Guidelines for Writing Reports in Econometrics
Pr. Ph. Polomé, Université Lyon 2
Publié-en-ligne : Feb. 1, 2013

Report Content

A scientific report is neither journalistic, nor political nor literary. It does not try to convince but strives to present verifiable facts, reproducible results, falsifiable theories or data. It is not useful to write long discourse on how important your topic is ; it is sufficient to indicate the problem at hand and who had treated it before.
Occam's razor principle always applies : the simplest explanation/format/text is preferred.
The topic can be absolutely anything that lends itself to econometric analysis. There must be a form of causality relation, that is you must want to estimate the effect that some variables have on a dependent variable. It can be original work or it can be based on a paper/ article, preferably a scientific one. In that case, the article must be cited and you should not only replicate some of its results, but also provide some results of your own. For example, a paper may present results using an Instrumental Variables estimator and more sophisticated GMM estimators; the student might then be interested in replicating the IV results and compare them to OLS, but not venture with GMM. In all cases, the source of the data MUST be identified and I must have access. A list of data sources is here.
The report starts with an introduction stating the objective of the research. Then the data must be described, first globally (what is the origin of the database) and then each variable that will be used in the report. Summary statistics must be provided in a table.
Next, present the econometric model formally, comment on the adequate estimator and then present the econometric results in a table (see format below). Discuss the results, test relevant hypotheses accordingly with your dataset and model.
Conclude briefly by recapitulating the main results and pointing out the weaknesses and ideas for improvement.

Style

Text format

Sources & Plagiarism

To avoid being accused of plagiarism, it is necessary to supply the source as soon as you state something that is neither your own result/opinion nor universally known or accepted, that is to say when one :

Handing in the report

Reports for Lectures in Applied Econometrics

Reports are individual and no longer than 5 pages. The data must be cross-section, thus no time series, so that macro or finance topics should normally be avoided; panel data cannot be used either.

Reports for Panel Data Econometrics

Reports are no longer than 10 pages and must make use of panel data. For M2 EQUADE, reports are by groups of 3 students. For M2 GAEXA, reports are individual.

How I Grade