Packaging CRA loans into securities.
The recent news about CRA (Community Reinvestment Act) mortgage-backed securitizations has created a great deal of interest among the holders of these loans. Is this new MBS market niche a short-term phenomenon or will it open a major new capital market for CRA loans?
The answer may lie in how well the initial transactions were received. Both the First Union ($416 million) securitization in November 1997 and the Mellon deal ($328 million) in April 1998 were several times oversubscribed. What is even more interesting is who bought the loans. Was it banks seeking CRA investment credit? No, it wasn't. The overwhelming participants were money managers and insurance companies buying the loans strictly because of their investment appeal.
Because of the publicity generated by the recent Mellon and First Union transactions, a number of banks holding sizable CRA portfolios have inquired about the feasibility of securitizing their loans. This article will detail the specific characteristics of a CRA securitization as well as, hopefully, answer some questions to determine if a transaction is right for your institution. The authors worked on the most recent CRA transactions and will share some of their insights.
There are three important elements to consider in the securitization process:
* The pros and cons of holding the loans vs. selling them.
* The credit quality of your portfolio and how it will be valued by either the rating agencies or the government-sponsored enterprises (GSEs).
* The prepayment behavior of your portfolio.
Some basic elements of CRA loans
Under the Community Reinvestment Act guidelines, a bank gets credit for originating loans or buying on a whole loan basis; but you get no credit for holding the loans. Conversely, if you sell your loans through a securitization, a buyer can get CRA investment credit if a percentage of the loans is based in the buying bank's trade area.
Talking with a number of banks, we are surprised at the size of many portfolios and, frankly, the circumstances under which banks hold on to the loans. The prevailing misconception we deal with in almost every instance is banks believing they must hold their CRA portfolio, or that they will incur a large loss if they try to sell them.
The perception is that the best a bank can do is 90 to 93 cents on the dollar. When dealing with a $300 million portfolio, that could result in close to a $25 million writedown on a whole loan sale. Most banks or their mortgage companies are not willing to take that one-time hit to their balance sheet. They are more likely to hold the loans, accept the inherent risk, charge off any foreclosures as they occur, and hope for close to a break-even proposition. Occasionally the CRA loans are lumped together with a group of conventional or conforming loans and sold on a whole-loan basis to the GSEs. More often, the banks simply keep them in portfolio.
What has changed in the marketplace? Foremost is the current interest rate environment. With 30-year mortgage rates close to a decade low, most of the CRA portfolios automatically have an above-market weighted average coupon (WAC) rate. Many of the portfolios we see are in the 7.5 percent to 7.75 percent WAC range.
A second misconception regarding CRA loan sales is the lack of mortgage insurance (MI). Many banks had the notion beat into them that the loans are unsellable without mortgage insurance or pooled insurance or some level of recourse. Nothing could be further from the truth in the securities market. While the presence of MI does help alleviate some risk concerns and improve subordination on the marginal loans, it is not a requirement on the portfolio as a whole. Indeed, the entire portfolio in many cases can be sold without recourse. The necessary levels of credit enhancement, as needed, can be handled internally by sub-ordinating junior tranches to the seniors.
Understanding credit quality
Whether you sell a portfolio on a whole-loan basis, have it subordinated by one of the rating agencies or have it wrapped by one of the GSEs, perceived risk on the loans is going to affect the pricing. Three factors come into play when measuring risk on any portfolio: payment history, credit history and loan-to-value (LTV) ratio. What is different with CRA loans is the role each element plays separately and their linkage as a whole.
The quick and easy way for rating agencies to review any mortgage-backed securities portfolio is to see how many loans are current (or have few incidences of lateness), have a credit score higher than a 660 FICO and have an LTV of 80 percent or less. From a sample database of more than 50,000 CRA mortgage loans, here are some insights into each of these elements.
* Loan to value. Because CRA loans by design tend to be made to people with limited financial means, most of the borrowers have less than 20 percent equity in their homes. There is a disproportionately high number of loans with 95-plus percent LTVs. In fact, the average CRA portfolio is made up of at least 30 percent high-LTV loans. This is a critical area of understanding that we continue to address with the rating agencies.
Traditionally rating agencies view LTV as the single most important determinant of default. It is most important at the time of origination and less so after the third year. While we do not dispute these assumptions, LTVs have to be analyzed within the context of the affordable-loan situation. Three or 4 percent equity on a $50,000 house is significant to a family of limited financial resources. In relative terms, $1,500 to $2,000 could easily mean three to four months of advance rent payments in their previous housing situation. However, we have learned that going below 2 percent equity can be problematic for the portfolio.
While we cannot yet draw any definitive conclusions on the correlation of default to very low LTVs, we do know from speaking to many lenders and a number of housing advocates that LTVs higher than 98 percent have a significantly higher incidence of default and foreclosure. To graphically illustrate the LTV story, we've drawn a random sample of 20,216 CRA loans. Three-quarters or 75-3 percent of the sample have been seasoned for more than 24 months.
In our sample shown in Figure 1, the column to the left denotes LTV levels. The first numbers alongside the bar graphs depict the percentage of that LTV segment that has been 90 days delinquent within the past 22 months. The number to the right of the percent is the total number of loans in the segment. For instance, in the [less than]60 percent LTV segment, 1.8 percent of the 1,925 loans have had at least one delinquency during the past 12 months.
Obviously, there are more delinquencies with the higher LTV loans than the lower, but there is no tight linear correlation between the LTV levels. Delinquency rates increase along with the LTV levels, but not proportionately. As a result, the use of default models traditionally used for conforming loans have to be adjusted for CRA affordable loans. The amount of the down payment in the LTV equation is relative to the financial capability of the borrower.
* Credit scores. While credit scores can be an analytical tool with conforming loans, their effectiveness is limited with CRA loans. Unfortunately, CRA loans do not fit neatly into the standard credit score framework. We believe a broader array of credit analysis data is needed to get a clearer perspective of the situation.
In every CRA portfolio, a high percentage of the loans will have less than a 660 FICO score. Depending on the portfolio, the percentage could vary from 20 to 25 percent. And a disproportionately large percentage can be below a 620 FICO score. Do we automatically exclude or severely discount these loans? Absolutely not.
A low credit score (i.e., less than 620) can result from two situations: either negative incidents reflected on the credit file or a thinner credit file. When explaining the credit quality of a portfolio to a rating agency or GSE, it is essential to go beyond credit scores. Every individual credit file can potentially contain more than 330 credit attributes. Information on all trade lines, bank, retail and gas card utilization are individual components that need to be reviewed. Yes, all of these elements figure into the credit-scoring process. But when a FICO score is less than 660, and particularly below 620, it is essential to do some analysis and risk modeling on the available attributes.
* Payment history. While some credit-score purists might take issue with our comments in the preceding section, payment history for CRA loans tracks consistently close to the risk curves of conforming loans. The greatest risk of default occurs in the earlier years and declines after the fourth or fifth years - but the reasons for default can vary. Most delinquencies and defaults occur primarily because of financial reasons. The same reasoning applies to CRA loans, but under different circumstances.
In a conventional or conforming loan situation, a delinquency is caused because of a substantial loss of income or mismanagement of personal finances. With CRA borrowers, ownership of credit cards, for instance, occurs at substantially lower rates than the norm. Where we see the greatest adverse impact on payment stability is the presence of personal finance trades (i.e., personal credit lines or installment loans). While loss of income can occur due to job changes, there is a less frequent occurrence with CRA loan defaults.
In many cases, purchasing a home puts the borrower in a more favorable financial position than renting. It is quite common for a first-time homebuyer using a CRA loan to have been shouldering a rent payment that consumed 40 percent to 50 percent of his or her gross income. The purchase of a home including tax and insurance can move those ratios into the mid-30s.
Where do most payment problems occur? Usually, the problems stem from poor upfront planning and counseling. Hence, one of the key factors we look for in a CRA portfolio is whether the borrower completed a GSE-accredited homebuyer education program. The best of these programs help the individual plan for emergencies that can arise with homeownership.
Payment history is the single most important determinant of how well a CRA portfolio will perform. For instance, in looking at a segment of 4,858 loans within a portfolio of 6,240 loans, the common thread is that all the loans were current on their mortgage payments and had zero delinquencies for the past 12 months. However, the credit scores were fairly evenly dispersed along the four standard quartile levels (620 or less, 620-659, 660-719, 720 or more). Loans with 95 percent or greater LTVs comprised more than 30 percent of the portfolio.
However, looking beyond the credit scores into other individual credit trade lines, we found that all bank card and trade lines were used only at 40 percent of their high-balance limit and an astounding 89 percent of the segment, regardless of credit score, had never been delinquent on any trade lines.
Although 27 percent of the portfolio had FICO scores less than 660, would we automatically discount or eliminate them? Absolutely not. These borrowers generally pay their mortgages on a timely basis and do an excellent job of managing their entire financial situation. Nevertheless, there are those individuals who are habitually 30 days late on their mortgage payment. Fortunately we see a self-cure rate in about 70 percent of these cases. Some borrowers are just not credit savvy and are always late. The challenge in the securitization process is to identify the loans that will cure vs. those that could deteriorate further.
When considering the credit score, LTV and payment history, we put the greatest weight by far on the last variable. LTV is relative to the financial assets and capabilities * of the individual. Credit scores can be helpful and indicative when the FICO score is higher than 660; below that score one has to look at the underlying credit variables. Payment history speaks for itself. To many lower-income homeowners and CRA borrowers, being able to own a home is a near-sacred obligation. A family will do almost anything to meet that monthly mortgage payment. The challenge is to explain the appropriate levels of risk combining the three critical attributes during the securitization process.
CRA MBS' appeal to investors
CRA-backed securities are attractive to mortgage investors because of their very stable prepayment behavior. Because pre-payments [TABULAR DATA FOR FIGURE 2 OMITTED] are unlikely to accelerate if interest rates decline, these securities consistently outperform their traditional mortgage-backed counterparts on a total-rate-of-return basis.
Why do CRA securities exhibit this behavior? The answer is straightforward. Because CRA loan programs are targeted to meet the special needs of low- and moderate-income borrowers, loan attributes and prepayment behavior reflect the unique characteristics of this demographic segment of the community. Figure 2 compares the typical underwriting criteria and loan attributes of loans from CRA programs to those for agency conforming loans.
It is clear from Figure 2 that CRA loans differ from agency conforming loans in three important respects:
* CRA borrower demographics reflect the low- to moderate-income segment.
* CRA borrowers receive favorable underwriting standards.
* CRA loan attributes reflect the low- to moderate-income segment (i.e. small loan sizes, high LTVs, first-time purchases).
Based solely on these fundamental differences we would expect CRA prepayments to be slower and less interest rate sensitive than equivalent agency conforming loans.
CRA borrowers are less sensitive to refinancing opportunities
Figure 3 compares the average prepayment rate of CRA and agency loans conditional on the economic incentive to refinance (data from April 1996-July 1997). Clearly, CRA borrowers were less sensitive than agency borrowers given equivalent refinancing opportunities during the time period we observed. In fact, no significant premium coupon/origination year cohort of CRA loans exhibited a single month prepayment rate higher than 20 CPR during the observation period despite sometimes large economic incentives to refinance.
The structural and behavioral characteristics of CRA loans/borrowers that account for this finding are highlighted below:
* Favorable loan terms are so attractive they essentially "lock" borrowers into their loans and limit refinancing alternatives. An important feature of CRA origination programs is that they provide very favorable financing terms for low-and moderate-income borrowers. In most of the portfolios we analyze, those loans with LTVs greater than 80 percent do not have private mortgage insurance (PMI). This occurs for two reasons: the bank is already holding the loan and bearing most of the risk, and the lender does not want to add to the total monthly payment of the borrower. Unfortunately, the absence of PMI coupled with an LTV greater than 80 percent is a combination that is not acceptable under normal Freddie Mac/Fannie Mae underwriting guidelines. Although there is no structural impediment to refinancing (such as a prepayment penalty), CRA borrowers do not have access to the liquid refinancing market available to most conforming borrowers. In addition, because of their unique financing terms and low average balances, they are less likely to be solicited by lenders to refinance.
* Small balances and high LTVs reduce the borrower's economic gain in a refinancing transaction. The impact of loan size on prepayments is well recognized in the MBS markets. Our empirical studies have shown that agency pools with small average loan sizes tend to be much less prepayment sensitive than equivalent pools with large loan sizes. The reason is that the economics of refinancing a small balance are less attractive when rates fall, assuming all borrowers face fixed refinancing transaction costs.
* Low- and moderate-income borrowers are more payment sensitive than interest rate sensitive. The low-income borrower population is much more likely to have limited access to funds and/or have limited desire or ability to pay the out-of-pocket expenses associated with a refinancing transaction. CRA lenders indicated that because "affordable" housing programs were targeted toward people buying homes, borrowers looking to refinance faced similar refinancing transaction costs as their conforming counterparts (CRA closing costs run in the 2 percent to 3 percent range). We suspect that a similar but less concentrated demographic effect in Fannie Mae/Freddie Mac low loan balance pools contributes to the stable prepayment behavior in that sector (loan size and income are highly correlated).
The advantage of CRA securities is that investors know with complete certainty that the loans are originated to low-and moderate-income borrowers either on an income basis or a census tract basis. Investors also know the exact distribution of loan sizes in the pool and the exact terms under which the loans were originated. This should give investors added comfort with respect to future prepayment behavior.
Finally, we found that the baseline seasoning ramp (the average CRA prepayment for near-current coupon loans conditional on the age of the loan) for CRA loans was 24. percent below that of agency conforming collateral [ILLUSTRATION FOR FIGURE 4 OMITTED]. We believe two underlying factors are responsible for this behavior:
* The mobility rates of low-income borrowers are substantially lower than those of high-income borrowers.
* Borrowers with less equity in their properties are less likely to move or trade up to another property.
CRA valuation results
From the investor's perspective, the most attractive feature of CRA-backed securities is that they are less negatively convex than agency conforming mortgages and many home equity securities. Convexity refers to the price/yield curve; negative convexity means that the price/yield curve flattens as interest rates decline). CRA securities tend to benefit from limited extension risk (they are priced to a slow base prepayment speed) and less call risk because of limited refinancing alternatives for the borrowers. To evaluate CRA-backed securities, we calibrated our agency conforming prepayment model in two ways: 1) we slowed the housing turnover component of the model to be consistent with the slow baseline speed of CRA loans; 2) we tempered the CRA refinancing function to account for the CRA's lower sensitivity to interest rates.
Figure 5 compares the projected prepayment sensitivity profiles of CRA loans to agency conforming loans. Using the CRA adjusted model, we can then compare CRA nominal and option-adjusted spreads to other sectors in the mortgage and asset-backed markets. Our results show that CRA securities compare favorably to equivalent agency securities on both a nominal spread basis and an option-adjusted spread basis. (For more details on CRA valuation results, see the October 2, 1997, Bear, Stearns publication, Securities Backed by CRA Loans: A New Product for Mortgage and Asset-Backed Investors.)
Better loans stay in longer
Under declining interest rate scenarios, the better-quality loans in a conforming loan portfolio will self-select out through refinancing. Borrowers with better credit scores (660 and greater) realize that the cost/benefit of refinancing works to their advantage as rates decline. At the other end of the spectrum, borrowers with lower credit scores either cannot get refinanced or would have to do so at a cost that would be unfavorable. The net result in a declining rate environment is that the better loans refinance out of the portfolio at a higher rate, leaving a higher percentage of inferior loans behind.
Bear, Stearns and Financial Modeling Concepts are in the midst of a new prepayment study. We have discovered some interesting refinance behavior that benefits the buyers of CRA securities. In a recent study of 6,393 CRA loans, 62.3 percent had FICO credit scores greater than 660 in August 1997. When we rescored the portfolio in February 1998, during one of the lowest interest rate periods of the decade, we found that nearly 89.6 percent of the August group maintained or improved their score levels. Best of all, not one of these loans refinanced during the same period. The net benefit for investors is that while declining interest rates negatively impact the quality of conforming loan portfolios, CRA portfolios are not nearly as adversely impacted.
To securitize or not
Because we've probably given you more than enough statistical data, graphs and charts, we shall leave you with a brief checklist to help you decide whether securitizing your CRA loans is an attractive option.
* If you can get 97 cents on the dollar or more, and depending on your loss tolerance, the financial consideration could be right for doing a CRA loan securitization. Quite possibly, your portfolio could bring you a 101 execution and a profitable sale.
* If more than 85 percent of the loans in your portfolio are current in terms of payments and had only zero through two 30-day delinquencies during the past 12 months, you are well within the correct payment history ranges.
* If you are setting aside inordinately high loan loss reserves against your balance sheet, you should consider freeing up the capital for more productive purposes.
* Forget about FICO scores and high LTV levels. Almost everyone evaluating your portfolio assumes that the scores will be low (many 660 and less) and LTVs will be high (90 percent and greater).
* Mortgage insurance is a "nice to have" amenity, but not a "need to have;" credit enhancements can be added later through subordinated securities.
* Do not be too quick to take a whole loan offer. Chances are you could be leaving money on the table by not going the securitization route. Always ask for proposals from your underwriter to act as principal as well as agent, and get proposals from at least two sources.
Until just recently, the market for buyers of CRA loans has been very limited - traditionally other banks or GSEs on a whole-loan basis. Today, a much larger and broader investment audience increasingly wants these loans. Education of the investor and rating agency communities has done much to correct the misconceptions about these loans. In the majority of cases, CRA loans can be a safe and stable investment. For instance, borrowers go through a more rigorous homebuyer education process than traditional borrowers. The loan denominations are smaller. The borrowers overall carry fewer debts (e.g., credit cards). Counseling is often available for those who are delinquent with their payments.
This past year at the annual Mortgage Bankers Association of America convention in New York, the chairmen of both Fannie Mae and Freddie Mac predicted that the segments of the population holding the greatest potential for first-time homeownership are recent immigrants and lower-income borrowers - both squarely within the CRA category. When we enter the new millennium and look back to the volume of CRA securitizations since 1997, we'll probably wonder why everyone waited so long to do something so logical.
Ned Brown is president and managing partner of Financial Modeling Concepts, Inc., Jersey City, New Jersey. FMC developed the first credit analytical models for CRA portfolios and has created the largest proprietary database of CRA loans. FMC works with clients to improve CRA portfolio credit enhancement terms and portfolio pricing.
Dale Westhoff is senior managing director of Bear, Stearns & Co. Inc., New York. He has been ranked No. 1 in prepayment analysis by Institutional Investor for five consecutive years. Bear, Stearns pioneered and is the leading underwriter of CRA mortgage-backed securitizations.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Community Reinvestment Act|
|Author:||Brown, Ned; Westhoff, Dale|
|Article Type:||Cover Story|
|Date:||May 1, 1998|
|Previous Article:||Prepayments swamp subprime lenders.|
|Next Article:||First Union's booming business.|