Paired Sales in the Chicago Condo Real Estate Market

Chicago Condo Garden

With all of the discussion regarding the CU (Collateral Underwriter) and using regression analysis to support adjustments, I thought I would share a recent, good old fashioned Paired Sales Analysis (Matched Pairs) to determine an adjustment for a Garden unit vs. a high first floor unit.  The idea behind paired sales is to find 2 or more sales that sold around the same time, in the same location, that are the same with the exception of one feature.  The difference between to the two sales price should give you a supportable adjustment.  The best thing about condo conversions in Chicago, is that when they are converted from apartments, the building and units are often rehabbed and resold around the same time.  Even better, they are often rehabbed by the developer with similar finishes in each of the units.  When the garden unit has the same floor plan as the units above it, it creates a perfect data set to extract a matched paired sales adjustment.  You have similar location (same address), they typically sell with a few months of each other, and are similar in bed and bath count, GLA, etc.

I was appraising a garden unit in on the northwest side of Chicago.  While I was able to find other recent sales of garden units for comparable sales, I needed to use a non garden unit to bracket a particular feature.  After pulling 3 recent garden unit sales, I found out when the buildings were converted and look up the sales prices of each of the units at that time.  Here is the data below:

Chicago real estate apprasier

As you can see above, I have four different paired sales to analyze.  One building actually provided two sets as the two units both sold again recently within 2 months of each other.  The percentage difference ranged from 9.2% – 15.3%.  After average all four percentages, I was able to come up with a 12.5% adjustment for the difference between a garden unit and the unit above it.  When the CU comes back and I am asked why I gave such a large (or small, who knows with CU) adjustment for floor level, I can cut and paste this into the addendum.  Or better yet, just put in into the report in the first place and hopefully avoid the hassle.

While this ended up working out really nice and neat for this appraisal, my fellow appraisers will attest to the fact that finding good paired sales in all circumstances in just a pipe dream.  I just thought I would share one example in which it worked out really well.  I have been testing many of the recent regression tools available for appraisers and have found a couple to be really promising.  Once I learn all of the nuances and techniques, I believe they will be extremely helpful in many situations.  But those of you who have tested them know, they often give some crazy results and other methods will still be necessary.  My favorite so far is PAIRS, by Gandysoft.  Please leave a comment and let me know which software you are finding to be the most useful so far.

Rowe Appraisal Group specializes in real estate appraisals for divorce, estates (date of death), pre-listings and more throughout the Chicagoland area.  If you have any home appraisal related questions, please call us at (847) 863-5776 or email

Social tagging: > > > > >
  • Conrad Meertins

    Nice reminder of the benefits, simplicity and defendability of the paired sales analysis! Thanks!

  • Thanks Conrad!

  • Great blog post Paul. Going way back in time to when the units were converted to find paired sales is creative and sounds like something that I would to do find paired sales. Great job, I wish all appraisers were willing look this hard for sales to support adjustments. The only thing I have to add is that if all the units had the same seller, the paired sale might not be ideal. However, I see that you looked at more than one development, so I think you’re covered in this analysis. The reason that I point this out is because there as a building in my area where the developer priced all of the units at $10,000 increases in price per floor level during the original absorption of the new units. However, when those units resold, the adjustment per floor was much larger, even though overall value had not changed much. The developer in my example had a price that was different than value and it is likely because he was trying to move hundreds of units in a particular amount of time. Maybe the developer in my example needed the upper and higher priced units to move faster than the market average to meet sale projections.

  • Out here is SoCal there is little value difference in the first three floors that can be directly attributed to floor location. The exception is ‘garden level’ that opens to a common pool or interior garden area, these typically command a measurable premium. Above the firest few floors we begin to see a difference in view amenity and can measure it with reasonable accuracy. To what do you attribute the lower value of a garden level unit in Chicago? Keep up the great work #RoseAppraisalGroup !

  • Thanks for the comment Mike. Garden units in the Chicago market are typically less desirable as they are partially below grade, typically have smaller windows and therefore less natural light, are more prone to flooding, and are also less safe as they are easily accessible from the exterior. As far as view goes, they rarely have a beneficial view as they typically face the street in the front, another similar building on each side, and usually just an exterior door in the rear with no window. I guess it goes to show just how different market can be based on geography. I wonder if the CU will know that it should be an adverse location in the Midwest while possibly a beneficial location in California : )

  • Thanks Gary! Your point is well taken regarding builder controlled listing and sale prices when the development was originally offered. That is one of the reasons I included the most recent paired sale for 4824 Hoyne. As you can see when it sold in 2007 there was a 14% difference and then when the two units again sold in 2014 there was just over 9% difference. Definitely a significant variance, but I think with the multiple buildings and multiple data point, there’s enough there to support the adjustment. Thanks again for your feedback. It’s always appreciated and I continue to learn new things and different appraisal techniques from reading your great blog.