We ran a couple of posts last week about the declining daily trading volume in US equities the last few years (during a time in which share prices have climbed).
Kid Dynamite had a couple of good comments — one under each post — arguing that it could simply be the result of two trends in some combination.
1) There was a trading frenzy during some of the worst days of the crisis, and volumes were just correcting in its aftermath; and volumes are actually still just a tad above their pre-crisis levels.
2) To some extent you would expect higher volumes when the market was plummeting such lows — the same dollars were being traded. But looking at daily dollar-value traded would probably reveal that nothing big had occurred.
We’re flagging them again here because it seems he was right, certainly about the second point:
That chart comes from the Credit Suisse Trading paper that we’d been awaiting, and which was going to explain an alleged inverse correlation between volumes and share prices.
But it seemed to us that this inverse correlation had only existed since about 2008, not before — and as it turns out, that’s the period of time that Credit Suisse Trading is focusing on anyways:
Of course, this inverse correlation is necessary to get the mostly flat dollar-value traded amount in the last few years. Which leads the analysts to ask:
So what does this tell us about the future? Does this imply that the only way we’ll increase trading activity is to endure a sustained bear market?
Their answer, which we hadn’t previously given any thought to, is that a big reason for the recent relationship between prices and volume is the paltry number of stock splits in recent years.
Here’s their explanation:
1) Citi provided a real example of what actually happens after a reverse split – share price went up and volume went down proportionally (Exhibit 6). Assuming that the reverse would happen (higher volume with a lower price) following a regular stock split is a natural extension because asset managers are dealing with the same $value investment. They would have to trade more shares to move the same amount of money.
2) We showed on page 1 how lower prices tend to be inversely related to volumes. When prices drop, there is more uncertainty, forcing traders to adjust positions more frequently as they reassess risk levels. The opposite is true when prices rise.
3) Our own study of how splits – and, importantly, reverse splits – have impacted market volumes over the past few years suggests that if each day’s shares traded was adjusted by any relevant split ratio, today’s volume would be almost 10% higher than it is today (Exhibit 7). Of course, this assumes that trading decisions are unaffected by splits, and the results are dominated almost entirely by C, but they’re still compelling. A large number of regular splits – in line with the 100 per annum average of years past – might have a similar effect in driving volumes back up.
An interesting hypothesis and maybe it’s right, though we’d note that the reliance on the single example of Citi’s reverse stock-split is less than an ideal sample size.
The analysts also give a nod to some academic evidence that companies which split their stock end up with abnormally higher returns, as the split signals management confidence and makes the stock potentially more accessible to a new class of investors. But we remain somewhat sceptical about each of these points (and at any rate it can’t be healthy for companies to start splitting stock just to send a signal of management confidence that isn’t supported by actual management confidence).
And again, Credit Suisse appear to be assuming that the inverse correlation between prices and volumes from the past few years will hold — but they haven’t established anything like a definitive causal connection here.
The suggestive evidence, though, is admittedly interesting. Essentially the analysts have found that bid-ask spreads on stocks tend to start rising beyond a certain price, as market makers try to keep the spreads above a certain basis point limit. (E.g. if the minimum one-cent spread is fine for a $10 stock, it’s harder to justify for a $100 stock.) …
In fact, what we show in Exhibits 8 & 9 is that after a certain price, market makers will increase the spread (in cents) to keep the basis point spread above a certain floor. Our analysis suggests that there might actually be a sweet spot for stocks around $50-$70. We also find that, quite the opposite from hurting liquidity, posted size at the NBBO actually declines for expensive stocks.
This means that what might otherwise be a company’s argument to allow a stock’s price to rise indefinitely – that they will benefit from better liquidity and tighter spreads as the price goes up – may not be realizable after all. Another potential impediment to splitting is gone.
Other reasons they give for contributing to falling volumes are more commonly known: many fewer IPOs now than in the 90s, more companies going private, and more company buybacks the last couple of years — in other words, de-equitisation.
At this point, and with the admission that we’re no expert despite our interest in the topic, we think it’s too early to rule out anything. But some of you are experts, so have at it in the comments.
The mysteriously falling US trading volumes – FT Alphaville
Falling daily trading volumes – not so mysterious? – FT Alphaville