April 2023, by Paul Rayburn

In a recent discussion, the subject of how we select data for analysis came up. It was in the context of removing outliers in price from our subject market analysis. The point made was that we should be thinking more like a buyer and clearly, if the subject was worth just under a million they probably were not looking at 2 million dollar houses. The easy or most obvious answer may be to simply filter by price, or is it? Filtering by price can be acceptable if you already have an identified market segment and can confidently remove those outlier sales from the set. What was not obviously clear from the conversation was just how important it is to not simply filter by price when gathering that initial data and making the initial market condition analysis.

We are taught that one of the first things we do in our process is to identify the relevant subject market segment. The next step in the process is to select comps (or if you’re a data analyst, a competitive market data set) and then make date of sale adjustments as applicable before making any other adjustments if any adjustments are required.

Depending on the market an appraiser is working in, they may have a lot or a little in terms of quantity in sales. Sifting through thousands of sales may not be practical and one of the most obvious ways to do this is to filter the data, and the price seems like a reasonable place to start. Most anybody even remotely familiar with a market can come up with a probable range of what the subject market may be in. Let us say we are looking at property that is worth about a million dollars. Surely, most buyers of a million-dollar home would not be looking at two million-dollar homes. So then if we are supposedly reflecting what reasonable market participants are contemplating then we should be producing credible results. Would that million-dollar buyer look between $900,000 and $1,100,000? well, that’s reasonably probable.

Here’s the problem with filtering by price before making market condition adjustments. That decision could significantly skew relevant date of sale market trend/ condition adjustments. In the following example, I chose a large market segment. This data set was generated purely for this discussion and not specifically for how I would analyze a particular subject market. Generally, the more data points available the more consistent the data. I chose several neighbourhoods around the Central Okanagan that were reasonably similar in appeal. I chose a fairly long date range and some might question that, but in my market and many other markets we often run into assignments where we need to or should look at comps that are one, two years prior sales, or even more. To avoid complex issues of market trend trajectories I choose a market date range where the values were consistently climbing to keep this example reasonably simple. Although the early part of 2020 saw some minor price declines, they very quickly reversed course. I might have filtered to June 2020 and the results would probably be even more accurate. Market condition adjustments can be one of the most reliable adjustments if made properly. Unfortunately, most major lenders request sales within 90 days or maybe 180 days, and that has pressured many appraisers into relying only on very recent sales however, for obvious reasons that is rarely practical, particularly in the markets like we are currently in, some where there are almost no sales and in other markets there are no sales, so we may need to look to historical trends across larger data sets to find actual comparable data.

The best way to present the issue is by visualization of the data we are filtering. In the search field boxes the simple input of max price $1,100,000 and min price $900,000. seems fairly innocent but when presenting the results visually we can see the issue. I did purposefully choose this price range as it was the most egregious example and visually striking. Obviously, any appraiser or other data analyst worth their salt would find a price change of around 2% over the last three years to be an error. If we expand the price range to $750,000 to $1,500,000 we get a slightly more plausible rate of 37% but that is still not accurate. So what is going on? When we plot the data out you can see what is happening here.

In the following, Chart 1 is all of the data even remotely indicative of a market. The formula in the lower right corner of the charts is the price per day from the slope of the data. For example in chart 1 is $693.03 per day and our data date range is 850 days so that’s 82%. That indicates an 82% increase over the period but that is a very broad market and may not really represent a more conforming market so as noted we want to refine that market.

In Chart 1a we can visually see what data was left out by filtering by price alone. The data in the triangles contain significant and relevant data which would be excluded by searching this tight of price range initially.

In Chart 2 we see the issue of filtering by price graphically presented and the calculation is 2%, which obviously not very useful.

Chart 2a is the modified broader search by price showing the 37%.

In Chart 2b is the data filtered by price down $750,000 to $1,500,000 and then filtered to a supposedly more competitive market segment of 2,000 to 3000 square feet and lot sizes .15 to .5 acre. and year built 2005 and newer. The other parameters may or may not be valid for a given market but for the simplicity of this demonstration, we can use those parameters as at least one method of paring down the data. This is still skewed by the initial price filter but does indicate a more plausible adjustment of 60%.

If we filter the data first by those parameters of physical similarity, we can get a more reliable initial data set and then, if need be, remove outliers with a more surgical approach. Chart 3 is filtered by building size and acres with 75% adjustment.

Thats probably a fine set for date of sale adjustment but there is still an issue of brand new homes bringing up the rate because they did not exist at the beginning of the set so filtering by year built might give us a more similar market trend adjustment and also eliminate a few more sales to make the final selection of comparables simpler. The final set in chart 4 is filtered to year built between 2010 and 2017 bringing the adjustment down to 73%. This now gets us to a more manageable set for individual analysis of specific properties and similarities with the subject. The outliers were also eliminated by the filtering in this case and did not need to be removed individually although, on occasion, that may still be required. It is possible the subject is similar to one of our outliers but we can still analyze those in conjunction with this more reliable date of sale condition-adjusted data. We may also search for more specific characteristics and even include some comps that we left out in the approach used for the market trend date of sale adjustments with the ability to compare to our adjusted data.

The HPI for this same period is 60%. An appraiser may be faced with the challenge of deciding if the HPI is reasonably accurate for the subject market. The average HPI price for this broad market is $971,000 but does it represent the relevant market for the subject? Looking at more refined markets may be too limiting. The HPI is not adaptable for size, quality, or condition, and it only provides an adjustment for the most common homes in that market. If you were analyzing a $600,000 property, a 2 million dollar lakeshore home, or vacant land, that would not be well represented by the HPI and on occasion, the HPI has produced erroneous results, after all, it is a program designed by humans, and errors are possible. Therefore, we should have our own methods for comparison.

With advances in Automated Value Models (AVM’s) taking away much of the lending business for typical properties, appraisers are faced with the increasing challenge of analyzing more complex markets. However, we also have, or should have, increased access to data. Our ability to interact with that data in a meaningful way has become increasingly important.

Leave a Reply

Your email address will not be published. Required fields are marked *