ASSET LIQUIDITY, STOCK LIQUIDITY, AND OWNERSHIP CONCENTRATION: EVIDENCE FROM THE ASE

ASSET LIQUIDITY, STOCK LIQUIDITY, AND OWNERSHIP CONCENTRATION: EVIDENCE FROM THE ASE Ghada Tayem*, Mohammad Tayeh**, Adel Bno** * Correspondng author: Department of Fnance, School of Busness, The Unversty of Jordan, Amman 11942, Jordan. Tel. +96265355000. Emal address: g.tayem@ju.edu.jo ** Department of Fnance, School of Busness, The Unversty of Jordan, Amman 11942, Jordan. Abstract Ths paper examnes how ownershp concentraton nfluences the relaton between stock lqudty and asset lqudty. Lqud assets reduce uncertanty of assets n place and hence mprove stock lqudty. However, lqud assets are less costly to turn nto prvate benefts compared to other assets. Therefore, lqud assets may result n ncreasng the uncertanty of assets n place rather than reducng t. In ths paper we examne the mpact of asset lqudty on stock lqudty condtonal on a company s ownershp structure usng the context of Jordan. Jordanan companes lsted n the ASE are mostly characterzed by hghly concentrated ownershp. In the absence of nvestor protecton, concentrated ownershp allows shareholders wth large ownershp stakes to exercse control over the frm and hence may result n ncreasng the uncertanty of assets n place. The uncertanty regardng the usage of lqud assets n cashrch frms leads to greater uncertanty regardng the frm s cash flows and hence lower stock lqudty. The fndngs of ths study show evdence that as ownershp concentraton ncreases asset lqudty becomes negatvely related to stock lqudty. Keywords: Stock Lqudty; Asset Lqudty; Ownershp Concentraton; Shareholders; Jordan JEL classfcaton: G14; G31; G32 Acknowledgment The authors kndly acknowledge the generous fnancal support receved from the Deanshp of Academc Research (DAR)/The Unversty of Jordan. The DAR/The Unversty of Jordan had no nvolvement n the conduct of ths research or the preparaton of ths artcle. The authors would lke to thank partcpants at the 34 th Internatonal Busness Research Conference 2016 held at the Imperal College of London for provdng useful comments. All remanng errors are ours 1. INTRODUCTION Ths paper examnes the relaton between asset lqudty and stock lqudty usng a sample of frms lsted on the Amman Stock Exchange (ASE). Specfcally, t examnes f the lqudty of a frm s assets carres to the lqudty of the fnancal clams on those assets. Lqud assets reduce the uncertanty of assets n place and hence mprove stock lqudty (Gopalan et al., 2012). However, the extant lterature assumes that the nterests of the frm s agents are algned and therefore a frm determnes ts lqud assets such that the value of the frm, through mprovements n ts stock lqudty, s maxmzed (Gopalan et al., 2012). In ths study, we relax ths assumpton by lookng at how ownershp concentraton affects the relatonshp between asset and stock lqudty. Specfcally, we argue that excess lqud assets ncrease the scope of large shareholders dscreton and hence may result n ncreasng the uncertanty of assets n place rather than reducng t. Therefore, we expect that the senstvty of stock lqudty to asset lqudty s negatve for companes wth concentrated ownershp. To the best of our knowledge ths s the frst study to examne whether and how ownershp concentraton nfluences the relatonshp between asset and stock lqudty. The lterature on the mpact of a frm s nvestment choces on stock lqudty s only recent wth a small number of papers examnng ths ssue (Gopalan et al., 2012; Chen et al., 2013; and Charoenwong et al., 2014). Gopalan et al. (2012) formalze a theoretcal model that shows how manageral nvestment decsons can affect stock lqudty by convertng lqud assets nto llqud ones. The authors hypothesze that more cash lowers valuaton uncertanty assocated wth assets n place, and therefore more cash mproves stock lqudty. Ths s consstent wth the argument that lqud assets, such as cash and ts equvalents, are subject to less nformaton asymmetry and hence are easer to value than other assets such as fxed assets and growth optons (Kothar et al., 2002 and Aboody and Lev, 2000). Therefore, frms wth hgher level of asset lqudty are expected to have lower valuaton uncertanty and hence hgher stock lqudty. Gopalan et al. (2012) fnd that for a panel data of all 48

Compustat frms durng the tme perod 19622005 and after controllng for determnants of stock lqudty, there s a postve and sgnfcant relatonshp between the alternatve measures of asset lqudty and those of stock lqudty. Charoenwong et al. (2014) report nternatonal evdence n 47 countres that supports the fndng n Gopalan et al. (2012) of a postve mpact of asset lqudty on stock lqudty. In addton, Chen et al. (2013) use the methodology n Faulkender and Wang (2006) to study the varaton of the value of corporate lqud assets wth that n stock lqudty. They fnd that excess returns are postvely related to cash holdngs and that the value of lqud assets ncreases for llqud frms. However, ths study proposes that asset lqudty affects stock lqudty negatvely when a frm s ownershp structure s taken nto account. Free cash flows ncrease managers, and by extenson largecontrollng shareholders, power due to the exstence of more resources under ther control (Jensen, 1986). In addton, t s less costly to turn lqud assets nto prvate benefts compared to other assets (Myers and Rajan, 1998). Therefore, and n the absence of nvestor protecton, large shareholders have ncentves to approprate cash holdngs. The greater uncertanty over the usage and redeployment of cash and lqud assets entals greater uncertanty over the frm s future cash flows (Charoenwong et al., 2014). Traders antcpate ths uncertanty of cashrch frms controlled by large shareholders and therefore trade ther stocks at a premum. The drecton of the relaton between asset lqudty and stock lqudty, therefore, can be resolved emprcally. In ths paper, we examne ths relaton emprcally usng a sample of Jordanan nonfnancal frms lsted n the ASE durng the perod 20012012. The dependent varable n our analyss s stock (ll) lqudty. To test our proposton, we employ dfferent alternatve measures of stock llqudty and another measure of stock lqudty. The measures of stock llqudty are: the mplct bdask spread proposed by Roll (1984) as estmated by Hasbrouck (2009); the proporton of zero tradng days proposed by Lesmond et al. (1999); and the llqudty measure proposed by Amhud (2002). Moreover, we employ the turnover rato whch s a measure of stock lqudty (Brennan et al., 1998; Datar et al., 1998; Chorda et al., 2001; and Avramov and Chorda, 2006). We follow Gopalan et al. s (2012) methodology to construct the asset lqudty measures and we modfy these measures to take nto account shortterm debt. Fnally, we employ a set of control varables based on the emprcal work on the determnants of the lqudty of ndvdual assets (Demsetz, 1968; Tnc, 1972; Branch and Freed, 1977; Stoll, 1978; and Easley et al., 1987). The man ndependent varable of nterest n ths paper s asset lqudty. The measurement of asset lqudty for the purposes of ths study s emprcal analyss follows closely the approach dscussed n Gopalan et al. (2012). In order to construct the asset lqudty measures we rank a frm s assets based on ther degree of lqudty and assgn to each asset class a lqudty score between zero and one. Then, for each frm we compute a weghted average of the lqudty scores across the dfferent asset classes. The weghts are based on the proporton of each asset class scaled by the lagged value of total assets. As we wll explan n detals n the methodology secton, we defne three alternatve measures of asset lqudty by varyng the lqudty scores assgned to each asset class n the ntal step. In addton, we propose another measure of asset lqudty based on the dea that nvestors take nto account net cash poston, cash mnus shortterm labltes, when they assgn a value to the frm s stock. In our analyss, we control for varables that have been documented to affect stock lqudty n the lterature. We nclude frm fxed effects to control for unobservable frm characterstcs that affect stock lqudty. To test f our results are robust to controllng for endogenety, we employ System Generalzed Method of Moments (SystemGMM) estmator proposed by Arellano and Bond (1991). Our ntal fndngs show nconsstent evdence of a postve relaton between asset lqudty and stock lqudty n the ASE. However, we obtan more consstent results when we ntroduce the nteracton term between asset lqudty measures wth the ownershp concentraton measure. The results show, n about half of our specfcatons, that asset lqudty measures are negatvely and sgnfcantly related to llqudty measures and postvely and sgnfcantly related to the lqudty measure. These results ndcate that asset lqudty s postvely related to stock lqudty. The results also show that the nteracton term s postvely related to stock llqudty measures and negatvely related to the stock lqudty measure ndcatng that lqud assets n companes wth (wthout) large shareholders reduce (enhance) stock lqudty. Ths result ndcates that nvestors beleve that excess cash n companes wth large shareholders ncrease the scope of large shareholders dscreton whch leads to greater uncertanty about future assets and hence lower stock lqudty. Therefore the senstvty of stock lqudty to asset lqudty s negatve for companes wth hgh concentrated ownershp. Stock lqudty s an mportant feld of study as lqudty s n tself a reducton n the cost of tradng and an ndcator of the degree of stock market development (DemrgüçKunt and Levne, 1996). In addton, the extant evdence shows that an ncrease n stock lqudty ncreases frm value by reducng ts cost of equty (Amhud and Mendelson, 1986). However, there s lttle evdence on the mpact of corporate nvestment decsons on the lqudty of stocks and vrtually no evdence from the ASE. Ths research ams to fll ths gap by studyng whether and how the composton of frm assets of companes lsted on the ASE nfluence ther stock lqudty. Therefore, ths study extends the US evdence presented n Gopalan et al. (2012) and the nternatonal evdence presented n Charoenwong et al. (2014). More mportantly, ths study contrbutes to the extant lterature by provdng the frst evdence on the nfluence of ownershp structure on the relatonshp between stock and asset lqudty. We fnd that n frms wth large shareholders the senstvty of stock lqudty to asset lqudty s negatve. Overall, our fndngs ndcate that ownershp structure s an mportant determnant of the assetstock lqudty relaton. 49

The rest of the paper s organzed as follows. The next secton presents the lterature related to the measurement of stock lqudty and asset lqudty. Secton 3 presents the research model and data s descrbed n Secton 4. Results and analyss are dscussed n Secton 5 and the concluson s presented n Secton 6. 2. VARIABLE MEASUREMENT 2.1. Lqudty Measurement and Determnants The lterature suggests several varables that capture the stock lqudty. These varables are explaned next. 2.1.1. BdAsk The bdask spread s the most popular measure of lqudty and s wdely used to measure lqudty n the market mcrostructure lterature (e.g. Amhud and Mendelson, 1986; Chorda et al., 2000; and Venkataraman, 2001 among others). Moreover, the bdask spread reflects three cost components: order processng costs, nventory costs, and nformaton asymmetry costs. However, t s deemed a nosy measure, because large trades have a tendency to happen outsde the spread and small trades have a tendency to happen nsde the spread (Brennan and Subrahmanyam, 1996). Accordng to the data avalablty, we calculate ths measure and construct ndvdual frm spread usng daly data. Ths measure s computed n two stages. Frst we calculated a frmspecfc quoted bdask spread and a proportonal quoted spread, whch s the quoted bdask spread dvded by the mdpont of the quote for stock n day t as follows: j qspr t = ask t bd t (1) pqspr t = (ask t bd t ) ((ask t + bd t ) 2) where, ask t s the ask prce for stock at day t, bd t s the bd prce for stock at day t. Then, the average ndvdual stock s quoted spread and proportonal quoted spread s computed each year to construct a yearly lqudty seres. The yearly lqudty seres of quoted spread and proportonal quoted spread s computed as follows: QSPR t = ( 1 N ) (ask t bd t ) j j PQSPR t = ( 1 N ) (ask t bd t ) ((ask t ) + bd t 2) where, N s the number of tradng days n a gven year of stock. 2.1.2. Zero Proporton of Tradng Days Lesmond et al. (1999) suggest a stock llqudty measure derved from daly stock returns. Stock llqudty measure called the Zero Proporton, s the proporton of tradng days wth zero returns for stock durng a year to the total tradng days n a gven year: (2) (3) (4) Zero = Tradng Days wth Zero Returns Total Tradng Days 2.1.3. Prce Impact The prce mpact (known as Kyle s lambda) s utlzed as a proxy for lqudty n order to capture the depth dmenson of lqudty whch s the mean of the market s ablty to absorb and execute large orders wth a low prce mpact. We measure the prce mpact through llqudty rato, whch s defned as the rato of daly absolute stock returns over the tradng value as proposed by Amhud s (2002). It can be nterpreted as the daly prce response assocated wth one dollar of tradng volume, whch s the opposte of the lqudty rato that s used n the market mcrostructure lterature (such as Cooper et al., 1985; Berkman and Eleswarapu, 1998). The man feature of ths measure over other dfferent measures of lqudty s that t requres just daly data to be computed and can be utlzed to construct a seres that could span a long tme perod. Ths measure s frst calculated for each stock n the sample, that s, the prce mpact for stock at day t s gven as follows: pmpact t = R t TValue t where, R t s the return for stock at day t and TValue t s the tradng value for stock at day t. Then, the average of the ndvdual stocks prce mpact s computed each day to construct a yearly lqudty seres as follows: PIMPACT t = ( 1 N ) pmpact t where, N s the number of tradng days n a gven year of stock. 2.1.4. Tradng Actvty Tradng actvty measures are wdely accepted among researchers (see Brennan et al., 1998; Datar et al., 1998; Chorda et al., 2001; Avramov and Chorda, 2006 among others) because they are hghly assocated wth the bdask spread and other measures of lqudty. We defne the turnover rato as the product of the dvson between the tradng value and the market captalzaton. Usng daly data on ths measure we construct an ndvdual frm turnover rato by computng the average ndvdual stocks turnover rato as follows: TOV t = ( 1 N ) TValue t t j where, TValue t s the tradng value for stock at day t, t s the market captalzaton for stock at day t, and N s the number of tradng days of stock. j (5) (6) (7) (8) 50

2.2. Asset Lqudty Measurement The major ndependent varable n our study s asset lqudty. We follow Gopalan et al. s (2012) methodology to construct asset lqudty measures. For a gven frm, we rank ts asset classes based on ther degree of lqudty and assgn a lqudty score between zero and one to each of them. Second, we calculate a weghted average of the lqudty scores across the dfferent asset classes for each frm. The weghts are based on the proporton of each asset class scaled by the lagged value of total assets. Dependng on the lqudty scores assgned to each asset class n the frst step, ths methodology yelds three alternatve measures of weghted asset lqudty (WAL) score for each frm, explaned next. 2.2.1. WAL1 The WAL1 measure s crude and assumes that assets other than cash have no lqudty. We then calculate WAL1 as follows: WAL1 t = Cash & Equvalents t 1 + Other Assets t 0 In addton, we modfy WAL1,t nto two ways as follows: WAL1A t = Cash & Equvalents t Bank Debt t Other Assets t 1 + 0 WAL1B t = Cash & Equvalents t Short Term Debt t Other Assets t 1 + 0 (9) (10) (11) where, bank debt and loans refer to short term maturty bank debt and short term maturty loans. Ths modfcaton takes nto account the practce among Jordanan frms to borrow n the short run as a mean of cash management. Frms subject to sudden cash shortages borrow from banks usng credt lnes or delay payment to ther supplers. Ths practce n essence turns shortterm debt nto negatve cash. 2.2.1. WAL2 We assgn a lqudty score of one to cash and cash equvalents and 0.5 to noncash current assets because noncash current assets are the second most lqud assets after cash. All other assets are assgned a score of zero. We calculate WAL2 as follows: WAL2 t = Cash & Equvalents t 1 + Non Cash CA t 0.5 + Other Assets t 0 (12) 2.2.3. WAL3 The thrd weghted asset lqudty WAL3 measure looks further nto longlved assets. Longlved assets can be classfed nto tangble and nontangble assets. We assgn a lqudty score of one to cash and cash equvalents, 0.75 to noncash current assets, 0.5 to tangble fxed assets, and zero to nontangble assets. We then compute WAL3 as follows: WAL3 t = Cash & Equvalents t 1 + NonCash CA t 0.75 + Tangble Fxed Assets t 0.5 + Other Assets t 0 3. STUDY METHODOLOGY (13) The am of ths study s to examne the relaton between asset lqudty and stock lqudty usng a sample of Jordanan frms. In addton, we condton the relaton between asset lqudty and stock lqudty on the level of ownershp concentraton. To examne these relatons we emprcally test the followng equaton: (IL)LIQ jt = λwal kt + γ t + θ t 1 WAL kt + δ j Χ mt + ν + u t (14) m where, (IL)LIQ jt are our measures of llqudty/lqudty whch nclude: the quoted spread, proportonal spread, proporton of zero tradng days, prce mpact and fnally turnover rato; WAL kt are our measures of asset lqudty; t represent the sum of the percentage ownershp of the largest three shareholders ownng 5% and more; t WAL kt s our man varable of nterest that represents the nteracton between ownershp concentraton and the measures of asset lqudty; Χ mt s a vector of control varables that ncludes the frm s sze, rato, frm s proftablty and prce nverse. Followng Stoll (2000) and Charoenwong et al. (2014) we nclude the followng control varables. We nclude, defned as the log of total market captalzaton, to control for the frm's sze effect. We nclude market to book rato () to control for growth opportuntes. In addton, we nclude return on assets to control for the frm s operatng performance. We also nclude the nverse of the stock prce to control for the dscrete tck sze effect. The Operatonal defntons of the varables dscussed so far are presented n Table 1. Equaton 14 s estmated usng two alternatve models: fxed (wthn) effects and SystemGMM. The fxed effect (wthn) model deals wth unobservable frmspecfc effects ν, whch, change across frms but s fxed for a gven frm through tme (Wooldrdge, 2002). However, asset lqudty and stock lqudty are lkely to be endogenous as frms wth growth opportuntes may have hgh asset lqudty and stock lqudty (Gopalan et al., 2012). Falng to control for ths source of endogenety wll lead to based estmators. To deal wth ths ssue we employ the System Generalzed Method of Moments (GMM) estmator proposed by Arellano and Bond 51

(1991). Ths procedure uses lagged values to nstrument for asset lqudty and estmates the regresson usng the GMM procedure. Table 1. Summary of Varable Defntons Varables Quoted Proportonal Zero Prce Impact Turnover Rato WAL1 WAL1A WAL1B WAL2 WAL3 Sze Proftablty Prce Inverse Proxy The dfferences between ask prce and bd prce. The quoted bdask spread dvded by the mdpont of the quote. The proporton of tradng days wth zero returns to total tradng days n a gven year. The mpact of order flows on prces calculated as a rato of absolute return to tradng value. Turnover measure of tradng actvtes, whch s calculated by dvdng tradng value over the market captalzaton. A measure of asset lqudty that assgns a lqudty score of one to cash and cash equvalents multpled by a weght equal to the proporton of cash and cash equvalents scaled by the lagged value of total assets. A measure based on WAL1 but that deducts short term bank debt from cash. A measure based on WAL1 but that deducts short term debt from cash. A measure of asset lqudty that assgns a lqudty score of one to cash and cash equvalents and 0.5 to noncash current assets and zero score to all other assets. Each score s multpled by a weght computed as the proporton of each asset class scaled by the lagged value of total assets. A measure of asset lqudty that assgns a lqudty score of one to cash and cash equvalents, 0.75 to noncash current assets, 0.5 to tangble fxed assets, and zero to nontangble assets. Each score s multpled by a weght computed as the proporton of each asset class scaled by the lagged value of total assets. The percentage of shares held by the largest three owners who hold 5% or more of outstandng shares. The logarthm of total market captalzaton (). Market to book value rato () defned as book value of total assets mnus book value of equty plus market value of equty dvded by book value of assets. Earnngs before nterest and tax (EBIT) dvded by total assets. The nverse of the closng prce. 4. DATA DESCRIPTION Ths paper uses a sample of nonfnancal Jordanan companes that are publcly traded on the Amman Stock Exchange (ASE) over the perod 20022012. The data s collected from three sources. Data on stock tradng are obtaned from the ASE s Tradng Fles and data on fnancal tems are obtaned from the ASE s Company Gudes. Tradng Fles comple market and tradng related data and s publshed by the ASE at the end of each tradng day. The Company Gude comples fnancal data tems obtaned from fnancal statements of frms lsted n Table 2. Summary Statstcs the ASE and s publshed by the ASE at the end of each fscal year. Data on ownershp s collected manually from the Corporate Gudes for the perod 20022007 and from the frm s annual reports thereafter. It s mandated that lsted frms on the ASE dsclose the names of owners wth a stock holdng equal or above 5%, the numbers of declared shares and the correspondng percentage of ownershp for each owner. The tradng, fnancal and ownershp data are matched usng the frm s dentfer. The next table presents some descrptve statstcs of the key varables n the study. Table 2 reports descrptve statstcs for a sample of nonfnancal Jordanan frms lsted n the ASE over the perod 2002 2012. Tradng data s collected from the Tradng Fles ssued by the ASE. Fnancal data s collected from the Corporate Gudes ssued by the ASE. Ownershp data s collected from the Corporate Gudes for the perod 20022007 and from the fnancal statements of lsted companes thereafter. Varables Mean Medan SD Mn Max Skewness Kurtoss Quoted 0.365 0.181 0.467 0.017 3.370 2.847 13.826 Proportonal 0.248 0.141 0.267 0.013 1.295 1.542 4.813 Zero 0.403 0.382 0.215 0.080 0.843 0.267 1.930 Prce Impact 0.000088 0.000041 0.000142 0.0000013 0.001238 4.387 28.092 Turnover Rato 0.001 0.000 0.002 0.000 0.013 3.535 17.864 WAL1 0.062 0.028 0.096 0.000 0.587 3.035 13.691 WAL1A 0.010 0.008 0.145 0.543 0.587 0.297 7.174 WAL1B 0.100 0.087 0.196 0.837 0.509 0.270 4.393 WAL2 0.262 0.253 0.147 0.013 0.770 0.662 3.377 WAL3 0.575 0.581 0.165 0.067 1.099 0.424 3.776 53.566 53.00 18.475 7.9 98.38 0.106 2.684 Sze 16.419 16.331 1.266 12.972 22.011 0.435 4.426 1.367 1.230 0.558 0. 3.782 1.333 5.262 Proftablty 0.040 0.040 0.071 0.309 0.360 0.096 7.119 Prce Inverse 0.661 0.500 0.566 0.023 4.000 2.703 12.974 52

Table 3. Correlaton Matrx Table 3 shows the correlaton between the varables used n the study. The sample conssts of nonfnancal Jordanan frms lsted n the ASE over the perod 20012012. Varable defntons are presented n Table 1. a, b, and c ndcate sgnfcance at the 1%, 5%, and 10% respectvely. Q P Zero Prce Impact Turnover Rato WAL1 WAL1A WAL1B WAL2 WAL3 Sze Proftablty Prce Inverse Q 1 P 0.767 a 1 Zero 0.424 a 0.450 a 1 P. Impact 0.148 a 0.361 a 0.350 a 1 Turnover 0.12 b 0.028.338 a 0.16 a 1 WAL1 0.242 a 0.066 0.104 b 0.017 0.024 1 WAL1A 0.197 a 0.079 c 0.124 b 0.007 0.060 0.791 a 1 WAL1B 0.091 c 0.009 0.032 0.030 0.038 0.378 a 0.510 a 1 WAL2 0.184 a 0.109 b 0.007 0.007 0.026 0.650 a 0.366 a 0.073 1 WAL3 0.092 b 0.064 0.018 0.021 0.040 0.434 a 0.200 a 0.002 0.845 a 1 0.068 0.16 a 0.027 0.18 a 0.361 a 0.098 b 0.107 b 0.158 a 0.108 b 0.060 1 Sze 0.596 a 0.420 a 0.238 a 0.053 0.185 a 0.251 a 0.161 a 0.098 b 0.179 a 0.168 a 0.455 a 1 0.202 a 0.026 0.120 b 0.14 a 0.262 a 0.264 a 0.207 a 0.058 0.202 a 0.167 a 0.365 a 0.334 a 1 Proft. 0.30 a 0.14 a 0.20 a 0.062 0.273 a 0.18 a 0.120 b 0.055 0.19 a 0.14 a 0.364 a 0.38 a 0.48 a 1 P.Inverse 0.043 0.151 a 0.213 a 0.098 b 0.202 a 0.070 0.107 b 0.019 0.067 0.082 c 0.127 b 0.160 a 0.101 b 0.14 a 1 Table 4. Fxed Effects Model Table 4 reports estmaton results of the stock lqudty model usng frm fxed effects. The sample conssts of nonfnancal Jordanan frms lsted n the ASE over the perod 20012012. Varable defntons are presented n Table 1. tstatstcs are n parentheses. ***, **, * ndcate sgnfcance at the 1%, 5%, and 10% respectvely. Quoted Proportonal Zero Prce Impact Turnover Quoted Proportonal Zero Prce Impact Turnover WAL1 0.0423** 0.0538** 0.0116** 0.0961*** 0.0226 (2.04) (2.24) (2.47) (3.46) (0.80) WAL1A 0.116 0.386 0.0796 0.587** 0.153 (0.55) (1.59) (1.62) (2.00) (0.52) 0.00929** 0.00327 0.0057*** 0.0257*** 0.0396*** 0.00956** 0.00312 0.00557*** 0.0256*** 0.0388*** (2.17) (0.66) (5.81) (4.42) (6.70) (2.26) (0.64) (5.74) (4.41) (6.67) 0.234*** 0.0825 0.0402** 0.0595 0.394*** 0.211** 0.0727 0.0396** 0.0388 0.385*** (2.68) (0.82) (2.04) (0.51) (3.32) (2.43) (0.73) (2.02) (0.33) (3.27) 0.703*** 0.733*** 0.0816*** 0.25 0.236 0.746*** 0.747*** 0.0838*** 0.25 0.253 (5.94) (5.36) (3.04) (1.56) (1.46) (6.33) (5.53) (3.15) (1.58) (1.58) Proftablty 1.125* 2.454*** 0.123 0.672 0.131 1.** 2.815*** 0.104 1.179* 0.0798 (1.89) (3.57) (1.00) (0.92) (0.18) (2.59) (4.29) (0.89) (1.68) (0.11) Prce Inverse 0.479*** 0.223* 0.0284** 0.459*** 0.*** 0.426*** 0.210* 0.0277** 0.464*** 0.488*** (4.64) (1.87) (2.01) (5.50) (5.68) (4.28) (1.84) (1.99) (5.57) (5.83) Observatons R 2 0.3844 0.1695 0.1122 0.1484 0.2157 0.3713 0.1681 0.1031 0.1361 0.2132 WAL2 0.0217 (0.32) WAL3 0.00953** (2.25) 0.207** (2.39) 0.747*** (6.31) Proftablty 1.461** (2.54) Prce Inverse 0.433*** (4.30) Observatons R 2 0.3709 0.1 (1.30) 0.00298 (0.61) 0.086 (0.86) 0.759*** (5.58) 2.718*** (4.11) 0.239** (2.07) 0.1661 0.018 (1.18) 0.00559*** (5.75) 0.0402** (2.05) 0.0783*** (2.93) 0.124 (1.04) 0.0286** (2.03) 0.1002 0.251*** (2.78) 0.0255*** (4.42) 0.0366 (0.32) 0.311* (1.96) 0.857 (1.21) 0.445*** (5.35) 0.1443 Table 4. Contnued 0.00461 (0.05) 0.0389*** (6.67) 0.382*** (3.25) 0.247 (1.54) 0.0744 (0.10) 0.486*** (5.78) 0.2126 0.0343 (0.37) 0.00946** (2.23) 0.204** (2.33) 0.750*** (6.29) 1.468** (2.56) 0.434*** (4.31) 0.371 0.0585 (0.55) 0.00289 (0.59) 0.0882 (0.88) 0.751*** (5.47) 2.794*** (4.24) 0.226* (1.96) 0.1627 0.0342 (1.60) 0.00567*** (5.83) 0.0370* (1.88) 0.0746*** (2.77) 0.121 (1.02) 0.0293** (2.09) 0.1029 0.300** (2.36) 0.0247*** (4.27) 0.0139 (0.12) 0.328** (2.05) 1.026 (1.46) 0.*** (5.37) 0.1396 0.0997 (0.78) 0.0391*** (6.71) 0.394*** (3.32) 0.228 (1.41) 0.134 (0.19) 0.493*** (5.86) 0.2139 53

5. RESULTS AND ANALYSIS To begn our emprcal analyss we test whether on average there s a postve or a negatve relaton between asset lqudty and stock lqudty by estmatng a fxed effects model as specfed n equaton 1. In order to account for the mpact of other varables we estmate the relaton between asset lqudty and stock lqudty ncludng a set of control varables. We don t report the estmaton results usng the varable WAL1B to save space, however, the results are qualtatvely smlar to ones usng WAL1A. We employ the 20 dfferent combnatons of asset lqudty and stock lqudty measures. Snce four measures of stock lqudty, namely quoted spread, proportonal spread, zero rato and prce mpact, are n fact measures of stock llqudty the sgn of the relaton between asset lqudty and stock lqudty s opposte to the sgn of the coeffcent. However, the turnover rato s a measure of stock lqudty and hence the sgn of the relaton between asset and stock lqudty s smlar to the sgn of the coeffcent. We report the results of the base model n Table 4 (See page 55). Each column n Table 4 reports the estmaton result usng an alternatve measure of (ll) lqudty. Each fve columns n Table 4 report the estmaton results for a dfferent measure of asset lqudty: WAL1; WAL1A; WAL2; and WAL3. The results of the model estmates usng the varable WAL1B are not reported to save space, however, they are qualtatvely smlar to the results usng WAL1A. In case the dependent varable s Quoted, Proportonal and Prce Impact, the sgn of the coeffcents of asset lqudty measures reported n Table 4 are negatve (wth two exceptons). However, they are postve n case the dependent varable s Zero. On the other hand, n case the dependent varable s Turnover the sgn of the coeffcents of asset lqudty measures are negatve. The sgns of the coeffcents of stock lqudty when usng Quoted, Proportonal and Prce Impact as the dependent varable ndcate that asset lqudty s postvely related to stock lqudty. However, the sgns of the coeffcents of stock lqudty when usng Zero and Turnover ndcate asset lqudty s negatvely related to stock lqudty. The only specfcatons where the coeffcents of asset lqudty measures are consstently sgnfcant are when the dependent varable s Prce Impact. However, the coeffcents of asset lqudty measures are sgnfcant when usng Quoted, Proportonal n one specfcaton, usng WAL1, and when usng Zero n one specfcaton, when usng WAL1. The coeffcents on the asset lqudty measures are nsgnfcant n case of usng Turnover n all specfcatons. Overall, the results reported n Table 4 are mxed. These mxed results call for further examnaton as they ndcate that two effects may coexst due to the nfluence of another varable on the assetstock lqudty relaton. One partcular varable of nterest that can nfluence the asset stock lqudty relaton s the frm s ownershp structure. More lqud assets mply more dscreton for agents controllng the frm whch leads to greater uncertanty about future assets and hence lower stock lqudty. Therefore, the average relaton between asset lqudty and stock lqudty may be subject to the nfluence of a frm s ownershp structure. In order to examne the possble nfluence of a frm s ownershp structure on the asset stock lqudty relaton, we nclude an nteracton term between the frm s ownershp concentraton, approxmated by the ownershp of the largest three shareholders, and ts stock lqudty measure. We report the results of the modfed model n Table 5. The estmatons reported n Table 5 reveal some nterestng results. Frst, the sgn of the coeffcents of asset lqudty measures are negatve (wth lttle exceptons) n specfcatons usng llqudty measures as ther ndependent varable and postve n specfcatons usng the lqudty measure. These results ndcate that on average asset lqudty s postvely related to stock lqudty. Second, the average postve mpact of asset lqudty on stock lqudty s reversed when consderng the frm s ownershp structure. The nteracton term between and each of the asset lqudty measures, carry a postve sgn (wth lttle exceptons) n specfcatons usng llqudty measures as ther ndependent varable and a negatve sgn n specfcatons usng the lqudty measure. Specfcatons where the coeffcents of the nteracton term are consstently sgnfcant are when the dependent varable s Prce Impact. The coeffcents of asset lqudty measures are sgnfcant when usng Zero n one specfcaton, and when usng Turnover n another specfcaton. These results show evdence that although the average mpact of asset lqudty on stock lqudty s postve; the mpact becomes negatve n frms wth large ownershp concentraton. To deal wth the endogenety between asset lqudty and stock lqudty we employ the System Generalzed Method of Moments estmator proposed by Arellano and Bond (1991). Ths procedure uses lagged values to nstrument for asset lqudty and estmates the regresson usng the GMM procedure. We report the results n Table 6. The results are smlar to the ones reported n Table 5. The sgn of the coeffcents of asset lqudty measures are negatve (wth lttle exceptons) n specfcatons usng llqudty measures as ther ndependent varable and postve n specfcatons usng the lqudty measure, whch ndcates that on average asset lqudty s postvely related to stock lqudty. In addton, the nteracton term between and each of the asset lqudty measures s postve (wth lttle exceptons) n specfcatons usng llqudty measures as ther ndependent varable and negatve n specfcatons usng the lqudty measure. The coeffcents of the nteracton term are mostly sgnfcant when usng Turnover, and more often than not sgnfcant when usng Quoted, Proportonal, and Prce Impact. However, the coeffcents of asset lqudty measures are sgnfcant when usng Zero n one specfcaton. These results supports the noton that the average mpact of asset lqudty on stock lqudty s postve, however, the mpact becomes negatve n frms wth large ownershp concentraton. 54

Table 5. Stock Lqudty and Asset Lqudty Condtonal on Ownershp Concentraton: Frm Fxed Effects Table 5 reports estmaton results of the stock lqudty model ncludng an nteracton term between and each of the asset lqudty measures. The sample conssts of nonfnancal Jordanan frms lsted n the ASE over the perod 20012012. Varable defntons are presented n Table 1. tstatstcs are n parentheses. ***, **, * ndcate sgnfcance at the 1%, 5%, and 10% respectvely. Quoted Proportonal Prce Quoted Proportonal Prce Zero Turnover Zero Impact Impact Turnover WAL1 0.0612 0.104* 0.0091 0.324*** 0.150** (1.22) (1.80) (0.80) (4.84) (2.18) WAL1* 0.0004 0.0011 0.0004** 0.0049*** 0.0037*** (0.41) (0.96) (1.99) (3.72) (2.75) WAL1A 0.384 0.37 0.0502 2.165** 0.969 (0.52) (0.44) (0.30) (2.20) (0.98) WAL1A* 0.0094 0.0142 0.0006 0.0300* 0.0213 (0.70) (0.93) (0.19) (1.68) (1.19) 0.0112* 0.0083 0.0078*** 0.0491*** 0.0572*** 0.0100** 0.0038 0.0056*** 0.0273*** 0.0401*** (1.78) (1.15) (5.40) (5.78) (6.59) (2.34) (0.78) (5.68) (4.65) (6.78) 0.236*** 0.0766 0.0372* 0.0921 0.418*** 0.215** 0.0659 0.0395** 0.0474 0.391*** (2.70) (0.76) (1.89) (0.80) (3.54) (2.48) (0.66) (2.01) (0.41) (3.32) 0.702*** 0.730*** 0.0800*** 0.231 0.223 0.743*** 0.743*** 0.0837*** 0.244 0.248 (5.92) (5.34) (3.00) (1.48) (1.39) (6.30) (5.50) (3.14) (1.54) (1.55) Proftablty 1.112* 2.419*** 0.102 0.436 0.0482 1.441** 2.753*** 0.101 1.005 0.0438 (1.86) (3.52) (0.83) (0.60) (0.07) (2.50) (4.18) (0.85) (1.42) (0.06) Prce Inverse 0.480*** 0.227* 0.0295** 0.472*** 0.491*** 0.431*** 0.217* 0.0275* 0.454*** 0.*** (4.64) (1.90) (2.10) (5.74) (5.83) (4.31) (1.89) (1.97) (5.45) (5.73) Observatons R 2 0.3847 0.1718 0.1214 0.1787 0.2312 Table 5. Contnued 0.3722 0.1702 0.1031 0.1424 0.2161 WAL2 WAL2* WAL3 WAL3* Proftablty Prce Inverse Observatons R 2 0.052 (0.39) 0.0006 (0.26) 0.0106* (1.78) 0.206** (2.38) 0.746*** (6.27) 1.444** (2.49) 0.435*** (4.31) 0.3710 0.169 (1.10) 0.0015 (0.52) 0.0055 (0.80) 0.0877 (0.88) 0.755*** (5.54) 2.682*** (4.03) 0.242** (2.09) 0.1667 0.0071 (0.24) 0.0006 (0.96) 0.0065*** (4.73) 0.0410** (2.09) 0.0769*** (2.87) 0.111 (0.92) 0.0284** (2.02) 0.1024 0.836*** (4.75) 0.0128*** (3.85) 0.0474*** (5.90) 0.0199 (0.17) 0.280* (1.80) 0.55 (0.78) 0.441*** (5.40) 0.1762 0.368** (2.04) 0.0082** (2.40) 0.0529*** (6.43) 0.371*** (3.17) 0.227 (1.42) 0.122 (0.17) 0.484*** (5.78) 0.2243 0.147 (0.73) 0.0033 (1.01) 0.0067 (1.32) 0.211** (2.41) 0.756*** (6.33) 1.537*** (2.66) 0.425*** (4.21) 0.3728 0.145 (0.63) 0.0037 (0.99) 0.0002 (0.04) 0.0802 (0.80) 0.757*** (5.51) 2.871*** (4.32) 0.216* (1.86) 0.1651 0.0439 (0.96) 0.0002 (0.24) 0.0055*** (4.72) 0.0367* (1.85) 0.0749*** (2.77) 0.125 (1.04) 0.0294** (2.09) 0.1031 0.820*** (3.04) 0.0098** (2.18) 0.0331*** (4.78) 0.0069 (0.06) 0.312* (1.96) 0.796 (1.12) 0.442*** (5.31) 0.1501 0.0477 (0.17) 0.0028 (0.61) 0.0415*** (5.92) 0.388*** (3.26) 0.224 (1.38) 0.0688 (0.10) 0.492*** (5.83) 0.2146 55

Table 6. Stock Lqudty and Asset Lqudty Condtonal on Ownershp Concentraton: SystemGMM Table 6 reports estmaton results of the stock lqudty model ncludng an nteracton term between and each of the asset lqudty measures and usng SystemGMM. Varable defntons are presented n Table 1. zstatstcs are n parentheses. ***, **, * ndcate sgnfcance at the 1%, 5%, and 10% respectvely. a ndcates sgnfcance at the 1%. Quoted Proportonal Prce Quoted Proportonal Prce Zero Turnover Zero Impact Impact Turnover WAL1 0.298*** 0.342*** 0.0232 0.506*** 0.268* (3.19) (3.05) (0.97) (3.80) (1.94) WAL1* 0.0038** 0.0045** 0.00085* 0.0097*** 0.0068** (2.11) (2.09) (1.82) (3.74) (2.53) WAL1A 1.869 2.017 0.165 3.038* 0.0644 (1.55) (1.46) (0.59) (1.87) (0.04) WAL1A* 0.0397* 0.0517** 0.0031 0.0531* 0.0061 (1.73) (1.96) (0.58) (1.72) (0.20) 0.0506*** 0.0357** 0.0115*** 0.0838*** 0.0659*** 0.0200** 0.0269*** 0.007*** 0.0354*** 0.0336*** (3.88) (2.28) (3.41) (4.48) (3.39) (2.31) (2.71) (3.47) (3.03) (2.93) 0.276** 0.0473 0.0774** 0.0799 0.613*** 0.543*** 0.747*** 0.160*** 0.787*** 0.338* (2.01) (0.29) (2.16) (0.40) (2.96) (3.99) (4.78) (4.94) (4.21) (1.84) 1.095*** 1.273*** 0.0496 0.724** 0.146 1.468*** 1.329*** 0.266*** 1.064*** 1.123*** (4.80) (4.66) (0.84) (2.21) (0.43) (8.62) (6.79) (7.03) (4.87) (5.24) Proftablty 0.805 3.634** 1.128*** 5.121** 3.912* 0.143 1.199 0.342 3.989** 2.084 (0.57) (2.14) (3.08) (2.52) (1.86) (0.11) (0.84) (1.25) (2.53) (1.34) Prce Inverse 0.428** 0.305 0.0849** 0.806*** 1.076*** 0.642** 0.365 0.0395 0.549* 0.345 (2.00) (1.19) (2.05) (3.51) (4.52) (2.50) (1.24) (0.71) (1.70) (1.09) Observatons ArellanoBond Sargan Test 1.53 60.65 1.01 51.58 0.47 64.90 0.41 117.75 a 0.12 83.92 a 0.88 53.01 0.98 56.43 0.65 101.91 a 0.21 82.15 a 0.01 130.92 a Table 6. Contnued WAL2 WAL2* WAL3 WAL3* Proftablty Prce Inverse Observatons ArellanoBond Sargan Test 0.227 (1.14) 0.0012 (0.40) 0.0188* (1.71) 0.318** (1.97) 0.978*** (4.74) 1.079 (0.77) 0.675** (2.37) 1.14 80.34 a 0.190 (0.87) 0.0076 (1.28) 0.0039 (0.32) 0.190 (1.06) 0.905*** (3.96) 1.157 (0.74) 0.522* (1.65) 0.99 74.74 a 0.0646 (1.25) 0.0007 (0.94) 0.0063** (2.21) 0.0544 (1.27) 0.0827 (1.54) 0.739** (2.00) 0.131*** (3.13) 0.54 97.61 a 0.219 (0.78) 0.0032 (0.76) 0.0329** (2.13) 0.331 (1.43) 0.223 (0.77) 2.688 (1.34) 0.892*** (3.94) 0.02 165.34 a 0.611** (2.12) 0.00998** (2.31) 0.0223 (1.41) 0.497** (2.09) 0.0718 (0.24) 1.003 (0.49) 0.787*** (3.38) 0.47 160.54 a 0.0908 (0.33) 0.0036 (0.90) 0.0059 (0.67) 0.0791 (0.69) 1.241*** (6.05) 2.129* (1.75) 0.876*** (3.58) 1.12 85.60 a 0.363 (1.19) 0.0104 (1.38) 0.0036 (0.37) 0.153 (1.21) 1.273*** (5.62) 2.042 (1.52) 0.392 (1.45) 1.18 71.12 0.0060 (0.08) 0.0002 (0.20) 0.0013 (0.56) 0.0307 (0.97) 0.0049 (0.09) 1.070*** (3.26) 0.136*** (3.02) 0.48 87.78 a 0.315 (0.76) 0.0034 (0.58) 0.0107 (0.82) 0.0703 (0.41) 0.550* (1.82) 6.958*** (3.88) 0.940*** (3.82) 0.18 126.19 a 0.307 (0.72) 0.0059 (0.97) 0.0004 (0.03) 0.396** (2.25) 0.34 (1.10) 4.012** (2.19) 0.831*** (3.30) 0.45 121.64 a 56

As for the control varables, we fnd strong evdence that ownershp concentraton s negatvely related to lqudty. The varable s postvely and sgnfcantly related to our proxes of llqudty measures, except for Proportonal, and negatvely and sgnfcantly related to our lqudty measure, Turnover. Ths result ndcates s negatvely related to lqudty. In addton, we fnd that the frm s market value s postvely and sgnfcantly related to Quoted and negatvely and sgnfcantly related to Turnover, ndcatng that s negatvely related to lqudty. Ths result s consstent wth the evdence reported n Gopalan et al. (2012). However, s sgnfcantly and negatvely related to Zero, ndcatng that s postvely related to. Ths result supports the fndngs reported n Charoenwong et al. (2014). We also fnd that s postvely and sgnfcantly related to Proportonal, Quoted and Zero. These fndngs ndcate that s negatvely related to lqudty. The frm s proftablty, Proftablty, when sgnfcant, s negatvely related to measures of stock lqudty and postvely related to the measure of stock lqudty ndcatng that the frm s proftablty s postvely related to lqudty. Fnally, Prce Inverse s negatvely related to tradng costs especally Quoted. However, t s postvely related to other llqudty measures, Zero and Prce Impact and negatvely related to Turnover. 6. CONCLUSION Ths paper nvestgates the mpact of asset lqudty on stock lqudty usng a sample of frms lsted on the ASE durng the perod 20012012. In ths study, we examne how ownershp concentraton affects the relatonshp between asset and stock lqudty. Excess lqud assets ncrease the scope of large shareholders dscreton and hence may result n ncreasng the uncertanty of assets n place rather than reducng t. Free cash flows ncrease largecontrollng shareholders power due to the exstence of more resources under ther control. In addton, t s less costly to turn lqud assets nto prvate benefts compared to other assets. Therefore, n the absence of nvestor protecton, large shareholders may have ncentves to approprate lqud assets. The uncertanty regardng the usage of lqud assets n cashrch frms leads to greater uncertanty regardng the frm s cash flows and hence nvestors trade the stocks of cashrch frms controlled by large shareholders at a premum. Therefore, the senstvty of stock lqudty to asset lqudty s expected to be negatve for companes wth concentrated ownershp. The results show some evdence that asset lqudty measures are negatvely related to tradng costs, prce mpact and the proporton of zero tradng days and postvely related to the turnover rato. These results ndcate that on average asset lqudty s postvely related to stock lqudty. In addton, the results ndcate that the nteracton between asset lqudty measures wth the ownershp concentraton measure s postvely related to llqudty measures (negatvely related to the turnover rato). Ths result ndcates that lqud assets n companes wth (wthout) large shareholders reduce (enhance) stock lqudty. Excess cash n companes wth large shareholders ncrease the scope of large shareholders dscreton whch leads to greater uncertanty about the frm s future cash flows. Therefore the senstvty of stock lqudty to asset lqudty s negatve for companes wth concentrated ownershp. The evdence presented n ths paper, although nconclusve, s mportant as t shows that ownershp structure nfluences the relatonshp between stock and asset lqudty. Future research can reexamne ths ssue by nvestgatng other economes wth varyng degrees of nvestor protecton. REFERENCES 1. Aboody, D. & Lev, B. (2000), Informaton Asymmetry, R&D, and Insder Gans, The Journal of Fnance, Vol. 55 No. 6, 27472766. 2. Amhud, Y. (2002), Illqudty and Stock Returns: CrossSecton and TmeSeres Effects, Journal of Fnancal Markets, Vol. 5 No. 1, 3156. 3. Amhud, Y. & Mendelson, H. (1986), Asset Prcng and the BdAsk, Journal of Fnancal Economcs, Vol. 17 No. 2, 223249. 4. Anderson, T. W. & Hsao, C. (1981) Estmaton of Dynamc Models wth Error Components, Journal of the Amercan statstcal Assocaton, Vol. 76 No. 375, 598606. 5. Arellano, M. & Bond, S. (1991), Some Tests of Specfcaton for Panel Data: Monte Carlo Evdence and an Applcaton to Employment Equatons, The Revew of Economc Studes, Vol. 58 No. 2, 277297. 6. Avramov, D. & Chorda, T. (2006), Asset Prcng Models and Fnancal Market Anomales, Revew of Fnancal Studes, Vol. 19 No. 3, 10011040. 7. Branch, B. & Freed, W. (1977), BdAsked s on the AMEX and the Bg Board, The Journal of Fnance, Vol. 32 No. 1, 159163. 8. Berkman, H. & Eleswarapu, V. R. (1998) ShortTerm Traders and Lqudty: A Test Usng Bombay Stock Exchange Data, Journal of Fnancal Economcs, Vol. 47 No. 3, 339355. 9. Brennan, M. J., Chorda, T. & Subrahmanyam, A. (1998), Alternatve Factor Specfcatons, Securty Characterstcs, and the CrossSecton of Expected Stock Returns, Journal of Fnancal Economcs, Vol. 49 No. 3, 345373. 10. Brennan, M. J. & Subrahmanyam, A. (1996), Market Mcrostructure and Asset Prcng: On the Compensaton for Illqudty n Stock Returns, Journal of Fnancal Economcs, Vol. 41 No. 3, 441 464. 11. Charoenwong, C., Chong, B. S. & Yang, Y. C. (2014) Asset Lqudty and Stock Lqudty: Internatonal Evdence, Journal of Busness Fnance & Accountng, Vol. 41 No (34), 435468. 12. Chen, T., Duong, H. N., Rhee S. G. & Veeraraghava M., (2013), Stock Lqudty and the Value of Cash Holdngs, Monash Unversty, Australa. 13. Chorda, T., Roll, R. & Subrahmanyam, A. (2000), Commonalty n Lqudty, Journal of Fnancal Economcs, Vol. 56 No. 1, 328. 14. Chorda, T., Subrahmanyam, A. & Anshuman, V. R. (2001), Tradng Actvty and Expected Stock Returns, Journal of Fnancal Economcs, Vol. 59 No. 1, 332 15. Cooper, K. S., Groth, J. C. & Avera, W. E. (1985), Lqudty, Exchange Lstng, and Common Stock Performance, Journal of Economcs and Busness, Vol. 37 No. 1, 1933. 16. Demsetz, H. (1968), The Cost of Transactng, Quarterly Journal of Economcs, Vol. 82 No. 1, 33 53. 57

17. DemrgüçKunt, A. & Levne, R. (1996), Stock Markets, Corporate Fnance, and Economc Growth: An Overvew, The World Bank Economc Revew, Vol. 10 No. 2, 223239. 18. Datar, V. T., Nak, N. Y. & Radclffe, R. (1998), Lqudty and Stock Returns: An Alternatve Test, Journal of Fnancal Markets, Vol. 1 No. 2, 203219. 19. Easley, D., & O Hara, M. (1987), Prce, Trade Sze, and Informaton n Securtes Markets, Journal of Fnancal Economcs, Vol. 19 No. 1, 6990. 20. Faulkender, M. & Wang, R. (2006) Corporate Fnancal Polcy and the Value of Cash, The Journal of Fnance, Vol. 61 No. 4, 1957 1990. 21. Gopalan, R., Kadan, O., & Pevzner, M. (2012), Asset Lqudty and Stock Lqudty, Journal of Fnancal and Quanttatve Analyss, Vol. 47 No. 2, 333364. 22. Hasbrouck, J. (2005), Tradng Costs and Returns for US Equtes: The Evdence from Daly Data, NYU Stern School Department of Fnance Workng Paper, New York Unversty. 23. Jensen, M. C. (1986), Agency Cost of Free Cash Flow, Corporate Fnance, and Takeovers, Amercan Economc Revew, Vol. 76 No. 2, 323329. 24. Kothar, S. P., Laguerre, T. & Leone, A. (2002), Captalzaton versus Expensng: Evdence on the Uncertanty of Future Earnngs from Captal Expendtures versus R&D Outlays, Revew of Accountng Studes, Vol. 7 No. 4, 355382. 25. Lesmond, D. A. (2005), Lqudty of Emergng Markets, Journal of Fnancal Economcs, Vol. 77 No. 2, 411452. 26. Myers, S. C. & Rajan, R. G. (1998), The Paradox of Lqudty, The Quarterly Journal of Economcs, Vol. 113 No. 3, 733771. 27. Roll, R. (1984), A Smple Implct Measure of the Effectve BdAsk n an Effcent Market, The Journal of Fnance, Vol. 39 No. 4, 1127 1139. 28. Stoll, H. R. (1978), The Prcng of Securty Dealer Servces: An Emprcal Study of Nasdaq Stocks, The Journal of Fnance, Vol. 33 No. 4, 11531172. 29. Stoll, H. R. (2000), Presdental address: Frcton, The Journal of Fnance, Vol. 55 No. 4, 14791514. 30. Tnc, S. M. (1972), The Economcs of Lqudty Servces, Quarterly Journal of Economcs, Vol. 86 No. 1, 7993. 31. Venkataraman, K. (2001), Automated Versus Floor Tradng: An Analyss of Executon Costs on the Pars and New York Exchanges, The Journal of Fnance, Vol. 56 No. 4, 14451485. 32. Wooldrdge, J. M. (2002). Econometrc Analyss of Cross Secton and Panel Data, MIT Press. 58