Does Corporate Headquarters Location Matter for ... - CiteSeerX

6 downloads 57925 Views 1MB Size Report
1Sheldon B. Lubar School of Business, University of Wisconsin-Milwaukee, ... Authors' contact information: Gao, [email protected], (414) 229-2547; Ng, ...... AT&T, AOL, to Citigroup and Lehman as well as a diverse constituency of both ...
Does Corporate Headquarters Location Matter for Corporate Financial Policies? Wenlian Gao, Lilian Ng, and Qinghai Wang1

Preliminary and Incomplete

July 2006

1

Sheldon B. Lubar School of Business, University of Wisconsin-Milwaukee, Milwaukee, P.O. Box 742, WI 53201-0742. Authors’ contact information: Gao, [email protected], (414) 229-2547; Ng, [email protected], (414) 229-5925; and Wang, [email protected], (414) 229-4775.

Does Corporate Headquarters Location Matter for Corporate Financial Policies?

Abstract This paper studies the impact of corporate headquarters location on corporate financial policies. We show that firms exhibit conformity in their financial policies to those of geographically proximate firms, and that the location of corporate headquarters helps explain the cross-sectional variation of corporate policies in the United States. This location effect is robust to state regulations on corporate takeover and payout and to the impact of local financial market conditions. The results show that non-economic factors such as local culture and social interactions among corporate executives are important determinants of corporate financial policies.

Keywords: Financial policy, state regulation, bank condition, social interaction

1

Introduction

Significant differences in corporate financial policies persist across countries. Recent studies show that country-specific factors still play a crucial role in explaining the cross-country differences in corporate financial policies,1 even after controlling for economic and financial factors, including legal protection of investors. One explanation for the importance of country factors is that differences in institutional settings across countries are an important determinant of corporate financial policies (see Booth et al. (2000)). Another explanation is that varying non-economic factors such as culture contribute to cross-country differences in both the institutional settings and corporate financial policies (see Stulz and Williamson (2003)). In this paper, we explore whether non-economic factors affect corporate financial policies of U.S. firms. By restricting our sample to U.S. firms, we can easily control for differences in institutional settings that can influence corporate financial policies. Using corporate headquarters locations as proxies for non-economic factors that can affect corporate policies, we focus on the local components of these factors such as local culture and social interactions among corporate executives.2 We study whether corporate headquarters locations are related to important corporate financial policies such as capital structures, financing policies, and payout policies. Our sample includes large U.S. corporations that have greater access to external finance and have well established financial policies. The sample period spans from 1988 to 2003. For the sample firms, we examine corporate financial policy, including capital structure variables (financial leverage, interest coverage, cash holdings, the level and likelihood of net long-term debt issues, and the level and likelihood of net equity issues) and payout policy variables (the level and likelihood of common dividend payout, the level and likelihood of share repurchase, the level of total payout). We use Metropolitan Statistical Area (MSA) to define the location of corporate headquarters. In our empirical analysis, we employ the fixed effects of MSAs as a 1

See Rajan and Zingales (1995) and Booth et al. (2000) for studies on capital structure, La Porta et al. (2000) on dividend policies, and Levine (2001) on corporate financing choices. 2 Following standard definition, culture is “transmission from one generation to the next, via teaching and imitation, of knowledge, values, and other factors that influence behavior” (see Boyd and Richerson (1985)). Social interactions, on the other hand, can help transmit such “knowledge, values, and other factors that influence behavior” within generation and particularly within social networks.

1

proxy for the effect of local communities on corporate policies and examine the joint significance of these metropolitan areas on corporate policies, after controlling for firm specific and timevarying characteristics, year fixed effects and industry fixed effects. We find strong evidence that corporate headquarters locations have a significant impact on corporate financial policies. Firms exhibit conformity in their financial policies to those of other local firms, and corporate headquarters location helps explain the cross-sectional variations of corporate policies in the United States. Corporations located in the same metropolitan areas exhibit similar leverage ratios and have similar policies of cash holdings. These firms also tend to follow similar patterns of issuing equity and debt. While corporate headquarters locations have less impact on the amount of payouts by area firms, corporations located in the same metropolitan areas show commonality in decisions on whether to pay dividends and repurchase shares. We control for various local economic factors that could contribute to the community effect we document. Specifically, we incorporate state laws and banking sector development into our specification. Given the fact that all public corporations are subject to the business corporation statutes of the state where they are incorporated, state statutes are acknowledged to be an important determinant in corporate policies (Bebchuk and Cohen (2003) and Wald and Long (2006)). Following Wald and Long (2006), we create an antitakeover index and a payout restriction variable for each state to proxy for the variation in antitakeover statutes and payout restriction laws across states. Antitakeover statutes have impacts on firms’ financial policy. The state law on payout restrictions requires firms to maintain a minimum level of capital-to-debt ratio when they pay dividend or buy back their own stocks, thus affecting both corporate capital structure and payout policies. Our analysis demonstrates that, despite the fact that state statutes have significant effects on some financial policies, the fixed effects of local community remain significant. Specifically, we find that antitakeover statutes do not reduce firms’ debt use, which is consistent with the findings of Wald and Long (2006). Instead, we find that antitakeover statues are positively related to the level of net long-term debt issues and negatively related to the likelihood of net equity issues. As 2

expected, payout restrictions are negatively related to firm leverage, net long-term debt issues, and the likelihood of net long-term debt issues. Also, payout restrictions are negatively related to net equity issues. Accordingly, payout restrictions are positively related to interest coverage, an alternative measure of capital structure. As for payout policies, payout restrictions have positive effect on cash dividend payout and the likelihood of paying cash dividend. The second set of local economic variables we employ is related to local financial market conditions. If corporate financial decisions are affected by local financial market, particularly the banking sector, local financial market conditions can contribute to the community effect. Existing literature has documented a strong link between the functioning of financial sector and economic growth at state level (Jayaratne and Strahan (1996), Black and Strahan (2002)). More recent studies show local banking conditions affect individual firm’s financial and investment policies (Zarutslie (2006)). We thus conjecture that the variation of financial sector development at the metropolitan level may affect the fixed effects of local community on financial policies. We construct three variables to measure the commercial bank conditions in each metropolitan area. One is the average ratio of nonperformance loan, weighted by each bank’s commercial loans outstanding, a proxy for bank loan quality. The second is the ratio of the sum of commercial loans outstanding from all commercial banks in each metropolitan area to the sum of sales from all the sample firms in the same area, a proxy for the size or depth of the financial sector. The third is Herfindal index of commercial loans, a measure of bank competition. The results show that, though bank conditions are significantly related to some firm policies, they do not alter the significance of the local community fixed effects. In particular, we find that the ratio of non-performance loan is negatively related to net long-term debt issues and the likelihood of net equity issues, and positively related to the level of share repurchase. The financial depth has significantly positive effect on leverage and negative effect on cash holdings. Moreover, the financial depth has significantly negative effect on cash dividend payout but positive effect on the likelihood of share repurchase. Under higher degree of banking market competition, firms are more likely to issue long-term debt. We further integrate both state statutes and banking development into the fixed effect regression specification and find that the fixed effects of local 3

community are still strongly significant. The robustness of the community effect after controlling for firm characteristics, industry effect, state regulation and local financial market condition leads us to further explore no-economic factors as possible explanations. A number of recent studies find that personal attributes of corporate executives and particularly their management ”styles” have significant impact on corporate policies.3 Stulz and Williamson (2003) show differences in the broadly defined culture contribute to differences in financial systems and financial policies across countries. Since corporations’ decisions are constrained and shaped by managers’ embeddedness in social networks (Granovetter (1985)), social interactions among corporate executives can influence the decisions of corporate executives and drive the local commonality in corporate policies. Similarly, local culture can affect the behaviors of both managers and investors and help to shape corporate financial policies. In our empirical analysis, we focus on the two aspects of non-economic factors and more importantly we assess the relative contribution of the two factors to the observed evidence. The local community effect we document could reflect the impact of social interactions among firm managers on corporate decisions. Local community facilitates managerial interaction through two channels: face-to-face information transmission and easy observational learning. Managers who work in the same geographic area normally have many opportunities to network and build valuable relationship with peers, exchanging ideas and learning from each other’s experience. For example, they may attend the same CEO clubs or conferences, or they may be members of the same regional business leadership associations, such as local charitable organizations and chambers of commerce. ”The country club cliche - that much business gossip is traded over golf games - is in fact surprising accurate, according to discussions with directors” (Davis and Greve (1997)). The local community effect can also be shaped by local culture. Culture is ”transmission from 3 For example, personal characteristics of CEOs and other top executives can affect acquisition or diversification decisions, dividend policy, interest coverage, and cost-cutting policy (Bertrand and Schoar (2003)), investment policy (Malmendier and Tate (2005)), financial policy (Malmendier, Tate, and Yan (2005)), and corporate risk management (Beber and Fabbri (2005)).

4

one generation to the next, via teaching and imitation, of knowledge, values, and other factors that influence behavior.” (see Boyd and Richerson (1985)) Social interactions, on the other hand, can help transmit such ”knowledge, values, and other factors that influence behavior” within generation and particularly within social networks. The impacts of local culture and social interactions can overlap considerably. Culture and social factors are slow-moving components in the social structure and they shape value and behavior of managers. Firm managers are embedded in social structures and that their social interactions also reflect the social networks they belong to and the social contexts in which they live. To examine the difference of the impacts of local culture and social interactions on corporate policies, we rely on the intuitive fact that social interactions are more important at the local or community level, while local culture should be more important at the regional level. We employ the nine regions defined by Census Bureau as an aggregate proxy for local culture effect. The fixed effects of regions could further capture any possible variation with respect to income characteristics, industrial composition of the employed labor force, and such noneconomic factors as demographic, social, and cultural characteristics among regions. In addition, we identify measurable component or proxies for local culture at the regional level and directly examine the impact of those proxies on corporate policies. We find that the differences in culture and socioeconomic characteristics can account for part of the variation in corporate policies. Specifically, the fixed effects of regions are significant or marginally significant for most of the financial policies, except leverage, interest coverage, and the likelihood of equity issues. However, the region fixed effects cannot subsume the fixed effects of metropolitan area (except for the financial policy of interest coverage). This suggests that most of the financial policies are affected not only by social interactions of corporate managers but also by culture and socioeconomic factors. We further examine the role of culture in determining corporate financial policy. The role of culture in economics is well established (Weber (1930) and Landes (2000). We use three culture variables, trust, church attendance, and the percentage of Protestant in each region, which are obtained from World Value Surveys (WVS). Trust is the percentage of people who think most 5

people can be trusted. Church attendance is defined as the percentage of people who attend church at least once a week. Results suggest that cultural variables are jointly significant for cash holdings, the likelihood of long-term debt issuance, the likelihood of paying dividend and the likelihood of buying back shares. Our findings contribute to the literature along several dimensions. First, our study show significant local commonality in corporate financial policies that are unexplained by firm and industry characteristics. Second, the study contributes to our understanding of non-economic determinants of corporate financial policies. Existing studies show that cultural differences across countries contribute to differences of financial systems and corporate financial policies and corporate executives have significant impact on corporate policies. While our study follow the insights of the above two strands of literature, our analysis provides a link (or medium) between the two. We show that local culture and social interaction among corporate executives are important determinants of corporate financial policies. The remainder of the paper is organized as follows. Section 2 describes the sample selection process and defines the main variables of interest. Section 3 discusses the empirical method, examines the significance and magnitude of local community effects on corporate financial policy. Section 4 checks possible omitted variable problem, including statutes on antitakeover and payout restrictions at state level and bank conditions at metropolis level. Section 5 investigates interpretations to the local community effect and argues that social interactions among firm managers can account for the local component in corporate financial policies. Section 6 concludes.

2

Data

Our sample includes publicly traded U.S. firms that are covered by Compustat over the period 1988 to 2003. We exclude financial and utility firms (with industry code 4900-4999 and 60006999), because their financial policies are subject to the impact of regulation. We also delete firms with assets less than 20 million dollars from the sample, as these smaller firms are more 6

likely to face various constraints on their financial policies. Similarly, because corporate financial policies for younger firms are more likely to be affected by policies established before the firms become public, we require that a firm stay in the CRSP data for at least 5 consecutive years before entering the sample. Finally, we get 39,287 firm-year observations for 4,118 different firms. We define a firm’s local community as the Metropolitan Statistical Area (MSA) where the firm is headquartered. This definition is justified by the fact that metropolitan areas are usually clusters of firm headquarters (Davis and Henderson (2004)). According to the definition of the Office of Management and Budget (OMB), MSAs have “at least one urbanized area of 50,000 or more population, plus adjacent territory that has a high degree of social and economic integration with the core as measured by commuting ties”. Usually, MSAs consist of one or more entire counties from one single state or across a couple of states. To identify the MSA where a firm is headquartered, we obtain the current state and county of firms’ headquarters from Compustat annual file, and then use the Compact Disclosure database to check all the firms that have ever relocated during our sample period. Employing the state/county combination as a link, we next merge Compustat annual file with the Metropolitan Areas and Components data released by OMB as of 1993. To classify an MSA as a major metropolitan area in the sample, we require that at least 20 sample firms each year are located in the metropolitan area and that there are at least 200 firm-year observations over the whole sample period. In total, we identify 27 metropolitan areas in our sample that meet the above criterion. Our analysis focuses on the impact of each metropolitan area on the corporate financial polices of firms located in the area. We use the remaining firms that do not belong to any of the 27 metropolitan areas as the reference sample in our analysis. We employ the Compustat annual file to obtain firms’ financial information. For our sample of firms, we construct financial policy variables and control variables. We divide the financial policy variables into two broad categories - capital structure and payout policy. The capital structure category includes leverage, interest coverage, cash holdings, the level and likelihood of net long-term debt issues, and the level and likelihood of equity issues. The payout policy 7

category includes the level and likelihood of common dividend, the level and likelihood of share repurchase, and the level of total payout.4 The definition of these variables is listed in Figure 1. Furthermore, throughout this study, our analysis controls for three firm-specific variables: size, return on assets, and market to book ratio. Size is the logarithm of total assets. Return on assets (ROA) is EBITDA (#18) deflated by total assets. Market to book ratio, a proxy for growth opportunities, is defined as total assets (#6) minus the book value of common equity (#60) plus the year-end closing price (#24) times the number of shares outstanding (#25) over total assets (#6). Panel A of Table 1 provides summary statistics for the financial policy variables. The data are winsorized at 99.5 percent to reduce the impact of outliers on the findings. Firm leverage has a median of 21.1 percent and 25th and 75th percentile values of 4.6 percent and 36.9 percent, respectively. The median of interest coverage, an alternative measure of firm leverage, is 5.78 with a standard deviation of 206.21. While there is a large cross-sectional variation in firm leverage, a large portion of firms use a significant amount of debt. Cash holdings have a mean of 39.1 percent and a median of 7.2 percent, suggesting a highly skewed distribution. On average, net long-term debt issues of firms is 3.3 percent and their net equity issues is 5 percent. 34.1 percent of firms, on average, issue long-term debt and 57.8 issue equity. On payout policy, firms have a mean dividend payout of 0.7 percent, share repurchase of 2.6 percent, and total payout of 5.5 percent. About 31.8 percent of firms pay dividend and 38.5 percent have share repurchase activities. Panel B of the same table shows the statistical distribution of the 3 control variables. The median firm asset value is $182 million (standard deviation = $3,712 million), median market to book ratio is 1.416 (standard deviation = 1.612), and median return on assets is 0.037 (standard deviation = 0.163). It is evident that the 3 variable vary widely across the sample firms.

4

Common dividend payout, share repurchase, total payout, net long-term debt issues, and net equity issues are set to 0 if the value is missing.

8

3

Location and Corporate Financial Policies

3.1 Methodology This subsection discusses the main methodology employed to evaluate the impact of headquarters location on corporate financial policies. In our analysis, we perform a baseline regression by regressing each financial policy variable on year fixed effects, industry fixed effects, and the control variables. Next, we incorporate the fixed effects of headquarters locations and examine the joint significance and explanatory power of firm locations. More specifically, we estimate two regressions for each dependent variable: yit = αt + γIN D + βXit + it

(1)

yit = αt + γIN D + βXit + λM SA + it

(2)

where yit represents a financial policy variable, αt is the year fixed effect, γIN D are industry fixed effects, Xit are firm-level control variables, and it is an error term. We employ Fama-French (1997) 43 industry classifications. λM SA in Equation (2) are fixed effects of firm locations. We use 27 location dummy variables for the 27 metropolitan areas. All sample firms not included in the 27 metropolitan areas form the reference sample. It is crucial to control for the industry fixed effects in both models since the existing literature suggests that many industries tend to cluster around a geographic area due to the consideration of positive externalities. For example, many high technology firms tend to cluster in California and, on average, they tend to have lower leverage. Controlling industry effects ensures that the fixed effects of firm locations are not simply picking up sector characteristics. Finally, we use clustered standard errors to adjust for the correlation within a firm over time in our pooled analysis.

3.2 Baseline results Table 2 reports F tests and adjusted R−squares from the estimation of models (1) and (2) on financial policy variables. For each financial policy variable, the first row shows F −statistics 9

from the joint significance test of industry fixed effects and adjusted R−square from (1), and the number of observations employed in the estimation. The second row reports the F −statistics from the joint significance tests of industry fixed effects and of metropolitan area fixed effects, and adjusted R−square from (2). All regressions include year fixed effects and the three firmlevel control variables: logarithm of total assets, market to book ratio, and the rate of return on assets. We find substantial evidence that firm location has a significant impact on corporate policies, even after controlling for industry effects. Column 4 of the table shows that the F −statistics are large enough to reject the joint hypothesis that metropolitan areas bear no effects on all the financial policy variables we consider. Interestingly, the significant metropolitan area effects do not subsume the industry effects, which are also highly significant across all regressions. Furthermore, the results show a wide variation in the explanatory power of the metropolitan area effects, suggesting that varying location impacts on various financial policies. For capital structure variables, firm locations strongly influence leverage, cash holdings and the decision to issue new equity. For example, adding metropolitan area fixed effects to regression model (1) helps improve the adjusted R−square or the pseudo R−square by 1.7 percent for leverage, 3.8 for cash holdings, 1.4 for the likelihood of net equity issues. For corporate payout variables, the location effects of corporate headquarters have a substantial impact on the level of dividend payout and the dividend payout dummy. Adjusted R-squares increase by 1.1 percent for the level of cash dividend payment and by 2.5 percent for the likelihood of paying cash dividends. Increases in the adjusted R−square, however, are less than 1 percent for all other financial policy variables, and are particularly low for the level of share repurchase and the level of total payout. Literature on payout policy suggests that firm managers are more concerned with the stability of dividend payout, and hence they tend to smooth out dividend payments. By contrast, share repurchases are more volatile and also are more sensitive to economic conditions than dividend payout. Thus, they are less likely to exhibit any systematic pattern across metropolitan areas.

10

Finally, in Table 3, we also plot the distribution of the coefficient estimates on metropolitan area fixed effects. Table 2 shows that firm locations in the 27 metropolitan areas jointly play a role in corporate financial policies, while Table 3 shows how location effects on each financial policy variable vary across the metropolitan areas. It is evident that the variation in the magnitude of location fixed effects is economically large. For example, the first row associated with leverage shows that the range between the 25th and 75th percentile is 0.054 and the range between the minimum and maximum is 0.134. The median leverage for all sample firms is 0.211. For cash holdings, we observe a range of 0.031 between the 25th and 75th percentile, while the sample median is 0.072. For payout policies, we highlight the level of cash dividend payment as an example. The difference between the 25th and 75th percentile is 0.004, compared with the median common dividend of 0 in our sample.

3.3 Robustness checks In this section, we conduct a series of robustness checks, including the significance of the metropolitan area fixed effects over time, alternative estimation technique, and alternative industry definition. We find that our baseline evidence of location effects on financial policy is robust. 3.3.1 Time series analysis We first report results from the cross-sectional regressions year by year and then the pooled regression results for three subperiods. In the cross-sectional analysis, for each year, we include industry fixed effects, metropolitan area fixed effects, logarithm of total assets, market-to-book ratio and ROA in the regression. We then determine the number of years in which the metropolitan area fixed effects are significant at 5 and 10 percent confidence levels; the respective results are reported in Columns 2 and 3 of Table 4, with time-series average adjusted R−square in the last column. The results show patterns similar to those of Table 2. The variables that are highly significant tend to remain significant in most of the years. We further examine the robustness of our results over time by estimating pooled cross11

sectional regressions for each of the three subperiods: 1988-1992, 1993-1998, and 1999-2003. The results are presented in Table 5. Generally, the results are fairly stable over time. For most of the financial policy variables, the metropolitan area fixed effects are consistently significant over different time periods. There are only a few exceptions. The metropolitan area fixed effects are insignificant for both interest coverage and the likelihood of net long-term debt issues during the period 1988 to 1992. For the level of share repurchase, the fixed effects of metropolitan area are only marginally significant over the subperiods 1993 to 1998 and 1999 to 2003. Furthermore, the metropolitan areas are not jointly significant for the level of total payout over the subperiod 1999 to 2003. 3.3.2 Alternative estimation techniques The pooling of cross-sectional and time-series data in our regressions may create correlation of errors at the firm level. Instead of clustering the standard errors at the firm level, we employ an alternative technique - Newey-West (1987) specification with a lag of one. The unreported results indicate that the F −statistics for the joint significance of metropolitan area fixed effects are even higher than those reported in Table 2. Another way to address the possible serial correlation at the firm level is to collapse the data at the area-firm level. Starting with the firm-year data, we regress all financial policy variables of interest on the year fixed effects, industry fixed effects, and control variables at the firm level. Then, we extract firm-year residuals from the above regressions and collapse these residuals by the area-firm level. Finally, we estimate the metropolitan area fixed effects in the collapsed residuals. We find that our baseline evidence is robust to this alternative estimation technique. 3.3.3 Alternative industry definition We replicate the regressions in Tables 2 and 4 by controlling for the fixed effects of 2-digit SIC industry classifications from Compustat, instead 43 Fama-French industry classifications. The unreported results indicate that the joint effect of 2-digit industry classifications is also significant for all corporate policy variables. There is little variation in the significance test 12

of metropolitan area fixed effects. Thus, our basic results are robust to alternative industry definitions.

3.3.4 Alternative reference firms Finally, we exclude all firm-year observations for which firms are located in metropolitan areas with less than 20 firms in a year, or with less than 200 firm-year observations over the sample period, instead of taking such firm-year observations as the reference in our previous analysis. We get 31,136 observations for 3,316 different firms that are located across 27 metropolitan areas. Then we re-estimate model (2) and obtain similar results.

4

State Regulation, Economic/Financial Conditions, and Location Effects

Thus far, the geographic location of corporate headquarters exhibits a substantially significant impact on firms’ financial policies, and the magnitude of location effects is economically significant. As discussed in the introduction, the significant location effect could represent the impact of local “fundamentals” on corporate financial policies. In this section, we examine the impact of these local “fundamentals” on financial policies. Particularly, we investigate whether our results are driven by state regulations and local economic and financial conditions.

4.1 State Regulations and Financial Policies In the U.S., all public corporations are subject to the business incorporation statutes of the state where they are incorporated. Bebchuk and Cohen (2002) provide evidence that state antitakeover statutes affect firms’ decision on where to incorporate. Wald and Long (2006) further demonstrate that firms choose their incorporation state according to state laws and capital structure needs. Therefore, we conjecture that firms headquartered in the same metropolitan area may have incorporated in states with certain state statutes, which may partly account for the local community effect in financial policies. 13

We consider two types of state regulations in our study. One is state statutes on antitakeover, and the other is payout restrictions. State antitakeover laws are believed to have an impact on increasing the entrenchment level of the incumbent managers, while payout restrictions capture the potential conflict of interest between shareholders and debtholders. The state antitakeover laws that we consider here are referred to as “the second generation of antitakeover laws” and most states enacted some of these laws in late 1980s. State antitakeover statutes are recognized to have an impact on firms’ capital structure and payout policies by increasing the managerial entrenchment level. Zwiebel (1996) and Novaes and Zingales (1995) argue that managers use leverage to reduce the threat of a hostile takeover. Thus, if a firm is shielded from hostile takeover, its manager will prefer a lower leverage. Garvey and Hanka (1999) provide empirical evidence that firms issue less debt and reduce their leverage over time after their incorporation state enacted more antitakeover statutes which are supposed to increase managers’ entrenchment level. They further point out that the reduction in debt funding results in the reductions in share repurchase and dividend payout. In contrast, Wald and Long (2006) and Litov (2005) find that firms incorporated in states with more antitakeover statutes use more debt finance and have higher leverage ratios. The two authors attribute the difference to self-selection in firms’ decision to choose their incorporation state (Wald and Long (2006)) and sample selection procedure (Litov (2005)). The payout restriction statute typically requires a minimum ratio between the amount of book capital and debt before making a dividend payment or share repurchase. Firms subject to payout restrictions are limited in the amount of debt they can issue. Wald and Long (2006) document that payout restriction laws have an impact on capital structure. Specifically, firms incorporated in states with stronger payout restrictions (i.e., higher minimum capital-to-debt ratio requirement) use less debt. Hu and Kumar (2004) document that both the likelihood and the level of payouts are positively related to factors that increase executive entrenchment levels, because, by committing to higher payouts, entrenched managers can protect themselves from disciplinary sanctions by outsiders. John and Knyazeva (2006), however, find that more antitakeover protections increase dividend payout but lower share repurchase. They attribute 14

the differential effect of governance quality on repurchase to the fact that share repurchase largely acts as a discretionary tool to distribute excessive temporary cash and that absence of antitakeover protections will urge managers to dispense cash to shareholders through repurchase. Information on firms’ incorporation state is available from Compustat that only provides the current state of incorporation. We therefore gather details on historical reincorporation decision by searching Mergent online. The information about statutes for each state is from McGum, Pamepinto, and Spector (1989) for the late 1980s and from Gartman (2000) for the period 1990 to 2003. Following Bebchuk and Cohen (2003), we construct an antitakeover protection index for each state by assigning one point to every specific statute in place and 0 otherwise. The antitakeover protection index is the sum of points assigned to five state antitakeover statutes, namely control share, fair price, no freeze-outs, poison pill endorsement, and constituencies. A higher protection index indicates that the state provides more antitakeover protection to firms incorporated in the state. However, most state antitakeover provisions allow companies to “opt out” of coverage by stating their intention in their charters. For example, Romano (1993) reports that most Pennsylvania firms choose to opt out of the Pennsylvania statute. As in Wald and Long (2006), we use the information from IRRC to incorporate some firms’ decision to opt-out of antitakeover statutes. For the statutes on payout restrictions, we use the information provided in Wald and Long (2006) and construct a variable, “Restriction”, in the same way as they do. That is, for California and Alaska, it is equal to 1.25, for Delaware, Maine, Oklahoma, and South Dakota, it is equal to 0, and for the remaining states, it is equal to 1. Subsequently, we incorporate the two state regulation variables into the baseline specification model (2) and estimate the regression model for each policy variable. The results are presented in Table 5. Overall, the significance of the local community effect on financial policies is not affected by state regulations. The fixed effects of metropolitan areas are still highly significant for all the financial policy variables. Compared with the results shown in Table 2, the F values of the joint significance of metropolitan areas are slightly reduced for all the financial policy variables. Moreover, adding the state regulation variables does not enhance the explanatory 15

power of the regression. The adjusted R−square for each regression is almost the same as from the estimates of model (2). For example, the adjusted R−square for the regression on leverage is 23.3 percent for the estimate with state regulation variables, while it is 23.2 percent for the estimate of model (2). Thus, the two state regulation variables only contribute an increase of 0.1 percentage points in the explanatory power. For the regression on cash holdings, the R−square even keeps constant as in models (2). The state regulation variables are significant in only a few regressions. The index of antitakeover statutes is statistically significant only for the likelihood of net equity issues and the coefficient is negative. This is consistent with the findings of Harford, Mansi, and Maxwell (2005) that firms with weak shareholder rights use less equity financing and more debt financing. In our case, the positive coefficient on the level of net long-term debt issues is only marginally significant. This can be attributed to the lower cost of equity associated with well-governed firms, given the documented higher price-earnings ratio for these firms (Gompers, Ishii, and Metrick (2003)). In contrast, weak governance firms have a lower cost of debt (Klock, Mansi, and Maxwell (2004)) due to the alignment of interests between managers and bondholders in these firms as both entrenched managers and bondholders dislike risky projects. As for other variables of financial policies, state antitakeover statute has no significant effect on them. However, they do have the expected sign. First, antitakeover statutes are negatively related to leverage ratio, which is consistent with the findings of Wald and Long (2006) and Litov (2005). Secondly, antitakeover statutes are negatively related to cash holdings. This is consistent with the findings of Klock, Mansi, and Maxwell (2004) who argue that weak governance firms dissipate their cash reserves far more quickly than do managers of firms with strong shareholder rights, primarily through acquisitions. Thirdly, antitakeover statutes are positively related to payout policies, which is consistent with the findings of Hu and Kumar (2004). The variable of payout restrictions has a significantly negative coefficient in the regression on firms’ leverage ratio, which indicates that strict state statute on payout restrictions tends to lower firms’ leverage. This result is consistent with the evidence provided in Wald and Long (2006). Correspondingly, the statute on payout restrictions is positively related to interest coverage. In 16

addition, this law has a negative effect both on the level of long-term debt issues and the level of equity issues, since, payout restrictions statute imposes some upper limit on firms’ liabilities. Moreover, it has a negative effect on the likelihood of issuing long-term debt. To our surprise, payout restrictions law is positively related to the level of common dividend payment and the likelihood of dividend payment occurrence. In contrast, the level of total payout is negatively associated with the law, which seems to be driven mainly by the level of share repurchase. However, neither effect is significant. To summarize, incorporating state statutes on antitakeover and payout restrictions into our models has no material effect on the significance of local community effects on financial policies. Moreover, state statutes have little contribution to the explanatory power of the model.

4.2 Bank Conditions and Corporate Financial Policies In this subsection, we consider the impact of local economic and financial condition variables on our finding of location effects. Corporate financial policies are likely to be affected by local economic conditions. More important, corporate capital structure policies are more susceptible to local financial market conditions. Up to 1978, the U.S. banking system was segregated, with 50 banking systems, one per state. The passage of the Reigle-Neal Act in 1994 made interstate banking a bank right and the banking system became much more integrated. However, the development of banking system may still vary in different metropolitan areas of the country. In our analysis, we integrate banking conditions into our specification to ensure that the local community effect is not a simple proxy for the variation in financial sector development across metropolitan areas. In unreported results, we examine the relation between regional economic growth and corporate financial policies and do not find any significant results. The functioning of a local financial sector has been documented to have an impact at the firm level. Banks have a comparative advantage in acquiring private information on borrowers and in monitoring their actions. They can also modify the relative supply of different securities to some extent. Peterson and Rajan (1995) provide evidence that young firms receive more institutional finance and are more indebted in concentrated markets than in competitive markets, while 17

older firms exhibit the opposite pattern. Zarutslie (JFE forthcoming) confirms Peterson and Rajan (1995)’s findings and further demonstrates that bank competition has significant effect on individual firm’s financial policy. With increased competition of banking market, younger firms would use less debt, and the pattern reverses for established firms. These studies mainly focus on small or privately held firms, while our study examines relatively large firms. Though large U.S. corporations shift to the securities market to fulfill their financing needs and exhibit decreasing reliance on bank credit, banks continue to perform a critical function in providing liquidity to large corporations, especially during economic turmoil (Saidenberg and Strahan (1999)). To assess the development of the financial sector in each metropolitan area, we focus on commercial banks and obtain their balance sheet information from their Reports of Condition and Income which is required by the Federal Deposit Insurance Corporation (FDIC). These reports are available quarterly over our study period 1988 to 2003. For our study, we only need the bank accounting information at the end of each calendar year. FDIC provides the Primary Metropolitan Statistical Area (PMSA) code of bank location. We manually check the domain of each PMSA of bank location and the domain of each MSA of firm location respectively, and then merge the bank accounting information with our basic sample described in Section 2. We construct three variables to denote bank condition, i.e., nonperformance loans, commercial loans to sales, and Herfindal index of commercial loans. We use nonperformance loans as an indicator for bank lending quality which is defined as the fraction of total loans classified as ”nonperforming”. Following Jayaratne and Strahan (1996), all loans 90 days or more past due but still accruing and non-accrual loans are classified as nonperforming loans. Next, we compute the weighted average of nonperforming loans for all commercial banks headquartered in the same metropolitan area, taking each bank’s commercial loans as the weight. Commercial loans are the sum of commercial and industrial loans (C&I loans) and commercial real estate loans. Commercial loan category measures the flows of bank credit to industries and thus is ”likely to be closely linked to commercial investment and economic conditions” (Jayaratne and Strahan (1996)). 18

The ratio of commercial loans to sales, a proxy for bank depth or volume of bank lending, is measured by the ratio of total volume of commercial loans in each metropolitan area to the total volume of firm sales in each metropolitan area. It measures the size of banking sector to the size of the economy. Total volume of commercial loans are derived by summing up the commercial and industrial loans held by all commercial banks in each metropolitan area, whereas total volume of firm sales are derived by summing up sales of all firms in each metropolitan area. Herfindal index of commercial loans, a measure of the degree of bank competition, is constructed by summing over the squared market share of commercial loans from each individual commercial bank in a metropolitan area. The descriptive statistics for the bank condition variables are displayed in Panel C of Table 1. The average ratio of nonperformance loans is 0.6 percent across the metropolitan areas where the sample firms are headquartered, and the 25th and 75th percentile is 0.2 percent and 0.8 percent, respectively. The ratio of commercial loans to sales has a mean value of 0.291 and standard deviation of 0.279. The Herfindal index of commercial loans is averaged at 0.219, with the 25th and 75th percentile at 0.101 and 0.302. The summary statistics of the bank condition variables suggest that banking market efficiency varies across metropolitan areas of the country. We enhance the basic specification model (2) by incorporating the three variables of bank conditions and present the regression estimates in Table 6. The results show that bank conditions are significantly correlated with some firm policies. The level of net long-term debt issues and the likelihood of net equity issues are negatively related to the ratio of nonperforming loans, a bank health indicator, while the level of share repurchase is positively related to the ratio of nonperforming loans. Firm leverage is positively related to the ratio of commercial loans to sales, a measure of bank depth, since firms may be inclined to use more bank loans if the size of the bank sector is relatively large. Firm cash holdings are negatively related to bank depth. The precautionary incentive for cash holdings suggests that firms can use cash reserves to finance their investment activities if other sources of funding are not available or excessively expensive (Keynes(1934)). With abundant bank loans available, firms would find it easier to make shortterm loan arrangement and accordingly reduce their cash holdings. The variables of financing 19

sources are negatively correlated with bank depth, but none of them is significant. For payout policies, the bank depth variable has a significant negative coefficient in the regression on the level of common dividends, while it has a significant positive coefficient in the logit regression on the likelihood of share repurchase. Herfindal index, a proxy for bank competition, is only significantly related to the likelihood of net long-term debt issues, which may be interpreted as, in more concentrated credit markets, firms are less likely to issue long-term debt. Basically, taking into account the bank condition neither reduces the significance of the local community effects nor increases the model’s explanatory power substantially. For instance, in the regression on leverage, the F value of the significance of metropolitan area fixed effects reduces only 0.65 compared with the results from the benchmark regression shown in Table 2 and the local community effect is still highly significant. The adjusted R−square remains 23.2 percent, the same as the explanatory power of the benchmark specification. We also try altering the specification of the bank health measure and depth of financial structure. For example, we replace the weighted-average nonperformance loans and the ratio of commercial loans to firm sales by the weighted-average of charge-off and the ratio of total bank assets to firm assets, respectively. These replacements yield no material change in our findings. In a word, the evidence from Table 6 suggests that, though the variables of bank conditions have significant impact on some financial policies, overall, bank condition does not change the significance of local community effect. So far, we have investigated the impact of state regulations and the impact of bank conditions on firm policies separately. Now we consider these two factors together and integrate the two sets of variables into one model to see if their combination has any impact on the local community effect in firm policies. We report the regression estimates in Table 7. In the specific regression on each policy variable, state regulations and bank conditions have almost the same effect as when they are regressed on each policy variable independently. To highlight an example of firm leverage, the law on payout restrictions has a significantly negative coefficient of -0.019, and the ratio of commercial loans to sales has a significantly

20

positive coefficient of 0.031, which is exactly the same as the results shown in Tables 5 and 6. In addition, there is not much increase in the explanatory power of the specification. For instance, the adjusted R−square is 23.4 percent, compared with 23.2 percent from the basic specification model (2). The most important thing is that the fixed effects of metropolitan area are still jointly significant, which indicates that the combination of the two considerations does not explain away the local community effect in corporate financial policies.

5

Culture, Social Interaction, and Location Effects

5.1 Should non-economic factors matter? The previous sections have documented that the location effect on corporate financial policies is strongly statistically significant and economically important. The results are robust with respect to state regulation variables and proxies for local financial sector condition. We now proceed to explore non-economic explanations to this location effect. The location effect we document could capture the influence of non-economic factors such as culture and social interactions on corporate decision making. The role of culture in economics is well established. The literature can be traced back to Weber (1930) who argues that cultural changes inspired by the Protestant Reformation helped to explain the rise of capitalism in Western Europe and America and more recently, Landes (2000) confirms Weber’s argument. U.S. is a geographically large country. Culture and other socioeconomic factors vary substantially across regions. For example, Southern culture has been generally more conservative than that of the North and such cultural differences could have impact of corporate policies. The location effect is potentially related to the immediate institutional environment or the social context where the firm is embedded. Granovetter (1985) argues that firm managers’ embeddedness in social networks serves as a major channel of conveying information and ideas about firm behavior. The conformity demonstrated in financial policies among local firms, could be driven by social interactions among corporate managers. This proposition can be further justified by two facts. Firm managers are acknowledged to imprint their mark on a wide range of corporate policies. 21

For example, Bertrand and Schoar (2003) document that personal characteristics of CEO and other top executives, education and birth cohort, have significant effect on firms’ investment policy, financial policy, cost-cutting policy and performance. Beber and Fabbri (2005) find that foreign currency risk management is also tied to CEO individual characteristics, such as education, gender, tenure, birth cohort and previous working experience, etc. Malmendier and Tate (2005a, 2005b) provide evidence that, distinct from observable personal characteristics, managerial overconfidence matters in determining firms’ investment policy and financial policy. Hence, corporate decision making is, to a large degree, rooted in firm managers’ background. On the other hand, corporate managers are deemed as a group of people who are extremely socially active and they may be influenced by network contacts in decision making. Operating in an uncertain environment, firm officials may look to their peers for ideas about appropriate strategies or mimic one another’s behavior through direct contact. Recent studies have suggested that social interaction with peers has tangible effects on a wide range of firm activities from charitable action (Galaskiewicz and Wasserman (1989), Marquis, Glynn, and Davis (2005)), political contributions (Mizruchi (1989) and (1992)), acquisition decision (Haunschild (1993)), to adoption of antitakeover procedures (Davis and Greve (1997)). The social interaction effect should be especially important for local firms because: First, geographic proximity facilitates face-to-face interaction and makes contact/relationship easier to start and maintain. Survey data indicates that the correlation between distance between friends and frequency of contacts is 64 percent (Jaffe, Trajtenberg, and Henderson (1993)). Actually, many metropolitan areas have various kinds of business leadership associations, such as the Executive Council of New York, whose members constitute 3,500 New York metropolitan business leaders, ranging from AT&T, AOL, to Citigroup and Lehman as well as a diverse constituency of both enterprise and emerging growth companies. Secondly, geographic proximity facilitates observational learning even with no direct contact. Simple exposure to the strategies of other firms may prompt firms to adopt similar strategies and to align their activities with other firms in the local geographic community. Naturally, we posit that social interaction matters for corporate financial policy, i.e., firm managers may get some input from their peers and take into account when they make

22

firm financial strategies.

5.2 Culture, social interactions, and corporate financial policies To test for the presence of culture and social interaction effect on corporate financial policies, we first attempt to separate the effect of social interaction and the effect of culture and then use explicitly defined culture variables in the empirical analysis. We first employ the classification of nine census regions and identify each firm’s region as the census region where the firm’s is headquartered. The fixed effects of regions capture any possible variation with respect to income characteristics, industrial composition of the employed labor force, and such noneconomic factors as demographic, social, and cultural characteristics among regions. Though both metropolitan areas and census regions can capture some variation in socioeconomic characteristics, it is evident that census regions serve better for this purpose. For example, some metropolitan areas even belong to one state, so it is hard to imagine that there is much variation in socioeconomic characteristics between such two areas. In contrast, as for the capability to capture the interaction among managers, metropolitan areas work better than census regions. We believe that, social interaction is really a local phenomenon, the intensity of interaction among firm managers at regional level would decline substantially due to the distance. Thus, integrating both the fixed effects of metropolis and the fixed effects of regions into estimation enables us to isolate the effect of managerial interaction from the possible effect of socioeconomic factors. To help understand if there is any pattern in corporate financial policies across regions, Panel A of Table 8 presents, for each region, the average of residuals retrieved from the regressions controlling for firm-specific characteristics, year fixed effects, industry fixed effects, state regulation and bank conditions. It shows that regions with higher leverage, such as South Atlantic, East South Central, and West South Central, tend to have lower interest coverage, lower cash holdings, more long-term debt issues, and less payout, whereas regions with lower leverage, such as New England and North West, exhibit the opposite pattern. The region of Rocky Mountains is an exception. It has higher leverage, higher interest coverage ratio, and higher dividend payment which is financed with exceptional higher long-term debt issues and net equity issues. 23

The regression results are reported in Table 9. There are several points that deserve attention. First of all, the fixed effects of industry are still significant, which suggests that neither the fixed effects of metropolitan nor the fixed effects of region can subsume the industry effect on firm policies. Secondly, the integration of the fixed effects of regions does not alter the significance of the fixed effects of metropolitan areas, though the F value of the joint test declines to some extent. For example, the F value of the joint significance of metropolitan areas, for the variable cash holdings, decreases from 11.55 (in the specification with no region fixed effects) to 4.51 (in the specification with region fixed effects). Hence, the argument that the similarity in financial policies is caused by managerial interaction has been justified. The only exception is interest coverage. In the previous analysis, the metropolitan areas are jointly significant in the regression on interest coverage. However, after the fixed effects of regions are controlled, the metropolitan areas turn out to be insignificant any more, which indicates that managerial interaction is not a significant determinant in interest coverage. Thirdly, the significance of the join effect of regions varies with different financial policies. In particular, the fixed effects of regions are insignificant for interest coverage, total payout, and the likelihood of equity issuance, which means that the socioeconomic characteristics have no significant effects on these policy variables. However, the region fixed effects are significant for cash holdings, the level of common dividends and the likelihood of paying common dividend, the likelihood of conducting share purchase, the level of equity issuance, the level of net long-term debt issuance and the likelihood of net long-term debt issuance, and marginally significant for leverage and the level of share repurchase. Accordingly, we may conclude that these financial policies are affected by regional socioeconomic characteristics. Overall, Table 9 provides solid support for the social interaction hypothesis by showing that the fixed effects of local community are not a proxy for regional socioeconomic characteristics. We have seen that regional characteristics play a significant role in some financial policies. However, the fixed effects of region capture all kinds of variation across regions. To get a more explicit idea about what kind of regional characteristics works, we further check the role of culture in determining corporate financial policy. Specifically, we examine two aspects of 24

culture, trust and religion. La Porta, Lopez-de-Silanes, Shleifer and Vishny (1997) document that trust can promote cooperation within large organizations. Guiso, Sapienza, and Zingales (2005) examine the role of trust in the international setting and find that lower level of trust toward a country is associated with less trade with that country, less portfolio investment and less direct investment in that country. The claim that religious activity can affect economic performance has been affirmed by abundant evidence (See Iannaccone (1998) for a survey.) We obtain culture information from World Value Surveys (WVS). The surveys provide the census region where the interviews were conducted. The surveys were conducted in 1990, 1995, and 2000 respectively. Since culture is pretty stable over time, we use 1990 wave for the period 1988 to 1992, 1995 wave for the period 1993 to 1997, and 2000 wave for the period 1998 to 2003. We construct three variables, trust, church attendance, and the percentage of Protestant for each region. Trust is defined as the percentage of respondents who answered ”yes” to the question: ”Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?” The intensity of religious beliefs is proxied by the frequency of church attendance which is coded based on the question: ”Apart from weddings, funerals and christenings, about how often do you attend religious services these days?” We define church attendance as the percentage of participants who attend church at least once a week. As for religious denomination, we find the percentage of people who belong to Protestant based on the question: ”Do you belong to a religious denomination?” The summary statistics of cultural variables by region are displayed in Panle B of Table 8. The cultural variables vary much across regions. West north central has the highest level of trust, 0.451, then east north central ranks the second, while east south central has the lowest level of trust, 0.268. East south central has the highest church attendance rate, 54 percent of people attend church at least once a week. In New England, only 35 percent of people attend church at least once a week. West North Central has the highest percentage of Protestants, whereas it is only 18.6 percent for New England. Table 10 shows the regression estimates. We use F test to examine the joint significance of culture variables on each financial policy. Results suggest that cultural variables are jointly 25

significant for cash holdings, the likelihood of long-term debt issuance, the likelihood of paying dividend and the likelihood of buying back shares. The level of trust is only significantly related to the likelihood of paying dividend. Regions with high level of trust are more likely to pay dividend. The intensity of church attendance has negative effect on cash holdings, the likelihood of equity issues, cash dividend payment ratio, and positive effect on the likelihood of issuing debt. But only the relation with cash holdings is significant. The percentage of Protestants is positively related to leverage and negatively related to interest coverage. Both relationships are only marginally significant. In addition, the percentage of Protestant is positively related to the ratio of long-term debt issues and the likelihood of long-term debt issues, while it is negatively related to cash holdings and the likelihood of share repurchase. It suggests that regions with more Protestants tend to use more debt, hold less cash, and less likely to conduct share repurchase. The fixed effects of local community are strongly significant for all the financial policies (except for the ratio of share repurchase for which the local community effect is significant at 10 percent). Combining with the results shown in Table 8, it is evident that, though the culture variables are jointly significant for some financial policies, the fixed effects of region capture something not restricted to culture. Unfortunately, we do not have enough information to identify what else plays at the regional level other than culture.

6

Conclusion

This paper documents a significant local community effect on corporate financial policies. Moreover, the local community effect is consistently significant over different time periods and robust to alternative industry classifications and various estimation techniques. We further demonstrate that this local component is robust against state statutes on antitakeover and payout restrictions and banking market development at community level. We then propose a hypothesis to interpret the local community effect, i.e., the local community effect can be ascribed to social interactions among firm managers. We argue that firm managers who work close by may spread information to each other, observe each other’s management style, and then influence each other’s policy making. We test this hypothesis by differentiating local community effects 26

from region fixed effects, caused by regional socioeconomic characteristics, particularly culture. The evidence from our analysis supports the social interaction hypothesis. Our study provides a new dimension for future studies to examine corporate policy. Our findings suggest that, other than fundamental firm-specific characteristics or observable managerial traits, social interaction among firm managers also acts as a determinant in policy making. We believe that it would be very interesting to investigate into other corporate policies, such as investment policy, and corporate governance structure such as board structure, managerial compensation, to see whether and to what extent they exhibit conformity for firms located in the local geographic community.

27

Figure 1 Variable

Definition

Capital Structure Category Leverage

Sum of long-term debt (Compustat data item #9) and debt in current liabilities (#34) over total assets (#6)

Interest coverage

Earnings before depreciation, interest, and tax (#13) over interest expense (#15)

Cash holdings

Cash and short-term investments (#1) standardized by total assets

Net long-term debt issues

Long-term debt issuance (#111) minus long-term debt retirement (#114) scaled by total assets

Net long-term debt issues dummy

1 if a firm’s net long-term debt issues are greater than 0 in a year, and 0 if otherwise

Net equity issues

Sale of common and preferred stock (#108) minus any purchase of common and preferred stock (#115), scaled by total assets

Net equity issues dummy

1 if a firm’s net equity issues are greater than 0 in a year and 0 if otherwise

Payout Policy Category Common dividends

Ratio of common dividends (#21) over total assets

Common dividend dummy

1 if a firm pays common dividends in a year and 0 if otherwise

Share repurchase

Purchase of common and preferred stock (#115) minus any reduction in the redemption value of preferred stock (#56), then scaled by total assets

Share repurchase dummy

1 if a firm repurchases common stock in a year, and 0 if otherwise

Total payout

Common dividends plus purchase of common and preferred stock minus any reduction in the redemption value of preferred stock deflated by total assets

28

References Bebchuk, Lucian A., and Alma, Cohen, 2003, Firms’ decisions where to incorporate, Journal of Law and Economics 46, 383-425. Beber, Alessandro, and Daniela Fabbri, 2005, Who times the foreign exchange market? Corporate speculation and CEO characteristics? University of Lausanne and FAME, Working paper. Bertrand, Marianne, and Antoinette Schoar, 2003, Managing with style: The effect of managers on firm policies, Quarterly Journal of Economics 118, 1169-1208. Black, Sandra E., and Philip E. Strahan, 2002, Entrepreneurship and the availability of bank credit, Journal of Finance 67, 2807-33. Booth, Laurence, Varouj Aivazian, Asli Demirguc-Kunt, and Vojislav Maksimovic, 2001, Capital structures in developing countries, Journal of Finance 56, 87-130. Boyd, Robert and Peter H. Richerson, 1985, Culture and the evolutionary process. Chicago: University of Chicago Press. Davis, Gerald F., and Henrich R. Greve, 1997, Corporate elite networks and governance changes in the 1980s,American Journal of Sociology 103, 1-37. Davis, James, and J. Vernon Henderson, 2004, The agglomeration of corporate headquarters, Working paper, Brown University. Fama, Eugene F., and Kenneth French, 1997, Industry costs of equity, Journal of Financial Economics 43, 153-193. Fama, Eugene F., and Harvey Babiak, 1968, Dividend policy: An empirical analysis, Journal of the American Statistical Association 63, 1132-1161. Galaskiewicz, J., and S. Wasserman, 1989, Mimetic process within an interorganizational field: An empirical test, Administrative Science Quarterly 34, 454-479. 29

Gartman, Grant A., 2000, State Takeover Laws. Washington D.C.: Investor Responsibility Research Center. Garvey, Gerald T., and Gordon Hanka, 1999, Capital structure and corporate control: The effect of antitakeover statues on firm leverage, Journal of Finance 54, 519-546. Gompers, P., L. Ishii, and A. Metrick, 2003, Corporate governance and equity prices, Quarterly Journal of Economics 118, 107-155. Granovetter, Mark S., 1985, Economic action and social structure: The problem of embeddedness, American Journal of Sociology 91, 481-510. Guiso, Luigi, Paola Sapienza, and Luigi Zingales, 2005, Cultural biases in economic exchange, Working paper. Jayaratne, Jith and Philip E. Strahan, 1996, The Finance-Growth Nexus: Evidence from Bank Branch Deregulation, Quarterly Journal of Economics 101, 639-70. Harford, Jarrad, Sattar A. Mansi, and William F. Maxwell, Corporate governance and a firm’s cash holdings, Working paper, University of Washington. Haunschild, P. R., 1993, Interorganizational imitation: The impact of interlocks on corporate acquisition activity, Administrative Science Quarterly 38, 564-592. Hu, A. and P. Kumar, 2004, Managerial entrenchment and payout policy, Journal of Financial and Quantitative Analysis 39, 759-790. Jaffe, Adam B., Manuel Trajtenberg, and Rebecca Henderson, 1993. Geographic localization of knowledge spillovers as evidenced by patent citations, The Quarterly Journal of Economics 108, 577-98. Jayaratne, Jith, and Philip E. Strahan, 1996, The finance-growth nexus: Evidence from bank branch deregulation, The Quarterly Journal of Economics 111, 639-670. John, Kose, and Anzhela Knyazeva, 2006, Payout policy, agency conflicts, and corporate governance, New York University, Working paper. 30

Keynes, J.M., 1936, The general theory of employment, interest and money, Mcmillan, London. Klock, Mark, Sattar Mansi, and William F. Maxwell, 2005, Does corporate governance matter to bondholders? Journal of Financial and Quantitative Analysis 40, 693-719. Kusnadi, Yuanto, 2004, Corporate governance mechanisms and corporate cash holdings, EFA 2005 Moscow Meetings Paper. La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny, 1997, Trust in large organizations, American Economic Review 87, 333-338. La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny, 2000, Agency problems and dividend policies around the world, Journal of Finance 55, 1-33. Landes, D., 2000, Culture makes almost all the difference. In: Harrison, L.E., Huntington, S.P. (Eds.), Culture matters. Basic books, New York, NY. Levine, Ross, 2001, Financial structure and economic growth: A cross-country comparison of banks, markets, and development, Cambridge, MA: MIT Press. Lintner, John, 1956, Distribution of incomes of corporations among dividends, retained earnings, and taxes, American Economic Review 46, 97-113. Litov, Lubomir P., 2005, Corporate governance and financing policy: New evidence, Working paper, New York University. Malmendier, Ulrike, and Geoffrey Tate, 2005, CEO overconfidence and corporate investment, Journal of Finance 60, 2661-2700. Malmendier, Ulrike, Geoffrey Tate, and Jun Yan, 2005, Corporate financial policies with overconfident managers, Working paper. Marquis, Christopher, Mary Ann Glynn, and Gerald F. Davis, 2005, Community isomorphism and corporate social action, Academy of Management Review, Forthcoming. 31

Mizruchi, M.S., 1989, Similarity of political behavior among large American corporations, American Journal of Sociology 95, 401-424. Mizruchi, M.S., 1992, The structure of corporate political action: Interfirm relations and their consequences, Cambridge: Harvard University Press. McGurn, Patrick S., Sharon Pamepinto, and Adam B. Spector, 1989, State Takeover Laws, Washington D.C.: Investor Responsibility Research Center, 1989. Morgan, Donald, Bertrand Rime, and Philip E. Strahan, 2004, Bank integration and state business cycles, Quarterly Journal of Economics 119, 155-85. Newey, W. K., and K. D. West, 1987, A simple positive semi-definite, heteroskedasticity and autocorrelation consistent covariance matrix, Econometrica 55, 703-8. Novaes, Walter, and Luigi Zingales, 1995, Capital structure choice when managers are in control: Entrenchment versus efficiency, NBER working paper No. W5384. Petersen, Mitchell A., and Raghuram G. Rajan, 1995, The Effect of credit market competition on lending relationships, Quarterly Journal of Economics 110, 407- 444. Pirinsky Christo, and Qinghai Wang, 2005, Does corporate headquarters location matter for stock returns? Journal of Finance, Forthcoming. Rajan, Raghuram G., and Luigi Zingales, 1995, What do we know about capital structure? Some evidence from international data, Journal of Finance 50, 1421-60. Romano, Roberta, 1993. The Genius of American Corporate Law. Washington D.C.: AEI. Saidenberg, Marc R., and Philip E. Strahan, 1999, Are banks still important for large businesses? Current Issues in Economics and Finance 5, Federal Reserve Bank of New York. Stulz, Rene, and Rohan Williamson, 2003, Culture, openness, and finance, Journal of Financial Economics 70, 313-349. 32

Wald, John K., and Michael S. Long, 2006, The effect of state laws on capital structure, Journal of Financial Economics, Forthcoming. Weber, Max, 1930, The Protestant Ethnic and the Spirit of Capitalism. Harper Collins, New York. Zarutskie, Rebecca, 2005, Evidence on the Effects of bank competition on firm borrowing and investment, Journal of Financial Economics, Forthcoming. Zwiebel, Jeffrey H., 1996, Dynamic capital structure under managerial entrenchment, American Economic Review 86, 1197-1215.

33

Table 1 Summary Statistics The sample period is 1988-2003. The full sample consists of 39,287 firm-year observations for 4,104 different firms. Assets are in millions of dollars. Variable Panel A: Financial policies Leverage Interest coverage Cash holdings Net long-term debt issues Net equity issues Net long-term debt issues dummy Net equity issues dummy Common dividends Share repurchase Total payout Dividend dummy Share repurchase dummy Panel B: Control variables Assets Market to book ratio Return on assets Panel C: Bank condition variables Nonperformance loan Commercial loans to sales Herfindal index

Obs.

Mean

Median

STD

Q1

Q3

39,119 34,655 39,271 39,287 39,287 39,287 39,287 39,287 39,287 39,287 39,287 39,287

0.243 37.63 0.391 0.033 0.050 0.341 0.578 0.007 0.026 0.055 0.318 0.385

0.211 5.78 0.072 0.000 0.001 0.000 1.000 0.000 0.000 0.000 0.000 0.000

0.222 206.21 1.105 0.077 0.141 0.474 0.494 0.018 0.086 0.161 0.466 0.487

0.046 2.26 0.019 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

0.369 14.95 0.284 0.025 0.013 1.000 1.000 0.007 0.009 0.045 1.000 1.000

39,287 39,019 39,253

1218 1.943 0.002

182 1.416 0.037

3712 1.612 0.163

65 1.068 -0.012

657 2.144 0.077

438 438 438

0.006 0.291 0.219

0.004 0.219 0.197

0.008 0.279 0.146

0.002 0.122 0.101

0.008 0.371 0.302

34

Table 2 Location Effects on Financial Policies This table presents the results from fixed effects panel regressions with clustered standard errors. For each dependent variable, independent variables in row 1 include year fixed effect, industry fixed effects, logarithm of total assets, marketto-book ratio, and return on assets, while those of row 2 add metropolitan area fixed effects. Columns 2 and 4 report Fstatistics for the joint significance of industry fixed effects and metropolitan area fixed effects. Columns 3 and 5 show pvalues for the corresponding tests. The number of constraints is 42 for industry fixed effects and 27 for location fixed effects.

Financial policy Leverage

Interest coverage

Cash holdings

Net long-term debt issues

Net equity issues

Net long-term debt issues dummy

Net equity issues dummy

Common dividends

Share repurchase

Total payout

Dividend dummy

Share repurchase dummy

Industry

Metropolitan

effects

areas effects

Adjusted R-square

Number of Obs.

12.55 8.97