This paper studies the consequences of a change in the tick size in the Paris Bourse. It induced a decrease in depth at the quotes, but did not alter the spread. A coarse pricing grid encourages traders to submit and expose limit orders. Tighter grid induces frequent undercutting strategies.
This paper studies the consequences of a change in the tick size in the Paris Bourse. It induced a decrease in depth at the quotes, but did not alter the spread. A coarse pricing grid encourages traders to submit and expose limit orders. Tighter grid induces frequent undercutting strategies.
This paper studies the consequences of a change in the tick size in the Paris Bourse. It induced a decrease in depth at the quotes, but did not alter the spread. A coarse pricing grid encourages traders to submit and expose limit orders. Tighter grid induces frequent undercutting strategies.
Why Markets should not Necessarily Reduce the Tick Size
David Bourghelle a and Fany Declerck b, c a IAE, Lille 1 University b Toulouse University March 2002
Abstract: This paper studies the consequences of a change in the tick size in the Paris Bourse. Some stocks experienced a decrease in the tick size. It induced a decrease in depth at the quotes. However, in contrast with results obtained for US markets, this neither generated a change in the bid-ask spread nor a reduction in liquidity provision for large trades. Other stocks experienced an increase in the tick size. Again, this did not alter the spread, but it increased the depth at the best quotes. We also find that a relatively coarse pricing grid encourages traders to submit and expose limit orders, while a tighter grid induces frequent undercutting strategies. JEL Classification: G14 G18 G19 Key words: tick size, liquidity supply, spread, market depth and order exposure.
c Correspondance: Fany Declerck Universit de Toulouse 1 Manufacture des Tabacs MF 308 GREMAQ 21 alle de Brienne 31000 Toulouse; Voice: +33.(0)561-128-570; Fax: +33.(0)561-225-563; Email: fany.declerck@univ-tlse1.fr We are indebted to Bruno Biais for very insightful discussions, suggestions and advice. We thank P. Alphonse, R. Bellegarde, E. De Bodt, M. Demarchi, R. De Winne, R. Gillet, P. Hazart, V. Remay, R. Van Ness, seminar participants at the IDEI (Fourth Toulouse Finance Workshop), the PASFI (Lille Business School Workshop) and the AFFI and NFA meetings for valuable comments and suggestions. We are grateful to Euronext -Paris for providing data. All remaining errors are our own. 2 1. Introduction Competitive pressure recently led US and Canadian exchange to carry out large decreases in tick sizes. This minimum allowable price variation determines the minimum market bid-ask spread, which is important in determining transaction costs. US average execution costs decreased significantly from 38.10 to 24.55 basis points following the 1997 reform to sixteenths and the decimalization program. On January 4, 1999, Euronext Paris adopted a new pricing grid with significant changes in the smallest allowable price increment. The new grid was intended to reduce execution costs and improve market quality. There is a significant debate between academics, regulators and practitioners about the optimal tick size. In 1994, after a four- month evaluation program in which tick sizes were reduced by 50%, a large drop in trading activity forced the computerized limit order market of the Hong Kong Stock Exchange to revert to the original tick size for the low priced stocks. Besides, and more recently, a few weeks only after the switch to decimal pricing for the 3,535 NYSE listed stocks on January 29, 2001, some big institutional traders strongly requested an increase of the minimum tick size from one to five cents, clearly stressing the potential advantages of a coarser pricing grid. In fact, the net effect of tick size changes on execution costs and liquidity supply may depend on market design. For example, the ability to undercut on a fine grid may be especially profitable for floor traders and specialist on the NYSE, and could generate winners' curse problems for investors away from the floor 1 .
1 Within a specialist market like the NYSE, the privileged information set of the specialist and the floor traders is such that they can take advantage of undercutting the best quotes. Conversely, frequent value-motivated based undercutting strategies are expected within a transparent computerized limit order market. 3 From a theoretical point of view, since the tick size is the lower bound of the bid-ask spread, Harris (1994) predicts that a reduction in the tick size would decrease the quoted spread 2 . Nevertheless, the reduction in the spread could also decrease order exposure because liquidity provision is less profitable and more risky. As a consequence, the quoted depth would also decline. Cordella and Foucault (1999) develop a model that suggests that a zero tick does not minimize trading costs, as traders are more willing to quote the competitive spread when the tick size is large. Also, a coarse pricing grid helps to enforce time priority and to protect limit orders from adverse selection [Harris (1991,1994) and Seppi (1997)]. Numerous papers have empirically investigated the consequences of reduction in tick size on market quality 3 . As mentioned in Harris (1991, 1994), Seppi (1997) and Biais et al. (2001), a tightening of a pricing grid would generally result in a reduction in bid-ask spreads. Finally, while the change in tick size has improved the liquidity for small size orders, Goldstein and Kavajecz (2000) and Lipson and Jones (2001) find that institutional traders were worse off because they had to bear an increase in trading costs following the decline in depth throughout the entire order book. While these papers considered the effects of the tick size change in specialist markets like the NYSE, the present paper studies the consequences in an electronic limit order market. Actually, fully displaying limit orders in this environment provides free trading options to (parasitic) quote matchers who can trade in front of large traders. Yet, a tighter pricing grid may not adversely affect the cumulated depth in a computerized order-driven market if the exchange offers facilities
2 Harris (1994) estimated a cross-sectional discrete model to assess the impact of a reduction in tick size. He projected that quoting in sixteenths would result in a decrease in spreads (quoted depth) of 38% (16%) for stocks with price under $10. He also predicted an increase in daily volume of 34% for these stocks. 3 See Harris (1997) and Van Ness et al (1999) for an extensive review of these studies. 4 to liquidity providers to partially hide their orders, and/or if traders shift their limit orders to prices further from the quotes. Inversely, even if a coarser pricing grid could result in a larger cost of immediacy, it may increase propensities of traders to expose their orders, thus balancing the cost effect and improving market liquidity. The new price schedule at Euronext Paris offers a natural opportunity to explore theses two issues simultaneously. Before the switch, although the relative tick size for stocks over 2,000FRF (French Franc) was only 0.05%, it reached 0.20% for stocks over 500FRF (the minimum bid-ask spread was then 1FRF). The new pricing grid was designed so that the relative price increment did not exceed 0.10% of share prices. Consequently some stocks benefited from a reduction of their relative tick size (what we will call DTS or decreasing tick size sample) and some others experienced an increase in the relative tick size (ITS or increasing tick size sample). Thus, we can examine both the relaxation and the tightening of the constraint that the tick size imposes within the same schedule. Our main results are the following. With regard to execution costs, and in contrast with previous studies, we do not observe any significant decrease in the relative spread for the decreasing tick size sample. It seems that the old pricing grid did not cause artificially wide spreads before the switch 4 . While the average relative tick went from 0.0989% to 0.0558%, the average spread to tick ratio (the average number of times the tick size is contained in the average quoted spread) rose from 3.9 to 6.5. Yet, since a smaller relative tick size allows traders to obtain priority at less cost, limit order submission inside the best quotes (on the best quotes) increases (decreases) significantly for the decreasing tick size sample. On the contrary, limit order submission inside the best quotes (on the
4 Even if the relative tick were rather different among stocks before the change over to sixteenths on the NYSE, AMEX or NASDAQ, all US studies showed a significant decrease in proportional quoted spreads after the change. 5 quotes) decreases significantly (is unchanged) for the increasing tick size sample. Indeed, we clearly observe a change in the traders behavior to protect themselves from front-running strategies. For the decreasing tick size sample, hidden quantities and the daily number of modified orders significantly increase. On the contrary, the proportion of hidden quantities significantly decreases for the increasing tick size sample. A larger relative price increment makes undercutting strategies more expensive and so encourages investors to expose their orders. Changes in order submission strategies alter quoted and possibly market depth. First and as expected, stocks that benefit from a reduction in the relative tick size show evidence of a significant drop in the depth at the best quotes. Thus, as mentioned previously, small orders do not get a reduction in their proportional cost of immediacy. However, the average proportion of trades for which the quoted depth at the best quotes is not as sizeable enough to allow complete execution significantly increases. Yet, it appears that institut ional traders do not suffer from damaging trading conditions since the weighted average spread for a standard-sized block is unchanged. The finding that a decreasing tick size does not increase trading costs for large trades is in marked contrast with previous US studies (Jones and Lipson, 2001). On the other hand, stocks that experienced an increase in the relative tick size show evidence of a significant rise in the depth at the best quotes but no change in the cumulated depth. This paper is organized as follows. The next section discusses the hypothesis tested in this paper. Section 3 describes the trading system and presents data used in the paper. Section 4 provides our empirical results. Some concluding remarks are contained in section 5. 6 2. Empirical hypothesis The minimum price variation is important in determining bid-ask spread, liquidity, order exposure, and obviously quoted and cumulated depth. The economic insights, which provide the background of our investigation, are the following. 2.1. Bid-ask spread The tick size induces the minimum bid-ask spread, which measures transaction costs. In the absence of constraints, market forces result in a bid-ask spread based on determinants such as trading volume, risk, information and competition. If the price grid available on an exchange is such that the minimum price increment is larger than the minimum spread available, a reduction in the tick size would lead to a reduction in the cost of immediacy thus encouraging trading [Harris (1994)]. On the contrary, a raise in trading costs should follow tick size increases. Empirical implication 1: If the current tick size is a binding constraint, a tick size reduction could lower the quoted bid-ask spread, otherwise it would have no impact. On the other hand, a raise in trading costs should follow tick size increases. 2.2. Liquidity supply and order exposure If time priority is enforced, traders have to improve the best quote to step ahead of the book. Harris (1996) and Angel (1997) demonstrate that with a finer pricing grid, the cost of acquiring order precedence through price priority is marginal. So time priority protects limit orders against quote-matching and front-running. This competition takes place when traders submit limit orders inside the spread. Therefore a smaller tick size allows traders to obtain priority at a lower cost. Empirical implication 2: The number of orders submitted inside the spread would increase (decrease) after a tick size reduction (increase). 7 A reduction in the minimum price increment facilitates undercutting and quote- matching behavior. To cut down order exposure, investors will probably cancel and/or modify their orders more often when the tick size decrease. They can also split their orders into several small orders. The ratio of market orders to limit orders submitted by retail investors could also increase for stocks with a smallest tick size. Finally, limit order traders will probably use more frequently hidden quantities 5 . Empirical implication 3: Traders will reduce their order exposure after a tick size reduction. Conversely, order exposure will increase after an increase in tick size. 2.3. Quoted and cumulated market depth If the tick size is a binding constraint on the spread, the latter is artificially wide and it is useful to submit limit orders in the book. While a smaller tick size may narrow quoted bid-ask spreads, it could adversely affect liquidity provision. The difference between the prices at which a liquidity provider is willing to buy (bid) and sell (ask) represents the revenue per share for providing immediacy to market orders. Thus, as liquidity supply is less profitable after a reduction in tick size, smaller sizes could be offered at the quotes. Empirical implication 4: A smaller (larger) tick size will decrease (increase) quoted depth. As noted in Lau and McInish (1995) and Goldstein and Kavajecz (2000), a reduction in tick size could lead liquidity providers to reduce depth not only at the best prices but away from the quotes too. As a consequence, it seems important to check if large liquidity demanders are not worse off by suffering from an increase in transaction costs. For instance on the NYSE, Jones and
5 Euronext Paris allows traders to display only a portion of their orders. The remaining size is hidden. Harris (1997) finds that the proportion of hidden orders is larger for stocks trading with a fine tick size in Paris.
8 Lipson (2000) find that institutional trading costs increase with the change to sixteenths, as the depth decreases throughout the entire order book. The net effect of a tick size change on the cumulative depth and the cost of trading large order sizes is hence uncertain. The cumulative depth (aggregate quoted size) at a certain price could remain constant if liquidity providers shift their limit orders to prices further from the quotes. Moreover, for the stocks that experienced an increase in relative tick, a change in cumulated depth could be observed if investors expose more orders. Empirical implication 5: A smaller tick size does not make larger-sized orders more expensive. 3. Trading system and data 3.1. Market design This study investigates the impact of the tick size changes at Euronext Paris, which is based on a computerized limit-order trading system. Orders are prioritized for execution in terms of price and time: orders for each security are ranked by price limit as they enter the system. For example, buy orders specifying a higher limit are executed before orders with lower limits. Secondly, orders are ranked in chronological order: two buy or sell orders at the same price will be executed in the order in which they arrive on the central book. In an order-driven market, there is no designated market maker who has the obligation to provide liquidity. By submitting a limit order, a trader is providing to other market participants the ability to execute against his limit order. So, limit orders provide liquidity to those who demand immediacy (market order traders). 9 The trading day is ten hours 6 , beginning at 7:15 a.m. and ending at 5:30 p.m. Paris local time. From 7:15 a.m. to 9:00 a.m., the market is in pre-opening phase and orders are fed into the centralized order book without any transactions taking place. The market opens at 9:00 a.m. The central computer automatically calculates the opening price or call auction price at which the largest number of bids and asks can be matched. From 9:00 a.m. to 5:25 p.m., trading takes place on a continuous basis, and the arrival of a new order immediately triggers one (or several) trade(s) if a matching order exists on the other side of the book. From 5:25 p.m. to 5:30 p.m., the market is in its pre-closing period. As in the pre-opening session, orders are fed into the order book. The market closes at 5:30 p.m. with a call auction that determines closing prices. Trading is anonymous. Cancellation of orders may be done at any time. Starting on January 4, 1999, the new price grid sets a sliding scale of tick size in which the tick varies from 0.01 for prices up to 50, 0.05 for prices between 50.05 and 100, 0.10 for prices between 100.10 and 500, and 0.50 for prices above 500. Euronext introduced rules on block trading to allow immediate and full execution of such trades at a guaranteed price derived from rates available on the central market. Block trades must take place at a price which falls within the weighted average spread for a standard-sized block. The weighted average Bid-Ask spread (WAS) represents the price for blocks that exceed the Normal Market Size (NMS), and is calculated for a given quantity of shares in real time by taking the average of bid and ask prices for all orders placed in the central order book, weighted by the number of shares displayed at the successive bids and asks.
6 For our examination period, the continuous trading session begins at 10:00 a.m. and ends at 5:05 p.m. 10 3.2. Data Our stock sample consists of 37 stocks traded in the continuous market and which belonged continuously to the CAC 40 index during our examination period 7 (see Table I for details related to the sample firms). The component stocks are the most actively traded and provide a reasonable representation of the market, since they account for about 74% of the market capitalization, and about 85% of market volume. We use the Euronext database (BDM) intra-day five best quotes, orders and transaction prices about all stock for each day in our examination period. Information is time-stamped to the second. Call auction opening and closing prices are such that traders who submit orders may use different strategies than those who submit to the continuous market. To account for abnormal trading patterns and procedures around the start of each day and the close of the trading day, the opening and closing five minutes of trading are excluded. We therefore examine only trades and quotes data available between 10:00 a.m. and 5:00 p.m. while the market was open for continuous trading. Our examination period extends from 50 trading days (October 15, 1998) prior to the switch through 50 trading days (March 15, 1999) after the switch. We exclude observations 10 trading days around the change to eliminate any unusual trading behavior associated with the adoption of the new price grid. We call the 45 trading days preceding the window the pre-event period and the 45 trading days following the window the post-event period 8 . Sample sizes during the period include 4 544 814 trades, 8 076 428 orders, and 896 704 cancellations of orders.
7 We avoided 3 stocks that were included in or excluded from the Index between October 15, 1998 and April 15, 1999. 8 As some organizational and regulatory changes occurred in 1998 and 1999 at Euronext Paris (changes in opening hours, adoption of a fixing procedure for the market closing), we did not use any other one separate pre- event or post-event period to control for any transition period caused by adjustment behahior of market participants. 11 3.3. Trading activity The first half of 1999 years at Euronext Paris followed the same trend as 1998 years exceptional figures, once again beating a succession of records as the CAC40 index went past the 4,600-point mark for the first time, showing an increase of 15% for the six- month period. Transaction volumes reached 323.46 billion during the first half, a 26.51% progression from 1998s first half, which had then set a record. The total number of trades from January to June was 25.96 million an increase of 17.5%. The daily average of 2.6 billion in transaction volumes also sets a record, representing a 22.9% increase compared to the first six months in 1998. Total transaction volume for the first eight months was 437.46 billion, a 30% increase over the same period in 1998, with 34.9 million trades being registered (+22.6%). On the whole year, the CAC 40 index increased by 48.7%, that is the second most important increase since 1987. Summary statistics for our sample firms are shown in Table 2. The equally- weighted statistics are obtained by calculating the average by firm on all the trading days, and then by calculating the average over firms. The summary statistics are measured over the post-event period. The stocks are divided into four quartiles based upon average daily volume. PLEASE INSERT TABLE 2 HERE Average market capitalization is 17 billion with a maximum of 76 billion, which is 25 times the minimum. The prices range is large: the minimum is 30 and the maximum is 656, with an average value of 161. Average daily volume is 39 million, with a minimum (maximum) of 2 (373) million. Trading volume for a time interval can be divided into two 12 components, the number of trades and the average trade size. It appears that the daily trade size increases with trading activity : 14,787 for the first quartile and 39,096 for the last quartile 9 . The average bid-ask spread is 0.45 and the average percentage spread is around 0.2967 % during the sample period. The percentage spread is a better measurement of transaction costs since it is independent of the price level. Angel (1997) reports that the median percentage spread is 0.32% for the Dow Jones Industrial Average index stocks. Recently, Venkataraman (2001) find that the quoted spreads in Paris (0.26 percent) are lower than spreads on similar NYSE stocks when the tick size at the NYSE is an eighth (0.31 percent), but higher than NYSE spreads after the reduction in tick size at the NYSE to the sixteenth (0.24 percent). Bid and ask depths measure the number of shares that can be sold or bought without causing significant price movements. As expected, the percentage spread decreases, and quoted depth increases as one moves to upper quartiles. 4. Empirical results 4.1. The new pricing grid and relative tick Table 3 and 4 show details related to the pricing grid available before and after the switch. They indicate the number of CAC 40 index stocks trading in each bracket, and the maximum relative tick that can be observed (the ratio of the price increment to the minimum share price in each category). With Table 3, we note that the maximum relative price increment was for a stock priced at 5.05 FRF before the switch (1%), but never exceeds 0.10% with the new price grid.
9 As informed investors want to maximize the profit they can obtain from their private information, they may hide their trading activity by splitting one large trade into several small-sized trades [Kyle (1985) and Admati and Pfleiderer (1988)]. Jones et al. (1994b) investigate how daily price volatility could be explained by daily number of trades and average trade size. They find that the number of trades appears to provide more explanation for the volatility-volume relation than average trade size. 13 Table 4 (panel A) shows that before January 4, 1999 not far from 90% of the stocks were quoted with a price level expressed as an integer multiple of 0.1 FRF or 1 FRF (around 57% with a tick size of 1 FRF). After the change, this clustering effect is not observed any more (panel B). The tick size constraint is more likely to be binding at the inside quotes for liquid and/or low priced stocks. Indeed, given a large relative tick the proportion of one-tick spreads (bid-ask spreads equal to the price increment) should be high. PLEASE INSERT TABLE 3 AND TABLE 4 HERE Table 4 shows a large proportion of one-tick bid-ask spreads. This fact induces that the tick size was a binding constraint in the pre-event period (almost 40% for the 1 FRF tick size). With the switch, numerous stocks seem to have benefited from a reduction in the proportion of one- tick spreads. PLEASE INSERT TABLE 5, TABLE 6 AND FIGURE 1 HERE In fact, as Tables 5 and 6 indicate, a close examination of the stock sample allows us to separate 26 stocks for which a reduction in the relative tick became effective after the switch from 11 other stocks bearing an increase in the relative price increment. For the stocks trading in price ranges for which the relative tick decreased after the switch (DTS sample), the proportion of one-tick spreads remained stable or decreased dramatically, and as expected the quoted depth generally decreased. For the increasing tick size sub-sample (ITS sample), the larger relative price increment became largely binding (for example, the proportion of one-tick spreads rose from 8.64% to 25.67% for stocks trading between 50 and 76.25), suggesting important changes in market quality and liquidity provision. 14 Indeed, the switch provides a unique opportunity to investigate not only the effects of a tick size reduction, but allows for separate examination of tick size increases. In the next point, we begin by analyzing standard measures of bid-ask spreads. 4.2. Bid-ask spreads We compute the daily time weighted average proportional spread, i.e. the average of the bid- ask spread to midpoint weighted by the proportion of the day that the spread was quote. Harris (1994) shows that the impact of the tick size reduction on liquidity is not uniform across stocks because it depends on the trading activity. So, we partition the sample into four sub-samples based on daily average volume. To test the hypotheses, we first calculate the time-series averages in pre-event and post-event periods for each stock. Next we calculate the cross-sectional means (or median for each sub-sample) and standard errors for each period. A paired t-test for the full sample is performed to determine whether the change in the variable is significant. As the sub- samples are small (9 or 10 stocks), we also performed a non-parametric Wilcoxon test for each one 10 . PLEASE INSERT TABLE 7 AND FIGURE 2 PANEL A HERE In accordance with theoretical work and empirical analysis of tick size change, we anticipate a decrease (increase) in both spread and depth following a decrease (increase) in relative price increment. In marked contrast to previous studies, we do not observe any significant decrease in the relative spread for the decreasing tick size sample. Table 7 (panel A) shows that the time
10 Moreover, the same results were found when we used both parametric and non-parametric tests for the full sample. 15 weighted relative spread is even unchanged for the two samples whatever the quartile 11 . Expressed as a proportion of the midpoint, quoted spreads remain on average at a 0.30% level. Consequently a reduction in tick size does not always lead to reduced transaction costs. Our hypothesis was that a smaller price increment would have enhanced the capacity of investors to improve quoted prices because the old pricing grid was too coarse, impeding competition and causing artificially wide spreads. It seems that the old pricing grid was already fine enough, so that the reform did not have much effect on spreads 12 . Nevertheless, liquidity providers made up for a reduction in the per-share proportional revenue with an increase in the average number of tick contained in the quoted spread (panel B and panel C) 13 . Surprisingly, another result which differs completely from the usual findings, is that the time weighted proportional spread is not altered by the change for the increasing tick size sample. The null hypothesis that the proportional spread is equal before and after the switch for the ITS sample is not rejected at the 1% level, suggesting that the increase in relative tick did not lead to rising execution costs. Otherwise, Table 7 panel B presents statistics of the proportion of one-tick spreads for the four volume quartiles in the pre-event and post-event periods. It appears clearly that the constraint was far less binding in the pre-event period for the ITS sample (7.17% of spreads equal to the price increment for the ITS sample and 31.12% for the DTS sample). Therefore, the number of one-tick spreads rose significantly to attain 20.55% after the switch for
11 This result holds when we exclude some stocks that passed through a price bound which makes them trade with a different tick for a short while. 12 As observed in Table 6, the average relative tick for the stock that experienced a tightening of the grid was already thin at a 0.0989% level. Yet, Chan and Hwang (1998) showed that within an order driven market environment (the Hong-Kong stock Exchange), the tightening of a quite thin pricing grid led to a significant decrease in proportional spreads. 13 This behavior seems to hold for all the stocks in the DTS sample whatever the price range. 16 the ITS sample. The constraint became binding for the ITS sample in the post-event period, revealing more incentives of some traders to be the first one to quote the narrowest possible spread quickly, and consequently enjoy time priority at this relatively advantageous price 14 . Conversely, as the proportion of one-tick spreads decreased to a 15.69% level, and as the average spread-to-tick ratio rose significantly (panel C) in the DTS sample, liquidity providers are less prompted to be the first one to quote the competitive spread. This empirical evidence is consistent with the theoretical analysis of Cordella and Foucault (1999), and supports the hypothesis that the price increment which minimizes the cost of immediacy and leads investors to post the competitive spread (improve market efficiency) is not zero. 4.3. Liquidity supply and order exposure If time priority is enforced on a market, traders have to improve the best quote to step ahead of the book. With a finer pricing grid, the cost of acquiring order precedence through price priority is reduced. Therefore, a smaller (larger) tick size is likely to decrease (increase) limit order submission on the best quotes. Moreover, time priority protects limit orders against quote- matchers. A reduction in the minimum price increment exposes limit orders traders to undercutting behavior. On the other hand, a rise in relative tick allows investors to control for risk. As limit orders compete with market orders for order flow, changes in limit order activity may be closely related to changes in market quality. In this section, we present some summary statistics on the distribution of order flow across order types and order exposure. In Table 8 and 9, we sort on orders type (market orders or limit orders), hidden quantity and order state. We present the total daily order flow for each order type and also the proportion of each order type
14 Biais et al. (2001) 17 relative to all orders. We test for changes in the daily number of orders and the daily proportions. Table 8 is related to the decreasing tick size sample (DTS sample) and Table 9 to the increasing tick size sample (ITS sample). PLEASE INSERT TABLE 8 AND TABLE 9 HERE There are several noteworthy points here. First, whatever the sample, market orders account for only 11% of orders while limit orders account for about 86% of orders. Secondly we observe for the decreasing tick size sample, a statistically significant increase in the daily total number of orders, the number of market sell orders, and the relative number of market orders. Furthermore, the relative and absolute number of limit orders increase significantly at a 1% level. Conversely, with regards to the ITS sample, only the daily number of market and limit buy orders slightly increases. Also those changes have to be carefully considered, it could be that the frequency of limit orders is greater when the tick size is thin, as suggested by Cordella and Foucault (1999). Perhaps more important are the changes in hidden quantity between pre and post-event periods. The CAC system allows traders to partially disclose their orders. Rule gives, at a given price, precedence to displayed orders size over hidden size. Indeed, this system hides the remaining size and displays it only after the previous displayed size is executed. Harris (1994) argues that a reduction in the tick size could lead to lower displayed depth, as liquidity providers hide more of their orders to avoid quote matchers. Harris (1996) investigates the empirical relation between tick size and order exposure by using orders data from Euronext Paris and the Toronto Stock Exchange. He shows that, unless price volatility is high, larger ticks are associated with greater order exposure when the order is not expected to stand long. As we expected, hidden quantity as expressed as an average daily number of orders or as a percentage of order volume increases significantly after the switch for the DTS sample. This 18 behavior suggests that investors want to hide more often their trading interest. In the same way, we observe a significant rise in the average daily number of modified orders, which reveal an attempt to create uncertainty about the reasons why they want to trade. For the ITS sample, the use of hidden quantity, expressed both as a percent of all orders or of order volume, decreases significantly after the switch. As the tick size is now a binding constraint for the bid-ask spread, costly undercutting practices are reduced and order exposure becomes less risky. As a reduced (larger) relative tick makes price priority less (more) costly, limit order submission inside the best quotes is likely to increase (decrease). Table 10 shows clear evidence of this behavior. The proportion of orders submitted inside the spread significantly increases (decreases) after the reduction (rise) in tick size. At the same time, the average daily proportion of orders submitted at the quotes significantly decreases for the DTS sample (the post-event level is about twice as small as before the adoption of the new pricing grid). On the other hand, a larger relative tick encourages order exposure inside the book rather than on the best quotes (no significant change for the latter). PLEASE INSERT TABLE 10 HERE These results suggest an important change in the trading behavior of investors, especially as concerns the determinants of orders aggressiveness. In fact, it brings empirical evidence that a finer pricing grid reduces the willingness of limit order traders to provide market liquidity at quotes. 4.4. Quoted and aggregate depth As observed previously, a smaller tick size affects liquidity provision. Even if the relative bid-ask spread is unchanged, less limit orders are offered at the best quotes after the switch. We 19 therefore calculate the depth at the bid (ask) quotes, as the average of the value of shares available for immediate exchange. Furthermore, the aggregate quoted size at a certain price could remain constant if liquidity providers shift their limit orders to prices beyond the best quotes. Moreover, if investors display more trading interest or if some new liquidity providers want to take advantage of a larger relative tick size, we expect an increase in cumulated depth. Finally, it is important to check if large liquidity demanders (institutional) are not worse off in suffering from an increase in transaction costs for large trades. 4.4.1 Quoted depth and the cost of trading small orders As indicated in Table 11, and consistent with the empirical hypothesis, we find that the quoted liquidity as measured by the time weighted euro depth available at the quotes decreases significantly for the decreasing tick size sample. For instance, across all firms, the average euro bid and ask depths decline by about one quarter. The biggest change occurs in firms with high volume: a drop of 38%. PLEASE INSERT TABLE 11 AND FIGURE 2 PANEL B HERE We also calculated the change in the average proportion of trades for which the quoted depth was not as sizeable enough to allow complete execution at the quotes. We found a large rise for the stocks that experienced a reduced relative tick (from 23% to 32%) and no significant change for the ITS sample. While this result may not necessarily indicate an increasing cost of immediacy for the former, some buyers (sellers) have to walk up (down) the book to get complete execution of their orders 15 .
15 Harris (1997) also shows that the tick size reduction results in a significant decrease in the quoted depth on the Toronto Stock Exchange (as mentioned by Harris (1996), the TSE uses a pure price-time order precedence hierarchy to arrange trades as does Euronext Paris). 20 Thus our results suggest that investors refrain from displaying trading interest at the quotes after a tick size reduction. Conversely, the quoted liquidity rises dramatically after the switch for the increasing tick size sample (more than 40% increase for the entire sample) 16 . Indeed, traders are more willing to provide liquidity at the quotes after an increase in the tick size, probably because front-runners have to bear a considerable cost to get price priority by improving quotes. 4.4.2 Cumulated depth and the cost of trading large orders As recognized by Harris (1994), a decrease in sizes at the best quotes does not indicate a decrease in the total market depth. Yet, Goldstein and Kavajecz (2000) reconstruct the NYSE limit order book around the tick size reform. For the 100 stocks they examine, depth decreases throughout the book. Lau and McInish (1995) study the reduction in tick size from $0.50 to $0.10 at the Singapore Stock Exchange, and find that the sum of all quoted sizes is reduced after the tick size decrease. Jones and Lipson (2000) find that even if the cost of retail orders declines after a tick size reduction, the cost of institutional orders is greater with a smaller tick. Seppi (1997) also predicts that a smaller tick size benefits to small traders but is costly for large traders. In fact, the depth at the best quotes can not be considered as an adequate measure of liquidity. Thus, Biais et al. (1995) show that for Euronext, the provision of liquidity at the quotes is only a small portion of the overall depth in the book 17 . To assess the tick size reduction impact on medium-sized orders execution costs, we calculate an average depth-to-trade size ratio for each trade before and after the switch. The depth-to-trade size ratio indicates the number of times the average trade size is contained in the
16 Nevertheless, changes are not statis tically significant when we partition the sample into four sub-samples based on daily average volume. 17 Unfortunately, we can not make direct evidence of a potential change in the quantity of shares posted at limit orders away from the quotes (cumulated depth) as the five best limits of the book are only available after the switch. 21 average quoted depth. As observed in Table 12, we document a large drop in the depth to trade size ratio (- 47%) for the decreasing tick size sample, and no significant change for the increasing tick size sample. It confirms that the reduction in tick size significantly reduced the quoted depth, but it did not affect average trading conditions. PLEASE INSERT TABLE 12 HERE This result can be reinforced by analysing potential changes in market conditions for large (institutional) traders. Euronext introduced rules on block trading to allow immediate and full execution of such block trades at a guaranteed price derived from the central market. The Weighted Average Spread for a standard-size block (WAS) represents the price for blocks that exceed a defined Normal Market Size (NMS). It is continuously calculated by the Paris Bourse for a given quantity of shares in real time by taking the average of bid and ask prices for all orders placed in the central order book, weighted by the number of shares displayed at the successive bids and asks. Indeed, the quoted WAS, which is calculated on 10 limits on average, is 1.72% for trading 50 times the average trade size without hidden orders. Declerck (2000) shows that reserved (or hidden) orders represent up to 60% of disclosed orders, which means that the disclosed market depth is much lower than the true market depth. On average, the effective depth at the best limits is about three times higher than the displayed depth. The effective WAS, which takes into account the hidden quantities entered into the system, is only on average 0.80%. PLEASE INSERT TABLE 13 HERE Table 13 does not show evidence of any significant change in the WAS for the two samples. This result implies that institutional trading costs neither increased nor decreased after the adoption of the new pricing grid at Euronext Paris. In marked contrast with previous US studies, 22 large traders did not suffer from damaging trading conditions consecutive to a reduction (or a rise) in the relative tick. 5. Conclusions This paper examines the impact of a tick size change in the Paris Bourse. By analyzing change in bid-ask spreads, order exposure and market depth, we deliver some insights on the importance of the tick size for the "economics" of liquidity provision in an electronic limit order book market. The change in the pricing grid allows us to examine the effects of a tick size reduction, but gives us the opportunity for separate examination of tick size increases. This is undeniably one of the main contributions of this paper. In contrast with previous studies, we do not observe any significant decrease in the relative spread for the firms that experienced a reduced tick size. It appears that the old pricing grid was already rather fine. Thus, it appears that a reduction in tick size does not necessarily lead to reduced execution costs. Besides, and as expected, limit order submission inside the best quotes (on the best quotes) increases (decreases) significantly, and investors use more hidden quantity orders to reduce exposure. Finally, we show evidence of a significant drop in the quoted depth. However, institutional block traders do not suffer from damaging trading conditions. These last results are in marked contrast with previous empirical evidence on US markets and might reflect differences in market design between European and US exchanges. Otherwise, we do not observe any significant rise in the relative spread for the firms that experienced an increasing tick size. Limit order submission inside the best quotes significantly decreases, and investors use fewer hidden quantity orders. Thus, a coarser grid makes undercutting strategies more expensive and encourages investors to expose their orders. 23 Moreover, we find a significant rise in the quoted depth, showing evidence of a larger display of liquidity. By increasing the per-share rent, a larger relative tick makes liquidity supply more profitable and probably attracts new limit order traders in the market. To attract liquidity demanders, designers of trading systems have to stimulate investors to fully display their orders. A relatively coarse pricing grid does not always result in excessively large spreads, but enhances quoted depth, encourages liquidity providers to expose their trading interest and stimulate investors to quote the competitive spread. Our analysis indicates that regulators may be well advised to avoid reducing tick size if they want to attract liquidity providers, and if order exposure is profitable for a market. Yet, the significant competition between trading mechanisms highlights the need for future research related to the consequences of tick size on trading costs and the dynamics of liquidity supply. 24 References Ahn, H.J., Cao, C.Q. and H. Choe, 1998, Decimalization and competition among stock markets: evidence from the Toronto stock exchange cross- listed, Journal of Financial Markets 1, 51-87. Ahn, H.J., Cao, C.Q. and H. Choe, 1996, Tick size, spread and volume, Journal of Financial Intermediation 5, 2-22. Angel, J.J., 1997, Tick size, share prices and stock splits, Journal of Finance 52, 615-681. Anshuman, V.R. and A. Kalay, 2000, Can splits create market liquidity? Theory and evidence, Working paper, Boston College. Bacidore, J.M., Battalio, R. and R. Jennings, 2001, Changes in order characteristics, displayed liquidity, and execution quality on the NYSE around the switch to decimal pricing, Working Paper, NYSE. Bacidore, J.M., 1997, The impact of decimalization on market quality: an empirical investigation of the Toronto stock exchange, Journal of Financial Intermediation 6, 92-120. Bessembinder, H., 2000, Tick size, spreads and liquidity: an analysis of NASDAQ securities trading near 10$, Journal of Financial Intermediation 9, 213-239. Bessembinder, H., 1997, Endogenous changes in the minimum tick: an analysis of NASDAQ securities trading near ten dollars, Working paper, Arizona State University. Biais, B., Glosten, L. and C. Spatt, 2001, The microstructure of stock markets, Working paper, Toulouse University. Biais, B., Hillion, P. and C. Spatt, 1995, An empirical analysis of the limit order book and the order flow in the Paris Bourse, Journal of Finance 50, 1655-1689. Bollen, N.P.B. and R.E. Whaley, 1998, Are "teenies" better? Low-priced shares are the greatest beneficiary, Journal of Portfolio Management, Fall Issue, 10-24. 25 Brown, S., Laux, P. and B. Schachter, 1991, On the existence of an optimal tick size, Review of Futures Markets 10, 50-72. Chakravarty, S., Harris, S. and R. Wood, 2001, Decimal trading and market impact, Working paper, University of Memphis. Chan, L.K.C. and C.Y. Hwang, 1998, The impact of tick size on market quality: an empirical investigation of the stock exchange of Hong-Kong, Working paper, Hong-Kong University. Chordia, T. and A. Subrahmanyam, 1995, Market making, the tick size and payment for order flow: theory and evidence, Journal of Business 68, 543-575. Cordella, T. and T. Foucault, 1999, Minimum price variations, time priority and quotes dynamics, Journal of Financial Intermediation 8, 141-173. Crack, T.F., 1995, Tinkering with ticks: choosing minimum price variation for the US equity markets, Working paper, Indiana University. Declerck, F., 2000, Analyse des meilleures limites du carnet d'ordres: application la Bourse de Paris, Ph.D. Dissertation, Universit de Lille 2. Goldstein, M.A. and K.A. Kavajecz, 2000, Eighths, sixteenths, and market depth: changes in tick size on liquidity provision on the NYSE, Journal of Financial Economics 56, 125-149. Griffiths, M., Smith, B., Turnbull, D. and R. White, 1998, The role of tick size in upstairs trading and downstairs trading, Journal of Financial Intermediation 7, 393-417. Harris, L.E., 1991, Stock price clustering and discreteness, Review of Financial Studies 4, 389- 415. Harris, L.E., 1994, Minimum price variations, discrete bid-ask spreads and quotation sizes, Review of Financial Studies 7, 149-178. Harris, L.E., 1996, Does a large minimum price variation encourage order exposure? Working paper, University of Southern California. 26 Harris, L.E., 1997, Decimalization: a review of the arguments and evidence, Working paper, University of Southern California. Harris, L.E., 1999, Trading in pennies: a survey of the issues, Working paper, University of Southern California. Hasbrouck, J., 1998, Security bid/ask dynamics with discreteness and clustering: simple strategies for modeling and estimation, Working paper, New York University. Huang, R.D. and Stoll H.R., 1999, Tick size, bid-ask spreads and market structure, Working paper, Vanderbilt University. Huson, M., Kim, Y. and U. Mehrotra, 1997, Decimal quotes, market quality, and competition for order flow: evidence from the Toronto Stock Exchange, Working paper, University of Alberta. Jennings, R., 2001, Getting "pennied": the effect of decimalization on traders' willingness to lean on the limit order book at the New York Stock Exchange, Working paper, NYSE. Jones, C.M. and M.L. Lipson, 2001, Sixteenths, direct evidence on institutional execution costs, Journal of Financial Economics 59, 253-278. Jones, C.M., Kaul, G. and M.L. Lipson, 1994a, Information, trading and volatility, Journal of Financial Economics 36, 127-154. Jones, C.M., Kaul, G. and M.L. Lipson, 1994b, Transactions, volume, and volatility, Review of Financial Studies 7, 631-651. Lau, S. and T. McInish, 1995, Reducing tick size on the stock exchange of Singapore, Pacific- Basin Financial Journal 3, 485-496. McInish, T.H. and R.A. Wood, 1995, Hidden limit orders on the NYSE, Journal of Portfolio Management 21, 19-26. McKinnon, G. and H. Nemiroff, 1999, Liquidity and tick size: does decimalization matter? Journal of Financial Research 22, 287-299. 27 Niemeyer, J. and P. Sandas, 1996, Tick size, market liquidity and trading volume: evidence from the Stockholm Stock Exchange, Working paper, Stockholm School of Economics. Porter, D.C. and D.G. Weaver, 1997, Tick size and market quality, Financial Management 26, 5- 26. Ricker, J.P., 1998, Breaking the eighth: sixteenths on the New York Stock Exchange, Working paper, San Francisco University. Ronen, T. and D.G.Weaver, 1998, The effect of tick size on volatility, trader behavior, and market quality, Working paper, Zicklin School of Business. Ronen, T. and D.G. Weaver, 1999, Teenies anyone? the case of the American Stock Exchange, Working paper, Baruch College. Schultz, P., 1998, Stock splits, tick size and sponsorship, Working paper, University of Notre Dame. Seppi, D.J., 1997, Liquidity provision with limit orders and a strategic specialist, Review of Financial Studies 10, 103-150. Van Ness, B.F., Van Ness, R.A. and S.W. Pruitt, 2000, The impact of the reduction in tick increments in major US markets on spreads, depth and volatility, Review of Quantitative Finance and Accounting 15, 153-169. Venkataraman, K., 2001, Automated versus floor trading: an analysis of execution costs on the Paris and New York Exchanges, Journal of Finance 56, 1445-1485. 28 Table 1 Sample firms
Price () Capitalization (billion ) Sector
Panel A: Decrease in the Relative Tick Size Accor 212.90 7.36 Hotels, restaurants and tourism Air Liquide 140.93 12.85 Energy and commodities Alcatel 108.40 24.18 Electricity, electronics Axa / Uap 122.79 44.98 Insurance Canal + 266.20 7.31 Advertising, Media, Multimedia Cap Gmini 149.81 10.90 Data processing and computer applications Dexia France 128.05 5.11 Banking Elf Aquitaine 115.01 29.16 Energy and commodities Eridania Beghin Say 133.02 3.84 Agribusiness Groupe Danone 232.84 17.09 Agribusiness Lafarge 83.57 8.98 Building materials Lagardre 34.08 4.56 Conglomerate Legrand 200.06 5.27 Electricity, electronics LVMH 220.64 19.04 Conglomerate Michelin 39.19 4.60 Automotive equipment Peugeot SA 138.55 7.10 Automotive Pinault Printemps 153.12 20.90 Distribution, general consumer goods Renault 38.04 10.30 Automotive Rhne-Poulenc 43.23 17.57 Chemistry, pharmaceuticals and cosmetic Saint-Gobain 139.05 11.92 Building materials Sanofi 157.79 18.32 Chemistry, pharmaceuticals and cosmetic Sodexho Alliance 156.10 6.50 Environment and public services Suez Lyonnaise des Eaux 173.07 27.05 Environment and public services Thomson-CSF 30.10 5.79 Aerospace and defense Total 104.16 23.70 Energy and commodities Vivendi 236.71 39.26 Advertising, Media, Multimedia
This table reports summary statistics for the CAC 40 stocks around the adoption of the new pricing grid on January 04, 1999. The cross-sectional means and standard errors (in parentheses) are reported. Stock price and market capitalization are year-begin 1999 figures. Trading activity, quoted spread and depth measures are calculated from January to March, 1999 intra-day transactions and quotes data. All averages are equally-weighted means across firms. The intra-day information is from Euronext database. Stocks in the sample are partitioned into four sub-samples based on average daily volume in the pre -event period. The quoted spread is the difference between the euro ask and euro bid quotes. The proportional spread is the quoted spread in euro divided by the midpoint. Ask and bid depth are the euro number of shares available on the best quote.
30 Table 3 The pricing grid in the pre-event and post-event periods
This table presents details related to the pricing grid available before (panel A) and after the switch (panel B) on January 04, 1999. Maximum relative tick size is the ratio of the price increment to the minimum share price for each category. Minimum relative tick size is the ratio of the price increment to the maximum share price for each category.
This table provides the number of CAC 40 stocks traded in each tick size category during pre- event period from October to December, 1998 (panel A) and during post-event period from January to March 15, 1999 (panel B). The proportion of one-tick spreads is the ratio of the number of one-tick spreads to the total number of spreads. Information is from Euronext database.
32 Table 5 The new pricing grid and changes in the relative tick size
This table reports liquidity measures by price level for the CAC 40 stocks around the adoption of the new pricing grid on January 04, 1999. All averages are equally-weighted means across firms. The intra-day information is from Euronext database. The proportion of one-tick spreads is the ratio of the number of one-tick spread to the total number of quoted spread. The quoted depth is the euro number of shares available on the best bid and ask quotes. Results are based on standard comparison t -test.
*** denotes significance at 1% level ** denotes significance at 5% level
33 Table 6 The new pricing grid, decreasing and increasing tick size samples (firm by firm analysis)
This table reports average measures of relative tick size around the adoption of the new pricing grid on January 04, 1999. Some stocks benefited from a reduction of their relative tick size (DTS sample) but some others experienced the effects of an increasing tick size (ITS sample).
This table provides data on spreads measures from the limit order book for the CAC40 stocks during periods from October to December, 1998 (pre-event period) and January to March, 1999 (post-event period). The cross-sectional means and standard errors (in parentheses) are reported. The average change between the two periods is also reported. Statistics are time -weighted for each firm and equally weighted across firms. Stocks in the sample are partitioned into four sub-samples based on average daily volume in the pre-event period. The proportional quoted spread is the quoted spread divided by the midpoint. The proportion of one-tick spreads is the ratio of the number of one-tick spreads to the total number of spreads. Quoted spread to tick ratio is the average number of tick included in the average euro bid-ask spread. Tests of significance are paired t-tests for the full sample and the non-parametric Wilcoxon test for each sub-sample.
All firms Volume quartiles Mean Std 1 - Low 2 3 4 - High Panel A: Quoted proportional spread Before 0.3032% (0.1027%) 0.4287% 0.2930% 0.2657% 0.2106% After decrease 0.3075% (0.0979%) 0.4187% 0.3367% 0.2629% 0.2006% % change 0.00% -0.01% 0.04% 0.00% -0.01%
*** denotes significance at 1% level ** denotes significance at 5% level
35 Table 8 The daily order submission across types - Decreasing tick size sub-sample
This table presents some summary statistics on the distribution of daily order submissions across order types and order exposure related to the decreasing tick size sample for the CAC 40 stocks during periods October to December, 1998 (pre -event period) and January to March, 1999 (post-event period). The cross-sectional means and standard errors (in parentheses) are reported. All averages are equally-weighted means across firms. The intra-day information is from Euronext database. Stocks in the sample are partitioned into two sub-samples based on buy or sell side. Total daily order flow for each type of order (market or limit order), hidden quantity and order state and also as a proportion of total order submissions is obtained from intra-day orders. Tests of significance are paired t -tests.
Buys Sells Before After Before After Panel A: All orders Number of orders 1,144 (601) 1,312 *** (533) 1,179 (642) 1,331 *** (629) Panel B: Executed order activity Number of orders 673 (423) 767 ** (359) 676 (422) 714 (378) Percentage of all orders 56.13% (5.39%) 56.33% (3.98%) 54.46% (5.32%) 51.05% *** (5.20%) Panel C: Hidden quantity Number of orders 138 (43) 169 *** (53) 144 (44) 174 *** (48) Percentage of all orders 13.77% (3.91%) 13.96% (3.30%) 14.09% (3.62%) 14.63% (2.97%) Percentage of order volume 70.75% (2.06%) 71.94% *** (2.37%) 72.48% (2.18%) 73.01% (2.09%) Panel D: Order type Limit orders Number of orders 985 (485) 1143 *** (440) 990 (511) 1144 *** (516) Percentage of all orders 86.85% (2.40%) 87.74% *** (2.24%) 85.25% (3.81%) 86.70% *** (3.55%) Market orders Number of orders 136 (100) 145 (81) 150 (107) 148 (94) Percentage of all orders 11.26% (2.08%) 10.55% ** (2.04%) 11.79% (3.10%) 10.65% *** (2.96%) Panel E: Order state Cancelled 125 (41) 129 (37) 119 (44) 138 *** (42) Modified 189 (63) 224 *** (66) 177 (56) 215 *** (51) Blank 168 (98) 174 (86) 170 (94) 249 *** (107) Filled 673 (423) 767 ** (359) 676 (422) 714 (378)
*** denotes significance at 1% level ** denotes significance at 5% level 36 Table 9 The daily order submission across types - Increasing tick size sub-sample
This table presents some summary statistics on the distribution of daily order submissions across order types and order exposure related to the increasing tick size sample for the CAC 40 stocks during periods October to December, 1998 (pre -event period) and January to March, 1999 (post-event period). The cross-sectional means and standard errors (in parentheses) are reported. All averages are equally-weighted means across firms. The intra-day information is from Euronext database. Stocks in the sample are partitioned into two sub-samples based on buy or sell side. Total daily order flow for each type of orders (market or limit order), hidden quantity and order state and also as a proportion of total order submissions is obtained from intra-day orders. Tests of significance are paired t-tests.
Buys Sells Before After Before After Panel A: All orders Number of orders 1,073 (485) 1,274 ** (675) 1,584 (1,645) 2,094 (2,881) Panel B: Executed order activity Number of orders 548 (287) 714 *** (400) 949 (1,147) 1,280 (2,025) Percentage of all orders 49.50% (4.44%) 54.50% *** (3.47%) 52.93% (5.58%) 53.29% (5.98%) Panel C: Hidden quantity Number of orders 183 (58) 176 (76) 169 (49) 165 (58) Percentage of all orders 18.24% (3.17%) 15.02% *** (3.43%) 15.57% (4.14%) 12.99% *** (3.86%) Percentage of order volume 72.54% (1.66%) 71.37% ** (2.90%) 74.30% (1.75%) 72.36% *** (2.09%) Panel D: Order type Limit orders Number of orders 947 (412) 1 125 ** (583) 1 290 (1236) 1 642 (1943) Percentage of all orders 88.81% (1.87%) 88.88% (2.19%) 84.42% (5.77%) 85.31% (6.68%) Market orders Number of orders 108 (63) 128 ** (81) 239 (347) 355 (736) Percentage of all orders 9.59% (1.54%) 9.52% (1.92%) 12.48% (4.56%) 11.71% (5.12%) Panel E: Order state Cancelled 134 (40) 144 (69) 132 (71) 154 *** (86) Modified 260 (106) 240 (130) 217 (77) 220 (98) Blank 131 (69) 152 (78) 243 (154) 338 * (210) Filled 548 (287) 714 *** (400) 949 (1147) 1,280 (2025)
*** denotes significance at 1% level ** denotes significance at 5% level
37 Table 10 Order aggressiveness
This table provides data on order aggressiveness for the CAC 40 stocks during periods October to December, 1998 (pre-event period) and January to March, 1999 (post-event period). Proportions on each class are reported. The average change between the two periods is also reported. The designation in the book means that the order is submitted at a lesser price than the best price on the book. On the best quote means that the order is submitted at a price equal to the best price. Inside the spread means that the order is submitted at a greater price than the best price. And finally, marketable limit order denotes that the limit order will be immediately executed at the best bid price for a sell order and at the best ask price for a buy order. Statistics are time -weighted for each firm and equally weighted across firms.
Decreasing tick size sub-sample Increasing tick size sub-sample 1998 1999 % change 1998 1999 % change Marketable orders 29.40% 26.01% - 3.39% ** 26.63% 27.26% 0.63% * Inside the spread 12.69% 19.38% 6.69% *** 19.84% 15.61% - 4.23% *** On the best quotes 8.01% 4.22% - 3.78% *** 4.41% 4.36% - 0.05% In the book 49.91% 50.39% 0.48% * 49.12% 52.77% 3.65% **
*** denotes significance at 1% level ** denotes significance at 5% level * denotes significance at 10% level 38 Table 11 The quoted depth
This table provides data on liquidity measures from the limit order book for the CAC 40 stocks during periods October to December, 1998 (pre -event period) and January to March, 1999 (post-event period). The cross-sectional means and standard errors (in parentheses) are reported. The average change between the two periods is also reported. Statistics are time -weighted for each firm and equally weighted across firms. Stocks in the sample are partitioned into four sub-samples based on average daily volume in the pre-event period. Average depth is the euro number of shares available on the best quotes. Tests of significance are paired t-tests for the full sample and the non- parametric Wilcoxon test for each sub-sample.
All firms Volume quartiles Mean Std 1 - Low 2 3 4 - High Panel A: Average of Euro Ask Depth Before 81,088.53 (58,838) 31,543.80 67,699.16 61,733.59 174,860.84 After decrease 62,189.96 (32,896) 28,267.04 56,335.80 60,957.29 109,059.00 % change -23.31% ** -10.39% * -16.79% ** -1.26% -37.63% **
Before 46,929.01 (22,749) 29,252.24 39,313.85 50,965.78 69,529.76 After increase 68,876.51 (41,052) 34,215.65 54,793.26 80,901.98 109,603.66 % change 46.77% *** 16.97% 39.37% * 58.74% 57.64% Panel B: Average of Euro Bid Depth Before 73,084.71 (53,082) 29,393.69 59,500.27 55,876.57 157,718.19 After decrease 55,628.07 (29,543) 24,918.61 48,632.83 55,345.46 98,780.72 % change -23.89% ** -15.22% ** -18.26% * -0.95% -37.37% **
*** denotes significance at 1% level ** denotes significance at 5% level * denotes significance at 10% level
39 Table 12 The depth to trade size ratio
This table provides depth to trade size measures from the limit order book available at each trade for the CAC 40 stocks during periods October to December, 1998 (pre-event period) and January to March, 1999 (post-event period). The analysis is conducted on all trades. The cross-sectional means and standard errors (in parentheses) are reported. The average change between the two periods is also reported. Statistics are time-weighted for each firm and equally weighted across firms. The depth to trade size ratio indicates the number of times the average trade size is contained in the average quoted depth.
Ratio depth to trade size 1998 1999 Mean Std Mean Std % change Decreasing tick size sub-sample 38.52 (21.34) 20.54 (6.44) -46.69% *** Increasing tick size sub-sample 30.85 (20.31) 26.18 (20.40) -15.12%
*** denotes significance at 1% level 40 Table 13 Institutional block trading and weighted average bid-ask spread
This table analyzes changes in the weighted average bid-ask spread for the CAC 40 stocks during periods October to December, 1998 (pre-event period) and January to March 08, 1999 (post-event period) 18 . The cross-sectional means and standard errors (in parentheses) are reported. The average change between the two periods is also reported. All averages are time-weighted for each firm and equally weighted across firms. The intra-day information is from Euronext database. The Weighted Average Bid-Ask Spread (WAS) represents the price for blocks that exceed to the Normal Market Size (NMS). Euronext calculates the WAS for a given quantity of shares in real time by taking the average of bid and ask prices for all orders placed in the central order book, weighted by the number of shares displayed at the successive bids and asks. Tests of significance are paired t -tests.
18 We cannot take into account data after March 08, 1999 because some stocks experienced a NMS change. 41 Figure 1 Changes in the relative tick size by price range
These figures report measures by price level for the CAC 40 stocks around the adoption of the new pricing grid on January 04, 1999. We observe an increase in the relative tick size for prices between 50 and 76.25 and for prices above 500. We observe a decrease in the relative tick size for all other price ranges.
Quoted depth () 0 50 100 150 200 250 300 350 400 450 ]15.25;30 ] ]30;50 ] ]50;76.25 ] ]76.25;100 ] ]100;143 ] ]143;200 ] ]200;500 ] > 500 Price range E u r o
d e p t h
( 0 0 0 ) 1998 1999
42 Figure 2 Trading costs by volume quartile
This table provides data on liquidity measures from the limit order book for the CAC 40 stocks during periods October to December, 1998 (pre-event period) and January to March, 1999 (post-event period).
Panel A: the relative spread and the number of orders on the best quotes
43 Panel B: euro ask and bid depth on the best quotes