What the housing market statistics hide from us

Every week we’re bombarded with numbers that tell us how the financial world is doing. House prices are up or down. Shares gained. A company flourished.

But those figures leave out some of the most important information: the houses that didn’t sell, the investment that failed, the company that disappeared. Yet their absence can make markets, policies and even our own lives look healthier than they are.

Economists call this ‘selection neglect’, or the tendency to assume the information we can see as if it were the whole picture, when much of reality has already been filtered out. It’s like trying to find out what matters politically by talking only to your friends, assuming everyone else shares the same concerns.

A surprising place reveals just how powerful selection neglect is: fine wine auctions. Wine is hard to value, sceptics cannot bet against it, because there is no way to short sell a case of Bordeaux, and crucially, when a bottle fails to reach its reserve at auction (which happens on an exact day), that failure is recorded.

This meant I could measure what the official record leaves out, and so I assembled more than a million auction transactions: nearly 4000 wines, 58 auction houses around the world, going back to the late 1990s.

The first thing I learned is that failure is not the exception. At Langton’s, Australia’s leading wine auction house, almost half of the fine wine offered in a typical month does not sell.

The second thing I learned was what that filtering does to the numbers. By estimating what the unsold bottles were likely worth, I found the typical wine is worth about 20 percent less than the sale records suggest. The prices we see are not a fair sample of the market. They are the market’s highlight reel.

The bias predicts the future for potential investors. Wines whose recent sale prices flatter the auction house go on to earn sharply negative returns over the following months, as reality catches up. A wine that fails at auction subsequently underperforms similar wines that sold at auction. The silence of a failed sale was information that was there all along; it just takes a while for the market to hear about it.

This is not a quirk of the fine wine market. In other work, my co-authors and I have found the same pattern in share and corporate bond markets from over a century ago.

You don’t need to care about wine to care about this. If we translate this idea to real estate, the machinery is identical. A house, like a bottle of wine, only enters the statistics when it sells.

Think about a cooling housing market. Sellers rarely respond by immediately cutting their asking price. Instead, many simply withdraw the property and wait.

Because the house never sells, it never appears in the official sale-price statistics. The only properties recorded are the ones that successfully sold. As a result, the median sale price can look surprisingly stable, when demand is weakening.

Meanwhile, the auction that passed in on your street told you something real about what buyers are willing to pay, but it won’t appear in any official figure. Failure carries information, but our statistics are built to exclude it. A market could be considerably weaker than any published number suggest, because the failed sales have disappeared.

Once you notice the filter, you see it everywhere. Fund managers advertise the track records of surviving funds; the duds disappear from the averages.

Social media amplifies the good news stories; turning $10,000 into $10,000 trading shares and so on. You won’t read about the people who lost money trading, whose businesses failed, who exited the market. Business books distil the habits of wildly successful companies without asking whether failed firms followed the same advice.

Politicians point to the visible successes of their policies; the counterfactuals never hold a press conference. You won’t see media headlines about the plane that lands safely.

None of this means the numbers are lies, or that property or wine is a bad investment. It means the numbers reveal more insights than the answers to the questions we’re asking. “How did the things that sold perform?” is not the same as “How did this market perform?”, and the gap between the two is widest precisely in markets like housing, where sellers can simply refuse to sell at prices they don’t like.

There is no app that strips out this filter. But there are habits worth developing, and it costs nothing: don’t just look at the outcomes you can see. Ask what conditions were necessary for those outcomes to appear in the data in the first place. Which people, properties, businesses, investments, wines, or experiences are missing because they failed, withdrew, disappeared, or were never recorded?

In wine, in housing, in careers and companies: the most telling price is often the one nobody paid for, and nobody wants to talk about.