Scribd Logo
Upload

Welcome to Scribd!

  • 5 views

    AMFE Module 2

    Uploaded by

    Aruneema

    Copyright:

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

    AMFE Module 2

    Uploaded by

    Aruneema
    5 views10 pages

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    5 views10 pages

    AMFE Module 2

    Uploaded by

    Aruneema

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    You are on page 1of 10

    Applied Macro and

    Financial Econometrics
    Moving Average Process

    Course Instructor:
    Dr. Devasmita Jena
    Introduction
    • What is DGP of Yt?
    • There are few process (models) that may have generated Yt
    • Lets again draw a parallel with cross-section analysis
    • What is production function of a firm looks like?
    • Cobb Douglas? CES?
    • Although we draw a parallel with cross-section analysis, TS models break the
    assumption
    • Xt and Yt don’t cause each other even if ut, ut-1,…ut-k are independent of
    each other
    1st order Moving Average Process or MA(1) Process
    • ---(1); and are constants and is the white noise
    • y at time t is equal t, a constant, plus moving average (akin to weighted average)
    of current and past error
    • ---(2)
    • ---(3)
    • 1st autocovariance:
    • Higher order autocovariance are all zero:
    • Since mean and autocovariances are not function of time, MA(1) process is
    stationary stochastic process
    • 1st order autocorrelation:
    • 2nd order autocorrelation:
    • Higher autocorrelations are all zero
    MA(1) Process: Autocorrelations
    • For different , we will get different values of 1st order autocorrelation
    • +ve :
    • +autocorrelation
    • Large value of is likely to be followed by a larger than average value of
    • Small value of is likely to be followed by a smaller than average value of
    • -ve :
    • -ve autocorrelation
    • Large value of is likely to be followed by a smaller than average value of
    • Autocorrelation Function (ACF): Plot of autocorrelation against lags
    • Question: Is MA(2) stationary?
    qst order Moving Average Process or MA(q) Process

    • jth autocovariance, when j= 1,2,…q:


    •]
    • Terms involving s at different dates have been dropped because their product has
    expectation zero, and
    • jth autocovariance when j>q are all zero:
    • Since there are no ∈s with common dates in the definition of and the expectation is zero
    • Therefore:
    • =
    • =0; for all j>q
    MA(q) Process
    • For example, for MA(2):

    • =…=0
    • Is MA(q) process stationary?
    • What is the ACF of MA(q) process?
    Infinite order, MA(∾) Moving Average Process

    • --- (1)
    • ’s are coefficients of infinite order MA series
    • Eq. (1) is covariance stationary iff --- (2)
    • This is square summability condition
    • More stronger condition will be – absolute summability: --- (3)
    • (3) => (2) but not vice versa
    Absolute Summability implies Square Summability
    • If is absolutely summable, then Ǝ an N < such that
    • || < 1 for all j ≥ N => < || for all j ≥ N
    • Then + < +
    • Blue part is finite as N is finite
    • Orange part is finite as absolutely summable
    • Therefore,
    Square Summability doesn’t imply Absolute Summability
    • Let’s take a series that is square summable but not absolutely summable

    • Note, > for all x>j =>


    • Let

    • = log (N+1) – log(1) = log (N+1)


    • When N, this series diverges
    • The series, however, is square summable, since for all x<1

    • This converges to 2 as N∾, hence, is square summable


    Moments of MA(∾) Process

    • An MA(∾) process with absolutely summable coefficient has absolutely


    summable autocovariance

    You might also like