Don't Sell Everything Online

We are sure you have probably seen the TikToks and memes about people who sold their houses on Zillow well above market value and then turned around and bought them under market value. Yes, it was real and yes it resulted in some issues for Zillow. Housing giant Zillow has had to cut roughly 1/4th of its workforce and close down its home buying company, Zillow Offers, as a result of this phenomenon. What happened? One word: iBuying.

What is iBuying? Exactly what it sounds like: iBuying is an online company takes over the buying process for home sales. In other words, it allows users to sell their homes online relatively quickly with minimal hassle. How does that work? Well, typically the seller will get an offer within 24-72 hours and then a check in the mail within 5 business days of the offer being accepted. This process enables homes to be sold quickly and, in this case, at a significantly higher value. This is part of the reason that sometimes you will see houses listed online and sold in a 24 hour period. Unfortunately for Zillow, this buying and selling method did not go as planned.

What went wrong? According to Zillow, the unpredictability in forecasting homes exceeded that they had anticipated and resulted in too much volatility. Hence they had to shut down iBuying and make labor cuts. Potentially, this issue is centered around a statistical model with extremely wide confidence intervals and standard deviation that allowed for massive variations ranging in the hundreds of thousands instead of the tens of thousands. This suspicion is based on the high range that iBuying was allowing houses to be bought versus sold at. Typically confidence intervals combined with standard deviation will help determine the usability of the model and the "healthy" or expected range for the predicted output is.

In case you were curious and wanted to get a feeling for building your own housing market model, you can find a lot of housing data from the US census website. While it does not go down to the census tract level, it does provide information that is helpful in understanding certain populations. For example, you could see ethnic diversity, approximate income, and education levels within zip codes. You can also use this information (over time and taking phenomenon such as seasonality in to account) to predict what the housing market may do. Of course, do your research prior to putting your model together and also check your variables for high confidence intervals or multicollinearity.

Don't be too harsh on new applicants if they came from Zillow, especially if they are real estate agents. We have also previously written about the current state of the job market and applicants and it looks like there will be a fair number of people now looking for work. More than likely, Zillow as a company will be ok, so nothing to worry about there.