The funded research investigates the role of expectations in financial decisions by asking two separate empirical questions. First, what systematic patterns appear in the expectations that the investor holds about relevant random variables? Second, can we establish that her expectations play a causal role for choices? The main tool in addressing these questions is the collection of data from laboratory experiments and large-sample household surveys. We also develop novel econometric methods and behavioral models of portfolio choice.The majority of the planned studies add to the literature on biases in expectations. They conform to the revealed-expectations paradigm of expected utility, but ask whether the underlying expectations are suboptimal in predictable ways: (i) Are rational expectations violated systematically more in markets where agents need to make complicated inferences about fundamental information – e.g. where the inference occurs at the interim stage, before agents can observe realized prices and other agents’ choices (like in Rational Expectations Equilibrium)? (ii) Are agents able to successfully process the covariance of returns in portfolio-choice problems? (iii) Are investment herds excessively large and stable because people do not realize that other agents whom they observe also rely on others? (iv) Why do many people embark in risky gambling strategies under the false perception that they will stop after a few losses? The answers to these questions will contribute to the understanding of investment choices generally, and unnecessary risk exposure in particular.In further studies, we ask about the causal link between expectations and choices. This will contribute to the evaluation of decision-theoretic models but can also inform economic policy, as the causal link from expectations to choices is an important component in the design of policy campaigns that work via affecting beliefs. We introduce artificially created instrumental variables into choice situations and use their exogenous influence to identify causal effects.