"Most people who are not in a coma have already refinanced," joked Doug Duncan, chief economist for the Mortgage Bankers Association of America, at a speech last month. Indeed, the new national pastime in the United States seems to be mortgage refinancing, as low interest rates and rising home values have sent homeowners flocking to lenders in record numbers. "Refis", as they are commonly called, have had a measurable and substantial impact on the financial condition of the average homeowner, the value of our nation’s housing stock, and the growth in the economy. What is less well understood is the way this activity affects portfolios of mortgage-backed securities. Depending on the portfolio, the effects could be substantial. This paper attempts to explain—to a non-practitioner—how this works and makes an attempt to handicap future trends in prepayments.
Refis have been on a roll. In its latest report, the Mortgage Bankers Association said its index of applications to refinance an existing mortgage came in at 5433.4 in the week ended January 17, 2003, up from 4507.6 in the prior month and 2768.3 in the prior year. The MBA refi index hit an all-time high of 6926.9 in October; anytime the index is over 1000 it is considered to be a refinancing boom, and the index has been over 1000 for 2 years, the longest such stretch, and it has averaged over 4000 since July 2002.
In December the Federal Reserve published a report on mortgage refinancing during 2001 and the first half of 2002 in the U.S. "In many cases," the report begins, "refinancing has resulted in a lower interest rate and a reduction in monthly mortgage payments, which have allowed homeowners to spend or save that portion of their incomes no longer dedicated to servicing their mortgage debt." According to the report, 96% of the refinancings during the period obtained a lower rate, with the average interest rate declining from 8.65% to 6.82%. At the same time, the maturity of the average refinanced mortgage was 29 months longer than that of the average original.
"When they have refinanced," the report continues, "many homeowners have liquefied some of the equity they accumulated in their homes by borrowing more than they needed to pay off their former mortgage….They used the funds raised in so-called cash-out refinancings to make home improvements, to repay other debts, or to purchase goods and services or other assets." According to the report, almost half of all homeowners who refinanced took equity out of their homes. The average amount "cashed-out" was about $27,000. Tables I and II below set forth details on the size of the cash outs and the use of proceeds. The net effect of lower rates and raising capital through cash-outs increased consumption materially and helped boost GDP growth.

There is another side to the refinancing boom: It has a profound impact on mortgage investors. To explain, let’s begin by reviewing some basics on the mortgage market. When a bank or thrift or mortgage finance company originates a mortgage loan, it can either hold the mortgage in its own portfolio or it can sell the mortgage into the secondary market. The secondary market serves two primary functions in the mortgage industry: First, the cash that the secondary market investors pay originators for mortgages can be used to originate more mortgages for home buyers. Second, the mortgages that are purchased are pooled and used to create bonds called mortgage-backed securities (MBS).
When a mortgage investor buys a mortgage-backed security, he or she is buying a bond that represents an ownership interest in these pools of mortgage loans. In the most basic MBS structure, the principal and interest that the homeowner-borrower is scheduled to pay each month, plus any unscheduled payments—such as prepayments from refinancings—are typically passed through to the MBS investors according to pool ownership percentages. A borrower can refinance for economic reasons or personal reasons, but refinancing activity is primarily driven by changes in interest rates. Chart 1 illustrates how refinancing, as measured by the Mortgage Bankers Association’s Refinancing Index, increases when interest rates fall and decreases when they rise.
Mortgage-backed securities are different from most other fixed-income instruments because of the homeowner’s option to prepay, or call, his or her loan in whole or in part at any time before maturity. As a result, the owner of an MBS is short a call option to the borrower. The prepayment option makes the average term, yield and performance of an MBS uncertain because the investor is not sure when principal will be returned. For a plain vanilla pass-through MBS, the average term of a mortgage-backed security shortens as rates fall. This results in the MBS underperformance relative to treasuries because principal is returned, through prepayments to the MBS, sooner than expected and the MBS is not outstanding as long as originally expected. The opposite is also true: the average term of a MBS lengthens as rates rise, thereby increasing its value relative to treasuries because principal is returned later than originally expected and the bond is outstanding longer than expected. (Most other fixed income securities rise in value as rates fall and vice versa; this characteristic of MBS is called negative convexity.)
The market has developed two methods by which prepayment speed is defined. Prepayment information, which takes into account only prepayments and not regularly scheduled mortgage payments, is reported by Fannie Mae and Freddie Mac and awaited by the market every month. The two most commonly used measurements of prepayment speed are the Constant Prepayment Rate (CPR) or the percentage of the Public Securities Association series (PSA). The CPR starts with the percentage of the outstanding balance of a mortgage pool or pool of mortgages that prepays on a monthly basis and then presents it on an annualized basis. For example, a $100 MBS pool in which $1 prepays in a month has a 1% monthly single month mortality; the CPR annualizes this rate to arrive at a CPR of 11%. The PSA works on the same principle but takes into account the impact of the age of a mortgage on its chance of prepayments.

Chart 2 above, prepared by the UBS Warburg, shows that the propensity to prepay is affected by how much financial incentive exists for the borrower. The horizontal axis calculates the incentive to refinance as the rate on existing mortgages less the current rate on a new mortgage. The vertical axis is a prepayment rate calculation like the CPR. As the difference in the old and the new rate widens, the prepayment rate will increase, but there is a leveling off once the incentive gets to 150 to 200 basis points.
Prepayments drive MBS performance. The way in which prepayments drive MBS performance depends on the speed of those prepayments and the investor’s estimate of the speed at purchase, but the most important determinant of the way MBS performance is affected by prepayments is the price paid for the security: In general, if an MBS is purchased at a discount to par, faster prepayments will improve its yield. If a mortgage- backed security is purchased at a premium, faster prepayments will reduce its yield. Last, if an MBS is purchased at par, its yield is unaffected by prepayments.
To see how prepayments affect performance, let’s look at a specific security currently owned in one of our portfolios: FN 607777, so named because the pool of mortgages underlying the security has been numbered 607777 and it has been guaranteed as to payment of principal and interest by Fannie Mae. It is a pass-through MBS in which the weighted average coupon paid by the underlying mortgage-holders is 7.8% but pays a 7% coupon to the MBS holder (the difference principally goes to the servicer and to Fannie Mae). Issued on November 1, 2001, it had an original balance of $14,540,734, consisting of 98 different loans all originated by Countrywide from 30 different states —the top five in order of weighting currently are Texas (21%), California (19%), Arizona (5.5%), Missouri (4%) and Florida (3.8%). Of the original issue amount, at the end of December there was 75% still outstanding, or $10.9 million.
Recall that prepayments decrease the total yield on a bond purchased at a premium. This is because over the life of the bond, that premium has to be amortized. The faster prepayments come in, the faster that premium is amortized, and amortization eats into yield. The daunting-looking wall of numbers in Table III below lays out the effect of increasing or decreasing prepayment speeds—CPR in the right hand column—on Total Yield on an investment of $10 million of FN 607777 at three different prices: a discount price of 98 (shown below as .98), a premium price of 103 (shown below as 1.03) and par. As can be seen in the column headed Amortization (gain/loss), while the coupon is always 7% (divide interest by face), prepayments either help, hurt, or have no effect on yield depending on the price paid.

The CPR for FN 607777 in the tables above is as reported. It is interesting to observe how the CPR changes month to month. For example, in June 2002 the CPR was close to zero, while just a month later the annual CPR shot up to 33.5. And during that month of July 2002, the bond bought at 103 saw its total yield fall to 5.772% from 6.972% because the amortization of the premium went from a small step in June to a giant leap in July. Simple bond math for a typical corporate bond with a stated maturity shows us that a bond bought at par will have a yield to maturity equal to its coupon and a bond bought at a premium will have a yield to maturity lower than its coupon. For a mortgage-backed security, however, if prepayments turn out to be faster than expected, the yield to maturity of a security bought at a premium will be even lower. Conversely, if the prepayments on an MBS bought at a premium are slower than expected, the yield to maturity will turn out to be higher.
In a low yield environment like we have today, most of the MBS available in the market is priced at a premium. With refinancing rates at an all-time high, then, the prepayment effect on yields is the biggest challenge facing an MBS portfolio manager. Clearly, a portfolio with an average price of 105 will have a harder time holding its yield in times of high refi activity than a portfolio with a lower average price. As with all investment choices, there is a trade-off between buying a security at a premium as opposed to buying a security at par or at a discount. In the same interest rate environment, premium securities have less duration than discount or par securities. The result of having a lower duration is less price volatility on the underlying security and therefore the portfolio. Additionally, premium securities will provide more yield pick-up in the event of a slow down in prepayments. Moreover, the market values of certain types of securities, specifically, interest-only (IO) strips and principal-only (PO) strips, are more vulnerable to changes in interest rates because of their structures. Regardless, every manager in this interest rate environment has some premium exposure in his or her portfolio. The challenge is how to manage that exposure.
The average price paid for the securities in our portfolios is approximately 102. We manage our exposure a few ways, including seeking to acquire MBS as close to par as possible, purchasing fixed rate mortgages which normally experience capital gains in declining interest rate environments as well as acquiring securities that are less affected by prepayments, like collateralized mortgage obligations, or CMO floaters, a type of MBS. CMOs divide a pool of mortgage loans into multiple tranches that allow for shifting of prepayment risks from slower-paying tranches to faster-paying tranches. This is in contrast to pass-through securities where all investors share equally in all prepayments. When purchasing pass-through securities, we also look at the geographical regions represented in the underlying pools, as some regions historically prepay faster than others. For example, according to Bear Stearns, in November 2002 the upper Midwest region was the fastest prepaying region and, of the big mortgage issuing states, Michigan, Illinois and California the fastest prepayers; the slowest prepaying geographical regions were the middle Atlantic and the South, with Pennsylvania and New York the slowest prepaying of the big issuing states. We also look at who originated the underlying mortgages, because those originated by some originators prepay faster then others.
Other mortgage-backed investors have attempted to use other strategies for minimizing the effects of changes in prepayments, including utilizing derivatives, but no strategy can completely insulate an investor from prepayment risks. As we have seen in the market, there are some strategies that expose investors to other, unwanted risks (see our piece on Beacon Hill "When a Safe Harbor Fund Isn’t", October 23, 2002). Also, with prepayment risk comes reinvestment risk. Reinvestment risk exists due to the fact that principal will be paid back quicker in times of falling interest rates and thus result in the holder having to either remain uninvested or invest that principal in lower yielding assets.
Where are we in the refinancing cycle? Graph 3, below, illustrates an important indicator. The graph plots on the left-hand Y-axis the average coupon on outstanding mortgages versus the current mortgage rate. The right-hand Y-axis plots the difference between those two rates. That spread moves in a rough cycle as refinancing brings down the average outstanding mortgage rate, and the spread is now at the wide end of the cycle: At the end of December, the 30-year mortgage rate was 6% and the average coupon on outstanding mortgages was 6.58%, for a "mortgage gap" spread of 58 basis points.

Once the effects of today’s interest rate environment work through the system, we would expect to see a slowdown in refinancing activity. The tight compression of newly- refinanced mortgages around the 6% range may mean that as rates rise, refinancing activity slows precipitously. As Chart 4 shows, as the "mortgage gap" narrows the refinancing index drops.

Investors in MBS strategies must be aware of the effect of prepayments on their portfolios, whether it is in a REIT or a managed account or a fund. More importantly, investors in MBS strategies need to know how their portfolio manager is managing that risk. Looking forward, in the short run we can expect to see relatively high levels of prepayments as long as interest rates stay where they are. The latest prepayment reports show continued vigorous refinancing throughout the holiday period. The plateau in speeds that was reached in the fourth quarter probably reflects a ceiling on the capacity of the industry, with many lenders running at maximum capacity and, in many cases, extending their commitment periods. With a long-tailed pipeline, with many of the recently created MBS pools again refinanceable today, and with mortgage rates still at or near historic lows, we can expect the current refi wave to stay at or near the levels of the last quarter for the next few months. Beyond that, keep an eye on rates.
January 23, 2003
By Jeremy Diamond, Rose-Marie Lyght
Annaly Capital Management/FIDAC
This commentary is neither an offer to sell, nor a solicitation of an offer to buy, any securities of Annaly Capital Management, Inc. ("Annaly"), FIDAC or any other company.
All information contained herein is obtained by Annaly from sources believed by it to be accurate and reliable. However, such information is presented “as is,“ without warranty of any kind, and Annaly, in particular, makes no representation or warranty, express or implied, as to the accuracy, timeliness, or completeness of any such information or with regard to the results to be obtained from its use. While Annaly has attempted to make the information current at the time of its posting on the site, it may well be or become outdated, stale or otherwise subject to a variety of legal qualifications by the time you actually read it.
©2003 by Annaly Capital Management, Inc./FIDAC. All rights reserved. No part of this commentary may be reproduced in any form and/or any medium, without express written permission