اطلاعات، سرعت هزینه در مقایسه با تجارت آف، و جهت مسیریابی تصمیم گیری در بازار سهام ایالات متحده

Prior research indicates that both execution speed and cost are important to traders, but that these two dimensions of execution quality are negatively related across U.S. equity markets. In our paper, we examine how U.S. equity traders, who are (un)informed about future price changes, trade-off between speed and cost in their order-routing decisions. We find that informed traders are more likely to choose trading systems that allow them to trade-off lower cost for faster speed; whereas, uninformed traders are more likely to choose trading systems that allow them to sacrifice speed for lower costs. Our results indicate that traders have varying preferences for the different dimensions of execution quality based on their information levels. These differences subsequently influence order-routing decisions.

U.S. equities trade in multiple markets and research shows that execution speed and cost are negatively related across markets. For example, market centers that provide faster (slower) execution tend to have higher (lower) trading costs (e.g., Boehmer, 2005). While all traders presumably have a preference for minimizing their cost of trading, some traders might have an equal (or even greater) preference for minimizing their time to execution. Faster trading, among other things, better enables traders to capitalize on information. Whether or not informed traders prefer faster (slower) execution to higher (lower) trading costs remains an unresolved issue in the financial literature. For example, many theoretical models assume that informed traders prefer to submit market orders (or to execute immediately), but empirical research suggests the opposite, namely that informed traders prefer limit orders (e.g., Kaniel and Liu, 2006). If execution speed and cost are both important to traders but negatively related across trading systems, then which dimension of execution quality do traders prefer? How does this subsequently influence their order-routing decision? Our paper provides some insight for answering the above questions. It focuses on speed vs. cost trade-offs, and subsequent order-routing choice, with respect to trader information content. We infer trader information content by analyzing market prices following order execution. If, on average, prices rise (fall) following buy (sell) orders, we conjecture that these orders are more likely to be submitted by informed traders. On the other hand, if prices fall (rise) following buy (sell) orders, we conjecture that these orders are more likely to be submitted by uninformed traders. How (un)informed traders make order routing decisions has important implications for market design, market prices, and transaction dynamics. Using proprietary data about active traders, who directly access/provide liquidity in U.S. equity markets, we find that informed traders are more likely to trade-off lower costs for faster speed in their order-routing decisions. Traders who are less/not informed about future price direction are more likely to make the opposite order-routing choice (i.e., to trade speed for lower costs). Our paper focuses on a trader's order-routing decision in the Nasdaq marketplace. On Nasdaq-listed stocks, competing market makers in the Nasdaq market center provide a significant and stable source of liquidity. Traders can execute against market-maker displayable quotes by using automated execution systems, or, for larger orders, they can negotiate with market makers for trading at different (favorable) prices. An alternative to trading with a market maker is to place a limit order in the marketplace and, effectively, compete with market makers to trade at better prices.1 Traders can display their limit order in the consolidated quote stream by placing it on the Nasdaq stock market or on any number of competing electronic trading systems.2 The multiple execution options available to traders routinely force them to make important tradeoff decisions with respect to execution cost and execution speed. While some traders may prefer speed to cost or vice versa, economic theory generally assumes that informed traders have a tendency to execute their orders quickly so they can exploit short-lived information. That is, order-routing options that offer fast execution (whether on Nasdaq or on a competing electronic trading system that is part of the consolidated quote stream) should attract more informed traders. We find traders are much more likely to experience lower trading costs when they use market-maker trading systems that offer price improvement opportunities (i.e., negotiated trading systems). But, these execution systems result in a much slower execution because human intermediation is involved in the trading process. Traders can execute faster if they use market-maker automated execution systems, but execution over these systems typically results in much higher trading costs because traders are forced to pay the bid–ask spread. We find orders executed over automated market-maker trading systems convey more information than orders executed over negotiated market-maker trading systems. These findings suggest that informed traders are more willing to route their orders over trading systems that trade-off lower costs for faster speed. An alternative to trading with a Nasdaq market maker is to route a limit order to an electronic order book. While traders can avoid the bid–ask spread by posting a limit order in the marketplace, this opens them to adverse selection costs when (if) execution occurs. Both informed and uninformed traders choose this option. We find that the information content behind a limit order is highly dependent upon how the order is priced in relation to the national best bid and offer (NBBO) at the time of submission. Execution time declines and the information content of an order rises as traders submit more aggressively priced limit orders. For example, marketable limit orders execute the fastest and convey the most information. Our study contributes to three streams of literature which examine issues related to trading speed (costs) and informed trader behavior in securities markets. Theoretical models generally classify traders as either informed or uninformed (i.e., liquidity traders) based on their information content. Informed traders play a significant role in the price discovery process and understanding the motives and actions of this trader group has long been an important focus of microstructure studies. First, our study contributes to literature which examines multiple dimensions of execution quality and trader preferences for the different dimensions. A growing academic literature has begun to use new data sources to examine execution quality along more than one dimension.3 Studies indicate that traders are concerned about both execution speed and execution costs (e.g., Boehmer et al., 2007), but these two dimensions of execution quality are negatively related (e.g., Battalio et al., 2003 and Boehmer, 2005). Because the data (e.g., market center execution quality reports) used in these studies do not identify individual traders, as noted by Boehmer (2005), it is difficult to assess trader preferences for the different execution quality dimensions or how traders trade-off between speed and costs.4 For example, execution speed and cost cannot be measured from trader original order, which is presumably how traders assess execution quality. Furthermore, trader order submission strategies are not observable (e.g., original order size, order routing choice, order type choice, etc.). This complicates inferences about whether traders prefer faster (slower) trading to lower (higher) costs. While prior studies document differences in execution quality across markets using anonymous trade-level data, our study differs in that we aim at gaining a better understanding of which execution quality dimension (un)informed traders prefer. We analyze trader-defined order-level data, which allow us to link trader (ex ante) order submission decisions to subsequent (ex post) measures of execution speed and costs. Second, our study contributes to that literature which examines trader choice of order type. Theoretical models have traditionally assumed that informed traders submit market orders and uninformed traders submit limit orders (see, for example, Glosten, 1994, Kyle, 1985 and Rock, 1990). In competitive securities markets informational advantages erode rapidly and traders must act quickly (or submit a market order) in order to profit from their short-lived information. In spite of the fact that the commonly held belief among many that informed traders prefer to use market orders, recent empirical research seems to indicate the opposite, or that informed traders prefer limit orders. For example, Kaniel and Liu (2006) find that limit orders are more informative than market orders, although they admittedly use an information horizon for informed traders that is, on average, longer than what most theoretical models assume. Anand et al. (2005) find support for informed trader use of limit orders in the short-term. They find that informed traders submit more limit orders towards the end of the day, and that limit orders submitted by institutions convey more information than those submitted by individuals. Although empirical studies portray informed traders as being patient, or having a preference for limit orders, these studies examine trading in a manually-driven, centralized market setting with two order types (market and limit orders).5 Our study examines trading in an electronically-driven, multi-market setting with several order types (i.e., the existing U.S. equity market structure). This allows us to examine the link between information and order type through the order routing decision. Finally, our study contributes to that literature which examines quoting and trading across U.S. equity markets. In particular, several studies have used transaction-level data to examine differences between Nasdaq and Electronic Communication Networks (ECNs). ECNs are electronic trading systems that automatically match buy and sell orders at specified prices. The financial literature has found that Electronic Communication Networks (ECNs) quotes (e.g., Huang, 2002) and trades (e.g., Barclay et al., 2003 and Goldstein et al., 2008) are more informative than Nasdaq market-maker quotes and trades. Barclay et al. (2003) argue that informed traders have a preference for trading on ECNs because trading is fast and completely anonymous. Do trading systems offering speed and/or full anonymity attract more informed traders? It is difficult to answer this question by examining anonymous quote and trade data on Nasdaq and ECNs because execution speed is unobservable and both markets offer traders similar execution services. For example, Nasdaq market makers have long offered automated or fast execution (with pre-trade anonymity) options for public traders.6 By using proprietary rather than transaction-level data about individual traders, we can measure execution speed directly and then test whether trading systems (both Nasdaq and ECNs) that offer fast execution and anonymity do indeed attract more informed traders. The remainder of our paper proceeds as follows. In the next section, we discuss the Nasdaq market structure and develop our main hypothesis. Section 3 describes the three data sources used in our paper. In Section 4, we examine which order-routing choices traders make, and in Section 5, we examine the multiple dimensions of order execution quality and information across the different Nasdaq order-routing systems. In Section 6, we analyze our findings at the trader-level rather than at the order-level. Section 7 provides concluding remarks.

Execution quality is important to traders in financial markets. Yet, how traders assess execution quality is unclear because there are multiple ways to do so. Perhaps the two most important dimensions of execution quality are the speed at which an order is executed (i.e., execution speed) and the cost incurred with executing (i.e., execution cost). While prior studies indicate that both execution cost and execution speed are important, research results show that these two dimensions of execution quality are negatively related. For example, market centers that report faster (slower) fill rates also tend to report higher (lower) trading costs (Battalio et al., 2003 and Boehmer, 2005). If traders are forced to sacrifice one dimension of execution quality for another, then do traders have a preference for a particular dimension? In our paper, we provide some insight for answering this question by analyzing the order-routing decisions of individual traders on Nasdaq-listed stocks. We find that execution quality directly influences order-routing decisions and traders have varying preferences for the different dimensions of execution quality. Traders who are more informed about future price direction are more willing to route their orders over trading systems that offer higher costs for faster speed. On the other hand, traders who are not (less) informed about future price direction are more willing to route their orders over trading systems that sacrifice speed for lower costs. The trade-off between speed and cost is quite significant. For example, controlling for various order characteristics, market condition changes, market structure changes, stock characteristics, trader characteristics, and endogeneity concerns, we find that orders routed over automated market-maker trading systems are more informed than orders routed over negotiated market-maker trading systems. Furthermore, orders routed over automated market-maker trading systems execute, on average, 81% faster than orders routed over negotiated market-maker trading systems (i.e., human intermediated trading systems). Automated trading systems are clearly faster than negotiated trading systems, yet human intermediated trading systems can offer lower costs because trading is not anonymous and market makers have ample time to uncover an individual's motivation for trade. We find that effective spreads, on average, are 82 basis points lower on negotiated market-maker trading systems than on automated trading systems. Our study examines how individual's trade-off between speed and cost in their order-routing decisions. Whether or not individual preferences towards execution quality vary with market structure is an interesting question for future research. Our study is focused on trading in Nasdaq-listed stocks. Trader preferences toward execution quality, and/or the extent to which traders are willing to trade-off one dimension for another, might vary in other market settings with different trading structures than Nasdaq.