Introduction Directory UMM :Data Elmu:jurnal:J-a:Journal of Empirical Finance (New):Vol7.Issue5.2000:

1. Introduction

The rapid development of electronic communication has led to a profound restructuring of exchange-based security trading. Electronic trading systems have attracted a rapidly growing share in trading. The overall cost of electronic trading systems is lower than that of floor-based systems. In addition, electronic trading systems permit remote access of traders. Yet, the coexistence of floor and electronic trading systems indicates that each system has its strengths and weak- nesses. The purpose of this paper is to provide some insight into this issue by analyzing information diffusion in both systems. The quality of information diffusion is important because it affects traders’ ability to react quickly to new information and to protect themselves against adverse trading with insiders. Therefore, information diffusion should affect trading activities in both systems. The variety of existing screen-based trading systems as well as of floor trading systems makes it impossible to define each system precisely. Yet most electronic systems share one important feature which is the anonymity of trading, i.e. names of traders do not appear on the screen. 1 Information diffusion related to names and observable behavior of traders is lacking in electronic trading systems. Floor trading systems provide this information. All traders can observe and talk to each other. 2 Anonymous electronic trading systems try to make up for this lack of information by offering traders insight into the limit order book. Most floor trading systems do not reveal this information. As both trading systems differ in many features, it is impossible to ascribe a decisive role to this information differential in the competition between both systems. Yet this differential may be important. One method to look into this issue is to relate the information differential to certain characteristics of the trading situation. We argue that the information differential does not unambiguously support the competitive strength of one over the other system, but that such support depends on the trading situation. The main hypothesis advanced in this paper states: The information value, provided by the insight into the limit order book in the electronic trading system, relative to the information value of observing traders in the floor trading system, declines when the intensity of private and public information arrival increases. In other words, the hypothesis states that information diffusion in a floor trading system relative to that in an electronic trading system renders floor trading more attractive in times of high information 1 An exception is the APT-system of the LIFFE. This electronic trading system is in operation only before and after floor trading. In this system, floor trading is imitated; therefore names of traders appear on screens. Trading volume is modest. 2 These informational advantages of floor trading systems make it difficult for floor traders to switch to anonymous electronic trading systems. Swiss stock traders, for example, complained about the gap in information diffusion between floor and electronic trading when floor trading was replaced by Ž . electronic trading in Zurich in 1996 Cathomen, 1996 . ¨ intensity. Therefore, trading activity in a floor system should grow at a higher rate than in an electronic system when more information arrives. The basic idea behind this hypothesis is that in times of low information intensity, only a few transactions take place providing only little information. Therefore, the insight into the limit order book in the electronic system provides more information and, thus, renders trade in the electronic system more attractive. In times of high information intensity, however, the high trading frequency implies a steady flow of transactions data so that the limit order book information has little value. Furthermore, observing other traders and exchanging opinions on the floor provide valuable information. As a consequence, information diffusion in the floor system is faster making trade in this system more attractive. Since information intensity is not observable, we follow Admati and Pfleiderer Ž . 1988 who argue that information arrival is reflected in the time patterns of price volatility and trading volume. High price volatility indicates high information intensity. Therefore, a testable implication of our main hypothesis is that the market share in trading volume of the electronic system should be inversely related to price volatility. This hypothesis is tested by analyzing the trade of the Ž . Bund-Future contract at the DTB Deutsche Terminborse, renamed EUREX and ¨ Ž . at the LIFFE London International Financial Futures Exchange . The Bund-Future is a futures contract on a national German Government Bond with an annual coupon of 6 and residual maturity of 8.5 to 10 years at contract expiration. This contract is traded in almost identical design at the DTB, an electronic exchange, and at the LIFFE, a floor-based exchange. If our hypothesis is correct, then the DTB’s market share in trading activity should be inversely related to price volatility. Our analysis of transactions data from 1991 to 1995 confirms our hypothesis. However, the support declines over time. This is not surprising since low information intensity situations characterized by few transactions per unit of time become rare. Hence, the explanatory power of price volatility as a proxy for information intensity declines over time. Ž . A similar study has been done by Fremault Vila and Sandmann 1997 . They Ž analyze the trade of the Nikkei Stock Index Future at the SIMEX Singapore . International Monetary Exchange , an open outcry market, and the Osaka Securi- ties Exchange, a computerized market. In contrast to us, they find that the computerized market attracts additional trading volume in periods of high price Ž . volatility. Fremault Vila and Sandmann 1997 hypothesize that this result may be explained by the markets’ relative sizes instead of their trading mechanisms. Since the Osaka trading volume exceeds the SIMEX volume by about 60, it could be that in times of very active trading orders preferably go to the more liquid market. The paper is structured as follows. In Section 2, both trading systems will be briefly portrayed and the hypotheses about traders’ preferences for one or the other trading system will be developed. Section 3 presents and discusses the test results from the trade of the Bund-Future contract at the DTB and the LIFFE. Conclu- sions are offered in Section 4.

2. Traders’ preferences for electronic vs. floor trading