Date of This Version



Some regulatory reforms do not change just a specific signal that can be represented by a quantitative continuous variable, such as a tax rate, a price cap, or an emission threshold. The standard theory of reform in applied welfare economics (going back to contributions by e.g. Ramsey, Samuelson and Guesnerie) asks the question: What is the marginal effect on social welfare of changing a policy signal? However, reforms such as privatization, unbundling or liberalization of network industries are often described by ‘packages’ shifting a policy framework. It is increasingly frequent in the empirical evaluation of such reforms to use categorical variables, often in polytomous form, for instance describing unbundling steps (vertical integration, accounting, functional, legal, ownership separation) on a discrete numerical scale, such as those proposed by the OECD and other international bodies. We review recent econometric literature evaluating regulatory reforms using such variables (40 papers) and we discuss some methodological issues arising in this context.