Income Tax Treatment and Labour Supply in a multi-level hierarchical Difference-in-Differences model

Ignoring the possible hierarchical clustering of the data that frequently characterises the structure of labour markets implies that studies of the effects of income tax changes on labour supply use less than necessary information on the variability of the labour response. Estimation efficiency is reduced and relevant relationships affecting the agents’ reaction to net wage changes remain undetected. Motivated by the desire to implement an estimation procedure that accommodates a nested hierarchical statistical structure of labour supply macro data into of a causal-effect framework, we propose a novel multilevel DiD model that can estimate labour responses to exogenous tax hikes taking the above hierarchical structure into consideration. Using Italy as a case study, we examine the labour response to exogenous income tax changes using a hierarchical DiD model modified to account for the existence of different sources of variation of the data (regional and provincial labour markets) as well as for various possible clustering of the data (territorial, age and gender). We compare results obtained from various nested and non-nested procedures and show that our multilevel variant of the DiD model generates gains in efficiency with respect to approaches that ignore the clustering nature of the labour data. The hierarchical multilevel DiD procedure permits to qualify labour response in terms of cluster membership and to shed light on aspects of the tax issues not highlighted by current literature.