[img_assist|nid=1581|title=San Diego Resale Home: Median Price vs. Sales 24 and 30 Mos. Prior|desc=|link=node|align=left|width=400|height=267]
Sales are a leading indicator of prices, for obvious reasons: low demand = low prices, ultimately. It appears that sales 24-30 months prior foreshadow prices. Thus, it makes sense to keep track of sales, because changes in sales level may manifest themselves in prices 24-30 months later.
Obviously, sales are not the sole leading indicator for prices.
[img_assist|nid=1583|title=San Diego Resale Home: Median Price vs. Notices of Default 36 Mos. Prior|desc=|link=node|align=left|width=400|height=267]
There’s a nice strong inverse relationship between notices of default (NODs) and prices 36 months later. In my simple four factor model (not yet reflecting LS’ greed variable!), lagged NODs have the strongest impact on price.
[img_assist|nid=1584|title=San Diego Resale Home: Median Price vs. Fed Funds Rate 18 Mos. Prior|desc=|link=node|align=left|width=400|height=267]
This is the last predictor (besides time) in my simple four factor model. To me, the ridiculously low Fed funds rate helped explain the run up in prices in the last few years.
The Feds funds rate 18 months prior has the least weight in my model on predicting prices.
It is interesting that the last time the Feds lower rates it did not cause a run up like the recent one. You may need to factor in the availability of ARMs, Option-ARMs and lower underwriting standards.
My conclusion based on my analysis — when NODs start going down and sales start going up, one has a heads up that prices will begin moving up (within 24-30 months).
The tricky part is that when NODs start going down and sales start going up, (1) the economy will still be in a shambles and (2) it will have been after a particularly ugly recession (depression this time). Remember, newspapers and public sentiment are not leading indicators, but lag reality, big time. It will take real guts to buy after the bloodletting, when the only thing telling one to buy are some numbers.
Some may choose to rely on human sentiment to judge whether it’s time to buy. I’m a data-driven guy, and will buy when the numbers — price/rent, NODs, and sales — say things are priced fairly and will be going up.
jg, these are great charts! You could remove the seasonal volatility in sales by plotting “percent change in sales year over year” instead of the absolute number. What would that look like?
Since the median price is a 1-2 year lagging indicator, it makes sense that your data would trail the median price by 2 years. Does the model show the early 90’s downturn, the late 90’s upturn, and August 2004 downturn?
.
