1 2

What is econometrics?
I Literal meaning: economic measurement: “econo-metrics”.
INTRODUCTION I But the scope of econometrics is much wider.
I Two popular definitions of econometrics:
I “Econometrics may be defined as the social science in which
Hüseyin Taştan1
the tools of economic theory, mathematics, and statistical
1 Yıldız
Technical University
inference are applied to the analysis of economic phenomena”.
Department of Economics (A.S. Goldberger, 1964).
These presentation notes are based on I “...econometrics may be defined as the quantitative analysis
Introductory Econometrics: A Modern Approach (2nd ed.) of actual economic phenomena based on the concurrent
by J. Wooldridge.
development of theory and observation, related by appropriate
methods of science”. (P. Samuelson)
14 Ekim 2012

3 4

Scope of Econometrics Components of Econometrics

I Developing statistical methods for the estimation of economic I Econometric inputs:
relationships, I Economic Theory
I Mathematics
I Testing economic theories and hypothesis, I Statistical Theory
I Evaluating and applying economic policies, I Data
I Forecasting, I Computers (CPU power)
I Interpretation
I Collecting and analyzing nonexperimental or observational
I Econometric outputs:
data.
I Estimation - Measurement
I Inference - Hypothesis testing
I Forecasting - Prediction
I Evaluation
5 6

Why Do We Need Econometrics? Classical Methodology in Econometrics

I We learned statistical methods so why do we need a separate I Formulation of theory or hypothesis,
discipline? I Specification of economic (mathematical) model,
I The reason is as follows: econometrics focuses on the analysis I Specification of econometric model,
of nonexperimental economic data. I Collecting data,
I Nonexperimental data (or observational data) are not I Estimation of parameters,
obtained through controlled experiments on economic agents
(consumers, firms, households, sectors, countries, etc.)
I Hypothesis tests,
I Experimental data are collected in laboratory environments
I Forecasting/Prediction),
in natural sciences. I Evaluation of results for policy analysis or decision making.
I Although some social experiments can be devised it is usually (Gujarati, p.3)
impossible to conduct economic experiments.
I Unlike statistical methods employed in natural sciences,
econometrics develops special methods to handle
nonexperimental data.

7 8

ECONOMIC MODEL ECONOMIC MODEL vs. ECONOMETRIC MODEL

Example 1 - Economic Model of Crime Economic Model Example 2 - Job Training and Worker
Productivity
y = f (x1 , x2 , x3 , x4 , x5 , x6 , x7 ),
wage = f (educ, exper, training),
f functional form (not yet specified)
Description of variables wage: hourly wage (in dollars)
educ: level of education (in years)
I y: hours spent in criminal activities, exper : level of workforce experience (in years)
I x1 : “earnings” for an hour spent in criminal activity, training : weeks spent in job training.
I x2 : hourly wage in legal employment, Econometric Model: f Linear specification
I x3 : other income,
I x4 : probability of getting caught, wage = β0 + β1 educ + β2 exper + β3 training + u
I x5 : probability of being convicted if caught,
I x6 : expected sentence if convicted,
I x7 : age
9 10

ECONOMETRIC MODEL: Linear Specification ECONOMETRIC MODEL: Linear Specification

Econometric Model Example 2 - Job Training and Worker Econometric Model Example 2 - Job Training and Worker
Productivity Productivity

wage = β0 + β1 educ + β2 exper + β3 training + u wage = β0 + β1 educ + β2 exper + β3 training + u

Components of econometric model: Components of econometric model:

I u: random error term or disturbance term I β0 , β1 , β2 , β3 : parameters of the econometric model
I Random error term u contains influence of factors that are I These are unknown constants.
not included in the model. It also contains unobserved factors I They describe the directions and strengths of the relationship
such as innate ability or family background. between wage and factors affecting wage included in the
I No matter how comprehensive the specified model there will model.
always factors that cannot be included in the econometric I For example, we may be interested in testing H0 : β3 = 0
model. We can never eliminate u entirely. which says that job training has no effect on wage.

11 12

ECONOMETRIC DATA TYPES ECONOMETRIC DATA TYPES

I Cross-sectional data Cross-sectional data:
I Time series data I consists of a sample of individuals, households, firms, cities,
I Pooled cross-section states, countries, or a variety of other units, taken at a given
point in time
I Panel data (longitudinal data)
I Significant feature: random sampling from a target population
I Generally obtained through official records of individual units,
surveys, questionnaires (data collection instrument that
contains a series of questions designed for a specific purpose)
I For example, household income, consumption and
employment surveys conducted by the Turkish Statistical
Institute (TUIK/TURKSTAT)
13 14

Cross-sectional data example: Wage Data (GRETL data ECONOMETRIC DATA TYPES
set: wage1.gdt) Time series data:
A Cross-sectional data set on wages and individual I consists of observations on a variable or several variables over
characteristics time.
Obs. No wage educ exper female married I Chronological ordering
1 3.10 11 2 1 0 I Frequency of time series data: hour, day, week, month, year
2 3.24 12 22 1 1
3 3.00 11 2 0 0 I Time length between observations is generally equal
4 6.00 8 44 0 1 I Examples of time series data include stock prices, money
5 5.30 12 7 0 1 supply, consumer price index, gross domestic product, annual
6 8.75 16 9 0 1 homicide rates, and automobile sales figures.
. . . . . .
. . . . . .
. . . . . .
524 4.67 15 13 0 1
525 11.56 16 5 0 1
526 3.50 14 5 1 0

16
A Time Series Data Example: GRETL: prminwage.gdt
ECONOMETRIC DATA TYPES
Pooled cross-section:
I consists of cross-sectional data sets that are observed in
different time periods and combined together
I At each time period (e.g., year) a different random sample is
chosen from population
I Individual units are not the same
I For example if we choose a random sample 400 firms in 2002
and choose another sample in 2010 and combine these
cross-sectional data sets we obtain a pooled cross-section data
set.
I Cross-sectional observations are pooled together over time.
 18
A Pooled Cross-sectional Data Example
ECONOMETRIC DATA TYPES
Panel Data (longitudinal data):
I consists of a time series for each cross-sectional member in
the data set.
I The same cross-sectional units (firms, households, etc.) are
followed over time.
I For example: wage, education, and employment history for a
set of individuals followed over a ten-year period.
I Another example: cross-country data set for a 20 year period
containing life expectancy, income inequality, real GDP per
capita and other country characteristics.

20
A Panel Data Example
Causality and the Notion of Ceteris Paribus
I In testing economic theory usually our goal is to infer that one
variable has a causal effect on another variable.
I Correlation may be suggestive but cannot be used to infer
causality.
I Fundamental notion: Ceteris paribus: “other relevant factors
being equal”
I Or “holding all other factors fixed”
I Most economic questions are ceteris paribus by nature.
I For example, in analyzing consumer demand, we are interested
in knowing the effect of changing the price of a good on its
quantity demanded, while holding all other factors (such as
income, prices of other goods, and individual tastes) fixed.
I If other factors are not held fixed, then we cannot know the
causal effect of a price change on quantity demanded.
21 22

Causality and the Notion of Ceteris Paribus Ceteris Paribus Example: Effects of Fertilizer on Crop Yield
I Therefore, the relevant question in econometric analysis is “do I Suppose the crop is wheat. We are interested in measuring the
we control sufficient number of factors?” impact of fertilizer on wheat yield (production).
I Are there other factors that are not included in the model? I Obviously there are several factors that affect the production
I Can we say that other components are held fixed? of wheat such as rainfall, quality of soil and presence of
parasites.
I In most serious applications the number of factors is immense
so the isolation of the effect of any particular variable may I We need to control these factors in order to determine the
seem hopeless. But, if properly used, econometric methods ceteris paribus impact of fertilizers.
can help us determine ceteris paribus effects. I To do this we can devise the following experiment: divide the
land into equal pieces (such as one acre) and apply different
amounts of fertilizer to each land plot and then measure the
wheat yield.
I This gives us a cross-sectional data set where observation unit
is land plot.
I We can apply statistical methods to this data set to measure
the impact of fertilizers on crop yield.

23 24

Ceteris Paribus Example: Effects of Fertilizer on Wheat Ceteris Paribus Example: Measuring the Return to
Yield Education
I How do we know the results of this experiment can be used to I Question: How can we measure the return to education?
measure the ceteris paribus effect of fertilizer? I If a person is chosen from the population and given another
I Can we be sure that all other factors (quality of land plots for year of education, by how much will his or her wage increase?
example) are held fixed? I This is also a ceteris paribus question: all other factors are held
I It is generally very difficult to observe the quality of soil. fixed while another year of education is given to the person.
I But we can still use ceteris paribus notion I There are several factors other than education that affect
I Amounts of fertilizers should be assigned to land plots wages: experience, tenure, innate ability, gender, age, region,
independently of other plot features such as quality of plots marital status, etc.
I In other words, other characteristics of plots should be ignored
when deciding on fertilizer amounts.
25 26

Example: Measuring the Return to Education Ceteris Paribus Example: Measuring the Return to
I Similar to fertilizer example we can design the following Education
hypothetical experiment: I People choose their education levels. Thus, individual
I Social planner has the ability to assign any level of eduction characteristics will be correlated with the level of education.
to any person.
I For example, people with more innate ability tend to have
I The planner chooses a group of individuals from population
higher levels of education.
and randomly assign each person an amount of education:
some are given high school education, some are given 4-year I Workers with higher levels of education tend to have higher
college education, etc. wages.
I Subsequently the planner measures wages for each individual. I It becomes difficult to isolate the impact of education from
I If levels of education are assigned independently of other the impact of innate ability on wages.
characteristics that affect productivity (such as innate ability I How much of this effect comes from education? How much
or experience) then we can measure the impact of education from innate ability?
on wages correctly.
I Of course such an experiment is impossible to conduct.
I Even though we cannot obtain an experimental data, we can
obtain observational data set that contains information on
wages, education, experience and other personal
characteristics (e.g. from TUIK household employment
surveys)
27 28

Ceteris Paribus Example: The Effect of Law Enforcement Ceteris Paribus Example: The Effect of Law Enforcement
on City Crime Levels on City Crime Levels
I Does the presence of more police officers on the street deter I It almost impossible to find two cities identical in all respects.
crime? I But this is not necessary in econometric analysis.
I Ceteris paribus question: If a city is randomly chosen and I We just need to know if the data on crime rates and number
given, say, ten additional police officers, by how much would of police officers can be viewed as experimental.
its crime rates fall? I In most cases this is not the case, data is observational.
I Or: If two cities are the same in all respects, except that city I The size of police force is determined by city authorities who
A has ten more police officers than city B, by how much
probably take into account several other city characteristics.
would the two cities’ crime rates differ?
I The problem is a little bit more complex: Does the size of
police force affect the amount of crime or vice versa?
I The amount of crime and police force are simultaneously
determined.