Cash Holdings, Corporate Governance and Financial Constraints

We examine the relation between cash holdings, corporate governance and financial constraints. We find that firms with weak shareholder rights hold less cash, in contrast to the predictions of agency theory. This result is partly due to the positive correlation that exists between shareholder rights measures and the degree of financial constraint faced by the firm. We show that control enhancing mechanisms reduce cash holdings of financially constrained firms. Control/ownership deviation gives these firms extra flexibility, enabling them to issue shares without blockholders losing control, and provides an alternative to high cash holdings.


INTRODUCTION
Determining the level of cash holdings is one of the most important financial decisions a manager has to make. When there is an inflow of cash, the manager may decide to distribute it to shareholders as dividends or through a share repurchase, invest it, or save it to cover future needs. This last solution is far from negligible: Bates, Kahle, and Stulz (2009) document that cash holdings have increased from an average of 11% of total assets in 1980 to 23% in 2006.
Cash holdings may serve commendable purposes, allowing the firm to seize profitable investment opportunities as they arise (Keynes, 1936); or they may, on the contrary, indicate agency problems (Jensen, 1986). In the first case, cash gives the firm flexibility, a particularly valuable resource when under financial constraints. But once the firm has ready access to capital markets, advantage in holding cash is less clear. Such situations may relate to the second case, where large cash holdings result from management's desire for greater discretion to finance any investments that may arise, even if they are value-reducing, because an increase in the firm's size tends to increase their own power.
The objective of this article is to compare two hypotheses for cash holdings: flexibility for financially constrained firms (firms with limited access to capital markets), and agency problems. We show that these hypotheses are not mutually exclusive. Measuring potential agency conflicts alternatively by the wedge between voting rights and cash flow rights and the number of anti-takeover provisions in place at French corporations, we show that the least financially constrained firms are also the firms with the weakest shareholder rights. These firms hold less cash than financially constrained firms, often characterized by higher shareholder rights. Our results shed light on the findings in several previous studies that the firms with the Electronic copy available at: https://ssrn.com/abstract=2154575 weakest shareholder rights had less cash, although their agency conflicts should lead them to hold large cash reserves.
Our study is based on past theoretical and empirical research that has attempted to explain the level of cash holdings on the basis of financial constraints or agency conflicts. One hypothesis argues that financially constrained firms should have higher cash holdings. For example, Kim, Mauer and Sherman (1998) develop a model in which firms' decisions depend on a tradeoff between costs related to external financing and to low returns on cash holdings, and the ability to make profitable future investments using accumulated cash. Firms with the highest external financing costs, which are more likely to be financially constrained, should hold more cash. Opler, Pinkowitz, Stulz and Williamson (1999) find that firms with the greatest access to capital markets, such as large firms and those with high credit ratings, and therefore the least financially constrained tend to hold lower ratios of cash to assets. Almeida, Campello and Weisbach (2004) show that in financially constrained firms cash levels are sensitive to cash flows, whereas in financially unconstrained firms the level of cash holdings is independent of cash flow.
The second hypothesis refers to agency theory, in which firms with severe agency conflicts are expected to have the highest cash holdings. In the presence of agency costs resulting from managerial discretion, managers, instead of paying out available funds to shareholders, will prefer to keep funds within the firm, even when available investments are value-reducing. Many studies, that measure agency costs by an index of shareholder rights, confirm this hypothesis.
The existing empirical studies fall into two main categories and lead to contradictory findings.
In the first category are comparative analyses that use cross-country samples and show that cash holdings are higher in countries with poor shareholder protection and less developed financial markets (Dittmar, Mahrt-Smith and Servaes, 2003). Studies in the second category examine a given country, and find that cash holdings increase with shareholder protection levels, as measured by a governance index (for example Harford, Mansi and Maxwell, 2008, for U.S.) To provide some explanation for these contradictory results, we examine them in the light of the presence of financial constraints. In the first case, access to external resources is difficult and costly for firms located in countries with poorly-developed capital markets. These firms tend to be financially constrained. The governance index used to assess investor protection is largely based on market development and the quality of the institutional and legal environment. There is thus a negative correlation between governance quality and the degree of firms' financial constraints. The results of cross-country studies emphasising governance characteristics in fact echo the results of studies of financially constrained firms: firms in less developed countries where governance quality is low are likely to be financially constrained, and therefore hold more cash (Khurana, Martin and Pereira, 2006). What remains to be done is to link the contradictory results for a given country to the degree of financial constraint. Our hypothesis is that minority shareholder rights measures and the firms' degree of financial constraint can be correlated variables. Harford et al. (2008) find a significant negative correlation between the size of the firm and its governance quality. The same correlation is noted in Gompers, Ishii and Metrick (2003): poorly governed firms are significantly larger than high-quality governance firms. If small firms are supposed to be more financially constrained than large firms, these findings entail a positive relationship between the degree of financial constraints faced by a firm and the quality of its governance.
In this paper, we examine this issue for French firms for the period 1998 to 2004. The typical French firm has concentrated ownership. Governance quality is assessed by two measures: the wedge between voting rights and cash flow rights and the number of anti-takeover provisions adopted. First, we find that firms with low governance quality have lower cash holdings. Second, we find that this result is partly due to the correlation between governance quality and financial constraints. The largest and least financially constrained firms are also those with the less concentrated ownership. The managers and blockholders of these firms seek protection through anti-takeover provisions, while small, highly-concentrated ownership firms need less protection against hostile buyers. We find that the wedge between voting rights and cash flow rights reduces the investment to cash flow dependence for financially constrained firms with limited access to external financing. It reduces the financial constraints for these firms by enabling them to issue new shares without forfeiting control, providing a source of flexibility that is an alternative to holding large cash reserves.
The remainder of the paper is organized as follows. Section 2 presents our hypotheses related to financial constraints and governance effects on cash holdings. Section 3 describes the corporate governance characteristics of French firms, and the data are discussed in Section 4.
Section 5 reports empirical results and our conclusions are contained in Section 6.

Financial Constraints and Cash Holdings
In a world of perfect capital markets, cash holdings are irrelevant. In such an environment, firms would have instant access to external financing. However, when there are transaction costs, 1 especially a fixed cost associated with raising capital, firms may find it useful to hold cash reserves. This incentive will be the greater when the firm is financially constrained, and has profitable investment opportunities and volatile cash flows. As a result, financially constrained firms are led to hold high levels of cash so as to be able to invest in any profitable project that arises. Financially unconstrained firms, on the other hand, have little to gain from cash holdings, as they have easy access to the capital markets, enabling them to take quiet advantage of profitable investment opportunities as they arise. Bates, Kahle and Stulz (2009)

Agency Conflicts, Governance and Cash Holdings
In the presence of agency costs, managers may stockpile cash to increase their discretion and pursue their own objectives at shareholder expense. They are thus in a position to seize investment opportunities without having to wait to raise funds, or to undertake value-reducing investments. If shareholders are concerned about potential agency conflicts, adding one dollar to the firm's cash holdings increases its value by an amount of less than one dollar, as Faulkender and Wang (2006), and Pinkowitz, Stulz and Williamson (2006) have shown. Dittmar and Mahrt-Smith (2007) also document that shareholders assign a lower value to an additional dollar of cash reserves when agency problems are likely to be greater. Opler et al. (1999) deduce several empirical predictions from the presence of agency costs resulting from managerial discretion. They argue that agency costs are greater when ownership Electronic copy available at: https://ssrn.com/abstract=2154575 is dispersed, so that widely-held firms are presumed to have greater cash holdings. Likewise, low-leverage firms, which are less subject to the discipline of capital markets, and firms protected by anti-takeover provisions, should hold more cash. While the negative relation between cash and leverage appears to be confirmed in most studies, tests of the other two predictions have produced mixed results.
The empirical tests of the effects of agency conflicts and governance quality on cash holdings have followed two main approaches. The first approach tests agency theory predictions on samples of countries with heterogeneous legal and institutional environments and varying degrees of capital market development. The second approach examines the effect of shareholder rights on cash holdings in a given legal context.
Using the first approach, Dittmar, Mahrt-Smith and Servaes (2003) analyse data for approximately 11,000 companies from 45 countries in 1998. They find that firms in countries with the lowest level of shareholder rights hold almost 25% more cash than firms in countries with the highest level of shareholder rights, and conclude that poorly-protected shareholders cannot force managers to disgorge excessive cash. However, these authors do not have firmlevel data on shareholder and governance features, and use only the country-wide measures of shareholder rights as developed by La Porta et al. (1998). Thus their study provides no clearcut conclusion at the firm level for the effect of agency costs on cash holdings. Pinkowitz, Stulz, and Williamson (2006) find that cash is less valuable in countries with poor financial and economic development, but their dataset uses a country-level measure of ownership concentration which once again does not allow for tests of the managerial agency problem at the firm level. Kalcheva and Lins (2007) analyse detailed ownership data for a sample of over 5000 firms from 31 countries and show that cash levels are higher when the management group and its family have effective control of a firm. The positive relation between cash holdings and effective managerial control is more pronounced when external shareholder protection is lower.
However, their study lacks detailed data on the divergence between the degree of control and cash flow rights. Thus, cross-country studies find that poor shareholder protection leads firms to hold more cash, but this conclusion is limited by the difficulty of obtaining the firm-level governance data that are needed to separate the effects of corporate governance from that of market development level and financial constraints on firms.
Within-country tests of agency theory predictions for cash holdings have mainly focused on the U.S. 2 Opler et al. (1999) find little support for the agency cost motive for cash holdings.
They find that management ownership has a positive effect on cash holdings only for low insider ownership (< 5%). They explain this result by arguing that shareholders in the U.S. enjoy strong protection and can therefore force managers to distribute excess cash. Mikkelson and Partch (2003) find no differences in the ownership structure of cash-rich firms and firms with normal cash levels and Opler, et al., (1999) find that anti-takeover amendments have no significant effect on cash holdings. This finding may result from two opposite effects: while anti-takeover devices may encourage entrenched managers to hoard more cash, they can also attract buyers who could use that cash to finance their acquisition. However, Harford (1999) finds that the likelihood of a firm becoming a takeover target is significantly negatively related to holding excess cash. Harford, et al. (2008) examine the relation between the management of cash holdings and an index of shareholder rights, developed by Gompers, Ishii, and Metrick (2003). In contrast to the cross-country research, they find that firms with weak shareholder rights have smaller cash reserves and dissipate cash more quickly, primarily through acquisitions, than managers of firms with strong shareholder rights. Harford, Kecskes and Mansi (2012) find that firms with longer investor horizons, an external governance mechanism, hold more cash. Their results are thus consistent with previous evidence that better governed firms hold more cash.

Financial Constraints and Corporate Governance Effects
Overall, the two approaches used in prior empirical tests of the agency effects on cash holdings lead to contradictory results. An alternative interpretation of these findings is that in countries with weak shareholder rights, firms find it more costly to raise external funds, and therefore, they hold more cash. For example, Rajan and Zingales (1998) and Demirguc-Kunt and Maksimovic (1998) find that industries or firms that rely on external financing exhibit greater growth in financially developed countries. Love (2003) shows that in countries with a lower level of financial market development firms hold more cash. Using firm-level data for 35 countries covering about 12,782 firms for the years 1994-2002, Khurana, Martin and Pereira (2006 find the sensitivity of cash holdings to cash flow decreases with financial development. Beck, Demirguc-Kunt and Maksimovic (2006) conclude that the impact of financial underdevelopment is more financially constraining on small than large firms. Dittmar et al. (2003) show that the impact of the market-to-book ratio on cash is lower in countries with weak shareholder rights, indicating that cash is not simply held to cover future investments, but due to managerial discretion. However, the lack of firm-level governance data prevents a disentangling of governance characteristics from financial constraints.
In within-country studies, the least financially constrained firms are also those with the most dispersed ownership, and therefore are firms that are the most likely to introduce antitakeover provisions. Thus, the finding that firms with weaker shareholder rights hold less cash may be explained by the fact that these firms encounter the weakest financial constraints, and therefore are the least in need of cash. Governance indexes such as those constructed by Gompers et al. (2003), which attribute considerable importance to anti-takeover amendments, also show positive correlation with firm size, which may be a rough measure of financial constraints.
Our hypotheses are therefore as follows: -firms with weaker minority shareholder rights, as measured by the number of antitakeover devices adopted and the control-ownership deviation, also tend to be the firms with few financial constraints, and both factors may explain their low cash holdings.
-For firms with a blockholder, the wedge between voting rights and cash flow rights of the main blockholder can alleviate financial constraints, by enabling them to issue new shares without the blockholder losing control.

THE GOVERNANCE CHARACTERISTICS OF FRENCH FIRMS The French Institutional Setting
In France, as in most Western European firms, ownership tends to be concentrated. Faccio and Lang (2002) show that 36.93% of European firms are widely-held, whereas in France this percentage decreases to 14%. Similarly, 44.29% of European firms are controlled by a single family, versus 64.82% in France. Although Faccio and Lang assume that 20% of the voting shares suffices for control, we use a more restrictive definition of control, considering a firm to be controlled when the largest shareholder has at least one third of voting rights, given that this level represents a blocking minority. Under French law, there are two types of shareholders' meeting. Decisions by ordinary meetings, which approve the accounts, appoint and dismiss directors, and approve bond issues. These measures require approval of a majority of 50% of voting rights. Decisions by extraordinary meetings, which are empowered to make all decisions amending the charter, require a two-thirds majority of voting rights.

Corporate Governance Characteristics
We use two measures of corporate governance quality. The first one is a governance index, the second one is the wedge between the voting rights and the cash flow rights due to differential voting right shares of the largest shareholder.
Governance index. French law allows various devices intended to discourage takeovers or dissociate voting rights from cash flow rights. These devices reduce minority shareholder rights.
We construct a governance index by adding one point for each provision that reduces minority shareholders' protection. The following types of provisions are taken into account: pyramid at the 33.33% threshold, dual class shares, double voting rights, a French specific mechanism for registered shares that have been held for more than x years (where x is between 2 and 4 years) 3 , voting caps, firm voting thresholds disclosure 4 , and limited partnership corporate structure. 5 Control/ownership divergence. We calculate the wedge between voting rights and cash flow rights as the ratio of (voting rights minus cash flow rights) to cash flow rights for the largest shareholder.

Sample Selection
Our sample incorporates all non-financial firms listed on Euronext Paris at the end of the years 1998, 2000, 2002 and 2004. Accounting and financial data are collected from the Compustat and Datastream databases. Governance data are collected manually from the firms' annual reports 6 , since computer-based databases supply only a limited amount of governance information, state shareholdings as percentages of capital rather than percentages of voting rights, and supply no information on the anti-takeover provisions needed for constructing our governance index. After eliminating firms for which required data are missing, our final sample contains 991 firm-years.

Definition of Variables
We use a number of proxy variables to test our hypotheses. To assess the relation between cash holdings and a firm's control structure, we estimate regressions in which the firm's cash holding is the dependent variable, measured as the ratio of cash and marketable securities to total assets. 7 The regressions include ownership variables and a variety of control variables that previous research has shown to be useful to explain variation in cash holdings. 8 Definition of financial constraints. We use four measures to assess the extent of the financial constraints faced by the firm. 9 First, we rely on Whited and Wu (2006) index (WW index), equal to: Where TLTD it is the ratio of the long term debt to total assets; DIV POS it is an indicator that takes the value of one if the firm pays cash dividends; SG it is firm sales growth; LNTA it is the natural log of total assets, ISG it is the firm's three-digit industry sales growth and CF it is the ratio of cash flow to total assets.
The WW index decreases with the level of financial constraints. A firm is considered financially constrained (unconstrained) if the WW index lies in the lowest (highest) three deciles of the distribution.
The second measure, inspired by Almeida et al. (2004), is based on the size of the firm's assets. A firm is considered financially constrained (unconstrained) if the size of its assets lies in the first (last) three deciles of the distribution.
The third measure of financial constraints we use is the stock market segment in which a corporation is listed. On the French market, deferred settlement is possible for a subset of highly liquid shares: this market is called the "Service à règlement différé" or SRD. The firms belonging to this segment are probably the least financially constrained. For all other shares, only cash trading is possible. The firms that are only cash traded are supposed to be financially constrained.
As a fourth measure, we consider that all family firms are financially constrained. Family firms are likely to face important financial constraints, as their access to capital markets may be limited by their desire to retain control. A family firm will hesitate to issue new shares if the controlling block is not in a position to participate in the new issue. We include a dummy variable, equal to one for all family-controlled firms in which the main shareholder is an individual or a group of individuals. Non-SRD and family dummies are broader measures of financial constraints, and lead to a larger number of firms being classified as constrained than the classification based on WW index or size. 10 Ownership variables. Ownership concentration could have important implications for potential agency costs, leading to two opposite cash holding predictions. Large shareholders can monitor managers more effectively, resulting in lower than expected agency costs (Shleifer and Vishny, 1986). From this perspective, ownership concentration should be associated with a lower cost of external financing, reducing the need to hold cash balances. However, Shleifer and Vishny (1997) argue that in controlled firms, there are agency costs between minority and controlling shareholders. Large shareholders can impose costs on other shareholders in the form of wealth redistribution, so that ownership concentration may lead to large cash holdings. We measure ownership concentration as the percentage of capital held by the largest shareholder. We further introduce a dummy variable that takes the value of one if a financial institution owns more than 5% of the capital. A financial institution shareholder can reduce the financial constraints on the firm, as it can facilitate access to loans and other external financing. We also include the governance index and the discrepancy between cash flow rights and control rights held by the controlling shareholder, as defined above.
Firm-specific variables. We measure cash flow as earnings before interest, dividends, and taxes, plus depreciation, divided by assets. Following Almeida et al. (2004), we hypothesize that financially constrained firms should show positive cash flow sensitivity to cash, while unconstrained firms' cash savings should not be systematically related to cash flow.
We employ the ratio of net working capital to net assets as a proxy for liquid asset substitutes and predict a negative relation between the firm's cash holdings and its liquid assets.
We include the ratio of capital expenditures to assets and expect a negative relation with cash holdings.
To control for leverage, we use the ratio of long-term debt plus short-term debt divided by total assets. We include the squared leverage ratio. As firms with high leverage are more likely to be financially constrained, they can be expected to increase their cash balances for precautionary motives, and the relation between cash holdings and leverage, which is negative at low levels of leverage, can become positive at high levels. Since financially constrained firms may also use more short-term debt, we include the ratio of short-term debt to total debt.
A growth firm, if faced with a cash shortage, has to give up valuable investments. As a proxy for firms' growth opportunities we use Tobin's Q, defined as the ratio of the market value of equity plus the book value of debt to the book value of assets. We include the R&D expenseto-sales ratio as a measure of the potential for financial distress costs and asymmetric information, and expect that firms with higher R&D expenses will hold more cash reserves.
Firms that do not report R&D expenses are considered to be firms with no R&D expenses.
Firm size is measured as the logarithm of the book value of assets. We expect a negative relation between cash and size because smaller firms are more financially constrained and face higher external financing costs than larger firms.
Dividend payouts may reduce cash holdings, or alternatively, a firm that pays dividends can raise funds by cutting the dividend and therefore needs less cash. We use a dividend dummy, that equals 1 if a firm pays a dividend for a given year and zero otherwise.
Industry and year dummies are included in all specifications.

Univariate Tests
In Table 1, descriptive statistics are reported for the total sample and subsamples of constrained and unconstrained firms, using the four classification criteria, WW index, size, stock market segment and family firms.
[Insert Table 1 here] The number of financially constrained firms varies from 30% to 66% depending on the reference criteria (Table 1). On average, French firms hold 14.1% of their assets in the form of cash. This percentage is 17.4% (19.2%, 14.9% and 14.7% respectively) for financially constrained firms based on the WWindex (size, market segment and family firms) criterion, and from 11.6 to 13.0% for unconstrained firms. Financially constrained firms are smaller capitalisation firms, under all criteria, with lower leverage, lower cash flow to assets ratio, higher R&D expenses and a higher working capital to assets ratio. This last result, which is robust when sector subsamples are considered, may have two explanations. First, the smallest firms are not as good at managing their working capital requirements or are less effective at transforming trade receivables and stocks into liquid assets. Second, they are more likely to come under pressure from customers for longer payment terms and from suppliers for short settlement times.
Financially constrained firms are also different from unconstrained firms in the ownership structure and governance characteristics (Table 2). They are more closely-held. Based on the WW index the main shareholder of a constrained firm holds 53% of the capital compared to 36% in unconstrained firms. The average difference between the percentage of voting rights and the percentage of cash flow rights is significantly lower for constrained firms. Constrained firms are mostly family controlled-firms Finally, only 10 to 17% (depending on the financial constraint measure) of constrained firms have a financial institution as a shareholder, compared to 25% to 30% of unconstrained firms. Thus, family-controlled firms are more often financially constrained, presumably to maintain voting control, so that their options for issuing new shares are limited. Group subsidiaries and other controlled firms, in contrast, have access to intragroup financing and dispersed-ownership firms have ready access to financial market resources, and thus are presumed to be less financially constrained.
The governance index, with an average 1.54, is 1.25 (2.03) for constrained (unconstrained) firms based on the WW index and 1.42 (1.76) on the family firm criterion. Constrained firms have significantly fewer anti-takeover provisions than unconstrained firms, except for greater use of double voting rights. Since they are more closely-held, they feel less need for defence against a possible buyer. Governance quality, as measured by the index, grows weaker as the firm is more financially unconstrained. This result is not specific to France (Harford et al., 2008or Gompers et al., 2003.
[Insert Table 2 here] These observations confirm that the cash ratio, but also the degree of financial constraint, decrease as the governance index rises, supporting our hypothesis that the impact of governance on firms' cash holdings partly results from the fact that they are subject to different financial constraints.

Governance and Financial Constraints
To complement the univariate analysis, we estimate several regression models. Table 3 reports the estimation results from the following regression model, for the full sample and subsamples of size constrained and unconstrained firms.
CASHi,t =  + governance characteristicsi ,t+ n firm specific variables + industry dummies + year dummies +I,t We use OLS regressions predicting cash/asset ratios over the period 1998-2004, using the independent variables described earlier (Table 3). The dependent variable is the ratio of cash and marketable securities to assets. Firms are allowed to enter and leave the panel. All models include industry and year indicators.
[Insert Table 3 here] The first column of Table 3 reports estimates using OLS regressions for the full sample without governance variables. Cash holdings increase significantly with the cash flow to assets ratio, the Tobin's Q, and the R&D to sales ratio. Cash holdings decrease significantly with the capital expenditure to assets ratio, the net working capital to assets ratio, the dividend dummy and the short-term debt ratio. We find a non-linear relation between cash holdings and leverage.
Overall, these results are similar to those reported by Opler et al. (1999) and confirm previous findings with the exception of size, which has no significant impact on cash holding. The next columns of Table 3 add governance variables to these regressions and the previous results are unchanged (models 2 and 3). The results indicate that there is no evidence of a potential conflict between majority and minority shareholders, given that the percentage of capital held by the largest shareholder has no impact on cash holdings. However, the presence of a financial institution with more than 5% of capital significantly reduces cash levels. This result is robust after controlling for the percentage of shares owned by the main shareholder or the fact that the firm is controlled versus widely-held. Monitoring by a major shareholder also has less of an effect on the level of cash holdings than the presence of a financial institution shareholder, even when only a small fraction of the capital is held. Thus, the involvement of a bank reduces financial constraints by facilitating both the firms' access to loans and raising capital on the markets. We find a negative relation between our governance quality measures (governance index or control/ownership divergence) and firm cash holdings, and therefore a positive relation between the strength of shareholder rights and cash reserves. This result is robust with respect to whether a main shareholder is present, the percentage that the main shareholder owns, and its identity. It is consistent with the findings of Harford et al. (2008) for U.S. firms, and contradicts the prediction by Opler et al. (1999) that firms protected by anti-takeover provisions are likely to hold more cash. Across the full sample, we conclude that poorly controlled managers do not appear to stockpile cash for their own private benefits.
The same regressions are estimated including an interaction term between the control/ownership wedge and four dummy variables indicating financially constrained firms (lowest three deciles of WW index, first three deciles of size, non SRD firms, family firms) or financially unconstrained firms (first three deciles of WW index or size) (models 4 to 9). We find that constrained firms hold significantly more cash than other firms (+2.2% for non-SRD firms, +2.6% for WWindex constrained firms, +3.5% for size constrainted firms and +3.1% for family firms). Further, the interaction term between wedge and the financial constraints measure has a negative and statistically significant coefficient for WW index and size constrained firms as well as family firms, meaning that the control/ownership wedge reduces cash only when firms are financially constrained. When we introduce an interaction term between the wedge and the dummy variable for WW or size index unconstrained firms, the coefficient becomes significantly positive and larger than the coefficient of the wedge itself, so that the net impact of the wedge on cash becomes positive. For unconstrained firms, divergence from one-share / onevote rule increases cash hoarding.
These results establish that governance quality (which is greater when there are few antitakeover provisions or smaller control/ownership divergence) has no impact or even a positive impact on the cash holdings of financially unconstrained firms, and a negative influence on the cash holdings of financially constrained firms.

Investment to Cash Flow Sensivity and Control-Ownership Deviation
We next examine the impact of governance on investment to cash flow sensitivity. According to Fazzari, Hubbard, and Petersen (1988), when firms face financing constraints, investment spending will vary with the availability of internal funds, rather than just with the availability of positive net present value projects. Even if several authors have challenged the interpretation that higher investment-cash flow sensitivity is evidence of financing constraints (for example Kaplan andZingales, 1997 andCleary, 1999), Allayannis and Mozumbar (2004) reaffirm the original results when negative cash flow observations are excluded. Table 4 presents several models regressing the capital expenditure to asset ratio on cash flow and several control variables. We first document that investment is sensitive to cash flow availability. We then introduce an interaction variable between cash flow and control-ownership wedge and find that the wedge between voting rights and ownership reduces the investment to cash flow sensitivity. To distinguish between financially contrained and uncontrained firms, we introduce interaction variables between cash flow, wedge and dummy variables for financially constrained and unconstrained firms. We find that the control-ownership wedge significantly reduces the investment to cash flow sensitivity for constrained firms, whatever classification we use to measure financial constraints. The wedge between voting rights and cash flow rights reduces the dependence of financially constrained firms on internal funds. In the next section, we examine a possible explanation for this result: the wedge enables financially constrained firms to issue new shares without blockholders forfeiting control.

Control-Ownership Deviation and Seasoned Equity Offerings
We estimate a binomial logit model of the decision to issue equity at least once during the period 1997-2005 (Table 5). Each coefficient provides an estimate of how an increase in a specified variable affects the marginal likelihood that a firm issues equity. We find that highly leveraged firms, with RD expenses, have a higher probability to issue equity, whereas closely held firms, with a financial shareholder, have a lower probability to issue equity. Our results indicate that constrained firms (WW index or market segment) issue significantly less during the period. The wedge between voting rights and cash flow rights increases the probability to issue equity for financially constrained firms (all measures except for family firms). This result means that the wedge reduces financial constraints for constrained firms. It makes it easier for them to issue equity without the controlling blockholder loosing the control of the firm. For financially constrained firms, control/ownership deviation represents an alternative source of flexibility to holding large cash reserves.

CONCLUSION
The cash held by firms may result from the need for financial flexibility to undertake profitable investment opportunities as they arise, but also from self-interested managers' desire to hoard cash for their own private benefits. Previous studies testing this last explanation have produced contradictory results. Within country studies highlight a positive relationship between governance quality and cash holdings, while cross-country studies, stress a negative one. We reconcile these two results by highlighting the links between financial constraints and shareholder rights. In cross-country studies, firms in developing countries, characterised by severe financial constraints and weak shareholder rights, hold large amounts of cash. We show that within the context of France, managers of the largest and least financially constrained firms introduce anti-takeover provisions, leading to low quality governance. These firms hold small amounts of cash. Financially constrained firms are often small, hold large amount of cash and have limited access to external financing. For these firms, the wedge between voting rights and cash flow rights reduces the investment to cash flow dependence and enables them to issue new shares without forfeiting control, providing a source of flexibility that is an alternative to holding large cash reserves. 1 See, for example, Keynes (1936) or Miller andOrr (1966) 2 Other results are available for the UK (Ozkan and Ozkan, 2004). 3 In contrast to dual-class shares, the voting right is attached not to the share but to the shareholder. If the shareholder with double voting rights sells the share, the buyer acquires a share with only one voting right. All shares, (with either single or double voting rights), are traded on the same trading line, so that the price of the shares is identical. 4 Voting thresholds place shareholders under an obligation to inform the company when their shareholding reaches a level of x% of the capital or voting rights (where x is 0.5% or more), under penalty of a 2-year suspension of voting rights. This obligation may be a deterrent for a hostile takeover bid. 5 Limited partnership firm is a corporate structure that has partners with unlimited liability who manage the firm, and passive partners who provide financing and whose liability is limited to their equity contribution. As the active partners' shares are non-transferable, and the charter can make removal of the managing partner practically impossible, use of this form of legal entity is a highly effective anti-takeover provision. 6 Governance data are partly taken from the database used in Ginglinger and Hamon (2012), with additional information collected from annual reports. 7 We checked the robustness of our results by measuring cash holding as the ratio of cash to total assets minus cash, and as cash-to-sales ratio. These alternative measures lead to the same conclusions. 8 Besides the transaction cost model, the static trade-off model (Myers, 1977) and the pecking order model (Myers and Majluf, 1984) also provide several testable hypotheses. 9 We do not use Kaplan and Zingales (1997) index, whose validity to measure financial constraints has been questioned by Almeida et al (2004) and Whited and Wu (2006). Another candidate is credit ratings; however there are too few corporations with a credit rating available on the French market. 10 We use an alternative measure to assess financial constraints: the firm's ability and/or intent to pay out funds to shareholders. We define three categories: financially constrained firms with no dividend payouts and no share repurchases in the year concerned, firms with either dividend payouts or share repurchases, and financially unconstrained firms with both dividend payouts and share repurchases. We also perform the tests using the ratio of payouts to assets, assigning to the financially constrained (unconstrained) group those firms in the bottom (top) three deciles of the annual payout distribution. For both measures, our results are qualitatively unchanged.

TABLE 1 Description of Firm Specific Variables
This panel provides summary statistics on key variables for our sample of 991 firm-year observations (end year 1998, 2000, 2002 and 2004), and sub samples of constrained and unconstrained firms. Constrained firms (WW Index) are those in the bottom three deciles of the WW Index distribution. Unconstrained firms (WW Index) are those in the top three deciles of the WW Index distribution. Constrained (unconstrained) firms according to the size are those in the bottom (top) three deciles of the size distribution. Constrained firms (Stock market segment) are firms for which only cash trading is possible. Unconstrained firms (Stock market segment) are firms belonging to the SRD segment. Constrained firms (Family Control) are family-controlled firms in which the main shareholder is an individual or a group of individuals. Unconstrained firms (Family Control) are other controlled firms and widely held firms. Cash/Assets is the ratio of cash and equivalents to total assets. Cash flow is earnings before interest & taxes plus depreciation & amortization. Net Working Capital is working capital minus cash and equivalents. Leverage is total debt over total assets. Short term debt is the ratio of shortterm debt to total debt. Q is measured as the market value of equity plus the book value of debt, divided by the book value of assets. R&D/Sales is the R&D expenses to sales ratio. Size is defined as the natural logarithm of total assets. Dividend Dummy equals 1 if the firm pays dividends. WW Index is the index constructed by Whited and Wu (2006) to assess the financial constraints faced by the firm. N is the number of firm-year observations. ***, **, * indicate coefficients significance level for the test for equality of means (Student test), and the test for equality of medians (Wilcoxon test): 1%, 5% and 10% respectively.

TABLE 2 Ownership and Governance Variables
This panel provides summary statistics of ownership and governance variables for our sample of 991 firm year observations (end year 1998, 2000, 2002 and 2004), and sub samples of constrained and unconstrained firms. Constrained firms (WW Index) are those in the bottom three deciles of the WW Index distribution. Unconstrained firms (WW Index) are those in the top three deciles of the WW Index distribution. Constrained (unconstrained) firms according to the size are those in the bottom (top) three deciles of the size distribution. Constrained firms (Stock market segment) are firms for which only cash trading is possible. Unconstrained firms (Stock market segment) are firms belonging to the SRD segment. Constrained firms (Family Control) are family-controlled firms in which the main shareholder is an individual or a group of individuals. Unconstrained firms (Family Control) are other controlled firms and widely held firms. SH1 is the percentage of capital held by the largest shareholder. VRSH1 is the percentage of voting rights held by the main shareholder. Wedge measures the control-ownership deviation for the largest shareholder and is equal to (VRSH1-SH1)/SH1. Widely-held is the percentage of widely held firms. Famcont is the percentage of family controlled firms. Othercont is the percentage of firms controlled by banks or insurance companies, by non-financial companies, and state-owned firms. Financial shareholder is a dummy that equals 1 if a bank or an insurance company holds more than 5% of the capital. Gov Index is an index which adds one for the following anti-takeover mechanisms. Dual Class Shares, Pyramid, Limited Partnership, Double Voting Rights, Voting Caps, Voting Thresholds are dummies that equal 1 if the mechanism is present. N is the number of firm-year observations. ***, **, * indicate coefficients significance level for the test for equality of means (Student test), and the test for equality of medians (Wilcoxon test): 1%, 5% and 10% respectively.

TABLE 3 Governance and Financial Constraints
This table reports the estimated coefficients from regressing cash holdings on governance variables and various control variables for the total sample (end year 1998, 2000, 2002 and 2004). The dependent variable in all models is the ratio of cash and equivalents to total assets. Cash flow is earnings before interest & taxes plus depreciation & amortization. Net Working Capital is working capital minus cash and equivalents. Leverage is total debt over total assets. Leverage² is the squared leverage ratio. Short term debt is the ratio of short-term debt to total debt. Q is measured as the market value of equity plus the book value of debt, divided by the book value of assets. R&D/Sales is the R&D expenses to sales ratio. Size is defined as the natural logarithm of total assets. Dividend Dummy equals 1 if the firm pays dividends. SH1 is the percentage of capital held by the main shareholder. Financial shareholder is a dummy that equals 1 if a bank or an insurance company holds more than 5% of the capital. Gov Index is the index measuring the number of anti-takeover mechanisms adopted by the firm. Wedge measures the control-ownership deviation for the largest shareholder and is equal to (VRSH1-SH1)/SH1. WW (size) Constrained equals 1 if the firm is financially constrained according to the WW Index (size). WW (size) Unconstrained equals 1 if the firm is financially unconstrained according to the WW Index (size). Family Control equals 1 if the firm is a family-controlled one. Non-SRD equals 1 if only cash trading is possible for the firm. Industry dummy variables are constructed for each industry, defined by industry classification of Euronext. All models are OLS models which use firms for which we have data for at least one year and include year and industry dummies. N is the number of firm-year observations. t-statistics are in parentheses. ***, **, * indicate coefficients significance level: 1%, 5% and 10% respectively. Intercept 0,296*** (9,484)

TABLE 4 Control-Ownership Deviation and Investment to Cash Flow Sensitivity
This table reports the estimated coefficients from regressing investment on governance variables and various control variables for the total sample and subsamples of constrained firms (end year 1998, 2000, 2002 and 2004). Constrained firms (WW Index) are those in the bottom three deciles of the WW Index distribution. Constrained firms (Family Control) are family-controlled firms in which the main shareholder is an individual or a group of individuals, including control by non-listed companies. Constrained firms (Stock market segment) are firms for which only cash trading is possible. The dependent variable in all models is Capital Expenditure/ Assets. Size is defined as the natural logarithm of total assets. Q is measured as the market value of equity plus the book value of debt, divided by the book value of assets. R&D Dummy equals 1 if R&D/Sales is positive, where R&D/Sales is the R&D expenses to sales ratio. Net Working Capital is working capital minus cash and equivalents. Cash flow is earnings before interest & taxes plus depreciation & amortization. SH1 is the percentage of capital held by the main shareholder. Wedge measures the control-ownership deviation for the largest shareholder and is equal to (VRSH1-SH1)/SH1. Interaction terms between Cash flow, Wedge and dummy variables indicating financially constrained firms (WW Constrained, Size constrained Family Control and Non-SRD) are included in different models. All models are OLS models which use firms for which we have data for at least one year and include year and industry dummies. N is the number of firm-year observations. t-statistics are in parentheses. ***, **, * indicate coefficients significance level: 1%, 5% and 10% respectively.

TABLE 5 Control/Ownership Deviation and Seasoned Equity Offerings
This table reports the estimated coefficients from regressing issuing equity on governance variables and various control variables for the total sample (end year 1998, 2000, 2002 and 2004). The dependent variable in all models is a dummy variable which takes the value of 1 if the firm has issued equity at least once during the period 1997-2005. Cash flow is earnings before interest & taxes plus depreciation & amortization. Leverage is total debt over total assets. Short term debt is the ratio of short-term debt to total debt. Q is measured as the market value of equity plus the book value of debt, divided by the book value of assets. R&D/Sales is the R&D expenses to sales ratio. Size is defined as the natural logarithm of total assets. SH1 is the percentage of capital held by the main shareholder. Financial shareholder is a dummy that equals 1 if a bank or an insurance company holds more than 5% of the capital. . Wedge measures the control-ownership deviation for the largest shareholder and is equal to (VRSH1-SH1)/SH1. WW (size) Constrained equals 1 if the firm is financially constrained according to the WW Index (size). Non-SRD equals 1 if only cash trading is possible for the firm. Family Control equals 1 if the firm is a family-controlled one. Interaction terms between Wedge and one of the dummy variables indicating financially constrained firms (WW Constrained, size, Non-SRD and Family Control) are included in different models. All models are Logit models which use firms for which we have data for at least one year and include year and industry dummies. N is the number of firm-year observations. P-values are in parentheses. ***, **, * indicate coefficients significance level: 1%, 5% and 10% respectively.