The null hypothesis, based on Colliard and Foucault (2012), is that liquidity
measures that account for exchange fees are unaffected for the feeneutral
group. Consistent with the null hypothesis, we find that the “cum fee”
trading costs, measured by the effective bid-ask spread plus (twice) the taker
fee, did not change, despite the decline in the “raw” bid-ask spread, which does
not include the taker fee.