Automated Underwriting Systems for FHA Mortgages Overview Fannie Mae and the FHA were created by the US Government to provide mortgage insurance to the mortgage industry. Mortgage insurance limits the amount of losses that a lender would incur in the event that a foreclosure occurs. In the mid- 1990s, Fannie Mae created Desktop Underwriter (DU) to reduce loan origination time and costs by automating the underwriting process. Eventually, the FHA followed suit with the release of the FHA TOTAL Scorecard system. This white paper discusses both DU and TOTAL Scorecard to explain their similarities, their differences and where third-party automated underwriting systems fit into the picture. Components of an Automated Underwriting System DU is one of the industry s first Automated Underwriting Systems (AUS). DU was created to help lenders determine if Fannie Mae would insure a loan. For each loan file submitted into that system, a DU certificate is returned to the lender. The certificate contains three decisions: the credit risk decision, the eligibility decision and messages to help the lender package the loan in a way that meets Fannie Mae s requirements. Credit Risk Scoring Module DU utilizes a mortgage-scoring model to determine the statistical credit risk for the loan. The credit risk decision is returned back to the lender as either Accept or Referred. Accept means DU has deemed the borrower creditworthy to receive mortgage insurance. Referred means either DU lacks the information to make a credit decision or the borrower has too many risk factors and manual underwriting by the lender is required. 1
Eligibility Module DU utilizes a rules-based engine to determine if the loan conforms to eligibility requirements established by Fannie Mae. While a borrower may be deemed creditworthy, the loan may still be uninsurable because the loan amount exceeds the maximum amount allowed by Fannie Mae. Eligibility decisions are returned as either Eligible or Ineligible. Messaging Module DU also utilizes a rules-based engine to provide messages to help the lender package the loan. These messages may be reminders for the lenders to validate additional factors. There may also be a requirement for the lender to submit additional documentation with the loan file. Messages are returned in the form of a list of sentences. Scoring vs. Rules-based A scoring module makes decisions through the use of mathematics. Variables are assigned values that increase or decrease the overall score. If the borrower s score surpasses a certain threshold, then they qualify. Whereas a scoring model uses mathematics, a rules-based engine uses Boolean logic. That is, it relies on a set of if-then-else conditions to determine if a particular result should be applied or not. For example, in the past, the maximum loan amount for Fannie Mae was $417,000. This rule may be written as IF LOANAMOUNT > 417000 THEN BORROWER IS INELIGIBLE. The key difference between the scoring model and the rules-based engine is that the scoring model looks at the overall picture whereas the rules-based engine looks only at particular variables. FHA TOTAL Scorecard FHA TOTAL Scorecard is a mortgage scoring system used to determine the statistical credit risk for a loan based on FHA s analysis. FHA TOTAL Scorecard does not contain a rules-based engine and therefore cannot render either the eligibility decision or the messages. FHA leveraged Fannie Mae s DU (and Freddie Mac s LP) system to create a complete AUS that lenders can use to retrieve credit decisions, eligibility decisions and messages for FHA insured loans. When a loan is 2
submitted through DU for an FHA loan, some of the data is transferred to TOTAL Scorecard for the credit decision. The remaining data is processed by DU s rulesbased engine operating according to FHA eligibility and messaging requirements. These datasets are combined together and a single report is returned to the lender. In return for the services that DU and LP provides, lenders pay a transaction fee for each loan submitted to these engines. Alternatives to DU and LP for FHA Transactions Until recently, only DU and LP had direct access to the FHA TOTAL Scorecard system. If a lender wanted to use the FHA TOTAL Scorecard system to evaluate a loan, they had to use DU and LP. Several years ago, the U.S. Congress issued a mandate requiring that the FHA allow external vendors to create alternative means by which decisions for FHA insured loans could be provided. Since FHA TOTAL Scorecard lacks a rules-based engine, FHA has only worked with vendors who have demonstrated the ability to implement an eligibility and messaging engine. As of this writing, only two vendors are currently approved to provide FHA decisions. Utilization rates for these new technologies are low even though they have been available to the marketplace for several years. There are several reasons for this. The first is education. Most lenders are unaware that there are alternative choices to using DU and LP to obtain automated decisions on FHA loans. The second is acceptance. People are comfortable with what they know and unless there is a compelling reason to change, they will not change. There is an assumption that if the lender obtains an FHA approval certificate through DU, it is somehow more reliable or legitimate than if it was obtained from an alternative vendor. The reality is the credit decision comes from the FHA TOTAL Scorecard itself regardless of whether the transaction goes through DU or an alternative vendor s AUS. Fannie Mae also does not provide the lender with any guarantee on the accuracy of the eligibility or messages. The lender is still responsible for the loan regardless of what the eligibility and messages say. A third reason is switching costs. It costs money to change a lending workflow to utilize a different vendor for specific types of loans. In this area, LOS companies can do more to mitigate this cost so that lenders can take advantage of this new technology that brings them more savings and better quality decisions. 3
The Benefits of Alternatives Fannie Mae and Freddie Mac charge users transaction fees for automated underwriting decisions. Unfortunately, the fees are set at a rate that deters users from fully taking advantage of automation. That is, these systems were created to automate parts of the loan approval process. Due to the high fees charged for a decision, lenders perform manual pre-screening on loans to determine if those loans would receive an approval from DU/LP to keep their marketing costs down. The consequence is the industry took two steps forward but one step back. By allowing alternative vendors to provide the same functions as DU/LP for FHA transactions at a lower cost, the FHA is actually promoting greater efficiency within the industry by motivating more lenders to use technology earlier in the process. As costs rise due to increased compliance requirements, lenders are under pressure to identify technology that they can leverage to lower their costs so that they can stay competitive. Lenders can leverage these alternative AUS to reduce their costs for FHA transactions by up to 80%. Alternative vendors also provide the FHA with more control over the quality of the messages that are returned to users. Prior to approving the system to be released to the marketplace, the FHA reviewed and advised these alternative vendors on the verbiage of the messages that are returned to users. So while these messages may not match verbatim the messages that are produced by DU/LP, the clarity is better because the expertise of the FHA was utilized in their creation. Some alternative vendors allow lenders to superimpose overlays on top of what the systems would return as a default. Thus a lender who is more conservative than even the FHA would be able to include their credit policies directly into the technology that is used by their staff. This eliminates a manual step performed by the underwriter to check the loan against the additional overlays. This also shifts the decisioning step earlier in the loan process. The result is a higher pull-through rate for lenders. Summary A lender who does a high enough volume of FHA transactions will benefit from the utilization of an alternative AUS to DU and LP. A proper analysis will tell the lender if the reduced fees and other benefits that a lender gains by using an alternative AUS justifies the one-time switching cost to change workflow. 4
Who we are LendingQB is a provider of 100% web browser-based, end-to-end loan origination software that offers residential mortgage banking organizations faster closing time and reduced costs per loan. LendingQB delivers a set of Best Practices called Lean Lending and performs the bulk of the work with a hands-on approach to the system configuration process. Lean Lending can be scaled to lenders of different sizes and reduces implementation time and cost. To learn more: visit www.lendingqb.com or call 888.285.3912 5