AVM of Commercial Apartments

[Terrel L. Shields]

I turned this one down. I just don't want to fool with it. But it had an AVM attached, performed in Dec. 21. Figures to about $190/SF for a 30+ year old apartment complex. I am familiar with it but I think this is not reasonable considering local land costs - the appraised value is about $1.5 million (reported below is the assessed value which is 20%)and building costs.

 

[TerryRohrer]

Am I missing any indication of the number of units in all that?

 

[George Hatch]

Set aside the obvious problems with the median sizes indicating to a dataset where most of the sales therein will not be directly comparable to this 2021yb property with 12,000sf located out away from town.

Price/sf is the least relevant unit of comparison there is for multi-family. The units don't even rent by the door. They rent by the number of bedrooms, which means a 1bd unit generates less rent and therefore has less contributory value than a 2bd.

 

[Terrel L. Shields]

Price/sf is the least relevant unit of comparison there is for multi-family.

Actually there is a reasonable relationship between the two here for sure. Smaller apartments are 2 bed, larger units 3. Habitually landlords here do not change rents except upon rare occasion. They only change rents when they get a new tenant. Therefore, there is a discrepancy in the economic rents from market rents. And rents are pretty uniform overall in the town. (This is a town of about 19,000.) The AVM used a huge series of multifamily sales from surrounding areas over the past 4 years to estimate the value and the R2 for their selection was like 0.25...which renders the analysis pretty meaningless. They used a huge number of sales that were completely non-comparable. 4 plexes, etc. So the AVM is a nothing burger to me. But interesting how it analyzed the data.
For instance, the list of sales listed them as "5+ units" but since I recently did some duplexes in this town, a number of the "sales" were, in fact, duplexes or a single fourplex. In fact, none of the sales had 5 or more units. Look at the building size. We don't have 300 SF apartments in this town.

 

[George Hatch]

Data qualification is the thing, regardless of property type. If none of these properties are directly comparable then it doesn't matter how they're analyzed.

 

[Terrel L. Shields]

AVM - garbage in, garbage out

 

[George Hatch]

Yeah, but that's actually not the AVMs fault. It's the fault of the quality of the database. If the database broke all those sales down by number of units and number of bedrooms and reported the gross/net incomes that would make the AVMs valuation a fair bit more usable. Perhaps not as close as a human appraiser could do it WRT handling abnormal incomes or rent control limitations or extra income sources; but still within reason.

 

[Vernon Martin]

This company is out of Honolulu?

 

[Terrel L. Shields]

This company is out of Honolulu?

I donno. They claimed I had done work for them before but I don't think so. Not under the name they used.

 

[Michael S]

It boggles the mind that nowhere in that dataset is the number of units. What user could look at this and thing it was anything but worthless garbage?

Frankly the square footage (gross or net rentable) of many apartments is suspect, but it's pretty hard to get the unit count wrong.