S&W, Chapter 8

Solution Exercises
 Exercise 8.1
250−243
(a) The percentage increase in sales is 100 ∙ 243 = 2.88%.
The approximation is 100 ∙ ln 250 − ln 243 = 2.84%.

255−243
(b) When 𝑆𝑎𝑙𝑒𝑠2014 = 255, the percentage increase is 100 ∙ = 4.94%.
243
The approximation is 100 ∙ ln 255 − ln 243 = 4.82%.

260−243
When 𝑆𝑎𝑙𝑒𝑠2014 = 260, the percentage increase is 100 ∙ = 7%.
243
The approximation is 100 ∙ ln 260 − ln 243 = 6.76%.

265−243
When 𝑆𝑎𝑙𝑒𝑠2014 = 265, the percentage increase is 100 ∙ = 9.05%.
243
The approximation is 100 ∙ ln 265 − ln 243 = 8.67%.

(c) The approximation works well when the change is small. The quality of the
approximation deteriorates as the percentage change increases.
 Exercise 8.2
(a) According to the regression results in column (1), the house price is expected to
increase by 63% (= 100% ∙ 0.00042 ∙ 1500 ) with an additional 1500 square feet
and other factors held constant. The 99% confidence interval for the percentage
change is 100% ∙ 1500 ∙ 0.00042 ± 2.58 ∙ 0.000038 = [48.294%; 77.706%].

(b) The coefficient in a log-log regression is interpreted as the percentage increase in the
dependent variable with a 1% increase in the independent variable. According to the
regression results in column (2), the house price is expected to increase by 69%
(= 0.69  100) when the house doubles in size and other factors held constant.

(c) The house price is expected to increase by 2.7% (100% ∙ 0.027 ∙ 1) when there is a view
and other factors held constant. The 99% confidence interval for this effect is
100% ∗ 0.027 ± 2.58 ∙ 0.028 = [−4.52%; 9.92%]
Estimated t-value of view is 0.964 and critical value is 2.58. As 0.964 < 2.58, the null
hypothesis that 𝛽𝑣𝑖𝑒𝑤 = 0 cannot be rejected. The effect is not significantly different from 0.
 Exercise 8.2 (ctd.)
(d) The house price is expected to increase by 0.72% (100%  0.0036  2 = 0.72%)
with two additional bedrooms while other factors are held constant. The effect is
0.0036
not statistically significant at a 5% significance level: 𝑡 = 0.037 = 0.09730 < 2.58.
The size and the number of bedrooms variables are likely to be correlated so it is
difficult to assess the relative significance of each. In column (3), it is size which is
statistically significant.

(e) The coefficient on condition is statistically significant at a 1% level

0.12
(𝑡 = 0.036 = 3.33 > 2.58).
 Exercise 8.3

(a) The regression functions for hypothetical values of the regression coefficients that
are consistent with the educator’s statement are: 𝛽1 > 0 and 𝛽2 < 0. When is
plotted against STR the regression will show three horizontal segments. The first
segment will be for values of 𝑆𝑇𝑅 < 20; the next segment for 20 ≤ 𝑆𝑇𝑅 ≤ 25;
the final segment for 𝑆𝑇𝑅 > 25. The first segment will be higher than the second,
and the second segment will be higher than the third.

(b) It happens because of perfect multicollinearity. With all three class size binary
variables included in the regression, it is impossible to compute the OLS estimates
because the intercept is a perfect linear function of the three class size regressors.
 Exercise 8.4
(a) With 2 years of experience, the man’s expected AHE is:
෣
ln 𝐴 𝐻𝐸 = 0.1126 ∙ 16 − 0.392 ∙ 0 + 0.0099 ∙ 0 ∙ 16 + 0.0186 ∙ 2
− 0.000263 ∙ 22 − 0.080 ∙ 0 − 0.083 ∙ 0 − 0.018 ∙ 1 + 1.335 = 3.1548
With 3 years of experience, the man’s expected AHE is:
෣ = 0.1126 ∙ 16 − 0.392 ∙ 0 + 0.0099 ∙ 0 ∙ 16 + 0.0186 ∙ 3
ln 𝐴𝐻𝐸
− 0.000263 ∙ 32 − 0.080 ∙ 0 − 0.083 ∙ 0 − 0.018 ∙ 1 + 1.335 = 3.1720
෣
∆ ln 𝐴 𝐻𝐸 = 3.172 − 3.1548 = 0.0172
∆𝐴෣𝐻𝐸 = 100% ∙ 0.0172 = 1.72%

(b) With 10 years of experience, the man’s expected AHE is:

෣
ln 𝐴 𝐻𝐸 = 0.1126 ∙ 16 − 0.392 ∙ 0 + 0.0099 ∙ 0 ∙ 16 + 0.0186 ∙ 10
− 0.000263 ∙ 102 − 0.080 ∙ 0 − 0.083 ∙ 0 − 0.018 ∙ 1 + 1.335 = 3.2783
With 11 years of experience, the man’s expected AHE is:
෣ = 0.1126 ∙ 16 − 0.392 ∙ 0 + 0.0099 ∙ 0 ∙ 16 +
ln 𝐴𝐻𝐸 0.0186 ∙ 11
− 0.000263 ∙ 112 − 0.080 ∙ 0 − 0.083 ∙ 0 − 0.018 ∙ 1 + 1.335 = 3.2914
෣
∆ ln 𝐴 𝐻𝐸 = 3.2914 − 3.2783 = 0.0131
෣ = 100% ∙ 0.0131 = 1.31%
∆𝐴𝐻𝐸
 Exercise 8.4 (ctd.)
(c) The regression in nonlinear in experience (it includes Potential experience2).

(d) Let 4 denote the coefficient on Potential Experience and 5 denote the coefficient
on (Potential Experience)2. Then, following the discussion in the paragraphs just
about equation (8.7) in the text, the expected change in part (a) is given by
4 − 55 and the expected change in (b) is given by 4 - 215. The difference
between these, say (b) - (a), is 165. Because the estimated value of 5 is
significant at the 5% level (the t-statistic for 5 𝑖𝑠 − 0.000263/0.000024 = −10.96),
the difference between the effects in (a) and (b) (=165) is significant at the 5%
level.

(e) No. This would affect the level of ln(AHE), but not the change associated with
another year of experience.
 Exercise 8.6

(a) (i) 𝑇𝑆 = 𝛽0 + 𝛽1 𝑙𝑢𝑛𝑐ℎ + 𝛽2 𝑙𝑢𝑛𝑐ℎ² + 𝑜𝑡ℎ𝑒𝑟 𝑟𝑒𝑔𝑟𝑒𝑠𝑠𝑜𝑟𝑠 + 𝑢

(ii) 𝐻0 : 𝛽2 = 0 𝐿𝐼𝑁𝐸𝐴𝑅 𝑣𝑠 𝐻1 : 𝛽2 ≠ 0 𝑁𝑂𝑁 𝐿𝐼𝑁𝐸𝐴𝑅

 t-test
 Exercise 8.7
(a) (i) ln(earnings) for females are, on average, 0.44 lower than for men.

(ii) The error term has a standard deviation of 2.65 (measured in log-points).

(iii) Yes. However, the regression does not control for many other factors (like firm
size, industry, profitability, experience and so forth).

(iv) No. In isolation, these results do not imply gender discrimination. Gender
discrimination means that two workers, identical in every way but gender, are
paid different wages. Thus, it is also important to control for characteristics of
the workers that may affect their productivity (education, years of experience,
etc.) If these characteristics are systematically different between men and
women, then they may be responsible for the difference in mean wages. (If this
were true, it would raise an interesting and important question of why women
tend to have less education or less experience than men, but that is a question
about something other than gender discrimination in top corporate jobs.) These
are potentially important omitted variables in the regression that will lead to bias
in the OLS coefficient estimator for Female. Since these characteristics were not
controlled for in the statistical analysis, it is premature to reach a conclusion
about gender discrimination.