## How to use fracdiff EViews add-in to perform the fractional Dickey-Fuller test

For posting your own programs to share with others

Moderators: EViews Gareth, EViews Moderator

bensalma
Posts: 2
Joined: Mon Nov 18, 2019 5:33 am

### How to use fracdiff EViews add-in to perform the fractional Dickey-Fuller test

Code: Select all

`  ' With EViews 9                                            'The objective of this program is to find the upper and the lower bound of the fractional integration parameter of an ARFIMA(0,d,0)  'by using a sequential standard dickey-Fuller test, based on the usuels tabulated values.                                            'This program is not flexible enough to use it in any situation (for example for d>=1). I am not an expert in programming.  'But this program shows how efficient the sequential procedure is.                                             'In this program we simulate a process ARFIMA(0,d,0) (For example d=0.6). 'Then, we assume that (d) is unknown and we try to find an upper and lower bound for (d) by using a sequential procedure of the Dickey-Fuller test.                                            'Try to perform the program several time for a fixed value of d. 'Change the value of (d) (for example d=0.3, d=0.4, d=0.5) and perfom the program several time for each value.create u 600'---------------------------------------------------------'Simulation of an ARFIMA(0,d,0) with d=0.6 '---------------------------------------------------------genr e=nrnde.fracdiff(d=-0.6,)rename e_diff xgenr y=x-@mean(x)vector (10) Accept_Reject_H0  '-------------------------------------------------------------' Set of sequential values of (d0,i) and (d0,i-1)'-------------------------------------------------------------for !i=1 to 10!d0=0.1*!i!d=-1+0.1*!i'------------------------------------------'Compute xi(t)=(1-L)^(d0,i-1)y(t)'------------------------------------------y.fracdiff(d=!d, )rename y_diff x!i'------------------------------------------------------------------------------------'Testing the null H0:d>=d0,i by means the t-stat of c(1)                          'coefficient in the model (1-L)^(d0,i)y(t)=c(1)*(1-L)^(d0,i-1)y(t-1)'------------------------------------------------------------------------------------equation eq!i.ls d(x!i) x!i(-1)if eq!i.@tstat(1)>-1.94 then Accept_Reject_H0(!i)=1 else Accept_Reject_H0(!i)=0endifdelete x!inext'--------------------------------------------------------'Find the lower and the upper bound of d'--------------------------------------------------------for !i= 1 to 9if  Accept_Reject_H0(!i)=1  and Accept_Reject_H0(!i+1)=0 then scalar Lower_bound_of_d=0.1*!iscalar Upper_bound_of_d=0.1*(!i+1)endifnext'--------------------------------' display results in table'--------------------------------table tab1setcolwidth(tab1,1,20)setcolwidth(tab1,4,20)tab1(1,1)="Table 1" tab1(2,1)= "Sequentiel F-DF test,"tab1(3,1)="on the nile series: "tab1(4,1)= " H0:d>=d0,i;  i=1 to 10"tab1(4,2)=" "tab1(4,3)=" "tab1(4,4)=" "tab1(4,5)=" "setline(tab1,5)tab1(9,1) = "d0,i"tab1(9,2) = "d0,i-1"tab1(7,3)= "DFt"tab1(8,3)="="tab1(9,3)="eq!i@tsat(1)"tab1(7,4)="DF" tab1(8,4)="distribution"tab1(9,4)="cv(5%)"  'critical value of the Dickey-Fuller distributiontab1(6,5)="Accept(=1)"tab1(7,5)="or"tab1(8,5)="Reject(=0)"tab1(9,5)="H0"setline(tab1,10)tab1(11,1) = "0.1"tab1(11,2) = "0.9"tab1(11,3) = eq1.@tstat(1)tab1(11,4)="-1.94"tab1(11,5)=Accept_Reject_H0(1)setline(tab1,12)tab1(13,1) = "0.2"tab1(13,2) = "0.8"tab1(13,3) = eq2.@tstat(1)tab1(13,4)="-1.94"tab1(13,5)=Accept_Reject_H0(2)setline(tab1,14)tab1(15,1) = "0.3"tab1(15,2) = "0.7"tab1(15,3) = eq3.@tstat(1)tab1(15,4)="-1.94"tab1(15,5)=Accept_Reject_H0(3)setline(tab1,16)tab1(17,1) = "0.4"tab1(17,2) = "0.6"tab1(17,3) = eq4.@tstat(1)tab1(17,4)="-1.94"tab1(17,5)=Accept_Reject_H0(4)setline(tab1,18)tab1(19,1) = "0.5"tab1(19,2) = "0.5"tab1(19,3) = eq5.@tstat(1)tab1(19,4)="-1.94"tab1(19,5)=Accept_Reject_H0(5)setline(tab1,20)tab1(21,1) = "0.6"tab1(21,2) = "0.4"tab1(21,3) = eq6.@tstat(1)tab1(21,4)="-1.94"tab1(21,5)=Accept_Reject_H0(6)setline(tab1,22)tab1(23,1) = "0.7"tab1(23,2) = "0.3"tab1(23,3) = eq7.@tstat(1)tab1(23,4)="-1.94"tab1(23,5)=Accept_Reject_H0(7)setline(tab1,24)tab1(25,1) = "0.8"tab1(25,2) = "0.2"tab1(25,3) = eq8.@tstat(1)tab1(25,4)="-1.94"tab1(25,5)=Accept_Reject_H0(8)setline(tab1,26)tab1(27,1) = "0.9"tab1(27,2) = "0.1"tab1(27,3) = eq9.@tstat(1)tab1(27,4)="-1.94"tab1(27,5)=Accept_Reject_H0(9)setline(tab1,28)tab1(29,1) = "1"tab1(29,2) = "0"tab1(29,3) = eq10.@tstat(1)tab1(29,4)="-1.94"tab1(29,5)=Accept_Reject_H0(10)setline(tab1,30)tab1(2,4)="Lower bound of d="+@str(Lower_bound_of_d)tab1(3,4)="Upper bound of d="+@str(Upper_bound_of_d)tab1(4,4)=@str(Lower_bound_of_d)+"=<d<"+@str(Upper_bound_of_d)show tab1`
sequential_fdf_test.prg
Program EViews to perform a sequential fractional Dickey-Fuller test on the demeaned Nile series
Hello Everyone
this is my first time participating in this forum. I want to share with EViews users how to use fracdiff Eviews add-in. I use fracdiff Eviews add-in, to perform the fractional Dickey-Fuller test. Here is the document link

https://mpra.ub.uni-muenchen.de/107445/ ... 107445.pdf
Attachments
nil_data.wf1
Nile series workfile
Last edited by bensalma on Mon Jun 07, 2021 11:59 am, edited 2 times in total.

bensalma
Posts: 2
Joined: Mon Nov 18, 2019 5:33 am