This year, we’ve already heard from two of the largest companies in the U.S., with some of the highest-priced stocks, that they will do significant (20 for 1) stock splits. So today, we update our work on why stock splits make sense for traders, investors, and issuers.
It's all about improving tradability by reducing frictions and distortions, so investors keep more of the returns on their purchases. That, in turn, helps reduce cost-of-capital for companies too.
An uptick in splits
As we’ve discussed before, stock splits have significantly declined since the dot com boom of the late 90s across all market cap sizes. The biggest drop seems to have occurred around 2008. For larger-cap S&P 500 stocks, over the same time, splits decreased from around 200 per year to just 20 per year.
Chart 1: Since Reg NMS became effective in 2007, automating trading, stock splits have declined
The drop around 2008 coincides with the global financial crisis, where stocks fell over 50% at their lows. However, the decade from 1999 to 2009 was also when trading and settlement systems were becoming automated; starting with decimalization in 1999 and Reg NMS in 2005, algorithmic trading took off. It's possible the widespread adoption of computers, which could chop orders into ever smaller pieces, removed the need to keep round lot values consistent for traders.
However, 14 years of un-split price appreciation has increased the average price of S&P 500 stocks. After trading in a range between $31-$50 before 2011, the average price has increased consistently to be just below $200 now. That has started to create new trading problems, as we discuss below.
Splits outperform around the period of the split
There are still a few good reasons why a company might want to split its stock.
One of the most compelling is economics. Research shows that if you can make a stock trade more cheaply, those stocks become more attractive to investors. That broadening of the investor base improves a stock’s valuation and reduces the costs of capital for the issuer. In fact, a study in 2009 found that liquidity improvements following stock splits reduced average companies’ cost of equity capital by 17.3%, or 2.4 percentage points per annum.
In short, splitting stocks changes how they trade fundamentally.
Below, we expanded our earlier research to cover the last 23 years, and it still shows stocks that split outperform the market significantly. Just announcing a split causes the average stocks to outperform the market by 3%, indicating the market expects gains even before tradability improves. Over the next 12 months, they outperform another 15%, adding to 18% in total.
Chart 2: Stocks doing splits outperform the market even a year later
How does tradability improve around a split?
Even though, mathematically, a split makes each 1-cent “tick” worth more (in percentage terms), the reality is most splits occur in stocks that already trade multiple ticks wide (see Charts 5 and 8).
Data shows that if you can make each tick matter more, bid-offer spreads actually get more competitive, causing spreads to fall an average 22%. In short, even though each tick is larger, there are fewer “empty” ticks between the bid and the offer.
Lower spreads, in turn, help increase trading, with value traded rising by an average of 18%. Combined, that makes it easier and cheaper for institutions to build larger positions.
Our earlier research suggests intraday volatility also falls, but tick data back to 1998 is difficult to obtain – so in our expanded analysis below, we look at the daily high-low ranges for stocks, which are virtually unchanged.
One finding that is consistent across decades of research, and counterintuitive given that liquidity improves and spreads tighten, is the finding that close-to-close volatility increases. There are many different theories that try, mostly unsuccessfully, to account for this effect.
When AAPL and TSLA last split their stocks in August 2020, that’s also exactly what happened too (see this study, Chart 6).
Chart 3: Splitting stocks changes their tradability
Why does tradability improve around a split?
A simple way to look at where market frictions are highest is to compare spreads across stocks with different prices. The data shows that spreads form in a U-shape, where stocks with prices that are too low or too high have much wider spreads than stocks priced “just right.”
Not all stocks need to split, though. Many tick-constrained stocks should instead reverse split.
Chart 4: Stocks see a U-shaped spread curve
The reason spreads form in a U-shape is because of some “one-size-fits-all” rules in U.S. markets. All “ticks” are 1-cent, and round lots are 100-shares, regardless of a stock's price. That makes the economics of trading very different, as we show below with links to our studies for those who want more detail:
- Low-priced stocks: 1-cent tick on a $2 stock creates a minimum legal spread of 0.50%, and a round lot costs $200.
For an investor, crossing the spread to buy and again to sell reduces their returns by a whole 1%. That’s a significant cost, and as a result, these stocks tend to attract spreads capture, leading to longer queues or more fragmentation and hidden trading. All of which makes it harder for investors to find liquidity to complete their trades. These stocks have a tick-constrained problem.
- High-priced stocks: In contrast, a 1-cent tick on a $2,000 stock costs only 0005% but a round lot costs $200,000.
Even the most liquid stocks in the U.S. trade with spreads closer to 0.01%. In short, 0.0005% return for spread capture is insufficient to attract market makers. It is also no deterrent to new buyers bidding 1-penny better (pennying) to get queue priority over an investor who has been waiting longer – and that increases opportunity and signaling costs for the first buyer.
In addition, we know that supply and demand for stocks forms in a V-shape, so the higher cost of a round lot, required to set NBBO in high-priced names, demands a wider spread. That, in turn, leaves space for many odd lots at better prices than the NBBO, which not all investors can see, and best-ex rules ignore, leaving small investors potentially worse off and companies doing buybacks struggling to post at the true best-bid.
What is the optimal stock price?
Clearly, there is a “sweet spot” where the price is neither too low nor too high.
Looking at Chart 4, we see the “optimal price” forms around where the tick (blue line) is half of the spread (orange line). That’s consistent with recent studies by academics, traders as well as regulators in Europe, who found that “an appropriate tick (leaves) spread between 1.5 ticks and 2 ticks for liquid securities and between 1.5 ticks and 5 ticks for less liquid securities.”
It's also consistent with our own research looking for the perfect stock price, which shows that, if you rank stocks by daily value traded:
- Not all stocks are the same. In fact, stocks with more liquidity see much tighter spreads.
- But at each level of liquidity, yellow dots (those stocks with spreads around 2 ticks) generally have tighter spreads than other dot colors.
- Tick-constrained stocks (blue dots) have wider spreads, with the most tick-constrained (largest circle size) having even wider spreads for similar liquidity (across any vertical slice).
- For high-priced stocks, as grey dots get darker (more ticks inside NBBO, despite it being easier to penny the market), spreads in basis points also get even wider, with black dots generally having the widest spreads of stocks with comparable liquidity (across any vertical slice).
Chart 5: The perfect stock price is different depending on a stock’s level of value traded
This means every stock has its own optimal stock price (brown arrow) that we can calculate using the liquidity of the stock. For example, a stock trading $1 billion each day should target a spread around 1 basis point. A stock price of $100 results in the 1-cent tick costing 1 basis point.
Using that, with some simple math, we can compute how many shares that company should have so their price is around $100-$200, meaning they trade with a 1-2 tick spread. We then use that to compute a split ratio.
Looking at what this means for real stocks in Chart 5, note that AMZN, AAPL and MSFT all trade roughly the same value each day ($10 billion). However, AMZN trades with a much wider spread (5bps) than AAPL and MSFT (both around 1bps). The chart also shows that AAPL and MSFT trade with a 1-2 cent spread (yellow dots), meaning they are near their optimal stock prices already. While AMZN, with a stock price over $3000, has a spread near $2.70 (black dot).
As we note below, AMZN has already announced a 20:1 split. Based on that, we expect its price to fall to around $150 and its spread to fall to around 1 cent, narrowing from 5 basis points to less than 1 basis point.
Research we have done shows that splitting to the optimal price is important, even for reverse splits, where we found that companies that split to near their “perfect price” saw spreads tighten, while those that split relatively far from their “perfect price” saw spreads widen.
Chart 6: Reverse splits that missed their perfect stock price saw tradability and spreads worsen
Which stocks cost investors the most additional trading costs?
The reason wider spreads cost investors is because, even though investors use sophisticated broker algorithms that allow them to capture some spreads, research suggests they need to cross spread 20% more often than they capture it (Chart 7) in order to complete trades.
We also know from other studies that institutional U.S. investors (U.S. pensions and mutual funds) make up 21% of the market by trading. So we know what proportion of daily volume represents trades where spreads add to institutional costs.
Chart 7: Even with sophisticated algorithms to work orders, investors mostly pay spread more than they capture it
We can also use the perfect stock price formula to estimate how much narrower a spread could be if it split to its optimum level. So, we know how much spreads could compress.
By multiplying the expected mid-spread saving by the net proportion of shares institutions cross the spread with, we estimate the costs to investors of suboptimal stock prices and artificially wide spreads.
We estimate that market-wide, investors are spending $4.9 million extra per day (in stocks trading over $1 million per day) due to artificially wide large spreads caused by stocks not splitting to their “perfect price.”
Our results show that the top around 20% of the costs comes from the top 20 stocks, which we show in Chart 8. The chart is ranked by savings, shown for each stock by the height of the green bar. Those savings are a combination of the amount of trading each stock does (green bar width) and how wide each stock's spread is now, in cents (purple dots).
Our model suggests that all stocks should see spreads compressing over 90% (lightest green shade).
Chart 8: Ranking stocks that would help investors most if they split
TSLA, with its high liquidity, ranks first, with an estimated excess cost to investors of over $200,000 per day. AMZN, with a much wider spread, but less activity, ranks second, adding over $170,000 each day to investors' costs. Alphabet’s two tickers rank fourth and fifth.
Of course, AMZN and Alphabet are two of the stocks that have already announced a split in the summer of 2022. Our model suggests their spreads will fall significantly after the split, with both likely trading with a bid consistently just 1-cent below the offer.
In addition, Tesla is set to vote on another split at its 2022 annual shareholder meeting, likely in the fall of 2022, after the stock gained approximately 140% since its last split back in August 2020. AAPL also split in August 2020 and has gained approximately 33% since
Table 1: Upcoming Splits
Why is this important?
Although it is easy to think that market structure doesn’t matter to investors, this research shows something as simple as the right stock price and spreads can save investors millions each day.
Lower trading costs boost liquidity, which improves investor returns.
And if U.S. stocks are more attractive to investors, company valuations will improve, and costs of capital will fall. It all adds up to making the U.S. market even more efficient for issuers and investors.