Assignment: Elementary Econometrics
Directions: Follow the instructions closely.
1. Table 1 shows regression results from an equation where wagei is the dependent variable. Each column has slightly different independent variables. Use 5% significance level for all testing. Use the table to answer the following questions:
(a) Interpret the coefficient on years of schooling in column 4. Use a t-test to test the null that education has negative effects on wages.
(b) Interpret the coefficient on female in column 4. Find a 95% confidence in- terval for the coefficient on female.
(c) Why is the coefficient on female different in each column?
(d) Write out the estimated equation in column 2. (Write the numbers in for the β's.)
(e) Use the estimated equation in column 2 to predict the wages for a man with 12 years education and a woman with 12 years education.
(f) What racial group is the reference (or omitted group) in columns 3 and 4? (g)Interpret the coefficient on race == W hite in column 3.
2. Use the data set nlsy97 big.dta to complete the following tasks:
(a) Write down an equation for estimating income as a function of age, high- est grade completed, gender, race, and parent's income during adolescents (income gross yr 1997).
(b) Which coefficient describes the difference in wages by gender? What sign do you expect to find?
(c) Estimate the equation and report the results in a table similar to table 2. (Include the R2 in your table. The STATA help section shows how to make the table.)
(d) Interpret the coefficient on highest grade completed.
(e) Use a t-test to test if the coefficient is different from zero.
(f) What is the R2 and what does it tell us about the regression?
(g) Give a prediction of wage for a 30 year old Hispanic woman with 14 grades completed whose household income was $40,000 when she was an adolescent.
1. Stata Help
The command to run an OLS regression in stata is "regress" or "reg" for short. Fol- lowing the command should be a list of variables starting with the dependent variable. So if I want to estimate y = β0 + β1x, I type "reg y x". Stata will return something like figure 2. In that example, βˆ0 = 2.541224 and βˆ1 = 1.947394. We will learn what the other numbers mean at a later point.
Making a Regression Table
1. Make sure the estout package is installed. (see hw 1) 2.Run regressions: "reg y x1 x2 x3"
3. store results: eststo r1
4. Make table: esttab r1 using t1.rtf, replace label star(* 0.10 ** 0.05 *** 0.01) se r2 ar2
Table 1: Wage Regressions on 2014 CPS Data
|
|
(1)
|
(2)
|
(3)
|
(4)
|
|
Wage
|
Wage
|
Wage
|
Wage
|
|
Female
|
-1.691*
(0.976)
|
-2.745***
(0.882)
|
-2.328***
(0.862)
|
-2.238***
(0.863)
|
|
Years of Schooling
|
|
2.197*** (0.160)
|
2.222*** (0.157)
|
2.198*** (0.157)
|
|
Age
|
|
|
0.249*** (0.0405)
|
0.229*** (0.0420)
|
|
race==White
|
|
|
-6.560* (3.417)
|
-6.857** (3.417)
|
|
race==Black
|
|
|
-10.45*** (3.645)
|
-10.51*** (3.640)
|
|
race==American Indian
|
|
|
-14.41** (6.017)
|
-14.16** (6.010)
|
|
race==Asian/Pacific Islands
|
|
|
-7.099* (3.768)
|
-7.636** (3.775)
|
|
race==Other
|
|
|
|
|
|
married
|
|
|
|
1.604* (0.906)
|
|
Constant
|
23.45*** (0.671)
|
-6.923*** (2.294)
|
-10.77** (4.309)
|
-10.26** (4.313)
|
|
Observations
|
812
|
812
|
812
|
812
|
|
Standard errors in parentheses
|
|
|
|
|
|
* p < 0.10, ** p < 0.05, *** p < 0.01
|
|
|
|
|