A Federal Reserve Bank built the closest thing to a bubble detector, and what it chose to call it is the most instructive fact on this page. The Globalization Institute of the Federal Reserve Bank of Dallas publishes quarterly exuberance indicators alongside its International House Price Database, computed with recursive right-tailed unit root tests developed by Peter C. B. Phillips and co-authors and known as SADF and GSADF. In the Dallas Fed's own words the statistics "detect and date periods of exuberance during which house prices (or conventional house price-to-fundamentals ratios) display explosive behavior", and such behavior "may occur when house prices are not based on housing market fundamentals, so these indicators provide a useful signal of emerging misalignments." Read that carefully. It says may, it says signal, and it says misalignment. An institution with the data, the statisticians and the mandate declined to say bubble, and a reader meeting the word in a headline is entitled to ask what test produced it.
What the indicators actually test is the relationship, not the level. The published series covers real house prices and the ratio of real house prices to real personal disposable income. Neither a high price nor a fast rise is by itself the signal; the signal is a price series behaving explosively relative to what households earn. That framing is the useful transfer to any individual market: the question worth asking is not "are prices high" but "high compared with what, and is that comparison getting worse."
Leverage is what makes a housing decline behave differently from an equity decline of the same size. Most homes are bought with a mortgage, so a buyer's own money is a fraction of the purchase price and a percentage fall in the property consumes a multiple of that fraction. A 20 percent fall on a home bought with 10 percent down removes the entire down payment twice over and leaves the owner owing more than the property is worth. Nothing about the owner's conduct is involved, and the practical consequences of that position, which are the loss of the ability to sell, refinance or move, are the subject of the negative equity page.
Three further features slow a housing decline down and stretch it out. Selling a home is expensive and takes months, so supply does not arrive at the market quickly when sentiment turns. Owners are reluctant to accept a price below what they paid and will often withdraw a listing rather than take it, which suppresses recorded sales before it suppresses recorded prices. And a home is also somewhere to live, so most owners simply stay, which is why housing downturns show up first as collapsing transaction volumes and only later as falling prices. An equity market reprices in a morning; a housing market can take years to finish.
The historical record is worth stating carefully, because the famous number is the wrong one. Computed from Robert Shiller's own long-run data series, United States home prices rose about 3.34 percent a year between 1900 and 2000 in nominal terms, and about 0.22 percent a year after inflation. Across the full series from 1890 the figures are about 3.39 percent nominal and 0.56 percent real. The commonly quoted "housing goes up about three and a half percent a year" is the nominal figure, and almost all of it is inflation. Two caveats travel with any of these. A national index describes no individual home and no individual market, some of which behaved very differently. And a price index is not a total return: it omits the costs of ownership, which flatters it, and omits the value of living in the property, which understates it.
The most recent United States episode gives the shape of an aftermath. Real national prices fell about 35.7 percent from their peak in December 2005 to their trough in February 2012, a figure independently cross-checked at about 35.9 percent from Federal Reserve economic data. Six years is the part worth noticing. The decline was not a crash in the sense of a single bad month; it was a long grind, over which the households caught in it were making payments on properties worth less than they owed.
Not every large rise is a bubble, and saying so is not fence-sitting. Prices can rise for reasons that are entirely about fundamentals: borrowing costs falling, which raises what a given income can bid; incomes rising; new construction constrained by land or regulation; or a genuine increase in the number of households wanting to live somewhere. Each of those can produce a durable higher price. The distinguishing feature of a bubble is that the rise is justified mainly by the rise itself, and that is precisely the thing that cannot be measured from the outside while it is happening.
