IS THERE AN OPTIMAL CLOSING PRICE MECHANISM?

Concerns regarding the prevalence of manipulation of closing prices have prompted substantial changes to the mechanisms used to close trading at stock exchanges around the world.  Closing prices are important for, among others, benchmarking and contracting purposes. Accordingly, their manipulation undermines two fundamental pillars of market efficiency: liquidity and price discovery. Traditionally, a stock’s closing price has simply been its last-traded or volume-weighted average price. However, amid mounting concerns over price manipulation, stock exchanges are increasingly turning to closing batch mechanisms to calculate closing prices. These mechanisms gather orders over a set period before the close, and then determine a closing price at which the maximum number of orders can be filled. This study identifies two types of closing batch mechanisms: the single-price call auction and the on- close facility. The latter is prominent in North American markets. This study also identifies four design characteristics that feature in the closing batch mechanisms used across global markets. These are: transparency; order flexibility; randomisation of the closing time; and price stabilisation facilities, such as auction extensions. The study utilises a newly-developed direct measure for the probability of closing price manipulation. It finds that, at the overall level, closing batch mechanisms produce more efficient closing prices than the last recorded trade. The provision of greater transparency during the pre-close period reduces the prevalence of closing price manipulation. However, there is a cost in terms of market efficiency, as greater transparency can create a disincentive to trade for those who in doing so may reveal an information advantage. Randomised closing times and extension periods improve market liquidity, the process of price discovery, and price stability. They also correspond with a reduction in closing price manipulation. In contrast, order flexibility produces no benefits to market efficiency or integrity around the close. A single-price call auction consolidates buy and sell orders for a stock over a pre-determined period ahead of the market close. At the time of the auction, regular trading ceases and all open buy and sell orders are ranked according to price. The price at which the maximum volume of open orders can be cleared is then fixed as the closing price. By comparison, the operation of an on-close facility commences immediately after the market opens, storing orders that have been designated for completion at the close of trade. These on-close orders can be entered any time during the trading session until a pre-specified time before the close. In contrast to a single-price call auction, on-close facilities operate in parallel with the continuous market until the close of trade. Consequently, on-close facilities offer greater and immediate visibility of the distribution of orders, and provide uninterrupted access to the market for the entire trading period. The study examines the introduction of closing batch mechanisms and their design on market efficiency across 20 stock exchanges. It finds that the introduction of a closing call auction leads to a reduction in liquidity in the continuous market prior to the close. This reflects a redistribution of trade completions as traders delay their orders to participate in the closing mechanism and ensure they receive the closing price. This change in market behaviour over the last two hours of trade has the effect of widening price spreads, which impacts market efficiency and, in turn, pushes up trading costs at this time. Nevertheless, a closing batch mechanism significantly improves the process of price discovery at the close. Moreover, trading data shows a statistically significant reduction in the prevalence of closing price manipulation, and by implication an improvement in closing price integrity, following the introduction of such a mechanism. For all measures of price efficiency, the data indicates that a closing batch mechanism is superior to the traditional close, where the last completed sale at the moment trade ceases represents the closing price. An on-close facility in particular shows the greatest improvement to liquidity and price volatility at the close. The study finds there are two key components of an effective on-close facility that improves market efficiency and integrity. These are a price-stabilisation mechanism, and a randomised closing time. A temporary extension of the pre-close period is a typical price-stabilisation mechanism that comes into operation when the estimated closing price exceeds stipulated price movement limits. It serves to encourage adjustments to open orders, thereby reducing the imbalance that caused the excessive price swings. A randomisation of market closing time entails trading ceasing during a time interval, rather than at a specific point. This reduces the ability of traders to place misleading orders in the moments before the market closes. The study has strong implications for exchanges and regulators. Although greater transparency reduces the incidence of manipulation, it can detract from market efficiency. However, if transparency is complemented with a randomised closing time, efficiency may be restored as any incentive to delay trading until immediately before the close may be outweighed by the possibility of waiting too long and thereby missing out on a trade. This research uses a unique data set of dark trades collected from the four trading venues offering dark executions in Canada. This is combined with the Thomson Reuters tick history databases to compare dark trading to the lit. Following extant research methods of Hasbrouck (1995) and Gonzalo and Granger (1995) we separate price discovery into two measures: (1).Information Leadership Share, measuring the extent to which a venue produces new price information; and (2) Price Discovery Efficiency, the extent to which a venue avoids chasing temporary price shifts. Echoing past research, we find that dark trading adds little value in terms of price discovery; contributing only 7.1% in terms of the Information Leadership Share. Furthermore, the introduction of price improvement rules in Canada reduces both the level of dark trading and the information content of dark trades. Analysing cross-listed securities our research shows that limiting dark trading in Canada made their price discovery less efficient compared to US trading, where dark trading caps do not exist. Price Discovery Efficiency reduced 19.4% after the introduction of the price improvement regulation. The research further seeks to explore the determinants of price discovery, specifically analysts’ recommendations and order book depth. We find that analyst coverage is no longer a predictor for price discovery, however, order book depth measures are strong predictors of price discovery. While these results demonstrate dark trading offers little in terms of price discovery, there may be a case that it can increase the aggregate price discovery of a markets, especially if there are differences in market regulation. This brings into question whether or not exchanges should allow dark traders to free ride off the price discovery and quotes of the lit market. Australia and Canada have both said no; but will Europe and the USA follow?