Okay, I understand that you aren't using Marshall improvements to sum utilities. So if Marshall improvement is just the question as to whether net sums in dollar values< cancel each other out, then what is the reason to think that aggregated Marshall improvements, for a given individual, result in utility gains more often than utility declines (for that individual)? Meaning, why should aggregated Marshall improvements lead to a Pareto improvement more often than lead to Pareto decline?