By Matteo Aquilina, Adviser for Financial Stability, Bank for International Settlements, and Khaladdin Rzayev, Senior Finance Lecturer, University of Edinburgh and Koç University
The speed race in financial markets carries a hidden price tag, and it does not affect every company in the same way.
Financial markets play a vital role in the real economy by efficiently allocating capital to its most productive uses. But as trading velocity approaches the speed of light, do these markets still fulfil their primary purpose? Despite the importance of this question, the debate around high-frequency trading (HFT) has largely stayed within the market itself, focusing on HFT’s effects on traders and measures of market quality, such as bid-ask spreads, price discovery and market depth.
In a study we published with other researchers, we take a different approach. Instead of looking at traders, we ask what HFT means for issuers: the firms trying to raise capital.
In a study we published with other researchers (Aquilina, Ibikunle, Rzayev and Wang, 2025), we take a different approach. Instead of looking at traders, we ask what HFT means for issuers: the firms trying to raise capital. We find that the answer is not uniform. While HFT raises the cost of capital on average, these effects vary significantly across firm characteristics, which has important implications for financial regulation.
The speed race and why it matters
High-frequency traders now account for a substantial portion of global trading volume, investing enormous resources to shave mere microseconds (millionths of a second), and increasingly nanoseconds (billionths of a second), off execution times—quantities that are difficult to comprehend for humans, but extremely important for computers. In modern markets, a competitive advantage requires superior speed just as much as superior information. But how does this technological arms race affect the corporations issuing equity? The cost of capital, which is the return investors require to hold a firm’s shares, is arguably the most important metric financial markets produce. It dictates corporate investment decisions, determines which projects receive funding and ultimately shapes economic growth and job creation. If HFT moves this underlying cost, its impact extends far beyond the trading floor and directly alters the real economy.
Natural experiments in market speed
Studying the effects of HFT presents a major methodological challenge: endogeneity. Since HFT firms naturally target stocks with specific traits, simple correlations between trading activity and market outcomes often reflect selection bias rather than true causality. We resolve this by treating two major technological upgrades at NASDAQ (National Association of Securities Dealers Automated Quotations) as natural experiments, specifically the 2005 rollout of co-location services and a critical 2010 improvement in data latency.
Both upgrades dramatically boosted HFT activity on the exchange by reducing execution times for firms located near NASDAQ’s matching engines. More importantly, the timing of these technical changes had nothing to do with the fundamental traits of the individual listed companies. As a result, they provide a setting similar to that of a controlled experiment, allowing us to isolate the direct impact of HFT on the cost of capital.
Building on these shocks, we use a difference-in-differences framework to compare the affected NASDAQ stocks against an unaffected control group of NYSE (New York Stock Exchange) stocks. To ensure that the comparison is robust, we match the samples based on firm characteristics known to drive the cost of capital, making the two groups highly comparable across all observable dimensions.
The headline finding: HFT raises the cost of capital
Using two cost-of-equity measures (based on the Capital Asset Pricing Model [CAPM] and the Fama-French three-factor model [FF-3]) alongside the weighted average cost of capital (WACC)—which includes the cost of debt in the estimate—our baseline results show that increased HFT activity leads, on average, to a significantly higher cost of capital for NASDAQ-listed firms relative to their NYSE counterparts (see Figure 1). The magnitudes are economically meaningful. Following the 2010 latency reduction, the cost of equity rose by roughly 4.1 percent (CAPM) to 7.5 percent (FF-3) relative to pre-shock levels, while WACC increased by 3.1 percent. This smaller effect on WACC is entirely expected: HFT primarily impacts equity markets, meaning the inclusion of debt components naturally dilutes the overall effect. We observe comparable results for the 2005 co-location upgrade, which produced a 7.3-percent increase in the Fama-French measure.
These findings are interesting, yet at first glance, they appear counterintuitive. Since HFT is widely credited with enhancing market liquidity, one might expect it to lower required returns. To understand this result, we have to look beneath the aggregate numbers: HFT’s impact varies drastically depending on the stock’s characteristics.
Two channels, two kinds of stocks
We identify two distinct channels through which HFT affects the cost of capital, and they operate in opposite directions.
We identify two distinct channels through which HFT affects the cost of capital, and they operate in opposite directions.
The first channel is the systematic-risk channel. Because HFT firms deploy highly correlated trading strategies, they tend to respond to the same market-wide signals simultaneously. This coordinated behaviour rapidly transmits aggregate market information into individual stock prices, effectively raising their betas.
Our results show that this channel has a significant effect on the cost of capital, but only for certain firms (see Figures 2 and 3). For low-beta stocks, the increase in HFT activity leads to significantly higher systematic risk and a higher cost of capital. In contrast, for high-beta stocks, we find no evidence that HFT meaningfully changes their financing costs.
The difference comes down to how much “new” information HFTs are actually bringing to the price. High-beta stocks are already highly sensitive to market conditions. After all, a high beta means that the stock moves in tandem with the market as a whole, so news is usually priced in quickly even without HFT. However, low-beta stocks have historically reacted more slowly to market-wide information. Because HFT algorithms monitor and trade thousands of stocks at once, they force market-wide news into these “slower” stocks almost instantly. This makes low-beta stocks move much more closely with the overall market than they used to, raising the risk premium that investors demand.
The effect within the low-beta group is further concentrated among the least liquid stocks. This is intuitive: Illiquid stocks tend to be less actively followed by analysts and less continuously traded, meaning they incorporate market-wide information more slowly. When HFT arrives, it forces these stocks into the broader information stream, increasing their co-movement with the market by the greatest margin.
A silver lining for the most liquid stocks
The second channel operates in the opposite direction and applies to a different segment of the market: the most liquid stocks. This is the liquidity-premium channel. High-frequency traders often act as modern market makers, but they typically concentrate their liquidity provision in stocks that are easiest to trade—those with high baseline liquidity and the capacity to unwind inventory rapidly. For these specific stocks, HFT reduces the liquidity premium that investors typically demand to hold a security.
Our data support this second effect. For the most liquid stocks, defined as the 30 percent of stocks with the lowest relative bid-ask spreads, our analysis consistently finds that HFT has a negative and statistically significant impact on the cost of capital. In other words, HFT reduces the cost of capital for these firms because it further enhances the liquidity that was already their defining characteristic.
Validating global trends
One potential concern with the US results is that high-frequency trading in American markets is deeply intertwined with market fragmentation. This raises a critical question: Are these effects on the cost of capital a result of speed itself or simply an artifact of the many different trading venues that HFT firms arbitrage?
To address this issue, we replicate our analysis using data from the Stock Exchange of Hong Kong (HKEX). It is one of the world’s largest stock exchanges, and the stocks traded on it are not traded anywhere else, so fragmentation is not an issue. We use the 2014 introduction of the Orion Central Gateway (OCG), a major technological upgrade that substantially reduced trading latency, as our exogenous shock. Firms listed on the Shanghai Stock Exchange (SSE) serve as the control group.
The Hong Kong results mirror those from the United States. We find that HFT raises the cost of capital on average, with the effect concentrated in low-beta stocks and amplified for those that are relatively less liquid. Meanwhile, the most liquid stocks in Hong Kong also experience a cost-of-capital reduction. This consistency across two very different market structures, one highly fragmented and one not, significantly strengthens the causal interpretation of our findings.
Implications for regulation and policy
The policy implications of our findings are significant and nuanced. We do not find support for a simple narrative that HFT is either unambiguously good or bad for the real economy. Instead, the results demonstrate that the effects are fundamentally heterogeneous and depend on the specific characteristics of the stocks in question.
For the most liquid, highly traded companies, HFT appears to be a net positive, reducing the cost of raising equity capital and thereby potentially facilitating investment. These are typically large-cap, blue-chip firms that already attract abundant analyst coverage and investor attention. However, for other stocks—often smaller, less-liquid companies with business models less naturally correlated with the market—HFT appears to impose a genuine cost.
These findings can help policymakers better understand the diverse impacts of HFT and consider how to balance market efficiency with cost-of-capital dynamics.
Disclaimer
The views expressed in this article are those of the authors and do not necessarily reflect the views of the BIS or its member central banks.
Matteo Aquilina leads the Bank for International Settlements’ (BIS’s) contributions and deliverables to the G20 and the Financial Stability Board (FSB). He joined the BIS in late 2022, initially working in the Financial Markets Unit. Prior to this, Matteo spent more than two years at the FSB Secretariat, focusing on non-bank financial intermediation. He previously held a variety of roles within the Chief Economist Department at the United Kingdom’s Financial Conduct Authority (FCA), where he worked for a decade. During his time at the FCA, he also served as Vice-Chair of the International Organization of Securities Commissions (IOSCO) Committee on Emerging Risks.
Khaladdin Rzayev is a Senior Lecturer in Finance whose research focuses on market microstructure and financial econometrics, including the application of machine learning techniques. His work has a particular emphasis on high-frequency trading, liquidity, and price discovery. He has published in leading academic journals, including the Journal of Financial Economics and the Journal of Financial Markets, and is affiliated with the University of Edinburgh, Koç University, and the London School of Economics.