Do non-choice data reveal economic preferences? Evidence from biometric data and compensation-scheme choice
研究利用可穿戴设备采集睡眠、日常活动中的生物特征数据，通过机器学习预测个体对竞争性或团队薪酬方案的选择，发现其预测准确度与性别和绩效相当，但日常环境数据贡献有限。
We investigate the feasibility of inferring economic choices from simple biometric non-choice data. We employ a machine learning approach to assess whether biometric data acquired during sleep, naturally occurring daily chores and participation in an experiment can reveal preferences for competitive and team-based compensation schemes. We find that biometric data acquired using wearable devices enable equally accurate out-of-sample prediction for compensation-scheme choice as gender and performance. Our results demonstrate the feasibility of inferring economic choices from simple biometric data without observing past decisions. However, we find that biometric data recorded in naturally occurring environments during daily chores and sleep add little value to out-of-sample predictions.