Panic season

For normal people in the U.S., August is about relaxing at the beach. But for grizzled practitioners of systematic equity strategies, August is the cruelest month. Ever since the “quant liquidity event” of August 2007, we spend August compulsively checking our phones and having nightmares about screens full of glowing red numbers.[1]

Even if systematic equities aren’t your thing, you need to be mentally prepared for an epic financial disaster over the coming three months. Historically, the period from August to October is when most U.S. financial crises have occurred, such as the stock market crash of October 1929.

Here’s a list of terrible things that have happened in the past fifty years in U.S. financial markets in the calendar months of August, September, and October:

• October 1987, stock market crash
• October 1997, Asian financial crisis
• September 1998, LTCM collapse
• September 2008, Lehman collapse

And here’s a list of terrible things in the other nine months of the year:

• March 2020, COVID

See the pattern?

Let’s do some back-of-the-envelope math. If we include August 2007, that’s five disasters occurring in August-October over the past fifty years, and one occurring November-July. So over the next twelve months, my non-rigorous analysis says there’s a 12% chance of an epic disaster, with a 10% chance of it occurring between August 2025 and October 2025 and a 2% chance between November 2025 and July 2026.

This seasonal pattern in disasters is not a recent development. Since the founding of the United States, its financial crises have disproportionately occurred between August and October. America’s first bubble, the “Scriptomania” of July and early August of 1791, was a speculative frenzy surrounding securities issued by the Bank of the United States. America’s first bubble was followed swiftly by America’s first financial crisis, when prices collapsed in mid-August 1791, and then America’s first bailout, as the Treasury stepped in with open-market purchases on August 17, 1791, as discussed in Miller (2018). Long before we had the Greenspan put or the Bernanke put, we had the Hamilton put, although Hamilton only succeeded in delaying what later blossomed into the Panic of 1792.

Subsequent financial panics in the U.S. all tended to cluster in the period August to October, when harvest occurred and money needed to flow from the big East Coast cities and into the Western agricultural regions. These included the Panic of 1857 (August), the Panic of 1873 (September), and the Panic of 1907 (October). As described by Sprague (1910):

With few exceptions all our crises, panics, and periods of less severe monetary stringency, have occurred in the autumn, when the western banks, through the sale of the cereal crops, were in a position to withdraw large sums of money from the East.

The tendency of financial panics to occur around October in the Northern Hemisphere has been noted at least since Jevons (1884), who blamed “the autumnal pressure in the money market.” Analyzing British markets, he said:

since 1825 all the severest pressures have either commenced or culminated in the last quarter of the year … prima facie evidence of a dangerous tendency in these months worthy of the deliberate attention of commercial men.

Jevons wanted the Bank of England to combat this seasonal “dangerous tendency.” Similarly, the U.S. Federal Reserve System was created in 1914 partly to prevent the seasonal occurrence of bank panics and to “furnish an elastic currency,” as discussed by Miron (1986).

The U.S. economy is no longer dominated by agriculture. So why do we still observe financial crises around harvest time? I’d say it reflects one enduring legacy of our agricultural heritage: summer vacations. Hong and Yu (2009) show that for North America and Europe, equity liquidity and trading volume are lower in August and September as market participants go on vacation.

My theory is that since liquidity is lower in August and September, we have a greater likelihood of observing a “liquidity event” then because markets are less able to absorb large orders. Let’s take the quant bloodbath of August 2007. One explanation is that a large Wall Street firm, stung by losses in subprime, tried to unwind a systematic equity strategy. Because equity markets were illiquid due to the absence of vacationing traders, these trades had high price impact and generated large losses for other market participants, causing them to also delever in a self-fulfilling cycle of unwinding.

While the summer vacation hypothesis is not a great fit for the stock market crash of October 19, 1987, it makes sense for many other crises, such as the Asian financial crisis in 1997 which began in Thailand in July and then worsened and spread through October.

Is there anything we should do in response to these historical patterns? Sell in May and go away? I doubt it. The optimal response to seasonal variation in return distributions is a complicated topic involving the relative strength of the seasonal patterns in risk versus reward as well as the trading costs of seasonally varying leverage.

I think the best we can do is to structure our portfolios to withstand adverse events, no matter when they happen to arrive. While disasters are more likely during panic season, they can strike in any month.

Let me conclude by quoting Mark Twain, who was familiar with the concept of financial disaster since he managed to destroy both his own fortune and his wife’s through various unwise investments. Here’s his view of stock market seasonality:

October. This is one of the peculiarly dangerous months to speculate in stocks. The others are July, January, September, April, November, May, March, June, December, August, and February.

Endnotes

[1] August 2007, like Satan himself, goes by many names. Quant quake. Quant meltdown. Quant crisis. Quantocide. Quantocalypse. Quant bloodbath. Quant extinction event. Quant liquidity crunch. For more details, see Daniel (2009).

References

Daniel, Kent. “Anatomy of a Crisis.” CFA Institute Conference Proceedings Quarterly vol. 26, (January 01, 2009): 11-21.

Hong, Harrison, and Jialin Yu. "Gone fishin’: Seasonality in trading activity and asset prices." Journal of Financial Markets 12, no. 4 (2009): 672-702.

Jevons, William Stanley. Investigations in currency and finance. Macmillan and Company, 1884.

Miller, Scott Christopher. “Never Did I See So Universal a Frenzy: The Panic of 1791 and the Republicanization of Philadelphia.” Pennsylvania Magazine of History and Biography 142, no. 1 (2018): 7-48.

Miron, Jeffrey A. "Financial panics, the seasonality of the nominal interest rate, and the founding of the Fed." The American Economic Review 76, no. 1 (1986): 125-140.

Sprague, Oliver Mitchell Wentworth. History of crises under the national banking system. No. 5624. US Government Printing Office, 1910.