Discount Points and Housing Prices: Comment

Lenders make FHA insured and VA guaranteed mortgage loans at a discount. One important reason for doing this is to make up for the difference between the ceiling-contract interest rate and the market interest rate. Because lenders are unable to extract this discount from the borrower-buyer legally, they charge the seller discount points, a percentage of the loan amount. The seller, in turn, attempts to shift this burden back to the buyer through a higher selling price than would be charged if the sale were financed conventionally. Zerbst and Brueggeman [Z&B, 3] question the extent to which discount points paid directly by house sellers are shifted to buyers through higher prices. They should be congratulated for asking this interesting question. Unfortunately, their model is incapable of answering it. This paper clarifies the problem with the Z&B approach, develops a model which allows the direct estimation of the proportion of points shifted, estimates that proportion, and tests whether the estimate is significantly different from unity (i.e., 100% shifted). Narrowly interpreting their own results, Z&B find that only about 43% of points shifted to FHA buyers.' While as little as 43% of discount points may be shifted during brief and peculiar episodes like the sample period of the Z&B study, it is not reasonable to compute the percentage, whatever it is, on the basis of the Z&B model. The problem with the Z&B model stems from not treating asking price as a function of the probability of a non-conventional loan. This is untenable because there are some neighborhoods in which FHA mortgages are likely and others in which FHA mortgages are unlikely. Furthermore, there are individual sellers who refuse to sell at the asking price with FHA financing if they expect to do better (i.e., net of points) with a conventionally financed sale. Recognition that some financing outcome can be reasonably anticipated by the seller gives rise to the possibility that sellers adjust their asking prices to reflect anticipated financing. Thus, the Z&B model can be shown to be consistent with any shifting experience, including complete shifting of discount points. To argue otherwise is to suggest either that sellers find it impossible or too costly to acquire information on these probabilities or that having this information sellers proceed to act irrationally.