Bubble predictions: better late than early
Reader comments often inspire new posts, and this is a good example. In my post on Krugman’s 2005 prediction of a housing bubble, a number of commenters pointed out that Dean Baker made the same call three years earlier, in 2002. The clear implication was the earlier was better, and that Krugman was late to the game—just copying Baker. I think that’s wrong.
Just so I am not misunderstood, this post is not a criticism of Dean Baker. Commenters sent me links to bubble predictions Baker made in 2002 and also 2005. I am going to argue that the 2002 prediction was neutral, neither particularly good nor bad, and the 2005 prediction was a good one. All in all a decent record, nothing that deserves criticism. Rather I’d like to focus on a narrow technical point, and argue his 2005 prediction was actually far superior, even though it came later.
Precisely what does it mean to predict a housing bubble? Are people predicting that one will occur in the future? That prices will rise very rapidly? Or are they predicting that one is already here, that prices are too high relative to market fundamentals? I think it is usually the latter. If the term ‘bubble’ is to have any meaning at all (other than the trite observation that prices have recently risen) there must be an implied prediction that in the not too distant future (i.e. not 100 years out) prices will fall back closer to their fundamental value. I’ve argued this point ad nauseum, and won’t repeat it here. My hunch is that people confuse these two issues, which is why many people assume it is easier to spot bubbles than it really is.
Here is what Dean Baker said in 2002:
This paper examines whether the increase in home prices can be grounded in fundamental economic factors or whether it is simply a bubble, similar to the stock market bubble. It concludes that there is a housing bubble. While this process can sustain rising prices for a period of time, it must eventually come to an end.
He does acknowledge prices might rise before dropping, which is of course what happened. But that comment is so vague that I take it as one of those things you almost have to say. After all, if prices have been rising fast, only a fool would predict an immediate and sharp decline, especially given that housing prices have a bit more momentum that stock prices. In a nutshell, I infer that he is mostly saying that housing prices have risen above their fundamental value and that at some point the real price of housing should drop to more reasonable levels.
In this paper from late 2005, he and David Rosnick again make a bubble prediction. This time much more accurately, in my view.
In my Krugman post, I used this graph to think about the accuracy of bubble predictions. I argued that those seeing a bubble in the US in 2005 were right, but in Britain, New Zealand and especially Australia they were wrong (thus far.)
If you just eyeball the data, to me it looks like these prices occurred in the US:
2002: 200
2005: 300
2006: 350
2010: 250
So it’s fair to say that 2005 bubble predictions turned out to be accurate. But what about 2002? Well the actual price seems to have risen about 25% in 8 years. That’s not too different from the overall inflation rate, and hence I’d say there hasn’t been much change in real housing prices. So I’d call that a neutral, where lower real prices in 2010 would be a win for Baker, and higher real prices would have been a loss.
I’d like to use an analogy, to suggest why it’s better to be late than early, why Krugman actually deserves credit for being late to the bubble party. I recall after the 1987 stock market crash that someone praised John K. Galbraith for having predicted a stock crash. He made the prediction in January 1987, when the Dow was around 1700. It then rose to 2700 in August, before crashing to 1700 in late October. So was Galbraith right? As this post shows, people seem to assume he was. But I’d say no, as his prediction really didn’t convey useful information:
1. If you sold stocks on his prediction, you would not have made money—even in the long run.
2. It was an implied prediction that stocks were overvalued relative to fundamentals. But today very few people would say the Dow was overvalued in 1987 at 1700, indeed if anything it might have been a bit undervalued. This shows how hard it is for even a very smart person to know whether something was overvalued in real time. I could say the same about Boston house prices in 1987, and I’m sure people living in Manhattan, London, Vancouver or San Francisco could provide similar examples of prices that once seemed insane, but now (even in this recession) actually look (in retrospect) like equilibrium prices.
I think a good prediction, a useful prediction, would be someone that predicted a stock market crash in August 1987, not January 1987. Those are the people who deserve credit if you (unlike me) believe market predictions aren’t just dumb luck.
I can think on one counterargument. One could argue that an early prediction might have resulted in public policy changes that prevented the worst of the housing bubble. But I favored those public policy changes even without being able to predict the bubble. And I’m claiming it’s not obvious there was a bubble in 2002. I’d hate to have public policy decisions based on inaccurate bubble predictions.
So from now on when someone tells you that Dean Baker predicted the housing bubble back in 2002, the correct response is “You think that’s impressive, well Krugman predicted it in 2005!” Enjoy the puzzled look in their eyes, and savor the thought that you are soon about to show your superiority by setting them straight. At least if you’re as big a jerk as I am.
PS. Take a look at the link discussing Galbraith. It was from 1994, and they assumed we were in the midst of another speculative bubble—when the Dow was trading in the 3500 to 4000 range. What do you want to bet that they said “I told you so” in 2003, after the crash brought prices down to 8000?