ECONOMETRICS
Chapter 1
Essentials of Econometrics (4th ED), Gujarati, D.N. & Potter, D.
WHAT IS ECONOMETRICS? “Econometrics?
Isn’t that difficult?”
Literally speaking, the word Econometrics means measurement in economics.
Many alternative pinpoint definitions of econometrics are possible (see Gujarati,
p. 1)
Econometrics is an application of statistics and mathematics
… aimed at identifying and quantifying economic relationships between two
sets of variables
(1) the dependent variables and
(2) the independent variables.
example:
economist: “If the government increases tobacco excise tax, consumers will cut
down on their tobacco consumption.”
econometrician: “If the government increases tobacco excise tax by 20%,
consumers will reduce their tobacco consumption by 1%.”
…WHAT IS ECONOMETRICS?
three basic types of econometric questions
1. descriptive
How much do men and women earn annually on average in the BD?
How long do recessions typically last?
How does medical insurance coverage vary with income?
2. forecasting
What will the global temperature be in 2040?
How long will the recession last this year?
What will the stock price of Google be in April, 2022?
3. causal (or structural)
If the BB lowers interest rates today, what will happen to inflation
tomorrow?
What is the effect of political campaign expenditures on voting outcomes?
Will spending a lot of money on highway construction get us out of the
recession?
note the cause and effect elements in the previous questions
STEPS IN AN ECONOMETRIC STUDY (EMPIRICAL ECONOMIC ANALYSIS)
Statement of theory or hypotheses
Collection of data
Specification of the mathematical model
Specification of the statistical or econometric
model
Estimation of the parameters of the chosen
econometric model
Checking for model adequacy
Tests of hypotheses derived from the model
Use the model for forecasting or prediction
STATEMENT OF THEORY OR HYPOTHESES
In any econometric study, an econometrician first tries
to find out what economic theory says about the
relationship he wants to model.
According to the law of demand, when the price of a
commodity increases (and other things are held
constant), buyers tend to buy less of the commodity
and, vice versa.
COLLECTING DATA
Before turning to the estimation of the regression model
(finding the values of the model parameters), we must first
obtain the relevant data on quantity demanded (Q) and
price (P) of the commodity whose demand function we are
studying.
Time Series
Cross-sectional &
Pooled cross sections
Panel (longitudinal) data
TIME SERIES
• observations on
economic variables
over time
• stock prices, money
supply, CPI, GDP,
annual homicide rates,
automobile sales
• frequencies: daily,
weekly, monthly,
quarterly, annually
• unlike cross-sectional
data, ordering is
important here!
CROSS-SECTIONAL
• 1 observation = information
about 1 cross-sectional unit
• cross-sectional units:
individuals, households, firms,
cities, states
• data taken at a given point in
time
POOLED CROSS SECTIONS
• both cross-sectional and
time-series features
• data collected in multiple
points in time
• ordering is not crucial,
year is recorded as an
additional variable
• often used to evaluate
the effect of a policy
change
• collect data before and
after the policy change
and see how the
relationship between the
variables changes
PANEL (OR LONGITUDINAL) DATA
Different years
Same • several cross-sectional
city
units, a time series for
each unit (time series
with equal length)
• unlike with pooled
cross sections, the
same units are
measured over
time
• more difficult /costly to
obtain the data
• have several
advantages over
(pooled) cross sections
SPECIFICATION OF THE MATHEMATICAL MODEL
The law of demand postulates an inverse relationship between
price and quantity demanded, it does not indicate the precise
form of the relationship.
[1]
In Eq. 1 the variable appearing on the left side of the equality sign is called
the
dependent variable and the variable(s) on the right side is called the
1
independent, or explanatory, variable(s).
intercept it gives the value of Q when P is zero.
2
slope which measures the rate of change in Q for
per unit change in P. 2
If the
2 law
0 of demand
1 0 holds, we would expect to be
negative , that is, and (who wouldn’t
demand a good when its price is zero).
SPECIFICATION OF THE STATISTICAL OR ECONOMIC
MODEL
P Q
0 78 90
1 70 80
70
2 69 60
50
3 63
Quantity
40
30
4 60 20
10
5 58 0
0 1 2 3 4 5 6
Price
SPECIFICATION OF THE STATISTICAL OR ECONOMIC
MODEL
Observed relationship between P and Q becomes inexact.
We need to make change in the mathematical model to
allow for the influence of all other variables affecting Q.
Let us denote the effect of all other factors on Q
except P as ‘u’ and re-write the demand function as
follows:
[2]
‘u’ is called as a random or stochastic (in statistical
lingo) error term. ‘u’ captures all those forces (besides
price) that might affect Q but not explicitly introduced in
the model
ESTIMATION OF THE PARAMETERS OF THE CHOSEN ECONOMETRIC MODEL
How we do find the numerical values (known as
estimates) of and ? We shall start knowing this a couple of
classes later
[3]
Note that we have put a hat “” on Q to remind us that EQ [3]
is the estimated demand function. As equation EQ [3] shows,
the estimated values of is 76 and that of is 3.9
THE FITTED LINE
Actual
Estimated
CHECKING MODEL ADEQUACY
Demand is affected by several factors.
Q 1 2 P 3 I u
Q 1 2 P u
Which model do we choose? First one is better because
it includes income of the consumer.
But can we (or should we) choose all factors affecting
demand?
Model chosen should be a reasonable representation of
reality.
TEST OF HYPOTHESIS/HYPOTHESES DERIVED FROM THE MODELS
The law of demand and thus the demand
function indicates that the coefficient of the
price variable is expected to be negative.
We need to look whether the observed value of the
2 , is negative or not. In statistical lingo,
coefficient,
we need to test the hypothesis 2that
0
Thistest provides us the statistical/empirical
evidence on the law of demand (the testing
procedure will be discussed later).
FORECASTING OR PREDICTION
Suppose the manufacturer wants to know
what the quantity demanded be if he were to
charge TK. 4.50, a price not shown in the
Table. Plug the value of TK. 4.50 for P on the right-
hand side of EQ [3]
EQ[4]
That is, the forecasted value of Q is about 59
units if the price were TK. 4.50.
USE THE MODEL FOR POLICY OR CONTROL PURPOSES
We have estimated the demand function. Now, suppose the
company management under consideration decided to
produce 95 units of the product maximum per year
given the man, machine and money.
Now, if the manufacturer wants to know what optimum price
of the commodity that would guaranty the sale of all the 95
units of the products, we can use the estimated demand
function in EQ [4] to answer his question as follows:
…USE THE MODEL FOR POLICY OR CONTROL PURPOSES
That is, if the price were 4.88 TK/unit the manufacturer can sell
the whole lot of production.
As these calculations suggest, an estimated
regression/econometric model can be used for control, or
policy purposes.
By appropriate policy mix the producer can manipulate the
control variable P to produce desired level of the target
variable Q.
