Using linear and nonlinear models, the authors examine two samples of bond funds--one sample designed to eliminate survivorship bias and a second much larger sample. Overall and for subcategories of bond funds, they find that bond funds underperform relevant indexes postexpenses. The authors' results are robust across a wide choice of models. They find that, on average, a percentage-point increase in expenses leads to a percentage-point decrease in performance. The nonlinear model weights closely match actual composition weights. The authors find no evidence of predictability using past performance to predict future performance for their unbiased sample. Copyright 1993 by University of Chicago Press.
Stored only when both papers are in the corpus — 464 citations exist upstream.