"This study estimates the extent to which a policy-induced increase in distributions at retirement crowds out dissaving from taxable assets.... The study shows that retirement distributions at age 70 vary discontinuously at a statutory threshold in exact day of birth. Using this discontinuity as a first stage, the study finds that taxable saving becomes more positive (or less negative) among those whose retirement distributions are induced to be higher. In the baseline specification, ratio at which induced retirement contributions crowd out taxable dissaving is estimated to be 0.42."