Order of operations for market trend analysis

[CGinMN]

Theoretically, seller concessions and any expenses made immediately after purchase should be accounted for first to make it cash equivalent. Then you apply the market conditions adjustment to that cash equivalent adjusted sale price. Only after the sale is made cash equivalent and current do you start adjusting for physical differences (site and improvements). If you go out of order you'll get a different result, which is typically a rounding error in most residential assignments.

 

[TerryRohrer]

Theoretically, seller concessions and any expenses made immediately after purchase should be accounted for first to make it cash equivalent. Then you apply the market conditions adjustment to that cash equivalent adjusted sale price.

Agreed.

Only after the sale is made cash equivalent and current do you start adjusting for physical differences (site and improvements). If you go out of order you'll get a different result, which is typically a rounding error in most residential assignments.

These statements would only apply if all adjustments were derived from the sales being adjusted. Otherwise, order of adjustments would only matter in the case where one or more adjustments were a percentage.

The following image is from The Appraisal of Real Estate, 15th Edition (2020). The first time I recall seeing an argument for applying other adjustments before using the adjusted sales to gauge changes in market conditions from those adjusted sales was in The Appraisal of Rural Property, 2nd Edition (2000, a joint effort by the AI and ASFMRA). If I remember correctly, it was well debated and the final text was a compromise, likely with the rural appraisers promoting the idea of gauging market conditions last when necessary due to a limited data set.

 

[CGinMN]

If the adjustments for physical differences were derived from a large data set, again, theoretically, you'd want to adjust that entire dataset for transactional characteristics first. You can do it without, but it's not as pure.

The reason your market conditions analysis and corresponding adjustment should be locked and loaded before you even start your sales comparison approach is that market analysis should be done earlier in the process, when you are looking at segment-level data, before the comps are narrowed, often in conjunction with the highest and best use analysis.

 

[alebrewer]

The adjustment for concessions or financing is at the top of the grid, and it is done independently of other adjustments to reflect the concession or financing impact on price - imo it does not affect the value whether we choose to develop it first or last.

Disagree. I believe that to be the case for market adjustments (the part about being done independently, that is), but concessions adjustments are applied (based on the promulgated definition) in a manner that approximates 'the market's reaction to the financing or concession based on the appraiser's judgement.' That's why - IMO - these have to be done last.

 

[alebrewer]

Once you solve for one thing, other things get easier to solve, and everything is interdependent.

And, like a puzzle, (at least for me) I find myself having to fine tune adjustments that have already been applied. IOW - apply site adjustments first, then find that the GLA adjustment factor is a bit off. Fix that, and you have to fine tune the site adjustments just a bit. Rinse and repeat until the range is as tight as possible.

 

[CGinMN]

What I am reading is that if you are doing an "in-grid" sensitivity analysis using the comps, that should be the last step. I agree with that.

But how many variables can you adjust via sensitivity tests without overfitting? The most pure is just one. The rest of the data should then ideally come from a larger data set. If you start making 3 or 4 different adjustments using in-grid sensitivity, you will find that different "sets" of adjustment rates will narrow the range similarly. For example, a $30 GLA with a $30,000 garage might narrow the range the same as a $75 GLA and $15,000 garage. Add a 3rd or 4th variable, and the tinkering might lead to a really good-looking grid, but is it overfitting?

 

[alebrewer]

For me, any element of comparison that is discrete, and is bracketed, is a potential suspect for sensitivity analysis. It's the same as using multi-variate regression. You're basically manually regressing the elements of comparison. I also agree that you run the risk of overfitting - just as you do with multivariate regression.

 

[ZZGAMAZZ]

The adjustment for concessions or financing is at the top of the grid, and it is done independently of other adjustments to reflect the concession or financing impact on price - imo it does not affect the value whether we choose to develop it first or last. That said, I usually do it first out of habit; sometimes, when checking the reports, I find I missed something in that realm and then will go back and address it.

There are times I find when, after making the adjustments, something glares out as "off" - not revealed until we made the total of adjustments.

Yes of course, but isn't it true that the SCA line item adjustments are listed sequentially in order of "significance" so that the adjustments should be made top to bottom (with subsequent fine tuning}. Seems to me that the "condition" adjustment is the largest in about 70% of every SCA, so it should be made first (but that's not based on industry standards).

 

[J Grant]

Yes of course, but isn't it true that the SCA line item adjustments are listed sequentially in order of "significance" so that the adjustments should be made top to bottom (with subsequent fine tuning}. Seems to me that the "condition" adjustment is the largest in about 70% of every SCA, so it should be made first (but that's not based on industry standards).

We should make terms of sale/concessions and then any terms of financing adjustments in the first two steps, but even if we make them last or out of order, the beauty of the computer doing the math is that it works from top own to the same $ amount either way. You can experiment with your own software to see it, put a comp on the grid, adjust for concessions first, then take that out and make the adjustment last; the computer mathematically makes the adjustments from top line on grid down anyway

 

[Fernando]

Not if I read the lease.

Sometimes in the listing, the seller will pay the rent for first year. What's that about? Free rent for the tenant? How do you adjust for cap rate?