Using a rich new data source, we explore the selection of economics Lecturers into the last UK Research Excellence Framework (REF) exercise. Only some one-in-two (54%) of these Lecturers were submitted to REF2014; 57% of men and 46% of women. The decision making of Institutions is found to be well approximated by a simplified selection approach; focusing on working papers and higher quality journal publications. Our results also reveal sizeable conditional differences in the probability of selection, especially so in departments with higher research rankings. More than half of the variance in selection probability remains unexplained, revealing considerable idiosyncrasies in the management of submissions and uncertainty across the discipline in this research assessment process.