Saturday, July 12, 2008

Falling from the Period of Financial Distress into the Panic and Crash

In 1972, Hyman Minsky described the "period of financial distress," in a paper in a journal that no longer exists (Reappraisal of the Federal Reserve Discount Mechanism, vol. 3, pp. 97-136), "Financial Instability: The Economics of Disaster." Charles P. Kindleberger picked up on this and followed Minsky's analysis in his famous book, _Manias, Panics, and Crashes: A History of Financial Crises_, the 4th edn of which appeared in 2000 (the last one by him; he is now dead, but the pubbers have others hacking away at it for more editions, gag). The period of financial distress is a gradual decline after the peak of a speculative bubble that precedes the final and massive panic and crash, driven by the insiders having exited but the sucker outsiders hanging on hoping for a revivial, but finally giving up in the final collapse. According to Appendix B of Kindleberger's 2000 edition, 37 of the 47great historical speculative bubbles exhibited such a period before the final crash, even though all the theoretical models predict a crash immediately following the peak with no such period.

In 1991 I published the first mathematical model of such a phenomenon in my book _From Catastrophe to Chaos: A General Theory of Economic Discontinuities_ (Kluwer, Chap. 5), still my most cited work at over 150 according to Google Scholar, although nobody seems to have noticed this particular contribution in the book. In 1997, I published a paper describing this model (and related matters) in a paper that reproduced a plenary lecture given in 1996 in Berkeley, "Speculations on Nonlinear Speculative Bubbles," Nonlinear Dynamics, Psychology, and Life Sciences, Oct. 1997, vo. 1, no. 4, pp. 275-300. This paper has never been cited. More recently I have coauthored a paper that has been under long review by a journal (now under a long revise and resubmit, still waiting for an answer) with Mauro Gallegati and Antonio Palestrini, "The Period of Financial Distress in Speculative Markets: Interacting Heterogeneous Agents and Financial Constraints" (available at my website:, that lays all this out in much more up-to-date mathematical modeling.

So, why am I boring all of you with this self-citation? Well, Dean Baker is constantly claiming credit for his forecasts of doom and gloom. It looks like we might be finally reaching the big crash in the US mortgage market after a period of distress that started last August (if not earlier). I and my coauthors are the only people to have provided actually formal models of this phenomenon, beyond the verbal and historical discussions provided by the brilliant Minsky and Kindleberger (both of whom I knew, but both of whom are now dead). I have been forecasting this in unpublished lectures all over the globe for years, but never have put it up into the blogosphere. So, I am claiming credit, to the extent it is due, although the basic ideas were clearly laid out earlier by Minsky and Kindleberger.

I will add one more story. Three years ago I presented an earlier version of the still-unpublished paper with Gallegati and Palestrini in Tokyo at Chuo University. In the middle of the presentation the biggest earthquake in 13 years hit Tokyo, in fact right at the moment I said the word, "crash." Some of the Japanese in the audience blamed me, not entirely humorously, for having caused it.


Myrtle Blackwood said...

"The period of financial distress is a gradual decline after the peak of a speculative bubble that precedes the final and massive panic and crash, driven by the insiders having exited but the sucker outsiders hanging on hoping for a revivial, but finally giving up in the final collapse..."

Looks like this PFD may last a decade or two ;-(

Anonymous said...


Which "crash" do you have in mind...the crash in the real estate bubble or the crash in the stock market? What about contemporaneous bubbles in other markets. For example, if (and I realize it's a big if) the run up in oil prices is seen as a bubble rather than a fundamental reflection of supply and demand, then do crashes simply result in new bubbles somewhere else?

Sandwichman said...

If it all falls apart on Monday, Barkley, I'll know who to blame. said...

It does seem to have held up ultimately on Friday, although Indymac is the second biggest bank failure in US history. Anyway, I am thinking in terms of the mortgage-related financial markets, which seem to have peaked with the crisis last August (although housing itself peaked earlier). I note that of the other ten bubbles described in Kindleberger, they were split about half and half between those that just crashed straight from their peaks (e.g. the Tuiipmania) and those that just gradually declined without any full-blown crash. Of course it would appear that the Fed et al are trying to achieve this last outcome, and maybe they will, but also perhaps at the expense of a long decline that keeps going.

Real estate itself is funny, because individuals often can hold off selling and so the declines can be kept more gradual and outright crashes can be more easily avoided. The poster boy on that one is Japanese real estate, which started declining in 1992, and has not quite clearly fully hit bottom still, although it appears that it did in Tokyo at least a few years ago (when I was visiting and gave my lecture described in the post).

The biggest single one day crash ever in US stock markets was on a Monday, Oct. 19, 1987. Things were very weird on the previous Friday.

Bubbles often happen in tandem, with a new one following as another crashes. Probably the most dramatic example is the interlinkage between the Mississippi Bubble in France and the South Sea Bubble in England, both very large ones for the economies of that time. So, the Mississippi Bubble peaked in December, 1719, which marked the takeoff time for the South Sea Bubble. The Mississippi Bubble crashed in May, 1720, which marked the peak of the South Sea Bubble, which itself then crashed in September, 1720. The economist Richard Cantillon is someone who made money speculating in both of those bubbles, one after the other. But sometimes there is no bubble after one crashes, e.g. the stock market in the US in 1929, which was followed by the Great Depression.

Anonymous said...

i would guess that bubbles in 'primitive capitalism' may not be quite the same as those developing because and out of system decline and decay.
while expansion of fictitious capital always has limits, the last decades have seen a progressive stretching. so, barkley, i would say that contractionary tone of the world and u.s. economy since early 1970s provoked a turn to finance, that this turn became institutionalized, began its ending years ago and should be no surprise (not to say you are surprised). said...


There is a funny thing about "fictitious capital," a phrase I love and which dates back to discussions in the wake of the collapses of the Mississippi and South Sea bubbles in 1720. One can never forecast precisely what those limits are.

In a wise paper published in 1957, Paul Samuelson said something along the lines of "we can never know how a far a bubble will go beyond its fundamental value, 10% or 10,000%, but fall eventually it must." [that is definitely a paraphrase, but close.]

Anonymous said...

Sorry that I can't get back to you on the Economists view. Mark Thoma kicked me off because I would not play the role of Uncle Tom to the liberal leftwing sociopaths. I know what's required, but I just don't want to do it.

Anonymous said...

That was from me, Aaron

Anonymous said...


I used the term 'primitive capitalism' as shorthand differentiation between pre-industrial and industrial capitalism. Aside from psychologies, bubbles developing within the latter - and its more developed credit system - seem to me of a different 'flavor', one related to a business cycle (and waves containing cycles) not present in the 'primitive'.

I agree with Samuelson that the market can remain irrational for an indeterminable period, or that accurately calling tops can be a fool's game, but at same time believe it possible to see increasing tension between the real and the financial as the latter can be driven through the former's weakening but only up to that magical 'point' (beyond which masses of fictitious capital are devlorized/destroyed).

From my perspective, mid-1990s weakness in Asian real economies followed by the financial crises marked a particular turn within the larger, more generalized, post 1970 weakening. Since then the struggle by the financial to save itself, and be saved, has IMO been ever more extreme, leading straight to the present debacle-in-process. said...


Bubbles have certainly evolved with institutions and technology. But the psychological dynamics have been remarkably constant. Accounts of the Dutch tulipmania of 1636-37 are remarkably familiar. BTW, as an aside on that one, it is one of the ones that Kindleberger categorized as a sudden crash from a peak. However, the story is more complicated. When the peak was hit and sharp declines started on Feb. 5, 1637, the authorities simply closed down the market for futures in tulips (yes, a derivatives market, not the actual market for actual tulips themselves), which did not reopen for several months. When it did, the prices were much lower, of course, so the data looks like this huge crash. But we do not know what would have happened if the markets had stayed open.


Sorry Mark is being rough. I think he is getting tougher with lots of folks. He censored a couple of my postings. But then, his blog has gotten very busy.

You are certainly free to play here, although you should probably be forewarned that the general atmosphere here is leftier than on Mark's.


Anonymous said...


I'll try again but with a single question: do you think that crises within merchant capitalism on one hand and an industrial capitalism that had subsumed the merchant form on the other have been the same?

What I'm saying is that overproduction crises have been specific to the latter, that these promote speculative activities, and do so within a quite different context than the 17th & 18th centuries, no matter the psychological similarities. said...


Under merchant capitalism you had crises due to bubbles without crises of overproduction. The tulipmania is a perfect example.

What has been curious is that the nature of the sources of overproduction under more advanced capitalism have not always been the same. Thus, during most of the 19th century, many of the overproduction crises were preceded by the collapses of speculative bubbles, whose aftermaths included damage to the financial system, whose problems then spilled over into the real sector. Thus one finds this as the major theory of business cycles in such writers as Mill, and Marx discussed this phenomenon also, especially in Vol. 3 of Capital.

In the 20th century, while such things still happened (and the Great Depression was preceded by the 1929 stock market crash), it seems that other factors often were more involved, although more recently problems coming out of the financial sector seem to have reappeared, with the recession of 2001 clearly a fallout of the collapse of the high tech stock market bubble.

kevin quinn said...

Barkley: Fortuitously, I had just checked your book out of the library the day before this thread began. What a wonderful book!

On bubbles, for years I've spent time on bubbles in the Money and Banking course. I have them read Keynes famous beauty contest stuff and portions of Schiller's Rational Exuberance. I try to get them to see that asset markets can be efficient in the sense of offering no predictable excess returns and still wildly inefficient if efficency means - as it should- that prices equal fundamental values. It's amazing how many basic texts elide this all-important distinction. At any rate, a student who had taken the course was in a finance course where she had the temerity to raise the issue of rational bubbles in a discussion of the efficient markets hypothesis. "Where did you learn that," said the professor, "in the Peoples Department of Economics?" said...

Yes, your point is very apt. I also find it werid that even most advanced texts in financial economics, such as Campbell and Lo, are pretty skimpy on such things. There is also the oddity that while they mention in their first chapter that it is well known that financial asset returns exhbit kurtosis, or "fat tails," that is "more extreme events than predicted by the Gaussian normal distribution," the entire rest of the book does all the theory assuming Gaussian normal distributions. Sheesh.

Oh, and thanks for the nice compliment on my book. It was rejected by 13 publishers before Kluwer finally accepted it.


Anonymous said...

good, we agree, though my recollection of the argument in cap vol 3 is that overaccumulation/overproduction of industrial capital and falling profit rate provoked the turn to speculative activities, i.e. the crisis in the real tended to preceed that in the financial, or that the overvaluing then collapse of claims grew out of the former.

i'm not sure that marx ever made the above quite so specific but it follows the logic of that volume as well as parts of the 2nd.

so lets add policies which facilitate and support 'financialization' and not hard to see that the gap between real and claims to the real can be stretched to extremes and that owners of claims seek to protect these and gain more through, say, allocation and (necessarily expansive) hedging activities.

then lets think of the post-1970 period as one of real global slowing, the slightly later volker tightenings performed more with an eye to improving real returns to finance, the overall process of financial deregulation, etc, and here we are trying to save what can't be saved.

Anonymous said...

forgot to add

recession had begun prior to the '29 crash, which certainly exacerbated. 'other factors' may have been the high degree of heterogeneity of overall capital stock (which dumenil and levy brought out in their '95 paper), increased sectoral disproportionalities related to rapid expansion of the new consumer durables sector, a never fully recovered rural economy...

for 2001 recession, nipa measured corp profits had peaked end 1997, perhaps not unrelated to the asian crisis. other hand, earnings continued to rise (which might not have happened had they been less 'managed'). memory is bad on this but seem to recall a fund manager telling me that his roe measures had also peaked at roughly that earlier date. simply, the slide had begun well before the market noticed. should people have exited at that earlier date? no but should have become increasingly cautious. said...


The recesssion in 1929 prior to the crash was very mild and had not been going on very long. It was post-crash policies that turned what had been a not very serious recession into a full blown Great Depression.

Anonymous said...

are you referring to the 1927 recession after which a sharp increase in value and number of mergers, another boom in industrial production and final bull market run (slightly bothered by higher customer loan rates and the frbny's discount rate increase)?

perhaps not a 'serious recession' but a foreshock.

imo, post-crash policies as causal shifts too far towards notions of cycles and crises as politically determined. said...


Not 1927, but the slowdown that began earlier in 1929. There had also been a crash of real estate in Florida in 1925, the great Florida land bubble.

Anonymous said...

Thanks Barkley,

I actually spent some months studying that decade and the previous a few years ago while conversations with relatives and others who lived through them were moreless fresh in my mind. The different experiences of one who was an NYC broker from mid 1920s until '31, another who worked for a small midwest business, and another from very small town central Florida were particularly instructive.


Anonymous said...

Dear Barkley,

I'm so glad that Google led me to your post---I was searching for some rational bubble material to include in next semester's graduate money course here at GMU.

Your '97 paper is on the syllabus now...

Anonymous said...

This is an old thread but somebody else responded and I looked at the old posts and wanted to respond too.

Think of an automobile engine. There might be one way for it to be ideally in tune, and a small variety of ways for it to be well-tuned. But there are lots of ways for it to run badly.

The carburetor can be too lean or too rich. The spark plugs may be fouled, or one of the spark plug lines may be broken or weakened. The distributor cap might have carbon traces that make it supply too weak a spark to some cylinders. the coolant may be low or the water pump broken, or maybe something has gotten onto the radiator so it doesn't radiate properly, or the fan belt may be broken or slipping. If the tires are underinflated the car doesn't get good mileage, and if they are overinflated they may blow out unpredictably.

There are lots and lots of ways for things to go wrong.

With a lot of experience you might notice particular patterns of wear. "In this model the front brakes often get misaligned so they wear out fast." Which means you use a lot of gas pushing against the brakes until they're gone. With tens of thousands of cars and years of experience you can find out about such things.

I'd expect it to be similar with economies, which are so much more complicated. Why would we expect them to have only a few ways to fail? There are so many ways to fail that other things equal we should expect them to mostly run in failure mode.

Some people think that economies have internal mechanisms set up to optimise them. That's possible. But anybody who has much experience with optimising compilers will expect that what the economy thinks is optimal might not fit what you'd think was optimal. We're still looking for the Do What I Want (DWIW) or Do What I Mean (DWIM) programming commands. Economies contain complex feedback loops, and there's no reason to suppose they optimise for things that any sane person would want. They supply feedback to get whatever random results they get, and they change when the people who benefit from the current results are overpowered by people who think they'd benefit from some other results.

I think the principles of systemantics might apply here. When you have a complicated system that stops working, and you can't figure out all the little details and see how to fix it, sometimes a good swift kick will get it to lurch back into action for awhile. That probably won't work twice but it might. When that stops working then it's time to junk the whole complex system and replace it with something simpler.

J Thomas