View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Federal Reserve Bank of Chicago

When is Inter-Transaction Time
Informative?
Craig Furfine

WP 2003-04

When is Inter-Transaction Time Informative?

Craig Furfine
Federal Reserve Bank of Chicago
(312) 322-5175
craig.furfine@chi.frb.org
February 27, 2003

Abstract
We investigate the information content of inter-transaction time and find that it varies both
across stocks and over time. On average, inter-transaction time is found to be informative
whenever stocks are sufficiently traded. The magnitude of the information content is found to be
larger for less liquid, but still fairly actively traded stocks. In general, trades arriving quickly
move prices more than trades arriving more slowly. Further, the information content of intertransaction time is negatively correlated with proxies for the amount of private information in the
trading of a particular stock. We then distinguish between trades in the same direction as the
previous trade from trades in the reverse direction and find that the price impact of a trade as
well as the information content of inter-transaction time is dependent on trade type. In general,
reversing trades are more informative. Further, same-direction trades arriving quickly move
prices more than same-direction trades arriving more slowly, but reversing trades arriving
quickly move prices less than reversing trades arriving more slowly.

According to market microstructure models, prices respond to trades because trades
convey information regarding the underlying value of the security. In its simplest interpretation,
when traders buy, price rises as market makers revise upward their estimate of the securities true
value. The reverse holds true for sell orders. The literature has documented many factors that
influence by how much a trade moves price, the so-called price impact of a trade. Because the
price impact of a trade is related to the perceived quantity of private information held by the
buyer or seller, the price impact of a trade will be related to the probability that the order comes
from an “informed” rather than an “uninformed” e.g. noise trader. Other factors that have been
identified in the microstructure literature are characteristics of the trade being executed. For
instance, the size of a trade might convey information about its information content and thus
influence its price impact.
The focus of this paper is the impact of a particular trade characteristic that determines
price impact, inter-transaction time. Market microstructure literature has argued that the time
interval between trades conveys information. In Admati and Pfleiderer (1988), for example,
discretionary liquidity traders try to avoid losing money to the better informed by clustering their
trading close together in time. Thus, the observation of multiple transactions occurring together
suggests the presence of predominantly uninformed traders. The empirical prediction of this
model would be that trades that arrive more rapidly have lower price impact on average. Contrast
this intuition with that modeled by Easley and O’Hara (1992). In their model, they allow for the
possibility that no new information exists and for informed traders to be in a hurry to trade in
order to take advantage of their information advantage. As a result, an increase in trading activity
indicates that information has arrived, and therefore, order flow is more informative when
transactions are occurring rapidly.

2

Because theoretical models have an ambiguous prediction as to the relationship between
the time between trades and the price impact of trading, deciding upon the “correct” model
becomes an empirical question. The empirical evidence gathered to date suggests that the
relationship between inter-transaction time and price impact depends on the market. In foreign
exchange markets, Lyons (1996) documents that trades are less informative when they occur
when transaction intensity is high, a finding consistent with the theoretical result of Admati and
Pfleiderer (1988). Lyons describes it as hot-potato trading whereby foreign exchange dealers
rapidly and repeatedly lay off unwanted inventory in response to an initial potentially informed
trade. Because inventory adjustment by dealers is not informative as to the fundamental value of
a currency, these trades do not generally move prices. Dufour and Engle (2000) find the opposite
empirical relationship in equity markets. In a study of actively traded stocks, they find that when
equity markets are most active, i.e., inter-transaction times are short, the dynamic impact of order
flow on prices is enhanced. Spierdijk et. al. (2002) explores whether the relationship between
price impact and inter-transaction time is present in a sample of very illiquid stocks. They find
that the information content of inter-transaction time is greater for illiquid stocks than for the
actively traded stocks examined in Dufour and Engle (2000). In US treasury markets, Furfine
and Remolona (2002) find results similar to Dufour and Engle (2000). That is, trades of US
Treasuries arriving more quickly tend to have a greater price impact.
The aforementioned empirical studies generally focussed on both a limited number of
securities and on a fairly limited sample period. Thus, the first contribution of the current paper
is to determine whether previous results can be generalized across time and across a larger
number of securities. To some extent, Spierdijk et. al. (2002) study of illiquid stocks addresses
the cross-security issue, but the findings of their study may also be difficult to generalize because

3

they focus only on very infrequently traded stocks. In the present paper, we examine 100 stocks
that essentially span the range of trading levels from those studied by Dufour and Engle (2000)
to those in Spierdijk et. al. (2002). The second and more fundamental contribution of the present
paper is to document how and try to explain why the information content of the time between
trades changes over time. This question is motivated primarily by the observation that equity
market trading volume has increased dramatically during the past decade. Figure 1 indicates the
number of trades of NYSE-listed companies increased from around 3.5 million in January 1993
to over 30 million in December 2001. The value of these trades has risen similarly, from around
$200 billion in January 1993 to nearly $800 billion by the end of 2001. During the market peaks
of early 2000, monthly trading value approached $1.2 trillion. To put these numbers into the
trading context used in this paper, consider NYSE-listed companies grouped into deciles based
on their average daily number of trades.1 Figure 2 indicates that infrequently traded stocks, e.g.
those in the 9th decile, traded approximately once every 23 minutes in January 1993, but by
December 2001 traded once every 5 minutes. The most frequently traded stocks saw a similar
decline in the average time between trades, from around once every 37 seconds to one trade
every 10 seconds. Figure 3 examines more closely the trading of stocks in the first decile. The
median time between trades, which was 22 seconds in January 1993, had declined to less than 7
seconds in December 2001. Twenty-five percent of trades of the most actively traded stocks
occur within 3 seconds of the previous trade.
Given the tremendous increase in the amount of trading, it may be easy to imagine that
the information content of the time between trades has declined as trading activity has increased.
To see why this might be the case, consider the following hypothetical example. Suppose that
1

Unlike the data in Figure 1, the data in Figures 2 and 3 are based on the sample of companies used in this study and
not on a sample of all NYSE-listed companies.

4

250 uninformed traders of a particular stock will transact randomly and uniformly during a 6.5
hour trading day. If there is an information event, an additional 50 “informed” traders will
transact. Market makers will see trades approximately every 93 seconds when there is no
information and every 78 seconds when there is. Suppose this 16% reduction in average intertransaction time is sufficient to inform a market maker that a new information event has
occurred. Now assume that several years later, “uninformed” trading in this stock has increased
to 750 trades per day, but the number of potential informed traders remains at 50. Evidence that
such a relative increase in the “uninformed” has occurred is supported by Easley et. al. (2001).
Market makers would then see an average inter-transaction time of 31 seconds when there is no
information and 29 seconds when there is. Given the variability of inter-transaction times around
their mean, it is conceivable that this difference is not considered economically meaningful to
convey information. Intuitively, as average inter-transaction times fall, one might believe that the
information content of time declines. Thus, in light of the observed increase in trading activity, it
is interesting to determine whether inter-transaction time remains informative.
The coming sections of the paper present the following empirical evidence on the
information content of inter-transaction time. First, using tick-by-tick data on a sample of 100
stocks over 9 years, we document that the time between trades generally conveys information,
but only when a stock is traded fairly actively. Second, we find that among those stocks for
which inter-transaction time is informative, the information content itself varies across time.
Typically, faster trading is viewed as more informative, but we document cases where the
reverse is true. Third, we find that for actively traded stocks, variation in the information content
of inter-transaction time is related to changes in average inter-transaction time and average price
impact. Specifically, an increase in the average time between trades or a decrease in average

5

price impact is correlated with an increase in the information content of inter-transaction time.
Finally, we document that the relationship between inter-transaction time and price impact is
dependent on whether a trade is in the same direction or in the opposite direction as the previous
trade. In particular, same-direction trades arriving quickly move prices more than same-direction
trades arriving more slowly. However, reversing trades arriving quickly move prices less than
reversing trades arriving slowly.
The paper is organized as follows. Section I describes the data used in the study. Section
II reviews the Dufour and Engle (2000) model of price discovery implemented in the paper.
Section III presents empirical results for the stock of Disney (ticker DIS) that is illustrative for
the remainder of the paper. Section IV presents results from the full sample of 100 NYSE-listed
companies. Section V explores an extension to Dufour and Engle (2000) methodology,
specifically, the importance of whether or not a given trade is in the same direction as the
preceding trade. Section VI concludes.

I.

The Data

The transaction data were extracted from the NYSE TAQ (Trades and Quotes) database
covering the 2268 trades days beginning January 4, 1993 and ending on December 31, 2001.
Because information on market capitalization was used to perform various robustness checks,
sample companies were also required to be included in the CRSP daily stock files over the same
period. To abstract from potential differences in the price impact of trading across different
exchanges, only firms listed on the NYSE for the entire sample were considered.2 We also
2

The study hopes to analyze the time-series behavior of a cohort of firms where cohorts are determined by a
measure of trading intensity. Many NASDAQ firms were very infrequently traded in 1993, yet traded virtually every
second by 2001, complicating the definition of cohorts. Thus, to alleviate this difficulty, only NYSE-listed
companies were included.

6

require our sample firms to trade under the same ticker symbol throughout the 9-year period.
Because our main interest is to measure how the price impact of trades in the shares of a given
firm changes through time, we want to mitigate other factors that are changing through time. In
particular, major corporate mergers, which may lead to a ticker change, may mask any secular
change. Following Hasbrouck (1991), we also impose a minimum price requirement on each
company’s stock. We require each stock to be trading for at least $5, on average, during both
January 1993 and December 2001. Also following Hasbrouck (1991), we require a minimum
level of trading activity. Stocks were required to trade, on average, at least 8 trades per day
during January 1993 and 39 trades per day in December 2001.3 We then selected 100 of the
remaining stocks randomly, and then grouped them into 10 deciles according to their average
time between trade over the entire sample, with decile 1 corresponding to the most frequently
traded stocks.
The data are then adjusted according to procedures common in the microstructure
literature. Following Hasbrouck (1991), we keep only New York quotes and consider multiple
trades on a regional exchange for the same stock at the same price and time to be one trade.
Then, the trade data (for each company and day) are sorted by time, with the prevailing quote at
transaction t defined to be the last quote that was posted at least five seconds before the
transaction (Lee and Ready (1991)).
A complete listing of the stocks used in this study is given in Table I. As could be
expected, there is a positive although far-from-perfect negative correlation between the average
time between trades and a company’s market capitalization. Generally, larger firms have stocks
that trade more frequently. As closer examination of decile 1 stocks will be forthcoming, these
3

Hasbrouck (1991) chose a threshold of approximately 8 trades a day for data in 1989. 39 trades per day in
December 2001 is the same percentile of the distribution of trading frequency as 8 trades in January 1993.

7

have been printed in bold. Also, note that the information regarding the time between trades of
each stock are listed in minutes for 1993, but are given in seconds for 2001.

II.

Empirical framework

The dependent variable of interest is the trade-to-trade return on a given stock. We denote
this return rt , and define it formally as the change in the natural logarithm of the midquote of a
given stock that follows the trade at time t. That is,
æ æ bid t +1 + ask t +1 ö
æ bid t + ask t
rt = 100ç lnç
÷ - lnç
ç
2
2
ø
è
è è

öö
÷÷ .
÷
øø

(1)

Following Hasbrouck (1991), we define the variable xt0 as an indicator of the trade direction of
the trade occurring at time t. If the trade is initiated by the buyer, the variable xt0 = 1 . If the trade
is initiated by the seller, then the variable xt0 = -1 . We assume trades at a transaction price
greater than the midquote were buyer-initiated and trades below the midquote were sellerinitiated. For trades at the midquote, xt0 is assigned to equal zero. We also define Tt as the time,
in seconds, between the trade at time t and the trade at time t-1.
We adopt the empirical specification of Dufour and Engle (2000), which allows both the
trade indicator and the time between trades to affect returns. Defining Dt as an indicator that
equals 1 if trade t occurs during the first 30 minutes of the trading day, Dufour and Engle
propose an empirical relationship between trades, inter-transaction times, and returns given by
equation 2.4

4

Dufour and Engle (2000) specify additional equations for xt0 as well as trading intensity Tt . This allows the
computation of impulse response functions to see how inter-transaction time affects the dynamic path of price
adjustment in response to a trade. Our focus in this paper is on the narrower question of how the relationship
between inter-transaction time and price impact changes over time. Thus, we examine only the single equation.

8

5

rt = å ai rt -i + l
i =1

r
open

5

[

]

D x + å g ir + d ir ln(1 + Tt -i ) xt0-i + n t
0
t t

i =0

(2)

Because purchases should put upward pressure on prices, we expect that g ir + d ir ln(1 + Tt -i )
should evaluate to be positive over the range of relevant values of T for some or all of the trade
lags i. This prediction follows from traditional microstructure theory. In Glosten and Milgrom
(1985), for example, market makers set a positive bid-ask spread as compensation for trades
made with counterparties with superior information. As a sequence of sell orders arrive, market
makers lower bid prices, incorporating the probability that the order flow implies that betterinformed investors believe the previous price was too high. The reverse occurs when a sequence
of buy orders arrives.
As indicated in the introduction, however, there are theories that suggest that d ir could be
either positive or negative. In their analysis of 18 actively traded NYSE stocks, Dufour and
Engle (2000) find that d ir , when statistically significant, is generally negative, meaning that
trades that occur with a shorter inter-transaction time generally lead to price adjustments larger
than those following trades with larger inter-transaction intervals. In other words, stocks become
less liquid when trades arrive faster. This empirical finding is consistent with the intuition of
Easley and O’Hara (1992). That is, when trades arrive more quickly, market makers upwardly
adjust the probability that an information event has occurred. Thus, the probability of receiving
an order from an informed trader has risen and therefore prices must adjust more in response to a
given trade.

9

III.

Results for Walt Disney

To set the stage for the full-sample results presented in Section IV, in this section we
analyze the results for a single actively traded stock. The stock we pick is that of the Walt Disney
Company (ticker DIS). Table II presents selected coefficient estimates from a least squares
estimation of equation (2), with results presented separately for stock trades that occurred during
March 1993 and also those from April 1998. Standard errors are adjusted according to White
(1980).5 The first three columns of Table II reveal results quite similar to those presented by
Dufour and Engle (2000). In particular, the coefficients on the first three lags of the trade
indicator are all positive and statistically significant, and the coefficients on the first two lags of
the interaction of the trade indicator and the time between trade variable are negative and
significant. Thus, for DIS during March 1993, trades arriving faster moved prices more. Contrast
this finding with the results from estimating the same equation using trading data from April
1998. Like before, the coefficient on the trade indicator variable is positive and significant at low
lag levels. The coefficient capturing the effect of the time between trades, however, is now
positive and statistically significant. That is, in April 1998, trades of Disney stock arriving faster
moved prices less.
The results of Table II suggest that market makers for Disney believed trades arriving
faster conveyed more information in March 1993, yet contained less information in April 1998.
One possible explanation is depicted in Figure 4. The dotted line in Figure 4 plots the sum of the
d ir coefficients from the estimation of equation 2, where the equation was estimated separately

for each of the 108 months between January 1993 and December 2001. The solid line depicts the

5

In these and all subsequent regressions, price changes across days are omitted, as are return observations in the
extreme 0.25% tails of the distribution. These latter observations occur mainly due to infrequent, yet obvious errors
in either the bid or ask price of the stock.

10

average time between trades for Disney over the sample period. At least two observations are
worth making about Figure 4. First, the observations for March 1993 and April 1998 were the
most extreme observations for the d ir coefficients over the sample period. Second, there appears
to be a negative correlation between changes in the average time between trades and the d ir
coefficients. That is, when the time between trades declines sharply, this is generally associated
with an increase in the estimated value of the d ir coefficients. That is, an increase in trading
activity tends to reduce the negative relationship between inter-transaction time and price impact.
In the extreme case of April 1998, the relationship between inter-transaction time and price
impact became positive.
Although wanting to be cautious from making conclusions based on one observation, the
observation for Disney in April 1998 is interesting in that it coincided with the company’s April
23rd announcement of a 3-for-1 stock split. Starting on that date, trading activity in Disney
increased dramatically. As shown in Figure 4, trades of Disney stock occurred every 21 seconds
in March of 1998. Disney’s average inter-transaction time fell to just over 5 seconds by July of
that year. To the extent that market makers perceived the increase in trading activity as reflecting
new interest in the stock caused by a pending stock split and unrelated to new fundamental
information regarding the proper price level, the results of Figure 4 make sense. That is, trades
that were arriving much faster were viewed, on average, to be less informative about the price.
Thus far, we have commented on the evolution of the d ir coefficients for the Disney
Company between 1993 and 2001. From an economic standpoint, it is perhaps more useful to
compare measures of price impact rather than values of coefficients. In figure 5, we plot an
estimate of the price impact of trading, g ir + d ir ln(1 + Tt -i ) , summed across all lags, evaluated for

11

different values of the time between trading.6 The dotted line in figure 5 plots the price impact of
a trade that has occurred at that month’s average inter-transaction time. As the dotted line
indicates, the average price impact of a trade varies significantly over time. It is generally lower
at the end of the sample than at the beginning, indicating a general increase in market liquidity
for DIS stock. Analysis of the movements of the time series of liquidity measures such as that
depicted by the dotted line would be comparable to the work of Chordia et. al. (2001), who
analyze the variation of liquidity of common stocks both in cross-section and over time.
The focus of the present paper, however, is not on movements of the average liquidity of
stocks, but rather on the relationship between the time between trades and price impact. In Figure
5, this can be seen as the difference between the solid line and the line ticked with boxes.
Consider again the two observations highlighted in Table II. In March 1993, a trade occurring
with the average time since the previous trade moved the price of Disney stock by 1.4 basis
points. A trade arriving quickly, here defined as one arriving at the 10th percentile of the intertransaction time distribution for Disney stock during March 1993, moved the price of DIS stock
by an estimated 1.5 basis points. “Slowly” arriving trades, defined as those arriving at the 90th
percentile of the inter-transaction time distribution, moved prices by only 1.32 basis points.
Contrast that with the finding for April 1998. Liquidity, in general, was higher in that trades
occurring at the average inter-transaction time moved prices by only 1 basis point. Fast arriving
trades, perhaps because they were associated with news of the pending stock split, were
considered to be relatively uninformative, and moved prices less, by less than ½ basis point.
Slowly arriving trades moved prices by 1.24 basis points.

6

This estimated price impact is an approximate calculation that neglects the possibly endogenous nature of intertransaction time as well as the feedback of past returns on trading.

12

To more formally examine the relationship between trading activity and the significance
of the time between trades, we need a proxy for the information content of inter-transaction time.
In the empirical results to follow, we define
5

[

r
r
y m º å g im + d im ln(1 + Tz )
i =0

]

z =10 th % tile
z = 90th % tile

(3)

as a measure of the information content of inter-transaction time. That is, y m is calculated by
r
r
evaluating the sum of the g im + d im ln(1 + Tz ) coefficients at the 10th percentile of the interr
r
transaction time distribution and subtracting the sum of the g im + d im ln(1 + Tz ) coefficients

evaluated at the 90th percentile. This quantity is a measure of the relative information content of
fast trades. Graphically, y m is the distance between the solid line and the line with boxes in
Figure 5 during month m. The quantity y m is positive whenever faster trades are estimated to be
more informative, and therefore move prices more than more slowly arriving trades. We
analogously define the average price impact of a trade during month m as
5

[

r
r
Lm º å g im + d im ln(1 + Tz )
i =0

]

z = average

.

(4)

Defining Tm as the average time between trades during month m and D as the first difference
operator, we then estimate equation (5) using least squares.7
4

4

4

i =1

i =0

i =0

ym = å ai ym-i + å bi DTm + å ci DLm + em

(5)

Coefficient estimates from equation (5) for the Disney Company are given in Table III.

7

The specification chosen was based on the finding that average inter-transaction time and average liquidity have a
notable downward trend, but the information content of inter-transaction time does not. In fact, no statistically
significant trend was found in the information content of inter-transaction time for any of the 100 firms in the
sample.

13

The coefficient on the contemporaneous change in the average time between trades is
positive and significant. Thus, for the Disney Company, a decrease in the level of trading on
average (e.g. an increase in the average time between trades) is associated with a higher
differential price impact of fast arriving and slow arriving trades. In other words, when trading
becomes slower on average, faster arriving trades are thought to convey relatively more
information, and therefore move prices more. In addition to being statistically significant, Table
III indicates that this simple empirical specification explains a significant part, 39%, of the time
series variation in the information content of inter-transaction time. The high degree of
explanatory power is not solely due to the presence of lagged dependent variables in the
estimation. As the second column of Table III indicates, such variables account for 23% of the
total variation.

IV.

Results for the full sample

The analysis for the Disney Company in Section III suggests two things about the
information content of the time between trades. First, the information content of inter-transaction
time is itself, time varying. Second, the information content of time appears to be related to
changes in the average arrival rate of trades. In this section, we replicate the empirical exercises
of Section III on the full sample of 100 NYSE stocks. This entails estimating equation 2 for each
of 100 stocks for each of 108 months during the sample. Tables IV and V attempt to summarize
the basic findings from these 10,800 regressions.
The columns of Tables IV and V refer to stocks in different deciles, arranged from the
most actively traded issues in column 1 to the least actively traded issues in column 10. The rows
correspond to averages taken across the 12 months in the given year. Each cell in Table IV and V

14

contain three items. In Table IV, the first entry in each cell is the average value of the sum of the
g ir coefficients averaged across the 10 stocks in the decile and across the 12 months of the given

year (120 values). These coefficients measure the price impact of the given trade at lag i that is
unrelated to the time since the previous trade. The second entry in each cell in Table IV is the
percentage of the 120 individual observations of the sum of the g ir coefficients that were
estimated to be positive and statistically significant. The final entry in each cell is the percentage
of individual observations of the first item that were estimated to be negative and statistically
significant. For example, the cell in the upper left-hand corner of Table IV indicates that the
average sum of the g ir coefficients for the ten stocks in the most actively traded decile during
1993 is 0.018. Of the 120 estimated values of this quantity, 99.2% (e.g. 119) were estimated to
be both positive and statistically significant, whereas none were estimated to be negative and
statistically significant. Each cell in Table V is analogous to its counterpart in Table IV, except
that the first entry in each cell refers to the sum of the d ir coefficients, which measures the price
impact of a given trade that is related to the time since the last trade. For example, reading from
the upper left-hand cell in Table V indicates that the average sum of the d ir coefficients for the
most actively traded stocks during 1993 was -0.119. Recall a negative number indicates that
faster arriving trades carry more information and thus, move prices more. The remaining entries
in the cell indicate that 1.7% of the observations (2 out of 120) were positive and statistically
significant and 24.2% (29 out of 120) were statistically significantly negative.
The results presented in Table IV document that the relationship between trades and
returns is robust across time and across stocks of different levels of trading. For stocks in the
more actively traded deciles, regression estimates are virtually all positive and statistically

15

significant. For most years, average estimates of the g ir coefficients are increasing with trading
inactivity, suggesting that less frequently traded stocks are less liquid (because a given trade
moves prices more). The degree to which the results are found to be statistically significant tends
to increase through time for each decile, likely reflecting, in part, the increase in the number of
observations (e.g. trades) over time.
The results in Table V suggest that the negative relationship between inter-transaction
time and price impact found for the Disney Company in March 1993 is not robust across stocks,
either within trading activity deciles or across time. For the most actively traded decile of stocks,
the fraction of observations where we estimate a statistically significant negative relationship
between the time between trades and the information content of a trade is 86.7% in 1996, but
only 24.2% in 1993 and 50.0% in 2001. Looking throughout Table V, the average value of the
sum of the d ir coefficients is always negative, but for many deciles, especially during the early
part of the sample, most of the estimated coefficients are not statistically different from zero.
Only since 1999 have more than half of the trading activity deciles found more than half of the
d ir coefficients to be negative and statistically significant.

To give some further meaning to the numbers in Tables IV and V, we plot in Figure 6 the
estimated price impact of a trade arriving at the 90th and at the 10th percentile of the intertransaction time distribution, averaged over the stocks in the first decile. Given the relationship
between the importance of the time between trades and the average time between trades for the
Disney Company, Figure 6 also plots the average time between trades for stocks in the first
decile. As was the case for DIS, Figure 6 indicates that the price impact of a trade varies over
time. Furthermore, the information content of a fast arriving trade relative to a slow moving trade
varies as well, and does appear to be related to changes in the average time between trades.

16

To test this relationship more formally across the entire sample, equation (5) was
estimated again, with observations pooled across stocks within a given trading activity decile.
The results are shown in Table VI. As was the case for the Disney Company, the information
content of inter-transaction time is positively correlated with changes in the average time
between trades for stocks in the 3 most actively traded and 5 out of the 6 most actively traded
deciles of stocks. Table VI also indicates that for stocks in every decile except the most actively
traded, there is a negative correlation between the information content of inter-transaction time
and the average price impact of a trade. That is, all else equal, when average liquidity in a stock
improves (e.g. price impact at average inter-transaction time falls), the information content of
inter-transaction time increases (e.g. fast trades move prices relatively more). Thus, changes to
the average time between trades and changes to the average price impact are correlated with the
information content of inter-transaction time in opposite ways.
These results may seem somewhat perplexing since a decline in the average time between
trades and a decline in the average price impact are both often considered to be associated with
increases in market liquidity. However, the results here suggest that, holding one measure
constant, these two empirical measures have a differently signed correlation with the information
content of inter-transaction time. Intuitively, this finding may be explained as follows.
Consider first the negative coefficient on average price impact. The interpretation of this
coefficient is that holding the average inter-transaction time constant, an increase in average
price impact leads to a decline in the information content of inter-transaction time. Holding
average inter-transaction time constant, however, is equivalent to holding the number of trades
constant. Thus, an increase in the average price impact of a trade accompanied by no change in
the number of trades implies that the quantity of private information in the market has increased.

17

The negative coefficient on average price impact therefore implies that an increase in the amount
of private information in the market is associated with a fall in the information content of intertransaction time. This may be because market makers rely less on inter-transaction time to
discern which traders have information when information is plentiful, i.e. informed traders are
relatively common.
Consider now the positive coefficient on average inter-transaction time. The
interpretation of this coefficient is that holding average price impact constant, an increase in
average inter-transaction time increases the information content of inter-transaction time.
Holding average price impact constant, however, is equivalent to assuming that each trade
contains the same amount of private information. Thus, an increase in inter-transaction time in
this environment implies that the quantity of private information in the market has declined
because there are fewer trades. To be consistent with the coefficient on average price impact,
lower levels of private information must correlate with a higher information content of intertransaction time. This is what the positive coefficient on average inter-transaction time is
revealing.
Thus, the coefficients on these two empirical measures logically enter with opposite
signs. Economically, the results suggest that when information becomes relatively scarce, intertransaction time becomes more informative.

V.

Robustness of results

To this point, the analysis has assumed that all trades have an equal impact on returns.
Hasbrouck (1991), however, originally proposed that the price impact of a trade may be a
function not only of the direction of the trade but also the size of the trade. That is, a large

18

purchase of stock might be considered more informative than a small one and thus might affect
prices more. To consider this, the analysis of Section IV was repeated, replacing the simple trade
indicator xt0 with a variable xt , defined as the log of the fraction of a company’s market value
that was being transacted. For example, a buy-order of 10,000 shares in a company with
10,000,000 shares outstanding would produce xt0 = 1 , but xt = -3. With this new specification,
the results of the previous section hold qualitatively. Figure 7 replicates Figure 6, although now,
estimated values of the price impact of trades with different inter-transaction times are now also
assumed to be of average size, where size is measured here as the log of the share of the given
company’s market value. Figures 6 and 7 are nearly identical.
Another possible extension of the Dufour and Engle (2000) estimation approach
incorporates the suggestion of Peng (2001), who argues that the information content of a trade
depends on whether or not the trade is of the same type as the preceding trade. Define a buy
order following a buy order or a sell order following a sell order as a same-direction trade.
Analogously, define reversing trades as a buy order following a sell order or a sell order
following a buy order. Peng’s (2001) intuition is that a market maker who sees a same-direction
trade does not know whether the trade contains more information than the first or whether it is
simply a response to the same information as the first. This holds true especially if the samedirection trade comes quickly after the preceding trade. In contrast, a reversing trade cannot, by
definition, simply be a second response to the same information that led to the preceding trade.
Thus, on average, reversing trades must be more informative than same-direction trades,
regardless of the time elapsed since the preceding trade. To explore the possible interaction
between inter-transaction time and trade type (e.g. same-direction or reversing), we estimate
equation 6 below, which is an enriched version of equation 2,

19

5

rt = å ai rt -i + l
i =1

r
open

Dt x

0
t

+ å [g
5

i =0

r
i

r
i

+ d ln(1 + Tt -i

)]x

0
t -i

5

[

]

+ å g irs + d irs ln(1 + Tt -i ) S t -i xt0-i + n 1,t (6)
i =0

where the dummy variable S t equals 1 when the trade at time t is same-direction and 0
otherwise. Thus, the specification in equation 6 allows the price impact and the time impact
coefficients to vary depending on whether the trade is same-direction or reversing.
Tables VII through X present the results from this estimation procedure that are
analogous to those presented in Tables IV and V. Tables VII and VIII refer to the trade impact
and time impact coefficients reported for reversing trades. Tables IX and X relate to trades in the
same direction as the previous trade. Like the results of Table IV, Table VII and IX indicate that
the price impact coefficients are typically positive for both types of trades. Further, there do not
appear to be any major differences between the size of the estimates. However, a comparison of
the results in Tables VIII and X with those from Table V highlight that the information content
of inter-transaction time varies notably across trade type (e.g. same-direction or reversing). Table
X, for example, indicates that for trades in the same direction as the previous trade, faster arrival
is associated with more information and therefore greater price impact. This finding is
qualitatively similar to that found in Table V. Quantitatively, however, the magnitude of the
information content of inter-transaction time is generally stronger, as indicated by numbers of
greater absolute value in Table X relative to Table V and are more often statistically significant,
especially in more recent years and for more actively traded stocks.
A symmetry argument would therefore lead one to believe that if the information content
of inter-transaction time is larger for same-direction trades than for an “average” trade, then it
must follow that the information content of inter-transaction time for reversing trades is lower
than that found for trades in general. Table VIII indicates, however, that the information content
of inter-transaction time is positive for reversing trades. That is, faster arriving reversing trades
20

are thought to contain less information and therefore move prices by less than reversing trades
that arrive after a longer wait.
To help visualize these empirical results, Figure 8 plots the estimated price impact at
various inter-transaction times for both same-direction and reversing trades for stocks in the most
actively traded decile. One immediate finding is that reversing trades are considered more
informative in that they have a higher price impact. This is consistent with Peng’s (2001)
argument that same-direction trades may be a response to stale information. Depending on the
sample month and the inter-transaction time, a reversing trade is estimated to move the price of
an actively traded stock by between 2 and 5 basis. A same-direction trade, however, moves
prices between 0.5 and 3 basis points. Figure 8 also illustrates that inter-transaction time affects
the price impact of trading in different directions for the two types of trades. That is, a samedirection trade with a low inter-transaction time moves prices by approximately 1 basis point
more than a same-direction trade arriving more slowly. Fast-arriving reversing trades move
prices by approximately 0.5 basis points less than a reversing trade arriving more slowly.
Intuitively, these results can be explained as follows. First, the finding that fast-arriving
same-direction trades move prices more than slow-arriving same-direction trades is consistent
with the belief that when same-direction trades arrive faster, a market maker increases the
probability of an information event having happened. Second, the finding that fast-arriving
reversing trades move prices less than slow-arriving reversing trades is consistent with fastarriving reversing trades signaling an increased presence of “uninformed” traders, who by
definition would be equally likely to buy or sell. Thus, market makers interpret rapid arrival of
reversing trades to be less informative than those arriving more slowly.

21

Tables XI and XII complete the robustness exercise by re-estimating equation (5)
separately for same-direction and reversing trades. Table XI reports the coefficient estimates
when the dependent variable is the same proxy for the information content of inter-transaction
time as was used to estimate (5), except that price impacts are constructed only for reversing
trades. That is, the dependent variable measures the difference between the top two lines of
Figure 8. Recall, however, that since fast-arriving reversing trades are considered less
informative than slow-arriving reversing trades, the dependent variable is always negative. Table
XII reports the analogous results for same-direction trades. In this case, the dependent variable is
positive, just as it was when equation (5) was estimated without consideration of whether a trade
was same-direction or reversing. The independent variables (e.g. average inter-transaction time
and average price impact) are calculated only for reversing trades or same-direction trades,
respectively. As Tables XI and XII indicate, the information content of inter-transaction time
remains positively correlated with changes in the average time between trades and negatively
correlated with average price impact. For same-direction trades, the interpretation remains the
same, namely that higher average inter-transaction time or lower average price impact correlate
with less private information. Less private information correlates with an increased information
content of inter-transaction time. To be more precise, fast trades are more informative when
private information is lower. For reversing trades, however, because the dependent variable is
negative, the interpretation of the coefficients reported in Table XI is somewhat different. The
interpretation is that a decrease in the quantity of private information is correlated with an
increase in a negative number, implying that the difference between the price impact of fast and
slow reversing trades becomes less.

22

VI.

Conclusion

In this paper, we apply the method of Dufour and Engle (2000) to a larger cross section
of stocks and a notably longer time series. Doing so allows us to document many features of the
role that inter-transaction time plays in the price discovery process. First, the information content
of inter-transaction time varies across stocks and across time. At any point in time and for
relatively actively traded stocks, trades that arrive faster generally move prices more than trades
that arrive more slowly. However, secular declines in inter-transaction time have not eliminated
the information content of inter-transaction time. Second, we find empirical measures that help
to explain the time-series behavior of the information content of inter-transaction time. Our
results suggest that when the level of private information in a market falls, inter-transaction time
becomes more informative in a particular way. Specifically, faster trading tends to move prices
more relative to slow trading. Finally, we document that the direction of a trade relative to the
previous trade is an important factor in determining a trade’s price impact. Same-direction trades
are generally less informative. Further, the correlation between inter-transaction time and price
impact is different for the two types of trades. Fast arriving same-direction trades move prices
more than slow-arriving same-direction trades, but fast-arriving reversing trades move prices less
than slow-arriving reversing trades.

References
Admati, Anat R. and Paul Pfleiderer (1988), “A theory of intraday patterns: volume and price
variability,” Review of Financial Studies 1, 3-40.
Chordia, Tarun, Lakshmanan Shivakumar, and Avanidhar Subrahmanyan (2001), “The CrossSection of Daily Variation in Liquidity,” mimeo.

23

Dufour, Alfonso and Robert F. Engle (2000), “Time and the Price Impact of a Trade,” The
Journal of Finance 55, No. 6, 2467-2498.
Easley, David, Robert F. Engle, Maureen O’Hara, and Liuren Wu (2001), “Time-Varying
Arrival Rates of Informed and Uninformed Trades,” working paper.
Easley, David and M. O’Hara (1992), “Time and the process of security price adjustment,” The
Journal of Finance 47, 905-927.
Furfine, Craig and Eli Remolona (2002), “Price discovery in a market under stress: the U.S.
Treasury market in fall 1998,” mimeo, Bank for International Settlements.
Glosten, Lawrence R. and Paul R. Milgrom (1985), “Bid, Ask, and Transaction Prices in a
Specialist market with Heterogeneously Informed Traders,” Journal of Financial Economics 14,
71-100.
Hasbrouck, Joel (1991), “Measuring the information content of stock trades,” The Journal of
Finance 46, No. 1, 179-207.
Lee, Charles M. C. and Mark J. Ready (1991), “Inferring Trade Direction from Intraday Data,”
The Journal of Finance 46, No. 2, 733-746.
Lyons, R. (1996), “Foreign exchange volume: sound and fury signifying nothing?” in Frankel, J.,
G. Galli and A. Giovannini, eds.: The Microstructure of Foreign Exchange Markets (University
of Chicago Press).
Peng, Liang (2001), “Trading Takes Time,” Yale ICF Working Paper No. 00-57.
Spierdijk, Laura, Theo E. Nijman, and Arthur H.O. van Soest (2002), “The Price Impact of
Trades in Illiquid Stocks in Periods of High and Low Market Activity,” mimeo, Tilburg
University.

24

White, Hal (1980), “A heteroskedasticity consistent covariance matrix estimator and a direct test
for heteroskedasticity,” Econometrica 48, 817-838.

25

Company name

A C M INCOME FUND INC
ALBERTO CULVER CO
ARCHER DANIELS MIDLAND CO
ALCATEL ALSTHOM
ALASKA AIRGROUP INC
ADVANCED MICRO DEVICES INC
A S A LTD
ATMOS ENERGY CORP
AMERICAN WATER WORKS INC
AMERICAN EXPRESS CO
BANDAG INC
BECKMAN COULTER INC
BRIGGS & STRATTON CORP
B J SERVICES CO
BELLSOUTH CORP
BRISTOL MYERS SQUIBB CO
BOWATER INC
B P PRUDHOE BAY ROYALTY TRUST
BURLINGTON RESOURCES INC
ANHEUSER BUSCH COS INC
CONAGRA INC
CIRCUIT CITY STORES INC
CADENCE DESIGN SYSTEMS INC
CINERGY CORP
COMERICA INC
COMPAQ COMPUTER CORP
CARLISLE COMPANIES
CENTURYTEL INC
DANA CORP
DILLARDS INC
DISNEY WALT CO
DELUXE CORP
DIME BANCORP INC NEW
DOLE FOOD INC
EMPIRE DISTRICT ELEC CO
FIRSTFED FINANCIAL CORP
FEDERAL NATIONAL MORTGAGE ASSN
G A T X CORP
GEORGIA PACIFIC CORP
HALLIBURTON COMPANY
HOUSEHOLD INTERNATIONAL INC
HITACHI LIMITED
HANCOCK FABRICS INC
HILTON HOTELS CORP

Ticker

ACG
ACV
ADM
ALA
ALK
AMD
ASA
ATO
AWK
AXP
BDG
BEC
BGG
BJS
BLS
BMY
BOW
BPT
BR
BUD
CAG
CC
CDN
CIN
CMA
CPQ
CSL
CTL
DCN
DDS
DIS
DLX
DME
DOL
EDE
FED
FNM
GMT
GP
HAL
HI
HIT
HKF
HLT

0.54
0.40
8.57
0.31
0.22
1.64
0.32
0.16
0.78
11.10
0.78
0.71
0.83
0.25
25.50
31.00
0.76
0.65
5.65
16.30
7.68
2.47
0.97
2.17
3.46
3.97
0.38
1.28
2.16
5.07
23.90
3.64
0.20
1.95
0.29
0.25
21.80
0.67
5.32
3.12
2.59
0.18
0.29
2.29

($ billions)

Market value

62.48
20.44
371.88
32.82
47.18
531.53
131.53
13.90
38.06
529.76
37.41
30.54
50.13
28.77
359.74
1633.39
30.54
48.42
166.53
265.93
220.58
255.68
116.91
74.36
108.46
603.23
10.12
63.86
61.19
166.09
991.88
123.16
79.49
75.31
16.53
7.52
413.52
31.11
235.00
211.42
84.32
19.16
23.03
96.89

(# of trans.)

Daily trading

6.80
21.05
1.10
12.92
9.35
0.95
3.69
28.29
11.27
0.85
12.01
14.35
8.47
19.94
1.14
0.28
14.71
9.11
2.63
1.58
1.95
1.86
4.51
5.52
3.93
0.76
37.00
6.58
6.83
2.60
0.44
3.39
5.99
5.91
23.68
39.78
0.99
14.08
1.88
2.02
5.16
26.28
17.40
4.83

Ave. time
between trades

1993

0.34
1.11
0.08
0.55
0.41
0.07
0.14
2.54
0.37
0.06
0.30
0.54
0.21
3.65
0.08
0.04
0.51
0.55
0.15
0.11
0.10
0.10
0.14
0.32
0.19
0.06
5.97
0.23
0.26
0.09
0.05
0.16
0.25
0.20
0.76
12.94
0.11
0.35
0.09
0.11
0.13
2.76
1.01
0.10

(minutes)

th

17.12
58.10
2.65
33.20
24.30
2.41
10.04
75.50
29.79
2.12
32.41
38.73
23.09
52.49
2.80
0.65
39.69
22.16
6.73
3.93
4.92
4.87
12.17
13.87
9.89
1.88
97.06
17.48
18.30
6.81
1.05
8.64
15.87
15.75
66.96
90.17
2.39
38.99
4.93
5.25
14.11
73.11
46.22
13.23

10 percentile of 90 percentile
time between off time between
trades
trades

th

1.20
1.46
9.49
1.37
0.77
5.39
0.19
0.87
4.18
47.60
0.29
2.70
0.92
5.12
71.60
98.70
2.61
0.32
7.54
40.00
12.80
5.41
5.36
5.32
10.20
16.60
1.12
4.63
2.06
1.28
42.20
2.76
4.27
1.50
0.41
0.44
79.60
1.58
6.34
5.62
26.50
0.79
0.24
4.03

($ billions)

Market value

216.75
264.06
892.91
856.95
330.68
3842.20
80.87
140.54
307.23
3538.40
62.61
324.22
201.18
1515.73
1850.04
2833.93
423.07
170.54
1129.28
1418.15
1097.13
992.71
813.81
552.29
908.65
3853.76
160.25
510.45
650.69
509.17
3586.39
421.44
481.36
292.58
77.75
123.04
2207.25
322.68
926.40
2886.46
1676.98
115.13
80.05
693.01

(# of trans.)

Daily trading

127.45
96.08
28.08
29.61
90.00
7.40
366.25
179.41
90.95
7.81
446.57
78.87
134.08
17.08
13.90
9.45
58.46
154.42
22.02
17.91
23.00
26.99
31.05
45.74
28.11
7.60
157.16
49.69
38.99
51.36
8.39
62.69
55.06
95.50
330.08
218.51
11.94
87.89
26.77
10.41
16.03
214.11
524.09
36.50

Ave. time
between trades

2001

7.51
4.82
2.98
3.09
5.11
1.37
9.39
6.30
3.94
1.88
16.48
4.28
5.05
2.10
2.31
1.94
3.35
5.85
2.72
2.10
2.57
2.60
3.18
3.88
2.70
1.62
5.89
3.20
2.83
3.00
1.73
3.99
3.38
4.36
8.97
6.14
2.13
4.05
2.74
1.92
2.39
4.85
38.95
2.79

(seconds)

26

318.75
253.00
67.64
71.24
240.25
16.30
1034.56
480.89
233.46
16.74
1193.62
204.69
359.68
41.24
31.85
20.75
150.71
422.08
52.53
43.88
55.92
65.38
76.00
113.10
68.77
16.56
418.53
126.27
96.61
131.94
18.22
161.21
141.62
255.13
904.58
599.95
27.15
230.83
66.67
23.17
37.02
572.96
1437.33
91.49

10th percentile of 90th percentile
time between
off time between
trades
trades

The data are based on a sample of 100 stocks taken from the TAQ and CRSP databases. Market values are taken in January 1993 and December 2001 from the
CRSP database. Trading statistics are averaged over the entire year (1993 or 2001) based on trades in the TAQ database. For 1993, time is quoted in minutes. For
2001, time is quoted in seconds.

Summary Statistics for Sample Firms

Table I

Company name

JACOBS ENGINEERING GROUP INC
JOHNSON & JOHNSON
KELLOGG CO
KEYCORP NEW
KOREA FUND INC
K MART CORP
DREYFUS STRATEGIC MUNICIPALS INC
LEGG MASON INC
LINCOLN NATIONAL CORP IN
L T C PROPERTIES INC
LIMITED INC
MASCO CORP
MEAD CORP
MURPHY OIL CORP
INCO LTD
NATIONAL FUEL GAS CO N J
NIAGARA MOHAWK HOLDINGS INC
NUVEEN PREMIUM INC MUNI FD 2 INC
NUVEEN INVT QUALITY MUNI FUND
NUVEEN SELECT QLTY MUNI FUND INC
NORFOLK SOUTHERN CORP
NETWORK EQUIPMENT TECHNOLOGIES
OAKWOOD HOMES CORP
PRECISION CASTPARTS CORP
PARKER HANNIFIN CORP
PUTNAM MASTER INCOME TRUST
P N M RESOURCES INC
CATALINA MARKETING CORP
PUGET ENERGY INC
RITE AID CORP
R G S ENERGY GROUP INC
ROHM & HAAS CO
ROYCE VALUE TR INC
SEARS ROEBUCK & CO
SEITEL INC
SYNOVUS FINANCIAL CORP
S P X CORP
STRIDE RITE CORP
STUDENT LOAN CORP
SAFEWAY INC
TEKTRONIX INC
THOMAS INDUSTRIES INC
TOLL BROTHERS INC
T R C COMPANIES INC
T X U CORP
TYCO INTERNATIONAL LTD NEW
UNIFI INC
U G I CORP
UNIVERSAL HEALTH RLTY INCM TR
VODAFONE GROUP PLC NEW
WESTPAC BANKING CORP
WALLACE COMPUTER SERVICES INC
WENDYS INTERNATIONAL INC
W G L HOLDINGS INC

Ticker

JEC
JNJ
K
KEY
KF
KM
LEO
LM
LNC
LTC
LTD
MAS
MEA
MUR
N
NFG
NMK
NPM
NQM
NQS
NSC
NWK
OH
PCP
PH
PMT
PNM
POS
PSD
RAD
RGS
ROH
RVT
S
SEI
SNV
SPW
SRR
STU
SWY
TEK
TII
TOL
TRR
TXU
TYC
UFI
UGI
UHT
VOD
WBK
WCS
WEN
WGL

0.64
28.80
14.80
2.87
0.33
9.44
0.56
0.25
3.06
0.08
9.96
4.94
2.25
1.61
2.43
1.00
2.72
0.52
0.61
0.51
8.88
0.17
0.30
0.38
1.58
0.47
0.44
0.37
1.54
1.80
0.91
3.64
0.20
17.00
0.07
0.93
0.24
0.99
0.57
1.37
0.69
0.10
0.50
0.05
9.45
2.10
1.91
0.67
0.11
0.95
0.03
0.66
1.29
0.76

($ billions)

Market value

33.23
927.54
238.71
107.37
42.49
827.74
37.96
10.89
72.26
9.07
437.18
143.64
81.81
31.94
60.41
57.68
227.62
29.53
28.70
36.43
127.36
34.80
47.48
31.44
46.08
52.89
53.30
22.88
100.85
98.55
42.37
46.11
20.96
472.19
25.81
29.38
19.90
132.17
23.06
129.49
48.81
7.36
71.24
6.52
285.21
67.45
75.02
48.03
15.80
72.28
26.25
26.90
187.12
36.30

(# of trans.)

Daily trading

12.77
0.48
1.77
3.94
14.78
0.52
10.76
35.10
5.84
40.93
1.03
2.93
5.30
14.58
7.30
7.30
1.79
13.01
13.54
10.93
3.22
12.97
10.31
13.97
9.34
7.51
7.66
21.40
4.76
4.21
10.18
9.42
18.65
0.91
21.06
15.12
21.33
3.17
26.67
3.56
9.79
42.32
6.76
44.50
1.44
6.25
6.74
8.59
25.78
7.58
17.46
14.77
2.26
11.46

Ave. time
between trades

1993

0.56
0.06
0.11
0.21
0.52
0.05
0.68
6.81
0.26
9.44
0.08
0.13
0.16
0.37
0.29
0.29
0.14
1.14
0.86
0.70
0.15
0.37
0.27
0.61
0.22
0.36
0.22
1.51
0.28
0.18
0.64
0.27
1.17
0.10
2.31
0.60
2.70
0.15
2.54
0.16
0.42
11.12
0.31
13.18
0.11
0.17
0.20
0.24
2.91
0.27
0.66
0.34
0.12
0.51

(minutes)

th

34.12
1.15
4.47
10.07
41.43
1.23
27.59
90.48
15.12
96.78
2.52
7.52
14.15
41.08
19.98
18.70
4.37
31.86
34.27
27.37
8.52
37.49
28.63
37.73
25.48
19.56
21.13
62.03
11.89
10.90
25.40
25.23
48.91
2.21
55.72
40.23
57.93
7.93
69.67
9.39
26.25
93.13
18.61
94.25
3.50
16.65
18.03
22.83
65.85
20.54
47.22
40.91
5.74
29.32

10 percentile of 90 percentile
time between off time between
trades
trades

th

1.78
181.00
12.20
10.30
0.65
2.72
0.59
3.20
9.20
0.12
6.31
11.20
3.06
3.81
3.08
1.96
2.84
0.57
0.51
0.48
7.07
0.12
0.05
1.46
5.38
0.34
1.09
1.91
1.90
2.61
1.30
7.63
0.61
15.40
0.33
7.30
5.53
0.27
1.61
21.00
2.37
0.38
1.57
0.41
12.50
127.00
0.39
0.83
0.27
23.20
0.18
0.78
3.05
1.41

($ billions)

Market value

Summary Statistics for Sample Firms

Table I continued

406.21
3173.42
688.33
973.65
60.07
1292.16
49.21
467.10
744.31
41.13
826.70
949.32
593.47
585.66
550.46
243.59
281.87
30.70
25.22
26.07
876.83
28.92
50.33
424.35
528.79
49.73
279.43
260.16
401.26
1315.07
135.58
755.90
91.10
1215.50
273.93
634.87
638.35
75.47
40.53
1615.92
536.14
25.97
517.54
197.44
1164.52
4485.28
86.09
127.74
60.33
1243.70
17.74
136.99
481.68
196.57

(# of trans.)

Daily trading

72.42
8.67
37.09
25.65
488.75
20.88
489.42
55.19
33.43
557.94
30.73
26.56
43.17
45.45
45.73
103.37
92.61
791.29
950.30
951.83
28.68
750.57
556.05
61.08
48.46
481.02
98.32
98.94
64.70
21.76
190.45
34.12
284.43
21.54
98.18
40.38
41.55
323.89
885.01
15.88
45.97
1222.07
51.83
301.83
21.87
6.46
265.51
196.95
861.04
20.14
1402.31
181.28
51.98
125.27

Ave. time
between trades

2001

3.19
1.91
3.18
2.90
12.23
2.52
7.55
3.38
2.72
9.70
2.81
2.48
4.01
2.87
3.01
4.99
4.81
12.89
24.47
19.36
3.04
24.83
21.73
4.00
3.95
15.21
4.71
3.88
4.16
2.30
7.49
3.04
10.06
3.07
4.85
3.28
2.54
12.27
112.19
2.88
3.02
229.00
3.10
10.77
2.81
1.35
8.84
7.35
110.00
2.66
142.78
7.20
3.05
4.09

(seconds)

27

189.59
18.64
91.33
62.58
1369.83
49.79
1397.87
140.40
84.11
1590.99
76.47
65.73
104.92
114.32
116.81
266.68
237.32
2218.20
2665.78
2614.28
70.63
2137.98
1528.77
153.12
122.00
1291.62
254.75
264.00
164.55
52.55
495.96
84.11
755.03
51.23
256.78
102.61
108.29
884.33
2283.08
36.28
116.82
3194.75
132.13
866.50
52.86
13.77
707.76
515.07
2169.16
47.14
3602.08
476.36
132.47
337.67

10th percentile of 90th percentile
time between
off time between
trades
trades

Table II
Estimated Coefficients for the Return Equation for the Disney Company (DIS)
Coefficient estimates and robust standard errors (in parenthesis) for the equation
5

5

i =1

i =0

[

]

rt = å ai rt -i + lr Dt xt0 + å g ir + d ir ln(1 + Tt -i ) xt0-i + n t
open
rt , is the change in the natural logarithm of the midquote of a given stock that follows the trade at time t,
xt0 is the trade indicator (1 for a buy, -1 for a sale, 0 if at midquote), Tt is the time (in seconds) between the
transaction at t and the transaction at t-1, Dt is an indicator that equals 1 if the trade is in the first 30
minutes of trading. The coefficients in columns 2-4 reflect all trades in DIS during March 1993. The
coefficients in columns 5-7 reflect all trades in DIS during April 1998.
March 1993
Lag number

Quote
Revision

Trade

( ai )

(g i )

0
1
2
3
4
5

April 1998

-0.0212
(0.0082)**
-0.0129
(0.0089)
0.0046
(0.0086)
0.0042
(0.0093)
0.0179
(0.0085)*

Trade *
Duration

r

0.0093
(0.0009)**
0.0075
(0.0009)**
0.0026
(0.0009)**
0.0012
(0.0009)
0.0016
(0.0009)
0.0011
(0.0009)

r

(d i )
-0.0010
(0.0003)**
-0.0010
(0.0003)**
-0.0002
(0.0003)
0.0002
(0.0003)
-0.0001
(0.0003)
-0.0002
(0.0003)

17587 observations
Adj. R2: 0.05

Quote
Revision

Trade

( ai )

(g i )

0.0040
(0.0056)
0.0254
(0.0058)**
0.0266
(0.0064)**
0.0286
(0.0058)**
0.0259
(0.0059)**

Trade *
Duration

r

0.0017
(0.0002)**
0.0005
(0.0002)**
-0.0002
(0.0002)
-0.0004
(0.0002)
-0.0009
(0.0002)**
-0.0011
(0.0002)**

r

(d i )
0.0013
(0.0001)**
0.0007
(0.0001)**
0.0005
(0.0001)**
0.0004
(0.0001)**
0.0002
(0.0001)*
0.0003
(0.0001)**

41047 observations
Adj. R2: 0.07

Robust standard errors in parentheses
* significant at 5%; ** significant at 1%

28

Table III

Estimated Coefficients for the Information Content of Inter-Transaction Time Equation for
the Disney Company (DIS)
Coefficient estimates and robust standard errors (in parenthesis) for the equation
4

4

4

i =1

i =0

i =0

y m = å ai y m -i + å bi DTm + å ci DLm + em
are reported in column 1. y m is the proxy for the information content of inter-transaction time in month m defined as
r
r
r
r
ym º å [g im + d im ln(1 + Tz )] z =90th %tile . g im and d im are estimated from equation 2. TZ is the z-percentile of the inter5

z =10 th % tile

i =0

transaction time distribution for month m. The values of y m are represented visually as the difference between the two
solid lines graphed in Figure 5. Tm is the average inter-transaction time during month m. Lm is the average price
r
r
impact of a trade, defined as Lm º å [g im + d im ln(1 + Tz )]
5

z =average

, and D represents first differences. Related empirical

i =0

specifications reported in columns (2)-(4).
Specification
Independent variable
Lags of y
a1
a2
a3
a4
Lags of DT
b0
b1
b2
b3
b4

(1)
0.2926
(0.1250)*
0.2587
(0.1003)*
0.2057
(0.0978)*
-0.0978
(0.1121)

(2)
0.3272
(0.1142)**
0.1720
(0.1043)
0.1587
(0.0959)
-0.0931
(0.0944)

(3)
0.3386
(0.1245)**
0.1496
(0.1122)
0.1518
(0.1008)
-0.0871
(0.0964)

0.0167
(0.0041)**
0.0031
(0.0061)
-0.0040
(0.0061)
-0.0061
(0.0048)
-0.0036
(0.0042)

0.0140
(0.0039)**

Lags of DL
c0

-0.1379
(0.0758)
c1
0.0260
(0.0971)
c2
0.0168
(0.1009)
c3
-0.0021
(0.1128)
c4
0.0479
(0.1002)
Constant
0.0004
(0.0003)
Observations
103
R-squared
0.39
Robust standard errors in parentheses
* significant at 5%; ** significant at 1%

(4)

0.0005
(0.0003)
104
0.23

0.0083
(0.0818)
0.0701
(0.1000)
0.0945
(0.0953)
-0.0211
(0.0886)
-0.0363
(0.0841)
0.0005
(0.0003)
103
0.25

-0.1257
(0.0655)

0.0011
(0.0002)**
107
0.12

29

i =0

i =1

year

0.018
0.992
0

0.022
0.983
0

0.019
0.992
0

0.026
1
0

0.022
1
0

0.019
0.992
0

0.018
1
0

0.021
1
0

0.014
1
0

1993

1994

1995

1996

1997

1998

1999

2000

2001

Most actively traded

0.023
1
0

0.042
1
0

0.036
1
0

0.041
1
0

0.048
1
0

0.058
1
0

0.051
0.975
0

0.056
0.975
0

0.052
0.933
0

0.029
0.992
0

0.061
1
0

0.054
1
0

0.062
1
0

0.062
1
0

0.072
1
0

0.063
1
0

0.071
0.992
0

0.065
0.975
0

0.032
1
0

0.068
1
0

0.065
1
0

0.069
1
0

0.072
1
0

0.092
0.983
0

0.104
0.958
0

0.12
0.875
0

0.111
0.8
0

0.036
1
0

0.066
1
0

0.06
1
0

0.062
1
0

0.07
0.933
0

0.091
0.95
0

0.101
0.908
0

0.1
0.725
0

0.091
0.658
0

…

Firm category

0.062
0.983
0

0.109
0.967
0

0.109
0.975
0

0.102
0.95
0

0.089
0.883
0

0.12
0.875
0

0.111
0.775
0

0.097
0.6
0

0.136
0.65
0

0.1
0.9
0

0.146
0.983
0

0.105
0.958
0

0.103
0.967
0

0.109
0.892
0

0.128
0.783
0

0.125
0.617
0

0.119
0.55
0

0.126
0.583
0

0.08
0.967
0

0.126
0.983
0

0.134
0.95
0

0.121
0.942
0

0.111
0.85
0

0.135
0.6
0

0.149
0.608
0

0.197
0.533
0

0.146
0.429
0

0.079
0.808
0

0.137
0.817
0

0.11
0.85
0

0.109
0.717
0

0.103
0.529
0

0.126
0.445
0

0.132
0.378
0

0.146
0.336
0

0.116
0.25
0.008

0.102
0.6
0

0.215
0.403
0

0.249
0.439
0

0.398
0.342
0

0.265
0.246
0

0.258
0.228
0

0.235
0.26
0

0.344
0.231
0.01

0.233
0.284
0

Least actively traded

estimated for the full sample of 100 stocks over the 108 months from January 1993 to December 2001. The first entry in each cell
represents the average value of the sum of the g ir coefficients for all firms in the given firm decile and the given year. The second entry is
the share of the 120 individual observations (10 firms x 12 months) that were statistically significant and positive. The third entry is the
share of the 120 individual observations that were statistically significant and negative.

5

5

rt = å ai rt -i + lropen Dt xt0 + å [g ir + d ir ln(1 + Tt -i )]xt0-i + n t

Entries in the table derive from the estimates of the g ir coefficients from the equation

Summary of Sign and Statistical Significance of the g ir Coefficients: Full Sample

Table IV

30

i =0

i =1

-0.119
0.017
0.242

-0.143
0.025
0.275

-0.173
0.042
0.625

-0.306
0.017
0.867

-0.152
0.108
0.6

-0.074
0.117
0.442

-0.094
0.108
0.533

-0.155
0.117
0.592

-0.087
0.092
0.5

1994

1995

1996

1997

1998

1999

2000

2001

Most actively traded

1993

-0.216
0.008
0.767

-0.371
0
0.742

-0.312
0.042
0.725

-0.373
0.008
0.742

-0.484
0
0.733

-0.662
0
0.858

-0.519
0
0.667

-0.545
0
0.55

-0.543
0
0.508

-0.26
0
0.733

-0.531
0
0.692

-0.453
0.008
0.667

-0.538
0.008
0.592

-0.585
0
0.592

-0.77
0
0.725

-0.562
0
0.592

-0.65
0
0.542

-0.625
0.008
0.492

-0.323
0
0.825

-0.633
0
0.75

-0.543
0.008
0.633

-0.543
0
0.533

-0.579
0
0.458

-0.789
0
0.467

-0.875
0
0.367

-0.899
0
0.258

-0.936
0.008
0.242

-0.347
0
0.733

-0.597
0
0.775

-0.547
0
0.675

-0.526
0
0.542

-0.563
0
0.5

-0.887
0
0.525

-0.925
0
0.417

-0.761
0.008
0.217

-0.76
0
0.2

…

Firm category

-0.597
0
0.642

-0.839
0
0.367

-0.891
0
0.333

-0.867
0.008
0.375

-0.681
0
0.417

-1.056
0
0.35

-0.886
0.017
0.242

-0.554
0
0.158

-0.919
0.008
0.283

-0.963
0
0.567

-1.335
0
0.45

-0.953
0
0.542

-0.987
0.008
0.508

-0.91
0
0.308

-0.975
0.008
0.25

-0.955
0.008
0.133

-0.837
0.008
0.142

-1.04
0.017
0.225

-0.735
0
0.658

-1.039
0
0.55

-1.122
0
0.475

-1.02
0.008
0.442

-0.883
0
0.283

-1.034
0.008
0.2

-1.077
0
0.2

-1.588
0.008
0.233

-1.103
0
0.21

-0.744
0.008
0.358

-1.249
0
0.358

-0.973
0.008
0.317

-0.893
0
0.225

-0.817
0
0.227

-1.124
0
0.193

-1.016
0
0.134

-0.851
0.008
0.109

-0.659
0.033
0.083

-0.87
0.008
0.275

-1.694
0.008
0.109

-2.124
0.026
0.07

-4.607
0
0.105

-1.787
0
0.105

-2.019
0
0.096

-1.555
0.01
0.125

-2.387
0.01
0.096

-1.581
0.009
0.155

Least actively traded

estimated for the full sample of 100 stocks over the 108 months from January 1993 to December 2001. The first entry in each cell
represents the average value of the sum of the d ir coefficients for all firms in the given firm decile and the given year. The second entry
is the share of the 120 individual observations (10 firms x 12 months) that were statistically significant and positive. The third entry is the
share of the 120 individual observations that were statistically significant and negative.

5

5

rt = å ai rt -i + lropen Dt xt0 + å [g ir + d ir ln(1 + Tt -i )]xt0-i + n t

Entries in the table derive from the estimates of the d ir coefficients from the equation

Summary of Sign and Statistical Significance of the d ir Coefficients: Full Sample

Table V

31

4

i =0

i =1

4

z =10 th % tile

5

[

DL

DT

-0.1857
(0.0588)**
0.0709
(0.0625)
0.0919
(0.0678)
-0.0021
(0.0618)
0.1001
(0.0589)
0.0030
(0.0005)**
1030

0.0059
(0.0016)**
0.0021
(0.0015)
0.0002
(0.0014)
0.0004
(0.0014)
-0.0008
(0.0013)

0.2778
(0.0468)**
0.1741
(0.0404)**
0.1409
(0.0403)**
0.2007
(0.0456)**

0.46

-0.0518
(0.0456)
-0.0286
(0.0460)
0.0117
(0.0471)
0.0676
(0.0461)
-0.0257
(0.0455)
0.0008
(0.0002)**
1030

0.0196
(0.0021)**
0.0075
(0.0020)**
0.0035
(0.0019)
0.0024
(0.0021)
-0.0016
(0.0019)

0.4011
(0.0438)**
0.1988
(0.0471)**
0.0821
(0.0510)
0.1582
(0.0470)**

Most actively traded

0.57

Robust standard errors in parentheses
* significant at 5%; ** significant at 1%

Obs
R-squared

Constant

c4

c3

c2

c1

Lags of
c0

b4

b3

b2

b1

Lags of
b0

a4

a3

a2

Lags of y
a1

0.37

-0.1970
(0.0653)**
0.1704
(0.0757)*
0.0307
(0.0745)
0.1042
(0.0893)
0.0087
(0.0749)
0.0061
(0.0009)**
1030

0.0081
(0.0020)**
0.0046
(0.0021)*
0.0030
(0.0021)
0.0032
(0.0020)
0.0005
(0.0019)

0.3471
(0.0505)**
0.1759
(0.0669)**
0.0467
(0.0538)
0.1314
(0.0538)*

0.36

-0.4805
(0.1184)**
0.1235
(0.1036)
0.0366
(0.0916)
-0.0862
(0.0940)
-0.0888
(0.1195)
0.0090
(0.0016)**
1030

0.0007
(0.0015)
0.0007
(0.0015)
0.0018
(0.0014)
0.0035
(0.0023)
-0.0025
(0.0010)*

0.2913
(0.0536)**
0.1313
(0.0677)
0.1430
(0.0595)*
0.0771
(0.0446)

0.34

-0.5425
(0.0983)**
-0.1093
(0.0971)
-0.0827
(0.1140)
0.2169
(0.0971)*
0.1921
(0.0812)*
0.0080
(0.0015)**
1030

0.0037
(0.0013)**
0.0036
(0.0014)*
-0.0003
(0.0012)
-0.0016
(0.0011)
-0.0034
(0.0012)**

0.2254
(0.0568)**
0.1671
(0.0489)**
0.1102
(0.0459)*
0.1665
(0.0526)**

…

0.30

-0.6752
(0.0835)**
-0.3309
(0.0928)**
-0.1727
(0.0852)*
-0.2129
(0.0742)**
-0.1396
(0.0631)*
0.0112
(0.0021)**
1030

0.0035
(0.0012)**
0.0018
(0.0011)
0.0009
(0.0010)
0.0018
(0.0009)
0.0024
(0.0009)*

0.2723
(0.0492)**
0.1496
(0.0479)**
0.1266
(0.0515)*
0.0963
(0.0468)*

Firm category

i =0

]
z =average

0.30

-0.5619
(0.1360)**
-0.3242
(0.1486)*
0.0826
(0.1526)
-0.1766
(0.1351)
-0.1272
(0.1164)
0.0158
(0.0056)**
1030

0.0012
(0.0008)
0.0011
(0.0009)
0.0002
(0.0009)
-0.0005
(0.0008)
0.0010
(0.0008)

0.3523
(0.1433)*
0.2241
(0.0758)**
0.0049
(0.0862)
0.0369
(0.0660)

r
r
inter-transaction time during month m. Lm is the average price impact of a trade, defined as Lm º å g im + d im ln(1 + Tz )

i =0

5

0.39

-0.7836
(0.0692)**
-0.3174
(0.0870)**
-0.1052
(0.0790)
-0.0256
(0.0871)
0.0240
(0.0623)
0.0171
(0.0029)**
1030

0.0012
(0.0007)
0.0019
(0.0006)**
0.0013
(0.0008)
0.0016
(0.0007)*
0.0014
(0.0006)*

0.2231
(0.0445)**
0.1886
(0.0579)**
0.1351
(0.0442)**
0.0745
(0.0566)

0.25

-0.6137
(0.0510)**
-0.3428
(0.0850)**
-0.3236
(0.0723)**
-0.0025
(0.0666)
0.0151
(0.0491)
0.0244
(0.0037)**
1030

-0.0006
(0.0007)
-0.0013
(0.0009)
0.0004
(0.0009)
0.0006
(0.0007)
0.0008
(0.0008)

0.1900
(0.0415)**
0.0365
(0.0408)
0.1554
(0.0367)**
-0.0167
(0.0568)

Least actively traded

, and D represents first differences.

0.20

-0.1748
(0.0249)**
-0.0340
(0.0119)**
-0.0262
(0.0101)*
-0.0119
(0.0073)
-0.0020
(0.0014)
0.0259
(0.0055)**
1000

-0.0004
(0.0006)
0.0007
(0.0007)
0.0002
(0.0008)
0.0002
(0.0008)
0.0010
(0.0007)

0.2220
(0.0437)**
0.0296
(0.0310)
0.0514
(0.0281)
0.0278
(0.0285)

r
r
r
r
ym º å [g im + d im ln(1 + Tz )] z =90th %tile . g im and d im are estimated from equation 2. TZ is the z-percentile of the inter-transaction time distribution for month m. Tm is the average

Each column reports coefficients for the given decile of firms. y m is the proxy for the information content of inter-transaction time in month m defined as

i =0

Coefficient estimates and robust standard errors (in parenthesis) for the equation y m = å ai y m -i + å bi DTm + å ci DLm + em .

4

Estimated Coefficients for the Information Content of Inter-Transaction Time Equation for the Full Sample

Table VI

32

i=0

i =1

i =0

5

Year

0.022
0.992
0

0.026
0.992
0

0.023
0.992
0

0.031
1
0

0.026
1
0

0.022
1
0

0.021
1
0

0.025
1
0

0.019
1
0

1994

1995

1996

1997

1998

1999

2000

2001

Most actively traded

1993

0.03
1
0

0.05
1
0

0.041
1
0

0.047
1
0

0.054
1
0

0.066
1
0

0.057
0.983
0

0.063
0.983
0

0.06
0.95
0

0.036
1
0

0.07
1
0

0.062
1
0

0.07
1
0

0.069
1
0

0.08
1
0

0.071
1
0

0.081
1
0

0.075
0.992
0

0.04
1
0

0.077
1
0

0.072
1
0

0.076
1
0

0.078
1
0

0.101
0.983
0

0.113
0.958
0

0.13
0.892
0

0.124
0.892
0

0.045
1
0

0.073
1
0

0.066
1
0

0.069
1
0

0.077
0.95
0

0.1
0.958
0

0.11
0.942
0

0.112
0.808
0

0.103
0.75
0

…

Firm category

0.073
0.992
0

0.117
0.983
0

0.117
0.975
0

0.111
0.975
0

0.102
0.908
0

0.135
0.908
0

0.127
0.833
0

0.11
0.633
0

0.158
0.725
0

0.115
0.917
0

0.159
0.992
0

0.113
0.975
0

0.111
0.975
0

0.118
0.933
0

0.14
0.825
0

0.138
0.725
0

0.134
0.608
0

0.139
0.65
0

0.094
0.958
0

0.137
0.983
0

0.145
0.967
0

0.13
0.958
0

0.12
0.85
0

0.148
0.658
0

0.162
0.633
0

0.22
0.6
0

0.161
0.442
0.008

0.093
0.883
0

0.149
0.85
0

0.12
0.875
0

0.117
0.75
0

0.114
0.575
0.008

0.14
0.5
0.008

0.155
0.35
0.008

0.16
0.342
0.008

0.148
0.325
0

0.117
0.642
0

0.231
0.4
0.008

0.268
0.45
0.067

0.25
0.367
0.075

0.211
0.275
0.083

0.24
0.283
0.125

0.281
0.283
0.167

0.216
0.208
0.2

0.203
0.375
0.05

Least actively traded

in the same direction as the preceding trade, 0 otherwise. The first entry in each cell represents the average value of the sum of the g ir
coefficients for all firms in the given firm decile and the given year. The second entry is the share of the 120 individual observations (10
firms x 12 months) that were statistically significant and positive. The third entry is the share of the 120 individual observations that were
statistically significant and negative.

estimated for the full sample of 100 stocks over the 108 months from January 1993 to December 2001. S t equals 1 if the current trade is

5

5

rt = å a i rt -i + lropen Dt xt0 + å [g ir + d ir ln(1 + Tt -i )]xt0-i + å [g irs + d irs ln(1 + Tt -i )]S t -i xt0-i + n 1,t

Entries in the table derive from the estimates of the g ir coefficients from the equation

Summary of Sign and Statistical Significance of the g ir Coefficients when Controlling for Trade Type

Table VII

33

i=0

i =1

i =0

5

0.16
0.408
0.008

0.16
0.317
0.008

0.158
0.367
0.008

0.245
0.475
0

0.239
0.667
0

0.234
0.7
0

0.24
0.8
0

0.257
0.708
0

0.489
0.992
0

1993

1994

1995

1996

1997

1998

1999

2000

2001

Most actively traded

0.575
0.975
0

0.247
0.408
0.008

0.166
0.267
0.05

0.038
0.142
0.075

0.081
0.158
0.067

0.04
0.108
0.092

-0.025
0.108
0.083

0.042
0.058
0.05

-0.026
0.075
0.067

0.659
0.858
0

0.221
0.275
0.033

0.135
0.15
0.058

0.022
0.075
0.075

0.091
0.108
0.083

0.051
0.05
0.1

0.104
0.092
0.067

0.092
0.083
0.067

0.005
0.075
0.058

0.678
0.867
0

0.026
0.133
0.017

0.006
0.133
0.042

0.015
0.033
0.042

-0.037
0.058
0.1

-0.077
0.033
0.075

-0.087
0.017
0.058

-0.003
0.033
0.025

0.072
0.058
0.05

0.707
0.808
0.008

-0.057
0.033
0.058

-0.112
0.05
0.1

-0.144
0.017
0.058

-0.095
0.033
0.1

-0.222
0.025
0.092

-0.326
0.008
0.092

-0.07
0.075
0.05

-0.124
0.025
0.083

…

Firm category

0.673
0.425
0

-0.008
0.125
0.025

-0.221
0.067
0.058

-0.207
0.067
0.042

0.094
0.075
0.033

-0.029
0.017
0.033

-0.041
0.042
0.042

0.19
0.033
0.033

0.008
0.042
0.092

1.058
0.3
0.008

0.286
0.008
0.042

-0.288
0.025
0.133

-0.335
0.033
0.167

-0.291
0.008
0.092

-0.334
0.017
0.1

-0.364
0.017
0.042

-0.093
0.033
0.058

-0.188
0.008
0.05

0.788
0.483
0.008

-0.203
0.017
0.075

-0.155
0
0.067

-0.378
0.025
0.075

-0.379
0.017
0.083

-0.271
0.033
0.033

-0.277
0.033
0.033

-0.636
0.033
0.067

-0.295
0
0.067

0.67
0.092
0

-0.339
0.008
0.058

-0.128
0.033
0.05

-0.376
0.008
0.083

-0.164
0.017
0.05

-0.417
0.017
0.075

0.039
0
0.033

0.009
0.033
0.075

0.454
0.067
0.017

0.799
0.125
0.008

0.02
0.025
0.033

-0.458
0.067
0.083

2.163
0.092
0.083

3.556
0.075
0.075

2.066
0.125
0.092

1.698
0.133
0.1

2.117
0.15
0.117

0.88
0.067
0.025

Least actively traded

in the same direction as the preceding trade, 0 otherwise. The first entry in each cell represents the average value of the sum of the d ir
coefficients for all firms in the given firm decile and the given year. The second entry is the share of the 120 individual observations (10
firms x 12 months) that were statistically significant and positive. The third entry is the share of the 120 individual observations that were
statistically significant and negative.

estimated for the full sample of 100 stocks over the 108 months from January 1993 to December 2001. S t equals 1 if the current trade is

5

5

rt = å a i rt -i + lropen Dt xt0 + å [g ir + d ir ln(1 + Tt -i )]xt0-i + å [g irs + d irs ln(1 + Tt -i )]S t -i xt0-i + n 1,t

Entries in the table derive from the estimates of the d ir coefficients from the equation

Summary of Sign and Statistical Significance of the d ir Coefficients when Controlling for Trade Type

Table VIII

34

i=0

i =1

i =0

5

Year

0.022
0.975
0

0.025
0.983
0

0.022
0.992
0

0.03
1
0

0.026
0.992
0

0.021
0.992
0

0.02
1
0

0.025
1
0

0.018
1
0

1993

1994

1995

1996

1997

1998

1999

2000

2001

Most actively traded

0.028
1
0

0.047
1
0

0.039
1
0

0.044
1
0

0.054
1
0

0.067
1
0

0.057
0.967
0

0.066
0.967
0

0.059
0.942
0

0.034
1
0

0.068
1
0

0.06
1
0

0.068
1
0

0.071
1
0

0.084
1
0

0.074
1
0

0.084
1
0

0.071
0.983
0

0.037
1
0

0.073
1
0

0.069
1
0

0.072
1
0

0.076
1
0

0.102
0.975
0

0.113
0.967
0

0.132
0.892
0

0.117
0.808
0

0.042
1
0

0.07
1
0

0.062
1
0

0.066
1
0

0.081
0.95
0

0.101
0.958
0

0.11
0.95
0

0.113
0.775
0

0.1
0.717
0

…

Firm category

0.069
0.992
0

0.119
0.967
0

0.113
0.975
0

0.107
0.933
0

0.097
0.883
0

0.132
0.9
0

0.126
0.825
0

0.102
0.625
0

0.147
0.683
0

0.114
0.942
0

0.158
0.983
0

0.113
0.967
0

0.106
0.958
0

0.114
0.908
0

0.142
0.808
0

0.145
0.733
0

0.138
0.608
0

0.143
0.633
0

0.087
0.95
0

0.135
0.967
0

0.151
0.967
0

0.127
0.9
0

0.119
0.833
0

0.146
0.642
0

0.163
0.617
0

0.228
0.583
0

0.15
0.417
0.008

0.089
0.817
0

0.16
0.883
0

0.122
0.842
0

0.115
0.717
0

0.113
0.567
0.008

0.136
0.517
0.008

0.152
0.342
0.008

0.15
0.308
0.008

0.131
0.258
0

0.123
0.675
0

0.231
0.442
0.008

0.27
0.45
0.05

0.285
0.383
0.05

0.206
0.317
0.067

0.277
0.325
0.083

0.391
0.317
0.133

0.22
0.308
0.15

0.193
0.367
0.042

Least actively traded

in the same direction as the preceding trade, 0 otherwise. The first entry in each cell represents the average value of the sum of the g ir
coefficients for all firms in the given firm decile and the given year. The second entry is the share of the 120 individual observations (10
firms x 12 months) that were statistically significant and positive. The third entry is the share of the 120 individual observations that were
statistically significant and negative.

estimated for the full sample of 100 stocks over the 108 months from January 1993 to December 2001. S t equals 1 if the current trade is

5

5

rt = å a i rt -i + lropen Dt xt0 + å [g ir + d ir ln(1 + Tt -i )]xt0-i + å [g irs + d irs ln(1 + Tt -i )]S t -i xt0-i + n 1,t

Entries in the table derive from the estimates of g ir + g irs coefficients from the equation

Summary of Sign and Statistical Significance of the g ir + g irs Coefficients when Controlling for Trade Type

Table IX

35

i=0

i =1

i =0

5

-0.21
0.008
0.467

-0.257
0.008
0.533

-0.277
0.025
0.733

-0.461
0.008
0.95

-0.311
0.05
0.833

-0.214
0.058
0.725

-0.248
0.025
0.85

-0.357
0.025
0.875

-0.333
0
0.958

1993

1994

1995

1996

1997

1998

1999

2000

2001

Most actively traded

-0.499
0.008
0.992

-0.726
0
0.992

-0.55
0.017
0.933

-0.55
0
0.867

-0.687
0
0.917

-0.871
0
0.95

-0.693
0
0.742

-0.731
0
0.708

-0.741
0
0.667

-0.552
0
0.975

-0.927
0
0.967

-0.745
0
0.842

-0.784
0.008
0.833

-0.748
0
0.742

-0.976
0
0.825

-0.751
0
0.667

-0.873
0
0.65

-0.889
0
0.675

-0.607
0
0.992

-0.973
0
0.95

-0.809
0.008
0.858

-0.782
0
0.725

-0.759
0
0.65

-1.025
0
0.625

-1.107
0
0.5

-1.148
0
0.383

-1.264
0
0.4

-0.648
0
0.975

-0.867
0
0.9

-0.766
0
0.817

-0.696
0
0.65

-0.728
0
0.6

-1.09
0
0.658

-1.115
0
0.5

-1.038
0
0.333

-1.036
0
0.333

…

Firm category

-0.878
0
0.8

-1.088
0
0.492

-1.098
0
0.483

-1.036
0.008
0.425

-0.958
0
0.483

-1.387
0
0.45

-1.175
0
0.4

-0.808
0
0.242

-1.263
0
0.342

-1.313
0
0.683

-1.653
0
0.533

-1.138
0
0.55

-1.144
0.008
0.508

-1.085
0
0.425

-1.202
0
0.317

-1.169
0
0.192

-1.169
0
0.192

-1.257
0.017
0.25

-1.054
0
0.808

-1.353
0
0.65

-1.39
0
0.533

-1.223
0.008
0.475

-1.067
0
0.333

-1.291
0.008
0.217

-1.286
0
0.258

-2.075
0
0.25

-1.269
0
0.183

-0.929
0
0.425

-1.385
0
0.392

-1.176
0.008
0.375

-0.995
0
0.258

-0.942
0.008
0.258

-1.285
0
0.183

-1.4
0
0.158

-0.927
0.017
0.192

-1.118
0.008
0.133

-1.148
0
0.292

-2.062
0.017
0.125

-1.764
0.075
0.083

-1.616
0.05
0.158

-1.114
0.058
0.142

-1.884
0.092
0.183

-1.623
0.125
0.183

-0.353
0.125
0.217

-1.237
0.05
0.183

Least actively traded

in the same direction as the preceding trade, 0 otherwise. The first entry in each cell represents the average value of the sum of the d ir
coefficients for all firms in the given firm decile and the given year. The second entry is the share of the 120 individual observations (10
firms x 12 months) that were statistically significant and positive. The third entry is the share of the 120 individual observations that were
statistically significant and negative.

estimated for the full sample of 100 stocks over the 108 months from January 1993 to December 2001. S t equals 1 if the current trade is

5

5

rt = å a i rt -i + lropen Dt xt0 + å [g ir + d ir ln(1 + Tt -i )]xt0-i + å [g irs + d irs ln(1 + Tt -i )]S t -i xt0-i + n 1,t

Entries in the table derive from the estimates of d ir + d irs coefficients from the equation

Summary of Sign and Statistical Significance of the d ir + d irs Coefficients when Controlling for Trade Type

Table X

36

4

4

4

i =0

i =0
5

i =0

z =10 th % tile

r
r
y m is the proxy for the information content of inter-transaction time in month m defined as ym º å [g im + d im ln(1 + Tz )] z =90th %tile .

i =1

y m = å ai y m -i + å bi DTm + å ci DLm + em .

DL

DT

0.66

-0.3432
(0.0220)**
-0.1615
(0.0237)**
-0.1110
(0.0260)**
-0.0275
(0.0233)
0.0043
(0.0182)
-0.0005
(0.0002)*
1030

0.0146
(0.0016)**
0.0033
(0.0016)*
0.0018
(0.0016)
0.0010
(0.0016)
-0.0017
(0.0015)

0.4026
(0.0435)**
0.2365
(0.0460)**
0.1037
(0.0483)*
0.1659
(0.0427)**

Most actively traded

-0.3770
(0.0371)**
-0.1671
(0.0440)**
-0.0541
(0.0443)
-0.0620
(0.0353)
-0.0163
(0.0272)
-0.0012
(0.0004)**
1030

0.0061
(0.0013)**
0.0013
(0.0015)
-0.0006
(0.0016)
-0.0000
(0.0014)
-0.0011
(0.0013)

0.3726
(0.0548)**
0.1879
(0.0579)**
0.0861
(0.0463)
0.1532
(0.0454)**

0.51

-0.3683
(0.0256)**
-0.1598
(0.0287)**
-0.0864
(0.0289)**
-0.0752
(0.0332)*
0.0300
(0.0274)
-0.0008
(0.0003)*
1030

0.0056
(0.0011)**
0.0022
(0.0011)*
0.0003
(0.0009)
-0.0005
(0.0009)
-0.0011
(0.0008)

0.3315
(0.0443)**
0.1962
(0.0407)**
0.1315
(0.0410)**
0.2000
(0.0415)**

0.55

0.39

-0.2708
(0.0522)**
-0.1086
(0.0625)
-0.1388
(0.0549)*
-0.1041
(0.0471)*
-0.0304
(0.0372)
-0.0005
(0.0007)
1030

0.0030
(0.0013)*
-0.0003
(0.0015)
0.0017
(0.0010)
0.0016
(0.0013)
0.0007
(0.0006)

0.2735
(0.0674)**
0.0835
(0.0831)
0.1623
(0.0628)**
0.0902
(0.0571)

0.40

-0.3602
(0.0292)**
-0.2207
(0.0382)**
-0.1123
(0.0371)**
-0.0917
(0.0430)*
-0.0106
(0.0250)
0.0001
(0.0006)
1030

0.0024
(0.0008)**
0.0028
(0.0009)**
-0.0002
(0.0008)
-0.0014
(0.0008)
-0.0017
(0.0008)*

0.2200
(0.0539)**
0.1805
(0.0475)**
0.1013
(0.0455)*
0.1572
(0.0548)**

i =0

5

…

z =average

0.46

-0.3644
(0.0377)**
-0.2269
(0.0515)**
-0.1470
(0.0451)**
-0.0997
(0.0373)**
-0.0407
(0.0268)
-0.0009
(0.0010)
1030

0.0030
(0.0009)**
0.0023
(0.0008)**
0.0016
(0.0008)*
0.0013
(0.0008)
0.0008
(0.0006)

0.2440
(0.0559)**
0.1285
(0.0575)*
0.1907
(0.0565)**
0.1213
(0.0545)*

Firm category

r
r
Lm is the average price impact of a reversing trade, defined as Lm º å [g im + d im ln(1 + Tz )]

Robust standard errors in parentheses
* significant at 5%; ** significant at 1%

Obs
R-squared

Constant

c4

c3

c2

c1

Lags of
c0

b4

b3

b2

b1

Lags of
b0

a4

a3

a2

Lags of y
a1

trades during month m.

0.58

-0.3610
(0.0251)**
-0.1802
(0.0438)**
-0.0640
(0.0481)
-0.1036
(0.0457)*
0.0131
(0.0313)
-0.0004
(0.0012)
1030

0.0007
(0.0005)
0.0005
(0.0005)
0.0001
(0.0005)
-0.0003
(0.0006)
0.0005
(0.0005)

0.56

-0.3547
(0.0187)**
-0.2237
(0.0282)**
-0.1317
(0.0290)**
-0.0827
(0.0294)**
-0.0233
(0.0187)
0.0031
(0.0013)*
1030

0.0008
(0.0005)
0.0020
(0.0006)**
0.0009
(0.0005)
0.0012
(0.0005)**
0.0012
(0.0004)**

0.2002
(0.0495)**
0.1519
(0.0484)**
0.1253
(0.0414)**
0.0630
(0.0443)

D represents first differences.

0.2933
(0.0385)**
0.2176
(0.0758)**
0.1122
(0.0488)*
0.2077
(0.0422)**

, and

0.40

-0.2613
(0.0180)**
-0.1805
(0.0254)**
-0.1258
(0.0249)**
-0.0550
(0.0222)*
-0.0274
(0.0153)
0.0009
(0.0014)
1030

0.0002
(0.0004)
-0.0003
(0.0006)
0.0006
(0.0005)
0.0006
(0.0005)
0.0006
(0.0005)

0.1704
(0.0404)**
0.1169
(0.0398)**
0.1676
(0.0411)**
0.0450
(0.0480)

Least actively traded

0.38

-0.0631
(0.0110)**
-0.0457
(0.0102)**
-0.0399
(0.0079)**
-0.0300
(0.0083)**
-0.0101
(0.0053)
-0.0046
(0.0047)
1000

-0.0000
(0.0005)
0.0008
(0.0006)
0.0003
(0.0006)
0.0007
(0.0007)
0.0015
(0.0006)*

0.1308
(0.0377)**
0.0442
(0.0444)
0.0788
(0.0390)*
0.1340
(0.0449)**

r
r
g im and d im are estimated from equation 6. TZ is the z-percentile of the inter-transaction time distribution for month m for reversing trades. Tm is the average inter-transaction time for reversing

Each column reports coefficients for the given decile of firms.

Coefficient estimates and robust standard errors (in parenthesis) for the equation

Estimated Coefficients for the Information Content of Inter-Transaction Time Equation for Reversing Trades

Table XI

37

4

4

i =1

i =0

i =0

i =0

z =10 th % tile

i =0

5

Lm

DL

DT

0.64

0.0335
(0.0511)
0.0302
(0.0535)
0.0245
(0.0615)
0.0680
(0.0532)
0.0382
(0.0555)
0.0013
(0.0003)**
1030

0.0277
(0.0029)**
0.0089
(0.0030)**
0.0064
(0.0027)*
0.0021
(0.0029)
-0.0046
(0.0025)

Robust standard errors in parentheses
* significant at 5%; ** significant at 1%

Obs
R-squared

Constant

c4

c3

c2

c1

Lags of
c0

b4

b3

b2

b1

Lags of
b0

a4

a3

a2

0.4457
(0.0433)**
0.1540
(0.0447)**
0.0876
(0.0475)
0.1778
(0.0417)**

Most actively traded

-0.1825
(0.0696)**
0.0850
(0.0775)
0.0976
(0.0799)
0.1319
(0.0800)
-0.0641
(0.0814)
0.0095
(0.0015)**
1030

0.0103
(0.0028)**
0.0048
(0.0028)
0.0005
(0.0029)
0.0020
(0.0025)
-0.0004
(0.0026)

0.3440
(0.0458)**
0.1820
(0.0673)**
0.0431
(0.0514)
0.1061
(0.0529)*

0.36

0.52

i =0

0.46

-0.3704
(0.0483)**
-0.0105
(0.0578)
0.0982
(0.0652)
-0.0446
(0.0673)
-0.0055
(0.0563)
0.0095
(0.0022)**
1030

-0.0013
(0.0011)
-0.0022
(0.0013)
-0.0007
(0.0013)
0.0009
(0.0011)
-0.0010
(0.0010)

0.3484
(0.0480)**
0.1585
(0.0698)*
0.1681
(0.0540)**
0.0404
(0.0467)

0.36

-0.6461
(0.0888)**
-0.2477
(0.0723)**
-0.0682
(0.0935)
0.1304
(0.1070)
0.0923
(0.0800)
0.0113
(0.0020)**
1030

0.0051
(0.0019)**
0.0050
(0.0017)**
-0.0005
(0.0017)
-0.0011
(0.0013)
-0.0031
(0.0014)*

Tm

i =0

…

0.25

-0.4569
(0.0645)**
-0.2516
(0.0756)**
-0.1548
(0.0635)*
-0.0730
(0.0632)
-0.0664
(0.0487)
0.0161
(0.0024)**
1030

0.0019
(0.0013)
0.0019
(0.0013)
0.0004
(0.0013)
0.0010
(0.0011)
0.0021
(0.0010)*

0.2013
(0.0442)**
0.1637
(0.0462)**
0.2000
(0.0463)**
0.0612
(0.0458)

z = average

-0.4400
(0.0736)**
-0.1397
(0.0744)
-0.1517
(0.0861)
-0.1984
(0.1117)
-0.0541
(0.0805)
0.0164
(0.0063)**
1030

0.0003
(0.0010)
0.0014
(0.0011)
-0.0006
(0.0011)
-0.0001
(0.0012)
0.0006
(0.0011)

0.4400
(0.1212)**
0.1740
(0.0511)**
-0.0428
(0.0795)
0.1176
(0.0511)*

0.33

. Parameters are estimated from equation 6.

TZ

is the z-

is the proxy for the

, and

0.31

-0.5153
(0.0603)**
-0.2129
(0.0701)**
-0.0393
(0.0704)
0.0094
(0.0665)
0.0434
(0.0392)
0.0201
(0.0034)**
1030

0.0003
(0.0008)
0.0010
(0.0008)
0.0010
(0.0007)
0.0010
(0.0008)
0.0005
(0.0006)

0.24

-0.4613
(0.0455)**
-0.2833
(0.0719)**
-0.1937
(0.0576)**
-0.0346
(0.0417)
0.0341
(0.0291)
0.0288
(0.0045)**
1030

-0.0011
(0.0008)
-0.0002
(0.0009)
0.0011
(0.0008)
0.0008
(0.0007)
0.0003
(0.0007)

0.1211
(0.0451)**
0.1074
(0.0393)**
0.1315
(0.0411)**
0.0232
(0.0460)

Least actively traded

represents first differences.

0.1860
(0.0477)**
0.1923
(0.0542)**
0.1466
(0.0441)**
0.1078
(0.0460)*

D

0.18

-0.0940
(0.0197)**
-0.0813
(0.0163)**
-0.0732
(0.0158)**
-0.0308
(0.0135)*
-0.0052
(0.0079)
0.0359
(0.0089)**
1000

0.0001
(0.0013)
-0.0002
(0.0015)
-0.0014
(0.0012)
-0.0001
(0.0013)
-0.0004
(0.0009)

0.0709
(0.0598)
0.0229
(0.0399)
0.1505
(0.0742)*
0.0703
(0.0604)

is the average inter-transaction time for same-direction trades during month

rs
rs
+ å [g im + d im ln(1 + T z )]
5

Firm category

z = average

0.2638
(0.0477)**
0.1560
(0.0480)**
0.0637
(0.0470)
0.1813
(0.0562)**

r
r
Lm º å [g im + d im ln(1 + T z )]

-0.2108
(0.0640)**
0.0410
(0.0651)
0.0434
(0.0585)
-0.0077
(0.0580)
0.0617
(0.0585)
0.0040
(0.0008)**
1030

0.0088
(0.0020)**
0.0032
(0.0019)
-0.0002
(0.0017)
-0.0006
(0.0019)
-0.0006
(0.0017)

0.3478
(0.0459)**
0.1911
(0.0389)**
0.0872
(0.0436)*
0.1952
(0.0429)**

is the average price impact of a same-direction trade, defined as

Lags of y
a1

m.

5

z =10 th % tile

r
r
rs
rs
y m º å [g im + d im ln(1 + Tz )] z = 90th % tile + å [g im + d im ln(1 + Tz )] z =90 th % tile
5

4

y m = å ai y m -i + å bi DTm + å ci DLm + em . Each column reports coefficients for the given decile of firms. y m

percentile of the inter-transaction time distribution for month m for reversing and same-direction trades, depending on which sum is being evaluated.

information content of inter-transaction time in month m defined as

Coefficient estimates and robust standard errors (in parenthesis) for the equation

Estimated Coefficients for the Information Content of Inter-Transaction Time Equation for Same-Direction Trades

Table XII

38

1400

30

1200

25

1000

20

800

15

600

10

400

5

200

0
Jan-93 Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01

0

trades in millions

$ volume (billions)

1600

35
# of trades (millions)

40

value in billions

Figure 1. Trading activity on the NYSE. The dotted line graphs the aggregate number of trades for all
stocks listed on the NYSE each month from January 1993 to December 2001. The solid line graphs the
dollar value of these same trades. Source: NYSE.

39

60
50
40
30
20
10
0
Jan-93 Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01
Cat. 1 (in seconds)

Cat. 5 (in minutes)

Cat. 9 (in minutes)

Figure 2. The evolution of trading activity across stocks. Plots the average time between trades for
stocks with different levels of average trading activity during each month between January 1993 and
December 2001. Firms in each category were based on the 100 firms listed in Table I and were grouped
into 10 deciles according to their total number of trades over the 9 year sample period. Stocks in category
one were the most actively traded and stocks in category ten were the least actively traded. Source: TAQ.

40

40
35
30
25
20
15
10
5
0
Jan-93 Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01
10th percentile

25th percentile

50th percentile

Figure 3. The changing distribution of inter-transaction time for actively traded stocks. Plots the 10th,
25th, and 50th percentile of the distribution of inter-transaction time for stocks in the most actively traded
decile of firms listed in Table I each month from January 1993 to December 2001. Source: TAQ.

41

0.004

45
40

0.003

35
0.002

30
25

0.001

20

0

15

-0.001

10
-0.002

5
0
Jan-93 Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01
average inter-transaction time

-0.003

sum of delta coefficients

Figure 4. Trade activity and the information content of inter-transaction time for the Disney
Company. The solid line plots the average time, in seconds, between trades in stock of the Disney
Company, ticker symbol DIS. The dotted line plots the sum of the d ir coefficients from the equation

rt = å ai rt -i + lropen Dt xt0 + å [g ir + d ir ln(1 + Tt -i )]xt0-i + n t
5

5

i =1

i =0

estimated each month from January 1993 to December 2001 using TAQ data for DIS.

42

0.025

0.02

0.015

0.01

0.005

0
Jan93

Jan94

Jan95

Jan96

Jan97

at average time

Jan98

Jan99

at 10th pctile

Jan00

Jan01

at 90th pctile

Figure 5. Estimated price impact of trading of stock in the Disney Company at different intertransaction times. The dotted line plots the sum of the g ir + d ir ln(1 + Tt -i ) coefficients estimated each
month from the equation

rt = å ai rt -i + lropen Dt xt0 + å [g ir + d ir ln(1 + Tt -i )]xt0-i + n t
5

5

i =1

i =0

evaluated at the mean inter-transaction time for that month. The two solid lines plot the same quantity, only
evaluated at the 10th and 90th percentile of each month’s inter-transaction time distribution.

43

0.03

60

0.025

50

0.02

40

0.015

30

0.01

20

0.005

10

0

0
Jan93

Jan94

Jan95

at 10th pctile

Jan96

Jan97

at 90th pctile

Jan98

Jan99

Jan00

Jan01

average inter-transaction time

Figure 6. Estimated price impact of trading at different inter-transaction times and average intertransaction time for actively traded stocks. The dotted line plots the sum of the g ir + d ir ln(1 + Tt -i )
coefficients estimated each month from the equation

rt = å ai rt -i + lropen Dt xt0 + å [g ir + d ir ln(1 + Tt -i )]xt0-i + n t
5

5

i =1

i =0

evaluated at the 90th percentile of the inter-transaction distribution and then averaged across all stocks in
the most heavily traded decile. The solid line is the analogous calculation for the 10th percentile. The boxed
line is the average inter-transaction time across all stocks in the most heavily traded decile.

44

0.03

70

0.025

60
50

0.02

40
0.015
30
0.01

20

0.005

10

0

0
Jan93

Jan94

Jan95

at 10th pctile

Jan96

Jan97

at 90th pctile

Jan98

Jan99

Jan00

Jan01

average inter-transaction time

Figure 7. Estimated price impact of trading at different inter-transaction times and average intertransaction time for actively traded stocks when the trade indicator measures the (log of the) share of
market value transacted. The dotted line plots the sum of the g ir + d ir ln(1 + Tt -i ) coefficients estimated
each month from the equation

rt = å a i rt -i + l ropen Dt x t + å [g ir + d ir ln(1 + Tt -i )]x t -i + n t
5

5

i =1

i =0

evaluated at the 90th percentile of the inter-transaction distribution and then averaged across all stocks in
the most heavily traded decile. The solid line is the analogous calculation for the 10th percentile. The boxed
line is the average inter-transaction time across all stocks in the most heavily traded decile.

45

0.06
0.05
0.04
0.03
0.02
0.01
0
Jan93

Jan94

Jan95

Jan96

Jan97

Jan98

Jan99

Jan00

Jan01

at 10th pctile, reversing

at 90th pctile, reversing

at 10th pctile, same direction

at 90th pctile, same direction

Figure 8. Estimated price impact of trading at different inter-transaction times for actively traded
stocks when the specification controls for whether each trade is same-direction or reversing. The
dotted line plots the sum of the g ir + d ir ln(1 + Tt -i ) coefficients estimated each month from the equation

rt = å ai rt -i + lropen Dt xt0 + å [g ir + d ir ln(1 + Tt -i )]xt0-i + å [g irs + d irs ln(1 + Tt -i )]S t -i xt0-i + n 1,t
5

5

5

i =1

i =0

i=0

evaluated at the 10th percentile of the inter-transaction distribution and then averaged across all stocks in
the most heavily traded decile. The solid line is the analogous calculation for the 90th percentile. The lines
with triangles and squares report the analogous values for reversing trades, namely
g ir + d ir ln(1 + Tt -i ) + g irs + d irs ln(1 + Tt -i ) evaluated at the 10th and 90th percentile of inter-transaction time
for same-direction trades.

46

0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
Jan93

Jan94

Jan95

Jan96

Jan97

Jan98

Jan99

Jan00

Jan01

at 10th pctile, reversing

at 90th pctile, reversing

at 10th pctile, same direction

at 90th pctile, same direction

Figure 9. Estimated price impact of trading at different inter-transaction times for stocks in the fifth
decile of trading activity when the specification controls for whether each trade is same-direction or
reversing. The dotted line plots the sum of the g ir + d ir ln(1 + Tt -i ) coefficients estimated each month from
the equation

rt = å ai rt -i + lropen Dt xt0 + å [g ir + d ir ln(1 + Tt -i )]xt0-i + å [g irs + d irs ln(1 + Tt -i )]S t -i xt0-i + n 1,t
5

5

5

i =1

i =0

i=0

evaluated at the 10th percentile of the inter-transaction distribution and then averaged across all stocks in
the fifth decile of trading activity. The solid line is the analogous calculation for the 90th percentile. The
lines with triangles and squares report the analogous values for reversing trades, namely
g ir + d ir ln(1 + Tt -i ) + g irs + d irs ln(1 + Tt -i ) evaluated at the 10th and 90th percentile of inter-transaction time
for same-direction trades.

47

Working Paper Series
A series of research studies on regional economic issues relating to the Seventh Federal
Reserve District, and on financial and economic topics.
Dynamic Monetary Equilibrium in a Random-Matching Economy
Edward J. Green and Ruilin Zhou

WP-00-1

The Effects of Health, Wealth, and Wages on Labor Supply and Retirement Behavior
Eric French

WP-00-2

Market Discipline in the Governance of U.S. Bank Holding Companies:
Monitoring vs. Influencing
Robert R. Bliss and Mark J. Flannery

WP-00-3

Using Market Valuation to Assess the Importance and Efficiency
of Public School Spending
Lisa Barrow and Cecilia Elena Rouse
Employment Flows, Capital Mobility, and Policy Analysis
Marcelo Veracierto
Does the Community Reinvestment Act Influence Lending? An Analysis
of Changes in Bank Low-Income Mortgage Activity
Drew Dahl, Douglas D. Evanoff and Michael F. Spivey

WP-00-4

WP-00-5

WP-00-6

Subordinated Debt and Bank Capital Reform
Douglas D. Evanoff and Larry D. Wall

WP-00-7

The Labor Supply Response To (Mismeasured But) Predictable Wage Changes
Eric French

WP-00-8

For How Long Are Newly Chartered Banks Financially Fragile?
Robert DeYoung

WP-00-9

Bank Capital Regulation With and Without State-Contingent Penalties
David A. Marshall and Edward S. Prescott

WP-00-10

Why Is Productivity Procyclical? Why Do We Care?
Susanto Basu and John Fernald

WP-00-11

Oligopoly Banking and Capital Accumulation
Nicola Cetorelli and Pietro F. Peretto

WP-00-12

Puzzles in the Chinese Stock Market
John Fernald and John H. Rogers

WP-00-13

The Effects of Geographic Expansion on Bank Efficiency
Allen N. Berger and Robert DeYoung

WP-00-14

Idiosyncratic Risk and Aggregate Employment Dynamics
Jeffrey R. Campbell and Jonas D.M. Fisher

WP-00-15

1

Working Paper Series (continued)
Post-Resolution Treatment of Depositors at Failed Banks: Implications for the Severity
of Banking Crises, Systemic Risk, and Too-Big-To-Fail
George G. Kaufman and Steven A. Seelig

WP-00-16

The Double Play: Simultaneous Speculative Attacks on Currency and Equity Markets
Sujit Chakravorti and Subir Lall

WP-00-17

Capital Requirements and Competition in the Banking Industry
Peter J.G. Vlaar

WP-00-18

Financial-Intermediation Regime and Efficiency in a Boyd-Prescott Economy
Yeong-Yuh Chiang and Edward J. Green

WP-00-19

How Do Retail Prices React to Minimum Wage Increases?
James M. MacDonald and Daniel Aaronson

WP-00-20

Financial Signal Processing: A Self Calibrating Model
Robert J. Elliott, William C. Hunter and Barbara M. Jamieson

WP-00-21

An Empirical Examination of the Price-Dividend Relation with Dividend Management
Lucy F. Ackert and William C. Hunter

WP-00-22

Savings of Young Parents
Annamaria Lusardi, Ricardo Cossa, and Erin L. Krupka

WP-00-23

The Pitfalls in Inferring Risk from Financial Market Data
Robert R. Bliss

WP-00-24

What Can Account for Fluctuations in the Terms of Trade?
Marianne Baxter and Michael A. Kouparitsas

WP-00-25

Data Revisions and the Identification of Monetary Policy Shocks
Dean Croushore and Charles L. Evans

WP-00-26

Recent Evidence on the Relationship Between Unemployment and Wage Growth
Daniel Aaronson and Daniel Sullivan

WP-00-27

Supplier Relationships and Small Business Use of Trade Credit
Daniel Aaronson, Raphael Bostic, Paul Huck and Robert Townsend

WP-00-28

What are the Short-Run Effects of Increasing Labor Market Flexibility?
Marcelo Veracierto

WP-00-29

Equilibrium Lending Mechanism and Aggregate Activity
Cheng Wang and Ruilin Zhou

WP-00-30

Impact of Independent Directors and the Regulatory Environment on Bank Merger Prices:
Evidence from Takeover Activity in the 1990s
Elijah Brewer III, William E. Jackson III, and Julapa A. Jagtiani
Does Bank Concentration Lead to Concentration in Industrial Sectors?
Nicola Cetorelli

WP-00-31

WP-01-01

2

Working Paper Series (continued)
On the Fiscal Implications of Twin Crises
Craig Burnside, Martin Eichenbaum and Sergio Rebelo

WP-01-02

Sub-Debt Yield Spreads as Bank Risk Measures
Douglas D. Evanoff and Larry D. Wall

WP-01-03

Productivity Growth in the 1990s: Technology, Utilization, or Adjustment?
Susanto Basu, John G. Fernald and Matthew D. Shapiro

WP-01-04

Do Regulators Search for the Quiet Life? The Relationship Between Regulators and
The Regulated in Banking
Richard J. Rosen
Learning-by-Doing, Scale Efficiencies, and Financial Performance at Internet-Only Banks
Robert DeYoung
The Role of Real Wages, Productivity, and Fiscal Policy in Germany’s
Great Depression 1928-37
Jonas D. M. Fisher and Andreas Hornstein

WP-01-05

WP-01-06

WP-01-07

Nominal Rigidities and the Dynamic Effects of a Shock to Monetary Policy
Lawrence J. Christiano, Martin Eichenbaum and Charles L. Evans

WP-01-08

Outsourcing Business Service and the Scope of Local Markets
Yukako Ono

WP-01-09

The Effect of Market Size Structure on Competition: The Case of Small Business Lending
Allen N. Berger, Richard J. Rosen and Gregory F. Udell

WP-01-10

Deregulation, the Internet, and the Competitive Viability of Large Banks
and Community Banks
Robert DeYoung and William C. Hunter

WP-01-11

Price Ceilings as Focal Points for Tacit Collusion: Evidence from Credit Cards
Christopher R. Knittel and Victor Stango

WP-01-12

Gaps and Triangles
Bernardino Adão, Isabel Correia and Pedro Teles

WP-01-13

A Real Explanation for Heterogeneous Investment Dynamics
Jonas D.M. Fisher

WP-01-14

Recovering Risk Aversion from Options
Robert R. Bliss and Nikolaos Panigirtzoglou

WP-01-15

Economic Determinants of the Nominal Treasury Yield Curve
Charles L. Evans and David Marshall

WP-01-16

Price Level Uniformity in a Random Matching Model with Perfectly Patient Traders
Edward J. Green and Ruilin Zhou

WP-01-17

Earnings Mobility in the US: A New Look at Intergenerational Inequality
Bhashkar Mazumder

WP-01-18

3

Working Paper Series (continued)
The Effects of Health Insurance and Self-Insurance on Retirement Behavior
Eric French and John Bailey Jones

WP-01-19

The Effect of Part-Time Work on Wages: Evidence from the Social Security Rules
Daniel Aaronson and Eric French

WP-01-20

Antidumping Policy Under Imperfect Competition
Meredith A. Crowley

WP-01-21

Is the United States an Optimum Currency Area?
An Empirical Analysis of Regional Business Cycles
Michael A. Kouparitsas

WP-01-22

A Note on the Estimation of Linear Regression Models with Heteroskedastic
Measurement Errors
Daniel G. Sullivan

WP-01-23

The Mis-Measurement of Permanent Earnings: New Evidence from Social
Security Earnings Data
Bhashkar Mazumder

WP-01-24

Pricing IPOs of Mutual Thrift Conversions: The Joint Effect of Regulation
and Market Discipline
Elijah Brewer III, Douglas D. Evanoff and Jacky So

WP-01-25

Opportunity Cost and Prudentiality: An Analysis of Collateral Decisions in
Bilateral and Multilateral Settings
Herbert L. Baer, Virginia G. France and James T. Moser

WP-01-26

Outsourcing Business Services and the Role of Central Administrative Offices
Yukako Ono

WP-02-01

Strategic Responses to Regulatory Threat in the Credit Card Market*
Victor Stango

WP-02-02

The Optimal Mix of Taxes on Money, Consumption and Income
Fiorella De Fiore and Pedro Teles

WP-02-03

Expectation Traps and Monetary Policy
Stefania Albanesi, V. V. Chari and Lawrence J. Christiano

WP-02-04

Monetary Policy in a Financial Crisis
Lawrence J. Christiano, Christopher Gust and Jorge Roldos

WP-02-05

Regulatory Incentives and Consolidation: The Case of Commercial Bank Mergers
and the Community Reinvestment Act
Raphael Bostic, Hamid Mehran, Anna Paulson and Marc Saidenberg
Technological Progress and the Geographic Expansion of the Banking Industry
Allen N. Berger and Robert DeYoung

WP-02-06

WP-02-07

4

Working Paper Series (continued)
Choosing the Right Parents: Changes in the Intergenerational Transmission
of Inequality  Between 1980 and the Early 1990s
David I. Levine and Bhashkar Mazumder

WP-02-08

The Immediacy Implications of Exchange Organization
James T. Moser

WP-02-09

Maternal Employment and Overweight Children
Patricia M. Anderson, Kristin F. Butcher and Phillip B. Levine

WP-02-10

The Costs and Benefits of Moral Suasion: Evidence from the Rescue of
Long-Term Capital Management
Craig Furfine

WP-02-11

On the Cyclical Behavior of Employment, Unemployment and Labor Force Participation
Marcelo Veracierto

WP-02-12

Do Safeguard Tariffs and Antidumping Duties Open or Close Technology Gaps?
Meredith A. Crowley

WP-02-13

Technology Shocks Matter
Jonas D. M. Fisher

WP-02-14

Money as a Mechanism in a Bewley Economy
Edward J. Green and Ruilin Zhou

WP-02-15

Optimal Fiscal and Monetary Policy: Equivalence Results
Isabel Correia, Juan Pablo Nicolini and Pedro Teles

WP-02-16

Real Exchange Rate Fluctuations and the Dynamics of Retail Trade Industries
on the U.S.-Canada Border
Jeffrey R. Campbell and Beverly Lapham

WP-02-17

Bank Procyclicality, Credit Crunches, and Asymmetric Monetary Policy Effects:
A Unifying Model
Robert R. Bliss and George G. Kaufman

WP-02-18

Location of Headquarter Growth During the 90s
Thomas H. Klier

WP-02-19

The Value of Banking Relationships During a Financial Crisis:
Evidence from Failures of Japanese Banks
Elijah Brewer III, Hesna Genay, William Curt Hunter and George G. Kaufman

WP-02-20

On the Distribution and Dynamics of Health Costs
Eric French and John Bailey Jones

WP-02-21

The Effects of Progressive Taxation on Labor Supply when Hours and Wages are
Jointly Determined
Daniel Aaronson and Eric French

WP-02-22

5

Working Paper Series (continued)
Inter-industry Contagion and the Competitive Effects of Financial Distress Announcements:
Evidence from Commercial Banks and Life Insurance Companies
Elijah Brewer III and William E. Jackson III

WP-02-23

State-Contingent Bank Regulation With Unobserved Action and
Unobserved Characteristics
David A. Marshall and Edward Simpson Prescott

WP-02-24

Local Market Consolidation and Bank Productive Efficiency
Douglas D. Evanoff and Evren Örs

WP-02-25

Life-Cycle Dynamics in Industrial Sectors. The Role of Banking Market Structure
Nicola Cetorelli

WP-02-26

Private School Location and Neighborhood Characteristics
Lisa Barrow

WP-02-27

Teachers and Student Achievement in the Chicago Public High Schools
Daniel Aaronson, Lisa Barrow and William Sander

WP-02-28

The Crime of 1873: Back to the Scene
François R. Velde

WP-02-29

Trade Structure, Industrial Structure, and International Business Cycles
Marianne Baxter and Michael A. Kouparitsas

WP-02-30

Estimating the Returns to Community College Schooling for Displaced Workers
Louis Jacobson, Robert LaLonde and Daniel G. Sullivan

WP-02-31

A Proposal for Efficiently Resolving Out-of-the-Money Swap Positions
at Large Insolvent Banks
George G. Kaufman

WP-03-01

Depositor Liquidity and Loss-Sharing in Bank Failure Resolutions
George G. Kaufman

WP-03-02

Subordinated Debt and Prompt Corrective Regulatory Action
Douglas D. Evanoff and Larry D. Wall

WP-03-03

When is Inter-Transaction Time Informative?
Craig Furfine

WP-03-04

6