Report the data analysis including statistical results and their discussion. Review the list of variables, codes, and descriptions and make any revisions needed for the estimated models.
Assess the quality of the statistical fit and clearly explain econometric problems addressed, the empirical findings, hypothesis test conclusions, and implications.
Explain any alternative models estimated and how any economically important results differed (3-6 double spaced pages).
Provide tables of summary statistics, appropriate scatter plots of the dependent variable against independent variables, and statistical results. Helpful suggestions for reporting statistical results and dealing with econometric problems:
• Place each results table right after the paragraph where it is first mentioned in the text. Include explanatory or transitional sentences before the table. Give each table a number and title.
• Don't simply copy STATA tables into the research report. Format each table. For statistical estimate tables, include the variable code, parameter estimate, standard error, and either the t-statistic or p-value for each estimated parameter. Also include R2, adjusted R2, and number of observations. See the table below for an example of how to report statistical results.
• Unless most of the estimated parameters are statistically significant, compute variance inflation factors to see if you have high multicollinearity. If any of the VIFs > 10, remove the variable with the highest VIF unless it is needed for theoretical consistency or your research objectives. Consider aggregating variables that are highly collinear but theoretically expected to be important (email the instructor for the procedure to do this if needed). Re-estimate the equation and then compute variance inflation factors again to see if they remain a problem. Repeat if needed until all VIFs are less than 10, or explain why the objectives of the research prevented that. Include the VIF tables either in the body of the report or the Statistical Estimation Appendix.
• If you are using cross-sectional data, test for heteroskedasticity. If it's a problem, compute robust standard errors.
• If you are using time-series data, test for autocorrelation. If it's a problem, estimate an AR(1) model. If you need guidance, see STATA - Helpful Hints or contact the TA.
• Do these tests and correct as needed before discussing your results. Then provide both statistical and economic interpretation of the results. Contrast them to prior literature.
Be thorough.
• If the R2 for the estimated model is very low, your model is likely misspecified which means your parameter estimates could be biased because the model omits some important independent variables or the variables are incorrectly specified. Consider other potentially relevant variables or specifications that are at least equally plausible and see if necessary data can be obtained.
• Specify and estimate some alternative (equally reasonable) models to determine whether your results are sensitive to modest changes in the model specification. Provide statistical and economic interpretation of the sensitivity of the results to the alternative model specifications after testing and adjusting for any violations of the standard statistical assumptions.
• Despite your best efforts, you may still get "unsatisfying results" in the sense they aren't very explanatory or are not what you expected. However, "unsatisfying results" are results. If your research turns up unexpected, insignificant, or generally "unpleasing" results, this is what you report. Your job is to be an objective, competent researcher who clearly presents the results of your research, even if they are not consistent with the theory you used to build your model. However, before doing so, consider whether you might have missed something important in the way you specified the model or used the data and explore some alternative reasonable specifications and tests. Discuss the possibilities in this assignment and examine them.
Attachment:- Ecom.zip