Clearing out my desk the other day, I came across a previously overlooked paper from the Financial Services Authority entitled "Guidance on the practice of Payment for Order Flow". The term was all but unknown to me until the US market maker, Knight Capital, got into trouble earlier this month. In the coverage I picked up that Knight is a very major exponent of Payment for Order Flow. Reflecting that I should be more familiar with the term, I glanced through the FSA paper. You will find it very easily on the FSA website. It's only nine pages long but it's a highly entertaining exercise in Zambooblement.
Right from the beginning it sets down in sermon-like tones one argument after another against Payment for Order Flow. As you reach the concluding section there's only one way it can go, which is against. Instead, it delivers a scorching handbrake turn: "This is not to say that [payments for order flow] are not permissible". Double negatives always denote contorted thinking. You can almost feel the (anonymous) author's pain? Who or what got to him?
Payment for order flow is what happens when a market maker pays a broker for an order to buy or sell a parcel of shares or any other kind of security, such as an option. These payments typically occur in securities where there is insufficient liquidity for orders to be processed instantly on screen.
The idea of a market maker paying for an order is a strange one. In the securities market one might expect business to be won or lost on the basis of the market maker's quotation. Isn't that integral to "best execution"? If the market maker is paying the broker a kickback - oops sorry I meant paying for order flow - then how can he be offering the best possible price?
Obtaining the best possible price is after all the bounden duty of the broker. After 20 years of rooting out Spanish practices, after all those Conduct of Business rules, after Mifid (the Market in Financial Instruments Directive, since you ask), how is there scope for a market maker to pay a broker for an order?
Well, apparently, there is plenty because it goes on all the time. In the US, Knight Capital seems to have built a huge business and makes no secret of paying vast sums to the brokers with whom it does business. Knight's recent difficulties had nothing to do with payment for order flow, and it will continue to rely on the practice following its rescue. Some of Knight's biggest rivals including UBS and Citadel are also major payers for order flow.
And in the UK, although we are much less conscious of it, payment for order flow seems to be an everyday practice, certainly occurring on a scale that prompted the Financial Service Authority's survey last year, resulting in the paper I found on my desk. In fact it discovered that for two market makers (it did not identify them) the total cost of payment for order flow is comparable to overall trading profit, "so clearly, it is a substantial cost".
As I say, there is a startling contradiction between the paper's report and its conclusions, but there is an even stronger contradiction in the FSA's Summary of Feedback from interested parties, which is available on the FSA website alongside the main paper. According to this, brokers favour the existing arrangements, which reflect "a service they provide to market makers… in return for which it is proper to charge a fee". However, market makers argue that they receive no service from brokers and that the commissions charged are "essentially a subsidy for the broker's client."
Whether it is a proper fee or a subsidy for the client, the FSA was in absolutely no doubt that payment for order flow must be disclosed "in a comprehensive, accurate and understandable way to clients, before the provision of the service… It would not be easy for a client to understand the arrangement unless it was described in terms of additional remuneration that is received by the broker".
Are you clear about that? Does your broker receive payment for order flow? Has he explained it to you in these terms? And is he cutting your dealing commissions as a result?
