Let G be (achieved points)/10+1.5.

Round G exactly to quarters of a grade to get your grade.

Solution to Homework Set 3

Managerial Economics Fall 2011

Conceptual and Computational Questions

6 points

2. The demand curve for a product is given by Qdx = 1, 000 2Px + .02Pz ,
where Pz = $400.
(a) What is the own price elasticity of demand when Px = $154? Is
demand elastic or inelastic at this price? What would happen to
the firms revenue if it decided to charge a price below $154?
At the given prices, quantity demanded is 700 units: Qdx = 1000
2154+.02400 = 700. Substituting the relevant information into
Px
1 point
the elasticity formula gives: EQx ,Px = 2 Q
= 308
700 = 0.44.
x
Since this is less than one in absolute value, demand is inelastic at
this price. If the firm charged a lower price, total revenue would 1 point
decrease.
(b) What is the own price elasticity of demand when Px = $354? Is
demand elastic or inelastic at this price? What would happen to
the firms revenue if it decided to charge a price below $354?
At the given prices, quantity demanded is 300 units: Qdx = 1000
2354+.02400 = 300. Substituting the relevant information into
Px
1 point
the elasticity formula gives: EQx ,Px = 2 Q
= 708
300 = 2.36.
x
Since this is greater than one in absolute value, demand is elastic
at this price. If the firm increased its price, total revenue would 1 point
decrease.
(c) What is the cross-price elasticity of demand between good X and
good Z when Px = $154? Are goods X and Z substitutes or complements?
At the given prices, quantity demanded is 700 units. Substituting the relevant information into the elasticity formula gives:
1 point
Pz
EQx ,Pz = .02 Q
= .02 400
700 = 0.011. Since this number is positive,
x
goods X and Z are substitutes.

1 point

7 points

3. Suppose the demand function for a firms product is given by

ln Qdx = 3 0.5 ln Px 2.5 ln Py + ln M + 2 ln A
where
Px = $10,
Py = $4,
M = $20, 000, and
A = $250.

(a) Determine the own price elasticity of demand, and state whether
demand is elastic, inelastic, or unitary elastic.
The own price elasticity of demand is simply the coefficient of
1 point
ln Px , which is 0.5. Since this number is less than one in absolute
value, demand is inelastic.
1 point
(b) Determine the cross-price elasticity of demand between good X
and good Y, and state whether these two goods are substitutes or
complements.
The cross-price elasticity of demand is simply the coefficient of
ln Py , which is -2.5. Since this number is negative, goods X and as in a) 2 points
Y are complements.
(c) Determine the income elasticity of demand, and state whether
good X is a normal or inferior good.
The income elasticity of demand is simply the coefficient of ln M ,
which is 1. Since this number is positive, good X is a normal as in a) 2 points
good.
(d) Determine the own advertising elasticity of demand.
The advertising elasticity of demand is simply the coefficient of 1 point
ln A, which is 2.

2 points

8. Suppose the true inverse demand relation for good X is P = a+bQ+e,

and you estimated the parameters to be a
= 10, b = 2.5, a = 1, and
b = 0.5. Find the approximate 95 percent confidence interval for the
true values of a and b.
The approximate 95 percent confidence interval for a is a
2a = 102. 1 point
Thus, you can be 95 percent confident that a is within the range of
8 and 12. The approximate 95 percent confidence interval for b is
b 2 = 2.5 1. Thus, you can be 95 percent confident that b is 1 point
b
within the range of -3.5 and -1.5.

7 points

9. The demand function for good X is qxd = a + bPx + cM + e, where

Px is the price of good X and M is income. Least squares regression
2

This should be -1.36/0.56

reveals that a
= 5.25, b = 1.36, c = 0.14, a = 6.19, b = 0.56,
and c = 0.05. The R-square is 0.24.
(a) Compute the t-statistic for each of the estimated coefficients.
9369.45
The t-statistics are as follows: ta = 11067.07
= 0.848; tb = 1.36
0.56 = each t one point
0.14
2.429; and tc = 0.05 = 2.80.
(b) Determine which (if any) of the estimated coefficients are statistically different from zero.
Since |ta | < 2 the coefficient estimate, a
, is not statistically different from zero. However, since |tb | > 2 and |tc| > 2, the coefficient
estimates b and c are statistically different from zero.
(c) Explain, in plain words, what the R-square in this regression indicates.
The R-square and adjusted R-square tell us the proportion of
variation explained by the regression. The R-square tells us that
24 percent of the variability in the dependent variable is explained
by price and income. The adjusted R-square confirms that fact
and the R-square is not the result of estimating too many coefficients (i.e. few degrees of freedom).

1 point
1 point
2 points

Problems and Applications

3 points

15. You are a division manager at Toyota. If your marketing department estimates that the semiannual demand for the Highlander is
Q = 100, 000 1.25P , what price should you charge in order to maximize revenues from sales of the Highlander?
To maximize revenue, Toyota should charge the price that makes de- 1 point
mand unit elastic. Using the own price elasticity of demand formula, 1 point
P
EQ,P = 1.25 100,0001.25P
= 1. Solving this equation for P implies
that the revenue maximizing price is P = $40, 000. 1 point

7 points

16. As newly appointed Energy Czar, your goal is to reduce the total
demand for residential heating fuel in your state. You must choose on
e of three legislative proposals designed to accomplish this goal: (a) a
tax that would effectively increase the price of residential heating fuel
by $2; (b) a subsidy that would effectively reduce the price of natural gas by $1; or (c) a tax that would effectively increase the price of
electricity (produced by hydroelectric facilities) by $5. To assist you in
your decision, an economist in your office has estimated the demand
for residential heating fuel using a linear demand specification. The regression results are presented in Table 3-1. Based on this information,
which proposal would you favor? Explain.

SUMMARY OUTPUT
Regression Statistics
Multiple R
0.76
R-Square
0.57
Adjusted R0.49
Square
Standard Error
47.13
Observations
25
Analysis of Variance
Degrees
Regression
Residual
Total

Intercept
Price of Residential Heating Fuel
Price of Natural
Gas
Price of Electricity
Income

of Freedom
4
20
24

Sum of Squares
60936.56
44431.27
105367.84

Mean Square
15234.14
2221.56

F
6.86

Significance F
0.03

Coefficients
136.96
-91.69

Standard Error
43.46
29.09

T-Statistic
3.15
-3.15

P-Value
0.01
0.01

Lower 95%
50.6
-149,49

Upper 95%
223.32
-33.89

43.88

9.17

4.79

25.66

62.1

-11.92

8.35

-1.43

0.17

-28.51

4.67

-0.05

0.35

-0.14

0.9

-0.75

0.65

Table 3-1
The estimated demand function for residential heating fuel is QdRHF =
136.96 91.69PRHF + 43.88PN G 11.92PE 0.05M , where PRHF is
the price of residential heating fuel, PN G is the price of natural gas,
PE is the price of electricity, and M is income. However, notice that
coefficients of income and the price of electricity are not statistically
different from zero. Among other things, this means that the proposal
to increase the price of electricity by $5 is unlikely to have a statistically
significant impact on the demand for residential heating fuel. Since
the coefficient of PRHF is -91.69, a $2 increase in PRHF would lead
to a 183.38 unit reduction in the consumption of residential heating
fuel (since 91.69 2 = 183.38 units). Since the coefficient of PN G is
43.88, a $1 reduction in PN G would lead to a 43.88 unit reduction in the
consumption of residential heating fuel (since 43.88 (1) = 43.88).
Thus, the proposal to increase the price of residential heating fuel by
$2 would lead to the greatest expected reduction in the consumption
of residential heating fuel.

best guess effect of each policy: 1 point each

4
statistical significance of each prediction: 1 point each
decision for increasing the price of residential heating fuel: 1
point

10 points

20. According to CNN, two dairy farmers challenged the legality of the
funding of the Got Milk?campaigns. They argued that the Got
Milk? campaigns do little to support milk from cows that are not injected with hormones and other sustainable agriculture products, and
therefore violate their (and other farmers) First Amendment rights.
The 3rd U.S. Circuit Court of Appeals agreed and concluded that dairy
farmers cannot be required to pay to fund the advertising campaigns.
One of the obvious backlashes to the National Dairy Promotion and
Research Board is reduced funding for advertising campaigns. To asses
the likely impact on milk consumption, suppose that the National Dairy
Promotion and Research Board collected data on the number of gallons of milk households consumed weekly (in millions), weekly price
per gallon, and weekly expenditures on milk advertising (in hundreds
of dollars). These data, in forms to estimate both a linear and a loglinear model, are available online at http: // www. cepe. ethz. ch/
education/ ManagerialEconomics in a file named hw3 q20.xls. Use
these data to perform two regressions: a linear regression and a loglinear regression.1 Compare and contrast the regression output of the
two models. Comment on which model does a better job fitting the
data. Suppose that the weekly price of milk is $3.10 per gallon and
the National Dairy Promotion and Research Boards weekly advertising expenditures falls 25 percent after the courts ruling to $100 (in
hundreds). Use the best-fitting model to estimate the weekly quantity
of milk consumed after the courts ruling.
Table 3-11 contains the output from the linear regression model. That
model indicates that R2 = .55, or that 55 percent of the variability in
the quantity demanded is explained by price and advertising. In contrast, in Table 3-12 the R2 for the log-linear model is .40, indicating
that only 40 percent of the variability in the natural log of quantity
is explained by variation in the natural log of price and the natural
log of advertising. Therefore, the linear regression model appears to
do a better job explaining variation in the dependent variable. This
conclusion is further supported by comparing the adjusted R2 s and
the F -statistics in the two models. In the linear regression model the
adjusted R2 is greater than in the log-linear model: .54 compared
to .39, respectively. The F -statistic in the linear regression model is
58.61, which is larger than the F -statistic of 32.52 in the log-linear regression model. Taken together these three measures suggest that the
linear regression model fits the data better than the log-linear model.
Each of the three variables in the linear regression model is statistically significant; in absolute value the t-statistics are greater than two.
In contrast, only two of the three variables are statistically significant
1

hw3 hints.m shows a way to do this using MATLAB.

2 points for the right

linear estimation,
2 points for the right
log-linear
estimation,
4 points for basing
the decision on the
F-stat, the t-stats
and the R-square (3
points for two of the
three, 2 points for
only one
argument).

this should be 6.52 - 1.61*3.1+....

in the log-linear model; the intercept is not statistically significant
since the t-statistic is less than two in absolute value. At P = $3.10
and A = $100, milk consumption is 2.029 million gallons per week
(Qdmilk = 6.521.61 3.10 + .005 100 = 2.029).

3 points

2 points

If this is only rounded to 0.0047, result is 1.985

23. The owner of a small chain of gasoline stations in a large Midwestern

town read an article in a trade publication stating that the own-price
elasticity of demand for gasoline in the United States is 0.2. Because
of this highly inelastic demand in the United States, he is thinking
about raising prices to increase revenues and profits. Do you recommend this strategy based on the information he has obtained? Explain.
The owner is confusing the demand for gasoline for the entire U.S.
with demand for the gasoline for individual gasoline stations. There
are not a great number of substitutes for gasoline, but in large towns
there are usually a very high number of substitutes for gasoline from an
individual station. In order to make an informed decision, the owner
needs to know the own price elasticity of demand for gasoline from
his stations. Since gas prices are posted on big billboards, and gas
stations in cities are generally close together, demand for gas from a
small group of individual stations tends to be fairly elastic.

1 point

2 points

Excel output

SUMMARY OUTPUT
Regression Statistics
Multiple R
0.739723793
R Square
0.547191289
Adjusted R Square
0.537855027
Standard Error
1.063235986
Observations
100
ANOVA
df
Regression
Residual
Total

Intercept
P
A

2
97
99

SS
132.5120801
109.6556639
242.167744

MS
66.25604004
1.130470762

Coefficients Standard Error

t Stat
6.519839045
0.823090822 7.921166011
-1.614382372
0.151475863 -10.65768723
0.004664382
0.001575003 2.961506969

F
Significance F
58.60924693
2.05053E-17

P-value
3.94495E-12
5.12726E-18
0.003849344

Lower 95%
Upper 95%
4.886231606 8.153446484
-1.915020029 -1.313744714
0.001538437 0.007790326

SUMMARY OUTPUT
Regression Statistics
Multiple R
0.633559439
R Square
0.401397563
Adjusted R Square
0.389055245
Standard Error
0.586876073
Observations
100
ANOVA
df
Regression
Residual
Total

Intercept
lnP
lnA

2
97
99

SS
22.40272212
33.40908188
55.811804

MS
11.20136106
0.344423524

Coefficients Standard Error

t Stat
-1.98867503
2.243299214 -0.886495666
-2.16951336
0.276091563 -7.857948763
0.910658367
0.370342406 2.458963249

F
Significance F
32.52205574
1.55316E-11

P-value
0.377543082
5.37032E-12
0.015705726

Lower 95%
Upper 95%
-6.441002994 2.463652934
-2.717478687 -1.621548032
0.175631206 1.645685527

Possible MATLAB code for regression

In the following output, the left column is output from the linear model,
and the right column is output for the loglinear model.
[num,txt,raw] = xlsread(hw3_q20.xls);
linear = num(:,1:3);
loglinear = num(:,5:7);
[n_row,n_columns] = size(linear);
X_lin = linear(:,2:3);
y_lin = linear(:,1);
X_log = loglinear(:,2:3);
y_log = loglinear(:,1);
stats_lin = regstats(y_lin,X_lin,linear);
stats_log= regstats(y_log,X_log,linear);
betas = [stats_lin.beta,stats_log.beta]
tstats = [stats_lin.tstat.t,stats_log.tstat.t]
pvals= [stats_lin.tstat.pval,stats_log.tstat.pval]
Rsquares = [stats_lin.rsquare,stats_log.rsquare]
adjRsquares = [stats_lin.adjrsquare,stats_log.adjrsquare]
Fstats = [stats_lin.fstat.f,stats_log.fstat.f]
Fstat_ps = [stats_lin.fstat.pval,stats_log.fstat.pval]
betas =
6.5198
-1.9887
-1.6144
-2.1695
0.0047
0.9107
tstats =
7.9212
-0.8865
-10.6577
-7.8579
2.9615
2.4590
pvals =
0.0000
0.3775
0.0000
0.0000
0.0038
0.0157
Rsquares =
0.5472
0.4014
adjRsquares =
0.5379
0.3891
Fstats =
58.6092
32.5221
Fstat_ps =
1.0e-10 *
0.0000
0.1553