A thriller about Zillow | The Economist

TIT TIME was fit. On November 2, just two days after Americans celebrated their scariest annual vacation, news of a suspicious death shocked the stock market. Zillow, a giant real estate and tech firm, has said it will shut down its massive instant buy, or i-buying business, which uses big data and algorithms to bid on homes in dozens of cities across America. , then sell them quickly on. The company expects to lose more than $ 500 million in the second half of 2021 after overpaying for thousands of homes. It will lay off a quarter of its 8,000 employees. It seemed like a business that should be in bad shape. All in all, it’s been a fantastic time to buy a home almost anywhere in America, if you can just pick one: Home prices have climbed 16-25% in the past 18 months. So why is Zillow’s online shopping business in the doldrums? And thriller?

Finding the right suspect is important for more reasons than the fate of Zillow itself. The i-buying business is one of many examples of companies using a platform to collect big data, analyze it using advanced techniques, and strengthen their algorithms to make a market run more smoothly. This trend has lowered transaction costs in many asset markets, from stocks and bonds to photographic equipment and clothing. The fate of Zillow’s online shopping business could indicate that using technology to buy and sell something as idiosyncratic as a house is too far away.

Consider the most serious suspect first: the real estate market. He was in an unusual state of flux. Initially, the covid-19 pandemic caused a freeze on all real estate transactions. Then the prices went crazy, increasing to all-time highs year over year in April. Without a doubt, volatile prices do not favor algorithms formed on historical data. Yet, in theory, higher prices should help i-buyers by making it harder to sell a home for less than what has been paid. The reverse, falling prices, might be a more likely culprit, but so far the data is mixed. A house price index compiled by the National Association of Realtors (NAR) finds that values ​​peaked in June 2021, 19% above pre-pandemic levels, and have since fallen 2.8 percentage points. another by S&P CoreLogic Case-Shiller suggests that prices continue to gallop. Both are published with a lag (the NAR runs until end of September, Case-Shiller until end of August), which means the evidence is inconclusive.

The next suspect is mathematical models. A handful of companies offer online shopping services, the first and largest of which is Opendoor, founded in 2014. They charge a fee for the services they provide: buying and selling homes immediately, hassle-free. The rapid entry and exit makes them more like market makers than real estate investors, who buy to keep. To be successful, online shopping businesses need two essential pieces of information: the current value of a home and a price forecast at the time of sale, usually two to three months into the future. To understand them, they need treasures of data, ranging from the precise location of a house, to the number of rooms it has, to the swimming pool or not. They compare them with the closest comparable homes that have been sold recently and look at recent trends to make a forecast. This allows them to make an “instant” offer to an owner. In the past, algorithms seem to have worked quite well. Mike DelPrete of the University of Colorado found they were offering homeowners about 1.4% less than market value, which isn’t a bad result for a quick, hassle-free sale.

Zillow boss Rich Barton said the big deal was the company’s forecast. He said he found himself unable to predict prices in three to six months. In particular, Zillow appears to have projected much more favorable conditions than those which materialized. In Phoenix, where house price appreciation has been particularly rapid but appears to be slowing, Zillow is offering housing on average 6.2% less than it paid for.

This problem is exacerbated by the inconstant economy of adverse selection. While the algorithms of online buying companies are great at valuing homes on average, they only need to be lagged a bit for the risk to tip down. Homeowners are unlikely to sell their home for much less than they think they can, but they will happily settle for a higher price than expected. Mr Barton revealed in a letter to shareholders on Nov. 3 that “higher than expected conversion rates” were part of the problem. A former Zillow employee claimed the company wanted around 50% of homeowners who asked for an offer to accept it, but up to 74% of offers made in recent months were accepted. Zillow bought nearly 10,000 homes in the third quarter, more than double the amount in the previous three-month period, which was itself more than double the amount in the first quarter. It is suspected that Zillow’s algorithm was making overly generous offers and that owners were rushing to take advantage of them.

This may have been a Zillow issue, not an i-buyer issue, however. Some of Zillow’s competitors seemed to realize before Zillow that the market was losing ground. OpenDoor and Offerpad, an i-buyer founded in 2015, both started bidding more cautiously about valuing their models around July, as price appreciation began to cool. When they announced their earnings on November 10, neither Opendoor nor Offerpad presented any issues similar to Zillow’s.

Interior work

Perhaps the killing blow was, in fact, self-inflicted. Zillow has aggressively grown its online shopping business. Opendoor has grown gradually. It was offering i-purchase services in just six markets after three years, taking the time to refine its algorithms. It is now operational in 44 markets. Zillow added almost as many markets in half the time. A former Zillow employee said Business intern that the management had wanted to catch up with Opendoor, the favorite. In order to compete, the employee said, the company pushed to offer generous offers to potential customers. He called it “Project Ketchup”. Now he’s got his own fake blood on his hands.

This article appeared in the Finance & Economics section of the print edition under the title “Home-icide”

Source link

About Marilyn Perkins

Check Also

Scottish banking pilot brings hope to branchless UK shopping streets

When Cambuslang in Scotland lost the last of its bank branches four years ago, the …

Leave a Reply

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