… mindmap¶
It would be great to show a modern roadmap in econometrics starting from mathematic foundations (linear algebra, calculus, probability) to econometrics to computationally intensive data processing tasks. I’ve seen this being approached as clusters of courses, Khan Academy has goals by subject, but I think there is more that can be done.
An (over)simplified view of econometrics curriculum¶
linear algebra, calculus, probability and statistical inference
OLS (assumptions, violations, fixes + estimatore quality)
limited depenedent variables + maximum likelihood
intrumental variables
time series, state space representation
panel data
classifications
systems of equations
Key areas¶
data structures (crosssection, time series, panel)
inference methods
model specification
estimation procedure
model evaluation
use cases
Additional topics¶
simulation (Monte Carlo, bootstrap)
transformations (PCA)
OLS Extensions:
GMM
2,3 stage OLS
quantile regressions
lasso, rigde
Estimation:
maximum likelihood
bayesian estimation
mcmc (see reddit post)
Time series:
time series, stationarity, unit root
state space representation, Kalman filter
fractional integration
seasonal adjustment
(vector) error correction model, VECM
structural breaks
User profiles¶
“Numerate biologists” - solve a domain problem in biology, psychology, social sciences
“Want to hit a ‘Run’ button” - quick results without thinking, typical of students
“I’m doing XYZ now!” - excited adopters, writing a piece on Medium full of acronyms
“Sane econometrics” - appropriate methods with clear, accessible explaination, rare trait
“Asymptotics” - publish evermore sophisticated articles to secure academic career
Discussion¶
Undergraduate Econometrics Instruction: Through Our Classes, Darkly. NBER/IZA and a criticism of G1/G2 goals