Understanding econometric modeling 2400-ZEWW809
First, methodology:
With just a few lines of R or Python code one can create simplistic econometric simulation. Programing skills required for such a task are limited to absolute minimum. Such simplistic simulation however, let us to answer the question if the model is correct or not. Usage such simulations can be extended to answering many methodological econometric questions, which are hard, or impossible to answer using mathematical derivations.
This way, in much simpler manner, one can get answers for much harder questions. Answers which can be both correct, as well as surprising in the context of econometric theory and practice.
Second, subject:
The main problem of econometric modeling can not be found among ideas known from basic courses: nonlinearity, autocorrelation, heteroskedasticity, or non-normality of model residuals. The fundamental problem of econometrics is variable selection, which will allow for the interpretation of the estimated parameter. For the last two decades the whole scientific branch dedicated to this topic emerged. It is called Causal Inference.
During this course, according to ability and willingness of the group, consecutive fundamental Causal Inference topics will be discussed. These topics gathered together can give clues which variables should be included in the model and why – and as a consequence allow for using words “cause” and “effect” correctly, instead of overinterpreting partial correlation as causation. It is huge shift of the level of the econometric modelling.
During the course, according to ability and willingness of the group, we will discuss following topics:
Fundamentals of Causal Inference:
• Confounder
• Mediator
• Collider
• M-Bias
• Butterfly Bias
Extension of identification problems:
• Attenuation Bias
• Suppressor
• Reversed Causality
• Sample Selection
Solutions for previously known problems:
• Experiment
• Instrumental Variable
• Front Door Criterion
• Regression Discontinuity Design
• Linear Regression
• Dynamic modelling
Type of course
Learning outcomes
KNOWLEDGE
• Knows fundamental econometrics and Causal Inference problems
• Knows simulation-based approach to assessment of correctness of econometric models
• Knows programming basics which allow for numerical experiments
ABILITIES
• Can independently and critically analyse econometric results
• Can use different datasets for own research
SOCIAL COMPETENCE
• Is critical towards econometric and statistical results in social sciences, can explain economic and social phenomena in terms of causality, learns to think, communicate and write in a logical and consistent manner.
Assessment criteria
The grade is based on the classroom activity.
Bibliography
Basic literature:
Hernan, M.A., & Robins, J.M. Causal Inference: What If (1st ed.). CRC Press, 2023.
Lewbel, Arthur. "The identification zoo: Meanings of identification in econometrics." Journal of Economic Literature 57, no. 4 (2019): 835-903.
Imbens, Guido. Potential outcome and directed acyclic graph approaches to causality: Relevance for empirical practice in economics. No. w26104. National Bureau of Economic Research, 2019.
Pearl, Judea, and Dana Mackenzie. The book of why: the new science of cause and effect. Basic Books, 2018.
Elwert, Felix, and Christopher Winship. "Endogenous selection bias: The problem of conditioning on a collider variable." Annual review of sociology 40 (2014): 31-53.
As well as supplementary materials for selected topics.
Additional information
Additional information (registration calendar, class conductors, localization and schedules of classes), might be available in the USOSweb system: