Wednesday, June 25, 2014

Case Shiller Index drops 85% this month


The media headlined the 11% increase in the index and realtors will happily use it to reassure buyers. But the real annual increase this month is 2.4% down from 15.4% last month. This index is volatile and noisy but a single month drop of this magnitude is unprecedented.  Either the index is right and something has happened in the housing market the users of the index need to understand or the index is weak. David Blitzer of S&P which publishes the Case Shiller decided that this month was the time to discuss the big picture of housing performance over the last several decades. This reflects a stunning lack of confidence in his index, a limited understanding of where housing is going and prudence.  However the index is not wrong, the market is changing.  

The monthly change shown in the graph below, illustrates the real volatility in the market.   This variation is a reaction to changes in the market. House prices fell in 2011. The market reacted to in restoration of lower FHA loan limits in November 2011 and rebounded, peaking in May 2012.  Prices declined in June but the euphoria about price increases lead investors and buyers to pile in, creating a peak in February 2013.  The investors started to leave at that point and rate of price increase declined. 
 The 12 month trend line is more realistic than the year over year but it is not reasonable given the variation of the data. The trend projection for next month is 7.5 which would require a sharp increase.
In June the most recent data is the April reporting.  The April closings consists of deals most of which were negotiated in February.  Deals in the previous month, March reporting for January  were at an annual rate of  15.4 up from the 9.8% negotiated in December. Short sales became more costly to the seller in January because the seller had to pay income taxes on the forgiven mortgage debt.  So there was a rush to close short sales before year end followed by homes taken of the market in January and February. However this is only a partial explanation for the erratic price data.

Further explanation can be developed by looking at the price changes of the individual cities in the following table, which compares the monthly increases, at seasonal adjusted annualized rates, for the March and April reporting.

These cities are grouped together with other cities that move in similar patterns, for ease of understanding.  For example Los Angeles grew at an Annualized rate of 14% in March but only 1.2% in April. The three California cities grew at only 0.5% in April 16 percentage points below March. The decline in the California made up 38% of the 20 city index drop in April.  After growing rapidly in March, Washington and New York decline in April caused 27% of the decline in April index.  Thus these five cities account for 65% of the change in the index.
The history of expansions in the U. S. is that at the end of the contraction the price is back to where it started in real terms unless the city is in economic decline or rise.  Mortgage rates, tighter credit demographics, student debt etc. all affect prices. But the three California cities are 90% above the start of the expansion.   Washington and New York average 50%and Dallas is 5% below. The high priced area will turn negative while other cities will do well. The Case Shiller index will turn negative, probably this year.




Saturday, June 21, 2014

The Pure statistical projection of Case Shiller




Suppose you are not satisfied with another opinion about what happen house prices. What would a statistical projection look like? The answer is illustrated below.  After 18 months of decline, the Case Shiller index turned positive in January 2012. The year over year comparison of growth looks very benign, growing very rapidly in 2012 and then growing more slowly in 2013 before turning down slightly.


The monthly change illustrates the real volatility in the market.  The market reacted to the restoration of lower FHA loan limits in November 2011 and rebounded, peaking in May 2012.  Prices declined in June but the euphoria about price increases lead investors and buyers to pile in, creating a peak in February 2013.  The investors started to leave at that point and rate of price increase declined.  This process was hidden from the view of market participants because the data lags events.  For example in June the most recent data is for February.  More importantly, the widely followed year over year comparisons masks this activity.  If you don’t understand why the monthly numbers vary, how can you forecast the future? 

The Case Shiller index is on the right scale, while the percent change in the monthly annualized and the year over year is on the left scale. The trend was calculated using 12 months of data. The most recent month was excluded. If it was included the trend line would decline more slowly.  Case Shiller data comes in three batches. So the first month of data is revised in the next two months. The change in the second month, when the majority of the data is available generates the largest change because the prices in the first month’s batch tend to differ from the average.

The 20 cities in the index vary sharply from the average.  Some of the cities move together, but still individual differences are still large.   The three California cities are declining more rapidly and the resort cities that were doing well are growing more slowly, while some cities that declined with the weak manufacturing industry are showing improved results.
  
 Improved employment and the housing inventory do not have much impact on the index in the near term but the prospect of a gain in the price of the house does.  This does not mean that all buyers are looking for gains, but when the prospects of price gain appear enough additional buyers, attracted by the improved investment potential, enter the market increasing price. It also works in reverse.  Price declines cause people to postpone purchases driving prices down. This leads to higher variability in higher priced markets.  
   

Because the variation up and down is so great in this period the regression is low – 38%.  However the trend is the best indication of the central tendency of the index.  The trend of the last year is an annual decline of 0.7% per month turning negative in March of 2015 reporting with wide swings in monthly results.  These large swings up and down are characteristic of a change in direction of an index which has yet to stabilize.  Thus the monthly results will vary widely from the central tendency until the market becomes more stable. The central tendency is like the forecast of a hurricane.  The cone of uncertainty increases the further out the projection goes.