60

CHAPTER 4
Exercise Solutions
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 61
EXERCISE 4.1
(a)
()
2
2
2
ˆ 182.85
1 1 0.71051
631.63
i
i
e
R
yy
=-=-=
-
∑
∑
(b To calculate R2 we need ∑( ) yi - y 2 ,
) ∑ ∑ ( ) y y y N y i i - = - = - × = 2 2 2 2 5930.94 20 16.035 788.5155
Therefore,
2 666.72

0.8455
788.5155
SSR
R
SST
===
(c) From
22
2ˆ()ˆ
1 1 R ei N K
SST SST
-σ
=-=-
∑
we have,
2

2
) 552.36 (1 0.7911)
ˆ 6.4104
SST R
σ= (20 2)
×-
 ==
-
(1
NK
-
-

Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 62

EXERCISE 4.2
(a)
ˆ 5.83 8.69
(1.23) (1.17)
y = + x∗
where
10
x
x∗ =
(b)
ˆ 0.583 0.0869
(0.123) (0.0117)
y∗ = + x ˆ
where ˆ
10
y
y∗ =
(c)
ˆ 0.583 0.869
(0.123) (0.117)
y∗ ∗ = + x ˆ
where and ˆ
10 10
yx
y x ∗∗= =
The values of R2 remain the same in all cases.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 63
EXERCISE 4.3
y b b x ˆ0 1 2 0 = + = + × = 1 1 5
(a)
6
n 2
2
0
1 ()
(b) var( ) 1 ˆ 14.9332 ⎞ ⎛ ⎞-
xx
f -
⎛
2
1 (5 1)
5.3333 1

2
()i
 Nxx
⎝⎠∑-
5 10
++⎜ =⎟ =σ++⎜ =⎟
⎝ ⎠
se( ) 14.9332 3.864 f = =
Using se( ) f from part (b)
(c)
and t t c = (0.975,3) = 3.182 ,
y t f ˆ0 ± = ± × = - c se( ) 6 3.182 3.864 (
6.295,18.295)
Using se( ) f from part (b)
(d)
and t t c = (0.995,3) = 5.841,
y t f ˆ0 ± = ± × = - c se( ) 6 5.841 3.864 (
16.570,28.570)
Using x x = = 0 1, the prediction is yˆ0 =1 1 1 2
+ × = , and
n
22
20
(e)
1 1 (1 1) ( )
var( ) 1 5.3333 1 6.340 ˆ
xx
f
⎛⎞-⎛⎞-
=σ++=++=⎜⎟⎜⎟

2
()i
Nxx 5 10
⎝⎠∑- ⎝ ⎠
se( ) 6.340 2.530 f = =
y t f ˆ0 ± = ± × = - c se( ) 2 3.182 2.530 ( 6.050,10.050)
Width in part (c) = - - = 18.295 6.295 24.59 ( )
Width in part (e) = - - = 10.050 6.050 16.1 ( )
The width in part (e) is smaller than the width in part (c), as expected. Predictions are
more precise when made for x values close to the mean.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 64
EXERCISE 4.4
(a) When estimating E( ), y0 we are estimating the average value of y for all observational
units with an x-value of
x0. When predicting y0, we are predicting the value of y for one
observational unit with an x-value of
x0. The first exercise does not involve the random
error
e0; the second does.
(b) E( ) ( ) ( ) b b x E b E b x x 1 2 0 1 2 0 1 2 0 + = + = β + β
()
2
120102012
22222
0 0
 2 2 2
222
22
0
0 2
2

2 2 2
2 2 0 0 0 2 2

var( ) var( ) var( ) 2 cov( , )

2
() () ()
Nxx xx xx
- - -
= + -
()(2)
()
() ()
()
12 i
1 xx
Nxx
xx xx
xxxx
- ⎞-
-
⎛ ⎟
⎛⎞-+
⎠-
()
i N
xx ⎝
⎠-
N
=σ+⎜ =σ+⎟⎜
⎝
i
iii
i
ii
bbxbxbxbb
xxxx
x x Nx x x x
+=++
σσσ
σ-+σ-
=+
∑
∑∑∑
∑
∑∑
∑∑
(c) It is not appropriate to say that E( ) y y ˆ0 0 = because y0 is a random variable.
[E( ) y x x e y ˆ
012012000= β + β ≠ β + β + = ] [ ]

We need to include
y0 in the expectation so that
E y y E y E y x x E e ( ) ( ) ( ) ( ) 0. ˆ ˆ 0 0 0 0 1 2 0 1 2 0 0 - = - = β + β - β + β + = ( )
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 65
EXERCISE 4.5
 (a If we multiply the x values in the simple linear regression model y = β + β + 1 2x e by 10,
) the new model becomes
12 ()
**
1222
10
10
where 10 and 10
yxe
x e x x ∗∗
⎛⎞β
=β+×+⎜⎟
⎝⎠
=β+β+β=β=×
The estimated equation becomes
ˆ 1 2 ( ) 10
10
b
ybx=+×⎛⎞⎜⎟
⎝⎠
Thus, β1 and b1 do not change and β2 and b2 becomes 10 times smaller than their
original values. Since e does not change, the variance of the error term var( ) e = σ2 is
unaffected.
Multiplying all the y values by 10 in the simple linear regression model y = β + β + 1 2x
(b
e
)
gives the new model
y x e × = β × + β × + × 10 10 10 10 ( ) ( ) 1 2 ( )
or
**
y∗ ∗ = β + β + 1 2x e
where
* * y y ∗= × ee=×
β = β × 10, 1 1 β = β × 10, 2 2 10, 10 ∗
The estimated equation becomes
yˆ ˆ ∗ = × = × + × y b b x 10 10 10 ( ) 1 2 ( )
Thus, both β1 and β2 are affected. They are 10 times larger than their original values.
Similarly, b1 and b2 are 10 times larger than their original values. The variance of the new
error term is
var( ) var 10 100 var( ) 100 e e e ∗ = × = × = σ ( ) 2
Thus, the variance of the error term is 100 times larger than its original value.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 66
EXERCISE 4.6
The least squares estimator for β1 is b y b x 1 2 = - . Thus, y = b b x 1 2 + , and hence (
(a) y, x )
lies on the fitted line.
Consider the fitted line yˆi i = + b x b 1 2 . Averaging over N, we obtain
(b)
yˆ = y x ˆi 1 1 ( ) b x b b N b x b b b b x 1 2 1 2 1 2 1 2 i i ( ) i
NNNN
∑ =+=+=+=+∑∑∑
From part (a), we also have y = b b x 1 2 + . Thus, y = yˆ .
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 67
EXERCISE 4.7
(a
yˆ0 2 0 = b x
)
Using the solution from Exercise 2.4 part
(b
(f)
)
()
()()
22222
22
ˆ (2.0659 2.1319 1.1978 0.7363
0.6703 0.6044 11.6044
SSE e = = + + + - i
+-+-=
∑
2222222 4 6 7 7 9 11 352
∑ yi = + + + + + =
2 11.6044

1 0.967
Ru = - = 352
(c)
( )( )
ˆˆ ˆ
(42.549)
yyyy σ
⎡⎤--
()() ˆˆ
ˆ ˆ 65.461 29.333 yy
yy
ii yy
yyyy
⎣ ⎦
= =
= =
2
22
2ˆ
ˆ2222
ˆ
0.943
ii
r
σσ×
--
∑
∑∑
The two alternative goodness of fit measures Ru2 and ryy 2ˆ are not equal.
(d) SST SSR = = 29.333, 67.370
{SSR SSE SST + = + = ≠ = 67.370 11.6044 78.974 29.333 } { }
The decomposition does not hold.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 68
EXERCISE 4.8
Simple linear regression results:
()()
*** ***
ˆ 0.6776 0.0161
(se) 0.0725 0.0026
y t t= +
Linear-log regression results:
()
()()
*** ***
(a) ˆ 0.5287 0.1855 ln
(se) 0.1472 0.0481
y t t= +
Quadratic regression results:
()()
2
*** ***
ˆ 0.7914 0.000355
(se) 0.0482 0.000046
y t t= +
(i) (ii)
2
0.4595
R=
2
0.2441
R=
2
0.5685
R=
(b)
-.8
-.4
.0
.4
.8
0.4
0.8
1.2
1.6
2.0
2.4
50 55 60 65 70 75 80 85 90 95
Residual Actual Fitted
Figure xr4.8(a) Fitted line and residuals for the simple linear regression
-0.8
-0.4
0.0
0.4
0.8
1.2
0.4
0.8
1.2
1.6
2.0
2.4
50 55 60 65 70 75 80 85 90 95
Residual Actual Fitted
Figure xr4.8(b) Fitted line and residuals for the linear-log regression
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 69
Exercise 4.8(b) continued
(b)
-.6
-.4
-.2
.0
.2
.4
.6
0.4
0.8
1.2
1.6
2.0
2.4
50 55 60 65 70 75 80 85 90 95
Residual Actual Fitted
Figure xr4.8(c) Fitted line and residuals for the quadratic regression
(iii) Error normality tests Jarque-Bera:
p-value =
0.870
Simple linear: JB = 0.279
p-value =
Linear log: JB = 1.925
0.382
Quadratic: JB = 0.188
p-value =
0.910
(iv) Values of R2 are given in part (a)
To choose the preferred equation we consider the following.
1. The signs of the response parameters β2 2 2 , and α γ : We expect them to be positive
because we expect yield to increase over time as technology improves. The signs of the
estimates of
β α γ 2 2 2 , and are as expected.
2. R2 : The value of R2 for the third equation is the highest, namely 0.5685.
3. The plots of the fitted equations and their residuals: The upper parts of the figures
display the fitted equation while the lower parts display the residuals. Considering the
plots for the fitted equations, the one obtained from the third equation seems to fit the
observations best. In terms of the residuals, the first two equations have concentrations
of positive residuals at each end of the sample. The third equation provides a more
balanced distribution of positive and negative residuals throughout the sample.
The third equation is preferable.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 70
EXERCISE 4.9
(a) Equation 1: yˆ0 = + × = 0.6776 0.0161 49 1.467
Equation 2: yˆ0 = + = 0.5287 0.1855ln(49) 1.251
Equation 3: yˆ0 = + × = 0.7914 0.0003547 (49) 1.643 2
(b) Equation 1:
m
1
dyt ˆ 0.0161
dt
=β=
Equation 2:
m
ˆ 1 0.1855
0.0038
49
dyt
dt t
α
===
Equation 3:
m
dyt 2 2 0.0003547 49 0.0348 ˆ1t
dt
=γ=××=
(c) Evaluating the elasticities at t = 49 and the relevant value for yˆ0 gives the following
results.
Equation 1:
n
1
0
49
ˆ 0.0161 0.538
ˆ 1.467
t
t
dy t t
dt y y
=β=×=
Equation 2:
n
ˆ 1 0.1855
0.148
ˆ 1.251
t
tt
dy t
dt y y
α
===
Equation 3:
n 22
10
2ˆ 2 0.0003547 49
1.037
ˆ 1.643
t
t
dy t t
dt y y
γ××
===
(d The slopes dyt
) dt

t
and the t
dt
elasticities
y
dy t
give the marginal change in yield and the
percentage change in yield, respectively, that can be expected from technological change
in the next year. The results show that the predicted effect of technological change is very
sensitive to the choice of functional form.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 71
EXERCISE 4.10
For households with 1 child
n
(a
1.0099 0.1495ln( )
)
WFOOD TOTEXP
=-
2
(se) (0.0401) (0.0090) 0.3203
( ) (25.19) ( 16.70)
R
t
=
-
For households with 2 children:
n
2
0.9535 0.1294ln( )
(se) (0.0365) (0.0080) 0.2206
( ) (26.10) ( 16.16)
WFOOD TOTEXP
R
t
=-
=
-
For
β2 we would expect a negative value because as the total expenditure increases the
food share should decrease with higher proportions of expenditure devoted to less
essential items. Both estimations give the expected sign. The standard errors for b b 1 2 and
from both estimations are relatively small resulting in high values of t ratios and
significant estimates.
(b) For households with 1 child, the average total expenditure is 94.848 and
()
()
12 []
12
ln 0.5461
b b TOTEXP ˆ
⎦
⎣ η=
==
+
1.0099 0.1495 ln(94.848)
-×
For households with 2 children, the
average total expenditure is 101.168 and
()
12 []
ln 1 0.9535 0.12944 ln(101.168) 1
 b b TOTEXP
++⎡⎤
⎣⎦-×+
η===
() 0.9535 0.12944
ln ˆ 0.6363 ln(101.168)
b b TOTEXP -×
+
ln 1 1.0099 0.1495 ln(94.848) 1
b b TOTEXP
++⎡⎤
-×+
12
Both of the elasticities are less than one; therefore, food is a necessity.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 72
Exercise 4.10 (continued)
0.0
0.2
0.4
0.6
0.8
3456
X1
WFOOD1
-0.4
-0.2
0.0
0.2
0.4
3456
X1
RESID
Figure xr4.10(a) Figure xr4.10(b)
(c) Figures xr4.10 (a) and (b) display the fitted curve and the residual plot for households with
1 child. The function linear in WFOOD and ln(TOTEXP) seems to be an appropriate one.
However, the observations vary considerably around the fitted line, consistent with the
low R2 value. Also, the absolute magnitude of the residuals appears to decline as
ln(TOTEXP) increases. In Chapter 8 we discover that such behavior suggests the existence
of heteroskedasticity.
Figures xr4.10 (c) and (d) are plots of the fitted equation and the residuals for households
with 2 children. They lead to similar conclusions to those made for the one-child case.
The values of JB for testing H0 : the errors are normally distributed are 10.7941 and
6.3794 for households with 1 child and 2 children, respectively. Since both values are
greater than the critical value χ = (0.95,2) 2 5.991 , we reject H 0 . The p-values obtained are
0.0045 and 0.0412, respectively, confirming that H0 is rejected. We conclude that for
both cases the errors are not normally distributed.
0.0
0.2
0.4
0.6
0.8
3.5 4.0 4.5 5.0 5.5 6.0
X2
WFOOD2
-0.4
-0.2
0.0
0.2
0.4
3.5 4.0 4.5 5.0 5.5 6.0
X2
RESID
Figure xr4.10(c) Figure xr4.10(d)
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 73
EXERCISE 4.11
Regression results:
n
(a) 51.9387 0.6599
VOTE GROWTH
=+
()() (se)
0.9054 0.1631
( ) ( )( )
2
57.3626 4.4060
t
Predicted value of VOTE in 2000:
n0
VOTE = + × = 51.9387 0.6599 1.603 52.9965
Least squares residual:
n
VOTE VOTE 0 0 - = - = - 50.2650 52.9965 2.7315
Estimated regression:
n
(b)
52.0281 0.6631
VOTE GROWTH = +
(se) (0.931) (0.1652)
Predicted value of VOTE in 2000:
n0
VOTE = + × = 52.0281 0.6631 1.603 53.0910
Prediction error in forecast:
n
f VOTE VOTE = - = - = - 0 0 50.2650 53.0910 2.8260
This prediction error is larger in magnitude than the least squares residual. This result is
expected because the estimated regression in part (b) does not contain information about
VOTE in the year 2000.
95% prediction interval:
n
(c) VOTE t f 0 (0.975,28) ± × = ± × = se( ) 53.091 2.048 5.1648 (42.513, 63.669)
The non-incumbent party will receive 50.1% of the vote if the incumbent party receives
49.9% of the vote. Thus, we want the value of GROWTH for which
(d)
49.9 52.0281 0.6631 = + ×GROWTH
Solving for GROWTH yields
GROWTH = -3.209
Real per capita GDP would have had to decrease by 3.209% in the first three quarters of
the election year for the non-incumbent party to win 50.1% of the vote.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 74
EXERCISE 4.12
Estimated regression:
(a) n
STARTS FIXED_RATE 0 0 = - × 2992.739 194.2334
 In May 2005: STARTS n = - × = 2992.739 194.2334 6.00 1827
In June 2005: STARTS n = - × = 2992.739 194.2334 5.82 1862
Prediction error for May 2005:
n
f STARTS STARTS = - = - = 0 0 2041 1827 214
Prediction error for June 2005:
(b) n
f STARTS STARTS = - = - = 0 0 2065 1862 203
Prediction interval for May 2005:
n
STARTS t f 0 (0.975,182) ± × = ± × = se( ) 1827 1.973 159.58 (1512, 2142)
(c)
Prediction interval for June 2005:
n
STARTS t f 0 (0.975,182) ± × = ± × = se( ) 1862 1.973 159.785 (1547, 2177)
Both prediction intervals contained the true values.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 75
EXERCISE 4.13
(a) Regression results:
() ( )( )
() ()()
se 0.0219 0.000013
t 484.84 46.30
ln( ) 10.5938 0.000596
PRICE SQFT
=+
The intercept 10.5938 is the value of ln(PRICE) when the area of the house is zero. This
is an unrealistic and unreliable value since there are no prices for houses of zero area. The
coefficient 0.000596 suggests an increase of one square foot is associated with a 0.06%
increase in the price of the house.
To find the slope d PRICE d SQFT ( ) ( ) we note that
2

d PRICE d PRICE dPRICE ln( ) ln( ) dPRICE 1

=×= ×=β
dSQFT dPRICE dSQFT PRICE dSQFT
Therefore
2
dPRICE
PRICE
dSQFT
=β×
At the mean
dPRICE 2 PRICE 0.00059596 112810.81 67.23
dSQFT
=β×=×=
The value 67.23 is interpreted as the increase in price associated with a 1 square foot
increase in living area at the mean.
The elasticity is calculated as:
2
1%
%
dPRICE dPRICE PRICE PRICE
SQFT SQFT
PRICE dSQFT dSQFT SQFT SQFT
Δ
β×=××==
Δ
At the mean,
elasticity = 0.00059596 1611.9682 0.9607 β × = × = 2 SQFT
This result tells us that, at the mean, a 1% increase in area is associated with an
approximate 1% increase in the price of the house.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 76
Exercise 4.13 (continued)
(b) Regression results:
() ()()
() ()()
se 0.1655 0.0225
t 25.20 44.65
ln( ) 4.1707 1.0066ln( )
PRICE SQFT
=+
The intercept 4.1707 is the value of ln(PRICE) when the area of the house is 1 square foot.
This is an unrealistic and unreliable value since there are no prices for houses of 1 square
foot in area. The coefficient 1.0066 says that an increase in living area of 1% is associated
with a 1% increase in house price.
The coefficient 1.0066 is the elasticity since it is a constant elasticity functional form.
To find the slope d PRICE d SQFT ( ) ( ) note that
2
ln( )
ln( )
d PRICE SQFT dPRICE
d SQFT PRICE dSQFT
==β
Therefore,
2 dSQFT
SQFT At the means,
dPRICE PRICE
=β×
2
112810.81
1.0066 70.444
1611.9682
dPRICE PRICE
dSQFT SQFT
=β×=×=
The value 70.444 is interpreted as the increase in price associated by a 1 square foot
increase in living area at the mean.
(c) From the linear function, R2 = 0.672 .
From the log-linear function in part(a),
 2 [] 2
92
9
9

2
cov( , ) ˆ 1.99573 10 0.715
yy ⎡⎤⎣⎦× =
ˆ[corr( , )]
Ryy
===
ˆ
var( ) var( ) 2.78614 10 1.99996 10 g
yy
×××
From the log-log function in part(b),
2 [] 2
92

2
cov( , ) ˆ 1.57631 10
yy ⎣⎦⎡⎤×
9 ˆ[corr( , )]
9 0.673 Ryy
= ===
ˆ
var( ) var( ) 2.78614 10 1.32604 10 g
yy
×××
The highest R2 value is that of the log-linear functional form. The linear association
between the data and the fitted line is highest for the log-linear functional form. In this
sense the log-linear model fits the data best.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 77
Exercise 4.13 (continued)
(d)
0
20
40
60
80
100
120
-0.75 -0.50 -0.25 0.00 0.25 0.50 0.75
Figure xr4.13(a) Histogram of residuals for log-linear model
0
20
40
60
80
100
120
-0.75 -0.50 -0.25 0.00 0.25 0.50 0.75
Figure xr4.13(b) Histogram of residuals for log-log model
0
40
80
120
160
200
-100000 0 100000 200000
Figure xr4.13(c) Histogram of residuals for simple linear model
Log-linear:
Jarque-Bera = 78.85, p -value = 0.0000
Log-Log:
Jarque-Bera = 52.74, p -value = 0.0000
Simple
Jarque-Bera = 2456, p -value = 0.0000
linear:
All Jarque-Bera values are significantly different from 0 at the 1% level of significance.
We can conclude that the residuals are not normally distributed.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 78
Exercise 4.13 (continued)
(e)
-0.8
-0.4
0.0
0.4
0.8
1.2
0 1000 2000 3000 4000 5000
SQFT
residual
Figure xr4.13(d) Residuals of log-linear model
-0.8
-0.4
0.0
0.4
0.8
1.2
0 1000 2000 3000 4000 5000
SQFT
residual
Figure xr4.13(e) Residuals of log-log model
-150000
-100000
-50000
0
50000
100000
150000
200000
250000
0 1000 2000 3000 4000 5000
SQFT
residaul
Figure xr4.13(f) Residuals of simple linear model
The residuals appear to increase in magnitude as SQFT increases. This is most evident in
the residuals of the simple linear functional form. Furthermore, the residuals in the area
around 1000 square feet of the simple linear model are all positive indicating that perhaps
the functional form does not fit well in this region.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 79
Exercise 4.13 (continued)
(f) Prediction for log-linear model:
n( )
()
2
12
2
exp 2 ˆ
exp 10.59379+0.000595963 2700+0.20303 2
203,516
PRICE b b SQFT = + + σ
=×
=
Prediction for log-log model:
n exp 4.170677 + 1.006582 log(2700)+0.208251 2 ( ) 2

188,221
PRICE = ×
=
Prediction for simple linear model:
n
PRICE = - + × = 18385.65 81.3890 2700 201,365
(g) The standard error of forecast for the log-linear model is
()
()
()
2
2
1 2700 1611.968
i
Nxx
⎢⎥⎣⎦-
∑
0.203034 1 0.20363
=σ++⎢ ⎥
2
20
1
se( ) 1 ˆ
880 248768933.1
xx
f
⎡⎤-
-
=++=
The 95% confidence interval for the prediction from the log-linear model is:

(n ( ) )
()
[]
exp ln( ) se (0.975,878)
exp 10.59379+0.000595963 2700 1.96267 0.20363
133,683; 297,316
ytf±
=×±×
=
The standard error of forecast for the log-log model is
1 ( ) 7.90101 7.3355 2
se( ) 0.208251 1
880 0.20876 85.34453
f
-
= + + =
The 95% confidence interval for the prediction from the
log-log model is

(n ( ))
()
[]
exp ln( ) se (0.975,878)
exp 4.170677 + 1.006582 log(2700) 1.96267 0.20876
122,267; 277,454
ytf±
=×±×
=
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 80
Exercise 4.13(g) (continued)
(g) The standard error of forecast for the simple linear model is
1 ( ) 2700 1611.968 2
se( ) 30259.2 1
30348.2
880 248768933.1
6
f
-
=++=
The 95% confidence interval for the prediction from the simple linear model is
()
()
ˆ0 (0.975,878)se 201,364.62 1.96267 30,348.26
141,801; 260,928
ytf±=±×
=
(h) The simple linear model is not a good choice because the residuals are heavily skewed to
the right and hence far from being normally distributed. It is difficult to choose between
the other two models – the log-linear and log-log models. Their residuals have similar
patterns and they both lead to a plausible elasticity of price with respect to changes in
square feet, namely, a 1% change in square feet leads to a 1% change in price. The loglinear model is
favored on the basis of its higher Rg2 value, and its smaller standard
deviation of the error, characteristics that suggest it is the model that best fits the data.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 81
EXERCISE 4.14
(a)
0
40
80
120
160
200
240
0 10 20 30 40 50 60
Figure xr4.14(a) Histogram of WAGE
0
10
20
30
40
50
60
70
80
1.0 1.5 2.0 2.5 3.0 3.5 4.0
Figure xr4.14(b) Histogram of ln(WAGE)
Neither WAGE nor ln(WAGE) appear normally distributed. The distribution for WAGE is
positively skewed and that for ln(WAGE) is too flat at the top. However, ln(WAGE) more
closely resembles a normal distribution. This conclusion is confirmed by the Jarque-Bera
test results which are JB = 2684 (p-value = 0.0000) for WAGE and JB =17.6 (p-value =
0.0002) for ln(WAGE).
(b) The regression results for the linear model are
n
0.2024 2
4.9122 1.1385
R=
WAGE EDUC = - +
 () ()()
se 0.9668 0.0716
The estimated return to education at the 2
100 100 11.15% 1.1385
mean 10.2130
b
WAGE
=×=×=
The results for the log-linear model are

(n) 0.2146 2
ln 0.7884 0.1038 R=
WAGE EDUC = +
() ()()
se 0.0849 0.0063
The estimated return to education
= b2 × = 100 10.38%.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 82
Exercise 4.14 (continued)
(c)
0
40
80
120
160
200
240
-10 0 10 20 30 40
Figure xr4.14(c) Histogram of residuals from simple linear regression
0
10
20
30
40
50
60
70
80
90
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
Figure xr4.14(d) Histogram of residuals from log-linear regression
The Jarque-Bera test results are JB = 3023 (p-value = 0.0000) for the residuals from the
linear model and JB = 3.48 (p-value = 0.1754) for the residuals from the log-linear model.
Both the histograms and the Jarque-Bera test results suggest the residuals from the loglinear model are
more compatible with normality. In the log-linear model a null hypothesis
of normality is not rejected at a 10% level of significance. In the linear regression model it
is rejected at a 1% level of significance.
(d) Linear model: R2 = 0.2024
[cov( , ) ˆ
Log-linear model: 6.871962
2
2
]2

yy
[corr( , )] 0.2246 ˆ
Ryy
====
g
var( ) var( ) 38.9815
5.39435 ˆ
 yy
×
Since, 2 2
Rg > R we conclude that the log-linear model fits the data better.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 83
Exercise 4.14 (continued)
(e)
-20
-10
0
10
20
30
40
50
0 2 4 6 8 10 12 14 16 18 20
EDUC
residual
Figure xr4.14(e) Residuals of the simple linear model
-1.6
-1.2
-0.8
-0.4
0.0
0.4
0.8
1.2
1.6
2.0
0 2 4 6 8 10 12 14 16 18 20
EDUC
residual
Figure xr4.14(f) Residuals of the log-linear model
The absolute value of the residuals increases in magnitude as EDUC increases, suggesting
heteroskedasticity which is covered in Chapter 8. It is also apparent, for both models, that
there are only positive residuals in the early range of EDUC. This suggests that there
might be a threshold effect – education has an impact only after a minimum number of
years of education. We also observe the non-normality of the residuals in the linear model;
the positive residuals tend to be greater in absolute magnitude than the negative residuals.
(f) Prediction for simple linear model:
n0
WAGE = - + × = 4.9122 1.1385 16 13.30
Prediction for log-linear model:
WAGE nc = + × + = exp 0.7884 0.1038 16 (0.4902 )/ 2 13.05 ( ) 2
Actual average wage of all workers with 16 years of education = 13.30
(g) The log-linear function is preferred because it has a higher goodness-of-fit value and its
residuals are consistent with normality. However, when predicting the average age of
workers with 16 years of education, the linear model had a smaller prediction error
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 84
EXERCISE 4.15
Results using cps_small.dat
(a), (b)
Summary statistics for WAGE
Sub-sample Mean Std Dev Min Max CV
(i) all males 11.525 6.659 2.07 60.19 57.8
(ii) all females 8.869 5.484 2.03 41.32 61.8
(iii) all whites 10.402 6.343 2.03 60.19 61.0
(iv) all blacks 8.259 4.740 3.50 25.26 57.4
(v) white males 11.737 6.716 2.07 60.19 57.2
(vi) white females 9.007 5.606 2.03 41.32 62.2
(vii) black males 9.066 5.439 3.68 25.26 60.0
(viii) black females 7.586 4.003 3.50 18.44 52.8
These results show that, on average, white males have the highest wages and black
females the lowest. The wage of white females is approximately the same as that of black
males. White females have the highest coefficient of variation and black females have the
lowest.
(c)
Regression results
EDUC
Sub-sample Constant 0.0967
(i) all males 1.0075 (0.0084)
(se) (0.1144) 0.1097
(ii) all females 0.5822 (0.0088)
(se) (0.1181) 0.1048
(iii) all whites 0.7822 (0.0065)
(se) (0.0881) 0.0744
% return R2
(iv) all blacks 1.0185 (0.0238)
9.67 0.2074
(se) (0.3108) 0.0987
(v) white males 0.9953 (0.0087)
(se) (0.1186) 0.1085
(vi) white females 0.6099 (0.0091)
(se) (0.1223) 0.0535
(vii) black males 1.3809 (0.0321)
(se) (0.4148) 0.1275
(0.0360)
10.97 0.2404
10.48 0.2225
7.44 0.1022
9.87 0.2173
10.85 0.2429
5.35 0.0679
(viii) black females 0.2428 12.75 0.2143
(se) (0.4749)
The return to education is highest for black females (12.75%) and lowest for black males
(5.35%). It is approximately 10% for all other sub-samples with the exception of all blacks
where it is around 7.5%.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 85
Exercise 4.15 (continued)
Results using cps_small.dat
The model does not fit the data equally well for each sub-sample. The best fits are for
all
females and white females. Those for all blacks and black males are particularly poor.
(d) The
2
t-value for testing H0 2 : 0.10 β = against H1 2 : 0.10 β ≠ is given by
0.1
b
t
(e)
2
se( )
b
-
=
We reject H0 if t t > c or t t < - c where t t c = (0.975,df ) . The results are given in the following
table.
Test results for
H0 2 : 0.10 β = versus H1 2 : 0.10 β ≠
Sub-sample t-value df tc p-value Decision
(i) all males
(ii) all females - 0.394 504 1.965 0.6937 Fail to reject H 0
(iii) all whites 1.103 492 1.965 0.2707 Fail to reject H 0
(iv) all blacks 0.745 910 1.963 0.4563 Fail to reject H0
(v) white males - 1.074 86 1.988 0.2856 Fail to reject H 0
(vi) white - 0.149 464 1.965 0.8817 Fail to reject H 0
females 0.931 444 1.965 0.3525 Fail to reject H 0
(vii) black males - 1.447 38 2.024 0.1560 Fail to reject H 0
(viii) black 0.764 46 2.013 0.4485 Fail to reject H 0
females
There are no sub-samples where the data contradict the assertion that the wage return to an
extra year of education is 10%. Thus, although the estimated return to education is much
lower for all blacks and black males, it is not sufficiently less to conclude conclusively it
is not equal to 10%.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 86
EXERCISE 4.15
Results using cps.dat
(a), (b)
Summary statistics for WAGE
Sub-sample Mean Std Dev Min Max CV
(i) all males 11.315 6.521 1.05 74.32 57.6
(ii) all females 8.990 5.630 1.28 78.71 62.6
(iii) all whites 10.358 6.275 1.05 78.71 60.6
(iv) all blacks 8.626 5.387 1.57 39.35 62.5
(v) white males 11.491 6.591 1.05 74.32 57.4
(vi) white females 9.105 5.648 1.28 78.71 62.0
(vii) black males 9.307 5.274 2.76 34.07 56.7
(viii) black females 8.129 5.424 1.57 39.35 66.7
These results show that, on average, white males have the highest wages and black
females the lowest. Males have higher average wages than females and whites have higher
average wages than blacks. The highest wage earner is, however, a white female. Black
females have the highest coefficient of variation and black males have the lowest.
(c)
Regression results
Sub-sample Constant EDUC
(i) all males 0.9798 0.0982
(se) (0.0543) (0.0040)
(ii) all females 0.4776 0.1173
(se) (0.0579) (0.0043) % return R2
(iii) all whites 0.7965 0.1040 9.82 0.1954
(se) (0.0428) (0.0032)
(iv) all blacks 0.6230 0.1066
(se) (0.1390) (0.0106)
(v) white males 0.9859 0.0988
 (se) (0.0561) (0.0042)
(vi) white females 0.5142 0.1152
(se) (0.0611) (0.0045)
(vii) black males 1.0641 0.0798
(se) (0.2063) (0.0157)
0.1327
(0.0138)
11.73 0.2479
10.40 0.2030
10.66 0.1800
9.88 0.2009
11.52 0.2453
7.98 0.1167
(viii) black females 0.2147 13.27 0.2569
(se) (0.1820)
The return to education is highest for black females (13.27%) and lowest for black males
(7.98%). It is approximately 10% for all other sub-samples with the exception of all
females and white females where it is around 11.5%.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 87
Exercise 4.15 (continued)
Results using cps.dat
The model does not fit the data equally well for each sub-sample. The best fits are for
all
females, white females and black females. That for black males is particularly poor.
(d) The
2
t-value for testing H0 2 : 0.10 β = against H1 2 : 0.10 β ≠ is given by
0.1
b
t
(e)
2
se( )
b
-
=
We reject H0 if t t > c or t t < - c where t t c = (0.975,df ) . The results are given in the following
table.
Test results for
H0 2 : 0.10 β = versus H1 2 : 0.10 β ≠
Sub-sample t-value df tc p-value Decision
(i) all males
(ii) all females - 0.444 1.961
(iii) all whites 4.023 1.961
(iv) all blacks 1.276 1.961
(v) white males 0.629 1.965
2435 0.6568 Fail to reject H 0
(vi) white - 0.296 1.961
females 3.385 1.961
(vii) black males - 1.284 1.972
(viii) black 2.370 1.969
females
2294 0.0001 Reject H 0
 4264 0.2019 Fail to reject H0
465 0.5294 Fail to reject H 0
2238 0.7669 Fail to reject H 0
2024 0.0007 Reject H 0
195 0.2005 Fail to reject H 0
268 0.0185 Reject H 0
The null hypothesis is rejected for females, white females and black females. In these
cases the wage return to an extra year of education is estimated as greater than 10%. In all
other sub-samples, the data do not contradict the assertion that the wage return is 10%.
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 88
EXERCISE 4.16
(a) Regression results:
n
65.503 0.003482
BUCHANAN BUSH
=+
() ()()
() ()()
se 17.293 0.000249
t 3.788 13.986
2
The R2 tells us that 75.35% of the variation in votes for Pat Buchanan are explained by
variation in the votes for George Bush (excluding Palm Beach).
(b) The vote in Palm Beach for George Bush is 152,846. Therefore, the predicted vote for Pat
Buchanan is:
n0
BUCHANAN = + × = 65.503 0.003482 152,846 598
( )2
se( ) 112.2647 1
11 116.443
f
66 2.0337296 10
1 152,846 41761.9697
-
=++=
×
The 99.9% confidence interval is
y t f ˆ0 (0.9995, 66) ± × = ± × = se 597.7 3.449 116.443 196, 999 ( ) ( )
The actual vote for Pat Buchanan in Palm Beach was 3407 which is not in the prediction
interval. The model is clearly not a good one for explaining the Palm Beach vote. This
conclusion is confirmed by the scatter diagram in part (c).
(c)
0
500
1000
1500
2000
2500
3000
3500
0 200 400 600 800 1000 1200
BUCHANANHAT
BUCHANAN
Figure xr4.16(a) Predictions versus actual observations on Buchanan vote
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 89
Exercise 4.16 (continued)
(d) Regression results:
n
0.6305 2
109.23 0.002544
R
BUCHANAN GORE
=
=+
() ()()
() ()()
se 19.52 0.000243
t 5.596 10.450
The R2 tells us that 63.05% of the variation in votes for Pat Buchanan are explained by
variation in the votes for Al Gore (excluding Palm Beach).
The vote in Palm Beach for Al Gore is 268,945. Therefore, the predicted vote for Pat
Buchanan is:
n0
BUCHANAN = + × = 109.23 0.002544 268945 793
( )2
se( ) 137.4493 1
11 149.281
f
66 3.188628 10
1 268,945 39975.55
-
=++=
×
The 99.9% confidence interval is
y t f ˆ0 (0.9995, 66) ± × = ± × = se 793.3 3.449 149.281 278, 1308 ( ) ( )
The actual vote for Pat Buchanan in Palm Beach was 3407 which is not in the prediction
interval. The model is clearly not a good one for explaining the Palm Beach vote. This
conclusion is confirmed by the scatter diagram below.
0
500
1,000
1,500
2,000
2,500
3,000
3,500
0 200 400 600 800 1,000 1,200
BUCHANANHAT2
BUCHANAN
Figure xr4.16(b) Predictions versus actual observations on Buchanan vote
Chapter 4, Exercise Solutions, Principles of Econometrics, 3e 90
Exercise 4.16 (continued)
(e) Regression results:
n
0.1004 2
0.0017 0.01142
R
BUCHSHARE BUSHSHARE
=
=-+
() ()()
() ()()
se
0.0024 0.00427
t
 0.710 2.673
-
The share of votes for George Bush in Palm Beach was 0.354827. Therefore, the predicted
share of votes in Palm Beach for Pat Buchanan is:
n
BUCHSHARE 0 = - + × = 0.001706 0.011424 0.354827 0.002348
The standard error of the forecast error is
1 ( ) 0.354827 0.554756 2
se( ) 0.003078 1
0.0032168
f
66 0.518621
-
=++=
A 99.9% confidence interval is given by
y t f ˆ0 (0.9995, 66) ± × = ± × = - se 0.002349 3.449 0.0032168 0.0087457, 0.0134437 ( ) ( )
There were 430,762 total votes cast in Palm Beach. Multiplying the confidence interval
endpoints by this figure yields (-3767, 5791) . The actual vote for Pat Buchanan in Palm
Beach was 3407 which falls inside this interval.