VECM model!

[Sul]

Hi guys,

URGENTLY!!!
I would like to ask about 2 questions on the error correction model:
1) I am analyzing inflation determinants, first long-run in money and labor markets separetely using Johansen cointegration test, and second VECM model by running it in so-called Dynamic OLS (my understanding is simple OLS with lags of explanatory variables).
Within the first framework, i.e., cointegration, I run Johansen test with real money, real output and deposit rate which represent real money balances. I got one cointegrating vector at 1% level and correct signs. However, the problem is with error correction term (ECT) itself (I obtained it by clicking Proc/Make Cointegration Group, is that correct?). it has the form of zigzag. Actually this zigzag pattern, I believe, came from zigzag real output which is seasonnaly was not adjusted since the test requires instead using centered seasonal dummies. I tried to seasonally adjust only real output data and the form of the ECT has changed. Now, I don't know what to do whether use zigzag shape ECT or seasonally adjust all variables and drop centered seasonal dummies. Please, could anyone give me advice on this matter?
2) In my coinegration test, a weak exogeneity condition is violated (alpha coefficients are not zero and are significant, restrictions on alpha coefficients were rejected at 1% level). But since i am using only ECTs from long-run relationships in my anlysis of the short-run dynamics (using dynamic OLS), can I just not to take into account this violation? If I can, what is the plausible explanation then? Somewhat similar was done by Sekine (2001) "Modeling and Forecasting Inflation in Japan".
Thanks in advance, please help!

[random_access]

i will just try here but please think about what i say here before doing anything
for the first issue can u test the ECT for stationarity if so test and see whether it is stationary or not. my question is why u need the ECT here?
i do not really know how to tackle the second issue. i hope that helped

[Sul]

Thanks, dear random_access for your reply.

Actually it is stationary, it is not an issue. I just think that ECT is biased because of strong seasonality of RGDP. ECT, in these case, is necessary to see whether disequilibrium in the money market is significant and how long it takes to adjust it in the short-run model.

But i am still curious about my questions. If anyone has tips, let me know pls.

[random_access]

so what is the expected shape of the ETC? if i may ask

[marwa nassar]

hello, I would like to know what are the steps of VECM, and what is its indications, as there were tutorials in the You tube, where the steps like that:
1. test for data stationarity.
2. Cointegration of data after choosing the lag length criteria
3. Run VECM.
4. Make a system ordered by variables.
5.then take the first equation and estimates it.
6. i will get the estimated coefficients as coeff.(1) represents the speed of adjustment which must be negative and p-value less than 0.05 to be significant, the other coefficients represent the short run causality and will be tested by Wald statisitcs and the chisq. should be less than 0.05 to be significant.
7. at the same time we should look at the f-stat less than 5% to say that the model is good.
8.finally we do Residual diagnosis and test if there is serial correlation, their normal distribution.
9. we will repeat the above steps with every variable in the model

So, are these steps correct??, and is the VECM ends by these steps??? i mean is there any other steps that should be done?, and does VECM proves what type of relationship between two variables.
sorry for bothering you