Abstract
Socialism in internal capital markets (SICM) is the phenomenon whereby multi-division firms underinvest in high-profit segments and overinvest in low-profit segments. This article modifies and extends the Scharfstein and Stein (Citation2000) model that explains SICM. The major modification is to abstract from rent-seeking behaviour – a central ingredient that is used by most SICM models and has the side effect of overturning SICM when the underlying rent-seeking game is slightly perturbed. It is shown that when division managers’ private benefits are not exactly propotional to their division's production profit, as assumed by Scharfstein and Stein (Citation2000), SICM arises due to the CEO's incentive to extract gain from the trade of capital between his division managers by equalizing their marginal private benefits, instead of accommodating rent-seeking. Furthermore, we show that ‘diversification discounts’ exist, increase in diversity and decrease in CEO ownership. The results are robust to endogenizing the investors’ choice of the total capital provided to the firm, which allows the conditions for aggergate underinvestment or overinvestment to be determined.
Notes
1For example, Chevalier (Citation2004) finds firms exhibit socialistic investment patterns even before they merge. Whited (Citation2000) argues that these results can be subject to measurement errors in the proxy for profitability – Tobin's q. Graham et al. (Citation2002) argue that the discount might happen before the merging of the segments. There are also many studies, for example Villalonga (2000), question whether the discount should be interpreted as value destruction.
2This is the case under Assumption 5. In this case, any increase in the private benefit of a division manager can be totally captured by the owner by a same amount of saving in cash payment to the division manager.
3This could be the case when, for example, the manager of the more profitable segment also has a more ‘productive’ technology to rent seek.
4For example, Bernardo et al. (Citation2001) allow for incentive contracts in a similar framework but find that both the less and more profitable segments will be underinvested, which does not represent SICM.
5See, for example, Harris and Raviv (Citation1996, Citation1998).
6Roughly speaking, the current model corresponds to ‘period 1’ of the SS model.
7Without loss of generality, we can assume that the segment managers do not own any shares because what matters is the total shares owned by each firm's CEO and segment manager(s).
8In contrast to the SS model, wage expense is included in the calculation of profit in the current model to better reflect common accounting standards. The results will not change if the setup of the SS model is adopted.
9This shows that the role of Assumption 5 is not to be substitute for the rent-seeking assumption.
10In SS φ represents a private benefit parameter, instead of ownership, as in our model.
11SS only point out the possibility of aggregate underinvestment.
12
I
2 cannot reach before I
1 reaches
because this implies case (iii).
13Note that we are restricting our attention to capital misallocation that is caused by moral hazard. In the case of adverse selection, there are some results in which overinvestment and underinvestment co-exist to induce truth-telling (Harris and Raviv (Citation1996)).
14Note also that, as in the SS model and Rajan et al. (Citation2000), the setup here is meant to capture the aspect of difference in segments’ investment opportunities, instead of other important aspects, of diversification.
15Formal analysis can be provided by the author upon request.
16In any case, the CEO does not divert any value because he never receive more than his reservation value.
17Note here that we can ignore Equation Equation8 because it is redundant given the condition that and Equation Equation22.