Data Conversion to Stationary. Model Two. EVIEWS

hi hi today we shall be developing sorry hi hi today we shall be developing a stationary variable you just won RIT today we are hi today we shall be developing stationary variable from non stationary variable sorry from non stationary variable and we shall use this stationary variable in different different time in different time series model such as var then arch and GARCH and so on so first we shall convert a variable into stationary then we can use this stationary variable in two different time series model such as far then arch GARCH and so on so first first I see my variable the variable is here you can see the variable yeah so you just wait what one minute I make it at just lightly sorry okay okay so this is my variable you can see GDP gross domestic product it is my variable excuse me and I can make it zoom for you that is the gym GDP variable and the sample sizes is from 1960 until 1999 and I shall first check whether our GDP variable is non station non stationary or not if it is non stationary then I shall convert this variable into stationary and when the variables are converted into stationary then we can use this stationary variable in time series model such as bar such as arch and guards and so on okay so what I do I open this variable as a result you can see the variable it is the GDP right cross domestic product and starting from 1960 until until until 2 0 1 2 right so that is my data from 1960 until until 2 0 1 0 that is my data and here but here out of out of 51 here my sample is from 1960 until 1990 so here my sample my range is 1960 ante to 0 1 0 but out of this total range my sample is from 1960 to 1990 that means I shall be using data until 1990 excuse me ok so so first I check whether our variable is non-stationary or not that I check first okay so first I check graphically graphically first I check whether our our GDP variable is non-stationary or not so so I go to quick then I choose the graph right I can make June for you I choose the graph right from here you can see I choose the graph ok then you can see I can make a Jew for you here I put GDP gross domestic product that I write here right so that I write here then I press ok okay so here I choose lines and symbols right then I press okay so here is the data our GDP data it is gradually going up from 1960 until 1990 it is gradually going up meaning that the data is not stationary it is non stationary okay so this one is the graphical presentation it is the graphical presentation now we have to check it using using test whether our variable is really non stationary or not and here we shall be using correlogram correlogram and also we shall be using actually LG LG box we can call it lb statistics that we shall be using to test whether our variable is stationary or non stationary okay so now now now who we can use this June box statistics or we can call it lb statistics same meaning so I go to quick then I choose series statistics then I choose correlogram right it is here kuroh gram then the variable name is GDP right that I put here I can make it do for you you can see I can put series name GDP so then I have to run okay here there are two thing label I can make it do for you so here is three things one is level then first difference that means after first after first differencing what happens and this one is the second difference meaning that after converting the variable into second difference then the variable is is stationary or not and what about at level meaning that data has not been into first difference or second difference so this one is the original data original data that is called level and suppose a if I choose first difference then the eview software will automatically convert the variable into first difference and then they will check whether first difference variable is stationary or not or if I choose the second difference then our software will automatically convert the variable into second difference and and then they will check whether second difference variable is stationary or no but here first I choose at Louisville meaning that the data that I have original data original data that I choose first and here I choose lakh lakh 10 why normally the guideline is normally we take 1/3 of the data as a lakh so our data is almost 30 so I take 1/3 that is 10 so that is some guideline normally we take 1/3 of the sample size ok now we are set we can proceed for testing at level I click OK then I click OK click OK so the result is coming up of the of the correlogram it takes some some time because my computer is very slow so it takes some time when also I personally also very slow I talk slowly all the time I am sorry for that is coming up where is the result is it coming up already so long okay it has come up here actually it is the outcome of the ku-ku-ku-ku hologram you can see from here I make it big for you hook it is the outcome so you can see the autocorrelation right you can see autocorrelation this one and this one is the autocorrelation right so what is this suppose I talk about this one this one means 0.766 means actually this one then what about this one this one ah sorry what about this one 0.77 one and that is autocorrelation coefficient it is actually here this one then what about the last one this is zero point zero four eight that is the autocorrelation coefficient this one means actually this one right so here you can see autocorrelation coefficient is gradually going down as a result the here it is also gradually going down so here going down as a result here is also going down and here I have taken 10 lakh so you see from here lag 1 lakh 2 lakh 3 lakh for until lakh 10 and normally I say when the autocorrelation gradually go goes down meaning that data our variable probably non stationary so so I am Telling again when the autocorrelation it is the autocorrelation right SC when it when it goes down gradually meaning that probably our data is non station and here is the Q statistics you can see the queue statistics it is the queue statistics right and it is the corresponding probability it is the corresponding probability of this queue statistics now what is our null hypothesis that we have to check first our null hypothesis is data is stationary and our alternative hypothesis is data is not stationary station excuse me I have some problem in my throat okay so our null hypothesis is data is stationary an alternative is data is not stationary okay and here you can see again here we shall be using junk box test here our task we shall be using Jun boxed statistics to test whether our variable is stationary or not okay now we see from here okay and here you can see excuse me and here we can see that Q statistics the queue statistics is how much here you can see our Q statistics is here right and here we choose the last one the last one of that queue statistics I go down so I choose this in ninety five point two six nine I choose the ninety five point two six nine of queue statistics right and I choose the corresponding probability value that means p-value so here the probability value is zero meaning that less than five percent meaning that we can reject the null hypothesis meaning that we reject the null hypothesis and accept the alternative meaning that our data is not stationary because we can reject the null hypothesis and we accept the alternative why because our p-value is less than five percent it is here so so and the thing is that what is the guideline the guideline is normally if the p-value or probability value is less than five percent we normally reject the null hypothesis but if the p-value more than five percent we do not reject null hypothesis rather we accept null hypothesis but here the p-value is less than five percent so we are rejecting null hypothesis and accepting alternative hypothesis meaning that data is not stationary so we cannot use this data in that time series model normally time series model need stationary data and here our data is non stationary okay and and and also we can see from here there is a another symptom the autocorrelation I just make it big you see from here this autocorrelation actually you can see there is two lines there is one line can I check this line and this line right one line and one line right there's two things okay normally when this we call it spike we call it spike spike means s Pik we call it spot I write it here spike we call it spike and normally when this spike just one minute when this spike are outside this two line you can see here is one line and here is one line and when the spikes are outside this line meaning that data is not stationary that is one symptom that is one symptom when the spikes are outside the line we can suspect that we can assume that our variable is not stationary and so that is that a graphical way and we have just proved that that Q statistics it is not significant that is significant because p-value is less than 5% meaning that indeed the variable is not stationary okay now we are now our target is to convert the variable into into stationary okay now the question is that how to convert a variable into stationary one way one way is to convert a variable into stationary stationary is first differenced that means we shall convert our variable into first difference and and probably after that our variable will become stationary okay so we convert our variable now then we check what happens okay we convert our variables and it is our variable you can see our gdb variable right okay our GDP variable is here we are here is our variable now we we shall convert the variable into first difference how okay so so here we have a variable called GDP I create one more variable suppose that is called dgtp suppose I choose new object to create a new variable new object then here I choose here I choose series right I series and here I put D GDP right so I can make it do for you you can say I choose series then I stood in D GDP right that means this one is the new variable that that I am creating now I am creating this new variable now that is D GDP okay and okay so so the new variable has been created that is D GDP excuse me and the question is that what does mean by D GDP actually D D GDP means first difference of GDP that is called D GDP okay now now now sorry now what is called D GDP GDP means here I am writing the command D GDP means D by GDP so that is the command in eviews that means the GDP GDP will convert to first difference automatically and this one is the command and after writing the command you just you just press on the enter you just press on the enter as a result GDP will be will be converted into first difference and the name of the new variable is D GDP okay sooner now we can check here is the our variable I can make it jus for you you can see here is the variable B GDP and this one is the GDP and this one is the D GDP and here GDP the data is non-stationary and here that D GDP actually first difference of GDP now now now we want to check whether this D GDP is stationary or not okay okay okay so okay what I can do I can show you the data I can obtain and I can show the data you can see that is our new variable D GDP right first difference of GDP and you can see the data from here right you can see the data from here right and and and here we are using until 1990 right so that is our data okay now I check i check using graphically I check using graphically whether this first difference variable has become stationary or not okay so i go to i go to quick I choose the graph right I choose the graph then here I put make it Jim for you here I put here I put D GDP da GDP right I put D GDP then I press ok then here I choose line and symbol right and press okay okay now you see the variable is D GDP and it is from 0 right 0 and the variable is like this right the variable is like this it is straight right on the 0 then I can say that probably our first difference variable has become stationary because it is on the zero right so the data is not like this or not like this the data is not going up or is not going down and there is no trend no trend no train so I can say that probably our data has become stationary but but we can check it using using our statistics that whether our variable is really stationary or not okay then I check it I go too quick then I choose the series statistics then I go to that correlogram as before and the variable name right now D GDP right I can make it jump for you the variable name is now D GDP then I press ok I press ok and here I choose at level because the variable has been already converted to first difference so here I put I choose I select the level because because the variable has been already converted to first difference and so we can check it and lag here I choose 10 right lag I choose 10 so I can show you the whole things here I choose level and the lag I choose 10 because the our sample is 30 I choose one third of the sample as a lakh okay then I press ok so this is the outcome of the statistics you can see from here this the outcome status is comes up right okay and you can see from here that is the things the result has come up so here is again here where here is a result of correlogram of B GDP right it is d GDP the correlogram of D GDP I can make it Jim what you can see the correlogram of not GDP it is d GDP meaning that first difference of GDP and here is the auto correlation coefficient and they have become so small right become so small and sometime becomes 0 right some become 0 right all are very small and we have 10 lakhs you can see all the 10 lakhs so that becomes very small as a result here is also very small you can see the data almost zero so there is no spikes there is no spikes right and all spike becomes zero so here is only this one actually this one and this one actually this one right and other becomes zero you can see other become zero so there is no spike we cannot see any spike but but normally what is the guideline if the spikes are within this two line then we can say probably the data is stationary right but here one is outside the spike this one is outside the line so so here one spike one spike is outside the line so normally the guideline is all spike should be within this two line but the things that this one is the graphical presentation that cannot make anything confirm but we can make it confirm by using our statistics and that is called what statistics that can make us confirm by using Jun box that is lb statistics whether our variable is stationary or not and our what is our null nollies data is stationary right that is our null hypothesis and we can check from here right we can check from here you can see from here okay and this is the queue statistics right this is the queue statistics and we choose the last one we choose the last one seven point three six zero two and corresponding probability value these are all probability value and and and here out of queue statistics I choose the last one seven point three six and the corresponding probability value the probability value is sixty nine point one percent sixty nine point one percent which is more than five percent so we cannot reject null hypothesis rather we accept null hypothesis meaning that data is stationary so meaning that our variable that is D GDP can be used for data analysis right such as I can run the real regression line and I can do the VAR model I can do the arch model with this D GDP variable because this variable has become stationary after first difference okay suppose suppose if this variable does not become stationary after first difference then then we should do we should do second difference we should do second difference to make it stationary make it stationary meaning that as long as the data is not stationary we should continue to do first difference if not then we should do the second difference and then we can use that data for time series analysis so that is the normal guideline but here but in our model after first difference the variable has become stationary so our first difference data such as D GDP we can use this D GDP for for analyzing the data thank you very much for being with me for a while

Channel: Sayed Hossain

Views: 63,880

Keywords: Education, regression, vecm, spss, eviews, variable, econometrics, sayed hossain, stationary variable

Id: OMg8GGSJfGI

Length: 39min 10sec (2350 seconds)

Published: Mon Dec 10 2012