Print
E-MailMerritt B. Fox is the Arthur Levitt Professor of Law at Columbia Law School; Lawrence R. Glosten is the S. Sloan Colt Professor of Banking and International Finance at Columbia Business School; and Sue S. Guan is Assistant Professor of Law at Santa Clara University School of Law. This post is based on their recent paper.
Drawing on microstructure and financial economics, our new paper, Spoofing and Its Regulation, seeks to clarify the considerable confusion among commentators and the courts with regard to a common kind of quote-driven manipulation, often referred to as “spoofing.” We conclude normatively that spoofing is a socially undesirable activity and provide a sounder rationale for finding it illegal under the Securities Exchange Act of 1934 than has been offered so far.
The Nature of Spoofing
Quote manipulation refers to the practice where a trader uses quotes—limit orders constituting binding commitments posted on an exchange, until canceled, to buy or sell a given number of shares at a stated price—in order to buy or sell shares at a more favorable price in a separate transaction going in the opposite direction. Once the separate transaction has been executed at the favorably changed price, the trader cancels these quotes, which she can usually do because they have yet to be executed against.
With spoofing, the trader’s manipulative quoting activity is at prices equal or inferior to the best quotes in the market for transactions going in the opposite direction of the transaction she really seeks to have execute. Suppose, for example, the manipulator wishes to buy a certain number of shares. Her spoof involves three steps. First, she submits a buy limit order (a bid quote) at a price equal to the NBB (national best bid). Second, she submits one or more sell limit orders (offer quotes), at prices equal to, or above, the NBO (national best offer), aggregating to a much larger number of shares than her bid. These sell limit orders are the manipulative quotes. They are intended to induce quoting and transacting behavior of other market participants that results in the manipulator’s buy limit order being executed against, something that otherwise would have been much less likely to occur. Third, the manipulator cancels her sell limit orders. Where the actually desired transaction is instead a sale, the manipulator does a mirror image of these steps, first submitting an offer at the NBO and then bids for a large number of shares at prices equal to or below the NBB.
Typically, the spoofer in fact undertakes a pair of spoofs going in opposite directions, thereby buying at the NBB and selling at the NBO. Since, at any one point in time, the NBO is always above the NBB, the spoofer on average makes a profit in an amount per share equaling the spread between the two. The profit on any one such pair of spoofs may be small, but the trick may be performed many times in a short period, such that some spoofers make tens of millions of dollars a year.
Spoofing is based on the general observation, confirmed by empirical studies, that the arrival of an offer for a large number of shares at a price equal to, or higher than, the pre-existing NBO, is followed by market participants acting in the same manner as if bad news had arrived about the issuer. Similarly, the arrival of a bid for a large number of shares at a price equal to, or lower than, the NBB, is followed by market participants acting as if good news arrived about the issuer. The spoof works because these participants believe that the spoofer’s opposite-side large quotes, which are anonymous, also carry such news when in fact they do not.
Normative Analysis: Spoofing is Socially Undesirable
Each of spoofing’s three steps—submitting the large quotes, the actual purchase or sale of shares, and the cancelling of the large quotes—is, by itself, a perfectly acceptable form of behavior that is at the core of any efficiently operating secondary market for securities. What commentators and courts have found problematic is the three steps being undertaken together, combined with the intent to have the opposite-side large quotes favorably influence the price at which the actual transaction occurs. But what is the social harm, if any, when the manipulator succeeds? We evaluate precisely who is hurt and who is helped if spoofing is left unregulated, and how this would change if spoofing were instead banned by law, comparing these two worlds in terms of economic efficiency and the fairness of the resulting wealth positions of the various market participants.
We find that, contrary to what many commentators and courts say, spoofing does not raise serious fairness issues. Nevertheless, the existence of the practice is undesirable because of a number of deleterious effects on economic efficiency. As discussed in more detail below, it reduces both liquidity and price accuracy. It also wastes productive resources on an activity generating no social gain. And, by lessening confidence in the market, it discourages some persons from participating who would benefit if they did.
- Ex post analysis. Consider first, ex post, who is helped and who is hurt by a pair of spoofs, the first involving a purchase at the NBB and the second a sale at the NBO. The hurt persons are professional liquidity suppliers, a service typically provided by high frequency traders (HFTs). The spoofer’s large offer quotes at or above the NBO induce the liquidity suppliers to cancel all their own bids at the NBB and to submit sell orders that execute against the spoofer’s bid, thus assuring that the spoofer’s desired purchase at the NBB occurs. The liquidity suppliers’ moves reflect their belief that the share price will fall and, consequently, their desire to clear out the remaining bids at the NBB so that they can submit new offers at that price or lower, something that would otherwise be prohibited. The spoofer then enters into a mirror-image set of actions. In the roundtrip, the spoofer collects the spread between the NBB and NBO times the number of shares bought and sold, and the HFTs, as the counterparties, lose this amount.
- Ex ante analysis. If the practice is allowed, market participants will be aware that spoofing will occur from time to time and adjust their behaviors accordingly. In particular, liquidity suppliers will widen the spread between their bids and offers.
One reason for the wider spread arises from the fact that large orders at or away from the best quote predict price changes, meaning that the presence of such orders is a signal that informed trading is going on. If spoofing is occurring from time to time, this signal gets muddied because, while a large quote could still reflect informed trading, it also could instead reflect a spoof. This muddying is a problem for liquidity suppliers. When a liquidity supplier in an anonymous market transacts with, unbeknownst to it, an informed trader, it on average experiences a loss. This adverse selection concern is the primary driver of the spread between a liquidity supplier’s bid and offer: making this spread when dealing with the uninformed traders covers a liquidity supplier’s losses when dealing with informed traders. The worse liquidity suppliers are at detecting the incidence of informed trading, the less able they are to change their quotes in ways that lessen these adverse selection losses. Anticipating more in the way of adverse selection losses, liquidity suppliers widen their spreads to cover these higher expected costs.
A second, though less certain, reason for wider spreads arises from the losses liquidity suppliers experience dealing with the spoofers themselves. These losses are not due to adverse selection. Rather, liquidity suppliers engage in these transactions as an opportunity to expand their businesses of quoting: doing so allows them to quote during a period of time when they otherwise would not have been able to. So the costs to a liquidity supplier of choosing to engage in this kind of activity is more like an overall cost for the ability to do expanded business, just like a larger computer or more staff. This cost may in the long run also make its way into the liquidity supplier’s spread, widening it further, since in the long run the liquidity supplier needs to cover costs to stay in business. Alternatively, however, it may just cut into the positive rents, if any, earned by the liquidity suppliers or by providers of some of their inputs.
- The deleterious effect of wider spreads on efficiency. Wider spreads, by making trading more expensive, reduce the liquidity of the shares involved, which in turn has multiple negative effects on economic efficiency. Less liquid shares raise a firm’s cost of capital. Less liquidity also impedes the ability of investors to construct their optimal portfolios. It reduces as well share price accuracy by making more costly, and hence discouraging, the activities of the fundamental value traders whose trades incorporate new information into prices. Less accurate share prices in turn undermine the effectiveness of the market’s mechanism for disciplining corporate management and the efficiency with which capital is allocated.
Legal Analysis
Some courts have found spoofing to be a form of manipulation in violation of both Exchange Act section 9(a)(2) and, pursuant to Rule 10b-5, section 10(b). Other case law has suggested the opposite. Often, the reasoning in the opinions going both directions has been confused and, where liability has been imposed, rather strained.
- Section 9(a)(2). Section 9(a)(2) outlaws “a series of transactions” that in effect have as their purpose manipulation. Though some courts have ruled otherwise, it is in our view too great a stretch to properly conclude that quoting and then cancelling before the quote is executed against involves engaging in a “transaction.”
- Rule 10b-5. Rule 10b-5(b) makes it unlawful for any person “to make any untrue statement of a material fact or to omit to state a material fact necessary in order to make the statements made, in light of the circumstances under which they were made, not misleading . . . in connection with the purchase or sale of a security.” We believe that Rule 10b-5 provides a more solid doctrinal basis for outlawing spoofing. Those courts finding spoofing to be a such a violation, however, have done little more than assert this conclusion.
We think good rationale can in fact be provided in support of this conclusion. When a spoofer tells her broker to submit the large opposite-side quote, she is in essence making the following statement: “I am prepared, unless and until I cancel, to be legally bound to buy or sell x amount of securities at y price.” This statement is communicated to the market by the posting of the quote on an exchange. The statement is literally true, but others viewing it would reasonably assume something more: that the submitter of the quote in fact wants someone to execute against this bid or offer. Based on this assumption, other market participants erroneously believe that the quote contains information about the future prospects of the issuer. As described above, the behavior induced by this belief allows the spoofer to profit. The spoofer’s failure to announce that she in fact does not want anyone to execute against her quote can thus be argued to be a materially misleading omission in violation of Rule 10b-5.
If, as we argue, the submission of quotes solely to move price violates Rule 10b-5’s prohibition against making misleading statements, how, in the absence of documentary evidence, can it be determined that this was the submitter’s sole purpose for doing so? Suppose a trader establishes a repeated pattern of submitting a large quote on one side of the market accompanied almost simultaneously by a much smaller quote that executes on the other side, followed by the cancellation the large quote. We believe that this would constitute sufficient circumstantial evidence to conclude that the sole intent of the cancelled large quotes was to get a more advantageous price for the transactions going the other way. This is even clearer when this pattern includes in most instances an immediately subsequent mirror image set of quotes such that the trader is reliably able to buy at the bid and sell at the offer.
The full paper is available here.
