Zillow (ZG) had the classic, if not archetypal, market advantage. It was the platform where buyers and sellers went to transact, it had home listings before anyone else, and it had access to the network of sellers from across the country. All of this, at least in theory, should have given Zillow a massive market advantage worth millions, maybe billions, of dollars.
This advantage was so obvious and so disturbing (again in theory!) that it inspired a TikToker in Las Vegas (one of the areas where Zillow’s advantage would be most, well, advantageous) to post a video calling Zillow out for an evil act of market manipulation. While he didn’t use those exact words or refer to Zillow by name, it was quite obvious what the target was and what his accusations were.
In short, the TikToker was calling out Zillow Offers, a program where Zillow would offer to buy sellers’ homes directly. Zillow would do this by using A.I. and machine learning to find the perfect price for the home, make sure they were buying below true market value, and selling at that sweet spot. Sounds like an easy way to make a lot of money, doesn’t it? It also seems like a morally dubious way to undercut (and possibly disrupt) the housing market. So Zillow would make a fortune and possibly create a systemic risk from its illicit profit seeking.
This was the popular narrative on Zillow; investors and markets, however, saw things differently. Zillow shares were down over 22% YTD at the end of November, and they had been falling for a long time before the TikTok viral video went live. Then, when Zillow suddenly announced early in November that it was shutting down Zillow Offers after losing over half a billion dollars due to the program, the popular narrative was not only disproven, but not really revisited.
So what really happened is a bit more complicated, a lot more interesting, and massively less insidious or evil than the TikToker suggested. Zillow attempted to be the buyer of last resort in the housing market by using short-term market price fluctuation data to determine the right ceiling for buying and the right floor for selling, essentially providing an opportunity for faster sales to sellers (good) and more inventory for buyers (again, good), while profiting from this market making opportunity. If this sounds familiar, you might be aware of the work of market makers like Citadel Securities, whose activity was lambasted (and misunderstood) publicly when Gamestop (GME) soared in early 2021 in the WallStreetBets-led MOASS (mother of all short squeezes). Some chose to paint this as a moral tale of David slaying the evil Goliath, when in reality it was two amoral market participants clashing (and one losing very badly).
Market makers have been attacked before. Famed financial journalist Michael Lewis wrote his book “Flash Boys” about high frequency trading (HFT) firms, with the heavily editorialized thesis claiming that market makers cost retail investors money (academic studies on the topic, however, suggest the exact opposite is true). While Flash Boys was popular, it was nowhere near as massive as his other books.
This is in large part because the moral crusade against market makers is murky and unsupported by data. While market makers can distort markets and create unfair distortions, this is more the exception than the rule and happens more in the short term; making markets worse is a bad business model! In reality, market makers (as in the case of HFT and Citadel) meant retail commissions ended and bid/ask spreads narrowed; and in the case of Zillow, it simply failed.
So, the first lesson from Zillow’s lesson is that not all corporate market making is evil, even if it takes some counterintuitive analysis to understand why. Secondly, and perhaps more importantly, it also tells us that being a market maker is very, very hard.
Few firms do HFT on markets because it involves very, very complicated math and expensive infrastructure. Similarly, no one had ever tried to be a market maker in real estate on a nationwide scale because of too much noise; there are so many variables and qualitatively driven changes to prices in real estate markets that no A.I. and no dataset had ever been enough to find a predictable and reliable money making signal among all of that noise. Zillow thought they were positioned to be the first to find that signal—and, intuitively, it made sense that, if anyone would be able to do it, they would.
And they failed.
That failure might be because their data wasn’t complete enough. It might have been because their A.I. wasn’t smart enough. It might have been due to a simple oversight—did they not account for the winner’s curse in auction theory, or did they miss some dataset that they have and didn’t incorporate?
Or, perhaps most heretical of all, was it because short-term price movements in the housing market are an unpredictable random walk and there is no signal at all?
Whatever the cause of the failure, they failed—and paid a lot for that failure. Which is a lesson that, no matter how much you might think a company has an advantage, and no matter how intuitively it makes sense that a business strategy will work, there are always many many reasons why your intuitive theory and appeals to reason could be very, very wrong.
