Thursday, October 19, 2017

The Crash of 1987 -- Happy 30th Anniversary

I was at the center of the October 19, 1987 stock market crash, when the U.S. equity market dropped 20% in a single day. The second chapter of my book, A Demon of Our Own Design recounts my experiences during the 1987 Crash, and provides an explanation of what went wrong.

The essence of the problem was a programmatic positive-feedback strategy, namely, the dynamic hedging program of portfolio insurance, coupled with the time disintermediation between the speed of the index futures market in Chicago and the more deliberate pace of those in the equity market in New York.

This is not so different structurally from what we have now with the time disintermediation between the ETFs and the underlying cash markets in the less liquid markets -- which is not observed when we have two-way flow -- coupled with various positive-feedback strategies such as volatility targeting and various flavors of what are essentially momentum strategies. Or other structural issues within the current financial system.

So, below is an edited version of Chapter 2 of A Demon of Our Own Design.

(If you have arrived here by going to my blog rather than to this post specifically, you need to click on the post to see the excerpt.)


Excerpts from A Demon of Our Own Design, Chapter 2: The Demons of ‘87 
Richard Bookstaber
....
Portfolio insurance was commercially developed by two Berkeley finance professors, Hayne Leland and Mark Rubinstein.  With John O’Brien, their marketing partner, they founded a management company, LOR, in 1981 to sell their technique.  Within a few years it was programmed for action in the computers of some of the largest investment firms in the world. At the start of each day LOR sent their portfolio manager clients hedging instructions based on their runs of the Black-Scholes model.  The managers did the hedging themselves.
In practice, the equity manager initiated a hedge against his equity portfolio, usually using the S&P 500 futures contract as the hedging instrument. ... As the portfolio increased in value and moved above the floor price, the hedge was reduced, allowing the portfolio to enjoy a greater fraction of the market gain.  As the portfolio declined in value, the hedge was increased, so that finally, if the portfolio value fell well below the floor price, the portfolio was almost completely hedged.  Thus the portfolio was hedged when it needed it, and was free to take market exposure when there was a buffer between its value and the floor value defined by the exercise price. 
...
Since the basic option technology for portfolio insurance was well known, other firms followed LOR into the market to provide this hedging advice. I spearheaded the effort at Morgan Stanley.
...
This activity engulfed segments of the firm that rarely related to one another. I was in the Fixed Income Division marketing an equity product to investment banking clients and then managing the resulting programs as a fiduciary in Morgan Stanley Asset Management.  I ran programs for some of the firm’s blue chip clients, including Chrysler, Ford and Gillette. This strategy was considered to be at the leading edge of market innovation; rather than buying an existing security, portfolio insurance was ushering in the brave new world of creating synthetic instruments on the fly through dynamic trading strategies. 
...
The equity market was ripe for the promise of portfolio insurance, because there was much to insure. From 1982 to its pre-Crash peak in August 1987 the Dow Jones Industrial Average went on a bull run that nearly tripled the index.  The U.S. economy cooperated providing five years of uninterrupted economic expansion.  By 1987 the market was moving forward at an exponential rate; from the start of the year to late August the Dow rose more than 40 percent.
By mid-October, though, the promise of portfolio insurance began to look like a very good idea. From the close on October 16th, the market seemed like it came from a totally different world.  The Dow had already fallen nearly 500 points from its August high of over 2700, washing away nearly half the year’s gains.  And then decline became free fall.  The week preceding the 19th, the market dropped 4% on two separate days: On Wednesday, October 14th the Dow dropped by a one-day record 95 points, and on Friday the 16th a new record was set with a drop of more than 100 points. 
On the trading floors at Morgan Stanley, equity trading turned into a spectator sport.  Throughout the latter part of that week fixed income traders and salesmen filtered down the stairs from the 32nd floor to the equity trading floor, standing around to watch the frenetic scene.  The equity markets benefited from the built-in structure of having listed exchanges for stocks, futures, and options, and usually were much calmer than the fixed income markets, where every bond the desk purchased had to be taken  out to many clients to find a new home.  The crowds watching this train wreck amplified the crisis mentality. This finally led Anson Beard, the head of the Equity Division for Morgan Stanley, to post signs declaring that “Unauthorized Personnel Loitering in the Trading Area are Subject to Immediate Dismissal.”  That took care of the riffraff, but the firm’s Managing Directors still found their way to the floor.
And if they were on the floor on Monday, the 19th, they got an eyeful. The open of the futures market at 9:00 a.m. that day started a cascade of selling.  A half hour later, the New York Stock Exchange opened to an apparently insatiable demand to sell stocks as the NYSE tried to keep pace with the selling of S&P futures in Chicago.  The imbalance of buying and selling demand was so severe that many stocks did not even open, and the rapid decline in the price of the stocks that did, coming on the heels of the previous week, left most investors frozen in their tracks.  ...  By the end of the day the market had suffered its worst one-day percentage drop in history, down over 22 percent.  The S&P futures fared even worse.  The program trading that normally linked the futures’ intraday prices to the S&P cash market could not keep up with the selling demand in the futures pit, so the futures dropped even further – nearly 29 percent.  Overnight the panic spread around the globe to other equity markets.  In the 18-hour period after the New York market open, wealth equal several years worth of global GDP was wiped from the face of the earth.
The postmortem of the 1987 Crash is filled with reams of reports that have tried to fit it into the efficient market, information-based mold of contemporary economic dogma. ... But none of the analysis stood out, either individually or in aggregate, as the source for any sort of major rethinking of the market. The 1987 Crash simply was not the result of a rational reaction to new information.  What sort of information could have led the market to drop more than 20 percent on the 19th, jump 12 percent early the next morning only to fall another 10 percent in the following few hours?  Nor was it a matter of herd psychology. The moon and stars did not align to lead broad segments of the market to wake up Monday morning and decide to dump their shares.  In fact, a select and concentrated set of firms generated the selling demand on the 16th and the 19th.
...
It didn’t take a genius to see that the source of the Crash was market illiquidity, illiquidity that was the unintended by-product of the new and wildly successful portfolio insurance strategy—and one that I had helped to popularize and implement.
...
If a hedge cannot be readily adjusted, then obviously all bets are off, and for a hedge to be properly executed, the hedging instrument has to be liquid: some counterparty needs to take the other side of the hedge.
If one small portfolio uses this sort of stop loss strategy, liquidity will not be an issue. If everyone in the market is trying to do it, it can become a nightmare, a little like everyone one a cruise ship trying to pile into a single lifeboat—it won’t float; neither did the market. And that, in a nutshell, is what happened.  On Monday morning, October 19, 1987, everybody who was running a portfolio insurance program pulled out the computer runs from Friday’s market decline.  The record drop on the previous Friday had caught some firms flatfooted, and on Monday morning they had to make hedge adjustments to the decline, pouring sell orders into the S&P futures pit by the truck full.  But other such trucks had started down the road a week earlier, barreling into the option and futures markets from a distant part of the equity trading floor that was run more by rumors and personal contacts than by computers and mathematical models.
...
By Monday morning everyone was lined up at the gate to be the first to get orders filled.  Portfolio insurance firms sold nearly half a billion dollars of S&P futures, amounting to about 30% of the public volume.  The futures prices dropped precipitously, and the stock market had not even opened. About 15 minutes into the futures market decline, we started to see inexplicable activity from an unexpected quarter, cash-futures arbitrageurs. Their poorly executed attempts to capitalize on the apparent chasm between the cash and futures prices would be the red flag that triggered the stampede in the NYSE.
The futures desk on the equity floor had one junior member whose job throughout the entire trading day was to sit with a phone cradled on his shoulder.  On the other end of the line was someone whose job during trading hours was to watch the pit and relay the latest trade or bid and offer to the guy on the desk.  When a new level was reached, or unusual activity occurred, he would announce it in a monotonic cadence that was thankfully blended into the background noise of the trading floor.  But on the morning of the 19th we were all ears.
The cash-futures spread was reaching levels that were many-fold what was generally needed to make an arbitrage worthwhile, so a host of cash-futures traders began to bring their orders to our program desk.  The only problem— and in our minds it was a big problem—was that the stock market was not even open yet.  The discrepancy these traders were observing was based on the current futures price versus the price of the stock market on Friday’s close.  The traders were basing their actions on stale data; there could be no telling where the stock market would actually open.  They put in orders to sell at the market price at the open, under the assumption that the open would be close enough to the Friday close to still make the discount in the futures contracts a profitable trade.
That was a big bet, and a far cry from the relatively low risk enterprise of the usual cash-futures trade. And in this environment, it was even more risky because when the stock market did open, it was an absolute lock to open down.  The execution of the program trade would then be complicated by the downtick rule, which proscribes short selling a falling stock.  The arbitrageurs wanted to buy the futures and sell the stocks short against them.  If the market is in free fall, up-ticks are few and far between, and there are many short sellers trying to squeeze in their execution.  It can take a long time to get a trade off.  In the meantime, the long futures position is being held unhedged.  If the market drops, the trader loses.
The portfolio insurance hedgers found the other side of the market in the cash-futures traders and market makers, and these traders in turn were depending on the stock market to hedge out their bets in the futures.  In effect the cash-futures traders were taking the market impact from the futures pit and transmitting it back to the individual stocks on the NYSE.  By buying the futures and then selling the individual stocks, the individual stocks would finally “feel” the impact that was being implied by the intense selling pressure in Chicago. In theory this arbitrage is a natural market mechanism for tying the cash and futures stock markets together, and up to Monday the 19th it worked smoothly.  But on the 19th the speed and magnitude of the normal smooth waves of selling metamorphosed into a tidal wave that rushed in from Chicago.  It was more than the stock market could absorb.
...
Program traders and arbitrageurs take positions on the S&P contracts trading in the futures pit while simultaneously taking opposite positions on the individual stocks that comprise the S&P on the NYSE.  When the S&P futures contract sells for less than the price of the basket of the individual stocks in the S&P, then the cash-futures arbitrageur buys the S&P and sends in orders to sell the individual stocks. If the price difference is greater than the transaction costs of doing this trade, then they make an almost certain profit.  This trade effectively transfers the stock market activities of the futures pit to the individual stocks on the NYSE.  It is here where things broke down in 1987, and broke down for a simple reason: stocks are not as liquid as futures.
The problem was that the traders in the S&P pit are mostly market makers, jammed together gesticulating and shouting out orders in hopes of scalping a few ticks.  They thrive because of their quick reactions to the market and their speed of execution.  By contrast, equity investors of the day who frequented the NYSE were not particularly focused on speed of execution, nor were they concerned with the minute-by-minute movement of the market. As the futures traders reacted to the market and the cash-futures arbitrage traders transmitted that activity to the NYSE floor, the flow hit a wall. The specialists could not unload their inventory into the stock market as quickly as the selling was shoveled over to them from the futures pit because their clientele was not glued to their screens, ready to react in mass.  The futures market was operating in broadband and the NYSE on dial up.
The specialists tried to elicit more buyers by dropping the price, but there was a limit to how much more buying interest they could attract. No matter how quickly the price was dropped, the decision making by the equity investors took time; not all equity portfolio managers sat glued to their screens, and unlike the twitch-quick futures pit traders, they made portfolio adjustments only after reasoned consideration.  With their limited capital the specialists were not willing to wait for the process to unfold, and their increasingly aggressive offers ended up backfiring.  Prices dropped so violently that many potential buyers started to wonder what was happening and backed off completely.  In pushing so hard, the specialists’ actions became cause for suspicion, which then fed and finally betrayed their own panic. The root dynamic was time disintermediation—the time frame for being able to do transactions was substantially different between the futures market and the equity market, yet these two markets had been linked together through market arbitrageurs.
On our trading floor we could see that the drop in price was not having the desired effect.  In fact, it was scaring investors away. One of our institutional clients in Boston was bullish on IBM and had discussed strategies for adding more of the stock to his substantial portfolio. His salesman tried to grab him as IBM started to tank, but he was off his desk in a meeting.  A second call could not locate him.  We could imagine him heading off to grab some coffee and leaf through his morning faxes, unaware that the markets had begun to slide down around him.
Back at the NYSE, a day’s worth of activity had passed in what seemed to be 10 minutes. The specialists were starting to panic. A flood of sell orders was coming in from the other side of the cash-futures arbitrage and there were no where near enough buyers coming in to take them off of his hands.  With price is his only tool, the specialist dropped IBM another point, and then two more points, to try to dredge up some buying interest.
The portfolio manager in Boston finally got back to his desk, saw the beating IBM was taking on the open and gave us a call. If IBM had been down a half point or a full point, he would have put in an order, and would have been provided the other side of the trade that the specialist needed. But with IBM and other stocks in freefall he hesitated to buy, waiting instead to get a read on what was going on with IBM and the market generally.  As he spoke with us, he was interrupted and returned to announce that because the downswing was accelerating, the fund’s director of equity investments asked that no trades be executed until all the portfolio managers could meet to assess the situation.
Since this client worked for an asset management firm with a long-term investment horizon, he could put on a position just as easily tomorrow as today.  He watched the market’s downward cascade with something likely approaching detached curiosity.  As the markets fell further, from 5 to 10 to 20 percent, he likely felt some panic.  But for the moment he and many of his counterparts were on the sidelines. Meanwhile for the specialist, more shares piled up in inventory with each passing minute.  Other specialists were faced with the same onslaught and prices fell all around, so now the IBM specialist found that eliciting buyers was even more difficult, since he had to compete with the other falling stocks for attention. 
It is not long before the offer price for IBM, Big Blue, the bluest of chips, was down 10 points from the open. The result was a disaster. The potential liquidity suppliers and investment buyers were being scared off by the higher volatility and wider spreads.  And, more importantly, the drop in price was actually inducing more liquidity-based selling.  With each point drop, the portfolio insurance programs triggered more selling, and the portfolio insurance managers threw more sell orders into the futures market. Because of the dislocation between the hair-trigger execution of the futures and the ponderous decision making on the cash equity side, compounded by the insufficient capital of the specialist to bridge the gap between the incoming supply and the timeframe of the potential buyers, the specialist dropped the price of IBM too quickly. The potential suppliers who could have taken on the selling demand – and who would have been willing to do so with modest price concessions, had the move been more gradual – got spooked, and the portfolio insurance hedgers demanded even more liquidity than they would have otherwise.
A price drop is normally the dinner bell for buyers. So precipitous drop should have had traders licking their chops.  But it doesn’t work that way; if prices drop too far and too fast, it sends the wrong signal to the potential liquidity suppliers.  Rather than taking the drop as an indication of liquidity demand, they viewed it as – or at least heavily weighed the possibility—that the fall was a result of some new market information.  In classic “market for lemons” style, they viewed themselves at an informational disadvantage, and elected to stay out of the market.
Replay this mutual fund manager’s reaction over many times, and you basically have the fault line of the crash of 1987. Selling demand increased as prices dropped because of the pre-wired hedging rules of the portfolio insurance programs. Supply dried up because of the difference in time frames between the demanders and suppliers.  By the time equity investors could have reacted to the prices and done some bargain hunting the specialists had moved prices so precipitously that these potential liquidity suppliers were scared away. The key culprit was the difference in liquidity because of the different trading time frames between the demanders and the suppliers in the two markets. If the sellers could have waited longer for the liquidity they demanded, the buyers would have had time to react and the market would have cleared at a higher price.
...
BAD GAMMA
The problems created by portfolio insurance were compounded by a feature of the option strategy being implemented.  When the portfolio is far from the floor price, the hedge is small and changes in the value of the portfolio require only a small adjustment in the hedge.  This is because the position has a low likelihood of needing to be protected and because there is a lot of room between the current portfolio value and the floor value to ramp the hedge up.  As the portfolio value drops and nears the floor, the size of the hedge increases and the amount of adjustment that is made for any change in the market increases as well.  Another way to think about this is that the change in the value of the option being created (because that is what the hedge is doing) with a change in the underlying security (in this case the market) varies depending on how far the option is from the exercise price.  In the mathematics of option theory this change in the amount of the hedge with a change in the price is called the option’s gamma.  As the market declined and brought the portfolio insurance programs closer and closer to the floor, the impact the portfolio hedges, bad as they were early in the day, got worse and worse.
Nothing could change the momentum. The lack of program trading caused the discount in the price of the futures relative to the cash equity market to reappear.
...
In the midst of the market disconnect, the inability to get timely execution, the vanquishing of many of the liquidity providers, and the fears of a total break in liquidity through an early market close, the portfolio insurance programs continued robotically to spit out sell orders, oblivious to anything but the current market level and the mathematical requirements of the hedge. In the last 75 minutes of the trading day, the DOW dropped by 300 points, three times as much in a little over an hour as it had in any other full trading day in history.

THE PHYSICS OF THE MELTDOWN
I spent the week chained to my desk, my eyes frozen on the Quotron screen as I struggled to maintain the hedges demanded of the portfolio insurance programs I ran. Prices were moving all over the place, swinging more violently minute by minute than they usually did in an entire day, and the spread required to buy or sell the S&P futures – still the most liquid instrument in the equity market – was a dollar or more, 20 times normal. I had to weigh the implications of holding off on a hedge adjustment on the one hand with the incredible transaction costs in executing in the market on the other.
The huge volatility of the market broke down all but the most fundamental relationships between the market securities.  The usual day-to-day world where investors cared about subtleties like corporate earnings or analyst forecasts dissolved as the energy of the market was turned up.  All stocks moved together; if it was a stock, it was sold.  The market hardly differentiated between domestic and foreign, small cap or large.  It was like plasma physics: As matter becomes hotter it becomes less differentiated.  The forces that bond atoms together in the form of molecules are overwhelmed, so that rather than having a myriad of different substances, we have the elemental building blocks of the atoms.  Turn up the heat further and the atoms themselves are melded into plasma, positively charged ions and negatively charged free electrons; matter in its most uniform and non-differentiated state, no longer hydrogen atoms and oxygen atoms, just a seething white-hot blur of matter.
Just as high energy physics creates a state that is no more differentiable than to say that it is matter, so the high energy in the financial markets created a world where securities were no more differentiable than that they contained risk.4
This melding even extended beyond stocks.  High yield bonds, which usually tracked fairly closely to Treasury Bonds, suddenly became simply high-risk bonds, and traded just like stocks.  Meanwhile, Treasury Bonds, the anti-matter of the world of risk, were grabbed in the flight to quality, and traded up in price.  This behavior demonstrates a characteristic I have observed and expounded repeatedly: as the market moves into crisis, the absolute value of the correlation of assets approaches one.  The problem is you cannot always predict ahead of time if the correlation will be one or negative one.  One asset might end up hedging another, or it might end up doubling your exposure.
...
The surprise for Morgan Stanley was that our biggest losses came not from equities, but from high yield bonds.  The flight to quality moved investments away from equities into Treasury bonds, with the result that as equity prices declined, bond prices shot upward.  In the aftermath of the 19th, the interest rate on 90-day Treasury bills dropped almost two percentage points to just over five percent and the benchmark 30-year Treasury bond shot up by over 11 points. Corporate bonds did not share in the shift to fixed income; they went south instead.
...

October 19th ushered in, spectacularly so, the context for a type of risk that would embrace the markets in the future.  At the root of the Crash was computer-assisted liquidity in the form of rapid, programmed, cash-futures execution, and an innovative analytical products in the form of the option model’s differential equations applied to construct synthetic put options.  The market was developing to allow lightning fast reactions while at the same time driving complex innovations in market instruments and strategies that required ever more time for investors to analyze and absorb. This combination of speed and complexity would be the source of many future crises.

No comments:

Post a Comment