Compensation Does Not Necessarily Turn a Kaldor-Hicks Improvement into a Pareto Improvement

I was just reading a paper that made the surprisingly common mistake of arguing that the purpose or effect of compensation (e.g., for governmental takings of property) is to turn an alleged Kaldor-Hicks improvement into a Pareto improvement. The argument is as erroneous as it is common. At best, compensation can turn an alleged Kaldor-Hicks (or "potential Pareto") into a more likely Pareto improvement.

Compensation, by itself, cannot achieve the strict Pareto requirements because, except in voluntary market transactions (a small subset of all transactions), the level of compensation is not based on losers' subjective valuations of their losses. In many cases, compensation is determined by policy-makers or judges, reflecting their own subjective sense of supposedly objective losses. Absent the ability to make interpersonal utility comparisons, non-market compensation cannot turn a Kaldor-Hicks improvement into a Pareto improvement, except accidentally.