Time Series Econometrics - PowerPoint PPT Presentation

1 / 12
About This Presentation
Title:

Time Series Econometrics

Description:

Stationary vs. Nonstationary time series. Time-domain vs. Frequency-domain approach ... Stationary ARMA Models. Outline of Lecture 1. What is the ARMA model? ... – PowerPoint PPT presentation

Number of Views:655
Avg rating:3.0/5.0
Slides: 13
Provided by: junso
Category:

less

Transcript and Presenter's Notes

Title: Time Series Econometrics


1
Time Series Econometrics
  • Junsoo Lee
  • Department of Economics
  • U.C.F.

2
What is Time Series Econometrics?
  • Stationary vs. Nonstationary time series
  • Time-domain vs. Frequency-domain approach
  • Other Classifications
  • Univariate vs. Multivariate models (VAR)
  • Linear vs. Nonlinear models
  • Observed vs. Unobserved Component models (Kalman
    Filtering model)
  • Classical vs. Baysian, ...

3
Popularity Revolution
  • Nelson (1972, AER), Ashley (1987, Intl J. of
    Forecasting), among others, showed that a simple
    time series model performs better than
    sophisticated macro simulation models
    (FRB-MIT-Penn).
  • UCF football team defeated OSU, Nebraska, UT,
    Notre Dame, UF, FSU,.
  • Time Series Econometrics Revolution in late 1980s
    ??!.
  • Phillips (JEC, 1986) Spurious Regression
  • Phillips Park (ET, 1987, 1988) Asymptotics

4
Lecture 1
  • Stationary ARMA Models

5
Outline of Lecture 1
  • What is the ARMA model?
  • Their Properties
  • Box-Jenkins type fishing
  • Estimation (MLE)
  • Transfer function Analysis
  • Intervention Analysis
  • Practical Exercise
  • Box-Jenkins Fishing (PEST, Eviews)
  • Intervention Analysis (RATS)

6
What is the ARMA model?
  • White Noise process
  • ARMA (p, q) Process
  • Definition
  • No unique form
  • AR(?) or MA(?) representation
  • Causal and invertible

7
  • ACF PACF describe their properties
  • ACF
  • ?(j) 0, j gt q, for MA(q) model
  • PACF
  • ?(j) 0, j gt p, for AR(p) model
  • Dynamic Multiplier and the long-run effect

8
  • S-step ahead prediction
  • Unit Root Process and ARIMA model
  • Slowly decaying ACFs...

9
Box-Jenkins Fishing Method
  • Model Identification
  • Detrend deterministic parts
  • Examine sample ACFs and PACFs
  • Eyeball examination
  • Information Criteria (Akaike, Schwarz, PIC,..)
  • Also, Usual tests (t, F, Wald..) and many others
  • Diagnostic check of randomness of residuals
  • Box-Pierce statistics

10
Estimation of ARMA models
  • AR(p) models OLS
  • Mann-Wald theorem
  • MA and ARMA models MLE
  • Prediction error decomposition method

11
Intervention Analysis
  • Introducing Zt (exogenous, dummy)
  • transfer function analysis
  • Abrupt or gradual change
  • Permanent or temporary change

12
Practical Exercises
  • Box-Jenkins Approach
  • Data gnp82.dat
  • Program PEST pest.zip or Eviews
  • Intervention Analysis
  • Data gnp82.dat
  • Program RATS gnp82.prg
Write a Comment
User Comments (0)
About PowerShow.com