Hyperparameters for monthly data? about bvar HOT 1 CLOSED

Hey,
Thanks for the kind words, glad that you find the package helpful.
That's a good question – it's important to think of the model and priors in the context of the data. In general, the standard setup should work with quarterly and monthly data. However, it may be worth investigating the sensitivity to $\alpha$, which pulls more distant lags closer to the prior mean (the prior variance includes the factor $s^{-\alpha}$). For monthly data, you may want less shrinkage than the default value of 2.
You can address this by either fixing alpha somewhere else (just set the mode via bv_alpha()), or by treating it as a hyperparameter. For this, just use bv_priors(hyper = c("lambda", "alpha")), and possibly tweak the range and prior for it.
Hope this helps, let me know if you have any more questions.
Best,
Nikolas

from bvar.