Number of lags too high

[Stefano Vertua]

Hi everyone,

I am writing as I am currently working on my bachelor's thesis with Stata.
I am doing a time series analysis.
I have found that 4 is the optimal number of lags, I have one cointegrating equation, and the variables are stationary at first difference.
However, my professor pointed out that, because I am estimating a VAR of dimension 5 with 4 lags, the estimated parameters are far more than the observations.
The problem is that I do not really know what to do. Because of my research topic, I cannot really change my data.
Therefore, what do you recommend me to do? To perform an additional test? To simply state this issue in the limitations of the thesis?

Thank you in advance for the help (I'm still a beginner in Stata)!

[Daniel Schaefer]

Hi Stefano,

I agree, it is a problem to have more parameters than observations! I don't know exactly what you are trying to do, so I can't really say what you ought to do instead. That being said, if you can't change the data, I'm wondering if you can change (or simplify) your modeling approach? For example, perhaps you can safely reduce the number of dimensions?

I would strongly suggest that you take any advice you see here, think it though yourself, then make an appointment with your professor and talk the problem through with them at length. I think your professor is your best resource here.