0% found this document useful (0 votes)
5 views

VAR in Time Series

The document provides an overview of Vector Autoregression (VAR) modeling, detailing its objectives, assumptions, and mathematical representation. VAR is a statistical model used to analyze dynamic relationships among multiple time series variables, treating all variables as endogenous. The document also discusses the key features of VAR models, their applications in various fields, and the conditions under which they are useful.

Uploaded by

Eugene Adu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
5 views

VAR in Time Series

The document provides an overview of Vector Autoregression (VAR) modeling, detailing its objectives, assumptions, and mathematical representation. VAR is a statistical model used to analyze dynamic relationships among multiple time series variables, treating all variables as endogenous. The document also discusses the key features of VAR models, their applications in various fields, and the conditions under which they are useful.

Uploaded by

Eugene Adu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 54

Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Time series Econometrics

Alexander Opoku

Department of Applied Economics

December 17, 2024

1 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Overview
1 Vector Autoregression (VAR)
2 Objectives of VAR Modeling
3 Assumptions of VAR Models
4 Mathematical Representation of VAR Models
5 Representing Equations in Matrix Form
6 VAR Model with One Predictor in Matrix Form
7 VAR Models: Higher Orders
8 Stationary VAR and Variable Selection
9 Wold’s Decomposition Representation
10 Wold’s Decomposition: Layman Explanation
11 Lag Selection Criteria

2 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Estimating GDP is an enormous task-doing so at county


levels is even tougher:

Due to data challenges. This means that GCP is the best


available measure of aggregate activity at the county but not
a perfect measure;
Prevalence of informalities. Many transactions or activities in
the economy, including many that go through the market are
never recorded officially; and
Difficulties in attributing economic activity to a specific
county unit. It gets worse when we start to account for
cross-county transactions.

3 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

What is VAR?
Vector Autoregression (VAR) is a statistical model used to
study the dynamic relationships among multiple time series
variables. It is particularly useful in analyzing systems where:
Variables influence each other over time.
Past values (lags) of one variable can affect the present and
future values of another.
Example:
Consider a macroeconomic system with GDP, inflation, and
interest rates:
How does a rise in interest rates today impact GDP and
inflation in the future?
How do past inflation rates influence current interest rates?
Key Feature: Every variable in the system is treated as
endogenous (both influenced by and influencing other variables).
This makes VAR suitable for uncovering complex interdependencies
in time series data.
4 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Mathematical Representation of VAR


The VAR model for k variables and p lags is expressed as:
p
X
yt = c + Ai yt−i + ut
i=1
Explanation of Terms:
yt : A k × 1 vector of variables at time t (e.g., GDP, inflation,
interest rates).
c: A k × 1 vector of constants (baseline levels of the
variables).
Ai : A k × k matrix of coefficients for lag i, capturing how
past values of the variables affect the current values.
ut : A k × 1 vector of error terms (white noise), representing
unexplained shocks or randomness.
Example: For a system with two variables (GDP and inflation),
the equations might look like:
5 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Key Features of VAR Models

Endogeneity: All variables are treated symmetrically,


meaning each variable is both influenced by and influences
other variables.
Lagged Structure: Captures how past values of variables
affect their current and future behavior.
Flexibility: No need to predefine dependent or independent
variables, allowing the data to determine relationships.
Dynamic Feedback: Models feedback effects, such as:
How GDP changes in response to past inflation.
How inflation reacts to previous changes in interest rates.
Shock Analysis: Enables analysis of how unexpected events
(e.g., an oil price spike) ripple through a system over time.

6 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Objectives of VAR Modeling

What is the Purpose of a VAR Model?


To analyze the dynamic relationships and feedback effects
between multiple time series variables.
To predict future values of these variables based on their
historical patterns and interactions.
To evaluate the impact of shocks to one variable on other
variables over time using Impulse Response Functions
(IRFs).
To perform Variance Decomposition to understand how
much variability in one variable is explained by others.

7 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Why Use VAR Models?


Versatility: Suitable for a wide range of fields, including
economics, finance, and social sciences.
Forecasting: Effective for predicting future trends based on
historical patterns.
Dynamic Relationships: Provides insights into how variables
interact over time.
Example of Use:
Macroeconomics: Studying the impact of monetary policy
changes (e.g., interest rate hikes) on inflation and
unemployment.
Finance: Understanding how past stock prices influence bond
yields and exchange rates.
Policy Evaluation: Helps policymakers evaluate the effects of
fiscal policies, such as government spending or tax cuts, on
economic growth.
8 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Applications of VAR Models

Macroeconomic Analysis:
Analyze interactions between GDP, inflation, interest rates,
and unemployment.
Example: How does a central bank’s interest rate hike impact
GDP and inflation over the next few quarters?
Finance:
Study relationships between asset prices, exchange rates, and
market indices.
Example: How do fluctuations in the SP 500 affect bond yields
or currency values?
Policy Analysis:
Evaluate the effects of government interventions, such as
stimulus packages or tax reforms.
Example: What is the impact of fiscal stimulus on
unemployment and consumer spending?
9 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

When is VAR Useful?


Conditions for Using VAR:
Interdependent Variables: When variables are
interdependent and influence each other over time.
Dynamic Relationships: When capturing feedback effects
between variables is important.
Exploratory Analysis: When the relationships between
variables are unclear and you want the data to reveal the
dynamics.
Examples of Use:
Analyzing how GDP, inflation, and interest rates interact
dynamically.
Examining how trade shocks affect economic growth and
exchange rates.
Analyzing the effects of monetary policy shocks on financial
markets.
10 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Overview of VAR Model Assumptions

Assumptions of VAR Models:


Linearity: The relationships between variables are linear.
Stationarity: All variables in the system are stationary.
No Serial Correlation: The residuals (errors) are not serially
correlated.
Homoscedasticity: The residuals have constant variance over
time.
Endogeneity: All variables are treated as endogenous (no
strict causal ordering is required).
Why These Assumptions Matter:
Violations can lead to biased estimates, unreliable forecasts,
and incorrect conclusions about relationships.

11 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Assumption 1: Linearity

Definition:
The VAR model assumes that the relationships between
variables are linear. Each variable is explained as a weighted
sum of its own past values and the past values of other
variables.
Example:

GDPt = c1 + a11 · GDPt−1 + a12 · Inflationt−1 + u1t

Inflationt = c2 + a21 · GDPt−1 + a22 · Inflationt−1 + u2t


Here, GDP and inflation are linearly related to their past values.

12 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Assumption 2: Stationarity

Definition:
The mean, variance, and autocovariance of each variable are
constant over time.
A non-stationary series (e.g., one with trends or changing
variances) can lead to spurious relationships.
Example:
Stationary: GDP growth rate (percent change in GDP) is
usually stationary.
Non-stationary: The level of GDP often exhibits trends and
needs to be differenced to achieve stationarity.

13 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Assumption 3: No Serial Correlation

Definition:
The residuals (errors) of the VAR model should not exhibit
serial correlation.
Serial correlation indicates that residuals from one time period
are correlated with residuals from another period.
Example:
If GDP residuals are serially correlated, this implies missing
variables or inadequate lag selection.

14 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Assumption 4: Homoscedasticity

Definition:
The variance of the residuals should remain constant over
time.
If the residual variance changes (heteroscedasticity), it can
lead to inefficient estimates.
Example:
In financial data, stock returns often exhibit
heteroscedasticity, with periods of high volatility followed by
periods of low volatility.

15 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Assumption 5: Endogeneity

Definition:
All variables in the VAR system are treated as endogenous,
meaning each variable is both influenced by and influences
other variables.
No variable is strictly dependent or independent.
Example:
GDP affects inflation, but inflation also affects GDP, creating
a feedback loop.
Benefit:
This assumption allows VAR to capture mutual dependencies
and dynamic interactions without imposing restrictive causal
assumptions.

16 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Multivariate Regression
Definition: Multivariate regression models explain the relationship
between multiple dependent variables and one or more predictors.
General Form:
Y = XB + E
Where:
Y: A n × m matrix of m dependent variables (e.g., sales of
different products),
X: A n × p matrix of predictors (e.g., advertising expenditure,
seasonality),
B: A p × m matrix of coefficients,
E: A n × m matrix of residuals (errors).
Key Feature: Multivariate regression accounts for correlations
between the dependent variables, allowing for more robust
modeling compared to multiple separate regressions.
17 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Multivariate Regression with One Predictor

Case: When there is only one predictor, the model simplifies to:

Y = XB + E

Where X is a single column vector.


Example:
 
y
Suppose Y = 1t represents sales of apples (y1t ) and
y2t
oranges (y2t ) over time, and X represents advertising spend.
The regression explains how advertising influences sales of
both apples and oranges.

18 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Multivariate Regression with Two or More Predictors

Case: With multiple predictors, the model becomes:

Y = X1 B1 + X2 B2 + · · · + E

Example:
 
y
Y = 1t represents sales of apples (y1t ) and oranges (y2t ),
y2t
Predictors X1 = advertising spend and X2 = pricing strategy.
The regression accounts for how advertising and pricing affect
sales simultaneously.

19 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Basic VAR Model with One Predictor

Definition: The simplest VAR model (VAR(1)) explains the


current value of a variable using its own past value (lagged) and
the lagged values of other variables in the system.
Form:
y1t = c1 + a11 y1t−1 + a12 y2t−1 + u1t
y2t = c2 + a21 y1t−1 + a22 y2t−1 + u2t
Apples-Oranges Example:
y1t = sales of apples today, y2t = sales of oranges today.
y1t−1 = sales of apples last week, y2t−1 = sales of oranges
last week.
Interrelation: Higher sales of oranges last week might indicate
greater overall market demand, boosting both orange and
apple sales today.

20 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Matrix Representation of VAR(1) Model


Matrix Form of VAR(1):

yt = c + A1 yt−1 + ut

Where:  
y
yt = 1t : vector of variables at time t,
y2t
 
c
c = 1 : vector of constants,
c2
 
a11 a12
A1 = : coefficient matrix,
a21 a22
 
u
ut = 1t : vector of white noise error terms.
u2t
Compact Form:
yt = c + A1 yt−1 + ut
21 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

What is a Matrix?

Definition: A matrix is a rectangular array of numbers arranged in


rows and columns. It is used to compactly represent data, systems
of equations, or relationships.
General Notation:
 
a11 a12 · · · a1n
 a21 a22 · · · a2n 
A= .
 
. .. . . .. 
 . . . . 
am1 am2 · · · amn

Where:
aij : Entry in row i and column j.
m: Number of rows.
n: Number of columns.
22 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Representing One Equation in Matrix Form


Scalar Form:
y = βx + ϵ
Matrix Form:
y = Xβ + ϵ
Where:  
y1
y2 
y =  . : Dependent variable.
 
 .. 
yn
 
x1
 x2 
X =  . : Predictor (independent variable).
 
 .. 
xn
23 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

β: Coefficient.
 
ϵ1
ϵ2 
ϵ =  . : Residuals.
 
 .. 
ϵn

24 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Representing Multiple Equations in Matrix Form


System of Equations:

y1 = β1 x + ϵ1

y2 = β2 x + ϵ2
Matrix Form:
Y = XB + E
Where:
 
y11 y12
y21 y22 
Y= . .. : Dependent variables (e.g., sales of Apples and
 
 .. . 
yn1 yn2
Oranges).

25 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Representing Multiple Equations in Matrix Form

 
x1
x2 
X =  . : Common predictor (e.g., advertising).
 
 .. 
xn
 
β
B = 1 : Coefficients.
β2
 
ϵ1
ϵ2 
E =  . : Residuals.
 
 .. 
ϵn

26 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Basic Form of VAR Model (VAR(1))


Definition: The simplest VAR model (VAR(1)) uses the current
value of each variable as a function of its own lagged value and the
lagged values of other variables in the system.
System of Equations:
y1t = c1 + a11 y1t−1 + a12 y2t−1 + u1t
y2t = c2 + a21 y1t−1 + a22 y2t−1 + u2t
Where:
y1t , y2t : Current values of variables (e.g., sales of Apples and
Oranges today).
y1t−1 , y2t−1 : Lagged values of variables (e.g., sales of Apples and
Oranges last week).
c1 , c2 : Constants (intercepts).
u1t , u2t : White noise error terms. 27 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Matrix Form of VAR(1) Model


Compact Representation:
yt = c + A1 yt−1 + ut
Expanded Form:
        
y1t c a a y1t−1 u
= 1 + 11 12 + 1t
y2t c2 a21 a22 y2t−1 u2t
Where:
yt : Vector of current values (y1t , y2t ).
c: Vector of constants (c1 , c2 ).
A1 : Coefficient matrix.
yt−1 : Vector of lagged values (y1t−1 , y2t−1 ).
ut : Vector of white noise error terms.
28 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Apples-Oranges Example for VAR(1)

Scenario: - y1t : Sales of Apples today. - y2t : Sales of Oranges


today. - y1t−1 : Sales of Apples last week. - y2t−1 : Sales of
Oranges last week.
System of Equations:

Applest = c1 + a11 · Applest−1 + a12 · Orangest−1 + u1t

Orangest = c2 + a21 · Applest−1 + a22 · Orangest−1 + u2t


Matrix Form:
        
Applest c1 a11 a12 Applest−1 u
yt = = + + 1t
Orangest c2 a21 a22 Orangest−1 u2t

29 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

VAR(2) Model
Question: How do the current values of variables depend on their
lagged values from one and two periods ago?
System of Equations:
(1) (1) (2) (2)
y1t = c1 + a11 y1t−1 + a12 y2t−1 + a11 y1t−2 + a12 y2t−2 + u1t ,
(1) (1) (2) (2)
y2t = c2 + a21 y1t−1 + a22 y2t−1 + a21 y1t−2 + a22 y2t−2 + u2t .

Matrix Form:
    " (1) (1) #   " (2) (2) #    
y1t c1 a11 a12 y1t−1 a11 a12 y1t−2 u
yt = = + (1) (1) + (2) (2) + 1t .
y2t c2 a21 a22 y2t−1 a21 a22 y2t−2 u2t

Compact Form:

yt = c + A1 yt−1 + A2 yt−2 + ut

30 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

VAR(3) Model
System of Equations:
(1) (1) (2) (2)
y1t = c1 + a11 y1t−1 + a12 y2t−1 + a11 y1t−2 + a12 y2t−2
(3) (3)
+ a11 y1t−3 + a12 y2t−3 + u1t ,
(1) (1) (2) (2)
y2t = c2 + a21 y1t−1 + a22 y2t−1 + a21 y1t−2 + a22 y2t−2
(3) (3)
+ a21 y1t−3 + a22 y2t−3 + u2t .
Matrix Form:
    " (1) (1) #   " (2) (2) #   " (3)
y1t c1 a11 a12 y1t−1 a11 a12 y1t−2 a a
yt = = + (1) (1) + (2) (2) + 11(3)
y2t c2 a21 a22 y2t−1 a21 a22 y2t−2 a21 a
Compact Form:

yt = c + A1 yt−1 + A2 yt−2 + A3 yt−3 + ut


31 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

VAR(p) Model
System of Equations:
p  
(i) (i)
X
y1t = c1 + a11 y1t−i + a12 y2t−i + u1t ,
i=1
p  
(i) (i)
X
y2t = c2 + a21 y1t−i + a22 y2t−i + u2t .
i=1
Matrix Form:
p
" #
    X (i) (i)    
y1t c1 a11 a12 y1t−i u
yt = = + (i) (i) + 1t .
y2t c2 a21 a22 y2t−i u2t
i=1
Compact Form:
p
X
yt = c + Ai yt−i + ut
i=1
32 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Summary of VAR Models

VAR(2): Includes lagged values from periods t − 1 and t − 2.

yt = c + A1 yt−1 + A2 yt−2 + ut

VAR(3): Includes lagged values from periods t − 1, t − 2, and


t − 3.
yt = c + A1 yt−1 + A2 yt−2 + A3 yt−3 + ut

VAR(p): Includes lagged values from periods t − 1 to t − p.


p
X
yt = c + Ai yt−i + ut
i=1

33 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

What is Stationary VAR?


Definition: A stationary VAR assumes that all variables maintain
constant statistical properties (mean, variance, autocovariance)
over time.
Why is Stationarity Important?
Non-stationary variables can lead to misleading results.
Stationary data ensures meaningful relationships in the VAR model.

yt = c + A1 yt−1 + ut
Example:
GDPt : GDP growth rate (stationary).
INFt : Inflation rate (stationary).
The VAR(1) model captures how GDPt and INFt influence each
other.
34 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Common Tests:

ADF Test, Null hypothesis: Variable has a unit root


(non-stationary).

PP Test: Adjusts for serial correlation and heteroscedasticity.

KPSS Test: Null hypothesis: Variable is stationary.

Example:

GDPt (non-stationary GDP levels) becomes ∆GDPt (stationary


GDP growth).

Use VECM if variables are cointegrated. A VECM is used when


variables are non-stationary but cointegrated, meaning they share a
long-term equilibrium relationship. Note: VECM modeling will be
covered in detail in the cointegration section of the course.

35 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Criteria for Variable Selection:


1 Economic Relevance: Include variables with plausible
relationships. Example: GDP, INF , IR (interest rate).
2 Stationarity: Variables must be stationary.
3 Data Quality: Ensure high-quality, consistent time series data.
4 Granger Causality: Include variables that are Granger-caused or
Granger-cause others. Granger causality tests whether past values
of one variable (X ) provide significant information for predicting
another variable (Y ). Note: Detailed testing procedures and
interpretation will be discussed later in the course.

36 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Scenario: Analyzing monetary policy effects on GDP growth.


Step-by-Step:
1 Identify Relevant Variables: GDPt : GDP growth, INFt : Inflation
rate, IRt : Interest rate.
2 Check Stationarity: Perform ADF test; take differences if
non-stationary.
3 Determine Lag Length: Use AIC or BIC criteria. Note: The
calculation and application of AIC/BIC will be explored in detail
later in the course.

Selected Variables:  
GDPt
yt =  INFt 
IRt

37 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Mathematical Representation of VAR(1)

VAR(1) Model:
yt = c + A1 yt−1 + ut
Expanded Form:

GDPt = c1 + a11 GDPt−1 + a12 INFt−1 + a13 IRt−1 + u1t ,


INFt = c2 + a21 GDPt−1 + a22 INFt−1 + a23 IRt−1 + u2t ,
IRt = c3 + a31 GDPt−1 + a32 INFt−1 + a33 IRt−1 + u3t .

Matrix Form:
        
GDPt c1 a11 a12 a13 GDPt−1 u1t
yt = INFt = c2 + a21 a22 a23
       INFt−1 + u2t  .
 
IRt c3 a31 a32 a33 IRt−1 u3t

38 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Wold’s Decomposition: Introduction


What is Wold’s Decomposition?
The Wold Decomposition Theorem states that any
covariance-stationary time series can be expressed as:

X
yt = µ + ψi εt−i + dt
i=0

This means that:


1 yt is composed of a deterministic component (µ + dt ).
P∞
2 Plus a stochastic component ( i=0 ψi εt−i ) capturing the effects of past
shocks.

Wold’s decomposition helps us analyze the dynamic structure of a


stationary VAR model.
39 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

VAR(p) Model and Lag Operator Notation


Vector Autoregressive (VAR) Model:
yt = c + A1 yt−1 + A2 yt−2 + · · · + Ap yt−p + ut
Where:
yt : Vector of endogenous variables.
Ai : Coefficient matrices.
ut : Vector of white noise errors.
Lag Operator Notation:
A(L)yt = c + ut
Where A(L) is the lag polynomial:
A(L) = (I − A1 L − A2 L2 − · · · − Ap Lp )yt = c + ut
Lagr polynomial form A(L)yt = c + ut
40 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Stationarity and Invertibility


Stationarity Condition:
For the VAR model to be stationary, the lag polynomial A(L) must
be invertible.
This occurs when the roots of the determinant of A(L) lie outside
the unit circle:
det(A(L)) ̸= 0 for |z| ≤ 1

Implication of Stationarity:
If stationary, the VAR can be rewritten as a Moving Average (MA)
process:

X
yt = µ + Ψi ut−i
i=0
Where Ψi are moving average coefficients derived from A(L)−1 .
41 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Deriving the MA Representation


From Lag Polynomial to MA Representation:

yt = c + A(L)−1 ut

Expand A(L)−1 :

A(L)−1 = (I + Ψ1 L + Ψ2 L2 + . . .

) ut
Substitute into the equation:

yt = c + ut + Ψ1 ut−1 + Ψ2 ut−2 + . . .

Interpretation:
yt is expressed as an infinite sum of past shocks ut−i .
Ψi : Matrices capturing how shocks propagate over time.
42 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Example: Inflation Time Series

Scenario: Inflation (INFt ) modeled as:

INFt = 0.8 · INFt−1 + εt

Decomposition:

INFt = εt + 0.8εt−1 + 0.82 εt−2 + · · ·

Interpretation:

Current inflation depends on an infinite weighted sum of past


shocks.

Recent shocks (εt ) have greater influence than older shocks (εt−2 ).

43 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Practical Implications of Wold’s Decomposition

Applications:

Impulse Response Functions: Trace the effect of a one-time


shock to one variable on others.

Variance Decomposition: Quantify the contribution of each


shock to the variance of forecast errors.

Forecasting: Use the MA representation to predict future values.

44 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Wold’s Decomposition: Basic lens


Imagine you’re tracking daily sales of apples.
The number of apples sold today depends on two things:
1 Predictable patterns (deterministic component): Weekends are
busier, and you sell more apples.
2 Unpredictable events (stochastic component): Sudden rainstorms or
promotions affect sales unexpectedly.

Wold’s Decomposition separates these two components.


What does it tell us?
The average sales (predictable trend).
The impact of past events (unpredictable shocks) on today’s
sales.
How far back past events continue to affect the present.
45 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Example: Apple Sales


Scenario: Your daily apple sales (yt ) follow this pattern:
yt = 100 + 0.6εt−1 + 0.3εt−2 + εt
Breaking it down:
100: Average sales per day.
εt : Today’s unexpected shock (e.g., sudden rain or a promotion).
0.6εt−1 : Yesterday’s shock still affects sales, but only 60
0.3εt−2 : The shock from two days ago affects sales by 30
Key Idea:
Today’s sales are influenced by both predictable trends and past
random events.
The impact of past shocks fades over time.
46 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Visualizing Wold’s Decomposition

Think of it like ripples in a pond:

When you throw a stone (today’s shock), it creates ripples.

Larger stones create bigger ripples (stronger shocks).

Closer ripples (recent shocks) have more impact than ripples from
stones thrown further back (older shocks).

What Wold’s Decomposition Does:

Measures how much past shocks (ripples) affect today.

Separates predictable patterns (like busy weekends) from random


events (like promotions).

47 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Why is Wold’s Decomposition Useful?


Applications in Real Life:
Understanding Patterns: Separate predictable events (e.g.,
weekends) from unpredictable ones (e.g., weather).
Forecasting: Use past trends and events to predict future sales or
outcomes.
Measuring Impact: Analyze how far back past events still
influence today’s outcomes.
Applications:
Impulse Response Functions: Trace the effect of a one-time
shock to one variable on others.
Variance Decomposition: Quantify the contribution of each
shock to the variance of forecast errors.
Forecasting: Use the MA representation to predict future values. 48 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Lag Selection Criteria: Introduction


What is Lag Selection?
In time series econometrics, choosing the optimal lag length is
crucial for model accuracy.
Lag selection criteria help determine the number of lagged terms
to include in a model (e.g., VAR, ARIMA).
Commonly used criteria:
1 Akaike Information Criterion (AIC)
2 Schwarz Information Criterion (SIC or BIC)
3 Hannan-Quinn Criterion (HQC)

Why is it important?
Ensures that the model is neither underfitted nor overfitted.
Helps balance goodness of fit and model complexity. 49 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Akaike Information Criterion (AIC)


Definition:

AIC measures the trade-off between the goodness of fit and the
complexity of the model.

Formula:  
RSS 2k
AIC = ln +
n n
Where: RSS: Residual Sum of Squares, n: Sample size, k:
Number of parameters (including intercept)

Interpretation:

The model with the lowest AIC value is preferred.

Penalizes models with more parameters to avoid overfitting.


50 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Schwarz Information Criterion (SIC or BIC)


Definition:
SIC (also known as BIC) adds a harsher penalty for model
complexity compared to AIC.
Formula:  
RSS k ln(n)
SIC = ln +
n n
Where: ln(n): Logarithm of the sample size acts as a stronger
penalty factor.
Interpretation:
The model with the lowest SIC value is preferred.
Useful for large datasets due to its stronger penalty for additional
parameters.
51 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Hannan-Quinn Criterion (HQC)


Definition:

HQC is a middle ground between AIC and SIC in terms of


penalizing complexity.

Formula:  
RSS 2k ln(ln(n))
HQC = ln +
n n
Where: ln(ln(n)): Logarithmic term reduces the penalty compared
to SIC.

Interpretation:

The model with the lowest HQC value is preferred.

Often used when neither AIC nor SIC provides a clear choice.
52 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Comparison of Criteria
Key Points:
All three criteria balance model fit and complexity but differ in
their penalties for additional parameters.
AIC: Less harsh penalty, may favor slightly overfitted models.
SIC/BIC: Harsher penalty, suitable for large datasets.
HQC: Moderate penalty, balances between AIC and SIC.
Selection Rule:
Choose the model with the lowest value of the respective criterion.
In software outputs (e.g., Stata, EViews, R):
AIC, SIC, and HQC are reported for various lag lengths.
The lag length corresponding to the lowest criterion value is optimal.
53 / 54
Vector Autoregression (VAR) Objectives of VAR Modeling Assumptions of VAR Models Mathematical Representation of VAR Mod

Example in Practice

Scenario: You are analyzing monthly GDP growth data using a


VAR model.

Using lag selection criteria:

Lag Length AIC SIC


1 12.345 13.456
2 11.789 12.345
3 12.012 13.123

AIC selects lag 2 (lowest value: 11.789), SIC selects lag 1 (lowest value:
13.456).

54 / 54

You might also like