Project topics and materials
Get Complete Project Material File(s) Now! »
Performance and Behavior of Family Firms: Evidence from the French Stock Market
Introduction
While, since Berle and Means, financial economists have devoted a lot of attention to large, listed and widely held corporations, it turns out that most firms around the world have a dominant owner, in many instances the founding family.1 In addition, founding families are often involved in the actual management of the firm. In our own sample, which comprises the set of all listed French firms, more than 60% of the firms are still managed by their founding family. Even among the largest US firms, Anderson and Reeb (2003) report that some 16% of their sample of S&P500 firms are still managed by their founders or descendants. Therefore, the premises of the Berle and Means model of the corporation where (1) the CEO is not an owner and (2) ownership is dispersed, do not apply to most firms around the world.
The relevant view on world capitalism is thus that the typical large listed firm is owned, and frequently managed, by a family. This new perspective calls forth a research agenda on the specific features of dynastic management and ownership. Do family firms maximize profit? Are they more prone to building family empire at the expense of minority shareholders? Are they too prudent, slow reacting? On the contrary, are they more cool-headed and better at avoiding fads? More generally, do they behave any differently from the widely held corporations that academics know so well?
To bridge this gap, this paper provides evidence on the performance and behavior of family firms in France.
We believe the French example is of interest for two reasons. First, France is a continental European country, and as a consequence its financial institutions and history differ markedly from English-speaking countries, where most systematic studies on family firms have been conducted so far.2 In particular, family firms are much more prevalent even among the largest firms and therefore much more representative of the economy than in the US. The second reason is that, in contrast to many continental European countries, France also has a lot of widely held firms, which tend to be very much US-like managerial firms – no dominant owner and an entrenched management. This gives us access to a large enough control group to compare family firms to.
We collected data on some 1,000 corporations listed on the French stock market over the 1994-2000 period. Our panel has information on the firm (employment, corporate accounts, acquisitions, stock returns) and on the founding family (ownership, management). This dataset is supplemented with information on acquisitions, stock returns and detailed information on the wage bill and skill structure.
First, we provide cross-sectional evidence on the relative performance of family firms. Looking at accounting profitability, we find that family firms significantly outperform non family firms. Consistently with the existing literature on “founder effects”,3 we find evidence that founder-managed firms are very profitable. Also consistent with available evidence from the US (Anderson and Reeb (2003), Amit and Villalonga (2006)), family firms run by an outside CEO outperform widely held corporations. It thus seems that in France as in the US, families as large shareholders are, on average, good for performance. Much more surprisingly, we find that managers who are descendants of the firm’s founder also tend to do better than non family firms, and even marginally better than professional managers in our sample. Therefore, even if we set founders aside, family firms (whether or not run by a descendant) consistently outperform non family firms in France. A causal interpretation of such cross-sectional evidence is difficult since only the best performing firms may be transmitted to heirs. A potential solution involves looking at transmissions of control in our data (as in P´erez-Gonz´alez (2006)). While we may have too few transitions in our sample to do statistically powerful tests, we nevertheless observe that (1) descendants typically do not inherit the management of the best firms and (2) descendants whose firms leave the stock market (de-list) do not systematically underperform. Endogeneity biases are thus not likely to be large in our sample.
We then try to understand these differences in performance by considering the various determinants of corporate performance. Higher labor productivity seems to be the most significant explanation for the higher performance of founder-managed corporations. Turning to the difference in performance between professionally managed and* heir-managed family firms, we look separately at the management of labor and capital inputs.
First, family firms run by both professionals and descendants are paying, on average, lower wages. Nevertheless, this result could simply stem from a different workforce composition in these family firms, e.g. a larger fraction of unskilled workers. To account for such a possibility, we match our firm-level data with employee-level social security records, which allows us to control for the labor force composition in experience, seniority and occupations. We find that professional CEOs in family firms pay lower wages because they indeed employ less skilled employees. On the contrary, even after controlling for their firm’s skill structure, descendants still pay low wages. We also find that labor demand in heir-managed family firms responds significantly less to industry sales shocks, i.e. that heir managers seem to be smoothing out employment across the business cycle. Overall, these results are consistent with a view of heir-managed family firms as providing their workers with long-term implicit insurance contracts. Such contracts allow them to pay lower wages for better skills. And indeed, heir-managed family firms exhibit, compared to professionally managed corporations, a higher level of labor productivity.
We then investigate differences in the management of capital. Professional CEOs pay lower interest rates on their debt and tend to operate at lower capital to labor ratios. Compared to heir-managed family firms, or even widely held corporations, large acquisitions made by outside CEOs destroy less shareholder value in the long run. Although a-structural and thus to be interpreted with caution, the broad picture emerging from these results indicates that zon (2006), which takes advantage of the richness of Danish data to carefully study causality. Our paper has the advantage to supplement these studies for a large country of continental Europe: we find that descendants do not do worse – even slightly better – than professional managers of family or widely held firms. As already stressed out, apart from the causal effect of family management, many selection, endogeneity and simultaneity biases could yet be explaining this cross-sectional correlation.
More interestingly, our paper also complements the existing family firms literature by looking at effects of family ownership/management on other dimensions of firm behavior. Our robust finding that family firms pay lower wages is, to our knowledge, new to this literature. It is reminiscent of existing evidence on the relationship between wage levels and the separation of ownership and control in corporations. A decade old literature recently surveyed by Bertrand and Mullainathan (1999), along with their own findings, indeed finds that managerial slack in organizations partly takes the form of higher wages among employees. The other novelty of our paper is the analysis on the difference in management styles between hired CEOs and descendants of the founders: the data are consistent with outside CEOs bringing financial expertise and reducing the waste of capital, while heirs being able to commit to long term employment and therefore obtaining lower wages from their employees. Such results can be related to Bertrand and Schoar (2003)’s analysis of American CEOs’ management styles: they find strong differences between individuals in terms of investment policy, acquisition policy and financing policy. In particular, MBA graduates tend to be more aggressive in terms of leverage and acquisition policy. Our own results suggest that family management/ownership might be yet another dimension to explain such heterogeneity in management styles.4
This paper is structured as follows. Section 2 presents the data construction and describes its content. Section 3 provides more systematic evidence on corporate performance. Section 4 looks at differences in corporate behavior between family and non family firms and between descendants and professional managers within family firms. Section 5 concludes.
Data Description
Our dataset is a panel of French listed firms over the 1994-2000 period. We restrict ourselves to non financial, non real-estate firms. The construction of this dataset uses 5 different sources. First, annual corporate accounts are retrieved from the DAFSA yearbooks. These books cover the set of all listed firms in France. These books also provide us with the identity of the management team, and the stakes held by the main shareholders. Second, we hand collect information on family management and ownership for most of these firms using various sources (newspapers, firm websites, annual report…). Third, we use social security records to retrieve firm level information on wages, occupational structure, age and seniority structure and gender composition. Fourth, we collect information on the major corporate acquisitions realized by these firms over the 1994-2000 period. To do this, we use the French extract of SDC platinium, a worldwide database on corporate transactions. Fifth, we obtain stock prices for these listed firms from Euronext for the 1991-2002 period.
Family Business on the French Stock Market
Our definition of a family firm is very close to the one used by Amit and Villalonga (2006). We report a firm as a family firm when the founder or a member of the founder’s family is a blockholder of the company. We also impose as an additional condition that this block represents more than 20% of the voting rights.5 We refer to Appendix A.1.2 for more detailed explanations on the construction of our family firm variable.
Following Anderson and Reeb (2003), we then break down our sample of firms into four categories. All firms that are not family firms are called widely held firms. The listed firms that are controlled by widely held firms also belong to this category.6 When a family firm is still managed by its founder, we refer to it as a founder-managed family firm. As is detailed in A.1, this category also entails firms owned and managed by a successful raider.7 Heir-managed family firms are family firms where the current CEO is a descendant of the company’s founder. Finally, when a family firm is run by an outside, professional CEO, we refer to it as a professionally managed family firm.
To be able to compute accounting profitability measures properly, we restrict our study to non financial, non real estate companies. There are 2,973 observations in our panel (some 420 firms each year), for which we were able to retrieve the firm’s family status.8 Table 1 reports the fractions of the various types of firms in our panel. These fractions are computed without weight (line 1), weighted with book value of assets (line 2) and weighted using total employment, as reported in the accounting data (line 3).
As is apparent from Table 1.1, 70% of all firms present in the sample are family firms. This is a very large number, compared to what previous studies found for English-speaking countries. Looking at US listed firms from the S&P500, Anderson and Reeb (2003) find 35% of family-controlled companies, although they use a slightly different definition of family ownership. Looking at the largest 500 listed Canadian firms, Morck, Strangeland and Yeung (1998) find a share of 50% of family firms. Our sample is more consistent with the investigations of Faccio and Lang (2002), who look at the ultimate ownership of listed firms in continental European countries: using various data sources, they find in 1997, for France, 64% of family firms. Thus, family ownership appears much more pervasive in France than in English-speaking countries, even Canada. The surprising fact is, however, that the bulk of these family firms is still founder-controlled, since these account for 31% of the total. In contrast, only 18% of all firms investigated by Morck et al. (1998) in Canada are still managed by the initial entrepreneur. It seems that the French stock market may display more mobility than the sheer fraction of family firms might suggest. But the family status is also very persistent: heir-managed firms account for a large share of the total (24%) in the same proportion as widely held firms. Last, less than a fourth of all family firms are managed by a professional CEO: hence, even after the founder retires, the norm seems to be that an heir takes over control. Of course, the real importance of family firms is overstated by these figures. Lines 2 and 3 of Table 1.1 highlight the relative small size of family firms. In weighted terms, widely held firms account for almost two thirds of all firms. Founder-controlled corporations are especially small and only account for 10% of total employment.
Do Family Firms Differ from Other Firms?
Table 1.2 allows to look for systematic differences between the four types of firms we have defined in the previous section. First, family firms grow, on average, much faster than non family firms, but this is mostly due to the contribution of founder-managed corporations. For these corporations, sales growth stands around 16%, instead of 9% for the average listed firm. A similar picture arises for the ratio of market to book value of assets.9
In contrast, when we look at accounting profitability, all types of family firms do better than widely held firms. Founder-managed firms are the most profitable ones. That founders do better in terms of profits, growth and valuation is consistent with the extensive literature documenting “Founder effects”(see, for a survey, Adams, Almeida and Ferreira (2005), Fahlenbrach (2005)). In a cross section, founders tend to run firms with outstanding performance, the question being whether they are inherently good managers, or whether those founders who manage to keep control are only those who perform well. Using various instruments, Adams et al. (2005) suggest that selection issues are minor, and that almost all of the founder effect may be interpreted in a causal way. Using US data on listed firms, they find a founder effect on ROA of around 3 percentage points in OLS regressions and of around 2 points when using their instruments. Our cross tabulation suggests it might be even larger in the French context, although a multivariate analysis needs to be run to estimate such an effect.
Even when we set founders aside, family firms are still more profitable than widely held corporations, although to a lesser extent. The result is particularly striking for return on equity (ROE10), but is also present when we look at returns on assets (ROA11). This is not too surprising as far as professionally managed family firms are concerned, as these companies have the advantage of having large, long term shareholders. Anderson and Reeb (2003), and Amit and Villalonga (2006) find similar results in the cross sections of their sample of large US listed firms. For France, the concern could be that large shareholders might be using their voting rights to pursue value destroying projects that grant them private benefits. Results from Table 1.2 suggest that the benefits from monitoring outweigh these potential costs of expropriation. Finally, the main surprise from Table 1.2 is that family firms run by descendants also outperform widely held corporations in terms of profitability. The existing literature on large US firms provides mixed evidence: while Anderson and Reeb (2003) have similar results, Amit and Villalonga (2006) and P´erez-Gonz´alez (2006) exhibit opposite ones. In Canada, Morck et al. (1998) find that heir-managed Canadian firms underperform all other types of firms. Overall, the balance of evidence from North American studies tilts in the direction of underperforming heir management.
The obvious problem with the univariate approach, however, is that family status in cross tabulations may be a proxy for other variables, notably age and size. That family firms are smaller than non family firms is confirmed by an examination of Table 1.2, which also provides a comparison of the various types of firms in terms of size, age and capital structure. On all accounts, widely held firms are much larger than family firms, and slightly older too. This conceals, however, a great heterogeneity between family firms. For example, family firms run by an outside CEO are those who resemble the most widely held firms, both in terms of age and size. Firms run by descendants are on average smaller, but older than the average corporation in our sample. As expected, firms still run by their founders are young and very small.
Table of contents :
Introduction
1 Performance and Behavior of Family Firms: Evidence from the French Stock Market
1.1 Introduction
1.2 Data Description
1.2.1 Family Business on the French Stock Market
1.2.2 Do Family Firms Differ from Other Firms?
1.3 Multivariate Evidence
1.3.1 Empirical Strategy
1.3.2 Family Firms Outperform widely held Firms
1.3.3 Discussion on Endogeneity Biases
1.4 Management Styles in Family Firms
1.4.1 Breaking down Corporate Performance
1.4.2 Family Firms Pay Lower Wages
1.4.3 Descendants Can Commit on Long Term Employment
1.4.4 Outside CEOs Are More Financially Literate
1.5 Summary and Leads for Future Research
2 Bottom-Up Corporate Governance
2.1 Introduction
2.2 Data and Measurement Issues
2.2.1 Datasets
2.2.2 Constructing an Internal Governance Index
2.3 Internal Governance and Corporate Performance
2.3.1 Basic Results
2.3.2 Robustness Checks and Causality
2.4 Internal Governance and Acquisitions
2.5 External Versus Internal Governance
2.6 Conclusion
3 Optimal Independence in Organizations
3.1 Introduction
3.2 The Model
3.2.1 Set-Up
3.2.2 Equilibrium Concept
3.3 Organizational Homogeneity affects Reactivity
3.3.1 Implementer’s Effort Choice
3.3.2 The Decision Maker’s Project Choice
3.3.3 Summary and Discussion
3.3.4 Organizational Design
3.4 Robustness
3.4.1 Delegation of Organizational Design to the Decision Maker
3.4.2 Allowing for Financial Incentives
3.5 Applications
3.5.1 Bottom-Up Corporate Governance
3.5.2 Public Administration and Ideological Bias
3.5.3 Organizing for Change
3.6 Conclusion
4 The Corporate Wealth Effect: From Real Estate Shocks to Corporate Investment
4.1 Introduction
4.2 Data
4.2.1 Accounting and Governance Data
4.2.2 Real Estate Data
4.2.3 Loan Contracts
4.3 Real Estate Prices and Investment
4.3.1 Empirical Strategy
4.3.2 Main Results
4.3.3 City level results
4.4 Collateral and Debt
4.4.1 Debt Issuance
4.4.2 Debt Contracts
4.5 Corporate Governance and Investment Performance
4.6 Conclusion
A Performance and Behavior of Family Firms 1
A.1 Data
A.1.1 Corporate Accounts
A.1.2 Family Ownership and Management
A.1.3 Employment Data
A.1.4 Stock Prices
A.1.5 Acquisitions
A.2 Additional Tables
A.3 Breaking down Performance
B Optimal Independence in Organizations
B.1 Proofs
B.1.1 Proof of Proposition 1
B.1.2 Proof of Proposition 2
B.1.3 Proof of Proposition 3
B.1.4 Proof of Proposition 4
B.1.5 Proof of Proposition 5
B.1.6 Proof of Proposition 6
B.2 Proof of Proposition 7
B.3 The True Role of Uncertainty
C The Corporate Wealth Effect
C.1 Construction of the ptLandi Variable
