The Art of Adjustments

Each home is unique, just like each of us. Age, updates, square footage, garage count, and other features affect the value of a home. This post explains how appraisers account for each of these differences.

How do you know what adjustments to make to the comps?

The primary method for valuing homes is through sales comparison. This method of valuation uses comps, which are sales of similar homes to the subject. Since no two properties are identical, adjustments to the comp data is necessary. For example, if the subject has a three car garage and the comp has a two car garage, an adjustment is made to the comp to make it comparable to the subject. Each comp is adjusted for every significant difference compared to the subject. The adjustments are to the sale price. After the adjustments are made, each comp has an adjusted sale price. These adjusted sale prices are used to value the subject. One might ask, how do you know how much to adjust for these differences?

There are many techniques available to support adjustments in the sales comparison approach. The two most common are "paired sales analysis" and "regression analysis". Paired sales analysis is the oldest and most recognized appraisal technique for supporting adjustments, and is generally considered the gold standard, but in recent years regression analysis has gained popularity because true paired sales analysis is often not possible. In the following paragraphs, I will provide a brief explanation of these techniques, and explain which technique I prefer, why I prefer it, and how I use it to support my adjustments.

Regression Analysis:

The major strength of regression relates to the large number of samples, and the weakness of regression relates to missing/inaccurate information and non-comp comparisons. Regression generally pulls raw data from large databases. The data is generally not verified, is incomplete, and contains errors. That is why a large sample is needed - It helps to lessen the skewing effect of the incomplete and inaccurate data. Real estate is particularly challenging for regression because it includes a large number of variables, many of which are not readily available. For example, typical appraisal forms include 24 line items for adjustments, and several of these lines include more than one item. Information on quality, condition, views, location (busy road, railroad tracks, golf course, etc.) and many other items are often not available in regression analysis. These unknowns result in skewed estimates of value for the known features (variables). Regression uses reliability indicators that show the likelihood of skewing. In the end, the methodology is very complex, confusing, beyond the expertise of most appraisers, and often unreliable. Were it possible and reasonable to clean-up the data used in the regression analysis, this would be a very useful technique.

Paired Sales Analysis:

In the strictest sense, paired sales analysis determines the value of a particular feature by comparing two properties that are identical in every way, except for the feature in question. For example, property A and B are identical in every way, except that property B has a fireplace and property A does not. Property B sold for $2,000 more than property A, so the value of the fireplace in this example is extrapolated to be $2,000. The problem with this method is that perfect paired samples are rarely available. For that reason, I use a modified form of the paired sales technique, using sensitivity analysis and grouped pairs analysis principles. The first step of this process is to adjust the best comps available, based on an understanding of market relationships from years of studying the market. By looking at the adjusted sale prices after step one, and their relationship to the different features, the adjusted sale price range can be tightened by refining the adjustments. For example, if after the adjustments are made in step 1, the comps with larger home sizes still have higher adjusted sale prices, may indicate that the square footage adjustments are too low. Or, when removing the site size adjustments, the adjusted sale price range is tightened, may indicate no preference is attributed to these differences by this segment of the market. Using this method, each adjustment is analyzed and refined until a reasonable adjusted sales price range is indicated.

Which method do you prefer?

Modified paired sales analysis is simply a more hands on approach using principles not unlike regression analysis. Paired sales analysis uses fewer samples than regression analysis, but the samples are based on the most comparable data after reviewing and excluding non comparables and outliers. The smaller sample means the data can be reviewed in more detail, can be verified and corrected by reviewing public records, MLS data and photos, plat maps, inspections and by talking to real estate agents as applicable and necessary. Smaller samples can be reliable if the outliers have been removed. They are generally much more reliable than larger samples of unverified, incomplete, and inaccurate data which use formulas that are often not transparent, and provide inconsistent/unreliable results that are beyond the expertise of most appraisers. For that reason, I generally use paired sales analysis techniques, as explained above, to support my adjustments.