Last week, Zillow Group announced the winding down of its iBuying business, which came as a shock to many in the industry. What could have gone wrong at one of the biggest names in the real estate world? And what does it mean for other companies?
Zillow bought too many houses at high prices and then couldn’t sell them at a profit.
What is “iBuying”?
In modern times, people often need to sell quickly, either to cash out of a house they don’t want or to relocate quickly to another part of the country for a job. For decades, there was one ingredient to a quick sale: selling at a low price to an investor, off-market. In the old days, that might have meant calling the number on a “we buy houses” sign on a telephone pole, and getting a quick offer for 30% off the market price. The iBuyer model improved on that by offering sellers competitive prices much closer to what they would get by listing with an agent.
The secret ingredient: technology and optimization. Namely: an automated valuation model, a reasonably good forecast of what prices will look like in 3-6 months, streamlined procedures to complete necessary repairs, and finally, a huge number of purchases. The last part means buying the “right” houses.
Automated valuations were too aggressive
Zillow’s automated valuations were too aggressive, causing them to buy at prices too high. Automated valuation models (AVMs) are algorithms that estimate the market value of a home. This works well for homes that are in common, standardized formats in highly homogenous areas. Think of the typical Sunbelt city and the vast tracts of similar homes.
Zillow uses sales and listings (i.e. asking prices) in its valuation model. The goal was probably to capture fast-moving market conditions, given the time it takes for homes to sell and reflect market conditions. However, this presents some major problems.
First, anyone can list his or her house for any price, regardless of market conditions. Despite the best efforts of agents to discourage the practice, some owners inevitably start with a high asking price and work their way down until they get offers. A listing of $50,000 over market across the street is not meaningless information, but it very well could be irrational and therefore not that informative. All too often, nearby listings tended to inflate Zillow Zestimates, and so contributed to Zillow buying homes at prices too high.
Forecasting with the crystal ball
In Zillow’s Q3 letter to shareholders, forecasting was cited as a major failure, writing that “We have been unable to accurately forecast future home prices at different times in both directions by much more than we modeled as possible.” The idea is to be able to predict prices in 3-6 months with enough accuracy to make money on the re-sale after repairs. At least, to get it directionally accurate. That was hard to do during this year’s fast-changing market conditions.
In the two quarters before closing down the iBuying business, Zillow bought more homes than it had in the last two years combined, according to MarketWatch. When the market was going “straight up” in most markets earlier this year, a buyer could almost could not go wrong. But then, price growth slowed, leaving a lot less room to make a profit, especially with repair costs and holding costs going up (see below). Also, as sales growth slowed, it became harder to sell the houses.
When a traditional investor wants to “flip” a house, they are planning to put some work into that house. That means replacing the kitchen, bathroom, and completing deferred maintenance on items such as the roof, water heater, and HVAC system. As a result, flippers have to buy low enough to make a profit. They may also allow a few months to complete work on the home. Labor shortages have made this more challenging, causing delays in finding workers to complete repairs for homes that were “iBought.”
This is especially true if too many of those homes required more work than expected. A computer model has difficulty assessing the interior condition, which can also vary widely, particularly if buying homes that are outside of a strict “buy box” where the odds of a somewhat decent interior condition are somewhat controllable.
Buying the “right” houses
The law of large numbers suggests that when dealing with a very large volume of home purchases, it’s a numbers game. Out of a random 1,000 purchases, most of them will work out without being too selective, as long as the market is going up. But what if it’s not? Then you have to get more picky. That means picky about local market conditions, local market risks, and home conditions.
Avoiding problems by using market condition data
While market conditions at the metro area may average out one way or another, this masks some very sharp differences in market conditions at the local level. By over-expanding, it is possible that Zillow was not fully equipped to account for these differences. We have a block-level index called the “short term investor index” that is specially formulated for short-term holds. It is based on several factors:
- Performance: recent appreciation in the area
- Demand factors: proximity to amenities and jobs
- Supply: limited new homes nearby (which compete with your home)
- Risks: trends in neighborhood deterioration in the last 3 years: education, crime, income; risk of loss from hazards such as fire and flood.
iBuying smart: avoiding downside risks
The important part is protecting from downside risk. When the market is surging, rising tide lifts all boats, until it doesn’t. What are the houses that will sell in “any” market, not just the “best” market? Some factors for an iBuyer to consider:
- Inventory in the area. What is the figure and how is it trending. Have we reached an inflection point?
- Access to jobs, shops, schools, highways, and (sometimes) transit?
- Health and safety risks
- Changing neighborhood conditions
- A diversified job base
These are all factors that we analyze in Locate Alpha.