Despite compelling evidence of widespread gain–loss-domain-dependent behavior, research on domain-dependent diversification is scarce. We recruited 251 experienced US retail investors to participate in a controlled experiment with the task to select portfolios that differ in asset correlation and, hence, diversification benefits in both the gain and the loss domain. We find evidence of domain-dependent diversification, both unconditional and conditional on benchmark portfolio preferences. Consistent with a loss-attention hypothesis, diversification errors are not observed in the loss domain but are clearly present in the gain domain (with much lower diversification relative to the benchmark).
Read full abstract