Reconsidering Returns with Panel Data

In this assignment you will use panel estimation methods to reconsider union wage effects and the returns to experience.

ü initial steps with data

The data for this exercise is available for downloading as VVDATA in MS Excel workbook format. There is a single worksheet with the data. Definitions are provided by Vella and Verbeek in the paper (p. 168). While the underlying source is the National Longitudinal Survey of Youth (NLSY) the panel data you will use are taken from the Journal of Applied Econometrics archive.

Create an unstructured EViews workfile from the spreadsheet then reshape the data as a balanced panel. YEAR is the time-series identifier and NR is the cross-section id. See EViews Help topics on Working with Panel Data for guidance.

Begin with the entire sample and note the total number of observations. Determine the mean values for WAGE, SCHOOL, EXPERIENCE, BLACK, RURAL and UNION and the simple correlations among these variables. Generate experience squared and also union interactions with all of these variables.

ü reconsidering union wage effects

Develop and present estimates using pooled OLS, fixed effects estimation and random effects estimation. See EViews Help topics on Panel Estimation for guidance. Your specification should use WAGE as the dependent variable and include SCHOOL, EXPERIENCE, EXPERIENCE SQUARED, BLACK, RURAL and UNION as the core set of explanatory variables.

Are you able to include all of these variables across all estimation methods? Explain. How does your estimate of the union wage effect change across estimation methods? Explain. Reestimate fixed effects with the union interaction terms included. Discuss the interpretation of the coefficients on these interaction terms.

What additional control variables in the data set would be most likely to alter the estimated wage effect of union membership?

How would you alter your estimation approach to address the potential endogeneity of the union choice?

ü reconsidering returns to experience

What are the returns to experience based on your estimates using pooled OLS, fixed effects estimation and random effects estimation? Discuss why these estimates may differ across the different estimation methods. Discuss the interpretation of lambda in the random effects model. EViews does not report this value but you can calculate it from the information provided. What is lambda for your random effects estimation? Evaluate the validity of the random effects assumptions using a variant of the Hausman test.

Which of your findings for the returns to experience would you argue to be the most valid?

ü related references

Wooldridge, "Advanced Panel Data Methods," in Introductory Econometrics: A Modern Approach , second edition, Thompson-Southwestern, 2003. Hausman, "Specification Tests in Econometrics," Econometrica, November 1978. Vella and Verbeek, “Whose Wages do Unions Raise? A Dynamic Model of Unionism and Wage Rate Determination for Young Men,” Journal of Applied Econometrics, March 1998. Swaffield, "Does Measurement Error Bias Fixed-Effects Estimates of the Union Wage Effect?" Oxford Bulletin of Economics and Statistics, September 2001. Barth, “Firm-Specific Seniority and Wages,” Journal of Labor Economics, July 1997. Parent, “Industry-Specific Capital and the Wage Profile: Evidence from the National Longitudinal Survey of Youth and the Panel Study of Income Dynamics,” Journal of Labor Economics, April 2000.

update 9/05/06