شفافیت نیمه تمام ، بازار نمایندگی، و کیفیت بازار ارز

This paper examines the effects of a semi-transparency event, the introduction of the electronic trading system (EBS), on the market quality of a typical dealership market – the FX market. We find that increasing transparency leads to smaller quote disagreement among dealers and higher trading volume, but the beneficial effects are bigger for uninformed dealers than informed dealers. We also find that information efficiency improves overall in the semi-transparent system; however, informed dealers are found to quote less aggressively in the more transparent market. We conclude that semi-transparency raises market quality in general, but that it is the uninformed dealers who benefit more from this increased quality.

Transparency is a fundamental issue in the design and regulation of financial markets. Since the information structure and liquidity providing mechanism are different in the centralized order driven auction market and the decentralized quote driven dealership market, the impact of changing transparency on market quality should be different in different market structures as well. For the dealership market, the theoretical model of Duffie, Garleanu, and Pedersen (2005) suggests that increasing market transparency should lead to higher market liquidity if there are multiple market makers. Recognizing that some of these dealers have private information, Golosov, Lorenzoni, and Tsyvinski (2008) find that uninformed dealers can learn the state of the world from informed dealers through transactions with them, but this comes at a cost (or rent for informed dealers), which can be reduced by increasing transparency. In general, these studies suggest that greater transparency should lead to improved market quality, but it benefits uninformed dealers at the expense of informed dealers. Furthermore, depending on how much information is revealed, the degree of transparency can be quite different. Intuitively, in a completely transparent market, informed dealers would hesitate to trade to avoid the loss of their information advantage. Conversely, in a totally opaque market, uninformed dealers will be taken advantage of, and are thus unwilling to trade. Hence, there is a trade-off between transparency and market quality, and an optimal quality should occur at some semi-transparency level. This view has been supported by theoretical papers such as Madhavan, 1995, Pagano and Roell, 1996, Lyons, 1996, July and Dumitrescu, 2008. Empirical examinations of changing transparency on market quality are not lacking. Some of the existing research, however, focuses on the centralized market. For example, Boehmer, Saar, and Yu (2005) examine the effect of increasing pre-trade transparency in limit order book by introduction of the OpenBook service on the New York Stock Exchange (NYSE). Madhavan, Porter, and Weaver (2005) study greater pre-trade transparency by publicly disseminating its limit order book in Toronto Stock Exchange increases in 1990. Other related research does focus on the dealership market, but examines full transparency events (i.e. revealing both quote and order book pre-trade or both transaction price and quantity post-trade). Chung and Chuwonganant (2009), for instance, examine the impact of revealing limit order book brought by SuperMontage on NASDAQ. Bessembinder, Maxwell, and Venkataraman (2006) examine the initiation of the Trade Reporting and Compliance Engine (TRACE) reporting system, which reveals full post-trade transaction information, on the corporate bonds market. Finally, these studies do not examine the effect on different dealers separately. The existing empirical examinations of semi-transparency are limited to a few experimental studies which are not equivalent to real market data. Moreover, different studies draw conflicting conclusions. Bloomfield and O'Hara (1999) use experimental results to conclude that semi-transparency (quote disclosure only) has little effect on price efficiency while only full-transparency (both quote and order disclosure) leads to better market quality. In contrast, Flood, Koedimk, and Mahieu (1999) report that a semi-transparent market leads to narrower spreads, higher volume, and slower price discovery. Thus, no research to date has empirically tested the theoretical conclusions introduced above with regard to the impact of semi-transparency on a dealership market.1 The experimental research also draws mixed conclusions. This paper adds to the literature by studying a semi-transparency event in a dealership market and specifically focusing on its impact on different types of dealers in terms of multiple aspects of market quality. Through the mid-1990s, the foreign exchange interdealer market primarily relied on a bilateral phone-based trading platform, in which access to market information was very limited. Two main electronic systems, Reuters system and EBS, were introduced to the market in 04/1992 and 09/1993, respectively. The electronic systems reveal the best live bids and offers available in the market before trade and publish the transaction price after trade, which greatly increases market transparency. However, the systems do not reveal the limit order book before trade and transaction quantity after trade.2 In this sense, the FX market only has pre-trade and post-trade semi-transparency. The impact of pre-trade and post-trade semi-transparency on market quality can therefore be determined by comparing the market qualities before and after the use of these electronic systems. To examine the impact, we obtain the data of tick-by-tick spot quotes3 for the Deutsche Mark–U.S. Dollar exchange rates around the introduction of the electronic systems. Among the two electronic systems, the EBS is the primary liquidity source for EUR/USD (DEM/USD before 1999).4Ding and Hiltrop (2010) also confirm that the introduction of the EBS generated much more significant impact on bid-ask spreads for DEM/USD than that of Reuters system. Accordingly, this paper only focuses on the EBS to test our hypotheses. There are multiple aspects of market quality. We follow Kaul and Sapp (2009) as well as experimental hypothesis proposed by Bloomfield and O'Hara, 1999 and Flood et al., 1999 to focus on volatility, volume, and price efficiency to show the impact of the semi-transparency event. We find that the semi-transparency leads to lower volatility and higher trading volume, and small dealers receive more beneficial effects than big dealers. We also find that the semi-transparency increases information efficiency. However, informed dealers are found to quote less aggressively in the more transparent market. Overall, the semi-transparency raises market quality, although uninformed dealers benefit more than informed dealers from this increased quality. These empirical findings support theoretical conclusions that semi-transparency improves market quality by Lyons, 1996, July and Dumitrescu, 2008, and also are consistent with experimental results obtained in Flood et al. (1999). These results also support the claim that greater transparency benefits uninformed dealers at the expense of informed dealers in the decentralized market, as suggested by Duffie et al., 2005 and Golosov et al., 2008. The rest of the paper is organized as follows: Section 2 surveys the literature and proposes general hypotheses to test; Section 3 introduces the data; market volatility, volume, and information efficiency are tested in 4, 5 and 6, respectively; Section 7 tests price discovery efficiency; Section 8 concludes.

the EBS, on the market quality of the FX market. The paper tries to answer whether the change from opacity to semi-transparency affects the market quality and if so, how different the effect is between informed (big) and uninformed (small) dealers. We tested basic market quality features, including volatility, trading volume, informational efficiency, and price discovery efficiency. The paper finds that semi-transparency brought by the new system leads to a decrease in market volatility and increase in trading volume, but informed dealers benefit less fromthe newsystemthan uninformed dealers. The paper also finds that informational efficiency improves overall after the introduction of the new system, but informed dealers tend to reveal less information in themore transparentmarket. Several robustness tests are conducted, and the results confirm our conclusions. Prior to our study, the impact of electronic trading systems on the FX market had not been studied extensively. Even now, further research is necessary in several key areas. Although quoting frequency turns out to be a reliable proxy for volume in our tests, it is still not as accurate as actual volume data. It would consequently be valuable if future studies retest our hypotheses with real volume data. With actual transaction data, future research can also test how different dealers' monetary welfare change after the new system. Moreover, current research is limited to the inter-dealer market, but it could be extended to the customer FX market. Such research could, for example, focus on Internet trading, which is gaining popularity and changing the information structure in the customer market.