Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
This paper estimates individual risk preferences based upon data that are generated by the same individuals acting in different institutions. The results show that the (estimated) numerical values of individuals' implied risk parameters are not stable within individuals across institutions. Furthermore, the ranking across subjects of the numerical values of individuals' implied risk parameters is not preserved across institutions. Copyright 2000 by Kluwer Academic Publishers