ساختار بازار، از هم گسیختگی و کیفیت بازار

This paper studies the impact of order flow fragmentation on market quality. Due to differences in market structure, order flow becomes more consolidated when stocks switch listings from a dealer market (NASDAQ) to an exchange (NYSE). We find that the post-switch improvements of market quality are related to the degree of order flow fragmentation on NASDAQ as well as the change of fragmentation after trading on the NYSE. After controlling and correcting for potential selection bias arising from a nonrandom sample, we find that order flow fragmentation affects market quality as predicted by finance theories. Our paper shows that order flow consolidation is particularly valuable for less liquid securities.

A key issue of interest to financial economists is why the same or similar securities have different trading characteristics in differently structured markets. This issue has raised questions about optimal market design. One of the goals of market design is to facilitate liquidity provision and price efficiency. Order handling rules, decimal pricing and new trading technology have narrowed the difference between the New York Stock Exchange (NYSE) and NASDAQ in terms of trading as well as market quality.1 A key difference of trading, however, still exists between these two types of market structures. On a dealer market, such as NASDAQ, order flows are usually more fragmented than on an exchange, such as the NYSE, where all buy and sell orders are consolidated and interact with each other. Indeed, a significant amount of trading of NASDAQ-listed stocks takes place on various ECNs and dealers.2 Many studies compare market quality across different market structures, but there is limited evidence explaining why observed differences exist in recent periods.3 This paper utilizes natural experiments of exchange switching to examine the impact of order flow fragmentation on market quality. Due to differences in market structure, NASDAQ stocks are traded by a large number of market venues, including NASDAQ SuperMontage, various ECNs, dealers, and regional exchanges, and therefore have a higher degree of order flow fragmentation than their NYSE peers. When NASDAQ stocks switch listing to the NYSE, order flows migrate from dealers and ECNs to the exchange and become more consolidated. Such natural experiments allow us to examine the impact of order flow fragmentation on liquidity provision and price efficiency. Using switching stocks in this study enables us to control for firm characteristics and remove potential influence due to an imperfect match. Market fragmentation has been widely studied in the literature. Theoretical work of Mendelson (1987) and Madhavan (1995) shows that fragmentation can result in reduced liquidity, higher price volatility, and violations of price efficiency. Empirical evidence on fragmentation and market quality is, however, inconclusive. Some studies find negative effects of decentralized, or “fragmented,” trading on market quality.4 Others show that fragmentation with competition does not hurt market quality.5Amihud et al. (2003) provide evidence that order consolidation improves liquidity and pricing, and Barclay and Hendershott (2004) show the positive impact of trading consolidation on liquidity. We examine the stocks of 39 companies that transferred from NASDAQ to the NYSE during 2002 and the first quarter of 2003. The stocks in our sample on average have a market capitalization of $1.4 billion each and trade 650,000 shares daily. They are not large and actively traded stocks if compared to index stocks and actively traded ETFs.6 We find that, as in earlier studies, the stocks that switch to the NYSE experience improvement in liquidity provision and price efficiency. We observe reductions in quoted and effective spreads as well as volatility. We also find the realized spreads drop sharply and are negative on the NYSE, suggesting more competitive liquidity provision on the NYSE than NASDAQ. Regarding execution speed, we find that the NYSE is faster for market orders, but NASDAQ is faster overall. Our study finds that market fragmentation explains the market quality changes. Using the SEC 11Ac1–5 execution data, we develop proxies to quantify order flow fragmentation. We find that the post-switch magnitude of market quality improvements is related to the degree of order flow fragmentation on NASDAQ as well as the change of fragmentation after trading on the NYSE. In addition, we find stock liquidity is negatively correlated with the post-switch reduction in volatility and execution cost, suggesting that the order flow consolidation is particularly valuable for less liquid securities. Compared to pre-decimal and pre-NASDAQ reform periods, the evidence in this paper shows that the NASDAQ-NYSE differences in quoted and effective spreads become smaller. Liquidity provision on the NYSE and NASDAQ has been studied intensively in the literature with the conclusion that the NYSE provides lower execution costs and volatility than NASDAQ.7 Recent evidence has shown that the NASDAQ's reform in 1997 has improved its market quality and narrows the spread difference between NASDAQ and the NYSE (Weston (2000) and Sapp and Yan (2003)). In Boehmer (2005), the magnitudes of differences of NYSE-NASDAQ matched samples are smaller than previously documented.8 We believe the narrowing differences between NASDAQ and the NYSE relate to increased competition within NASDAQ (order handling rules), reduction of minimum price variation (decimalization), higher degree of market transparency (SEC 11Ac1–5 Rule), as well as improved inter-market linkages among NASDAQ market centers, which help to reduce market segmentation and increase inter-market competition. Our finding suggests the incremental value of order flow consolidation on market quality above and beyond the aforementioned measures of improving competition, transparency, and market efficiency on NASDAQ. Furthermore, the average market quality differences that we have found in this paper, 3 cents (10 bp) on quoted spread and 3 cents (16 bp) in effective spread, are comparable to Boehmer (2005) and economically significant, particularly considering overall decreasing transaction cost and increasing trading volume.9 Our study supports the notion of a tradeoff in market structure between order flow consolidation and competition among market centers. There is evidence that the positive effect of competition may dominate in certain markets for intrinsically highly liquid securities.10 The stocks we examine in this paper are not the most liquid and actively traded securities. Electronic order routing technology and automatic execution may have improved the inter-market competition, but it appears that competition among these market centers does not dominate the benefits of order flow consolidation for these securities. Our findings illustrate the positive impact of order flow consolidation on market quality and price efficiency. Our paper finds that the extreme value volatility measurement complements the conventional return-variance volatility. Volatility measured by extreme value, such as high and low prices, is highly correlated with return-variance volatility and closely relates to market structure. Since our sample of switching stocks is not random and our study may involve a selection bias, we employ three different ways to address the possibility of sample selection bias, including a simple sample comparison, the use of a matched control sample, and the Heckman 2-stage selection model. We find our results are not affected by the sample selection bias. Our paper proceeds as follows. Section 1 introduces our sample and data for the stocks that switched markets, and describes our methodology. Section 2 presents the findings on changes in volatility and information efficiency of prices for switching stocks. Section 3 presents the evidence on quoted and effective spreads. Section 4 examines and rejects the hypothesis of selection bias for the switching stocks. Section 5 gives additional evidence for fragmentation effects, making use of cross-sectional differences among switching stocks. Section 6 concludes.

We study the impact of order flow consolidation on liquidity provision and market quality by using the natural experiments of exchange switching. Due to differences in market structure, NASDAQ stocks are traded by a large number of market venues and have a higher degree of order flow fragmentation than their NYSE peers. When NASDAQ stocks switch listing to the NYSE, order flows migrate from dealers and ECNs to the exchange and become more consolidated. Such natural experiments allow us to examine the impact of market fragmentation on liquidity provision and price efficiency. On average stocks have experienced improvement in market quality and price efficiency on the NYSE. We find that the pre-switch degree of order flow fragmentation on NASDAQ has a strong explanatory power for the post-switch market quality improvements on the NYSE, implying that companies with more fragmented trading on NASDAQ experienced larger improvements in market quality when switching to the NYSE, ceteris paribus. In addition, we also find stock liquidity is negatively correlated with the post-switch reduction in price inefficiency and execution cost, suggesting that the order flow consolidation is particularly more valuable for less liquid securities. Our results show that order flow consolidation has incremental value above and beyond the measures to improve inter-market competition, transparency, and efficiency. One key to the market quality of the NYSE is closely associated with the consolidation of order flows. These results underline the importance of order flow consolidation in a single primary market where buy and sell orders can interact competitively and prices can be discovered efficiently. These findings do not appear affected by a sample selection bias. Our study complements other work in the area of optimal market structure. Volatility and execution cost are important dimensions of market quality, but they are not the full story of it. Another dimension of market quality is related to a market’s ability to handle stress and liquidity shock. Barclay et al. (2003) demonstrate that the NYSE’s better performance over NASDAQ in handling liquidity shock and market stress is related to order flow consolidation in the auction market. Elliott and Warr (2003) show that the NYSE’s ability to adjust more quickly to liquidity shocks than NASDAQ stocks is due to NYSE’s market structure and consolidation of liquidity. Examining trading post the 9-11 event, Chung and Kim (2005) conclude that the NYSE’s market structure works more efficiently than NASDAQ in handling extreme market conditions and uncertainty. Studying fragmentation and a market’s ability to handle stress has many implications which we leave for future research. In short, while market center competition has important beneficial effects on market functioning, the economics of order flow consolidation appear to be a dominant factor in determining how well markets provide liquidity and form prices that allocate capital efficiently.