1. The paper should be an empirical investigation that employs estimation methods and test procedures we develop this quarter. You should estimate some interesting economic parameters and test hypotheses about them. Your report should focus on the econometrics that you do; it does not matter whether your results are "good" or "bad."
2. You must use a cross-section data set on micro-level entities. Datasets that include variation over time (e.g., yearly observations on the same person or country) are not allowed; they often introduce challenges that we will not have time to address in this course.
1) Data source. You must find your own data source - you cannot use one of the datasets from our textbook (or any other textbook).
2) Correct interpretation of regression coefficients. This includes:
a) Are you using a qualitative variable? If so, you must turn it into a dummy variable or set of dummies.
b) Are any of your variables in logs? If so, you must interpret the coefficients accordingly.
c) Interpret the sign of your coefficient. Does an increase in x cause y to go up or down?
d) Interpret the magnitude of your coefficient. This means you must tell me what the actual number means (ex: increasing education by one year leads to a $1.00 increase in wages) and tell me whether this seems like a large or small effect (ex: this is a large increase, as it represents 10% of the average wage of $10.00).
3) Correct inference about regression coefficients. This includes:
a) Correctly stating the null and alternative hypotheses that are used when testing for statistical significance.
b) Correctly interpreting the t-stat, p-value or confidence interval. Is your coefficient statistically significant?
c) What does it mean that your coefficient is statistically significant? (Ex: The coefficient on wages is statistically significant. This result shows that education does have an impact on wages.)
4) Test for heteroskedasticity
a) Conduct and interpret a test for heteroskedasticity.
b) If you do have heteroskedasticity, you must correct for it.
5) Discussion of your model and results.
a) What are some variables that might be omitted from your model?
b) Choose one potentially omitted variable. Using the formula for omitted variables bias, determine whether this omitted variable would bias your coefficient of interest upward or downward.