SIES Day 4: Advanced Econometrics Methods

The widely-used Ordinary Least Square (OLS) estimation method can introduce considerable bias into experiments and, consequently, ineffective policy implementation. How do researchers and policy makers address challenging situations such as heterogeneity and attrition? Which techniques provide consistent insights on whether to scale-up or discontinue a policy?

The morning of the fourth day of the Summer Institute of the Econometric Society (SIES) opened with a discussion led by Dr. Ismael Mourifié (University of Toronto) on these issues. He explored new developments in the field of Policy Evaluation, as well as the various problems encountered by researchers and policy makers in their work. Through his research, he demonstrated that the necessity of considering compliers and “defiers” in an experiment. According to a research project completed in the U.S. about the impact of job training impact on personal income, only 6% of those from the treatment group (that is, the group benefitting from the job training) complied. Consequently, Dr. Mourifié also provided examples of “defiers,” citing a study that found students from wealthy and educated parents are less likely to prioritize education, as they might find a personal safety net in wealth inheritance, as opposed to education attainment. Consequently, the study suggested that students from poor and uneducated households are more likely to excel in school, especially if their family lives in close proximity to educated people. Depending on the historical and social contexts, “defiers” may be more prominent in some geographic areas than others.

SIES continued to explore complex topics in the afternoon, as Dr. Firmin Doko Tchatoka from the University of Adelaide presented the ARIMA method and advanced models in Times Series Forecasting. Through his demonstration, Tchatoka revealed that within practical economy, methods involving causality are preferable to ones involving simple Machine Learning algorithms, which are more suitable to systems. Similarly, dynamic modeling provides more realistic results for practical economy.

Top: Dr. Ismael Mourifié delivers his lecture. Below: ASE students listen to the SIES lectures.