ECON2280 Introductory Econometrics

Topic 1: What is Econometrics and Why Do We Need It?

Yiming Cao

The University of Hong Kong

January 20, 2025

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 1 / 65

Lecture Plan

Motivation: Why Econometrics? (15mins)

Concept of Econometric Analysis (20mins)

Overview of Topics (20mins)

Data Structure (20mins)

A Note on Causality (15mins)

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 1 / 65

 Motivation

Experiments in Scientific Discovery

Galileo’s Pendulum Experiments

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 2 / 65

 Motivation

Galileo’s Pendulum Experiments

Objective: To understand what factors affect the period of a pendulum (time it

takes to complete one full swing).

Variables: Length of the thread, weight of the bob, angle of release, etc.

Controlled experiments: Varying one factor at a time while keeping other factors
constant.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 3 / 65

 Motivation

Data Collection and Analysis

Period vs. Mass Period vs. Mass

(Length = 0.6m) (Mass = 100g)

Period unaffected by mass. Period increases with length.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 4 / 65

 Motivation

A Theoretical Model
In high school, we learned that:
s
L
T = f (L) = 2π
g

Taking the natural logarithm of both sides:

1 1
ln T = ln 2π + ln L− ln g
2 2
In this relationship: ln T is the dependent variable (Y), ln L is the independent
variable (X ), ln 2π and 21 ln g are constants (α), and 21 is the coefficient (β) of ln L.
That is:
Y = α + βX
ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 5 / 65
 Motivation

Experimental Results vs. Theoretical Model

Theoretical Model Prediction Actual Experimental Results

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 6 / 65

 Motivation

Potential Sources of Error?

Measurement error in Y: The measured period may not be exactly equal to the
true period.

Measurement error in X: The measured length may not be exactly equal to the
true length.

Other factors: The period may be affected by other factors that we did not
consider (e.g., air resistance).

Random noises: The period may be affected by random noises that we cannot
control.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 7 / 65

 Motivation

A “Statistical” Model

Accounting for the errors:

Y = α + βX + u
where u is the error term.

The error term captures all the factors that affect the dependent variable but are
not included in the model.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 8 / 65

 Motivation

Fitting the Model to Data

Fitted line: ln T = 0.5084× ln L + 0.6946

1 1
≈ × ln L + ln 2π − ln g
2 2

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 9 / 65

 Motivation

Can We Do This in Economics?

Objective: To understand the relationship between economic variables.
What is the effect of consumption on GDP?
What is the effect of education on income?
What is the effect of minimum wage on employment?

What are the challenges?

Economic variables are not easy to measure
(e.g., GDP, inflation, unemployment, etc.).
Economic variables are affected by many factors
(e.g., GDP is affected by consumption, investment, government spending, net
exports, etc.).
We cannot vary one single variable at a time while holding other variables constant
(e.g., we cannot increase consumption while holding investment constant).

Solution: Econometrics!
ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 10 / 65
 Econometric Analysis

Lecture Plan

Motivation: Why Econometrics? (15mins)

Concept of Econometric Analysis (20mins)

Overview of Topics (20mins)

Data Structure (20mins)

A Note on Causality (15mins)

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 11 / 65

 Econometric Analysis Definition

What is Econometrics?

Econometrics is the use of statistical methods to analyze economic data.

Econometricians typically analyze nonexperimental data.

Goals:
Estimating relationships between economic variables.
Testing economic theories and hypotheses.
Evaluating and implementing government and business policy.
Make predictions.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 12 / 65

 Econometric Analysis Steps

Steps in Empirical Economic Analysis

Formulate the question of interest.

Specify the economic model/theory (formal or informal)

Determine the econometric model

Collect the data

Estimate the econometric model.

Test hypotheses

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 13 / 65

 Econometric Analysis Examples

In the case of Galileo’s pendulum experiments

Question of interest: What factors affect the period of a pendulum?
Physics theory:
1 1
ln T = ln 2π− ln g + ln L
2 2
Statistical model:
ln T = α + β ln L + u
Data: Period and length of the thread.
Estimating the fitted line:

ln T = 0.6946 + 0.5084 ln L

This is a simple linear regression model.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 14 / 65
 Econometric Analysis Examples

The Economics of Crime (Becker, 1968)

Question of interest: What factors affect crime rate in a society?

Economic theory (Becker, 1968):

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 15 / 65

 Econometric Analysis Examples

The Economics of Crime (Becker, 1968)

Question of interest: What factors affect crime rate in a society?
Economics theory (Becker, 1968): y = f (x1, x2, x3, x4, x5, x6, x7)
Econometric model:

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 16 / 65

 Econometric Analysis Examples

Job Training and Worker Productivity

Question of interest: What is the effect of additional training on worker
productivity?
Economics theory:

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 17 / 65

 Econometric Analysis Examples

Job Training and Worker Productivity

Question of interest: What is the effect of additional training on worker
productivity?
Economics theory: wage = f (educ, exper , training )
Econometric model:

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 18 / 65

 Econometric Analysis Practice and

Practice and Discussion

Write down a question of interest that you have.

What is the effect of X on Y?

Write down an economic theory that you think can answer your question.
What are the factors that should be considered?

Write down an econometric model that you think can answer your question.
How can you measure the factors in your economic model?
What are the factors that you cannot measure?

Share your questions and models with your classmates around you.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 19 / 65

 Topics

Lecture Plan

Motivation: Why Econometrics? (15mins)

Concept of Econometric Analysis (20mins)

Overview of Topics (20mins)

Data Structure (20mins)

A Note on Causality (15mins)

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 20 / 65

 Topics Estimation

How to Estimate the Econometric Model?

Model: Y = α + βX + u
Data from a random sample of n observations: {Yi , Xi }ni=1

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 21 / 65

 Topics Estimation

How to Estimate an Econometric Model?

Model: Y = α + βX + u
Data from a random sample of n observations: {Yi , Xi }ni=1
Estimation: Ŷ = α̂ + β̂X

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 22 / 65

 Topics Estimation

How to Estimate an Econometric Model?

Model: Y = α + βX + u
Data from a random sample of n observations: {Yi , Xi }ni=1
Estimation: Ŷ = α̂ + β̂X
What is the “best” line that fits the data?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 23 / 65

 Topics Interpretation

How to Interpret an Econometric Model?

What if we have more than one independent variable?

Y = α + β1 X1 + β2 X2 + ... + βk Xk + u

How to interpret the coefficients?

What if the variables are not linear? e.g., ln Y = α + β ln X + u

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 24 / 65

 Topics Fitness

How to Evaluate the Goodness of Fit?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 25 / 65

 Topics Evaluation

How Do We Know if the Estimated Coefficients are “Correct”?

Recall the Galileo’s pendulum experiments:

Theoretical model:
1 1
ln T = ln 2π − ln g + ln L
2 2
Estimation:
ln T = 0.6946 + 0.5084 ln L

Is this estimation “correct”?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 26 / 65

 Topics Evaluation

How Do We Know if the Estimated Coefficients are “Correct”?

Econometric model: Y = α + βX + u

Estimation: Ŷ = α̂ + β̂X

Do we have α̂ ≈ α and β̂ ≈ β? Under what conditions?

What may lead to α̂ ̸= α and β̂ ̸= β? Solutions?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 27 / 65

 Topics Evaluation

Would the Estimated Coefficients be the Same if We Use a

Different Sample?
“True” model: Y = 1 + 2X + u
Estimates from four different samples:

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 28 / 65

 Topics Inference

How Do We Know if the Effect Is Zero Or Not?

Does X have an effect on Y?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 29 / 65

 Topics Specifications

What if the Model is Not Standard?

What if the relationship between variables is not linear?

Logarithm: Y = α + β ln X + u, ln Y = α + βX + u, etc.
Polynomial: Y = α + β1 X + γX 2 + u, etc.
Interaction: Y = α + βX1 + γX2 + δX1 X2 + u, etc.

What if the variables are not continuous?

Binary: Grade=1 if pass, Grade=0 if fail.
Ordinal: Grade=4 if A, Grade=3 if B, etc.
Categorical: Asia, Europe, America, etc.

What if the model/variables are misspecified?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 30 / 65

 Topics Data Issues

What if the Data is Not “Good”?

What if the data contains missing values?

What if the data contains outliers?

What if a variable is not observable?

What if the measurement of a variable is inaccurate?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 31 / 65

 Topics Summary

We Will Cover All These Topics in This Course!

Lecture 2 (Estimation): Finding the “best” line that fits the data.
Lecture 3 (Interpretation): Interpreting the estimated coefficients.
Lecture 4 (Fitness): Evaluating the goodness of fit.
Lectures 5–6 (Evaluation): Evaluating the correctness of the estimated coefficients.

Lecture 5 (Unbiasedness): Expected value of the estimated coefficients.

Lecture 6 (Efficiency): Variance of the estimated coefficients.
Midterm

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 32 / 65

 Topics Summary

We Will Cover All These Topics in This Course!

Lecture 7 (Inference): Testing the estimated coefficients against a hypothesis.

Lecture 8–9 (Specifications): Dealing with non-standard models and variables.
Lecture 8: Fuctional forms
Lecture 9: Categorical variables
Lecture 10: Dealing with data issues.
Lecture 11: Heteroskedasticity
Lecture 12: Time series data
Final Review

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 33 / 65

 Data Structure

Lecture Plan

Motivation: Why Econometrics? (15mins)

Concept of Econometric Analysis (20mins)

Overview of Topics (20mins)

Data Structure (20mins)

A Note on Causality (15mins)

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 34 / 65

 Data Structure

Types of Economic Data

Cross-sectional data

Time series data

Pooled cross sections

Longitudinal/panel data

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 35 / 65

 Data Structure

Cross-sectional Data

Data of individuals, households, firms, cities, states, countries, or other units of

interest at a given point of time or in a given period.

Examples: GDP of countries in 2019, unemployment rate of cities in 2019, GPA of

students in this classroom, etc.

Observations are likely independent of each other.

Violations of independence may arise.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 36 / 65

 Data Structure

Cross-sectional Data: Illustration

Table 1.1: Cross-sectional data set on wages and other characteristics

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 37 / 65

 Data Structure

Cross-sectional Data: Illustration

Table 1.2: Cross-sectional data on growth rates and country characteristics

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 38 / 65

 Data Structure

Time Series Data

Data of a variable or several variables over time.

Examples: GDP of China from 1952 to 2019, unemployment rate of the US from
1948 to 2019, etc.

Observations are typically serially correlated.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 39 / 65

 Data Structure

Time Series Data: Illustration

Table 1.3 Time series data on minimum wage, unemployment, and related data for Puerto Rico

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 40 / 65

 Data Structure

Pooled Cross Sections

Two or more cross sections are combined in one data set.

Cross sections are drawn independently of each other.

Pooled cross sections are often used to evaluate policy changes.

Example: Evaluating effect of change in property taxes on house prices.

Random sample of house prices for the year 1993.
A new random sample of house prices for the year 1995.
Compare before/after (1993: before reform, 1995: after reform).

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 41 / 65

 Data Structure

Pooled Cross Sections: Illustration

Table 1.4: Pooled cross sections on two years of housing prices

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 42 / 65

 Data Structure

Longitudinal/Panel Data

The same cross-sectional units are followed over time.

Panel data have a cross-sectional and a time series dimension.

Panel data can be used to account for time-invariant unobservables.

Panel data can be used to model lagged responses.

Example: City crime statistics; each city is observed in two years.

Time-invariant unobserved city characteristics may be modeled.
Effect of police on crime rates may exhibit time lag.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 43 / 65

 Data Structure

Longitudinal/Panel Data: Illustration

Table 1.5: Two-year panel data set on city crime statistics

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 44 / 65

 Data Structure

Practice

What type of data is used in the following studies?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 45 / 65

 Causality

Lecture Plan

Motivation: Why Econometrics? (15mins)

Concept of Econometric Analysis (20mins)

Overview of Topics (20mins)

Data Structure (20mins)

A Note on Causality (20mins)

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 46 / 65

 Causality Motivation

Causal Relationships between Variables

We are interested in the causal relationship between variables.

How does thread length affect the period of a pendulum?
How does minimum wage affect employment?
What is the effect of education on income?

We write down the model as if the relationship is indeed causal:

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

But we never observe the causal relationship

We only observe the “co-movement”, or association between variables.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 47 / 65

 Causality Motivation

Is the Relationship Causal?

Does the sun heat the rock?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 48 / 65

 Causality Motivation

Is the Relationship Causal?

Does the cat cause the collapse?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 49 / 65

 Causality Motivation

Is the Relationship Causal?

Does consuming ice cream cause shark attacks?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 50 / 65

 Causality Motivation

Is the Relationship Causal?

Does eating more chocolate cause more people to win the Nobel Prize?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 51 / 65

 Causality Motivation

Is the Relationship Causal?

Individuals with higher levels of education tend to have higher incomes. Does
education cause income?

Districts with more police tend to have lower crime rates. Does police cause crime?

Countries with higher GDP tend to have lower unemployment rates. Does GDP
cause unemployment?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 52 / 65

 Causality Motivation

Why Do We Care About Causality?

Scientific discovery (e.g., what causes the wealth of nations?)

Business decision (e.g., what will happen if we increase the price of our product?)

Policy making (e.g., what will happen if we increase the minimum wage?)

Policy evaluation (e.g., what is the effect of China’s land reform?)

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 53 / 65

 Causality Concept

How Do We Think About Causality?

All regression models are associational models. There is no causality in regression

models.
Causality is a conceptual issue, not a statistical issue.
The possibilities when we observe an association between two variables X and Y :
X causes Y .
Y causes X .
X and Y are both caused by a third variable Z .
Correlation is a coincidence.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 54 / 65

 Causality Concept

Inferring Causality from Data

The notion of ceteris paribus (all else equal).

The notion of counterfactual (what would have happened if...).

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 55 / 65

 Causality Concept

Inferring Causality from Data

What would have happened if...?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 56 / 65

 Causality Concept

How to Achieve Ceteris Paribus?

Controlled experiments: Varying one factor at a time while keeping other factors
constant. (e.g.,: The Galileo’s pendulum experiments)
Other factors are held constant by design.
Randomized controlled trials (RCTs): Randomly assigning subjects to
treatment and control groups. (e.g.,: Randomly assigning students to different
class sizes)
Everything else is the same on average between the two groups.
Natural experiments: Random events that affect some people but not others.
(e.g.,: Just barely passing a test and just barely failing a test)
Control variables: We can control for other factors by including them in the
regression. (e.g.,: Including other factors that affect crime rates in the regression)
This is the magic of econometrics!
We will discuss this in detail in this course.
ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 57 / 65
 Causality Examples

Example: Effect of Fertilizer on Crop Yield

Question: By how much will the production of soybeans increase if one increases
the amount of fertilizer applied to the ground?

Other factors: quality of land, rainfall, presence of parasites, and so on.

Ideal experiment: Choose several one-acre plots of land; randomly assign

different amounts of fertilizer to the different plots; compare yields.

Feasibility: Can be achieved in a randomized control trial, amount of fertilizer

applied is unrelated to other factors influencing crop yields.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 58 / 65

 Causality Examples

Example: The Return to Education

Question: “If a person is chosen from the population and given another year of
education, by how much will his or her wage increase?”

Ideal experiment: Choose a group of people; randomly assign different amounts

of education to them (infeasable!); compare wage outcomes.

Problem without random assignment: amount of education is related to other

factors that influence wages (e.g. intelligence).

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 59 / 65

 Causality Examples

Example: Law Enforcement and Crime Rate

“If a city is randomly chosen and given ten additional police officers, by how much
would its crime rate fall?”

If two cities are the same in all respects, except that city A has ten more police
officers than city B, by how much would the two cities’ crime rates differ?

Ideal experiment: Randomly assign number of police officers to a large number

of cities

The problem of reverse causality: crime rate may affect the number of police
officers.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 60 / 65

 Causality Examples

Example: Minimum Wage and Unemployment

“By how much (if at all) will unemployment increase if the minimum wage is
increased by a certain amount (holding other things fixed)?”

Ideal experiment: Government randomly chooses minimum wage each year and
observes unemployment outcomes.

Other factors: In reality, the level of the minimum wage will depend on political
and economic factors that also influence unemployment.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 61 / 65

 Causality Examples

Another Example: Impact of the Course

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 62 / 65

 Causality Practice

Practice

Think about a causal question of interest that you have.

What is the ideal experiment that you would like to conduct?

What are the factors that you need to hold constant?

Is the ideal experiment feasible?

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 63 / 65

 Conclusion

Lecture Plan

Motivation: Why Econometrics? (15mins)

Concept of Econometric Analysis (20mins)

Overview of Topics (20mins)

Data Structure (20mins)

Causality (20mins)

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 64 / 65

 Conclusion

Next Lecture
Topic 2 Estimation: Finding the “best” line that fits the data.

Please read Chapters 2-1, 2-2, 3-1, 3-2a of the textbook.

ECON2280: Introductory Econometrics Topic 1: Introduction Jan 20, 2025 65 / 65