I'm going to morph several different threads that I've read over the years into my question to you and definitely go on a tangent....
Market extracted/supported adjustments vs. refine my value conclusion in the reconciliation vs. rational adjustments (which I deem to be common sense)...
1. If market participants provided conflicting opinions.
This is a reconciliation issue.
In a study I recently completed for a research paper, our office interviewed a number (50+/-) market participants regarding various sales of homes that had energy efficient (primarily a PV system). These included the agents who represented the buyer and seller, as well as agents of homes sold that would have been considered comparable to the property with the system.
The responses were not uniform. But I'm not sure I'd consider them "conflicting". In general, the participants put the value of such systems at "neutral" to perhaps 5% max.
Most considered it a positive amenity even if they couldn't put a $ on it. However, in some cases where the market was extremely tight (low inventory), some said the house would have sold at that price with or without the energy-efficient enhancements.
So, assume that the market participant input indicated a value impact of 0 to 5%. The PV analysis I complete has, to the best of my recollection, fallen within that range.
Survey says 0-5%. Math analysis shows 3.5%.
I think any of us can reconcile those two indications to a reasonable and credible consideration on how the enhancements impact the final value opinion.
2. If market data/extraction is inconsistent/inconclusive/inaccurate/etc.
I'm going to eliminate the "inaccurate" label and stick with "inconsistent" and "inconclusive".
In a CE class I teach (
Supporting the Adjustments: Let's be Reasonable) I show a number of examples where, by using several techniques (paired sales, historical trend, cost, income, and agent survey), that it is more likely than not that the analyses will not be uniform. But what they typically do is provide direction (positive or negative) and a range. This is a data issue and, unfortunately and in most cases, we do not have 100% of all the data necessary to explain all the value differences (if we did, then in theory, all analyses would point to the same result).
Again, this is a reconciliation issue; each analysis needs to be evaluated based on the quality/quantity of the data.
Assume that there are two side-by-side developments. One sells for an average of $1,000,000 and the other sells for $1,200,000. On paper, some differences can be seen (the higher sale development has slightly larger lot sizes; but we are talking a difference of 8,000sf vs. 12,000sf). Average and median GLAs are not that much different (2,500sf +/- vs. 2,650sf). Driving through the neighborhood, one can see the the higher-selling development has superior exterior finishes, landscaping (as a rule), and more houses in that neighborhood have 3-car vs. 2-car garages.
Looking at the MLS photos, it is apparent that the overall finish (and quality) of the higher-selling development is superior (but not over the top).
Talking to the agents, nearly all state that the higher selling homes are superior in quality.
Does the slightly larger, GLA, lot size and 3-car garage explain the $200k difference? Likely not (and in this case, no). It explains some, but not all.
Can the quality difference explain the value difference? Probably.
Assume that GLA, lot size, and 3-car garage differences can explain about $75k of the difference. The remainder is $125k.
Using MSV and comparing Class D home (average) vs. Class D home (good), the difference (including multipliers and a 10% EI) is $53/sf.
Assume Depreciation is 18% (Effective age is 10, TEL is 55, REL is 45). Depreciated cost difference is $44.
One analysis indicates a $50/sf difference ($125k for quality).
The other indicates $44/sf difference ($110k for quality).
Market participant survey indicates the value difference is due to quality.
Is there enough to conclude a quality adjustment between $100k and $125k (8% to 10%)? I think so.
You want to go a little lower? ($75k is 6%). I wouldn't fret that.
You want to go a little higher (12% is $145k)? That might be pushing it, but if that is your conclusion, make the case for it.
Is the analysis credible? (yes, I think so).
Is an adjustment between 8% to 10% reasonable? Yes, I think so.
Pick your point, explain your rationale, and go forward.
What if a pool is analyzed and the results are $0 contributory value to $20k contributory value?
You want to conclude that a pool has no contributory value? There is data to support that.
You want to conclude that a pool has $20k contributory value? There is data to support that as well.
Most might conclude a pool has some value, more than zero, but not more than $20k, and may simply split the difference and value the pool at $10k.
Is that credible and reasonable? Certainly sounds so to me.
In the first example, we have evidence of a positive value influence (quality); everyone agrees that the higher selling development is superior in quality to the lower selling development, and that quality matters. So, we have a basis for the direction of the adjustment; it is going to be downward in quality for the superior project comparables vs. our subject. The rationale for making an adjustment is supported.
How much should we adjust? In my example, anywhere between $100k to $125k is consistent with the data; making it a bit lower isn't necessarily unreasonable, going much higher begins to stretch the test of reasonableness.
In the second example, we have evidence of no contributory value vs. some ($20k). Again, the analysis has set the range, and the appraiser will reconcile that data to a point. IMO, anywhere from zero to $20k is credible and $10k is certainly reasonable.
Now, what if the analysis shows opposite results? One set of analyses shows a negative location adjustment and another shows a positive location analysis?
What can I tell you? That's when the appraiser's local knowledge and experience makes all the difference in the world. I would expect the report to describe the results of the analysis and then make a determination of how it concluded what to do. An appraiser (IMO) could convince me that a positive, negative, or no adjustment might be appropriate in that case. The better the report's ability to communicate the rationale of the adjustment, the more likely I'm going to be convinced. That is one of the things that differentiates us from an algorithm.
3. If the appraiser is reluctant/won't apply adjustments to the grid...
Why apply an "unsupported" characteristic/element value to the subject in the reconciliation?
The question implies (I would say it is explicit) that if an adjustment is not applied in the grid, then it must be "unsupported"?
I reject that out of hand.
An adjustment is an adjustment is an adjustment: regardless if it is considered quantitatively in the grid or qualitatively in the reconciliation.
Unfortunately, in the real world of retail residential-mortgage appraisals, the expectation is to "bracket" many of the elements of comparison. Sometimes that makes perfect sense. Other times, not so easy to do.
Interestingly, when the adjustment is negative (say, a cost to reconvert a finished garage back to a garage), most users don't have an issue with that. But if the adjustment is positive, that triggers the "where's the bracket" request.
The reconciliation process I use results in the same outcome regardless of where the adjustment is applied (in the grid or in the reconciliation). The contributory value of the amenity is identified, analyzed, considered, and its consideration does affect the value. This is the same result of an adjustment in the grid. And, most important (IMNSHO) the process is reasonable and credible.
There are certainly times when the expectation (users and peers) would be to make an adjustment in the grid. I'm fine with that. There are other times when making the adjustment in the reconciliation makes equally good sense, if not more.
The expectation should be that when applying different types of analyses to the same problem, the results are not going to be identical. Unless we have all the information necessary to evaluate the problem for each analytical technique, differences should be the expectation.
Usually, the trend-indication (negative to neutral, neutral to positive, all negative, or all positive) is consistent. The direction and support for an adjustment is supported: the specific adjustment-point may have some variance. This is where the appraiser's reconciliation of the analyses plays a critical role.
When analyses indicate conflicting results (positive for one and negative for another), then this situation more than any other calls for appraiser judgment and expertise. The reconciliation of the quality/quantity of the data along with the appraiser's knowledge should provide a credible conclusion and a reasonable result (adjust or no adjustment, and if so, how much?).
Reconciling differences in the quality of the data of a particular analysis (and how that impacts the reliability of the results) is the one thing an appraiser can do better than an algorithm (IMO).
Hopefully that addresses your questions.
