Federal Deposit
Insurance Corporation

From: charles warren 
Sent: Friday, November 21, 2008 2:08 PM
To: regs.comments@occ.tres.gov; regs.comments@federalreserve.gov; Comments;
regs.comments@ots.tres.gov; regcomments@ncua.gov
Cc: sf.nancy@mail.house.gov; president
Subject: proposed interagency appraisal and evaluation guidelines - comment

Among rule based systems there are two themes. One is supervisory; rule 
makers make and enforce rules. The other is adversarial. Contestants 
agree on rules which may be enforced to insure fair or equal play. The 
first is weak because there are almost always more players who have an 
interest in evasion than there are referees to catch them. The second is 
robust because there are usually about equal interests at stake.

An example of the first is arms control. Saddam Hussein exploited the 
limitations of even the most stringent arms control regime. An example 
of the second is the Geneva Accord. Few countries flout the Geneva 
Accord because it is in their interest to see it applies to them as well 
as their adversaries.

Let's take a less fraught pair of examples relating to sport. Sailboat 
racing has two sets of rules. One is related to handicapping boats 
depending on their different design characteristics. The second is 
related to conduct on the race course. The first is supervisory and the 
second adversarial.

Sailboats are faster or slower depending on a number of factors, 
principal of which are waterline length, weight, power and stability. 
Since I have been involved in racing boats there have been a succession 
of clever rules which were devised by brilliant people with the intent 
of making a formula that reliably handicapped the differences between 
different boats. They have all degenerated into encouraging the 
construction of rule-based boats. In other words, instead of equalizing 
the scoring of a disparate fleet of boats, they have resulted in 
construction of a more or less homogenous fleet optimized to the rule. 
There are lots of smart naval architects and racers with the money to 
encourage them to try to outsmart the rule. As the weaknesses of the 
rule usually lead to odd boats, one rule succeeds another. The 
International Offshore Rule was thoroughly discredited by the disastrous 
losses of boats designed to it in the Fastnet Race of 1979. Hope 
triumphs over experience, however, and it has had several successors.

Conduct on the race course is obviously adversarial. It is governed by 
published rules which are internationally recognized. They start with 
basic sorts of things. In general boats to your right have the right of 
way. You must avoid collision. If you don't, beside possibly being 
liable for repair costs, you will very probably be disqualified from the 
race. This rule regime is pretty stable. Every five years the rule 
making committee takes a look at how they've worked and makes some small 
amendments. They also work pretty well in practice. Any contestant may 
protest any other for an infringement. The decision is made speedily, 
usually the night after the event, and locally, at the race-sponsoring 
yacht club, by a committee of peers, members of a knowledgeable race 
committee. The immediacy and certainty of the process as well as the 
high probability of its outcome yields a high level of compliance. Even 
at high profile events there are usually few protests. There is an 
appeal process, too. Appeals are read carefully both by serious 
competitors and by the racing rules committee in its periodic rule 
revision process.

The tenet that I draw from these examples is that transparent 
adversarial rules are robust. My suggestions that relate to the proposed 
OCC, OTS, FDIC, etal. appraisal rules are made with that in mind.

Within the context of lending, appraisal is part of the underwriting 
process. Its role as a cross-check has been degraded as the system has 
had no room for cross-checks. The way to make money was to make loans. 
Not making loans was not making money. In the event, because everybody 
acted that way, now everybody is losing incredible amounts of money. 
That was predictable.

Now, specifically, the proposed rule only "allows" the sale contract to 
be shared with the appraiser. Let us suppose that the lender chooses not 
to, but orally informs the appraiser of a sale price. There have been 
sales in the last cycle in which as much as 15% of the nominal price was 
immediately rebated to the buyer. Presumably that rebate would be in the 
contract. How likely would it be that the lender would report it orally 
at the same time as the nominal gross price?

The assumption here is that purchase appraisals are a large problem. Of 
course they may be. But refinance and construction appraisals have 
absolutely no direct reference to the market. Purchases do, the buyer 
and the price. In general buyers are not stupid. Unless there is some 
tangible incentive to do otherwise they will pay a fair price. In home 
resales my experience is that is true over 90% of the time. Allowing 
lenders to deprive appraisers of that source of information is not going 
to improve the underwriting process.

It is a good idea to make appraiser independence a major goal of the 
rule. There are probably two simpler ways to do that. The first is to 
put appraisal in the audit department of the institution rather than 
loan origination. The second is to require selection of contract 
appraisers by a random or sequential process from licensed appraisers in 
a geographic area.

The assumption behind the proposed rule is that taking the appraisal 
quality decision to the highest level will improve the result. This is 
not universally so. Had a memorable termination interview in which the 
corporate head of underwriting informed me that my job as reviewer was 
to rubber stamp all appraisals as acceptable. The quality control 
decision was his. This interview was held in the presence of the 
corporate counsel who almost apologetically confirmed that policy was 
consistent with the wishes of the president-CEO. Maybe the result would 
have been the same if I had been responsible to the chief auditor.

But the follow up on that internal determination is to homogenize the 
contract appraisals so that none undermine the generation of fee income. 
Another memorable conversation I had was with a review appraiser of an 
ex-client. It turned out that I had really angered one of their 
borrowers and that was why I wasn't getting any more work from them. If 
appraiser selection was either sequential by license number or random in 
a geographic area, that sort of homogenization would be difficult if not 
imposssible.

Additional standards in excess of licensure is a dangerous concept. Yes, 
achieving, for instance, senior membership in an appraisal organization 
is worthy and hopefully represents an advance in knowledge and general 
expertise over licensure. Similarly, course work related to specialized 
topics may improve suitability for some assignments. But unless such 
standards are explicit and externally verifiable, others, such as the 
willingness and ability to please the client, may also weigh in the 
balance to the detriment of quality.

Here's a little one. Photos of the property securing the loan are now 
available from several online databases. The moment of time that these 
images capture is often poorly defined. If photos are going to be any 
sort of assistance to evaluating collateral, there has to be some 
assurance that they represent the property at a time contemporary with 
that decision.

Automated Valuation Models (AVMs) are another slippery problem. They 
have proved out for property taxation. But property taxation is an 
adversarial game. Using AVMs for lending is going to be a supervisorial 
game. Nobody makes money by saying no. AVMs will allow the proliferation 
of bad decisions faster than they can be detected let alone suppressed 
by a limited number of supervisors. This assumes that the supervisors 
will be as technically qualified as those who will seek to evade 
supervision. Imagine pitting a GS13 with an MBA against a team of math 
PhDs with bonus plans. Remember the Unibomber is a math PhD.

Assessed value is a stronger reed. Assessment is adversarial. The 
assessor wants an accurate value. The property owner wants a low one. 
Presumably the local jurisdiction wants a high one, but can't usually 
explicitly say so. With the exception of states like California, those 
interests come out with a pretty well equalized result. Yes, the 
relationship of assessed value to market value will vary for many 
reasons. But it may not be necessary for the lending institution to 
evaluate that. It is probably evaluated at the state level to oversee 
and supervise assessment generally. It may also be a matter of public 
record at the local level. If the institution is to be tasked with 
relating assessed value to market value, the rules better be explicit, 
and the measures for doing so better be related to publicly available 
information.

This is a bit outside the scope of property appraisal, but reference to 
rating agencies to determine the quality and value of real estate based 
securities seems a bit quaint. Perhaps verification that there is a 
market maker, a liquid market and the price of the securities might be 
more to the point.

Sincerely,

Charles B. Warren, ASA
urban real property
Valuation and Consultation
San Francisco
http://www.charlesbwarren.com
415.433.0959