Discount or Premium? Diversification, Firm Value, and Capital Budgeting Efficiency∗

We model the internal capital market of a conglomerate where headquarters must rely on information reported by division managers to allocate limited resources across competing projects. Managers may exaggerate project quality to attract more capital, which limits the extent of winner-picking in capital budgeting. Focus (correlated projects) helps headquarters to infer information regarding project quality but makes winner-picking less crucial, whereas diversification (distinct projects) facilitates winner-picking but reduces managers’ incentive to report truthfully. We characterize conditions under which diversification improves and destroys firm value, and show that neither allocation efficiency nor firm value varies with the degree of diversification in a uniform way. Capital budgeting and diversification practices are subtle consequences of firms’ organization and governance, rather than ex-ante causes of value destruction or creation. Our theory reconciles conflicting empirical findings and offers insights to further testing. ∗Preliminary draft. We thank seminar participants at the Hong Kong University of Science and Technology and the 2009 Hong Kong Joint Finance Research Workshop for helpful comments and suggestions. †Department of Finance, School of Business and Management, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, (852) 2358-5051, [email protected] ‡The Fuqua School of Business, Duke University, Durham, NC 27708, (919) 660-7891, [email protected] Corporate diversification is a much debated subject in both academic research and industrial practices. Despite a large amount of research in this area,1 whether diversification creates or destroys firm value remains highly contested. While the earlier research showed that corporate diversification lowers shareholder value,2 recent work casts doubts or even refutes this finding.3 For instance, Campa and Kedia (2002) find that accounting for firms’ self-selection and endogeneity bias lowers the diversification discount and sometimes yields a premium. Villalonga (2004) reports that using a U.S. Census Bureau establishment-level dataset produces a robust diversification premium for a sample of firms where using COMPUSTAT segment data yields a diversification discount. While Fan and Lang (2000) document an integration discount in that vertical relatedness destroys firm value, they point out that this discount depends on the breadth of diversification, and using the same measures, Fan and Goyal (2006) find that vertically-related mergers create value. Equally controversial is the discussion concerning the cause of such diversification discount (or premium). Inspired by the empirical evidence that internal capital markets often exhibit “socialism,” where capital is allocated equally across divisions despite some divisions having better investment opportunities than others,4 many theories have been proposed to explain diversification discount in terms of agency conflicts and dysfunctions of internal capital markets.5 Others attempted to reconcile the discount with firm-value maximization and rational economic outcomes.6 Nevertheless, besides a vague concept of “synergy,” none of these existing theories accounts for the presence and possible sources of diversification premium, especially in relation to the operation of internal capital markets. Thus, linking value implication of diversification to the efficiency of capital budgeting invites more questions such as how empirically observed cross-subsidizations come about affecting firm value, and whether there is a causal relationship between mis-allocation of internal resources and diversification discount. To the best of our knowledge, no consensus has been reached regarding whether diversification creates or destroys firm value, and no theory can Martin and Sayrak (2003) provide a survey on this topic. Lang and Stulz (1994), Berger and Ofek (1995), Comment and Jarrell (1995), Servaes (1996). Examples are Whited (2001), Graham et al. (2002), Campa and Kedia (2002), Schoar (2002), Chevalier (2004), Villalonga (2004), among others. See, e.g., Lamont (1997), Stein (1997), Scharfstein (1998). Examples are Denis et al. (1997), Stein (1997), Rajan et al. (2000), Scharfstein and Stein (2000), etc. See, for instance, Montgomery and Wernerfelt (1988), Maksimovic and Phillips (2002), Gomes and Livdan (2004), and Almeida and Wolfenzon (2005).