee, "The Privatization of Fannie Mae and Freddie Mac: Implications for Mortgage Industry Structure," HUD Studies (May 1996), p. 315, footnote 3, states: "Remarkably, they offer no estimate of how much they think prices exceed marginal costs."

19. Hermalin and Jaffee, "The Privatization of Fannie Mae and Freddie Mac," pp. 227, 241-253. See also Goodman and Passmore, Market Power and the Pricing of Mortgage Securitization.

20. The policy of using the GSEs to increase affordable housing, however, conflicts with the government's intention that the enterprises operate safely, without risk of failure. See John C. Weicher, "The New Structure of the Housing Finance System," Review, Federal Reserve Bank of St. Louis, vol. 76 (July-August 1994), pp. 47-65.

21. "Bank Regulators Back Taking More Flexible Approaches to Risk," American Banker (February 26, 1996), p. 4; Steven Grenadier and Brian Hall, Risk-Based Capital Standards and the Riskiness of Bank Portfolio: Credit and Risk Factors, Working Paper No. 5178 (Cambridge, Mass: National Bureau of Economic Research, 1996).

22. 12 U.S.C. 1716, 106 Stat. 3994.

23. Letter from James L. Bothwell, Director, Financial Institutions and Market Issues, General Accounting Office, to Richard K. Armey, Majority Leader, U.S. House of Representatives, March 25, 1996.

---