GLA Adjustment Using Sensitivity Analysis

[ltyrni]

I have a question regarding GLA adjustment that I am wondering if some of you experienced appraisers could help me with. I have attached my table for sensitivity analysis as I have performed it below.

Question 1: I adjusted the comparable properties for site (lot sqft), age, condition, and differences (bed, bath, garage stalls) before adjusting for GLA. I have read some old forum posts on this site which a poster had a table for GLA sensitivity analysis that I downloaded. In that document; it said to adjust for lot size, outbuildings, and garage differences before adjusting for the GLA.

Should I adjust for GLA last as I did in mine below, or should it be adjusted for before other variables? When should GLA be adjusted for?

Question 2: If you look below, you will see I used Standard Deviation of the data set to choose a square footage adjustment. Is this allowable and credible in your opinion? I also have a chart using Coefficient of Variance, which is just another side of the same coin.

In the example I had found on this site, it used Max-Min. The two numbers gave a different square footage adjustment; $45 for the StDev and $57 for the Max-Min. Should I be using Max-Min instead?

Question 3: When I create the table, should I use a smaller variable adjustment like $1 instead of $5? Is that too large of a gap?

I really appreciate any help you all can give me! I am new and if this is posted in the wrong forum, please let me know and I will correct my next posts. There will likely be a few.

About me. I am a new associate general appraiser. I am located in Iowa with an agricultural background. I graduated from college with a degree in Business Management and second major in History in 2011. I spent the past 7 years operating and administering our 6,000 acre family farm. I have an agricultural chemical business as well. Changes in our farm and structure will allow me to be more hands off which creates time to do other things. I want out of the chemical business, mostly because it requires an intensive workload for 3 months in spring and early summer, about 1,000 hours in that time frame. Also rashes... strange rashes can appear, lol.

With that background, I primarily want to focus on farmland valuation. I have a supervisor, but largely am on my own. I am fine with that; I am a self-starter, quick to learn, enjoy working, and expect quality out of everything I do. What I primarily need is help with is residential valuation and supporting all adjustments in a credible way that is acceptable to the professional standards. My questions will initially revolve around that and I look forward to learning from you all and maybe forming contacts!

Another note: I know I could have a narrower deviation, where it would actually match with the $57, if I removed comp #5. Comp #5 was included for bracketing, but perhaps I have enough with the others. It is the second closest home, but several of the others are 45-60 miles away. Small rural towns all, 2-3 bed, 2-3 bath, basements of various states of finish, 2 car garage, ranch styles built between 1976 and 2010. Subject is 1996 ranch, unfinished basement, 2 car garage (barely), 3 bed, 2 bath, C4, Q4.

 

[J Grant]

Just will share some thoughts...any item we want to find adjustment for we adjust for other major items first to isolate the item we are adjusting for, (sf, view, or site/upgrades etc). You were correct adjusting for major items first.

I don't derive adjustments from statistical data so I won't comment on that aspect. Keep in mind the purpose of an adjustment is to narrow the range between comps to find value for our subject. With additional sf of comps to subject we are looking at contributory value of the sf. A Q4 house costs less to build and tends to be lower price for the dwelling, thus the $ per sf contributory value tends to be commensurately lower. Imo, $40 a sf narrows the range better and is more typical of many markets, but it's your market might show different. Comp 4 and comp 5 are both outliers on high or low end so whether you will keep them as comps or not ...you can always throw one out (or both out) and if need be you can change your $ per sf adjustment on the grid as well. We have to start somewhere. I'll often put all my adjustments on the grid, then may end up changing one to better reflect the market and narrow the range. don't be afraid to change things on your appraisal till the signature!

 

[ltyrni]

Thanks for the reply. You're right, I could throw out 4 and 5. I should as it would narrow the range significantly. I should still be bracketed as well in contributory attributes, looking over my grid.

So are all of your adjustments qualitative then? What process or rationale supports them?

I am hoping to find the simplest, credible manner to make adjustments. I have seen where some people say they adjust within the grid, and I would like to do that. My market area is dispersed small towns of 500-5,000 people by in large for residences. There can be a lot of variability. I've run several regressions on large data pools to kick out numbers for adjustments, but I honestly don't like them too much. Things like age equals $800 per year, bedroom 3 equals $8,000, bathroom 3 is $4,000, etc... But they are really generic and I don't think they reflect as well as a paired-sales analysis would.

If you just adjust within your grid based on what is contained within, what process do you perform?

The adjustments for my sales within the grid in the assignment I am referencing are currently adjusted by comparing to each other. I don't know if it would be considered "credible" or "proven" though.... For example, my regression analysis of a broad pool says bathroom 3 worth $4,000. The grid tells me bathroom 3 is worth nothing. I agree with the grid. Why is a 3rd full bath in a 2 bedroom house worth extra money? Or a 3rd full bath in a ~1,400 sqft house anyway?

But if you, or anyone else, makes adjustments in grid, how similar do the properties need to be for a paired analysis? If I have a 1,500 sqft and a 1,600 sqft ranch, built in 1986 and 1984, 2 bed, vinyl siding, 1 car garage, Q4, C4, BUT one has 2 baths and the other only 1. Are those two suitable for use as paired-sales for the value of a 2nd bathroom? Or does the fact that they are two years apart in age and 100 sqft in area different preclude their use for that purpose; ie that they are not perfectly alike?

 

[Restrain]

Give you another viewpoint from our market (SW FL). We have run enough similar homes (thank goodness for generic housing) that depending on the quality of,the home, we have a pretty good feel for garage spaces, etc. some things often don’t require adjustments (2 br vs 3-4 br due to age of populace and market). Also, over the decades, if homes are relatively similar in age, we don’t bother with age adjustment but rather deal with condition and updates. We can use the sales to pull out the GLA adjustment as well as updates/remodeling.

Now having worked rural East Texas I know what you are facing. You are not going to be able to use statistical modeling to support your adjustments. Just way too few comps, way too many differences (ages, amenities, sites, locations). So, whenever you can find a paired sale, put it in your file. You are going to have to rely on “that works “ analysis and consistency in your report. Then look at your listings. If after adjusting for everything, if your listings are lower than the value, back to the drawing board.

Now let’s look at your two sales. 1500-1600 SF will have a marginal size differential. The age should not be an issue, So if the condition is similar, should be able to give a feel for the 2nd bath. 1 bath in a 3BR should have both physical and functional obsolescence and that should be reflective in the sales data. Conversely 3 baths in a 1600 SF home may be a superadequacy and may not be a premium, unless you are dealing with 2 master suites.

Hope my ramblings helps.

 

[J Grant]

Thanks for the reply. You're right, I could throw out 4 and 5. I should as it would narrow the range significantly. I should still be bracketed as well in contributory attributes, looking over my grid.

So are all of your adjustments qualitative then? What process or rationale supports them?

I am hoping to find the simplest, credible manner to make adjustments. I have seen where some people say they adjust within the grid, and I would like to do that. My market area is dispersed small towns of 500-5,000 people by in large for residences. There can be a lot of variability. I've run several regressions on large data pools to kick out numbers for adjustments, but I honestly don't like them too much. Things like age equals $800 per year, bedroom 3 equals $8,000, bathroom 3 is $4,000, etc... But they are really generic and I don't think they reflect as well as a paired-sales analysis would.

If you just adjust within your grid based on what is contained within, what process do you perform?

The adjustments for my sales within the grid in the assignment I am referencing are currently adjusted by comparing to each other. I don't know if it would be considered "credible" or "proven" though.... For example, my regression analysis of a broad pool says bathroom 3 worth $4,000. The grid tells me bathroom 3 is worth nothing. I agree with the grid. Why is a 3rd full bath in a 2 bedroom house worth extra money? Or a 3rd full bath in a ~1,400 sqft house anyway?

But if you, or anyone else, makes adjustments in grid, how similar do the properties need to be for a paired analysis? If I have a 1,500 sqft and a 1,600 sqft ranch, built in 1986 and 1984, 2 bed, vinyl siding, 1 car garage, Q4, C4, BUT one has 2 baths and the other only 1. Are those two suitable for use as paired-sales for the value of a 2nd bathroom? Or does the fact that they are two years apart in age and 100 sqft in area different preclude their use for that purpose; ie that they are not perfectly alike?

Omg, those are great pairs !! Since it is rarely possible to find perfect matched pairs, adjust for the differences and what is left is a paired sales for comparison and extraction of adjustment for an item.If there is no clear or measurable difference in market such as bedroom count I won;t adjust and explain why.

Land values are their own topic esp with acreage if you appraise farms/rural am sure you have that down ...Restrain answered nicely about residential attributes. What supports the adjustments are the results of comps on the grid , are the adjustments reasonable, do they have a relation to cost or repairs needed...also review additional sales, talk to RE agents etc. If one appraises similar homes in same subdivisions don't have to invent the adjustment from scratch each time but they should be credible for the properties used. Dollar per sf difference has some relation to quality of house and cost to build...we assume a well informed buyer for MV. A Q4 home that might cost $125 a sf to build will have a lower contributory value adjustment per sf than a Q2 house that is $400 a sf. I can't swear by a formula as each appraisal is different, but I often find one third roughly of cost as a good start point for $ per sf contributory value adjustment.

Picking the right comps is all important and I usually spend more time picking the comps than sweating over adjustments, because when you have the right comps, even if there are differences, the adjustments reveal themselves in development.

 

[ltyrni]

I will look for and file those away. I've seen a few different ones so far and have an excel page opened with them haphazardly listed there. I am hoping that utilizing a regression model for different variables, and coefficient of variance on the mean of the greater pool creating a bracket, and then some of these pairs would be adequate to support the adjustments. For example, my regression model on the pool, adjusted for age, currently gives me a square footage adjustment of $66.68 a square foot with an r2 of 0.4068.... So awful correlation... The mean and one standard deviation is $65.42 and $34.20. But if that is only secondary support, alongside noted pairs and anything I can find in my grid, it should be sufficient to support the adjustment. Like in the above sensitivity analysis, with a $57 per sqft adjustment.

Thanks for your ramblings Restrain, they do help!

 

[ltyrni]

I will remember the 1/3 and see how well that correlates with my market. So I guess I won't get too much quicker at picking comps then, lol. I thought I was spending an inordinate amount of time on them. This house appraisal has taken as long as the farm ones! I figured they would take half the time. But it could just be my limited data pool combined with a greater sense of comfort and competence with agland as opposed to residential. Appreciate your help J Grant.

 

[Michigan CG]

Welcome to the forum. I still maintain my Iowa CG license. Not sure where you are located as Des Moines and the Quad Cities are different from the remainder of the state. You state that you have been involved with the family 6,000 acre family farm. I would bet you have more knowledge about farms than most people can even imagine (I do farm work).

I strongly suggest you pursue this avenue and take classes (if you can find them) from ASFMRA. You most likely know things that would take years to teach a new appraiser about farms. I am guessing you know a lot about soil types, drainage, drain tile, slope and productivity. This is stuff that takes years to learn.

Question 1: I adjusted the comparable properties for site (lot sqft), age, condition, and differences (bed, bath, garage stalls) before adjusting for GLA. I have read some old forum posts on this site which a poster had a table for GLA sensitivity analysis that I downloaded. In that document; it said to adjust for lot size, outbuildings, and garage differences before adjusting for the GLA.

I have appraised residential properties in three states and rural residential properties in three states. You are NEVER going to convince me that you can derive any meaningful adjustment for bedroom count.........never, unless the subject property is in Des Moines and then I am still going to doubt that adjustment.

Sensitivity analysis is not typically derived from regression but you can do it that way.

Generally, sensitivity analysis is attributed to Mark Ratterman, MAI, SRA from his multiple books that are published by the Appraisal Institute. His two latest books are Value By Comparison which was just updated in 2018. I STRONGLY suggest you but the latest edition. The guy is considered the premier writer of residential books.

........now...........

STOP with the regression; you are doing it wrong!!!

You need 30 data points otherwise the data is completely MEANINGLESS. If you are working in a rural part of Iowa you will have very little success or credibility with regression. If you want to take the BEST class about stats take the Appraisal Institute class (Stats and Finance)........ The author and instructor of that class is great; I last took it on-line. I assure you, you will throw out all that regression you have posted here as it is completely worthless and whoever taught you that also needs to take that class.

 

[ltyrni]

I am actually reading the PDF version of "An Introduction to Statistics for Appraisers" presently. I ordered that along with several others from AI today because I thought everything I was doing in these adjustments regression was a load of BS... So in addition to that, I have a couple on AVM's and Market Analysis for RE coming as well. I looked up the one you suggested and will order it too. I do like to read and learn, so I appreciate the suggestion! I also planned on signing up for that very class as well, in the January 15th period after I could read and digest that Intro book and hopefully the others as well.

So in your experience as a rural appraiser, what residential qualities are typically worth adjusting for?

I really want to make it to some of the ASFMRA courses. They had some in DSM, but they stuck them in the last week of June, which I absolutely could not make at the time because of my spraying obligations. They do have an education week scheduled for late July in Omaha that I saw. I intend to contact them about it and see what they will be offering. I am actually pretty interested in the Farm Management Accreditation as well.

I am very knowledgeable about production agriculture, both cow-calf and row-crop. I can tell a person about the components of soil: silt, sand, clay, and loam; soil taxonomy and modifiers, what the cation exchange capacity conveys, soil ratings like CSR2 or NRCS's Soil Capability Classification (which I am currently using for cost approach), what a conservation plan is, different farm programs, difference between a narrow-based, grass back, and broad-based terraces, pattern tiling, implements of husbandry, growth stages in corn and soybeans, estimating yields, costs of production, green-tagging, artificial insemination, heterosis in cattle or crops, death loss, germ rates, the differences in cash and futures markets, basis and merchandising, how to use the board of trade as a risk management tool, futures and options, technical analysis, forward contracting, hedge to arrive, grain bins, legs, conveyors, dryers, axial flow or centrifugal flow, WASDEs and Cattle on Feed reports, the balance sheets of major commodities, how ethanol or trade impacts farm profitability and thusly effective purchasing power, in my specific area, recreational properties and the economic importance of the white-tail deer, lol, or a hundred other things. It's not all relevant to the task of an appraiser, but to be immersed in agriculture at a scale definitely makes me very comfortable with the idea of appraising farm values and assessing the forces that can impact value.

That's great you are still certified in Iowa! I am actually on the prowl for additional supervisors... There are a mere 3 Certified Generals within 60 miles of me... I'm lucky my supervisor even took me on, but I don't know how motivated he is to have an associate... I am to the south of Des Moines: Osceola, Creston, Leon, Mt. Ayr, area is my primary market if you are familiar with that part of the state. My mother was a long-time Farm Loan Manager for the FSA in several counties and I am very familiar with many of the bankers in the area. It's all small community banks down here where I can speak with the bank president himself. There are actually as many farm appraisals, between transactions, trust valuations, and refinance, as there are home appraisals in this area. Not to hock myself out or anything, but I am definitely looking for some knowledgeable mentors to split fees with and help me develop better methodology and consistency in approach. I can get business.... But I digress, another time.

Thanks for your suggestions and advice Sir!

P.S. Something to prove I farm: This is my sprayer one morning on Memorial Day Weekend. It was awful muggy and hot that weekend, there is still a haze in the air at that time and I had to wait for the inversion to lift. I was spraying Status, a safened Dicamba, and a generic glyphosate for rr resistant waterhemp and shattercane in some v4 corn. The inversion can make the dicamba get up and walk around, but the haze makes for a pretty picture, like an add for Deere! lol

 

[Michael Boyd]

I use Redstone for regression analysis, matched paired sales and group data analysis. I rarely find through matched paired sales, group data and or available vacant land sales that an adjustment for lot sizes are supported. I can not believe the typical buyer says "I want this 7,000 sf lot over the 6,500 sf lot." I also use group data analysis for age/effective age adjustments but basically select sales comparable to the subject in age and condition. GLA adjustments, I use different dollar amount per sf to see if the range of values are tightened.