One thing that I am always asked about is the value of servicing. Let’s break apart the price of a mortgage loan into a few components. At a very simple level, a mortgage is a loan for a home and that home is used as collateral. These loans are pooled into bonds called mortgage backed securities. These bonds are rather large, and an institution that holds a bond doesn’t want to collect and process hundreds of payments ever month from John and Suzie Borrower. The solution is to have a servicer handle the payment collection and other clerical tasks involved in mortgage lending.
Also, the holder probably purchased the bond because there is some level of guarantee from an agency like Fannie Mae (FNMA). These two things, servicing and bond guarantee, are very important aspects of mortgage banking as they add liquidity to the loans we originate.
All residential mortgage loans can be broken into different value components. The security piece, the servicing, and the guarantee fee are the three most important components.
The security piece represents the majority portion of the loan and it is owned by the security holder. The servicing piece, often a .25% of the interest rate, is paid to the company who collects and remits borrower payments, manages escrow accounts, handles collection and foreclosure issues, etc. Most Agency products go into Mortgage Backed Securities (MBS) and are backed at some level by one of the Agencies (Fannie Mae, Freddie Mac, Ginnie Mae). The Guarantee Fee (GFee) is the Agency’s cut of the transaction.
Below is an simplified example of the value components of a 5.375% 30yr Fannie Mae loan.
5.000 FNMA Security
0.250 Servicing Fee
0.125 G Fee
5.375 Consumer Note Rate
The security value is determined by the market just like the price of a stock. The guarantee fee is set by the contract between the lender and the investor and/or Agency.
The value of servicing … well that is a bit more complicated. The value of servicing is the price at which a servicer will pay for the right to service and collect the servicing fee over the life of a loan. There are plenty of variables that make this estimation of future cash flows difficult. Most importantly, how long will a loan or a pool of loans last? Second, what fee income can a servicer derive from the loan(s)? Finally, what are the costs to service this loan? If one can answer these questions, then anticipated revenues and costs can be used to create a future cash flow model that can be discounted to present value dollars. Without a crystal ball however, assumptions have to be made in order to create these models.
These models are very complex and analysts try to adjust for the many variables that can affect these future cash flows. One could right an entire book on the valuation of servicing. For purposes here, I believe we should focus on the following three variables.
The most significant factor that affects value of servicing is runoff. The longer a mortgage pays monthly without paying off, the more cash flows the servicer collects. A mortgage can pay off at any time, and when it does the servicing instantly becomes worthless.
Simply put, the longer the servicer collects fee income, the more money the SRP is worth. So what does a servicer use to estimate the life of a loan?
The first thing is the interest rate of the loan in relation to the market rate. A 5.00% loan in a 6.00% market will likely have a longer life than a 6% loan in 5% market. The other is socioeconomic and real estate law trends that tend to fall within state lines. In general, all other things being equal, borrowers in some states are more sophisticated than borrowers in other states and will tend to refinance quicker. Also, there are state laws that cause some portfolios to runoff slower. For example, Texas state law makes it more difficult for a borrower to refinance to get cash out. Another factor is the cost to refinance. In Texas, title insurance is very expensive while costs to refinance in Colorado a relatively low. Servicers continually back test their portfolios and use historical payoff speeds to determine which states runoff fastest.
The second factor affecting the value of servicing is the float income earned on escrow balances. When are borrower makes his or her PITI payment to the servicer, the taxes and insurance go into an interest bearing account until the taxes and insurance are due. Insurance is almost always paid annually and therefore the insurance escrow balance will average 6 times the monthly insurance payment over the year. The taxes that are put in escrow also build up over time. The average balance of taxes depends on how often they are paid to the state. In Texas, taxes are paid annually and therefore there is an annual average balance of 6 months taxes. With the 1-2 months cushion, the tax multiple for Texas is 7.5, the highest of all states. Additionally, some state laws require the servicer pay the borrower interest for the money sitting in escrow, and as you can guess this hurts the servicing value in those states. Lastly, the amount of taxes collected varies by state. Again, Texas has an SRP advantage due to comparatively high percentage Texans pay in real estate taxes. If you have ever wondered why you are charged a .25 to waive escrows, it’s due to the loss of expected float interest on escrow funds.
The third factor is delinquencies. Obviously a delinquent loan is more costly to service than a good performing loan. If a loan ends up going into foreclosure, the servicer’s costs skyrocket. Not only due delinquency trends follow state lines, but foreclosure laws differ amongst states and therefore so does the costs of foreclosure. One side note – you will see that delinquencies can affect government SRPs more than conforming do to remittance differences. On government loans, the servicer has to remit P&I to the security holder whether or not the borrower has made their payment. If you own a highly delinquent GNMA pool, you can run through your cash pretty quickly while advancing P&I payments for all your delinquent borrowers. Florida is an excellent example of this. Delinquencies have run very high in Florida of the last few years, and during that time we have seen a significant decrease in Government SRPs but only a slight decrease in Conventional SRPs in this state.
As I mentioned earlier, servicing valuation models are extremely complex and contain many more variable than just the three above. I discussed these three topics as they have a significant impact on servicing values, and they also help explain why loans from one state are valued higher than similar loans from another.