In the last few decades, there has been a proliferation of equity trading systems, among which dark pools have rapidly grown in popularity. Dark pools accounted for 13.66% of US equity trading volume in November 2021, as opposed to 7.51% in 2008. In November 2021, dark venues executed 18.84% of pan-European on-venue turnover, and dark MTFs hit a market share high. In a dark pool, investors can buy and sell stocks without publicly displaying their orders.
As opposed to traditional stock exchanges, these dark venues lack pre-trade transparency. Traders in dark pools submit buy and (or) sell orders, and trades are executed using prices derived from the exchanges.
As markets are becoming fragmented and a significant part of liquidity is increasingly unobserved, the long-held view that prices are a valuable source of information is called into question.
Do prices at the exchanges continue to convey meaningful information? Indeed, dark pools have raised regulatory concerns in that they may harm price discovery in financial markets. For instance, in 2014, then-SEC-Chair
Mary Jo White remarked that
"we must continue to examine whether dark pool trading volume is approaching a level that risks seriously undermining the quality of price discovery provided by lit venues." As recent as August 2021, on CNBC's Squawk Box, the SEC-Chair Gary Gensler pointed out that dark pools have been increasingly common during the recent rise in retail investing and asserted that the agency is "looking very closely" at market structure and dark pools of trading hidden from the public until execution to promote transparency. He was responding to allegations from retail traders that hedge funds and other "big actors" may be manipulating markets.
Effect on information efficiency of prices
Although dark trading has existed in stock markets for quite some time now in the form of either over-the-counter
(OTC) decentralized trading or special hidden order types on exchanges, the emergence of dark pools as alternative trading systems operating fully outside transparency requirements has attracted enormous trade volumes. Given the prominence of the issue, there has been significant interest in academic research in recent years to investigate the effect of dark trading on the informational efficiency of prices. Theoretical studies have yielded conflicting results. Empirical papers also differ in their results on the impact of dark pools on price and market quality measures.
While it is essential to study how dark trading influences the ability of asset prices to aggregate diverse information already available in the market, it is of utmost significance to analyze how the amount of new information produced by traders at the outset is affected by the presence of a dark pool. The literature has given significant attention to the former question.
Still, the latter issue of the causal effect of dark pool trading on costly information acquisition remains largely unexplored.
Given that a substantial portion of the liquidity is anticipated to be hidden, does the incentive to acquire costly information about stock fundamentals increase compared to the benchmark of a centralized trading institution with full pre-trade transparency? How does the relative usage of the dark pool vis-Ã -vis the lit exchange depend on the relative strength of private information held by an investor? Are informational efficiency of prices and market quality necessarily degraded when dark trading occurs?
Research methodology: using laboratory experiments
We design an experimental asset market with the endogenous acquisition of costly information.
We assume two equally likely states of nature, A and B, and a single asset, namely, an Arrow-Debreu security that provides a payoff only in state A. Before trading, some individuals can acquire costly, imperfect signals about the state of nature. Signals are binary and are independent and identically distributed (i.i.d.), conditional on the state. In the experiments, we implement markets with either low or high precision of signals. We investigate two market structures, one with a single lit exchange where all order submissions by a trader are observable to other traders, and another where two parallel trading venues exist. In the latter institution, in addition to the lit exchange, traders can submit orders and transact in a dark pool where order submissions of other investors are unobservable.
Transaction prices in the dark pool are derived from the existing buy and sell offer prices in the lit exchange. Participants in our experiments are recruited from the population of undergraduate and graduate students.
Experiments provide a useful complement to theoretical and empirical studies on the consequences of dark pool trading. In the laboratory, one can employ a trading mechanism close to the one used in actual markets while still having the ability to control and change variables to allow clean causal inferences. The novelty of our analysis stems from the fact that we consider fragmented markets in the laboratory where traders can exchange an Arrow-Debreu security either in a lit market with a fully transparent limit order book or in a dark pool where order submissions are hidden.
Effects of addition of a dark pool
The impact of trading in dark pools on information acquisition, informational efficiency of prices, and market outcomes depends crucially on the informativeness of the signals acquired by investors.
There is a strong negative crowding out effect on liquidity at the lit exchange irrespective of the information precision of signals.
With market fragmentation, traders now use both lit and dark venues for their transactions. The aggregate trade volume remains the same with and without a dark pool and the additional venue now substitutes transactions from the lit exchange. There are two positive effects of the addition of a dark pool: information acquisition effect whereby investors acquire a greater number of signals, and a sorting effect wherein traders with stronger signals use the lit exchange relatively more than either traders with moderate signals or uninformed ones.
Costly information acquisition
It is observed that dark trading encourages costly information acquisition among investors, significantly so in the high information precision environment. There are two reasons for this observation.
First, investors would anticipate that a sizeable number of orders will now be traded off-exchange given that markets are fragmented, resulting in less information being impounded into prices of transactions at the exchange. This would incentivize traders to acquire a larger number of signals. This is because information is valuable irrespective of whether precision is low or high. Informed traders outperform the uninformed ones given that prices are not fully revealing in markets with low precision. Even when they fully reveal the available information in several of the markets with high information precision, it takes time to do so which allows informed traders to outperform uninformed ones, and an incentive to acquire information is present in all our markets. Second, it seems likely that some traders would acquire more information in the hope that they will trade in the dark pool and be able to prevent the information leakage that would have resulted from placing orders in the lit exchange. Whether or not these traders are able to use the dark pool for performing their transactions though depends on several endogenous factors, like the execution risk and wait time in the dark pool.
The sorting effect is observed only under the high information precision environment. In this setting, the strongly informed traders have close to perfect information, cluster on the heavy side of the market, and suffer lower execution probabilities in the dark pool. As a result, they are more likely to migrate to the lit exchange for faster execution.
With low precision of signals, the proportion of traders with strong signals in the market is much lower, and information risk is substantial, and hence the sorting effect disappears.
Efficiency and level of dark trading
With high precision of signals, at modest level of dark trading, the combination of information acquisition effect and sorting effect outweighs the crowding out effect. As the dark market participation goes up, the crowding out effect grows stronger and eventually exceeds the positive effects. This results in a non-linear relationship between the informational efficiency of prices and the level of dark trading in the market. In contrast, when information precision is low, none of the positive effects are observed, and only the negative effect is present. Even then, price efficiency does not decline significantly with dark trading and only starts to deteriorate when dark market participation is very high.
Are informed traders better off with the introduction of a dark pool?
Informed traders are able to outperform the unintormed ones in markets with low precision as well as high precision. In general, the extent of outperformance is greater when information precision is low. Traders with strong signals can reap substantial benefits when compared to uninformed ones. However, the moderately informed traders get similar profits to the uninformed traders. At the same time, the availability of an additional dark venue for trading in itself does not affect the earnings of each type of trader (whether strongly informed, moderately informed, or uninformed).
It is further observed that, while informed traders' earnings are not affected significantly by their relative use of the dark pool vis-Ã -vis the lit exchange, uninformed traders who have a higher net dark transaction volume obtain higher profits. This indicates that uninformed traders face a higher adverse selection risk at the lit exchange than at the dark pool.
The addition of a dark pool allows them to use the additional trading venue to mitigate part of the adverse selection risk.
Reference:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4025127
