The 1987 stock market crash; myths and reality
Today’s the 26th anniversary. Here are some myths:
1. The October 19, 1987 crash was caused by . . .
It doesn’t matter what you put next, it’s wrong. Every theory of the October 19, 1987 crash that has ever been put forward has a fatal flaw. The theories always imply other similar crashes should have occurred on other occasions. But the October 19, 1987 crash is totally unique—nothing remotely like it has ever happened before or since (in the US.) Thus all theories of its cause are wrong. Or perhaps I should say “useless,” which is actually what we mean by “wrong.”
2. The crash reduced wealth, and hence reduced aggregate demand. The economy slowed.
Nope, the Fed determines aggregate demand. Fed policy remained expansionary and thus the economy boomed after the crash, even though the decline of August – October 1987 was virtually identical to 1929 in almost every respect. The 1929 crash was followed by tight money, and that is why the economy did much better in 1988 than 1930.
3. Unlike the 1929 crash, the Fed cut interest rates sharply after the 1987 crash.
No, the Fed cut interest rates sharply during 1929-30. Interest rates tell us nothing about the stance of monetary policy.
4. Unlike 1929, stocks quickly recovered from the 1987 crash.
This is highly misleading, in a very interesting way. People sometimes cite this as a reason for the contrast between the severe recession of 1929-30 and the big boom of 1987-88. But that can’t be right, because the initial bounce back was quite similar in both cases. In December 4, 1987, stocks actually closed lower than on October 19. Stocks rose in the first 3 1/2 months of 1988, but they rose about the same amount in the first 3 1/2 months of 1930. So the first 6 months after each crash was quite similar, and yet the macro performance of the economy was already radically different. So you can’t use stock price bounce back to explain why 1930 was a deep depression and 1988 was a boom. The two crashes were also similar in that each had been preceded by a big multiyear stock price boom.
5. JK Galbraith correctly forecast the 1987 stock market crash.
Nope, he forecast a crash at the beginning of 1987, when stocks were about at the level they fell to after the crash. Useless “forecast.”
The stock market is correlated with RGDP, so stock market crashes do contain useful information. But they aren’t infallible. The stock market sometimes gets it wrong. And 1987 was one of those times.
Tags:
19. October 2013 at 11:47
Excellent blogging…yes the 1987 crash is sui generis…theories I have none, but I think some program trading fixes and market stops were introduced both institutionally and privately after that…if the 1987 event caused the Fed to loosen it was probably a stimulative event…
19. October 2013 at 11:49
” Every theory of the October 19, 1987 crash that has ever been put forward has a fatal flaw.”
If there is no apparent explanation for the 87 crash does that mean the emh is wrong?
19. October 2013 at 11:50
Most predictions are more impressive the earlier they are made. If I now correctly predict who will win the next Soccer World Cup it will be more impressive to get it right than if I do the same prediction when half the tournament has been played, half the team have been eliminated and it has become clear who is playing well.
When it comes to the stock market, it is, however, equally wrong to predict a crash too early as too late. Society mocks the “permanent high plateau” of Irving Fisher but does not mock the irrational exuberance speech of Greenspan even though they were both ridiculously wrong in retrospect.
To pick another of this blog’s topics: One day the China bears will say “we told you so”, even though they never did say anything of the sort of thing that actually happened.
19. October 2013 at 12:28
Mike, Certainly a little bit wrong.
Luis, Yup.
19. October 2013 at 12:55
The causes of the 1987 crash and subsequent recovery just isn’t that mysterious. Quoting ourselves:
19. October 2013 at 13:43
Ironman, I’m afraid that’s no explanation at all, for the reason I gave in my post. If that was a good explanation, this sort of crash would have occurred more than once.
19. October 2013 at 13:45
“It doesn’t matter what you put next, it’s wrong. Every theory of the October 19, 1987 crash that has ever been put forward has a fatal flaw. The theories always imply other similar crashes should have occurred on other occasions. But the October 19, 1987 crash is totally unique””nothing remotely like it has ever happened before or since (in the US.)”
It is not unique 100%. Crashes, recessions, and collapses don’t have to be all identical. Nothing in history is identical anyway, and yet we propose theories to account for past events.
The cause for the 1987 crash is artificially low interest rates and inflation prior to that. Since accelerating inflation is necessary to prop up an increasingly distorted capital structure, the Fed chose to tighten up instead of hyperinflation, and hence the correction at that point, instead of after the currency collapses and remonetization occurs.
Artificially low interest rates and inflation is the same cause for the subsequent 1991 recession, as the Fed loosened after the 1987 crash and bloew up another bubble.
It’s the cause for the 2008 crash.
It will be the cause of the next crash.
This blog post is just an evasion of the effects of one’s own advocacy.
19. October 2013 at 16:32
Every theory of the October 19, 1987 crash that has ever been put forward has a fatal flaw. The theories always imply other similar crashes should have occurred on other occasions. So, the question is not why there was some stock market downturn–they happen regularly–but why it was so intense.
According to this, it was of the same scale as the 1929 crash, just over one day instead of two.
http://en.wikipedia.org/wiki/List_of_largest_daily_changes_in_the_Dow_Jones_Industrial_Average
19. October 2013 at 16:40
(Imaginary) speech by Mark Carney: http://longandvariable.wordpress.com/2013/10/19/the-speech-mark-carney-should-have-given-but-didnt/
19. October 2013 at 17:09
I’m reminded of the claim that Ron Paul predicted the 1987 stock market crash.
http://www.youtube.com/watch?v=An4XrseNTNU
What he actually predicted was a recession or “depression” for 1987, but he still gets credit anyway. As if stock market crashes and recessions necessarily need to go hand and hand.
19. October 2013 at 17:11
We should have a drinking game every time an “austrian” shows up parroting ABCT nonsense.
“artificially low interest rates” – 1 shot
“increasingly distorted capital structure” – 2 shots
“hyperinflation” – 3 shots
etc
19. October 2013 at 17:34
Mike, you say:
If there is no apparent explanation for the 87 crash does that mean the emh is whrong?
Actually, quite the opposite. If the EMH holds true, then the cause of the crash must be unknowable…If there was something clearly identifiable that caused the crash, it would have been information that the market already had, and would have been priced into valuations at the time.
Markets crash, and that is completely consistent with the EMH.
19. October 2013 at 18:38
The generally accepted cause of the 1987 crash was portfolio insurance. This was the first crash (but not the last) caused by Myron Scholes and Fischer Black.
Here’s a young Robert Shiller writing in 1988:
http://www.nber.org/chapters/c10958.pdf
Also contributing may have been traders trying to frontrun the expected selling of portfolio insurers, and the adoption of automated trading to facilitate this (similar to the last 2011 flash crash).
Finally, a large number of market participants worried the 1982-1987 bull market was a bubble, and the economy was ‘due’ for a 60-year recurrence of the Great Depression.
Ever since 1987, stock options implied volatility has had a generally downsloping skew.
19. October 2013 at 18:43
This chart says it all: http://research.stlouisfed.org/fred2/graph/?g=nxq
19. October 2013 at 23:33
Scott, that kind of crash has occurred more than once.
The most recent episode occurred between 22 July and 22 August 2011. Stock prices had been elevated for 12 months preceding the crash, thanks in part to QE 2.0, the end of which was confirmed by the Fed on 22 June 2011. The proximate cause for the crash itself was a breakdown in debt ceiling negotiations that year, however notice that the market didn’t recover immediately after it was resolved on 2 August 2011, which is something we’re seeing happen in the market today. Hmmm, what was different between then and now? Ah, yes! Today’s QE is continuing (for now)!
But in 2011, the end of QE 2.0 had shifted forward looking expectations for earnings on the stock market lower, which is why stock prices shifted downward to a lower upward trajectory, rather than bounce back to the level they had been.
If you want more distant history, pre-dating the modern era of the stock market, 1946 comes immediately to mind. Here, after having been elevated for 12 months preceding the crash, stock prices began falling sharply (chart) on 29 May 1946.
Note the timing of the crash comes just over one-year after V-E Day, which Americans celebrated in the weeks afterward by rushing out and buying Victory bonds, which was itself kind of an impromptu QE-type event (who says you need the Federal Reserve to do that job?!) Note also that stock prices didn’t stop falling until well into September 1946, which both coincides with the one-year anniversary of V-J Day, which a year earlier, marked the end for all that Victory bond buying. The reason for the one-year delay in this case before stock prices shifted downward from the QE shut-off effect is because those war bonds had to be held for a year before they could be redeemed.
1987 began with a similar surge in bond buying, which set records in January 1987, complete with a coinciding surge in stock prices. Just like in August 2010, just like in May 1945. Stock prices remained elevated for months on expectations of improved earnings, but ultimately crashed when the forward-looking expectations that had initially supported higher stock prices changed and reset themselves to more realistic levels, shifting downward to a lower upward trajectory.
The only thing that makes the 1987 crash stand out is how far stock prices fell in such a short period of time, which perhaps has more to do with the advent of large scale computerized trading, which occurred without the circuit breakers we have today to allow for human intervention when prices change dramatically, which were first instituted in the aftermath of that crash.
The 1987 crash really isn’t mysterious, nor was it an isolated incident.
20. October 2013 at 00:25
“1. The October 19, 1987 crash was caused by . . .
It doesn’t matter what you put next, it’s wrong.”
In economics, more frequently than any other field of study I can think of, the most correct answer, if not the correct answer, is “all of the above”. Given that, naturally, assigning a single cause to a particular event is almost always going to be wrong. But, that never prevents an economist from trying to assign single factor causes or pretending that there was one. I guess that’s just the way the mind works. Admitting that an event had multiple causes is not conducive to advancing one’s notoriety, nor is it helpful in promoting one’s pet objectives.
20. October 2013 at 04:29
This is not to question anything that you have said in your blog. It’s just that after this year’s Nobel Prize to people who laid “the foundation for the current understanding of asset prices I drew graphs connecting Corrected Money Supply and the S&P 500. If nothing else, the graphs look interesting.
http://www.philipji.com/item/2013-10-20/the-connection-between-corrected-money-supply-and-asset-prices
20. October 2013 at 06:01
Lorenzo, Yes, but one day makes all the difference. Oct. 19 really was a freak occurrence.
Saturos, What’s his point?
Doug, That’s wrong, as the news could have occurred on that day.
Steve, That explanation is just as useless as all the others. It predicts there should have been other comparable crashes, but there aren’t any. It’s a literally useless theory. And it doesn’t tell us why stocks fell so much on that particular day. It tells us nothing at all. It’s just empty words.
Tommy, Good point.
Ironman. I’m afraid you are wrong. Stocks fell 20% on October 19th, 1987. Nothing like that ever happened before or since. Nothing even close. The fact that other interesting things happened during other longer periods of time has no bearing on my claim that October 19th is unique. It would be one thing if stocks fell slightly more than any other time on the 19th, but they fell far more than any other time. Even with “fat tails” that’s a freakish event.
Vivian, Good point.
Thanks Philip.
Everyone, Theories have to be more than “Just so Stories.” If you have a theory that explains only one observation, it’s not a useful theory. Computer trading? Why not the 1986 tax reform? Or 1000 other options.
20. October 2013 at 07:03
“It predicts there should have been other comparable crashes, but there aren’t any.”
What about the 2010 Flash Crash?
http://en.wikipedia.org/wiki/2010_Flash_Crash
The DJIA drops 1000 in a day, most of it within minutes. The proximate trigger was pictures on TV of Greek riots, which triggered fears of another economic meltdown, and was exacerbated by stop-loss selling and instability in the fragmented electronic exchanges.
Interestingly, the DJIA recovered 600 points by the end of the day, but subsequently hit new lower lows within 2 months. Everything happened on an accelerated time scale.
In 1987, there were fundamental triggers, too. A weak dollar, a weak trade deficit, a brand new untested Fed Chairman Greenspan, and Iranian missile strikes a few days prior. Add in the certainty that porfolio insurers would be sellers into weakness and their faulty assumption about continuously available liquidity (as per Black-Scholes).
The pattern is the same: fundamental trigger, new untested market strategy/structures, and inside knowledge on the ‘Street’ about weak sellers who will be forcibly liquidated.
1987 was unique in magnitude, but not in dynamic.
20. October 2013 at 07:39
Steve, First of all the crash you cite wasn’t even in the same ballpark to 1987. Not even close. Second, the factors you cite (Greenspan, weak dollar, trade deficit, etc), are bad explanations for two reasons. First of all, these things are not even contractionary factors. And second, they were already fully priced into the market by the 18th. So they won’t work. And if they were the reason we’d see 20% declines on other days too. We don’t even see 15% declines. Ever.
It’s weird that I seem to be almost the only person in the comment section who sees 1987 as refuting the EMH. And I’m a fan of the EMH!
20. October 2013 at 08:15
Actually, I’m agreeing that 1987 refutes EMH. I think lots of things refute EMH.
“And second, they were already fully priced into the market by the 18th.”
Those things may have been priced in, but the liquidation of portfolio insurers was not. So the fundamental event creates a cascade that gets priced in many times over. These things always have a trigger. Markets are good at figuring out direction, but sometimes struggle with magnitude.
Even folks like George Soros got “Great Depression Fever”. Here’s what Soros said at the time: “Technically, the crash of 1987 bears an uncanny resemblance to the crash of 1929. The shape and extent of the decline and even the day-to-day movements of stock prices track very closely.”
20. October 2013 at 08:41
Another commonality between 1987 and 2010, is that the price reporting function broke down. Traders in 1987 talk about dark screens and two hour delays receiving price reports. The same thing happened (on a faster time scale) in the 2010 flash crash.
If you want to panic a trader, tell him the prices he has are two hours old. Maybe the market should be halted whenever the price reporting function breaks down (rather than using price related circuit breakers).
20. October 2013 at 09:22
Scott – I wouldn’t read too much into the 1-day time frame of the 1987 crash, since stock prices are scale-invariant. What matters instead are the shifts in price level for the event from beginning to end.
Even in considering the time scale of the event, 1987 is only unique because of the increasing use of algorithmic-based trading that enabled the decline to take place within such a compressed period of time. In those days, there were no circuit breakers to arrest a short-term cascade failure (aka sudden crash in stock prices), as these weren’t instituted until after the event. They have since proven successful in arresting such large scale cascade events (such as the 6 May 2010 “flash crash”), which is why the 1987 event has yet to be matched.
20. October 2013 at 10:25
Steve, Those things have happened on other occasions, why did prices fall so much in 1987?
Ironman, Even if that were right, it would not explain why nothing like 1987 happened before 1987. There are far more trading days before than after.
I don’t know what scale invariant means. Stocks tend to show larger changes over longer time periods, which is exactly my point.
20. October 2013 at 10:47
Daniel:
“We should have a drinking game every time an “austrian” shows up parroting ABCT nonsense.”
In order to cure alcoholics, we should recommend they drink a shot every time you display an understanding not only ABCT itself, but the foundation of ABCT, which is economic calculation. Heck, we can even throw in a whole Mickey if you display evidence you have even read the original source material.
20. October 2013 at 11:36
No thanks, I’m not interested in pseudoscience.
20. October 2013 at 11:46
“Those things have happened on other occasions, why did prices fall so much in 1987?”
It’s not very satisfying but it’s like asking why didn’t the bridge collapse before? The bolt was stressed before, but it never quite sheared off. 1987 had records by 2x for NYSE volume, and presumably CME volume, program trading baskets, and portfolio insurance were also at records.
Also, accurate and orderly price reporting is really important. Take the 2010 flash crash. When all the prices are delayed, and the only thing you know is that the DJIA is down 1000, you pull back from the market UNLESS you are on margin in which case you sell indiscriminately. Needless to say, this produces a very one-sided market.
Once things are halted, prices are up to date, and you see Proctor and Gamble trading for a penny, you know it’s a market structure problem, not an economic one. So traders can act like monkeys banging on the green buy button for free bananas, and the market goes back up.
Days later, the market has a hard time sustaining recovery, because the simple knowledge of a market structure problem casts a pall over risk taking.
20. October 2013 at 15:17
To be somewhat snarky, did future expected cash flows really fall or risk premium/interest rates really rise enough to justify a 20% movement? The violence was odd because of automatic selling algorithms, but in general the movement is not too dissimilar to bear markets like mid to late 70’s, early-00’s and 2009. I think 2009 went down too much while the early-00’s was more a correction to reality.
I agree with most of the post though. The stock market crashes or bear markets in general may or may not indicate a large drop in RGDP. But I don’t like how the Fed is seemingly credited for doing a far better job in 1987 than in 1929. Like many Market Monetarist arguments, it seems too lazy and too circular. Basically it’s stated as an assumption that the central bank is the only variable. But the natural rate of interest is pushed down by all sorts of exogenous factors, such as the lack of deposit insurance. If a central bank has some highly bimodal distribution of NGDP, where NGDP is fine above the zero-bound but declined rapidly with natural interest rates below the zero-bound, then it’s blind luck that on central bank does not see a big NGDP drop and another does. I believe this describes Australia, which didn’t see natural rate drop below the zero bound for reasons having nothing to do with central bank policy. If you put the Fed there, they would have looked just as good. Same thing goes for exchanging 1929 and 1987 Feds.
20. October 2013 at 22:02
The 1929 crash had a lot to do with the Smoot-Hawley act looking like it would pass. Also, the tariffs enacted in 1930 left farmers with no one to export to due to retaliatory tariffs. As a result, incomes to rural America fell which put pressure on local Midwestern banks. The bankruns spread from there and wiped out a lot of money regardless of the Fed cutting rates and injecting liquidity. A great deal of the crash from 29-33 was about real effects creating monetary effects and showcasing the impotence of the Fed. I have no doubt that the economy would go through a pretty rough patch today if we suddenly had a tariff change like that regardless of what the Fed did.
21. October 2013 at 05:13
Steve, Bridges collapse all the time, that’s why we have good models of bridge collapses. The US stock market only falls more than 15% once, that’s why we don’t have good models of 20% stock market plunges in one day.
Matt. I don’t follow your argument. The zero bound was not an issue in 1930. And there is nothing “circular” about MM arguments. If you think there is, you need to take a closer look.
21. October 2013 at 05:15
John, You said;
“The 1929 crash had a lot to do with the Smoot-Hawley act looking like it would pass.”
That’s false, indeed during October 1929 the prospects for Smoot Hawley passing got WEAKER. That’s what I found in my research and that’s what a new study found, which will be published in the near future.
21. October 2013 at 08:00
Scott – I’ll look forward to your upcoming paper on the impact of Smoot-Hawley on the October 1929 crash.
In this application, scale invariance means that the magnitude of changes in stock prices does not depend upon either the time scale or the price scale. It simply doesn’t matter if you’re looking at price fluctuations over a year, month, week or within a single day – if you plot the price vs time data on the same size chart, erasing the price and time scales, you won’t be able to tell the difference between an event that happened in a single day and an event that played out over a period of years.
Another way to describe this characteristic is to say that stock prices have fractal properties. This Scientific American article written by the godfather of fractal mathematics provides a good general introduction to the concept.
21. October 2013 at 09:48
Well, my first response was to the idea that the 1987 crash was somehow efficient. Clearly, it wasn’t. Or if it was, then the market wasn’t efficient before. No way there was news that changed cash flows or interest rates/risk premiums 20%.
Then I realized the post dealt more with how the economy in general fares after such a crash. My impression is the Fed had simply no control over NGDP with the gold standard in 1929. They were at the mercy of whatever happened in the economy, IF they didn’t leave the gold standard. But I also think central banks today treat the zero-bound similar to the gold standard then: something that, once in effect, takes the economy out of the Fed’s hands. The Fed did eventually do something months after deflation at the zero-bound, but they always call the measures “extraordinary.”
That means there is a bimodal distribution of NGDP under the same Fed in 1980-2013 depending on what happens in the underlying economy. The rates in 1987 were far from the zero-bound and of course there was no gold standard. Therefore however much the crash decreased velocity was easily offset by the Fed. If it did push natural interest rates below the zero-bound or if, say, LTCM and then Lehman and other banks failed in the 90’s, an alternative history is possible where we had the 2008-09 NGDP crash a decade or two earlier.
The distribution of NGDP under the Fed SHOULDN’T be bimodal certainly, but that’s my point. It is bimodal and I wouldn’t necessarily credit the Fed in 1988 or the Australian central bank today for succeeding where today’s Fed failed. I think all three banks had the same distribution of NGDP based on natural interest rates, but in 1988 or Australia today, they’re lucky enough to not have natural interest rates below zero.
21. October 2013 at 17:40
Ironman, My paper was published in 1991, but it will also be in my upcoming book. I was referring to another paper.
I understand fractals, and that supports my point. It’s far more common for stocks to change by 20% over a month, than over a day. So it does matter. It suggests October 19th was very unusual. Especially since no other day even come close. Assuming normality it would be one in a zillion. Of course asset prices have fat tails, but even so . . .
Matt, There’s some truth to that, but of course the Fed isn’t really powerless today, and indeed wasn’t even powerless in 1929. The 1929 crash was actually (partly) caused by tight money.
22. October 2013 at 18:29
Daniel:
‘No thanks, I’m not interested in pseudoscience.”
Is that what you call “What I don’t understand”?
Don’t worry, you’re not needed.
15. December 2014 at 22:53
Similar crashes happen quite often but in different forms.
The article FORECASTING ECONOMIC CRISES explains very clearly the root cause of the 1987 crash. The mystery is now solved.
Click the link below to read it:
https://www.smashwords.com/extreader/read/453058/3/forecasting-economic-crises
Y T Wong
25. November 2015 at 13:40
[…] we would not look at interest rates; that is a totally misleading indicator. As Sumner tells us in this post, “Interest rates tell us nothing about the stance of monetary policy.” In context, he […]
26. November 2015 at 05:05
[…] we would not look at interest rates; that is a totally misleading indicator. As Sumner tells us in this post, “Interest rates tell us nothing about the stance of monetary policy.” In context, he is saying […]
1. March 2016 at 00:12
[…] TheMoneyIllusion » The 1987 stock market crash; myths … – 38 Responses to “The 1987 stock market crash; myths and reality” benjamin cole 19. October 2013 at 11:47. Excellent blogging…yes the 1987 crash is sui generis … […]