Stock weighting and nontrading bias in estimated portfolio returns

Liu and Strong (2008) note that researchers often employ a simple (but incorrect) averaging approach that induces significant error into estimated buy‐and‐hold portfolio returns. This study explores the additional challenges that arise when stocks are subject to nontrading. We develop a decomposition of the total bias in estimated return into the components attributable to the stock weighting approach and the treatment of nontrading. While the latter is shown to be negligible, the former can approach 150 basis points per month. Our empirical analysis of Australian equities shows that the simple averaging approach tends to overstate the size and book‐to‐market effects, and understate the momentum effect.