Managerial ownership and firm performance: an analysis using switching simultaneous-equations models

Abstract

This paper uses a switching simultaneous-equations model to examine the relation between managerial ownership and firm performance. The model includes a multinomial logit for the firm's choice among three regimes of large-block ownership, which can be argued as the choice among different degrees of controlling-minority structures, and three simultaneous-equations systems of managerial ownership and performance for each ownership regime. The paper argues that the choice of ownership regimes is the firm's endogenous decision as a reflection of the firm-specific organizational and transactional attributes, and hence the impact of managerial ownership on performance varies across firms belonging to different regimes. Empirical results show that family involvement in the management and significant related-party transactions are important factors to determine the firm's choice of ownership regimes. Evidence also indicates that the patterns of the relation between managerial ownership and firm performance, in the sense that the inflection points for the impact of managerial ownership turning from positive to negative, are markedly different across ownership regimes. Interpretations consistent with the endogeneity of managerial ownership are provided.

Acknowledgements

I thank the Taiwan National Science Council (project NSC-91-2415-H-032-012) for financial support. I also wish to express my gratitude to Ching-Jung Liu, who is now assistant research fellow of Taiwan Institute of Economic Research, for her excellent work on the data collection.

Notes

¹ Some researches further stress the interaction of managerial ownership with other mechanisms in serving the function of corporate governance. For example, Agrawal and Knoeber (1996) estimate a seven-equation simultaneous model including Tobin's Q, managerial shareholdings, and five other control mechanism equations. Chung and Pruitt (1996) model relations between managerial ownership, compensation, and Tobin's Q. Chen and Steiner (2000) estimate a model with Tobin's Q, managerial ownership, and analyst coverage jointly determined within the system. Generally, they conclude that the appropriate monitoring mechanisms vary systematically across firms to reflect the trades between costs and benefits of different mechanisms for the firms.

² In their study of East Asia countries by identifying the company's ultimate control, Claessens et al. (2000) find that for Taiwan's non-widely held companies, 79.8% of sample companies have top managers being the relatives of the controlling shareholder's family, 43.3% of them are controlled by a single shareholder, and 49% and 8.6% of them are controlled through pyramid structures and cross-holdings respectively.

³ The switching regression model is usually used in the field of the labor economics. For example, there are many studies analyzing the wage differentials between part-time and full-time workers because both types of workers are distinguished by hours worked and because hours worked and wages are interdependent. In a recent study, Manrique and Ojah (2003) also apply the switching regression model to the demand for housing. Compared with a single wage regression for part-time and full-time regimes or a single expenditure equation in each housing regime, this paper has a simultaneous-equations model in each ownership regime.

⁴ Recent literature about business groups has focused on whether affiliated firms have superior or inferior performance than unaffiliated firms. In addition to the value-added view of Khanna and Palepu (2000), there is a competing view of value-eroded in the studies of Johnson et al. (2000) and Johnson and Mitton (2003). Group affiliates have a greater lack of transparency in the locus of control over companies than unaffiliated firms, and hence are more insulated from external monitors and usually indulge in using political connections to solicit privileges from the government or to leave financial intermediaries no incentive to monitor. These problems are particularly severe if the groups are owned by controlling families that attempt to protect their privacy and have little information about internal activities disclosed to the public.

⁵ Several other possible segments are tested but not reported in the section of empirical results. The signs and significance levels of the coefficients differ in some cases. This lack of consistency in the results of alternative specifications, as mentioned in several previous studies, does cast doubt on the general applicability of the piecewise linear approach, which has relatively weak theoretical support. However, in this paper, using the first quartile and the midpoint as the turning points produces results that coincide with the results when the curvilinear specification is used. It provides supplementary evidence for better understanding of the relation between managerial ownership and firm performance.