Subject: ECONOMETRIC PROBLEMS REVISITED:

MODEL MIS-SPECIFICATION.
CLASS ACTIVITIES:

Tutorial Exercises:
1. Question for Discussion No. 2
You have been hired by “Indo”, the new Indonesian automobile manufacturer, to build
a model of US car prices in order to help the company undercut US prices. Allowing
Friedmaniac zeal to overwhelm any patriotic urges, you build the following model of the
price of 95 different American–made 2004 US sedans (std errors in parenthesis):

Model A: ̂P𝑖 = 3.0 + 0.24 Wi + 1.1Ti + 5.8 Ci + 0.19Li

(0.07) (0.4) (2.7) (0.20)
tC - 1.697 1.697 1.697 1.697
𝑅̅2 = 0.91; F=97

where
Pi = the list price of the ith car (thousands of dollars)
Wi = the weight of the ith car (hundreds of pounds)
Ti = a dummy equal to 1 if the ith car has an automatic transmission, 0 otherwise
Ci = a dummy equal to 1 if the ith car has luxury option, 0 otherwise
Li = the size of engine of the ith car

a) Your firm‘s pricing expert hypotheses positive signs (one sided) for all the slope
coefficients in Model A. Test the expectations at the 10-percent significance level.
Evaluate the hypothesis of the equation.
b) What econometric issues appear to exist in Model A? In particular, does the size
of the coefficient of C cause any concern? Why? If required, how can we reduce
coefficient value (convert fraction) in regression model?
c) You decide to test the possibility that L is an irrelevant variable by dropping it and
rerunning the equation, obtaining the following Model T equation. Which model
do you prefer? Why? Any suggestion about model specification?
Model T: 𝑃̂ = 18 + 0.29 Wi + 1.2Ti + 5.9 Ci
(0.06) (0.30) ( 2.6)
𝑅̅2 = 0.94; F=120
d) What is the reasonable minimum sample for correlation/ OLS analysis?
e) Given, the mean value of disturbance in Model A & T are 5.25 & 0. Provide your
comment.
f) What is an economic model? Which important issues are to be considered for
appropriate economic model specification?

1
2. Question for Discussion No. 4: Assume that you have been hired by the surgeon of the
United States to study the determinants of smoking behavior and that you estimate the following
cross-sectional model based on data for all 50 states (standard errors in parentheses)
Ĉ𝑖 = 10 – 9.0 Ei + 1.0Ii - 0.04 Ti - 3.8 Vi + 1.6 Ri
(3.0) (1.0) (0.04) (1.0) (0.6)
t = ? ? ? ? ?
̅𝑅2 = 0.40 N = 50 (states)
Ci = the number of cigarettes consumed per day per person in the ith state
Ei = the average years of education for persons over 21 in the ith state
Ii = the average income in the ith state
Ti = the tax per package of cigarettes in the ith state
Vi = the number of video ads against smoking aired on the three major networks in the
ith state
Ri = the number of radio ads against smoking aired on the five largest radio networks in the ith
state
a) Hypothesize expected signs. Comment on final decision if the coefficients are
significant / insignificant by developing and testing (at the 10-percent level)
hypotheses for the coefficients of the variables in this equation. Do u have any
unexpected sign?
b) Do you appear to have any irrelevant variables? Do you appear to have any
omitted variables? Explain your answer.
c) Let’s assume that your answer to part b was yes to both. Which problem is more
important to solve first-irrelevant variables or omitted variables? Why?
d) One of the purposes of running the equation was to estimate effectiveness of
antismoking advertising on television and radio. What is your conclusion?
e) Interpret result of Ti. Is it logical for price inelastic product demand? Provide
comment. Given, tax increases by 20% but demand decreases by only 1%,
estimate ‘Elasticity of Demand’.
f) If you drop Ti, you find the same R square. Is it acceptable?
g) Based on economic model specification, do you recommend an additional
variable to check & resolve the issue of omitted variable?
h) Interpret the constant term (alpha) of the model.
2. Question for Discussion No. 6
You have been retained by the “Expressive Expresso” company to help them decide where to build their
next “Expressive Expresso” store. You decide to run a regression on the sales of the 30 existing
“Expressive Expresso” stores as a function of the characteristics of the locations they are in and then use
the equation to predict the sales at the various locations you are considering for the next store. You end
up estimating (standard errors in parentheses)
̂Y𝑖 = 30 + 0.1X1i + 0.01X2i + 10.0X3i + 3.0X4i
(0.02) (0.01) (1.0) (1.0)
Calculated t = ? ? ? ? Where: t critical is 2.45
Yi = average daily sales (in hundreds of dollars) of the ith store
X1i = the number of cars that pass the ith location per hour
X2i = average income in the area of the ith store
X3i = the number of tables in the ith store
X4i = the number of competing shops in the area of the ith store
a) Hypothesize expected signs, and write one sided & two sided hypothesis, and
test the significance at the 1-percent level for each of the coefficients.
2
 b) What problems appear to exist in the equation? What evidence of these problems
do you have from the model?
c) What suggestions would you make for a possible second run of this admittedly
hypothetical equation? (Hint: recommend the inclusion of a potentially omitted
variable.)

3. Question for Discussion No. 8

For each of the following situations, determine the expected bias (positive/negative)
introduced by omitting a variable on regressand:
a) In an equation for the demand for peanut butter, the impact on the coefficient of
disposable income of omitting the price of peanut butter variable. (Hint: Start by
hypothesizing signs.)
b) In an earning equation for workers, the impact on the coefficient of experience of
omitting the variable for age.
c) In a production function for airplanes, the impact on the coefficient of labour of
omitting the capital variable.
d) In an equation for daily attendance at outdoor concerts, the impact on the
coefficient of the weekend dummy variable (1=weekend) of omitting a variable
that measures the probability of precipitation at concert time.
4. Question for Discussion No. 9
Most of the examples so far have been demand-side equations or production functions,
but economists often also have to quantify supply-side equations that are not true
production functions. These equations attempt to explain the production of a product (for
example, Brazilian coffee) as a function of the price of the product and various other
attributes of the market that might have an impact on the total output of growers.
a) What sign would you expect the coefficient of price to have in a supply side
equation? Why?
b) What other variables can you think of that might be important in a supply-side
equation? Why?
c) Identify the difference between Lag variable and Lead variable in a regression
equation. What is lead–lag effect?
d) What are the important issues that must be considered for evaluating appropriate
model specification of a regression model?

5. Question for Discussion No. 10

Regressing GNP on the various definitions of money supply, we obtain the results shown
in following Table:

3
The monetarists or quantity theorists maintain that nominal income (i.e., nominal
GNP) is largely determined by changes in the quantity or the stock of money,
although there is no consensus as to right definition of money. Given the results in
the preceding table, consider these questions:
a) Which definition of money seems to be best fitted to nominal GNP?
b) Since the r2 terms are uniformly high, does this fact mean that our choice for
definition of money does not matter?
c) If the Fed wants to control the money supply, which one of these money measures
is a better target for that purpose? Can you tell from the regression results?
d) Based on relevant literature, propose an extended economic regression model
showing ‘GNP of X country’.
e) Distinguish between Narrow Money (M1) & Broad Money (M3).
Among the above four equations, which types of money (M1 to M4) are most
liquid, least liquid & the maximum? Which one is ‘High-Powered Money’ of the
economy?