Economics has experienced an evolution from Neoclassical welfare economics, with an emphasis on normative issues and the identification of market failures in a world with rational agents, towards public choice (with an emphasis on government failures with self-interested rulers) and modern political economy, which is similar to public choice but more neutral in terms of the moral implications of the tension between market and government imperfections. All these approaches have in common that they assume that the relevant agents are fully rational. Behavioural economics changed this, and with the help of psychology, assumes more realistically that agents have predictable psychological biases, with difficulties to optimize, self-control problems and non-standard (more social than individual) preferences. In the first years of behavioural economics, though, the emphasis has been on the bounded rationality of consumers and citizens, assuming implicitly that rulers could intervene to de-bias them. Of course, rulers (politicians, regulators) are biased themselves, after all they are also citizens and consumers. Hence the need of models that more realistically assume that all agents may be affected by bounded rationality. Of course, the relevant biases are different for different agents and time periods. Some of the agents may be aware of their biases and others not. This gives scope for the design or the evolution of institutional mechanisms that recognise the biases and, to the extent that realistic socio-political processes allow it, try to achieve results that improve the welfare of societies.