CHAPTER 4

Managing Risk

AS DAVID MAXWELL LIKED TO SAY, “FANNIE MAE WAS POLITIcal to its bone marrow.” It hardly could have been otherwise. It had the benefit of a special congressional charter that gave it valuable advantages over purely private companies, and it operated in a mammoth financial market—residential mortgages—whose smooth functioning was vital to the health of the American economy. If that weren’t enough, it was located in Washington, D.C.

A Unique and Controversial Charter

Fannie Mae’s 1954 charter, as amended in 1968, gave it the mission “to provide supplementary assistance to the secondary market for home mortgages by providing a degree of liquidity for mortgage investments, thereby improving the distribution of investment capital available for home mortgage financing.” To ensure that its efforts were focused on low- and moderate-income home buyers, it was limited to the purchase or guaranty of U.S. residential mortgages below a certain dollar amount (an amount that in 1981 was allowed to increase each year, based on the change in an index of national home prices). It could participate only in the resale or secondary mortgage market, with a primary market lender—typically a bank, thrift, or mortgage bank—that chose to do business with it. And it had to maintain a presence in all parts of the country, in bad times as well as good.

Balancing these restrictions—and to enable the company to attract the private capital necessary to carry out its mission—the charter contained benefits that fell into two broad categories: those that created the perception of a special relationship with the U.S. government and those that lowered the cost or increased the marketability of the company’s securities. Charter attributes falling into the first category included authorization for the secretary of the Treasury to purchase up to $2.25 billion of Fannie Mae debt at Treasury’s discretion, an exemption from state and local income taxes, and the appointment by the president of the United States of five members to Fannie Mae’s board of directors. Charter attributes that benefitted the company’s securities included the facts that they were eligible to be purchased or sold by the Federal Reserve in open market operations, were eligible for unlimited investment by national banks and most state banks and thrift institutions, and were considered government securities for purposes of the Securities Act of 1933 and the Securities Exchange Act of 1934 and were therefore exempt from SEC registration requirements. All of these benefits were intentional, and collectively they enabled Fannie Mae to do business on a scale and at a cost that would not have been possible without them.

From the time it became a shareholder-owned company, Fannie Mae faced criticism and pressures from three main sources: free-market advocates, actual and potential competitors, and the two principal bank regulators, the Federal Reserve and the Treasury. Free-market advocates opposed Fannie Mae on ideological grounds, because it used government-provided advantages to channel more funds into housing than would have been available otherwise. Actual and potential competitors of Fannie Mae were concerned about the company’s market power; they found it in their commercial best interest to ally with the free-market viewpoint and did. The Federal Reserve and Treasury were mistrustful of a profit-making entity with the implied backing of the government but not controlled by it, and they felt Fannie Mae’s government sponsorship gave it the ability to expand its business at will (to the disadvantage of the competitive position of the banks they regulated) as well as an incentive to take excessive risk.

Each of these groups of critics sought something different. The free-market ideologues wanted Fannie Mae not to exist at all or to be “fully privatized.” Most of Fannie Mae’s competitors wanted the company to be more tightly regulated and its business powers to be limited and strictly defined (though a few favored full privatization). And the Fed and Treasury, at least initially, were focused on constraining Fannie Mae’s size and risk taking. Any of the company’s critics or opponents seeking changes to its role in the market, risk, or regulation in a manner that advanced their interests came to Washington to try to get them.

Maxwell became Fannie Mae chairman seven months after Ronald Reagan’s election as president and after Reagan had filled key positions at Treasury, the Office of Management and Budget (OMB), and the Council of Economic Advisers with committed free-market conservatives. The philosophy of the Reagan administration toward both the rescue of the thrifts and the plight of Fannie Mae was made clear in the Report of the President’s Commission on Housing, published in 1982. That report said, “The Commission foresees a future in which government should be a participant in housing finance only in those areas where the private sector cannot provide needed services at a reasonable cost to borrowers.” The report left no doubt about the role envisioned for Fannie Mae and Freddie Mac in that future: “after FNMA’s [Fannie Mae’s] financial condition clearly has stabilized,” it said, the two GSEs “should become entirely private corporations.”

The policies of the Reagan administration toward Fannie Mae were based purely on ideology. No one on Wall Street or in the commercial banking industry was yet asking for any restraints to be put on us or for any of our agency attributes to be pared back (that would come later). Wall Street in fact was quite happy with us. They underwrote and made markets in the debt we issued, and we were in the process of initiating a new mortgage-backed securities program they would profit from as well.

We were fortunate that in the early stages of Reagan’s first term, proposals to cut back Fannie Mae’s agency attributes were not high policy priorities. Had we been in their gun sights, we may well not have survived. A crucial element of our turnaround was quickly doing more types of profitable new business, much of which required HUD approval. HUD was considered a backwater in the administration. The administration paid little attention to the department, and the new HUD secretary, Sam Pierce, paid little attention to us. We had a good friend and ally in HUD’s undersecretary, Don Hovde, who had been president of the National Association of Realtors and knew us well. We would bring our proposals for new programs—ranging from adjustable-rate mortgage purchases to multifamily investments to our new mortgage-backed securities business—to Hovde, who would put them in front of Pierce and ask Pierce to sign them. He invariably did. It was more than a little ironic that while the heads of Treasury and the OMB privately were telling Maxwell that Fannie Mae should be done away with, we were walking through the front door at HUD and obtaining the administration’s permission to expand.

Fannie Mae barely turned profitable in 1983, and to our critics within the administration, this was a signal that it was safe to come after us (notwithstanding the fact that a subsequent rebound in interest rates pushed us back into the red in 1984).

The assault started with the Office of Management and Budget and its director, David Stockman. A three-term congressman from Michigan before being tapped to run the OMB at the age of 35, Stockman was a close ally of Jack Kemp, a leading supply-side economist and coauthor of the Kemp-Roth tax cuts that were the centerpiece of the Reagan economic platform. Stockman’s preferred policy objective for Fannie Mae was full privatization immediately, but he knew Congress would not support that idea and so instead proposed to charge a “user fee” on the debt we issued to fund our mortgage purchases. The theory behind user fees was that because our charter benefits enabled us to borrow more cheaply in the capital markets than fully private firms, the government ought to be entitled to attach a fee to our debt as compensation for those benefits. Once put in place, the user fee could be raised over time to a level where eventually the cost of our debt and the debt issued by our competitors would be equal.

We took our case against user fees to our allies and Congress. Our best argument was a purely economic one: any fees added to our debt would cause us to raise the yield requirements on the mortgages we purchased, and as such it would be home owners who ultimately would bear the cost. To make that point clear, we called user fees a “home ownership tax.” That proved to be very effective labeling, and we were able to defeat user fees on Capitol Hill. But it was a constant battle. User fees were proposed every year, in varying amounts and with varying rationales. For us it was the vampire issue; we never could definitively kill it.

A second political problem for Fannie Mae in the mid-1980s was the increasingly vocal opposition of the California savings and loans. The S&L industry had been granted significantly broader asset powers in the Garn–St. Germain Act, but the large California S&Ls elected to stay concentrated in residential mortgage lending. They considered a healthy Fannie Mae a serious threat to their business. The executives who engaged most actively against us were Jim Montgomery, CEO of Great Western Financial, and Herb and Marion Sandler, co-CEOs of Golden West Financial. They argued that we had outlived our business purpose because the two conditions we had been chartered to alleviate—credit-driven housing cycles and regional mismatches between the need for and the supply of mortgage credit—had been addressed by thrift deregulation. They seconded the administration’s contention that our charter gave us unfair advantages over fully private firms, and they maintained that our portfolio exposed the U.S. taxpayer to excessive risk. They also aggressively supported user fees.

Full Privatization Gets Full Consideration

Fierce opposition to our charter from the Reagan administration and the California thrifts, along with the growing realization that there were no obvious ways to make this opposition disappear, led Maxwell to seriously investigate full Fannie Mae privatization in 1987.

We had looked at privatization before. In 1983 we had hired Lazard Frères to help us investigate privatization alternatives for discussion with our board, in an effort we called Project One. Project One took place while we still were struggling to return to profitability, however, and even with Lazard’s help we were unable to come up with any practical version of a fully private Fannie Mae worth recommending to our directors. We shelved Project One in the fall of 1984.

Three years later, several things were different. We were solidly profitable and had growing confidence in our future prospects. We had just raised over $100 million in fresh capital in February of 1987, selling six million shares of stock in the United States and two million shares in Europe. And we were strongly motivated by a desire to escape the political pressures we were facing. Maxwell hired Shearson Lehman Brothers to be the consultant on the new privatization study, largely because he wanted the assistance of the head of their Washington, D.C., office, Jim Johnson. Johnson had accompanied Maxwell and me on the road show for our European equity issue, and I had gotten to know him then. I was appointed Fannie Mae’s lead on the project shortly after it began.

Applied to Fannie Mae, the term “privatization” was a misnomer. Technically, we were private: we were owned by our shareholders. What our opponents and critics meant by Fannie Mae privatization was giving up some or all of the benefits of our federal charter while retaining our business restrictions. That would have been suicidal. The one conclusion we did come to in Project One was that a Fannie Mae bereft of agency attributes but limited to the residential mortgage business would not have been able to survive. For all of the talk by the California S&Ls about Fannie Mae’s “unfair advantages,” thrifts had their own federal benefit—federal deposit insurance—along with two captive federal agencies, the Federal Home Loan Banks and Freddie Mac, both under the wing of a supportive regulator, the Federal Home Loan Bank Board. Commercial banks had similar federal ties. It might have been expedient for thrifts and banks to ignore their federal benefits when referring to themselves as “private” and pressing their cases in the political arena, but for us to have ignored them when putting together a privatization business plan would have been foolhardy.

What we called the Shearson study had three components: developing versions of our existing businesses that could operate without the implicit support of the federal government, identifying new asset powers that would give us business and risk diversification while drawing on expertise we either already had or believed we could acquire, and outlining a feasible path of transition from the government-sponsored Fannie Mae to the new private company.

I took the lead on the non-government-sponsored enterprise versions of our portfolio investment and credit guaranty business. We and the Shearson team concluded that if a private Fannie Mae had a portfolio business at all, it would have to be far smaller than the $90 billion we then held, and that it would be more difficult to manage and be able to operate only in certain favorable market environments. The credit guaranty business of a private Fannie Mae would be smaller as well, because it would provide less value to lenders. Investors would not be willing to pay as much for mortgage-backed securities guaranteed by a private Fannie Mae, and the higher capital a private Fannie Mae would have to hold would raise the guaranty fees it charged.

The virtual certainty of having much smaller existing businesses in a non-GSE Fannie Mae raised the stakes for finding new products to offer, new businesses to enter, and new corporate forms of operation as a private company. Together with the Shearson staff, we investigated a number of possibilities, many of which seemed promising. But when we looked at implementation issues, we understood what we were up against.

Any proposal for Fannie Mae privatization required legislation to replace our existing charter with a new one. We believed it would be possible to find support for giving up our government-sponsored enterprise status, although the Realtors and the National Association of Home Builders—and probably the Mortgage Bankers Association—would fight it because we would be replacing a GSE Fannie Mae that was very valuable to their business with one that would do much less for them. But we became convinced that there was little chance of getting backing from anyone for any reasonable form of new business powers, for fear we might become competitors with them. Further, there were daunting transitional challenges involved in an orderly liquidation of the GSE Fannie Mae as the new private entity was trying to establish itself. Advocates of Fannie Mae privatization already were talking about imposing an exit fee on the new non-GSE Fannie Mae to prevent the transfer of any monetary value we may have received and retained as a result of our federal charter.

All of these uncertainties and imponderables proved too much to overcome. Once we took a privatization proposal to Congress, we would have no control over the outcome. We had a fiduciary duty to our shareholders to safeguard their interests and could not in good faith initiate a privatization process we had come to believe had so little chance of success. After a year and a half of work, we shut the Shearson study down. It had led us to precisely the same place as Project One: a realization that the status quo for Fannie Mae was preferable to any alternative we or anyone else had been able to come up with.

Our work on privatization crystallized the situation we faced. There was no realistic possibility of changing Fannie Mae in a manner that would appease our ideological critics and competitors and at the same time maintain our value to home buyers and our shareholders. Whether we liked it or not, political opposition would be a permanent aspect of our existence. To fulfill our charter mission, we would have to be able to both run our business successfully and defend ourselves in Washington against attempts to make running that business more difficult.

Congress had not seemed particularly receptive to the argument of the California S&Ls that we had unfair advantages over thrifts and banks that needed to be pared back, and it certainly didn’t agree with our ideological opponents that the government should not be involved in mortgage finance at all. But Congress did take seriously the notion that the benefits we provided to home buyers might not be worth the risks our activities posed to taxpayers. Fannie Mae’s near-failure already had given Maxwell an economic reason to make the company world class in managing our two principal business risks, interest rate and credit risk. Increasingly, we came to understand that to maintain strong and bipartisan support in Congress, we had a compelling political reason to do exactly the same thing.

Rethinking Interest Rate Risk

Interest rate risk was our most difficult challenge. Fannie Mae had been created to serve as a secondary market purchaser of the 30-year fixed-rate mortgage (30-year FRM), which was the government’s solution to the problems borrowers had with short-term balloon mortgages during the Depression. Thirty-year FRMs are extremely consumer friendly. Borrowers with a 30-year FRM who make a fixed monthly payment for 30 years can pay off their mortgage entirely, but the loan includes an option that allows them to prepay it at any time without penalty (although they do have to pay certain fees to obtain a new loan). When interest rates go up, borrowers can hold onto the loans they have—which at that point have below-market interest rates—and when interest rates go down, they can pay them off and take out new ones that cost less. It’s a terrific deal, which is why since adjustable-rate mortgages became widely offered in the early 1980s, U.S. home owners have preferred FRMs to ARMs by more than three to one.

The prepayment option that makes a 30-year FRM so attractive to borrowers is what makes managing its risk so difficult for lenders. The most comprehensive measure of the life of a fixed-income security, or bond, is its duration. While a bond’s maturity measures only the date on which the final repayment of principal is received, its duration incorporates all of the interest and principal payments from that security and weights them by when they occur. Large payments made early in the life of a bond shorten its duration, and large payments made late lengthen it. For most bonds, the duration calculation is straightforward, but for 30-year FRMs, it is complicated by the fact that very few mortgages remain outstanding for their entire lives. Typically, they get repaid—either when the borrowers sell their homes, or when interest rates fall and they refinance. And exactly when the mortgage gets repaid has a huge impact on its duration, because that repayment is by far the largest single cash flow from the mortgage that takes place.

The challenge of managing a group of fixed-rate mortgages is, in a nutshell, that their durations are not knowable. Mortgage durations depend on interest rates, which can’t be reliably predicted. Managers of fixed-rate mortgages therefore need to be able to protect themselves from changes in mortgage durations by reacting to their effects as they occur. If they are unable to do so—as was the case with Fannie Mae and the thrifts in the late 1970s and early 1980s—serious problems can ensue.

Ideally, the durations of a company’s assets and debt should be equal. When Maxwell became chairman of Fannie Mae, the company had a mismatch, or “gap,” of three years between the average duration of its mortgages (five years and two months) and the average duration of its debt (only two years and two months). We set a goal of reducing the portfolio’s duration gap to less than one year as quickly as market conditions permitted. We made relatively little progress through the end of 1984, but when mortgage rates fell in 1985 and 1986, we finally were able to get the duration gap down to under a year. Once there, our strategy was to initiate a series of “rebalancing” actions—adding fixed-rate or adjustable-rate mortgages and shorter-or longer-term debt—in response to future interest rate changes to keep the duration gap below one year, with a goal of eventually achieving a match.

I had joined Fannie Mae in 1982 as chief economist but began working with the company’s senior business officers on strategic and financial management issues almost immediately. In 1985 I was given responsibility for recommending interest rate risk management strategies for the portfolio. Then, in late 1987, Maxwell decided to put a single executive in charge of all aspects of the portfolio business—determining what types and amounts of mortgages to buy, when to buy them, how to fund the purchases initially, and when and how to rebalance the portfolio as interest rates changed. He asked me to take that position, and I did.

Having a dedicated portfolio group enabled us to take a fresh look at all aspects of the business. We kept coming back to its structure. Active rebalancing could keep the portfolio’s risk within acceptable limits, but it was not particularly efficient and could be very expensive if interest rates were volatile. Borrowers of fixed-rate mortgages had the advantage of a powerful option to repay or not repay the loan at their discretion. The fixed-term debt that we issued, in contrast, would stay on our books until it matured or we repurchased it.

The solution to this asymmetry turned out to be the obvious one. Since there were options embedded in the fixed-rate mortgages we bought, we needed to have options embedded in at least some of the debt we used to finance those mortgages. Option-based, or “callable,” debt in fact existed; it had been common in the corporate bond market for decades. Callable debt has a stated final maturity but also a specified earlier date on or after which the issuer can redeem, or “call” it. Callable debt would be more expensive to issue initially, but over time it had the potential to lower our rebalancing costs by an even larger amount and to greatly reduce the sensitivity of the portfolio’s net income to interest rate changes. Callable debt, however, had not been used to any significant extent by Fannie Mae or any other government-sponsored agency. In order to make it the foundation of our new portfolio strategy, we would have to reinvent the agency debt market.

Fannie Mae had a selling group of about 20 Wall Street firms, each of which had an “agency desk” specializing in the sales and trading of our and other government agencies’ securities. The traders and salespeople staffing these desks were highly effective at selling our noncallable (or “bullet”) debt. But callable debt involved options, which involved mathematics and quantitative analysis, and we had to give the staff on our dealers’ agency desks both the tools and the training to sell it. As we did that, we experimented with their corporate finance departments on different combinations of final maturities and call dates to find out which best met our risk management needs while appealing to the broadest range of investors.

On the investor front, we got a serendipitous break from Michael Milken, the pioneer of the junk bond phenomenon of the mid-1980s. It had not been uncommon during that time for holders of highly rated corporate bonds to see their bonds suddenly downgraded—and have their prices drop sharply—after the company that issued them fell victim to a hostile takeover by a junk-bond financed acquirer. This “downgrade risk” had caused new issuance of callable corporate bonds to all but dry up. Just as we were starting our callable agency debt program, we stumbled upon this large base of traditional corporate debt investors thrilled to be offered an alternative callable security from a high-quality issuer with negligible risk of falling prey to a leveraged buyout.

While developing our callable debt issuance capability, we worked to define specific risk management objectives that incorporated callable debt funding. As a shareholder-owned company, we did intend to take some interest rate risk, but we also believed that equity investors would reward us for a steady pattern of earnings growth. We made two important decisions. First, we determined that financing the portfolio with a roughly equal mix of callable debt and bullet debt would give us the best chance of achieving the combination of strong earnings growth, high returns, and low earnings volatility we sought. Second, we set strict internal guidelines for rebalancing the portfolio, linked to changes in its duration gap. As long as our duration gap remained within a range of plus or minus six months, we would not rebalance. (Mortgage durations longer than debt durations made the duration gap positive; the reverse was a negative gap.) When the duration gap was between 6 and 12 months—positive or negative—we would undertake actions to bring it back to 6 months or less, but with no set time limit in order to accomplish the rebalancing as cost effectively as possible. And if our duration gap ever exceeded 12 months, we would conduct immediate rebalancing to bring it back to 12 months or less by the next monthly reporting period, irrespective of cost. These guidelines would remain in effect for over a decade.

Devising standardized callable debt structures, developing a callable debt investor base, and reorienting our dealers to be able to distribute callables through their agency desks and to make active markets in them afterwards took us over two years. But by the beginning of 1990, we had done it. In 1990, 70 percent of the long-term debt we issued was callable, and soon the callable market was broad and deep enough to finance rapid portfolio growth. We were able to grow our portfolio by 50 percent during 1992 and 1993. At the end of 1993, the portfolio was $190 billion in size, its duration gap was a negative two months—very close to a match—and over half of it was financed by callable debt.

The portfolio business had been transformed completely from what it was when Maxwell first came to the company. In his bestselling book Good to Great, noted business author Jim Collins said:

With the use of callable debt, defined risk management objectives, and disciplined rebalancing guidelines, we now could legitimately claim that Fannie Mae was far better able to manage the risk of holding fixed-rate mortgages in portfolio than any other financial institution, including commercial banks and thrifts. In defending ourselves against our critics and detractors in Congress, this would be an invaluable advantage.

Credit Has Its Price

It was an irony of Fannie Mae’s charter restrictions—which limited us to dealing in residential mortgages—that Maxwell’s sole alternative for saving his company from the effects of excessive interest rate risk was to buy ARMs and other higher-risk products that significantly increased the amount of credit risk we took. Fannie Mae’s credit losses rose from a negligible $5 million in 1981 to $170 million in 1985. In just four years, loans acquired to give ourselves enough time and income to fix our interest rate risk problems were causing serious problems with credit risk. But the strategy worked. In 1985 the portfolio’s net interest income—the interest income on our mortgages less the interest expense on our debt—finally turned positive. Together with our growing fee income, positive net interest income enabled us to report a profit in 1985, even with our sharply higher credit losses. That year we made a thorough study of the company’s defaults to determine which risk factors were primarily responsible for them. Our analysis identified some obvious culprits, and in August 1985 we announced a sweeping set of revisions to our underwriting guidelines to address them.

Underwriting was one aspect of our credit risk management we needed to address; pricing was the other. When we purchased a loan for portfolio, our pricing theoretically included compensation for credit risk as well as interest rate risk, but the credit risk amount was nowhere calculated or specified. While we might have been able to get away with this in the portfolio, we could not with our credit guaranty business, where we were paid a monthly fee for guaranteeing the timely payment of interest and principal on the loans that went into our mortgage-backed securities. Initially we picked guaranty fees that were deliberately conservative, but we knew that at some point we would need a way to determine those fees with more precision.

I took the lead on that task. In 1985 I set up a financial analysis group and staffed it with individuals with expertise and experience in mortgage credit analysis and financial modeling. Fannie Mae’s size, national presence, and longevity gave us enormous amounts of historical information on single-family loan performance by product and risk type (which we had used to determine where and how to tighten our underwriting standards). We set out to use this information to assess how economic and financial variables affected the credit performance of the different types and characteristics of mortgages we would be guaranteeing.

The mechanics of credit guaranty pricing are straightforward. You make estimates of the incidence of default and the severity of loss on the loans that go bad (which constitute your losses) and projections of prepayment rates on the loans that stay good (which produce your income), and then you determine a fee that is high enough to cover your expected losses, prepayments, and administrative costs while providing a market return on your invested capital. As we approached this task, however, we faced a complication. Fannie Mae was not in the mortgage-backed securities business when the company was privatized in 1968 and HUD was designated as its regulator. We had a statutory capital requirement for our portfolio investments, which were held as assets on our balance sheet, but no capital requirement for MBS guarantees, which were not. Before we could determine how to price our credit guarantees, we first had to figure out how much capital to put behind them.

Having just survived a brush with insolvency from excessive interest rate risk—and having also seen how quickly our credit losses could rise—our goal was to devise a capitalization framework for our single-family guaranty business that could withstand a defined level of worst-case, or “stress,” credit losses. To quantify that loss amount, we generated several hundred random paths for future interest rates and home prices, and then we employed the results of our loan performance analyses to project mortgage prepayments, defaults, and loss severities in each of the individual scenarios. We defined the stress loss amount as the one that was just worse than 99 percent of all the other loss amounts produced by this exercise. Put differently, the capital that would back our mortgage-backed securities guarantees would come from a loss scenario our analytic group estimated had only a 1 percent chance of occurring.

Once we had our methodology for setting capital requirements for our credit guarantees, pricing them became simple. There was only one combination of initial capital and guaranty fee that simultaneously would produce exactly our target return on capital in the loss scenario our credit pricing model assessed to be most likely, and also was just sufficient to cover the credit losses in the 1 percent probability stress scenario we wished to protect ourselves against. That was the amount of capital we would hold and the guaranty fee we would seek to charge.

Defaults from loans made before we tightened our underwriting standards in 1985 continued to drive our credit losses higher through 1988, when they peaked at $315 million, or 11 basis points as a percentage of our combined portfolio and mortgage-backed securities outstanding. From that point, however, our credit losses declined, both absolutely and as a percentage of our book of business. By 1993 our total credit losses had fallen to $244 million, or only four basis points as a percent of our book, which by then had grown to two and a half times the size of our 1988 book. That same year our average single-family MBS guaranty fee rate was 21 basis points. Our underwriting was solid, our guaranty fee pricing was comfortably covering our credit losses, and the capital we held to back our credit guarantee business was growing rapidly. As was the case with our management of interest rate risk, in 1993 we could credibly claim in Congress and elsewhere that our management of mortgage credit risk was second to none.