Long Case Study 1 – Determinants of a basketball player's Wage
Background
Sport is a big industry in the United States. The average value of an NBA basketball team in
2022 was $2.86 billion. It is of significant interest how much its players earn. Data was
collected in 1996 from 269 players, including their annual wages and information about
them. A researcher wants to study the relationship between salary, game participation, and
activity.
Model
The model being estimated is shown below with the list of variables. The dataset to use is
wage_de.xls.
wag e i=β 0 + β 1 game s i+ β 2 point si + β 3 ag e i+ β 4 blac k i +ui
The list of variables and their definition are below.
i - i = 1, 2, 3…, 265th player
wag e i - The annual wage of player i in thousands ($’000)
game s i- The average number of games played each year by player i
point s i- The average points scored per game by player i
ag ei - The age of player i
blac k i - A dummy variable indicating if player i is black. It equals 1 if the player is
black and 0 if not.
Copy the questions on the next page and write the answers in a separa te document. Please
include a cover sheet with the group number and members' full names with their student IDs.
The STATA output is on a separate pdf document, Output_LongCaseStudy1.pdf.
Questions (70 points)
1. Write the estimated model based on the output. Do not include the standard errors to
reduce difficulties in writing the equation. (4 points)
^ game s i +110.956 ^
wag e i=−1848.359−3.19552 ^ ag e i+ 67.35674 ^
point s i+ 83.77618 ^ blac k i
2. Write an interpretation of the model explaining the impact of the independent variables
with their coefficients. Take note of the unit of measurement for the wage. (4 points)
A rise of 1 unit in the average number of games played each year will result in a decrease of
3.19552 thousand dollars ($3,195.52) in the annual wage of player, ceteris paribus.
A rise of 1 unit in the average points scored per game by player will result in an increase of
110.956 thousand dollars ($110,956) in the annual wage of player, ceteris paribus.
A rise of 1 unit in the age of player will result in an increase of 83.77618 thousand dollars
($83,776.18) in the annual wage of player, ceteris paribus.
A rise of 1 unit in the average points scored per game by player will result in an increase of
110.956 thousand dollars ($110,956) in the annual wage of player, ceteris paribus.
If the player is black, it will result in an increase of 67.35674 thousand dollars ($67,356.74) in
the annual wage of player, ceteris paribus.
3. Is the model significant in explaining the basketball player's wage? Conduct the test at
5%, showing the null and alternative hypothesis, critical statistic and conclusion. (4 points)
H 0: β 1=β 2=β 3=β 4 =0
H 1: At least one is not equal to 0
( ) ( 0.5113
4 )
2
R
k
F−statistic ( 4 ,264 )= = =69.07
[ ( 1−R2 )
( n−k −1 ) ][ (1−0.5113 )
( 269−4−1 ) ]
F−critica l (0.05 , 4 ,264 )=2.7
Because F−statist ic> F−critical , reject the null hypothesis at 5% significance level.
p−value=0.0000
4. Which independent variables are statistically significant at 5%? State the null
hypothesis, the confidence interval, the conclusion and whether the variable affects wage.
(20 points)
H 0: β 1=0
H 1: β 1 ≠ 0
β 1−β 1 −3.19552−0
t−statistic= = =−1.21
σ1 2.644506
√ n1
t−critica l 0.05=± 1.96
level ⇒ game s i is statistically insignificant.
Because t−statist ic=−1.21>t−critical=−1.96, accept the null hypothesis at 5% significance
H 0: β 2=0
H 1: β 2 ≠ 0
β 2−β 2 110.956−0
t−statistic= = =13.16
σ2 8.429568
√ n2
t−critica l 0.05=± 1.96
level ⇒ poin t i is statistically significant.
Because t−statist ic=13.16> t−critical=1.96, reject the null hypothesis at 5% significance
H 0: β 3=0
H 1: β 3 ≠ 0
β 3−β 3 83.77618−0
t−statistic= = =6.53
σ3 12.8327
√ n3
t−critica l 0.05=± 1.96
⇒ age i is statistically significant.
Because t−statist ic=6.53> t−critical=1.96, reject the null hypothesis at 5% significance level
H 0: β 4 =0
H 1: β 4 ≠ 0
β 4− β 4 67.35674−0
t−statistic= = =0.61
σ4 109.8431
√ n4
t−critica l 0.05=± 1.96
⇒ blac k i is statistically insignificant.
Because t−statist ic=0.61<t−critical=1.96, accept the null hypothesis at 5% significance level
5. The first assumption of the OLS model is that the parameters are linear. What problems
would be with interpreting the model if this assumption was unmet? (4 points)
6. The second assumption of the OLS is that the observations were from a random sample.
Can the model estimates be helpful for a wide range of players? Explain why or why not. (4
points)
7. The third assumption is that there is variation in the explanatory variable. What does
this mean for the model? (4 points)
8. Does the model suffer from multicollinearity? State the statistic used with your
explanation. (4 points)
9. Does the model suffer from misspecification? Conduct the test at 5% from the STATA
output (5 points)
10. Does the model suffer from heteroskedasticity? There are two tests to conduct. Perform
each one at 5%. (10 points)
11. Based on the answers from questions five to ten, are the tests in questions two & three
reliable? Explain why or why not. (7 points)
