Most pre-crisis explanations of the various corporate governance systems have considered the separation between ownership and control to be an advantage of the Anglo-American economies. They have also attributed the failure of other countries to achieve these efficient arrangements to their different legal and/or electoral systems. In this paper we compare this view with the co-evolution approach based on the hypothesis that politics and corporate governance influence each other, generating complex interactions of financial and labour market institutions. Countries cluster along different complementary politics-business interaction paths and there is no reason to expect, or to device policies for, their convergence to a single model of corporate governance. We argue that this hypothesis provides a more convincing explanation of the past histories of major capitalist economies and can suggest some useful possible scenarios of their future institutional development. Bayesian model comparison suggests that the co-evolution approach turns out at least as influential as the competing theories in explaining shareholder and worker protection determination