Rob Coulter: Property Revaluation and Subjectivity
I had a very helpful conversation with a gentleman from the property revaluation vendor for Tiverton last night, and I learned quite a bit about the process. By the way, he was very patient and cordial, and I was very impressed with him, even if we may arrive a different conclusions. I agree with Justin that this should be an exploratory dialogue, and I do not pretend to have all the answers either.
The truth is somewhat in the middle of what I’m reading from comments here. As far as I can tell, they are using a multivariable regression computer model. It is susceptible to error because the sample sizes are not statistically robust enough for so many variables. When this happens, judgment calls have to necessarily be made. In a sense, the “methodology” has not changed, but there are still many, many variables calling for subjectivity on the appraiser’s part.. I don’t want to go so far as to call these judgment calls arbitrary, but there is definitely enough play between the joints for the appraiser to “skew” (if that’s the right word) a result based on assumptions being made.
Although it sounds fancy and complicated, the idea of using multivariable regression is to let the computer try to find the impact of one variable while holding all others constant. It’s like algebra on acid. This can’t be done by hand when there are dozens and dozens of variables, as there are, here, so we let a computer do it.
I do not believe that it is incorrect to use this type of modeling, but there are two very important qualifiers:
- The model only works if there are enough samples for each variable. I’m not sure there are here.
- More importantly, this model and all models have assumptions built into them. These assumptions are necessary for any model but are at the end of the day subjective and subject to dispute. For example, I learned last night that (roughly speaking) all taxpayers are taxed at nearly one acre of land no matter how much less they have, and owners with additional acres are only taxed at a very low cost per acre. So if you own one-third of an acre or one full acre, you pay about the same tax on land. Do you think that’s fair? Maybe yes, maybe no, but these are some of the assumptions that lurk behind the “methodology.”
There are myriad assumptions and they have a major impact. They do not involve only objective things such as acreage and square footage, but multiplying factors applied based on the style of the house. These are very subject to debate. For example, I argued that a solar panel on a roof should add value to a house based on fuel costs. The vendor suggested that it might detract because of decreased curb appeal. I replied that a new buyer could simply remove the panel. And so on. You can see how very quickly a lot of error and assumptions can creep into a system that otherwise sounds so impressive.
Again, I want to stress that I don’t think anyone is trying any funny business here. I was very impressed with the vendor, and I also very much respect David Robert, Tiverton’s tax assessor. But I do have experience with multivariable regression, and if that is the model behind this, I can tell you that we can’t trust it wholesale. I don’t think the “methodology” has changed, but there are many assumptions under this methodology that can be adjusted and are essentially subjective.
Rob Coulter is a member of the Tiverton Budget Committee as well as Tiverton Citizens for Change.
So “methodology” was clearly the wrong word for me to use. The question remains: was one of the subjective judgment calls such as to skew taxation toward waterfront properties unfairly (say by including different towns or by counting sales farther back chronologically?
So what is the proper method to achive a fair and objective (as much as possible) revaluation?
A problem without an alternative solution is nothing more than a poor solution that we are stuck with.
Justin: This will be my last post here but please, what is it with you? Why was anything skewed???? What proof do you have? Are you this cynical all the time? I think Monique, Jeff, and Rob has explained the process extremely well but for some reason you’re hung up on ‘values being skewed’. Please provide the details of what information you’ve been given or told that supports your wild claim.
David,
I think the difficulty, here, is that you’re convinced I’ve got some ulterior scheme going. It’s no doubt a reasonable reaction, having had more experience with the villainized image of me perpetuated by some of my friends about town than with, well, me, but it’s causing you to misread my intent. Actually, your assertion that I’m continuing to make any claims whatsoever suggests that you’re plain misreading.
I know of people whose property values went up while most went down, and general discussion has suggested that there was something distorted in the revaluation process that produced that result. I’m not asserting an unfair skew, I’m asking about one.
Toward the end of figuring the whole thing out, I’ve asked some very specific questions that you’ve demurred from answering. It’s probable that the reason you’re doing so is that you think I’m laying some sort trap that will snap if you give me an imprecisely worded answer. I’m not.
I just want to know whether particular neighborhoods differed in the groups of sales that determined their values. For example, if a waterfront neighborhood had no sales in the past year, while a working class neighborhood had a dozen, the working class values would be determined by those sales, but the waterfront property might require an assessment based on 2007 or 2006 sales, or perhaps Little Compton sales.
That seems to me to be one way in which the revaluation would be methodologically unfair. But I’m not asserting that it is.
Please believe me that this isn’t all part of my master plan to drive the town into poverty so that my legions of theocratic followers can sweep in and subject all the nice people of Tiverton to slavery. (That specific plan of mine hinges much more extensively on landcaping trends.)
PMFJI but despite the sophisticated models used to determine property values such as multivariable regression (I can barely pronounce it) the bottom line is that the process is imprecise and much of it is subjective.
One could revalue using different staff and get different results and they’d both be ‘correct’. It’s the nature of the beast and we’d better come to grips with the problem. The strange results one might perceive are unfortunately part of the problem with using property value as the basis for local revenue generation (unless there is malfeasance and I am certain that no one is saying that.)
Either we accept it or change the fundamentals. See my website.
I’m not sure I fully understand the complaint. Assuming the data they are using to build the model is accurate and the sample size is large enough (the biggest problem), using a regression analysis would entail fewer assumptions than any other kind of assessment.
For instance, you don’t have to try to figure out whether a solar panel adds or subtracts from the value of the home, you simply plug in the data from home sales and you can find out empirically how the presence of a solar panel affected the price.
Not to mention there are a lot of great things a town could do with that kind of data. For instance, if you made the coefficients public, homeowners could use that information to decide whether certain upgrades are worth doing. You could get an estimate for how much an extra bathroom would add to you home’s value (if you were planning to sell) or about how much it would cost in increased property taxes.
The only big problem is the size of the data set, but this can be fixed by including data from other areas. That would be pretty difficult with something like “ocean views” since the climate and many, many other things factor into that, but the value people place on other items should be pretty consistent throughout the region.
I first met Dave when he came to Tiverton in Nov 2005 (the previous revaluation year). Dave gave a public information revaluation talk which I attended. Dave explained that if say a sale was 9 months before 31 Dec 2005, that the sale value would be time propagated to the 31 December value. This (if I recall correctly) would be accomplished by say looking at the “trend” so to speak. For example, if a sale was for $500,000 on 31 March, and if it was generally accepted that sale prices decreased 5% between 31 March and 31 December, then the value of that property on December 31 would be $475,000 (5% less). I imagine this number would be an input to iterative analysis – or at least this is the idea – that sale prices are temporally propagated to the 31 December date. This said, the market peaked in 05/06, stalled, and then the bottom fell out in Fall 2008. My reading is that the “stall” generated less than normal sales data. And that this is compounded by the Fall 2008 meltdown, all coming off a market high in 05/06. The only comment I have about “out of town” sales are that maybe they could be used as a sanity check in the absence of Town data, sort of like asking yourself “does this answer make sense” after doing a physics problem. For example, assume two comparable properties that straddle the Tiverton/Little Compton town line. The taxes on the Little Compton property are $4,000 and the taxes on the Tiverton property are $8,000. Assuming a 5% investment yield, a lump sum of $80,000 is needed to generate the tax difference of $4,000. A buyer trying to decide which house to buy must discount the Tiverton property by $80,000 so that… Read more »
My own experience highlights one of Jeff’s points.
When I began thinking of purchasing a house, I constructed a spreadsheet of all the tax rates of towns I’d be intersted in buying. When taxes were considered, the same monthly mortgage payment would buy a $270k house in East Providence, a $300k house in Tiverton, or a $330k house in Westport…guess where I started looking…