Case-Shiller Housing Index: 18 Month Forecast

This is the 4th in my Case-Shiller Housing Index series. If you want to start at the beginning here is the 1st, 2nd, and 3rd in the series.

In the previous post on this series, I developed and demonstrated the seasonal ARIMA model I developed to fit the Case-Shiller 10-city housing index. One important concept in doing statistics and developing statistical models is that the model you end up gives you insights into the process that produced that data. The statistical forecasting model for the Case-Shiller I think fits very well and the fact that it does fit so well suggests some specific conclusions:

1. When this type of ARIMA model fits, it suggests that housing prices as measured by the Case-Shiller index have what can be thought of as inertia or momentum.
2. This means that once housing prices are moving in one direction, they tend to stay moving in that direction.
3. If a change in direction occurs, it occurs over many months or even years. Housing prices (Case-Shiller index) can't just abruptly move reverse course 15% in a month. If they are to change direction, it will take a while to occur.

So Where are Housing Prices Going?

So this graph shows what the ARIMA forecasting model forecasts for the next 18 months. The red series is the forecast and the blue lines are the 95% confidence intervals. Bottom line: the forecasting model sees further declines in the housing market. Specifically:

1. 1.8% drop to 151.82 in the next index release (coming out in the next few weeks).
2. Due to the seasonal effect, the rate of drop of the index will slow down to near 0.5% or even 0% in May, June, July, and August.
3. In April 2010, the index is forecasted to drop 16.4% down (compared to the Feb '09 index) to 129.23. This looks like April 2010 could be the bottom in the housing market but really too early to tell.

So of course this forecast is not exact. There is certainly uncertainty. This model works very well when the trend is consistent and as the trend shifts, the forecast will evolve over the course of a few months, in fact this forecast has shifted up some over the past few months which is why I am thinking the bottom could be coming next year. But the confidence intervals give you an understanding of how much this forecast could shift.

What is the Best Case Scenario for Housing Prices?

The upper 95% confidence interval gives a good estimate of the best case scenario for the Case Shiller. If home prices put it in reverse now, that gives you an understanding of how quickly they could really turn up.

This best case scenario has the Case-Shiller 10 city index trying to bottom soon and moving mostly sideways between now and April of 2010 and then making progress upward.

Remember, this should be viewed as a best case scenario. According to the forecasting model, the most likely scenario is still down for the next year. Like everyone I would like to believe that housing prices have bottomed but the reason I like statistical methods for forecasting is that they are as objective as it gets and they are one of the best tools to cut through the caveman bias.

Next I will examine how my forecast compares to some other forecasts and then after that, I will examine the new Case-Shiller ETF's coming out next week from MacroMarkets UMM and DMM.

...