In 1985, Demsetz and Lehn argued both that the optimal corporate ownership structure was firm-specific, and that market competition would drive firms toward that optimum. Because ownership was endogenous to expected performance, any regression of profitability on ownership patterns would yield insignificant results. To test this hypothesis, we use the zaibatsu dissolution program from late-1940s Japan as a natural experiment: an exogenous shock to the equilibrium ownership structure. Through that program, the US-run occupation removed the more prominent shareholders from many of the most successful Japanese companies. By focusing on the way firms and investors responded to the mandated selloff, we accomplish two goals: (a) we avoid the endogeneity problem that has plagued much of the other research on the subject, and (b) we clarify the equilibrating dynamics by which competitive markets move firms toward their optimal ownership structure. With a sample of 637 Japanese firms for 1953 and 710 for 1958, we confirm the equilibrating mechanism behind the Demsetz-Lehn hypothesis: between 1953 and 1958, the ex-zaibatsu firms did retructure their ownership patterns. As of 1953, the unlisted ex-zaibatsu and new firms still had not been able to negotiate the transactions necessary to approach their profit-maximizing ownership structures. Even the listed firms had not fully undone the effect of the occupation-induced changes on managerial practices. By 1958 the firms had done this, and the earlier correlation between profitability and ownership disappeared. By then, firm profitability showed no correlation with ownership, whether under linear, quadratic, or piecewise specifications. We further find no evidence that ex-zaibatsu firms sought to strengthen their ties to banks over 1953–1958.