Is September the cruelest month?

That is from Greg Ip, reporting the work of Luc Laeven and Fabian Valencia (pdf), feel free to regard it as a spurious correlation!


Does that make the recession now a "Never Ending September"?

It's not the only way we're stuck in September...

Ancient pattern of harvest in the Northern hemisphere is partly to plane as individual plans are discinfirmed collectively. Friedman and Schwartz discuss the seasonal adjustment pattern in "A monetary history".

I particularly like the way it resembles a one fingered salute.

Good news, Congress has introduced legislation to permanently remove the month of September from the calendar, thus preventing around half of bank crises.

I'm not pleased with the replacement though...

Lousy Smarch weather...

1. Bad harvest numbers cause problems.

2. Learning that you child's teacher is a moron causes hopeless despair for the future. A sell-off in the market results and traders build cabins in the woods.

3. Fall sucks, and you don't have enough food/profit to last you the winter. Sell-off ensues.

How many unique years make up that September spike? 40 years isn't very long to begin with. I haven't read the whole PDF, but page 11 figure 3 shows the 2008 crisis with about a quarter of the crises. I'm guessing they're counting September 2008 for a good chunk of these crises. That may explain the entire aberration. (There's a link to the full data but I'm on my phone.)

A comment on The Economist blog says 22 of the 25 Sept crises are from 2008. I can't believe Greg Ip, Justin Wolfers, Tyler Cowen and many others are actually spreading this graph hunting for some hidden meaning.

Oops, page 10, and I think it's absolute values, not percentages. But the more I look at this the lamer the above graph gets. The paper is supposed to analyze well over 100 crises, but just eyeballing the graph I see less than 40 non-September crises. So it looks like not only is September 2008 an outlier in a short data range, but the data isn't even complete for other crises in the range!

I had no clue that the world has had that many banking crises!

"seven years has gone so fast
wake me up when September ends"

That's when markets realize that winter is coming.

(apologies if that last wisecrack was too dry and/or esoteric...)

Also interesting is the rising trends towards the end of the last 2 quarters of the year. What is accounted for in quarters and yearly?

I don't know if the graph is build upon a solid and complete set of data.
In any case, there might be a correlation, but it'l be hard to establish a causation between Septembers and bank crisis. Or vice versa for that matter..

I remember reading about this phenomenon in A monetary history of the US. At the time I thought, "gee, the 2008 crisis started in the fall too". I think there may be something important here, but I'll be damned if I know what it is. I asked a trader friend about it, and he thought it had something to do with either regulators or traders being on summer vacation. I couldn't think of the causal mechanism, maybe everybody comes back at once and tries to adjust their portfolios at the same time then they all misinterpret the volatility. Not sure, but it seemed as plausible as any. I guess then you'd expect to see the same thing around Christmas.

Federal fiscal year begins in October. People looking to cash out beforehand?

Hmm, doesn't this mean one could profit from this by short selling or buying short-term put options on bank bonds or stocks in June-August?

If this arbitrage isn't possible, then you should also see bank stocks and bond going suddenly up in value at the end of September once it's clear that acrisis isn't coming, and then gradually come down as September comes closer: is this the case?

Totally stupid paper, this plot means nothing. It's all coming from September 2008, and the fact that baking crises were so correlated across countries at that time.

Comments for this post are closed