Use the lag length criteria to determine this. I suggest you use the Akaike Information Criterion (AIC) because Gujarati and Porter (2009) noted that the AIC is slightly superior based on forecasting performance within a regression model, not only for in-sample analysis but also for out-of-sample analysis. Since your study focuses on the use of VECM estimation techniques then the AIC is the way to go. This is very easy to construct particularly if you use Eviews or Stata.
In the alternative, some studies are of the view that if your data is collated on an annual basis, then a lag length of one is appropriate, while quarterly data should have a lag length of a maximum of 4 and monthly data should have a maximum of 12. This implies that if you want to follow this rule, then start by formulating a lag length that would suit your data based on the number of years. Mind you, you may still need to do some parsimonious reduction to get the optimal lag length so I suggest you go with the first option and let the computer work out the appropriate lag length for you.
Dear Clement Menard,
Given that you are utilizing daily data, I would suggest employing a monthly cycle, which would entail a maximum lag length of 30. As Olajide O. Oyadeyi Olajide correctly stated, the maximum lag length increases when shorter time periods are considered (the maximum lag length with quarters should be 4, with monthly data 12). However, I would diverge from this approach for daily data and utilize 30; the optimal number can then be evaluated with an information criterion.
Best regards,
Florian
Use Akaike method to determine optimal lag length that is based on minimizing error in estimation.
The maximum lag length should be determined based on model criteria such as AIC (Akaike Information Criterion) or BIC (Bayesian Information Criterion), which balance model fit and complexity. Generally, you should start with a reasonable number of lags (e.g., 1 to 12) and use these criteria to select the optimal length.
The maximum lag length should be determined based on model selection criteria like AIC (Akaike Information Criterion) or BIC (Bayesian Information Criterion), which balance fit and complexity. Common practice involves testing a range of lags (e.g., up to 12) and choosing the length that minimizes these criteria.
The optimal lag length depends on the specific data and context, but a common practice is to use information criteria like AIC (Akaike Information Criterion) or BIC (Bayesian Information Criterion) to determine it. Generally, the maximum lag length is often set between 1 to 12 for monthly data or up to 4 for quarterly data, but it can vary based on the specific use case and dataset.
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Nucleic Acids
- RNA