Quantitative Data Analysis

Scope: Check for neglected nonlinearity in time series data or regression models.

Consequences of non-linearity when linear approaches are used: Biased estimates, poor predictions and inaccurate understanding of the relationships between variables. In all cases, misleading conclusions can be derived, while non-linear models have to be considered.

H₀: Linearity in “mean”

*

Decisions before the implementation of the test:

Implementation of the test: R package “tseries”

Other tests for checking non-linearity: Terasvirta test, Keenan’s test, Tsay’s test.

Sources

• Lee, T. H., White, H., & Granger, C. W. (1993). Testing for neglected nonlinearity in time series models: A comparison of neural network methods and alternative tests. Journal of econometrics, 56(3), 269-290.
• Franses, P. H., & Van Dijk, D. (2000). Non-linear time series models in empirical finance. Cambridge university press.