Tuesday, June 30, 2009

Case-Shiller Housing Index: Apr'09 Data and Forecast

Posted by CavemanForecaster at 9:32 AM
.
The Case-Shiller Housing Index came out this morning with the April '09 data. Here are some highlights and below you will see my updated chart along with an updated 18-month forecast based on my ARIMA model.
  • The 10-city composite was down just slightly compared to last month's data (March data) at a drop of 0.7%.
  • The year over year change for the 10-city composite is 20.6% down.
  • A number of cities (8 of the 20 tracked) posted positive gains for April over May. Those showing month over month gains were: San Fransisco, Denver, Washington D.C., Atlanta, Boston, Cleveland, Dallas, Seattle. Keep in mind there is a positive seasonal effect this time of year but it still is looking like a promising trend from last year.

Here is my updated 10-city Case-Shiller chart along with my 18 month forecast:

The forecast model shows prices to be just about flat the next few months. I see it as very possible we get a positive month over month change in the 10-city composite in the next few months which I think will make some big headlines and probably spur some potentially false hopes that the housing market has bottomed. Keep in mind the strong seasonal effect at play in this time of year.

The ARIMA forecast model is showing a year from now being down 11% from here at a point slightly higher than my last forecast I published a couple of months ago. But overall, the forecast model going forward is looking stable and consistent. The best case scenario looks like a roughly flat next year and going up from there as shown but the upper confidence interval on the forecast.

I will post some more for tomorrow looking more into individual city trends, etc.

For more of my past posts on the Case-Shiller including alot more background on my analysis methodology see here.

Labels: ,

4 comments:

  1. How far out are you willing to trust the ARIMA projections? It's my cursory understanding that they are best used for very short-term forecasts (obviously, any forecasting model should be better for short term projections, but I thought this was particularly true for ARIMA). 18 months seems like it's pushing it. Another interesting or useful model could be price to rent or price to income ratios:

    http://www.calculatedriskblog.com/2009/05/house-prices-real-prices-price-to-rent.html

    Although rents and incomes are falling, so that might hurt the ability to project home prices based on such ratios. Also, I don't think incomes are available on a monthly basis. But rents and incomes are believed to be two of the "fundamentals" of home prices.
    ReplyDelete
  2. Dave,

    It depends on the data, the data set, and what you want to get out of the ARIMA forecast. ARIMA models are certainly best for shorter term models but in the case of this ARIMA model for the Case Shiller, I have more confidence in it. It is one of the nicest ARIMA models I have created. You can look back at my previous posts on it. But I would not hold out for a specific future forecast value precisely but I believe this model can give a heads up on the general behavior and trend and hopefully give a heads up 6-12 months out when the Case Shiller is really changing trend. At least it did in the past and we will see this time. But certainly keep in mind the confidence intervals which is why I showed those on my forecast chart. But also keep in mind those are 95% confidence intervals so they will be wide.

    To summarize, I feel pretty confident in the general trend of the ARIMA model which is to expect home prices to be pretty flat in the next few months and then drop again into next March/April/May to new lows and then that could be the bottom. But in a few months it will be interesting to see what the updated ARIMA models say. I also find alot of value in the "best case" as I said above. The upper 95% confidence interval is the best case scenario that is possible and I feel pretty confident about that as well.

    I have thought about fitting a model to price to rent ratios and inventory as other ways to forecast. The other way is by looking at individual markets which I will post about tonight. I think of these will help to come to an estimate of where prices are going.
    ReplyDelete
  3. I thought of one more thing...

    Fannie and Freddie lifted a moratorium on foreclosures in April that had lasted about 4 months. The lower number of foreclosures during that time may have a distortionary effect on prices (with a lag to allow for the foreclosure process), which may have otherwise been lower due to a higher housing inventory. I have no idea to what extent though.

    http://washingtonindependent.com/37160/fannie-freddie-quietly-lift-moratorium-on-foreclosures
    ReplyDelete
  4. Dave, I do think the foreclosure moratorium has had an impact on home price data but I have not real clue either how big the impact has been. We may see an impact next month when the May data comes out. Should be interesting.
    ReplyDelete

Subscribe to: Post Comments (Atom)