Deep Dive - The Cost Approach

[Mejappz]

PS - site won't allow an XLS upload, so the TEL form is in PDF.

I think if you put it in a zip file it may come through.

 

[alebrewer]

The CA does not have much validity without extracting adjustments from market data. Neither does the income approach. The gist of the three approaches is they are all dependent upon market data. Sales are market data. Cost is market data. Rents and expenses are market data.

The cost books are not created out of thin air. They are polling national builders for costs. They are checking materials costs. They are analyzing building life, building repair costs and a lot of other things. Same with income, we have vendors who provide national indexes of cap rates, analyzing what buyers and sellers are actually doing. They are market data. Sales allows us to refine the obsolescences that sales can demonstrate.

The one place that the CA is forced to stand alone is the least reliable place for it. That's valuing a public building. There are few or no sales. There is the issue of what a private property owner would spend on a similar building and what the taxpayer has to pay. We had a building here that they wanted to re-roof. $125,00 bid but the city att'y said you have to use an architect or engineer because it was over $50k. In the end, they paid an engineer $50k to engineer a roof that ended up costing over $400,000. Made perfect sense to the government. Makes no sense to the taxpayer nor the market for a similar building.

No doubt the IA uses market data in it's development - hence my affinity for the IA. Do you think, though, that there is a difference between using market data to develop an approach and using market data to validate an approach? It seems to me that this is what makes an approach a 'stand alone' approach. Both the SCA and IA use market data to develop the approach - there is no need to validate with any other approach. From what I am hearing you say, the CA needs market data to validate the efficacy of the approach - which to me makes it a liability in any residential assignment (can't speak to commercial as that's not my world). IOW - if you need to develop the SCA to validate the accuracy of the CA, why expose yourself to scrutiny via developing the CA? If you don't develop it, you can't be held accountable for the methods and tools used for said development...

Been a while since I used the M&S, but as I recall, there are quality multipliers, cost multipliers, and local multipliers (there may be others, but again - it's been a while). So we see that the PPFN can pretty much be manipulated until we get a number that 'fits' our narrative with the SCA. Same for TEL, EA, contributory value of site improvements, % depreciation (which I know is related to TEL and EA - but not in a linear manner as many believe).

Which, I guess, kind of illuminates your illustration of the public building example - similar to the state park example I cited the other day. There are no sales with which to 'validate' the efficacy of the CA, so the CA is developed as a standalone approach. Aside from the site value, it seems to me everything else is just fairy dust...

 

[Terrel L. Shields]

Do you think, though, that there is a difference between using market data to develop an approach and using market data to validate an approach?

Not really. It's all market data. And why do all the textbooks teach all 3 approaches? Why does USPAP address all three approaches without caveat?

Reality is the one central approach is the income approach. Why do you buy real estate? For income or to avoid paying rent with some expectation of earning a residual upon selling. Why do you buy a building? Income or utilize in your business which makes income. The value in real estate is the income it either makes you or saves you.

 

[CGinMN]

As is see it the benefit of the cost approach in residential valuation is that serves as more support for the adjustments in the sales comparison approach. If I know land is worth $50,000 and that leaves $200,000 for the home (based on rough sales comparison of the subject, or from the most similar comp), and the RCN is $300,000, then I can know that the components are physically depreciated 30-35%, some more (short-lived) and some less (long-lived). Using a cost guide or cost breakdown from builders in your appraisal files, you can get a good idea of each component's share of the total cost. Solomons' cost "calculator" does this as well.

As a standalone approach, the CA takes a lot more work than the above to credibly develop. But when used for adjustment support it is another tool in the bag along with regression and paired data/sensitivity, from which you can reconcile within. Reconciling your adjustments from 2 or more approaches is a way to bulletproof your analysis. No need to put the full CA in your report, just summarize how it was used in the development of the SCA and keep it in the file.

 

[sputnam]

Every part of data used in the Cost Approach can be extracted from the market. Very few appraisers do that throughout the Cost Approach (I don't either). Instead they rely on building cost data provided by third party services. They use age/life to determine depreciation, etc. Every number that is used which is not extracted from market data, is one more point where error can creep in.

I do find that if an appraiser spends the time learning how to use his/her cost service of choice and extracts as much as is practical from market data, the CA is very reliable.

Recently reviewed an appraisal report where the appraiser stated something to the effect that the Cost Approach was considered unreliable and was given no consideration in the final opinion of value (that's what most of us do in residential appraisals, right?). Only thing was, the variance between the indication by the Cost Approach and the indication by Sales Comparison was less than $300 on an $800,000 property. I don't know whether the appraiser worked the Cost Approach or backed into it... either way, his comment was ridiculous.

 

[alebrewer]

I do find that if an appraiser spends the time learning how to use his/her cost service of choice and extracts as much as is practical from market data, the CA is very reliable.

As a standalone approach, or in conjunction with one or more of the other approaches? IOW - let's say you 'spent the time learning', do you think you could develop an accurate (however you define that) estimate of MV based JUST on the CA?

 

[alebrewer]

As is see it the benefit of the cost approach in residential valuation is that serves as more support for the adjustments in the sales comparison approach. If I know land is worth $50,000 and that leaves $200,000 for the home (based on rough sales comparison of the subject, or from the most similar comp), and the RCN is $300,000, then I can know that the components are physically depreciated 30-35%, some more (short-lived) and some less (long-lived). Using a cost guide or cost breakdown from builders in your appraisal files, you can get a good idea of each component's share of the total cost. Solomons' cost "calculator" does this as well.

As a standalone approach, the CA takes a lot more work than the above to credibly develop. But when used for adjustment support it is another tool in the bag along with regression and paired data/sensitivity, from which you can reconcile within. Reconciling your adjustments from 2 or more approaches is a way to bulletproof your analysis. No need to put the full CA in your report, just summarize how it was used in the development of the SCA and keep it in the file.

So it seems like I'm hearing from you and Terrel - both of whom I admire - that the CA isn't really useful as an 'approach' in resi appraisal, but moreso as a tool to assist in the other approaches? If so - I can get on board with that. My only caveat would be that (IMO) most appraisers don't understand how to do depreciated cost to estimate contributory value of an improvement; it's much easier to simply used paired sales, regression, or sensitivity analysis IMO. And even if they did understand how to develop a depreciated cost, wouldn't you still need to validate the findings with paired/grouped sales anyway? IOW - let's say you have a pool and want to estimate the contributory value of the pool. You could take the overall MV (which you'd have to get from the SC), subtract the site value and the improvements value and, theoretically, you'd be left with the contributory value of the pool. You could then estimate the cost of the pool new and calculate the estimated depreciation. But for what purpose? You've already estimated the contributory value when completing the SC...

 

[CGinMN]

My only caveat would be that (IMO) most appraisers don't understand how to do depreciated cost to estimate contributory value of an improvement; it's much easier to simply used paired sales, regression, or sensitivity analysis IMO.

Not gonna disagree with the first part, but in my opinion, for most assignments, if the goal is to use the CA to develop adjustments it takes 15 minutes if you use the right tools. There's no need to get in the weeds with market extraction for TEL and depreciation. Estimate the land value, the rest is the depreciated value of the improvements, then break down the improvements into their component parts to ballpark their contributory value. Whether the improvements contribute $240,000 or $260,000 won't significantly alter the adjustments. Then since you are 80% of the way there, you can fine-tune it with sensitivity.

Sensitivity as a stand-alone approach for adjustments usually relies upon a baseline understanding of depreciated cost, unless you have a very homogenous set of data with only minor differences. In that case, sensitivity is basically the textbook example of multiple pairs where everything works out neatly. When that is not the case, any appraiser can close the gap on their sales by making a $40,000 condition adjustment to Comp 1, but what if depreciated cost is telling me it's $15,000 at best? That other $25,000 is some other variable that I've missed or several other under/over adjustments for other characteristics. Having a baseline knowledge of cost makes sensitivity more credible.

Using paired sales as a stand-alone is not that easy, again, unless you are working with homogenous data. I know it is the appraiser's go-to explanation for how they derived adjustments, and it is what they teach in residential courses as the primary method for deriving adjustments, but I would argue developing every adjustment using paired sales is much more tedious than a ballpark cost. And even if you do your pairs in market area A, that doesn't mean they will apply in market area B, or in a year from now. To do paired sales correctly, you need to be doing it frequently and in many markets.

There are always going to be exceptions to the above, most often when functional obsolescence is at play, such as is common for pools or when properties are in transitional locations and a change in HBU may be approaching.

The cost approach as a standalone approach does take quite a bit of work to develop credibly. It is often not feasible for most mortgage financing assignments. The exception there is when the current use is close to the optimal use, and there is very little obsolescence. Comparing costs to prices of non-competing new builds in other market areas can be used to extract EI for the subject. If the subject lacks good direct comparisons in its market area, for example, when all the direct comps are all 10+ years old, then it might make sense to place some weight on the cost approach. Once you have to market extract TEL, all forms of obsolescence, and EI, and, to do it well, reconcile these conclusions from several comps the cost approach for residential usually becomes less reliable than any one sale in the SCA.

 

[alebrewer]

The exception there is when the current use is close to the optimal use, and there is very little obsolescence.

I'd disagree a bit and say that the exception is when there is no obsolescence or depreciation. As stated in an earlier post, when you annualize the depreciation on a 1-5 year old home, it will result in an understatement of the TEL (and hence the REL), as the front end of an improvement's life cycle bears a heavier depreciation load than the back end. Would you agree?

 

[alebrewer]

Sensitivity as a stand-alone approach for adjustments usually relies upon a baseline understanding of depreciated cost, unless you have a very homogenous set of data with only minor differences.

I've always performed sensitivity analysis in conjunction with other techniques (grouped sales/regression), so that - theoretically - you've solved all the other residual issues prior to engaging the analysis for the bracketed feature you're applying sensitivity to. More often than not, there tends to be some fine tuning as the analysis develops - e.g., I'll use sensitivity for GLA, then maybe have to go back and tweak the sensitivity for site (or the paired sales adjustment for garage/bath/etc.).