Professional Documents
Culture Documents
Spring 2014
J. Berry
Liyuan Cui
Xin Fan
1 n
(Yi Y )2
n 1 i=1
X N(0, 1/4)
(a) Graph the PDF of X
(c) Find P(X = 1), and draw the corresponding region on your graph from part (b).
Solution: The corresponding region of P(X = 1) is the vertical line drawn in part (a).
The area of a vertical line equals 0. Thus, we have:
P(X = 1) = 0.
3. (15 points) Suppose {X1 , ..., Xn } is an i.i.d random sample of size n drawn from a distribution
with mean and variance 2 . Define the following two estimators for :
b1 = X
b2 =
1 n
Xi
n 1 i=2
n
= 1 E(Xi ) =
b1 ) = E(X)
E(
n i=1
b1 ) = E(
b1 ) = 0
Bias(
b2 ) = E(
E(
1 n
1 n
n1
E(Xi ) =
X
)
=
E(Xi ) =
i
n 1 i=2
n 1 i=2
n1
b2 ) = E(
b2 ) = 0
Bias(
n
2
Var(X
)
=
i
n2
n
n
1
2
n1
Var(X
)
=
Var(X
)
=
i
i
(n 1)2 i=2
(n 1)2
n1
(c) Find whether or not these two estimators are consistent. (Hint: it may be helpful to
n
b2 as n1
X Xn1 )
express
Solution:
=
plim(1 ) = plim(X)
n
X1
n1
plim( 1 )X1 =
plim( 2 ) = plim(
X) + plim( ) = plim(
)plim(X)
n1
n
n
n
Thus, both estimators are consistent.
(d) Which one do you prefer? Explain.
Solution: Given both estimators are both unbiased, we prefer the one that is more
efficeint. We consider an estimator is more efficient if it has smaller variance.
Bias(1 ) = Bias(2 )
Var(1 ) =
2
2
< Var(2 ) =
n
n1
Thus, we prefer the estimator 1 , which has smaller smaller variance and is more efficient.
4
0
else
(b) the marginal pdfs g(x) and h(y)
Solution:
2
g(x) = 10 (x + y)dy = (xy + y2 )|10 = x + 21 0 x 1
3
2
E(X) = 10 x g(x)dx = 10 (x2 + 2x )dx = ( x3 + x4 )|10 =
7
By Symmetry, E(Y ) = 10 y h(y)dy = 12
7
12
(e) E(XY ) [You do not have to compute the integrals, just set them up]
Solution:
E(XY ) = 10 10 xy f (x, y)dxdy
(f) Cov(X,Y ) [Use the fact that E(XY ) = 13 ]
Solution:
7
7
1
Cov(X,Y ) = E(XY ) E(X) E(Y ) = 13 12
12
= 144
(g) Are X and Y independent random variables? Why or why not?
Solution:
No, they are not independent. f (x, y)6=g(x)*h(y)
5. (35 points) New Jerseys minimum wage rose from $4.25 to $5.05 per hour on April 1, 1992.
As an economist, you are hired by the state government to evaluate the impact of the law on
employment. To do this, you collect data on the employment of 321 fast-food restaurants in
5
r
t = 4NJ
= 1.09
= 9.67/
2
321
NJ
nNJ
Solution:
Note that this is a double-sided test, the p value for t-statistic value,1.09, is0.8621.
Thus, the p-value is 2 (1 (1.09)) = 0.2758
(c) What must you assume for the difference in employment in NJ across the two waves to
be a causal estimate of the change in employment due to the minimum wage increase?
Solution:
We need to assume that all other factors need to remain unchanged and only the minimum wage changes affect the employment changes.
(d) In order to improve the causal interpretation of your estimates, you decide to test
whether the change in employment in NJ was different from the change in employment
in PA. Using a two-sided test and 5% confidence level, conduct the test, by constructing
the following:
i. The null hypothesis
Solution:
H0 : 4NJ = 4PA
Ha : 4NJ 6=4PA
ii. The test statistic
Solution:
+ n PA
PA
[0.59(2.16)]0
q
11.112
9.672
321 + 78
= 2.01
nPA
[0.067, 5.43]
7
321
78
The confidence interval represents the range of values such that, in 95% of random
samples, the interval contains the true difference in means.