by Tanta on 4/09/2008 11:07:00 AM

The Center for Responsible Lending released a fascinating paper yesterday, "Steered Wrong: Brokers, Borrowers, and Subprime Loans." Using a nationwide database of prime and subprime loans, both retail and brokered, originated in the 2004-2006 period, the authors "paired" retail and brokered loan transactions with similar risk characteristics (such as FICO, LTV, DTI, and loan amount, among others), to compare interest costs over a one-year, four-year, and life of loan (30-year) horizon. The results are no doubt stunning as well as surprising to a certain reliably-stunned-and-surprised contingent of the regulatory and securitization enclaves:

We find significant differences between broker and lender pricing on home loans, primarily on mortgages originated for borrowers with weaker credit histories. During the first year of the loan, borrowers with credit profiles in the subprime range pay statistically more for brokered loans than they would have if they had obtained their loan directly from a lender. Over a four-year period, a typical subprime borrower pays over $5,000 more, and over the 30-year life of the loan, the cost gap grows to almost $36,000. . . .

Significant disparities exist between broker and lender pricing. After matching loans on objective factors that affect interest rates, the analysis reveals that interest payments were significantly higher on broker-originated mortgages in the majority of risk categories we examined. Disparities are greatest for subprime borrowers. For people with weaker credit, brokers consistently charged higher interest rates than retail lenders. A typical subprime borrower was slated to pay $5,222 more during the first four years of a $166,000 mortgage compared to a similar borrower who received a loan directly from a lender. Over thirty years, this borrower would pay $35,874 more in interest payments, equivalent to an interest rate approximately 1.3 percentage points higher than a similar borrower with a retail loan.

Cost disparities grow greater after initial years. For subprime borrowers, significant disparities are apparent even during the first year of the loan. However, because so many subprime mortgages come with short-term introductory rates that rise substantially when they adjust, the cost disparities become more pronounced after the first four years of a loan. Prime borrowers generally do not pay more for brokered loans. In general, people with higher credit scores—those who received prime loans—did not pay higher interest on broker-originated loans. In fact, some borrowers with very high credit scores who received loans from brokers achieved modest savings, although long-term savings were largely limited to fixed-rate loans.

The study zeroes in on two basic "market mechanisms" in play: the fact that borrowers with weaker credit are less informed about competitive rates and less able to analyze the often complex products they are offered than prime borrowers, and that brokers are more inclined to raise "profit margins" on low-volume business (subprime) and maintain thinner margins on high-volume business (prime). What I find most refreshing about the approach is that it mostly dismisses the issue of "better disclosures" as a regulatory side-track. That's a long-time hobbyhorse of mine, too.

Given the prevalence of brokers in today’s market and their impact on the ability of borrowers to build and maintain home equity, we propose specific policy recommendations below.

First, yield spread premiums and prepayment penalties should be banned on subprime mortgages.

Second, lenders should be more accountable for the actions of mortgage brokers originating loans in their names and investors should share responsibility. Third, mortgage brokers should have a
fiduciary responsibility to the borrowers they serve. . . . Notably, our proposals do not focus on increased disclosures. While improved disclosures and increased financial literacy are laudable goals, the magnitude of the problems identified in this paper indicate the need for a direct and immediate response. Moreover, experience and findings from behavioral economics as well as findings related to limited financial literacy among consumers suggest that improved disclosures are likely to have limited effect, at best.

I strongly recommend the entire paper to those of you with an interest in the matter. I want to suggest something that the authors of this paper do not suggest, but which I think cries out for further study.

The CRL paper does comment on the possible causes for retail lenders' pricing practices in regard to subprime loans:

People with weaker credit scores naturally pay more for mortgages than people with strong scores. However, it is very difficult for borrowers with weaker credit or less experience in financial matters to know precisely how much more is appropriate, especially since, unlike prime rates, subprime rates are not generally publicly available. In addition, subprime loans tend to be much more complex than the fixed-rate mortgages that have long dominated the prime market, making their costs more difficult for borrowers to compare. Accordingly, we hypothesize that brokers have been able to take advantage of this situation by
emphasizing maximum revenues per loan for subprime borrowers. While retail lenders are probably not immune from these dynamics, we believe the effects on the costs of retail loans are less pronounced due to more regulation, better internal controls, and concerns about reputational risk.

Depository retail loan pricing has been subject to "fair lending" as well as safety and soundness examination--not just regulation--for quite some time. What I doubt most consumers and observers realize is that since at least the early 90s, retail lenders used loan origination systems that "integrated" the lender's rate sheet and the loan application; the "rate sheet" is simply a printed version of rates and prices calculated by the system based on imported market price data, established tables of profit margins by branch, and "rules" or calculations that adjust rates for lock periods, "buy up" or "buy down" rates by calculating premium or discount prices for higher or lower rates, and apply various loan-level adjustments to rates or prices for individual loan characteristics. Importantly for our present purposes, these systems "know" when a loan is priced above or below the required rate sheet price--this is known as an "overage" or "underage," meaning that the price or points of the loan are more less than the required price. Not only do they "know" this, they also store this information, which can then be pulled into reports and analyzed for frequency, patterns, and possible discriminatory practices. There just aren't many "rogue loan officers" in a retail environment in which all loans must be "priced" with the computer, and in which a pattern of "overages" will result in a report on someone's desk sooner rather than later.

A practice that developed, formally or informally, in the 90s was often called the "point bank" or "price bank." Individual originators or branches were basically allowed to charge overage or underage on individual loans, as long as the monthly or quarterly pipeline "balanced" to zero or thereabouts. An overage or two (or three . . .) could be "banked" and used in essence to subsidize a concession or two on other loans.

Point banks, in my view, were complete disasters. A whole lot of lenders found--sometimes with the assistance of their examiners--that the concessions were being offered to, precisely, those "best customers" who could shop around or negotiate or quite possibly were just part of the loan officer's social or business network. (They were frequently RE agents, builders, or other mortgage market participants the loan officers wanted to "reward"). Meanwhile, the overages that paid for those concessions were being charged to the least informed borrowers with the weakest "bargaining position": weak credit borrowers, first time homebuyers, and small-balance loans (the latter of which don't pay handsome commissions in dollars, but are just as much work for a loan officer as a larger loan).

In other words, weaker and less informed (and possibly more financially prudent) borrowers were "subsidizing" the rates offered to the strongest, best informed, and quite possibly most debt-heavy borrowers. Insofar as the largest loans were most likely to get a concession, and the smallest loans most likely to be charged overage, it generally required several small loans with overage to each single large loan with a concession to make the "point bank" balance. It does not take long for this pattern to appear on the reports. There was a time, at least, when it didn't take long for regulators to issue some threats regarding this behavior. Retail originators basically lost interest in the "point bank" idea as corporate policy.

I bring all this up because brokers often argue--at least in the comment section of this blog, they do--that they offer highly competitive rates on prime business, indeed, often better rates than a retail lender would offer the same customer. The CRL study backs this up to a limited extent:

[B]orrowers with higher credit scores pay virtually no additional interest and occasionally pay modestly less interest when they receive their loan from a broker versus a retail lender, and borrowers with lower credit scores pay more interest. What’s more, the figures illustrate that the additional interest paid in brokered loans increases dramatically as credit scores decline. The figures also help clarify that the results vary considerably between loans that would be considered prime and those that are subprime.

In essence, retail lenders charge slightly more for prime loans than brokers, and much less for subprime loans. (Note that this analysis is precisely attempting to control for individual risk factors by pairing comparable loans; the difference cannot therefore be due solely to brokers originating the lowest-quality end of the subprime spectrum more frequently than retail lenders.)

My suggestion is that we ask whether brokered mortgage lending is, in practice, running on the old "point bank" model that retail lenders largely abandoned, in which "strong" borrowers are basically subsidized by "weak borrowers." The defenders of "risk based pricing" have always argued that loan-level pricing adjustments are fundamentally fair, because "average" pricing would mean that the highest-quality borrowers were paying more to "subsidize" lowest-quality borrowers. Yet here we have a study that weeds out the "risk-based pricing effect," and that shows that in fact in brokered business the weakest borrowers seem to be paying more than their fair share of risk, with the savings passed through to those borrowers who can demand competitive rates, whether that is because of their credit profile or simply their social profile.

The CRL study touches on the question of the "brokered loan premium," which also needs further analysis and study, in my view:

Finally, we note that part of the explanation for higher rates could arguably arise as a consequence of expectations for worse loan performance among brokered loans. In other words, under this explanation, lenders who expect worse loan performance on brokered loans charge higher baseline rates to “price in” the added risk associated with a loan originated by a broker. Alexander et al offer an analysis of subprime loans consistent with this theory, but end up with a net effect of just half a percentage point.37 Moreover, they do not distinguish whether this increase is associated primarily with increased default risk (which they do detect) or yield spread premiums. Even if the entire amount were attributable to increased default risk, the reported half percentage point difference would be less than half of the 1.3 effective percentage point difference we report here. Moreover, since we control for the vast majority of risk factors, we believe our analysis already directly controls for much of this concern. In addition, to the extent that broker-origination itself is the risk factor that may cause higher delinquencies we question whether the risk factor could be related to the very additional expenses we identify here. In addition, Jackson and Burlingame do explicitly isolate yield spread premiums to identify differences.38 Consequently, the differences they report between brokered and retail pricing have their origin with the broker and cannot be the result of higher prices built in by a lender to account for anticipated worse loan performance.

This is a profoundly important question begging for sophisticated empirical work and analysis, since it gets to the very heart of the matter from a consumer standpoint. If there is such a thing as "brokered transaction risk" that is logically and analytically separate from "borrower default risk," that is to say that there is something about the brokered loan transaction or broker business model that results in higher defaults or (more likely) increased prepayments from churning that eat into lender profitability on loans, all other things being equal in terms of the underlying loan characteristics. The idea of a "brokered loan premium," then, would signal that wholesalers are pricing this in by charging higher rates/prices to brokered loans than to retail loans across all credit categories/loan types.

This immediately raises some troubling questions: If there is a "brokered loan premium," are consumers aware that they are paying higher credit costs by using a broker? If there isn't a "brokered loan premium," then the pricing differentials identified must be due solely to brokers' practices, not the wholesaler's policies. Do consumers understand, then, that by using brokers they are quite possibly subject to pricing practices that differ substantially, at least on the aggregate, from retail pricing practices?

We need to bear in mind that even twenty years ago, there were relatively few nationwide lenders, there was no internet, and there were substantial pockets of the country in which there simply wasn't an easily-accessed depository retail lender who offered a wide array of mortgage product, and specifically subprime credit. Brokers filled the market need in those areas by being able to access out-of-state or niche lenders on the borrowers' behalf. In such an environment, paying a "brokered loan premium" hardly seemed unfair to consumers.

Yet we are long since past such an environment, in my view. Insofar as a "brokered loan premium" exists in areas already saturated by Megabank branches on every corner offering mortgage credit all the way down the spectrum from prime to subprime, and insofar as the internet offers consumers access to direct lenders anywhere at any time, it is distressing to think that borrowers are still paying a premium for the services of an intermediary. This is particularly troubling given that consolidation in the wholesaler business means, in many markets, that brokers are all relying on the same dozen or so mega-lender rate sheets; I see little evidence that brokers are still able to find "inefficiencies" in the delivery of mortgage credit to underserved areas, or seek out small niche lenders to make them available to borrowers who could not otherwise find them--not when the big wholesalers are all offering "niche" products.

No doubt a lot of people will be troubled by the implication I am making that more than a few "good" borrowers got better rates than they probably should have, courtesy of those "bad" borrowers who paid not only their own "risk premium" but part of someone else's, as well. But I am more and more convinced that this has occurred at least in those markets that saw the explosive growth of subprime originations in the period in question. We are clearly seeing higher rates and prices paid by prime borrowers now, in the throes of the credit crunch, than we have in the past. A certain segment of the market (and nearly all the media) wishes to see this as an "overreaction" to "subprime problems," meaning that those most excellent borrowers are "paying too much" and that prices should return to "normal" once everyone realizes that the world hasn't ended. I have my doubts about that.

If, in fact, "normal" prime or "super prime" pricing for the last several years was subsidized by an unprecedented growth in subprime originations--for which there is certainly evidence of some "steering" of decent-quality loans into those high-rate subprime products--then "normal" pricing won't be back if we lose some 20% of the applicants who can be made to keep pricing "competitive" for those who know how to and can "compete." It seems to me, at least, that further research and analysis of this question is time much better spent than, say, tinkering with new disclosure forms.