Joint vs. Individual performance in a dynamic choice problem

This paper compares the relative ability of individuals and pairs to solve a finite, stochastic lifecycle problem that requires borrowing and saving to achieve the rational benchmark. We find that pairs significantly outperform individuals, especially when allowing subjects to account for past mistakes along conditionally-optimal consumption paths. Joint decision-makers out-earn individuals by about 23%. Though pairs and individuals both overreact to income and wealth balances, these distortions are twice as large for individuals. Analyzing chat data reveals that pairs bargain to balance idiosyncratic consumption preferences, which reduces consumption errors. We estimate consumption heuristics at the observation level and study their dynamics. We show that about half our subjects (or pairs of subjects) stick to heuristics for the majority of the experiment. These ‘stable’ subjects significantly outperform their ‘unstable’ counterparts in the dynamic optimization task. Finally, we provide suggestive evidence that subjects who have a nuanced view of debt outperform subjects who think of debt as always bad, even after controlling for cognitive ability.