Block Trades and Associated Price Impact: International Evidence On The Two Asymmetries
Block Trades and Associated Price Impact: International Evidence On The Two Asymmetries
Block Trades and Associated Price Impact: International Evidence On The Two Asymmetries
a
Finance Discipline, Faculty of Economics and Business, University of Sydney,
NSW, 2006, Australia.
b
School of Accounting and Finance, Faculty of Commerce, University of Wollongong,
NSW, 2522, Australia
Abstract
Previous literature has documented two interesting findings related to block trades,
predominantly in the US. The first is a directional asymmetry following block trades,
whereby continuations follow purchases, and reversals follow sales. The second is a
permanent price impact asymmetry, with block purchases having greater impact on
stock prices. This paper examines if these two asymmetries are robust across
confirm the existence of the directional asymmetry in four markets, and a permanent
price impact asymmetry in all five markets. Transition to quote data resolves the
∗
Corresponding Author. Finance Discipline, School of Business, University of Sydney, NSW, 2006,
Australia. Tel: +61 2 9227 0889 Fax: +61 2 9351 6461 Email: afri1432@ usyd.edu.au. This research
has been funded by the Cooperative Research Centre for Technology Enabled Capital Markets. The
authors thank SIRCA for the provision of data and Joel Fabre, Vito Mollica, Teddy Oetomo, Joel Grant
and Zaffar Subedar for useful comments.
1
1. Introduction
The price impact of block trades has been an extensively studied area in extant
understanding of the impact that their trades have. In this paper, we aim to gain this
international setting.
block trades impact stock prices. Going as far back as Kraus and Stoll (1972), when a
large parcel of stock is either purchased or sold, the underlying stock price has been
shown to experience some form of price adjustment. The degree of price adjustment,
however, has varied. Holthausen, Leftwich and Mayers (1987) show that the impact
to a block purchase on the NYSE was 161 basis points. Chan and Lakonishok (1993)
find the corresponding impact to be only 34 basis points. Gemmill (1996), on the
Aitken and Frino (1996) find an impact of 51 basis points. The degree of price
Although the concept of block trades impacting on the stock price is accepted,
the exact nature of this impact is far from settled. There are two main issues that have
arisen in the literature. The first, and perhaps most perplexing, is the directional
asymmetry following block transactions. Stock prices continue to rise after block
purchases, whilst there is a partial reversal following block sales.2 The second, and
1
Over 70 percent of trading on stock markets worldwide is institutionally based (Jain, 2003).
2
See Holthausen et al. (1987, 1990), Choe, McInish and Wood (1992), Chan and Lakonishok (1993,
1995), Keim and Madhavan (1995), Aitken and Frino (1996) and Bonser-Neal, Linnan and Neal
(1999).
2
arguably most important issue, is that of the permanent price impact asymmetry. The
overall impact to block purchases is greater than for block sales.3 It is these two
(Holthausen et al. 1987, p90 and Chan and Lakonishok, 1993, p175) and a “key
puzzle” (Chan and Lakonishok, 1993, p197). The apparent price continuation
following block purchases, and partial reversal following block sales, suggests that
block sellers pay a liquidity premium, while block buyers do not. The literature has
effects may be stronger for purchases than sales; specialists are more willing to
accommodate sales rather than purchases; and short-term imbalances in supply and
demand (see Chan & Lakonishok, 1993 and Saar, 2001). To date, none of these
attention. While it has been consistently shown that block buys move prices up, and
block sales move prices down, the degree of price movement, together with the
continuation / reversal just described, has lead to a disparity in measured price effects,
with purchases having a greater impact than sales. Saar (2001) argues that the history
of the price performance of the stock has significant influence. He finds that a long
price run-up prior to a block transaction can actually lead to a negative asymmetry
(sales move more than buys). Chiyachantana, Jain, Jiang and Wood (2004) find that
the performance of the market in which the stock trades is most important. During a
bull period, block purchases will have greater price impact, whilst bearish periods will
3
The documentation of these two asymmetries has predominantly been based
on data captured from the NYSE. Studies including Tinic and West (1974), Neal
(1992), Pagano and Roell (1996) and Nimalendran and Petrella (2003) demonstrate
that depth, liquidity and execution costs vary considerably across specialist based,
auction based and dealer based trading systems. If the measurement of price impact is
affected by the trading system employed, then the existence and nature of the two
The few studies conducted outside of the NYSE already exhibit this variation.
Gemmill (1996) on the London Stock Exchange finds little to no permanent price
impact for block sales, and reversals following both purchases and sales. Aitken and
Frino (1996) on the ASX find that the continuation and reversal pattern following
block purchases and sales exists even in an event-study methodology. As the NYSE
employs a specialist market structure, the asymmetrical stock price responses may be
The aim of paper is to explore if these two asymmetries are robust across
impact of block trades from five of the largest exchanges, covering the three major
trading regions (North America, Europe and Asia-Pacific), we are able to determine if
(i) the directional asymmetry exists in a range of markets, and (ii) the existence and
nature of any permanent price impact asymmetry in these same markets. Going
dataset and subsequent sample used. Section 3 sets out the research design, Section 4
4
presents the empirical results, while Section 5 presents several additional tests.
from trades executed in the following markets - New York Stock Exchange (NYSE)
in the US, the London Stock Exchange (LSE), Xetra (from Germany) and Euronext in
Europe and the Tokyo Stock Exchange (TSE) in the Asia-Pacific Region.4 These
markets are selected as they constitute the largest markets, based on market turnover,
Sample trades are limited to the largest 200 stocks, by total turnover, in each
ensure consistency block trades are defined as the largest one percent of on-market
transactions for each stock, by volume, in each calendar year, over the sample period.
The data used in this paper is sourced from a Reuter’s database for the period January
1, 1999 to December 31, 2002. In order to convert all monetary values into US
dollars, daily exchange rate data is sourced from DataStream over the relevant sample
period.
are classified as being buyer or seller initiated using the trade direction algorithm set
out in Ellis, Michaely and O’Hara (2000). Trades which execute at the ask quote are
classified as buyer initiated. Similarly, trades which execute at the bid quote are
classified as seller initiated. Trades which execute at neither the ask or the bid are
4
Euronext is a conglomeration of the Paris, Brussels, Amsterdam and Lisbon Stock Exchanges.
5
classified using the tick test.5 If the last price change was an uptick, the trade is buyer
initiated; if the last price change was a downtick, the trade is seller initiated. If the
last price change was a zero tick, then the block price is compared to the price two
(and then three) trades before. If the third trade price does not result in a
classification, the trade is then removed from the sample. Bid and ask quotes at the
time the trade was executed are used as both Bessembinder (2003) and Peterson and
Sirri (2003) demonstrate the accuracy of trade classification rules are improved as the
To assist in the comparability of results, all trade values are converted into US
Dollars using day-average exchange rates. Within several tables throughout the
remainder of this paper, results are presented across sub-divisions of the sample. The
largest 200 stocks (by turnover) are further segregated into 4 equal groups of 50
stocks, for each calendar year, again based on turnover. For example, columns 2 and
3 in Table 1 represent the descriptive statistics for the least liquid quartile of stocks in
the sample, for buyer- and seller- initiated transactions respectively. The final two
columns represent results for the entire sample of 200 stocks (again, buyer- and seller-
initiated, respectively).
Table 1 presents descriptive statistics for block trades for all of the five
markets included in the sample. Looking at the number of trades analysed, the NYSE
has approximately 967,000 purchase and 675,000 sale transactions over the four year
period. The LSE has approximately 530,000 purchases and 535,000 sales, Xetra has
374,000 purchases and 330,000 sales, while Euronext has 501,000 purchase and
481,000 sale transactions. The volume of trading in the Asia-Pacific region is much
more subdued, with the TSE having only 69,000 purchase and 58,000 sale
5
Lee and Ready (1991) report a high degree of correlation in trade direction classified by the tick rule
and quote based methods.
6
transactions, although only two years of data (from January 1, 2001 to December 31,
volume of block trading is concentrated in the largest quartile. For all exchanges,
apart from the NYSE, approximately 50 percent of trading occurs in the largest
quartile of the sample. Still looking at the number of trades, apart from the NYSE (in
which there are approximately 290,000 more purchase transactions), there is a strong
parity in the number of buyer- and seller- initiated transactions within each market.
Turning to the average size of block trades, the average trade value for NYSE
trades is approximately $2.6 million. Looking at the European markets, the average
trade values for the LSE are approximately $1.8 million, for Xetra $1.6 million and
$1.25 million for Euronext. In the Asia-Pacific region, the TSE has average trade
values of approximately $1.7 million, which is larger than for both Xetra and
Euronext.
Finally, the spreads immediately prior to each block transaction indicates that
Xetra has the tightest spreads, with averages around the 0.19 percent level. Euronext
has average spreads of approximately 0.22 percent. The average spread on the NYSE
measures approximately 0.25 percent. The average spreads for both the LSE and TSE
are over 0.3 percent. Looking across the size sub-divisions, the most liquid stocks
have tighter spreads, with sales in the most liquid group on Xetra having an average
Overall, it can be seen that the NYSE is the dominant exchange, both in terms
of trade frequency and trade size. The three European markets are relatively similar in
terms of their trading activities. The TSE, whilst having significantly fewer
transactions, has trade sizes directly comparable to the European markets. Although
7
the largest 200 stocks within each of the markets are not in general affected by
liquidity concerns, the largest 50 stocks within each market clearly dominate all other
stocks, both in terms of trading frequency and trade size. Most importantly, buyer-
and seller- initiated transactions, across all size sub-divisions, and across all markets,
are relatively similar. Any differences between these types of transactions are not
3. Research Design
calculating the price effects, the pre-execution benchmark is taken to be the opening
transaction price on the day of the block trade. The post-execution benchmark is
taken to be the closing transaction price on the day of the block trade.
With these pre- and post- execution benchmarks, the price impacts are
calculated as follows.
In order to test whether bid-ask effects influence price behaviour following block
trades, the analysis based on transaction prices is repeated using prevailing quotes
8
the ask quote, and block sales at the bid quote, the quotes in the above three equations
represent either the opening ask (bid) or closing ask (bid) for block purchases (sales).
Reliance on quote returns should eliminate any potential bid-ask effects in transaction
price data.
4. Empirical Results
Panel A of Table 2 presents the results for the total effect calculated using
transaction price data for the largest 200 stocks in each exchange. For the NYSE, the
total effect indicates significantly positive returns for buyer initiated transactions, and
significantly negative returns for seller initiated transactions. The reaction to sales
appears to be greater than for purchases (-0.5488 percent versus 0.1803 percent), and
The results for the European markets are quite similar to the NYSE.
Purchases are associated with significantly positive returns, sales with significantly
negative returns. Seller initiated transactions again appear to have larger impacts, and
this is most evident in Xetra (0.3357 percent for buys and -0.6650 percent for sales).
This pattern holds in all but the smallest size sub-division of the samples. Moving on
to the TSE, as with the other four markets, the significantly positive / negative
reaction to buyer- / seller- initiated transactions is evident. Unlike with the other
markets, the TSE indicates similar reactions for both purchases and sales, and this
One of the main aims of this paper is to explore the directional asymmetry that
has been documented previously. Panel B of Table 2 presents these results, again for
9
the largest 200 stocks. Starting with the NYSE, there are continuations in the stock
price following buyer initiated transactions (significantly positive returns), and partial
reversals in the stock price following seller initiated transactions (again, significantly
positive returns). This continuation / reversal pattern is consistent across the four size
sub-divisions. This confirms that the NYSE exhibits the asymmetric behaviour.
The results for two of the three European markets have a strong resemblance
with the NYSE results. For Xetra and Euronext, the continuation following
purchases, and partial reversal following sales is evident (although the continuation
Euronext). The LSE results indicate significant reversals following both buyer- and
seller- initiated transactions, which hold in each of the four size sub-divisions. This
was expected, however, as Gemmill (1996) found very similar results when analysing
the London market. Finally, for the TSE, the continuation following purchases and
partial reversal following sales is evident in the aggregated results, although some of
the four size sub-groups indicate that these continuations / reversals are insignificantly
impact, and explore if any asymmetries exist, as well as the direction they occur.
Panel C of Table 2 presents the results for the permanent effect. The difference figure
with the permanent effect in Table 2 is calculated by subtracting the absolute value of
the permanent effect for sales from the absolute value of the permanent effect for
purchases. A positive difference thus indicates that purchases have larger permanent
effects, whilst a negative difference indicates that sales have larger permanent effects.
Commencing with the NYSE, purchases are associated with a permanent effect of
0.2471 percent, while sales have a -0.4633 percent impact. The difference of -0.2162
10
percent is significantly different from zero, indicating that block sales have larger
asymmetry for all markets. The results for the LSE indicate that the mean permanent
effect for buyer initiated transactions is significantly negative, while the mean
permanent effect for seller initiated transactions is significantly positive. Even with
this unusual finding, the test of the difference indicates that block sales have
significantly larger permanent effects.6 With the other two European markets, both
Xetra and Euronext indicate a permanent price impact asymmetry, with seller initiated
transactions having greater impact. This holds across all size sub-divisions, except for
the smallest group on Euronext which has an asymmetry favouring block purchases.
Moving on to the Asia-Pacific region, the TSE has significantly positive and
negative permanent effects for block purchases and sales, respectively. Unlike with
the other four markets, the reaction to purchases is significantly larger than the
reaction to sales. This holds in all but the second size sub-division (which has a
documented is clearly evident in four of the five markets analysed. With the market
that does not conform, the LSE provides results which are consistent with previous
findings. The permanent price impact asymmetry is clearly evident in some form
(either favouring buys or sells) in all five stock exchanges. The asymmetries are
6
Effectively, the positive impact for block sales is smaller than the negative impact for block
purchases.
11
4.2 Quote Data Results
To control for any microstructural biases in the transaction data, the same
three price effects are recalculated using ask quotes for purchases, and bid quotes for
sales. These results, calculated for the largest 200 stocks in each exchange, are
presented in Table 3. The total effect in Panel A is significantly positive for block
purchases, and significantly negative for block sales, in each of the five markets. This
holds across the four size sub-divisions, and is consistent with the transaction price
moving to quote data. For the NYSE, the continuations following purchases remain
with quote data, although the magnitude of the continuation is now greater. However,
the reversals following block sales have transformed into continuations. The
for sales. These continuations are robust across the four size sub-divisions. There are
now symmetrical stock price reactions following block purchases and sales on the
NYSE.
The three European markets are remarkably similar to the NYSE. For Xetra
and Euronext, the reversals following sales have turned into continuations measuring
purchases remain, as with transaction prices. The reversals following buys and sells
on the LSE are now significant continuations, measuring 0.1051 percent for
purchases, and -0.0581 percent for sales. These continuations hold across all four size
sub-divisions. Finally, the reversals following block sales on the TSE have
transformed into continuations measuring -0.1759 percent and this holds in all four
12
size groups. Overall, when quote data is used, the stock price continues to rise after
block purchases, and fall after block sales, for all five exchanges.
As the temporary effect has changed considerably, the permanent effect (and
associated asymmetry) may also have changed. The results are presented in Panel C
of Table 3. The difference between buys and sells is as discussed previously. For the
NYSE, the asymmetry favours block sales, consistent with the transaction price
results, and holds across all four size sub-divisions. Both Xetra and Euronext, as with
transaction prices, indicate that block sales have larger permanent effects.
The permanent effect for block purchases on the LSE now show significantly
positive returns (0.4037 percent), while block sales have significantly negative returns
(-0.4740 percent), leading to an asymmetry which favours block sales. The size sub-
division results for the three European markets are consistent. The TSE, in contrast to
the transaction price results, now shows that block sales are associated with
significantly larger permanent effects, which holds in all but the smallest size group.
To summarise, the transition to quote data has impacted on the results. All of
the five markets now indicate that stock price continuations follow both block
purchases and block sales. The directional asymmetry is now redundant. However,
the permanent price impact asymmetry still remains. In all five markets, block sales
are associated with significantly larger permanent price effects. The removal of
microstructural biases from return measurement has lead to significant changes in the
13
5. Additional Tests
A study by Chiyachantana, Jain, Jiang and Wood (2004) shows how the
behaviour of the underlying market affects the price impact of block trades.
Effectively they showed that during bull markets, block purchases will have larger
permanent price effects, whilst during bear markets, block sales will have larger
permanent effects. To explore if the trades included in the current sample react in
similar ways, we collected market return data (using relevant MSCI indexes on
DataStream) for each of the five exchanges. We then calculate 48 monthly returns,
and place block trades executed in months which experienced positive returns in one
sample, and block trades executed in months which experienced negative returns in
the other sample. The total, temporary and permanent effects are then calculated
using bid and ask quotes for the bull and bear market samples separately. The
average value) are presented in the first few columns, and are consistent with the
original summary statistics presented in Table 1. Panel A presents the results for bull
market months. The total effects are all positive for purchases and negative for sales.
There are stock price continuations following purchases and sales, although the
continuation following sales on the NYSE is not significantly different from zero. For
the NYSE and the three European markets, the permanent effect for block purchases
While the TSE still indicates that sales have larger permanent effects, the
magnitude of the difference (-0.0283 percent) is much smaller than the original results
14
(-0.0951 percent) in Table 3. Panel B presents the results for bear market months. In
each of the five markets analysed, block sales have significantly larger permanent
price effects. Consistent with Chiyachantana, Jain, Jiang and Wood (2004), block
purchases have larger and block sales have smaller price impact during bull markets,
while block sales have larger and block purchases have smaller price impact during
bear markets.7
selection of the largest one percent of trades (by volume) can be considered a
each trading day in the sample, the average daily turnover over the previous three
months is calculated. Each trade value is then compared to this average daily
turnover, and if the trade represents at least 2.5 percent of average daily turnover, it is
classified a block trade. The three price effects are then calculated for this new
reduced (for example, the NYSE originally had 1,643,155 block trades, and now has
338,350 block trades). Accordingly, the average volumes and values have increased
considerably (for example, the average sell value on Xetra is now approximately $7.1
million, up from $1.6 million). The total effects are still positive for purchases and
negative for sales. There are still significantly positive continuations following
15
permanent price impact asymmetry has changed, with the LSE now indicating that
block purchases have larger permanent price effects (the NYSE, Xetra and Euronext
difference between the permanent effect for block purchases and sales on the TSE.
Overall, the results seem relatively robust with the alternative block trade definition.
All price impacts are calculated using either the opening or closing quotes. It
may be that in the time between the open and the trade, or between the trade and the
close, other market activities are also affecting the stock price. To isolate the impact
of each trade, we narrow the event window around each order. We substitute the
opening quotes with the quotes in existence five trades and ten trades before the block
trade, and the closing quotes with the quotes in existence five trades and ten trades
after the block trade.8 We then recalculate all price impacts using these ‘new’ pre-
The five trade results are presented in Panel A. As the sample is identical to
that used in Table 1 through Table 3, no summary statistics are reported. The total
effect, as with all cases, is significantly positive for purchases and significantly
negative for sales. There are significantly positive continuations following purchases
larger for sales, although the difference is insignificantly different from zero for both
Euronext and the TSE. The ten trade results (in Panel B) are very similar, except that
the difference between the permanent effect for purchases and sales is significantly
8
If the order is placed in the first five or ten trades of the day, or executes in the final five or ten trades
of the day, the opening and closing quotes are left in place.
16
negative for all five markets. The choice of pre- and post- execution benchmarks is
The classification of trades into buys and sells noted the possibility of trades
not occurring at the bid or ask quote. Thus, some of the block trades may have
executed within the quotes. Calculating the price effects using the ask quote for
purchases and the bid quote for sales may not always be accurate. To make a more
general microstructural adjustment, the three price effects are recalculated using the
midpoint of the opening, block and closing quotes. The results are presented in Table
7. The results are very similar to the earlier findings. The total effect is positive
(negative) for buys (sells). There are continuations following both buys and sells.
The permanent price impact asymmetry favours block sales for each of the five
markets. The use of quote midpoints leads to the same conclusions as with bid and
ask quotes.
6. Summary
Much of the previous literature analysing the price impact of block trades has
utilised data sourced from trading on the NYSE, which operates predominantly as a
specialist market. Several previous studies demonstrate how depth, liquidity and
execution costs vary considerably across specialist based, auction based and dealer
based trading systems. To explore how these factors affect the price impact of block
trades, data encompassing five of the largest stock exchanges, from the three major
trading regions around the world - North America, Europe and Asia-Pacific, is used to
17
The overall findings in this paper support the directional asymmetry in four of
the five markets. In the other market, the LSE indicates that a partial reversal in the
stock price follows both block purchase and sale transactions. The permanent price
markets. The stock price continues to rise after purchases and fall after sales. Even
with continuations following both purchases and sales, the permanent price impact
asymmetry still remains, favouring block sales in all five markets. These findings are
18
References
Block Trades on the Australian Stock Exchange: A Note, Abacus 32, 54-61.
Chiyachantana, C.N., P.K. Jain, C. Jiang and R.A. Wood, 2004, International
Evidence on Institutional Trading Behaviour and Price Impact, Journal of Finance 59,
869-898.
Choe, H., T.H. McInish and R.A. Wood, 1992, Market Microstructure Effects
Memphis.
the London Stock Exchange Under Different Publication Rules, Journal of Finance
51, 1765-1790.
19
Holthausen, R., R. Leftwich and D. Mayers, 1990, Large Block Transactions,
the Speed of Response, and Temporary and Permanent Stock-Price Effects, Journal of
Holthausen, R., R. Leftwich and D. Mayers, 1987, The Effect of Large Block
Keim, D.B. and A. Madhavan, 1997, Transaction Costs and Investment Style:
Keim, D.B. and A. Madhavan, 1996, The Upstairs Market for Large-Block
Studies 9, 1-36.
Keim, D.B. and A. Madhavan, 1995, The Anatomy of the Trading Process:
Kraus, A. and H.R. Stoll, 1972, Price Impacts of Block Trading on the New
Lee, C.M.C. and M.J. Ready, 1991, Inferring Trade Direction from Intraday
Neal, M., 1992, A comparison of transaction costs between market maker and
20
Nimalendran, M., and G. Petrella, 2003, Do ‘thinly-traded’ stocks benefit from
auction and dealer markets with informed trading, Journal of Finance 51, 579-611.
Estimates Using Trade and Quote Data, Journal of Financial Markets 6, 259-280.
Tinic, S., and R. West, 1974, Marketability of common stocks in Canada and
the USA: A comparison of agent versus dealer dominated markets, Journal of Finance
29, 729-746.
21
Table 1
Summary Statistics
This table reports the number of observations, mean and median share volume, trade values and proportional spreads prior to each trade for the
block trades analysed in this paper. Statistics are presented for both buyer- and seller- initiated transactions for the largest 200 stocks, by
turnover, for each year, in each of the five largest stock exchanges covering the three major trading regions around the world - North America,
Europe and Asia-Pacific. The largest 200 stocks in each calendar year are further divided into quartiles, again based on turnover. Group 1
(smallest) represents stocks ranked 151 to 200, Group 2 stocks 101 to 150, Group 3 stocks 51 to 100 and Group 4 (largest) stocks 1 to 50. The
final two columns represent the results for all 200 stocks combined. Block trades are defined as the largest one percent of on-market
transactions, by volume, in each of the largest 200 stocks, in each calendar year, for each market, for the period January 1, 1999 to December 31,
2002. All monetary values have been converted to US dollars using day-average exchange rates.
22
Table 1, continued
Xetra (German)
No. Trades 12,446 11,915 25,706 22,588 73,986 62,620 262,152 233,615 374,290 330,738
Avg. Volume 7,434 8,071 10,306 11,874 20,156 27,186 27,362 31,221 24,104 28,301
Med. Volume 4,395 4,628 6,406 6,442 11,515 11,500 10,900 10,400 10,167 10,000
Avg. Value 229,279 240,671 364,752 398,321 792,490 1,025,893 1,908,933 1,977,341 1,526,340 1,626,795
Med. Value 155,431 150,443 218,621 210,191 403,826 384,266 724,718 685,030 595,493 570,980
Avg. Spread 0.6595 0.6328 0.4372 0.4296 0.2535 0.2416 0.1371 0.1272 0.1981 0.1877
Med. Spread 0.5526 0.5115 0.3231 0.3150 0.1698 0.1668 0.0892 0.0869 0.1128 0.1095
Euronext
No. Trades 19,931 19,251 51,119 48,068 119,872 114,846 310,911 299,141 501,833 481,306
Avg. Volume 18,685 20,867 13,008 12,685 17,887 18,190 24,295 25,424 21,392 22,244
Med. Volume 3,427 3,867 5,000 5,000 8,442 8,591 10,000 10,000 9,607 9,827
Avg. Value 355,244 401,025 409,432 426,063 774,856 791,172 1,627,288 1,635,200 1,249,092 1,263,683
Med. Value 148,005 156,101 180,076 176,955 344,521 342,163 660,773 650,617 491,212 487,534
Avg. Spread 0.6022 0.6005 0.3664 0.3779 0.2356 0.2415 0.1569 0.1614 0.2147 0.2197
Med. Spread 0.4357 0.4382 0.2601 0.2699 0.1658 0.1709 0.1115 0.1174 0.1371 0.1416
TSE
No. Trades 11,453 9,325 13,131 10,708 14,831 11,995 29,303 25,906 68,718 57,934
Avg. Volume 134,136 134,091 125,091 137,501 123,121 125,816 128,853 120,855 127,777 127,089
Med. Volume 80,000 77,000 49,000 50,000 64,000 64,400 58,000 56,100 60,150 60,000
Avg. Value 756,426 710,831 949,336 936,974 1,272,381 1,204,570 2,610,829 2,488,695 1,695,406 1,649,853
Med. Value 519,056 486,852 648,411 617,847 845,325 796,867 1,446,024 1,393,273 897,260 864,320
Avg. Spread 0.3561 0.3011 0.3554 0.2955 0.3293 0.2796 0.3226 0.2871 0.3359 0.2894
Med. Spread 0.2350 0.2210 0.2090 0.1889 0.2099 0.1982 0.2153 0.2094 0.2167 0.2055
23
Table 2
Price Effects Surrounding Block Trades Calculated Using Transaction Prices
This table reports the total effect (Panel A), temporary effect (Panel B) and the permanent effect (Panel C), all in percentage, calculated using
transaction prices. Results are presented for both buyer- and seller- initiated transactions for the largest 200 stocks, by turnover, for each year, in
each of the five largest stock exchanges covering the three major trading regions around the world - North America, Europe and Asia-Pacific.
The largest 200 stocks in each calendar year are further divided into quartiles, again based on turnover. Group 1 (smallest) represents stocks
ranked 151 to 200, Group 2 stocks 101 to 150, Group 3 stocks 51 to 100 and Group 4 (largest) stocks 1 to 50. The final two columns represent
the results for all 200 stocks combined. Statistical significance emanates from the test of whether the price effect (and difference) is significantly
different from zero. Block trades are defined as the largest one percent of on-market transactions, by volume, in each of the largest 200 stocks,
in each calendar year, for each market, for the period January 1, 1999 to December 31, 2002.
Xetra 0.5850** -0.9185** 0.4643** -0.9962** 0.3869** -0.6546** 0.2968** -0.6229** 0.3357** -0.6650**
Euronext 0.5547** -0.2402** 0.3337** -0.5703** 0.2800** -0.4300** 0.1747** -0.4197** 0.2312** -0.4300**
TSE 0.4717** -0.4232** 0.3906** -0.4469** 0.4115** -0.4311** 0.6436** -0.5957** 0.5166** -0.5063**
24
Table 2, continued
NYSE 0.0728** 0.0875** 0.0677** 0.0981** 0.0671** 0.0828** 0.0639** 0.0804** 0.0667** 0.0854**
LSE -0.2459** 0.3071** -0.2522** 0.3168** -0.2727** 0.3434** -0.2731** 0.3073** -0.2651** 0.3165**
Xetra 0.0590** 0.1480** 0.1263** 0.2408** 0.0352** 0.0800** 0.0397** 0.0730** 0.0454** 0.0885**
Euronext 0.1239** 0.1787** 0.0309 0.0322* 0.0457** 0.0412** 0.0008 0.0116* 0.0184** 0.0209**
TSE 0.0067 0.0689** 0.0281 0.0221 0.0438** 0.0185 0.0038* 0.0696** 0.0153** 0.0521**
25
Table 3
Price Effects Surrounding Block Trades Calculated Using Bid and Ask Quotes
This table reports the total effect (Panel A), temporary effect (Panel B) and the permanent effect (Panel C), all in percentage, calculated using
ask quotes for purchases and bid quotes for sales. Results are presented for the largest 200 stocks, by turnover, for each year, in each of the five
largest stock exchanges covering the three major trading regions around the world - North America, Europe and Asia-Pacific. The largest 200
stocks in each calendar year are further divided into quartiles, again based on turnover. Group 1 (smallest) represents stocks ranked 151 to 200,
Group 2 stocks 101 to 150, Group 3 stocks 51 to 100 and Group 4 (largest) stocks 1 to 50. The final two columns represent the results for all
200 stocks combined. Statistical significance emanates from the test of whether the price effect (and difference) is significantly different from
zero. Block trades are defined as the largest one percent of on-market transactions, by volume, in each of the largest 200 stocks, in each calendar
year, for each market, for the period January 1, 1999 to December 31, 2002.
Xetra 0.5059** -0.3873** 0.4114** -0.5421** 0.2804** -0.2883** 0.1799** -0.3420** 0.2255** -0.3417**
Euronext 0.4230** -0.1426** 0.1990** -0.4414** 0.2014** -0.3232** 0.1622** -0.3152** 0.1808** -0.3228**
TSE 0.3506** -0.2988** 0.2633** -0.2997** 0.2526** -0.2866** 0.4251** -0.4053** 0.3445** -0.3441**
26
Table 3, continued
NYSE 0.1070** -0.0761** 0.1088** -0.0572** 0.0928** -0.0707** 0.1460** -0.0521** 0.1479** -0.0608**
LSE 0.1898** -0.1543** 0.1387** -0.0860** 0.1391** -0.0777** 0.0467** -0.0013 0.1051** -0.0581**
Xetra 0.3770** -0.2368** 0.2178** -0.2399** 0.2297** -0.2156** 0.1784** -0.1593** 0.1989** -0.1837**
Euronext 0.3001** -0.0072 0.1860** -0.2066** 0.1557** -0.1162** 0.0655** -0.1488** 0.1135** -0.1411**
TSE 0.1037** -0.1373** 0.0776** -0.1702** 0.0358* -0.1667** 0.0950** -0.1964** 0.0804** -0.1759**
27
Table 4
Price Effects Surrounding Block Trades During Bull and Bear Markets
This table reports the number of transactions, average volume, average value, total effect, temporary effect and permanent effect, all in
percentage, calculated using ask quotes for purchases and bid quotes for sales. Results are presented for the largest 200 stocks, by turnover, for
each year, in each of the five largest stock exchanges covering the three major trading regions around the world - North America, Europe and
Asia-Pacific. Trades executed in months where the relevant MSCI index experienced positive returns are placed in the bull-market sample
(Panel A), whilst trades executed in months where the MSCI Index experienced negative returns are placed in the bear-market sample (Panel B).
Statistical significance emanates from the test of whether the price effect (and difference) is significantly different from zero. Block trades are
defined as the largest one percent of on-market transactions, by volume, in each of the largest 200 stocks, in each calendar year, for each market,
for the period January 1, 1999 to December 31, 2002.
28
Table 4, continued
29
Table 5
Price Effects Surrounding an Alternative Sample of Block Trades
This table reports the number of transactions, average volume, average value, total effect, temporary effect and permanent effect, all in
percentage, calculated using ask quotes for purchases and bid quotes for sales. Results are presented for the largest 200 stocks, by turnover, for
each year, in each of the five largest stock exchanges covering the three major trading regions around the world - North America, Europe and
Asia-Pacific. Statistical significance emanates from the test of whether the price effect (and difference) is significantly different from zero.
Block trades are selected if they represent at least 2.5 percent of average daily trading value calculated over the previous three months, in each of
the largest 200 stocks, in each calendar year, for each market, for the period January 1, 1999 to December 31, 2002.
30
Table 6
Price Effects Surrounding Block Trades Using Alternative
Pre- and Post-Trade Benchmarks
This table reports the total effect, temporary effect and permanent effect, all in
percentage, calculated using ask quotes for purchases and bid quotes for sales.
Results are presented for the largest 200 stocks, by turnover, for each year, in each of
the five largest stock exchanges covering the three major trading regions around the
world - North America, Europe and Asia-Pacific. In Panel A, the pre- and post
execution benchmarks are taken to be the quotes in existence five trades before and
after the block, while in Panel B they are the quotes in existence ten trades before and
after the block. Statistical significance emanates from the test of whether the price
effect (and difference) is significantly different from zero. Block trades are defined as
the largest one percent of on-market transactions, by volume, in each of the largest
200 stocks, in each calendar year, for each market, for the period January 1, 1999 to
December 31, 2002.
Permanent Effect
Total Effect Temporary Effect
(Difference)
Buys Sells Buys Sells Buys Sells
Panel A: Five Trades
0.1199** -0.1711**
NYSE 0.0750** -0.1511** 0.0449** -0.0200**
(-0.0512**)
0.3992** -0.4786**
LSE 0.3895** -0.4677** 0.0096** -0.0109**
(-0.0794**)
0.1689** -0.2268**
Xetra 0.1241** -0.1689** 0.0447** -0.0579**
(-0.0579**)
0.1656** -0.1703**
Euronext 0.1210** -0.1195** 0.0446** -0.0507**
(-0.0047)
0.1796** -0.1882**
TSE 0.1489** -0.1215** 0.0307** -0.0667**
(-0.0086)
0.1412** -0.2411**
NYSE 0.0814** -0.2122** 0.0597** -0.0289**
(-0.0990**)
0.4651** -0.5442**
LSE 0.4083** -0.5146** 0.0568** -0.0296**
(-0.0791**)
0.2118** -0.2991**
Xetra 0.1219** -0.2306** 0.0898** -0.0684**
(-0.0873**)
0.2310** -0.2558**
Euronext 0.1374** -0.1841** 0.0935** -0.0717**
(-0.0248*)
0.2062** -0.2684**
TSE 0.1653** -0.1798** 0.0409** -0.0885**
(-0.0622**)
** Indicates statistical significance at the 0.001 level
* Indicates statistical significance at the 0.01 level
31
Table 7
Price Effects Surrounding Block Trades Calculated Using Quote Midpoints
This table reports the total effect, temporary effect and permanent effect, all in
percentage, calculated using quote midpoints. Results are presented for the largest
200 stocks, by turnover, for each year, in each of the five largest stock exchanges
covering the three major trading regions around the world - North America, Europe
and Asia-Pacific. Statistical significance emanates from the test of whether the price
effect (and difference) is significantly different from zero. Block trades are defined as
the largest one percent of on-market transactions, by volume, in each of the largest
200 stocks, in each calendar year, for each market, for the period January 1, 1999 to
December 31, 2002.
Permanent Effect
Total Effect Temporary Effect
(Difference)
Buys Sells Buys Sells Buys Sells
0.3074** -0.4463**
NYSE 0.1789** -0.4235** 0.1258** -0.0228**
(-0.1388**)
0.2241** -0.2938**
LSE 0.1200** -0.2212** 0.1041** -0.0726**
(-0.0697**)
0.4437** -0.5441**
Xetra 0.3185** -0.4689** 0.1252** -0.0752**
(-0.1004**)
0.2571** -0.4270**
Euronext 0.1373** -0.3268** 0.1197** -0.1002**
(-0.1698**)
0.4777** -0.5528**
TSE 0.3957** -0.3971** 0.0819** -0.1551**
(-0.0745**)
** Indicates statistical significance at the 0.001 level
* Indicates statistical significance at the 0.01 level
32