The doggedly ascientific nature of the “science” of economics remains a wonder to behold. More power to you, John.

“A shock of this magnitude would have resulting in much worse catastrophe ”

Wrong, 100 times wrong.

A shock of this magnitude in the 1970’s would have resulted in bankruptcies of more financial institutions – but much less shock to the overall system. More rich white guys would have experienced a “catastrophe” – but the system as a whole would experience LESS unemployment, not more. We can’t gain any jobs until the bad debt is purged.. through bankruptcy.

Ken Houghton:

dsquared is not discussing vega. He is saying that volatility itself is low sometimes, and high other times. That is the essential characteristic of volatility.. it changes.

So I understand the position that these economists find themselves in… I empathize… but I will say one thing that distinguishes what I did from what they (and so many other anti-Keynesians) are doing these days: I admitted that the data did not support my theory. It’s the intellectually honest thing to do. Yet on TV and in written commentary we far too often see arguments that seem to cherry pick their metrics, or the date ranges for their data.

So what if they are?

Macroeconomic phenomena are based on the actions of quantum particles. That doesn’t make quantum-based macro a useful or productive approach to science. And if the only way you could get any interesting results out of your attempt to make quantum-based macro-economics work is by making overly-simplistic assumptions about both macroeconomics and quantum theory, then quantum-based macro would be actively harmful.

Reductionism in science is based not only on the simplistic claim that phenomena at level A are aggregates of phenomena at level B. But also on the claim that useful things can be learned about the laws that apply at level A by study of the laws that apply at level B.

For example, look at the problems Fama has in understanding what a bubble is.

But perhaps we should substitute the word “fluctuation” since (re)defining volatility appears to attract undue attention from the living dead.

“Many words could be used to describe these responses—and have been—but ‘judicious’ and ‘moderate’ are not among them.”

I’m a computer scientist, not an economist, but I’ve read four (good) books thus far on our ongoing economic disaster, and I don’t believe “judicious” or “moderate” appeared in any of them…

There’s a further big problem shown by the graphs in Brad DeLong’s page, which I mentioned in the chapter. Most of the graphs seem to show a long-term decline in volatility since the 1950s, interrupted temporarily in the 1970s and early 1980s, as argued by Blanchard and Simon. But, again, this implies that they can’t be matching anything of interest. That’s why no-one much accepted the Blanchard-Simon interpretation of the data, as against the main story where the Moderation began in 1985.

I completely agree with both of your posts. I am not saying that it is inappropriate to look at other data—quite the opposite: look at as much data as possible. There is no “One True Measure” of macro volatility. It is completely reasonable to “cross-check” and ask whether the pattern survives in some data sets / samples but not others. Absolutely – kick every tire, cut the data every which way you see fit. Just don’t rig the game so that you habitually exlude results which conflict with preconcieved notions of how the world should be.

Also, Daniel, while I agree that if one metric shows the great moderation and a “bunch of others” don’t then you might doubt the result, I think the pattern shows up in most of the other disaggregated output measures (that is if you look at investment types, consumption types, etc. you still see the decline in volatility though I have to admit I’m not sure of this since this is not my area). If this is true, it’s probably that there are a bunch of series that show the moderation and a few that don’t. Judging from DeLong’s pictures, employment measures seem to tell a different story for instance.

There’s nothing much wrong (indeed, it’s best practice) to look for evidence that a relationship or effect found in the analysis of one dataset can also be observed in other data sources.