Econometric paper (needs Eview)

Micro Paper A: Multiple Regression Consulting -- Spring 2016

You have a friend who is considering quitting her job and taking a year off to further her education. She knows that you study economics and econometrics, and decides to consult you, thinking you’ll provide advice that is based on real-world data and evidence. Your job is to submit a report (1-2 written pages) that describes the steps you took to address your friend’s decision, and discusses the recommendation you make to your friend, as well as any caveats or cautionary notes you may have regarding the evidence you present. In addition to the 1-2 page written report, you should also attach any regression results, tables or figures you think would be helpful in understanding the data or analysis. You may use the concepts you have learned in class to talk about other interesting things as well. Make sure to explain your answers (and the relevant economic concepts) in a way that a non-statistician can understand. Your explanations should make sense to someone who has not taken Econ 140A.

You should use data from the National Longitudinal Survey of Youth (NLSY). The NLSY is a survey of a nationally-representative sample of individuals age 16-43. Information about these individuals is contained in the dataset “nlsy2000_Spring2016.wf1” in the data folder on Gauchospace. Because your friend lives in California, and doesn’t plan on leaving the west coast work or school, you should limit the sample to the western U.S. for all analyses. You can do this using the command “smpl if Regw = 1”.

 Note that if you want to further limit your sample in any way, you can invoke two limitations using the “smpl” command: suppose you wanted to restrict the sample to married people in the western region (not that you necessarily would want to for this project), you would type: “smpl if Regw=1 and Married=1”. Also, you can type “smpl @all” to return to the full sample.

Refer to the questions below for the specifics of the assignment. Note that you should write this as a cohesive report, not a homework assignment. That is, the points below (a,b,c,d) are given as a guide to what you may need to do for the project, not as individual components to be addressed separately.

Note that the data contains the following variables (you may not need all of them):

 Age: age of the respondent at the time of the interview

 Agesq: age squared

 ASVABC: measure of cognitive ability

 Classgov: dummy variable for government employee

 Classpri: dummy variable for private sector worker

 Classse: dummy variable for self employed

 Earnings: hourly earnings in 1996 dollars

 Educ: years of completed education

 Ethblack: dummy variable for blacks

 Ethhisp: dummy variable for Hispanics

 Exper: years of experience

 Expsq: years of experience squared

 Height85: height in inches in 1985

 Hours: average number of hours worked per week

 Id: a unique id number for every individual

 Married: dummy variable for being married

 Male: dummy variable for being male

 Regnc: dummy variable for living in the North (central)

 Regne: dummy variable for living in the Northeast

 Regs: dummy variable for living in the South

 Regw: dummy variable for living in the West

 Siblings: number of siblings

 Single: dummy variable for being single

 Sm: mother’s years of completed schooling

 Sf: father’s years of completed schooling

 Tenure: years working for current employer

 Tenuresq: tenure squared

 Time: years since 1980

 Urban: dummy variable for living in an urban area

 Weight: individual’s weight in pounds

 Year: year of survey

1. You are advising a friend who is considering quitting her job and taking a year off to further her education, and then returning to the same employer. This means that in the future, she will have one more year of education, but one less year of experience and tenure on the job than otherwise. In terms of future earnings, do you recommend this? Is the answer statistically clear or is it subject to significant doubt? To answer these questions, you will need to (at least) do the following:

a. Build a model that can be used to predict earnings given education, experience and tenure. Discuss how you set up your model, what variables you chose to include, why you chose to include them, and so on.

b. Determine the difference in hourly wage brought about by this decision.

c. Test one or more hypotheses. It is up to you to figure out what hypothesis/hypotheses you should test, but you should set up the hypothesis/hypotheses such that they directly address the question you seek to answer.

d. Make a recommendation. Hint: is considering the change in hourly wage enough, or are there other factors that you should consider?

2. The same question, only now your friend will not return to the same employer. This means that her tenure on the job will reset to zero. Assuming your friend has tenure on the job typical of people in your sample, what is your recommendation?

GRADING: This assignment is worth 10 points. An approximate grading rubric is shown below. 2 points: Answer the questions completely 4 points: Justification of answer (clear, correct, and complete explanations in non-technical language) 2 points: Come up with a theoretically justifiable model 2 points: Clarity of writing, structure, grammar