REGRESSION DISCONTINUITY DESIGN: AN APPLICATION IN ECONOMICS

Abstract

Regression Discontinuity Design (RDD) is developed as an alternative to ex-post facto experiments. It is an approach causally estimating treatment effects in non-experimental designs. The Regression Discontinuity Methodology comprises a treatment and a control group and a threshold value depicting a known cutoff criterion which is employed to separate treatment and control groups. The logic behind the assignment process depends on the condition that if the value of assignment variable for each observation is above the cutoff value, the observation is assigned to treatment group; otherwise, it is assigned to control group. The difference occurred during the assignment process creates a discontinuity or a jump in the estimation process. The RDD methodology starts with graphical illustration revealing the distribution of data based on a cutoff criterion. Then, the direction of the discontinuity in the relationship between assignment and outcome variables is determined representing the main effect of the estimation process. The estimation process is supported by graphical illustration, parametric and nonparametric estimation methods. Furthermore, RD designs are used to obtain not only main effects but also interaction effects. This thesis aims at explaining research design with a theoretical framework, evaluating this main approach through an application of economic growth of Turkey and hence analyzing the relationship between terrorism and economic growth at province level in Turkey. In line with this, a theoretical discussion, a descriptive analysis, and a case study related with the dynamics of terrorism and economic growth constitute the outline of this thesis.