In Chap. 13 of The Calculus of Consent, Buchanan and Tullock argue that redistribution through the tax and transfer system may be viewed as income insurance agreed to at the constitutional stage. This idea has been eagerly taken up by well-known adherents of the welfare state (e.g., Arrow 1963; Rawls 1971; Mirrlees 1974; Diamond and Mirrlees 1978; Varian 1980; Sinn 1995; Lindbeck 2006; and Blundell 2006).1 What exactly have Buchanan and Tullock asserted, and what exactly does the argument imply? Buchanan and Tullock (1962: 192) start by suggesting that, in the absence of external effects and the insurance motive, “the individual, at the constitutional level, would never choose to collectivize the redistribution of real income among the members of the group”. This is doubtful. Since the social contract, by ending anarchy and violence, favors the weak over the strong, the strong may insist that some part of the gain from cooperation be redistributed in their favor. The strong also have a better bargaining position than the weak because, according to the Pareto distribution of income, they are less numerous. Both considerations imply that the contracting parties will agree on some redistribution from the poor to the rich. But this is not what Buchanan and Tullock intend to justify or explain. They want to explain “the decisive empirical evidence” that “in almost every society some collectivization of income redistribution is to be found” (Buchanan and Tullock 1962: 192), and the redistribution they mean is from the rich to the poor. They argue that under their assumption that all externalities are internalized, social contract theory can explain such redistribution if and only if the latter is viewed as income insurance. They do not consider the possibility that redistribution is not based on a social contract or that, owing to information and transaction costs, many externalities cannot be internalized through side payments.