A discussion of risk at Fannie Mae and Freddie Mac

A story in the March 6th American Banker led with the following sentence: “Fannie Mae and Freddie Mac are among the safest and soundest financial institutions on the planet, according to an analyst at Moody’s Investors Service.”

That doesn’t say much for our planet (or Moody’s, but that’s another story).

Fannie and Freddie do have indisputable strong points. Due to the implied backing of the United States government, they enjoy virtually unlimited access to the capital markets at funding costs that are below market. Together they dominate their market-their combined book of business of $2.6 trillion at March 31, 2002 constitutes 46% of the $5.5 trillion in home mortgage debt outstanding, up from 35% in 1996. They pay no local taxes, only national. Their credit experience has been terrific. They are adept and public and governmental relations. They each have great websites. Because of all these, each has enjoyed tremendous growth and profitability, with 2001 ROEs of 23.1% (Freddie) and 25.4% (Fannie), and 2001 net margins of 56% (Freddie) and 63% (Fannie).

If we were writing that American Banker article, however, we would have led with “Fannie Mae and Freddie Mac are two of the most highly leveraged, least diversified, under-reserved and over-politicized financial institutions on Earth.&rdquo What are the risks to Fannie and Freddie?

To answer this question, we review their financial condition. We spent most of our analytical time on Fannie Mae since it discloses more historical data than Freddie and is bigger (although Freddie is catching up-in 1996, Fannie, with $351 billion in assets, came in at twice the $173.8 billion of Freddie, and by 3/31/02 Fannie’s $808 billion in footings was only 25% more than Freddie’s $646 million). It is fairly safe to say, however, that with only a few exceptions, what goes for Fannie goes for Freddie, too.

If you look at Table 1-“The Evolution of Fannie Mae”-a few things jump out:

Fannie Mae is a big FAST-GROWING mortgage bond fund: In 1989, Fannie Mae earned net interest income of $1.2 billion and guaranty fees of $408 million, for a ratio of 2.9:1. In 2001, the company earned interest income of $8.1 billion and guaranty fees of $1.5 billion, a ratio of 5.5:1. In the first quarter, that ratio stood at 6.0:1. Net interest income has grown 17% annually over that 12 year span, while guaranty fees have grown just 11% annually. In the last five years ended 2001, net interest income has a CAGR of 18% while guaranty fees have grown just 4%.

A look at the pre-tax line and the net line show clearly that the growth engine of the company has been net interest income. Pre-tax income has grown about $5.3 billion from 1993 to 2001, and net interest income has grown $5.5 billion. Our conclusion is that guaranty fees roughly cover expenses while the marginal net interest income dollar basically falls to the bottom line. Interestingly, the company seems to be getting less for its guaranty than it used to: 2001 guaranty fees as a percentage of year-end net MBS (MBS outstanding less MBS in portfolio) were 17 basis points, the lowest in the 13 years.

The balance sheet leaves the tracks of this strategy. Every schoolboy knows that the GSEs have been eating their own cooking for some time now, buying or retaining more and more of the mortgage business volume that comes through their doors. As we calculate it, in 1989 Fannie kept just 5.1% of all MBS outstanding. At December 31, 2001, Fannie had retained $431.5 billion, or 33.4% of the $1.29 trillion in FNM MBS outstanding. At March 31, that percentage had increased to 33.9%. As a result, the MBS in portfolio has shown a CAGR of 35% from 1989 to 2001, a faster growth rate then MBS outstanding, which compounded at a 16% clip over the same time. The balance sheet as a whole has compounded 17% annually over the same span. (Freddie’s balance sheet has compounded at an even more heroic rate, nearly 29% per annum over the five years ended 2001.) According to FNM, in 2001 88% of the portfolio was fixed-rate.

Fannie Mae is a big LEVERED mortgage bond fund: At March 31, 2002, assets covered equity 38.93x. Put another way, equity as a percentage of assets was 2.57%. Factor in all of the MBS outstanding, which would include all of the MBS in the marketplace that Fannie guarantees, and the ratio to equity is 65.3x. What is interesting to see in looking at the history of Fannie’s leverage is that from 1989 to about 1996/97, the leverage ratios were on an improving or stabilizing trend, and it has been downhill ever since.

Fannie Mae has had miraculous credit results: Although single-family serious delinquency rates have been consistently in a range of .45% to .62% since 1992 (with an uptick in the latest year), the company’s credit loss ratio has steadily declined to almost nothing: In the first quarter of 2002 the company’s credit loss ratio was a measly .005%, or five-tenths of a basis point. The company’s allowance for losses-a balance sheet item-has barely budged since 1992. From 1989 to 1992 the company almost doubled its allowance to $780 million, but as a percentage of outstanding MBS, the allowance has only shrunk, from .2% of outstandings in 1989 to .059% of outstandings in the first quarter of 2002. Interestingly, since 1998 the company has disclosed a provision for losses-line 4 in the table, an income statement item-that is actually positive. This is because the recoveries from losses have exceeded the provision, with the end result being an almost flat allowance.

One item we noted from the company’s 2001 Information Statement was the fact that “Fannie Mae employs strategies to reduce loss exposure through resolutions other than foreclosure,” including workout alternatives and pre-foreclosure sales. FNM does not disclose the extent to which such policies have helped goose the credit numbers, but it does proudly say that in 2001, “loan workouts outpaced foreclosed property acquisitions for the third year in a row.” Outpace by how much? Unclear. But in 2001, the company disclosed a slight increase in foreclosed single-family property acquisitions, from 14,351 in 2000 to 14,486 in 2001.

Credit enhancements, which include primary loan-level mortgage insurance, pool mortgage insurance, recourse arrangements with lenders and other contracts, together provided for against credit losses on 33% of single family loans at 12/31/01 vs. 38% at 12/31/00. The primary reason for the drop-off is that the number of loans with a loan to value below 80% fell. The primary reason the number of loans with LTV below 80% fell is “property values enabled some borrowers with Fannie Mae loans to cancel their outstanding mortgage insurance.” However, credit enhancements absorbed a higher percentage of credit losses in 2001-85%, or $435 million out of $512 million in gross losses-than in 2000, when 80%, or $349 million out of $435 million in gross losses was absorbed by credit enhancements.

Seven mortgage insurers, all rated AA or higher by S&P, provided 96% of the total coverage in force (on $314 billion of single family loans). The seven are not named. Additionally, the unpaid balance of single family loans where Fannie Mae has recourse to lenders for losses is $42 billion; 59% of this amount is with investment grade counterparties (41% is with non-investment grade). 96% of the company’s Liquid Investment Portfolio was rated A or better.

Fannie’s ballyhooed disclosure regarding derivatives exposure reveals little but concentrated exposure: The remarkable thing when reading the disclosure is that the company didn’t disclose this information before, as any decisions to disclose are “voluntary.” Anyway, FNM says that it uses “only the most straightforward types of derivative instruments such as interest-rate swaps, basis swaps, swaptions, and caps, whose values are relatively easy to model and predict.” Further, Fannie says it uses derivatives primarily “as a substitute for notes and bonds it issues in the cash debt markets.”

In terms of notional amount outstanding and credit exposure, FNM is still a piker relative to large banks. The notional balance of the derivatives book stood at $533 billion at 12/31/01, up from $325 billion at 12/31/00 and $251 billion at 1/1/00. The latest figure represents 70% of debt outstanding, and 41% of total mortgage book, compared to 50% and 30% for 2000 and 45% and 26% in 1999. Of the $533 billion notional, 40% are pay fixed (avg. 6.21%)/receive variable (avg. 2.47%), and 41% are caps and swaptions. Swaptions give FNM the option to enter into swaps at a future date, which mirrors the economic effect of callable debt.

All of the firm’s derivatives contracts are OTC (and not exchange-traded), which makes makes them less liquid and subject to greater credit risk. Over 99% of the transactions in 2001 were with counterparties rated A or better. While the company has executed derivatives transactions with 23 different companies, eight counterparties represented 78% of total notional amount outstanding.

The exposure to credit loss on derivatives is estimated by calculating the cost, on a present value basis, to replace at current rates all outstanding derivatives contracts in a gain position. Fannie’s exposure at 12/31/01 was $766 million, against which the company held $656 million in collateral, for a net exposure of $110 million. This net replacement cost represents less than 2% of Fannie’s 2001 pre-tax income. Just five counterparties accounted for 98% of the exposure on derivatives.

95% of the derivative credit loss exposure is on contracts that have five years or more to maturity. Longer-term contracts pose greater credit risk than shorter-term contracts.

There is not a lot of information on prepayment speeds. Here is one fact: At 12/31/01, 85% of the gross unpaid balance of mortgages outstanding were 7.99% or lower, with 36% at 7% or lower. At 12/31/00, 76% were at 7.99% or lower, with 26% at 7% or lower.

Fannie Mae’s interest spread in the first quarter of 2002 was the highest since 1990: For the first quarter, the yield on FNM’s investments came in at 6.49% against a borrowing cost of 5.52%, to net a spread of .97%. Its net interest margin, never below the 1.01% clocked in 1999 and 2000, was 1.15% in the period.

Fannie Mae’s tree can’t grow to the sky. Fannie sees itself continuing its growth, bounded only by the growth in the mortgage market itself. We decided to see what FNM would look like in five years if we grew the relevant line items for 2001 at the compound rates seen from 1996-2001. This very rough exercise shows a company with net interest income at about 10 times guaranty fee income, a book of business of about $2.6 trillion, and on-balance sheet leverage of 90 times.

Table 2, with a summary of Freddie Mac’s last six years, tells basically the same story as the Fannie table. In Table 3, we did some simple addition to provide a sense of the order of magnitude of Fannie’s and Freddie’s dominance of the mortgage market. If we also grow FRE’s book of business from 2001 by its growth rate of the last 5 years, the total of the two would equal $4.8 trillion. If we apply the growth rate of total mortgage debt outstanding over the past five years to see where national mortgage debt would be in five years, the GSE book would grow to 60% of all mortgage debt outstanding by 2006, a staggering amount.

Graphs 1-3 attempt to put the business of Freddie and Fannie in context. As we’ve alluded to, Freddie and Fannie constitute a growing and dominant position in the mortgage market. According to the Fed’s Flow of Funds data, home mortgage debt outstanding stood at $5.5 trillion at the end of the first quarter of 2002, a 10.5% increase over the prior comparable period. For the five years ended Dec. 31, 2001, the CAGR of home mortgage debt outstanding is 8.5%. Contrast that with the growth in FNM’s book of business outstanding-clearly the agency is growing faster than the market.

On the credit front, the Mortgage Bankers Association data show that the MBA delinquency rate is on a deterioriating trend. Graph 4 shows that since hitting its recent low of 3.72% in March 2000, the rate has risen to a peak of 4.87% in 9/01, easing slightly to 4.65% in March 2002. The interesting compare here is the FNM delinquency rate. It has always been lower than the MBA index, which FNM attributes to a number of factors-quality of the borrowers, type of collateral. But what interests us in this matter is not the relative size of the delinquency rates but the relatively low volatility in the number for FNM.

The credit backdrop to all this is found on the next three graphs. Graph 5, owners’ equity as a percentage of household real estate, reveals a leveraging trend. Graph 6, debt service as a percentage of disposable personal income, shows that the easing of the debt service burden that began after the fourth quarter 1986 turned up in 1992-93 and is back to mid-80s levels. The last data point from the fourth quarter of 2001 shows an increase to 14.3%. Graph 7, mortgage debt service as a percentage of overall debt service, peaked in 1992-93, which shows that the easing of the debt service burden to that point may have been more a function of deleveraging of other forms of consumer credit. That process bottomed as consumers started to use their credit cards more in the mid to late 90s. We should watch this ratio as well, because the more prominent mortgage debt is within a rising overall debt service burden the more at risk it is.

The last two graphs show the GSEs in the context of the larger fixed income market. Graph 8 shows that GSE debt (the obligations of the entities themselves as opposed to the mortgages) is becoming a larger component of the debt markets. In 1990, GSE debt only represented 4.6% of overall gross public debt. In the first quarter of 2002, GSE debt was 22.8% of the overall debt number. Just as important, as Graph 9 points out, foreign central banks have shown an increasing appetite for agency paper (we suspect that to foreign holders the agencies are like buying a risk-free Treasury with extra yield-what’s not to like?). The conclusion here is that funding this hungry maw of growth is requiring a greater and greater constituency of lenders.

The question that is begged from the preceding, is So What? Nothing has slowed this juggernaught so far. What are the risks to FNM? First, leverage: Observers of the credit markets can appreciate the complexity of managing a bond fund that is levered at 39x, on the asset side and the debt side of the balance sheet-and, just as important (if not more so), the off-balance sheet challenges of running a derivatives book to hedge exposures and understanding the safety and soundness of the mortgage insurers behind the one-third of the book that enjoys credit enhancement. Just as one example, the upheavals of September and October 2001 illustrated the risk of the convexity trade to FNM, as duration shortened dramatically on the asset side as rates dropped and FNM found itself with its assets to liabilities match move to 10 months. FNM scrambled mightily to correct this move, although the company wouldn’t disclose what actions it took. Similar sharp moves in rates would pose difficulties.

A related point is the diversification of the FNM and FRE investment portfolio. There is none. Oh, there is geographic diversification and there is some structure diversification, but there is no product diversification and the credit risk is essentially of one type. Does that make the portfolio more or less risky? Easier or harder to hedge? LTCM was a large leveraged investment portfolio with an extremely diverse set of assets, and their genius failed when they all went wrong at the same time. FNM is a large leveraged investment portfolio with basically one kind of asset. FNM’s genius may be that it picked an asset and a structure that can never go wrong; we say never say never.

Second, credit: The company acknowledges that its credit experience will deteriorate in the coming months. “Credit losses may finally increase this year, after five years of declines. But we don’t expect them to rise by a large dollar amount,” said CFO Tim Howard on the 4th quarter conference call. With a de minimis allowance for losses already, there doesn’t seem to be much room for error and credit changes will more likely surprise to the upside.

Third, size: How will FNM be able to maintain its growth trajectory? It depends on two basic things-the availability of funding for growth and the ability of the originators to create product. As for the former, we believe that as long as the GSE story is intact, FNM and FRE will continue to enjoy easy and unexamined access to the markets. Any questioning of that story would change this state of affairs. As for the latter, the GSEs now account for over three-quarters of all conventional and conforming mortgage debt extended in this country. As they reach for more assets and into newer businesses like subprime and multifamily, this puts more pressure on their insurers, counterparties and other credit enhancement vehicles.

Fourth, headline risk and regulation: As public servants like to say in their more grandiose moments, “I serve at the pleasure of my constituents.” FNMA and FRE serve two masters with different objectives-the federal government that chartered and regulates them, and the stock market that looks for growth. The last five years of financial results would lead any observer to conclude that management of these two entities has been listening more to the second master, possibly to the detriment of the first. The bigger, more leveraged and more complex Fannie and Freddie become, the more the first master will take notice.

New disclosure on Fannie Mae’s website reveals that Chairman Franklin Raines owns stock and options that at today’s price of $78 are worth about $23 million (assuming 100% vesting). Certainly nothing to sneeze at, but not a fortune in modern-day terms. However, if the stock were to get to 100, Raines’ holdings would be worth $62 million; $120 per share gets him to close to $100 million. Now we’re talking. Can anyone doubt that management is incented to keep up the appearance of Fannie Mae as a growth stock?

“Market participants,” wrote Laurie Goodman in the 2/26/02 issue of UBS Warburg’s Mortgage Strategist, “have stopped reacting to poorly thought out, poorly substantiated articles which display a lack of understanding about MBS business.” She might be right about market participants (for now), but the WSJ op-ed pieces, poorly written and argued as they might be, were on to something. The fact remains that there are other parties that have just as much ability to affect the business of the agencies as “market participants.” We don’t believe it will take much drilling down to make the regulators and the market take notice and start calling for change. They already have. We don’t doubt that most of the GSEs many and far-flung constituents-except maybe the common stock holders-would be a little more comfortable with slower growth, better disclosure about risk counterparties and a profile that wasn’t so leveraged to ever-rising housing prices.

June 21, 2002
By Jeremy Diamond, Executive Vice President
Annaly Capital Management