We combine experimentally elicited preferences with administrative micro data to study actual financial decision-making. Firstly, we simultaneously elicit and estimate risk and time preferences in a real-life context, with horizons up to 10 years, for more than 1000 pension fund participants. We estimate a present-bias factor of 0.84, an annual discount rate of 1.1%, and a CRRA utility curvature of 0.97. Secondly, using an expected utility framework, we show that the individually estimated preferences explain actual retirement decisions up to 82% of our sample for a utility indifference of at most 2% annual certainty equivalent consumption. Freedom of choice by means of a front-loaded annuity creates annual potential welfare gains up to 2.77%, but realized welfare gains are lower or even negative.