, AIs, humans and rats: decision-making and economic welfare

Critics of economics often highlight two related issues: the empirical falsity of the ‘homo economicus’ assumption of rational, self-interested maximisation; and the ethical consequences of models based on this assumption. Yet many experiments in biology show non-human creatures often seem to behave as if they were rational maximisers, suggesting that context rather than cognitive capacity is important for determining behaviour. The critique of rational choice poses a less serious methodological challenge to economics than is sometimes thought. However, economists do need to respond to the ethical critique that decisions and policies based on the assumption of rational self-interested maximisation change the norms of individual behaviour for the worse. This paper argues that economics has become divorced from ethics because for a century it has dealt only with ordinal, not cardinal, welfare rankings and has thus ruled out interpersonal comparisons. While enabling economists to separate normative from positive analysis, this separation protocol has left welfare economics both internally contradictory and unable to address major societal decisions, even though welfare economics is used constantly in limited ways, such as cost-benefit analysis. This separation reflects empirically inaccurate assumptions concerning preference formation and the conditions of supply and demand (but not the rational choice assumption) in the foundational welfare economic theorems. Economics must urgently revisit welfare economics, particularly in the context of modern economies in which individuals are increasingly interdependent, and the assumptions required for the fundamental welfare theorems therefore increasingly invalid.