PDF Ebook The Organizational Efficiency of Internal Capital Markets
The relationship between multidivisional structure and performance is a central question in corporate strategy, as the multidivisional form is one of the most prevalent organizational structures around the world (Chandler 1973, Armour and Teece 1978). To gain insight into the effects of multidivisional organization, researchers have proposed investment as a mediating mechanism between structure and performance (Bower 1970, Williamson 1975), conceptualizing the firm as an internal capital market for divisional projects in need of funding. Despite the broad interest in the multidivisional form, however, there is no consensus about the specific effects of divisionalization on investment and performance. On the one hand, theories of internal capital markets reach radically different conclusions regarding their investment efficiency, often classifying results into the ‘bright side’ (e.g., Stein 2002) or the ‘dark side’ (e.g., Scharfstein and Stein 2000) of internal capital markets. On the other hand, multidivisional firms typically have a multibusiness scope, so that the effects of ivisionalization are hard to disentangle from those of diversification (Hoskisson 1987, C¸ olak and Whited 2007). As a result, Liebeskind’s (2000) proposition that the effects of divisionalization, reflected in the activity of an internal capital market, may depend on its organization remains an open question.
Empirical research attempting to shed new light on this theoretical discussion faces two clear challenges. First, because firms endogenously choose their organizational structure, it is hard to assess causality when observing correlations between structure and investment returns. Second, even if the effect of multidivisional structure is statistically clear, various competing mechanisms could account for such influence and require a close investigation.
To fill the gap in understanding the effect of multidivisional structure on investment returns and its underlying mechanisms, this paper provides an empirical investigation of the movie industry in the United States. Movies are an appropriate setting for the question of interest for at least three reasons. First and foremost, the setting allows for the observation of divisionalization without diversification, a key empirical advantage for identification. Starting in 1992, some of the largest movie distributors pursued acquisitions and internal developments, and subsequently operated in the same industry with two different types of divisions: major and specialty. This case of divisionalization allows for clearer inference than most research settings in which divisionalization is accompanied by expansion into other industries, thus avoiding confusion between the distinct effects of organizational structure and diversification.
Second, quantitative information on investment and market performance is observable for thousands of divisional projects (movies) in a long time-series and a rich cross-section helping to effectively control for alternative factors. This project-level detail provides clear advantages over most prior work that has analyzed divisionalization without observing the internal reality of each corporate division or the micro components of economic value. Third, institutional aspects of the movie industry such as the divisional operation of formerly independent distributors after entering a multidivisional structure, or the project-specific match between distribution and production companies motivate two intuitive identification strategies to analyze the effects of multidivisional structure on investment and performance.
The first identification strategy employs distributor-level fixed effects in regressions of project-level investment and performance on a multidivisional structure dummy. The identification of key parameters is obtained from comparing outcomes for the same distribution entity (i.e., a firm turned into a division) before and after the multidivisional structure is enacted. Using this fixed effects approach, I find that multidivisional structure is strongly associated with larger production and advertising investments. In contrast, multidivisional structure does not influence either box office revenue or ancillary revenue from the home video (e.g., DVD) market, two key components of a movie’s market performance. This stark contrast uncovers a negative relation between multidivisional structure and investment efficiency.
To probe the causal interpretation of the fixed effects models, I adopt an alternative identification strategy using detailed data on project-team configuration to instrument for the influence of multidivisional structure. By arguing and probing the validity and refutability of three variables — the prior experience of team members with major distributors other than the current distributor, the variety of prior interactions of team members with distributors other than the current distributor, and the prior exposure of team members to film genres other than the current genre — as not linked to unobserved drivers of investment returns but directly affecting the match of the team with a multidivisional structure type, I find these variables suitable to instrument for structure and assess its effects. The results based on this instrumental variables approach confirm that multidivisional structure increases total investment but does not affect total revenue. Taken together, the findings of the two identification strategies suggest a decrease in investment returns enacted by a multidivisional structure.
Download
Posted in :
