We revisit the saving behavior of elderly singles using an adversarial structural estimation framework by Kaji et al. (Econometrica 91:2041–2063, 2023). The method bridges the simulated method of moments (SMM) and maximum likelihood estimation by embedding a flexible discriminator—implemented as a neural network—that adaptively selects the most informative features of the data. Applying this approach to the model of De Nardi et al. (J Polit Econ 118:39–75, 2010) with AHEAD data, we show that including gender and health histories in the discriminator improves the identification and precision of bequest motives. The resulting estimates reveal that bequest motives explain between 13% and 19% of late-life savings across all permanent-income quintiles, not only among the rich. The adversarial estimator precisely disentangles bequest motives from precautionary saving motives. These findings suggest that heterogeneity in health-related survival expectations is another important source of identifying variation for distinguishing bequest and precautionary saving motives.
Kaji et al. (Wed,) studied this question.