Page 181 - 34-2

NTU Management Review Vol. 34 No. 2 Aug. 2024

Order Choices, Order Execution Quality, and Trading Volume:
Evidence from Reductions in the Call Auction Interval

Yi-Heng Tseng, College of Management, Yuan Ze University

1. Research Purpose/Objective

We focus on the policy impact of the reductions of auction interval from 25 seconds

to 20 seconds (hereafter denoted as Reform I) since July 1, 2013, and from 10 seconds to
5 seconds (hereafter denoted as Reform II) since December 29, 2014, on the Taiwan Stock
Exchange (TWSE). Using the intraday data obtained from the TWSE, the present study
empirically examines whether more frequent auctions alter the order choices, including
order arrivals, order cancellations, order sizes, and degree of order aggressiveness;

empirically tests whether higher auction frequencies are detrimental to the order execution
quality of individual investors (symbolizing the uninformed traders); and empirically
checks whether shorter auction intervals raise the trade value of individual stocks. In fact,

discussions on these issues are of great importance to regulators, practitioners and market
participants, thus motivating our research interests.

2. Research Design/Methodology

This study adopts 120 large cap stocks and 120 small cap stocks on the TWSE as the
research sample, with the data period covering 15 days before and 15 days after each of
the two reform dates. The large cap stocks consist of component securities of the FTSE
TWSE Taiwan 50 Index and Mid-Cap 100 Index, and the small cap stocks are picked up

according to both conditions of the market capitalization and stock turnover following
Duong, Kalev, and Krishnamurti (2009). For each trading day, we employ the sample
observation during periodical call auctions (from 9:00 to 13:25). In total, we investigate
127,825,423 observations of the intraday order book records.

Our research framework is arranged as follows. To investigate the policy impact of
reductions of the auction interval, we compute 14 dependent variables (without loss of
generality, denoted as y hereafter) for regression estimations, with intraday time-series
t

173