Introductory

2.

a.

regression line:

ATT=34857.3+4103.639MANNY+2281.722PM+5631.682WKND+4029.017PROM

+8081.094TEAM

. estat dwatson

Durbin-Watson d-statistic( 6, 81) = 1.29974

Referring to DW Table, dL=1.364 du=1.624

H0:ρ=0 vs. H1:ρ≠0

Since 1.29974<1.364, we reject H0. There is a serial correlation problem in this regression.

b .

Regression line:

ATT=36601.73+3911.541MANNY+2487.561PM+5678.212WKND+4120.441PROM+4430.108TEAM

c.

d.

H0:ρ=0 vs. H1:ρ≠0

Referring to DW Table, dL=1.338 du=1.653

1.338<1.522606<1.653, the test is inconclusive.

3.

a.

We test: ∆GDPt=β0+γGDPt-1+ρt+α∆GDPt-1+μt

H0: γ=0 vs. H1:γ<0

test statistic =-0.028

t*=-3.130, test statistic >t*, we can’t reject H0.

So there is a unit-autoregressive root in the AR(1) model of GDP.

b.

We can transform the data to growth rate, there are two approaches:

(1) Growth=(GDPt-GDPt-1)/ GDPt-1*100%

(2) Growth=(log(GDPt)-log(GDPt-1))*100%

c.

The estimates of coefficients is 0.4937808,

adjusted R2 is 0.2404.

d.

p=2, AIC=714; P=3, AIC=703; for larger p, AIC continues decreasing, however, the amount of decreasing is quite small. If we consider other factors such as R-square and BIC, we’ll find that p=3 is the best choice. So lag order p=3

e. Use the AR(3) model

We get the forecast of GDP growth rate for 2011 1st quarter is 1.270991%.

4.

a.

16215/31857=50.9%

b.

Estimated equation:

Hours=-17.3465-2.242341kidcount+0.9380073educ+2.089186age-0.0282374agesqr-0.0747539nonmomi

The coefficient on kidcount is -2.242341. Ceteris paribus, a woman who has one more children spends 2.242341 hours less weekly in working.

H0:βkidcount=0 v.s. H1:βkidcount≠0

t=-19.35 p>|t|=0.000, we reject H0, βkidcount is significant at 5% significant level.

c.

F(2,31854)=106.69, Prob>F=0.0000, multi2nd and samesex are jointly significant.

d.

(1) The 2SLS coefficient of educ is 0.9164616, the OLS coefficient of educ is 0.9380073. they have a small difference....