Relative efficiency of the internal capital market in a multi-division firm

Research on the efficiency of the internal capital market (ICM) of multi-division firms has not been conclusive. While most studies use data developed from Compustat and find that the ICM is inefficient, a few studies examine data developed from government or trade publications and find an efficient ICM. Measures of investment opportunities (Tobin's q) and segments (SFAS 14 data) used in the Compustat-based studies have been criticized in the literature as biased due to measurement error. This dissertation suggests a theoretical model of the internal capital market based on game theory and develops hypotheses to test those factors that would resolve the "social dilemma" of the division managers and create an efficient internal capital market. A unique dataset, involving a combination of hand-collected and Compustat data that addresses the measurement issues of previous studies, is collected to test these hypotheses and then test the efficiency of the internal capital market to the external market. Empirical testing was hindered by the lack of explanatory power of multivariate tests and therefore only limited evidence is available to examine the hypotheses proposed.