PowerPoint Presentation
ECON 4510/5510 Unions and Collective Bargaining:
Data Assignment 8
Re-run the union wage regression (from data assignment 7) with different combinations of variables until the union wage gap is as small as possible.
What factors primarily explain why union and nonunion wages are different?
ECON 4510/5510 Unions and Collective Bargaining:
Data Assignment 8
See audio instructions for Data Assignment 7 for information about regression analysis, how to install the regression tools in excel, and how to run a regression in excel and interpret the results.
ECON 4510/5510 Unions and Collective Bargaining:
Data Assignment 8
In multivariate regression analysis, when you add an independent variable, you are examining the dependent variable “holding that independent variable constant” (because including that independent variable controls for its effect).
ECON 4510/5510 Unions and Collective Bargaining:
Data Assignment 8
We have reason to believe wages are higher in the northeast and workers in the northeast are more likely to be in unions.
When the independent variables for region of residence are not included in the regression, the unions independent variable picks up the effect of being in a union on wages and some of the effect of region of residence on wages (some of the positive effect of living in the northeast on wages).
When the independent variables for region of residence are included, they pick up the effect of region of residence on wages and the unions independent variable picks up the effect of being in a union on wages (but it does not pick up the effect of region of residence on wages, because the region of residence independent variables pick that up).
ECON 4510/5510 Unions and Collective Bargaining:
Data Assignment 8
In my analysis for this assignment, I adjust the wage regression from assignment 7 by adding independent variables to control for region of residence.
ECON 4510/5510 Unions and Collective Bargaining:
Data Assignment 8
Here are the answers when adding controls for region of residence.
Coefficients Standard E
or t Stat P-value
Intercept 19.590 0.526 37.233 0.000
union 2.115 0.578 3.656 0.000
male 5.502 0.445 12.374 0.000
black -6.551 0.534 -12.274 0.000
hispanic -4.271 0.633 -6.747 0.000
northeast 3.542 0.715 4.955 0.000
south 0.995 0.594 1.675 0.094
west 3.051 0.719 4.245 0.000
ECON 4510/5510 Unions and Collective Bargaining:
Data Assignment 8
When holding gender, race/ethnicity, and region of residence constant (e.g., when comparing union and nonunion workers of the same gender and ethnicity in the same region), the union wage premium falls to $2.11.
So, a portion of the reason why union workers earn more than nonunion workers is that union workers are more likely to reside in the regions of the U.S. with higher wages.
ECON 4510/5510 Unions and Collective Bargaining:
Data Assignment 8
Next, I’m going to control for industry of employment in the union wage regression.
ECON 4510/5510 Unions and Collective Bargaining:
Data Assignment 8
Here are the answers when also controlling for industry of employment.
Coefficients Standard E
or t Stat P-value
Intercept 19.742 0.579 34.109 0.000
union 1.956 0.581 3.365 0.001
male 5.736 0.461 12.431 0.000
black -6.692 0.534 -12.539 0.000
hispanic -4.399 0.632 -6.956 0.000
northeast 3.621 0.717 5.051 0.000
south 1.075 0.596 1.805 0.071
west 3.112 0.722 4.313 0.000
manufacturing -0.045 0.681 -0.066 0.948
trade -3.245 0.714 -4.543 0.000
healthcare 1.016 0.652 1.558 0.119
ECON 4510/5510 Unions and Collective Bargaining:
Data Assignment 8
When holding gender, race/ethnicity, region of residence, and industry constant (e.g., when comparing union and nonunion workers of the same gender and ethnicity in the same region who are in the same industry), the union wage premium falls to $1.95.
So, a portion of the reason why union workers earn more than nonunion workers is that union workers are more likely to work in industries with higher wages.