微信客服：xiaoxionga100
微信客服：ITCS521
程序代写案例-ECOM30002
时间：2022-06-21
Course Information Course Structure Aims and Objectives Regression Interpretation
ECOM30002
Econometrics 2
Lecture 1
Introduction to the Course
Matthew Greenwood-Nimmo
matthew.greenwood@unimelb.edu.au
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Information
Teaching Team
Course Coordinator:
Matthew Greenwood-Nimmo
Oce Hours: Thursdays 10:00-12:00, 312 FBE Building
Email: matthew.greenwood@unimelb.edu.au
Tutor Coordinator:
Daniel Tiong
Oce Hours: TBA, 419 FBE Building
Email: daniel.tiong@unimelb.edu.au
Tutors:
Details available via the LMS
Note that tutors will start holding oce hours from week 3
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Information
Overview of the Course
According to Merriam-Webster, econometrics is “the application of
statistical methods to the study of economic data and problems”
In this course, we will study a number of statistical methods:
We will start with the classic ordinary least squares approach applied to
cross-sectional data
We will study the assumptions underlying the OLS estimator and the
problems that arise when these assumptions are violated
We will learn how to address some of these problems
We will study models designed to exploit panel data, where observations
vary over both cross-section units and time periods
We will wrap up with a section on dynamic econometric models
We will have a strong focus on real-world applications throughout
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Information
Assumed Knowledge
Prior knowledge of statistics/econometrics at level 2 is required
Most of you will have completed one or more of:
ECON20003 Quantitative Methods 2
ECOM20001 Econometrics 1
MAST20005 Statistics
ECOM30001 Basic Econometrics
These courses all cover random variables, expectations and
conditioning, single and multiple linear regression and statistical
inference using common distributions, among other topics
You need to have mastered this material – we will refresh some of this
content but not in sucient depth to act as a substitute for prior
knowledge
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Information
Assumed Knowledge
You will encounter a moderate amount of maths in this course,
although our focus is more on applications than econometric theory
You will need to be comfortable with simple matrix algebra (such as
handling matrix transposition, inversion and multiplication)
You will have a chance to brush up on this in the first few tutorials
The statistical package that we will use is R / RStudio
Prior knowledge of R is not necessary but a willingness to learn by
self-study is crucial
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Information
Reference Materials
The main textbook for this course is:
Stock, J. and Watson, M. (2015), Introduction to Econometrics,
Updated 3rd Edition, Pearson
There are plenty of copies available in the library
For some of the course content, you may find the following optional
textbook useful:
Hill, R.C., Griths, W.E. and Lim, G.C. (2018), Principles of
Econometrics, 5th Edition, Wiley.
When working with R, you will need to consult:
The R help documentation, which can be accessed using ?command name
Great instructions on using R
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Course Structure
The main course content is delivered in 23 x 1 hour lectures (including
today but not the final review session)
The lectures can be broken down into three sections:
1 OLS, instrumental variables and related methods. Properties of
estimators. Predictive and causal models. Omitted variables and
simultaneity. Statistical inference.
2 Panel data analysis, including pooled OLS and fixed e↵ects regression.
3 Time series analysis, including single-equation and vector autoregressive
models.
The lectures are supported by a sequence of 12 x 1 hour tutorials,
starting this week
You are expected to complete part A of the tutorial problem sets in
advance and to work on part B during the tutorial
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Course Structure
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Course Structure
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Assessment
Four in-semester group assignments:
Groups of between 1 and 4. More details to follow via the LMS.
Released via the LMS in weeks 2, 5, 8 and 11
Due by 4pm on Wednesday of weeks 4, 6, 9 and 12
Each assignment is worth 7.5% of the final mark
Limit of 600 words for each assignment
A two-hour final exam worth 70% of the final mark
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Support for Students A↵ected by Travel Restrictions
Sadly, some students enrolled in this course are unable to attend
lectures and tutorials due to the ongoing travel restrictions
The entire teaching team is committed to helping a↵ected students to
e↵ectively engage with the course. We will:
Make lecture recordings available via the LMS shortly after each lecture
Make tutorial recordings available via the LMS for at least the first 4
weeks of semester
Strive for an expedited turnaround of queries posted to the Ed
discussion forum
Run additional tutor consultations once the travel ban is lifted
Release the first assignment one week early to ease time pressures facing
a↵ected students
Monitor the situation closely and take additional measures if necessary
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Support Services
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Aims and Objectives
Introduction
Aims
To highlight the di↵erence between the conditional mean and the causal
interpretations of a regression model.
Objectives
On completion of this lecture, students should:
Understand the distinction between the conditional mean and the
causal interpretations of a regression model.
Have a refreshed understanding of marginal e↵ects.
Reading: No specific reading.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Aims and Objectives
Overview
1 Conditional Mean Interpretation
2 Causal Interpretation
3 Marginal E↵ects
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Does Chocolate Make You Smarter?
Messerli, F.H. (2012) “Chocolate Consumption, Cognitive Function, and
Nobel Laureates,” New England Journal of Medicine, 367: 1562–1564.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
A Linear Regression
## lm(formula = Nobels ~ ChocPerCap, data = choc_data)
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -8.2499 5.1813 -1.5923 0.1278
## ChocPerCap 3.0805 0.7366 4.1820 0.0005
## R-squared: 0.4793
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
A Linear Regression
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
How is this Interpreted?
We will consider two interpretations of this regression:
1 A statistical conditional mean interpretation
2 A causal interpretation
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Conditional Mean Interpretation
A statistical interpretation
Population Regression Function:
E(Nobelsi|ChocPerCapi) = 0 + 1ChocPerCapi
Sample Regression Function:bE(Nobelsi|ChocPerCapi) = ˆ0 + ˆ1ChocPerCapi
= 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
On average, countries with higher per capita chocolate consumption
also have more Nobel prize winners relative to population
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
t
conditional mean
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Conditional Mean Interpretation
A statistical interpretation
Population Regression Function:
E(Nobelsi|ChocPerCapi) = 0 + 1ChocPerCapi
Sample Regression Function:bE(Nobelsi|ChocPerCapi) = ˆ0 + ˆ1ChocPerCapi
= 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
On average, countries with higher per capita chocolate consumption
also have more Nobel prize winners relative to population
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
t
group of stats .
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Conditional Mean Interpretation
A statistical interpretation
Population Regression Function:
E(Nobelsi|ChocPerCapi) = 0 + 1ChocPerCapi
Sample Regression Function:bE(Nobelsi|ChocPerCapi) = ˆ0 + ˆ1ChocPerCapi
= 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
On average, countries with higher per capita chocolate consumption
also have more Nobel prize winners relative to population
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Conditional
✓
model
NO causality = no cause of some prerequisites leading to an result
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Conditional Mean Interpretation
In some applications, the conditional expectation is interpreted as a
prediction
e.g. the predicted number of Nobel prizes per 10 million people in a
country whose annual chocolate consumption is 6 kg/person is:
bE(Nobelsi|ChocPerCapi = 6) = ˆ0 + ˆ1 ⇥ 6
= 8.250 + 3.080⇥ 6
= 10.233.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
*
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Causal Interpretation
An economic/physical/behavioural interpretation
A causal equation:
Nobelsi = 0 + 1ChocPerCapi + Ui
\Nobelsi = 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
Higher chocolate consumption causes more Nobel prizes to be won?
A regression without sensible causal interpretation may still be useful
for means/prediction.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
causal disturbance
term
v1
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Causal Interpretation
An economic/physical/behavioural interpretation
A causal equation:
Nobelsi = 0 + 1ChocPerCapi + Ui
\Nobelsi = 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
Higher chocolate consumption causes more Nobel prizes to be won?
A regression without sensible causal interpretation may still be useful
for means/prediction.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Causal Interpretation
An economic/physical/behavioural interpretation
A causal equation:
Nobelsi = 0 + 1ChocPerCapi + Ui
\Nobelsi = 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
Higher chocolate consumption causes more Nobel prizes to be won?
A regression without sensible causal interpretation may still be useful
for means/prediction.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Marginal E↵ects
Marginal e↵ect: e↵ect due to a one-unit change in an explanatory
variable, holding all else constant.
Conditional mean interpretation:
1 is the marginal e↵ect of ChocPerCap on E(Nobelsi|ChocPerCapi)
Causal interpretation:
1 is the marginal e↵ect of ChocPerCap on Nobelsi
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Marginal E↵ects
Marginal e↵ect: e↵ect due to a one-unit change in an explanatory
variable, holding all else constant.
Conditional mean interpretation:
1 is the marginal e↵ect of ChocPerCap on E(Nobelsi|ChocPerCapi)
Causal interpretation:
1 is the marginal e↵ect of ChocPerCap on Nobelsi
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Marginal E↵ects
Marginal e↵ect: e↵ect due to a one-unit change in an explanatory
variable, holding all else constant.
Conditional mean interpretation:
1 is the marginal e↵ect of ChocPerCap on E(Nobelsi|ChocPerCapi)
Causal interpretation:
1 is the marginal e↵ect of ChocPerCap on Nobelsi
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Multiple Regression
Consider individual wages modelled using education, work experience
and IQ
We’ll consider a random sample of 3,010 males taken from Card (1993)
Card, D. (1993) “Using Geographic Variation in College Proximity to
Estimate the Return to Schooling.” NBER Working paper no. 4483
## Dependent variable: wage
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -516.66 56.76 -9.10 0
## educ 47.30 3.16 14.95 0
## exper 28.95 1.77 16.31 0
## IQ 2.22 0.40 5.54 0
## R-squared: 0.1681
“Returns to education”?
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
might not be
good causal
ratable
.
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Multiple Regression
In the PRF
E(wagei|educi, experi, IQi) = 0 + 1educi + 2experi + 3IQi
1 is the marginal e↵ect of education on average wages, holding
experience and IQ constant.
Is 1 also the causal marginal e↵ect of education on wages?
Thinking about this requires thinking about the role of omitted
variables, among other things.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Multiple Regression
In the PRF
E(wagei|educi, experi, IQi) = 0 + 1educi + 2experi + 3IQi
1 is the marginal e↵ect of education on average wages, holding
experience and IQ constant.
Is 1 also the causal marginal e↵ect of education on wages?
Thinking about this requires thinking about the role of omitted
variables, among other things.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
In the Next Lecture
Omitted variables
The E↵ect of Including/Excluding a Regressor
The Law of Iterated Expectations
The Omitted Variables Formula
Interpretation
Application to the Gender Wage Gap
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
ECOM30002
Econometrics 2
Lecture 1
Introduction to the Course
Matthew Greenwood-Nimmo
matthew.greenwood@unimelb.edu.au
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Information
Teaching Team
Course Coordinator:
Matthew Greenwood-Nimmo
Oce Hours: Thursdays 10:00-12:00, 312 FBE Building
Email: matthew.greenwood@unimelb.edu.au
Tutor Coordinator:
Daniel Tiong
Oce Hours: TBA, 419 FBE Building
Email: daniel.tiong@unimelb.edu.au
Tutors:
Details available via the LMS
Note that tutors will start holding oce hours from week 3
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Information
Overview of the Course
According to Merriam-Webster, econometrics is “the application of
statistical methods to the study of economic data and problems”
In this course, we will study a number of statistical methods:
We will start with the classic ordinary least squares approach applied to
cross-sectional data
We will study the assumptions underlying the OLS estimator and the
problems that arise when these assumptions are violated
We will learn how to address some of these problems
We will study models designed to exploit panel data, where observations
vary over both cross-section units and time periods
We will wrap up with a section on dynamic econometric models
We will have a strong focus on real-world applications throughout
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Information
Assumed Knowledge
Prior knowledge of statistics/econometrics at level 2 is required
Most of you will have completed one or more of:
ECON20003 Quantitative Methods 2
ECOM20001 Econometrics 1
MAST20005 Statistics
ECOM30001 Basic Econometrics
These courses all cover random variables, expectations and
conditioning, single and multiple linear regression and statistical
inference using common distributions, among other topics
You need to have mastered this material – we will refresh some of this
content but not in sucient depth to act as a substitute for prior
knowledge
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Information
Assumed Knowledge
You will encounter a moderate amount of maths in this course,
although our focus is more on applications than econometric theory
You will need to be comfortable with simple matrix algebra (such as
handling matrix transposition, inversion and multiplication)
You will have a chance to brush up on this in the first few tutorials
The statistical package that we will use is R / RStudio
Prior knowledge of R is not necessary but a willingness to learn by
self-study is crucial
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Information
Reference Materials
The main textbook for this course is:
Stock, J. and Watson, M. (2015), Introduction to Econometrics,
Updated 3rd Edition, Pearson
There are plenty of copies available in the library
For some of the course content, you may find the following optional
textbook useful:
Hill, R.C., Griths, W.E. and Lim, G.C. (2018), Principles of
Econometrics, 5th Edition, Wiley.
When working with R, you will need to consult:
The R help documentation, which can be accessed using ?command name
Great instructions on using R
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Course Structure
The main course content is delivered in 23 x 1 hour lectures (including
today but not the final review session)
The lectures can be broken down into three sections:
1 OLS, instrumental variables and related methods. Properties of
estimators. Predictive and causal models. Omitted variables and
simultaneity. Statistical inference.
2 Panel data analysis, including pooled OLS and fixed e↵ects regression.
3 Time series analysis, including single-equation and vector autoregressive
models.
The lectures are supported by a sequence of 12 x 1 hour tutorials,
starting this week
You are expected to complete part A of the tutorial problem sets in
advance and to work on part B during the tutorial
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Course Structure
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Course Structure
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Assessment
Four in-semester group assignments:
Groups of between 1 and 4. More details to follow via the LMS.
Released via the LMS in weeks 2, 5, 8 and 11
Due by 4pm on Wednesday of weeks 4, 6, 9 and 12
Each assignment is worth 7.5% of the final mark
Limit of 600 words for each assignment
A two-hour final exam worth 70% of the final mark
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Support for Students A↵ected by Travel Restrictions
Sadly, some students enrolled in this course are unable to attend
lectures and tutorials due to the ongoing travel restrictions
The entire teaching team is committed to helping a↵ected students to
e↵ectively engage with the course. We will:
Make lecture recordings available via the LMS shortly after each lecture
Make tutorial recordings available via the LMS for at least the first 4
weeks of semester
Strive for an expedited turnaround of queries posted to the Ed
discussion forum
Run additional tutor consultations once the travel ban is lifted
Release the first assignment one week early to ease time pressures facing
a↵ected students
Monitor the situation closely and take additional measures if necessary
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Course Structure
Support Services
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Aims and Objectives
Introduction
Aims
To highlight the di↵erence between the conditional mean and the causal
interpretations of a regression model.
Objectives
On completion of this lecture, students should:
Understand the distinction between the conditional mean and the
causal interpretations of a regression model.
Have a refreshed understanding of marginal e↵ects.
Reading: No specific reading.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Aims and Objectives
Overview
1 Conditional Mean Interpretation
2 Causal Interpretation
3 Marginal E↵ects
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Does Chocolate Make You Smarter?
Messerli, F.H. (2012) “Chocolate Consumption, Cognitive Function, and
Nobel Laureates,” New England Journal of Medicine, 367: 1562–1564.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
A Linear Regression
## lm(formula = Nobels ~ ChocPerCap, data = choc_data)
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -8.2499 5.1813 -1.5923 0.1278
## ChocPerCap 3.0805 0.7366 4.1820 0.0005
## R-squared: 0.4793
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
A Linear Regression
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
How is this Interpreted?
We will consider two interpretations of this regression:
1 A statistical conditional mean interpretation
2 A causal interpretation
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Conditional Mean Interpretation
A statistical interpretation
Population Regression Function:
E(Nobelsi|ChocPerCapi) = 0 + 1ChocPerCapi
Sample Regression Function:bE(Nobelsi|ChocPerCapi) = ˆ0 + ˆ1ChocPerCapi
= 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
On average, countries with higher per capita chocolate consumption
also have more Nobel prize winners relative to population
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
t
conditional mean
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Conditional Mean Interpretation
A statistical interpretation
Population Regression Function:
E(Nobelsi|ChocPerCapi) = 0 + 1ChocPerCapi
Sample Regression Function:bE(Nobelsi|ChocPerCapi) = ˆ0 + ˆ1ChocPerCapi
= 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
On average, countries with higher per capita chocolate consumption
also have more Nobel prize winners relative to population
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
t
group of stats .
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Conditional Mean Interpretation
A statistical interpretation
Population Regression Function:
E(Nobelsi|ChocPerCapi) = 0 + 1ChocPerCapi
Sample Regression Function:bE(Nobelsi|ChocPerCapi) = ˆ0 + ˆ1ChocPerCapi
= 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
On average, countries with higher per capita chocolate consumption
also have more Nobel prize winners relative to population
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Conditional
✓
model
NO causality = no cause of some prerequisites leading to an result
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Conditional Mean Interpretation
In some applications, the conditional expectation is interpreted as a
prediction
e.g. the predicted number of Nobel prizes per 10 million people in a
country whose annual chocolate consumption is 6 kg/person is:
bE(Nobelsi|ChocPerCapi = 6) = ˆ0 + ˆ1 ⇥ 6
= 8.250 + 3.080⇥ 6
= 10.233.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
*
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Causal Interpretation
An economic/physical/behavioural interpretation
A causal equation:
Nobelsi = 0 + 1ChocPerCapi + Ui
\Nobelsi = 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
Higher chocolate consumption causes more Nobel prizes to be won?
A regression without sensible causal interpretation may still be useful
for means/prediction.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
causal disturbance
term
v1
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Causal Interpretation
An economic/physical/behavioural interpretation
A causal equation:
Nobelsi = 0 + 1ChocPerCapi + Ui
\Nobelsi = 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
Higher chocolate consumption causes more Nobel prizes to be won?
A regression without sensible causal interpretation may still be useful
for means/prediction.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Causal Interpretation
An economic/physical/behavioural interpretation
A causal equation:
Nobelsi = 0 + 1ChocPerCapi + Ui
\Nobelsi = 8.250
(5.181)
+ 3.080
(0.737)
ChocPerCapi
Higher chocolate consumption causes more Nobel prizes to be won?
A regression without sensible causal interpretation may still be useful
for means/prediction.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Marginal E↵ects
Marginal e↵ect: e↵ect due to a one-unit change in an explanatory
variable, holding all else constant.
Conditional mean interpretation:
1 is the marginal e↵ect of ChocPerCap on E(Nobelsi|ChocPerCapi)
Causal interpretation:
1 is the marginal e↵ect of ChocPerCap on Nobelsi
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Marginal E↵ects
Marginal e↵ect: e↵ect due to a one-unit change in an explanatory
variable, holding all else constant.
Conditional mean interpretation:
1 is the marginal e↵ect of ChocPerCap on E(Nobelsi|ChocPerCapi)
Causal interpretation:
1 is the marginal e↵ect of ChocPerCap on Nobelsi
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Marginal E↵ects
Marginal e↵ect: e↵ect due to a one-unit change in an explanatory
variable, holding all else constant.
Conditional mean interpretation:
1 is the marginal e↵ect of ChocPerCap on E(Nobelsi|ChocPerCapi)
Causal interpretation:
1 is the marginal e↵ect of ChocPerCap on Nobelsi
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Multiple Regression
Consider individual wages modelled using education, work experience
and IQ
We’ll consider a random sample of 3,010 males taken from Card (1993)
Card, D. (1993) “Using Geographic Variation in College Proximity to
Estimate the Return to Schooling.” NBER Working paper no. 4483
## Dependent variable: wage
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -516.66 56.76 -9.10 0
## educ 47.30 3.16 14.95 0
## exper 28.95 1.77 16.31 0
## IQ 2.22 0.40 5.54 0
## R-squared: 0.1681
“Returns to education”?
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
might not be
good causal
ratable
.
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Multiple Regression
In the PRF
E(wagei|educi, experi, IQi) = 0 + 1educi + 2experi + 3IQi
1 is the marginal e↵ect of education on average wages, holding
experience and IQ constant.
Is 1 also the causal marginal e↵ect of education on wages?
Thinking about this requires thinking about the role of omitted
variables, among other things.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
Regression Interpretation
Multiple Regression
In the PRF
E(wagei|educi, experi, IQi) = 0 + 1educi + 2experi + 3IQi
1 is the marginal e↵ect of education on average wages, holding
experience and IQ constant.
Is 1 also the causal marginal e↵ect of education on wages?
Thinking about this requires thinking about the role of omitted
variables, among other things.
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
Course Information Course Structure Aims and Objectives Regression Interpretation
In the Next Lecture
Omitted variables
The E↵ect of Including/Excluding a Regressor
The Law of Iterated Expectations
The Omitted Variables Formula
Interpretation
Application to the Gender Wage Gap
Lecture 1 Introduction to the Course ECOM30002 Econometrics 2
