Adherents of Fundamental Indexing (FI) suggest that it is more protable to base portfolio weights on indirectly size-related indicators like accounting data rather than directly on market caps. In noisy markets a la Roll (1984), it is argued, underpriced stocks overperform but are underweighted and vice versa, implying a `drag' which FI claims to avoid. Mixed into the debate is the question whether mispricing is partly identiable or not, i.e. whether a policy of actively increasing the small-cap weights and vv helps. Carhart style regressions are unable to explain the extra return, but that conclusion is not robust across variant models, and there are substantial doubts about the constantness of factor sensitivities. We investigate the latter issue via cross-sectional regression of weight shifts, and find that not only the weight shifts are much larger than necessary to avoid drag, but the cross-sectional patterns are also quite variable over time. In short, there are style shifts and they are unstable. To estimate the benefits from drag avoidance, purged of style shifts without having to rely on generalized FF regressions, our procedure is to sort stocks on size into vigintiles, and compare within each vigintile the performance of FI-weighted returns to equally-weighted (EW) returns, which should be immune to drag too without much style shift. We find that within-vigintile EW portfolios are style neutral w.r.t. market and value, and do not meaningfully outperform VW portfolios. Thus, avoiding drag is not why FI does well: drag is empirically unimportant. Most or all of the prima facie benets must be from time-varying style shifts
