whiteness

Durbin-Watson test for whiteness (no serial correlation) of VAR residuals

Syntax

   [dw,pval] = whiteness(X,E)

Arguments

See also Common variable names and data structures.

input

   X          multi-trial time series data
   E          residuals time series

output

   dw         vector of Durbin-Watson statistics
   pval       vector of p-values

Description

Returns the Durbin-Watson test statistic dw along with p-values pval for a multivariate regression of time series data X with residuals E (may be single- or multi-trial). This routine computes the statistic separately for each individual variable in X.

A standard rule of thumb is that dw < 1 or dw > 3 indicates a high chance of residuals serial correlation; this implies poor VAR model fit.

NOTE: To test for significance you should correct for multiple null hypotheses, or test for false discovery rate; see significance.

References

[1] J. Durbin and G. S. Watson, "Testing for Serial Correlation in Least Squares Regression I", Biometrika, 37, 1950.

[2] A. Bhargava, L. Franzini and W. Narendranathan, "Serial Correlation and the Fixed Effects Model", Review of Economic Studies, 49, 1982.

See also

significance | mvgc_demo_stats

Copyright notice

[(C)] Lionel Barnett and Anil K. Seth, 2012. See file license.txt in root directory for licensing terms.