Why do people facing low risks tend to purchase more insurance than people facing high risks? Why do attempts to provide affordable health insurance in developing countries typically result in very low enrolment rates? These behavioural patterns are puzzling given objective risk estimates (e.g. life expectancy, prevalence of health risks). The rational expectations paradigm assumes that people base their decisions on these objective risk estimates, but is this really so? Research in psychology shows many cognitive biases in belief formation, challenging the assumption of rational expectations. And even if expectations are rational, beliefs may deviate from risk estimates because of private information. Measuring (subjective) beliefs is crucial to understanding people?s behaviour. Moreover, since Keynes (1921), we know that the confidence an agent has in his beliefs can be as important as the beliefs themselves. Ellsberg (1961) introduced the term ?ambiguity? to describe situations in which decision makers are unsure about the risks they are facing. He showed how the classical expected utility model fails to describe agents? behaviour under ambiguity. The present research program will measure and model expectations in economics, using the most recent insights from the ambiguity literature. The first part of the program will develop new methods to analyze behaviour and to measure beliefs and confidence. These new methods will, for instance, allow us to reveal people?s true expected longevity to better understand pension decisions. The second part will apply the new methods. I will incorporate people?s real subjective beliefs about risks and life expectancy in the analysis of their insurance and pension decisions. I will also show how one of the new methods can be applied to how people evaluate their life satisfaction. The results of the research program will improve policy design by taking into account what people really believe when making their decisions.