October 19, 2002
On July 29, 2002, HUD released a set of
proposals to substantially change the ways in which home loans are originated in
the US. As usual, the proposals were open for comment, and many thousands
of them were received. Mine was among them, and is shown below with minor
For the most part, HUD's proposals are very
well thought out, and would benefit consumers in every segment of the home loan
market. They would especially benefit less-sophisticated borrowers, who most
The first proposal would change the way in
which the compensation of mortgage brokers is reported. The objective is to make
a broker�s total compensation transparent to borrowers.
The major concern underlying the proposals for revealing broker compensation is
"rebate abuse" � the practice of steering unwary borrowers into
high-rate loans on which they should receive a rebate from the lender but don�t.
A rebate is negative points. Points are an upfront charge to the borrower
expressed as a percent of the loan, and a rebate is an upfront credit to the
borrower from the lender. Rebates are used to cover settlement costs.
For example, the loan officer�s price sheet
shows 6% at zero points, 5.75% at 2 points, and 6.25% at a 2-point rebate. If
the borrower is willing to pay 6.25% without argument, the rebate is retained by
the loan provider, which could be a mortgage broker or a lender. Most of the
attention has been directed at rebate abuse by brokers.
Lenders working through mortgage brokers
(called "wholesale lenders") transmit their price information to
brokers, not to borrowers. With few exceptions, borrowers are not privy to this
information. Borrowers are quoted prices by brokers that include the broker�s
For example, the lender�s quote to the
broker is 6% plus a rebate of 1%, and the broker�s quote to the borrower is 6%
plus a broker fee of .5%. The broker�s total compensation is 1.5%, 1% from the
lender rebate and .5% from the borrower fee, but the borrower does not know
about the 1% rebate unless the broker tells him. By the time borrowers become
aware of rebates retained by brokers, they are often too far along in the
transaction to back out.
Rebates collected by mortgage brokers are
known in the trade as "yield spread premiums" -- terminology that is
designed to obscure rather than reveal.
Rebate abuse is not practiced by Upfront
Mortgage Brokers (UMBs). UMBs set their total compensation in advance, revealing
and passing through the price quoted by the lender. But there are only 47 of
Under the proposed new rules, rebates would be reported on the GFE as a payment
by the lender to the borrower. The borrower would have to authorize the rebate
to be paid to the broker � as if it was coming directly out of the borrower�s
pocket. In effect, all brokers would become UMBs. The result will be greater
borrower resistance to rebate abuse by brokers.
It is not surprising that most brokers are against it, claiming an injustice on
two grounds. First, they ask why, of all the many intermediaries in our economy,
they alone should be required to disclose their wholesale prices? Grocers, for
example, don�t disclose the cost to them of the produce they sell, and nobody
Brokers would have a point if borrowers could
comparison shop for home loans as easily as they can shop for groceries. In such
case, they would have no need to know broker compensation in order to protect
themselves. Broker compensation is relevant only because many borrowers are
intimidated by the difficulties of shopping for a home loan, and place
themselves in the hands of a broker.
The brokers� second argument, that no
comparable disclosure rule is proposed for retail lenders, has more force. The
same disclosure rule should be applied to all loan providers who receive
rebates, which includes some smaller lenders who mark up the prices quoted to
them by wholesale lenders in the same way as brokers. But even if HUD extended
its disclosure rule to cover these lenders, it would leave the major retail
The good news is that HUD has a proposal for
disclosure of "interest rate dependent payments" which, if properly
implemented, would curb rebate abuse by lenders. This will be discussed below.
REVISING THE GFE
HUD�s proposal to make borrower payments to
mortgage brokers transparent is one part of a broader plan to make the Good
Faith Estimate (GFE), on which lenders and mortgage brokers disclose settlement
costs, more useful to borrowers as a shopping tool.
The GFE Now:
The existing GFE lists each individual settlement charge, which encourages
borrowers to focus on individual charges. I continually receive letters from
borrowers asking, e.g., "What is such and such a charge for?" "Is
$400 reasonable for this charge?" "Is it negotiable?" Such
questions distract them from what should be their major focus, which is the
total of shoppable settlement charges.
The GFE is also open-ended, inviting lenders
to add new charges, which some do. In addition, the existing GFE makes no
distinction between charges that lenders can, and those they cannot, control.
All are "estimates" subject to change. And they often are changed,
after borrowers pass the point of no return, and always to the borrower�s
Consolidation and Commitment:
Under the proposed GFE, settlement costs are consolidated into several major
groups, and only the total is reported for each group. The new GFE also limits
the extent to which the costs may change. These limits are different for
services controlled by the lender and services for which the borrower may shop
Settlement charges that cannot be changed
"except in unforeseeable and extraordinary circumstances..." include
"origination charges", which are all charges by the lender and broker;
"lender required and selected third party services", which are
all third party services required by the lender (such as appraisals) where the
lender selects the service provider; and "title services and title
insurance" when selected by the lender. Lenders will no longer be able to
manipulate these charges with impunity.
Charges for services required by the lender
for which borrowers can shop third-party providers, including title-related
services, can vary up to 10%. The same leeway applies to reserves for escrows.
Other settlement charges can vary as
circumstances dictate because they are not subject to lender manipulation. These
include hazard insurance and owners title insurance, which borrowers always
select; and per diem interest -- interest for the period between the closing
date and the first day of the following month -- which is determined by formula.
Interest Rate Dependent Payments (IRDP):
Another important feature of the proposed GFE is a table that shows the interest
rate and points selected by the borrower, and two other rate/point combinations
available to him. (Points are an upfront charge expressed as a percent of the
loan). The purpose is to let borrowers know their options. One such combination
is a higher rate with lower points, the other is a lower rate with higher
points. HUD is not clear, however, on how loan providers select these rate/point
The selection should aim at protecting
borrowers from rebate abuse by lenders. Borrowers are already protected
against rebate abuse from mortgage brokers by the new rules for broker
disclosure, discussed above. Those rules, however, do not protect borrowers
against rebate abuse by lenders who don�t reveal their rebates.
The most effective way to curb rebate abuse
by lenders would be to have the first column of the IRDP table show the interest
rate that is closest to zero points, the second and third columns show rates
that are plus and minus �% from the rate in the first column, and the fourth
column show the combination selected by the borrower. This would put borrowers
on notice that they should receive compensation for a higher rate. It would also
lead to consistent treatment among loan providers, making it easier to shop.
Interest Rate on the GFE:
The proposed GFE also requires loan providers to show the interest rate,
mortgage insurance and APR. In contrast to the settlement costs, however, the
new GFE says nothing about the extent to which the loan provider is committed to
these terms. Hence, it is not clear whether the new GFE would curb the most
pervasive abuse in the home loan market: "float abuse".
Assume that after shopping prices at several lenders, Jane Doe selects lender X
and submits an application. The prices quoted by X, upon which Jane based her
decision, "float" with the market until they are locked by the lender.
Floating is mandatory between the initial
price quote and the time when the lender is willing to lock. This period can
range from a day to several weeks or longer, depending on the lender�s
requirements to lock, and on how long it takes Jane to comply.
Jane might elect to wait, even after the
lender is willing to lock, if she believes that interest rates will fall. This
begins a voluntary float period, which can run until a few days before the
At the end of the float period, lender X is
supposed to lock at the market price. But since the market price is what X says
it is, the process is extremely vulnerable to abuse.
In principle, lender X should lock at the
price that X would quote to Jane�s identical twin if the twin walked through
the door on the lock date as a new customer shopping the exact same deal. In
practice, Jane may get a higher price than her twin because Jane is at least
partially committed while her twin is only shopping.
Float abuse is a violation of the twin
sibling principle. It consists of understating the decline in interest rates
that occurred during the float period, or overstating the rise. For example, a
loan provider instructs its loan officers to deduct one point from the market
price quoted to a shopper, and add a point to the market price on the lock day.
Float abuse is pervasive, practiced by mortgage brokers as well as lenders, and
Using the GFE to Curb Float Abuse:
The new GFE eliminates the incentive for brokers to practice float abuse. Since
the new GFE fixes the mortgage broker�s fee to whatever the borrower has
agreed to pay, the broker cannot profit by overstating the price on the lock
day. All that would do is benefit the lender at the expense of the borrower.
Under HUD�s current proposal, however,
float abuse by lenders is not curbed by the GFE at all. Lenders can adjust the
interest rate shown on the GFE to the "market rate" on the lock date
with the same impunity they enjoy now. That is disappointing, but hopefully HUD
will remedy it.
HUD need not, and should not set rules that
limit the ability of lenders to adjust the rate set on the GFE. That could be a
disaster. But with little risk HUD could require lenders to show how the rate
will be determined on the lock day, and leave it for competition to do the rest.
The requirement should be divided into two
categories: "conditions", and "market adjustment."
Conditions are future events that must be fulfilled for the rate to remain
valid. The fewer and more reasonable the conditions, the more attractive is the
rate on the GFE. For example, borrowers will prefer a rate subject to "an
appraisal of $300,000 or more" to one subject to a "satisfactory
In a similar way, borrowers will prefer
lenders who provide an objective procedure for implementing the twin sibling
principle. Perhaps the best would be a web-based pricing program on which
borrowers can price their own deal on any day. A low-tech equivalent would be to
identify the borrower�s price niche on the lender�s daily price sheet,
repeating the process on the lock date. Lenders who say "Trust us to give
you the market rate on the day you lock", will not last long.
MORTGAGE PRICE AGREEMENT (GMPA)
The most far-reaching of HUD�s proposals is
to authorize lenders and others to offer borrowers complete (or almost complete)
packages of loans and settlement services at a single price. This is permissive
rather than obligatory. Lenders who package would use a Guaranteed Mortgage
Price Agreement (GMPA) in lieu of the proposed new obligatory GFE, which I
The major purpose of the GMP is to drive down settlement costs. Under existing
arrangements, competition in the markets for settlement services is
"perverse" -- it tends to drive up prices, or prevent them from
falling in response to deployment of more efficient technology. Perverse
competition arises whenever one party selects the seller of the service and
another party pays for it.
For example, lenders select the mortgage
insurer but borrowers pay the premiums. Instead of competing for customers by
lowering prices and improving service, service providers compete for the favor
of the lenders.
While direct "kickbacks" to lenders
for the referral of business are illegal, mortgage insurers and others have
found legal ways to accomplish the same thing. These include the provision of
services to lenders at favorable prices, or the sharing of insurance premiums
with reinsurance affiliates owned by lenders. Such practices increase the costs
of service providers and keep the prices charged to borrowers from falling.
GMPs could be offered by lenders or other entities such as real estate companies
or title insurers. A package must include a loan at a guaranteed interest rate
plus a guaranteed dollar price for all settlement services excepting per diem
interest, hazard insurance, and escrows. Packagers can deal freely with their
own affiliates and are exempt from kickback prohibition rules.
The rationale is that competition among
packagers will force down the prices they pay for services. Competition will be
effective because the price of a package will consist solely of the interest
rate plus a single dollar price for all settlement services. This will make it
easy for borrowers to shop and compare. It won�t work, however, if borrowers
remain vulnerable to float abuse -- the practice of understating the rate when
quoting to shoppers, and overstating it on the day the rate is locked.
HUD is focused on preventing float abuse with the GMP. It proposes a rule that
the interest rate can change between the quote date and the lock date only
"based on observable market changes, or based on other data or appropriate
means to ensure the guarantee". This is deliberately vague as HUD is
looking for further guidance before trying to pin it down.
The way to pin it down is to establish the
twin sibling principle I noted above. This says that the packager must lock at
the rate that it would quote to the borrower�s identical twin if the twin
walked through the door on the lock date as a new customer shopping the exact
The twin sibling principle allows the lender
full leeway to adjust to general changes in the market, and also to narrower
changes applying to certain types of borrowers or transactions. What this
principle does not allow is the lender to lock at a higher rate solely because
the borrower is too far committed to back out.
There are at least half a dozen methods
packagers could use to comply with the twin sibling principle, some better than
others from a borrower�s perspective. Nonetheless, it would be a mistake for
HUD to define the methods that are "acceptable". It would be far
better simply to require the packager to explain the method that is used.
Competitive pressures will immediately favor the packagers who provide the best
There are other problems. Among the prices
that will be consolidated are mortgage broker fees, mortgage insurance premiums,
and points. All are problematic.
Mortgage Broker Fees:
Under HUD�s proposal, mortgage broker fees are included in the GMP price.
Hence, a broker involved in helping a borrower select from among competing GMPs
must accept whatever broker fee each packager has decided on in advance.
Brokers, however, sell an individualized
service to borrowers, and the broker�s investment of time varies from loan to
loan. The two parties should be free to negotiate the price of the broker�s
service. If prices are preset by packagers, brokers may opt not to play.
The competition that HUD is looking to drive
down settlement costs will not work as effectively if brokers are not available
to help borrowers assess competing packages. If brokers don�t play,
furthermore, it reduces the ranks of potential packagers. The only practical way
for most non-lender institutions to become packagers is to work through mortgage
HUD should explicitly recognize a distinction
between retail and wholesale packages. A wholesale package would be one offered
through mortgage brokers. The GMP agreement used by wholesale packagers would
include a slot for a broker fee, which would allow brokers to negotiate their
fee with borrowers.
The broker�s fee would be transparent, just
as it is in the proposed new GFE. By adding the broker fee to the lender�s
wholesale package price, borrowers could easily compare a package obtained
through a broker with one offered directly by a packager.
NOTE: I discovered after writing the above
that HUD was assuming that brokers would themselves become packagers, in which
capacity they could add any fee for themselves they wished. But whether or not
this will happen is very unclear. And even if some brokers can become successful
packagers, others will prefer to work as package assemblers, which is closer to
their traditional function.
Mortgage Insurance Premiums: Mortgage
insurance paid for with a financed single premium (the premium is included in
the loan) fits neatly into HUD�s single price scheme. For example, a single
premium of 2.35% on a $100,000 loan would amount to $2350, which would be added
to total settlement costs.
Premiums paid monthly, however, don�t fit.
For example, the monthly premium on the same loan would be .39%/12, or $32.50 a
month. If monthly premiums are permitted in the package, the GMP will have to
include the premium as a separate price -- as is the practice now.
HUD could permit only single premiums. This
has the additional merit that single premiums are less costly to borrowers than
monthly premiums. For example, if the interest rate on the loan with a single
premium of $2350 is 8%, the monthly payment would increase by $17.25, of which
$15.66 is additional interest that is tax deductible. None of the $32.50 monthly
premium is deductible. [Note: Why single-premium mortgage insurance is not used
more widely is discussed in What
Must Be Disclosed About PMI].
The alternative is to allow monthly premiums,
and prompt borrowers to use the APR in selecting among GMPs. The APR is a single
measure of credit cost that takes account of mortgage insurance premiums,
whether paid upfront or monthly. The case for this approach is strengthened by
the fact that consolidating points also would cause serious problems.
would consolidate points -- upfront credit charges expressed as a percent of the
loan -- in total settlement charges. This would eliminate the current practice
of quoting interest rates in increments of 1/8%, and using points to adjust to
small changes in the market. If points are frozen in the package price, the 1/8%
convention will be discarded and lenders will quote rates to 3 odd decimals,
such as 6.274%. That isn�t a disaster, just an inconvenience.
More serious is that borrowers must know the
points because points are tax deductible. Indeed, they should know the points
early on because taxes can figure importantly in selection decisions. All the
calculators on my web site designed to help borrowers make the right selections
take account of taxes.
If points are not consolidated, each GMP
would have an interest rate, points, and other settlement costs. On some GMPs,
monthly mortgage insurance premiums would be a fourth price. And if real estate
services are broken out from lender services, as proposed below, it would be yet
another price. Nonetheless, this would be far more manageable than the 20 to 30
separate cost items the typical home purchaser must deal with today.
In sum, HUD has taken consolidation too far
-- some fees should be separately identified. These include broker fees, because
borrowers should be able to negotiate them separately with brokers; points,
which borrowers need to know for tax and other reasons; and monthly mortgage
insurance premiums, which can�t be added to other settlement charges because
they are paid over time.
exceptions to consolidation highlight the need for a single measure of cost to
the borrower, which pulls together all the separate price components. In
principle, such a measure is already available to borrowers. It is called the
"Annual Percentage Rate", or APR, and it is a part of Truth in Lending
(TIL) disclosures, which are administered by the Federal Reserve. Until now,
however, HUD, has never included an APR, or anything like it, in its required
HUD includes an APR in its proposed new GMP,
and also in its proposed revision to the GFE. In addition, HUD would include
everything else of value on the TIL, such as prepayment penalties, while leaving
behind much junk. It is reasonable to conclude that HUD is looking to displace
the Federal Reserve in mortgage loan disclosure, although this would have to be
sanctioned by Congress.
Divided responsibility between the HUD and
the Fed has worked very poorly. For example, it has never been possible to
reconcile the figures on the TIL with those on the GFE. If Congress centralized
responsibility in HUD, it could be a blessing, but only if HUD redesigns the APR
so that it fits into the new scheme.
The APR as it is currently defined is close
to worthless and few borrowers use it. For one thing, there is no rhyme or
reason to the settlement cost items that are included. For example, the current
APR does not include the cost of a credit report, appraisal, or lender
inspection, all of which are included in both the GMP and the proposed GFE
(where they are consolidated into Lender Required and Third Party Services). On
the other hand, per diem interest is included in the APR but not consolidated in
the GMP or GFE.
If the cost items covered by the APR don�t
correspond exactly with the cost categories HUD is using in the GMP and GFE, the
borrower will be confused and the APR will be worthless. On the other hand,
changing the definition while Truth in Lending is still in force would mean that
there would be a Fed APR and a HUD APR, which might be even more confusing. If
HUD can�t find a way to get rid of the Fed APR, it should consider renaming
its measure � perhaps to "true interest cost" or TIC.
HUD also needs to fix the other major
weakness of the Fed�s APR, which is the assumption that all loans run to term.
In fact, very few loans run to term, and an increasing proportion are paid off
within 5 years. The APR is a treacherous guide for such borrowers.
For example, John Doe in borrowing $300,000
for 30 years is choosing between 6.5% with zero APR fees and 6% with $12,000 in
fees. The APR on the 6.5% loan is 6.5%, regardless of when the loan is paid off.
On the second loan, the Fed�s APR calculated over 30 years is 6.39%,
suggesting that this is the better choice. But an APR calculated over 5 years is
6.98%, leading to the opposite conclusion.
By far the best way to set the time period
over which the APR is calculated is to leave it up to the individual borrower.
Lenders and brokers will resist this, although modern mortgage technology makes
it extremely simple to do. A reasonable compromise would be to require two TIC�s,
one at term and one at 5 years. This would be an immense step forward.
One Package Versus Two: HUD�s
proposal to consolidate all settlement services into a GMP would be a
substantial break with the ways in which mortgages are delivered today, and how
it will work out exactly is not clear.
Lenders expect to profit from having the
complete freedom to package the services of their own affiliates, without having
to warn borrowers about the relationship or offer them alternatives. Lenders
without affiliates would be free to use their buying clout to negotiate the most
favorable terms with independent third-party service providers. HUD is gambling
that competition between GMP packagers will force them to pass on most of the
cost savings to borrowers.
The way to make this less of a gamble, closer
to a sure thing, is to expand the number of players, and there is a very simple
way to do this. Instead of requiring one package that includes everything, HUD
could allow two packages, which could be offered separately or together.
One package, offered only by lenders, would
consist of lender-related services. These are services provided directly by the
lender, or by third parties, such as independent appraisers, at prices known by
the lender. This package would have the same rate guarantee as the GMP, but the
price guarantee would cover only lender-related services.
The second package, offered by title
insurance, real estate or other non-lender firms, would consist of all real
estate-related services. These are all the services needed in settlement of the
real estate transaction. The price of real estate-related services would also be
guaranteed. Both groups of packagers would have the same type of exemptions
(from restrictions on referral fees, for example) as GMP packagers.
This breakdown corresponds to a natural
division of labor between lenders and real estate service providers. Lenders by
themselves can�t guarantee the prices of real estate services because they are
not themselves involved in that process. Lenders offering GMPs, therefore, will
subcontract with firms involved in real estate to provide real estate service
packages for inclusion in GMPs. There is no good reason why these firms should
not be able to offer the same packages to borrowers, directly or through
The two-package approach would materially
increase the number of competitive options available to borrowers. Many lenders
reluctant to offer GMPs will be willing to offer lender packages because, except
for the guarantee, it is what they do now. A few forward-looking lenders already
guarantee their settlement service package. Real estate firms that are
disinclined to become subcontractors to lenders offering GMPs will take
advantage of the opportunity to develop their own distribution networks.
Under the two-package approach, borrowers
could buy a complete GMP, or they could buy separate lender and real estate
packages. It would be a simple matter to compare the price of a GMP package to
the sum of the prices on a lender package and a real estate package.
Indeed, mortgage brokers would do the
arithmetic for them. Putting the two packages together is a natural mortgage
broker function. It is similar to what they often do now in combining a first
and second mortgage in one deal. The lender package should include a slot for
the broker�s fee, which would be negotiated with the borrower and fully
Copyright Jack Guttentag 2003