Friday, July 18, 2008

How Much Further Does House Prices Have To Fall?

A question many people ask at this point is how much further does U.S. housing bust have to go before it bottoms? If we look at residential investment spending, it has already fallen from its historical peak of 6.3% of GDP during the second half of 2005 to 3.8% during Q1 2008, a number which will surely fall further during the second quarter. This also means that it is already below its 1980-2000 average of 4.3% of GDP. However, during previous recessions in 1982 and 1991, residential investments fell as low as 3.2% and 3.4%, and with the greater inventory overhand, it is likely to fall below that level this time.

Perhaps even more interesting than how much further the residential construction sector has to fall is how much further house prices have to fall. Further significant decline in house prices could threaten not just the 3-4% of GDP constituted by residential construction, but also threaten consumer spending, not to mention bank stability.

There are of course different ways to assess just how much house prices have risen, and therefore implicitly how much they might have to fall to once again make them reasonable and sustainable. Since there are two competing price indexes, OFHEO and Case-Schiller and since they diverge significantly, one has to first decide which one of them to use. As I've written before, OFHEO probably understated both the increase during the boom and the decline during the bust while on the other hand Case-Schiller likely overstated the increase during the boom and the decline during the bust. Because however I think the OFHEO index is probably somewhat closer to the truth and because the net worth statistics issued by the Fed is based on it, I will use OFHEO.

One is to look at nominal house prices, but I would argue that this is misleading because of the significant general price inflation and the significant increase in nominal income levels.

My favourite indicator is instead house values to disposable income, which I used in the article from November 2004 when I predicted the current bust. I then pointed out that historically, housing values have tended to fluctuate between 135% and 150% of disposable income, but that the value during Q2 2004 (the latest number available during that writing) had risen to 184%. I now see that the values have been retroactively upwardly revised somewhat, but it still wasn't higher than roughly 188% with the current data, with the historical range being more like 138 to 153%. It was 183% in Q4 2003 and 192% in Q4 2004. It reached a historical peak of 206% during 2006. Starting late 2006, and continuing at an accelerating rate during 2007 and 2008, this number has now started to dive. Even so, during the latest available quarter (Q1 2008), it was still as high as 188%, which is to say it have fallen back to the level of Q2 2004.

In order to get back within the historical range, house prices will have to fall at least 19% relative disposable income. And were house prices to fall to the lower end of the historical range, then they would fall some 27% relative to disposable income. Compare this with the mere 9% relative decline from the peak we have seen so far. Or saw until Q1 2008. Using the likely Q2 2008 numbers, more of the needed decline will have been done away with (even excluding the temporary boost to disposable income from the tax rebates). But even so, it is clear that house prices will have to fall a lot more in relative terms to get reasonable and that most of the decline is ahead of us.

It might be argued that because interest rates are so low now, prices won't have to fall as much. Well, that will probably prevent a decline to the lower end of the range, but given the loss of confidence in housing, a return to at least the higher end of the range seems likely. That will not mean a 19% nominal decline if nominal income continues to grow, but the decline in relative value will still imply a decline in what should matter for saving decisions, real wealth.

This picture would probably not be very different even using Case-Schiller. The reason for this is that while the decline so far have been greater using the Case-Schiller index, the initial boom was also greater, meaning that a much greater decline now would be motivated if Case-Schiller would have been used.


Anonymous rich t said...

Hi Stefan -- Excellent blog -- I've really been enjoying it since I recently discovered it, especially the stuff on the money supply.

I've looked into the home price indexes a lot so I thought I'd throw my two cents in. In my opinion the Case-Shiller index is the far more accurate of the two. You are right that it only covers 10 cities (though there is a 20 city version and a quarterly nationwide version I think) and that could overstate the boom/bust.

However, within those cities it is substantially more accurate. The OFHEO index only counts homes sold with conforming loans, meaning that it leaves out all the homes with subprime and other "new age" mortgages, as well as those over the conforming limit (a bit over $400k until recently). This is where all the action is, pricewise, and the OFHEO index thus provides a very distorted view by leaving these properties out.

All in all, my guess is that whatever distortions are caused by looking at just 10 or 20 cities are probably smaller than the distortions caused by looking at only conforming loans.

Thanks for the great blog.


12:43 AM  

Post a Comment

<< Home