A Regression User’s Handbook Studenmund Chapter 11 11.1 A Regression User’s Checklist (pp. 391-392) 11.2 A Regression User’s Guide (Table 11.2) For each econometric problem, understand • • • • Its nature Its consequences for OLS estimation How to detect it How to attempt to get rid of it – For example, both multicollinearity and irrelevant variables can cause regression coefficients to have insignificant tstatistics 11.3 Running Your Own Regression Report “Econometrics is best learned by doing, not by reading books, listening to lectures, or taking tests.” “Get your hands dirty” Running your own regression project … 1. 2. 3. Choosing a topic Applying the 6 steps in regression analysis Writing your research report 6 Steps of Regression Analysis (p.64)Ch.3 1. 2. 3. 4. 5. 6. Review the literature & develop the theoretical model. Specify the model: Select the independent variables & the functional function. Hypothesize the expected signs of the coefficients. Collect the data. Estimate & evaluate the equation. Document the results. Common Elements in a Research Report a. b. c. d. e. f. g. h. i. j. An abstract (executive summary, >100 words) A brief introduction of the dependent variable and states the research goal. A short review of the relevant literature. An explanation of the model specification. A description of the data & data sources. A presentation of each estimated specification. A detailed analysis of the regression results, including any econometric problems encountered. A short conclusion that includes any policy implications or recommendations. A reference section. An appendix for tables that report regression results. 11.5 The Ethical Econometrican Nothing could be further from the truth! Principle: spend plenty of time to develop the theoretical underpinning and little time to “pursue” alternative estimations of that model. Reality: rare for a first attempt at estimation to be totally problem free. We are tempted to continue running regressions until you likes the way the result looks. Recommendation: report alternative estimated equations Can you handle the truth???? Q:? Why is there a Nobel Prize for economics and none for other social sciences? A:? They already have one in literature. Three econometricians went out hunting, and came across a large deer. The first econometrician fired, but missed, by a meter to the left. The second econometrician fired, but also missed, by a meter to the right. The third econometrician didn't fire, but shouted in triumph, "We got it! We got it!" More Jokes @ faculty.tamucc.edu/jlee/4310 Top 10 Reasons to Learn Econometrics Good Luck!! Ch. 6 – Model Specification Incorrect Specification Results: Explanatory variables correlated with errors Check: stepwise regression Bias Variance Omitted variables yes yes Irrelevant variables no yes Ch.7 -- Functional form To improve fit – curve – dummy variables – lagged independent variables Ch. 8 -- Multicollinearity Check: VIF Bias Variance No yes Ch 9 – Serial correlation Result: errors not random Check: DW statistics Remedy: GLS Bias Variance No yes Ch. 10 – Heteroskedasticity Result: variance not constant Check: Park or White Test Remedy: WLS or HAC Bias Variance No yes Term Paper Due: December 11, 12:00 Noon Format: Double-spaced contents; cover page includes an abstract of 150 words or less. Send your paper via email to jim.lee@tamucc.edu and attach a Word file with a file name: ECON4310_LastName.doc, where LastName is your Last Name. Please DO NOT SEND a file with a docx file extension (Vista format or newer). You will receive acknowledgement of receipt by the end of that day; otherwise, that means I will not have received your email. Your paper may be screened at Turnitin. What else? Economic Scholars Program – Presentation on March 27, 2009 – At the Dallas Fed – College will most likely pay for your travel costs & expenses – Details: dallasfed.org/educate/events/2009/09esp.cfm