Test for Equal Forecast Accuracy under Unconditional Heteroskedasticity

Seminar abstract

Heteroskedasticity is a common feature in empirical time series analysis, and this presentation will consider the effects of unconditional heteroskedasticity in statistical tests for equal forecast accuracy.

In such a context, it is proposed that two new Diebold-Mariano-type tests for equal forecast accuracy which have two key properties.

First, like the original Diebold-Mariano test, the new tests have size that is robust to heteroskedasticity.

Second, the new tests have the potential to achieve power improvements relative to the original Diebold-Mariano test for a quite general class of loss differential series.

The size validity and potential power superiority of the new tests are studied theoretically and in Monte Carlo simulations. The new tests are applied to competing forecasts for the GBP/USD exchange rate, and also survey-based forecasts for US output.

Booking

This seminar is free to attend with no need to register in advance.

We warmly encourage you to share with friends, colleagues and classmates.

Join this Seminar

Speaker bio

David Harvey is a Professor of Econometrics at the University of Nottingham and a fellow of the Granger Centre for Time Series Econometrics.

His current research interests are in the area of time series econometrics, in particular tests for bubbles, forecast evaluation and predictive regressions.

His work has been published on many top econometrics/economics journals, including;

• Journal of Econometrics,
• Journal of Business & Economic Statistics, 
• Econometric Theory,
• Journal of Applied Econometrics,

as well as many more.