اثر مزایده فراخوان پایان بر کیفیت بازار و استراژی های تجارت

We study the effects of the introduction of a closing auction (CA) on the microstructure on the continuous trading phase in Borsa Italiana and Paris Bourse. We postulate and compare several empirical predictions based on both standard Kyle-type models and more recent models of limit order book. We find that while the CA has no effect during most of the day, its effect on the last minutes of trading is dramatic. We document a sharp decline in volume, associated with a significant reduction in spread and volatility, and an increase in aggressiveness of liquidity suppliers during the last minutes. We show that the differences in the Reference Price algorithm between Milan and Paris have a significant effect: the CA attracts greater volumes when the Reference Price is equated to the CA price.

In December 2001 Borsa Italiana, an order-driven market, introduced a Closing Call Auction at the end of the continuous auction market. The objective of this innovation in market design was to increase the quality of the market at the end of the trading day. The Italian experiment was not isolated: practically all the main European exchanges introduced a Closing Call Auction (e.g. Deutsche Börse, Paris Bourse, and the London Stock Exchange) and, more recently, the NASDAQ market introduced a closing-batch mechanism at the end of the trading day. In the call auction, consolidation of order flows may reduce the price impact of a trade. Furthermore, the enhancement of information revelation could improve the price discovery process and, by reducing intraday volatility, result in increased price stability. Finally, comparing the two systems, one should consider the relative benefits that some traders could obtain by moving from the discriminatory pricing rule of the continuous auction to the uniform pricing rule, which governs the batch system. The success of this change in market architecture depends on how effectively traders develop new strategies to deal with this new trading opportunity. Using data from both the Italian and the French stock markets, we show that the effects of the introduction of the call auction are concentrated in the very last minutes of the continuous phase. We observe a significant reduction in quoted spread, volatility, trading volume and average trade size. We also find strong support for the empirical implications of the existing theoretical models that focus on the impatience of traders prior to close (Kaniel and Liu (2006), Foucault (1999) and Foucault et al. (2005)), but not much support for the predictions of Admati and Pfleiderer (1988). This does not invalidate the latter model, but suggests that impatience and its effects in this case are more powerful than the information asymmetry ones. The introduction of the closing auction makes liquidity demanders less impatient and induces liquidity suppliers to offer liquidity at narrower bid-ask prices. The reduction in market orders and in the bid-ask bounce reduces volatility; this makes limit order submitters even more willing to supply liquidity. It follows that in terms of the agents’ order-submission strategies, the proportion of Non-aggressive orders decreases, and the share of the orders submitted at or inside the BBO increases. Using a data set on the agents’ order submissions, we test the empirical implications on order aggressiveness. The results obtained for the different groups of stocks of the Borsa Italiana strongly confirm these predictions. We also find evidence that the volume at the CA is much more sensitive to the intraday price movements in Milan than in Paris. We attribute this fact to the bias introduced by the BIt computation of the Reference Price. This suggests that by equating the Reference Price to the closing auction Price, as is done on Paris Bourse, BIt may increase the volume of trading at the closing auction, and make the closing price more efficient.