When to pay attention to Case-Shiller and when not to

One positive byproduct of this real estate/financial crisis has been the elevation of the S&P/Case-Shiller Home Price Indices to the status of standard measure of housing prices and the relegation of the National Association of Realtors' average and median sales price numbers to sideshow. This is mostly a good thing, because the NAR numbers are skewed by what kind of houses are selling. If lots of high-price homes sell in a month and few low-price ones do, the average and the median price go up for the month even if all those houses are selling for less than they would have a month before. Economists Chip Case and Bob Shiller found a way to get around this problem by comparing sales prices over time for the same houses, then making adjustments to get a monthly figure. As a result, the house-price bust showed up in the Case-Shiller data long before it did in the NAR's, which were skewed by the fact that, early in the decline, high-end markets held up better than low-price ones.

Compiling data the Case-Shiller way takes a little while, though, which leads to strange reports like this one today from Reuters:

U.S. home prices in July rose for the third straight month, surpassing forecasts and suggesting that the housing market is stabilizing after a three-year plunge. ... The data relieved investor concerns about the impact of a weak housing market on the economy and U.S. stocks opened higher.

Hmmm, where did these concerns come from? Well, in part from last week's existing home sales report from the NAR, which showed an unexpected drop in sales and a decline in average and median prices. Not in July, mind you, but in August. The NAR's figures are simpler to compile, so they come out just a few weeks after the end of the month in question. The Case-Shiller numbers released today, meanwhile, are almost two months old. Now I don't know how meaningful the bad news from the NAR will turn out to be—there's a lot of noise in these numbers. But Case-Shiller data from July shouldn't really dispel any concerns raised by NAR data from August, should it?