The friction-free weighted price contribution

•We propose a friction‐free version of the WPC that is consistent and reliable.•We show that trading frictions can severely bias the original WPC approach.•The bias is mostly due to bid‐ask bounce. Midpoint serial correlation contributes.•A negative bias for small‐size non‐aggressive trades is the most notorious.•We provide consistent evidence of the bias for the NYSE.

We introduce a methodology to obtain friction-free estimates of Barclay and Warner's (1993) Weighted Price Contribution (WPC). With this new approach, we verify recent simulation results suggesting that trading frictions may severely bias the WPC approach. We use high frequency data from a European electronic order-driven market to show that frictions generate a sizable downward bias in the WPC of non-aggressive small-size trades. The bias increases in periods of significant price discovery, and is due to both bid-ask bounce and serial correlation in the quote-midpoint changes. We show that our results extend to the US case.