Shiller On The Housing Bubble

Calculated Risk takes note of an NPR appearance by Robert Shiller this past Friday.  You won't be surprised by the headline:

Yale Professor Predicts Housing 'Bubble' Will Burst

I trust that you will neither be surprised that most of what Professor Shiller has to say is quite sensible and balanced.  (Similar thoughts were expressed in a Wall Street Journal editorial published on Thursday.)  But this statement struck me:

Although home prices have gone up a lot in the recent years, they are just the same houses, right? There is no change in the services they provide, its just the value we put on them.   

I'm not so sure about that.  Here is the usual scary picture of price-rent ratios:

Here's a picture of the ratio, when size and quality of the housing stock are controlled for:

Of course, this doesn't speak to the relevant question of whether homeowners' balance sheets can withstand the pressure of price declines and increases in interest-expense.  But it does, to me, look like a picture in which people are responding to price incentives -- i.e. low interest rates -- to increase their flow of housing services.  If that's a bubble, then what isn't?

Comments

[edhopper]

The second chart seems to be less stastically significant. A value such as "quality" is arbitrary and can skew the results. I put much more stock in the first graph that shows the normal rent to price rate is abnormal.

[K]

"But it does, to me, look like a picture in which people are responding to price incentives -- i.e. low interest rates -- to increase their flow of housing services. If that's a bubble, then what isn't?"

Good point, but doesn't it presume that all incentives are created equal (with special emphasis on "created"). What would these "incentives" look like without central bank intervention?
(Enjoy your blog very much. Regards.)

[edward]

Interesting, this gives the impression that for some reason housing got to be really cheap in the mid ninetees, then has been recovering historic levels. Of course, given the way the graph is moving at some point it has gone through these levels - this was true in the UK in about 2001, recovering the previous 'real high' in 1992 (or something) since then it's nearly all 'froth'. Something similar will undoubtedly happen in the US if the heat isn't taken out of the market.

[CalculatedRisk]

Dr. Altig, the OFHEO index is based on repeated transactions of the same home. Although this index has some questions, by using the same house issues of size, quality and location are removed (unless someone adds a pool, etc.)

The Census bureau series is subjective and has some serious problems. They try to adjust for location (like a busy street) but they ignore the bigger issue of location (Hawaii and Utah are both in the West).

Best Regards!

[Dave Altig]

CR --

I'm basically appealing to the argument by the new York Fed's Jonathon McCarthy and Dick Peach: "... comparisons of the various home price series suggest that a significant portion of price increases in some series— including the OFHEO index—can be attributed to increases in quality. As our analysis shows, the home price index used can have dramatic implications for one’s assessment of whether a home price bubble exists. Of the indexes available, we believe that the constant-quality new home price index is most appropriate for this assessment because it is the only one that
explicitly controls for changes in quality over time."

I grant that the methods used to construct the quality adjustments are legitimate topics of debate. But they *are* constructed from hedonic -- i.e statistical -- analysis. I don't think it is quite fair to call them "subjective."

But, in any event, my main point is that I'm just not convinced that it is so obvious that the housing market is out of line with what we would expect given interest rates. This was the point made by McCarthy and Peach and the Chicago Fed's Richard Rosen in an article you linked to last week. (I should probably cite these articles in an update.)

Of course, K's question remains: Are interest rates low as a result of some less-than-desirable constellation of policies?

Thanks, as always, for the input.

[CalculatedRisk]

Thanks for the feedback. I'm looking forward to your interest rate post tomorrow!

I've been puzzling over several questions:
1) What is the best measure of the bubble (I'm assuming there is a bubble).
2) What is the impact of interest rates on assets (especially housing)? Dr. Rosen's article was very interesting.
3) and your last point: Are interest rates low due to "less-than-desirable" policy?

Dr. Stephen King argued that point today:
http://news.independent.co.uk/business/comment/story.jsp?story=644563
"Interest rates may be peaking at unusually low levels, but this can only mean that the underlying foundations for this economic recovery are unusually weak: the whole edifice is in danger of subsiding under the weight of excessive - and policy-induced - debt."

Which Stephen King is more frightening?
Best Regards!

[Erik R]

Why does one need to "adjust" a price-to-rent index for size of the house? How does one do it? Don't both price and rent go up as a house increases in size?

[Secret Admirer]

You are misinterpreting the constant-quality price-to-rent ratio.

Let’s start from the beginning. The value of a house has two components. First, the value of having access to three bedrooms, a garage and a roof over your head (the consumption value) and the value of owning an asset that you think will appreciate over time (the investment value). The price-to-rent ratio is actually an investment/consumption value ratio.

Now, let’s take a closer look at this constant-quality index. The index measures the value of a new home what has the same characteristics as a home built in 1976. If we were in 1976, this would be a house with the most modern, desirable features. It would probably be “the” best house in the neighborhood. In other words, it would be more than a simple roof over your head, it would be an asset admired by others. That would give you a high investment/consumption ratio.

Fast forward 30 years. What we consider modern and desirable has changed. Your house does not have a cathedral ceiling and who knows what other amenities are missing. Don’t be sad though. Your house may not be in high-demand but it still has consumption value. You still have your bedrooms and roof after all. The question is, why should people pay a premium for such a house when more “modern” and “desirable” homes are being built in the neighborhood? They won’t. This means, your 1976 house will have a “lower” investment/consumption ratio than it did in 1976.

Aha! This means the constant-quality price-to-rent ratio will “naturally” trend downward over time. All increases will be temporary. Let’s go back to your second chart. What does the recent increase from 0.85 to 0.96 signify? Well, that is the sign of the bubble you have been looking for.

By the way, the second chart is definitely more accurate than the first one, IF interpreted properly.

[Gerard MacDonell]

Dean Baker's piece argues, convincingly in my view, that home improvment spending is far too low a ratio to the replacement value of the housing stock to explain the gap between Census and OFHEO.

That Dr. Baker is about to burst a vessel while arguing his point only slightly detracts from his credibility, in my opinion. So far as I understand, there is a strong preference for the OFHEO among those who have studied the various imperfections in these two price measures. For example, I believe your colleagues at the Fed prefer the OFHEO.

Not to sound snarky, but this reminds me a bit of the "controversy" around whether we should use the payroll or household survey to measure employment growth. Both measures are imperfect but it is a bit misleading to imply that we really can't know which is the more accurate. Payrolls for jobs, OFHEO for home prices. No?

[kharris]

"What is the best measure of the bubble?" Nope, can't do it. There cannot be a "best" measure of something not well specified. What do you really want to measure? Risk of widespread price decline? Likely extent of average price decline? Extent of divergence in prices from some long-run norm? Gotta know what you want to measure before you measure it.

[mousebender]

Is the constant-quality index data available online?