ECON 345
Problem Set 6
Aarav Agarwal V00990817
The diminishing return on roe means that as return on effect increases the
marginal increase in salary decreases, that is it increases at a decreasing
rate.
∂ log Salary
=0.0215−0.00016 roe
∂ roe
When roe is 50 the effect is 0.0215-0.00016(50)=0.0135 a positive effect
on salary
When roe is 100 the effect is 0.0215-0.00016(100)=0.0055 a much
smaller positive effect on salary. This generality is not necessary as the t
stat for roe2is -0.31 which is statistically insignificant that means there is
no statistical evidence hat suggests the effect of return on equity is
diminishing.
0.003 0.003
i)The turn around point would be = =$ 21,428.57
(2∗0.0000000070) 0.000000014
ii) it would be good to include the quadratic term as its t stat is -1.89
which is significant at the 5% level
−0.000000070
=1.891 ,
0.0000000037
cv =−1.70 , df =29
iii) coefficients and SEs get 1000x
rdintens=2.613+0.30 salesbil−0.007 0 salesbil ( 0.429 ) ( 0.14 )( 0.0037 )n=32 R =0.1484
2 2
iv) mathematically both the equations yield the exact same relationship
however there would be a preference for the equation in part iii due to the
fewer 0s making it more readable.
i) Holding all else fixed
∆ log ( wage )=β 1 ∆ educ + β 2 ∆ educ∗pareduc= ( β 1+ β 2∗pareduc ) ∆ educ
∆ logwage
divide both sides by ∆ educ¿ =(β 1+ β 2∗pareduc)
∆ educ
The expected sign of beta 2 would be positive as we would
expect a higher wage outcome for children in school with more
educated parents, this maybe because we expect the more
educated the parents the better the quality of learning at home.
ii) The difference in estimated return on education is 0.00078(32-
24)=0.0062 or 0.62 percentage points this means that there is a
0.62 percentage point increase in the return of education when
the child’s parents have gone to college as opposed to only high
school.
iii) Ho: betaeduc∗pareduc=0 Ha :beta educ∗pareduc ≠ 0
−0.0016
tstat= =−1.33 , not significant at 10 %
0.0012
i) If we do include attendance as a explanatory variable then
we will be measuring the effects of alcohol holding the
student’s attendance fixed because attendance is probably
correlated with drinking, therefore it would not be
preferable to keep attendance fixed, if we do not include
attendance then we will be looking at the effect of alcohol
on GPA that is not explained by attendance, for this
analysis it would be better to leave it out.
ii) This would be a good inclusion as these variables are
indicators of student’s abilities and may indicate a pattern
in sat and hs performance and alcohol consumption.