Outlines of CH-2:

 Regression Analysis
 Simple & Multiple Regression
 Assumptions
 Estimation method(OLS)& Their properties (Precision)
 Application Example

St. Mary's University

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 2. The Concept of Regression Analysis

 Regression analysis is concerned with the study of the dependence of

one variable (the dependent variable) on one or more other variable(s)
(the explanatory variables) with a view of estimating and/or predicting
the mean value of the population of the former in terms of the known
or fixed (in repeated sampling) values of the latter..

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 ….Regression and related concepts
1). Statistical Vs. Deterministic Relationships

 In regression analysis, we are concerned with statistical dependence among variables (not

functional or deterministic relationship), that is why we essentially deal with random or

stochastic variables that use probability concepts or distribution. Hence, we couldn’t deal

deterministic r/nship.
2) Regression Vs. Causation

 Regression does not necessarily imply causation, b/c statistical relationship cannot logically

imply causation rather causation comes from theory, intuition, common-sense etc.
3) Regression Vs. Correlation

 In regression analysis there is an asymmetry,( the dependent variable is assumed to random,

while the explanatory variables are non-random).

 In correlation analysis, we treat variables symmetrically (both variables are assumed to be

random).
( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 2.1. Simple Regression
 Simple regression is an analysis of a single dependent variable, Y,
as a function of a single independent variable, X.
PRF Vs SRF
 A population regression function (PRF) hypothesizes a theoretical
(population) relationship between a dependent variable and a set of
independent or explanatory variables.
Intercept w/c is fixed & unknown. Slope (parameter) w/c is fixed & unknown

𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜀𝑖 (2.1)

Dependent variable,
explained variable, Error term, disturbance,
response variable,… Independent variable, unobservables,…
explanatory variable,
regressor,…

(
Econometric for finance by Teshome H. MSc in Econometrics ) October 3, 2023
 ..cont’d
 The sample regression function(SRF) is an equation that represents the
relationship between the Y variable and X variable(s) that is based
only on the information in a sample of the population.
Fitted(predicted)value Residuals

𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜀𝑖 (2.2)

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 …cont’d
Why we introduced error term?
We introduce the error tem to captures the effect of:
1). Omitted variables: in reality, a given variable is influenced by large
number of factors and all can’t be included b/c of:
 Some factor may not be known

 Some factor may not be measurable (e.g. test, preference, motivation etc. ).

 Availability of data may not be adequate

2) Measurement error in the dependent & independent variables

3) Model Misspecification error
4) Sampling error
5) Erratic nature of human being: human behaviour may deviate from
the normal situation to a certain extent in unpredictable way

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 2.1.1. Assumptions of Simple Regression (Gauss-Markov Assumptions)

1) The regression model is linear in the parameters

 The classical assumed that the model should be linear in the

parameters regardless of whether the variables are linear or not.

 Linear in the parameters means that the parameters are not

multiplied together, divided, squared or cubed etc.
E.g. 1: 𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜀𝑖 is linear in both parameters & variables.
Eg. 2:. 𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝐼 + 𝜀𝑖 is non-linear in parameters.
1

2) The explanatory variable (X) are non-stochastic, but variability in value is necessary
3) No model specification and measurement error:

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 ..cont’d

4) E(𝜀𝑖 )=0 Zero Mean

5) var(𝜀𝑖 )=σ2 , Homoskedasticity =constant variance of error term

6) E(𝜀𝑖 𝜀𝑗 )=0 No serial correlation

 No serial correlation implies no observable relationship or pattern between the current
value of a variable and its past value.

7) E(X,i 𝜀𝑖 )=0 No Endogeneity

 Endogeneity occurs when the explanatory (X) may be influenced by the
explained (Y) or both may be jointly influenced by an unmeasured third.
So we assume no endogeneity issues.
8) 𝜀𝑖 ~N(0, σ2 ), normal distribution of the error term
 Time Invariability of the Unknown Parameters (β0, β1 ,…, βk, σ2)

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 2.1.2. Methods of Estimations(Ordinary Least square)

 After specifying the model and set their assumptions, the next step is estimation of

the parameters.

 The most widely used methods of estimation are:

 Methods of ordinarily Least Squares (OLS)

 Method of Maximum Likelihood

 Method of moments

 However, OLS is used as a default in regression b/c it is intuitively appealing &

mathematically much simpler than other methods & has some desirable properties.

 The principle of least squares is based on the residuals(𝜀𝑖 ) which play an

important role in the regression diagnostics.
 Therefore, the principle of least squares is to estimate the regression line, which
𝑛 2
minimizes the sum of squared residuals( 𝜀
𝑖=1 𝑖 ).
( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 ..cont’d Deriving the OLS Estimators

 Recall the SRF, since the PRF is not directly observable: 𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜀𝑖

 The the residuals: 𝜀𝑖 = 𝑌𝑖 − 𝛽0 − 𝛽1 𝑋𝑖

 The squared sum of residual is given by

𝑛 2 𝑛
𝑖=1 𝜀𝑖 = 𝑖=1(𝑌𝑖 − 𝛽0 − 𝛽1 𝑋𝑖 )2
𝑛 2
 OLS estimate the coefficients (𝛽0 and 𝛽1 ) in such a way that 𝑖=1 𝜀𝑖 is
minimum.
𝑛 2
 To find the values( 𝛽0 and 𝛽1 ) that minimize 𝑖=1 𝜀𝑖 , apply first order
conditions (FOC) w. r. t. 𝛽0 and 𝛽1 .

 The necessary condition for intercept term (β0 ):

𝑛 2 n
𝜕 𝜀
𝑖=1 𝑖
=0  2 Yi − β0 − β1 Xi −1 = 0
𝜕β0
i=1

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 …cont’d
n n
− i=1 Yi + nβ0 + β1 i=1 X i = 0 , divided both side by 2
n n

Yi = nβ0 + β1 Xi
i=1 i=1
𝑌 = β0 + β1 X
𝛃𝟎 = 𝐘 − 𝛃𝟏 𝐗 (2.3)
 The necessary condition for slope coefficient β1 (continued):
𝑛 2 n
𝜕 𝑖=1 𝜀𝑖
=0  2 Yi − β0 − β1 Xi −Xi = 0
𝜕β1 i=1
n n n

− Xi Yi + β0 Xi + β1 Xi 2 = 0
i=1 i=1 i=1
n n n

Xi Yi = β0 Xi + β1 Xi 2 (2.4)
i=1 i=1 i=1
 Then, substitute equations (2.3) into equation (2.4) and obtained:
n n n

Xi Yi = 𝑌 − β1 X Xi + β1 Xi 2
i=1 i=1 i=1

(
Econometric for finance by Teshome H. MSc in Econometrics ) October 3, 2023
 …cont’d
n n n n

Xi Yi = 𝑌 Xi − β1 X Xi + β1 Xi 2
i=1 i=1 i=1 i=1
n n n n

Xi Yi = 𝑌 Xi + β1 Xi 2 − β1 X Xi
i=1 i=1 i=1 i=1
n n n n

Xi Yi − Y Xi = β1 Xi 2 − X Xi
i=1 i=1 i=1 i=1

n n n
i=1 Xi Yi −Y i=1 Xi i=1 Xi Yi −nYX
β1 = n X 2 −X n X = n X 2 −nX2 , or
i=1 i i=1 i i=1 i

n n n
i=1 Xi Yi −nYX i=1 Xi −X Yi −Y i=1 xi yi Cov X,Y
β1 = n X 2 −nX2 = n 2 = n x 2 (2.5) Or β1 =
i=1 i i=1 Xi −X i=1 i Var X

 Where yi = Yi − Y & xi = X i − X, they are in deviation form

 Interpretation of the coefficients: β1 measures the partial effect of the
regressor X on Y. E.g. β1 = 0.36, a one unit change in the explanatory
variable result in a change of the dependent variable by 0.36.

(
Econometric for finance by Teshome H. MSc in Econometrics ) October 3, 2023
 2.1.3. Properties of OLS estimators

 Under the Gauss-Markov Theorem. the least squares estimators (OLS) should

satisfy Best, Linear, and Unbiased Estimator (BLUE) properties.

Property 1: Linearity: least squares estimators(𝛽0 & 𝛽1 ) should be linear estimator

of the true (population) value of 𝛽0 & 𝛽1 .

Property 2:Unbiasedness: least squares estimators (𝛽0 & 𝛽1 ) should be on average

equal to the true (population) parameters 𝛽0 & 𝛽1 .

Mathematically,
𝐸 β1 = β1

Similarly, 𝐸 β0 = β0

Here, ‘E’ is the expectation operator

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 ..cont’d
Property 3: Best: Minimum Variance
 The efficient property of any estimator says that the estimator is
the minimum variance unbiased estimator.
 OLS estimators have the least variance among the class of all
linear unbiased estimators.
 M𝐚𝐭𝐡𝐞𝐦𝐚𝐭𝐢𝐜𝐚𝐥𝐥𝐲
𝜎2 ∗
var β1 = 2 < 𝑣𝑎𝑟 𝛽1
𝑥𝑖

𝑋𝑖2 ∗
var β0 = 𝜎 2 2 < 𝑣𝑎𝑟 𝛽0
𝑛 𝑥𝑖

Where : 𝑥𝑖 = 𝑋𝑖 − 𝑋, in deviation form and 𝑋𝑖 is in observed form, 𝛽0∗ & 𝛽1∗ are an
estimator obtained by other than OLS and hence, 𝑣𝑎𝑟 𝛽0∗ & 𝑣𝑎𝑟 𝛽1∗ are their variance,
respectively.
( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 .cont’d
Property 4: Consistency
 An estimator is said to be consistent if its value approaches
the actual, true parameter (population) value as the sample
size increases.
 An estimator is consistent if it satisfies two conditions:
 It is asymptotically unbiased

 Its variance converges to zero as the sample size increases.

Mathematically,

Plim 𝛽 ~ 𝛽
𝑛→∞

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 …cont’d
The variance of error term(𝝈𝟐 )
 It is necessary to estimate the variance of the errors terms(𝜎 2 ) from sample counterpart
𝜀𝑖2
because the population error term is not directly observable.𝜎𝜀2 = 𝑛−2 = 𝑆𝑆𝑅 𝑛 − 2

where SSR denotes the sum of the squared residuals and K is the number of regressors
(including the constant).
 Note: The quantity 𝜎𝜀 is also called the Standard Error (S.E.) of the regression or the
Root Mean Squared Error (RMSE).

Thus, the reliability or precision of the estimators (β0 &β1 ) is given by its
standard error:
𝑋𝑖2
SE β0 = 𝜎𝜀
𝑛 𝑥𝑖2

1
SE β1 = 𝜎𝜀
𝑥𝑖2

 The smallest standard error implies, the more reliability of the estimator .
(
Econometric for finance by Teshome H. MSc in Econometrics ) October 3, 2023
 Application example for simple regression

 Given the following a sample data on consumption expenditure denoted by ‘Y’

(dependent variable) and income denoted by ‘X’ (independent variable) as
follows:
obs Yi Xi
1 10 30
2 20 50
3 30 60

 A) find a simple linear regression coefficient(β0 & β1 ) or function; Y = f(X)

 B) Interpret your result.
 C) Predict the value of Y when X is 45.

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 ..cont’d
Solution
A) To fit the regression equation we do the following computations.
n Yi Xi Yi Xi Xi2
1 10 30 300 900
2 20 50 1000 2500
3 30 60 1800 3600
Sum =60 =140 3100 7000
Mean Y = 20 140
X=
3

n XY  ( X )( Y )
3(3100)  (140)(60)
ˆ1    0.64
3(7000)  (140) 2
n X 2
 ( X ) 2

ˆ0  Y  ˆ1 X  20  0.64(140 / 3)  10

 Thus, the fitted regression function is given by:

(
Econometric for finance by Teshome H. MSc in Econometrics ) October 3, 2023
 ..cont’d

B) Interpretation:
 The value of the intercept term,-10, implies that the value of the dependent
variable ‘Y’ is – 10 when the value of the explanatory variable is zero.

 The value of the slope coefficient (β1 ) is a measure of the marginal change
in the dependent variable ‘Y’ when the value of the explanatory variable
increases by one. That is as income increase by birr 1, consumption
expenditure increase by 0.64 cents. 3
C). Predict expenditure when income is birr 45.

= -10+(0.64)(45) =18.8
 That means when X assumes a value of 45, the value of Y on average is expected to
be 18.8.

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 2.2. Multiple Regression
 Motivation for multiple regression
 Incorporate more explanatory factors into the model

 Explicitly hold fixed other factors that otherwise would be in

 Allow for more flexible functional forms

 Thus, multiple linear regression (MLR) is a statistical technique that uses several
explanatory variables to predict the outcome of a response variable.

 The general multiple regression model can be written in the population as:

(2.1)

Econometric
20 ( )
for finance by Teshome H. MSc in Econometrics October 3, 2023
..cont’d

 Interpretation of the multiple regression model

By how much does the dependent variable change if the j-

th independent variable is increased by one unit, holding
all other independent variables and the error term constant

 The multiple linear regression model manages to hold the values of other
explanatory variables fixed even if, in reality, they are correlated with the
explanatory variable under consideration „Ceteris paribus“-interpretation

 It has still to be assumed that unobserved factors do not change if the

explanatory variables are changed

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
..cont’d

 Example: Family income and family consumption

Other factors

Family consumption Family income Family income squared

 Model has two explanatory variables: inome and income squared

 Consumption is explained as a quadratic function of income
 One has to be very careful when interpreting the coefficients:

By how much does consumption Depends on how

much income is
increase if income is increased by already there
one unit?

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 …cont’d
 In a simplified(vector-matrix) form equation (2.1) can be written as:

𝑦 = 𝑋𝛽 + 𝜀 (2.2)

where 𝑦 𝑛𝑥1 , dependent variables for n observations

X = 𝑛𝑥 denotes the number of regressors (including intercept

term).

𝛽 𝐾𝑥1 a vector of parameters

𝜀 𝑛𝑥1 vector of error terms

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 2.2.1. Assumptions of Multiple CLRM

 The assumptions are the same as in the simple regression, except

the assumption of Full rank condition/no multicollinearity.
Assumption 1: Linearity: The model is linear with respect to
the parameters 𝛽1, .., 𝛽K .
Assumption 2: Full rank condition or identification
 X is an n × K matrix with rank K.
Interpretation
 There is no exact relationship among any of the independent
variables in the model . No perfect Multicollinearity
 The columns of X are linearly independent(no singularity).

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 …cont’d
Assumption 3: Exogeneity
 The regressors X1 , X2 , . . , Xk are distributed independently of the random
disturbance term
i.e. 𝐸 𝜀 𝑋 = 𝐸 𝜀 = 0
Implications
 The explanatory variables are not predictors of the error terms.

 If the strict erogeneity condition satisfied, so that 𝐸(𝑌/𝑋) = 𝑋𝛽

Assumption 4: Spherical error terms(homoscedasticity + non-autocorrelation)

 1. 𝑉𝑎𝑟 𝜀𝑖 𝑋 = 𝐸 𝜀𝑖 2 = 𝜎𝜀2 ,

 2. 𝑐𝑜𝑣 𝜀𝑖 𝜀𝑗 𝑋 = 𝐸 𝜀𝑖 𝜀𝑗 = 0, for i≠j

Notes:
 1 The condition of constant variances is called homoscedasticity.

 2 The uncorrelatedness across observations is called non-autocorrelation.

(
Econometric for finance by Teshome H. MSc in Econometrics ) October 3, 2023
 ..cont’d
 Comments: The two assumptions (homoscedasticity and nonautocorrelation)
imply that:
𝐸 𝜀−𝐸 𝜀 𝜀−𝐸 𝜀 ′ =𝐸 𝜀−0 𝜀−0 ′ =𝐸 𝜀 𝜀 ′

𝜀1 𝜀1 2 𝜀1 𝜀2 … 𝜀1 𝜀𝑛
𝜀2 𝜀2 𝜀1 𝜀2 2 … 𝜀2 𝜀𝑛
= 𝐸 . 𝜀1 𝜀2 . . 𝜀𝑛 =𝐸 ..
.
𝜀𝑛 𝜀𝑛 𝜀1 𝜀𝑛 𝜀2 … 𝜀𝑛 2

𝐸 𝜀1 2 𝐸 𝜀1 𝜀2 … 𝐸 𝜀1 𝜀𝑛 𝜎2 0 … 0
𝐸 𝜀2 𝜀1 𝐸 𝜀2 2 . . . 𝐸 𝜀2 𝜀𝑛 0 𝜎2 … 0 2𝐼
= = = 𝜎
.. ...
𝐸 𝜀𝑛 𝜀1 𝐸 𝜀𝑛 𝜀2 … 𝐸 𝜀𝑛 2 0 0 … 𝜎2
( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 ..cont’d

Assumption 5: The disturbance term 𝜀 is multivariate normality

distributed.
Given assumption 1-4, it is true that,
𝜀𝑖 ~𝑁𝐼𝐼𝐷(0, 𝜎 2 )
where NIID stands for normally independently and identically distributed.
Comments
 1 Normality is not necessary to obtain most of the results
presented below.
 2 Assumption 5 implies assumptions 3 (exogeneity) and 4
(spherical disturbances).
( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 2.2.2. Estimation of Multiple Regression Model (Using Matrix Approach)

 Since the PRF is not estimable, we use the sample counter

part as defined by
𝑌 = 𝑋𝛽 + 𝜀
The residual is, 𝜀 = 𝑌 − 𝑋𝛽 ,
 OLS estimator used by minimizing the sum of squared
residuals (SSR), 𝑛𝑖=1 𝜀𝑖2 = 𝜀12 + 𝜀22 + 𝜀32 + ⋯ + 𝜀𝑛2
𝜀1
𝜀2
= 𝜀1 𝜀2 . . 𝜀𝑛 . = 𝜀 ′𝜀
.
𝜀𝑛
 Therefore,
𝜀𝑖2 = 𝜀 ′𝜀 = (𝑌 − 𝑋𝛽 )′(𝑌 − 𝑋𝛽)
( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 …cont’d
𝜀′𝜀 = 𝑌 ′ 𝑌 − 𝛽′ 𝑋 ′ 𝑌 − 𝑌 ′ 𝑋𝛽 + 𝛽′ 𝑋 ′ 𝑋𝛽
 Since 𝛽 ′ 𝑋 ′ 𝑌 is scalar (1x1), it is equal to its transpose; i.e., =
𝜀′𝜀 = 𝑌′𝑌 − 2𝑌 ′ 𝑋𝛽 + 2𝛽′ 𝑋 ′ 𝑋𝛽
 Minimizing 𝜀 ′𝜀 with respect to the elements in 𝛽
𝜕 𝜀′𝜀
= −2𝑋 ′ 𝑌 + 2𝑋′𝑋𝛽
𝜕𝛽
Theorem 1: Assume 𝐴 is symmetric matrix, 𝐴 = 𝐴′ or +𝐴′ = 2𝐴 .
Then
𝜕 𝑋 ′ 𝐴𝑋 ′
 = 𝐴 + 𝐴 𝑋 = 2𝐴𝑋 = 2𝑋 ′ 𝐴
𝜕𝑋
𝜕𝑎′𝑋
 = 𝑎 , and
𝜕𝑋
𝜕 𝑌′𝑋 𝐵
 = 𝑋′𝑌
𝜕𝐵
𝜕𝛽′𝑋′𝑋𝛽
 Thus, = (𝑋 ′ 𝑋) + (𝑋′𝑋)′ 𝛽 = 2𝑋′𝑋𝛽, since 𝑋′𝑋 is
𝜕𝛽
symmetric, (𝑋 ′ 𝑋) = (𝑋′𝑋 )′, so (𝑋 ′ 𝑋) + 𝑋 ′ 𝑋 ′ = 2𝑋′𝑋
( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 ..cont’d

 Equating the expression to null vector 0, we obtain:

−2𝑋 ′ 𝑌 + 2𝑋 ′ 𝑋𝛽 = 0
 System of normal equation in matrix form:
𝑋 ′ 𝑋𝛽 = 𝑋 ′ 𝑌
 Assumption X has full column rank. This means no column of
X is a linear combination of the other columns. This in turn
implies 𝑋 ′ 𝑋 −1 exists.
 Then pre-multiplying both sides by this inverse gives us the
following equation.
𝜷 = 𝑿′ 𝑿 −𝟏
𝑿′ 𝒀 (2.3)
Econometric for finance by Teshome H. (MSc in Econometrics) October 3, 2023
 2.2.3. Statistical Properties of the OLS estimates (Matrix Approach)

 Just like simple linear regression, multiple regressions OLS estimators

also satisfy the BLUE property.

 Linearity: the sample estimate(𝛽 ) should be a linear estimator of the

population parameter (β)

 Unbiasedness: on average 𝛽 should be equal β. .i.e. 𝐸 𝛽 = 𝛽

 Best estimator (Minimum variance): The variance of OLS

estimator (𝑉𝑎𝑟 𝛽 = 𝜎𝜀2 𝑋 ′ 𝑋 −1 ) should be smallest as compared with
any other estimators.

𝐸 𝜀′ 𝜀
 Where 𝜎𝜀2 can be estimated from the sample variance, 𝜎 = 𝐸 𝜎 2 2
=
𝑛−𝑘

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 Application example
 Example: Le see the effect of debt and income on consumption
expenditure over 15 observations defend by a model :
𝑐𝑜𝑛𝑠𝑒𝑥𝑝𝑒𝑛𝑑 = 𝛽0 + 𝛽1 𝑑𝑒𝑏𝑡 + 𝛽2 𝑖𝑛𝑐𝑜𝑚𝑒 + 𝜀
 and the following data have been calculated from the
original X’s.

, 𝜀 ′ 𝜀 = 10.96

 A) Calculate the coefficient estimates and their standard errors?

Solution
 Recall from equation(2.3.)
𝛽 = 𝑋′𝑋 −1 𝑋 ′ 𝑌

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 ..cont’d
𝛽0 𝟏. 𝟏
= −𝟒. 𝟒
 𝛽 = 𝛽1 = 𝑋 ′ 𝑋 −1
𝑋′𝑌 = 𝟏𝟗. 𝟖𝟖
𝛽2
 To calculate the standard errors, we need an estimate of

 The variance-covariance matrix 𝛽 is

 The variances are on the leading diagonal:

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023
 ..cont’d
 . The fitted model with its standard error is given by:

 As we observed from the result, the debt coefficient is

negative, while the income coefficient is positive.
 Thus, a 1 birr increase in debt results a fall in consumption
expenditure by birr 4.4, assume income constant.
 A 1 birr increase in income results an increment in
consumption expenditure by birr 19.88, assume debt constant.

( )
Econometric for finance by Teshome H. MSc in Econometrics October 3, 2023