Uncategorized May 9, 2023

IRES & RECO Data fidelity

A recent discussion between a few of us market watchers concerns our data integrity and what sources we should be using. Should we use data only from the IRES system or should we use data from both systems with the de-duplication system enabled. In most of my stats, I only track IRES data. This is a vestigial decision from the days before it was a possibility to include data from both systems. With IRES only data going back to 2004, some of which cannot be re-created, I’m hesitant to now include REColorado data with duplicates removed as that may invalidate some comparisons to previous years.

After my recent discussion, I felt I should at least look at how much data I might be missing by ignoring RECO data. How many duplicates is the IRES auto de-duplication system missing and why? How much of what is happening locally is being reported only in REColorado? Like most of the stats questions I ask myself, the answer is muddy and harder to reach the more closely you look at the data.

The first surprise, the problem of duplicates in the system seems to have gotten much better since the last time I took a detailed look. Less than 1.4% of the homes on the market are duplicated between the systems and not automatically removed by the de-duplication algorithms. This is much better than my recollection. I wonder if this can be attributed to the better data share agreement between the two systems and brokers feeling they no longer need to belong to both and enter homes into both. When looking at the Q1 2023 sales and current listings, there were only 29 duplicates not removed that should have been, an error rate of 1.22%. When I looked at the reasons for these erroneous duplicates not being removed, the most common reason was bad data entry by brokers, then the mismatch between A/B and P statuses between the two systems and finally Attached Homes were more likely to have duplicates due to how the two systems sometimes format addresses differently.

The dividing line between the two systems remains Broomfield, where just more than 75% of the listings and sales are reported solely in REColorado. Areas with more new home sales have higher percentages of REColorado only listings. Charts for Active listings and Q1 sales by category below.

I still feel that the long-term record of IRES only data I have is worth maintaining. The somewhat large numbers of RECO only listings and sales though has me continuing to ponder that decision. I’ll probably keep the status quo for now and take another look in the future and see how the mix is changing over time. Someday there will be a big enough change of some sort to have to give up on the idea of data consistency and take a more expansive, but possibly less accurate, view of our local market.

Nice to see everything outside starting to turn green, a good sign!