MSc Economics Thesis 2019

The repository contains the version as presented at the defense. Although graded well, the work suffers from methodological issues. The main issue is that it relies too much on a STATA package named “xsmle” which is ill-suited to address the simultaneity problem. A better approach would be to do the computations of the xsmle command manually and to introduce instrumental variables. Belotti, F. Hughes, G. and Piano Mortari, A. (2017). Spatial Panel Data Models Using Stata. The Stata Journal. Volume 17. Issue 1. Pages 139-180.

A second issue is the strategy taken to create a spatial multinomial logit model. This approach is creative but invalid.

An alternative and more promising approach to the project would be the use of a difference in differences design that exploits the characteristics of (two) neighbouring municipalities being similar but different in their value for the variable of interest.

Finally, some of the special characters in the document are wrongly displayed on Github. When downloaded this is resolved.