Association for Public Policy Analysis & Management

The European economic policy monitoring and coordination process, also known as the European Semester, requires a systematic analysis of structural and fiscal reforms undertaken by the EU member States. In order to replicate tax reforms enacted or to simulate alternative tax policy scenarios two conditions must be met: (i) the policy reform simulated must mimic with exactitude real policy reforms enacted or proposed (ii) the analysis must account for possible changes in the tax bases (i.e. the behavioural impact of tax reforms) and interaction with the rest of the economy (i.e. the feedback effect). The objective of this paper is to describe recent methodological developments for analysing the impact of tax reforms combining a microsimulation model covering all EU countries and a Dynamic Stochastic General Equilibrium (DSGE) macro-model in order to meet the above conditions. For that, we combine the microsimulation model EUROMOD, augmented with an econometrically estimated labour supply model, with the new Keynesian DSGE model of the European Commission QUEST. The labour supply function is first estimated econometrically using the micro-data employed in the microsimulation model whereby individuals decide on the number of hours worked represented by a set of discrete choices and considering among those choices the decision not to work in order to account for the extensive margin of labour supply. Individual-specific characteristics as well as joint decision making process within the household are both accounted for in these estimations. The estimated labour supply function is used to predict the change in the number of hours worked caused by a change in disposable income resulting from a given tax policy reform simulated using the EUROMOD model. We compute the new implicit tax rates resulting from the tax cuts, which are fed in turn into the QUEST model as permanent fiscal policy shocks. Importantly, we calibrate labour market related parameters and exogenous variables in QUEST as close as possible to the predictions of the microeconometric discrete choice model, giving due account to aggregation issues. This is the case of the labour supply elasticities and the non-participation rates. From QUEST we are able to obtain the medium-term – three years – general equilibrium projections for employment, wages and prices, which are imputed back in EUROMOD in order to obtain the final effects on tax revenues and disposable incomes. We illustrate the usefulness of our approach by simulating recent tax reforms undertaken in three EU countries: Italy, Belgium and Poland.  We estimate two alternative scenarios, one under which the behavioural response to tax changes over the medium term is ignored and another scenario where this behavioural/micro-dimension is embedded into the macro-model. We compare the changes in tax revenues and disposable income resulting from these two scenarios and discuss the relevance of considering behavioural reaction to analyse the impact tax policy reforms in the European case. We show that accounting for the behavioural and macroeconomic feedback effects.