Time Series Basics (2) Fin250f: Lecture 3.2 Fall 2005 Reading: Taylor, chapter , 3.9(skip 3.6.1)
Outline Linear stochastic processes Autoregressive process Moving average process Lag operator Forecasting AR and MA’s The ARMA(1,1) Trend plus noise models Bubble simulations
Linear Stochastic Processes Linear models Time series dependence Common econometric frameworks Engineering background
AR(1) Autoregressive Process, Order 1
AR(1) Properties
AR(m)
Moving Average Process of Order 1, MA(1)
MA(1) Properties
MA(m)
AR->MA
Lag Operator (L)
Using the Lag Operator
An important feature for L
MA -> AR
Forecasting the AR(1)
Forecasting the AR(1): Multiperiods
Forecasting an MA(1)
The ARMA(1,1): AR and MA parts
ARMA(1,1) with L
Forecasting 1 Period
ARMA(p,q)
Why ARMA(1,1)? Small, but persistent ACF’s Comparing the AR(1) and ARMA(1,1)
AR(1) ACF’s
ARMA(1,1) ACF’s
Adding an AR(1) to an MA(0) (Trend plus noise)
Why Is This Useful? (Taylor 3.6.2) Returns follow a combination process Sum of: Small, but very persistent trend Independent noise term
Trend Plus Noise
Parameter Example A small big A = 0.02,
Trend Plus Noise ACF
Temporary Pricing Errors Bubbles(3.6.1)
AR(1) Difference
Variance Ratio
Return Autocorrelations
An Example
Bubble Price Simulation
Return ACF
Outline Linear stochastic processes Autoregressive process Moving average process Lag operator Forecasting AR and MA’s The ARMA(1,1) Trend plus noise models Bubble simulations