This piece was co-authored by Atharva Nirdesh Varshney, MSc Business Analytics graduate at Warwick Business School.
There are two, much-publicised trends at play in equity markets. One is the shrinking of the settlement cycle, the period between executing a trade and exchanging securities for money. The other is extension of the trading day.
Longer trading days and faster settlement cycles may seem unrelated but they both potentially have major implications for market liquidity, if taken to the point where markets operate 24/7 and trading/settlement happens instantaneously. There is a potential
model for dealing with the liquidity issue but before discussing the solution it is necessary to understand the nature of the problem.
Last year saw the US shorten the standard settlement cycle from two business days after the execution of a trade (T+2) to one (T+1). This was copied by other markets including Canada and Mexico.
Recently the European Union and the UK have announced plans to move from T+2 to T+1, a move attributed to the need to remain competitive. Settlement cycles have historically existed because of the need to carry out all operational tasks that need to take
place between agreeing a sale and exchanging funds for securities. As well as all transactions to be netted.
Automation allowed the settlement cycle in the US to shrink from T+5 to T+3 in 1995 and to T+2 in 2017. The move to T+1, however, required a fundamental change to business process rather than automation per se.
A key part of the overall trading process is stock lending. When a trader wishes to ‘short sell’ (attempt to profit from falling share prices), they need to borrow the stocks that they sell in order to deliver them to buyers.
Lenders, however, have the right to ask for the stocks lent back at any point. This is known as issuing a recall. Usually, a lender recalls shares after they have sold them. The borrower needs to obtain shares, and deliver them back to the lender before
the lender has to deliver them. This process came close to breaking due to the move to T+1 and was only fixed by extending the recall processing time to 11:59 pm.
Shrinking the settlement cycle further makes life even more problematic for short sellers. In general borrowing stocks needs to be agreed and the borrow settled in an even shorter cycle than the sale process. If the settlement cycle shrinks to T-Instant,
i.e. trades settle instantly, the current model of short selling would break.
Problems would arise for buyers of equities as well, because many of them borrow the funds used for purchases. Funding is currently a process done one or two days in advance based on expected settlements of trades.
In the existing model of prolonged settlement cycles, the movements of securities and funds from multiple transactions can be netted together, reducing the total amount of funds and shares that need to be delivered. Fundamentally sellers could only sell
shares in their possession and buyers could only buy with the funds at hand, greatly reducing market liquidity, and there would be no netting.
Though the length of settlement cycles has gradually reduced over the course of time, the big driver for T+1 ultimately came from the largely unregulated ‘wild west’ of cryptocurrency trading.
Cryptocurrency markets operate 24/7 and many of the traders in those markets also became traders of stocks on platforms such as Robinhood. The ‘pump and dump’ crypto trading techniques were applied to trading low value, often near bankrupt shares, the so-called
’meme stocks’.
The volume of trading on meme stocks and their price volatility caused major instability in the overall equity market in 2021. The length of the settlement cycle and the degree of market volatility meant Robinhood had to make peak margin payments of $3 billion,
an amount exceeding its capital.
Only a quick infusion of new funds into Robinhood stopped a major disaster for the market. Reducing the settlement cycle to T+1 was done to reduce systemic risk by reducing the duration of settlement risk and the amount required to fund margin calls.
Ironically, it is the demands of those that have ‘grown up’ trading crypto 24/7 that is now driving the trend towards longer trading days for more conventional assets.
Robinhood is leading the way again. It has partnered with Blue Ocean ATS (BOATS) to allow 24-hour trading of equities during weekdays.
The New York Stock Exchange (NYSE) has also announced plans to extend trading hours with its new
ARCA exchange. This may seem like progress but the problems with extended trading hours and market liquidity are well known. BOATS highlighted the following risks for potential customers
as:
- Greater price volatility
- Lower liquidity
- Wider bid/ask spreads
- News announcements out of hours leading to exaggerated and unsustainable effects on price
- Risk of changing prices – the prices determined by out-of-hours trading are ignored by the main markets when they open in the morning
All of these risks essentially could be looked on as aspects of low market liquidity.
The problems of market liquidity have long been understood in equity trading. Stock exchanges and related trade venues operate within fixed hours. Generally, the smaller the exchange and the less liquid the market the shorter the trading hours. A very small
exchange such as Zambia’s Lusaka Stock Exchange lists only 20 shares and is only open for two hours per day. By contrast NYSE is open from 9.30am to 4pm.
The main reason for restricting trading hours is fundamentally the same reason exchanges were invented in the first place, i.e. concentrating as many buyers and sellers together in the same place at the same time.
Though that ’place’ is now virtual rather than physical, concentration of buyers and sellers makes a market operate more smoothly and provides better price discovery. If buyers and sellers come to market at different times, there is likely to be more dramatic
price movements and greater spreads. Even larger markets, such as the London Stock Exchange, reduced the length of trading hours in 2020 to maximise liquidity.
So, what approach could potentially facilitate longer and faster trading? Regardless of the technology used for implementation the most important thing is to deal with the problems of market liquidity upfront.
Extended hours could be dealt with by switching over to a model entirely based on auctions. The period between auctions would be a function of the security traded and the time of day.
For heavily traded securities during normal business hours, an auction may run every few seconds, for a less liquid security in the middle of the night, orders would feed into the auction over the course of hours. Using an appropriate pricing model in the
auction could also smooth out some of the extreme volatility currently found in out-of-hours trading.
Auctions for buying and selling, would by definition require automated auction models to provide funding (based on using existing security positions as collateral) and borrowing/lending of securities. The world is already moving towards more effective intra-day
funding using tools such as BNY Mellon’s intra-day tri-party service but an intra-day borrow/loan market would require changes in the way that market operates such as greater use of fixed term trades.
As well as the liquidity issue there would also be a requirement to drive out all the operational friction in the processing of trades. Currently settlement cycles are also needed to resolve problems such as mismatched trades, incorrect static data and errors
in inventory data. There are already proven models for reducing post-trade friction to a minimum. These include the Turkish and Indian markets which have much simpler models of recording and transferring ownership of securities and as result less error prone
operational processes. Their techniques could form the foundation for a new trading model.
Finally, going against the dreams of decentralised trading, it would require trading for any given security to be concentrated into a single auction process rather than spread across multiple exchanges, ATFs and Dark Pools.
Gillmore Centre for Financial Technology members have created a paper that describes the overall
algorithms to implement this type of model as well as implemented a working prototype of the
interacting auctions.
The model also has some other surprising features. Speeding up settlement requires slowing down trading from the current milli-seconds but this would create a fairer market; one where those ordinary traders would not be at a disadvantage to the algo fund
that makes money by having faster connections and more powerful systems. Operational friction can also be essentially eliminated, potentially saving billions in costs.
This is the latest in the Gillmore Centre Series, in which authors from the Gillmore
Centre of Financial Technology at Warwick Business School examine new innovations in fintech from an academic perspective. Keep an eye out for more articles from the Gilmore Centre to learn more about new developments in the field.