topic about FARMAFIT subrountine in SAS/IML Software and Matrix Computations

about FARMAFIT subrountine

Hanyu — Thu, 20 Feb 2014 15:53:34 GMT

I want to estimate a ARFIMA model of realised volatility using real series (not simulated ones) . Here is my code.

proc iml;

use square;

read var{square}into exper ;

lag=(0:5);

d=0.4;

phi=0.5;

ma=0.3;

sigma=0.06;

call farmafit(d,phi,ma,sigma,exper);

Sas generate a error saying insufficient number of observations. I have 787 observation for this time series.

Anybody know why? Thank you in advance.

Rick_SAS — Thu, 20 Feb 2014 16:11:42 GMT

My guess is that you don't have 787 observations. You have 1 (which is certainly insufficient). Print the number of rows for expe like thisr:

print "number of rows=" (nrow(exper));

You might also try the following example and report what happens:

proc iml;

use sashelp.air;

read all var{air} into exper;

close sashelp.air;

call farmafit(d,phi,ma,sigma,exper);

print d, phi, ma, sigma;

Hanyu — Thu, 20 Feb 2014 18:51:36 GMT

Thank you for your response. I have tried print "number of rows=" (nrow(exper)); using sashelp.air;

The return number is 144. The estimation of the ARFIMA on sashelp.air reports d=1.15 phi ma don't exist. sigma=1118.

Rick_SAS — Thu, 20 Feb 2014 19:53:36 GMT

I meant for you to print the number of rows for your example, not for mine.

Hanyu — Thu, 20 Feb 2014 19:57:03 GMT

I will post it tomorrow if that is okay for you as it is pretty late in UK.

Hanyu — Fri, 21 Feb 2014 10:36:57 GMT

Thank you so much Rick. I have solved my issue. Unfortunately there is no long memory feature of my series.