Case-Shiller Wants You To Ignore Some Housing Data

On Tuesday the S&P/Case-Shiller home price index will offer its monthly look at what happened to home prices in February. The reading isn’t likely to show a great month, in part because the index uses a three month moving average, and so it will show home price measurements for December, January and February–three of the weakest months for home sales since home prices hit bottom one year ago.

But there’s another reason that the reading may not be as positive as the last few reports: The committee that releases the index has told everyone to stop paying attention to the seasonally adjusted figures for the time being. The reason: The growing share of distressed sales over the past three years had increasingly skewed the formula used to calculate seasonal adjustments, which are used help smooth out regular distortions that arise because spring and summer months are typically stronger than the fall and winter periods that bring weaker sales activity.

In January, the Case-Shiller 10-City composite index registered a seasonally-adjusted 0.4% monthly gain, even though the unadjusted data registered a 0.2% decline, and the fourth decline in a row. Writes David Rosenberg, chief economist at Gluskin Sheff:

Now it would be one thing if January was an unusually weak seasonal month for home prices deserving of an upward skew from the adjustment factors; however, from 1998 through to 2006, they rose in each and every January and by an average of 0.6%. But what happened is that home prices collapsed in each of the past three Januarys — by an average of 1.8%, or a 25% annual rate. And, seasonal factors typically weigh the experience of the prior three years disproportionately so what looks like steady gains in housing prices may be little more than a statistical mirage.

Over at his Matrix blog, New York appraiser and housing analyst Jonathan Miller says something similar. Seasonality in New York City “ran amok” after the financial crisis, he writes: “Contract peak moved forward 90 days for the first time in the 25 years I’ve been tracking the market, from May-June to August-September, which will then screw up year over year comparisons.”

Because we’re no longer in a “normal” housing market, the predictability of seasonal adjustments has disintegrated. Instead of making comparisons to a month-ago period with a seasonal adjustment, the S&P index committee is now recommending that observers make year-ago comparisons where seasonal shifts aren’t going to distort trends.