Algorithmic trading reshapes market liquidity by altering who supplies liquidity, how quickly orders arrive, and how that liquidity behaves under stress. Academic and regulatory research shows both net benefits in normal conditions and vulnerabilities during stress. Heather Hendershott of University of California Berkeley, Charles M. Jones of Columbia Business School, and Albert J. Menkveld of VU University Amsterdam find that algorithmic trading can narrow bid-ask spreads and increase displayed depth, indicating improved short-term liquidity provision. Their work attributes this to faster and more continuous quoting by automated market participants.
How algorithmic trading changes liquidity provision
Algorithmic systems, including high-frequency trading, act as rapid market makers that continuously post and update quotes. This increases the number of price points and the speed at which trades can be executed, reducing immediate transaction costs for many investors. Research by Maureen O'Hara of Cornell University and Marcos López de Prado of Cornell University highlights that algorithmic liquidity reduces the cost of trading for uninformed investors because automated strategies can manage inventory and hedging more efficiently than slower human traders. This improvement is especially pronounced in highly liquid, developed equity markets where order flow is dense and predictable.
Algorithms also enhance liquidity through sophisticated arbitrage that aligns prices across venues and instruments, improving price discovery. Empirical evidence reported by regulatory analysis of the 2010 flash event and subsequent studies by Andrei Kirilenko of the U.S. Commodity Futures Trading Commission shows that algorithmic participation contributes to rapid incorporation of information into prices, which is a component of functional liquidity. Faster price discovery does not always mean more resilient liquidity.
Risks, fragility, and market stress
Algorithms often supply liquidity only when markets behave within expected statistical norms. When volatility spikes or order flow becomes toxic, automated market makers can withdraw quotes in milliseconds, producing a sudden contraction in displayed liquidity. The U.S. Commodity Futures Trading Commission team led by Andrei Kirilenko documented how rapid withdrawal of liquidity amplified the 2010 Flash Crash, illustrating how liquidity droughts can arise when many algorithms respond in similar ways to shocks. Such behavior increases short-term price impact and can cascade into broader instability.
Regional, cultural, and structural market differences shape these dynamics. Emerging and thinly traded markets with fewer algorithmic participants can see larger relative benefits from a single automated liquidity provider but become more vulnerable if that provider exits. Market microstructure and regulation in different territories influence incentives, for example by setting maker-taker fees or offering protected venues, which affects whether algorithms post genuine liquidity or primarily exploit passive order-flow strategies.
Consequences extend beyond transaction costs. Better continuous liquidity supports tighter integration of global capital markets, lowering financing costs for issuers in mature markets. Conversely, fragility during stress raises concerns for retail investors and pension funds exposed to sudden price moves and for market integrity more broadly. Policymakers and exchanges respond by implementing measures such as circuit breakers, minimum resting times, and enhanced pre-trade controls to mitigate withdrawal risks, as discussed in reports from the U.S. Securities and Exchange Commission and the Financial Stability Board. Balancing the efficiency gains of algorithmic liquidity with safeguards against ephemeral liquidity remains a central challenge for market design.