Econometrics for Finance

(MBA422)

Tigabu Molla (PhD)

Department of Economics
Faculty of Business and Economics
Woldia University
2022
 Abstract
• Overview:
Econometrics has become an integral part of teaching & research in
modern economics and business. Along with microeconomics and
macroeconomics, it has been taught as one of the three core courses in
most undergraduate and graduate economic programs. The importance
of econometrics has been increasingly recognized and econometric
tools and methods have been widely employed in empirical studies. This
is a graduate level course and focuses on the classical paradigm of
econometrics whereby we use sample to infer about corresponding
population. Topics to be studied include nature of econometrics,
specification, estimation, and inference in the context of models that
include then extend beyond the standard linear multiple regression
framework. It incorporates theoretical background and practical skills in
econometrics used to answer economic/business/financial questions
using data on entities observed at one or multiple points in time.
• Objective:
Preparation to read and carry out empirical social science
research using modern econometric methods.
 Learning outcomes
At the end of this course you should:
̶ have knowledge and skills of regression analysis
̶ know the theoretical background and assumptions for
standard econometric methods
̶ be able to use SPSS/STATA to perform an empirical
analyses
̶ be able to interpret and evaluate outcomes of an
empirical analysis
̶ be able to read and understand empirical studies (like
journal articles that make use of econometric methods)
̶ be able to make use of econometric techniques in your
own academic and other empirical works, for example
in your master’s thesis
 Course content
Chapter 1: Introduction to (financial)
econometrics

Chapter 2: Regression analysis

 Simple regression model
 Multiple regression model
 Regression analysis with qualitative data
 Regression analysis with time series data
 Regression analysis with panel data
 SPSS/STATA application
 Course delivery and evaluation
 Teaching methods:
 Student-centred
 Practicals:
 Software
 STATA
 Questions and review as requested
 Problem Sets: (more details later)
 Readings:
1. Books:
̶ Wooldridge, Jeffery M: Introductory Econometrics
̶ “Introductory Econometrics for Finance” by Chris Brooks.
Cambridge.
̶ Greene, W., Econometric Analysis, 8th Edition, Prentice Hall, 2017
2. Notes and materials
3. A few articles
Cont.
Prerequisites:

 A previous course that used linear regression

 A background in statistics and mathematics

Course Requirements:
 Problem set and article review (30%)
 Midterm (20%)
 Final exam (50%)
 Chapter 1
The Nature of Econometrics and Economic Data
Structure:
1.1 The nature of econometrics
What is econometrics?
Why econometrics?
1.2 Definition of financial econometrics
1.3 Major uses of (financial) econometrics
1.4 Data characteristics (types of data)
1.5 Classical Inference versus Bayesian Inference
1.6 What is Regression Analysis?
1.7 Methodology of econometrics
̶ Model formulation
̶ Model estimation
̶ Model evaluation
 1.1 The Nature of Econometrics
What is Econometrics?
̶ Econo-metrics: ”measurement in economics”. (literal
meaning)
̶ Econometrics is the science and art of using economic
theory, mathematics, statistical techniques and data in
the analysis of economic/business/financial issues.
̶ Econometrics helps us to use information from
financial/management/economic theory and data in
appropriate techniques to make economic decisions
 Cont.
It helps us to identify the direction and magnitude of
relationships between variables in economic relationship
Listing variables in economic relationship is not enough;
we need to answer the how much question. For effective
policy we must know the amount of change needed for a
policy instrument to bring about the desired effect:
 Economic theory suggests relationships between economic
variables.
Example: the demand function for a good Q = f(P, T, Pr, Y, N)
Where Q is quantity demanded, P own price, T taste/preference, Pr
price of related goods, Y income and N number of buyers
 How to a) test whether this relationship exists, and b)
provide a magnitude for the effect?
 This is where econometrics comes in.
1.2 Definition of financial econometrics

 Financial econometrics is a subset of

econometrics and is referred to as the application
of econometrics to problems in finance.
What makes financial econometrics
distinct?
̶ Financial econometrics uses financial theory, math,
statistical tools and financial data to address
financial issues.
̶ That is, it is the use of financial data and financial
theory to address financial issues that makes it
distinct.
What makes financial econometrics distinct?

What are financial theories Financial

econometrics uses financial theory,
math, statistical tools and financial
data to address financial issues.
 What are financial issues?
 Cont.
What are financial issues?
̶ Finance is concerned with how to allocate scarce
resources across assets overtime in order to
earn a return.
• What should we invest in (capital budgeting)?
• Should we use cash (equity) or incur debt (capital
structure)?
• Working capital, for example, liquidity issues)?
̶ These financial issues are addressed to maximize
return
̶ However, since the future is unknown, this makes
finance difficult.
 Returns in financial modeling
Since decisions are made in face of uncertainty,
econometric methods are important in finance.
1. Expected return of a share is the sum of the
earnings per share and expected percentage capital
gain.
E ( Pt 1 )  Pt E ( Dt 1 )
E (r )  
Pt Pt
2. CAPM
The CAPM states that the expected return on any
stock i is equal to the risk-free rate of interest, Rf , plus
a risk premium.
E(Ri) = Rf + βi[E(Rm)−Rf ]
Cont.

̶ We use financial theories and financial data

along with the tools of mathematics and
statistical principles to address these financial
issues
What are financial theories?
These are theories that deal with finance.
What are financial data?
Financial data have special characteristics.
What are the Special Characteristics of Financial Data?

• Frequency & quantity of data

Stock market prices are measured every time there is a
trade or somebody posts a new quote.

• Quality
Recorded asset prices are usually those at which the
transaction took place. No possibility for measurement
error but financial data are “noisy”.
 Uses of financial econometrics

 Testing theories in finance,

 Determining asset prices or returns,
 Testing hypotheses concerning the relationships
between variables,
 Examining the effect on financial markets of
changes in economic conditions,
 Forecasting future values of financial variables
Examples of questions we would like to study:
 What is the effect of advertising on sales?
 What is the effect of risk on return?
 What is the effect of a retraining program on the duration of
unemployment?
 What is the effect of an additional year of education on
earnings?
 How is stock price volatility related to macroeconomic factors?
 Testing whether financial markets are weak-form
informationally efficient.
 Is the CAPM consistent with the reality?
 Do earnings or dividend announcements have effect on stock
prices?
How does (financial) econometrics address such questions?
 Cont.

How does (financial) econometrics address such

questions?
It integrates (unifies) economic theory,
mathematics, statistics and economic data for a
real understanding of economic/business issues
Economic theory
• Economic theory only suggests a qualitative economic
relationship, while econometrics gives empirical (quantitative)
content to most economic theory
Mathematics
• Summarizes the essence of a theory in concise manner without
empirical measure of the theory.
• Ensures logical consistency and coherency in economic theory
itself (checks whether the reasoning process of an economic
theory is correct)
• mathematical modeling is a necessary path to empirical
verification of an economic theory.
• Econometrics gives consistency between theory and stylized facts
Statistics
The tools and methods of statistics provide the operating principles
for analyzing data.
Data
• Data are realizations of stochastic setting (economy)
• We use economic data to ascertain the consistency of economic
theory with reality.
1.3 Major uses of Econometrics (purpose)
1. Estimating relationships between variables
EX: the relationship between education and wages or between
risk and return
2. Testing theories in economics/finance.
Ex: Testing theories of capital structure (static trade-off, pecking
order, …)
3. Policy evaluation.
Ex: Evaluating the impact of community based health insurance
or safety net program
4. Forecasting financial variables (asset prices,
returns…)
Ex: What will be the inflation rate in 2015 E.C in Ethiopia?
1.4 Data Characteristics
All empirical analysis requires data.
Different techniques are required to analyze
different types of data.
Data characteristics:
̶ Quantitative versus qualitative data
̶ Cross sectional, time series and panel data
̶ Experimental versus non-experimental data
̶ Nominal, ordinal and scale variables
 Cont.
a. Quantitative versus Qualitative
̶ Quantitative variables measure "quantities" such as
price, sales volume, weight or income.
̶ Qualitative variables are used to model "either/or"
situations and might be used to model membership
in one of several groups such as homeowner or
non-homeowner, employed/unemployed,
male/female, accurate or inaccurate income tax
returns
̶ Dependent and independent variables can be
quantitative or qualitative variables.
̶ Example: Consider a possible relationship between
salary, years of employment and gender. This model
might be formulated as:
Salary = β0 + β1 years employed + β2 Gender
 Cont.
b. Time Series, Cross Sectional and Panel Data
Cross section data is data collected for multiple entities at one
point in time
Each observation is a new individual, household, firm, etc.. With
information at a point in time.
Examples: Data on expenditures, income, hours of work,
household composition, assets, investments, employment, etc..
Cross sectional data is usually a random sample of the
underlying population.
• E.g. Randomly select 500 people from the population of all working
people in Ethiopia.
• Ordering of observations does not matter.
If the data is not a random sample, we have a sample-selection
problem.
Cross sectional data
Cross section data is data collected for multiple entities at
one point in time
Each observation is a new individual, household, firm, etc..
With information at a point in time.
Examples: Data on expenditures, income, hours of work,
household composition, assets, investments, employment,
etc..
Cross sectional data is usually a random sample of the
underlying population.
• E.g. Randomly select 500 people from the population of all working
people in Ethiopia.
• Ordering of observations does not matter.
If the data is not a random sample, we have a sample-
selection problem.
A cross-sectional dataset on wages and other individual
characteristics
Time series
• Time series data is data collected for a single entity at multiple
points in time
• Time series data consists of observations on a set of variables
over time.
• Separate observation for each time period. Ordering of
observations does matter.
• Typically Macroeconomic measures: Yearly GDP of Ethiopia for a
period of 20 years, quarterly data on Inflation and unemployment
rate in Ethiopia from 1970-2011, Prices, daily Birr/US dollar
exchange for the past year, Interest Rates, etc..
• Financial data: Stock Prices, Bonds and other financial instruments
at frequencies that range from minute to minute up to annual.
• Since not a random sample, different problems to consider.
• Trends and seasonality will be important. (E.g.: monthly ice cream
sales; wheat production.)
Time series dataset on minimum wage, unemployment &
related variables for Puerto Rico
Panel (longitudinal) data
Panel data is data collected for same multiple entities at
multiple points in time
Has both cross sectional and time series dimensions
Similar to pooled cross sections with one important
difference.
 Responses related to the same cross=sectional unit (individual,
firm, country) over time.
Consists of time series for each cross-sectional unit.
Allows to control for unobserved characteristics of the
cross-sectional unit (e.g. ability).
Data is usually structured by cross sectional unit over time.
• All observations on unit 1 over time, then observations on unit 2
over time, ...
Panel data
Cont.

c. Non-experimental versus Experimental Data

̶ Non-experimental data:- typical in the social sciences.
Observations drawn from a system not subject to
experimental control.
̶ Experimental (common in natural sciences, but
experimental data are becoming more commonly
used in economics)
examples: Physics/chemistry, Negative income tax
(different tax rates, direct subsidies), Health insurance,
Influence of housing allowance, Split cable--different
commercials
 Cont.

d. Nominal, ordinal and scale variables

• Nominal. A variable can be treated as nominal when its
values represent categories with no intrinsic ranking
For example,
• Department of the company in which an employee works
• Region, religious affiliation,.....
• Ordinal. A variable can be treated as ordinal when its values
represent categories with some intrinsic ranking
For example,
• Levels of satisfaction in a certain service giving organization
• Grade, Attitude scores, preference rating scores,… .

• Scale. A variable can be treated as scale when its values

represent ordered categories with a meaningful metric, so
that distance comparisons between values are appropriate.
• Examples of scale variables include weight in kg, age in years ,
income in thousands of dollars.
 Why econometrics?
Economic data is (in the main) non-experimental and
hence non-deterministic.
• It is observed from the world around us; they are outcomes of
uncontrolled experiments
• It is not obtained from ’test tube’ experiments.
• This means we can rarely ’hold constant’ all factors in the
economic model.
• This means that all variables must be treated as random

E.g.: What is the effect of an extra year of education on

wages, holding all other factors constant (ability, age, sex,
experience, industry etc.)?
o It is impossible to hold these factors constant in reality.
(We cannot force people to undergo or not undergo
extra education. And we certainly cannot keep age
constant while adding another year of education...)
o Even if it was possible, it would be unethical in many
cases. E.g.: What is the effect of a reduction in
unemployment benefits on the probability of finding a
job?
 1.5 Econometrics: Paradigm
Classical Inference versus Bayesian Inference
Paradigm: Classical Inference

Population Measurement

Econometrics
Characteristics
Imprecise inference about the Behavior Patterns
entire population – sampling
theory and asymptotics Choices
Paradigm: Bayesian Inference

Population Measurement

Econometrics
Characteristics
Sharp, ‘exact’ inference about only
the sample – the ‘posterior’ density.
Behavior Patterns
Choices
 1.6 Regression analysis
regression analysis is concerned with the conditional
expectation (the average value) of the dependent (response)
variable given the explanatory variables
Less commonly, the focus can also be on a quantile, or other
location parameter of the conditional distribution of the
dependent variable given the explanatory variables.
̶ For the case of a conditional quantile, one usually calls it the quantile
regression function.
̶ In all cases, a function of the explanatory variables is loosely called
the regression function.
̶ Conditional mean analysis or regression analysis is one of the most
popular statistical methods in econometrics
cont

The term “regression” was coined by

Galton (1877, 1885) in the nineteenth
century to describe a biological
phenomenon. The phenomenon was that
the heights of descendants of tall
ancestors tend to regress down towards a
normal average, a phenomenon also
known as regression toward the mean.
1.7 Methodology in (financial)
econometrics
Stages of econometric analysis:
1. Model formulation (specification)
̶ What is the right model to use to
analyze economic relationship?
2. Model estimation
3. Model evaluation (diagnosis)
Tigabu Molla (PhD) 37
1. Formulating a model
In this stage, there are three important tasks to fulfill:
A) Developing hypothesis
• The dependent and independent (explanatory)
variables which will be included in the model.
• We form a priori theoretical expectations about
the size and sign of the relationship between
variables (parameters of the function).
• Depends on theory, past experience, other
studies, ‘Common sense’

Tigabu Molla (PhD) 38

Cont.
B) Specification of the mathematical model of the
hypothesis.
Expressing the hypothesized relationship in
mathematical form with which economic
phenomena will be explored empirically.
 A mathematical model expresses the
relationship in deterministic (exact) way
C) Specification of the econometrics model of the
hypothesis.
Expressing the hypothesized relationship
in econometric model
 An econometric model expresses the
relationship in stochastic (inexact) way 39
Tigabu Molla (PhD)
An econometric model is a complete
specification of economic and
statistical behavior.
We include the stochastic
component ε in the mathematical
(deterministic) model to form an
econometric (stochastic) model as

Tigabu Molla (PhD) 40

a. Stochastic (econometric) versus non-
stochastic (deterministic) relationships

Tigabu Molla (PhD) 41

 b. The Disturbance (Error) Term, ε

Disturbance term plays a pivotal role in econometrics.

It accounts for all the factors that affect the
dependent variable that we have not explicitly
accounted for in our regression function.
Omission of variables from the function. For example,
education is not the only determinants of wage.
 Randomness of human behavior
Unavailability of data
Wrong or imperfect specification of the functional form
Errors of aggregation
Errors of measurement
Tigabu Molla (PhD) 42
c. Population regression versus sample regression
i) Population regression
The true econometric model that relates the
dependent variable (variable to be explained) with the
variable (s) that are expected to explain this effect is
called the population regression model (PRM)
The true function that relates the conditional
mean of the dependent variable with the
independent variable (s) is called the
population regression function (PRF)
Tigabu Molla (PhD)
43
 Cont.
Population regression curve is the locus of the conditional
means or expectations of the dependent variable for the
fixed values of the explanatory variable X
PRF tells us how the average value of y changes with x. It
does not say that y = β0 + β1X for everyone in the
population. It refers to the average tendency.
• Some observations will be below the average and some will be
above the average, depending on the influence of u in each
specific case.
• Example: Y= β0 + β1 X+ ε (PRM)
Y= E(Y|X)+ ε (regression identity)
E (Y / X )   0  1 X population regression function
E(/X)  0 expected value of the error term

 Population regression model is a true model but not directly

observable or not feasible to compute. Tigabu Molla (PhD)
44
• The random variable ε represents the part of Y that is
not captured by E(Y |X). It is usually called a noise or a
disturbance, because it “disturbs” an otherwise stable
or deterministic relationship between Y and X. On the
other hand, the regression function E(Y |X) is called a
signal.
• The property that E(ε|X) = 0 implies that the regression
disturbance ε contains no systematic information of X
that can be used to predict the expected value of Y. In
other words, all information of X that can be used to
predict the expectation of Y has been completely
summarized by E(Y |X).
• The condition E(ε|X) = 0 is crucial for the validity of
economic interpretation of model parameters

Tigabu Molla (PhD) 45

 Parameters and variables
There are two types of objects in this PRM:
- Variables
- Parameters
A variable is a factor, trait or condition which assumes
a variety of values in a particular problem (one whose
values is not known yet)
̶ There are dependent and independent variables in a
model.
Parameter:- is a quantity that characterizes a population
(true) relationship and that can be estimated from sample
data.
 parameters are also called constants/unknowns/coefficients 46
Tigabu Molla (PhD)
 Cont.
Intercept parameters are parameters which describe the
value that the dependent variable will take when
independent variables are all equal zero
 Slope parameters (also called slope coefficients) are
parameters that describe the impact that independent
variable has on the dependent variable. For example, β1 in
the wage model shows the effect on Wages for one
additional year of Education.
The statistical & economic significances are important
Example: Y = β0 + β1 X + ε
• Y is dependent variable
• X is independent variable
• β0 is the intercept term
• β1 is the slope coefficient 47
Tigabu Molla (PhD)
 ii) sample regression function (SRF)

The PRF (E(Y/X)=β0 + β1X) is a true model but not directly

observable or not feasible to compute.
We therefore estimate it from the SRF
Example:
• Imagine we wish to find the relationship between father’s
height and son’s height.
• Imagine we had data on ALL the world’s father’s and son’s
heights.
• Then we can get the Population Regression Function!
son' s height i     (father' s height ) i   i

• But we can’t observe the heights of ALL fathers and sons in

the world throughout time.
• So we have to do with a sample! 48
Tigabu Molla (PhD)
 As we do not have the luxury to have data on the
population, we rely on the sample
SRF is the sample counterpart of PRF:

Yˆ  ˆ0  ˆ1 x sample regression function

Y  Yˆ  ˆ  ˆ0  ˆ1 x  ˆ
where
Yˆ is called fitted/predicted value of actual value Y
ˆ  estimator of 
0 0

ˆ1  estimator of 1
ˆ  Y  Yˆ  actual value  predicted value  estimator of  & is
called residual.
Are results of the SRF the same as that of the PRF? The
answer is no due to sampling fluctuation
The task is to make sure that the estimators are as close as
the population parameters 49
Tigabu Molla (PhD)
PRF shows population average tendency
whereas
SRF sample average tendency Tigabu Molla (PhD)
50
Population & Sample Regression Models
(classical regression)

Population Random Sample

Unknown y  ?0  1 x  ˆ
Relationship
$
y  0  1 x   $
$
$ $
$
$

Tigabu Molla (PhD) 51

 2. Model estimation (second stage)
Model estimation refers to estimating
(calculating) the values of parameters based on
empirical data that has a random component
Parameters describe an underlying physical setting
but they are unknown and hence should be
estimated
Two tasks:
1. Estimating the model using appropriate
econometric techniques (estimators).
2. Collection of data on the variables of the model
Tigabu Molla (PhD) 52
 Cont.
The choice of estimator (estimation method)
depends on:
a) type of data we use in our study;
b) nature of the dependent variable
c) functional form
We use an estimator to calculate parameter
estimates.
An estimator is a formula for calculating parameter
estimates or for estimating parameters.
An estimate is a numerical value we obtain after
substituting data/sample in the estimator.
53
Tigabu Molla (PhD)
Types of estimators:
Ordinary least squares method
Method of moments
Maximum likelihood method

We typically use Greek letters such as  , and  2 to

denote unknown parameters of an econometric
model
Estimators are typically denoted by putting a hat ^,

tilde ~or bar -over the corresponding letter, e.g.,
are estimators of ˆ and ~
Estimators work in relevant environment which is
determined by assumptions.
Tigabu Molla (PhD) 54
 3. Model evaluation (third stage)
The process of checking the adequacy of the model
against a range of criteria and possibly returning to the
model formulation stage.
The evaluation consist of deciding whether the
estimates of the parameters are theoretically meaningful
and statistically satisfactory.
The evaluation of the model is based on economic a
priori criteria, statistical criteria, econometric criteria and
the forecasting ability of the model.
1. Economic a priori criteria:
These criteria are determined by economic theory and
refer to the size and sign of the parameters of
economic relationships. 55
Tigabu Molla (PhD)
Cont.
ii. Statistical criteria (first-order tests):
These are determined by statistical theory
aim at the evaluation of the statistical reliability of the
estimates of the parameters of the model.
Correlation coefficient test, standard error test, t-test, F-
test, and R2-test are some of the most commonly used
statistical tests.
iii. Econometric criteria (second-order tests):
aim at the detection of the violation or validity of the
assumptions of the various econometric techniques.
They serve as a test of the statistical tests i.e. they
determine the reliability of the statistical criteria; they
help us establish whether the estimates have the
desirable properties
Tigabu Molla (PhD) 56
Summary: Flow chart for the Steps of an
Empirical Study

Tigabu Molla (PhD) 57

 Formulating a model: Example
Example: A study on returns to education
A model of human capital investment predicts that getting
more education should lead to higher wages.
Formulating a model (first stage in econometrics)
• Hypothesis: more education leads to higher wages
• Mathematical model
Wages = β0 + β1Education (exact relationship)
• Econometric model
wages = β0 + β1 Education + ε (inexact relationship)
ε is called the disturbance (error, stochastic) term
β0 + β1 Education is called the systematic component and ε
the random component
58
Tigabu Molla (PhD)
Cont.
How can we study if the evidence of the
data supports economic theory (the
model of human capital investment)?
How to approach this question?
 Let us look at the following data set as
example :
US national survey of 528 people in the labor force
that already completed their education.
 Scatterplot of this dataset is drawn on next
page

Tigabu Molla (PhD) 59

Formulating a model

• People with the same years of education have different

hourly wages.
• There is a distribution for the hourly wages conditional
on the years of education. 60
Tigabu Molla (PhD)
• How can we study if the evidence of the data
supports Economic Theory?
• A possibility is to look at means of wages conditional
on the years of Education.

61
Tigabu Molla (PhD)
• Conditional Mean Function: Hourly Wages and Education

• We can see that the mean of wages vary with the years of Education.
• Hence, the object that we are interested in studying is the mean of wages
given the years of Education: E[Wages|Education]. 62
Tigabu Molla (PhD)
Exercise
Discuss the econometric stages we follow in
addressing the following issues:

1. The effect of advertising on sales

Tigabu Molla (PhD) 63