Concentrated control and corporate value: a comparative analysis of single and dual class structures in Canada

Abstract

This study directly examines the empirical relationship between corporate value and three distinct ownership structures using data from Canada, where the security laws and shareholder protection conditions are similar to those of the US (La Porta et al., 1999) but corporate control tends to be more concentrated (Holderness et al., 1999). Ownership structure is classified in three ways: dual class firms, single class closely-held firms and widely-held firms. The focus of this article is to test for the impact of concentrated control on corporate value using either dual class or single class closely-held ownership structure. The empirical results, using both fixed and random effects estimation methods, show that after controlling for size, financial leverage, percentage of outside directors and industry differences, dual class companies sell at a significant discount compared to closely-held single class companies. Consistent with Claessens et al. (2002), and Gompers et al. (2004) dual class structure in Canada lessens corporate value because it lowers shareholder and manager alignment and increases agency problems. We also find that pyramid structure has a negative impact on value in both dual class and single class closely-held companies.

Notes

¹ For a comprehensive review of the literature on ownership and corporate value, see Demsetz and Villalonga (2001).

² Investors can hold a noncontrolling portion of the superior voting shares if the shares are publicly traded.

³ Extant cross-country studies of voting and cash flow rights (e.g. Lins, 2003; Nenova 2003) suggest that there are differences in impact on value across countries. These studies do not, however, provide complete details as to what makes the legal environment of each country different for dual class companies.

⁴ It is assumed in this article that shareholders bear all the agency costs because it is expected that the bigger the agency costs, the smaller the Q-ratio. However, one interesting issue that has not been sufficiently addressed in the corporate governance literature is who bears the agency costs.

⁵ We also classify companies as being widely-held, when the management group controls more than 10% of the votes but holds less votes than institutional block holders. For all such companies, the management group holds less than 20% of the votes.

⁶ As discussed in Goergen et al. (2005), since the controlling family often holds a large block of superior voting shares, such shares are often much less liquid than restricted shares.

⁷ It should be noted that the shareholder disagreement and legal cases cited in Amoako-Adu and Smith (2001) do not mean that there are no legal disputes and concerns in either single class closely-held or widely-held firms. The recent US legal cases against the executives of Enron, Worldcom, Tyco and Adelphia are examples of management accounting manipulations and corporate governance malfeasance in single class firms.

⁸ Though the Debt/Equity variable was measured using book values to ensure its exogeneity, as a robustness check, the models were retested with market value of debt versus equity but the results were similar to those of the book value measure.

⁹ Given that the ‘between’ estimator averages variable observations, random effect estimator can reduce the bias caused by the measurement error by averaging the measurement errors. On the other hand, the random effect estimator can be biased if the explanatory variables are correlated with the residuals, while the fixed estimator is always unbiased (because it allows for dummies for different intercepts). Further, the fixed effect estimator can be more robust to selection bias problems compared to the random effect estimator.

¹⁰ Pyramid structure is also referred to as tunnelling.