The New USPAP: MARS Required

[RCA]

Don't know if I will renew my license this year or next, but I am taking the required courses, including the latest USPAP.
====
No, of course, the latest USPAP does not specifically require that appraisers use MARS regression.

But all that new verbiage on non-discrimination - and associated laws, including case law, there are some notable repercussions.

For example:

A: School Districts:

1. You have to be very careful making adjustments for school districts, i.e., you cannot mention quality or test scores. Yet appraisers know that, all things being equal, parents with elementary- or secondary-school-age children are very concerned about moving into a school district with high-quality schools, as evidenced by quality ratings and score levels.

2. What is allowed is objective adjustments that use ONLY school district boundaries. Your only option is regression. And with so many other variables involved, only MARS regression will give you the needed accuracy. You put the school district ID into the MARS regression, and if there is a difference for some school districts, it will generate an adjustment. No reason given, other than it is just Market Characteristics.

B. Neighborhood/Market Area Crime Rates

1. Again, this is an area that will leave you wondering whether you can do the right thing.
2. You can use MARS regression on MLS IDs or Zip Codes, along with their crime rates, if the data comes from objective, publicly sourced statistics. Your reasoning and support should focus solely on the area and its borders, without any mention of people.
3. You can have a policy that allows you to avoid servicing certain areas due to high crime statistics, such as "We don't service areas where the murder rate is > 10 per 100,000 residents for safety reasons." That would, for example, allow San Francisco (4.5 murders/100,000 residents, but exclude Oakland (16-36 murders/100,000 residents)
4. You pretty much have to avoid using protected characteristics in your reports. And whatever you do, it should make sense and only involve non-protected characteristics.

Note: The way you would do this is have a table of MLS IDs or zip codes, plus crime stats, then create a crime stat column in your listings spreadsheet for crime and enter the corresponding crime rate. In fact, you might have one for the murder rate and another for the theft rate.

C. The Difficulty
1. You can't use so-called "proxies" for protected characteristics. But what is or is not considered a proxy? - That's where case law (e.g., historical redlining litigation) and the actual application of regulations (e.g., State Board disciplinary cases) have an impact.

2. In doubt, ask ChatGPT, Grok or Claude.

 

[Fernando]

What a misleading thread title.

 

[George Hatch]

Most SFR appraisal assignments don't require neighborhood adjustments for locational factors relating to social factors. I realize that's just one example, but it also doesn't take a sophisticated analysis and massive amounts of data to parse for most other factors, either.

 

[RCA]

Most SFR appraisal assignments don't require neighborhood adjustments for locational factors relating to social factors. I realize that's just one example, but it also doesn't take a sophisticated analysis and massive amounts of data to parse for most other factors, either.

That's nonsense. I use MLS IDs for most counties, if I can get them (since some counties don't have them, though other options exist). There are in fact, usually adjustments for these different areas. But, of course, your comps should preferably come from the same MLS area, in which case there wouldn't be an adjustment. But it often happens that you have to extend your comparable area further out to get a decent set of comps. In my hometown of Pacifica, you have distinctly different areas. South Pacifica is much sunnier and warmer than North Pacifica; you have inland areas without an ocean view, and the more coastal areas with magnificent views. North Pacifica is a quicker commute to work, and so on.

If you don't use MARS, if you have NEVER used MARS, then you don't know the difference. You live at a certain level of ignorance. I have seen the appraisals of so-called Expert Witnesses. Flying by the seat of their pants. I have worked in some respected offices and seen many other reports. Don't you think I know? I do, exactemant.

As to your statement "locational factors relating to social factors," here's the problem: there is almost ALWAYS some correlation between location and racial characteristics. And the law and regulations say absolutely NOTHING about correlation levels. So, when you say "relating," we can take that to mean there are no adjustments to be made among locational factors. That is nonsense. 12% of blacks live in district A, and 14% of blacks live in district B, so B has more blacks than A. This kind of relationship exists everywhere. So, that first statement above is nonsense. Even if you succeed in stating more precisely what you really mean (I would assert you perhaps don't really know what you are talking about), - you will run into other problems.

So, in conclusion, there is almost always a relationship between locational and social factors. Per USPAP, such relationships, based on social factors involving protected characteristics, are to be avoided in appraisals. And yet, appraisals must be adjusted to market conditions. And, in practice, "market characteristic" should avoid protected characteristics, per anti-discriminations laws (FHAct, ECOA, Civil Rights Act, 1866, etc.)

The primary factor that trips up appraisers is the additional avoidance of "proxies" as replacements for protected characteristics. A somewhat grey area that can lead to contradictory requirements for the appraiser, which must be resolved through case law and other jurisdictional proceedings.

 

[alebrewer]

No, of course, the latest USPAP does not specifically require that appraisers use MARS regression.

That's correct - it doesn't. Just finished the USPAP Instructor recertification course. No mention of MARS, Python, or R even once... Of course, if MARS WAS required by USPAP (in whatever alternate universe that might happen), the appraiser population would be reduced to 1 person overnight...

 

[Fernando]

That's nonsense. I use MLS IDs for most counties, if I can get them (since some counties don't have them, though other options exist). There are in fact, usually adjustments for these different areas. But, of course, your comps should preferably come from the same MLS area, in which case there wouldn't be an adjustment. But it often happens that you have to extend your comparable area further out to get a decent set of comps. In my hometown of Pacifica, you have distinctly different areas. South Pacifica is much sunnier and warmer than North Pacifica; you have inland areas without an ocean view, and the more coastal areas with magnificent views. North Pacifica is a quicker commute to work, and so on.

I hardly use comps from North and South Pacifica. Completely different neighborhoods.
Only time would be rentals since few SFR rental data available.

If you don't use MARS, if you have NEVER used MARS, then you don't know the difference. You live at a certain level of ignorance. I have seen the appraisals of so-called Expert Witnesses. Flying by the seat of their pants. I have worked in some respected offices and seen many other reports. Don't you think I know? I do, exactemant.

Pacifica is one of easiest place to appraise in Bay Area. No need to use MARS. Just stick with comps in same neighborhood.

 

[RCA]

This is a case where USPAP is failing us again. It doesn't have to spend so much time on discrimination, if any at all. The latest 7-hour course literally beats the issue of discrimination to death, - while absolutely nothing about 'REAL' appraisal. All the course does is try to cover-up its weaknesses as an appraisal standard.

USPAP is crap!!

USAP should focus on appraisal accuracy. Then there wouldn't be anything but objectivity, and the racial characteristics of the people involved wouldn't play any direct role. But, of course, to do so, MARS would be required, as well as an overall jump in competency.

But good news! AI is laying off software engineers. What are the poor out-of-work software engineers going to do? Get an appraiser's license - that might be a good idea. But then again, the old vested guard will resent the move, and stupid **** makes a mess out of discrimination issues rather than deal with the real problems.

 

[ZZGAMAZZ]

Don't know if I will renew my license this year or next, but I am taking the required courses, including the latest USPAP.
====
No, of course, the latest USPAP does not specifically require that appraisers use MARS regression.

But all that new verbiage on non-discrimination - and associated laws, including case law, there are some notable repercussions.

For example:

A: School Districts:

1. You have to be very careful making adjustments for school districts, i.e., you cannot mention quality or test scores. Yet appraisers know that, all things being equal, parents with elementary- or secondary-school-age children are very concerned about moving into a school district with high-quality schools, as evidenced by quality ratings and score levels.

2. What is allowed is objective adjustments that use ONLY school district boundaries. Your only option is regression. And with so many other variables involved, only MARS regression will give you the needed accuracy. You put the school district ID into the MARS regression, and if there is a difference for some school districts, it will generate an adjustment. No reason given, other than it is just Market Characteristics.

B. Neighborhood/Market Area Crime Rates

1. Again, this is an area that will leave you wondering whether you can do the right thing.
2. You can use MARS regression on MLS IDs or Zip Codes, along with their crime rates, if the data comes from objective, publicly sourced statistics. Your reasoning and support should focus solely on the area and its borders, without any mention of people.
3. You can have a policy that allows you to avoid servicing certain areas due to high crime statistics, such as "We don't service areas where the murder rate is > 10 per 100,000 residents for safety reasons." That would, for example, allow San Francisco (4.5 murders/100,000 residents, but exclude Oakland (16-36 murders/100,000 residents)
4. You pretty much have to avoid using protected characteristics in your reports. And whatever you do, it should make sense and only involve non-protected characteristics.

Note: The way you would do this is have a table of MLS IDs or zip codes, plus crime stats, then create a crime stat column in your listings spreadsheet for crime and enter the corresponding crime rate. In fact, you might have one for the murder rate and another for the theft rate.

C. The Difficulty
1. You can't use so-called "proxies" for protected characteristics. But what is or is not considered a proxy? - That's where case law (e.g., historical redlining litigation) and the actual application of regulations (e.g., State Board disciplinary cases) have an impact.

2. In doubt, ask ChatGPT, Grok or Claude.

Please clarify your message about "proxites," i.e., what that means in appraisal context with an example or two. BTW, the audience remains out still concerning their perspective of yours, although I personally started to re-learn regression because your perspective if accurate is defensible. My only concern, as a blue collar tradesperson, is the scenario in which clients might not be persuided of the crediblity of a scholarly approach that you suggest, and reject results of an assignment simply because they don't understand the underlying quan logic...

 

[George Hatch]

In my experience appraisers are sensitive to economic incentive. And are resistant to all forms of authority and coercion, including coercion to take sides in their work.

"Sure, I can make the effort that you're asking for. What's the fee?"​

 

[djd09]

and dont talk about crimes rates or gangs and such..the unethical stakeholders maybe offended