Time Series MT 2.1 Update

The new Time Series MT 2.1 Update enhances current GAUSS time series modeling functionality, with tools for testing and estimation of nonlinear time series.

Newly included and supported models include:

New tools facilitate extensive data analysis and goodness of fit evaluations:

  • New ARIMA lag selection procedure including AIC, SBC, HQC, and FPE.
  • Data aggregation procedure facilitating end-of-period, period average, or beginning-of-period aggregation.
  • Data standardization allowing for demeaning and data scaling using standard deviations.
  • Parameter instability detection including:
    – Rolling and recursive OLS estimations
    – Chow forecast test
    – CUSUM and CUSUM squared tests
    – Hansen-Nyblom test

Tools for accommodating individual modeling needs:

  • Parameter structures for flexible user control of TAR modeling including:
    – User specified AR ordering
    – Specific lag omission
    – Number of Monte Carlo repetitions
  • Flexible user control of structural break modeling including:
    – Maximum number of breaks
    – Partial structural break modeling
  • Optional screen and graphics output.

GAUSS Time Series output includes (when applicable):

  • TAR estimates and test values including:
    – SupLM, ExpLM, AveLM, SupLMs, ExpLMs, and AveLMs test statistics
    – Monte Carlo generated p-values
    – Estimated coefficients for both regimes
    – Estimated threshold variable lag and value
    – Regime specific and total model error variance
  • Global structural break estimates including:
    – Estimated break dates and total SSR for all number of breaks less than or equal to maximum number of breaks specified