Probate & Property Magazine
The Value of Sunshine
The Efficacy of the Real Estate Settlement Procedures Act Disclosure Strategy
By Mark Shroder
Mark Shroder is Associate Deputy Assistant Secretary for Research, Evaluation and Monitoring with the Office of Policy Development and Research at the U.S. Department of Housing and Urban Development. An earlier version of this article appeared in Cityscape. No opinion in this article should be interpreted as an expression of the policy of the U.S. Department of Housing and Urban Development.
The Real Estate Settlement Procedures Act (RESPA), 12 U.S.C. §§ 2601–2617, regulates the provision of services involved in the sale of most single-family homes in the United States. Its declared intent is to protect the consumers of those services, but very little empirical research has been done to determine whether the statute efficiently accomplishes its intended goals. This article takes an initial step toward answering that question by analyzing a small sample of FHA-insured loans. Although the sample is too small to provide definitive conclusions, the analysis raises substantial questions about the basic assumptions underlying the RESPA regime and suggests that further research is needed.
RESPA’s Regulatory Scheme
RESPA regulates the conduct of ser-vice providers when a single-family home is bought with a loan that (1) comes from a federally insured depository institution or a federally regulated lender, (2) is insured by a federal agency, or (3) will be sold to Fannie Mae, Ginnie Mae, or Freddie Mac. RESPA therefore governs most single-family mortgages in the United States.
Two of the primary purposes of RESPA are to provide “more effective advance disclosure to home buyers and sellers of settlement costs” and the “elimination of kickbacks or referral fees that tend to increase unnecessarily the costs of certain settlement services . . . .” 12 U.S.C. § 2601(b)(1), (2). The statute attempts to address these goals through the following substantive provisions:
• Section 4 provides that the U.S. Department of Housing and Urban Development (HUD) must prescribe a standard form (HUD-1) for the statement of settlement costs. The person conducting the settlement, who is usually an attorney or escrow agent, must give the form to the borrower at settlement or, on request, one day before. Id.
• Section 5 states that within three days of receiving a loan application, the lender must make a good faith estimate (GFE) of the settlement costs the buyer is likely to incur. The GFE does not have a prescribed form but usually follows in part the format of the HUD-1, in which the corresponding actual costs are detailed. Id. § 2604.
• Section 8 states that no person may give or receive a kickback, fee, or any other thing of value in return for referring business to a settlement provider. The penalty for violating this provision is a fine of up to $10,000 or a prison term of not more than one year. The prohibition does not apply to payments for ser-vices actually performed. Id.
A Sunshine Regulation
RESPA is primarily a form of sunshine regulation because it regulates disclosure of information to customers, rather than the quality or cost of the service itself. RESPA does have a criminal penalty for kickbacks, id. § 2607(d), but one should not overstate its deterrent power. Nobody ever goes to jail for RESPA violations, and the maximum statutory fine of $10,000 has not been raised since 1974. Thus, the consumer dealing with an unethical provider must look essentially to the disclosure requirements and his or her own resources for protection.
Regulation by disclosure is a common strategy in the United States. Examples include requirements for financial disclosure by publicly held corporations and banks, labeling laws for food and pharmaceuticals, statutes requiring prior notice of plant closings, informed-consent prerequisites for medical experiments, and obligations on car repair garages to provide initial estimates of the cost of repair.
The common theme of such regulation is that the less-informed party must receive some minimum information from the more-informed party before the transaction is considered lawful. The government does not prohibit any type of transaction as inherently unfair; it merely requires disclosure to allow consumers to make informed decisions. Louis Brandeis stated the rationale in a 1913 quote much beloved of lawyers and regulators: “Publicity is justly commended as a remedy of social and industrial diseases. Sunlight is said to be the best of disinfectants; electric light the most efficient policeman.”
The language of the Brandeis quotation is loose. Sunlight is not the most powerful of disinfectants (try chlorine or iodine), nor do electric lights have the power to arrest perpetrators. Publicity is often a less costly policy instrument than enforcement, however, so we can read Brandeis as arguing that sunshine regulation may have greater net social benefits, given enforcement costs, than any regulatory alternative, including laissez-faire.
Sunshine is a poor remedy, however, if ignorance is not the problem. Consumer ignorance is certainly a factor in the market because most people do not buy and sell real estate more than half a dozen times in their lives, but ignorance is not the only factor. Two other problems should be noted—timing and control.
Timing is a factor because the buyer and seller usually endure considerable awkwardness or inconvenience if the closing is delayed. Many sales are motivated by divorce, disability, or death, and many purchases by new jobs or household formation. Much may be extracted from people in a hurry, even if they are well-informed.
Consumers also lack control over the actual settlement costs. In a typical transaction, the settlement expenses include not only the costs of the loan itself (such as loan origination fees or discount points) but also the costs of routine services such as the title search, title insurance, appraisal, home inspections, and credit checks on the buyer. Because most of these services are demanded by the ultimate lender or insurer, often the mortgage banker or broker will select the responsible agent to perform them. Yet the banker or broker does not bear the cost of the services; instead, the costs are customarily shared by two parties—the buyer and the seller. Thus, the normal vigilance that consumers maintain over their own spending may be relaxed not only by ignorance and urgency but also by the reflection that somebody else will pay for part of it.
To evaluate the efficiency of RESPA’s approach to this combination of problems, it is helpful to consider alternative approaches. In 1974, the leading legislative alternative was “lender-pay,” under which the lender would be responsible for all settlement costs. Although lenders would pass these costs on to the borrower (and indirectly to the seller), competition among lenders would give each lender a strong incentive to control settlement provider fees.
In 1998, HUD and the Federal Reserve Board endorsed a somewhat modified version of the lender-pay proposal, under which the lender would be required to offer all settlement services as a single package at a fixed price. The lender would have been permitted to offer other pricing alternatives but required to offer a package price as well. If the borrower selected a package price, unexpected deviations in the cost of services would come at the expense of the banker/broker, who would, however, have both the knowledge and the incentive to control them.
Some large lenders advocate package price legislation because they consider section 8 of RESPA an obstacle to profitable offerings of fixed-price closing fees. The prohibition on kickbacks, referrals, and unearned fees might be applied to the volume discounts they would try to obtain from the service providers in a fixed-price context.
Smaller lenders and service providers oppose this proposal, for essentially the same reason. The ability of large lenders to demand and obtain volume discounts would enhance their competitive position relative to smaller lenders and would reduce the prices charged by service providers.
Finally, any consideration of regulatory alternatives should take into account the role of the states. State law defines the rights of the property owner and whether those rights have transferred. Title services flow almost entirely from the rules of the property game defined by the several states, while “lending” services are defined by national standards. The fees for these services are usually distinct.
Thus, there are at least three conceptual alternatives to the status quo—laissez-faire, lender-pay, and state law reform. An empirical research agenda can be defined around a set of propositions that represent necessary conditions for the status quo to be better than these alternatives.
An Empirical Agenda
Four questions are critical in determining whether RESPA’s disclosure requirement, by itself, is an efficient regulatory strategy: (1) whether lending and title fees are large enough to be worth regulating; (2) whether the good faith estimate is an unbiased and consistent estimator of lending and title fees on the HUD-1; (3) whether state law has only a negligible influence on fees (such that national regulation is the most desirable solution); and (4) whether disclosure makes the market “fair” by ensuring that the personal characteristics of buyers and sellers do not influence the fees they pay.
This article examines those four propositions in the context of a small and unrepresentative sample of closings in June 1997 that resulted in FHA loan guarantees. No assurance is given that this particular sample is a fair representation of the U.S. housing market in 2007 because the purpose of the sample was to demonstrate only what would be learned from a definitive empirical study. Nonetheless, the author does not expect the results will surprise experienced practitioners and does not expect subsequent studies to materially undermine them.
The settlement fees analyzed here are those that belong in the 800, 1100, and 1300 series of the HUD-1, with a few exceptions. The intent is to exclude obvious choice variables, either seller’s choices or buyer’s. Excluded fees are real estate agent commissions and points paid to the lender, whether for “origination” or for buying down the interest rate. Payment of the FHA mortgage insurance premium (MIP) at closing (“upfront MIP”) rather than over time is also excluded. Taxes to state and local governments are not within the scope of RESPA, and funds paid into escrow are not within the scope of this article.
1. Lending and Title Fees Are Large Enough to Be Worth Regulating
If lending and title fees were trivial, then mandated disclosure would add to cost without adding enough to consumer surplus to make regulation worthwhile. In the real estate closings studied in the sample, the lending and title fees were in fact significant amounts.
First, a word on methodology: measurement of transaction costs needs to be scaled to the size of the transaction. The putative sales price of the home may be misleading because the seller may agree to a lower (or higher) price in return for a smaller (or larger) share of the closing costs. By agreeing to a $1,000 increase in his or her share of the fees, for example, the buyer can perfectly compensate the seller for a $1,000 reduction in the sales price. In this article, the transaction is scaled by the “value to seller” (VTS), which is the net change in the seller’s financial assets as a result of the transaction; that is, net cash plus payoff of debt. “Debt” includes all mortgages, unsecured debt, ex-spouse’s share of the proceeds, delinquent property taxes, unpaid child support, and similar obligations. Debt excludes payments for improvements to the property to meet FHA standards or the buyer’s demands; such payments also are not counted as settlement fees.
VTS averaged $84,278 in the sample while total lending and title fees averaged $2,060, or 2.4%. As indicated above, these fees do not include either points or commissions to real estate agents. Overall, lending and title fees varied between $692 and $5,671, or between 0.7% and 10.6% of the VTS.
The relationship between VTS and fees is linear in a very rough way. At least one important fee, the title insurance premium, is charged per dollar of the sales price or mortgage principal. Other significant fees, however, are theoretically based on the time the agent spends on the service, that time being largely unrelated to the size of the transaction.
The substantial portion of settlement fees not related to house price has negative social consequences. It rewards the buyer for engaging in larger transactions—for buying a new home, for example, rather than purchasing and renovating an existing one, or for buying in high-priced jurisdictions rather than low-priced ones. In terms of equity, it favors buyers and sellers of more costly homes over less costly ones. Buyers and sellers of cheap homes tend to have lower incomes than buyers and sellers of expensive units.
In short, lending and title fees are large in absolute terms, have a structure that may distort the housing market, and will tend to work a disproportionate burden on the least affluent participants in the market. If regulation would do any good—a point that cannot be taken for granted—these fees are worth regulating.
2. The Good Faith Estimate Is an Unbiased and Consistent Estimator
If lenders disclosed expected costs but their predictions were generally wrong, mandated disclosure would not be an effective approach. The closings studied in the sample suggest that lender predictions were generally accurate but subject to wide variations in a few individual cases.
The GFE is present in only 47 of the 146 FHA binders studied in the sample because the FHA stopped requiring a copy of the GFE for underwriting purposes in 1996. A regression of the actual lending and title fees on the lender’s predicted lending and title fees performs poorly and suggests serious bias. The linear least-squares estimator suggests that the amount actually paid was $290 plus about 83 cents for every dollar predicted in the GFE (but with an R-squared of .276). A perfect estimator would yield an intercept of zero dollars and a slope of 1 dollar for every dollar predicted in the GFE. The F-statistic for (0,1) being the true intercept and slope coefficients, respectively, is 189, so the null hypothesis that the true parameters are, in fact, 0 and 1 at conventional confidence levels can be rejected.
To put these results in plain English, the GFE is right “on average,” but many GFEs are off by a lot. The average value of the difference between the test variable and the GFE estimate is just 75 cents, but the mean of the test variable is $1,832 and the mean of the GFE estimate is $1,332. In this sample, most buyers got, on average, small overestimates (29 cases too high out of 47) and a minority received, on average, large underestimates. The average absolute value of the error is $328, so the typical estimate is off by about 18%.
Many lenders seem to prefer small overestimates of the title and lending fees to make sure the buyer will have enough money on hand to close. More troubling are the minority of cases in which very large underestimates occur.
3. State Law’s Influence on Fees
If state law had a major role in influencing fees, then state reform would be, at least, a necessary complement to the federal statute and, possibly, a replacement for it. Perhaps somewhere in America, a property attorney believes that state law has little to no influence on fees, but common sense would warn any analyst that title fees are highly sensitive to state law, if only because the clarity of state law determines the clarity of the title that is being transferred.
It is not easy to make the common- sense case with data from a small sample. In visual examination of mortgage insurance binders, two states stood out. Title determinations in New Jersey seem to require much higher attorney’s fees and much higher premiums per dollar of title insurance coverage than in other states. The office of the state treasurer has assumed responsibility for title insurance in Iowa, and, from casual inspection, premiums appear to be lower there by hundreds of dollars than they are in other states. Regression results tend to support this impression.
4. Does RESPA Make the Market “Fair”?
RESPA seems designed to promote equity among principals so that all purchasers of settlement services have a common minimum access to relevant information. Although social science has no universally accepted definition of “fairness,” a working definition for empirical purposes might be that people with different identifiable characteristics should not pay different fees for the same services unless those characteristics are linked to higher costs of service provision. If individual characteristics not linked to cost are associated with differences in the level of fees, then disclosure regulation might protect some principals less efficiently than other forms of regulation would.
The study disclosed wild variation in the detailed fees charged. For example, the credit report is a standard national, largely automated, service that typically cost about $50 in 1997, but charges in the sample range from $25 to $100. What can explain these differences?
Let’s begin by comparing title fees and lending fees. One reasonable hypothesis is that fees for title services are completely independent of fees for lending services because these fees are quite distinct in character. If that assumption is correct, there would be zero correlation between one fee category total and another.
A second reasonable, and quite different, hypothesis is that compensation occurs within the overall transaction, an apparent overcharge on one line effectively paying for other services. In that case, there would be a negative correlation between one fee category total and another.
The third hypothesis might be termed the Eli Wallach proposition, from the reasoning of the bandit leader in The Magnificent Seven—“If God had not meant them to be sheared, he would not have made them sheep.” In this hypothesis, a sheep can be sheared on one side for lending fees and on the other side, too, for title fees because some people are candidates for high fees in both title and lending. It would follow that there would be a positive correlation between these categories.
Holding VTS constant, $1 more of lending fees translates into another 24 cents’ worth of title fees as well. Cross-fee compensation is not occurring in this sample, and the fees for different services are not independent of each other. This result does not prove but is consistent with the Eli Wallach proposition.
Let’s try a more comprehensive approach, and attempt to explain the level of title and lending fees as a whole. Consider the hypothesis that title and lending fees are a function of the VTS, of state law, and of buyer and seller characteristics.
Successful builders and developers who plan on a large number of similar transactions can capture whatever economies of scale exist in lending and title processes. For example, some new home sales do not appear to have appraisal or survey fees, possibly because of special banking arrangements that the seller has made, and some appear to have reduced title fees.
A seller can be considered “troubled” if the sale appears motivated by a divorce or if there is substantial delinquency on property taxes. This variable could raise fees by one of three routes—by increasing the complexity of the transaction; by heightening the time pressure on the seller, who must pay bonuses to speed up the process; or by reducing the seller’s resistance to agent opportunism.
When an above-average interest rate loan is brokered to the lender, the latter may pay the broker a premium outside of closing. Mortgage brokers—but not bankers—are obliged to report all such payments on the HUD-1. These “service release,” “above par,” or “yield spread” premiums are substantial, ranging from 1% to 4% of the loan principal in this sample. Perhaps such premiums are paid in exchange for discounts on closing fees; other things being equal, their disclosure on the HUD-1 should strengthen the bargaining position of the buyer. If, however, they merely reflect the buyer’s naïveté or some high value of time relating to his or her situation, there will be no compensation in lower lending fees.
One version of the model also regresses fees on credit score variables. Credit scoring has become a standard, although far from definitive, method of summarizing and evaluating the large amounts of data in a borrower’s credit record. Lenders differ in their loan standards, and some are more willing to work with lower-scoring applicants than others. Borrowers with bad credit presumably represent more work for the lender and a higher risk that the loan will not pass muster with the FHA. It seems plausible that higher-risk borrowers might be sorted with higher-cost lenders because the lower-cost lenders would screen out applicants with poorer credit so they can remain competitive in serving higher credit, cheaper-to-serve borrowers.
In the regressions based on the model summarized above, lending and title fees go up about one penny per dollar of VTS, with a $1,200 intercept indicating a substantial fixed cost unrelated to the transaction scale. Coefficient estimates for New Jersey and Iowa are consistent with previous findings but do not reach statistical significance.
Fees for new home sales average about $400 less than fees for sales of existing homes, all things being equal. Thus, the current institutional arrangements for property transfer amount to a differentially higher tax on existing homes—one might say an unplanned suburbanization policy.
Sale by a troubled seller leads, on average, to another $1,000 to $1,100 of fees, presumably at the seller’s own expense. It is difficult to understand how marital or property tax troubles could so inflate title costs in this
FHA borrowers do not appear to receive any relief in fees when they borrow at above-market rates because the coefficient on the presence of a service release premium is insignificantly positive. (Note that lack of disclosure of such spreads by mortgage bankers makes an insignificant result more probable.) It is suggestive that the coefficient takes a relatively high positive value in these circumstances rather than the negative value that indirect compensation would dictate. It seems the premium must reflect either exploitation of the buyer’s ignorance of the market or an urgent need on the buyer’s part for some unmeasured characteristic of performance, such as speed, by service providers.
The pattern of coefficient signs for the credit variables is roughly consistent with the notion of sorting among lenders suggested above, with high-scoring borrowers paying a bit less, low-scoring borrowers a bit more, and borrowers with no scores quite a bit more in fees. None of these coefficients, however, reaches statistical significance.
Some Tentative Conclusions
At first glance, lending and title fees seem to be appropriate targets for regulation. The federal government created and, through a variety of means, maintains the long-term amortizing home mortgage market. The services for which fees are paid are often federally mandated, and title services are intended to verify that state law on the transfer of property is satisfied. Thus, the federal and state governments require that lending and title services should be performed, and these governments have some responsibility for the orderly functioning of the market for services and the underlying market for housing.
This argument does not necessarily support the current RESPA regime, a form of sunshine regulation implicitly founded on the proposition that the only problem in the market is consumer ignorance, to be solved by federal action. Consumer ignorance might not be the only problem in the market, and nonfederal action might be preferable.
Sunshine regulation raises costs and may or may not benefit consumers, but from the small sample analyzed in this article, only tentative conclusions about those benefits are possible. Admittedly, the topic deserves a much deeper research effort than this one, but the research conducted to date suggests the
• Title and lending fees are worth regulating, if regulation can be efficient.
• The GFE is a reasonably good guide to fees for most people, but, for some borrowers, realized fees are much higher than the estimates. If any form of regulation is needed, its benefits might well reside primarily in the protection of this latter group.
• State action may strongly influence title fees. Actions to improve the clarity, simplicity, and accessibility of title records could lower fees in many locations.
• Disclosure may not make the market fair, in that buyers’ and sellers’ characteristics seem to lead to differences in fees for transactions of equal size. Lending and title fees paid for new home sales are notably lower than fees for existing homes, presumably because builders and developers can capture some economies of scale. To the extent that section 8 of RESPA inhibits lenders from realizing such economies and passing them on to consumers, this inhibition represents a previously unrecognized distortion in the housing market. It lowers the prices of new (mostly suburban) homes relative to existing residences.
Buyers’ and sellers’ characteristics of other sorts also affect market outcomes. Transactions in which the seller pays off an ex-spouse or is seriously delinquent in property taxes seem to generate much higher fees. There does not appear to be any reduction in fees in transactions in which the mortgage broker receives a “service release” or “above-par premium” from the ultimate lender for obtaining an above-market interest rate; in a fair market there probably would be a fee reduction.
This author first suggested a sunset clause for RESPA 10 years ago. Mark Shroder, Issues in Settlement Regulation: RESPA at 23, Ill. Real Est. Letter 10, 13 (Spring 1997), available at www.