Ownership, liquidity and investment

These effects should increase a firm’s reliance on internal funds when making investment decisions, thus resulting in an increased sensitivity of investment to cash flow.. If managers ty

Trang 1

RAND Journal of Economics

Vol 29, No 3, Autumn 1998

er levels of insider ownership, and I find some evidence that investment—cash flow sensitivities decrease slowly with insider holdings after a certain point I argue that these results are inconsistent with the hypothesis that free-cash-flow problems cause the widely noted sensitivity of investment to cash flow The results are consistent with the presence of asymmetric-information problems in the capital markets that are height- ened when managers have a strong incentive to maximize shareholder returns

1 Introduction

m A large empirical literature supports the hypothesis that liquidity—the availability

of internal funds—is an important determinant of investment Using the methodology

developed by Fazzari, Hubbard, and Petersen (1988), a number of studies demonstrate

that cash flow is a more important determinant of investment for firms that are a priori identified as the most likely to be constrained by internal funds when they invest.’ Additionally, Lamont (1997) demonstrates that capital expenditures by nonoil subsidiaries of oil firms dropped significantly following the oil price crash of 1986 This occurred despite the fact that the nonoil subsidiaries’ investment opportunity sets were not adversely affected by the price crash Taken as a whole, the articles in this literature provide substantial evidence that firms invest as if there is a perceived wedge between

the cost of internal and external funds However, the source of this difference in the

perceived cost of funds has not yet been fully explained

The difference in cost between internal and external funds is generally interpreted

as evidence that there is a premium on external funds arising from contracting and information problems as modelled by Greenwald, Stiglitz, and Weiss (1984) and Myers

* Michigan State University; hadlock @bus.msu.edu

I thank Daron Acemoglu, David T Brown, Robert Carpenter, Judy Chevalier, Steve Fazzari, Chris James, Jay Ritter, Mike Ryngaert, Jeremy Stein, Rene Stulz, seminar participants at Columbia, Florida, Michigan, MIT, NYU, Northwestern, Purdue, Texas, and the Federal Reserve Banks of Chicago and Cleve- land, two anonymous referees, and especially David Scharfstein for helpful comments and suggestions This article is a revised version of parts of my MIT Ph.D dissertation and was supported by a National Science Foundation graduate fellowship All errors remain my own

' Vogt (1994) and Carpenter (1995) find generally inconclusive evidence in their tests between these two explanations In very special samples both Blanchard, Lopez-de-Silanes, and Shleifer (1994) and Jung, Kim, and Stulz (1996) present evidence that they argue is more consistent with the managerial-overspending

argument

Trang 2

488 / THE RAND JOURNAL OF ECONOMICS

and Majluf (1984) According to this explanation, the relationship between liquidity and investment is typically a symptom of underinvestment; firms pass up some projects with positive net present value (NPV) because of the inflated cost of external funds

An alternative interpretation of the difference between internal and external funds is

not that external funds are too expensive, but rather that internal funds are too inex-

pensive from a manager’s viewpoint These agency problems are emphasized by Jensen (1986) in his free-cash-flow theory and are formally modelled by Stulz (1990) Ac- cording to this explanation, the relationship between liquidity and investment is a symptom of overinvestment; managers overspend internal funds on unprofitable projects

These two explanations of the relationship between liquidity and investment have very different implications for our understanding of both corporate financial policy and macroeconomic movements in investment It is therefore an important matter to deter- mine their relative empirical importance However, in many of the previous tests for the presence of financing constraints, these two competing explanations are observa- tionally equivalent.”

In this article I attempt to empirically distinguish between these two competing explanations of why liquidity affects investment My tests are based on examining how managerial ownership, or more generally the alignment of interests between managers

and shareholders, affects the sensitivity of investment to cash flow The tests are based

on the following intuition If the sensitivity of investment to cash flow is caused by a managerial preference to overinvest internal funds, then as managers’ interests become

more aligned with those of shareholders, they should internalize more of the financial

consequences of their overinvestment decisions This should result in a lower propensity to waste internally generated cash and thus a decreased sensitivity of investment

to cash flow If instead the sensitivity of investment to cash flow is caused by asymmetric information in the capital markets, then as managers’ interests become more aligned with those of shareholders, they should internalize more of the mispricing on external funds Consequently, managers will be more hesitant to raise external funds, and in equilibrium the premium on external funds will become even greater These effects should increase a firm’s reliance on internal funds when making investment

decisions, thus resulting in an increased sensitivity of investment to cash flow

In my empirical analysis I use managers’ ownership stakes in their firms as my measure of the alignment of interests between managers and shareholders The empir-

ical work of Morck, Shleifer, and Vishny (1988) and McConnell and Servaes (1990)

suggests that the relationship between insider holdings and the alignment of interests between managers and shareholders is not monotonic The available evidence suggests that at low levels of insider ownership, increased holdings do proxy for an increase in alignment of interests At higher levels of insider ownership, however, there is evidence

that increased holdings are not associated with increased alignment, an effect that is

generally attributed to the entrenchment properties of high levels of insider holdings

I take this possibility into account in my empirical analysis by allowing insider holdings to affect investment—cash flow sensitivities in a nonlinear way If managers typically overinvest internal cash, I expect investment—cash flow sensitivities to initially decrease with insider shareholdings and then to level off or increase as entrenchment effects set in If asymmetric information typically raises the cost of external funds, I expect to observe the opposite comparative-statics properties If capital markets are perfect and the availability of internal funds does not effect on investment choices,

2 A partial list of studies in this area includes Hoshi, Kashyap, and Scharfstein (1991), Oliner and

Rudebusch (1992), Whited (1992), Fazzari and Petersen (1993), Himmelberg and Petersen (1994), and Cal-

omiris and Hubbard (1995)

Trang 3

HADLOCK / 489 then insider shareholdings should exhibit no systematic relationship with the correlation between investment and cash flow.’

I examine the investment behavior of a large panel of Value Line-listed firms in the 1970s Previous authors have used Value Line samples from this period to examine the relationship between liquidity and investment (Fazzari, Hubbard, and Petersen, 1988) and the relationship between insider holdings and Tobin’s g (McConnell and Servaes, 1990) I find that investment—cash flow sensitivities are systematically related

to insider shareholdings My estimates imply that for a firm with no insider holdings, the sensitivity of investment to cash flow is not significantly different from zero As insider holdings increase, this sensitivity rises sharply I find evidence that these sen- Sitivities stop increasing in ownership and may in fact decrease slowly with insider holdings after holdings reach a certain level, although my confidence in the exact point this occurs is not high.t These comparative-statics properties are most significant for the firms in the sample with the highest Tobin’s q values, which are the firms that should be most affected by asymmetric-information problems I conclude that my empirical results support asymmetric-information explanations of the difference between internal and external funds and are inconsistent with both free-cash-flow theory explanations and perfect-capital-markets explanations

The article is organized as follows Section 2 discusses the theoretical motivation behind the tests, and Section 3 presents the basic empirical specification Section 4 outlines my sample-selection procedures and describes the data Section 5 presents the

empirical results, and Section 6 concludes

2 Managerial incentives and investment

m= =Free-cash-flow problems Jensen (1986) argues that when a firm’s managers have more cash than is needed to fund all of the firm’s profitable investment projects (i.e., free cash flow), there is an incentive for the managers to invest the excess cash in unprofitable projects This incentive to overinvest derives from the fact that the managers are not the owners of the firm and thus may reap personal benefits from over- investing without fully internalizing the costs of the investment decision borne by outside shareholders According to the theory, if managers did not have free cash flow, they would not invest as much, because accessing the external capital markets provides some discipline on managerial behavior.’ For a firm with free cash flow, this theory would suggest that the firm’s managers will follow an investment rule of spending a large fraction of the firm’s internally generated funds This investment behavior will naturally generate a positive correlation between investment and cash flow

Let @ denote the degree to which a firm’s managers internalize the returns to outside shareholders when making investment decisions The free-cash-flow problem arises because a is too low and, consequently, managers Overinvest Thus, according

to the theory outlined above, we would expect firms with low-a managers to exhibit

a substantial sensitivity of investment to cash flow Consider now a firm with free cash flow but with high-a managers who make investment decisions that maximize the returns to outside shareholders The managers of this firm will invest at the efficient

*Oliner and Rudebusch (1992) find that ownership is not linearly related to investment—cash flow sensitivities for a set of 84 firms My analysis suggests that their lack of significant findings may be due to the fact that they do not account for potential nonlinearities

‘In a recent study, Houston and James (1996) find evidence of a similar nonlinear relationship between

ownership and investment—cash flow sensitivities in a panel of Compustat-listed firms from 1980 to 1993

* Exactly how the external capital markets discipline a manager who desires to raise funds to overinvest

is unclear Presumably the threat of a takeover or the effect on a manager’s reputation makes accessing external capital for poor projects very costly to a manager,

Trang 4

level, and therefore they will ignore the amount of free cash flow in determining investment levels Thus, the investment behavior of a firm with high-a managers will not depend substantially on variations in cash flow These observations imply that if firms have free cash flow and their managers have a propensity to overinvest it, we should expect firms with higher-a managers to exhibit comparatively low investment— cash flow sensitivities

O Asymmetric-information problems Myers and Majluf (1984) demonstrate how

a firm may underinvest in the presence of asymmetric-information problems in the capital market A key assumption in their analysis is that managers act in the interests

of existing shareholders when making investment and financing decisions Let a denote the degree to which managers internalize the returns to existing shareholders when making a financing choice Myers and Majluf (1984) essentially assume that @ is typically quite high Dybvig and Zender (1991) have pointed out that if a is low, the underinvestment problem is alleviated My test of the asymmetric-information explanation of the relationship between investment and cash flow is based on this observation that the underinvestment problem worsens as a increases, which should imply that investment—cash flow sensitivities are increasing in a Since the exact reasoning behind this test of asymmetric-information theories is more subtle than that behind the test for free-cash-flow problems, I sketch here a simple illustrative model of investment under asymmetric information where investment is a function of both @ and cash flow The model generates the testable implication that investment—cash flow sensitivities are increasing in a

The timing of the model is as follows At date 0 the firm has assets in place, A, and realizes a cash flow from operations of c At date 1 the firm’s manager makes a financing decision by choosing a quantity e of external funds to raise from an equity issue.’ After raising the external funds, she chooses an investment level ¡ < c + e At date 2 the firm is liquidated for a sum equal to the value of the date-0 assets in place, plus the returns from the date-1 investment, plus any unused date-1 investment funds There is no discounting, and the gross returns from investing at a level i = i, at date

1 are gi, where g > 1 For simplicity I assume that investing at any level above i, generates no additional returns above gi,, and I further assume that managers are constrained to invest at or below the efficient level i,.* Finally, I assume c < /,, so that the firm needs to raise external funds to invest efficiently

All information is common knowledge except the value of the firm’s assets in place at the time of the equity-issue decision The manager knows the exact value of

A, which is either A, (a high-type firm) or A, (a low-type firm), where 4, > A, The market knows only that the firm is a high type with probability p and a low type with probability (1 — p) The financing decision at date 1 can be modelled as a signalling game Managers first announce e, the quantity of external funds they will raise The market then forms its beliefs about the firm’s type and prices equity so that a fraction s(e) of the firm must be sold in new equity to raise e I look for perfect Bayesian

6 Dybvig and Zender (1991) suggest that optimal managerial contracting can eliminate the underinvestment problem Persons (1994) argues that these contracts will be ineffective because of renegotiation possibilities

7 To highlight the role of asymmetric information, the firm in the model uses equity to finance its new investment projects Similar results would obtain if the firm issued risky debt to finance its new investment projects

8 This assumption is intended to provide an upper bound on the manager’s potential investment choice The use of i, for this upper bound is made for simplicity The analysis below should make it clear that the tradeoffs that describe equilibrium investment would be the same under more general assumptions about the region of potential investment choices.

Trang 5

YADLOCK / 491 equilibria of this signalling game that satisfy the intuitive criterion refinement condition

of Cho and Kreps (1987) The market’s belief of the probability that the firm is a high type following a fund announcement e is denoted by (e) Given a set of beliefs and

an announcement e, the market’s expected value of the firm’s assets in place following the announcement can be written as A(e|) = t(£)A, + (1 — (@))A,

To complete the model, the manager’s utility function must be specified I assume that the manager acts partially in the interest of existing shareholders, but I also assume that she has a preference for size that may affect her investment and financing decisions

To model this, when managers are making their investment decision I assume that part

of their utility payoff can be represented by a times the return to existing shareholders, where a > 0 The term a reflects the fact that compensation policies, ownership stakes, and forces in the managerial labor market tend to align the interests of managers and shareholders In addition to this component of the manager’s utility payoff, I assume the manager has a preference for size that can be represented by a utility payoff of yi when investing at a level i, where y > 0.°

Given the manager’s preferences as outlined above, it will always be the case that the manager invests all of the external funds she raises, thus i = e + c Competition

in the securities market then implies that the fraction of equity sold to raise e given a set of beliefs yw by the market must satisfy s(e|u) = e/[A(e|w) + s( + ø)] The objective of a manager of a type-j € {h, / } firm is to choose i € [c, i,] to maximize

a[l — sự — c|)][A, + gi] + +ịỉ Œ)

To derive equilibrium investment levels, it is useful to substitute i = e + c and

the expression for s(e|,2) given above into (1) The manager’s payoff when she raises

e in external funds is then

as a result of security mispricing

The investment/financing signalling game presented above is similar to the project- scaling models of Krasker (1986) and Daniel and Titman (1995) As is typical of these models, the unique equilibrium is a separating equilibrium where the low type raises funds to invest at the efficient level i,, and the high type cuts back from the efficient level and invests at some level i < i, because of the high cost of external finance In equilibrium the high type raises external funds exactly up to the point where the low type is indifferent between following her equilibrium Strategy and cutting investment

to sell overpriced shares Using (2), this point is characterized by

(A, ~ AG = ©)

° This is intended to represent the fact that the manager will typically have concerns other than maxi- mizing the returns to current shareholders An assumption that would generate identical implications is that the manager places some weight on investing at the efficient level i, to build a reputation as a manager who invests efficiently

Trang 6

cowering nner

The comparative-statics properties of this model can be intuitively derived from inspection of (3) The left side of (3) is the utility loss to the low-type manager from investing inefficiently The right side of (3) is the utility gain to the low-type manager from selling overpriced shares An increase in @ results in a larger proportional increase

in the right side of (3), because higher-a managers put relatively more weight on any security mispricing that occurs With a higher a, the manager of a low-type firm is more eager to fool the market, while the manager of a high-type firm is less willing

to sell securities to a fooled market The preceding observations establish that if a increases and i is unchanged, then the right side of (3) will be larger than its left side Thus, to restore an equilibrium following an increase in a, investment by the high-type firm (i) must decrease, and consequently 7, < 0

Equation (3) can be used to intuitively derive the fact that investment by the high- type firm will be sensitive to its internal cash flow, that is, i, > 0 To see this, note that if investment by the high type does not change following an increase in cash flow, the right side of (3) decreases This reflects that the low type’s incentive to pretend to

be the high type has decreased, since with more internal cash the low-type firm sells fewer overpriced funds when pretending to invest like the high-type firm This allows the high-type firm to increase investment up to the point where the low type is again tempted to pretend she is the high type

To establish the intuition for why the sensitivity of investment to cash flow is increasing in a for the high-type firms, consider the case where a firm’s cash flow fluctuates between some c; < i, and c, = i, For all values of a it will be the case that

i = i, when c = c, When the firm has sufficient cash to fund all positive-NPV in- vestments there is no need to go to the external markets, and thus @ plays no role in the investment decision When c = c,, investment by_the high-type firm will fall more for higher-a firms, reflecting the finding above that i, < 0 Thus for firms with cash flow fluctuating between c, and c,, investment will fluctuate more for the higher-a firms In the Appendix I prove the general result that i,, > 0

The results concerning investment in this model are summarized in the following proposition, whose proof is in the Appendix

Proposition 1 If A, > i,, then there is a unique pure-strategy equilibrium that satisfies the intuitive criterion The low-type firm raises i, — € and invests at the efficient level i, The high-type firm raises 1 — ¢ and invests i < i,, where i is given by equation (3) The following comparative-statics properties of the equilibrium investment by the high- type firm hold: i, < 0, 7 > 0, and i, > 0

3 Empirical specification

= Ownership as a proxy for a The theoretical discussion above serves to motivate the empirical tests As @ increases, investment should become more sensitive to cash flow if firms have asymmetric-information problems and less sensitive to cash flow if firms have free-cash-flow problems A natural empirical proxy for a firm’s œ is the ownership interest of its managers The value of a manager’s shareholdings is typically large relative to annual compensation, and more importantly it is generally much more sensitive to shareholder returns than compensation is (Jensen and Murphy, 1990)

Morck, Shleifer, and Vishny (1988) and McConnell and Servaes (1990) present

evidence that is consistent with increases in insider holdings proxying for increases in

a at low levels of insider holdings At higher levels of insider holdings, however, there

is evidence that increases in insider holdings may not proxy for increases in a This evidence is commonly attributed to the entrenching properties of high ownership stakes When managers control a substantial fraction of the shares in their firm, it can be difficult for the board to remove the management team or for an outside party to

Trang 7

HADLOCK / 493

successfully acquire the firm These effects can weaken the incentive of managers to maximize shareholder value by insulating them from the external forces that otherwise would direct them toward value maximization Thus, as ownership stakes enter the range where entrenchment begins to become feasible, it is possible that increased ownership actually weakens managers’ incentives to maximize shareholder returns Taking into account the above considerations, two points should be noted First, variations in ownership should be a good proxy for variations in a when ownership is below the point where any potential entrenchment effects might set in Thus empirical results based on ownership variations within the set of firms with fairly low ownership should have a clear and unambiguous interpretation Second, a can be viewed as a potentially nonlinear function of ownership At low levels of ownership, a will be increasing in ownership, but with the onset of entrenchment effects, a may increase more slowly or actually decrease with ownership Thus, ownership variations within the range where entrenchment is important are likely to have a smaller effect on investment—cash flow sensitivities than their effect when ownership is below that range, and these effects may even be of opposite sign

Regression specification The empirical tests can now be specified For firms with asymmetric-information problems, investment—cash flow sensitivities should initially increase with ownership and then level off or decrease with ownership when entrenchment effects become important For firms with free-cash-flow problems, investment— cash flow sensitivities should exhibit the Opposite comparative-statics properties with respect to ownership Finally, in a perfect capital markets world there should be no systematic relationship between ownership and investment—cash flow sensitivities

To investigate the role of ownership on investment—cash flow sensitivities, I use

the same basic approach as Fazzari, Hubbard, and Petersen (1988), and Hoshi, Kashyap,

and Scharfstein (1991) These authors regress investment on Tobin’s g, other controlling variables, and cash flow They interpret differences in the importance of cash flow between different groups of firms as evidence of financing constraints In this article, rather than grouping firms by the level of insider ownership, I exploit the continuous nature of shareholdings data by using regression terms that interact cash flow with Ownership

Since the above discussion suggests that it is likely that ownership affects investment—cash flow sensitivities in a nonlinear way, I allow the slope of the interaction of cash flow with ownership to change at the point where entrenchment effects are likely

to become important It is unclear exactly where this point is, but based on the findings

of Morck, Shleifer, and Vishny (1988) I initially pick 5% Later I experiment with other possible choices as well as a quadratic specification

With the above considerations in mind, the basic regression specification I employ is

Z = (controls) + BQ + g,Cash flow , g Cash flow, vs

Cash fl + p,—* -OwNnGs +e,

where / is investment, K is the replacement cost of the firm’s capital stock, Q is

a measure of Tobin’s g, OWNER is the percentage of equity held by insiders, OWNLS = min(5, OWNER), and OWNGS = max(0, OWNER — 5) If firms typically face asymmetric-information problems, I expect 8, to be positive and, if entrenchment effects are important, B, to be smaller than B, and possibly negative If firms have free- cash-flow problems, I expect B, to be negative and, if entrenchment effects are important, B, to be greater than Ø, and possibly positive If asymmetric-information problems

Trang 8

and free-cash-flow problems do not affect investment decisions, I expect 8; and Ø; to equal zero

¡n Endogeneity of ownership The empirical tests proposed above implicitly assume that the level of insider holdings for a firm is exogenously set and is thus independent

of the level of asymmetric-information problems or free-cash-flow problems.'° To take into account the possible endogeneity of ownership, in the analysis below I attempt to control for factors that may potentially be related to both the firm’s ownership structure and the severity of these problems However, it is still possible that insider ownership varies systematically with inherently unobservable factors related to the severity of asymmetric-information problems or free-cash-flow problems Thus it is worth considering here the potential effects of endogeneity on the interpretation of the results, as well as some relevant findings in the existing empirical literature

The issue of whether ownership is determined endogenously to optimally balance off various costs and benefits is ultimately an empirical question According to the arguments outlined above, one might expect firms with severe asymmetric-information problems to choose particularly low ownership levels, while firms with severe free- cash-flow problems would choose moderate or high ownership levels But it is not clear that actual ownership structures are in fact set optimally by taking into account the potential costs and benefits Since managers have substantial discretion over their choice of holdings, it may be difficult for outside shareholders to compel them to choose a certain ownership level

Several pieces of empirical evidence suggest that suboptimal ownership structures and incentive schemes do exist Kaplan (1989) finds evidence consistent with the change in management incentives and ownership following management buyouts resulting in significant increases in corporate performance Additionally, Slovin and Sush-

ka (1993) document significant valuation changes when firms experience exogenous shocks to their ownership structure arising from the unexpected death of a corporate insider These findings cast doubt on the hypothesis that ownership levels and management incentives are always optimally set Consequently, it would appear unlikely that ownership is frequently chosen to minimize the consequences of asymmetric-information problems or free-cash-flow problems

If ownership levels did adjust to reflect the severity of asymmetric-information problems, one would expect the firms with the worst such problems to choose low levels of insider holdings These firms would be likely to show up in the data as firms with high investment-cash flow sensitivities and low insider ownership This would bias us against finding the comparative-statics properties outlined above in the empirical test for asymmetric-information problems where I assumed exogenous ownership Despite this down- ward potential bias, the evidence below is consistent with the asymmetric-information theories Thus, if ownership adjusts to counteract asymmetric-information problems, the evidence in support of asymmetric-information theories is even stronger.''

If instead ownership levels adjust to minimize the consequences of free-cash-flow problems, one might expect the firms with the worst free-cash problems to choose moderate or high insider ownership levels These firms would then appear in the data with high investment—cash flow sensitivities and moderate or high insider ownership levels This would tend to weaken the comparative statics outlined above for testing

0 This assumption of exogenously set incentives is common in the empirical literature Jensen and Murphy (1990) and Morck, Shleifer, and Vishny (1988) are two prominent examples

Denis and Sarin (1997) find that changes in insider ownership are typically associated with changes

in top management Since asymmetric information 1s presumably related to the nature of a firm’s assets rather than the identity of its manager, their finding casts doubt on the hypothesis that ownership is set to minimize asymmetric-information problems

Trang 9

HADLOCK / 495

free-cash-flow problems under the assumption of exogenous ownership It is even possible that this behavior could generate the comparative statics predicted above for the

asymmetric-information case As the evidence below is consistent with these asym-

metric-information comparative statics, considering this potential competing explanation would appear to be particularly important

There are, however, several reasons to doubt this competing explanation First, as

discussed above, existing empirical evidence does not support the hypothesis that ownership levels are optimally determined Second, if moderate ownership firms are the ones with the most severe free-cash-flow problems, this would imply that moderate- ownership firms should have relatively low Tobin’s q values, a finding that is exactly the opposite of what the previous literature has documented Finally, this explanation predicts that insider ownership should exhibit a stronger systematic relationship to investment—cash flow sensitivities for sets of firms where free-cash-flow problems are most likely to be present, for example firms with low Tobin’s q values The evidence below comparing different sets of firms does not support this prediction

The results below should be interpreted with these concerns about the endogeneity

of ownership in mind I have argued that the possibility of endogenous ownership is unlikely to change the basic interpretation of the results However, this discussion and the results below certainly do suggest the importance of a more complete understanding

of the factors governing both optimal and actual insider ownership levels

4 Data and sample selection

= = =Sample selection The initial sample is drawn from the universe of all firms listed

in the Value Line Investment Survey for the first quarter of 1976." I chose this sample for several reasons First, previous researchers have analyzed Value Line samples from the 1970s and have identified (1) an important role for cash flow in investment (Fazzari, Hubbard, and Petersen, 1988), and (2) a systematic relationship between Tobin’s q and ownership (McConnell and Servaes, 1990) Second, there are reasons to believe that the power of the tests will be maximized by examining the mid-1970s In particular, Gertler and Hubbard (1988) and Kashyap, Lamont, and Stein (1994) identify 1974~—

1975 as a period with historically high investment—cash flow sensitivities Additionally, Gertler and Hubbard argue that financial constraints are countercyclical Finally, there were fewer hostile control transactions during this period than in the 1980s Since an active market for corporate control may limit a manager’s propensity to overinvest, if investment—cash flow sensitivities are driven by free-cash-flow problems, it is more likely that this relationship could be detected in the 1970s."3

Since I have only a single-year cross section of ownership data, investment regressions estimated over a long period will use unreliable ownership data However, estimates derived over a short period are likely to be imprecise As a compromise, I use a four-year period in most of the results reported below The results using three- year and five-year windows are similar and are discussed briefly below While the Value Line ownership data were published in early 1976, the original source of this data was primarily the 1975 corporate proxy statements issued in the spring of 1975 Thus I estimate investment regressions over the period 1973-1976, which is roughly centered around the time that the ownership data should be most accurate

'? The Value Line Investment Survey is a well-known investment advisory publication that publishes reports and recommendations on a broad-based set of publicly traded firms As of 1979, the Value Line universe of firms represented 96% of trading volume on all U.S stock exchanges (see Bernhard, 1979)

'3 Although it should be noted that Houston and James (1996) find similar results to those reported

below for their sample over the period 1980-1993

Trang 10

cea HNL TNT AEE NIST ALR A ST

I restricted the initial Value Line sample to firms incorporated in the United States

that were traded on the NYSE, AMEX, or over the counter I also excluded regulated

firms, financial firms, and firms with ambiguous or missing data on insider ownership

I required sample firms to be included on the 1984 Compustat primary-supplementary-

tertiary, over-the-counter, or research tapes, and I required the Compustat variables

necessary for the regression analysis to be available for every year from 1973 to 1976

Firms involved in significant merger and acquisition activity over the sample period

were eliminated.'* Finally, to eliminate outliers in the independent variables, all firms

where Q (tax-adjusted Tobin’s q), cash flow/K, or lagged sales/K were in the top or

bottom percentile of the sample in any firm-year were eliminated.'> After these selection

procedures, the final sample consisted of a balanced panel of 435 firms for the years

1973-1976 Of these 435 firms, 318 (73%) were in manufacturing industries (SIC codes

2000-3999)

6 Summary statistics The variable OWNER is defined to be equal to the total

percentage of equity held by individuals that Value Line identifies as officers, directors,

management, or insiders, plus all significant family holdings.'® As reported in Table 1,

the median level of insider holdings is 10% and the mean is 17.03% Table 2 presents

summary statistics for the entire sample and for subsamples based on insider-holdings

quartiles.'7 Several of the variables are normalized by K, the beginning-of-year replace-

ment cost of the firm’s physical capital One variable that is clearly related to insider

ownership is firm size, as measured by the book value of assets at the beginning of

1973 The median firm in the lowest insider-ownership quartile is more than three times

as large as the median firm in any other ownership quartile Since firm size could proxy

for the severity of informational asymmetries or free-cash-flow problems, in some spec-

ifications below I control for size to check that the results do not simply reflect the

correlation between insider holdings and size

Interpreting differences in the summary statistics on the reported variables as ev-

idence for or against the presence of financial constraints is problematic Kaplan and

Zingales (1997) suggest that high cash balances indicate that a firm is likely not to be

financially constrained However, Calomiris and Himmelberg (1995), Calomiris, Him-

melberg, and Wachtel (1995), and Houston and James (1996) all present evidence

suggesting that firms that are a priori more likely to be financially constrained tend to

hold larger stocks of cash and working capital Since the relationship between financing

costs and the levels of endogenous financial variables such as cash balances and lev-

erage is not clear, inferences based directly on the summary statistics would appear to

be of limited use

14 | eliminated firms where in any year expenditures on acquisitions were greater than 10% of the firm’s

assets or where the Compustat footnotes indicated that the firm’s sales reflected acquisition activity Addi-

tionally, I examined the Moody's manuals for firms where equity outstanding in any year changed by 10%

and eliminated the firm if the change was related to a merger, acquisition, or spinoff

15 “Cash flow” is net income after extraordinary items minus preferred dividends plus depreciation plus the annual change in deferred taxes “Lagged sales” are sales in the year prior to the observation year The

capital stock (K) and tax-adjusted Tobin’s g (Q) are beginning-of-the-year figures calculated using the al-

gorithm presented by Salinger and Summers (1983) with minor modifications

16 Since outside directors generally hold small stakes (Denis, Denis, and Sarin, 1997), OWNER should

be a good proxy for the holdings of insiders For firms where Value Line identified insider holdings as “less than 1%,” I assigned a value to OWNER of 56% since this was the mean insider holdings for a subset of these firms that appear in Hermalin and Weisbach (1991)

‘7 Investment is gross annual expenditures on property, plant, and equipment Details of the construction

of K and Q are available from the author Tax-unadjusted Tobin’s q is defined to be (market value of common stock + book value of preferred stock + book value of long-term debt + current liabilities — current assets)/

(book value of all assets — current assets)

Trang 11

directors, management, or insiders, as

well as family holdings of board mem- bers The median of OWNER is 10 and the mean is 17.03

by Fazzari, Hubbard, and Petersen (1988) for their class-3 firms It would be premature

to interpret this coefficient as evidence of liquidity constraints, for the standard reason that cash flow may be proxying for the quality of the firm’s investment opportunities over time This is why I emphasize below the difference in estimated cash-flow sensitivities for firms with varying levels of ownership

Column 2 includes terms interacting cash flow with insider holdings as in the specification outlined in (4) above I refer to this as the baseline specification.'® The estimated coefficient on cash flow of —.022 is small and insignificantly different from zero This implies that for a firm with no insider holdings, investment is insensitive to cash flow The estimated coefficient on cash flow interacted with QWNL5 of 065 is significant at the 1% level and indicates that as insider holdings increase from zero, the sensitivity of investment to cash flow increases rapidly The estimate implies that

a firm with 5% insider holdings invests 33 cents more per dollar of internally generated cash than a firm with no insider holdings

The coefficient on cash flow interacted with OWNGS5 of —.0057, which is signif-

icant at the 1% level, implies that the sensitivity of investment to cash flow decreases slowly with insider holdings when insider holdings are greater than 5% The estimates imply that as a firm’s insider holdings increase from 5% to 25%, the firm’s cash flow coefficient will decrease from 307 to 193 A test of the difference in the estimated

'SIf cash flow is interacted linearly with ownership without separate terms for ownership below and above the 5% level, the coefficient on the ownership variable is small, negative, and insignificant This result

is similar to the findings of Oliner and Rudebusch (1992)

Tiêu đề	Ownership, Liquidity, and Investment
Tác giả	Charles J. Hadlock
Trường học	RAND
Chuyên ngành	Economics
Thể loại	Journal Article
Năm xuất bản	1998

Định dạng
Số trang	22
Dung lượng	1,38 MB