11/13/2024
Econometrics Application of Econometrics
• The most common application of econometrics is the
Econometrics is based upon statistical methods for forecasting of important (macroeconomic) variables
estimating economic relationships, testing economic such as interest rates, inflation rates, and Gross
theories, and evaluating and implementing government Domestic Product (GDP).
and business policy.
• Econometric methods can be applied to any branch
of applied economics to test economic theories or to
test relationships among variables that have
importance for business decisions or policy analysis.
3 4
Experimental and non-experimental data
Why evolved Econometrics?
• What is non-experimental data?
• Econometrics evolved as a separate discipline from Non-experimental data are not collected through
mathematical statistics because Econometrics controlled experiments on individuals, firms, or other
focuses on the problems inherent in collecting and units.
analyzing non-experimental data. Non-experimental data often called observational data or
retrospective data, because the researcher is a passive
collector of the data.
• Experimental data?
Experimental data are often collected in laboratory
environments in the natural sciences but they are much
difficult to obtain in social sciences.
5 6
11/13/2024
Steps in empirical analysis
• Suppose we want to study return to education. The
• Empirical analysis uses data to test a theory or to functional relationship may be as follows:
estimate a relationship.
𝑤𝑎𝑔𝑒 = 𝑓(𝑒𝑑𝑢𝑐, 𝑒𝑥𝑝𝑒𝑟, 𝑡𝑟𝑎𝑖𝑛𝑖𝑛𝑔)
• For testing economic theories, a formal economic
model is constructed. • The choice of variables is determined by economic
An economic model consists of mathematical equations theory as well as data considerations.
that describe various relationships. Formal economic • After we specify an economic model, we need to
modeling is sometimes the starting point for empirical
turn it into an Econometric model.
analysis.
But it is also common to use economic theory less formally • An econometric model for wage may be
and rely entirely on intuition or economic reasoning.
𝑤𝑎𝑔𝑒 = 𝛽 + 𝛽 𝑒𝑑𝑢𝑐 + 𝛽 𝑒𝑥𝑝𝑒𝑟 + 𝛽 𝑡𝑟𝑎𝑖𝑛𝑖𝑛𝑔 + 𝑢
7 8
• The term 𝑢 contains unobserved factors, which is • Once an econometric model has been specified,
called error term or disturbance term. various hypothesis of interest can be started.
• Dealing with this disturbance term is the most • Empirical analysis requires data. Data should be
important component of any econometric analysis. collected following random sampling technique.
• In our example, 𝑢 contains • After data collection, econometric model needs to be
innate ability, estimated by econometric method.
quality of education,
• After estimation, we can formally test the set
family background or other.
hypothesis.
• Also, sometimes econometric model is used to make
predictions.
9 10
11/13/2024
The structure of Economic data
Economic data sets come in a variety of types.
1. Cross-sectional data Obs. No. Wage Education Experience Gender Marital status
($) (Years) (years)
• A cross-sectional data set consists of a sample of 1 3.1 11 2 1 (male) 0 (unmarried)
individuals, households, firms, cities, countries, or a 2 3.2 12 10 1 (male) 1 (married)
variety of other units, taken at a given point in time. 3 3.0 8 7 0 (female) 0 (unmarried)
• An important feature of cross-sectional data is that
we can often assume that they have been obtained
by random sampling from the population.
11 12
2. Time series data
Obs. No. Year Minimum Unemployment GNP
• A time series data set consists of observations on a wage
variable or several variables over time. 1 1950 0.20 15.4 878.7
• Example: Stock price, sales, consumer price index, 2 1951 0.21 16.0 925.0
GDP, money supply 3 1952 0.23 14.8 1015.9
• A key feature of time series data is that time series
observations can rarely be independent across time.
13 14
11/13/2024
4. Panel or longitudinal data
3. Pooled cross sections
A panel data or longitudinal data set consists of a time
• Pooled cross section data can be formed by series for each cross-sectional member in the data set.
combining two or more cross sectional data. Case Individual Time Wage Education Experience Gender Marital status
Suppose that two cross-sectional household surveys (year) ($) (years)
are taken place in Bangladesh, one in 2010 and 1 1 1990 3.1 11 2 1 (male) 0 (unmarried)
another one in 2015. In order to increase the sample
2 1 1991 3.3 11 3 1 (male) 0 (unmarried)
size, we can form a pooled cross section by combining
these two years. 3 2 1990 3.2 12 10 1 (male) 1 (married)
4 2 1991 3.4 12 11 1 (male) 1 (married)
• One key feature to remember is that the cross- 5 3 1990 3.0 8 7 0 (female) 0 (unmarried)
sectional observations may be different over time. 6 3 1991 3.1 8 8 0 (female) 0 (unmarried)
15 16
