How Algorithmic Trading Undermines Efficiency in Capital Markets

Abstract

I. INTRODUCTIONIn 2012, traders in the United States submitted over two billion offers to buy and sell securities on major national exchanges, resulting in around seventy-four million completed trades. As a point of comparison, at the height of the internet boom in 2000, traders sent out only around five million quotes and, from these, concluded three million trades. In other words, in 2012, traders were submitting almost 460 times the number of quotes than they did in 2000. Rather than reaching a bargain 60% of the time, as was the case in 2000, traders did so in just 3% of cases in 2012, sending out more than thirty-one quotes for every completed trade.1 An uptick in trading appetite over the years is almost certainly not an explanation for this staggering rise in the number of orders.2 Rather, these statistics are indicative of a much larger and more fundamental transformation underway in U.S. markets. Reflecting the rise of computer technology, today's marketplace has come to rely heavily on automation and algorithms as an essential part of the trading process. In place of humans submitting orders and routing and processing trades, firms can delegate these tasks to algorithms, computerized instructions that transact in accordance with a firm 's pre-set strategy.3 Untethered from the limitations of human cognition, algorithms enable trades to occur at high speed and high volume using pre-programmed decision rules to identify trading opportunities. With the aid of algorithms, traders can deploy a more powerful array of quantitative techniques, statistics, and financial modeling as part of the buying and selling process than previously possible.4These changes prompt fresh reflection about the ability of markets to continue to perform their most basic function: supplying capital to the real economy. In both law and finance, scholars have debated market function and regulation through the lens of efficiency.5 How well markets are able to reflect information in securities prices often serves as a proxy as a rough-and-ready measure of market health.6 It is easy to understand why; efficient markets help investors cheaply investigate the most optimal destination for their capital.7 When prices reflect information accurately, investors can assess their risks and rewards more precisely and better direct their capital to the most worthwhile investments-showcasing allocative efficiency.8 While predicting the future performance and intrinsic value of investments is near impossible, informational efficiency offers a mechanism, albeit an imprecise one, to make beneficial allocative choices with the information at hand.9 Better informed markets can help capital to reach those companies and areas of the economy that are likely to use it most productively. Put simply, informational efficiency in the market should help bring about allocative efficiency for the economy.10This Article challenges this conventional wisdom. It argues that algorithmic trading creates trade-offs in the relationship between informational and allocative efficiency. Though algorithms help markets make gains on several measures of informational efficiency, they also create costs for their ability to allocate capital productively. With the considerable reliance that law and regulatory policy place on prices in matters of allocative importance, this decoupling is significant for bedrock assumptions underlying corporate and securities regulation.At first glance, the shift to automation holds much promise for efficient markets-and, by extension, for capital allocation. Using algorithms, traders can harness data, deploy complex analyses, and submit orders at will to strategically execute their desired strategy.11 Rather than searching extensively for the ideal trade, or waiting for the best time to send out orders, traders can rely on algorithms to do this hard work for them.12 Transaction costs can diminish and so too the cost-benefit threshold at which traders might enter markets with their private reserves of information and insight. …

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 53 citations exist upstream.