Abstract
Self-reported productivity and satisfaction have become central metrics in evaluating work-from-home (WFH) policies, yet their reliability remains largely unexamined. Despite growing scrutiny of WFH efficacy, assessments continue to rely heavily on subjective evaluations, creating a persistent gap between perceived and objective productivity measures within the working-from-home literature. This study investigates whether retrospective self-reports of productivity during WFH are systematically biased due to recollection inaccuracy, particularly through anchoring biases in memory recall. A two-wave survey data sample consisting of 772 home-workers during the 2020 shift to the home office examines recollection accuracy as well as the underlying mechanism. Using a five-factor productivity scale, within-subject analyses explore consistency within and between multiple waves and evaluate the predictive value of the targeted score compared to the current (inaccurate) score. The signed rank test shows that recollection scores consistently underestimate past scores (one factor: p < .002; four factors: p < 0.0004). The recollected scores are closer to the current score than the targeted score (for all factors: p < .0004) and have a greater magnitude impact on the recollection scores than the targeted score (difference in OLS coefficients ranging from.164 to.606). Exploration of trends additionally suggests that, although the absolute scores are influenced by the current reference point, the relative changes seem consistent over time. These findings highlight the risks of relying on self-reported retrospective productivity measures in shaping WFH policies. The observed biases help contextualize the ongoing discrepancy between optimistic self-reports and more pessimistic objective measures of WFH productivity. Without accounting for recollection biases, assessments of WFH effectiveness may be flawed, potentially leading to suboptimal or counterproductive policy decisions.