The adverse selection component of exchange traded funds

International Review of Financial Analysis 19 (2010) 65–76

Contents lists available at ScienceDirect

International Review of Financial Analysis

The adverse selection component of exchange traded funds Patricia Chelley-Steeley a,⁎, Keebong Park b a b

Aston University Business School, Aston University, Birmingham, Great Britain, B4 7ET, United Kingdom Department of Business Administration, Keimyung University, Sindang-dong, Daegu City, 704-701, South Korea

a r t i c l e

i n f o

Article history: Received 23 April 2008 Received in revised form 9 September 2009 Accepted 9 September 2009 Available online 13 October 2009 JEL Classification: G15

a b s t r a c t The aim of our paper is to examine whether Exchange Traded Funds (ETFs) diversify away the private information of informed traders. We apply the spread decomposition models of Glosten and Harris (1998) and Madhavan, Richardson and Roomans (1997) to a sample of ETFs and their control securities. Our results indicate that ETFs have significantly lower adverse selection costs than their control securities. This suggests that private information is diversified away for these securities. Our results therefore offer one explanation for the rapid growth in the ETF market. © 2009 Published by Elsevier Inc.

Keywords: Spread Adverse selection costs Exchange traded funds

1. Introduction Adverse selection losses arise when uninformed traders, such as market makers or liquidity traders, trade with the informed. The market maker recoups these losses by including in the spread a component to compensate for adverse selection losses (see for example, Copeland & Gali, 1983). In this paper we use the magnitude of the adverse selection component of the spread to measure adverse selection losses and information asymmetry.1 We show that these costs are lower for exchange traded funds than for individual securities suggesting that there is less information asymmetry between traders in ETF markets than in markets for individual securities. Since their introduction in 1993, the market for Exchange Traded Funds (ETFs) has grown rapidly. Ackert and Tian (2001) note that the most actively traded ETFs (Diamonds and the NASDAQ-100 Index Tracking Stock) have become especially popular and rank consistently as the most active issues on their exchanges. Additional evidence of their importance is given by Boehmer and Boehmer (2003) who show that by 2001, the three most heavily traded ETFs generated average daily trading volume of about $5 billion. A number of reasons have been suggested for the popularity of ETFs. Hegde and McDermott (2004a,b) attribute the growth of ETFs to the ease with which investors can obtain portfolio diversification benefits, at

⁎ Corresponding author. E-mail addresses: [email protected] (P. Chelley-Steeley), [email protected] (K. Park). 1 The more private information a trader has the greater are the profits he can extract from the uninformed. Higher adverse selection losses are therefore associated with higher information asymmetries between the informed and uninformed. 1057-5219/$ – see front matter © 2009 Published by Elsevier Inc. doi:10.1016/j.irfa.2009.09.003

low transaction costs, in comparison to trading a portfolio of underlying stocks. Although each ETF reflects the collective performance of a portfolio, there is no uncertainty about their redemption value. This contrasts with the findings of Neal and Wheatley (1998) who found that closed-end funds had similar levels of information asymmetry to individual securities. Moreover, since short selling is permitted for ETFs, arbitrage between the ETF and component securities ensures that ETFs have negligible premiums/discounts eliminating this potential source of information asymmetry. For example, Pennathur, Delcoure and Anderson (2002) and Engle and Sarkar (2006) find that ETFs closely mirror their underlying index and have low tracking errors while Harper, Madura and Schnusenberg (2006) show that tracking errors associated with ETFs are lower than for matched closed-end funds. Moreover, since short selling is permitted for ETFs arbitrage between the ETF and component securities can easily be undertaken. These advantages are reinforced by the comparatively low trading costs associated with ETFs that were identified in studies by Elton, Gruber, Comer and Li (2002) and Boehmer and Boehmer (2003) that examined both quoted and effective spreads. In this paper we draw attention to an alternative explanation for low bid-ask spreads and high levels of trading activity associated with ETFs. An ETF is an example of a basket security, the characteristics of which were outlined by Subrahmanyam (1991). One important characteristic of a basket security is that adverse selection costs are diversified away, reducing the adverse selection component of the spread relative to component securities. The effect of lower trading costs associated with the basket encourages the concentration of trading in the basket security, accentuating liquidity, and driving trading costs down further. We test this hypotheses for the first time using the spread decomposition models of Glosten and Harris (1988) and Madhavan,

66

P. Chelley-Steeley, K. Park / International Review of Financial Analysis 19 (2010) 65–76

Richardson and Roomans (MRR) (1997), two of the most widely used spread decomposition models. Our results indicate that ETFs have substantially lower adverse selection costs than control securities showing that ETFs diversify private information. Our results offer an explanation for why these markets have grown so rapidly since their inception. The diversification of private information in the basket leads to concentration of trades by liquidity traders in the market for the basket security so that losses with informed traders are reduced. The remainder of this paper is set out as follows. In Section 2 we explain why ETFs diversify adverse selection costs and review the Glosten–Harris and the Madhavan–Richardson–Roomans (MRR) spread decomposition model. Section 3 discusses our data. Section 4 provides the results and Section 5 offers a summary and conclusion to the paper. 2. Measuring the adverse selection component of the spread Subrahmanyam (1991) showed that a basket security generates lower adverse selection losses than component securities. Using the framework of Kyle (1984, 1985) and Admati and Pfleiderer (1988), Subrahmanyam (1991) models the strategic interaction of discretionary and non-discretionary liquidity traders in a market with informed traders. Both groups of liquidity traders can choose to execute their portfolio trades either in the market for the basket or trade in the underlying security markets. Subrahmanyam shows that a basket security with identical payoffs to component securities has lower adverse selection losses than individual securities. This outcome arises because the impact of security-specific private information is diversified away in a basket security.2 Although Subrahmanyam (1991) considered stock index futures as an example of a basket security, the diversification of adverse selection costs will arise for any basket security that has high levels of transparency if liquidation and the creation of the basket can easily be achieved. We argue that an ETF has the characteristics of a basket security outlined by Subrahmanyam (1991). 2.1. The Glosten–Harris model The first attempt at decomposing the spread was made by Glosten and Harris (1988). This model has since been used to study a range of issues influenced by adverse selection costs. Using the Glosten–Harris model Chiyachantana, Chiraphol, Christine, Taechapiroontong and Wood (2004) showed that adverse selection costs fell after Regulation Fair Disclosure was introduced. Examinations of the relationship between block ownership and adverse selection costs undertaken by Sarin, Atulya, Shastri and Shastri (2000) and Jiang and Kim (2005) show that adverse selection costs increase with institutional ownership. Jiang, Kim and Wood (2002) show that ADR's experience lower adverse selection costs following a stock split. The Glosten–Harris model has also been used by McInish and Van Ness (2002) and Ahn, Hamao and Ho (2002) to measure the evolution of adverse selection costs during the trading day. The Glosten–Harris model is based on the following representation of intrinsic and observed transaction prices. pt = mt + Q t Ct ;

ð1Þ

mt = mt−1 + Q t Zt + Ut ;

ð2Þ

Ct = c0 + c1 Vt ;

ð3Þ

Zt = z0 + z1 Vt :

ð4Þ

where, pt is the observed transaction price, mt is the intrinsic value of the security and Vt is the number of shares in the transaction at time t. Ut captures the arrival of public information and any rounding error. Q t is a trade indicator that takes the value +1 if the transaction is buyer initiated and − 1 if the transaction is seller initiated. The adverse selection component is Zt and the order processing component is Ct , where both are linear functions of Vt. Solving for the price change and incorporating the equations for Zt and Ct provide the following equation that can be estimated using OLS. Δpt = c0 ΔQ t + c1 ΔðQ t Vt Þ + z0 Q t + z1 Q t Vt + Ut ;

ð5Þ

The bid-ask spread is then measured as the sum of the order processing and adverse selection components, 2(c0 + c1Vt) and 2(z0 + z1Vt) respectively. The percentage adverse selection component of the spread is 2(z0 + z1Vt)/[2(c0 + c1Vt) + 2(z0 + z1Vt)]. 2.2. The MRR model The model developed by Madhavan et al. (1997) allows the estimation of the adverse selection component and a trading frictions component that captures both inventory and order processing costs. This model has been used by McInish and Van Ness (2002) to document intraday adverse selection costs on the NYSE and by Ahn et al. (2002) to measure these costs on the Tokyo exchange. Hatch and Johnson (2002) used the MRR model to examine the impact of specialist firm acquisitions on market quality. Hegde and McDermott (2004a,b) use the MRR model to examine whether the introduction of ETFs for the Dow Jones Industrial Average (Diamonds) and the NASDAQ 100 index (Q's) altered the liquidity of the underlying stocks. In the MRR model order flow is assumed to exhibit first order autocorrelation (ρ) and trades can take place within the quotes with a probability equal to θ. MRR shows that transaction price changes can be written as pt −pt−1 = α + ðϕ + λÞQ t −ðϕ + ρλÞQ t−1 + μ t

ð6Þ

Q t is the trade indicator variable that now also takes on a value of 0 if the trade is at the midpoint, α captures the constant drift in prices and ut is a composite error term that captures the change in price due to new information and errors due to price discreteness. The parameter vector (α, θ, ϕ, λ, and ρ) is estimated using Hansen's Generalized Method of Moments (GMM). The implied spread can be computed as S = 2ðθ + ϕÞ

ð7Þ

When the percentage adverse selection and the percentage market friction components are computed as spread proportions, the two components, can be expressed as %θ =

2θ 2ðθ + ϕÞ

order processing costs

ð8Þ

%ϕ =

2ϕ 2ðθ + ϕÞ

adverse selection costs

ð9Þ

3. Data and summary statistics

2 Subrahmanyam (1991) demonstrates this result for risk neutral traders but Gorton and Pennacchi (1993) show that the diversification of private information is robust to alternative assumptions about the risk preferences of traders.

Our ETF sample is drawn from ETFs that are comprised of US registered companies. We exclude ETFs that include constituents of non-US firms to ensure that our spread measures do not reflect an information disadvantage that US investors may have in trading foreign equities. Subrahmanyam (1991) shows theoretically that when informed traders have private information about a common factor (such as an industry or market-wide factor) adverse selection costs in the basket

P. Chelley-Steeley, K. Park / International Review of Financial Analysis 19 (2010) 65–76

will be diversified to a greater extent. Thus broad-market ETFs will contain less adverse selection costs than industry-wide ETFs but it is necessary that informed traders have private information about the common factor. A practical issue raised by Madhavan (2000, p. 6) is that “it is unlikely a trader has market-wide private information”.3 This means that market-wide ETFs will not have lower adverse selection costs than industry-wide ETFs. This is an empirical issue that we explore by comparing the adverse selection costs of industry and broad-market ETFs. Differences will depend upon the extent to which traders are informed about the common factor. We distinguish between broad-market ETFs and industry-based ETFs using the same classification procedure adopted by NASDAQ. We examine 39 broad-market ETFs and 38 industry-wide ETFs. We estimate the spread decomposition models for the period October 1st to December 31st 2005 using intraday price, quote and volume information obtained from TAQ. To screen the files for mistakes we employ a similar screening procedure used previously by Huang and Stoll (1996) and Chung, Chuwonganant and McCormick (2004). We exclude quotes if either the ask or bid price is less than or equal to zero; quotes if either the ask size or the size is less than or equal to zero; quotes if the bid-ask spread is greater than $5 or less than zero; quotes and trades before the open and after the close; trades if the price or volume is less than or equal to zero; the trade price, pt, if |(pt − pt − 1)/pt − 1|N0.5; an ask quote, at, if |(at − at − 1)/at − 1|N0.5; a bid quote, bt, if |(bt − bt− 1)/bt − 1|N0.5. We determine the direction of a trade by following the procedure developed by Lee and Ready (1991). This requires us to match quotes at time t to a trade that takes place five seconds later. A trade is assumed to take place at the bid (ask) if the midpoint price 5 s earlier is above (below) the transaction price. These trades are then classified as +1 and −1 respectively. Since the MRR model applies a three way classification we also denote trades at the midpoint as a 0 in accordance with Madhavan et al. (1997). We match to each ETF a control stock using the method reviewed by Madhavan (2000) and applied by Hatch and Johnson (2002). This allows us to select a control stock for each ETF that minimizes the function outlined in Eq. (10). The security characteristics used in the matching procedure are volume, price, and return volatility that were shown by Stoll (2000) to explain the cross-sectional variation in spreads. For the industry ETFs we also ensure that the matched control stock comes from the same industry by using the two digit SIC code. Once a control security is selected it cannot be chosen for another ETF. As a robustness check we also make comparisons with control securities that are matched using volume instead of the three component model below. 

ðVolC −VolE Þ ðVolC + VolE Þ=2

2

 +

2  2 ðPC −PE Þ ðσC −σE Þ + ðPC + PE Þ=2 ðσC + σE Þ=2

ð10Þ

Where, VolC (VolE) is the average daily volume of a control stock (ETF), PC (PE) is the average daily price of the control stock (ETF), σC (σE) is the daily return volatility of a control stock (ETF). Eq. (10) is calculated using the test period 1st of July to 30th of September 2005. Once the controls are selected we estimate the adverse selection models for the control securities and the ETFs using the period 1st October to 31st December 2005. Table 1 panel (a) provides a list of the broad-market ETFs along with their control stocks. Table 1 panel (b) provides a list of 3 A trader cannot obtain information about the industry and market as a whole that other traders cannot access so it is not possible to be privately informed about a common factor.

67

the industry-wide ETFs and their control securities. This table shows that the values from this matching procedure are below unity suggesting that the controls are a good match for the ETFs. Hegde and McDermott (2004a,b) showed that when a security becomes a component of an ETF the adverse selection costs of the security decline. In Table 1 we highlight with an * those control securities that are constituents of an ETF. In all only ten of the thirty nine broad-market control securities are not constituents of our ETF sample and five control securities out of the thirty eight industry-wide ETFs.4 This suggests that our sample is predominately drawn from securities that are likely to have lower adverse selection costs than non constituent stocks. A finding that ETFs have lower adverse selection costs than controls is likely to be even stronger in the general population of securities. 4. Results In Table 2 panels (a) and (b) we provide information about the liquidity of the ETFs and the matched control stocks. These tables show that trades in ETFs are over twice the size of trades in the control securities. It is also noted that the spread measures associated with broad ETFs are considerably smaller than for the control securities in both absolute and percentage terms. The mean quoted spread of the broad ETFs is 0.13 (0.21) while the mean spread of the control securities is 0.41 (0.63). Similarly, the industry-wide ETFs suggest that the spread is higher in the control stocks. The mean quoted spread of the industry ETFs is only 0.14 (0.40) but the spread of the controls is 0.27 (0.77). Although calculated, for brevity we do not report the time-weighted spreads of each ETF and control. However, differences between the quoted and effective spread of the ETFs and the control securities are also reflected in the time-weighted spread. The mean time-weighted spread for the ETF samples and control samples are 0.29 and 0.70 respectively. This difference is found to be statistically significant using both a t-test and a Wilcoxen signed rank test.5 The results from the estimation of the adverse selection models for each of the ETFs and the control securities we employ are contained in Tables 3 and 4. 4.1. Glosten and Harris In Table 3(a) and (b) we provide the results from estimating adverse selection costs with the Glosten–Harris model for the broadmarket ETFs and the industry-wide ETFs respectively. In Table 3(c) we report the results for the combined samples. In the Glosten–Harris model co and c1 measure the order processing component of the spread. The c0 coefficient measures the fixed component and c1 measures the variable component as a function of trading volume. The estimates of c0 are almost uniformly positive and significant for all ETFs and control securities. This indicates that the order processing costs unrelated to firm size tend to be positive for both groups of securities. When the c1 estimates are negative it indicates that order processing costs decline with trade size. Although both ETFs and the control securities display a tendency for the c1 coefficients to be positive, fewer of these coefficients are significant for ETFs. This shows that for both ETFs and control securities order processing costs do not tend to fall with trade size. The coefficients z0 and z1 measure the adverse selection component of the spread. The z0 coefficient measures the part of the adverse selection component that is independent of trade size and the z1 4 The identification of control securities uses volume as an important control characteristic. Volume in ETFs is very high in comparison to that of securities. There is a tendency within our control procedure to choose high volume securities as a control. High volume securities are also more likely to be ETF constituents. 5 The mean time-weighted spreads of the broad ETFs and their controls is 0.39 and 0.79 respectively. The mean time-weighted spreads for the broad-market ETFs and their controls are 0.20 and 0.63. Both of these differences are statistically significant.

68

Panel (a) Broad-market ETFs and controls Symbol

Inception

Product name

Tracked index

Volume

Control

Company name

Volume

Dist

DIA DSG DSV ELV IJH IJJ IJK IJR IJS IJT IVE IVV IVW IWB IWD IWF IWM IWN IWO IWP IWR IWV JKH JKJ JKK JKL MDY PEY PWC PWO QQQQ VB VBK VBR VO VTI VTV VV VXF

20-Jan-98 25-Sep-00 25-Sep-00 25-Sep-00 22-May-00 24-Jul-00 24-Jul-00 22-May-00 24-Jul-00 24-Jul-00 22-May-00 15-May-00 22-May-00 15-May-00 22-May-00 22-May-00 22-May-00 24-Jul-00 24-Jul-00 17-Jul-01 17-Jul-01 22-May-00 02-Jul-04 02-Jul-04 07-Jul-04 06-Jul-04 04-May-95 09-Dec-04 01-May-03 01-May-03 10-Mar-99 26-Jan-04 26-Jan-04 26-Jan-04 26-Jan-04 31-May-01 26-Jan-04 27-Jan-04 27-Dec-01

DIAMONDS streetTRACKS DJ Wilshire Small Cap Growth ETF streetTRACKS DJ Wilshire Small Cap Value ETF streetTRACKS DJ Wilshire Large-Cap Value ETF iShares S&P MidCap 400 Index Fund iShares S&P MidCap 400 Value Index Fund iShares S&P MidCap 400 Growth Index Fund iShares S&P SmallCap 600 Index Fund iShares S&P SmallCap 600 Value Index Fund iShares S&P SmallCap 600 Growth Index Fund iShares S&P 500 Value Index Fund iShares S&P 500 Index iShares S&P 500 Growth Index Fund iShares Russell 1000 iShares Russell 1000 Value iShares Russell 1000 Growth iShares Russell 2000 iShares Russell 2000 Value iShares Russell 2000 Growth iShares Russell Midcap Growth Index Fund iShares Russell Midcap Index Fund iShares Russell 3000 iShares Morningstar Mid Growth Index iShares Morningstar Small Core Index iShares Morningstar Small Growth Index iShares Morningstar Small Value Index MidCap SPDRs POWERSHARES E T F TRUST PowerShares Dynamic Market Portfolio PowerShares Dynamic OTC Portfolio Nasdaq-100 Index Tracking Stock Vanguard Small Cap VIPERs Vanguard Small-Cap Growth VIPERs Vanguard Small-Cap Value VIPERs Vanguard Mid-Cap VIPERs Vanguard Total Stock Market VIPERs Vanguard Value VIPERs Vanguard Large-Cap VIPERs Vanguard Extended Market VIPERs

Dow Jones Industrial Average Dow Jones Wilshire Small Cap Growth Index Dow Jones Wilshire Small Cap Value Index Dow Jones Wilshire Large-Cap Value Index S&P MidCap 400 Index S&P MidCap 400/BARRA Value Index S&P MidCap 400/BARRA Growth Index S&P SmallCap 600 Index S&P SmallCap 600/BARRA Value Index S&P SmallCap 600/BARRA Growth Index S&P 500/BARRA Value Index S&P 500 Index S&P 500/BARRA Growth Index Russell 1000 Index Russell 1000 Value Index Russell 1000 Growth Index Russell 2000 Index Russell 2000 Value Index Russell 2000 Growth Index Russell MidCap Growth Index Russell MidCap Index Russell 3000 Index Morningstar Mid Growth Index Morningstar Small Core Index Morningstar Small Growth Index Morningstar Small Value Index Standard & Poor's MidCap 400 Index Dividend Achievers 50 Index Dynamic Market Intellidex Index Dynamic OTC Intellidex Index Nasdaq-100 Index MSCI U.S. Small Cap 1750 Index MSCI U.S. Small Cap Growth 1750 Index MSCI U.S. Small Cap Value 1750 Index MSCI U.S. Mid Cap 450 Index MSCI U.S. Broad-market Index MSCI U.S. Prime Market Value Index MSCI U.S. Prime Market 750 Index Wilshire 4500 Completion Index

6956065 2025 3048 7009 169623 201975 97709 1289262 176246 94821 247568 768209 239596 477685 675448 734865 23289239 864953 893392 132775 113631 115660 10479 8153 4482 7284 2050306 161148 76239 30390 80030105 19803 14712 18067 30070 130410 25357 40312 16689

MO* UTL Y* STU* HB CYN* CBSH* NOC* MRBK ESE* MCY* CB* UB* TMK* BCR* LH* GE* CMA* ABC* IDXX* WPS EEP COKE* SFSW* BF BARI LEH BFIN* KMR CFFN* INTC* TRH* YANB WSFS* GBL* PDX* ERIE* IBOC* BANF*

ALTRIA GROUP INC UNITIL CORP ALLEGHANY CORP DE STUDENT LOAN CORP HILLENBRAND INDS INC CITY NATIONAL CORP COMMERCE BANCSHARES INC NORTHROP GRUMMAN CORP MERCANTILE BANKSHARES CORP E S C O TECHNOLOGIES INC MERCURY GENERAL CORP NEW CHUBB CORP UNIONBANCAL CORP TORCHMARK CORP BARD C R INC LABORATORY CORP AMERICA HLDGS GENERAL ELECTRIC CO COMERICA INC AMERISOURCEBERGEN CORP I D E X X LABORATORIES INC W P S RESOURCES CORP HOLDING CO ENBRIDGE ENERGY PARTNERS LP COCA COLA BOTTLING CO CONS STATE FINANCIAL SERVICES CORP BROWN FORMAN CORP BANCORP RHODE ISLAND INC LEHMAN BROTHERS HOLDINGS INC BANKFINACIAL CORP KINDER MORGAN MANAGEMENT LLC CAPITOL FEDERAL FINANCIAL INTEL CORP TRANSATLANTIC HOLDINGS INC YARDVILLE NATIONAL BANCORP WSFS FINANCIAL CORP GAMCO INVESTORS INC PEDIATRIX MEDICAL GROUP ERIE INDEMNITY CO INTERNATIONAL BANCSHARES CORP BANCFIRST CORP

6943895 2306 5215 9403 171178 201515 110748 1371482 172236 77898 199573 813104 277196 501623 690060 687262 20435200 879142 896554 140180 117257 71156 8920 8661 4604 7146 1871851 161813 74134 33019 49778952 21171 11115 21606 33953 161601 25329 39429 17106

0.4060 0.9445 0.8248 0.4556 0.1321 0.0628 0.1440 0.0289 0.0303 0.4095 0.2992 0.2448 0.1387 0.0921 0.2337 0.1696 0.4562 0.0350 0.0103 0.2017 0.2224 0.3307 0.7676 0.3752 0.0465 0.4162 0.1793 0.0242 0.0851 0.1712 0.7576 0.0187 0.3871 0.1791 0.4970 0.5978 0.0437 0.4010 0.4857

P. Chelley-Steeley, K. Park / International Review of Financial Analysis 19 (2010) 65–76

Table 1 The Sample of ETFs and control securities.

Panel (b) Industry-Wide ETFs and Controls Inception

Product Name

Sector/ tracked industry

Control

Company Name

BDH BHH HHH IAH IBB IDU IGE IGM IGN IGV IGW IIH IYC

04-Apr-00 24-Feb-00 24-Sep-99 25-Feb-00 05-Feb-01 12-Jun-00 22-Oct-01 13-Mar-01 10-Jul-01 10-Jul-01 10-Jul-01 25-Feb-00 12-Jun-00

BroadBand B2B Internet Internet Internet architecture Health Utilities Natural Res Technology Technology Technology Technology Internet Infrastructure DJ U.S. Consumer Services Index

173621 44428 407012 81576 977192 69692 79937 47970 102742 174032 246871 59964 43092

ACTL* QVDX* CERN* PQE IVGN* UIL* PVA* LIFE MRCY* IMN* ATK* AMSWA* BL

ACTEL CORP QUOVADX INC CERNER CORP PROQUEST CO INVITROGEN CORP U I L HOLDINGS CORP PENN VIRGINIA CORP LIFELINE SYSTEMS INC MERCURY COMPUTER SYSTEMS IMATION CORP ALLIANT TECHSYSTEMS INC AMERICAN SOFTWARE INC BLAIR CORP

176076 55826 390190 109364 914890 52253 72575 41324 117866 212600 261171 54876 47593

0.3309 0.1325 0.1184 0.1477 0.0178 0.6036 0.5103 0.5603 0.3221 0.1343 0.0730 0.2144 0.6740

IYE IYG IYH IYJ IYK

12-Jun-00 12-Jun-00 12-Jun-00 12-Jun-00 12-Jun-00

Natural Res Financial Health DJ U.S. Industrials Index DJ U.S. Goods Index

118121 29085 151492 23892 42301

MIDD* BOKF* UTR* VTRU ROG*

MIDDLEBY CORP B O K FINANCIAL CORP UNITRIN INC VERTRUE INC ROGERS CORP

90222 38275 168439 24972 66490

0.4661 0.8227 0.3340 0.8792 0.7640

IYM

12-Jun-00

75651

CW*

CURTISS WRIGHT CORP

82823

0.4489

IYT

06-Oct-03

DJ Transportation Average Index

136176

FLA

FLORIDA EAST COAST IND INC

99534

0.4444

IYZ

22-May-00

Communications

239857

WFSL*

WASHINGTON FEDERAL INC

251932

0.1853

OIH

06-Feb-01

BROADBAND HOLDRS TRUST B 2 B INTERNET HOLDRS TRUST INTERNET HOLDRS TRUST INTERNET ARCHITECTURE HOLDRS TR iShares Nasdaq Biotechnology iShares Dow Jones U.S. Utilities Sector Index Fund iShares Goldman Sachs Natural Resources iShares Goldman Sachs Technology iShares Goldman Sachs Networking iShares Goldman Sachs Software iShares Goldman Sachs Semiconductor INTERNET INFRASTRUCTURE HLDRS TR iShares Dow Jones U.S. Consumer Services Sector Index Fund iShares Dow Jones U.S. Energy Sector Index Fund iShares Dow Jones U.S. Financial Services Index Fund iShares Dow Jones U.S. Healthcare Sector Index Fund iShares Dow Jones U.S. Industrial Sector Index Fund iShares Dow Jones U.S. Consumer Goods Sector Index Fund iShares Dow Jones U.S. Basic Materials Sector Index Fund iShares Dow Jones Transportation Average Index Fund iShares Dow Jones U.S. Telecommunications Sector Index Fund OIL SERVICE HOLDRS TRUST

5491767

IBM*

5905220

0.3713

RKH RTH SMH UTH VAW VCR

26-Jun-00 02-May-01 26-May-00 26-Jun-00 26-Jan-04 26-Jan-04

REGIONAL BANK HOLDRS TRUST RETAIL HOLDRS TRUST SEMICONDUCTOR HOLDRS TRUST UTILITIES HOLDRS TRUST Vanguard Materials VIPERs Vanguard Consumer Discretionary VIPERs

655056 3672783 18421435 379027 11656 7929

ZION* TGT* DELL* EQT* SCL* FSCI*

592771 3765931 16852963 350667 11654 9067

0.6953 0.4172 0.0325 0.42407 0.8235 0.8148

VDE VFH

29-Sep-04 26-Jan-04

Vanguard Energy VIPERs Vanguard Financials VIPERs

Regional Bank Retail Semiconductor Utilities Natural Res MSCI U.S. Investable Market Consumer Discretionary Index Natural Res Financial

INTERNATIONAL BUSINESS MACHS COR ZIONS BANCORP TARGET CORP DELL INC EQUITABLE RESOURCES INC STEPAN CO FISHER COMMUNICATIONS INC

VPU XLB XLE XLK XLU XLV XLY

26-Jan-04 22-Dec-98 22-Dec-98 22-Dec-98 22-Dec-98 22-Dec-98 22-Dec-98

Vanguard Utilities VIPERs Select Sector SPDR SECTOR SPDR TRUST Select Sector SPDR Select Sector SPDR Select Sector SPDR Select Sector SPDR

Utilities Materials Energy Technology Utilities Health care Consumer Discretionary

17187 2496942 14027384 1050464 1934689 831734 1036512

DJ U.S. Basic Materials Index

Oil Services

Volume

39581 7275

TNC* COBH* CHG* MAS* XOM* CVC* SO* PFG* CTL*

TENNANT CO PENNSYLVANIA COMMERCE BANCORP IN C H ENERGY GROUP INC MASCO CORP EXXON MOBIL CORP CABLEVISION SYSTEMS CORP SOUTHERN CO PRINCIPAL FINANCIAL GROUP INC CENTURYTEL INC

Volume

Dist

36556 5380

0.3720 0.7376

32182 2265406 18119117 1240800 1906203 929846 1118901

0.5452 0.1456 0.1135 0.1877 0.0063 0.2522 0.0158

P. Chelley-Steeley, K. Park / International Review of Financial Analysis 19 (2010) 65–76

Symbol

Inception is the date an ETF became active for the first time. Product name is the name by which an ETF is known in the marketplace. Control is the abbreviated name of the control stock used thereafter. Company name is the name of the  2  2  2 ðVolC −VolE Þ C −PE Þ C −σE Þ + ðPCðP+ + ðσCðσ+ Subscripts C and E represent Control securities and ETFs, respectively. PE Þ = 2 σE Þ = 2 ðVolC + VolE Þ = 2

control stock. Volume represents trading volume. Dist is a computed measure of a chosen control stock that minimizes

VolC (VolE) is the average daily volume of a control stock (ETF). PC (PE) is the average daily price of the control stock (ETF). σC (σE) is the standard deviation of daily return of control stock (ETF). *indicates that a control security is also a component of an ETF used in our sample.

69

70

P. Chelley-Steeley, K. Park / International Review of Financial Analysis 19 (2010) 65–76

Table 2 Summary statistics. ETF

Trade price

Trade size

Percent Quoted S

Effective S

Percent effective S

Control

Trade price

Trade size

Quoted S

Percent quoted S

Effective S

Percent effective S

Panel (a) Broad-market ETFs and controls DIA 105.78 1560 0.03 DSG 79.24 377 0.32 DSV 61.38 575 0.25 ELV 68.97 504 0.22 IJH 71.58 652 0.09 IJJ 68.41 756 0.11 IJK 73.36 1020 0.12 IJR 56.80 1354 0.11 IJS 62.91 634 0.12 IJT 113.67 695 0.21 IVE 63.61 659 0.08 IVV 122.56 1261 0.08 IVW 58.45 663 0.10 IWB 66.84 935 0.10 IWD 68.34 1280 0.11 IWF 50.61 1611 0.09 IWM 65.64 3142 0.02 IWN 65.37 1490 0.11 IWO 68.27 1336 0.10 IWP 91.64 976 0.13 IWR 86.18 722 0.12 IWV 71.07 1011 0.10 JKH 75.91 638 0.12 JKJ 71.64 769 0.20 JKK 66.91 658 0.13 JKL 70.33 813 0.24 MDY 130.62 1422 0.07 PEY 14.93 927 0.07 PWC 44.98 598 0.15 PWO 48.76 550 0.22 QQQQ 40.05 4362 0.01 VB 58.98 699 0.13 VBK 57.57 552 0.21 VBR 61.05 398 0.17 VO 63.13 572 0.13 VTI 121.04 595 0.19 VTV 56.26 748 0.11 VV 54.84 710 0.10 VXF 88.98 630 0.29 Mean 70.94 996 0.13 StdDev 23.04 742 0.07 Median 66.91 722 0.12

0.03 0.40 0.41 0.32 0.13 0.16 0.17 0.19 0.18 0.18 0.13 0.07 0.16 0.15 0.16 0.18 0.04 0.17 0.15 0.14 0.14 0.14 0.16 0.28 0.20 0.35 0.05 0.48 0.32 0.45 0.03 0.23 0.35 0.28 0.20 0.16 0.20 0.19 0.33 0.21 0.11 0.18

0.01 0.09 0.07 0.06 0.03 0.04 0.05 0.03 0.04 0.07 0.03 0.02 0.03 0.03 0.03 0.02 0.01 0.03 0.03 0.04 0.04 0.04 0.04 0.06 0.04 0.05 0.02 0.03 0.05 0.08 0.01 0.05 0.06 0.06 0.05 0.05 0.04 0.04 0.09 0.04 0.02 0.04

0.01 0.11 0.12 0.09 0.04 0.05 0.06 0.05 0.06 0.07 0.04 0.02 0.06 0.04 0.04 0.05 0.02 0.05 0.04 0.05 0.04 0.05 0.06 0.08 0.07 0.08 0.01 0.18 0.12 0.16 0.02 0.08 0.11 0.09 0.07 0.04 0.07 0.07 0.10 0.07 0.04 0.06

MO UTL Y STU HB CYN CBSH NOC MRBK ESE MCY CB UB TMK BCR LH GE CMA ABC IDXX WPS EEP COKE SFSW BF BARI LEH BFIN KMR CFFN INTC TRH YANB WSFS GBL PDX ERIE IBOC BANF Mean StdDev Median

74.07 25.61 297.21 221.97 47.07 70.80 52.66 55.41 56.83 43.04 58.85 92.66 67.98 53.30 65.68 50.36 34.74 57.55 76.36 68.84 54.91 47.34 44.91 37.53 73.58 34.99 120.51 14.11 47.33 33.83 24.94 61.93 35.12 61.52 44.84 80.18 52.65 29.71 81.25 65.44 50.77 54.91

740 266 200 189 391 329 218 561 217 327 256 322 436 525 415 504 982 338 361 215 348 355 186 267 385 332 394 332 351 220 3309 204 196 220 187 285 272 297 151 412 502 327

0.17 0.45 2.20 1.99 0.29 0.28 0.10 0.27 0.09 0.34 0.45 0.27 0.17 0.30 0.77 0.20 0.04 0.30 0.41 0.14 0.32 0.44 0.56 0.12 0.29 0.45 0.49 0.15 0.15 0.13 0.02 0.93 0.34 0.36 0.26 0.36 0.15 0.12 1.23 0.41 0.46 0.29

0.23 1.76 0.74 0.89 0.63 0.39 0.19 0.49 0.15 0.79 0.77 0.29 0.25 0.57 1.17 0.40 0.12 0.53 0.54 0.20 0.58 0.92 1.26 0.32 0.40 1.28 0.41 0.99 0.31 0.38 0.06 1.51 0.96 0.59 0.57 0.45 0.28 0.40 1.61 0.63 0.43 0.53

0.05 0.15 0.73 0.70 0.06 0.08 0.04 0.05 0.03 0.09 0.09 0.05 0.05 0.07 0.12 0.04 0.01 0.06 0.07 0.05 0.06 0.07 0.20 0.04 0.07 0.17 0.06 0.06 0.05 0.05 0.01 0.10 0.11 0.13 0.09 0.10 0.05 0.04 0.50 0.12 0.16 0.06

0.06 0.60 0.25 0.31 0.13 0.11 0.07 0.09 0.06 0.21 0.16 0.05 0.07 0.13 0.19 0.07 0.02 0.10 0.10 0.07 0.11 0.16 0.45 0.10 0.10 0.48 0.05 0.39 0.10 0.14 0.03 0.17 0.32 0.21 0.19 0.12 0.09 0.13 0.66 0.18 0.15 0.12

Panel (b) Industry-wide BDH 18.67 BHH 2.51 HHH 64.77 IAH 35.41 IBB 75.43 IDU 76.47 IGE 85.74 IGM 46.81 IGN 30.92 IGV 40.57 IGW 59.31 IIH 3.81 IYC 58.77 IYE 83.64 IYG 110.72 IYH 61.43 IYJ 56.50 IYK 52.95 IYM 48.80 IYT 72.23 IYZ 23.19 OIH 120.78 RKH 137.18 RTH 95.27 SMH 36.02 UTH 113.63 VAW 57.64 VCR 52.07 VDE 71.76

0.44 3.69 0.24 0.30 0.23 0.17 0.43 0.35 0.29 0.23 0.21 3.13 0.26 0.18 0.17 0.21 0.23 0.34 0.26 0.18 0.20 0.14 0.15 0.29 0.11 0.16 0.25 0.26 0.43

0.03 0.02 0.03 0.03 0.06 0.04 0.12 0.06 0.03 0.03 0.04 0.02 0.05 0.05 0.06 0.04 0.04 0.06 0.04 0.04 0.01 0.05 0.05 0.09 0.01 0.05 0.05 0.05 0.09

0.13 0.69 0.04 0.08 0.08 0.06 0.14 0.12 0.09 0.08 0.07 0.63 0.08 0.06 0.05 0.07 0.08 0.11 0.09 0.06 0.06 0.04 0.04 0.09 0.03 0.05 0.09 0.10 0.12

ACTL QVDX CERN PQE IVGN UIL PVA LIFE MRCY IMN ATK AMSWA BL MIDD BOKF UTR VTRU ROG CW FLA WFSL IBM ZION TGT DELL EQT SCL FSCI TNC

13.99 2.68 89.36 31.00 67.34 49.20 56.28 35.10 20.40 42.99 73.27 5.96 38.79 77.29 45.95 46.02 36.35 37.87 58.64 42.53 23.15 83.95 73.40 54.39 31.19 37.59 25.40 46.16 45.19

288 682 269 358 339 209 267 240 345 254 377 455 187 206 203 226 178 215 288 244 289 491 272 514 1960 434 637 174 377

0.05 0.07 0.27 0.18 0.09 0.40 0.85 0.19 0.06 0.28 0.30 0.20 0.62 0.29 0.17 0.37 0.26 0.44 0.30 0.22 0.04 0.26 0.08 0.17 0.03 0.25 0.29 0.49 0.38

0.34 2.56 0.30 0.57 0.13 0.82 1.51 0.55 0.29 0.65 0.41 3.43 1.60 0.38 0.37 0.80 0.72 1.18 0.51 0.53 0.16 0.31 0.10 0.31 0.09 0.67 1.15 1.05 0.85

0.02 0.02 0.12 0.06 0.03 0.08 0.12 0.06 0.02 0.07 0.08 0.06 0.10 0.10 0.06 0.08 0.10 0.09 0.07 0.08 0.01 0.03 0.02 0.03 0.01 0.05 0.09 0.17 0.08

0.11 0.91 0.13 0.21 0.04 0.16 0.21 0.17 0.10 0.17 0.11 1.10 0.27 0.14 0.13 0.17 0.27 0.24 0.12 0.18 0.06 0.04 0.03 0.06 0.03 0.15 0.37 0.37 0.18

ETFs and 1472 1644 867 820 1211 585 445 645 704 1105 1035 1212 1211 565 446 664 644 539 591 1251 1277 920 1385 2007 1791 979 404 1034 348

Quoted S

controls 0.08 0.09 0.15 0.11 0.17 0.13 0.37 0.16 0.09 0.09 0.12 0.12 0.15 0.15 0.18 0.13 0.13 0.18 0.12 0.13 0.05 0.17 0.20 0.28 0.04 0.18 0.14 0.14 0.31

P. Chelley-Steeley, K. Park / International Review of Financial Analysis 19 (2010) 65–76

71

Table 2 (continued) ETF

Trade price

Trade size

Panel (b) Industry-wide ETFs and VFH 55.28 405 VPU 65.45 362 XLB 28.56 4464 XLE 49.45 2295 XLK 21.01 1379 XLU 31.41 1958 XLV 31.01 1350 XLY 32.34 2091 Mean 59.96 1138 StdDev 31.40 772 Median 55.28 1034

Quoted S controls 0.24 0.20 0.03 0.04 0.02 0.04 0.04 0.05 0.14 0.08 0.13

Percent Quoted S

Effective S

Percent effective S

Control

0.42 0.31 0.10 0.07 0.11 0.12 0.11 0.16 0.40 0.74 0.23

0.10 0.06 0.01 0.01 0.01 0.01 0.01 0.02 0.04 0.03 0.04

0.17 0.09 0.03 0.03 0.04 0.04 0.03 0.05 0.11 0.14 0.08

COBH CHG MAS XOM CVC SO PFG CTL Mean StdDev Median

Trade price 33.30 46.45 29.25 57.89 25.24 34.61 48.76 32.98 43.24 20.14 42.53

Trade size

Quoted S

Percent quoted S

Effective S

Percent effective S

239 225 681 766 1179 602 422 375 417 332 289

0.82 0.54 0.17 0.08 0.14 0.10 0.24 0.14 0.27 0.20 0.24

2.48 1.17 0.57 0.14 0.54 0.29 0.50 0.44 0.77 0.73 0.54

0.27 0.17 0.04 0.02 0.03 0.02 0.03 0.03 0.07 0.05 0.06

0.83 0.37 0.15 0.03 0.11 0.06 0.07 0.10 0.21 0.24 0.15

This table presents the mean spread values for broad-market ETFs and control stocks. Trade price is the mean sample traded price. Trade size is the mean sample trade size.The quoted S is the difference between the quoted ask and the quoted bid price. The effective S is the difference between the midpoint price and the subsequent transaction price. Standard deviation is the standard deviation of the spread estimates. Percent refers to percentage values of the spread.

measures the element linearly related to trade size. In most cases the z0 estimates associated with the market-wide and the industry-wide ETFs are negative or insignificant. However, for the control securities estimates of z0 are almost always significantly positive. The mean adverse selection cost is insignificant for both the broad and the industry-wide ETFs while for both sets of control securities the mean estimates of zo are positive and significant. This suggests that the adverse selection component is much less important for the ETFs than for the control securities. The model of Easley and O'Hara (1987) allowed informed traders to choose to trade larger amounts at any given price. A prediction of this model is that the z1 coefficients should be positive to ensure that the costs of engaging in larger trades reflect the likelihood of higher adverse selection costs. The results for both the ETFs and the control securities indicate no clear relationship between trade size and adverse selection costs. Some of the z1 coefficients are negative but few are statistically significant (any sign) for either the ETFs or the control securities. For the broad-market ETFs four of the z1 coefficients are positive6 and significant and five for the control securities. A similar picture is apparent for the industry-wide ETFs as only five z1 coefficients are significantly positive but eight are for the control securities.7 The high number of negative z0 and z1 coefficients associated with the ETFs cause overall estimates of the percentage adverse selection component of the spread to be negative for many of the ETFs. Over half of the adverse selection components are negative for the ETFs. In contrast percentage adverse selection costs for the control securities are almost always positive. In general the results from this model suggest an absence of adverse selection costs associated with ETFs but positive adverse selection component costs associated with the control securities. The mean adverse selection cost across the broadmarket and industry ETFs are insignificant but are about 30% for both groups of control securities and are therefore comparable to those identified by Glosten and Harris (1988) and Neal and Wheatley (1998) for individual securities. The Wilcoxon test indicates that the costs associated with all samples of ETFs are significantly lower than those of the control stocks. When selecting control securities using our alternative control procedure, which matches securities based on only volume, the results of Table 3 are corroborated. When we compare the mean adverse selection costs of the broadmarket ETFs and the industry-wide ETFs a Wilcoxon test suggests that differences between the two samples are not significant (p-value is 0.236). This suggests that although it is theoretically possible to

6 7

This rises to six at a 10% level. This rises to nine at a 10% level.

diversify the information asymmetry of a common factor we find no evidence to suggest that this is happening within our sample. 4.2. MRR In Table 4(a), (b) and (c) we report the results when estimating the MRR model using the controls and the ETFs, respectively. In the MRR model estimates of ϕ indicate dollar inventory costs while θ provides estimates of dollar adverse selection costs. Estimates of ϕ for the ETFs are almost uniformly positive and almost always significant. For the controls about two-thirds of these coefficients are positive and significant. These results indicate that average dollar inventory costs are higher for the ETFs than for the control securities since the mean value of ϕ is 0.0181 and 0.0165 for the broad-market and industry-wide ETFs, respectively, as opposed to 0.0069 and 0.0072 for their controls. Average values of θ are almost always positive for the control securities and for the ETFs. The higher values of θ associated with the control securities cause the average percentage adverse selection component of the spread to be noticeably higher in the control securities when compared to either the broad-market or the industry-wide ETFs. The MRR model suggests that for broad-market ETF's the mean adverse selection component is about 17.5% of the spread but for the controls it is almost 55%. For the industry-wide ETF's the mean adverse selection component is 24% and nearly 61% for the controls. Overall the adverse selection component of the spread is almost 21% for the ETFs and is about 58% for the controls. When selecting control securities using our alternative procedure using only volume we find that our results corroborate those presented in Table 4. Comparisons of the mean percentage adverse selection costs of broad and industry-wide ETFs suggest that the broad-market ETFs have slightly lower adverse selection costs than the industry-wide ETFs. A Wilcoxon test undertaken on estimates of adverse selection costs suggests that the difference between broad and industry ETFs is not significant (p-value 0.085). This suggests that market-wide ETFs do not have significantly lower adverse selection costs than the industry-wide ETFs. Overall our results have shown that ETFs have lower adverse selection costs than individual securities. We therefore support the assertion that ETFs are a good example of a basket security proposed by Subrahmanyam (1991) and allow the diversification of private information. 5. Conclusions Since introduction there has been a rapid growth in the market for exchange traded funds. Potential reasons for the popularity of ETFs include the relatively low trading costs associated with ETF portfolio investment. We propose an alternative explanation for the rapid growth of ETF markets and offer an explanation for the low costs

72

Table 3 Model estimates using Glosten and Harris. ETFs

C0

C1

Z0

Mean volume

Implied spread

Percent ASC

Controls

C0

C1

Z0

Z1

Mean volume

Implied spread

Percentage ASC

0.043*** − 0.0412 0.2500 7.8182 − 0.1825* 0.2438* 0.2675* 0.0388 0.1045 0.4847 − 0.0707 0.157*** 0.0046 − 0.1064* − 0.0238 0.0067 0.019*** − 0.0034 0.0534 − 0.1060 − 0.0134 0.0137 0.7526 − 4.2818 − 0.3820 0.9747 − 0.0276 0.0597 0.2435 0.8211 0.013*** − 0.2304 − 2.4315* − 0.0688 0.5681 0.2856 − 0.2183 1.1789 − 0.4550 0.1477 1.5352 0.0137 0.3368 0.4499

1560 377 575 504 652 756 1020 1354 634 695 659 1261 663 935 1280 1611 3142 1490 1336 976 722 1011 638 769 658 813 1422 927 598 550 4362 699 552 398 572 595 748 710 630 996 742 722

0.0145 0.1107 0.0734 0.0840 0.0291 0.0459 0.0619 0.0206 0.0453 0.0756 0.0327 0.0186 0.0413 0.0329 0.0259 0.0205 0.0110 0.0183 0.0173 0.0633 0.0577 0.0439 0.0859 0.0408 0.0510 0.0378 0.0217 0.0190 0.0434 0.0506 0.0073 0.0646 0.0680 0.0502 0.0382 0.0494 0.0358 0.0360 0.0575 0.0436 0.0232 0.0413 b 0.0001 0.4254

0.2350 0.1601 − 0.2168 − 0.4321 0.0078 − 0.0544 0.0036 0.0499 − 0.1038 − 0.1590 0.0320 0.3761 0.0060 −0.0389 0.0016 − 0.0755 0.2421 0.2753 0.3028 0.0169 0.0081 − 0.0212 0.1601 − 0.5641 − 0.2424 − 0.0061 0.3795 − 0.0348 − 0.0064 0.0333 0.0942 0.0980 − 0.0654 0.0754 −0.1755 − 0.1507 − 0.0781 0.0001 − 0.1461 − 0.0003 0.1913 0.0016 0.5224 b 0.0001

MO UTL Y STU HB CYN CBSH NOC MRBK ESE MCY CB UB TMK BCR LH GE CMA ABC IDXX WPS EEP COKE SFSW BF BARI LEH BFIN KMR CFFN INTC TRH YANB WSFS GBL PDX ERIE IBOC BANF Ave. StdDev Med Sign

0.0071*** 0.0258*** 0.0496 0.4653*** 0.0056*** 0.0084*** 0.0082*** 0.0042*** 0.0118*** 0.0136*** 0.0064*** 0.0070*** 0.0080*** 0.0047*** 0.0040*** 0.0065*** 0.0043*** 0.0049*** 0.0036*** 0.0137*** 0.0111*** 0.0146*** 0.0626*** 0.0407*** 0.0128*** 0.0071*** 0.0083*** 0.0100*** 0.0155*** 0.0101*** 0.0043*** 0.0371*** 0.0229*** 0.0477*** 0.0072 0.0174*** 0.0100*** 0.0107*** 0.0266*** 0.0266 0.0735 0.0100 b0.0001

0.042*** − 3.4163 135.10 − 394.65 − 0.815 − 1.936** 0.553*** 0.496*** 1.122*** − 0.6757 − 0.4372 0.593*** 0.586** 0.1576 0.745*** 0.362*** 0.013*** 0.788*** 1.067*** 3.027*** 0.39 1.9161 0.5908 − 36.571 0.3496 1.5869 0.74*** 0.617** − 1.5212 0.9453 − 0.0007* − 2.8386 4.1124 1.2557 95.23*** 0.9422 0.699*** 0.691** 9.1099 − 4.5900 69.54 0.5908 0.0034

0.0013*** 0.0137*** 0.3963 −0.0490 0.0024* 0.0045*** 0.0046*** 0.0030*** 0.0052*** 0.0033** 0.0048* 0.0044*** 0.0057*** 0.0019** 0.0039*** 0.0023*** 0.0002*** 0.0028*** 0.0045*** 0.0080*** 0.0048*** 0.0065*** 0.0164*** − 0.0045 0.0042*** 0.0083*** 0.0063*** 0.0023*** 0.0029* 0.0039*** 0.0005*** 0.0024 0.0131*** 0.0207*** 0.0114 0.0130*** 0.0055*** 0.0047*** 0.0154*** 0.0144 0.0636 0.0045 b0.0001

0.0009 26.872** − 637.69 220.63 0.8195 3.203*** 0.594*** − 0.38*** − 0.071 − 0.0412 1.207 0.0064 − 0.6592 − 0.376** − 0.95*** − 0.0658 0.011*** − 0.2267 − 0.165 − 1.2763* − 0.3616 − 1.4508 −9.8877 11.4517 0.4953 − 1.7448 − 0.42*** 0.3128 2.8083 − 0.0409 0.017*** 10.5293 − 6.1452 − 0.1376 − 58.068 − 2.2052 − 0.0198 0.4196 9.3352 − 11.119 109.43 − 0.0412 0.5224

740 266 200 189 391 329 218 561 217 327 256 322 436 525 415 504 982 338 361 215 348 355 186 267 385 332 394 332 351 220 3309 204 196 220 187 285 272 297 151 412 502 327

0.0168 0.0914 0.8541 0.7685 0.0159 0.0265 0.0260 0.0146 0.0345 0.0333 0.0228 0.0232 0.0273 0.0129 0.0155 0.0178 0.0092 0.0158 0.0169 0.0441 0.0319 0.0426 0.1546 0.0585 0.0350 0.0306 0.0294 0.0253 0.0378 0.0284 0.0097 0.0820 0.0712 0.1371 0.0510 0.0600 0.0315 0.0315 0.0895 0.0759 0.1574 0.0314 b 0.0001

0.1554 0.4557 0.6274 − 0.0207 0.3391 0.4167 0.3602 0.3880 0.3029 0.1950 0.4465 0.3777 0.3961 0.2604 0.4455 0.2532 0.0573 0.3468 0.5299 0.3503 0.2922 0.2849 0.1889 − 0.0503 0.2730 0.5145 0.4175 0.1941 0.2071 0.2723 0.1066 0.1093 0.3345 0.2996 0.0203 0.4081 0.3529 0.3131 0.3772 0.3003 0.1605 0.3097 b 0.0001

P. Chelley-Steeley, K. Park / International Review of Financial Analysis 19 (2010) 65–76

Panel (a) Broad-market ETFs and their control securities DIA 0.0054*** 0.1030***6 0.0016*** DSG 0.0474*** −2.5666 0.0089 DSV 0.0450*** − 0.6326 − 0.0081 ELV 0.0616*** − 1.7935 − 0.0231** IJH 0.0142*** 0.302*** 0.0002 IJJ 0.0242*** 0.0161 − 0.0014*** IJK 0.0308*** − 0.0062 − 0.0002 IJR 0.0098*** 0.0301 0.0005** IJS 0.0250*** 0.0534 −0.0024*** IJT 0.0441*** −0.4885 −0.0063*** IVE 0.0156*** 0.333*** 0.0006* IVV 0.0058*** − 0.0178 0.0033*** IVW 0.0205*** 0.0510 0.0001 IWB 0.0170*** 0.0676 − 0.0005 0.0127*** 0.201*** 0.0000 IWD IWF 0.0110*** 0.0072 − 0.0008*** IWM 0.0040*** 0.064*** 0.0012*** IWN 0.0066*** 0.0186* 0.0025*** IWO 0.0057*** 0.175*** 0.0026*** IWP 0.0311*** 0.056 0.0006 IWR 0.0285*** 0.1615 0.0002 IWV 0.0222*** 0.1563 −0.0005 JKH 0.0370*** − 1.4917 0.0064 JKJ 0.0385*** − 8.5779 − 0.0082 JKK 0.0295*** 3.4322 − 0.0059 JKL 0.0193* − 0.6496 −0.0004 MDY 0.0064*** 0.236*** 0.0041*** PEY 0.0098*** 0.0507 −0.0004* PWC 0.0220*** − 0.2628 − 0.0003 PWO 0.0247*** − 0.6814* 0.0004 QQQQ 0.0033*** 0.007*** 0.0003*** VB 0.0283*** 1.2173 0.0033 VBK 0.0350*** 2.248** −0.0009 VBR 0.0235*** −0.7621 0.0019 VO 0.0225*** − 0.0794 −0.0037*** VTI 0.0285*** − 0.1856 − 0.0039*** VTV 0.0194*** − 0.1644 − 0.0012 VV 0.0184*** − 0.6346 −0.0008 VXF 0.0329*** 0.0042 −0.0039 Ave. 0.0228 −0.2564 − 0.0009 StdDev 0.0136 1.6577 0.0050 Med 0.0222 0.0161 −0.0002 Sign b0.0001 0.3368 1 SignedRank 0.0015 0.0646 b 0.0001

Z1

Panel (c) Full sample Ave. 0.0198 StdDev 0.0132 Med 0.0188 Sign b0.0001 Signed rank 0.0003

0.0660 1.4351 0.0445 0.0038 0.0947

− 0.0004 0.0041 0.0000 0.5666 b0.0001

0.0358 − 0.0599 − 0.1846** 0.0932 0.0289 − 0.7441* − 1.1152 0.1909 0.1988 − 0.1031 0.0322 0.0063 − 0.0179 − 0.1694 −1.9979 −0.2769 0.0968 − 0.1630 −0.0299 0.0497 0.1272 0.2310*** −0.2121*** 0.0256* 0.0378*** −0.0586 −1.3716 −0.0757 −0.04545 0.3485 4.0007 − 0.0026 0.0519*** 0.0095* 0.0008 −0.0017 0.0256*** − 0.0391 0.8226 0.0008 1 0.8767

1472 1644 867 820 1211 585 445 645 704 1105 1035 1212 1211 565 446 664 644 539 591 1251 1277 920 1385 2007 1791 979 404 1034 348 405 362 4464 2295 1379 1958 1350 2091 1138 772 1034

0.0175 0.0093 0.0189 0.0138 0.0254 0.0432 0.1140 0.0532 0.0291 0.0285 0.0360 0.0085 0.0515 0.0379 0.0724 0.0489 0.0503 0.0598 0.0520 0.0326 0.0176 0.0319 0.0343 0.0195 0.0072 0.0298 0.0452 0.0622 0.0563 0.0418 0.0574 0.0089 0.0126 0.0083 0.0112 0.0107 0.0108 0.0343 0.0230 0.0319 b0.0001 0.6517

0.0777 − 0.0648 0.1726 0.0839 0.0914 −0.0350 − 0.0432 −0.0267 − 0.0401 0.0613 0.0635 0.0070 − 0.1257 0.1400 0.2984 − 0.0418 −0.0344 − 0.0554 −0.0636 −0.0532 0.0416 0.2478 0.1922 0.2757 0.1872 0.2005 − 0.3446 − 0.0515 − 0.1315 0.0094 − 0.1414 0.1796 0.2130 0.1736 0.1524 0.0291 0.0169 0.0448 0.1364 0.0289 0.3240 b 0.0001

ACTL QVDX CERN PQE IVGN UIL PVA LIFE MRCY IMN ATK AMSWA BL MIDD BOKF UTR VTRU ROG CW FLA WFSL IBM ZION TGT DELL EQT SCL FSCI TNC COBH CHG MAS XOM CVC SO PFG CTL Ave. StdDev Med Sign

0.0052*** 0.0050*** 0.0203*** 0.0087*** 0.0076*** 0.0057* 0.0090*** 0.0199*** 0.0066*** 0.0088*** 0.0130*** 0.0088*** 0.0140*** 0.0313*** 0.0114*** 0.0039*** 0.0201*** 0.0088*** 0.0122*** 0.0189*** 0.0054*** 0.0058*** 0.0079*** 0.0069*** 0.0041*** 0.0074*** 0.0434*** 0.0461*** 0.0150*** 0.0669*** 0.0320*** 0.0038*** 0.0050*** 0.0042*** 0.0049*** 0.0048*** 0.0037*** 0.0137 0.0139 0.0087 b0.0001

0.0568 1.2360 0.0081 0.4222 0.6443

1065 755 816

0.0391 0.0234 0.0370 b0.0001 0.7350

0.0216 0.1683 0.0078 0.2067 b 0.0001

Ave. StdDev Med Sign

0.0203 0.0535 0.0088 b0.0001

0.324*** − 0.0349 0.721*** − 0.0605 0.774*** 8.4205 − 1.1589 − 0.0955 0.2369*** 2.4455** 0.4851 0.0388 − 1.5332 − 3.36*** 1.88*** 2.1465 1.1058 − 0.5088 5.178* − 0.2942 0.294*** 0.345*** 1.224*** 0.136*** 0.017*** − 0.0164 − 4.251 − 6.4991 2.9422* − 14.8013 − 15.2052 0.074*** 0.026*** 0.045*** 0.055*** −0.0097 0.0802 − 0.5088 4.2210 0.0550 0.1877

− 2.6031 49.6255 0.3349 0.0018

0.0020*** 0.0017*** 0.0081*** 0.0017 0.0041*** 0.0138*** 0.0070** 0.0073*** 0.0023*** 0.0073*** 0.0094*** 0.0024*** 0.0071*** 0.0138*** 0.0059*** 0.0073*** 0.0113*** 0.0142*** 0.0071* 0.0050 0.0012*** 0.0019*** 0.0050*** 0.0024*** 0.0007*** 0.0042*** 0.0127 0.0146*** 0.0054 0.0165* 0.0056 0.0014*** 0.0015*** 0.0019*** 0.0018*** 0.0015*** 0.0021*** 0.0059 0.0046 0.0050 b0.0001

0.0103 0.0455 0.0047 b0.0001

0.0137 0.01711 0.4645* − 1.0776 0.161*** − 11.3625 0.8167 − 0.2297 0.126** − 3.147** 0.7866 − 0.3717 10.359*** 2.834** − 2.311*** − 4.93** −3.6185 3.2548 − 9.303** 7.4279 0.232** 0.138*** 0.0305 0.0123 0.03*** − 0.2393 8.4379 23.4804* −2.9631 7.7471 5.2492 0.0072 0.017*** − 0.025* − 0.0007 −0.1838 − 0.1484 0.8577 5.6376 0.0171 0.3240

288 682 269 358 339 209 267 240 345 254 377 455 187 206 203 226 178 215 288 244 289 491 272 514 1960 434 637 174 377 239 225 681 766 1179 602 422 375 417 332 289

0.0144 0.0133 0.0574 0.0199 0.0239 0.0378 0.0318 0.0542 0.0181 0.0321 0.0457 0.0220 0.0454 0.0899 0.0344 0.0212 0.0620 0.0472 0.0363 0.0515 0.0136 0.0159 0.0264 0.0188 0.0099 0.0230 0.1174 0.1275 0.0408 0.1621 0.0706 0.0105 0.0131 0.0122 0.0133 0.0124 0.0115 0.0394 0.0352 0.0264 b 0.0001

0.2734 0.2544 0.2884 0.1260 0.3484 0.6052 0.4587 0.2660 0.2620 0.4076 0.4241 0.1990 0.3949 0.3191 0.3160 0.5839 0.3426 0.6315 0.2448 0.2666 0.1911 0.2560 0.3785 0.2614 0.1571 0.3558 0.3072 0.2944 0.2106 0.2164 0.1893 0.2771 0.2287 0.3038 0.2652 0.2283 0.3489 0.3102 0.1154 0.2734 b 0.0001

−5.2887 78.2286 0.0037 0.9088

415 425 324

0.0581 0.1161 0.0310 b 0.0001

0.3051 0.1395 0.2972 b 0.0001

P. Chelley-Steeley, K. Park / International Review of Financial Analysis 19 (2010) 65–76

Panel (b) Industry-wide ETFs and their control securities BDH 0.0081*** − 0.0475 0.0007** BHH 0.0049*** 0.0177 − 0.0002 HHH 0.0076*** 0.322*** 0.0018*** IAH 0.0064*** −0.092 0.0005 IBB 0.0112*** 0.303*** 0.0012*** IDU 0.0221*** 0.3719 − 0.0003 IGE 0.0597*** − 0.4704 −0.0020* IGM 0.0275*** − 0.3034 − 0.0008 IGN 0.0153*** −0.12 − 0.0008** IGV 0.0132*** 0.148*** 0.0010* IGW 0.0168*** 0.0946* 0.0011** IIH 0.0042*** 0.0464 0.0000 IYC 0.0290*** 0.0331 − 0.0032** IYE 0.0157*** 1.013*** 0.0028*** IYG 0.0247*** 1.5275 0.0117*** IYH 0.0255*** 0.0509 − 0.0008* IYJ 0.0255*** 0.7892 − 0.0009 IYK 0.0316*** − 0.1519 −0.0016* IYM 0.0280*** −0.501*** − 0.0016** IYT 0.0172*** 0.0151 − 0.0009 IYZ 0.0085*** −0.0183 0.0002 OIH 0.0117*** 0.449*** 0.0037*** RKH 0.0132*** 0.526*** 0.0036*** RTH 0.0067*** 0.176*** 0.0026*** SMH 0.0029*** 0.058*** 0.0006*** UTH 0.0113*** 0.672*** 0.0030*** VAW 0.0284*** 4.7962 −0.0072* VCR 0.0336*** − 0.8617 −0.0015 VDE 0.0316*** 0.702 − 0.0035** VFH 0.0202*** 1.1335 0.0000 VPU 0.0313*** 4.0424 −0.0055* XLB 0.0035*** 0.022*** 0.0009*** XLE 0.0048*** 0.089*** 0.0012*** XLK 0.0034*** 0.037*** 0.0007*** XLU 0.0046*** 0.084*** 0.0008*** XLV 0.0052*** 0.0204* 0.0002 XLY 0.0052*** 0.042*** 0.0000 Ave. 0.0168 0.4060 0.0002 StdDev 0.0124 1.0765 0.0030 Med 0.0132 0.0059 0.0002 Sign b0.0001 0.0026 0.3240 Signed Rank 0.0571 0.8073 b0.0001

The model estimated is ΔPt = c0ΔQ t + c1Δ(Q tVt) + z0Q t + z1Q tVt + Ut, where Pt is the security price at time t, Q t is the transaction indicator variable (+1 for a buyer initiated transaction −1 for a seller initiated trade). Vt is the trade size. c0 is the inventory cost unassociated with trade size. c1 is the inventory cost related to trade size. zo is the adverse selection cost unrelated to trade size while z1 reflects adverse selection costs related to trade size, the c1 and z1 coefficients are multiplied by 1000000. Implied spread is [2(c0 + c1V) + 2(z0 + z1V)]. Implied is the magnitude of the spread implied by the model estimates, percentage ASC is the estimated percentage of the spread due to adverse selection costs. Dollar adverse selection cost is percent adverse selection cost times average quoted spread. An ***indicates that a coefficient is significant at a 1% level and ** at a 5% level and * at a 10% level.

73

74

P. Chelley-Steeley, K. Park / International Review of Financial Analysis 19 (2010) 65–76

Table 4 Model estimates using MRR. ETFs

Inventory

Adverse selection

Implied spread

Percent adverse selection

(φ)

(θ)

2(φ + θ)

2θ/2(θ + φ)

Controls

Panel (a) Broad-market ETFs and their control securities DIA 0.0052*** 0.0091*** 0.0286 DSG 0.0068 0.0109 0.0354 DSV 0.0133 0.0083 0.0433 ELV 0.0223* 0.0018 0.0481 IJH 0.0166*** 0.0035*** 0.0402 IJJ 0.0190*** 0.0025*** 0.0431 IJK 0.0203*** 0.0021 0.0447 IJR 0.0184*** 0.0025*** 0.0417 IJS 0.0212*** 0.0006 0.0436 IJT 0.0287*** − 0.0050** 0.0472 IVE 0.0246*** −0.0022*** 0.0448 IVV 0.0207*** 0.0007* 0.0428 IVW 0.0205*** 0.0008 0.0426 IWB 0.0197*** 0.0011 0.0417 IWD 0.0195*** 0.0009** 0.0408 IWF 0.0187*** 0.0009*** 0.0392 IWM 0.0155*** 0.0010** 0.0331 IWN 0.0141*** 0.0022*** 0.0326 IWO 0.0118*** 0.0038*** 0.0311 IWP 0.0126*** 0.0034* 0.0320 IWR 0.0133*** 0.0034*** 0.0334 IWV 0.0146*** 0.0032** 0.0355 JKH 0.0148 0.0090 0.0477 JKJ 0.0218 0.0029 0.0496 JKK 0.0324 − 0.0045 0.0559 JKL 0.0293* 0.0015 0.0615 MDY 0.0212*** 0.0052*** 0.0527 PEY 0.0168*** 0.0080*** 0.0496 PWC 0.0167*** 0.0072*** 0.0479 PWO 0.0172*** 0.0058*** 0.0462 QQQQ 0.0169*** 0.0049*** 0.0437 VB 0.0128** 0.0097 0.0449 VBK 0.0140*** 0.0088 0.0456 VBR 0.0134*** 0.0096** 0.0460 VO 0.0217*** 0.0003 0.0441 VTI 0.0199*** 0.0045** 0.0488 VTV 0.0201*** 0.0033 0.0468 VV 0.0192*** 0.0047 0.0478 VXF 0.0190*** 0.0064 0.0509 Ave. 0.0181 0.0037 0.0435 StdDev 0.0054 0.0038 0.0070 Med 0.0187 0.0033 0.0441 Sign b0.0001 b0.0001 b0.0001 0.1450 SignedRank b0.0001 b0.0001

0.6345 0.6154 0.3857 0.0749 0.1721 0.1169 0.0936 0.1199 0.0295 −0.2130 −0.0992 0.0338 0.0382 0.0536 0.0459 0.0476 0.0615 0.1362 0.2415 0.2135 0.2014 0.1785 0.3781 0.1187 −0.1600 0.0473 0.1973 0.3229 0.3026 0.2534 0.2253 0.4299 0.3852 0.4175 0.0148 0.1835 0.1411 0.1965 0.2531 0.1767 0.1813 0.1721 b0.0001 b0.0001

MO UTL Y STU HB CYN CBSH NOC MRBK ESE MCY CB UB TMK BCR LH GE CMA ABC IDXX WPS EEP COKE SFSW BF BARI LEH BFIN KMR CFFN INTC TRH YANB WSFS GBL PDX ERIE IBOC BANF Ave. StdDev Med Sign

Panel (b) Industry-wide ETFs and their control securities BDH 0.0049*** 0.0158*** BHH 0.0051*** 0.0120*** HHH 0.0184*** 0.0032*** IAH 0.0157*** 0.0036* IBB 0.0144*** 0.0044*** IDU 0.0154*** 0.0048*** IGE 0.0237*** 0.0043** IGM 0.0244*** 0.0026 IGN 0.0233*** 0.0016** IGV 0.0197*** 0.0029*** IGW 0.0183*** 0.0044*** IIH 0.0167*** 0.0037*** IYC 0.0182*** 0.0010 IYE 0.0165*** 0.0050*** IYG 0.0145* 0.0078 IYH 0.0183*** 0.0044*** IYJ 0.0239*** − 0.0001 IYK 0.0262*** − 0.0017 IYM 0.0280*** −0.0016 IYT 0.0245*** 0.0006 IYZ 0.0203*** 0.0010*** OIH 0.0167*** 0.0088*** RKH 0.0155*** 0.0071*** RTH 0.0131*** 0.0082*** SMH 0.0124*** 0.0081*** UTH 0.0097*** 0.0101***

0.7653 0.7006 0.1475 0.1854 0.2340 0.2362 0.1532 0.0950 0.0646 0.1285 0.1953 0.1828 0.0538 0.2314 0.3502 0.1944 − 0.0028 − 0.0692 − 0.0623 0.0228 0.0452 0.3442 0.3144 0.3851 0.3952 0.5081

ACTL QVDX CERN PQE IVGN UIL PVA LIFE MRCY IMN ATK AMSWA BL MIDD BOKF UTR VTRU ROG CW FLA WFSL IBM ZION TGT DELL EQT

0.0413 0.0342 0.0431 0.0386 0.0375 0.0402 0.0560 0.0540 0.0499 0.0451 0.0454 0.0408 0.0385 0.0430 0.0447 0.0454 0.0477 0.0491 0.0527 0.0501 0.0425 0.0511 0.0451 0.0427 0.0411 0.0396

Inventory

Adverse selection

Implied spread

Percent adverse selection

(φ)

(θ)

2(φ + θ)

2θ/2(θ + φ)

0.0047*** 0.0054 0.0274 0.0264 0.0126*** 0.0016 0.0028*** 0.0049*** 0.0049*** 0.0040 0.0041 0.0039*** 0.0085*** 0.0135*** −0.0050*** 0.0051*** 0.0133*** 0.0069*** 0.0011*** 0.0063*** 0.0126*** 0.0019 0.0037 0.0137** −0.0031 −0.0096** 0.0053*** − 0.0007 0.0060*** 0.0023*** 0.0069*** 0.0126 0.0213*** 0.0101** 0.0125 0.0059 0.0028*** 0.0098*** 0.0008 0.0069 0.0074 0.0053 b0.0001

0.0169*** 0.0195*** 0.0303 − 0.0223 0.0137*** 0.0152*** 0.0128*** 0.0112*** 0.0160*** 0.0104*** 0.0107** 0.0119*** 0.0054*** −0.0044*** 0.0168*** 0.0087*** 0.0149*** 0.0074*** 0.0079*** 0.0159*** 0.0156*** 0.0140*** 0.0334*** 0.0045 0.0175*** 0.0229*** 0.0094*** 0.0133*** 0.0076*** 0.0162*** 0.0069*** 0.0018 0.0274*** 0.0284*** 0.0178 0.0134* 0.0119*** 0.0108*** 0.0192*** 0.0131 0.0095 0.0134 b 0.0001

0.0432 0.0497 0.1153 0.0081 0.0528 0.0337 0.0313 0.0323 0.0417 0.0289 0.0296 0.0316 0.0278 0.0182 0.0237 0.0276 0.0565 0.0286 0.0180 0.0444 0.0564 0.0318 0.0742 0.0364 0.0289 0.0265 0.0292 0.0251 0.0271 0.0370 0.0276 0.0288 0.0973 0.0771 0.0606 0.0387 0.0294 0.0413 0.0399 0.0399 0.0213 0.0318 b0.0001

0.7823 0.7836 0.5251 − 5.4898 0.5209 0.9034 0.8193 0.6948 0.7656 0.7225 0.7250 0.7546 0.3854 − 0.4878 1.4174 0.6333 0.5285 0.5190 0.8782 0.7171 0.5529 0.8817 0.9005 0.2473 1.2123 1.7228 0.6406 1.0555 0.5582 0.8766 0.5000 0.1276 0.5625 0.7373 0.5876 0.6926 0.8069 0.5234 0.9615 0.5448 1.0505 0.7171 b0.0001

0.0017*** 0.0051*** 0.0014*** 0.0066*** 0.0054*** 0.0051 0.0053 0.0047*** 0.0047*** 0.0060*** 0.0024 0.0022*** 0.0019 0.0010 0.0024** 0.0178*** 0.0139*** 0.0024 0.0039 0.0046 0.0138*** 0.0118*** 0.0163*** 0.0162*** 0.0021*** 0.0031***

0.0069*** 0.0091*** 0.0187*** 0.0083*** 0.0086*** 0.0107 0.0132** 0.0127*** 0.0131*** 0.0139*** 0.0108*** 0.0077*** 0.0140*** 0.0241*** 0.0141*** 0.0184*** 0.0186*** 0.0172** 0.0162** 0.0124** 0.0185*** 0.0100*** 0.0216*** −0.0074*** 0.0139*** 0.0111***

0.0173 0.0283 0.0401 0.0298 0.0278 0.0317 0.0368 0.0349 0.0355 0.0397 0.0265 0.0199 0.0318 0.0501 0.0331 0.0724 0.0649 0.0392 0.0402 0.0341 0.0645 0.0434 0.0756 0.0177 0.0319 0.0285

0.8033 0.6433 0.9325 0.5576 0.6150 0.6750 0.7144 0.7279 0.7365 0.6997 0.8184 0.7790 0.8777 0.9609 0.8537 0.5071 0.5729 0.8777 0.8041 0.7291 0.5720 0.4590 0.5701 − 0.8395 0.8706 0.7803

P. Chelley-Steeley, K. Park / International Review of Financial Analysis 19 (2010) 65–76

75

Table 4 (continued) ETFs

Inventory

Adverse selection

Implied spread

Percent adverse selection

(φ)

(θ)

2(φ + θ)

2θ/2(θ + φ)

Panel (b) Industry-wide ETFs and their control securities VAW 0.0138** 0.0063 0.0402 VCR 0.0156** 0.0083 0.0478 VDE 0.0190*** 0.0087*** 0.0553 VFH 0.0189*** 0.0076** 0.0530 VPU 0.0178*** 0.0064 0.0485 XLB 0.0141*** 0.0046*** 0.0374 XLE 0.0138*** 0.0047*** 0.0369 XLK 0.0126*** 0.0043*** 0.0337 XLU 0.0103*** 0.0042*** 0.0290 XLV 0.0094*** 0.0036*** 0.0260 XLY 0.0086*** 0.0032*** 0.0236 Ave. 0.0165 0.0050 0.0430 StdDev 0.0055 0.0036 0.0077 Med 0.0165 0.0044 0.0430 Sign b0.0001 b0.0001 b0.0001 SignedRank b0.0001 b0.0001 0.1091

0.3127 0.3489 0.3135 0.2852 0.2637 0.2465 0.2523 0.2536 0.2891 0.2784 0.2701 0.2409 0.1764 0.2465 b0.0001 b0.0001

Panel (c ) Full sample Ave. 0.0173 StdDev 0.0055 Med 0.0171 Sign b0.0001 Signedrank b0.0001

0.2079 0.1806 0.1969 b0.0001 b0.0001

0.0043 0.0037 0.0040 b0.0001 b0.0001

0.0432 0.0073 0.0437 b0.0001 0.0342

Controls

SCL FSCI TNC COBH CHG MAS XOM CVC SO PFG CTL Ave. StdDev Med Sign

Inventory

Adverse selection

Implied spread

Percent adverse selection

(φ)

(θ)

2(φ + θ)

2θ/2(θ + φ)

0.0079 0.0220*** − 0.0038 0.0107 0.0089 0.0089*** 0.0187*** 0.0135*** 0.0036*** 0.0101*** 0.0184*** 0.0123 0.0064 0.0127 b 0.0001

0.0419 0.0703 0.0207 0.0387 0.0344 0.0285 0.0599 0.0419 0.0267 0.0330 0.0516 0.0390 0.0152 0.0349 b0.0001

0.3788 0.6269 −0.3645 0.5528 0.5196 0.6267 0.6234 0.6446 0.2712 0.6141 0.7144 0.6083 0.3344 0.6446 b0.0001

0.0395 0.0184 0.0339 b0.0001

0.5757 0.7835 0.6937 b0.0001

0.0130 0.0131* 0.0142* 0.0087 0.0083 0.0053*** 0.0113*** 0.0075*** 0.0097*** 0.0064*** 0.0074*** 0.0072 0.0049 0.0054 b0.0001

0.0070 0.0063 0.0053 b0.0001

0.0127 0.0081 0.0131 b 0.0001

The ϕ is the dollar inventory cost, and θ is the dollar adverse selection cost. The estimated implied spread is calculated as S = 2(θ + ϕ). The percentage adverse selection component of the spread is 2θ/2(θ + ϕ). Dollar adverse selection cost is computed by the average quoted spread times estimated percent adverse selection costs. Dollar adverse selection cost is percent adverse selection cost times average quoted spread. An *** indicates that a coefficient is significant at a 1% level and ** at a 5% level and * at a 10% level.

associated with ETF trading. We suggest that one factor that has contributed to the popularity of ETFs is that they offer lower levels of adverse selection than investments in individual securities. We show that an ETF has the characteristics of a basket security proposed by Subrahmanyam (1991) and an important prediction of a basket security is that information asymmetry about the underlying securities can be diversified away. This causes the basket to have lower adverse selection costs than the weighted averages of the component securities. To explore this hypothesis we apply the Glosten–Harris and MRR spread decomposition models to a sample of ETFs and appropriately selected control securities. The adverse selection costs extracted from both of these models indicate that ETFs have significantly lower adverse selection costs than the control securities. Our findings have a number of important implications for market participants. Because ETFs offer lower information asymmetry than individual securities, uninformed traders should experience lower adverse selection losses in these markets than in the market for individual securities. This benefit is reflected in the adverse selection component of the spread which ensures that trading in ETFs is cheaper than for the underlying securities. These effects are likely to be reinforced as liquidity traders migrate to the lower cost market. Although the study of intra-day patterns is beyond the scope of this paper an interesting issue for future work would be to examine the intra-day pattern of adverse selection costs for both ETFs and comparable control securities. Prior work by McInish and Van Ness (2002) has shown that for individual securities the adverse selection component of the spread is most elevated at the open and then tends to decline throughout the trading day mirroring the same patterns as those discovered for the time-weighted bid-ask spread. In future work it would be interesting to examine whether ETFs display the same intraday patterns as individual securities. References Ackert, L., & Tian, Y. (2001). Arbitrage and valuation in the market for Standard and Poor's Depository Receipts. Financial Management, 20, 71−88.

Admati, A. R., & Pfleiderer, P. (1988). A theory of intraday patterns: Volume and price variability. Review of Financial Studies, 1, 3−40. Ahn, H. -J., Hamao, Y., & Ho, R. (2002). The components of the bid-ask spread in a limitorder market: Evidence from the Tokyo Stock Exchange. Journal of Empirical Finance, 8, 125−142. Boehmer, B., & Boehmer, E. (2003). Trading your neighbor's ETFs: Competition or fragmentation. Journal of Banking & Finance, 27, 1667−1703. Chiyachantana, Chiraphol, N., Christine, X., Taechapiroontong, J., & Wood, R. A. (2004). The impact of regulation fair disclosure on information asymmetry and trading: An intraday analysis. The Financial Review, 39, 549−577. Chung, K. H., Chuwonganant, C., & McCormick, D. T. (2004). Order preferencing and market quality on NASDAQ before and after decimalization. Journal of Financial Economics, 71, 581−612. Copeland, T., & Gali, D. (1983). Information effects and the bid-ask spread. Journal of Finance, 38, 1457−1469. Easley, D., & O'Hara, M. (1987). Price, trade size, and information in securities markets. Journal of Financial Economics, 19, 69−90. Elton, E. J., Gruber, M. J., Comer, G., & Li, K. (2002). Spiders: Where are the bugs? Journal of Business, 75, 453−472. Engle, R. E., & Sarkar, D. (2006). Premiums-discounts and exchange traded funds. Journal of Derivatives, 8, 1−13. Glosten, L. R., & Harris, L. E. (1988). Estimating the components of the bid-ask spread. Journal of Financial Economics, 21, 123−142. Gorton, G. B., & Pennacchi, G. G. (1993). Security baskets and index-linked securities. The Journal of Business, 66, 1−27. Harper, J. T., Madura, J., & Schnusenberg, O. (2006). Performance comparison between exchange-traded funds and closed-end country funds. Journal of International Financial Markets, Institutions and Money, 16, 104−122. Hatch, B. C., & Johnson, S. A. (2002). The impact of specialist firm acquisitions on market quality. Journal of Financial Economics, 66, 139−167. Hegde, S. P., & McDermott, J. B. (2004). Firm characteristics as cross-sectional determinants of adverse selection. Journal of Business Finance and Accounting, 31, 1097−1124. Hegde, S. P., & McDermott, J. B. (2004). The market liquidity of diamonds, Q's and their underlying stocks. Journal of Banking and Finance, 28, 1043−1067. Huang, R. D., & Stoll, H. R. (1996). Dealer versus auction markets: A paired comparison of execution costs on NASDAQ and the NYSE. Journal of Financial Economics, 41, 313−357. Jiang, C. X., & Kim, J. -C. (2005). Trading costs of non-US stocks on the New York Stock Exchange: The effect of institutional ownership, analyst following, and market regulation. Journal of Financial Reseach, 28, 439−459. Jiang, C. X., Kim, J. -C., & Wood, R. A. (2002). The change in trading activity on volatility and adverse selection component: Evidence from ADR splits. Journal of Multinational Financial Management, 12, 323−345. Kyle, A. S. (1984). Market structure, information, futures markets, and price formation. International agricultural trade: advanced readings in price formation, market structure and price instability (pp. 45−64).. Kyle, A. S. (1985). Continuous auctions and insider trading. Econometrica, 53, 1315−1336.

76

P. Chelley-Steeley, K. Park / International Review of Financial Analysis 19 (2010) 65–76

Lee, C. M. C., & Ready, M. J. (1991). Inferring trade direction from intraday data. The Journal of Finance, 46, 733−746. Madhavan, S. A. (2000). Market microstructure: A survey. Journal of Financial Markets, 3, 205−258. Madhavan, A., Richardson, M., & Roomans, M. (1997). Why do security prices changes? A transaction-level analysis of NYSE stocks. Review of Financial Studies, 10, 1035−1064. McInish, T. H., & Van Ness, B. F. (2002). An intraday examination of the components of the bid-ask spread. The Financial Review, 37, 507−524.

Neal, R., & Wheatley, S. M. (1998). Adverse selection and bid-ask spreads: Evidence from closed-end funds. Journal of Financial Markets, 1, 121−149. Pennathur, A. K., Delcoure, N., & Anderson, D. (2002). Diversification benefits of iShares and closed-end country funds. Journal of Financial Research, 25, 541−557. Sarin, Atulya, Shastri, K.A., Shastri, K., 2000, Ownership structure and stock market liquidity, Working Paper, University of Pittsburgh. Stoll, H. (2000). Friction. Journal of Finance, 55, 1479−1514. Subrahmanyam, A. (1991). A theory of trading in stock index futures. Review of Financial Studies, 4, 17−51.