The U.S. Residential Mortgage Market: Sizing the Problem and Proposing Solutions

Transcription

1 The U.S. Residential Mortgage Market: Sizing the Problem and Proposing Solutions Laurie S. Goodman Senior Managing Director Amherst Securities Group, LP New York City T The U.S. housing market remains depressed and faces significant challenges to recovery. Demand is being constrained by frictions in the credit process and cannot absorb the oversupply of homes that will continue to flood the market as a result of foreclosures. Bulk sales of foreclosed homes and principal modifications to loans are likely to be necessary to close the gap in supply and demand and revive the housing market. he housing market crisis that has ravaged the U.S. economy for the past few years is likely to deepen unless appropriate measures are put in place to stem the oversupply of housing and stimulate demand; as many as 18 percent of U.S. homes with a mortgage may be in danger of defaulting. To frame the problem, I will begin by providing some background on the size and composition of the U.S. mortgage market, and then I will quantify the magnitude of the crisis in terms of the mismatch between supply and demand. I will discuss how tighter lending standards and proposed housing legislation are further deepening the problem and stalling recovery. In my opinion, the most viable ways to reduce existing and potential oversupply are bulk sales to investors and principal modifications to existing mortgages. Finally, I will outline what I believe it will take to restart the private label securitization market. U.S. Mortgage Market The size and composition of the U.S. mortgage market is shown in Table 1. Of the $16.6 trillion total value, about $6.3 trillion is equity and about $1.4 trillion is mortgage debt. About half of the mortgage market, $5.4 trillion, is composed of mortgage-backed securities (MBS) guaranteed by Fannie Mae, Freddie Mac, and Ginnie Mae. Another $2.9 trillion of mortgages are unsecuritized first liens held at commercial banks, savings institutions, and credit unions and in Fannie Mae and Freddie Mac portfolios. Of the remaining mortgages, $1.2 trillion are in the private label universe and $.9 trillion are second liens. This presentation comes from the Fixed-Income Management 211 conference held in Boston on October 211 in partnership with the Boston Security Analysts Society, Inc. As Table 1 shows, agency MBS dominate the mortgage market. Fannie Mae, Freddie Mac, and Ginnie Mae provide a liquid secondary market for MBS; Fannie Mae and Freddie Mac also purchase MBS for their own investment portfolios. Fannie Mae and Freddie Mac are government-sponsored enterprises (GSEs) that guarantee mortgage principal and interest payments; these insured mortgages Table 1. Size and Composition of the U.S. Mortgage Market as of June 211 Amount ($ billions) Value of U.S. housing market Equity $6,246 Debt mortgages 1,396 Size of U.S. mortgage market 1st Liens, unsecuritized $2,857 2nd Liens 94 Private label 1,185 Agency MBS 5,449 Size of private label universe Subprime $368.9 Option ARM Prime Alternative assets Size of agency MBS market Fannie Mae $2,622 Freddie Mac 1,635 Ginnie Mae 1,192 Notes: MBS are mortgage-backed securities. ARM is adjustable rate mortage. Sources: Based on data from the Federal Reserve as of 2Q211, Fannie Mae, Freddie Mac, Ginnie Mae, and CoreLogic; calculations are by Amherst Securities. 24 MARCH Amherst Securities Group cfapubs.org

2 The U.S. Residential Mortgage Market are sold into the secondary market, freeing up capital that lenders can use for more loan originations. Ginnie Mae is a wholly government-owned entity that guarantees loans issued by the Federal Housing Administration (FHA) and the U.S. Department of Veterans Affairs (VA) that are bundled into MBS. Fannie Mae MBS constitute nearly half of the agency mortgage market, Freddie Mac MBS are about 3 percent, and Ginnie Mae MBS are approximately 22 percent. Although the private label universe is just more than 1 percent of the mortgage universe, 4 percent of all defaults are private label mortgages. Not surprisingly, there has been near zero origination in private label mortgage securities recently. Nearly all recent originations have been in the agency market. Bank portfolio origination is still occurring but at levels far below historical averages because of the tight credit standards. In 21, approximately 63 percent of mortgage production was originated by Fannie Mae and Freddie Mac and another 26 percent was originated by FHA and VA. Bank portfolio loans were about 12 percent of total origination. Size of the Housing Problem I am going to make the case that of the approximately 55 million homes in the United States with a mortgage, more than 1 million of them, which is a huge number of homes, are in danger of defaulting over the next six years. There are three categories of loans: nonperforming loans, reperforming loans, and always performing loans. Nonperforming loans are 6 or more days past due, and reperforming loans used to be 6 or more days delinquent but are now current (or 3 days past due), usually because of loan modifications. Always performing loans can be divided into three categories: 2.6 million have a mark-to-market loan-to-value ratio (LTV) of more than 12 percent, 5.4 million have an LTV of 1 12 percent, and 38.6 million have an LTV of less than 1 percent. Following, I will estimate the eventual default rate on each of these groups of loans. Nonperforming Loans. There are about 4.5 million nonperforming loans, which includes both private label mortgages and nonprivate label mortgages (GSE, FHA, VA, and bank portfolio loans). My estimate for the currently nonperforming loan bucket is based on the fate of nonperforming loans from two years ago: 4.5 percent of those loans have been liquidated, 42.7 percent remain nonperforming, 16.7 percent are reperforming, and.1 percent were voluntarily prepaid. From these data, a reasonable estimate of the eventual default rate and liquidation of these loans is 9 percent, with a lower bound of 8 percent. That translates into 3.6 million 4.1 million housing units. One of the reasons why the stock of nonperforming loans is so large is that those units are simply not being liquidated. The liquidation lag is very long. The average time period from when a borrower makes the last payment to when a property is liquidated is about 24 months, as shown in Panel A of Figure 1. That time period is longer in Florida and other judicial states, where foreclosures are processed through the courts, and shorter in California and other nonjudicial states, where foreclosure procedures are determined by state statute. Panel B in Figure 1 shows the percent of liquidations that were more than 24 months delinquent during the time period. In Florida, more than 7 percent of the liquidated loans were delinquent for more than two years; in California, about 5 percent were. Large numbers of very delinquent loans remain in the nonperforming bucket and will eventually be liquidated. In addition, Panel A of Figure 1 includes only loans that have actually been liquidated; for example, a Florida loan that has been in the pipeline for 3 months and has not yet been liquidated is not included in the figures. Loans remain in the reperforming bucket for long periods of time before they eventually liquidate. Reperforming Loans. In the reperforming loans group, there are 3.8 million loans. Unfortunately, these modified loans have fairly high redefault rates. The annualized monthly transition rate, the rate at which these mortgages are redefaulting, is 47.2 percent per year. About 2.6 percent per year are being voluntarily prepaid. If the ratio of defaults to defaults-plus-prepays remains constant, 94.7 percent of these mortgages will eventually default. The transition rate, however, has slowed for reperforming loans, as shown in Figure 2. In early 29, redefault rates were in the 6 9 percent range; now they are in the 4 5 percent range. Once a loan pays for a while, it is more likely to perform. To estimate the eventual default rate of reperforming loans, I reviewed the fate of reperforming loans 24 months ago: 6.8 percent have been liquidated, 36.8 percent remain nonperforming, 54.3 percent are reperforming, and 2.1 percent are voluntarily prepaid. The success rate of the reperforming bucket two years later is 56.4 percent, which includes loans that have been remodified for a second or third time. That translates into an eventual failure rate of 5 65 percent. cfapubs.org MARCH

3 CFA Institute Conference Proceedings Quarterly Figure 1. Liquidation Lags, December 26 August 211 Months 35 A. Months Delinquent at Liquidation Dec/6 Aug/7 Apr/8 Dec/8 Aug/9 Apr/1 Dec/1 B. Percent of Loans More Than 24 Months Delinquent at Liquidation Percent Dec/6 Aug/7 Apr/8 Dec/8 Aug/9 Apr/1 Florida Judicial (e.g., FL) California Nonjudicial (e.g., CA) All Dec/1 Sources: Based on data from CoreLogic and 11data; calculations are by Amherst Securities. Always Performing Loans. In the always performing loan bucket are the 2.6 million loans with a mark-to-market LTV greater than 12 percent, 5.4 million loans with an LTV of 1 12 percent, and 38.6 million loans with an LTV of less than 1 percent. These loans represent borrowers who have never been two payments behind. Figure 3 shows that equity is the single most important determinant in whether a loan will default or not. Loans that are prepaid do not default, and loans that default are not prepaid. Figure 3 also shows that prime loans with LTVs greater than 12 percent 26 MARCH 212 cfapubs.org

4 The U.S. Residential Mortgage Market Figure 2. 1 Redefault Rates of Various Types of 24 7 Vintage Reperforming Loans, February 29 August 211 A. Prime Feb/9 Aug/9 Feb/1 Aug/1 Feb/11 B. Alt-A Feb/9 Aug/9 Feb/1 Aug/1 Feb/11 27 Vintage 26 Vintage 25 Vintage 24 Vintage (continued) have almost no volatility in prepayments; they are fairly stable at a low level. The question is, What is the likelihood that a borrower that has never been two payments behind actually ends up defaulting? The likelihood is actually reasonably high for high- LTV borrowers. The three-month annualized monthly new default transition rate for always performing loans measures the likelihood that a mortgage will fall two payments behind for the first time. For always performing loans with a mark-to-market LTV greater than 12 percent, 13.8 percent of the borrowers fell two payments behind for the first time over the last three months on an annualized basis. If default transition rates and prepayment rates continue at current levels, 74 percent of those borrowers will eventually default. More reasonably, default transition rates will burn out over time; hence, my reasonable estimate of default for this bucket is 4 percent, and my lower bound estimate is 25 percent. For always performing loans with a mark-tomarket LTV of 1 12 percent over the past three cfapubs.org MARCH

5 CFA Institute Conference Proceedings Quarterly Figure 2. 1 Redefault Rates of Various Types of 24 7 Vintage Reperforming Loans, February 29 August 211 (continued) C. Subprime Feb/9 Aug/9 Feb/1 Aug/1 Feb/11 1 D. Option ARM Feb/9 Aug/9 Feb/1 Aug/1 Feb/11 27 Vintage 26 Vintage 25 Vintage 24 Vintage Note: The redefault rate is annualized. Sources: Based on data from CoreLogic and 11data; calculations are by Amherst Securities. months on an annualized basis, borrowers have exhibited about a 7 percent chance of falling two payments behind for the first time (transitioning to default) and a 7 percent chance of voluntarily prepaying. I use a lower bound estimate of 1 percent and a reasonable estimate of 15 percent as the eventual default rate for this borrower group. A loan that does not default and does not prepay in Year 1, however, has another chance to default or prepay in Years 2, 3, or 4. Of the 38.6 million units of always performing loans with a mark-to-market LTV of less than 1 percent, borrowers are becoming 6 days delinquent for the first time at a rate of 2.4 percent per year. They are voluntarily prepaying at 13 percent per year. If the ratio of defaults to defaults-plus-prepays 28 MARCH 212 cfapubs.org

6 The U.S. Residential Mortgage Market Figure 3. 4 Prepayment and Default Transition Rates for Prime Fixed-Rate Loans, December 28 August 211 A. Prepayment Rate Dec/8 Aug/9 Apr/1 Dec/1 B. Default Transition Rate Dec/8 Aug/9 Apr/1 Dec/1 <=1 LTV LTV >12 LTV Note: The prepayment rate and default transition rate are annualized. Sources: Based on data from CoreLogic and 11data; calculations are by Amherst Securities. remains constant, 15 percent of these borrowers will eventually default. I assume 5 percent as a reasonable estimate and 4 percent as a lower bound estimate. The reason that borrowers who seem to have equity have a 5 percent default rate is that not all the borrowers in this bucket actually have equity. These loans are being classified by their LTV, not their combined loan-to-value ratio. A borrower may be in the less-than-1-percent-ltv bucket because he or she has 95 percent LTV, but with a second lien, the combined LTV is actually 12 percent. When the numbers for all five groups of loans are added together, the reasonable estimate is that about 1.4 million borrowers are in danger of default and 8.3 million are in the lower bound estimate. These figures assume no change in overall housing prices, interest rates, or any other factors. Clearly, this problem is enormous. cfapubs.org MARCH

7 CFA Institute Conference Proceedings Quarterly Mismatch in Supply and Demand As I mentioned earlier, nonperforming loans are not declining and are creating quite an overhang. The growth in shadow inventory is depicted in Figure 4. In this figure, shadow inventory is narrowly defined as loans that are more than 12 months delinquent and loans that are in foreclosure. There are about 2.8 million of these loans and about 4, real estate owned (REO) loans (i.e., loans that have already been foreclosed and are now owned by the lender), totaling 3.2 million units in overhang. Approximately 1, of these loans are being sold per month, so it will take 32 months just to move the existing supply of distressed homes that are more than 12 months delinquent. The mismatch between supply and demand is huge, especially with about 1.4 million homes, or a lower bound estimate of 8.3 million, at risk of default over the next six years. If the reasonable estimate of 1.4 million units and the lower bound of 8.3 million units is divided by six (to obtain an annual number) and a half-million units of new construction per year are added, the total annual supply is between 1.88 million and 2.23 million units. If nothing changes, the question is, what can demand absorb? Demographics can account for demand of.6 million units. Housing formation during the past few years has been about.5 million units. The Joint Center for Housing Studies at Harvard University estimates 1.2 million units in new household formation over the next decade. I assume a 5 percent homeownership rate for this group, which reflects, in part, the fact that 7 percent of the 1.2 million units of annual housing formation will be from the minority population, which has lower rates of homeownership. The homeownership rate for the total population drops from 66 percent to 62 percent after factoring in borrowers who have been living in their home and have not paid their mortgage in more than a year. A 5 percent homeownership rate for new household formation leads to 6, units of demand. To this count must be added 4, units from obsolescence and 2, units from second home purchases. Subtracting the 1.2 million units in annual demand from the 1.88 million 2.33 million units in supply leaves a housing mismatch of.7 million 1 million units annually or 4.1 million 6.2 million units over the next six years. Frictions in the Credit Process Home prices are down more than 3 percent from their peak, and primary mortgage rates are at the lowest level since the 195s. The ability of the median family to afford the median price home is better than it has been in a generation. The reason people are not buying homes is that they cannot qualify for a mortgage. The MBA purchase index is at its lowest level in 15 years. Existing home sales are low, and new home sales are at their lowest level since the 196s, when the index was first tabulated. Figure 4. Loan Count 3,5, Number of Loans in the Shadow Inventory, 1Q29 2Q211 3,, 2,5, 2,, 1,5, 1,, 5, 1Q9 2Q9 3Q9 4Q9 1Q1 2Q1 3Q1 4Q1 >12-Month Delinquent and Foreclosure Real Estate Owned 1Q11 Loans Sold 2Q11 Sources: Based on data from CoreLogic Prime Servicing Database, CoreLogic Securitized Loan Database, 11data, the Federal Deposit Insurance Corporation (FDIC), Fannie Mae, Freddie Mac, and FHA; calculations are by Amherst Securities. 3 MARCH 212 cfapubs.org

8 The U.S. Residential Mortgage Market Tighter Lending Standards. The housing market is struggling with the oversupply of housing, limited demand, and restrictive credit conditions. Of the borrowers who had a mortgage in 27, 19 percent would no longer qualify today for a mortgage because they were 9 days delinquent on their previous loan. Additionally, the FICO (Fair Isaac Corporation) credit scores required to qualify for a mortgage have become increasingly higher. Fannie Mae and Freddie Mac are both requiring higher FICO scores to originate loans, and bank portfolio standards are very similar to GSE standards. From 29 to 21, the average GSE loan had an original LTV of 67 and an original FICO score of 762; the average bank portfolio loan had an original LTV of 66 and an original FICO score of 756. With few borrowers able to afford a minimum down payment of 2 percent, people are only able to buy homes because Ginnie Mae has become the major outlet for purchasing a home and the FHA has played a critical role. Ginnie Mae mortgages are primarily backed by FHA loans, although VAbacked loans comprise an important share of the mortgages. Ginnie Mae comprises about 3 percent of total agency issuance. Ginnie Mae, however, comprises 55 6 percent of the purchase volume and approximately 2 percent of the refinancing volume. About 6 7 percent of the mortgages issued by Ginnie Mae were for purchase activity, with the balance being from refinance activity. For Fannie Mae and Freddie Mac, purchase activity is 2 3 percent of their recent activity. At this point, FHA and VA are primarily home-purchase vehicles; Fannie Mae and Freddie Mac are primarily refinancing vehicles. Proposed Housing Legislation. In addition to higher required FICO scores and LTVs, there are other frictions in the credit process, including less flexibility on down payment requirements. Moreover, every housing market reform that is being contemplated would tighten credit standards even more to ensure that only good loans are securitized. The GSEs are considering raising their loan level pricing adjustments and/or their guarantee fees. The U.S. Department of Housing and Urban Development (HUD) is considering tightening the debtto-income requirements on FHA loans. In addition, risk retention guidelines are being drawn up, as required under the Dodd Frank legislation. Basically, institutions would be required to retain 5 percent of the risk on all loans that are not considered qualified residential mortgage (QRM) loans. A QRM loan meets certain lending standards and is expected to be high quality. QRM and GSE loans would be exempt from the proposed risk retention guidelines. The QRM requirements, as originally proposed, for debt-toincome ratios, LTVs, product type, and credit quality are quite strict. Of the GSE loans originated in 29, only 3 percent of total production would meet QRM requirements 3 percent in a year in which the GSEs originated loans with low LTVs and good FICO scores. If the debt-to-income standards on QRM loans had been relaxed, an additional 25 percent of 29 borrowers would have qualified. If the LTV standards were relaxed, an additional 15 percent would qualify. The concern is that if QRM requirements are implemented as drafted, a bank may reserve the right to make only QRM loans because these are mortgages in which the borrower has, by definition, shown the ability to pay. Thus, banks would be immune from being sued for making unqualified mortgages, which would tighten credit availability even more. Solutions The mismatch in supply and demand and the restrictive credit conditions have created a death spiral in housing prices. Falling home prices lead to homeowner defaults. Homeowner defaults increase the supply of distressed homes, which increases the shadow inventory. Therefore, more borrowers are underwater, which, in turn, increases the shadow inventory and leads to further drops in home prices, more homeowners underwater, and more defaults. Of course, lenders only want good mortgages, and thus credit standards keep getting tighter. On the other side of the equation is rental demand. Rental demand is rising rapidly. Foreclosures are causing more evictions, more evictions are creating more renters, and more renters are leading to higher rental prices and lower affordability of homes. To close the gap in supply and demand of 4 million 6 million housing units over the next six years, two things must happen: The demand for housing must be increased by encouraging investor participation, and the supply of housing must be reduced by having more sustainable loan modifications, which means principal reductions. Bulk Sales. Investor participation can help reduce the overhang of supply through the purchase of large blocks of distressed properties that are rented back to homeowners. Rental income is up substantially 4 percent on a year-over-year basis. The multifamily vacancy rate has declined considerably from about 8.3 percent in 29 to just less than 7 percent. The increased demand for rental housing cfapubs.org MARCH

9 CFA Institute Conference Proceedings Quarterly is directly correlated with the homeownership rate. The homeownership rate has fallen to about 66 percent from 69 percent, and many of these former homeowners have become renters. From an investor s point of view, rental yields are more attractive relative to the purchase price of the investment (the home) than they have been in a long time. Rental yields compare favorably with other alternatives in an environment where the competing uses of funds are very limited that is, interest rates are extremely low across the board. If distressed properties were sold in bulk sales, there would be a mechanism by which the homeowner could be converted to a renter and the distressed home converted to a rental. Shadow inventory would decrease, and homeownership demand would be unaffected. This scenario would avert the deepening of the death spiral in housing prices. On 15 August 211, the U.S. Treasury, FHA, and the Federal Housing Finance Agency issued a joint request that solicited proposals, due back by 15 September, on how to structure bulk sales to investors. They expected very few proposals because of the tight deadline, but they received a surprising 4, responses that they are currently evaluating. In my opinion, bulk sales should be at the metropolitan statistical area (MSA) level. All of the REO and nonperforming loans from FHA mortgages and from Fannie Mae and Freddie Mac mortgages in a given geographic area, such as Indianapolis, should be bundled into large groups of 1 2 loans, and each group of properties should be sold on an allor-nothing basis. Selling properties individually has two problems. First, it would discourage large-scale investors from developing national rental organizations. If an investor, for example, accumulates 3 properties in Atlanta, 12 in Indianapolis, and 6 in Cleveland, the investor cannot hire property managers or rental agents on an efficient scale. And with only a few properties in hand, investors would be reluctant to incur the costs to build out an organization in a given geographic area, in the hope that they could accumulate more properties in that area. Second, selling properties one by one would be a very slow process. There are currently 32 months of existing overhang. Investors must be confident when they buy properties that they are within sight of the bottom. Investors will feel certain the bottom is near only when there is a bulk sale that actually clears a fair amount of the overhang. Even without this consideration, bulk sales may be the best execution because every day that a property sits vacant, it is not being maintained and is deteriorating in terms of price. Moreover, the lender is incurring the tax and insurance costs on the property. I believe investor participation is essential to the housing market s recovery, and bulk sales are the best way to accomplish this goal. Fannie Mae, Freddie Mac, and the FHA would achieve a higher price on these bulk sales if conservative financing was included. Loan Modifications. The second thing needed to revive the housing market is a successful loan modification plan. Modifications have evolved quite a bit since 28, when most of the modifications were capitalization modifications. Capitalization modifications were not very meaningful because they did not reduce a borrower s interest rate or principal; they simply extended the term of the loan, the delinquent amount was added to the principal balance, and the borrower was pronounced current. Since 28, interest rate modifications, in which the interest rate of the loan is reduced, have become the most common type of modification. Recently, the number of principal modifications has grown but still remains in the minority. One measure of the default rate on modified loans is the cumulative redefault rate per unpaid balance at modification. This measure shows that modification success has been improving over time for two reasons. The first reason is that payment relief is much more significant. There are a lot more interest rate reductions and principal reductions and fewer capitalization modifications. I believe the first reason modifications are more sustainable or of better quality is that the Home Affordable Modification Program (HAMP) provided a blueprint for substantial pay relief. Second, borrowers are now being offered a three-month trial period. If the borrower defaults during the trial period, the modification is not counted as a redefault. The HAMP has not served the number of borrowers it had hoped to because it has been somewhat unsuccessful in converting trial modifications to permanent modifications. Under the HAMP, a borrower s front-end debt-to-income ratio (DTI) goes from 45 percent before the loan modification to 31 percent after a decline of 31 percent. This modification generates an average saving of nearly $6 a month on an average payment of $1,4 per month. The problem is the program only considers front-end DTI, defined as taxes plus insurance plus the first mortgage payment relative to income. The program does not look at the total debt burden, or back-end DTI. Median back-end DTI declined from 79 percent before the modification to 62 percent afterward, still an unsustainable level. By contrast, the FHA requires a back-end DTI of no more than 43 percent under its standard loan programs. 32 MARCH 212 cfapubs.org

10 The U.S. Residential Mortgage Market The HAMP has also modestly increased negative equity. Median LTVs under the HAMP increased from 12 to 125 percent. That increase is just loan to value, not combined loan to value. A disproportionate number of these borrowers have second mortgages. The result is that successful permanent modifications to date are well under half of the total HAMP modifications started. This result is unfortunate because what it takes to implement a successful modification program is known. Figure 5 shows the cumulative redefault rate by the type of loan modification, measured by unpaid balance at modification. Capitalization-modified loans performed the worst, interest rate modified loans are in the middle, and principal-modified loans performed the best. A natural question concerns the moral hazard issue relating to a principal reduction: Will principal reduction modifications lead to higher defaults? It is important to realize that borrowers respond to financial incentives under the present interest rate reduction program. In February 29, an upcoming modification program was announced; subsequently, owner-occupied default rates increased far more rapidly than non-owneroccupied defaults because most defaults contain a strategic element. So, a moral hazard already exists under the present program. Principal Modification. Banks know that principal modification is powerful, as evidenced by the amount of principal reduction they are doing for their own portfolio loans versus loans they service in which the risk is owned by other entities. Of the loans in bank portfolios, 25 percent received a principal reduction in the first quarter of 211; among the loans insured by government agencies or in private label securitizations, very few loans received principal reductions. The six-month redefault rate for loans in bank portfolios is significantly lower than the redefault rates for loans serviced by these institutions but owned by other types of investors. Solutions are still being sought for a principal reduction program that at least partially mitigates the moral hazard issue. One major servicer has found a solution through a shared appreciation mortgage. Consider the following scenario: It is highly likely that a delinquent borrower with a 15 percent LTV will default. He or she is offered an opportunity for a write-down of 35 percent of the value of the property, to 115 percent LTV, in return for sharing a percentage for example, 5 percent of any future price appreciation with the lender. Realize that the principal does not have to be written all the way down to 1 percent LTV; a write down to 11 or 115 percent will be sufficient to reduce Figure 5. Cumulative Redefault Rate by Type of Loan Modification, as of June Months after Modification Capitalization Modification Principal Modification Interest Rate Modification Note: Calculation is based on unpaid balance at modification. Sources: Based on data from CoreLogic and 11data; calculations are by Amherst Securities. cfapubs.org MARCH

11 CFA Institute Conference Proceedings Quarterly defaults substantially. I believe solutions that involve shared appreciation features can go a long way toward eliminating the moral hazard issues. In addition to the payment relief and the type of modification, the timing of the modification is also very important. If borrowers are modified when they are less than 2 months delinquent, the modification success rate, as measured by the 12- month redefault rate, increases dramatically. If loans are not modified until borrowers are 12 months delinquent, the modification success rate is much lower. In this case, the borrower has been making zero payments for the past year, so any type of payment is a much greater financial burden. Finally, the borrower needs substantial pay relief to lower the redefault rate. Table 2 summarizes the 12-month redefault rate of the three types of modifications under different payment relief amounts and periods of time delinquent. For example, the second row shows a loan modification scenario in which the borrower is offered a principal modification and a 2 4 percent payment reduction when he or she is less than two months delinquent. According to the 21 failure rate, the 12-month redefault rate on those borrowers is only 14 percent. If modification is not offered until after the borrower is more than 12 months delinquent, the default rate after one year is 5 percent. This difference in success rates is enormous. Restarting the Private Label Securitization Market Finally, I would like to talk about the five things that I believe are necessary to restart the private label securitization market. First, the regulatory rules of engagement need to be established. Risk retention and the specifications for QRM loans must be defined so that institutions know what is expected of them in terms of risk retention. Second and most important, the securitization has to be economic. Right now, credit standards are very narrow. Banks are bidding aggressively for portfolio loans at 7 percent LTV. If that is what is being originated, the bank portfolio bid is going to be better than the securitization bid because banks essentially have a near-zero cost of funds. In addition, the rating agencies need to be explicit regarding their rating criteria. In order to regain some credibility, they need to be transparent in terms of exactly how the rating process works so that when a pool of mortgages is securitized, the issuer knows exactly what percentage will be rated AAA and what percentage will be allocated to the subordinate tranches. This transparency allows the issuer to have more certainty about the economics of the transaction. At the present time, the single largest reason securitizations are not occurring is because they are not economic. Table 2. Twelve-Month Redefault Rates for Various Loan Modification Scenarios, Payment Relief (%) <=2 Mo. 3 6 Mo Mo. >12 Mo. <=2 Mo. 3 6 Mo Mo. >12 Mo. <=2 Mo. 3 6 Mo Mo. >12 Mo. Principal modification >4 39% 61% 72% 73% 2% 32% 4% 54% 8% 15% 24% 39% <= Interest rate modification > < = Capitalization modification > <= Note: is delinquent. Sources: Based on data from CoreLogic and 11data; calculations are by Amherst Securities. 34 MARCH 212 cfapubs.org

12 The U.S. Residential Mortgage Market A third requirement is that governance standards for securitizations need to be set and conflicts of interest addressed. Investors must be able to access the loan files, and it must be very clear how these files can be accessed. In addition, credit standards must be loosened because there is a lack of demand with the current credit standards. Lastly, origination expertise has to be developed for the private label securities market. Underwriting criteria need to be redeveloped, and title perfection, credit approval, and documentation processes all must be rebuilt. It is going to be a long time before the private label securitization market is able to be restarted. Conclusion Solving the housing crisis will be a process. First, investor participation is needed to absorb the overhang of shadow inventory. This step is best done through a bulk sales program of REO and nonperforming loans. Bulk sales would avert a major death spiral in home prices by allowing investors to buy a sufficient number of homes to build out a property management organization. It is clear that many borrowers must transition to become renters. In many cases, investors who buy nonperforming loans will find that keeping the delinquent homeowners in their homes as renters will be the best solution for both investor and borrower. Financing for bulk sale investors would also help because leverage would enable them to pay more for the properties and yield a higher return. Second, successful loan modifications are necessary principal write-downs in particular. Redefault rates on modified loans that use principal write-downs is much lower than the redefault rates for loans that have received capitalization or interest rate modifications. If the industry can overcome the moral hazard associated with principal write-downs, these modifications can significantly stem the flow of shadow inventory. Without decisive action to curtail further declines in home prices, as many as one in five borrowers are in danger of losing their homes. This article qualifies for.5 CE credits. cfapubs.org MARCH

13 CFA Institute Conference Proceedings Quarterly Question and Answer Session Laurie S. Goodman Question: There are political aspects associated with increasing housing demand and decreasing supply through bulk sales or changes in loan modifications, but which scenario do you think is most likely considering the political aspects? Goodman: I think there will be at least some bulk sales in the next year, although I cannot predict the scale they will be on. With regard to principal reductions, I think this solution is too politically charged. There will be some backlash about reducing the principal for so-called deadbeat borrowers. The solution to principal modification is shared appreciation, but I do not think it can be implemented on a national level. At some point, there will be a begrudging realization that principal reductions are the most effective form of modifications. They may be used slightly more in private label securitizations. When delinquent loans are transferred to special servicers, these servicers are more apt to write down principal. Another major issue is that it is difficult to determine how to implement principal reductions on loans held by Fannie Mae and Freddie Mac because they are in conservatorship. Their mission is to conserve assets, and if they reduce the principal on a loan, they will absorb the cost of it, whereas if the borrower defaults, the mortgage insurer would share that burden with them. If the mortgage insurer did not share that burden, the GSEs could often send back the loan to the originator. At the moment, no viable solutions exist for sharing the cost of principal reduction among Fannie Mae or Freddie Mac and the originators and mortgage insurers. Question: Because Fannie Mae and Freddie Mac are in conservatorship, would it be possible to transform them into the largest rental agencies in the country instead of implementing a bulk sales program? Goodman: That is a possibility, although they have no real experience in this area. A private investor group that deals exclusively in that area would be more motivated to do a good job. It is important to realize that a large amount of money is being raised for the absorption of bulk sales; hence, it should be expected that aggressive bidding between private investor groups would occur. This scenario should allow the GSEs and FHA to achieve good execution on the bulk sales. Question: Do you believe that there is any prospect for sustainable economic recovery without tackling these housing issues first? Goodman: No. The housing market will continue to be a drag on the economy until the housing issues are resolved. Question: What is your estimate for home prices over the next 12 months? Goodman: I believe they will decline about 5 7 percent, which is in line with the consensus estimate. I think there will be bulk sales and investor bids for these sales that will lend some support to the housing market. Question: How many of the 8 million 1 million housing units that you estimate borrowers will eventually default on need to be cleared to reduce shadow inventory to a manageable level? Goodman: Because there are a limited number of new homes being constructed and there is going to be some demand based on demographic need, eliminating 4 million of the 8 million 1 million potential foreclosures would likely solve the oversupply problem. The rest of the supply could be worked out over a number of years because new home construction is so low. Question: Do you have any thoughts on whether there will be regional disparities as the housing market recovers? Goodman: There will be huge regional disparities, primarily because of judicial versus nonjudicial resolution issues. California, which is a nonjudicial state, has gone a long way toward clearing the oversupply of homes because homes can actually move through the foreclosure process. Florida, a judicial state, has made much less progress in clearing the supply because of the court process. Therefore, the overhang in Florida is much worse, and the potential for home price declines is much greater. 36 MARCH 212 cfapubs.org