INTERNATIONAL EVIDENCE ON ALGORITHMIC TRADING
Ekkehart Boehmer Kingsley Fong Julie Wu
March 14, 2012
Abstract We use a large sample from 2001 – 2009 that incorporates 39 exchanges and an average of 12,800 different common stocks to assess the effect of algorithmic trading (AT) intensity on liquidity in the equity market, short-term volatility, and the informational efficiency of stock prices. We exploit the first availability of co-location facilities to identify the direction of causality. We find that, on average, greater AT intensity improves liquidity and informational efficiency, but increases volatility. The volatility increase is robust to a range of different volatility measures and it is not due to more “good” volatility that would arise from faster price discovery. These patterns are widespread and are not limited to a few markets, but they vary in the cross-section of stocks. In contrast to the average effect, more AT reduces liquidity in small stocks; has little effect on the liquidity of low-priced or highvolatility stocks; and leads to greater increases in volatility in these stocks. Finally, during days when market making is difficult, AT provide less liquidity, improve efficiency more, and increase volatility more than on other days.
Boehmer is from EDHEC Business School, 393 Promenade des Anglais, 06202 Nice, France (
[email protected]). Fong is from Australian School of Business, UNSW, Sydney, NSW 2052 (
[email protected]). Wu is from the Terry College of Business, University of Georgia, Athens, GA 30605 (
[email protected]). We thank seminar participants at the Bank of England, Bocconi University, EDHEC Business School, EDHEC Risk Institute London, ESSEC, Monetary Authority of Singapore, National University of Singapore, University of Houston, TCU, and SMU for their helpful comments.
Abstract We use a large sample from 2001 – 2009 that incorporates 39 exchanges and an average of 12,800 different common stocks to assess the effect of algorithmic trading (AT) intensity on liquidity in the equity market, short-term volatility, and the informational efficiency of stock prices. We exploit the first availability of co-location facilities to identify the direction of causality. We find that, on average, greater AT intensity improves liquidity and informational efficiency, but increases volatility. The volatility increase is robust to a range of different volatility measures and it is not due to more “good” volatility that would arise from faster price discovery. These patterns are widespread and are not limited to a few markets, but they vary in the cross-section of stocks. In contrast to the average effect, more AT reduces liquidity in small stocks; has little effect on the liquidity of low-priced or highvolatility stocks; and leads to greater increases in volatility in these stocks. Finally, during days when market making is difficult, AT provide less liquidity, improve efficiency more, and increase volatility more than on other days.
1.
Introduction By most accounts, high frequency trading (HFT) represents the majority of trading volume in
today’s markets. HFT refers to orders submitted by algorithms that emit orders or order cancellations in reaction to market updates or other events within milliseconds. Mainly because of their overall importance, but also because HFT strategies are neither transparent nor well understood, there is substantial public policy interest in this issue. Market regulators around the world debate whether HFT should be regulated, and place increasing scrutiny on order submission strategies and their effects in markets that are associated with HFT. Despite this debate and a recent flurry of theoretical and empirical work in this area, we still face many open questions. In this paper, we take a very basic but comprehensive approach that contributes new evidence to this debate. We follow Hendershott, Jones, and Menkveld (2010) and infer proxies for algorithmic trading (AT), a precondition for HFT, from measures that are derived from the intensity of order-related message traffic. We use nine years of intraday security-level quote and trade data for 39 markets around the world. This sample covers an average of 12,800 firms, excluding the U.S. This new and comprehensive sample allows us to exploit variation in algorithmic trading intensity in the cross-section of stocks and the cross-section of markets. We have several objectives. First, we describe the relationship between algorithmic trading and market quality, measured in terms of liquidity, informational efficiency, and short-term volatility. While some studies of HFT have looked outside the U.S. (see Hendershott and Riordan, 2009, or Menkveld, 2010), they are based on relatively small samples. Even the most comprehensive study so far, Hendershott, Jones, and Menkveld (2010), does not use data beyond 2006 and its main analysis is based on a change in trading protocol that happened in 2003. These dates arguably precede the real growth of HFT. Overall, we have little understanding of how the relationship
between AT and market quality evolves over time, especially outside of the U.S. Our within-country analysis sheds new light on these issues. Second, we exploit the presence of several separate cross-sections of firms. We investigate whether features that are known to affect order submission strategies, such as market cap, share price, and idiosyncratic volatility, also impact the effects of AT on market quality. Third, existing evidence suggests that HFT provides liquidity to other traders, and that fast traders act as informal market makers (Brogaard, 2011b; Jovanovic and Menkveld, 2011). In contrast to exchange-regulated market makers, however, informal market makers are not subject to affirmative obligations, such as a requirement to always provide liquidity on both sides of the market. Therefore, it is likely that informal market makers withdraw from the market when conditions get too difficult. Kirilenko et al. (2011) show that this happened during the “Flash Crash” of 2010; in this paper, we pose the more general question whether algorithmic traders reduce the intensity of their market making strategies when they become more costly to implement. We find that greater AT intensity is, on average, associated with more liquidity, whether measured at the transaction level or at the daily level, faster price discovery, and greater volatility. These results are remarkably consistent across different markets. They are robust to using different econometric models for estimation and to different measures of volatility. To link AT causally to market quality, we use co-location events as instruments. Co-location events allow fast traders to physically locate their computer hardware next to the exchanges’ computer to minimize data turnaround times. These events are essential in facilitating AT and represent exogenous shocks to AT that do not directly affect market quality. We use these events as instruments for AT and find evidence that supports causality from AT to market quality–more AT improves liquidity and efficiency, but increases volatility.
2
Volatility is important to traders and issuers. Greater volatility makes limit orders more costly, and may discourage some traders from supplying liquidity. As a result, future liquidity may decrease or the price attached to liquidity risk may increases. Issuers dislike volatility because higher volatility may lower share prices or make subsequent equity issue more difficult. Certain types of volatility could be desirable. For example, prices change faster in response to new information, and volatility could be higher, when markets are more efficient. It is thus conceivable that the greater efficiency that is associated with more AT also produces higher volatility. In this case, the elevated volatility could be desirable because it would be associated with faster price discovery. To address this issue empirically, we hold constant each stock’s level of informational efficiency, but we still find that AT increases volatility. Therefore, it is unlikely that the AT-induced change in volatility is due to faster price discovery. Our second objective is to assess the cross-sectional determinants of AT’s effect on market quality. While the average effect of AT on market quality is positive, we also find substantial skewness in AT. This suggests that many firms either do not experience AT or are subject to negative consequences. We believe that it is important to understand the cross-sectional determinants of the benefits and costs of greater AT intensity. Specifically, stocks that are larger in terms of market capitalization, have higher share prices, or have low volatility are typically also easier to trade. It is easier to provide liquidity in these stocks, and the trading intensity is high enough to allow for significant algorithmic activity. In particular, high-frequency market making strategies are probably easier to implement in these stocks than in small, low-priced, high volatility stocks. To address these issues, we divide stocks into terciles based on market cap, price, and volatility within each market, and allow the effects of AT to differ according to these characteristics. We find that indeed much of the benefits of high AT intensity accrue to large, high-priced, and low-volatility stock. When AT increases, liquidity actually declines for the smallest terciles of stocks and remains unchanged for low-volatility and low-price stocks. The main costs associated with 3
AT, an elevated level of volatility when AT intensity is high, are significantly higher in stocks that are small, low-priced, or have high volatility. Our third objective is to take a closer look at the market-making strategies that are common among algorithmic traders. We cannot observe actual strategies, nor can we identify specific traders that might employ them. We only observe the aggregate effect of AT and can measure how it changes over time at the stock level. We exploit this advantage of our data by designing a timeseries proxy for days when market making strategies are likely to be more costly. Then we examine whether the effects of AT differ on those difficult market-making days. We find that when market making is more costly, AT provides less liquidity; increases the fraction of informed trading, which also enhances efficiency; and their activity increases volatility more than it does on easy marketmaking days. Overall, our results show that algorithmic trading often improves liquidity, but this effect is smaller when market making is difficult and for low-priced or high-volatility stocks. It reverses for small cap stocks, where AT is associated with a decrease in liquidity. AT usually improves efficiency. The main costs associated with AT appear to be elevated levels of volatility. This effect prevails even for large market cap, high price, or low volatility stocks, but it is more pronounced in smaller, low price, or high volatility stocks. Finally, on days when market making is more costly, AT induces less liquidity improvement and increases volatility more than on regular trading days. In future research, we will use characteristics of markets, such as details of the trading protocol, to identify whether the effects of AT are subject to variation across markets. At the country level, we will relate metrics of financial market development, economic growth, stringency of securities regulation, and market activity to costs and benefits of AT. Finally, we will examine whether links exist between the intensity of algorithmic trading and the development of securities markets in a general sense (see Menkveld, 2010). These analyses will help guide policy decisions by identifying which firms or markets, if any, may benefit from regulatory restrictions on fast trading. 4
Our paper is organized as follows. We review the theoretical and empirical literature in the next section. In section 3, we discuss our data and define the key variables we use. We discuss our empirical design in section 4 and present our results in section 5. The final section concludes.
2.
Literature on algorithmic and high frequency trading HFT is a quite recent phenomenon and has experienced precipitous growth over the past
decade. A precondition for HFT is that trading algorithms are available and can be efficiently implemented. While algorithmic trading (AT) can come from either agency algorithms or from proprietary HFT, more AT will most likely imply more HFT. This is because HF traders compete for interacting with agency algorithms, either to supply liquidity or to create short-term alpha (see Hasbrouck and Saar 2011). Only in the 2000s have information technology and market structure developed into an arena that facilitates fast, automated trading. In the U.S., this is mainly a consequence of limit order display rules that were implemented during the 1990s and, in particular, Reg NMS in 2005. Other factors also play their roles, including the NYSE’s 2003 change to autoquote (mandatory automatic quote updates, as opposed to manual updates initiated by specialists), the development of fast markets that compete with the traditional venues, and the increase in capital available for proprietary trading. Other markets, including Europe, have also adapted trading protocols to facilitate HFT, mostly in the second half of the decade. Here, regulation also played a key role – MIFID, for example, provides a framework for off-exchange trading that set the stage for more AT (see Menkveld 2010). Despite being a young literature, analyses of HFT and algorithmic trading reveal an interesting dichotomy. Several theoretical and empirical models analyze HFT’s effects on market quality measures, including execution costs, volatility, and informational efficiency. While theoretical models mostly predict negative (or mixed) consequences of having fast traders in the market, the average effects estimated in empirical results tend to be positive. 5
A. Theory Hoffman (2010) extends Foucault’s (1999) limit order market and allows algorithmic and human traders to compete. Their ability to react faster to new events allows algorithmic traders to evade the adverse selection that is associated with stale limit orders. In this model, the effect of introducing algorithmic traders has ambiguous effects on trading volume and the price impact of human traders, but it decreases the profits of human traders. Considering the overall effect, Hoffman shows that in most cases human traders are strictly worse off when algorithmic trading is widespread. Cartea and Penalva (2011) design a model with liquidity traders, market makers, and HFT. They find that HFT increase overall trading volume, but also volatility and the price impact of liquidity traders. Market makers come out even–they lose market share (and thus revenues) for liquidity provision to the HFT, but are compensated with higher rewards for their remaining liquidity supply. The cost for the higher rewards to market making, and for the greater revenues to HFT, are all born by the liquidity traders. McInish and Upson (2011) arrive at similar conclusions using a different mechanism. In their model, strategic fast traders are the first to learn about quote updates and use this privileged information to trade at stale prices with slow traders. Here, too, does HFT activity increase trading costs for (slow) liquidity traders. In Jarrow and Protter’s (2011) model, HFT also observe order-flow information faster than other traders. They show that when demand curves are downward sloping, HFT’s activity affects price and creates a temporary mispricing that HFT can profitably exploit. In this case, the detrimental effect lies in less efficient pricing in addition to a transfer from slow to fast traders. A similar wealth transfer arises in an earlier model by Brunnermeier and Pederson (2005). They allow traders to follow order anticipation strategies (“predatory trading” in their model), a strategy that requires the ability to predict order flow in real time at high frequency and is easily implemented as a trading algorithm. Order anticipators attempt to predict large uninformed orders and then trade ahead of these orders, in the same direction. This increases the costs for the large
6
liquidity trader, who will end up trading at relatively inferior prices, perhaps even with the order anticipator. Brunnermeier and Pederson show that this leads to price overshooting and that it withdraws liquidity from the market when it is most needed (by the large trader). As a result, a wealth transfer occurs from the large liquidity trader to the order anticipator. Moreover, they show that the low-liquidity event can trigger systemic liquidity shocks for other traders and markets, thereby multiplying the negative consequences the order anticipator imposes on the market. The models discussed so far generally predict higher costs to uninformed and/or slow liquidity traders in the form of a greater price impact and pre-trade information leakage. Greater execution costs essentially involve a wealth transfer from slow to fast traders, but this does not necessarily have welfare implications. Biais, Foucault, and Moinas (2010) make an elegant argument in this regard. They show that HFT can generate gains either from trade or from adverse selection, which would arise from their faster access to information. But a social planner would only consider gains from trade, not from adverse selection. As a result, HFT overinvest in technology, which leads to socially undesirable outcomes. Overall, existing theoretical models agree that HFT has undesirable consequences for liquidity traders, informational efficiency, and volatility, and these effects may well result in lower social welfare. Jovanovic and Menkveld (2011) also study welfare implications of high frequency trading. In their model, middlemen intermediate between fast limit order and slow market order traders. Depending on parameter values, their entry may increase or decrease trading volume, and also has a mixed effect on welfare.
B. Empirical studies The recent spread of HFT has spurred a number of empirical studies that examine its consequences. Their inferences are easiest to synthesize by first categorizing the type of data that each study uses. The ideal data to analyze the consequences of HF and algorithmic trading would allow identification of trader (account) identifiers, which in turn allow the researcher to observe each 7
trader’s strategy across stocks and over time. To date, only one study (Kirilenko et al. 2011) has access to this type of data, and it is limited to the trading in index e-minis around the flash crash of May 6, 2010. To date, there is no academic study of equity trading that uses data where the researcher can directly identify trader-level order submission strategies and their consequences for algorithmic or HF traders, either over time or across stocks. Researchers follow one of two approaches: infer the portion of algo/HF trading from intraday data; or use data where HF traders are identified as a group. We discuss advantages and disadvantage of both approaches below. The most basic approach uses standard intraday transactions data and either develops proxies for HFT, or infers their actions from the speed with which traders react to market events. On the downside, these approaches do not exactly measure HFT or AT– instead, they infer it from the data with relatively unknown consequences for the quality of inference. But the advantage to these approaches is that they permit construction of broad and long panels that allow fairly general inferences. We adhere to this approach in this article, and closely follow Hendershott, Jones, and Menkveld (2010) in using message counts as a proxy for AT activity. Hasbrouck and Saar (2011) and Egginton, VanNess, and VanNess (2010) instead infer HFT activity from periods of apparent highfrequency activity. The former identify episodes of orders that react within milliseconds to market updates. The latter examine high-activity intervals, defined as one-minute periods where the quotesper-minute count exceeds a historical average by 20 standard deviations (and the trading day as a whole is not too different, defined as being less than two standard deviations away from the mean). These samples lose breadth relative to the message-count sample, but are able to study periods where HF activity takes actually place. The second category of data provides summary information about the type of trader. For example, Brogaard (2010, 2011a) and Hendershott and Riordan (2011) use a 2008-2009 Nasdaq sample that reveals the aggregate order flow generated by 26 HFT firms that account for about three quarters of sample trading volume. Here, the advantage is that actual HFT can be observed for 8
a random sample of 120 stocks. Potential drawbacks include the selection of HFT firms, which have been picked by the exchange that provided the data and, presumably, have been willing to have their order flows disclosed to academics. Because HF strategies are typically considered sensitive both from a legal and competitive perspective, this selection process could conceivably result in orders that are more benign than a random sample of HFT orders. There are also other potential issues that complicate inference from this dataset. First, the sample of 26 HFT firms does not include any of the large proprietary trading desks that, allegedly, are responsible for a substantial portion of HFT. Second, we do observe orders that the sample traders have submitted to other markets. Third, the high percentage of trading volume of those 26 firms is of some concern. High trading volume is not necessarily representative of HF traders – instead, they are typically characterized as traders with relatively low volume, but a very high ratio of order messages to trades (see Kirilenko et al. 2011 and Hasbrouck and Saar 2011). 1 Overall, while these data are currently the most informative about HFT strategies in equities, they also have significant shortcomings that complicate inference. In summary, the broadest data, which in principle would allow the strongest inferences, makes the least clear distinction between HF, algorithmic, and slow trading. In the other extreme, data sets that identify actual HF activity tend to be either small or not necessarily representative for other reasons. Moreover, some of these available data sets are subject to endogeneity concerns, because it is generally not easy to identify whether causality goes from market quality to HFT activity, of from HFT activity to market quality. Against these basic concerns about the data, most but not all results document positive effects of HFT. Hendershott, Jones, and Menkveld (2010) show that algorithmic trading is associated
1
The sample used by Hendershott and Riordan (2009) also falls into this category and it is not subject to a selection concern. They use a short sample of exchange-classified algorithmic trades at Deutsche Boerse. Also similar is Menkveld’s (2010) sample, who uses brokerage identities to infer the trades by a single HFT in the European market. These samples allow inferences about algos and HFT, respectively, but are limited to relatively narrow samples.
9
with better liquidity and faster price discovery. They use the 2003 introduction of autoquote at the NYSE as an instrument to argue that algorithmic trading causes these market quality improvements. Brogaard (2010, 2011a, 2011b) uses the 2008-2009 Nasdaq-selected data on 26 HFTs and shows ambiguous effects on volatility, but improvements in liquidity. Based on HFT activity inferred from millisecond-level responses, Hasbrouck and Saar (2011) find improvements in volatility, spreads, and depth when these fast traders are active. Using the same data as Brogaard, Hendershott and Riordan (2011) document that HFT play an important role in price discovery. Additionally, for a much smaller Deutsche Boerse sample that is not subject to selection concerns, Hendershott and Riordan (2009) find that algorithmic trading makes prices more informative. On the negative side, Kirilenko et al. (2011) argue that HFT worsened (but did not cause) the May 6, 2010 Flash Crash. Because this is the only study that can see exactly what HFT do, it carries significant weight among the empirical work we have so far. Dichev, Huang, and Zhou (2011) find that trading per se generates excess volatility, suggesting that HFT can lead to undesirable levels of volatility. Hasbrouck and Saar (2009) are the first to document the “fleeting” nature of many limit orders in electronic markets, and question the traditional view that limit orders provide liquidity to the market. This argument raises questions about the quality or usefulness of often short-lived liquidity that HFT supply. Consistent with this concern, Egginton, VanNess, and VanNess (2011) show that periods of extremely active quoting behavior are associated with degraded liquidity and elevated volatility. Importantly, they show that such episodes are surprisingly frequent. While there are good economic reasons for such quote-bunching to occur as a benign by-product of HF liquidity provision, as Hasbrouck and Saar (2011) argue, it is also possible that it arises as a consequence of intentional “quote stuffing.” McInish and Upson (2011) examine trading around quote changes and compare fast and slow responses. They find that fast traders strategically leave stale orders on the book and that slow traders often interact with these at prices that are inferior to those available elsewhere. Finally, Chaboud et al. (2009) look at HFT in the foreign exchange market and document that the correlation among algorithmic “machine” orders is much higher than the correlation among 10
“human” orders. This raises questions regarding the contribution of algorithms to the transmission of systemic risk. Overall, we make two observations. First, the empirical evidence does not seem consistent with the negative theoretical predictions regarding the consequences of HFT. Instead, on average, algorithmic and HF traders appear to increase liquidity and price discovery. But other empirical work raises concerns about the quality of liquidity, effects on volatility, and about disparities between traders’ response times that suggest a wealth transfer from slow to fast traders. We believe that these observations demand additional analysis of the broader issues related to algorithmic and HF trading. In this paper, we examine how algorithmic trading is related to market quality and contribute new large-sample, cross-country evidence to this literature.
3.
Data We obtain intraday quotes and trades from the Thomson Reuters Tick History (TRTH)
database. Our initial sample includes all non-U.S. common stocks covered in the database. We identify the primary listing market for each of these stocks and drop trading in stock that takes place on all other markets. This filter produces stocks trading on 40 primary equity exchanges in 37 countries. 2 We obtain data on daily returns, daily high and low prices, trading volume, security details and financial statement data from Datastream and WorldScope. TRTH, supplied by the Securities Industry Research Centre of Asia-Pacific (SIRCA), provides access to the data feeds from various stock and derivatives exchanges transmitted through the Reuters Integrated Data Network (IDN). The trades, which include odd lots, and quotes are time stamped to the millisecond and the system covers more than 5 million equity and derivatives
2
China has three exchanges (Hong Kong, Shenzhen, and Shanghai), Japan has two (Osaka and Tokyo), all other countries have exactly one exchange included in the sample.
11
instruments around the world. TRTH organizes data by the Reuters Instrument Code (RIC). Each RIC code is unique worldwide and is associated with a list of RIC characteristics such asset class (e.g. equity), market, currency denomination, the first and the last data date, and the ISIN and SEDOL where applicable. Each RIC is associated with a data history that describes its changes over time, including changes in currency denomination and company name. Each RIC-market combination is associated with exactly one ISIN. 3 Datastream identifies securities by DSCODE, which is unique to a security- trading venue combination. We retain only the trading location identified as “primary quote.” In Datastream, this refers to the primary listing location. We are interested in the primary trading location, but in all markets except Germany the two coincide. In Germany, we use XETRA rather than Frankfurt, the primary listing location, because XETRA handles roughly 90% of volume for most stocks. Each DSCODE is associated with a comprehensive list of static securities information including the ISIN. Together with the “primary quote” designation, each DSCODE is associated with exactly one ISIN. To merge the two data sources we proceed as follows. For each exchange, we obtain the high, low and last trade price for each RIC from TRTH. We find the corresponding trading venue on Datastream and identify the unadjusted daily price, market capitalization, and the adjustment factor (for dilution) for each primary quote DSCODE. Then we match each ISIN associated with a DSCODE with the corresponding ISIN associated with a RIC. The resulting sample consists of all common stocks in TRTH that have an ISIN assigned, are denominated in the primary-market home currency,
3
The RIC for equity has the structure of company code (often, but not always, the same as the local ticker) plus a non-common stock security class modifier called the brokerage character, and followed by “.” and the exchange code. The brokerage character varies by market and we obtain the brokerage character information from TR’s date sensitive market and securities reference system “Speedguide.”
12
and trade in the primary market. We validate the match by comparing the Datastream prices with TRTH prices on the first common data date after adjusting for currency differences 4. TRTH trade and quote data have qualifiers that contain market specific codes denoting the first trade of the day, auction trades, and irregular trades (such as off-market trades or option exercises). In computing intraday bid-ask spreads, effective spreads, returns, and related measures, we only use regular trades and quotes during the continuous trading period and exclude auction trades and irregular trades. Fong, Holden and Trzcinka (2011) provide further TRTH data validation against other international intraday data bases and find TRTH data to be highly accurate. Brockman, Chung and Perignon (2009) compile an intraday trade and quote database from Bloomberg over the 21 month period from October 1, 2002 to June 30, 2004 (455 trading days). The average difference in effective spreads between the two databases over the same period is 3 basis points, again speaking to the accuracy of TRTH. Official trading hours differ across exchanges and over time. We determine each exchange’s historical trading hour regime by examining the trade frequency across all stocks in the exchange at 5 minute intervals. We identify the opening and closing of regular trading from spikes and drops in trading activity across all stocks at each exchange. We cross-check our algorithm against the trading hour regime and the trading mechanism entries listed in Reuter’s Speedguide and the Handbook of World Stock, Derivative and Commodity Exchanges.
A. Variables Our objective in this analysis is to measure the effect of algorithmic trading on market quality. We use variables that describe several dimensions of market quality, focusing on liquidity, volatility,
4
TRTH prices are historical prices in the original currency. Datastream unadjusted prices are historical prices in the current currency unit, e.g. French stocks prior to 1999 were traded in French Franc. We convert Datastream prices to Euro equivalents.
13
and informational efficiency. We describe these variables in this section, along with our proxies for algorithmic trading activity. Execution costs We compute several standard measures of liquidity and execution costs. For each stock, we have the best quoted spread throughout the trading day. For a given time interval s, the relative quoted spread, standardized by the quote midpoint, is defined as RQSs = (Asks - Bids) / ((Asks + Bids)/2),
(1)
where Asks is the best ask quote and Bids is the best bid quote in that time interval. When aggregating over a trading day we use time-weighted averages of RQS. To take into account possible price improvement, potentially arising due to hidden liquidity, we compute the relative effective spread, standardized by the quote midpoint at the time of the trade. The RES on the k th trade is defined as RESk = 2Dk (ln(Pk) - ln(Mk)),
(2)
where Dk is an indicator variable that equals +1 if the k th trade is a buy and -1 if the k th trade is a sell, Pk is the price of the k th trade and Mk is the midpoint of the national consolidated best bid and offer (BBO) prevailing at the time of the k th trade. RES measures the total price impact of a trade. We decompose this price impact into a permanent (information-related) component, RPI, and a transitory component, the relative realized spread, RRS. We follow standard practice and base both components on the quote midpoint that prevails five minutes after the trade. RRS on the k th trade is defined as RRSk = 2Dk (ln(Pk) - ln(Mk+5)),
(3)
14
where M(k+5) is the midpoint five-minutes after the k th trade. RRS can be interpreted as the reward for providing liquidity. The permanent component, RPI, is defined as RPIk = 0.5 (RESk -RRSk) = 2Dk (ln(Mk+5) - ln(Mk)),
(4)
and measures the change in quote midpoints that is attributable to the information content of the trade. When aggregating over all trades during a particular day, we produce either trade or (localcurrency) volume-weighted averages of RES, RRS, and RPI. When aggregating across stocks, we equally weight trade-weighted daily averages and volume-weight volume-weighted daily averages. This produces two alternative measures for each market quality dimension that are either consistently equal-weighted or consistently volume-weighted. The latter may give a better picture of profit opportunities in the whole market, but the former may provide a better picture if the costs of doing the largest trades differ fundamentally from the costs of completing small trades. Our last liquidity measure is Amihud illiquidity measure, estimated as the absolute value of daily return divided by the contemporaneous daily dollar trading volume. A larger Amihud ratio indicates that a given volume moves prices by a larger magnitude, and implies lower liquidity. Volatility Our primary measure of volatility is the intra-day range between the highest and lowest prices of a day, standardized by the daily closing price. This measure is useful because it reflects intraday fluctuations in share prices that may trigger or result from algorithmic trading. In addition, we compute four different measures or realized volatility. As lower-frequency measures we employ the absolute value of daily returns, |R|, and return squared, R2. Similarly, we compute analogous measures for daily market-adjusted returns. Finally, we use the log of intra-day return variances computed for 10-minute and 30-minute returns. We aggregate within days (if applicable) and across stocks analogously to the execution cost measures above.
15
Informational efficiency Following Boehmer and Kelley (2010), we compute intraday measures of informational efficiency. For most of our analysis we rely on intraday measures of quote midpoint autocorrelation, which should be zero at all horizons if prices follow a random walk. Deviations from zero in either direction indicate partial predictability. We first estimate quote midpoint return autocorrelations for each stock-day, based on 30-minute intervals (results are qualitatively identical for 5, 10, 20, and 60minute return intervals) based on the results from Chordia, Roll, and Subrahmanyam (2005). We use |AR30| to denote the absolute value of this autocorrelation. Algorithmic / high frequency trading High frequency algorithmic activity is generally associated with fast order submissions and cancellations (see Hasbrouck and Saar, 2010). The proxy used by Hendershott, Jones, and Menkveld (2010) reflects this concept, and we follow them and use AT, the negative of trading volume in USD100 divided by the number of messages, as our proxy for algorithmic trading activity. It represents the negative of the dollar volume associated with each message (defined as either a trade or a quote update). An increase in this measure reflects an increase in algorithmic activity. Our measure of AT differs in an important way from the one used by HJM, who have access to order-level messages. For our world-wide sample, we only have access to a subset of these messages. We only observe each exchange’s best quotes and trades, rather than all order-related messages. This means that we cannot directly capture one important dimension of HFT activity, the high ratio of order (submission and cancel) messages to trades that is so characteristic of many popular HFT strategies. But the HFT strategies that are mentioned, for example, in the SEC 2010 concept release involve activity at the BBO, rather than behind it. Therefore, we believe that HFT activity in our BBO-trade data set is highly correlated with HFT activity in an order-trade data set. We address this issue explicitly by repeating HJM’s time-series and panel results using the U.S. portion of our sample constructed from the TAQ database. The time series of AT during the 16
overlapping period is very similar to the one presented in HJM. Our replication of the panel estimated using only NYSE activity yield qualitatively identical results, despite the differences between the two tests. 5 Given this result, we have little reason to expect our AT proxy to deliver substantially different results than the HJM proxy.
B. Descriptive statistics For inclusion in our final sample we impose a few additional data requirements. We drop Ireland, where fewer than 30 stocks are listed prior to 2008. We exclude stocks that have data for fewer than 21 trading days during the sample period. To illustrate the breadth of our sample, Table 1 lists the number of stocks for each market. For the average year, our sample includes about 12,800 firms and we have substantial variation across markets. Over the sample period, the number of listed firms increases by 140% and, in 2009, represents an aggregate market capitalization of USD 16.7 trillion. We first describe the time-series distribution of message traffic (the number of all quote changes and all trades), which is the main component of our algorithmic trading proxy. For each country, Table 2 lists the median number of messages in 2009 (the most recent year in our sample), the mean number of messages, and the change in both between 2001 and 2009. We make several important observations. First, message traffic grows steeply over time in all but three markets. This is consistent with AT/HFT playing an increasingly important role around the world. Second, the mean exceeds the median in all markets, and often by an enormous margin. This indicates that a few firms are responsible for a disproportionate chunk of the message traffic. For example, Shanghai has the most messages per stock-day in 2009, and the mean of 13,045 is about 30% larger than the median. At Euronext Amsterdam, however, the mean of 8717 is about 14 times larger than the median.
5
Our analysis differs from theirs in the following ways. First, our AT measures are based on BBO messages rather than all order-level messages. Second, our time series covers 2000-2009 rather than 2001-2006. Third, we use the full panel rather than only the period surrounding the implementation of the NYSE’s 2003 implementation of “autoquote.”
17
Third, growth from 2001 to 2009 differs substantially across markets. The increase in daily message traffic ranges from -47% in Athens to 3707% in Brussels. These two observations have important implications for our test design. They suggest that there is substantial variation of algorithmic trading intensity both across firms and over time. Our empirical approach will account for both. Especially useful is the cross-firm variation, because it allows us to conduct cross-sectional analysis of how AT affects market quality. In Figures 2 and 3, we present the average time-series development in our measures of liquidity, efficiency, and volatility. For each market-day, we first compute an equally weighted mean across firms, and then compute equally weighted means for each market-year. In the figures, we plot the annual time series of averages across markets. The quoted and effective spreads, TWQS and RES, in Figure 2 show similar patterns. For example, RES begins at 25bp in 2001 and declines to 14bp by 2007. Afterwards, it increases again to about 23bp, probably as a result of the financial crises around the world. For TWQS, the range is from 60bp in 2001 to 30bp in 2007, and then an increase to 52bp. When we decompose RES into its transient (RRS) and permanent (RPI) components, we again find very similar patterns. Both components decrease until 2007, and then increase again. We also observe that RRS exceeds RPI in every year by more than 50%, and the difference appears to grow over time. Figure 3 show the daily relative price range, a proxy for intraday volatility. It is almost flat from 2001 to 2006 around 0.025, and then begins to increase above 0.04 in 2008. This provides an interesting contrast to the liquidity measures that have a pronounced “U” shape over time. Finally, the lower graph in Figure 3 shows |AR|, the quote midpoint autocorrelation of returns measured over 10 and 30 minute intervals. Both measures decline over time, indicating an improvement of informational efficiency over time.
18
4.
Methodology A. Country-specific analysis Our first objective is to identify, for each of the 39 countries, the effects of algorithmic
trading intensity on market quality, represented by measures of liquidity, volatility, and informational efficiency. We will document this relationship in panel regressions that control for firm and day fixed effects. These fixed effects prevent us from interpreting systematic patterns in market quality across firms or secular patterns over time as the result of algorithmic trading. To estimate these relationships at the country level, we use two approaches. We employ a panel regression of the form
MQit = αi + γt + βATit + δ_Xit + εit,
(5)
where the αi are firm fixed effects, the γt are day fixed effects, AT is our proxy for algorithmic trading, and X is a vector of control variables. This vector includes share turnover, 1/price, market value of equity, the lagged dependent variable, and the daily price range standardized by the daily closing price (a proxy for volatility). All explanatory variables are lagged by one period to ensure that they are predetermined. Finally, all continuous variables are standardized in the cross-section to make coefficients comparable across countries. For inference within countries we use standard errors that are robust to cross-sectional and time-series heteroskedasticity and within-group autocorrelation based on Arellano and Bond (1991). Across-market inference is based on an equal-weighted means of the 39 market-specific coefficients and simple cross-sectional t-statistics also based on these 39
19
observations. This approach should be conservative, because all inference is based only on the 39 market-specific observations. 6 In some of our analysis, we differentiate observations according to cross-sectional firm characteristics including market cap, volatility, and share price. Unless stated otherwise, we determine each day, separately for each market, the lowest and highest tercile of firms based on the most recent 20 trading days. We assign “LOW” and “HIGH” dummies to firms in these terciles, respectively. We augment our regression model (5) with the two interactions between AT and each dummy. The interaction coefficients capture the potentially differential effect of AT on market quality in the LOW or HIGH terciles relative to the middle tercile. The total effect of AT for LOW firms is given by the sum of the coefficient on AT and the coefficient on AT*LOW. We interpret results for the HIGH dummy analogously.
B. Instruments for algorithmic trading Our second objective is to establish that AT and market quality are not spuriously related to each other. To identify an exogenous change in AT we rely on the first instance of “co-location” in each country. 7 Co-location refers to locating a trader’s order submission algorithm physically close to a trading center’s computer. This minimizes response times for data feeds from the exchange and also for order messages from the trader. The introduction of co-location happens at different dates across markets. Because co-location applies to all firms within a market simultaneously, we use a between estimator at the market level. Specifically, we compute the market-value weighted averages for all variables within each market, standardize the resulting time series within each
6
As a robustness check, we use a Fama-McBeth model. For each day, we estimate a cross-sectional regression analogously to equation (5) within each market, but without the time effects. For each market, we then compute the time-series average of each coefficient. Inferences within countries use Newey-West standard errors. Across-country inference uses cross-sectional t-statistics as in the main analysis. This approach produces qualitatively identical results that are not tabulated. 7 Other possible instruments include the introduction of direct market access for traders, DMA, or other updates to the trading protocol that imply a structural change in how AT / HFT is done.
20
market, and then perform a two-stage generalized instrumental variable estimation. In the first stage, we regress our AT proxy on time-fixed effects and a co-location dummy. 8 In the second stage, we estimate
MQct = αc + γt + βAT*ct + δ_Xct + εct,
(6)
where the αc are market fixed effects, the γt are day fixed effects, AT* is the vector of predicted values from the first-stage regression, and the other variables are market-specific weighted averages of the control variables as described above. For inference we use standard errors that are robust to cross-sectional and time-series heteroskedasticity and within-group autocorrelation based on Arellano and Bond (1991).
5.
Regression results A. Within market effects of algorithmic trading As described in section 4, we conduct a two-dimensional analysis within each market using
daily two-way fixed-effects regressions. We conservatively conduct global inferences, but to illustrate our estimation we present market-specific results in the appendix. The table shows the 39 countries, whether they are G7 or OECD members, and coefficient estimates and p-values for each market. The p-values are based on dynamic Arellano-Bond standard errors. For example, the effect of AT on RES is -0.01 for Xetra stocks. Since all variables are standardized each day, this means that a one-standard deviation increase in AT is associated with a 0.01 standard deviation decline in RES. But this effect is heterogeneous across markets. For example, the effect is 0.044 standard deviations in Shanghai, but only 0.002 in Singapore.
8
Adding the remaining explanatory variables to the first stage leaves inferences unchanged but results in less precise estimates.
21
Panel A of Table 3 presents a summary of the liquidity-related coefficients that we estimate for each market. We omit coefficients on firm fixed effects, daily fixed effects, and control variables. These are estimated as described in equation (5) but not tabulated. The mean effect of AT on RES is 0.013, which means that a one-standard deviation increase in AT implies a 0.013 standard deviation decrease in relative effective spreads. The associated t-statistic is -7.4, using the cross-sectional standard error across the 39 markets. AT reduces (improves) RES in 92% of the markets. 9 We find consistent results for each of the liquidity measures – an increase in AT is associated with decreases in RQS, RES, and the Amihud measure. Finally, we document that more AT reduces both the information content of trades (RPI) and the transitory price impact (RRS). This suggests that AT does not necessarily increase the intensity of informed trading. These results suggest that, on average, greater AT intensity is associated with improved liquidity. In Panels B, C, and D, we assess whether the effect of AT varies with market cap, share price, or the 20-day return volatility. For each measure, we contrast the effect for the lowest tercile (“LOW”) with the effect of the largest tercile (“HIGH”). Within each market, we first determine the lowest and highest terciles and accordingly define the LOW and HIGH dummies as described in the data section. In Table 3, we report the mean coefficient of AT, which now represents the effect of AT on liquidity for the middle tercile of the sort variable, and the two interactions. We also report the total effects for LOW and HIGH stocks. In Panel B, we sort by market cap so the LOW and HIGH dummies represent firm size within each market. Looking at Amihud illiquidity, the largest firms have a marginal coefficient of -0.006 that is significant at the 5% level. This implies that the largest firms experience a reduction in
9
While not tabulated, the corresponding coefficient averages using the Fama-MacBeth approach yield identical inferences regarding the coefficient signs. We note, however, that the magnitude of the AT effect is much larger for the FMB models than for the panel estimation that we present in Table 3. For example, the mean effect of AT on RES is -0.082 standard deviations with FMB, which is six times larger than the estimate from the panel regressions. The differences arise from different treatment of time effects. FMB allows slope coefficients to vary across days, while the two-way panel nets out an aggregate time trend.
22
Amihud that is 0.006 standard deviations greater than the reduction of 0.007 standard deviations experienced by firms in the middle terciles. The total effect of AT on large firms is the sum of these coefficients, -0.013, presented in the last row of Panel B. With a t-statistic of -7.3, the total effect is statistically significant at the 1% level. The results for the other liquidity measures indicate that the AT effect in the large terciles are not significantly different from that in the middle terciles, but AT in the largest firms is still associated with a significant liquidity improvement. This is quite different for small firms: the marginal effect of AT is positive, which means that AT tends to widen spreads in small firms relative to the effect it has for middle-tercile firms. This suggests that AT provides less liquidity improvement in small firms than in other firms. In fact, the total effect of AT is positive for small firms, which means that greater AT intensity in small firms increases transactions costs and therefore reduces liquidity in these firms. Panel C presents a similar cross-sectional analysis according to share price. The marginal effects (the coefficients on the interactive terms) are mostly not significant, which indicates that neither the LOW nor the HIGH terciles are significantly different from the middle tercile. We note, however, that the effect of AT on liquidity partly vanishes for low-priced stocks – AT does not reduce RQS or RES, where the total effects of AT have t-statistics of -0.7 and -0.6, respectively, and significantly worsens Amihud illiquidity for low-priced stocks. Panel D looks at the effects of AT on liquidity sorted by each stock’s 20-day return volatility. Similar to the price sorts, we find the liquidity-enhancing effect of AT vanishes for high-volatility stocks. Here, the marginal effect of AT is significantly more positive for high-volatility stocks, implying that they experience a significant lower liquidity benefit when AT increases. Table 4 looks at the effects of AT on informational efficiency. On average, we show significantly negative coefficients in Panel A. This suggests that more AT reduces |AR|, implying greater improvement in informational efficiency. The results are quite similar for both measures of autocorrelation. 23
Panels B, C, and D show that the AT effects on efficiency vary in the cross section. Generally, AT has an efficiency-enhancing effect in all terciles, independent of the sort variable. One exception is the tercile of small firms, which experience no efficiency improvement when AT increases. Share price has no statistically significant effect on the AT-efficiency relation. High volatility firms, however, experience significantly greater efficiency improvement than low-volatility firms when AT becomes more intense. Table 5 summarizes the AT coefficients for regressions that explain short-run volatility. AT increases volatility and the results are uniform across volatility proxies, whether we look at intraday realized volatility, daily realized volatility, or the standardized intraday price range. Moreover, the coefficients are consistently positive across most markets. The percentage of positive coefficients ranges from 67% (variance of 10-minute returns) to 79% or more in all other models. This finding is largely consistent with Brogaard (2011a), who also finds that more HFT can result in greater volatility in certain cases. Given that AT improves informational efficiency, it is conceivable that the elevated volatility associated with more AT reflects faster price adjustments to new information. In this case, the higher volatility could be desirable, because it reflects new information rather than noise. It is also possible that narrower spreads, also associated with greater AT, are associated with smaller quoted sizes so that subsequent trades result in trade prints that experience lower execution costs, but result in greater price fluctuations. This could also be desirable, because traders value narrow spreads. To control for both possibilities, we add lagged |AR30|, our main efficiency measure, and lagged RES, a measure of liquidity, to model (5) whenever we estimate effects on volatility. Controlling for potential sources of “good” volatility in this way does not change our inference. Therefore, it is difficult to attribute the elevated volatility that is associated with AT to faster reflection of news or to tighter spreads.
24
The remaining panels of Table 5 reveal that greater AT intensity is associated with greater volatility in each firm tercile, whether sorted on market, price, or volatility. The effect is significantly stronger, however, for stocks that are small, low priced, or have volatile returns. Taken together, our results indicate that AT improves informational efficiency for most stocks, but liquidity improvements are limited to the two largest terciles of the firm-size and volatility distributions. In contrast, the negative effects of AT on intraday and realized volatility are significant for all firms independent of the market cap, price, or volatility category they fall into.
B. Difficult market-making days Market making strategies account for an important fraction of algorithmic and HF strategies (Kirilenko et al., 2011), although it is unclear to what extent their importance varies over time. Because the traders who supply this liquidity are not subject to the same affirmative obligations that force “traditional” market makers to provide liquidity at all times, regulators are concerned that liquidity provided by these strategies is not stable over time (see the 2010 SEC Concept Release on Equity Market Structure, http://www.sec.gov/rules/concept/2010/34-61358.pdf). In the context of our analysis, one might ask how our result change on days when making markets is costly. To identify such difficult market-making days, we rely on a simple proxy that is based on daily returns. Market making is easiest when prices do not change and buyers are equally likely to arrive as sellers. Market making is more difficult when a trading day is one sided. For example, if buyers are more aggressive than sellers on a particular day, prices are likely to increase and market makers are likely to end the day with a short inventory (or realize unusually large losses by covering at the prevailing high prices). It would be yet more difficult if the positive return/positive order imbalance day was followed by another day with positive returns and order imbalances. On this second day, market makers face additional losses and would be more reluctant to provide liquidity, especially for buyers.
25
Based on this argument, we create a proxy for difficult market-making days in the following way. For each individual stock, we identify all days where the daily return has the same sign as the previous day’s return. In addition, we require the two-day cumulative return to exceed the 20-day historical mean by at least one standard deviation. This eliminates smaller return episodes that probably do not have much effect on liquidity supply. We create dummy variable that is one on the second day of these episodes, because we expect market making to be unusually difficult on this day. To estimate how AT affects market quality on difficult days, we proceed analogously to our cross-sectional analyses above. We expand model (5) by interacting HARD dummy with AT. The results in Table 6 show the mean coefficients for liquidity, efficiency, and volatility effects in separate panels. Panel A shows that AT provides significantly lower liquidity benefits on difficult days. AT’s effect on quoted spreads is not different than on other days, but effective spreads and the Amihud measure improve significantly less on difficult days. The total effect still implies an improvement in terms of lower RES, but the total effect of AT during difficult days on Amihud is not statistically different from zero. We also see that AT is associated with significantly more information content, as represented by RPI, on difficult market-making days. This suggests that AT on difficult days leads to increased price impacts, which would be consistent with costly market making on such days, and it could explain why AT has less effect on liquidity on these days. This is also consistent with the negative interaction coefficients in Panel B, which imply that AT improves efficiency more on difficult days. If AT is associated with more information on difficult days, then it makes sense that prices become more informative. Panel C presents the effect of AT on volatility on difficult days. The interaction terms have significantly positive coefficients for all measures except the 10 and 30 minute return variances. For the other five volatility variables, AT increases volatility substantially more than on regular trading days. Taken together, these results show that the effect of AT on market quality are different on days when market making is difficult: AT provides less liquidity, brings more informed order flow 26
that improves efficiency more, and elevates volatility by more. This indicates that AT involves different strategies on difficult days than on other days, and these differences imply significant changes in how AT affects market quality.
C. Instrumental variable estimation Co-location facilitates algorithmic trading without directly affecting market quality. We use a time-varying dummy within each market to indicate, if ever, the availability of co-location. We use this dummy as an instrument for algorithmic trading in a between-markets panel as described in section 4.B. The results in Table 7 are uniformly consistent with the within-market analysis: algorithmic trading improves market quality and efficiency, but elevates short-run volatility. Each of the liquidity measures in Panel A declines as algorithmic trading increases, implying narrower spreads and smaller Amihud price impacts. The efficiency measures in Panel B also decline, implying greater efficiency, although AT1 does not affect the 30-minute autocorrelation significantly. Panel C shows that AT increases each of the volatility measures, although squared returns are not significantly affected. Overall, these estimates mirror the within-market ones. Importantly, despite the reduced power, most estimates remain significant and suggest that algorithmic trading causally improves liquidity and efficiency, but worsens volatility.
6.
Conclusions We use intraday message traffic in 39 major stock markets to analyze the effects of
algorithmic trading on these markets. We cover a broad cross-section of stocks in these trading centers and create measures of algorithmic trading intensity, a precondition for high frequency trading, and several market-quality measures including liquidity, efficiency, and volatility metrics. We estimate the relationship between algorithmic trading and market quality and explore whether cross-sectional factors that are known to affect trading are related to the effect that AT has on market quality. 27
Quite consistently across all markets, algorithmic trading improves liquidity and efficiency, but it worsens volatility. The effect on volatility cannot be attributed to more efficient prices that adjust faster to new information, and hence create higher volatility. We use co-location events, which represent exogenous shocks to AT, as instruments and show that these effects arise because AT causally affects market quality. We then document that the average effects conceal important cross-sectional differences in the effect that AT has on markets. AT generally enhances efficiency, but AT can systematically have negative effects on liquidity. For example, AT reduces liquidity in small stocks. Moreover, AT improves liquidity less in low-priced stocks or high-volatility stocks than in high-priced or lowvolatility stocks. The effects of AT on market quality vary over time. We create a proxy for days on which market making is difficult. Compared to regular days, AT improves efficiency more, but worsens volatility more than on regular trading days. Most importantly, AT improves liquidity significantly less on difficult market making days, a time more it is presumably most needed. Overall, our results support prior results that attribute liquidity-enhancing and efficiencyenhancing effects to algorithmic and HF trading. We complement these results with evidence that AT’s liquidity provision does not apply to all firms and not on days when market making is unusually costly. If market making strategies dominate algorithmic trading, then our results suggest that especially for the smallest tercile of firms and on days when market making is costly, algorithmic traders appear to trade according to strategies that do not primarily focus on market making. Perhaps equally importantly, we document that AT systematically increases volatility, which imposes costs on other market participants. In evaluating the costs and benefits of algorithmic trading in markets dominated by high frequency trading, market observers should take into account these effects.
28
References Ang, A. and K. Kjaer, 2011, Investing for the Long Run, Working paper, Columbia. Amihud, Y., 2002. Illiquidity and Stock Returns: Cross-Section and Time-Series Effects, Journal of Financial Markets 5, 31-56. Arellano, M. and S. R. Bond, 1991, Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations, Review of Economic Studies 58, 277–297. Barnes, A. W., 1911, History of the Philadelphia Stock Exchange, banks and banking interests, Cornelius Baker. Biais, B. and P. Woolley, High frequency trading, Working paper, IDEI Toulouse. Biais, B., T. Foucault, and S. Moinas, 2010, Equilibrium algorithmic trading, Working paper, IDEI Toulouse. Boehmer, E., and E. Kelley. 2009. Institutional Investors and the Informational Efficiency of Prices. Review of Financial Studies 22, 3563-3594. Brockman, P., D. Y. Chung, and C. Perignon, 2009, Commonality in Liquidity: A Global Perspective, Journal of Finance and Quantitative Analysis 44, 851-882. Brogaard, J., 2010, High frequency trading and its impact on market quality, Working paper, University of Washington. Brogaard, J., 2011a, High frequency trading and volatility, Working paper, University of Washington. Brogaard, J., 2011b, The activity of high frequency traders, Working paper, University of Washington. Brunnermeier, M. and L. H. Pedersen, 2005, Predatory trading, Journal of Finance 60, 1825-1863.
29
Cartea, A. and J. Penalva, 2011, Where is the Value in High Frequency Trading? Working paper, Universidad Carlos III de Madrid. Chaboud, A., B. Chiquoine, E. Hjalmarsson, and C. Vega, 2009, Rise of the machines: Algorithmic trading in the foreign exchange market, Working paper, Federal Reserve Board. Chordia, T., R. Roll, and A. Subrahmanyam. 2005. Evidence on the Speed of Convergence to Market Efficiency. Journal of Financial Economics 76: 271-292. Dichev, I., K. Huang, and D. Zhou, 2011, The dark side of trading, Working paper, Emory. Egginton, J., B. F. VanNess, and R. A. VanNess, 2011, Quote stuffing, Working paper, University of Mississippi. Fong, K. Y. L., C. W. Holden, and C. Trzcinka, 2011, What are the best liquidity proxies for global research? Working paper, Indiana University. Hasbrouck, J. and G. Saar, 2009, Technology and Liquidity Provision: The Blurring of Traditional Definitions, Journal of Financial Markets 12, 143-172. Hasbrouck, J. and G. Saar, 2011, Low-latency trading, Working paper, NYU. Hendershott, T. and R. Riordan, 2009, Algorithmic trading and information, Working paper, UC Berkeley. Hendershott, T. and R. Riordan, 2011, High frequency trading and price discovery, Working paper, UC Berkeley. Hendershott, T., C. Jones, and A. Menkveld, 2010, Does algorithmic trading improve liquidity? Journal of Finance 66.
30
Hoffmann, P., 2010, Algorithmic Trading in a Dynamic Limit Order Market, Working paper, Universitat Pompeu Fabra. Jarrow, R. and P. Protter, 2011, A Dysfunctional Role of High Frequency Trading in Electronic Markets, Working paper, Cornell. Jovanovic, B. and A. Menkveld, 2011, Middlemen in limit-order markets, Working paper, NYU. Khandani, A. and A. Lo, 2008, What happened to the quants in August 2007? Journal of Financial Markets. Kirilenko, A., A. S. Kyle, M. Samadi, and T. Tuzun, 2011, The Flash Crash: The Impact of High Frequency Trading on an Electronic Market, Working paper, University of Maryland. McInish, T. and J. Upson, 2011, Strategic liquidity supply in a market with fast and slow traders, Working paper, University of Memphis. Menkveld, A., 2010, High frequency trading and the new-market makers, Working paper, Free University of Amsterdam. Philippon, T. and E. Pagnotta, 2011, Competing on speed, Working paper, NYU. SEC, 2010, Concept Release on Equity Market Structure, Securities and Exchange Commission Release No. 34-61358.
31
Table 1. Number of stocks listed on sample markets
2001 26 19 561 66 58 36 33 280 62
2002 32 48 603 69 59 35 36 285 62
2003 49 63 642 74 65 36 80 266 63
2004 53 77 776 76 73 36 114 263 64
322 75 521 195 135 174
335 78 578 199 144 188
528 233 50 449 47 156 76 110
331 79 636 201 153 198 659 612 262 48 556 65 175 82 102 67 700 461 269 158 197 26 562 160 279 1,888 311 34 49
348 94 678 207 161 200 655 696 324 57 613
2005 55 80 907 77 80 37 126 294 70 13 432 105 708 219 172 196 645 751 418 60 661 112 215 104 130 80 782 492 361 181 214 144 629 232 341 2,116 431 39 99
2006 57 87 1,055 81 92 37 139 357 76 29 455 114 766 231 186 204 660 781 534 66 696 121 236 124 140 81 701 456 407 187 229 157 637 275 373 2,167 519 44 117
2007 60 91 1,274 86 145 44 147 434 80 30 455 124 810 231 194 209 670 795 589 72 720 126 235 132 144 81 751 458 436 191 243 159 632 296 378 2,161 573 49 119
2008 59 118 1,270 121 166 103 177 437 98 41 609 133 933 294 269 212 662 824 971 74 959 119 227 218 156 81 804 654 516 290 209 189 679 365 396 2,326 830 58 222
2009 57 115 1,240 119 166 101 166 400 91 42 583 129 1,052 294 257 221 667 804 865 81 961 113 221 203 156 81 816 714 510 283 213 185 698 346 403 2,296 822 56 229
Argentina Athens Australian Barcelona Sao paulo Brussels Copenhagen Dt Boerse Xetra Euronext Amsterdam Euronext Lisbon Euronext Paris Helsinki Hong Kong Istanbul Milan Jakarta Korea/ Daehan Kuala lumpur London Mexican Mumbai New Zealand Osaka Oslo Philippines Santiago Shanghai Shenzhen Singapore Johannesburg Stockholm Swiss Exchange Taiwan Tel-aviv Thailand Tokyo Toronto Wiener Borse Warsaw
567 503 222 159 184
571 241 52 513 59 162 80 100 65 631 462 245 157 195
417 149 236 1,699 277 36 44
498 148 254 1,794 295 35 47
Average Total number of stocks
242 253 274 308 321 342 361 423 420 8,729 9,377 10,681 11,718 12,833 13,700 14,451 16,913 16,797
194 90 114 74 773 482 321 168 204 131 600 193 309 2,009 369 36 64
32
Change % change 2001-2009 2001-2009 31 119% 96 503% 679 121% 53 81% 108 185% 65 182% 133 397% 120 43% 29 47% 42 316% 261 81% 54 72% 531 102% 99 51% 122 90% 47 27% 667 101% 276 52% 632 272% 31 61% 512 114% 65 137% 64 41% 127 167% 45 41% 81 126% 249 44% 211 42% 288 130% 124 78% 29 16% 185 711% 281 67% 197 132% 167 71% 597 35% 545 197% 20 56% 184 416% 202 8,067
140%
Table 2. Number of quote and trade messages per stock and year We count all intraday messages that represent trades or changes in the best quotes for each stock on all 39 markets in our sample. Then we compute equally weighted averages for each market and year. Median number of messages per stock in 2009 Shanghai Shenzhen Mumbai Korea/ Daehan stock ex Barcelona Taiwan Euronext Amsterdam Mexican stk ex (BMV) Tokyo stk ex Euronext Lisbon Milan Winener Borse AG Bovespa/SaoPaulo Brussels London Stock Exchange Hong Kong Istanbul Toronto Helsinki Thailand Stock Exchange Irish Stock Exchange Stockholmsborsen Jakarta SWX Swiss Exchange Warsaw Euronext Paris Oslo Bors Athens Tel-Aviv Singapore Exchange Kuala Lumpur Copenhagen Deutsche Boerse AG Xetra Buenos Aires South Africa Johannesburg Philippine Stock Exchange Australian Stock Exchange Santiago Osaka New Zealand
9,230 3,404 993 970 666 659 631 584 376 299 272 241 179 166 160 112 112 102 76 72 70 52 51 50 47 43 42 37 35 32 29 29 29 27 26 23 22 16 16 14
Mean number of Change of annual messages per medians 2001- Change of annual stock in 2009 2009 means 2001-2009 13,045 3,441 2,449 2,857 3,396 672 8,717 2,161 1,064 1,638 3,434 839 2,180 2,958 5,309 634 221 4,535 1,892 424 597 745 611 377 168 4,455 1,012 195 372 316 171 674 2,527 86 1,289 107 1,290 83 102 47
33
1702% 1177% 2051% 259% 205% 116% 884% 3299% 316% 657% 121% 760% 1834% 1090% 452% 110% -23% 281% 157% 106% 375% -30% 127% 144% 881% 93% 134% -87% 436% -5% -22% 134% -31% 67% 46% 246% 47% 260% 91% 63%
2366% 1135% 571% 126% 175% 43% 842% 1591% 273% 439% 484% 946% 3354% 3707% 1407% 163% -18% 1708% 424% 130% 1811% 138% 329% 533% 2674% 631% 801% -47% 544% 80% 69% 1721% 1611% -4% 1025% 187% 1108% 276% 392% 102%
Table 3. The relationship between algorithmic trading and market quality Our data cover 39 markets from 2001-2009. We first estimate, for each market, a two-way panel regression with firm and day fixed effects. We show the mean coefficients across the 39 markets, the associated t-statistic using the cross-market standard deviation, and in Panel A additionally the percentage of positive market-specific coefficients. Market quality measures include quoted spread (RQS), relative effective spread (RES), permanent price impact (RPI), and temporary price impact (RRS), and Amihud. AT is the negative of dollar trading volume ($100) per message. For message counts, we include all inside quote and trade messages. Control variables include daily share turnover, price range, 1/price, log market cap, and the first lag of the dependent variable, all measured at t-1. All continuous variables are standardized every day to have a mean of zero and a standard deviation of one within each exchange. In Panels B, C, and D we interact AT with two dummy variables, LOW and HIGH. In Panel B, the dummies indicate the smallest and largest market cap tercile based on moving average from the past 20 trading days, respectively. In Panel C, the dummies indicate the smallest and largest share price tercile based on moving average from the past 20 trading days, in Panel D the dummies indicate the smallest and largest volatility tercile based on the standard deviation of the 20 most recent daily returns. RRS
Amihud
-0.002 -0.7 54%
-0.018 -9.2 5%
-0.009 -6.5 15%
Panel B. The effect of AT for the smallest and largest market cap terciles Mean coefficient on AT -0.013 -0.013 Cross-sectional t-stat -4.6 -3.4 Mean coefficient on AT*LOW 0.038 0.040 Cross-sectional t-stat 4.7 4.7 Mean coefficient on AT*HIGH -0.001 -0.001 Cross-sectional t-stat -0.5 -0.4 Mean total effect for LOW 0.024 0.026 Cross-sectional t-stat 2.6 2.6 Mean total effect for HIGH -0.015 -0.015 Cross-sectional t-stat -8.4 -8.0
-0.001 -0.1 0.032 3.4 -0.003 -0.6 0.031 2.4 -0.004 -1.6
-0.022 -4.7 0.032 4.3 0.002 0.5 0.010 1.0 -0.019 -8.3
-0.007 -3.1 0.069 7.5 -0.006 -2.0 0.061 6.3 -0.013 -7.3
Panel C. The effect of AT for the smallest and largest price terciles Mean coefficient on AT -0.013 -0.013 Cross-sectional t-stat -5.9 -4.6 Mean coefficient on AT*LOW 0.009 0.009 Cross-sectional t-stat 1.5 1.4 Mean coefficient on AT*HIGH -0.001 -0.001 Cross-sectional t-stat -0.5 -0.2 Mean total effect for LOW -0.004 -0.004 -0.7 -0.6 Cross-sectional t-stat Mean total effect for HIGH -0.014 -0.013 Cross-sectional t-stat -8.3 -7.2
0.000 0.0 0.009 1.9 -0.003 -1.3 0.009 1.3 -0.003 -1.1
-0.020 -5.9 0.001 0.2 0.002 0.9 -0.019 -2.5 -0.017 -8.5
-0.008 -3.6 0.024 3.7 -0.003 -1.5 0.015 2.8 -0.012 -7.8
Panel D. The effect of AT for the smallest and largest volatility terciles Mean coefficient on AT -0.016 -0.016 Cross-sectional t-stat -8.7 -7.6 Mean coefficient on AT*LOW 0.003 0.003 Cross-sectional t-stat 2.5 2.3 Mean coefficient on AT*HIGH 0.018 0.020 Cross-sectional t-stat 5.6 6.0 Mean total effect for LOW -0.013 -0.013 Cross-sectional t-stat -8.3 -8.1 Mean total effect for HIGH 0.002 0.004 Cross-sectional t-stat 0.5 1.0
-0.004 -1.4 0.000 -0.1 0.026 6.6 -0.004 -1.7 0.022 3.9
-0.020 -8.6 0.004 2.9 0.010 3.0 -0.017 -9.0 -0.010 -2.3
-0.010 -5.5 0.003 2.5 0.010 2.8 -0.007 -3.0 0.000 0.1
RQS
RES
RPI
Panel A. Aggregate of market-specific, two-way panel regressions with firm and day fixed effects Mean coefficient on AT Cross-sectional t-stat Percent positive
-0.013 -9.1 5%
-0.013 -7.4 8%
34
Table 4. The relationship between algorithmic trading and informational efficiency Our data cover 39 markets from 2001-2009. We first estimate, for each market, a two-way panel regression with firm and day fixed effects. We show the mean coefficients across the 39 markets, the associated t-statistic using the cross-market standard deviation, and in Panel A additionally the percentage of positive market-specific coefficients. Efficiency measures are daily observations of the absolute value of intraday autocorrelations|AR##|, measured for quote-midpoint returns over 10 and 30 minute periods. AT is the negative of dollar trading volume ($100) per message. For message counts, we include all inside quote and trade messages. Control variables include daily share turnover, price range, 1/price, log market cap, and the first lag of the dependent variable, all measured at t-1. All continuous variables are standardized every day to have a mean of zero and a standard deviation of one within each exchange. In Panels B, C, and D we interact AT with two dummy variables, LOW and HIGH. In Panel B, the dummies indicate the smallest and largest market cap tercile based on moving average from the past 20 trading days, respectively. In Panel C, the dummies indicate the smallest and largest share price tercile based on moving average from the past 20 trading days, and in Panel D the dummies indicate the smallest and largest volatility tercile based on the standard deviation of the 20 most recent daily returns. |AR10|
|AR30|
Panel A. Aggregate of market-specific, two-way panel regressions with firm and day fixed effects Mean coefficient on AT Cross-sectional t-stat Percent positive
-0.017 -6.1 5%
-0.006 -4.5 23%
Panel B. The effect of AT for the smallest and largest market cap terciles Mean coefficient on AT Cross-sectional t-stat Mean coefficient on AT*LOW Cross-sectional t-stat Mean coefficient on AT*HIGH Cross-sectional t-stat Mean total effect for LOW Cross-sectional t-stat Mean total effect for HIGH Cross-sectional t-stat
-0.015 -2.9 0.004 0.7 -0.001 -0.3 -0.011 -1.9 -0.017 -6.0
-0.002 -0.6 0.010 2.3 -0.004 -1.2 0.008 1.4 -0.006 -4.4
Panel C. The effect of AT for the smallest and largest price terciles Mean coefficient on AT Cross-sectional t-stat Mean coefficient on AT*LOW Cross-sectional t-stat Mean coefficient on AT*HIGH Cross-sectional t-stat Mean total effect for LOW Cross-sectional t-stat Mean total effect for HIGH Cross-sectional t-stat
-0.016 -4.0 0.005 1.5 0.001 0.2 -0.011 -2.3 -0.016 -6.0
-0.004 -1.8 0.005 1.5 -0.002 -0.9 0.001 0.2 -0.006 -4.2
Panel D. The effect of AT for the smallest and largest volatility terciles Mean coefficient on AT Cross-sectional t-stat Mean coefficient on AT*LOW Cross-sectional t-stat Mean coefficient on AT*HIGH Cross-sectional t-stat Mean total effect for LOW Cross-sectional t-stat Mean total effect for HIGH Cross-sectional t-stat
-0.014 -5.0 0.000 -0.2 -0.012 -5.7 -0.014 -5.0 -0.026 -7.2
-0.005 -3.5 0.001 0.5 -0.004 -1.7 -0.005 -3.1 -0.009 -3.6
35
Table 5. The relationship between algorithmic trading and short-term volatility Our data cover 39 markets 2001-2009. We first estimate, for each market, a two-way panel regression with firm and day fixed effects. We show the mean coefficients across the 39 markets, the associated t-statistic using the cross-market standard deviation, and in Panel A additionally the percentage of positive market-specific coefficients. Volatility measures include |Ret|, |MktadjRet|, Ret^2, MktadjRet^2, the daily price range standardized by the daily closing price, and Ln(Ret##_Var), the log of the daily averages of the variances of 10-minute and 30-minute quote midpoint returns, respectively. AT is the negative of dollar trading volume ($100) per message. For message counts, we include all inside quote and trade messages. Control variables include daily share turnover, 1/price, log market cap, RES, |AR30|, and the first lag of the dependent variable, all measured at t-1. |AR30| is the absolute value of intraday autocorrelations measured for quotemidpoint returns over 30-minute periods. All continuous variables are standardized every day to have a mean of zero and a standard deviation of one within each exchange. In Panels B, C, and D we interact AT with two dummy variables, LOW and HIGH. In Panel B, the dummies indicate the smallest and largest market cap tercile based on moving average from the past 20 trading days, respectively. In Panel C, the dummies indicate the smallest and largest share price tercile based on moving average from the past 20 trading days, and in Panel D the dummies indicate the smallest and largest volatility tercile based on the standard deviation of the 20 most recent daily returns. |ret|
|mktadjRet|
Ret^2
MktadjRet^ Ln(Ret10_Va Ln(Ret30_Va 2 PriceRange r) r)
Panel A. Aggregate of market-specific, two-way panel regressions with firm and day fixed effects Mean coefficient on AT Cross-sectional t-stat Percent positive
0.033 7.4 87%
0.028 5.9 82%
0.025 6.6 85%
0.022 5.8 79%
0.045 7.9 87%
0.017 3.4 67%
0.029 4.9 79%
Panel B. The effect of AT for the smallest and largest market cap terciles Mean coefficient on AT 0.034 0.024 Cross-sectional t-stat 6.1 3.5 Mean coefficient on AT*LOW 0.030 0.029 Cross-sectional t-stat 3.3 3.2 Mean coefficient on AT*HIGH -0.001 0.002 Cross-sectional t-stat -0.4 0.5 Mean total effect for LOW 0.064 0.054 Cross-sectional t-stat 5.2 3.9 Mean total effect for HIGH 0.032 0.026 Cross-sectional t-stat 6.9 5.7
0.024 5.0 0.027 2.6 0.000 -0.1 0.050 4.1 0.023 6.3
0.018 3.2 0.025 2.4 0.003 0.9 0.042 3.2 0.021 5.6
0.049 5.5 0.034 4.0 -0.005 -0.9 0.083 5.9 0.043 7.6
0.023 2.7 0.025 2.6 -0.005 -0.7 0.048 3.3 0.018 3.5
0.034 3.9 0.029 2.7 -0.005 -0.7 0.063 4.2 0.029 4.8
Panel C. The effect of AT for the smallest and largest price terciles Mean coefficient on AT 0.034 0.027 Cross-sectional t-stat 6.4 4.6 Mean coefficient on AT*LOW 0.020 0.019 Cross-sectional t-stat 4.3 3.7 Mean coefficient on AT*HIGH -0.004 -0.002 Cross-sectional t-stat -1.2 -0.6 Mean total effect for LOW 0.054 0.046 Cross-sectional t-stat 6.4 4.9 Mean total effect for HIGH 0.030 0.025 Cross-sectional t-stat 7.3 5.9
0.025 5.8 0.017 4.0 -0.003 -1.3 0.042 6.2 0.022 6.5
0.021 4.8 0.015 3.5 -0.002 -0.8 0.036 5.2 0.019 5.7
0.044 6.1 0.020 3.3 -0.003 -0.6 0.064 5.5 0.042 8.1
0.015 2.5 0.025 3.5 0.002 0.3 0.040 3.8 0.017 3.4
0.028 4.1 0.026 4.1 0.000 -0.1 0.054 5.0 0.028 4.8
Panel D. The effect of AT for the smallest and largest volatility terciles Mean coefficient on AT 0.026 0.021 Cross-sectional t-stat 5.4 4.2 Mean coefficient on AT*LOW 0.002 0.000 Cross-sectional t-stat 0.9 0.0 Mean coefficient on AT*HIGH 0.042 0.042 Cross-sectional t-stat 6.5 7.0 Mean total effect for LOW 0.028 0.021 Cross-sectional t-stat 5.9 4.2 Mean total effect for HIGH 0.068 0.063 Cross-sectional t-stat 9.0 8.6
0.020 5.2 -0.001 -0.6 0.031 7.2 0.019 5.1 0.051 9.5
0.018 4.5 -0.002 -1.5 0.032 8.0 0.015 4.2 0.050 9.2
0.036 6.1 0.005 1.7 0.048 6.3 0.040 6.5 0.083 8.7
0.005 1.2 0.017 5.1 0.021 4.4 0.023 3.7 0.026 4.0
0.015 2.7 0.018 4.5 0.032 5.4 0.033 4.6 0.046 5.8
36
Table 6. The effect of algorithmic trading when market making is difficult Our data cover 39 markets 2001-2009. We first estimate, for each market, a two-way panel regression with firm and day fixed effects. We show the mean coefficients across the 39 markets, and the associated t-statistic using the cross-market standard deviation. Market quality measures include quoted spread (RQS), relative effective spread(RES), permanent price impact (RPI), and temporary price impact(RRS), and Amihud. Efficiency measures include daily observations of the absolute value of intraday autocorrelations|AR|, measured for quote-midpoint returns over 10 and 30 minute periods. Volatility measures include |Ret|, |MktadjRet|, Ret^2, MktadjRet^2, the daily price range standardized by the daily closing price, and Ln(Ret##_Var), the log of the daily averages of the variances of 10-minute and 30-minute quote midpoint returns, respectively. AT is the negative of dollar trading volume ($100) per message. For message counts, we include all inside quote and trade messages. Control variables include daily share turnover, Price range, 1/price, log market cap, and the first lag of the dependent variable, all measured at t-1. Regressions where the dependent variable is volatility do not include price range, but add RES, |AR30|. |AR30| is the absolute value of intraday autocorrelations measured for quote-midpoint returns over 30-minute periods. All continuous variables are standardized every day to have a mean of zero and a standard deviation of one within each exchange. We create an variable HARD that indicates days on which market making is difficult for a particular stock. Using daily returns, the HARD dummy equals one if a daily return has the same sign as the return on the previous day, and the absolute value of the 2-day return is at least one standard deviation larger than the 20-day trailing average of returns for that stock . We interact AT with HARD dummy in the regression. Panel A. Difficult market-making days and the effect of AT on liquidity
Mean coefficient on AT Cross-sectional t-stat Mean coefficient on AT*HARD Cross-sectional t-stat Mean total effect on HARD days Cross-sectional t-stat
RQS -0.013 -8.5 0.001 1.3 -0.012 -6.8
RES -0.013 -7.1 0.003 2.7 -0.010 -4.8
RPI -0.005 -2.0 0.014 5.5 0.009 2.3
RRS -0.015 -7.8 -0.007 -4.2 -0.023 -8.3
Amihud -0.011 -7.7 0.009 6.5 -0.002 -1.4
Panel B. Difficult market-making days and the effect of AT on informational efficiency
Mean coefficient on AT Cross-sectional t-stat Mean coefficient on AT*HARD Cross-sectional t-stat Mean total effect on HARD days Cross-sectional t-stat
|AR10| -0.015 -5.9 -0.004 -2.7 -0.019 -6.2
|AR30| -0.005 -3.5 -0.003 -2.6 -0.008 -4.9
Panel C. Difficult market-making days and the effect of AT on volatility
Mean coefficient on AT Cross-sectional t-stat Mean coefficient on AT*HARD Cross-sectional t-stat Mean total effect on HARD days Cross-sectional t-stat
|ret| 0.024 6.0 0.032 4.3 0.056 6.8
|mktadjRet| 0.017 4.1 0.037 5.4 0.054 6.7
Ret^2 0.016 5.2 0.030 5.8 0.046 7.2
37
MktadjRet^2 PriceRange 0.013 0.040 4.2 7.2 0.033 0.017 6.4 3.2 0.045 0.057 7.0 7.5
Ln(Ret10_ Ln(Ret30_ Var) Var) 0.016 0.027 3.3 4.7 0.003 0.006 0.9 1.2 0.020 0.033 3.2 4.5
Table 7. Instrumental variable estimation of the effect of algorithmic trading Our data cover 39 markets from 2001-2009. We aggregate all variables within each market by forming market-value-weighted averages across firms for each day within each market. We estimate a two-way panel over 39 markets and nine years using a generalized instrumental variable approach. We use AT, the negative of dollar trading volume (in $100) per message, as a proxy for algorithmic trading. As an instrument for algorithmic trading we use colocation, a dummy that is one on days when a market allows colocation. Market quality measures include quoted spread(RQS), relative effective spread(RES), permanent price impact(RPI), temporary price impact(RRS), quoted depth, |AR10|, and |AR30|, and volatility measures include |Ret|, |MktadjRet|, Ret^2, MktadjRet^2, the daily price range standardized by the daily closing price, and Ln(Ret##_Var), the log of the daily averages of the variances of 10-minute and 30-minute quote midpoint returns, respectively. Control variables include daily share turnover, price range, 1/price, log market cap, and the first lag of the dependent variable, all measured at t-1. Price range is not included in regressions where the dependent variable is volatility. Each model includes market and day fixed effects with t-values based on standard errors that are robust to cross-section and time-series heteroskedasticity and within-group autocorrelation (see Arellano and Bond (1991)). Dependent Panel A. Liquidity RQS RES RPI RRS Amihud
AT coefficient
t
-1.0087 -1.1546 -0.9285 -0.2718 -0.0001
-6.6 -6.2 -7.4 -1.8 -6.0
Estimate -0.0006 -0.0001
t -4.5 -0.5
0.0002 0.0001 0.0001 0.0000 0.0000 0.0053 0.0087
7.9 2.7 4.3 0.1 0.5 5.0 6.9
Panel B. Efficiency Dependent |AR10| |AR30| Panel C. Volatility (controlling for lag RES and lag|AR| PriceRange |Ret| |MktAdjRet| Ret^2 MktAdjRet ^2 ln(Ret10_Var) ln(Ret30_Var)
38
Figure 1. Annual average number of quotes and trades for each market-year We count all intraday messages that represent trades or changes in the best quotes for each stock on all 39 markets in our sample. Then we compute equally weighted means and medians for each market-year, and then compute means and medians across markets for each year. 2000 1800 1600 1400 Median across markets of annual mean across firms Mean across markets of annual mean across firms Median across markets of annual median across firms Mean across markets of annual median across firms
1200 1000 800 600 400 200 0 2001
2002
2003
2004
2005
2006
39
2007
2008
2009
Figure 2. Liquidity measures TWQS are time weighted relative quoted spreads (RQS), RES are relative effective spreads, RRS are 5-minute relative realized spreads, and RPI are 5-minute permanent price impacts. We compute all spread measures intraday for each stock. Then we compute the equally weighted average for each day, and then the equally weighted average for each market and year. The figures report the mean of the annual market averages. 0.07 0.06 0.05 0.04 Mean annual RQS
0.03 0.02 0.01 0 2001
2002
2003
2004
2005
2006
2007
2008
2009
0.03 0.025 0.02 0.015
Mean annual RES
0.01 0.005 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 0.018 0.016 0.014 0.012 0.01
Mean annual RRS
0.008
Mean annual RPI
0.006 0.004 0.002 0 2001 2002 2003 2004 2005 2006 2007 2008 2009
40
Figure 3. Volatility and efficiency The relative price range, RPR, is the intraday high minus low price, standardized by the daily closing price. |AR10| is the absolute value of the daily average 10-minute quote-midpoint return autocorrelations. We omit overnight returns and periods without price changes. The 30 and 60 minutes autocorrelations are computed analogously. We compute the equally weighted average for each market and year. The figures report the mean of the annual market averages. 0.045 0.040 0.035 0.030 0.025 Mean annual RPR
0.020 0.015 0.010 0.005 0.000 2001
2002
2003
2004
2005
2006
2007
2008
2009
0.350 0.300 0.250 0.200
Mean annual |AR10|
0.150
Mean annual |AR30|
0.100 0.050 0.000 2001
2002
2003
2004
2005
2006
2007
41
2008
2009
Appendix. Market-specific coefficients of algorithmic trading intensity. Our data cover 39 markets from 2001-2009. This table lists the coefficients of AT in 39 markets. We perform firm-day fixed effects panel regression within each market. Market quality measures include quoted spread (RQS), relative effective spread(RES), permanent price impact(RPI), temporary price impact(RRS), |AR10|, and |AR30|, and volatility measures include |Ret|, Ret^2, and the daily price range standardized by the daily closing price. AT is the negative of dollar trading volume ($100) per message. Control variables include daily share turnover, price range, 1/price, log market cap, and the first lag of the dependent variable, all measured at t-1. All daily variables are standardized every day to have a mean of zero and a standard deviation of one within each exchange. Regressions where the dependent variable is volatility do not include price range, but add RES, |AR30|. |AR30| is the absolute value of intraday autocorrelations measured for quote-midpoint returns over 30-minute periods. Firm fixed effects and (date) time dummies are included with t-values based on standard errors that are robust to cross-section and time-series heteroskedasticity and within-group autocorrelation (see Arellano and Bond (1991)). RQS RES RPI RRS |AR10| |AR30| |ret| Ret^2 PriceRange G7 OECD β p β p β p β p β p β p β p β p β p Xetra 1 1 -0.009 0% -0.010 0% 0.001 42% -0.022 0% -0.021 0% -0.011 0% 0.030 0% 0.019 0% 0.016 0% London 1 1 -0.008 0% -0.009 0% 0.005 0% -0.014 0% -0.099 0% -0.037 0% 0.030 0% 0.019 0% 0.040 0% Milan 1 1 -0.015 0% -0.008 0% 0.016 0% -0.029 0% -0.007 7% 0.011 1% 0.005 4% 0.000 86% 0.032 0% Osaka 1 1 -0.038 0% -0.039 0% -0.014 0% -0.030 0% -0.015 0% -0.004 42% 0.002 20% -0.003 9% 0.009 0% Euronext Paris 1 1 -0.009 0% -0.007 0% 0.011 0% -0.014 0% -0.013 0% -0.007 1% 0.013 0% 0.007 0% 0.050 0% Tokyo 1 1 -0.028 0% -0.028 0% -0.011 0% -0.021 0% -0.005 0% -0.001 16% 0.010 0% 0.011 0% 0.016 0% Toronto 1 1 0.000 47% -0.004 0% 0.004 0% -0.008 0% -0.012 0% -0.007 0% 0.028 0% 0.013 0% 0.043 0% Euronext Amsterdam 0 1 -0.013 0% -0.012 0% 0.038 0% -0.029 0% -0.011 2% -0.015 0% 0.046 0% 0.030 0% 0.068 0% Athens 0 1 -0.015 0% -0.011 0% 0.005 11% -0.026 0% -0.016 0% -0.009 2% 0.013 0% -0.002 49% 0.055 0% Australian 0 1 -0.005 0% -0.004 0% 0.009 0% -0.015 0% -0.012 0% 0.001 44% 0.028 0% 0.013 0% 0.056 0% Brussels 0 1 -0.036 0% -0.025 0% 0.010 6% -0.044 0% -0.015 1% -0.002 72% 0.034 0% 0.028 0% 0.088 0% Copenhagen 0 1 -0.008 0% -0.012 0% 0.010 0% -0.015 0% -0.011 0% -0.011 0% 0.046 0% 0.038 0% 0.054 0% Helsinki 0 1 -0.005 0% -0.008 0% 0.007 2% -0.012 0% -0.016 0% -0.011 0% 0.062 0% 0.045 0% 0.087 0% Hong Kong 0 1 -0.017 0% -0.016 0% 0.001 45% -0.023 0% -0.012 0% 0.001 47% 0.011 0% 0.003 3% 0.014 0% Euronext Lisbon 0 1 -0.014 0% -0.010 1% 0.001 92% -0.020 0% -0.005 57% 0.005 62% 0.039 0% 0.028 0% 0.072 0% Barcelona 0 1 -0.016 0% -0.011 0% 0.015 0% -0.032 0% -0.011 2% -0.006 24% 0.037 0% 0.024 0% 0.094 0% New Zealand 0 1 -0.015 0% -0.002 40% 0.005 27% -0.001 71% -0.018 0% -0.021 0% 0.039 0% 0.023 0% 0.094 0% Oslo 0 1 -0.012 0% -0.006 1% -0.001 75% -0.009 0% -0.024 0% -0.006 7% 0.039 0% 0.031 0% 0.055 0% Swiss 0 1 -0.015 0% -0.008 0% 0.004 9% -0.014 0% -0.039 0% -0.022 0% 0.033 0% 0.026 0% 0.053 0% Singapore 0 1 -0.005 0% -0.002 1% -0.003 5% -0.003 0% -0.008 0% 0.000 87% 0.016 0% 0.005 0% 0.022 0% Stockholm 0 1 -0.026 0% -0.023 0% -0.012 0% -0.026 0% -0.025 0% -0.012 0% 0.036 0% 0.022 0% 0.055 0% Vienna 0 1 -0.003 39% 0.007 3% 0.029 0% -0.019 0% -0.024 0% -0.010 9% 0.106 0% 0.089 0% 0.119 0% Argentina 0 0 -0.022 0% -0.019 0% 0.018 0% -0.037 0% -0.034 0% -0.016 1% 0.055 0% 0.045 0% 0.086 0% Thailand 0 0 -0.006 0% -0.011 0% -0.019 0% -0.010 0% 0.011 0% 0.004 17% -0.005 1% -0.006 0% -0.004 3% Mumbai 0 0 -0.014 0% -0.017 0% -0.015 0% -0.020 0% -0.008 0% -0.005 0% -0.017 0% -0.011 0% -0.013 0% Istanbul 0 0 -0.011 0% -0.012 0% -0.036 0% 0.007 1% -0.029 0% -0.013 0% 0.012 0% 0.012 0% 0.010 0%
42
G7 OECD Kuala Lumpur Korea Mexican Philippine Sao Paulo Santiago Shanghai Shenzhen Tel-aviv Taiwan Warsaw
0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0
RQS β -0.015 -0.003 -0.031 -0.005 -0.015 -0.007 -0.023 -0.009 -0.017 -0.012 -0.010
p 0% 3% 0% 0% 0% 2% 0% 0% 0% 0% 0%
RES β p -0.015 0% 0.004 1% -0.029 0% -0.005 1% -0.015 0% -0.007 4% -0.044 0% -0.032 0% -0.014 0% -0.015 0% -0.006 0%
RPI β p -0.009 0% -0.007 0% -0.010 1% 0.013 0% -0.008 0% 0.005 21% -0.043 0% -0.035 0% -0.014 0% -0.037 0% -0.001 81%
RRS β p -0.016 0% 0.006 0% -0.024 0% -0.013 0% -0.017 0% -0.008 2% -0.042 0% -0.044 0% -0.006 1% -0.020 0% -0.010 0%
43
|AR10| β p -0.002 11% -0.004 8% -0.028 0% -0.001 90% -0.034 0% -0.013 1% -0.016 0% -0.013 0% -0.026 0% -0.008 0% -0.006 2%
|AR30| β p 0.003 3% 0.004 8% -0.002 74% -0.003 66% -0.010 0% -0.009 12% -0.002 18% -0.001 43% -0.012 0% -0.001 45% -0.005 10%
|ret| β 0.007 -0.004 0.037 0.011 0.042 0.046 -0.038 -0.027 0.038 -0.013 0.015
p 0% 5% 0% 0% 0% 0% 0% 0% 0% 0% 0%
Ret^2 PriceRange β p β p -0.003 2% 0.016 0% -0.004 4% -0.022 0% 0.028 0% 0.079 0% 0.001 56% 0.006 5% 0.027 0% 0.069 0% 0.042 0% 0.060 0% -0.037 0% -0.050 0% -0.028 0% -0.036 0% 0.026 0% 0.060 0% -0.013 0% -0.027 0% 0.009 0% 0.025 0%
