Not logged in.
Quick Search - Contribution
Contribution Details
| Type | Master's Thesis |
| Scope | Discipline-based scholarship |
| Title | Matching Methods’ Performance in the Presence of Endogeneity |
| Organization Unit | |
| Authors |
|
| Supervisors |
|
| Language |
|
| Institution | University of Zurich |
| Faculty | Faculty of Business, Economics and Informatics |
| Date | 2018 |
| Abstract Text | When estimating causal effects using observational data, researchers often apply matching methods to eliminate systematic differences between treatment and control groups. These methods only account for observed differences between groups, but observational data is likely to lack some important variables that influence treatment assignment and the out- come. This can lead to bias in the estimated treatment effect due to endogenous treatment assignment. This thesis compares the performance of different matching methods when only a subset of the confounding variables is observed. I find that (1) restrictive methods, such as exact and coarsened exact matching, achieve the best balance, but yield the most biased treatment effect estimate; and (2) genetic matching retrieves the least biased estimate, but fails at sufficiently balancing the treatment and control groups when the correlation between the observed and unobserved confounders is high. Overall, the higher the level of correlation between the omitted variable and the observed confounders the less biased is the estimate of the treatment effect. At the same time, it gets more difficult to sufficiently balance the observed variables for the treatment and control group. |
| Export |
BibTeX
|