Internalization is one of the most difficult issues to approach in the U.S. equities markets. Many broker-dealers are unwilling to discuss the topic at all, and conversations with market participants frequently require the presence of lawyers, who carefully vet any responses. Given the nature of internalization, it is not difficult to understand such caution.

Internalization has engendered much debate, and much has been written about the impact of internalization on market quality, price discovery and overall fairness of the market. However, little has been published to quantify to what extent order flow is internalized and how much impact that internalization has on the quality of order execution. In a recent study, Celent sought to address this issue and determine both the prevalence of internalization in the U.S. equities markets as well as the impact internalization has on execution quality, if any.

In order to understand this area, it is necessary first to provide some definitions. Internalization involves a brokerage firm acting as the exclusive counterparty to its own customers' orders. The broker also becomes a dealer and trades against its customers' orders for its own account. The broker-dealer stands to make a profit as a result of such trading activity. In addition, the broker-dealer is able to avoid transaction fees levied by exchanges or ECNs. Other broker-dealers do not have the ability to compete for these trades, and these orders do not interact with other orders. Internalization is particularly prevalent in the Nasdaq markets, but is increasingly employed with NYSE issues as well.

While these internalized orders are not exposed to the market, broker-dealers are required to be diligent in assuring that their customers' orders are filled using the best execution quality. The general market practice is for broker-dealers to execute internalized orders at the national best bid or offer (i.e., the best price currently available on the primary markets). This practice is referred to as "price matching." While this is the general business practice, there is no hard-and-fast requirement for broker-dealers to match the NBBO, and the term "best execution" has not been defined clearly by U.S. regulators.

The reason that internalization is such a contentious topic is the obvious conflicts of interest inherent in the practice. As a broker, the firm has a fiduciary responsibility to obtain the best possible execution for its clients. However, as a dealer trading against the firm's own clients, there is strong incentive to give clients the worst prices possible while conforming to existing - and rather vague - rules on execution quality.

Complicating the issue further is the fact that execution quality can be measured using a number of different measures - some relatively simple, others far more involved and complex. In order to determine the impact of internalization, Celent analyzed execution quality along a number of measures, including execution speed, frequency of price improvement, average value of price improvement per share, effective spreads and realized spreads. We collected this data for the 16 largest brokerage firms in the U.S. and compared the achieved execution quality with the internalization rate for that firm. The results of this research are shown in Figure 2.

It is clear that there is no statistical correlation between the level of internalization and the quality of order execution. This means that firms that have higher internalization rates do not perform worse or better in terms of execution quality. This result should be reassuring since some market observers view internalization as a nefarious activity, ripe with conflicts of interest. From our analyses, it appears that brokerage firms have managed these conflicts of interest well. It also appears that the execution quality they achieve for their clients is, on average, as good as that of firms that do not internalize as actively. Indeed, the frequency of internalization provides no indication of the brokerage firm's execution quality.

It should be pointed out that these analyses have been performed at the individual firm level and have not taken into account the impact of high internalization rates on the overall market. It is clear that internalization does have an impact on the NBBO itself. Greater internalization translates into wider spreads, particularly for less-liquid issues. Therefore, internalization could be having a detrimental effect on prices on a marketwide basis. However, this question is a topic for future research.

---

Our Analyst

Octavio Marenzi, President and CEO, Celent Communications

Before he founded Celent Communications, Mr. Marenzi cofounded Meridien Research, where he was the research director of the firm's e-financial services practice. Previously, he was the head of IT planning for the asset management division of the Union Bank of Switzerland in Zurich.