Øyvind G. Schanke, CIO and Simon Emrich, Lead Analyst at Norges Bank Investment Management examine the rise of dark pools and highlight issues of concern to large longterm global investors.
Off-exchange trading has evolved and increased in importance in recent years. This is partially driven by the increased ‘institutionalisation’ of asset management. At the same time, the advent of computer-based trading and the emergence of new forms of liquidity providers such as high-frequency traders in an increasingly fragmented market has changed the nature of equity trading1.
‘Dark Pools’ have been in the news recently, and are often seen as one of the potential problem areas of modern market microstructure – no doubt due in part to their somewhat unfortunate name. As a large participant in asset markets globally, NBIM has a more differentiated view. ‘Dark pools’ cover a wide range of trading venues that are utilized at different stages of the investment process, either directly by the investor, or by a broker employed as an agent. The impact of dark pools on market quality needs to be analysed based on these different uses.
Do dark pools contribute to well-functioning markets?
For present purposes, we define this as supporting a market structure that maximizes natural liquidity (long-term, natural buyers and sellers can find each other with high probability) while minimizing cost (rent extraction by intermediaries such as high-frequency traders, exchanges and broker/dealers should not be excessive). Several dark pool characteristics can help in achieving these objectives:
• they can efficiently facilitate direct block trading between institutional investors,
• they can serve as competitive checks on exchange monopoly power, and
• they can be tailored to specific market participant requirements, and innovate rapidly.
These benefits have to be weighed against the potential efficiency drag that dark pools can introduce to the price discovery process. Dark pools operate without pre-trade transparency. This means that the residual volume that is traded on ‘lit’ exchanges becomes more informative. This can increase instantaneous volatility and the cost of price discovery.
Dark trading venues can be classified in a number of different ways – Butler (2007) segments 24 US dark pools into 16 different types, for example2. A classification by the stage of the investment process in which a venue is used is particularly useful. Some dark pools focus on direct block crossing, and typically appear early in an investor’s execution plan. These pools show large average trade size and low fill rates. Other dark pools – most of those operated by broker/dealers, as well as independents and HFT ping destinations – have smaller trade sizes (comparable to those in lit exchanges) and higher fill rates. These pools typically appear later in an investor’s execution plan, after the investor has delegated execution to a broker.
Block crossing venues are the modern-day equivalent of ‘upstairs trading’, which is probably as old as exchanges. Their potential utility has increased as global asset markets have become increasingly ‘institutionalized’3. Institutional liquidity demand is likely to be ultimately satisfied by the contra liquidity demand of another institutional manager. Block crossing venues serve to efficiently facilitate such trades. Ready (2013) shows that these venues account for between 5 and 8% of US large cap institutional flow4. One area of interest for us is whether this percentage can be increased.
For the remaining institutional flow, a broker is typically tasked with sourcing liquidity, with the understanding that this will lead to market impact. Typically, this involves a trading algorithm, which will break up a ‘parent order’ into small, sequential ‘child orders’. At this point, the number of venues that may be accessed increases significantly. In addition to traditional exchanges, these include the second type of dark pools described above. Trade sizes across these trading venues are typically comparable. However, the venues differ in probability of fill (from near-certainty for market orders sent to exchanges, to very low for most HFT ping destinations), toxicity (information leakage) and cost.
Subject to a broker’s best execution obligations, their algorithms’ venue routing decision will be driven by economics. This includes explicit access costs for the broker, as well as competitive considerations which can lead to excessive fragmentation. This increased competition is effective in limiting rent extraction by exchanges and liquidity providers. However, execution quality considerations mean the investor has to direct the broker on permissible venues and trading strategies. Execution benchmarks are an important tool, but in our experience more explicit venue selection criteria are also necessary. For example, we do not believe that the liquidity from HFT ping destinations is worth the information leakage costs. For other venues, as well as exchange-operated hidden books, the picture is more nuanced and requires constant monitoring of execution quality.
We believe that both block crossing venues and other dark pools can play an important role in ensuring well-functioning markets. In particular, they are useful in limiting rent extraction by intermediaries. Recent letters from US exchanges to brokers and investors indicate the effectiveness of this; however, we believe that there has to be a broader discussion on exchange costs – especially around data fees.
At the same time, transparency around the operating procedures of many dark trading venues is needed. Publication of Reg ATS forms by many of our US brokers is a useful first step; we are in favour of further transparency about operating procedures and available order types, particularly if they differ by client. If we do not feel we have sufficient transparency in a given trading venue, it will not be on our whitelist of permitted trading venues for our brokers.
We are also actively working on establishing and strengthening direct block crossing venues. We believe these trading venues should have greater prominence in today’s equity markets, and are a better reflection of the reality of ever-increasing institutionalisation of asset markets.
Dark pools are a valuable complement to exchanges, and are an essential part of today’s market structure. However, they can introduce avenues for novel forms of rent extraction through insufficient transparency. This is also not a new phenomenon – as always, it requires vigilance and a proactive approach by asset owners and managers.
Trading in Dark Pools – An Asset Manager’s Perspective
