Corporate Governance, Internal Financing and Investment Policy: Evidence from Anti-takeover Legislation

Transcription

1 Corporate Governance, Internal Financing and Investment Policy: Evidence from Anti-takeover Legislation Bill Francis, Iftekhar Hasan, Liang Song * Lally School of Management and Technology of Rensselaer Polytechnic Institute 110 8th Street - Pittsburgh Building, Troy, NY, U.S.A., First Draft: Jan This Draft: Jan Abstract Much of our understanding builds on the idea that firms investment policy depend on their financial position since there are financing frictions caused by agency problems and/or asymmetric information problems. But how agency problems individually influence the relation between internal financing and investment? This paper uses variation in corporate governance generated by state adoption of antitakeover laws to empirically isolate the effect of agency problems on the relation between internal financing and investment. We use differences-in-differences approach and find robust evidence of a rise in investment-cash flow sensitivity when managers are insulated from takeovers. These results are consistent with agency theory that incentive problems and costly monitoring of managerial actions can cause financing frictions and further affect investment-cash flow sensitivity. The results of firms less access to external capital market and decreased payout after passage of laws also support this view. Keywords: corporate governance, financing frictions, investment- cash flow sensitivity * Please address correspondence to Liang Song (songl@rpi.edu), Lally School of Management and Technology of Rensselaer Polytechnic Institute, 110 8th Street - Pittsburgh Building, Troy, NY, U.S.A., 12180, Phone: (518)

2 Corporate Governance, Internal Financing and Investment Policy: Evidence from Anti-takeover Legislation Abstract Much of our understanding builds on the idea that firms investment policy depend on their financial position since there are financing frictions caused by agency problems and/or asymmetric information problems. But how agency problems individually influence the relation between internal financing and investment? This paper uses variation in corporate governance generated by state adoption of antitakeover laws to empirically isolate the effect of agency problems on the relation between internal financing and investment. We use differences-in-differences approach and find robust evidence of a rise in investment-cash flow sensitivity when managers are insulated from takeovers. These results are consistent with agency theory that incentive problems and costly monitoring of managerial actions can cause financing frictions and further affect investment-cash flow sensitivity. The results of firms less access to external capital market and decreased payout after passage of laws also support this view. Keywords: corporate governance, financing frictions, investment-cash flow sensitivity 2

3 I. Introduction How financing frictions influence real investment decisions is a central point in contemporary finance research (Stein 2003). There are two explanations of the difference in cost between firms internal and external funds: asymmetric information problems as modeled by Greenwald et al.(1984) and Myers and Majluf (1984); agency problems proposed by Jensen (1986) and formally modeled by Stulz (1990). With imperfect information about the quality or riskiness of the borrowers investment projects, adverse selection leads to a gap between the costs of external financing in an uninformed capital market and internally generated funds. In the presence of incentive problems and costly monitoring of managerial actions, external suppliers of fund to firms require a higher return to compensate them for these monitoring costs and the potential moral hazard associated with managers control. And overinvestment by entrenched managers may place a discount on internal funds and overspend by undertaking even negative NPV projects because managers may derive more private benefits, which will further increase the difference in the cost between internal financing and external financing from the management s point of view. The existing literature provide substantial evidence that firms invest as if there is a perceived wedge between the cost of internal and external funds (see Hubbard (1998) for a survey). However, research is still needed to isolate the source of capital-market imperfections that influence firm investment decisions (Hubbard 1998) since these two theories have very different implications for our understanding of both corporate financial policy and macroeconomic movements in investment. We want to add to these kinds of literatures and isolate the effect of agency problem on financing friction by 3

4 introducing an exogenous measure of variation in corporate governance, passage of antitakeover laws to estimate how corporate governance affects the relation between liquidity and investment. Our results support the view by Jensen (1986), that agency problems is a very important determinant of capital market imperfection and further influence the relationship between liquidity and investment. Specially, we use passage of antitakeover laws as our measure of variation in corporate governance. If managers fear a hostile takeover and the resulting job loss, they may more closely pursue shareholder interests. After states adopt these laws, one of important corporate governance components, takeover threats decline and agency problems increase. we use the methodology developed by Fazzari et al.(1988) and differences-in-differences approach to get the results that investment-cash flow sensitivity statistically and economically increases in firms incorporated in the states, where anti-takeover laws passed, compared to our control group, firms incorporated in the states, where anti-takeover laws did not pass. We also find other supportive results of firms less access to external capital market and less payout after firms are insulated takeover. Our research extends the emerging empirical literature on the relationship between capital-market imperfections and investment (see Hubbard (1998) for a survey), pioneered and developed by Fazzari et al.(1988). They argue that the sensitivity of investment to internal funds should increase with the wedge between the costs of internal and external funds. Accordingly, one should be able to gauge the impact of financing constraint on corporate investment by comparing the sensitivity of investment to cash flow across samples of firms sorted on proxies for financial constraints, such as payout 4

5 policy, size and credit ratings. A number of recent articles, however, question this approach. Kaplan and Zingales (1997) argues that such an exercise is valuable only if the investment-cash flow sensitivity is monotonically increasing with respect to the degree of financing constraints and they are only able to identify constrained firms and not firmyears. This paper does not rely on an assignment of firms into financial constrained or not constrained categories according to a priori measure of financing constraints and cross-sectionally compare the investment-cash flow sensitivity. Because we focus on the time-series component by testing how investment-cash flow sensitivity will change after one important and direct determinant of financing friction, agency problem became worse within firms, we are less concerned about this issue. Second, we argue that investmentcash flow sensitivities can be used to gauge the effect of financing frictions on investment instead of the effect of financing constraints on investment. Finally, the time-series nature of our methodology also alleviates other Kaplan and Zingales concern that some risk-averse managers make some savings to prevent some future accidents. Erickson and Whited (2000),Gomes (2001) and Alti (2003) show that evidence of higher investment-cash flow sensitivities for constrained firms has been ascribed to measurement and interpretation problems with regressions including Q since the cash flow coefficient contains information about investment opportunities and the level of investment-cash flow sensitivities will be biased upwards for firms seen as constrained even in the absence of financing frictions. Because we use differences-in-differences approach and focus on the coefficient representing the difference of investment-cash flow 5

6 sensitivity between our sample group firms and control group firms, any bias that is systematically related to our variables will be differenced out. Our study is not the first attempt as testing the investment-liquidity relation within a specific corporate governance framework. Some articles,e.g. Kathuria and Mueller (1995), Hadlock (1998), Schaller (1993) for Canada, Degryse and de Jong (2006) for the Netherlands, Goergen and Renneboog (2001) for the U.K., Vogt (1994) and Cho (1998) for the U.S., use ownership concentration, firm size and so on to proxy agency problems. However, one problem associated with these researches is the potential endogeneity problems. Firms with better and worse corporate governance probably also differ on other, unobservable, dimensions since changes in governance within a firm may be accompanies by other unobservable changes (Bertrand and Mullainathan 2003). This problem will be even more severe in the context of testing relationship between investment and internal financing since Q may be related to ownership structure or firm size and definitely influence investment-cash flow sensitivity. And these agency problems measures are also related to the level of asymmetric information, so they can not really isolate the effect of agency problems on financial frictions and keep the level of asymmetric information problems constant. We attempt to deal with this endogeniety problem by using an exogenous measure, the passage of antitakeover laws to capture changes in corporate governance, which is also used in some recent articles (Bertrand and Mullainathan 1999; Garvey and Hanka 1999; Bertrand and Mullainathan 2003; Cheng, Nagar et al. 2005). After states adopt these laws, one of important corporate governance components, takeover threats decline and agency problems increase. At the same time, firms Q did no change; neither did the 6

7 level of asymmetric information. So we can isolate the effect of agency problems on financing frictions to see how agency problems influence the relation between internal financing and investment. Specifically, we use exogenous variation in corporate governance in the form of the second-generation anti-takeover laws passed between 1984 and 1991 on a state-bystate basis 2. By reducing management s fear of a hostile takeover and subsequent job loss, such laws weaken corporate governance and create more opportunity for managerial slack. In this paper, we empirically test the interplay between corporate governance and investment-cash flow sensitivity based on differences-in-differences approach using a sample of manufacturing firms drawn from COMPUSTAT between 1981 and Our investment equations resemble those of Fazzari et al. but include an interaction term that captures the effect of corporate governance on investment-cash flow sensitivities. In this paper, we find robust evidence of a rise in investment-cash flow sensitivity when managers are insulated from takeovers. When we control the level of asymmetric information, proxied by analyst forecast dispersion and other firm-level corporate governance measure, the number of analyst forecast 3, out results do not change qualitatively. Second, we use issue data form Securities Data Corporation (SDC) to examine the actual access of external capital markets before and after passage of antitakeover laws and find that firms tend to decrease their access of external capital markets following passage of antitakeover laws. Specifically, we find that the number of equity issues per firm year and amount raised scaled by market value per firm year decrease after state 2 More detailed description about anti-takeover laws can be found in several studies: Bertrand and Mullainathan (1999, 2003); Garvey and Hanka (1999); Cheng, Nagar et al. (2005). 3 We can not find other firm-level corporate governance measure since our sample period is

8 adoption of antitakeover laws. Garvey and Hanka (1999) s result that long term debt scaled by total assets decrease after firms are insulated take over also supports the view that firms less likely access to external capital markets after passage of antitakeover laws because of increased difference between internal financing and external financing caused by external fund suppliers required compensation for their monitoring costs and the potential moral hazard associated with managers control. Finally, we examine firms payout policy around the adoption of antitakeover laws and find that firms issue fewer dividends and also decrease their total payout including dividend and repurchase. This is consistent with the view that overinvestment by entrenched managers may place a discount on internal funds and managers will more internal capital and make less payout. The article is organized as follows: Section II develops hypotheses about relation between agency problems and investment-cash flow sensitivity. Section III describes the second-generation antitakeover laws. Section IV describes and summarizes our data. Section V reports the methodology and results. Section VI concludes. II. Determinants of Financing Frictions As we have discussed above, there are mainly two factors that influence how a firm s investment policy depends on its financial position: asymmetric information problems and agency problems. A. Asymmetric Information Problems 8

9 Asymmetric information may lead to the rejection of good investment opportunities because external financing may be deemed overly expensive by the management (Myers 1984; Myers and Majluf 1984). Since the management has superior information about the firm compared to outside investors and the market is less wellinformed about the firm s or the project s quality, the outside investors will demand a premium on the capital provided based on the level of asymmetric information. This mechanism may lead to adverse selection problems among the firms, who want to get external financing. Further, it will cause financing frictions, i.e. there is a difference in the costs between internal financing and external financing. As a result, a fraction of good investment projects which are not profitable enough to compensate for the excessively high cost of external financing are foregone. This is so called underinvestment problem. B. Agency Problems Another source of the investment-cash flow sensitivity is caused by agency problems (Jensen 1986; Stulz 1990). Corporate managers interests may not be perfectly aligned to the interests of shareholders as the private benefits derive form managing firms has been shown to be an increasing function of the corporation s size. So, management s corporate objective may be empire building rather than maximization of firm value. As a consequence, negative present value investments could also be undertaken. This is so called overinvestment problem. And overinvestment by entrenched managers may place a discount on internal funds and overspend by undertaking even negative NPV projects because managers may derive more private benefits, which will further increase the 9

10 difference in the cost between internal financing and external financing from the management s point of view. Another effect of agency problems is that in the presence of incentive problems and costly monitoring of managerial actions, external suppliers of fund to firms require a higher return to compensate them for these monitoring costs and the potential moral hazard associated with managers control. This has the same effect as asymmetric information and caused a wedge between internal financing and external financing. This paper uses variation in corporate governance generated by state adoption of antitakeover laws, which is an exogenous event, to empirically isolate the effect of agency problems on the relation between internal financing and investment. Since passage of laws is exogenous (Bertrand and Mullainathan 1999; Garvey and Hanka 1999; Bertrand and Mullainathan 2003; Cheng, Nagar et al. 2005), this measure did not influence the level of asymmetric information and other factors, which may affect investment-cash flow sensitivity. So we can make reasonable hypotheses that how agency problems will influence the relation between internal financing and investment: Hypothesis 1: After managers are insulated from take over, firms investment-cash sensitivity increase. If this increase in sensitivity stems from more difficulty to access to the capital markets, which is the second effect of agency problems on the relation between internal financing and investment, then firms may decrease their access of the capital markets following state adoption of antitakeover laws. 10

11 Hypothesis 2: After managers are insulated from take over, firms decrease access to external capital market. A lot of articles (Fazzari, Hubbard et al. 1988) use the level of dividend payout as an indicator whether firms are financial constrained or not, less payout means that firms are more likely financial constraint. According to the first effect of agency problems on the relation between internal financing and investment, overinvestment by entrenched managers may place a discount on internal funds. So the managers will keep more internal capital and make less payout. Hypothesis 3: After managers are insulated from take over, firms decrease payout. III. Antitakeover Laws More detailed description about anti-takeover laws can be found in several studies (Bertrand and Mullainathan 1999; Garvey and Hanka 1999; Bertrand and Mullainathan 2003; Cheng, Nagar et al. 2005). We will give a brief discussion as follows. In the midto-late 1980s, several states enacted antitakeover legislation, which increased the difficulty of a successful takeover for firms incorporated in those states. These laws are referred to as second-generation anti-takeover legislation and bring the firms incorporated in those states an exogenous event (Bertrand and Mullainathan 2003). The second generation laws can be divided into three types: (1) Control Share Acquisition (CSA), (2) Fair Price (FP) and (3) Business Combination (BC). Although these laws have some differences in various states, they share some similar themes. Control Share Acquisition laws entitle shares not held by the bidder firms the right to 11

12 decide whether the bidder s shares may vote on the take over. The bidder must disclose its identity, intent and terms of acquisition, basing upon which, other share holders can vote to entitle voting rights to the bidder s shares. Fair Price laws rule that the bidder must pay a fair price for the shares acquired in the take over action. This price will be calculated by a specific formulae and impede the bidders to pay a high price for shares. The Business Combination laws forbid a suitor from takeover activities for a specified number of years, unless with the target firm s board permission. About the degree of various types of antitakeover laws impact, Karpoff and Malatesta (1989) find that investor reaction was the most negative to the announcement of the passage of Business Combination. Based on that, Bertrand and Mullainathan (1999, 2003) argue that Business Combination laws were the most stringent of the three laws. Another point comes from Cheng, Nagar et al. (2005), they argue that the passages of various anti-takeover laws were stimulated by one another and further investors will have anticipated the passage of next law upon knowing the passage of the previous law. So they believe it is the first law that has the most influence on investors since the enactment of subsequent laws in a certain state is eased by the passage of the first piece of antitakeover laws. In our empirical section, our test gets the qualitatively same results under these two different approaches. IV. Data A. Sample 12

13 Our data sample is Compustat database and covers U.S manufacturing firms 4. Whited (1992) argues that the investment behavior in other sectors may differ substantially from the manufacturing sector. For example, government regulation influences the public utilities, transportation, and farming industries; and the financial service industries often use different accounting procedures. Since the first secondgeneration antitakeover law passed in 1983 and the last one passed in 1991, we extend our sample coverage two years before and three years after that period and focus on the period from 1981 to 1994 to include all possible related influence by antitakeover laws 5. We use capital expenditures (item 128) to measure investment. Cash flow is measured as the sum of earnings before extraordinary items (item 18) and depreciation (item 14). Sales is measured as sales (net) (item 12) to measure and cash is measured as cash (item 162). Investment, cash flow, sales and cash are all deflated by capital, which is measured as net property, plant and equipment (item 8) at the beginning of the fiscal year. Average Tobin s Q is measured as the market value of assets divided by the book value of assets (item 6), where the market value of assets equals the book value of assets plus the market value of common equity less the sum of the book value of common equity (item 60) and balance sheet deferred taxes (item 74). Tobin Q is calculated at the beginning of a firm s fiscal year. A firm s state of incorporation is also collected from Compustat, which refers to the state of incorporation in 1994 if the firm is still running or the last state of incorporation if the firm was dead before Bertrand and Mullainathan (2003) argued that anecdotal evidence showed that changed in state of incorporation are very 4 Two-digit SIC code between 20 and 39 inclusive. 5 In our robustness test, we change our sample period and our results do not change qualitatively. 13

14 rare, especially for the very large sample. They also verify that by checking a number of randomly selected firms. We exclude firms with Q is in excess of 10 from our sample since these firms are more likely subject to big corporate events like M&A and Abel and Eberly (2001) argued that linear investment equations at high levels of Q are poorly fitted. Firms for which the required Compustat data are not available in the entire sample period are also excluded. The first advantage to get a balanced sample is that we can get a stable data series and do not need to suffer the effect of some big corporate events like delisting or IPO on firms investment behavior. The second advantage is that we have less concern about influence of firms changes in state of incorporation since all firms in our sample are still running in The last advantage is that we implicitly eliminate very small firms, those for which linear investment models are likely in adequate (Gilchrist and Himmelberg 1995). We acknowledge that we can not avoid selection bias problem and leave it in the future research. In total, there are 2786 observations by 199 firms ranging from in our balanced sample. To control the level of asymmetric information, we get analysts current-fiscalyear EPS forecasts data form IBES and define AnalystForecastDispersion as the standard deviation of analysts current-fiscal-year EPS forecasts scaled by the absolute value of the mean forecast. To make our results more robust, we also define AnalystForecastDispersion as the standard deviation of analysts current-fiscal-year EPS forecasts scaled by fiscal year share price and our results don t change qualitatively. Because of our sample period, we can not find other firm-level corporate governance measure and the only one available is the number of analyst forecasts. More analyst 14

15 forecast means higher outside monitoring effect. We define the variable AnalystForecastNumber as the number of analyst forecast for a certain firm. Due to data limitation, we can only match half of our sample firms with these two variables. B. Summary Statistics Table I presents various antitakeover law passage times in various states, which is from Bertrand and Mullainathan (2003). We will focus long-run difference in investment-cash flow sensitivity pre- and post-passage unlike event study, so precise day of passage is not necessary. The distribution of our firms across states is shown in Table II. As shown, our sample is not equally distributed across states, 44 percent of firms are incorporated in Delaware. Insert Table I about here. Insert Table II about here. Table III presents descriptive statistics for regression variables used in our analysis. Our sample firms have mean capital million dollars and median million dollars, so they are not small firms. Average Q is 1.45, which is in the range between the reported Q of 1.2 by Kaplan and Zingales (1997) and the reported average Q of 1.5 by Polk and Sapienza (2002). Investments in our sample firms are all positive. Insert Table III about here. Table IV compare the mean of variables used in our regression analysis for companies incorporated in the states, which finally passed anti-takeover laws, before and after state adoption of antitakeover legislation. The result shows that average Q pre- and after-passage of antitakeover laws do not statistically change. Neither do sales and the 15

16 level of asymmetric information, AnalystForecastDispersion. This confirms that passage of antitakeover law is exogenous and it did not influence firms investment opportunity and asymmetric information problems. The result also shows that firms reserve more cash after management are insulated from antitakeover, which is consistent with agency theory that managers want to control more cash inside the firm. The number of analyst forecast increases a little bit and we will control it in our formal regression. Insert Table IV about here. V. Methodology and Result A. Differences-in-Differences Approach To examine the change in the sensitivity of investment-free cash flow before and after passage of antitakeover laws, we use the same differences-in-differences methodology as Bertrand and Mullainathan (Bertrand and Mullainathan 1999; 2003), which is very popular in labor economics and in some sense like event study methodology used in finance and accounting. The advantage of this methodology is that our control group for any given year is the set of states, which did not pass antitakeover laws at that time, even if they will pass the law later, and make our control group is not very small since it is very clear that most of companies are incorporated in the states passing laws from the table II. Our regressions are based on the methodology used in Fazzari et al.(1988), among others. To accommodate our research demand, we adjust the model in certain ways described below to test pre- and post-antitakeover laws effects. The regression equation has the following form: 16

17 I K i, t i, t 1 = a + BQ 1 i, t 1 + B 2 CF K i, t i, t 1 + B Treat 3 CF CF * Postlaw * Treat Sales Cash i, t i, t i, t i i, t 1 i, t 1 i * + B4 + B5 + B6 + B7 PostLawi, t * Treati + Ki, t 1 Ki, t 1 Ki, t 1 Ki, t 1 e i, t The dependent variable is investment ( ) scaled by, capital of the I i, t K i, t 1 preceding year. The independent variable and expected relationship with the dependent variables are: 1. Average Tobin Q ( ) is to control firms investment opportunities. Q i, t 1 Classical investment theory predicts a positive relation between Tobin Q and investment if Q correctly measures the firm s investment opportunities and if the firm invests according to these investment opportunities. 2. Free cash flow( CF i, t ) scaled by capital: Fazzari et al. (1988) argue that if we control for investment opportunities and there is a costly access to external capital markets, then there will be a positive relation between internally generated cash flow and investment. 3. Treat i is a dummy variable equal to one if a firm is incorporated in a state, which passed anti-takeover laws, otherwise zero. We do not individually include this variable into our regression equation since the firm fixed effect can naturally include that. 4. Free cash flow ( CF i, t ) multiplied by Treat i is to capture the difference in investment-cash flow sensitivity across our treatment group firms, firms in antitakeover states and our control group firms, firms in not antitakeover states without the effect of passage of antitakeover legislation.. 17

18 5. PostLaw i, t is a dummy variable equal to one after passage of the antitakeover law for a certain company and zero otherwise. In this paper, we use two different ways to define this dummy variable. First, Bertrand and Mullainathan (1999, 2003) use the passage of Business Combination laws to represent anti-takeover law passage time since they argue that Business Combination laws were the most stringent of the three laws. In this way, we define the dummy variable PostLaw i, t for each firm that takes a value of unity in the year when Business Combination laws has passed and the results presented in this paper are based on this approach. Second, Cheng, Nagar and Rajan (2004) use the passage of the first antitakeover law to represent antitakeover law passage time since they argue that the passages of various anti-takeover laws were stimulated by one another and further investors will have anticipated the passage of next law upon knowing the passage of the previous law. So they believe it is the first law that has the most influence on investors since the enactment of subsequent laws in a certain state is eased by the passage of the first piece of anti-takeover laws. In this way, we define the dummy variable PostLaw i, t for each firm that takes a value of unity in the year when the corresponding state s first antitakeover law has passed and we use this way as a robustness test. 6. Free cash flow ( CF i, t ) multiplied by the product of PostLaw i, t and Treat i is our key variable, which only capture the effect of passage of antitakeover laws on investment-cash flow sensitivity since the variable free cash flow ( CF i, t ) 18

19 multiplied by Treat i has captured all other difference between our treatment group firms and control group firms. According to our hypotheses, this variable should be positive, which means that firms investment is more sensitive to internal cash flow after states adopted antitakeover laws. 7. Sales ( Salesi, t 1 ) divided by capital ( K i, t 1 ) is a proxy for production to control for a possible accelerator effect. Hoshi and Kashyap (1991) and Lins et al. (Lins, Strickland et al. March, 2005) argue that liquidity variable might be a proxy for production effects that make no sense in empirical investment research if the variable sales is excluded from the regression equation. 8. Cash ( Cashi, t 1 ) relative to capital ( K i, t 1 ) is included in the regression since we want to control the effect that the sensitivity of investment to cash flow is likely to be lower if the firm has a lot of financial slack. We estimate a firm and year fixed-effect model and report p-values based on robust (white) standard errors that also incorporate clustering around each firm to account for a short of independence between the time series-observations within a certain firm. B. Investment-Cash Flow Sensitivity As we have discussed, the most important for our baseline analysis is the coefficient on the interaction term among free cash flow, PostLawdummy variable and Treat dummy variable, which control antitakeover law effect on investment to cash flow sensitivity. From Column 1 of Table IV, we find a large positive and significant coefficient on this interaction term, which means after passage of antitakeover laws, investment sensitivity of cash flow significantly increased. Moreover, this result is not 19

20 only statistically significant but also economically significant since the coefficient on this interaction is very large compared to the coefficient in front of cash flow and the sum of the coefficient of cash flow and that coefficient of the interaction between treat and cash flow. To make our results more robust, we use other regression specifications and our results do not change qualitatively. In column (2) we removed Beginning-of period Q as an independent variable from the regression leaving cash flow as the only independent variable following Kaplan and Zingales (1997); in column (3) we include the End-of- Period Q into our regression equation, which include the information from the cash flow during current period; in column (4) and (5), we add lagged cash flow since Hoshi, Kashyap, and Scharfstein (1991) argue that cash flow during the current period cash flow might reflect some investment opportunity information not included within Beginning-of- Period Q. Insert Table V about here. To control the effect of asymmetric information on investment-cash flow sensitivity, we add an interaction term; analyst forecast dispersion multiplied by cash flow and redo our baseline regression. Higher analyst forecast dispersion means higher level of asymmetric information. Our results in Table VI do not change quantitatively and become more significant. To control the effect of other firm-level corporate governance on investment-cash flow sensitivity, we add another interaction term; analyst forecast number multiplied by cash flow in Table VII. More analyst forecasts mean higher outside monitoring effect. Our results do not change either. Insert Table VI about here. 20

21 Insert Table VII about here. In table VIII, we present the baseline regression results on the effects of passage of laws on investment-cash flow sensitivity with more different specifications and our results are still robust 6. In column (1), all variables are winsorized to make sure our results are not affected by outliers. In column (2), only firms in antitakeover states are included to avoid firms in the states, where antitakeover laws did not pass, have different characteristics compared to firms in antitakeover states. In column (3), firms in Texas and California are excluded since during the sample period, shocks to the oil and defense industries drastically influence the performance of firms in Texas and California. In column (4) and (5), we change sample period to and to make sure our sample selection did not influence our results. Insert Table VIII about here. To summarize, we find robust evidence of a rise in investment-cash flow sensitivity when managers are insulated from takeovers even if when we control for the level asymmetric information and other corporate governance measure. These results are consistent with agency theory that incentive problems and costly monitoring of managerial actions can cause financing frictions and further affect investment-cash flow sensitivity. C. Access to External Capital Markets If this increase in sensitivity stems from more difficulty to access to the capital markets, then firms may decrease their access of the capital markets following state 6 Since bulk of firms (44 percent) is located in Delaware, we also need to test in the sample firms that are not incorporated in Delaware. However, our small sample size can not afford so drastic reduction to test change of investment-cash flow sensitivity. So we leave it in the future research. 21

22 adoption of antitakeover laws. To investigate this hypothesis, we first examine the equity issuance patterns of our sample firms, who incorporate in the states, where finally passed antitakeover laws. We gather the issue date, the dollar amount raised in seasoned equity offerings for our sample firms from SDC. Table IX shows decreased access of capital markets in the post-passage of antitakeover laws period and the trend tends to be quite pronounced and very significant based on t-test. Specifically, we find that the number of equity issues per firm year and amount raised scaled by market value per firm year decrease after state adoption of antitakeover laws. Insert Table IX about here. We also try to investigate the debt issuance patterns. Garvey and Hanka (1999) come to a conclusion that firms protected by "second generation" state antitakeover laws substantially reduce their use of debt, and that unprotected firm s do the reverse. which is presented in Figure 1. We can see an obvious trend of debt deduction as the passage of antitakeover laws after Since outside investors need some time to negotiate about the contract of long-term debt with firms, the change of long-term debt can not happen at once, so we can only see the decreasing trend in long-term debt after This result supports the view that firms less likely use external debt markets after passage of antitakeover laws because of increased difference between internal financing and external financing caused by external fund suppliers required compensation for their monitoring costs and the potential moral hazard associated with managers control. Insert Figure 1 about here. D. Payout Policy 22

23 According to the first effect of agency problems on the relation between internal financing and investment, overinvestment by entrenched managers may place a discount on internal funds. So the managers will keep more internal capital and make less payout. We examine firms payout policy around the adoption of antitakeover laws in Table X using differences-in differences approach. The dependent variable is dividend to earning ratio (dividends-common (item 21) / income before extraordinary (item 18)) and payout to earning ratio ((dividends-common (item 21) + purchase of common and preferred stock (item 115)) / income before extraordinary (item 18)). We also control Profitability [income before extraordinary items (item 18) +interest expense (item 15) + deferred taxes (item 50) if available]/ total asset (item 6), Growth Rate of Assets (Asset t -Asset t-1 )/Asset t and Asset t is total asset (item 6) and Firm Size (the percent of NYSE firms with the same or lower market capitalization), which are also used in Fama and French (2001). To construct dependent variable, firms with negative earning years are excluded. We find that firms issue fewer dividends and also decrease their total payout including dividend and repurchase. This is consistent with the view that overinvestment by entrenched managers may place a discount on internal funds and managers will keep more internal capital and make less payout. Insert Table X about here. VI. Conclusion The evidence we uncover in this article is strongly consistent with a link between financing frictions and corporate investment. As we hypothesize, we find that while agency problems increase, proxied by states adoption of antitakeover legislation, 23

24 investment-cash flow sensitivities statistically and economically increase. We also find that result of firms less access to external capital market and decreased payout after passage of laws. These results are consistent with agency theory that incentive problems and costly monitoring of managerial actions can cause financing frictions and further affect investment-cash flow sensitivity. Our results do not attempt to imply agency problems are the only determinant of financing friction. Asymmetric problems also have some explanatory power on the relation between internal financing and investment and we leave it in the future research. Our results also have important policy implications. There are several ways to improve firms financing and investment environment. One of them is to improve firms own corporate governance to reduce managers overinvestment problems and decrease financing frictions. 24

25 Figure 1 Long-Term Debt/Asset Trend Pre-and Post-Antitakoever Laws Figure 1 describes long-term Debt/Asset trend of firms in Delaware, firms in antitakeover states except Delaware and firms not in antitakeover states pre-and post-antitakeover Laws, which is from Garvey and Hanka (1999). Sample is 835 firms with complete Compustat and CRSP data, excluding utilities, and firms in states that passed a second-generation anti-takeover law before Anti-takeover states are those that passed second generation antitakeover laws in the period Control sample includes all firms not in antitakeover states. 25

26 Table I State AntiTakeover Legislation Table I describes various antitakeover law passage times in various states. (Source: Bertrand and Mullainathan (2003)). Business Combination Law Fair Price Law Control Share Acquisition Law Arizona 1987 Arizona 1987 Arizona 1987 Connecticut 1989 Connecticut 1984 Hawaii 1985 Delaware 1988 Georgia 1985 Idaho 1988 Georgia 1988 Idaho 1988 Indiana 1986 Idaho 1988 Illinois 1984 Kansas 1988 Illinois 1989 Indiana 1986 Louisiana 1987 Indiana 1986 Kentucky 1989 Maryland 1988 Kansas 1989 Louisiana 1985 Massachusetts 1987 Kentucky 1987 Maryland 1983 Michigan 1988 Maine 1988 Michigan 1984 Minnesota 1984 Maryland 1989 Mississippi 1985 Mississippi 1991 Massachusetts 1989 Missouri 1986 Missouri 1984 Michigan 1989 New Jersey 1986 Nebraska 1988 Minnesota 1987 New York 1985 Nevada 1987 Missouri 1986 North Carolina 1987 North Carolina 1987 Nebraska 1988 Ohio 1990 Oklahoma 1987 Nevada 1991 Pennsylvania 1989 Oregon 1987 New Jersey 1986 South Carolina 1988 Pennsylvania 1989 New York 1985 South Dakota 1990 South Carolina 1988 Oklahoma 1991 Tennessee 1988 South Dakota 1990 Ohio 1990 Virginia 1985 Tennessee 1988 Pennsylvania 1989 Washington 1990 Utah 1987 Rhode Island 1990 Wisconsin 1985 Virginia 1988 South Carolina 1988 Wisconsin 1991 South Dakota 1990 Wyoming 1990 Tennessee 1988 Virginia 1988 Washington 1990 Wisconsin 1987 Wyoming

27 Table II Summary of the adoption year of the first antitakeover law in different states and the number of sample firms incorporated in each state State Number of sample firms incorporated in the state Year of passage of the first secondgeneration antitakeover law State Number of sample firms incorporated in the state Alabama 1 New Jersey California 4 New York Colorado 1 North Carolina Delaware Ohio Florida 7 Oklahoma Georgia Oregon Illinois Pennsylvania Indiana Rhode Island Maryland Tennessee Massachusetts Texas 1 Michigan Virginia Minnesota Washington Missouri Wisconsin Nevada Year of passage of the first secondgeneration antitakeover law 27

28 Table III Summary Statistics for Our Sample of Manufacturing Firms in U.S.A. Table III presents descriptive statistics for regression variables used in our analysis. The sample consists of over for which we have sufficient accounting data to conduct pre- and post-passage of antitakeover laws test. Investment is measured as capital expenditures (item 128). Cash flow is measured as the sum of earnings before extraordinary items (item 18) and depreciation (item 14). All variables mentioned above are scaled by capital of the preceding year (item 8). Tobin s Q is measured as the market value of assets divided by the book value of assets (item 6), where the market value of assets equals the book value of assets plus the market value of common equity less the sum of the book value of common equity (item 60) and balance sheet deferred taxes (item 74). Tobin Q is calculated at the beginning of a firm s fiscal year. We use sales (item 12) to measure sales and cash (item 162) to measure cash AnalystForecastDispersion is the standard deviation of analysts current-fiscal-year EPS forecasts scaled by the absolute value of the mean forecast. AnalystForecastNumber is the number of analyst forecast for a certain firm. For each variable, we provide number of observations, mean, median, 1 st percentile value, 99 th percentile value and standard deviation. Variable Number of Obs Mean Median percentile percentile Standard Deviation Investment Cash flow Q Sales Cash Capital AnalystForecastDispersion AnalystForecastNumber st 99 th 28

29 Table IV Univariate Analysis Table IV compares the mean of variables used in our regression analysis for companies incorporated in the antitakeover states pre- and post-passage of antitakeover laws. Investment is measured as capital expenditures (item 128). Cash flow is measured as the sum of earnings before extraordinary items (item 18) and depreciation (item 14).Tobin s Q is measured as the market value of assets divided by the book value of assets (item 6), where the market value of assets equals the book value of assets plus the market value of common equity less the sum of the book value of common equity (item 60) and balance sheet deferred taxes (item 74). All variables mentioned above are scaled by capital of the preceding year (item 8). Tobin Q is calculated at the beginning of a firm s fiscal year. We use sales (item 12) to measure sales and cash (item 162) to measure cash. AnalystForecastDispersion is the standard deviation of analysts current-fiscal-year EPS forecasts scaled by the absolute value of the mean forecast. AnalystForecastNumber is the number of analyst forecast for a certain firm. The significance level is based on a twosample t test. Variable Mean of variable before passage of antitakeover laws Mean of variable after passage of antitakeover laws P value Investment 0.28*** 0.22*** 0.00 Cashflow 0.48 * 0.42 * 0.06 Q Sales Cash 0.22 *** 0.33*** 0.01 AnalystForecastDispersion AnalystForecastNumber 9.74* 10.78* 0.06 * significant at 10% level; ** significant at 5% level; *** significant at 1% level 29

30 Table V Effects of Passage of Laws on Investment-Cash Flow Sensitivity Table V presents the regression results on the effects of passage of laws on investmentcash flow sensitivity. The dependent variable investment is measured as capital expenditures (item 128). PostLaw equals one after passage of the antitakeover law for certain firm and zero otherwise. Treat equals one If a firm is incorporated in a state, which passed antitakeover laws; otherwise zero. Cash flow is measured as the sum of earnings before extraordinary items (item 18) and depreciation (item 14).Tobin s Q is measured as the market value of assets divided by the book value of assets (item 6), where the market value of assets equals the book value of assets plus the market value of common equity less the sum of the book value of common equity (item 60) and balance sheet deferred taxes (item 74). All variables mentioned above are scaled by capital of the preceding year (item 8). Tobin Q is calculated at the beginning and end of a firm s fiscal year. We use sales (item 12) to measure sales and cash (item 162) to measure cash. All regression equations include firm and year effect and report p-values based on robust (white) standard errors that also incorporate clustering around each firm. Dependent Variable=Investment (1) (2) (3) (4) (5) CashFlow 0.142*** 0.179*** 0.137*** 0.134*** 0.134*** (0.023) (0.028) (0.023) (0.025) (0.025) Treat*CashFlow ** ** ** ** ** (0.039) (0.045) (0.038) (0.038) (0.038) Treat*PostLaw*CashFlow 0.045* 0.051* 0.044* 0.054** 0.058* (0.026) (0.031) (0.026) (0.027) (0.030) Beginning-of-Period Q 0.088*** 0.082*** 0.077*** 0.072*** (0.013) (0.013) (0.012) (0.012) Sales 0.010*** 0.011*** 0.010*** 0.009*** 0.009*** (0.003) (0.003) (0.003) (0.002) (0.002) Cash 0.015* * (0.009) (0.010) (0.009) (0.007) (0.007) Treat*PostLaw (0.021) (0.022) (0.020) (0.020) (0.021) End-of-Period Q (0.010) Lagged Cashflow (t-1) 0.056*** 0.043** (0.022) (0.019) Lagged Cashflow (t-2) 0.033** (0.013) Constant 0.097*** 0.211*** 0.088*** 0.098*** 0.087*** (0.030) (0.030) (0.030) (0.027) (0.028) R-squared N. of observations Robust t-statistics in parentheses * p<0.10, ** p<0.05, *** p<

31 Table VI Effects of Passage of Laws on Investment-Cash Flow Sensitivity Controlling for the Level of Asymmetric Information Table VI presents the regression results on the effects of passage of laws on investmentcash flow sensitivity, which is the same regression with another control variable AnalystForecastDispersion multiplied by CashFlow. AnalystForecastDispersion is the standard deviation of analysts current-fiscal-year EPS forecasts scaled by the absolute value of the mean forecast. The details of definitions of all the other variables are reported in Table IV. All regression equations include firm and year effect and report p- values based on robust standard errors that also incorporate clustering around each firm. Dependent Variable=Investment (1) (2) (3) (4) (5) CashFlow 0.393*** 0.433*** 0.391*** 0.358*** 0.358*** (0.065) (0.062) (0.067) (0.062) (0.063) Treat*CashFlow *** *** *** *** *** (0.077) (0.079) (0.077) (0.072) (0.073) Treat*PostLaw*CashFlow 0.049*** 0.055*** 0.049*** 0.069*** 0.068*** (0.013) (0.016) (0.013) (0.019) (0.022) AnalystForecastDispersion*CashFlow (0.039) (0.041) (0.041) (0.039) (0.039) Beginning-of-period Q 0.066*** 0.065*** 0.060*** 0.060*** (0.014) (0.013) (0.013) (0.013) Sales (0.007) (0.007) (0.007) (0.006) (0.007) Cash (0.021) (0.025) (0.020) (0.019) (0.019) Treat*PostLaw * * (0.017) (0.017) (0.017) (0.018) (0.018) End-of-Period Q (0.011) Lagged Cashflow (t-1) 0.089*** 0.091*** (0.024) (0.025) Lagged Cashflow (t-2) (0.017) Constant 0.099* 0.166*** ** 0.108** (0.056) (0.048) (0.041) (0.052) (0.053) R-squared N. of observations Robust t-statistics in parentheses * p<0.10, ** p<0.05, *** p<