Loan Underwriting
Starting my professional journey as a field guy for Security Pacific Housing Services I think I have a solid ground level view of how underwriting really works. Particularly, I have direct experience in observing the effects of changes in underwriting policy on individual loan and portfolio performance. In my humble opinion you really can’t say you understand the assumptions behind underwriting policy until you’ve attempted to collect money in places like Okeechobbe County Florida and Shaboota, Mississippi or you’ve had to reposess a mobile home from the Sand Mountain area on the Alabama – Georgia line. I seriously think it should be written into Federal law that underwriting policy can’t be changed unless a field collector (5 years minimum experience) has the final say in validation of the assumptions used to make the change.
Enough of that and on to describe how policy worked for us in the 80′s and early 90′s vs. where we are today. Consumer loan underwriting had a major metamorphosis in the 80′s with the introduction of Asset Backed Securities in conjunction with a loophole in the tax code concerning the treatment of “Gain on Sale” as it relates to the issued securities (I’ll address that in part 2 of the series). As a result, lending institutions began to put together credit score cards that would force individual loans to conform to a standard pre-pay model therefore allowing the creation of “pools” of loans providing predictable pre-payments and cash flows. My focus in this note will be on loan underwriting for site built and manufactured housing.
The housing underwriting models of the late 80′s and early 90′s where largely based on Federal Housing Authority (FHA) Title I and Title II underwriting policy. FHA Title I is the manufactured housing program and FHA Title II is the single family real estate mortgage program. This model segmented the credit score card into several subsets of data and each subset carried a weight. The maximum score obtainable by adding up all the subsets was 100 points. I’ll use the Security Pacific Housing Services model as our example because it was obviously the most conservative in the late 80’s and the portfolios underwritten to this standard out preformed competing securities of the time:
1. Credit Score Maximum 35 points
2. Net Debt Ratio Maximum 20 points
3. Loan to Value Maximum 20 points
4. Employment
a. Current Maximum 5 points
b. Previous Maximum 10 points
5. Residence
a. Current Maximum 2 points
b. Previous Maximum 4 points
c. Previous Home Owner Maximum 4 points
Total Maximum Score 100 Points
Each of these subsets had very detailed procedures for how to tally points. The thought behind this card is very simple and is a result of fundamental underwriting questions: 1) Does the consumer have good credit, good income, strong equity?; 2) Does the consumer demonstrate stability through consistent employment and place of residence? As a lender we don’t like to see job hoppers and “new to the area” profiles. It is important to understand and never forget that underwriting is an art form, not an exact mathematical science. There still needs to be a human being in the loop that absorbs the data, closes his or her eyes, imagines the person being described in the profile and asks the question, “Will they pay me back?”
The fundamental slip down that ever slippery slope began here with the introduction of unfettered competition and the perceived need for and value of high growth. First slip occurred when the industry began changing the underlying requirements of the weights shown above. Then we began to change the weights themselves and finally we took underwriters completely out of the loop and digitized the entire process – underwriting via algorithm.
As competition increased, each institution was pressured to maintain market share. In the lending world this means you have to maintain both loan application count and the ratio of loans approved compared to the number funded. As a lender you present a pure commodity to the client – a loan. Being more competitive means you begin to do one of two things: 1) Reduce verification requirements which are perceived as “hassles”; 2) Loosen your underwriting standards to allow for higher approval rates. Not much wiggle room outside of those two choices for product differentiation.
The result was that the industry would punch and counter punch by first increasing allowable rates of “credit exceptions”. Perfect example – deals with medical collections showing up in a credit bureau. When I started in 1988, I was not allowed to approve a loan where the client’s credit bureau had a medical collection above $250. By 1998 the same client was allowed medical collections of up to $2,000. The argument was that there was no real effect on the client’s ability to pay. HOG WASH! These collections lead to judgments which lead to garnishments and corresponding reductions in disposable income.
The second issue was an incremental change in the way the industry looked at income. When I began my tenure, we calculated Net Debt Ratio, which was the total monthly gross income multiplied by the median tax rate divided into the total debt. The debt was calculated by summing the credit bureau, plus any debt shown on the credit application plus the house payment. Your point tally decreased as the net debt ratio approached 50%. Any client showing 52% or more net debt ratio was automatically turned down. One other metric that did not count toward the score but insured “affordability” could be demonstrated through calculation of net disposable income. The calculation was derived after tax monthly income minus total debt (including the house payment) minus a calculated cost of living allowance (Food, utilities, insurance, etc) In 1988, if the client had less than $200 a month in net disposable income they where automatically turned down.
By 1998, the debt ratio calculation had change to 40% of gross and exceptions where made up to 50% of gross. No disposable income calculation existed. In conducting comparisons of the two metrics we easily demonstrated why default rates in manufactured housing began to sky rocket as a result of this change alone in the late 90’s through 2002.
Another detrimental change was the reduction in the equity requirements and increased Loan to Value ratios. In 1995 I personally ran a test with 25 zero down loans. Great marketing, but horrible long term performance. None of these loans had loan credit card scores below B+ (85% where A or better) and after 36 months we had a 50% default rate. A normal default curve with 10% down payments demonstrated 17% default rates over 7.5 years.
Loan to value is the measurement of the gross loan amount to the current value of the property. The higher the loan to value ratio the higher risk you have of default. This occurs because it takes many more months of principle payments to put the client in a strong equity position for resale. In 1988, manufactured housing loans had maximum loan to value ratios of 125%. By 1998, the rates had climbed to as high as 145%. Real Estate loans where adjusted in the same manner by changing appraised value methodology and allowable comparison properties.
All of these changes occurred in parallel without proper testing. The argument was always, “This tweak surely will not affect overall performance enough to move the needle on a $100 million deal”. Always a “rational” explanation existed as competitive pressure mounted. “Slip, slip, .. sliden’ away!”
