ARCH estimation error

[Nandez]

Hi.

I'm working on an exercise where I have to obtain daily prices for the Norwegian and the US stock market indicies for the last 20 years. I used datastream to collect the numbers. For the Norwegian market I used OSEBX and SP500 for the US market.

One question that I have to answer is: Formally test for ARCH effects in the log returns series. Before I move on I must say that I only had estimation problem when testing for ARCH effects for SP500. The error msg I got: MA estimation requires a continuous sample.

My input when estimating equation: logrsp500 c ar(1) ma(1) (see workfile attahced)

logrsp500 = log return on sp500 data collected

Did the same procedure for my osebx data and I hade no problem with that.

eviews version: Std. edition jan 7 2010 build.

Thanks!

[startz]

You NAs in your sample. Hence it's not continuous.

[Nandez]

So do I need to collect new data and try or can I do sometthing to my existing data ?

Also look at my osebx workfile I see that i have NAs in the sample too, but no error when estimating my ARCH equation.

[trubador]

Missing values lead to a problem when they are scattered within the sample and hence resulting in discontinuity. EViews can handle them if they are only at the very beginning/end of the sample. If you like, you can get rid of missing values in your first data set as follows:

Code: Select all

pagecontract if sp500<>NA

If you are going to work on return series, make sure you transform the data first and use it for contraction:

Code: Select all

series ret = dlog(sp500) pagecontract if ret<>NA

[Nandez]

Missing values lead to a problem when they are scattered within the sample and hence resulting in discontinuity. EViews can handle them if they are only at the very beginning/end of the sample. If you like, you can get rid of missing values in your first data set as follows:

Code: Select all

pagecontract if sp500<>NA

If you are going to work on return series, make sure you transform the data first and use it for contraction:

Code: Select all

series ret = dlog(sp500) pagecontract if ret<>NA

Thanks trubador, it worked now! :)