September 9, 2013
FINRA's proposal to permit "inadvertent" wash sales (Release No. 34-70276 File No. SR-FINRA-2013-036) is not reasonably supported by FINRA's rule filing nor is it appropriate as a matter of public policy. The SEC should therefore disapprove the proposed rule change.
The rule filing by FINRA states that wash sales resulting from "inadvertent" algorithmic transactions originated by different programs operated by the same broker-dealer firm or by an affiliate (hereinafter, collectively, a "Firm") "can account for a material percentage (e.g., over 5%) of the consolidated trading volume in a security on a particular day, which can distort the market information that is publicly available for that security." FINRA adds as to such wash sales, "Furthermore, in these instances it appears that firms will continue to allow this type of trading to occur rather than incur the costs necessary to prevent it, even though the trading activity is resulting in instances where significant misinformation may be disseminated to the marketplace."
Rather than requiring that Firms which engage in and therefore profit from algorithmic trading, without providing any obvious benefits to the financial markets, to pay for the costs of complying with the federal securities law requirements pertaining to wash sales, FINRA seems to look the other way. FINRA, perplexingly, proposes to address this by amending its Rule 5210 to state, among other things, that "Transactions that originate from unrelated algorithms or separate and distinct trading strategies within the same firm would generally be considered bona fide transactions and would not be considered wash sales, even if the transactions did not result in a change of beneficial ownership, unless the transactions were undertaken for manipulative or other fraudulent purposes." The SEC had said immediately before this rule filing that "algorithmic trading resulting in executions with no change in beneficial ownership, even if unintended, raises concerns." Release No. 34-69751 File No. SR-NYSE-2013-29 (June 13, 3013) at 6 , n. 11. Does FINRA mean to suggest that its filing, blessing the continued operation of the same-Firm wash sale generating algorithms so long as they are not improperly motivated, adequately addresses those concerns? The filing is silent as to how the market volume distortions that attend operation of these program are to be prevented – and the new rule would do nothing in that regard. This from the nation's primary regulator of broker-dealer firms, at a time when mistrust of our securities markets is at or near an all-time high, is unnerving.
FINRA explains that, while industry advisory committees consulted by its staff about the proposed rule "recognized the problem FINRA was seeking to address, they indicated the need for FINRA to recognize that not all wash sales can be prevented," referring, presumably, to wash sales resulting from the operation of different trading algorithms from or within the same Firm. FINRA appears to agree, but does not explain why it agrees. This key underpinning of the rule proposal is not supported by any showing in the filing. Why FINRA agrees ought to be explained and supported in the filing so that commenters and the SEC can assess and address whether it is so. FINRA also ought to explain how it is able now, and will in the future be able, to surveil for compliance with the new rule, dependent as it is on the absence of any manipulative or other improper purpose when such wash sales are effected. This task is also not undertaken in the rule filing. Last, and most astoundingly, the rule proposal's implicit premise is that it is acceptable under the Exchange Act for the Firms to continue to artificially inflate reported volume in securities traded by their algorithms. Nothing is offered to explain this either.
For the foregoing reasons, the foundation for FINRAs proposal seems less than sound.
It is possible, one imagines, that the reason why not all such algorithmic wash sales can be prevented is that the Firms trading programs already trade as rapidly as human design permits (to carry out the strategy for which the algorithms were designed), leaving no time for cancellation or redirection of a proposed trade that will take place against opposite-side trading interest from someone other than what might be called a "companion" algorithm operated by the same Firm. That is, perhaps it can be shown to the SECs satisfaction that it is beyond the ability of the Firms to design or redesign their algorithms (say, by having them signal each other in advance as to what the others are simultaneously doing in placing orders or bids or offers) to avoid inter-program wash sales so that each algorithm (i) recognizes when a purchase or sale order is about to trade against another bid or offer or contra-side sale or purchase generated by a "companion" algorithm operated by the same Firm and (ii) either cancels the order or redirects it to another execution opportunity not presented by the "companion" algorithm. Whatever the reason, FINRA should show what it is and the basis for it in its filing.
If it is not impossible to prevent these inadvertent algorithmic wash sales effected between different trading programs operated by the same Firm, but merely expensive to correct, it seems obvious that the Firms – at least those that engage in generating a material number of these wash sales – should be forced to shoulder that expense. They should do that by augmenting their algorithms and supporting systems to prevent the occurrence of such same-Firm wash sales altogether rather than continue to mislead other market participants as to the accuracy of the market information as to trade volume being given to them. (Is accurate information as to volume and changes in trading volume important to investors and proper operation of the securities markets? That goes without saying. See, e.g., Section 11A(c)(1)(A) and (B) of the Exchange Act and analysis such as that presented in Dynamic Volume-Return Relation of Individual Stocks, Llorente, Michealy, Saar, and Wang, The Review of Financial Studies, Fall 2002, vol.15, no. 4, pp. 1005-1047 (http://web.mit.edu/wangj/www/pap/LlorenteMichaelySaarWang02.pdf ).) It also must be borne in mind that these inadvertent wash sales are not simple mistakes that occur from time to time, but instead occur regularly as a predictable, direct result of conduct in which the Firms have chosen to engage.
If it is impossible to prevent the occurrence of these wash sales, and their occurrence is an unavoidable adjunct to the operation of multiple algorithmic trading programs within the same Firm, the question remains why FINRA should be permitted, through SEC approval of this rule proposal, to allow the appearance of the markets for the securities traded by these algorithms to be so significantly distorted as they are by the repeated occurrence of these wash sales. Why should the investing public continue to be misinformed on a regular basis as to what reported volume in this or that security is in fact instead of as it appears to be when inflated by these non-trade trades? This is not a case where the SEC has determined that the type of trading at issue, algorithmic trading by means of different programs operated by the same Firm that incidentally generate a high percentage of wash sales, is and should be entitled to special consideration under the Exchange Act. Although this has sometimes has been the case with other trade types – e.g., arbitrage, which can be said to serve the over-arching Exchange Act purpose of price discovery – FINRA has not suggested any reason why the SEC should reach such a conclusion in this case. Needless to say, it would not be enough to point out that, if such same-Firm multiple algorithmic trading were to be stopped, volume in general would decline.
What FINRA proposes will not cleanse the markets of the false appearance of volume inherent in permitting these algorithmic wash sale transaction to continue. Quite the contrary. Even assuming that FINRA has some means of disentangling these wash sales from ordinary, bona fide transactions and testing whether they are not improperly motivated (and it is not at all clear how it proposes to do this – another topic not addressed in the rule filing, as noted above – without which the assurances that such wash sales, even if not legitimate, do little harm, give no real comfort), why is this a tolerable result?
FINRA does not tell us in its filing just how high a percentage of reported volume in a given security these wash sales sometimes constitute – only that the percentage can exceed 5%. What has FINRA done to investigate those that do exceed this already high percentage and how has FINRA satisfied itself in those cases that the wash sales were not improperly motivated and otherwise had no deleterious effect on the appearance of the market for or investors in the affected securities? This is especially troublesome when it is all too easy to imagine "intelligent" trading algorithms that learn from exogenous events or empirical data and can be programmed to respond by changing their trading strategies to serve some selfish purpose of the Firms that operate them under predictable circumstances. How will FINRA or the SEC be able to ascertain on a regular basis that this is not so in the case of these "accidental" wash sales?
There is no basis under the federal securities laws why FINRA must formulate its solution to the problem of same-Firm inadvertent algorithmic wash sales in a way that distinguishes between intentional and inadvertent wash sales (i.e., those undertaken with manipulative intent and those that are not). Instead, the courts for years have recognized as valid use by the SEC and FINRA of the rulemaking power conferred by the Exchange Act to define certain market conduct as "manipulative" or "deceptive" per se, forbidding more conduct than actually can be shown in each case to be manipulative or deceptive or to have been intended to create or had the effect of creating a "false or misleading appearance with respect to the market." FINRAs rule proposal would better serve the purposes of the Exchange Act if it were revised to do that in this case – forbidding the operation within the same Firm of multiple trading algorithms that produce wash sales (unless there is some identifiable low amount of such wash sales can be shown to have an immaterial impact on reported trade volume, which seems unlikely since otherwise the problem FINRA is attempting to address would not have attracted its attention). That algorithmic trading provides no demonstrated or generally agreed upon advantage to the operation of the market or the investing public only serves to strengthen this view.
What must be balanced here is, on the one hand, the cost to certain large Firms of either cleansing the market of the inappropriate, market distorting activity that those Firms generate by means of operating multiple, different algorithmic trading programs within the same Firm, improperly inflating volume and altering the appearance of the market artificially, or ceasing to operate the programs that produce such results and, on the other, the deception of other market participants (that is, the investing public) who now wrongly rely on the accuracy of trading volume information and market appearance being conveyed to them. The result contemplated by the proposed rule, simply stated, is out of keeping with what the anti-manipulative and market information provisions of the Exchange Act were designed to achieve.
Unless corrected, therefore, FINRA's rule proposal should be rejected as inconsistent with the Exchange Act and the public interest.