how to test heteroskedasticity of a time series in R?

Question

How can I test heteroskedasticity of a time series in R? I have heard of two tests McLeod.Li.test and bptest (Breusch-Pagan test). Can I use these two tests? and what are the differences and assumptions of these tests if I can use them?

Thanks

Answer 1

Breusch-Pagan test is for hetroscedasticity in regression model. It is testing the relationship between squared residuals and the covariates. You get more information in wiki

McLeod.Li.test is a test for the presence of conditional heteroscedascity. This test is used to identify the presence of ARCH/GARCH modeling. It is very similar to Ljung-Box test on squared residuals.

For time series modeling Mcleoid Li test is more appropriate heteroscedascity test than bptest.

As a suggestion, visual inspection is very important and this should be your first step before any test. Generate the residual graph and see the presence of heteroscedascity visually.