Econometrics and

Economic Data
Ani Katchova

© 2020 by Ani Katchova. All rights reserved.

Outline
• Define econometrics, economic models, and econometric models
• Types of economic data (cross-sectional, time series, pooled cross
sections and panel data)
• Causation versus correlation/relationships in econometrics

2
Economic models
• Econometric analysis
• Start with economic model
• Estimate an econometric model
• Economic models
• Analyze micro- or macro-variables
• Estimate relationships between economic variables
• Examples: demand and supply models, asset pricing models

3
Econometrics - definition and goals
• Econometrics definition
• Econometrics - use statistical methods to analyze economic data
• Econometric models – statistical models used in economics. They
specify the relationship between variables.
• Use econometric models to:
• Estimate relationships between economic variables
• Test economic hypotheses and theories
• Evaluate effectiveness of a new policy
• Forecast economic variables

4
Example of economic model
Economic model describing wage determinants for workers

𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 = 𝑓𝑓 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒, 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒, 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡

More generally,
𝑦𝑦 = 𝑓𝑓(𝑥𝑥1 , 𝑥𝑥2 , 𝑥𝑥3 )

f is an unknown function relating the variables.

5
Example of econometric model
Econometric model describing wage determinants for workers

𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 = 𝛽𝛽0 + 𝛽𝛽1 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 + 𝛽𝛽2 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 + 𝛽𝛽3 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + 𝑢𝑢

The functional form is specified (multiple regression model)

More generally,
𝑦𝑦 = 𝛽𝛽0 + 𝛽𝛽1 𝑥𝑥1 + 𝛽𝛽2 𝑥𝑥2 + 𝛽𝛽3 𝑥𝑥3 + 𝑢𝑢
y is the dependent variable
𝑥𝑥1 , 𝑥𝑥2 , 𝑥𝑥3 are independent variables,
𝛽𝛽0 , 𝛽𝛽1 , 𝛽𝛽2 , 𝛽𝛽3 are coefficients to be estimated
u is an error term representing unobservable factors affecting y
6
Goals of econometric model
𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 = 𝛽𝛽0 + 𝛽𝛽1 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 + 𝛽𝛽2 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 + 𝛽𝛽3 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + 𝑢𝑢

We can use this econometric model to test hypotheses

• What is the effect of education on wage represented by 𝛽𝛽1 ?
• Is 𝛽𝛽1 significantly different from zero? What is the magnitude of 𝛽𝛽1 ?

7
Types of economic data
• Econometric models use economic data
• Types of economic data
• Cross-sectional data
• Time series data
• Pooled cross sections
• Panel/longitudinal data
• Which econometric models we apply depends on the type of data
used

8
Cross-sectional data
• Data for people, households, businesses, countries, cities, etc.
• Data are at a given point of time/during a given period – no time dimension
• Typically denote the individual by i
• Widely used in microeconomics

9
Example: cross-sectional data on wages
Wage yi expersq
Hourly Female x4i lwage (experience
wage Educ x1i Exper x2i Tenure x3i Female=1 Married x5i (log wage) squared)
3.1 11 2 0 1 0 1.131402 4
3.24 12 22 2 1 1 1.175573 484
3 11 2 0 0 0 1.098612 4
6 8 44 28 0 1 1.791759 1936
5.3 12 7 2 0 1 1.667707 49
8.75 16 9 8 0 1 2.169054 81
11.25 18 15 7 0 0 2.420368 225
5 12 5 3 1 0 1.609438 25
Note i, yi, x1i, etc. Every observation/row is for person i.
10
Time series data
• Data can be macroeconomic, financial, etc.
• Examples: stock and bond prices, GDP, growth rates
• Observations are over time
• The time dimension can be annual, monthly, daily, etc.
• Time series may have trend (e.g. rising values over time), seasonality (e.g.
higher values in a given month), and cycles (e.g. every 3-5 years).
• Observations may be serially correlated (errors are correlated from one
period to the next).

11
Time series data on minimum wages
Prunemp x3t
Avgmin yt Avgcov x1t Prgnp x2t Umemployme
Year t Avg min wage Avg coverage GDP nt rate
1950 0.198 0.201 878.7 15.4
1951 0.209 0.207 925 16
1952 0.225 0.226 1015.9 14.8
1953 0.311 0.231 1081.3 14.5
1954 0.313 0.224 1104.4 15.3
1955 0.369 0.236 1138.5 13.2
1956 0.447 0.245 1185.1 13.3
1957 0.488 0.244 1221.8 12.8

Note t, yt, x1t, etc. Every observation/row is at time t.

12
Pooled cross sections
• Example: house prices in two periods but not the same houses are sold.
• Two or more cross sections
• Cross sections are drawn independently of each other
• Used to estimate effect of new policy, for example effect of new tax on house
prices

13
Pooled cross-sectional data on house prices
Year t Price yt Rooms x1t Baths x2t lprice y81
1978 60000 7 1 11.0021 0
1978 40000 6 2 10.59663 0
1978 34000 6 1 10.43412 0
1978 63900 5 1 11.06507 0
1981 49000 6 1 10.79958 1
1981 52000 5 1 10.859 1
1981 68000 6 2 11.12726 1
1981 54000 6 1 10.89674 1

Note t, yt, x1t, etc. Every observation is at time t. Before and after period.
14
Panel data or longitudinal data
• Example: employment data across individuals and over time
• Same cross-sectional units over time
• Have both cross-sectional i and time series t dimensions

15
Panel data on wages
Person id Year lwage Exper Educ Hours
i t yit x1it x2it x3it
13 1980 1.19754 1 14 2672
13 1981 1.85306 2 14 2320
13 1982 1.344462 3 14 2940
17 1980 1.675962 4 13 2484
17 1981 1.518398 5 13 2804
17 1982 1.559191 6 13 2530
18 1980 1.515963 4 12 2332
18 1981 1.735379 5 12 2116
18 1982 1.631744 6 12 2500

Note i, t, yit, x1it, etc. Every observation is for unit i at time t.

16
Causation versus correlation
• Causation: An additional year of education causes wages to increase
by a given amount, all else equal.
• Correlation: An additional year of education is associated with higher
wages.
• For most economic studies, a causation cannot be determined. It is
only correlation.

17
Example on causation vs correlation
• Causal effect of fertilizer on crop yield
• Will crop yield increase signficantly if fertilizer is applied?
• Implicit assumption: all other factors that influence crop yield such as quality of land
are held fixed
• Experiment
• Choose several plots of land; randomly assign different amounts of fertilizer to the
different plots and compare crop yield
• Valid experiment because fertilizer that is applied is not related to other factors
influencing crop yields
• Observational study
• Plots that have different levels of fertilizer and different yields. If there is a positive
correlation (plots with more fertilizer have more yields), does that mean that
applying more fertilizer will result in a higher yield?
• Hint: this is a correlation, but not causation.

18
Example of causation vs correlation
• Effect of education on wages
• If a person at random is given another year of education by how much will his/her
wage increase?
• Implicit assumption: all other factors that influence wages such as experience and
ability are held fixed
• Experiment
• Choose some people at random and assign them to get more education – not
feasible.
• Problem without random assignment: amount of education is related to other
factors that influence wages (such as intelligence)
• Observational study
• Data on people with wages and education. Interpretation: people with higher
education have higher wages (correlation). Interpretation is not: if a person gets
additional education, he/she will get higher wages (causation).

19
Econometrics and economic data – review
questions
1. Define econometrics, economic model, and econometric model.
What are some goals of econometric analysis?
2. What are the main types of economic data in econometrics? Give
examples for each type of economic data.
3. Which data/models have i or t dimension? How many dependent
and independent variables do econometric models have?
4. Does a regression imply that there is a causal relationship or
correlation between the dependent and independent variables?
Give examples.

20