Redistribution and Welfare: One Simulation of the Real Effects of the Reform of the IRPEF (Redistribuzione e Welfare: Una Simulazione degli Effetti Reali della Riforma dell'IRPEF)

Abstract

Empirical researchers are used to fix a common distribution of income before tax for all tax schedules being compared. Is this realistic for accurate intertemporal, or international, comparisons when, as a matter of fact, schedules operate on different original income distributions? It is well-known that what matters for the distributional impact of personal income taxation is the tax schedule and where the taxpayers are located. On this, Dardanoni and Lambert (2002) were able to formulate what could be a strategy a practitioner needs to perform if she or he wishes to take pre-tax inequality into account and to draw out correct and actual distributional implications of tax reforms. To summarise, their procedure acts on the pre- and post-tax distributions under analysis: it looks for an isoelastic transformation between the former and, if this were the case, corrects for the pre-tax distributional differences between the latter. Standard results on redistribution of Jakobsson and Kakwani are preserved under specific conditions on the structure of pre-tax income distributions and it achieves an 'independence of baseline' property. Critically, in Russo (2005) I proposed an application to Italian micro-data and personal income tax systems (IRPEF) between 1995 and 2000 and, according to Dardanoni and Lambert's requirements, I found that before-tax log distributions differ essentially by location and scale. In this new paper I focus on an interesting insight. If isoelasticity does hold between pre-tax income distributions through time, it seems reasonable enough to conclude that this will occur again. As a consequence, with some parameters, say a and b, reflecting size and inequality differences there should exist some isoelastic transformation able to transplant an unknown pre-tax income distribution into an already known one. According to this hypothesis an empirical researcher wishing to assess, now, which may be the outcomes of a tax reform, could proceed in two stages. Firstly, as usual, she or he can simulate those outcomes by using the most recent available distribution of pre-tax income. Secondly, in addition, she or he could take into account distributional differences by assuming different values for the parameters a and b. In this way, we may control for a range of potential distributional variations: the choice of parameter values is the 'variable' which could influence, together with the new tax structure, the redistributive effect. I propose the 2005 Italian personal income tax reform as a good candidate for this more comprehensive approach and, in particular, I focus on the second stage. As far as I know, following this method tax systems intertemporal comparisons have not been treated elsewhere before. I present and discuss empirical evidence arising from the methodology and the main 'potential' pure redistributive results with regard to the 2005 IRPEF reform versus the 2000 IRPEF.