Search code examples
rtime-seriesvariancerandom-walk

Pratical questions about the vrtest package

I want to perform Variance Ratio tests (Lo-MackKinlay, Chow-Denning) but I have some problem with the running of the commands.

  1. I have a price Index for 1957 to 2007. Do I need to perform the variance ratio tests on the level series or on the series of returns?

  2. How do you fix the kvec? It is a vector with the lags for which you want to do the test right?

So here is my output: 

> rcorr
 [1]  0.0000 -0.1077  0.4103 -0.0347  0.1136  0.0286  0.0104  0.0104  0.1915
[10] -0.0025  0.0665  0.2127  0.0116 -0.1288  0.1640  0.3089  0.2098 -0.1071
[19] -0.2079 -0.1082  0.0022  0.1419  0.0641 -0.0082 -0.1163 -0.1731  0.0260
[28]  0.0468  0.0882  0.2640  0.3946  0.2094  0.2754  0.0623 -0.3696 -0.1095
[37] -0.1463  0.0118  0.0152 -0.0103  0.0223  0.0379  0.0580 -0.0091 -0.0510
[46]  0.0765  0.0984  0.1250  0.0519  0.1623  0.2552
> kvec<--c(2,5,10)
> Lo.Mac(rcorr,kvec)
Error in y[index] : only 0's may be mixed with negative subscripts

Why do I get this error?

Solution

  • It is the same error as in your other question I just answered:

    kvec<--c(2,5,10)

    is the same as

    kvec <- -c(2,5,10)

    ie

    kvec <- -1 * c(2,5,10)

    Remove the second dash.