a forward commitment. At the program's outset, it was clear that the pricing structure provided a price that was too high for the current market. Fannie Mae agreed to adjust its price through a temporary concessionary pricing structure. EMI had also planned to offer a bridge-financing program to finance tax-credit equity pay-ins to the projects. The revenue from this program was an important part of EMI's projected operating revenue. Fannie Mae, however, was unable to offer a price on the bridge loan product that left room for a fee to EMI and could still be competitive in the marketplace. As a result, EMI became essentially a one-product program: the forward commitment on LIHTC transactions.

By June 1997, more than 21⁄2 years into the program, EMI had made commitments under the concessionary pricing to 13 projects, totalling $18.5 million in mortgages. The start was considerably slower than expected, largely due to sometimes lengthy negotiations between EMI and Fannie Mae over underwriting specifics, particularly concerning some of the special features of affordable projects (such as requirements of various types of subsidies and requests for waivers). In addition, Fannie Mae's documentation requirements were burdensome, particularly for small, nonprofit borrowers. In late 1996 EMI proposed fundamental program changes to Fannie Mae that it felt were necessary for doing business. In January 1997 EMI and Fannie Mae agreed to restructure the program, and Fannie Mae awarded EMI full authority to underwrite, grant provisional waivers for, and commit to loans of less than $3 million, for a total pool of $50 million. EMI was granted the authority to enter into affordable housing transactions of more than $3 million, but EMI is required to compete with other delegated multifamily lenders on their terms; EMI would not receive concessionary pricing on these transactions. The EMI/ Fannie Mae program is currently doing business under this new structure, focusing almost exclusively on the forward commitment for transactions under $3 million.

These two efforts have helped in addressing the question of how to involve a GSE in the secondary market for affordable multifamily mortgages. There are some fundamental differences in the programs, but both experienced slow starts and encountered many similar obstacles. The obstacles principally result from wide differences between the GSES' and their partners' assessments of the risks associated with the loans, as raised in the underwriting and pricing of the transactions. Both programs suffered under the weight of time-consuming negotiations that often resulted in processing costs that were higher than revenues. The discussion that follows provides a detailed description of the five problem areas for these programs: underwriting; pricing; information and expertise; the capacity issues associated with small, startup organizations working with two large institutions; and the financial viability of a business focused on providing mortgages for affordable rental housing.

Underwriting

The mortgage underwriting process is an area in which standards and information are crucial. Underwriting assesses the soundness of the mortgage as an investment. Uniform underwriting guidelines are essential to the operation of the secondary mortgage market because they assure the investor that-regardless of the individual underwriter or the location of the project-the assessment of the investment follows standard guidelines. The LIMAC and EMI experiences reveal several underwriting issues that are fundamental to the successful development of a secondary multifamily mortgage market:

Delegated underwriting, so that only one party underwrites the transactions.

■ Establishing underwriting standards appropriate to affordable multifamily developments, which are often relatively small projects developed by nonprofits.

■Determining a level of documentation that sufficiently ensures the project's soundness, yet does not overburden the borrower or the lender.

Delegated underwriting. Delegated underwriting assigns authority and responsibility for underwriting transactions to one party. Freddie Mac made it clear from the beginning of the LIMAC program that it had no intention of delegating underwriting; this was Freddie Mac's policy, and there was no exception for multifamily transactions. Instead, the LIMAC/Freddie Mac program was structured so that the originating lender, LIMAC, and Freddie Mac all underwrote each transaction. Underwriting for the one completed transaction by the three participants produced quite different results, indicating clearly divergent views of the risks associated with these properties. LIMAC and Freddie Mac tackled underwriting issues for more than a year for the one transaction they closed.

A major principle that was unique to the EMI/Fannie Mae program was that EMI was to be a delegated Fannie Mae underwriter. Because EMI was the lender, it was to be the sole underwriter. Before being granted delegated underwriter status, however, EMI had to qualify in each of Fannie Mae's five regional offices by satisfactorily completing at least three transactions per region under precommitment review, during which Fannie Mae and EMI each underwrote every transaction. Two years into the program, EMI was unable to gain approval for delegated status in any of Fannie Mae's regional offices, and there was no indication that it was to be released from precommitment review soon, even though it had completed eight transactions in one of the regional offices. The lack of delegated underwriting made the program too expensive and time consuming for both EMI and Fannie Mae. In addition, it was difficult to keep prospective customers in the pipeline because EMI was unable to make a clear commitment in a timely fashion.

One of Fannie Mae's concessions in the January 1997 program restructuring was to grant EMI immediate delegated underwriting status for $50 million of qualifying affordable multifamily mortgages of $3 million or less. Fannie Mae removed precommitment review requirements in all regions and granted EMI the authority to issue some waivers for exceptions. Making EMI a fully delegated underwriter was fundamental to testing whether a viable business could be built based on these kinds of transactions. Unfortunately, during the first 6 months of the restructured program, EMI needed to restaff and reorganize, which slowed program progress. It remains to be determined whether the restructuring will allow EMI to shorten its underwriting and processing time, thereby increasing its efficiency.

Establishing underwriting standards for multifamily affordable housing. The LIMAC/Freddie Mac program was guided by a Master Agreement, which, among other things, established a set of underwriting guidelines for the program to anticipate potential problems in underwriting. In practice, however, LIMAC and Freddie Mac usually disagreed on the implementation of those guidelines. Even when both claimed to be following the guidelines as meticulously outlined in the Master Agreement, they often reached different conclusions in their analyses. In addition, the lender for the completed transaction had held the loans in portfolio and had used a different set of underwriting guidelines and documentation requirements when it originated the mortgages.

Freddie Mac and LIMAC placed significantly different values on the properties underlying the loans. Just 3 months before closing, Freddie Mac's appraisals of the eight projects resulted in loan-to-value (LTV) ratios that averaged 91 percent; LIMAC's assessment of LTV ratios for the same projects averaged 71 percent (close to the lender's original assessment of 72 percent, which at the time had sparked criticism by local community development groups for being too high). When the transaction closed, Freddie Mac's final weighted average of the eight LTV ratios was 84 percent.

The discrepancy arose from differences in three assessments:

The estimation of project income.

The treatment of soft second mortgages.

■ Capitalization rates.

To determine project value, net operating income is divided by a capitalization rate, which reflects current market conditions, including interest rates and an assessment of risks associated with the projected rental income stream. To estimate projected income, LIMAC considered net operating income based on tax-credit-restricted rents as well as the flow of tax credits to the investor, arguing that tax credits were an important part of the return to the equity investor (four of the eight mortgages were for LIHTC projects). Freddie Mac considered unrestricted, or market, rents but did not include the flow of tax credits.

Many projects had soft second mortgages, which Freddie Mac treated as hard loans with fixed payment schedules, and which LIMAC treated more like grants. LIMAC's and Freddie Mac's capitalization rates were also far apart: LIMAC used rates that were between 12 percent and 12.8 percent; Freddie Mac's rates were as high as 16 percent. LIMAC and the lender believed that Freddie Mac undervalued properties located in low-income neighborhoods by failing to consider elements that substantially strengthened the projects, including the benefits of tax credits, the level of commitment of local public officials, and lender and developer histories. From the perspective of LIMAC and the lender, Freddie Mac's approach to underwriting these loans seemed to ignore some essential features of affordable rental housing that influence the risk associated with the projects and may have resulted in its overstating those risks.

The climate for the EMI/Fannie Mae program was considerably different. As the EMI program began, Fannie Mae was completing its new underwriting guidelines for affordable multifamily housing, incorporated in their Delegated Underwriting and Servicing (DUS) program. Fannie Mae's guidelines dealt explicitly with various issues in affordable rental housing. DUS specified guidelines for LTV ratios and debt service coverages (DSCs) appropriate to affordable rental housing, explicitly accounting for various government programs, such as LIHTC. Both EMI and Fannie Mae used the same appraisal. Net operating income was based on the restricted low-income rents, and capitalization rates for completed transactions varied but averaged approximately 10 percent. In recognizing the unique features of tax-credit transactions and that tax-credit and other affordable housing projects are inherently different from standard market-rate transactions, Fannie Mae's guidelines were a significant improvement over those used by Freddie Mac for the LIMAC program. The EMI program's defining document, the Special Purchase Agreement (SPA), revolved around Fannie Mae's new guidelines, making EMI well positioned to test the application and performance of Fannie Mae's new approach to this segment of the market.

In practice, EMI and its borrowers had few complaints about the program's underwriting guidelines. Although Fannie Mae's guidelines could still be a significant challenge for affordable housing, EMI found that a sufficient number of transactions could successfully meet those requirements. Yet continuing disagreements between EMI and Fannie Mae resulted in long processing delays and limited the volume of the program. Although EMI had expected to process applications within 3 months of receiving them, it regularly required 6 months or longer. In some cases EMI and Fannie Mae simply disagreed on the implementation of the guidelines, but the majority of problems originated in areas not

explicitly addressed in the guidelines, or in processing waivers or exceptions that can often arise with affordable housing. EMI found that its projects typically required waivers or exceptions, even with Fannie Mae's affordable housing guidelines. Much of the delay in processing arose from negotiating terms and gaining Fannie Mae's consent for waivers. Often EMI staff indicated that Fannie Mae staff did not understand the affordable housing market segment, whereas Fannie Mae staff indicated that EMI did not comply with documentation requirements.

Documentation. Both programs have suffered from burdensome documentation. The initially dense documentation for the LIMAC/Freddie Mac program became increasingly burdensome over time to both LIMAC and the lenders. More than one potential lender dropped out of the program because of the amount of documentation required. The only lender to complete a transaction under the program felt inundated by Freddie Mac documentation requirements and had hoped that LIMAC would do more to push Freddie Mac to simplify the mortgage documents and to assist in their compilation. Because the transaction involved eight existing loans, the lender already had several years' worth of loan origination documentation and servicing records for each loan. However, the information needed to be reformatted to fit Freddie Mac's documents. The lender was frustrated that LIMAC never provided a comprehensive list of what was needed. LIMAC believed that there was no one office or individual within Freddie Mac who was coordinating Freddie Mac's requirements; Freddie Mac's requests were piecemeal and transaction-specific rather than standardized for the program. As a result, the documentation seemed repetitive, disorganized, and endless. As LIMAC considered new transactions, there was no indication that Freddie Mac's documentation requirements would become less complicated or standardized. If anything, the documentation requirements increased over time. For example, Freddie Mac's forward commitment letter on the second, unsuccessful transaction was 22 pages with attachments.

The EMI/Fannie Mae program also required burdensome documentation, although to a lesser degree. From the outset, both borrowers and EMI staff complained of cumbersome documentation requirements. Particularly early on, EMI was aware that some borrowers withdrew preliminary applications because of burdensome documentation, reporting, and due diligence requirements. In our interviews, one borrower with considerable experience in multifamily housing said that she would happily pay a premium to avoid working with Fannie Mae. Many EMI borrowers with whom we talked were nervous about the paperwork requirements they faced. The EMI program had considerably more reporting requirements, especially during construction and project management, than those of local lenders. Some forms seemed incompatible with certain characteristics of their projects. As with LIMAC, some borrowers said that they expected EMI to be more of an advocate for them with Fannie Mae and were disappointed that EMI failed to challenge Fannie Mae's rules. Fannie Mae staff, though, were frustrated early on with EMI's unfamiliarity with their documentation. Staff in one regional office complained that they had to retype EMI documents to get the information in the correct Fannie Mae format.

The restructuring of the program granted EMI the ability to issue some provisional waivers without formally applying to Fannie Mae, which may ease the situation. Although Fannie Mae staff acknowledged that the heavy documentation and due diligence requirements could be problematic and that they would like to ease the burden, as of June 1997 there had been no changes to requirements.

The typical underwriting, documentation, and due diligence requirements of Freddie Mac and Fannie Mae transactions were particularly burdensome for programs targeted

at affordable multifamily housing. The smaller, nonprofit, and CDC developers that these programs intended to bring to the market were unprepared, and perhaps unwilling or unable, to meet the high costs of Freddie Mac's and Fannie Mae's due diligence requirements. Certainly our interviews with potential and actual program borrowers indicated that smaller organizations unfamiliar with Freddie Mac and Fannie Mae were most likely to have difficulty meeting the requirements. Given the importance of these smaller borrowers to providing affordable housing in many local markets, streamlined underwriting and documentation requirements are critical to the development of a secondary market for affordable rental housing.

Pricing

As discussed in the introduction, a major benefit of an active secondary mortgage market is increased efficiency in funding mortgages, which should decrease total costs for the borrower. The experiences of the LIMAC/Freddie Mac and EMI/Fannie Mae programs illustrate the difficulty inherent in offering a rate that works for the borrower but still compensates the lender and investor.

Making the price competitive. Exhibit 1 shows that the cost to the borrower consists of upfront fees and other components that make up the rate on the loan. In the LIMAC/ Freddie Mac program, the borrower paid an initial purchase fee of 1 percent of the loan amount to LIMAC. The rate on the loan consisted of the rate to the investor, plus 50 basis points (bps), or 0.5 percent, to LIMAC for credit enhancement, plus a guarantor fee to Freddie Mac determined by the risks of the transaction, plus a 25-bps loan-servicing fee to the lender.

Although the pricing of the EMI/Fannie Mae program had more elements, it was somewhat simpler because EMI was the lender and Fannie Mae was the investor. The upfront fees to EMI included a $1,500 application fee and a placement and commitment fee of 2 percent of the loan amount. Fannie Mae received a $1,500 postpurchase fee. In addition, Fannie Mae required a 2-percent mandatory delivery fee to be refunded when the loan was delivered. The loan rate was based on Fannie Mae's published rate for affordable housing, plus a forward commitment fee to Fannie Mae, plus a 50-bps loan-servicing fee to EMI.

For both LIMAC and EMI, generating sufficient income to cover operating expenses was a challenge, given the concentration on small affordable housing loans. As discussed earlier, underwriting these loans is often complicated by many anomalies. However, income is based on loan size. On a $1 million loan, LIMAC would generate $10,000 in fees plus 50 bps of the unpaid mortgage balance (UPB) per year for credit enhancement ($5,000 in the first year). For EMI, a $1 million loan would generate $21,500 in fees plus 50 bps per year in servicing ($5,000 in the first year).

The LIMAC/Freddie Mac program had some unique pricing problems. The rate on the loan had to compensate four participants: the investor, Freddie Mac, LIMAC, and the lender. The rate to the investor was determined first. Because the rate to the investor was locked in, any gap between the time LIMAC locked in the investor's rate and the time the rest of the pieces were in place created exposure to interest-rate risk, which could be expensive. In the one completed transaction, LIMAC committed a return to the investor 20 months before it closed. By that time, decreases in market interest rates meant that the investor seemed to be getting a return that was too high. As a result, the investor agreed to modify its return twice before the loans were delivered. Because Freddie Mac's fee was determined last, its fee was viewed as determining the viability of pricing, leaving Freddie