2
$\begingroup$

Scikit Learn has a guide for machine learning techniques. Matthias Vallentin did a remarkable cheat sheet/cookbook for probability and statistics.

Many heavy summaries of econometric methods are presented in textbooks and articles.

But is there a brief summary of econometrics? Like when one should use GMM, or 2SLS versus OLS, or what to analyze time series with, and so on. Like a map that can be given to an applied economist. Or at least a summary table of econometric methods?

Improve this question
$\endgroup$
2
  • 2
    $\begingroup$ Just to add on the summary. I guess many have heard of it but there is also "The Matrix Cookbook": mit.edu/~wingated/stuff_i_use/matrix_cookbook.pdf $\endgroup$
    – Plissken
    Nov 3, 2014 at 18:25
  • 1
    $\begingroup$ @Dan Very useful. Green (2012) also has a good summary of matrix algebra in the appendix. $\endgroup$
    – Anton Tarasenko
    Nov 3, 2014 at 20:11

4 Answers 4

Reset to default
5
$\begingroup$

There is indeed such a paper, written by some of the most eminent current econometricians.

Econometrics: A Bird's Eye View by John Geweke, Joel Horowitz and Hashem Pesaran

Don't think that this was widely circulated or read, but fits your bill, I think.

Improve this answer
$\endgroup$
4
$\begingroup$

I really like A Guide to Econometrics by Peter Kennedy. Each topic starts with a simple explanation, usually with diagrams, followed by technical notes with some math and references. Most of the book is structured around dealing with violations of the classical linear regressions model, such as endogeneity or time series.

At a higher level, and focused on micro topics, is Cameron and Trivedi's Microeconometrics: Methods and Applications, aimed at the applied researcher. It contains some Stata examples.

The first is a medium-length book, the second is a much longer one. Neither one contains too many proofs.

Improve this answer
$\endgroup$
4
$\begingroup$

Mostly Harmless Econometrics by Angrist and Pischke gets at the core idea of reduced form econometrics. These methods seek to use statistical models with believable assumptions that can establish causality when randomization is not always possible, rather than just association.

Examples:

  1. Instrumental Variables
  2. Regression Discontinuity Design
  3. Propensity Score Matching
  4. Randomized Field Experiments
  5. Natural Experiments
Improve this answer
$\endgroup$
2
$\begingroup$

Such a summary would in essence be a collection of rules of thumb. The problem with rules of thumb in statistics is that they are only safe to use when you know enough not to need the rule of thumb. Otherwise, you will use such a rule of thumb in situations where you should not. So, I am sure that such summaries you ask for exist, but you should only consult them if you already know enough to not need them.

Improve this answer
$\endgroup$
2
  • 1
    $\begingroup$ This is logical. But in practice it may be more helpful to let analysts start from some overview, which guides them to advanced topics. Many have to read and do data work without six semesters of econometrics. $\endgroup$
    – Anton Tarasenko
    Nov 3, 2014 at 15:07
  • 1
    $\begingroup$ A good rule of thumb (but see my answer...) is to only use methods you understand. This usually means to keep it simple, which is such a common rull of thumb (but...) that it has it's own abreviation KIS(S). $\endgroup$
    – Maarten Buis
    Nov 3, 2014 at 15:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.