VECM

HOW TO RUN VECM (vector error correction model) There are three steps to run VECM 1. Leg selection for the model, means how many lag must be select for our model. 2. Then co-integration, if we find cointegration than we can apply VECM otherwise unrestricted VAR model.

3. VECM STEP 1 LAG SELECTION a.

Open all variable as group , as VAR

\\\\\\\\ \\\\\\\\\\\\\\\\\\\\

b. Write the name of all variables and select lag 12 and ok like in above pic and following results will be appear

Now from these results go to view, then lag structure, then lag length

criteria. c. By SAEED AAS KHAN MEO THE SUPERIOR GROUP OF COL LEGES LAHORE PAKISTAN

d. From step (b) go to view of above results window, go to lag structure-lag length criteria and ok but chose lag length what u like 3, 4 what u like whe re SIC or AIC IS MINIMUM THAT LAG LENTH IS SUPERB OK (here you will find many crite ria all are best but it depend on you what you like, for example SIC, AIC etc.)

THIS WINDOW IS INDICATING THAT LAG LENTH TWO IS GOOD because if we add lag length three then AIC AND SIC VLAUE INCCRESE. And further most of criteria’s are criteria’s are suggesting 2 is optimal lag length, but if you are making decision according to SC 1 is optimal

All are best criteria but choose lag lengths according to criteria

So one more thing if AIC suggest

through you will make decision. Like if I am making final decision

8 lags, while SIC suggest 3 lags it

according to FPE then 2 is optimal lag lengths.

depend on

you what you

like both are true

e. STEP TWO JOHNSON CO-INTEGRESSION TEST

FROM above result we have decided that lag length two is good) Step 2.1 go to quick — group group statistic – statistic – Johnson Johnson co-integration remember for co-integration chose lag length which we decide in first step on lag length selection and ok and one thing more used current results output for the test if u use new window or then results results may suffer losses,

By SAEED AAS KHAN MEO THE SUPERIOR GROUP OF COL LEGES LAHORE PAKISTAN

Here I see six is suitable lag length, so next step is to find out that either there is cointegration or not between variables.so I go to quick of resulted window, the Johnson cointegration.

Explanation of this dialog box Deterministic trend assumption of test Practical guides: • use case 1 only if you know that all series hav e zero mean (unusual ( unusual in empirical studies); • case 5 may provide a good fit in -sample but will produce implausible forecasts out-of-sample.; • use case 2 if none of the series appear to have a trend; By SAEED AAS KHAN MEO THE SUPERIOR GROUP OF COL LEGES LAHORE PAKISTAN

• use case 3 if series are trending and you believ e all trends are stochastic; • use case 4 if series are trending and you believe some of them are trend stationary; • use case 6 if you are not certain which trend assumption to use (Eviews will help you determine the choice of the trend assumption).

Here we find that there are two cointegration equations in our model which is good for us Note: to conclude that there is cointegration exist or not only one cointegration equation is enough, Note :some time trace test indicate cointegration ,while maximum Eigen test indicate no cointegration this case you can conclude that there is cointegration among variable on the base of trace test as bench mark, same if trace showing no cointegration while maximum Eigen showing cointegration you can say there is cointegression . Note it’s better if both of test showing same results. Condition fulfill of cointegration for VECM now we can move ahead.


Now we have applied co-integression if we find co-integression the apply VECM and if u did not find cointegration then apply U-VAR model

Another example of cointegression having only one cointegration equation.

We find that there is one co-integrated equation which means their is a long run relation between variables, means all these variables move together,

Note; for the co-integration minimum one equation must be cointegrated, not necessary that at most, 1 or two or three are cointegrated. Note: if we find no cointegration we must apply unrestricted var model

Step three VECM Now we have found that there is a long run relation in our variables means our variables are cointegrated. From above output file we will go quick— quick—estimation of varAnd here ill select vector error correction then ill write first dependent dependent variable and the all independent variables like co2 oil gdp and I must select lag length which a selected from our first step minimum AIC and SIC which was two and last thing in the present window go to co -integression option select the cointegrated equation like in our above Johnson test we have find only one equation which which was cointegrated so we will chose 1. 1 . And ok Note vecm will tells us about the causality between variables Step2


when click ok this window will be open here if see there is no P value so to find out p value we go to proc— proc—make system – system – by by variable and ok

From above windows go to proc— proc —system- variable ok and a window will be open like t his


Now u must copy the only one equation e quation (D(CO2) = C(1)*( CO2(-1) - 0.002313245974*OIL(-1) 0.0002037688597*GDP(-1) + 0.3834446834 ) + C(2)*D(CO2(-1)) + C(3)*D(CO2(-2)) + C(4)*D(OIL(-1)) + C(5)*D(OIL(-2)) + C(6)*D(GDP(-1)) + C(7)*D(GDP(-2)) + C(8))the above equation and go to quick and estimate equation and past and ok

From here we will choose either here long run causality or not Long run causality: causality: note: c(1) if comes negative

and significant we can say there is long run causality running form oil and gdp to co2 , but in our case c(!) value is not negative and significant so we can say there is no long run causality Short run causality: now we will check short run causality with the help of Wald test.

How to check short run causality Now we will apply Wald test, so how to test Wald test? ……first of all we will decide that from which independent variable we are going to che ck causality with dependent variable like in the following example I’m going to check short run causality between between oil and co2.we co2.we go to view – view – coefficient coefficient diagnostic – diagnostic – Wald Wald


So we want to check causality between oil and co2 ,c(4)=c(5)=0

Now question is that how we decided c(4) and c(5) representing oil. So look below and remember above generated equation. Here c(6) c(6) is gdp that’s why i did no t include c(6) in oil .

equation (D(CO2) = C(1)*( CO2(-1) - 0.002313245974*OIL(-1) 0.0002037688597*GDP(-1) + 0.3834446834 ) + C(2)*D(CO2(-1)) + C(3)*D(CO2(-2)) + C(4)*D(OIL(-1)) + C(5)*D(OIL(-2)) + C(6)*D(GDP(-1)) + C(7)*D(GDP(-2)) + C(8))

Now go view – view – coefficient coefficient diagnostic – diagnostic –Wald Wald and past c(4)=c(5)=0 and ok if chi- square probability value comes more the 5% percent we will accept null hypothesis means no causality,but causality,but here we will reject null hyposthes and say there is cusality

And same for other variables 1.

Further u can check serial ser ial correlation and other diagnostic test. Thank you so much for being we with and sir sayed Hossain . By SAEED AAS KHAN MEO THE SUPERIOR GROUP OF COL LEGES LAHORE PAKISTAN

From this window check serial correlation and other diagnostic test.


VECM

Recommend Documents