Scribd Logo
Upload

Welcome to Scribd!

  • 43 views

    Stationarity TS PDF

    Uploaded by

    Justina Sasnauskaite
    The document discusses the stationarity condition for an AR(p) (autoregressive model of order p) time series model. It states that an AR(p) process is stationary if the roots of the characteristic polynomial, the inverse roots, and the eigenvalues of the companion matrix are all inside the unit circle. It demonstrates how to determine stationarity from the eigenvalues and links this to the roots of the characteristic polynomial. Finally, it provides an example of checking for stationarity in an AR(2) model of real house prices in Denmark.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Stationarity TS PDF

    Uploaded by

    Justina Sasnauskaite
    43 views24 pages
    The document discusses the stationarity condition for an AR(p) (autoregressive model of order p) time series model. It states that an AR(p) process is stationary if the roots of the characteristic polynomial, the inverse roots, and the eigenvalues of the companion matrix are all inside the unit circle. It demonstrates how to determine stationarity from the eigenvalues and links this to the roots of the characteristic polynomial. Finally, it provides an example of checking for stationarity in an AR(2) model of real house prices in Denmark.

    Original Title

    Stationarity TS.pdf

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document discusses the stationarity condition for an AR(p) (autoregressive model of order p) time series model. It states that an AR(p) process is stationary if the roots of the characteristic polynomial, the inverse roots, and the eigenvalues of the companion matrix are all inside the unit circle. It demonstrates how to determine stationarity from the eigenvalues and links this to the roots of the characteristic polynomial. Finally, it provides an example of checking for stationarity in an AR(2) model of real house prices in Denmark.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    43 views24 pages

    Stationarity TS PDF

    Uploaded by

    Justina Sasnauskaite
    The document discusses the stationarity condition for an AR(p) (autoregressive model of order p) time series model. It states that an AR(p) process is stationary if the roots of the characteristic polynomial, the inverse roots, and the eigenvalues of the companion matrix are all inside the unit circle. It demonstrates how to determine stationarity from the eigenvalues and links this to the roots of the characteristic polynomial. Finally, it provides an example of checking for stationarity in an AR(2) model of real house prices in Denmark.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 24

    university of copenhagen department of economics

    Econometrics II
    What is the Stationarity Condition
    in an AR(p) Model?
    Morten Nyboe Tabor
    university of copenhagen department of economics

    Outline: What is the Stationarity Condition in an AR(p) Model?

    1 What is the Stationarity Condition?

    2 Stationarity in Terms of the Eigenvalues of the Companion Matrix

    3 The Link Between the Roots of the Characteristic Polynomial and the
    Eigenvalues of the Companion Matrix

    4 ARMA Models and ARIMA Models

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 2/24
    university of copenhagen department of economics

    The Stationarity Condition

    • MA(q):
    yt = µ + εt + α1 εt−1 + α2 εt−2 + ... + αq εt−q .
    The MA(q) process is stationary by construction (sum of stationary
    terms).

    • MA(∞):
    yt = µ + εt + α1 εt−1 + α2 εt−2 + ...
    P∞
    The MA(∞) process is stationary if i=0
    αi2 < ∞.

    • AR(1):
    yt = δ + θyt−1 + εt .
    The AR(1) process is stationary if |θ| < 1.

    Note, that the stationarity condition is a restriction on the models’ parameters!

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 3/24
    university of copenhagen department of economics

    The Stationarity Condition in an AR(p) Model

    We consider the stationarity condition in two ways:


    1 In terms of the eigenvalues of the companion matrix.

    2 In terms of the roots of the characteristic polynomial.

    We illustrate the results with an AR(2) model.

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 4/24
    university of copenhagen department of economics

    Outline: What is the Stationarity Condition in an AR(p) Model?

    1 What is the Stationarity Condition?

    2 Stationarity in Terms of the Eigenvalues of the Companion Matrix

    3 The Link Between the Roots of the Characteristic Polynomial and the
    Eigenvalues of the Companion Matrix

    4 ARMA Models and ARIMA Models

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 5/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 6/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 7/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 8/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 9/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 10/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 11/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 12/24
    university of copenhagen department of economics

    Outline: What is the Stationarity Condition in an AR(p) Model?

    1 What is the Stationarity Condition?

    2 Stationarity in Terms of the Eigenvalues of the Companion Matrix

    3 The Link Between the Roots of the Characteristic Polynomial and the
    Eigenvalues of the Companion Matrix

    4 ARMA Models and ARIMA Models

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 13/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 14/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 15/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 16/24
    university of copenhagen department of economics

    The Stationarity Condition in an AR(p) Model

    The AR(p) process is stationary if:


    • The roots of the characteristic polynomial, z1 , z2 , ..., zp , are outside the
    unit circle.

    • The inverse roots of the characteristic polynomial, φ1 , φ2 , ..., φp , are inside


    the unit circle.

    • The eigenvalues of the companion matrix, λ1 , λ2 , ..., λp , are inside the unit
    circle.

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 17/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 18/24
    university of copenhagen department of economics

    The MA Representation of a Stationary AR(p) Process

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 19/24
    university of copenhagen department of economics

    Outline: What is the Stationarity Condition in an AR(p) Model?

    1 What is the Stationarity Condition?

    2 Stationarity in Terms of the Eigenvalues of the Companion Matrix

    3 The Link Between the Roots of the Characteristic Polynomial and the
    Eigenvalues of the Companion Matrix

    4 ARMA Models and ARIMA Models

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 20/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 21/24
    university of copenhagen department of economics

    Example: Danish Real House Prices


    [Data used: hprice.in7]

    • Estimating an AR(2) model for 1975:3–2001:4 yields

    pt = 0.012285 + 1.51607 · pt−1 − 0.53050 · pt−2 + residual.

    The lag polynomial is given by

    θ(L) = 1 − 1.51607 · L + 0.53050 · L2 ,

    with inverse roots 0.968 and 0.548, such that

    θ(L) = (1 − 0.968L)(1 − 0.548L)

    • One inverse root is close to unity 0.968 ≈ 1 and we estimate an


    ARIMA(1,1,0) for pt :

    ∆pt = 0.00115 + 0.5132 · ∆pt−1 + residual.

    The second inverse root is basically unchanged.

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 22/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 23/24
    university of copenhagen department of economics

    Econometrics II — Fall 2016 — Dynamic Models for Stationary Time Series — What is the Stationarity Condition in an AR(p) Model? — Slide 24/24

    You might also like