Corporate cash holdings: An empirical investigation of UK companies

Corporate cash holdings:An empirical investigation of UK companies This paper investigates the empirical determinants of corporate cash holdings for a sample of UK firms over the period

Corporate cash holdings:

An empirical investigation of UK companies

This paper investigates the empirical determinants of corporate cash holdings for a sample

of UK firms over the period 1984-1999 We present evidence of the significant relationbetween managerial ownership and cash holdings The results also suggest that the way inwhich managerial ownership exerts influence on cash holding decisions differs between firmswith ultimate controllers and those that are widely-held The results reveal that growth options

of firms, cash flows, liquid assets, leverage and bank debt are important in determining cashholdings In contrast, there is much less evidence that larger firms hold less cash Our analysisalso suggests that unobserved firm heterogeneity and endogeneity are crucial in analysing thecash structure of firms

1 Introduction

Why do firms hold large amounts of cash and cash equivalents? Various explanations havebeen put forward in an attempt to provide some answers to this question One majorexplanation is that cash provides low cost financing for firms According to this view, raisingexternal finance costs more in the presence asymmetry of information between firms andexternal investors (Myers and Majluf, 1984); costly agency problems such as underinvestmentand asset substitution (Myers, 1977; and Jensen and Meckling, 1976); and adjustment costsand other financial restrictions Therefore, managers trying to minimize the costs associatedwith external financing in imperfect capital markets may find it optimal to maintain sufficientinternal financial flexibility However, there are also potential adverse effects of cashholdings Central to this view is the argument that agency conflicts existing betweenshareholders and managers can be most severe when firms have large free cash flows (Jensen,1986) Managers can pursue their own interests at the expense of shareholders and cash servesthe interests of managers more than those of shareholders in this respect

Recently the investigation of cash holdings of firms has gained a great deal of attention inthe empirical literature An important strand of this literature has focused on the determinants

cash holdings for a sample of US companies They report that firms facing higher costs ofexternal financing and having more volatile earnings and firms with relatively lower returns

on assets have significantly larger proportions of liquid assets to total assets For similarfirms, Opler et al (1999) provide evidence that firms with strong growth opportunities andriskier cash flows, and small firms hold larger amounts of cash Finally, in a related paper,Pinkowitz and Williamson (2001) examine the cash holdings of firms from the United States,Germany, and Japan In addition to finding which are similar to those in Opler et al (1999)

1

The other important strand of this literature examines the relationship between cash holdings and corporate performance See, for example, Harford (1999), Mikkelson and Partch (2002), and also Opler et al (1999).

they document that the monopoly power of banks in Japan has a significant impact on cashbalances of Japanese firms.

The purpose of our paper is to contribute to this literature by examining the empiricaldeterminants of cash holdings for a sample of UK companies over the period 1984-1999 The

UK and US are often described as being similar with respect to ownership and controlstructures of companies They are also characterised as having similar institutional and legalframework It is, however, our view that there are distinct features of the corporategovernance system in the UK, which may lead to different inferences than those in the USwith regard to the cash holding behaviour of firms To take an example, to the extent thatlarge cash holdings serve managers’ interests, it is more likely in the UK that cash holdingswill increase with managerial shareholdings This is because, as we will argue, managers inthe UK appear to entrench themselves considerably against external market discipline

As a first contribution to this literature, we investigate the role of ownership structure indetermining corporate policies on cash holdings More specifically, we first empiricallyanalyse the nature of the relationship between managerial ownership and cash holdings.Furthermore, we investigate whether the presence of ultimate controllers in the firm has asignificant impact on the amount of cash it holds In addition, we extend our analysis byaddressing the potential interactions between managerial ownership and ultimate controllers

In particular, we examine the extent to which the presence and identity of the controllingshareholder affect the managerial behaviour towards cash holdings We argue that thepresence of a controlling shareholder can affect cash holding decisions of firms For instance,

if large holdings of cash serve controlling shareholders’ interests one would expect to observehigher cash holdings in firms with controllers Also, to the extent that the incentive and ability

to monitor managers change with the identity of controllers, the relationship betweenmanagerial ownership and cash holdings may depend on who the firm’s ultimate controller is.Our second contribution is that, distinct from previous empirical studies, the analysis ofthis paper explicitly deals with the endogeneity problem in testing the cash holdingshypotheses We think that the endogeneity issue in this context is important for several

reasons First, it is highly likely that observable as well as unobservable shocks affecting cashholdings can also affect some of the firm-specific characteristics such as market value ofequity Second, it is possible that observed relations between cash and its potential

determinants reflect the effects of cash on the latter rather than vice versa To control

efficiently for the potential endogeneity problem we utilise panel data and the GeneralisedMethod of Moments (GMM) estimation procedure, the combination of which allows us tooptimally choose instruments as well as to deal with firm heterogeneity

Our last contribution lies in the dynamic analysis of the cash holding decision Weincorporate the view that market imperfections such as adjustment costs may prevent firmsfrom adapting to new circumstances We utilise a partial target-adjustment model that allowsfor the possibility of delays in response of firms in adjusting their cash holdings We are notthe first to investigate the question whether firms have target cash holdings Opler et al.(1999), for example, estimate different target-adjustment models relating the firm’s actualcash holdings to its target cash holdings Their results provide evidence that firms have targetcash levels Our dynamic analysis is an attempt to complement rather than substitute theanalysis of Opler et al (1999) Our model incorporates all the firm-specific factors described

in the paper as relevant in determining cash holdings More importantly, in estimating thetarget-adjustment model we also control for unobservable fixed effects as well as time effects

To the extent that these effects are significant in the underlying target cash model and notcontrolled for, estimated coefficients of the target-adjustment model will be biased As notedabove, the approach adopted in our dynamic analysis also controls for the potential biases thatmay arise from endogeneity of regressors as well as random measurement errors

Our analysis reveals that ownership structure of firms plays an important role indetermining levels of cash UK companies hold We find evidence for the non-monotonicrelationship between managerial ownership and corporate cash holdings In addition, thenature of the relationship changes with the presence of ultimate controllers We also provideevidence of significant dynamic effects in the determination of firms’ cash holdings.Moreover, there is evidence that cash flow and growth opportunities of firms exert a positive

influence on cash holdings There is significant evidence for the negative impact of liquidassets The results also suggest that higher cash holdings are associated with lower levels ofbank debt and leverage in firms’ capital structure.

The paper is organised as follows In Section 2 we briefly discuss the main features of theownership and control structures of companies in the UK as distinct from the US Section 3reviews the relevant theory and derives the empirical hypotheses Section 4 describes thealternative estimation methods used in the paper Section 5 describes the construction of thedata set Section 6 presents the empirical results and finally Section 7 offers our conclusions

2 UK Institutional Features

Companies in both the UK and US are often described as being similar with respect to theirownership structures and as characterised as having similar regulatory systems For example,both the UK and US are often described as “market-oriented” countries with similar capital

countries are described as “outsider” systems in which ownership is dispersed amongst a largenumber of outside investors However, there exist important differences in the corporategovernance system and in the patterns of share ownership, which makes the conduct of ananalysis of cash holdings of UK companies interesting

The main features of the prevailing corporate governance system in the UK as distinctfrom those in the US can be summarised as follows First, the concentration of institutionalstock ownership is higher in the UK than in the US Nestor and Thompson (2000) report thatfinancial institutions in the UK hold 68 percent of the all shareholdings in 1994 as compared

to 46 percent in the US in 1996 Goergen and Renneboog (2001) argue that financialinstitutions adopt a passive stance towards disciplining firms’ management This, in turn,

2

The so-called market-oriented countries include Canada, United Kingdom and United States; and the bank

oriented countries include France, Germany, Italy and Japan See for a detailed discussion, for example, Hoshi et

al (1991), and Rajan and Zingales (1995).

coupled with significant shareholdings by directors, further increases in the power of directorsand creates its own type of agency problems, i.e high managerial discretion They argue thatthe passive stance of financial institutions is mainly due to the fact that they are not majorplayers from the agency perspective Goergen and Renneboog (2001) report that the average

of the largest shareholding owned by financial institutions is only 5.5 percent for their sample

of firms in 1992 We also report in Section 5 that 6.5 percent of non-financial firms in the UK

is controlled by widely-held financial institutions at the 20 percent ultimate control thresholdfor a sample of 780 firms Also, we find that the average value of control rights of the largestcontrolling financial institution is 20.2 percent This compares with the average values of38.35 and 26.67 percent for those firms that are ultimately controlled by widely-heldcorporations and families respectively

Second, managerial discretion is higher in the UK Franks et al (2001) report that highershareholdings by insiders in the UK lead to entrenchment rather than discipliningmanagement There are several potential reasons for this For example, they argue thatsubstantial directors shareholdings enable managers to hinder monitoring activities sought byother shareholders such as restructuring the firm’s board They also argue that in the UK therole of non-executive directors is quite different from that in the US Non-executive directors

claim that stronger minority investor protection in the UK discourages coalition ofshareholders This, in turn, coupled with less fiduciary obligations on directors but stricter

Finally, there is also a divergence of empirical evidence regarding the disciplining role oftakeovers Franks and Mayer (1996), in contrast to the findings of Martin and McConnell(1991) for the US companies, provide evidence that takeovers do not work as a corporategovernance mechanism for disciplining poor managers in the UK

To summarise, the agency problem between shareholders and managers seems to be moresevere in the UK than in the US and, with the lack of external market discipline and efficientmonitoring by financial institutions, managers are more likely to be entrenched And aboveall, the significance of managerial ownership would be greater in the UK than in the US andone would expect that this is reflected in the determination of cash holdings.

3 Theory and Empirical Hypotheses

3 1 Asymmetric Information, Agency Costs of Debt, Financial Distress

Existence of asymmetric information between firms and investors make external financingcostly Myers and Majluf (1984) argue that in the presence of asymmetric information firmstend to follow a hierarchy in their financing policies in the sense that they prefer internal overinformationally sensitive external finance Smith (1986) provides evidence that is consistentwith asymmetric information problems of external financing, i.e investors discount the value

of firms when they attempt to sell risky securities Myers and Majluf (1984) also argue thatasymmetric information problem is more severe for firms whose values are determined bygrowth options If a firm has investment opportunities that would increase its value whentaken and finds itself being short of cash, it may have to pass up some of these investments.Hence firms with such opportunities would hold greater amounts of cash in attempt to make itless likely that it will have to give up valuable investment opportunities in some states ofnature It is also important to note that firms with greater growth opportunities are expected toincur higher bankruptcy costs (Williamson, 1988; Harris and Raviv, 1990; and Shleifer andVishny, 1992) Growth opportunities are intangible in nature and their value fall precipitously

in financial distress and bankruptcy This would in turn imply that firms with greater growthopportunities have greater incentives to avoid financial distress and bankruptcy and hencehold larger cash and marketable securities

4

Franks et al (2001) find that factors such as capital structure of firms and new equity financing are more

Growth opportunities are also related to agency costs of debt arising from the conflicts ofinterest between shareholders and debtholders It is argued that growth firms face higheragency costs because firms with risky debt and greater growth opportunities are likely to pass

up valuable investment opportunities in more states of nature (Myers, 1977) Higher expectedagency costs in turn make external financing expensive Then it follows that firms withgreater growth options are more likely to accumulate cash to avoid costly external financing

as much as possible and to have financial flexibility

To proxy for growth opportunities of firms we use the market-to-book ratio defined as theratio of book value of total assets minus the book value of equity plus the market value ofequity to book value of assets

It is also suggested that large firms have less information asymmetry than small firms(Brennan and Hughes, 1991; and Collins et al., 1981) Therefore, small firms face moreborrowing constraints and higher costs of external financing than large firms (Whited, 1992;Fazzari and Petersen, 1993, and Kim et al., 1998) To the extent that size is an inverse proxyfor the degree of informational asymmetry and, in turn, cost of external financing, a negativerelation should be expected between size and cash holdings Finally, size of firms can also berelated to costs of financial distress Larger firms are more likely to be diversified and thusless likely to experience financial distress (Titman and Wessels, 1988) and smaller firms aremore likely to be liquidated when they are in financial distress (Ozkan, 1996) If this is thecase, small firms are expected to hold relatively more cash to avoid financial distress

We use the natural logarithm of total assets in 1984 prices as a proxy for the size of firms

3.2 Ownership and cash holdings

In this section, we discuss the impact that ownership structure of firms may potentiallyhave on their choices of cash and other marketable securities holdings

effective in disciplining management especially in poorly performing companies.

Managerial ownership Jensen and Meckling (1976) show that there are potential conflicts

of interest between managers and outside shareholders when managers own less than 100percent of the residual cash flow rights The potential for such problems in turn createsagency costs that are generally borne by the original shareholders The agency problembetween managers and shareholders can stem, for example, from managerial shirking andconsumption of perquisites that normally benefit managers Another potential agency problem

is due to free cash flow the firm generates in excess of the amount required to fund allvaluable investment projects Jensen (1986) argues that free cash flow presents seriouspotential conflicts because large amount of cash reserves can serve mainly managers’interests Managers have incentives to increase the amount of funds under their controlbecause this enables them to spend it as they wish, i.e squandering funds by consuming

may prefer that free cash flow be returned to them

Following Jensen and Meckling (1976) a large body of literature has developed to supportthe notion that managerial ownership can help align the financial interests of managers withthose of shareholders A great deal of this literature concerns the relationship betweenmanagerial ownership and performance of the firm It has been claimed that, with anincreased managerial ownership, a firm’s performance improves since managers are lesslikely to divert resources away from value maximisation At certain level of managerialownership, however, outside shareholders find it difficult to monitor the actions of managers

as management becomes entrenched Consequently, managers become more capable ofshirking and the consumption of perquisites may outweigh the loss they suffer from a reducedvalue of firm (Morck et al., 1988, and McConnell and Servaes, 1990) It is argued that thereexists a non-monotonic relationship between managerial ownership and the alignment ofshareholder and managerial interests

5 Blanchard et al (1994) provides evidence that firms experiencing cash windfalls tend to keep the resources inside and invest them in unattractive projects just to avoid giving up cash or having an outsider lay a claim on it.

Despite the theoretical arguments and some empirical evidence pointing to the significantrelationship between the managerial ownership levels and the alignment of the interests ofshareholders and managers, the extent to which managerial ownership impacts firms’ choices

of cash holdings has not been adequately investigated In this paper we investigate the extent

to which cash levels change with increasing levels of managerial ownership and whether thisrelationship is affected by other characteristics of ownership structure of firms

Large share ownership or block ownership We now turn to the question whether corporate

cash holdings are different in the presence of a large shareholder (controller) among the firm’sshareholders It is argued that one of the ways of controlling the agency problem betweenmanagers and shareholders is to effectively monitor managers to ensure that they act in theinterests of shareholders However, monitoring by an average shareholder is not free ofdrawbacks For example, for an average shareholder there may not be enough incentives tomonitor managers as the cost of monitoring is likely to outweigh the benefit Shareholdersbear all costs related to their monitoring acts while benefiting from monitoring only inproportion to their shareholding (Grossman and Hart, 1988) In contrast, large shareholders,having claims on a large fraction of the firm’s cash flows, can monitor managers moreeffectively and free riding problems involved in monitoring are mitigated Consequently, inthe presence of a large shareholder, managerial discretion is curbed to some extent and agencycosts between management and shareholders are reduced (Stiglitz, 1985; Shleifer and Vishny,1986) To the extent that this argument holds, the cost of external financing would be lower

While enhanced monitoring by large shareholders can help reduce some of the agencyproblems associated with management, there are also private benefits of control accrued to

6 The empirical evidence on the effectiveness of the monitoring by large shareholders is, however, mixed Mehran (1995) finds that use of executive compensation declines with the percentage of equity ownership by outside blockholders, which is interpreted as evidence of a significant role for blockholders in monitoring executives (see Holderness, (2002) for a detailed survey on the effectiveness blockholders in monitoring management) Franks et al (2001), on the other hand, report that large shareholders in the UK do not seem to discipline the management of poorly performing companies.

large shareholders, not necessarily shared by minority shareholders Recent studies emphasisethe potential conflicts of interest between the controlling shareholder and other shareholders.Shleifer and Vishny (1997) argue that when large owners gain nearly full control of thecorporation, they prefer to generate private benefits of control that are not shared by minorityshareholders (see also Faccio, Lang, and Young, 2001; and Holderness, 2002) Consequently,large shareholders might have the incentive and the opportunity to increase the amount offunds under their control to consume corporate benefits at the expense of minorityshareholders One way of doing it is obviously to accumulate large amounts of cash Withhigher levels of cash holdings it is less likely that controlling shareholders will relinquishcontrol and share the efficiency gains with outside shareholders These arguments suggest apositive relationship between large shareholders and cash holdings.

Identity of ultimate controllers Finally, it is possible that identity of a controlling

shareholder can determine the nature of monitoring that is provided by that shareholder That

is, the incentives to monitor the management may depend on the category of controllingshareholders For example, the direct involvement of controlling family owners in themanagement of the firm is more likely than that of financial institutions It can be argued thatthis would lead to higher agency costs regarding the relationship between managers andoutside shareholders In particular, they may want to keep their control over the firminefficiently long from the outside shareholders’ perspective In the context of our analysisthis means that family controlled firms would hold more cash and marketable securities thanother controlled firms On the other hand, to the extent that they are run by managers who arenot entrenched through a controlling ownership and have superior monitoring abilities,control by financial institutions may imply more and cost-effective monitoring of themanagement This may lead to a reduction in the frictions between managers and shareholdersincluding outside shareholders This could in turn make the agency costs associated withhigher levels of managerial ownership lower However, this could also lead to higher costs iflarge institutions do not exert sufficient monitoring and there are not other controlling

shareholders of the firm There is evidence, for example, in the UK of a passive stance taken

by financial institutions (Franks et al., 2001) Also, Faccio and Lasfer (2000) provideevidence that pension funds do not add monitoring value

3.3 Bank Relationship

It is often argued that bank financing is more effective than public debt in reducingproblems associated with agency conflicts and informational asymmetry (see, e.g., Diamond,1984; Boyd and Prescot, 1986; and Berlin and Loeys, 1988) This is mainly because of thecomparative advantage of banks in monitoring firms’ activities and in collecting andprocessing information Fama (1985) argues that banks have a comparative advantage as alender in minimizing information costs and can get access to information from a firm’sdecision process not otherwise publicly available Therefore, banks can be viewed asperforming a screening role employing private information that allows them to evaluate andmonitor borrowers more effectively than other lenders Thus, a bank’s willingness to provide

providing signals about the borrowing firms’ credit worthiness, existence of a bankrelationship would enhance the ability of firms to have access to external finance To theextent that these arguments hold, firms with more bank debt in their capital structures areexpected to have easier access to external finance This would, in turn, imply that such firmsshould have less cash and marketable securities Another reason why bank debt financingmight have a negative impact on firms’ cash holdings is that bank debt is more easilyrenegotiated when firms need to (see, e.g., Chemmanur and Fulghieri, 1994) By providingflexibility through renegotiation, bank debt can serve as a substitute for holding high levels ofcash and marketable securities

7 James (1987) and Mikkelson and Partch (1986) document that the announcement of a bank credit agreement conveys positive news to the stock market about the borrowing firms’ credit worthiness Billett et al (1995) reconfirms that, unlike public debt issuance, bank loan announcements are associated with positive borrower returns Additionally, Slovin and Young (1990) demonstrate that the presence of a banking relationship lessens the degree of expected underpricing associated with the initial public offerings of client firms.

3.4 Liquidity constraints and cash substitutions

The greater the firm’s cash flow variability, the greater the number of states of nature inwhich the firm will be short of liquid assets As noted earlier, it may be costly to be short ofcash and marketable securities if the firm has to pass up valuable investment opportunities.There is evidence that firms with cash shortfalls do indeed fail to take up some of the valuablegrowth opportunities For example, Minton and Schrand (1999) show that firms with highercash flow volatility permanently forgo investment rather than reacting to cash flow shortfalls

by changing the discretionary investment timing They also argue that a higher frequency ofcash flow shortfalls in the presence of capital imperfections increases a firm’s cost ofaccessing external capital This also adversely affect the level of investment Thus, firms withmore volatile cash flows are expected to hold more cash in an attempt to mitigate the expectedcosts of liquidity constraints

To the extent that there are substitutes for holding high levels of cash firms can use themwhen they have cash shortfalls For example, firms can use borrowing as a substitute forholding cash because leverage can act as a proxy for the ability of firms to issue debt (John,1993) Moreover, Baskin (1987) argues that the cost of funds used to invest in liquidityincreases as the ratio of debt financing increases, which would imply a reduction in cashholdings with increased debt in capital structure We therefore predict that there is a negativerelation between the firm’s cash holdings and its leverage However, one should note thathigher debt levels can increase the likelihood of financial distress In that case one wouldexpect a firm with a high debt ratio to increase its cash holdings to decrease the likelihood of

a financial distress This would induce a positive relation between leverage and cash holdings.Another substitution effect is due to other liquid assets besides firms may have It isreasonable to assume that the cost of converting non-cash liquid assets into cash is muchlower as compared with other assets To the extent that the firm has these assets it may nothave to use the capital markets to raise funds when it has a shortage of cash The proxy weuse for non-cash liquid assets is the ratio of net working capital minus cash to total assets

Finally, we include a dividend dummy in our regressions to control for the potential impact

of the firm’s dividend policy on its cash holdings It takes a value of one if firms paydividends and zero otherwise To the extent that firms that pay dividends can raise fundsrelatively easily by cutting their dividends, a negative relationship is expected between

dividend and cash holdings (Opler et al., 1999) However, it is possible that ceteris paribus

dividend-paying firms can also hold more cash than firms that do not as they may want toavoid a situation in which they are short of cash to support their dividend payments If this isthe case a positive relation can be observed

4 Empirical Specifications

In the following, we describe the empirical methods used in the paper to investigate therelation between corporate cash holdings and firm-specific characteristics includingownership structure, growth opportunities, size and leverage

4.1 Cross-sectional analysis

We begin our analysis by examining the relation between ownership structure and cashholdings by focusing on the question whether managerial ownership and characteristics ofultimate controllers of firms affect cash levels For this purpose, we estimate a cross-sectionalcash model using the average values of each of the firm characteristics (except variability andownership variables) over four years in an attempt to mitigate problems that might arise due

to short-term fluctuations or extreme values in one year We measure cash holdings (thedependent variable) in 1999 and the explanatory variables preceding the sample year in whichcash holdings are measured over the period 1995-1998 This is done to control for theproblem of endogeneity Using past values reduces the likelihood of observed relationsreflecting the effects of cash holdings on firm-specific factors (Rajan and Zingales, 1995)

4.2 Static panel data model

The static model of cash holdings takes the following general form

it t i it it

it it

it it

it it

DIVIDEND SIZE

MKTBOOK BANKDEBT

LEV LIQ

CFLOW

CASH

εααβ

β

ββ

ββ

β

++++

++

++

+

=

6 7

5 4

3 2

1

(1)

time-constant firm-specific effects and firm-constant time effects respectively It is assumed

impact on cash holdings They change across firms but fixed for a given firm through time In

economy-wide factors that are outside the control of firms such as prices

extent that there are time-constant fixed effects in the underlying model, estimatedcoefficients in a cross-sectional regression will be biased due to the correlation generatedbetween the regressors and error term The extent to which these unobserved effects remainrelatively stable over time, one could control for them by using a fixed-effects (within)estimator and obtain consistent coefficient estimates (see, e.g., Wooldridge, J.M., 2002)

4.3 Dynamic panel data model

We now proceed to motivate the dynamic model The static cash holding model implicitlyassumes that firms can instantaneously adjust towards the target cash level following changes

in firm-specific characteristics and/or random shocks In this paper, we adopt a more realisticapproach recognising that an adjustment process takes place, involving a lag in adjusting tochanges in the target cash structure The possibility of delays in the adjustment process can bejustified by the existence of adjustment costs causing the current cash structure not to beimmediately adjusted to a new desired cash structure (for a discussion in a capital structure

context see, e.g., Myers, 1984 and Fischer et al., 1989) We investigate these issues bymodelling the firm’s behaviour as a partial adjustment to a target cash ratio.

it k

kit k

Combining (2) and (3) yields

k

it kit k t

it t i it it

it

it it

it it

it it

u DIVIDEND

SIZE MKTBOOK

BANKDEBT LEV

LIQ CFLOW

CASH

CASH

++++

+

++

++

+

ααγ

γγ

γγ

γγ

γ

8 7

6

5 4

3 2

1 1

(5)

The presence of the lagged dependent variable in (5) makes allowance for the adjustment

by including time dummies in all panel data estimations

Despite its appeal, the dynamic specification in (5) involves several estimation problems.Even when unobservable firm-specific effects are not correlated with regressors, it is still

the error term are serially uncorrelated However, the first-difference transformation toeliminate fixed effects introduces correlation between the lagged dependent variable and

Another estimation problem that is not necessarily specific to the dynamic specificationarises because the firm-specific variables are unlikely to be strictly exogenous That is, shocksaffecting cash structure choices of firms are also likely to affect some of the regressors such

as market value of equity, liquidity, and leverage Moreover, it is likely that some of theregressors may be correlated with the past and current values of the idiosyncratic component

of disturbances

Finally, it is worth mentioning that, the regressors including the lagged dependent variablemay be subject to random measurement errors, which also induce biases in the estimates.This problem may be particularly relevant for all the variables in our analysis that are

8 The alternative to first-difference transformation is the within transformation that is commonly used in the literature Although controlling for the fixed effects, it introduces correlation between the lagged dependent variable and time-averaged idiosyncratic error term, leading to biased estimates It is shown that the bias falls

with the number of years T (see Nickell, 1981; and Chamberlain, 1982) This would however not be the case in our analysis as T is fairly small ranging from 5 to 16 Moreover, the bias induced by within transformation will remain in the presence of measurement errors even if T is large.

calculated from company balance sheet data and market data (see, Baltagi, 1995 for a detaileddiscussion of measurement error and panel data).

The problems outlined above advocates the use of an Instrumental Variables (IV)estimation method, where the lagged dependent variable and endogenous regressors areinstrumented using an appropriate set of instrumental variables This paper therefore employsthe GMM method of estimation which provides consistent parameter estimates by utilizinginstruments that can be obtained from the orthogonality conditions that exist between thelagged values of the variables and disturbances (see Arellano and Bond, 1991) It allows bothfor an MA(1) error structure and the heteroskedasticity of the disturbances across firms in thesample It is obviously subject to an optimal choice of instruments where the validity ofinstruments depends on the absence of higher-order serial correlation in the idiosyncraticcomponent of the error term Therefore, a test for the second-order serial correlation isreported In this context, we also report the statistic for the Sargan test of overidentifyingrestrictions, indicating whether the instruments and residuals are independent

5 Data Description

For our empirical analysis of corporate cash holdings we use a sample of publicly traded

UK firms from 1984 to 1999 Our initial sample is the set of all firms for which data are

available on the Datastream database Datastream provides both accounting data for firms

and market value of equity The panel data set for this study has been constructed as follows

observations for any variable in the model during the sample period were dropped Finally,from these firms, only those with at least five continuous time series observations during thesample period have been chosen These criteria have provided us with a total of 1,029 firms,which represents 12,960 firm-year observations We obtain information regarding the equityownership structure from two different sources Data for the shareholdings of directors were

collected from Datastream, consisting of beneficial as well as non-beneficial directors’

holdings, in which the latter refers to holdings by directors on behalf of their families andcharitable trusts Although managers do not obtain benefit from these holdings directly, theyusually have control rights (the mean value of non-beneficial managerial holdings in oursample is 1.7 percent) Given that ownership data were only available for the period 1999-

2000 we only included data for 1999 in our analysis Data for the ultimate controllers of firmswere obtained from Faccio and Lang (2002) and merged with the managerial ownership.Consequently, we were left with 780 matched firms for our cross-sectional analysis

Table 1 presents descriptive statistics for the main variables used in our analysis It revealsthat the mean cash ratio is 10.3 percent and the median value is 5.8 percent It seems thatthese values are in line with those reported for the US firms For example, Kim et al (1998)report that the mean and median values of the cash ratio, defined as the ratio of cash plusmarketable securities to total assets, are 8.1 and 4.7 percent respectively However, in theanalysis of Opler et al (1999), the mean ratio is reported as 17 percent whereas the mediancash ratio is 6.5 percent The higher mean value in their analysis is most probably due tonormalizing cash and marketable securities by total assets minus cash and marketablesecurities rather than total assets

[INSERT TABLE 1 HERE]

As reported in Table 1, the average managerial ownership for our sample of firms is 11.9percent (the median is 3.5 percent) Table 2 analyses the ultimate controllers of UKcompanies in our sample Companies are mainly divided into those that are widely held in thesense that they have no owners with significant control rights and those that have controllingowners We report results for two different cut-off levels, namely 10 and 20 percentthresholds The reported results rely on voting rights that may differ from cash flow rights for

a variety of reasons such as pyramiding or issuing different classes of shares (for a detaileddiscussion see Faccio and Lang, 2002) Controlling owners in Table 2 are further classified

into six categories: widely held corporation, financial institution, family, unlisted companies,state, and miscellaneous.

[INSERT TABLE 2 HERE]

In Panel A of Table 2 we present percentage and number of firms controlled by differentcategories of owners at two different cut-off levels At the 10 percent level, only 24.36percent of firms are widely-held Family-controlled firms comprise 26.67 percent of firms inour sample, which makes it the most important category 20.38 percent of firms are controlled

by widely-held financial firms Another important category is the unlisted companies thatcontrol 18.72 percent of firms at the 10 percent threshold Control by widely-heldcorporations and state is trivial (0.64 and 0.13 percent respectively) As expected, the controlstructure of firms at the 20 percent threshold is significantly different than that at the 10percent level At this more conservative cut-off, 67.44 percent of companies have nocontrolling owner The characteristics of controllers are, however, similar More specifically,family is still the most important category though it decreases to 15.64 percent Financialinstitutions now control only 6.53 of firms However, the decrease in the control of financialinstitutions at the higher cut-off level is more significant than that in family control Thepercentage of companies controlled by unlisted companies also drops dramatically from 18.72percent to 6.41 percent

In Panel B of Table 2 we report summary statistics on the control rights of the largestcontrolling owner in each category of controller The findings reveal that the average values

of control rights of the largest controlling shareholder are 38.25 percent and 29.52 percentwhen firms are controlled by widely-held corporations and family respectively The averagecontrol rights is only 0.72 percent for the widely-held firms

Finally, Panel C reports descriptive statistics on the discrepancy between ownership andcontrol rights The statistics are based on firms where the largest ultimate controller owns atleast 5 percent of control rights The largest ultimate shareholder’s average ratio of cash-flow