Jason Thomas, CFA

Department of Finance

George Washington University

December 14, 2009

Abstract

Some observers contend that the housing government-sponsored enterprises (GSEs), Fannie Mae

and Freddie Mac, were responsible for the financial crisis due to their financial support for the

subprime mortgage market. This paper reviews the available evidence and concludes that while

Fannie and Freddie played a role in the propagation of the financial crisis, they were hardly the

mortgages purchased and securitized by the GSEs. The GSEs responded by purchasing the AAA

pieces of subprime asset-backed securities (ABS) and later purchased large quantities of Alt-A

loans. However, the GSEs’ mortgage loan portfolios performed much better than the national

average and the losses that nearly brought down the financial system were on instruments wholly

unrelated to the GSEs. Moreover, critics of the GSEs focused on interest rate risk while the

problems that ultimately led to their insolvency related to market and credit risk.

1. Introduction

On September 7, 2008, the Federal Housing Finance Agency (FHFA), in conjunction

with the Treasury Department and the Federal Reserve Board of Governors, placed Fannie Mae

and Freddie Mac, the shareholder-owned housing government sponsored enterprises (GSEs) into

conservatorship. This legal status is similar to bankruptcy: top management was removed, the

remaining shareholders’ equity was nearly wiped out, and the government assumed day-to-day

operation of both enterprises. The government took this step because the losses at the GSEs had

accelerated in recent months and analyses performed by regulators suggested that they had

insufficient capital to withstand expected future losses.1

As part of the conservatorship, the Treasury Department pledged to invest up to $200

billion in both GSEs as necessary to ensure they maintained a positive net worth.2 This amount

was later increased to $400 billion on February 18, 2009.3 As of the 3rd quarter of 2009, Fannie

Mae had formally requested $55.6 billion and Freddie Mac had requested $50.7 billion in capital

injections, requiring taxpayers to provide $106.3 billion in financing just so the GSEs could

maintain a positive net worth (FHFA, 2009). Even with no additional losses, if the GSEs were

recapitalized so as to maintain core capital at pre-conservatorship levels, it is likely that

taxpayers would have to invest an additional $121 billion, although no policy decision on that

front has yet been made (FHFA Capital, 2009).

Given the magnitude of the losses at the GSEs and the fact that the financial crisis

seemed to have originated in the residential housing finance market, some commentators have

suggested that the GSEs bear most of the responsibility for the financial crisis. This contention

www.federalreserve.gov/newsevents/press/other/20080907a.htm, and

www.treasury.gov/press/releases/hp1129.htm.

2 See Treasury Senior Preferred Stock Purchase plan:

www.treas.gov/press/releases/reports/pspa_factsheet_090708%20hp1128.pdf.

www.ustreas.gov/press/releases/tg33.htm.

Page 3

is often augmented by a “we told you so” refrain from those who argued in the earlier part of the

decade that Fannie and Freddie posed a “systemic risk” to the financial system. That Fannie and

Freddie required an expensive taxpayer-financed rescue just years after they were identified as a

systemic risk provides an ephemeral confirmation for this line of reasoning.

However, closer analysis reveals this narrative to be quite unsatisfying. It is generally

agreed that the crisis began in the summer of 2007 in short-term collateralized lending markets,

not in the so-called “Agency” debt market. Moreover, it was the default rates on mortgages sold

into non-Agency pools that first caught investors’ attention and led to the failure of two Bear

Stearns-sponsored hedge funds in July 2007. If the GSEs were responsible for the financial

crisis, it seems odd, to say the least, that the onset of the liquidity crisis would occur in funding

markets they did not access and was precipitated by defaults on mortgages they did not purchase.

This paper relies on data from the Federal Reserve, the Government Accountability

Office (GAO), and Fannie Mae and Freddie Mac’s annual reports and monthly volume

summaries to assess the extent to which the data support or refute critics’ narrative about the

GSEs’ contribution to the financial crisis.

2. The GSE Policy Debate

Fannie Mae and Freddie Mac are Congressionally-chartered to provide liquidity to the

secondary mortgage market by buying mortgages from originators. They are unique in that they

are considered “Federal Government Agencies” even though they are owned by shareholders and

their debt is not explicitly guaranteed by the United States Treasury. The GSEs raise funds from

investors to purchase mortgages by issuing mortgage-backed securities (MBS) or what is

referred to as “Agency debt.” The acquired mortgages are then deposited in an off-balance sheet

special purpose vehicle (SPV) or sometimes held as whole loans on balance sheet. Although the

Page 4

GSEs retain the credit risk on the mortgage however its purchase is financed, the GSEs only

record the value of the mortgage on balance sheet when it is debt-financed. The GSEs also have

to record the value of the mortgage on balance sheet in those cases where they later issue debt to

repurchase the MBS originally issued to investors. The on-balance sheet portion of the GSEs’

mortgage holdings is called the “retained portfolio” and became the central issue in the policy

debate.

In April 2005, Federal Reserve Chairman Alan Greenspan testified before the Senate

Banking Committee to urge lawmakers to pass legislation to reform the regulation of Fannie Mae

and Freddie Mac. Greenspan put the challenge to lawmakers in stark terms: “To fend off

possible future systemic difficulties, which we assess as likely if GSE expansion continues

unabated, preventive actions are required sooner rather than later” (emphasis added). The

systemic crisis Greenspan suggested the Federal Reserve viewed as “likely” absent “preventative

actions” was thought to arise from the risks presented by the GSEs’ issuance of implicitly-

guaranteed debt to accumulate the retained portfolios. In addition to Greenspan’s testimony, the

Federal Reserve released two studies in 2005 that argued that much the market value of the

GSEs’ shareholders’ equity came from the implicit guarantee on the Agency debt (Passmore,

2005) and that the GSE portfolio purchases had a negligible impact on mortgage rates (Lehnert,

Passmore, & Sherlund, 2005). Greenspan recommended that Congress should enact a law to (1)

create a new regulator for Fannie Mae and Freddie Mac and (2) restrict the ability of the GSEs to

hold retained portfolios.

From 1998 to 2002, the combined mortgage portfolios of the GSEs more than doubled in

size, from $670 billion to $1.36 trillion (Monthly Funding Summaries). During this four year

period, retained portfolio purchases accounted for over 51% of the growth in the GSEs’ total

book of business. This was a 13 point increase from the end of 1997, when the portfolio

accounted for only 38% of total mortgage holdings. The GSEs increasingly relied on debt

instead of MBS to fund mortgages for two reasons: (1) it was more profitable to earn the spread

on mortgage coupons over Agency debt interest payments than it was to pass-through those

coupons (minus guarantee fees) to the holders of MBS; and (2) the fiscal position of the federal

government changed so dramatically at the end of 1990s that there was substantial foreign

investor demand for an alternative to Treasury securities. From 1995-1999, net foreign

purchases of Agency securities averaged $4.5 billion per month. Over the next five years, net

foreign purchases of Agency securities more than tripled to $14.9 billion per month (Treasury

International Capital, 2009). In its 1999 Annual Report, Fannie Mae predicted this shift, arguing

that its Agency debt will be “viewed by many investors as a liquid investment alternative to U.S.

Treasuries (page 89).” Roll (2004) argues that increases in GSE debt issuance were essential to

tap international capital markets and meet the growing demand for U.S. mortgage borrowing.

Unlike other issuers of asset-backed securities (ABS), Fannie Mae and Freddie Mac

retained the credit risk on their MBS in the event that the mortgages that collateralized the SPV

defaulted. As a result, increasing the share of mortgages financed through debt had no impact on

Fannie and Freddie’s exposure to credit risk. It did, however, increase the GSEs’ exposure to

interest rate risk because of the prepayment option embedded in mortgages. It was the interest

rate risk from funding negative convexity assets with debt that led Greenspan and others to argue

for limits on the size of the GSEs’ balance sheets. Jaffee (2003) argues that the GSEs interest

rate risk management systems were not robust to large movements in interest rates and also

failed to account for basis risk. Passmore (2005) estimates that market participants would

require the GSEs to hold much smaller portfolios and much more equity to deal with interest rate

Page 6

risk if their debt were not implicitly guaranteed. Hubbard (2003) finds Fannie Mae’s interest rate

risk management practices able to effectively manage interest rate risk over most conceivable

scenarios. Miller and Pearce (2003) and Miller (2006) find that the GSEs are less exposed to

large moves in interest rates than the thrift industry. Finally, Stiglitz, Orszag, and Orszag (2002)

find that the probability of an unfavorable interest rate move of the magnitude necessary to cause

a Fannie Mae default is substantially less than one in 500,000.

Interest rate risk is measured through a “duration gap” that provides a measure of the

effective duration of assets minus the effective duration of liabilities, adjusted for leverage.

When the duration gap is zero, interest rate risk is effectively “immunized” for any reasonable

change in interest rates. As seen in Chart 1, Fannie Mae and Freddie Mac had much different

tolerance for interest rate risk in the earlier part of the decade. In August of 2002, Fannie Mae’s

duration gap was –14 months, which suggests that its mortgage portfolio was prepaying at

speeds the risk managers did not anticipate, leaving the firm badly exposed to further declines in

interest rates. Interestingly, this dramatic deviation in risk did not show up in debt spreads, as

Page 7

the yield on Freddie Mac’s debt was actually 15 to 20 basis points wider than Fannie’s despite its

much safer risk profile.

In summary, the debate regarding GSE policy concerned the growth in the GSEs’

retained portfolios and the interest rate risk associated with leveraged investments in negative

convexity assets. Critics argued that the portfolios created systemic risks without providing

lower mortgage rates. Defenders of the GSEs argued that Agency debt was more attractive to

investors than MBS and that the interest rate risks were far overstated. The data from 2002-2003

suggest that Fannie Mae and Freddie Mac had much different interest rate risk management

systems in place but market participants did not penalize Fannie Mae for its relatively lax

standards. Interestingly, both sides in the debate seemed to agree that the credit risk posed by

mortgages holdings was trivial or otherwise undeserving of serious scrutiny. The reforms

recommended by the Federal Reserve would have allowed the GSEs to continue to aggregate

unlimited amounts of mortgage credit risk so long as these mortgage purchases were financed

through the issuance of MBS.

3. The Evolution of the U.S. Mortgage Market

As the policy debate raged on, the structure of the U.S. mortgage market was changing

dramatically. In 2003, Fannie Mae and Freddie Mac accounted for 52.3% of all residential

mortgage loans (Federal Reserve and Monthly Funding Summaries). Within three years, their

share of total outstanding mortgage debt (retained portfolio plus net MBS outstanding) had fallen

by 8 percentage points to 44%. During this same period, their retained portfolios declined to

13.6% of the total mortgage market from 23% in 2003.

This shift involve two related, but distinct, developments: (1) the share of total outstanding

mortgage debt financed by the issuance of non-Agency asset-backed securities (ABS) nearly

doubled over the course of only three years; and (2) the origination of non-traditional mortgage

products, like subprime and Alt-A loans, grew exponentially during this period. As seen in Chart

2, from 2003 to 2006, the net amount of mortgage debt held by asset-backed security issuers

(ABS Issuers) and “Finance Companies” increased dramatically. ABS Issuers are off-balance

sheet vehicles sponsored by banks and other financial intermediaries that finance mortgage

purchases by issuing notes to investors. Finance companies perform a similar function, but

generally issue notes that are the obligation of the company itself instead of being tied to the

underlying collateral. Finance companies can be a mortgage financier like GMAC, or an asset-

backed security conduit. The stock of mortgages used to collateralize non-Agency ABS (the

combination of the two subcategories) grew by more than 25% per year and exceeded 40%

growth in 2005 and 2006. From 2002-2006, the dollar value of outstanding mortgages financed

through non-Agency securitization increased by 239%, from $774 billion to $2.63 trillion. The

non-agency ABS market share roughly doubled during this period, from 13% of all residential

mortgages in 2002 to 26.7% of total household mortgage debt outstanding in 2006 (Fed Z-1,

Table 2 illustrates the growth in the non-traditional mortgage market. “Subprime” loans

are made to borrowers that have low credit ratings and other undesirable characteristics that

would make them unable to qualify for a traditional, prime mortgage. This segment of the

Page 11

Agency securitization market contracted.4 As shown in Chart 2, 2007 was also the year that

Fannie and Freddie’s combined book of business grew by over 10%, adding over $600 billion in

net new mortgage debt.

Not every mortgage loan purchased by ABS issuers was subprime or Alt-A; likewise, not

every subprime or Alt-A loan found its way into a non-Agency ABS pool. Yet, it is seems clear

that subprime and Alt-A mortgage markets would not have grown so dramatically without

securitization, and the non-Agency ABS markets would not have grown without mortgage loans

to serve as the underlying collateral for those securities. While it is not clear how much of the

net increase in mortgage debt came from subprime and Alt-A loans, it is clear that non-Agency

securitization financed 75% of the gross origination of these products (GAO, 2009).

It is important to recognize that this shift in the mortgage finance market also occurred in

the context of rapidly increasing housing prices in much of the country, but especially in

California, Arizona, Florida, and Nevada. As shown in Chart 4, by early 2005 the year-over-year

price increases in the ten and twenty city Case-Shiller composite index had been running at over

10% per year for three years. The dramatic price growth led some to argue that the move to non-

traditional products was driven by affordability issues (Kling, 2009), as much of the purchase

volume of subprime and Alt-A loans was likely homebuyers who could not have otherwise

purchased the home. This is consistent with previous Fed research (Feldman, 2001) which found

that Fannie and Freddie made a limited contribution to homeownership rates precisely because

4 It is important to note that the subprime and Alt-A loan originations in Chart 2 are volume or “flow” statistics that

capture the amount of new loans originated in a given year (GAO, 2009), while the non-Agency Securitization

measures the net change in the dollar value of mortgage debt held by these securitization vehicles. Therefore there

is not a one-to-one correspondence. Since, as shown in Table 3, over half of subprime loans and over one-third of

Alt-A loans were used to refinance existing debt, the volume statistics incorporate loans used to retire existing debt

in addition to net new mortgage obligations. Separating the net new debt from refinancing on these loans is

difficult, especially because over half of the subprime loans during this period were for “cash-out” refinancing. A

portion of these loans add incrementally to the aggregate amount of mortgage debt outstanding while most of the

proceeds go to retire existing debt.

Page 12

the marginal buyer could not afford the large down payments Fannie and Freddie generally

required.

However, it is just as likely as causality ran in the opposite direction. As is clear from

Table 2, only one-third of subprime originations and one-half of Alt-A loans were used to

purchase homes. The majority of subprime mortgages were actually cash-out refinancings,

Page 13

20% down payment only allows a borrower to leverage his savings five times; a 3% down

payment allows the same borrower to leverage his down payment 33-times. As confidence in

continued price increases and competition from new products reduced average down payments,

prices may have risen in response to the increased leverage in the system.

Whatever the functional form of the relationship between non-Agency ABS, subprime

mortgages, and housing prices, the alternative mortgage market came to a crashing halt in 2007,

after the year/year change in house prices turned negative. As seen in Chart 5 (Fed Z-1, 2007),

the annual net increase in mortgage debt had actually been lagging the net increase in the value

of household residential real estate from 2000 to 2005. This allowed more debt to be incurred

without disruption. That changed in 2006, when mortgaged debt increased by nearly $1 trillion,

on net, while the value of residential real estate grew by about $500 billion. In 2007, the

mortgage market added $600 billion in incremental debt but the value of the underlying

residential real estate collateral (including newly constructed homes) fell by $1.4 trillion, causing

the entire system to collapse.

Page 14

4. The GSEs Role in the Alternative Mortgage Market

The data in the previous section are entirely inconsistent with the hypothesis that Fannie Mae

and Freddie Mac were responsible for the ensuring crisis. Fannie Mae and Freddie Mac could

only purchase “conforming” mortgages judged to be of “investment” quality. Neither of these

terms is defined in law, aside from a formal cap on the size of the mortgages eligible for

purchase (Kling, 2009). According to annual reports, this generally meant that mortgage

borrowers had to have FICO scores of over 620, a down payment of more than 10%, and be fully

documented. Table 3 lists the percentage of loans low FICO or high LTV mortgages purchased

mortgages because these securities were “goals rich” (Lockhart, 2009). Fannie and Freddie were

required to meet affordable housing goals, set annually by the Department of Housing and Urban

Development (HUD) in accordance with The Federal Housing Enterprises Financial Safety and

Soundness Act of 1992. The purchase of non-Agency ABS backed by subprime mortgages

counted towards meeting these goals because the underlying mortgages were made to borrowers

Page 15

who otherwise could not have gained access to mortgage credit. Indeed, Fannie Mae suggested

in its 2006 Annual Report that “because subprime mortgages tended to meet many of the HUD

goals and subgoals,” the decline in subprime origination that year “has further limited our ability

to meet these goals” (page 21).

A typical non-Agency ABS involves a “waterfall” payment structure. Instead of a traditional

“pass-through,” the ABS note holders are paid according to their seniority, with senior tranches

getting paid first and suffering losses from defaults on the underlying collateral, if any, last. The size

of the tranches is chosen to achieve AAA ratings on the senior portion, with the amount of

subordinated securities adjusted as necessary to provide the overcollateralization required by rating

agencies (Hu, 2007). Fannie and Freddie were major buyers of the AAA piece of the ABS. The

remaining, junior pieces were then generally sold into collateralized debt obligation (CDO) pools.

According to Freddie Mac’s 2006 Annual Report, “more than 99.9 percent” of its non-Agency ABS

were rated AAA. Until 2007, Fannie Mae also bought exclusively AAA non-Agency ABS and

added only “limited amounts” of other investment grade, non-Agency ABS (Fannie Mae, 2007).

Neither GSE had any exposure to CDOs (Fannie Mae, 2007, p. 97; Lockhart, 2008).

Estimating the amount of subprime and Alt-A ABS purchased by the GSEs during this

period is difficult because of differing levels of disclosures provided. Freddie Mac provides the

dollar amount of “Other Mortgage Securities” purchased for each year, which include all non-

Agency ABS, including Prime Jumbo and other types of non-subprime ABS. From 2001-2007,

Freddie Mac acquired $527 billion of these securities, with the largest purchase volume coming in