I am currently analysing the relationship between monetary base and Unemployment and have constructed an ARDL model. When I use the BIC to determine the optimal lag length of my independent variable (monetary base), the model that is suggested only has one lag. I have a feeling this doesn't make very much economic sense. In the model with one lag, the independent variable isn't significant.
When I include 12 lags, the 5th lag of the independent variable is significant.
I have read that with monthly data, including 12 lags is reasonable.
Could I just include more lags than are suggested by the BIC in order to get a significant variable?
In general, the lag length is guided by the theory. Obviously, one lag is insufficient in this case. Try other tests, if the problem persists go with the theory.
Manhal Shotar Thanks! Do you have any recommendation, which other test would be useful? AIC?
Best,
Carl
Schwarz-Bayesian, (SBC), vector Hannan-Quinn, (HQC) and the likelihood ratio (LR) can also be used.
Dear Carl,
Why not try other criteria like:
AIC:Akaike Information Criterion
SC: Schwarz information criterion.
HQ:Hannan-Quinn information criterion.
Then compare with your results from BIC and go on with the best.
Regards
Dear Carl, you are using single independent variable with one lag in your analysis. Theoretically unemployment is not based only on monetary variable. So why not you search other explanatory variables?
Dear Chung Tin Fah , Dear Kehinde Mary Bello ,
thank you for your help! I have tried other criteria and they all seem to provide the same results, that are rather unreasonable in theoretical economic terms.
I have read , for example in the book of Wooldridge, that with monthly data, a rule of thumb could be to include 12 lags.
When I use this number, I get much more reasonable results.
Do you think it would be appropriate to go with this suggestion and disregard the information criteria?
I would be very thankful for your answers!
Best, Carl
Dear Yadav Sharma Gaudel ,
thank you for your help!
I also include other independent variables like GDP, consumer confidence, Stock prices, wages , etc. in my regression!
Dear Carl,
Kindly report the results you got from your findings.
Not all results go directly or in conformity with other journals or books.
That's why it's called a research, report your finding appropriately from your own personal analysis.
Differences always occur because the conditions surrounding research is different: source of data, type of country, techniques of analysis, the softwares used etc.
Give us the very outcome of your own analysis, that is the beauty of research.
Regards
BIC impose some penalty on lag length thus, it will most likely give u a lower lag length than AIC which imposes lesser restrictions. The lag chosen by the information criterion are optimal lags and a researcher can move around it..Considering the choice of 12 lags for monthly data, it is indeed a rule of them that 12 observation goes for monthly data , 4 for quarterly and 1 for yearly data. If you are to go for this assumption, be sure to have a reasonable sample size as failure to this will squeeze the degree of freedom and you will have plentitude of insignificant parameters. Moreover, you can try to apply a general to specific approach by taking different lags and see how your estimates look. In addition, after estimation try to implement am the requisite diagnostic test to see that your estimate is not suffering from something else different from lag selection before concluding the lag choice is the reason for insignificant parameters.
Best Regards.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publications