Revisiting Case-Shiller and seasonal adjustments
Yesterday, I mentioned that the housing data, while generally positive (the first uptick nationally in 34 months), contained some negative news due to seasonality.
The crux of the problem lies in the strong tendency for house prices to increase in the spring and summer because that is when demand for housing is greatest. Add in the fact that we have seen some distortions due to a rise in distressed sales in higher-income brackets due to the jumbo loan market meltdown. All of this means that one month’s data should be taken with a large grain of salt.
Nevertheless, the increase is the first uptick in prices nationally for some 3 years and it is broad-based. This has to be recognized.
For my part, I don’t like seasonal-adjustments because it adds in statistical noise. My preference is to look at trends via movements in year-on-year rates of change. And the trend here shows a clear plateauing. In January, year-on-year changes were 19.0%. Since then, the year-on-year changes have become progressively less negative. The change is now 17.1% year-on-year.
Does this mean house prices have hit bottom? Hardly, year-on-year declines are still massive at 17.1%. And the graph clearly shows the change in direction is rather slow. But, it does mean that house prices are bottoming because we have seen a trend in year-on-year changes from January through May that looks likely to continue into the present day. If this trend (of less negative data) holds for a few more months, say through September or October’s data, you could say that the housing market has stabilized. By next summer, the year-on-year numbers would be positive.
But, it is early days yet. At a a minimum, I would like to see the next few months data. However, the ultimate confirmation will come through the Winter 2010 data from a far less buoyant time of year. And this won’t be released until next Spring.
I should add that we are talking about nominal prices here, not inflation-adjusted prices where we could see declines for much longer.