The current issue and full text archive of this journal is available at

www.emeraldinsight.com/0307-4358.htm

MF
36,2
The effect of stock splits on
iShare exchange-traded funds
Pia Bandyopadhyay
College of Business Administration, California State University,
134 Long Beach, California, USA
James Hackard
Texas A&M University San Antonio, San Antonio, Texas, USA, and
Yiuman Tse
College of Business, University of Texas at San Antonio,
San Antonio, Texas, USA
Abstract
Purpose – The purpose of this paper is to examine the pre- and post-split behavior for trades and
quotes of iShare exchange-traded funds (ETFs) that split in June 2005. The objective is to determine
whether post-split changes in the bid-ask spread, trade turnover, average dollar-size trade, frequency
of small trades, trade price location, and order imbalance support either or both of the two widely
examined hypotheses for the motivation for share splits.
Design/methodology/approach – The impact of the iShares split around the split date was
studied, using the measures above to examine the support, if any, for each of two hypotheses, broker
promotion and/or the trading inconvenience, with regard to the sample and time period under study.
Findings – Bid-ask spread, average dollar order size, and frequency of small trades were found to
fail to reject the broker-promotion hypothesis, while the increase in post-split turnover fails to reject
the trading-inconvenience hypothesis. Changes in the trade-price-location parameter and in order
imbalance fail to support either hypothesis.
Practical implications – Because of the importance of basket securities in the determination of the
prices for listed securities, issuers of these securities, investors and regulators should be interested
whether the price behavior of splitting iShares is similar to that experienced in other securities.
Originality/value – Numerous studies in the literature have investigated the effects of stock splits
on individual securities, but it is believed, none has yet appeared studying the recent splits in iShares.
Keywords Bid offer spreads, Share prices, Share values, Investment funds, Stock exchanges
Paper type Research paper

Managerial Finance
Vol. 36 No. 2, 2010
pp. 134-159
# Emerald Group Publishing Limited
0307-4358
DOI 10.1108/03074351011014550
Introduction
Corporations issuing stocks and investment companies trading stocks face the
challenge of providing securities that can be purchased by a wide variety of investors
who wish to diversify their investment risk. Companies have used stock splits of shares
whose price is increasing to attract a broader base of shareholders. For existing
shareholders, the effect of the split is cosmetic, as their proportionate ownership is the
same after the split as before, and the company’s cash flows are unaffected. Stock splits
have been widely investigated in finance literature.
Stock splits have been common for individually listed securities, but only recently
have splits occurred in the exchange-traded-funds (ETF) market. In June 2005,
Barclays Global Investors undertook the first-ever share splits for 12 of the iShare
ETFs they had issued, thus allowing smaller initial investments in the relatively
higher-priced funds. Their goal was to cause the price per share of the iShare funds to
approximate the average share price of a US listed security (stock).
On June 9, 2005, share prices of the 12 iShare (ETFs) dropped to recognize the effect
of splits for those ETFs. We study pre-and post-split behavior of the measures
indicative of the motivations of the investors of these funds and of the brokers selling Stock splits on
the funds. We investigate two hypotheses that have been of interest in split literature; iShare exchange-
broker promotion and trading inconvenience. The broker-promotion hypothesis
(Angel, 1997; Schultz, 2000) is based on the premise that a post-split increase in relative traded funds
spreads motivates brokers to promote the splitting stocks to small investors. The
trading-inconvenience hypothesis posits that investors will postpone purchases until
after the ex-date in order not to deal with the inconvenience of the due-bill process, 135
which requires that stockholders who sell their shares between the record date of the
split and the ex-date of the split must provide the broker a document (the due bill)
stating that additional shares are due the purchaser as a result of the split.
Measures that could favor or reject either hypothesis include: relative bid-ask
spread, volume turnover, frequency of small trades, transaction price placement, and
order imbalance. We detect mixed results, with the increase in average daily spreads,
decline in average order size, and increase in frequency of small trades failing to reject
the broker-promotion hypothesis, while the increase in turnover following the split
fails to reject the trading-inconvenience hypothesis. Changes in trade-price-location
parameter and order imbalance fail to support either hypothesis.
The growing significance of trading in ‘‘basket securities’’ such as ETFs has meant
that changes in the prices of the individual securities upon which the ETFs are based
are often determined by price changes in the ETF markets. The iShares series of ETFs
issued by Barclays Global International track major market indexes and cover a broad
spectrum of investment market capitalization, investment styles, market sectors, and
investment options across domestic and global markets.
Because of the importance of basket securities in the determination of the prices
for listed securities, issuers of these securities, investors, and regulators should be
interested whether the price behavior of splitting iShares is similar to that experienced
in other security markets. A point of particular interest should be the difference
between the fees in ETF markets and those of regular mutual funds, determined by the
bid-ask spread and broker commissions.
Bid-ask spread is an implicit transaction cost that investors bear when buying
or selling an ETF. This transaction cost is approximately one-half of the ETFs spread.
Conventional mutual funds are exempt from spread fees since investors purchase at
the net asset value of the fund. Also, purchasing no-load funds directly from a fund
company eliminates broker commissions in mutual funds, while trading in most ETFs
involves broker-dealer commissions that can range between $10 and $50 per trade,
depending on the size of the trade, the assets invested, and the type of broker i.e.
discount or full service. All of these costs can be exacerbated through active trading or
by holding an ETF for short periods, or both.
Spread costs are particularly interesting since they give rise to a paradox. On the
one hand these can be minimized through a passive buy and hold strategy, but doing so
would defeat the advantage of intraday trading offered by ETF markets. ETFs were
originally constructed to provide a single security that tracks a major index, equity
sector, international market, region, country, or commodity. Intraday trading of ETFs
enables investors to buy or sell all of the securities that make up an entire market with
a single trade, thereby providing the flexibility to get into or out of a position at any
time throughout the day. Traders thus have the opportunity to track the direction of the
market, and execute accordingly, allowing active traders to take advantage of short-
term movements in the market while still maintaining a passive strategy.
MF With the overwhelmingly large number of ETFs that have recently begun trading,
36,2 investors now have a plethora of funds from which to choose[1]. Competition has
driven the cost of buying ETF shares to low levels, and brokers are under mounting
pressure to promote the funds, and be more aggressive when pitching their products.
Marketing of ETFs is a recently observed trend, and mutual funds outsold ETFs by
as much as 50 to 1 in the latter 1990s. Predictions of the Financial Research Corporation
are that by the year 2007 total ETF assets will approach $1 trillion, but that is still not
136 as large as the $7 trillion assets of the mutual funds industry in 2003. An explanation
for this is that, until recently, ETFs were benign instruments that nobody ‘‘sold’’, unlike
mutual funds that have always been actively promoted by the sell side. Competition
appears to have altered this outlook. Our results indicate that besides the intention to
increase customer base, a split provides brokers added incentive to promote an ETF to
investors, in order to profit from higher spreads.
Effective promotion entails active trading, leading to higher spreads for brokers,
since trading characteristics of ETFs resemble those of single stocks. At the same time,
they offer certain advantages over both mutual funds and stocks.
Advantages over mutual funds include:
. Continuous price determination and share pricing of ETFs vs end of day pricing
and trading of mutual funds.
. Purchase and sale of ETF shares on margin, which is not allowed for mutual
funds.
. Placement of stop and limit orders for ETFs, also not allowed for mutual funds.
Advantages of ETFs compared to owning a stock include:
. Holding ETF securities offers investors the advantage of maximum diversification
of firm-specific risk.
. Cost of informed trading is minimized with basket securities.
. iShares have no minimum required investment.
. Odd-lot trading of iShares is allowed.
Our research investigates the effects of the split in iShares at the time the resulting
share prices arising from the splits are reflected in the funds (the ‘‘ex-date’’ price). There
are numerous studies in the literature investigating the effects of stock splits on
individual securities, but to our knowledge, none has yet appeared studying the recent
splits in iShares. We investigate the effect of iShares splits in two broad categories:
transaction costs and trading characteristics around the event date. The analysis of
transactions costs employs the measurement of the bid-ask spread, which determines
the trader’s profit. The possibility of an increase in spreads following the split might
motivate a broker to promote the shares of a fund that has split. The effects of the splits
on number of trade and average size trade are summarized in Appendix 1.
The analysis of trading characteristics involves several variables. First, we
determine the volume turnover, which measures the volume of trades as a percentage
of the number of shares outstanding. This measurement indicates how actively a stock
is traded. Second, we look at the average trade size, computed as total amount of trades
(shares of the fund traded) divided by the number of transactions, and the frequency
of small trades (those of one or two ‘‘lots’’). A lot is the minimum number of shares that
may be traded at one time. For stocks the usual minimum lot size is 100 shares.
The change in average order size and frequency of small trades indicates whether Stock splits on
brokers are motivated to promote the splitting shares to smaller investors.
The third variable we study is the placement of the transaction price relative to the
iShare exchange-
bid and ask prices. Having more trades executed around the ask price would indicate traded funds
pressure to purchase, while more trades executed around the bid price would indicate
pressure to sell. The final variable we examine is order imbalance, which is the excess
of buy orders over sell orders, expressed as a percentage of total trades executed during
each day. A positive order imbalance is indicative of the pressure to purchase shares.
137
Recall that the objective of our analysis is to test the two major hypotheses
regarding the effects and motivations for stock splits; the broker-promotion hypothesis
and the trading-inconvenience hypothesis. If the evidence shows that brokers are
motivated to promote more investments by small investors after the split of the
iShares, and that the wider spreads following the split occur to the disadvantage of
these investors, this finding would suggest that regulators should step up their
supervision of brokers to insure that they act in the best interest of investors with
regard to share splits.
The rest of the study is organized as follows. The following section discusses prior
literature and develops the hypotheses. The next section describes the data and
preliminary descriptive statistics. The penultimate section lays out the methodology
and analyzes the evidence. The last section briefly summarizes and concludes the
study.

Prior literature and hypotheses

Several studies have investigated the effects of stock splits on individual securities,
primarily looking at the broker-promotion hypothesis and the trading-inconvenience
hypothesis. The broker-promotion hypothesis is based on the premise that a post-split
increase in relative spreads (Angel, 1997; Schultz, 2000) motivates brokers to promote
these stocks to small investors and profit from the higher spreads.
The trading-inconvenience hypothesis posits that investors prefer not to deal with
the inconvenience of the due-bill process and will postpone purchases until after the
ex-date. This is because the stock split makes it unattractive for investors to trade
the stock between the record date and the ex-date as they have to deal with attached
due bills, which are documents telling the broker that additional shares are due to the
purchasing shareholder after the split.
Studies advancing support for the broker-promotion hypothesis include Baker and
Gallagher (1980), Angel (1997), and Schultz (2000). Schultz (2000) identifies a significant
increase in the number of small buy orders after a split as evidence of the broker-
promotion hypothesis. Grinblatt et al. (1984) find significant cumulative abnormal
returns for the three-day period extending from one day before the ex-date until one day
after the ex-date. They explain these abnormal returns as the result of broker promotion.
Maloney and Mulherin (1992) support the broker-promotion hypothesis on the basis that
closing prices on or after the ex-date are closer to the ask than to the bid price, resulting
in a higher ex-split return due to an increase in small investor buying accompanied by
higher spreads.
Nayar and Rozeff (2001) offer support for the trading-inconvenience hypothesis.
They provide evidence that relates record-date returns, when-issued premiums, and
ex-date returns to lower stock prices that are probably a result of investor aversion to
splits. The trading-inconvenience theory is consistent with the microstructure view of
splits. Easley et al. (2001) report an increase in trading costs for uninformed investors
MF after a stock split. Conrad and Conroy (1994) provide evidence of changes in order
36,2 flow around the ex-date that may be considered an indication of investors avoiding
purchases of shares until after a split.
While the studies above examine splits in common stocks, the first study that
investigates splits in mutual funds is by Rozeff (1998). He finds that, unlike common
stocks, mutual funds are not associated with post-split capital appreciation, increase in
the number of shareholders, or an increase in total assets. The only effect of the split is
138 to bring the level of the average per-account number of shares back into the range of
non-splitting funds. Rozeff cites this as the only likely rationale for a mutual fund split.
Dennis (2003) examines the March 2000 two-for-one split of the Nasdaq-100 Index
Tracking Stock. He argues that the absence of a signaling effect in the index stock split
allows investigation of the pure liquidity effects of the split. He finds post-split
improvement in liquidity for smaller trades which helps smaller investors. However,
the split puts large traders at a disadvantage as the wider bid-ask spread results in
higher trading costs that cut into the profit of these traders. In an earlier study, Conroy
et al. (1990) provide evidence of post-split wider spreads in New York Stock Exchange
(NYSE) listed equities.
We investigate the recent splits in the increasingly popular iShares, which is
interesting for two reasons. First, the iShares are among the first ETFs to split; and
second, they comprise an important part of equity markets, as they offer investors the
advantages of traditional mutual funds in addition to the trading flexibility of a stock.
We test the two most popular hypotheses forwarded by the literature with respect to
splits: the broker-promotion hypothesis and the trading-inconvenience hypothesis.
We use measures of liquidity, order-flow characteristics, and volume turnover to
analyze the hypotheses. If the broker-promotion hypothesis holds we would expect to
find significant decreases in average buy-order size, corresponding increases in
frequency of small trades, and transaction prices closer to ask prices within the bid-ask
spread. Conversely, fewer buy orders, coupled with transaction prices closer to bid
prices, would lend support to the trading-inconvenience hypothesis. Inconsistencies in
evidence would suggest that the motives for fund managers in splitting iShares differ
from the motives surrounding a company manager’s decision to split common stock.
For example, Rozeff (1998) suggests that, all else equal, investors prefer a lower priced
fund and that managers respond to this demand by splitting the stock.

Data and price statistics

We examine 12 iShare funds that split in June 2005, all of which are traded on the
American Stock Exchange. Additionally, we designate a sample of 12 non-splitting
iShare funds as the control group. We repeat the analyses for the control group, to
compare and contrast the results between the splitting and non-splitting funds.
We select the 12 most actively traded funds from among those that did not split.
Appendices 1 and 2 list, respectively, our split sample and the control group, along
with their ticker symbols. The split sample also includes the split factors. Of the 12
funds in our sample, eight experience a two-for-one split while the remaining four
undergo a three-for-one split.
The chronology of the split event was as follows:
(1) June 6, 2005, the record date determining shareholders of record;
(2) June 8, the payable date, post-split shares were delivered to the Depository
Trust Company (DTC);
 (3) June 9, the ex-date, split-adjusted shares and prices were reflected in the market; Stock splits on
and iShare exchange-
(4) June 13, the due-bill redemption date, additional shares were delivered from the DTC traded funds
to shareholders of record and to those who purchased shares on June 6, 7, and 8.
We obtain from the NYSE’s Trade and Quote (TAQ) database intraday transaction data
for all trades and quotes between 9:30 AM and 4:00 PM for the trading days during the
period of our study. The TAQ data are filtered following procedures recommended by 139
Bessembinder (1999, 2000).
The period before the ex-date is the pre-period, and the period after the ex-date is
the post-period. We report results for ten-day pre- and post-periods[2]. Table I
summarizes information about the pre- and post-split prices of the splitting funds as
well as those for the 12 control funds. We report the high, low, mean, median, and
standard deviation of the prices over the sample period for each fund, as well as the
average for all of the funds. For the overall sample the average pre-split median price is
148.73 dollars while the corresponding post-split average median price is 64.44 dollars.
The funds with a split factor of 3.0 have an average pre-split median price of 180
dollars while those with a split factor of 2.0 report an average pre-split median price of
139 dollars (not shown in the table). For the control sample, median prices range from
10.32 dollars for the EWJ to 120.21 for IVV, with an average median price of 71.10
dollars. The average pre-split median price of the splitting funds is significantly higher
than the median price for all of the non-splitting funds. These observations point
toward price reduction as one possible motive for splitting.

Methodology and results

We study the impact of the iShares split on measures of liquidity, turnover, and order
imbalance around the split date. These measures allow us to examine the support, if
any, for the broker-promotion hypothesis and/or the trading-inconvenience hypothesis,
with regard to the sample and time period under study. Specifically, we analyze the
pre- and post-split turnover ratio, relative bid-ask spread, frequency of small trades,
average order size, and order imbalance, around the ex-date.

Relative spreads
Liquidity has been extensively discussed in the literature as one of the explanations for
a stock split that. Support for liquidity improvement has been presented by Baker and
Gallagher (1980), which provides survey-type evidence in favor of company managers’
intentions to improve liquidity through the result of the split and Muscarella and
Vetsuypens (1996) study, which investigates and provides evidence of improved
liquidity among splitting American Depository Receipts.
We analyze the liquidity effects of the iShare splits in terms of the relative spread,
which is defined as:

Sij ¼ ðAij  Bij Þ=Mij

where Aij and Bij are the quoted ask and bid prices and Mij is the midpoint of the bid and
ask prices. We designate the ex-date as Day 0 and report the results for ten days prior to
and following the ex-date. We use a similar convention for all subsequent analyses.
Table II presents the results of the spread analysis, reporting the figures for each
iShare fund in our sample, along with the average across all funds for each day. We
MF Pre-split Post-split
36,2 Lowest Median Highest Mean Std dev. Lowest Median Highest Mean Std dev.

Panel A: Descriptive statistics for split sample

EEM $209.77 $211.25 $211.92 $210.94 0.69 $70.02 $70.25 $70.66 $70.31 0.15
EFA 156.85 158.29 158.63 157.94 0.58 52.08 52.22 52.64 52.26 0.13
ICF 141.45 142.08 142.65 142.06 0.36 70.50 70.72 71.00 70.73 0.12
140 IGE 143.75 144.71 146.44 144.96 0.85 73.31 73.79 74.08 73.78 0.16
IJH 134.75 135.80 136.10 135.58 0.43 67.76 68.00 68.21 68.00 0.10
IJJ 130.43 131.46 131.76 131.30 0.39 65.73 65.88 66.08 65.89 0.07
IJK 136.94 138.00 138.55 137.82 0.46 68.79 69.05 69.23 69.05 0.09
IJR 161.15 162.32 163.05 162.12 0.66 54.00 54.15 54.42 54.16 0.09
IJS 120.32 121.36 121.66 121.20 0.39 60.52 60.69 60.96 60.71 0.09
IWM 123.25 124.38 124.70 124.07 0.51 62.09 62.25 62.55 62.27 0.09
IWN 187.38 187.89 189.18 188.23 0.61 62.90 63.00 63.19 63.02 0.08
IYR 126.20 127.28 127.66 127.27 0.28 63.19 63.31 63.60 63.32 0.09
Average 147.69 148.73 149.36 148.62 64.24 64.44 64.72 64.46
Panel B: Descriptive statistics for control group
EWJ $10.25 $10.32 $10.35 $10.30 0.03
IVV 119.60 120.21 120.41 120.05 0.29
IBB 64.28 64.66 64.95 64.60 0.17
IJT 106.63 107.69 108.10 107.46 0.46
IVE 61.81 62.13 62.23 62.06 0.13
IVW 57.37 57.70 57.88 57.66 0.13
IWD 66.59 66.76 67.07 66.81 0.17
IWF 48.25 48.42 48.75 48.44 0.12
IWO 62.84 63.33 63.62 63.23 0.29
IWS 115.58 116.20 116.48 116.13 0.27
IYE 73.58 74.76 75.13 74.52 0.47
IYH 60.90 61.04 61.62 61.12 0.18
Average $70.64 $71.10 $71.38 $71.03

Notes: This table summarizes information about the pre- and post-split prices of the splitting
funds as well as the two control funds. We report the high, low, mean, median, and standard
deviation of the prices over the sample period for each fund and also the average of these
statistics for all the funds. The stated goal of the split was to bring the average share price of the
ETFs in line with the average share price of a US listed security (stock), $42.21 at the time of the
Table I. split. While the average share of the iShare ETFs was still greater than that of the average US
Summary price statistics listed security, it was significantly closer than before split

report the t-statistics that test the significance of the difference between the mean of the
spread on a given day from the overall mean during the pre-event period. Results in
Panel A show a significant increase in the average relative spread, from 0.151 percent
on the day before the ex-date (Day 1) to 0.180 percent on the day of the split (Day 0),
with a t-statistic of 8.74. The increase in the relative spread persists throughout
the post-period, with the only exception occurring on Day 6, when the spread (0.139)
is actually lower than the pre-split spread. Panel B of Table II presents the spread
analysis for the control funds. The average spread for the control group varies
randomly between a minimum of 0.131 percent to a maximum of 0.176 percent. Unlike
the splitting funds, this group displays no consistent pattern over the 21 days.
The result of the relative spread analysis thus fails to reject the broker-promotion
hypothesis, implying that splitting a fund provides an incentive for promoting the fund
Panel A: Relative spread (%) for split sample
Day EEM IWN IJR EFA IGE IJK IJH IJJ IWM ICF IJS IYR Average t-stat

10 0.0838 0.1718 0.1361 0.0730 0.2172 0.1630 0.1910 0.1384 0.1097 0.1078 0.2743 0.0988 0.1471 0.65
9 0.0721 0.1074 0.1173 0.0729 0.2945 0.1031 0.1653 0.1156 0.0928 0.1141 0.2533 0.1266 0.1363 2.02
8 0.0753 0.1574 0.1137 0.0818 0.1555 0.1661 0.1066 0.1771 0.0905 0.1638 0.2277 0.1465 0.1385 1.47
7 0.0571 0.1886 0.1365 0.0723 0.2179 0.2825 0.2209 0.1594 0.1182 0.1068 0.2547 0.1254 0.1617 4.27
6 0.0663 0.1273 0.1440 0.0917 0.1882 0.2220 0.1300 0.1733 0.1082 0.1149 0.2276 0.1585 0.1460 0.39
5 0.0631 0.1182 0.1474 0.0965 0.1946 0.2089 0.1800 0.1871 0.1004 0.1128 0.2393 0.1550 0.1503 1.44
4 0.0870 0.0966 0.1737 0.1092 0.1778 0.1690 0.1806 0.1890 0.1229 0.1311 0.2327 0.1011 0.1476 0.77
3 0.0652 0.1128 0.1345 0.0515 0.1812 0.1966 0.2073 0.1763 0.0958 0.1106 0.1839 0.1059 0.1351 2.3
2 0.0620 0.1122 0.1471 0.0835 0.1824 0.1315 0.1573 0.1794 0.0913 0.0891 0.2202 0.1177 0.1311 3.28
1 0.0844 0.1285 0.1421 0.0825 0.2618 0.2217 0.2091 0.1765 0.0776 0.1062 0.2108 0.1075 0.1507 1.55
0 0.1365 0.1164 0.1631 0.1555 0.2387 0.2714 0.2541 0.2141 0.0917 0.1367 0.2195 0.1600 0.180 8.74
1 0.1380 0.1009 0.1905 0.1834 0.2889 0.2865 0.2233 0.2585 0.0553 0.1549 0.2504 0.1763 0.1923 11.81
2 0.0917 0.1068 0.3064 0.1553 0.2258 0.2461 0.2186 0.2649 0.0698 0.1334 0.2537 0.1098 0.1819 9.24
3 0.1278 0.1188 0.1595 0.1069 0.2500 0.2439 0.2254 0.2003 0.0815 0.1491 0.2650 0.1255 0.1711 6.6
4 0.0982 0.1013 0.1502 0.1115 0.2282 0.2911 0.2627 0.2055 0.0721 0.1202 0.2130 0.1289 0.1652 5.14
5 0.0852 0.0924 0.1416 0.1681 0.2210 0.2408 0.2088 0.1125 0.0720 0.1383 0.1926 0.1592 0.1527 2.04
6 0.0946 0.0947 0.1586 0.1037 0.2108 0.2039 0.1378 0.1631 0.0773 0.1284 0.1904 0.1078 0.1393 1.28
7 0.1264 0.0911 0.2106 0.1492 0.2051 0.1963 0.2511 0.1661 0.0757 0.1341 0.2458 0.1221 0.1644 4.94
8 0.0691 0.1284 0.1863 0.1338 0.3472 0.2441 0.1951 0.1650 0.0805 0.1254 0.2139 0.1194 0.1673 5.66
9 0.1043 0.0764 0.1988 0.1778 0.2780 0.2664 0.2931 0.2310 0.0774 0.1395 0.2217 0.1515 0.1847 9.94
10 0.0972 0.1025 0.2259 0.1182 0.2787 0.2741 0.2870 0.2226 0.0717 0.1280 0.2496 0.1409 0.1830 9.53
(continued)

Relative spread
Table II.
traded funds
iShare exchange-
Stock splits on

141
 MF
36,2

142

Table II.
Panel B: Relative spread (%) for control group
Day IVV EWJ IBB IWF IWD IVW IVE IWO IYE IYH IJT IWS Average t-stat

10 0.0714 0.1065 0.1307 0.2148 0.1678 0.1822 0.1971 0.3288 0.1835 0.1326 0.2504 0.1461 0.1760 6.86
9 0.0612 0.1049 0.1604 0.2020 0.1438 0.1970 0.1461 0.2429 0.1675 0.1692 0.2173 0.1094 0.1602 0.44
8 0.0696 0.1140 0.1020 0.1869 0.1686 0.1121 0.1104 0.2433 0.1946 0.1351 0.2397 0.0607 0.1448 5.81
7 0.0744 0.1155 0.1275 0.2159 0.1529 0.1517 0.1330 0.1903 0.2451 0.1460 0.2111 0.1470 0.1592 0.05
6 0.0802 0.1087 0.1043 0.1773 0.1461 0.1746 0.1766 0.2739 0.1674 0.1450 0.2212 0.1340 0.1591 0.01
5 0.0845 0.1160 0.1119 0.2106 0.1063 0.1970 0.1493 0.2069 0.2192 0.1090 0.1656 0.0945 0.1476 4.67
4 0.0791 0.1147 0.1759 0.1769 0.1102 0.1669 0.1945 0.3129 0.1481 0.1397 0.1690 0.0984 0.1572 0.76
3 0.0635 0.1269 0.1311 0.0803 0.1253 0.1475 0.1635 0.2262 0.2042 0.2335 0.2006 0.1239 0.1522 2.79
2 0.0964 0.1185 0.1453 0.2386 0.1696 0.1459 0.1248 0.3116 0.1685 0.1747 0.2243 0.1029 0.1684 3.79
1 0.0914 0.1806 0.1284 0.1928 0.1382 0.1635 0.1559 0.3008 0.1475 0.1794 0.2070 0.1087 0.1662 2.88
0 0.1041 0.1557 0.1427 0.1996 0.1587 0.1759 0.1510 0.1442 0.1783 0.1636 0.1999 0.1452 0.1599 0.34
1 0.0490 0.2072 0.1649 0.2034 0.1426 0.1418 0.1302 0.1157 0.1766 0.1826 0.1381 0.1409 0.1494 3.93
2 0.1029 0.1267 0.1618 0.2153 0.1939 0.2196 0.2013 0.1173 0.1892 0.1597 0.1806 0.0914 0.1633 1.72
3 0.1696 0.1187 0.1571 0.2425 0.1439 0.1911 0.1765 0.1001 0.1502 0.1670 0.1944 0.1231 0.1612 0.85
4 0.0837 0.1187 0.1477 0.2110 0.1231 0.1827 0.1537 0.1042 0.1758 0.1593 0.1663 0.1068 0.1444 5.95
5 0.0626 0.1685 0.1486 0.1692 0.1633 0.1779 0.1936 0.0869 0.1569 0.1106 0.1416 0.0949 0.1395 7.93
6 0.0761 0.1050 0.1212 0.1945 0.1549 0.1766 0.1278 0.0945 0.1653 0.1882 0.1803 0.0828 0.1389 8.18
7 0.0778 0.1494 0.1174 0.1576 0.1493 0.2184 0.1228 0.1079 0.1404 0.1024 0.2325 0.0873 0.1386 8.32
8 0.0566 0.1088 0.1248 0.1623 0.1315 0.1567 0.1918 0.0883 0.1408 0.1439 0.1724 0.0983 0.1313 11.26
9 0.0836 0.1223 0.1289 0.2090 0.1423 0.2285 0.1382 0.1130 0.1864 0.1439 0.2133 0.1073 0.1514 3.12
10 0.0867 0.1229 0.1377 0.2314 0.1711 0.2036 0.1340 0.0966 0.1636 0.1327 0.2519 0.1087 0.1534 2.30

Notes: This table presents the results of the spread analysis, frequently used as a measure of liquidity, with a lower spread being indicative of higher
liquidity. The relative spread for stock i at time j is computed as Sij ¼ (Aij – Bij)/Mij where Aij and Bij are the quoted ask and bid prices and Mij is the
midpoint of the bid and ask prices. While the lower post-split prices reflected in Table I result in lower absolute dollar spreads, the higher relative post-
split spreads below are indicative of decreased liquidity, consistent with the broker-promotion hypothesis. Panel A reports the results for the iShares that
split, and Panel B reports the results for the control group of non-splitting iShares. For panel A, the average of the post-split relative spreads is
significantly higher than the average of the pre-split relative spreads, consistent with the broker-promotion hypothesis. The results for Panel B (the non-
splitting iShare control group) show no consistent pattern over the reporting period
to small investors. This post-split increase in spread is consistent with the results Stock splits on
reported by Kadapakkam et al. (2005), Dennis (2003), Conroy et al. (1990), and Gray et al. iShare exchange-
(2003). An increase in spreads indicates that the split results in a decrease in aggregate
liquidity (i.e. higher spreads mean lower liquidity). We take a closer look at the liquidity traded funds
effects when we examine the frequency of small trades and order imbalance.

Turnover
Table III reports changes in volume turnover around the split date. If the trading-
143
inconvenience hypothesis prevails, a post-split increase in small buy orders would result
in an increase in turnover. Turnover is computed in the logarithmic form shown below,
which is used to minimize the impact of extreme observations (Kadapakkam et al., 2005):

Turnover ¼ 100  logð1 þ ðTrading Volume=Number of Shares OutstandingÞÞ:

The average percentage turnover increases from 0.24 percent on Day 1 to 0.58 percent
on Day 0 (t-statistic 6.85). The turnover for each day of the post-period remains higher
than the turnover in the days before the split, favoring the trading-inconvenience
hypothesis which predicts an increase in turnover due to pre-split lower trading volume.
A possible explanation provided in the literature and supported by our evidence is that
investors avoid the inconvenience of having to deal with due bills. However, since the
analysis looks at only daily aggregate turnover, we need to specifically look at the
frequency of small trades in order to determine if that contributes to the observed
increase in turnover. Our finding agrees with Dennis (2003), who detects a post-split
increase in share volume in the Nasdaq-100 index tracking stock.
The control group sample does not display any particular pattern. The average
turnover in the days before the split is about 0.31 percent. Beginning on the ex-date, June
6, the turnover appears to decline over the next four days, registering a minimum of 0.20
percent. However, the fifth day following the split exhibits a surge in the ratio to 0.58
percent. The sixth, seventh, and eighth days register an average of 0.43 percent, after
which the ratio drops to an average of 0.28 percent over the last two days. One possible
explanation for the low turnover immediately following the split is that investors
increase trading activity in the splitting funds in order to take advantage of the new price
per unit of these funds, and reduce the amount of trading in the non-splitting funds.

Order size and frequency of small trades

In this section we begin by investigating whether the split causes any changes in the
average order size. Under the broker-promotion hypothesis the higher spreads should
motivate brokers to promote the splitting fund to small investors, resulting in declining
buy-order size caused by an increase in the number of small trades.
This analysis requires us to classify trades as buy or sell orders. We employ the
widely used algorithm forwarded by Lee and Ready (1991) for this purpose, computing
the dollar-order size as the trade size in shares multiplied by the per-share price. A
post-split decline in average buy-order size indicates more buying by small investors.
Table IV summarizes the evidence for the average dollar-order size. Panel A presents
the results for the buy orders while Panel B presents the sell order results for the
splitting sample. Panels C and D present the control group results for buy and sell
orders, respectively.
Panel A shows that the buy-order size declines from a pre-period average of about
12,118,000 dollars (with a range from 8,349,000 to 17,049,000) to a post-period average
 MF
36,2

144

Table III.
Daily turnover
Panel A: Daily turnover (%) for split sample
Day EEM IWN IJR EFA IGE IJK IJH IJJ IWM ICF IJS IYR Average t-stat

10 0.202 0.081 0.087 0.204 0.260 0.062 0.040 0.164 0.438 0.134 0.088 0.555 0.193 1.51
9 0.212 0.052 0.093 0.106 0.221 0.345 0.057 0.108 0.977 0.261 0.125 0.437 0.250 0.29
8 0.146 0.053 0.294 0.102 0.085 0.068 0.058 0.096 0.678 0.128 0.042 0.344 0.175 1.90
7 0.188 0.081 0.246 0.139 0.486 0.072 0.070 0.189 0.651 0.160 0.066 0.875 0.269 0.11
6 0.510 0.544 0.215 0.090 0.725 0.118 0.600 0.133 1.459 0.111 0.132 0.703 0.445 3.88
5 0.193 0.039 0.076 0.047 0.195 0.079 0.138 0.115 0.363 0.124 0.087 0.233 0.141 2.63
4 0.285 0.413 0.077 0.045 0.088 0.124 0.090 0.287 0.699 0.186 0.040 1.404 0.312 1.03
3 0.223 0.322 0.072 0.114 0.417 0.157 0.036 0.290 0.788 0.115 0.086 0.487 0.259 0.09
2 0.398 0.278 0.136 0.107 0.134 0.126 0.526 0.202 0.749 0.226 0.102 1.246 0.352 1.90
1 0.345 0.047 0.086 0.252 0.140 0.076 0.071 0.104 0.618 0.156 0.117 0.868 0.240 0.50
0 0.784 0.164 0.147 0.270 1.522 0.143 0.096 0.197 1.290 0.273 0.855 1.264 0.584 6.86
1 0.326 0.119 0.176 0.509 0.684 0.198 0.063 0.185 1.069 0.296 0.093 0.671 0.366 2.19
2 0.592 0.147 0.108 0.142 0.359 0.157 0.070 0.221 0.995 0.246 0.144 0.971 0.346 1.76
3 0.833 0.205 0.290 0.269 0.606 0.157 0.126 0.179 0.961 0.406 0.190 0.913 0.428 3.52
4 1.189 0.245 0.247 0.260 0.445 0.176 0.105 0.800 2.036 0.293 0.165 1.029 0.582 6.82
5 0.948 0.364 0.366 0.225 0.416 0.202 0.311 1.600 1.623 0.290 0.205 0.626 0.598 7.16
6 1.719 0.204 0.227 0.490 0.522 0.177 0.233 0.730 1.654 0.203 0.265 2.846 0.773 10.89
7 0.701 0.760 0.148 0.209 0.639 0.230 0.115 0.422 2.032 0.157 0.222 0.955 0.549 6.11
8 0.461 0.331 0.194 0.236 0.364 0.188 0.240 0.337 1.103 0.245 0.137 1.864 0.475 4.53
9 0.455 0.596 0.179 0.186 0.889 0.188 0.138 0.251 1.034 0.244 0.144 1.028 0.444 3.87
10 0.791 0.405 0.256 0.254 0.294 0.158 0.120 0.237 1.928 0.361 0.231 1.396 0.536 5.83
(continued)
Panel B: Daily turnover (%) for control group
Day IVV EWJ IBB IWF IWD IVW IVE IWO IYE IYH IJT IWS Average t-stat

10 0.184 0.235 0.600 0.273 0.100 0.127 0.114 0.308 0.430 0.144 0.304 0.167 0.249 3.26
9 0.148 0.261 0.498 0.295 0.356 0.175 0.153 0.315 0.205 0.309 0.337 0.453 0.292 1.08
8 0.099 0.191 0.675 0.134 0.139 0.472 0.317 0.280 0.142 0.145 0.769 0.384 0.312 0.07
7 0.076 0.245 0.619 0.275 0.292 0.188 0.145 0.203 0.433 0.323 0.297 0.234 0.278 1.81
6 0.110 0.263 0.663 0.255 0.350 0.392 0.140 0.332 0.530 0.475 0.369 0.292 0.348 1.70
5 0.057 0.125 0.846 0.271 0.306 0.499 0.135 0.127 0.272 0.513 0.410 0.472 0.336 1.13
4 0.058 0.222 1.331 0.352 0.305 0.295 0.151 0.182 0.215 0.258 0.334 0.188 0.324 0.53
3 0.045 0.157 1.260 0.344 0.251 0.240 0.252 0.365 0.327 0.235 0.396 0.322 0.350 1.80
2 0.154 0.275 0.755 0.398 0.208 0.122 0.161 0.591 0.484 0.271 0.261 1.057 0.395 4.08
1 0.031 0.174 0.653 0.182 0.141 0.229 0.184 0.190 0.229 0.157 0.266 0.607 0.254 3.02
0 0.047 0.157 0.600 0.189 0.144 0.190 0.268 0.239 0.425 0.144 0.249 0.317 0.247 3.33
1 0.119 0.151 0.397 0.234 0.177 0.143 0.141 0.171 0.337 0.166 0.435 0.134 0.217 4.85
2 0.077 0.228 0.241 0.203 0.117 0.146 0.139 0.174 0.193 0.263 0.244 0.397 0.202 5.62
3 0.063 0.160 0.291 0.413 0.261 0.160 0.185 0.378 0.136 0.136 0.397 0.390 0.248 3.32
4 0.047 0.266 0.454 0.224 0.206 0.140 0.205 0.222 0.254 0.164 0.305 0.542 0.252 3.07
5 0.056 0.116 1.756 0.888 0.129 0.176 0.071 0.517 0.757 0.452 0.706 1.365 0.582 13.51
6 0.060 0.211 1.158 0.242 0.181 0.120 0.139 0.722 0.752 0.454 0.157 1.111 0.442 6.46
7 0.095 0.164 2.304 0.195 0.335 0.134 0.239 0.241 0.608 0.263 0.227 0.480 0.441 6.37
8 0.185 0.192 1.454 0.150 0.477 0.162 0.117 0.350 0.282 0.718 0.272 0.450 0.401 4.37
9 0.114 0.185 0.884 0.238 0.370 0.289 0.070 0.150 0.512 0.173 0.116 0.234 0.278 1.79
10 0.083 0.167 0.691 0.261 0.254 0.201 0.173 0.145 0.573 0.241 0.170 0.573 0.294 0.97

Notes: This table reports changes in volume turnover around the split date. Turnover is computed in the logarithmic form shown below in order to
minimize the impact of extreme observations

Turnover ¼ 100  logð1 þ ðTrading Volume=Number of Shares OutstandingÞÞ

If the trading-inconvenience hypothesis prevails, a post-split increase in small buy orders would result in an increase in turnover. The average percentage
turnover for iShares that split (Panel A) increases from 0.24 on Day 1 percent on to 0.58 percent on Day 0 (t-statistic 6.85). The turnover for each day
of the post-period remains higher than the turnover in the days before the split, thus favoring the trading-inconvenience hypothesis. The control group
funds (Panel B) do not display any particular pattern, but from June 6 (the ex-date) onwards exhibit an overall decline in the turnover ratio

Table III.
traded funds
iShare exchange-
Stock splits on

145
 MF
36,2

146

Table IV.
Daily order size
Panel A: Average dollar order size for buy orders (in ’000)
Day EEM IWN IJR EFA IGE IJK IJH IJJ IWM ICF IJS IYR Average t-stat

10 10,913 25,496 7,319 36,466 3,771 2,274 4,257 2,740 11,605 5,824 14,814 4,449 10,827 1.54
9 10,949 7,257 8,315 14,448 6,576 5,763 4,918 5,721 38,232 9,160 6,151 4,935 10,202 2.29
8 8,911 4,635 31,068 26,493 2,212 6,522 5,728 5,430 19,450 7,768 1,638 7,831 10,640 1.76
7 9,975 55,300 25,620 22,835 7,755 3,806 6,614 7,374 18,614 9,682 2,859 10,612 15,087 3.55
6 12,214 32,700 10,924 19,828 13,609 5,154 41,827 6,238 39,936 6,585 4,193 11,375 17,049 5.89
5 9,637 8,422 3,771 14,408 6,141 4,092 11,466 4,893 13,952 8,804 3,020 11,579 8,349 4.50
4 22,147 4,3917 5,054 15,089 2,776 3,152 3,162 13,474 26,896 5,916 2,673 12,424 13,057 1.12
3 10,512 29,095 4,667 17,272 22,292 12,635 4,910 6,234 25,049 7,599 6,150 6,436 12,738 0.74
2 20,738 15,350 7,755 8,349 3,741 3,764 46,614 3,892 17,313 12,492 4,264 11,320 12,966 1.01
1 17,190 5,000 7,515 48,481 5,059 2,562 7,877 2,729 8,424 6,114 5,136 7,039 10,260 2.22
0 8,737 4,261 2,057 6,797 16,365 1,582 3,192 1,872 8,657 3,777 5,854 5,576 5,727 7.63
1 3,606 4,418 4,894 9,758 4,371 4,201 2,105 3,017 6,243 2,252 1,425 4,504 4,233 9.42
2 8,797 4,564 3,339 4,902 4,276 1,157 3,264 2,227 10,071 5,978 2,047 6,790 4,784 8.76
3 14,594 7,139 5,053 8,427 3,564 2,421 2,618 1,804 14,475 7,965 2,124 4,096 6,190 7.08
4 15,507 2,592 3,747 9,114 3,117 4,501 2,379 12,622 13,928 2,757 2,399 2,896 6,297 6.95
5 10,772 5,295 5,408 5,444 2,101 3,786 2,115 5,938 13,479 5,523 2,477 5,261 5,633 7.74
6 17,635 6,110 2,504 23,369 3,895 2,068 1,839 4,147 16,585 4,489 6,636 5,598 7,906 5.03
7 9,274 12,085 2,370 7,103 7,698 4,683 1,762 3,155 14,662 2,786 2,603 4,199 6,032 7.27
8 5,258 3,763 4,953 8,244 2,122 1,148 2,213 3,866 13,284 3,887 2,178 3,301 4,518 9.07
9 6,963 6,185 7,864 5,222 3,590 1,734 1,653 3,158 11,450 5,389 1,747 3,031 4,832 8.70
10 6,496 7,964 5,738 7,399 3,622 2,590 4,058 2,157 11,018 4,846 2,826 4,708 5,285 8.16
(continued)
Panel B: Average dollar order size for sell orders (in ’000)
Day EEM IWN IJR EFA IGE IJK IJH IJJ IWM ICF IJS IYR Average t-stat

10 16,945 9,803 9,373 25,601 12,289 7,680 2,711 5,720 16,342 5,982 2,808 12,033 10,607 1.55
9 12,266 6,843 7,078 31,421 2,174 10,569 3,515 2,669 23,333 18,214 2,558 5,968 10,550 1.62
8 9,213 15,891 10,529 26,964 2,825 3,063 3,836 6,444 12,732 12,734 3,919 10,684 9,903 2.39
7 9,678 7,149 17,515 29,464 13,418 7,011 4,172 5,633 18,287 9,250 4,165 13,067 11,567 0.41
6 21,186 37,230 11,213 15,568 31,136 2,565 77,694 3,741 23,966 5,078 3,847 11,539 20,397 10.11
5 8,990 6,966 12,892 8,043 7,999 2,761 8,124 6,421 9,732 6,663 7,309 6,863 7,730 4.97
4 14,689 10,915 13,099 12,602 1,857 6,194 10,889 3,989 11,290 9,368 2,831 13,291 9,251 3.16
3 17,258 15,391 8,310 22,373 4,407 5,229 3,114 7,064 16,484 4,242 2,187 11,938 9,833 2.47
2 15,777 26,726 15,409 51,852 3,265 5,545 26,751 5,385 19,174 26,016 3,231 12,594 17,644 6.83
1 29,073 7,776 7,544 49,702 3,258 6,662 4,494 3,693 9,567 6,508 4,321 6,552 11,596 0.37
0 16,300 3,614 3,254 20,631 1,092 5,328 1,988 3,534 9,871 4,768 11,438 10,053 7,656 5.06
1 4,988 5,917 2,623 34,165 4,337 3,037 1,914 3,868 8,667 9,028 2,178 3,792 7,043 5.79
2 6,567 9,470 3,156 8,090 3,760 3,522 2,165 3,182 7,986 4,596 2,935 4,243 4,973 8.26
3 10,325 8,085 11,287 13,009 9,461 2,404 4,241 3,067 8,854 3,367 2,873 2,721 6,641 6.27
4 13,575 5,090 2,787 7,686 2,431 9,46 2,756 3,908 14,489 4,367 4,238 2,311 5,382 7.77
5 7,693 9,975 5,557 8,626 4,919 7,366 2,338 8,872 12,464 2,955 3,070 4,466 6,525 6.41
6 20,290 8,023 3,374 22,947 3,512 4,058 2,642 5,246 12,218 4,712 3,820 5,850 8,058 4.58
7 14,514 13,836 3,668 9,505 2,117 4,573 2,617 3,391 18,014 3,000 6,223 3,572 7,086 5.74
8 6,110 3,952 2,445 12,495 1,843 6,680 5,222 2,934 9,697 5,472 1,983 6,093 5,411 7.73
9 5,057 18,960 3,373 18,183 8,470 3,639 4,828 3,074 12,616 4,721 2,429 3,198 7,379 5.39
10 10,574 7,788 3,500 8,747 2,471 4,811 2,825 3,074 16,353 13,667 6,170 3,801 6,982 5.86
(continued)

Table IV.
traded funds
iShare exchange-
Stock splits on

147
 MF
36,2

148

Table IV.
Panel C: Average dollar order size for buy orders for control group (in ’000)
Day IVV EWJ IBB IWF IWD IVW IVE IWO IYE IYH IJT IWS Average t-stat

10 12,490 2,797 11,569 3,000 4,791 2,343 1,434 9,083 3,166 3,810 3,807 4,329 5,218 0.93
9 21,267 3,675 3,198 8,700 9,756 3,090 2,577 5,754 1,984 4,722 3,308 12,507 6,711 5.99
8 3,681 2,983 6,941 4,431 4,829 9,349 1,844 8,479 1,211 1,412 8,541 3,081 4,732 0.72
7 5,035 1,771 4,819 3,945 4,532 2,034 2,538 4,266 3,126 4,642 3,454 8,732 4,075 2.95
6 3,975 6,450 3,782 2,500 7,366 2,028 4,227 3,110 4,628 4,833 2,279 8,160 4,445 1.69
5 4,605 2,012 4,280 9,323 6,384 4,092 1,816 5,487 2,532 2,260 4,287 9,065 4,679 0.90
4 4,361 3,280 8,188 7,185 3,560 1,847 1,675 3,789 2,743 5,347 3,831 5,926 4,311 2.15
3 6,808 2,984 7,380 2,103 3,886 3,602 5,716 2,662 2,436 5,354 4,749 13,749 5,119 0.59
2 13,138 4,934 5,558 4,461 8,330 1,889 4,681 12,539 3,653 2,682 2,765 10,740 6,281 4.53
1 3,273 4,040 5,992 2,719 6,632 2,213 3,038 6,594 2,100 1,139 2,460 6,320 3,877 3.62
0 5,522 4,230 5,062 5,831 4,125 1,609 6,234 5,486 2,726 2,069 3,288 11,017 4,767 0.60
1 5,807 3,024 3,163 4,362 7,163 2,315 2,106 3,408 3,107 3,772 2,739 4,698 3,805 3.86
2 17,361 6,285 2,843 2,082 2,584 2,376 1,478 4,026 1,802 3,055 1,851 9,428 4,598 1.18
3 4,551 2,545 1,394 14,847 5,205 1,489 1,899 9,371 1,529 2,864 4,635 6,997 4,777 0.57
4 7,386 6,421 2,530 4,658 5,286 2,508 6,766 4,552 2,133 3,896 3,613 36,653 7,200 7.65
5 5,995 777 6,615 23,542 3,428 4,173 2,066 4,738 2,241 5,019 4,614 8,314 5,960 3.44
6 7,728 3,847 6,061 4,181 5,014 3,362 3,024 10,632 3,275 6,074 2,104 12,281 5,632 2.33
7 8,540 4,324 7,111 2,943 4,873 3,908 7,692 6,073 2,483 2,260 3,618 7,233 5,088 0.49
8 7,294 3,666 3,351 3,317 8,231 3,539 3,108 3,709 1,390 13,490 2,587 7,718 5,117 0.58
9 7,281 7,084 2,848 5,006 5,575 2,377 1,764 6,649 2,028 1,792 1,957 6,974 4,278 2.26
10 7,424 3,322 3,602 2,599 11,056 4,094 2,957 3,408 2,311 2,981 3,120 24,964 5,987 3.53
(continued)
Panel D: Average dollar order size for sell orders for control group (in ’000)
Day IVV EWJ IBB IWF IWD IVW IVE IWO IYE IYH IJT IWS Average t-stat

10 8,224 3,534 3,261 7,781 2,982 3,048 3,564 9,640 2,311 1,972 3,322 5,938 4,631 7.09
9 4,541 3,757 3,381 4,120 14,187 3,365 2,154 17,903 1,788 3,782 3,339 7,991 5,859 0.38
8 2,971 3,141 5,077 2,332 4,662 2,450 8,358 26,463 1,482 4,115 7,536 7,592 6,348 3.35
7 3,914 5,576 5,271 5,641 11,185 1,739 2,543 7,011 3,063 4,245 4,629 5,348 5,014 4.77
6 11,463 4,182 2,663 4,387 6,631 13,327 2,269 3,562 3,219 4,197 3,137 7,620 5,555 1.47
5 6,282 1,598 3,784 5,145 4,108 3,264 2,585 5,290 2,405 5,691 4,953 17,451 5,213 3.55
4 6,489 1,594 10,342 4,835 7,234 8,426 2,070 13,969 2,362 3,641 3,953 3,241 5,680 0.71
3 3,474 1,508 5,154 2,971 13,455 1,725 5,489 5,638 3,318 3,878 4,934 7,297 4,903 5.44
2 21,697 3,939 3,324 12,349 4,772 1,788 2,167 2,4013 4,750 3,222 3,500 23,322 9,070 19.92
1 3,931 2,201 3,094 3,118 4,596 6,639 3,147 4,568 1,812 1,956 3,251 30,053 5,697 0.61
0 5,463 3,783 2,550 2,911 4,288 2,972 3,113 3,231 5,496 1,959 1,853 9,754 3,948 11.25
1 6,043 4,780 3,542 4,574 4,161 4,344 4,767 7,671 1,625 3,546 8,435 4,419 4,825 5.91
2 4,466 4,878 2,285 8,133 5,801 4,538 3,309 3,337 2,333 8,000 3,202 15,498 5,482 1.92
3 10,953 3,134 5,659 6,429 8,709 4,930 4,505 2,926 1,674 1,588 6,090 21,288 6,490 4.22
4 4,622 5,034 3,575 2,776 5,258 5,993 2,469 3,455 1,365 1,815 4,784 11,612 4,397 8.52
5 6,680 1,882 5,818 8,471 3,253 5,237 2,060 4,552 2,371 3,528 9,272 14,173 5,608 1.15
6 9,468 3,404 4,379 4,727 4,721 2,456 2,036 11,033 2,303 3,515 2,302 12,593 5,245 3.36
7 11,149 4,212 7,584 4,945 7,617 2,940 3,739 4,362 2,831 2,365 2,551 6,323 5,052 4.54
8 8,131 3,656 5,138 4,503 7,587 3,327 3,688 9,325 2,385 4,489 1,851 22,319 6,367 3.47
9 10,916 2,360 4,670 7,430 11,021 6,624 1,192 3,498 2,544 2,350 2,840 7,717 5,264 3.25
10 4,947 4,065 3,718 4,674 4,945 4,224 3,035 5,035 4,389 4,882 3,699 14,241 5,154 3.91

Notes: This table summarizes changes in trading pattern around the split date. If the broker-promotion hypothesis holds then the higher spreads should
motivate brokers to promote the splitting fund to small investors, resulting in declining buy order size caused by an increase in the number of small
trades. Panel A presents the results for the buy orders while Panel B presents the sell orders. Panel C presents the buy and sell orders for the funds in
the control group. Panel A shows that the buy-order size declines from a pre-period average of about $12,118,000 to a post-period average value of about
$5,585,000, consistent with the broker-promotion explanation. Panel B also shows a decrease in the sell-order size, from a pre-slit average of $11,908,000
to a post-split average of $6,649,000. The t-statistics remain significant over the entire post-split period. There is no consistent pattern showing a
decrease in buy or sell-order size for the control group (Panel C); The dollar-order size is computed as the trade size in shares multiplied by the share
price.

Table IV.
traded funds
iShare exchange-
Stock splits on

149
MF value of about 5,585,000 dollars (with a range from 4,233,000 to 7,906,000). This result
is consistent with the broker-promotion explanation of increased broker incentive to
36,2 promote the splitting stock to small investors in order to take advantage of the wider
spreads. However, from Panel B we also note a decrease in the sell-order size that
declines from a pre-slit average of 11,908,000 dollars (with a range of 7,730,000-
20,397,000) to a post-split average of 6,649,000 dollars (with a range of 4,973,000-
8,058,000). The t-statistics remain significant over the entire post-split period. Since the
150 broker-promotion hypothesis makes no prediction regarding a decline in sell-order size,
there could be other factors influencing trading patterns around splits. A similar
observation is noted by Kadapakkam et al. (2005).
This evidence appears to be consistent with Ohlson and Penman’s (1985) discussion
of a post-split increase in the number of noise traders. They attribute this ‘‘clientele
factor’’ (advanced by Black, 1986) to the lower post-split price. Evidence of an increase
in return volatility, and hence uninformed trading following a stock split, has been
provided by Ohlson and Penman (1985), Conroy et al. (1990), and Dubofsky (1991).
The fact that we observe declines in order size on both the buy and sell side could
be an indication of increased noise trading, which securely rules out any information
effect of the split.
While ‘‘information’’ or ‘‘signaling’’ should not be an issue in the ETF market, recent
articles in the Wall Street Journal discuss a crucial shift in the behavior pattern of these
securities. Some current trends reveal new risks, unexpected results, and serious
performance deviations from the respective benchmarks. Given such discrepancies, the
results from our analysis do not necessarily indicate any contrarian behavior in our
sample.
The buy and sell order size for the control group does not follow the declining
example of the splitting sample. The buy order size varies randomly between a
minimum value of 3,805,000 dollars and a maximum of 7,200,000 dollars. The average
over the sample period is about 5,079,000 dollars. The sell side numbers are generally
higher than the buy side. The minimum sell order size is 3,948,000 dollars while the
maximum is 9,070,000 dollars, and the average over the sample period is about
5,514,000 dollars.
We next examine possible changes in trading patterns around the ex-date of the
split from the perspective of the frequency of small trades. We follow Kadapakkam et al.
(2005) in computing the frequency of small trades before and after the event on a split-
adjusted basis. For example, for a two-for-one split, we compare the frequency of one-
lot trades before the split with the cumulative frequency of one- and two-lot trades after
the split. An increase in the post-split frequency of small trades would corroborate the
evidence provided by the buy-order size in the previous section and lend further
support to the broker-promotion hypothesis.
Table V reports the results for the analysis of the frequency of small trades. We note
a post-split increase in the percentage of small trades. The ex post average frequency of
small trades increases to about 61.5 percent (ranging from 57.4 to 65.0 percent), from an
ex ante average of about 45.4 percent (ranging from 41.0 to 48.3 percent). This evidence
further substantiates our results on the post-event decline in buy-order size. Clearly,
both these findings lend support to the broker-promotion hypothesis.

Transaction price placement and order imbalance

To examine the effect of order flow around the ex-date we analyze two variables, the
trade-price location parameter and order imbalance. First, we examine the trade-price
Day EEM IWN IJR EFA IGE IJK IJH IJJ IWM ICF IJS IYR Average t-stat

10 32.806 59.574 55.046 22.910 52.459 45.714 64.706 45.370 50.281 40.260 42.857 37.179 45.764 0.22
9 45.541 74.510 33.333 37.358 48.485 30.702 63.235 50.549 47.917 43.210 49.231 43.979 47.338 1.06
8 53.208 52.000 41.259 28.796 51.351 48.148 39.344 42.857 42.418 40.909 56.522 34.043 44.238 0.60
7 41.503 31.034 36.154 27.737 38.596 50.000 42.857 41.284 38.043 49.254 54.167 41.799 41.036 2.32
6 47.030 85.044 25.238 30.483 41.071 53.247 46.377 41.739 40.351 39.437 39.560 48.630 44.851 0.27
5 49.718 85.393 42.424 35.204 34.286 52.500 43.478 42.169 47.575 38.710 52.542 34.328 46.527 0.63
4 30.233 84.025 38.158 35.366 58.824 41.509 56.364 44.371 45.372 50.000 40.909 16.732 45.155 0.11
3 35.503 78.226 50.427 39.818 36.170 38.000 58.000 42.949 58.299 60.759 45.455 36.232 48.320 1.59
2 30.667 72.414 42.105 41.667 47.059 45.333 37.662 41.772 49.783 40.000 37.037 28.521 42.835 1.36
1 31.200 79.787 41.667 28.309 53.061 52.083 50.000 56.140 50.000 40.816 45.570 42.038 47.556 1.18
0 58.876 76.316 75.419 57.026 75.325 76.119 65.882 52.206 69.515 58.772 28.395 52.679 62.211 9.04
1 71.394 71.186 74.233 54.377 69.388 74.667 67.073 51.042 77.760 53.571 64.286 50.980 64.996 10.54
2 50.114 75.556 63.415 65.333 56.250 79.167 69.841 61.146 63.092 54.545 62.745 57.143 63.196 9.57
3 57.895 81.197 67.669 57.447 61.111 64.634 53.846 65.217 53.537 43.860 63.704 56.150 60.522 8.14
4 52.802 89.547 71.250 64.348 73.684 70.175 61.458 54.217 45.281 59.690 51.765 75.042 64.105 10.06
5 52.545 71.429 61.345 56.000 67.368 66.667 60.870 48.872 52.875 52.174 61.157 77.477 60.732 8.25
6 38.131 86.636 67.249 55.457 66.337 62.338 68.263 56.452 45.308 53.846 59.091 61.911 60.085 7.90
7 48.034 71.186 58.108 57.477 61.111 58.462 61.250 61.307 41.463 64.130 51.852 54.006 57.365 6.44
8 51.168 56.534 67.526 55.689 82.069 81.690 59.420 60.116 40.995 47.727 54.717 42.357 58.334 6.96
9 63.861 93.294 64.063 60.160 52.336 64.557 65.556 54.610 43.689 60.952 60.000 50.195 61.106 8.45
10 49.367 93.400 66.000 52.147 73.239 60.656 52.577 68.817 43.646 52.475 71.429 75.883 63.303 9.63

Notes: This table summarizes the frequency of small trades before and after the split, following Kadapakkam et al. (2005) in computing the frequency of
small trades before and after the event on a split-adjusted basis. An increase in the post-split frequency of small trades would corroborate the evidence
provided by the buy-order size in the previous table and lend further support to the broker-promotion hypothesis

adjusted small trades

Frequency of split-

(in percent)
Table V.
traded funds
iShare exchange-
Stock splits on

151
MF location parameter (Lij) as defined by Keim (1989). Lij for firm i at time j is defined as
36,2 Lij ¼ ðPij  Bij Þ=ðAij  Bij Þ

where Pij is the transaction price, and Bij and Aij are the prevailing bid and ask prices,
respectively. A value of Lij between 0.5 and 1 indicates a trade that occurred at or near
the ask quote and a value between 0 and 0.5 indicates a trade that occurred at or near
152 the bid quote. Thus, on any given day a location parameter greater than 0.5 indicates
more trades nearer the ask price (i.e. more buy orders) which we would expect under
the broker-promotion hypothesis. Conversely, a trade location parameter less than 0.5
(i.e. more sell orders) implies a deficiency of buy orders, perhaps owing to the
inconvenience of dealing with due bills attached to splits, as suggested by the trading-
inconvenience hypothesis.
Table VI shows the placement of the trade price within the bid-ask spread. Panel A
presents the results for the splitting sample of iShares. The average value of the location
parameter on the ex-date is 0.58 (t-statistic of 6.64) which is significantly higher than the
previous day’s average of 0.47. This indicates a significantly greater number of buy
orders on the ex-date. However, on subsequent days we observe no consistent pattern for
the position of the location parameter. The pre- and post-event averages are close to 0.5
(about 0.52 and 0.51, respectively) indicating the absence of buying or selling pressure.
Thus the evidence does not appear to support either of the two hypotheses.
Panel B presents the results of the control group. Similar to the splitting sample,
there appears to be a surge in buy orders, 0.58 (t-statistic of 7.95), on the day of the split.
This is a significant increase from the previous day’s location parameter, 0.43. Apart
from this, there is no consistent pattern in the variation of buy and sell orders.
Next, we investigate whether the split affects order imbalance. Order imbalance
measures the difference between the number of buy orders and the number of sell
orders as a percentage of the total buy and sell orders. It is defined by

OIMij ¼ ðNBUYij  NSELLij Þ=ðNBUYij þ NSELLij Þ

where NBUYij and NSELLij represent the number of orders to brokers to buy or sell the
iShare i on a given day j. A positive value for OIMij indicates a greater number of buy
orders than sell orders and would be considered evidence favoring the broker-
promotion hypothesis.
Table VII presents the results for the analysis of order imbalance. Our findings
parallel those from the analysis using the location parameter. For the splitting sample
shown in Panel A we note a positive OIM on Day 0 of 0.27, which is significantly higher
(t-statictic ¼ 7.51) than the pre-split average on Day 1 of about 0.12. This
demonstrates a tendency for trades on Day 0 to execute near the ask price, which is
indicative of more buy orders. This observation corresponds to our findings for the
location parameter. However, this effect dissipates over Day þ1 which records an
approximate average of 0.12 indicating a reversion to more sell orders. Consequently,
we do not document any consistent pattern of either buying or selling tendencies. Thus
the evidence from the order-imbalance analysis is unable to provide support for either
hypothesis.
Panel B shows the results for the control group of funds. Similar to the results of the
splitting sample, we note a switch from sell orders on Day 1 (an average of 0.18
with a t-statistic of 9.2), to buy orders on Day 0 (an average of 0.18 with a t-statistic of
Panel A: Average location parameter for split sample
Day EEM IWN IJR EFA IGE IJK IJH IJJ IWM ICF IJS IYR Average t-stat

10 0.458 0.359 0.369 0.756 0.563 0.581 0.544 0.227 0.436 0.428 0.382 0.494 0.466 6.12
9 0.620 0.538 0.594 0.621 0.462 0.839 0.626 0.587 0.377 0.493 0.675 0.609 0.587 7.83
8 0.643 0.434 0.521 0.621 0.748 0.601 0.530 0.420 0.443 0.577 0.581 0.682 0.567 5.51
7 0.599 0.427 0.490 0.655 0.223 0.530 0.529 0.585 0.577 0.514 0.564 0.581 0.523 0.43
6 0.656 0.596 0.432 0.460 0.505 0.318 0.594 0.409 0.635 0.606 0.435 0.525 0.514 0.56
5 0.699 0.593 0.452 0.710 0.334 0.539 0.656 0.471 0.400 0.478 0.631 0.319 0.524 0.51
4 0.466 0.333 0.374 0.485 0.682 0.249 0.455 0.353 0.392 0.647 0.676 0.647 0.480 4.56
3 0.795 0.230 0.514 0.631 0.516 0.246 0.612 0.489 0.357 0.615 0.647 0.565 0.518 0.13
2 0.523 0.483 0.620 0.597 0.534 0.621 0.532 0.500 0.316 0.538 0.689 0.542 0.541 2.55
1 0.528 0.425 0.457 0.451 0.440 0.568 0.435 0.544 0.468 0.365 0.537 0.448 0.472 5.45
0 0.630 0.530 0.555 0.706 0.765 0.588 0.476 0.490 0.536 0.574 0.490 0.576 0.576 6.64
1 0.563 0.381 0.654 0.431 0.536 0.413 0.429 0.372 0.373 0.508 0.454 0.472 0.466 6.22
2 0.480 0.389 0.470 0.580 0.529 0.246 0.350 0.661 0.480 0.331 0.565 0.690 0.481 4.43
3 0.637 0.464 0.489 0.656 0.462 0.344 0.298 0.547 0.430 0.620 0.801 0.753 0.542 2.63
4 0.679 0.165 0.624 0.681 0.474 0.655 0.486 0.715 0.453 0.491 0.662 0.381 0.539 2.29
5 0.642 0.552 0.566 0.468 0.623 0.703 0.456 0.586 0.346 0.442 0.784 0.591 0.563 5.13
6 0.830 0.568 0.393 0.726 0.416 0.368 0.538 0.319 0.415 0.498 0.354 0.365 0.483 4.24
7 0.573 0.502 0.360 0.513 0.640 0.448 0.353 0.397 0.505 0.480 0.513 0.360 0.470 5.66
8 0.718 0.276 0.477 0.655 0.342 0.509 0.498 0.568 0.433 0.330 0.216 0.539 0.463 6.48
9 0.740 0.769 0.448 0.596 0.393 0.402 0.351 0.692 0.476 0.485 0.315 0.566 0.519 0.03
10 0.554 0.553 0.412 0.608 0.436 0.488 0.379 0.517 0.443 0.500 0.733 0.302 0.494 2.94
(continued)

price within the bid-ask

Placement of transaction

spread
Table VI.
traded funds
iShare exchange-
Stock splits on

153
 MF
36,2

154

Table VI.
Panel B: Average location parameter for control group
Day IVV EWJ IBB IWF IWD IVW IVE IWO IYE IYH IJT IWS Average t-stat

10 0.380 0.425 0.361 0.559 0.411 0.652 0.484 0.469 0.468 0.592 0.739 0.619 0.513 0.35
9 0.332 0.491 0.606 0.229 0.507 0.699 0.504 0.769 0.369 0.668 0.762 0.852 0.566 6.49
8 0.546 0.506 0.579 0.501 0.266 0.559 0.371 0.715 0.471 0.610 0.737 0.675 0.545 4.03
7 0.620 0.494 0.531 0.357 0.414 0.721 0.402 0.686 0.514 0.474 0.689 0.415 0.526 1.88
6 0.407 0.421 0.552 0.399 0.512 0.755 0.360 0.399 0.482 0.567 0.706 0.627 0.515 0.59
5 0.344 0.451 0.309 0.357 0.691 0.743 0.266 0.666 0.548 0.701 0.736 0.563 0.531 2.44
4 0.388 0.306 0.388 0.318 0.547 0.763 0.313 0.500 0.484 0.551 0.597 0.664 0.485 2.99
3 0.453 0.342 0.322 0.654 0.490 0.769 0.319 0.733 0.454 0.431 0.715 0.640 0.527 1.92
2 0.286 0.469 0.493 0.443 0.328 0.600 0.290 0.600 0.294 0.422 0.759 0.555 0.462 5.72
1 0.223 0.355 0.484 0.331 0.376 0.712 0.240 0.529 0.518 0.307 0.598 0.531 0.434 8.98
0 0.427 0.374 0.595 0.511 0.567 0.758 0.426 0.427 0.668 0.665 0.804 0.717 0.578 7.95
1 0.718 0.303 0.371 0.337 0.384 0.669 0.400 0.587 0.618 0.351 0.694 0.563 0.500 1.26
2 0.337 0.463 0.516 0.393 0.379 0.266 0.227 0.452 0.525 0.572 0.644 0.602 0.448 7.31
3 0.445 0.562 0.646 0.549 0.472 0.236 0.270 0.452 0.402 0.679 0.725 0.702 0.512 0.15
4 0.479 0.465 0.556 0.474 0.494 0.432 0.270 0.448 0.622 0.440 0.715 0.518 0.493 2.10
5 0.444 0.755 0.616 0.512 0.464 0.550 0.369 0.326 0.726 0.430 0.693 0.624 0.542 3.76
6 0.604 0.413 0.645 0.498 0.390 0.520 0.448 0.338 0.603 0.672 0.244 0.708 0.507 0.41
7 0.309 0.332 0.448 0.569 0.414 0.373 0.572 0.456 0.343 0.622 0.374 0.601 0.451 6.95
8 0.726 0.412 0.668 0.529 0.451 0.332 0.354 0.401 0.387 0.597 0.532 0.612 0.500 1.20
9 0.171 0.413 0.743 0.559 0.471 0.357 0.710 0.358 0.502 0.368 0.647 0.594 0.491 2.27
10 0.216 0.520 0.727 0.234 0.380 0.408 0.251 0.606 0.547 0.525 0.414 0.588 0.451 6.93

Notes: This table shows the placement of the trade price within the bid-ask spread using the following measure of placement Lij: Lij for firm i at time j
is defined as Lij ¼ (Pij – Bij)/(Aij – Bij) where Pij is the transaction price, and Bij and Aij are the prevailing bid and ask prices, respectively. A value of Lij
between 0.5 and 1 indicates a trade that occurred at or near the ask quote which would be supportive of the broker-promotion hypothesis. Conversely, a
trade location parameter less than 0.5 (i.e. more sell orders) supports the trading-inconvenience hypothesis. Panel A presents the results for the splitting
sample of iShares. The average value of the location parameter on the ex-date is 0.58 (t-statistic of 6.64) which is significantly higher than the previous
day’s average of 0.47, indicating a significantly greater number of buy orders on the ex-date. On subsequent days we observe no consistent pattern for
the position of the location parameter. The pre- and post-event averages of 0.52 and 0.51, respectively, indicating the absence of support for either
hypothesis
Panel A: Average order imbalance for split sample
Day EEM IWN IJR EFA IGE IJK IJH IJJ IWM ICF IJS IYR Average t-stat

10 0.080 0.231 0.087 0.595 0.133 0.188 0.143 0.567 0.162 0.360 0.667 0.026 0.093 6.04
9 0.286 0.030 0.242 0.269 0.107 0.827 0.397 0.333 0.193 0.089 0.619 0.118 0.228 5.87
8 0.239 0.250 0.065 0.272 0.929 0.385 0.375 0.236 0.088 0.436 0.289 0.413 0.236 6.16
7 0.074 0.481 0.072 0.271 0.370 0.188 0.053 0.269 0.149 0.169 0.304 0.191 0.062 0.28
6 0.330 0.215 0.133 0.094 0.137 0.088 0.390 0.212 0.299 0.391 0.047 0.269 0.122 1.93
5 0.426 0.282 0.043 0.421 0.515 0.568 0.545 0.073 0.213 0.123 0.280 0.375 0.138 2.54
4 0.049 0.224 0.360 0.130 0.548 0.462 0.434 0.270 0.221 0.248 0.317 0.287 0.062 4.89
3 0.620 0.497 0.310 0.304 0.048 0.292 0.395 0.074 0.281 0.342 0.389 0.254 0.131 2.28
2 0.022 0.061 0.292 0.152 0.143 0.315 0.013 0.200 0.343 0.156 0.333 0.121 0.055 0.54
1 0.059 0.033 0.333 0.146 0.167 0.217 0.344 0.143 0.084 0.362 0.120 0.159 0.121 7.04
0 0.309 0.138 0.251 0.501 0.684 0.193 0.525 0.076 0.048 0.303 0.139 0.248 0.272 7.51
1 0.087 0.273 0.156 0.114 0.075 0.268 0.210 0.250 0.279 0.050 0.147 0.108 0.115 6.83
2 0.121 0.036 0.119 0.305 0.194 0.463 0.258 0.461 0.064 0.233 0.060 0.413 0.005 2.38
3 0.396 0.269 0.039 0.392 0.270 0.315 0.341 0.368 0.215 0.413 0.712 0.632 0.174 3.86
4 0.404 0.735 0.420 0.438 0.011 0.532 0.136 0.379 0.189 0.089 0.375 0.260 0.111 1.53
5 0.295 0.069 0.221 0.101 0.444 0.630 0.156 0.213 0.383 0.046 0.508 0.227 0.160 3.36
6 0.667 0.106 0.300 0.514 0.179 0.178 0.013 0.446 0.232 0.147 0.326 0.304 0.045 4.25
7 0.144 0.018 0.343 0.017 0.045 0.127 0.304 0.122 0.002 0.011 0.158 0.276 0.065 4.97
8 0.483 0.634 0.043 0.430 0.479 0.125 0.129 0.059 0.138 0.205 0.402 0.055 0.073 5.28
9 0.504 0.467 0.169 0.244 0.019 0.343 0.341 0.243 0.015 0.051 0.290 0.064 0.024 1.67
10 0.067 0.163 0.031 0.231 0.029 0.192 0.417 0.103 0.115 0.082 0.273 0.527 0.050 4.42
(continued)

Order imbalance
Table VII.
traded funds
iShare exchange-
Stock splits on

155
 MF
36,2

156

Table VII.
Panel B: Average order imbalance for control group
Day IVV EWJ IBB IWF IWD IVW IVE IWO IYE IYH IJT IWS Average t-stat

10 0.254 0.160 0.582 0.162 0.397 0.391 0.089 0.207 0.154 0.171 0.513 0.280 0.027 3.07
9 0.257 0.023 0.308 0.635 0.040 0.542 0.202 0.576 0.121 0.415 0.720 0.806 0.215 6.68
8 0.304 0.024 0.262 0.333 0.466 0.282 0.219 0.515 0.000 0.101 0.494 0.435 0.172 4.97
7 0.342 0.013 0.100 0.211 0.288 0.553 0.016 0.344 0.287 0.213 0.452 0.151 0.082 1.34
6 0.138 0.142 0.193 0.159 0.210 0.578 0.250 0.271 0.043 0.200 0.449 0.433 0.096 1.88
5 0.135 0.096 0.418 0.261 0.471 0.665 0.635 0.158 0.441 0.574 0.349 0.224 0.111 2.52
4 0.196 0.409 0.323 0.358 0.123 0.469 0.435 0.100 0.137 0.121 0.257 0.365 0.029 3.15
3 0.375 0.320 0.511 0.367 0.093 0.703 0.419 0.409 0.041 0.138 0.403 0.345 0.105 2.28
2 0.365 0.072 0.028 0.078 0.405 0.247 0.478 0.381 0.521 0.026 0.561 0.114 0.056 4.23
1 0.500 0.305 0.057 0.440 0.523 0.468 0.633 0.104 0.049 0.573 0.156 0.015 0.180 9.22
0 0.342 0.282 0.374 0.061 0.224 0.605 0.156 0.059 0.518 0.200 0.595 0.400 0.178 5.20
1 0.339 0.427 0.111 0.419 0.107 0.055 0.034 0.245 0.118 0.167 0.423 0.302 0.018 1.25
2 0.057 0.092 0.147 0.426 0.361 0.307 0.591 0.095 0.048 0.182 0.314 0.183 0.078 5.14
3 0.247 0.132 0.443 0.068 0.045 0.736 0.588 0.024 0.107 0.408 0.351 0.564 0.018 1.24
4 0.000 0.075 0.229 0.010 0.058 0.075 0.385 0.184 0.124 0.057 0.479 0.164 0.022 1.08
5 0.030 0.589 0.255 0.178 0.089 0.278 0.100 0.270 0.593 0.203 0.402 0.380 0.185 5.49
6 0.050 0.197 0.335 0.097 0.264 0.380 0.069 0.321 0.119 0.457 0.323 0.474 0.045 0.14
7 0.211 0.336 0.076 0.214 0.161 0.406 0.059 0.065 0.053 0.365 0.194 0.294 0.039 3.54
8 0.480 0.187 0.448 0.185 0.146 0.402 0.222 0.219 0.396 0.072 0.093 0.394 0.008 1.64
9 0.721 0.190 0.509 0.224 0.289 0.293 0.532 0.313 0.090 0.325 0.118 0.000 0.070 4.79
10 0.646 0.010 0.463 0.673 0.348 0.143 0.573 0.123 0.009 0.288 0.395 0.077 0.200 10.05
Notes: This table presents the results for the analysis of order imbalance, which is defined by:

OIMij ¼ ðNBUYij  NSELLij Þ=ðNBUYij þ NSELLij Þ;

where NBUYij and NSELLij represent the number of orders to brokers to buy or sell the given iShare i on a given day j. A positive value for OIMij
indicates a greater number of buy orders than sell orders and would be considered evidence favoring the broker-promotion hypothesis. Our findings are
similar to those for the location parameter. For Panel A we note a positive OIM on Day 0 of 0.27 is significantly higher (t-statictic ¼ 7.51) than the pre-
split average on Day 1 of about 0.07. This demonstrates a tendency for trades to execute near the ask price on Day 0 which is indicative of more buy
orders, which corresponds to our findings for the location parameter. This effect dissipates over Day þ1, after which we do not document any consistent
pattern of either buying or selling tendencies, thus providing no evidence to support either hypothesis
5.2). Apart from this, the prevalence of buy or sell orders on a given trading day does Stock splits on
not exhibit any consistent trend. iShare exchange-
A possible explanation for the observation in the splitting fund sample may lie in
the fact that a split is an anticipated event. Brooks et al. (2003) provide an explanation traded funds
for this phenomenon in their investigation of anticipated vs unanticipated events.

Conclusions 157
Studying the effects of share splits for 12 iShare ETFs issued by Barclay Global
Investors, we examine the two hypotheses that have been advanced for the pattern of
trade-and-quote price and activity behavior surrounding stock share splits. The broker-
promotion hypothesis suggests that brokers will influence their small investors to delay
purchasing shares until after a split in order to profit from the increased bid-ask spreads
arising from the split. The trading-inconvenience hypothesis entails reluctance of
investors to deal with due bills, resulting in postponing purchases until after the ex-date.
We find that there are increases in spreads and frequency of small trades and the
decrease in average trade size experienced by the split iShares that are consistent with
the broker-promotion hypothesis, while there is a significant increase in daily turnover
after the split that is consistent with the trading-inconvenience hypothesis. Neither
hypothesis is supported by the trade-price location parameter or by trade imbalance in
the post-split period. This suggests that stock splits, being anticipated events, allow
market makers and small investors to be able to predict order flows and adjust their
positions accordingly. Overall, we conclude that there is more support for the broker-
promotion hypothesis than for the trading-inconvenience hypothesis in the iShare split.

Notes
1. As of 2005 global ETF assets grew 35 percent to about US $417 billion. The year 2005
saw the introduction of 119 funds, 53 more than the previous year (Source: Financial
Post, February 20, 2006).
2. We repeat the analysis for 60-day pre- and post-periods. Since results remain unchanged
we report only results of the ten-day periods for conciseness.

References
Angel, J.J. (1997), ‘‘Tick size, share price and stock splits’’, Journal of Finance, Vol. 52, pp. 655-81.
Baker, H.K. and Gallagher, P.L. (1980), ‘‘Management’s view of stock splits’’, Financial
Management, Vol. 9, pp. 73-7.
Bessembinder, H. (1999), ‘‘Trade execution costs on the Nasdaq and NYSE: a post reform
comparison’’, Journal of Financial and Quantitative Analysis, Vol. 39, pp. 387-407.
Bessembinder, H. (2000), ‘‘Tick size, spreads, and liquidity: an analysis of Nasdaq securities
trading near ten dollars’’, Journal of Financial Economics, Vol. 9, pp. 213-39.
Black, F. (1986), ‘‘Noise’’, Journal of Finance, Vol. 41, pp. 529-43.
Brooks, R.M., Patel, A. and Su, T. (2003), ‘‘How the equity market responds to unanticipated
events’’, The Journal of Business, Vol. 76, pp. 109-33.
Conrad, J.S. and Conroy, R.M. (1994), ‘‘Market microstructure and the ex-date return’’, Journal of
Finance, Vol. 49, pp. 1507-19.
Conroy, R., Harris, R. and Benet, R. (1990), ‘‘The effect of stock splits on bid-ask spreads’’, Journal
of Finance, Vol. 45, pp. 1285-95.
MF Dennis, P. (2003), ‘‘Stock splits and liquidity: the case of the Nasdaq-100 index tracking stock’’,
The Financial Review, Vol. 38, pp. 415-33.
36,2
Dubofsky, D.A. (1991), ‘‘Volatility increases subsequent to NYSE and AMEX stock splits’’,
Journal of Finance, Vol. 46, pp. 421-31.
Easley, D., O’Hara, M. and Saar, G. (2001), ‘‘How stock splits affect trading: a microstructure
approach’’, Journal of Financial and Quantitative Analysis, Vol. 36, 2001.
158 Financial Post (2006), February 20.
Gray, S., Smith, T. and Whaley, R.E. (2003), ‘‘Stock splits: implications for investor trading costs’’,
Journal of Empirical Finance, Vol. 10, May, pp. 271-303.
Grinblat, M.S., Masulis, R. and Titman, S. (1984), ‘‘The valuation effects of stock splits and stock
dividends’’, Journal of Financial Economics, Vol. 13, pp. 461-90.
Kadapakkam, P.-R., Krishnamurthy, S. and Tse, Y. (2005), ‘‘Stock splits, broker promotion, and
decimalization’’, Journal of Financial and Quantitative Analysis, Vol. 40, pp. 873-95.
Keim, D.B. (1989), ‘‘Trading patterns, bid-ask spreads and estimated stock returns: the case of
common stocks as calendar turning points’’, Journal of Financial Economics, Vol. 25, pp. 75-97.
Lee, C.M.C. and Ready, M.J. (1991), ‘‘Inferring trade direction from intraday data’’, Journal of
Finance, Vol. 46, pp. 733-46.
Maloney, M.T. and Mulherin, J.H. (1992), ‘‘The effect of splitting on the ex: a microstructure
reconciliation’’, Financial Management, Vol. 21, pp. 44-59.
Muscarella, C.J. and Vetsuypens, M.R. (1996), ‘‘Stock splits: signaling or liquidity? The case of
ADR ‘solo splits’’’, Journal of Financial Economics, Vol. 42, pp. 3-26.
Nayar, N. and Rozeff, M.S. (2001), ‘‘Record date, when issued and ex date effects in stock splits’’,
Journal of Financial and Quantitative Analysis, Vol. 36, pp. 119-39.
Ohlson, J.A. and Penman, S.H. (1985), ‘‘Volatility increases subsequent to stock splits: an
aberration’’, Journal of Financial Economics, Vol. 14, pp. 251-66.
Rozeff, Michael (1998), ‘‘Stock splits: evidence from mutual funds’’, Journal of Finance, Vol. 53,
pp. 335-49.
Schultz, P. (2000), ‘‘Stock splits, tick size and sponsorship’’, Journal of Finance, Vol. 55, pp. 429-50.

Appendix 1. The iShares funds that split in June 2005

Table AI is a list of the iShare funds that split in June 2005, and the number of shares that were
exchanged after the split for each share that existed before the split. All of the splitting funds
were traded on the American Stock Exchange.

Fund name Ticker Split factor

iShares MSCI Emerging Markets Index Fund EEM 3.0

iShares Russell 2000 Value Index Fund IWN 3.0
iShares S&P SmallCap 600 Index Fund IJR 3.0
iShares MSCI EAFE Index Fund IFA 3.0
iShares Goldman Sachs Natural Resources Index Fund IGE 2.0
iShares S&P MidCap 400/BARRA Growth Index Fund IJK 2.0
iShares S&P MidCap 400 Index Fund IJH 2.0
iShares S&P 400/BARRA Value Index Fund IJJ 2.0
iShares Russell 2000 Index Fund IWM 2.0
iShares Cohen & Steers Realty Majors Index Fund ICF 2.0
iShares S&P SmallCap 600/BARRA Value Index Fund IJS 2.0
Table AI. iShares Dow Jones US Real Estate Fund IYR 2.0
Appendix 2. iShares funds that did not split in June 2005 Stock splits on
Table AII is a list of iShare funds that did split in June 2005, used as a control group to insure that
the effects attributed to the splitting of the funds were not, in fact, the result of some other
iShare exchange-
economic event that affected all iShare funds. traded funds

Fund name Ticker

159
iShare MSCI Japan Index EWJ
iShare S&P 500 Index IVV
iShare Nasdaq Biotechnology IBB
iShare SPSC600 Growth Index IJT
iShare SP 500 Value Index IVE
iShare SP 500 Growth Index IVW
iShare Russell 1000 Value Index IWD
iShare Russell 1000 Growth Index IWF
iShare Russell 2000 Growth Index IWO
iShare RusseIl MidCap Value Index IWS
iShare Dow Jones US Energy IYE
iShares Dow Jones US Health Care IYH Table AII.

Corresponding author
Pia Bandyopadhyay can be contacted at: pbandyop@csulb.edu

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com

Or visit our web site for further details: www.emeraldinsight.com/reprints