Should We Be Afraid of the Dark? Dark Trading and Market Quality

Abstract

We exploit a unique natural experiment-recent restrictions of dark trading in Canada-and proprietary trade-level data to analyze the effects of dark trading. Disaggregating two types of dark trading, we find that dark limit order markets are beneficial to market quality, reducing quoted, effective and realized spreads and increasing informational efficiency. In contrast, dark midpoint crossing systems do not benefit market quality. Our results support recent theory that dark limit order markets encourage aggressive competition in liquidity provision. We discuss implications for the regulation of dark trading and tick sizes.

Increasing the Tick: Examining the Impact of the Tick Size Change on Maker‐Taker and Taker‐Maker Market Models

“…Our third empirical prediction pertains to the role of the tick size and measures of hidden liquidity. Recently, several studies (see Kwan, Masulis and McInish, ; Comerton‐Forde, Malinova and Park, ; Foley and Putniņš, ; Comerton‐Forde, Gregoire and Zhong, ) investigate the role of tick size and trading activity that takes place in dark pools (i.e., off‐exchange). Trading in dark venues allows market participants to bypass the lit market tick constraint and obtain a finer pricing grid .…”

Connectivity costs and price efficiency: An event study

“…As per Lo and Mackinlay (1988) comparing the variance estimates across different sampling periods (adjusted to a common unit of time) tests if a price series follows a random walk and if a market is efficient. The variance ratio is computed as in Foley and Putniņš (2016): where and are variances of k second, and jk second midquote returns for a given day, respectively. If the price series follows a random walk, the value of the variance ratio metric should be equal to 0.…”

“…Finally, as in Foley and Putniņš (2016), we combine the information contained in autocorrelation measurements with , and variance ratios with combinations ( j , jk) of (1 s, 10 s), (10 s, 60 s), and (1 min, 5 min) into a single metric, autocorrelation factor and variance ratio factor, by computing their first principal component. Coefficient estimates of regressions with autocorrelation and variance ratio factors as dependent variables are reported in Table 5 and provide further support for our main results.…”

Intermarket Competition: Evidence from Trading Venue Short Sales

“…For studies of variants of trade-at rule implementations in Canada and Australia, see Comerton-Forde, Malinova, and Park (2015) andFoley and Putnins (2016).22 For more information regarding these categories, see the appendix.…”