Do We Remember the Pain of Bubbles Past?

In a recent EconTalk podcast, host Russ Roberts and Yale's Robert Shiller got into a good discussion over the origins of the subprime crisis, which Shiller largely attributes to bubble psychology. But Roberts offered an interesting test:

Shiller: It's this mysterious failure to perceive the possibility that home prices could fall that I think is the biggest single explanation for the events.

Roberts: I think that's part of it perhaps. Although, again, many of us were alive in the 1980's and do remember a time when housing prices -- certainly in the markets that had gone the craziest -- did fall, but that was a long time ago. People do forget it.

But do people forget, or is there such a thing as "regional memory" in areas where home prices fell during the 80's and early 90's?

To illustrate, the drop in crime since the 1980's is now a part of New York City's local conventional wisdom, brought up in conversations and written about again and again in news stories. And it makes perfect sense, since safety is a major consideration in where a person chooses to live.

Does the same phenomenon happen for local real estate prices? For example, do Los Angeles residents, where home prices got very bubbly in the late 1980's, remember this pain? And if so, did it help in preventing prices from getting as bubbly this time around?

Shiller believes that if we can learn how bubble psychology forms, we'll be able to recognize it earlier and eventually learn how to prevent manias from happening again and again. If the cities that experienced a big price drop in the 80's and 90's were less bubbly this time around, implying some sort of medium-term regional learning, then maybe the task won't be as hard.

To get an idea, here's a list of cities that had experience with "the possibility that home prices could fall." The data is from the FHFA home price index which only includes information on prime loans.

Data collection for the six cities listed at the bottom started too late to capture any of the rise in prices, so only the bust is shown.

There are at least a couple of ways to read the table.

In support of regional memory: The magnitude of booms was generally larger the first time around than the second. We can't say that that's due to bubble memory, since a number of other factors could have driven regional home prices, but it's worth mentioning that for the entire group of cities and metropolitan areas tracked by the FHFA index (close to 400), real estate prices grew more rapidly during 90-00's boom than the 70-80's boom.

Against regional memory: Once the current real estate downturn is over, a number of cities in the table will have experienced bigger retrenchments than during the previous cycle, suggesting a greater overshoot on the way up.

The latest issue of the Atlantic has a great discussion by Virginia Postrel on the experimental research into bubble memory. (HT: Baseline Scenario.) The findings seem to support the case that people do learn from previous bubbles, except when underlying fundamentals change.

Looking again at the table, the only cities that are experiencing bigger falls now than in the downturn are those located in California. Why was speculation greater there? Is it because California is a "star" real estate area and bubbles are just more likely to appear there? Or did securitization - the underlying fundamental change - play a greater role in driving prices higher in the prime markets of California than in the other cities?

Comments

[John Lounsbury]

The size of bubbles should be analyzed in real terms (i.e., adjusted for inflation). I'll make up some inflation numbers:

If the cummulative inflation during the 70's and 80's was 100%, the rise in Boston would be 129%, inflation adjusted (compared to 375% nominal).

If the cummulative inflation during the 90's and 2000's was 50%, the rise in Boston would be 83%, inflation adjusted (compared to 175% nominal).

Adjusting for inflation the two bubbles will be more equal in magnitude.

Zubin, you wrote "The magnitude of booms was generally larger the first time around than the second. We can't say that that's due to bubble memory, since a number of other factors could have driven regional home prices, but it's worth mentioning that for the entire group of cities and metropolitan areas tracked by the FHFA index (close to 400), real estate prices grew more rapidly during 90-00's boom than the 70-80's boom."

This appears at first glance to be contradictory. The only rationale for the statement I can come up with is that the bulk of the later period gain occurred over only a few years (say 2002 to 2006) whereas the earlier bubble was spread out more over two decades. Is my interpratation correct?

2008 Dec 09 02:34 PM