I am having a countries panel data on macro variables<gdp, inflation etc) and a continuous quality indicators variable . My models suggests that the quality indicator is a factor for gdp to reach its potential output. I would like to construct an index of cost as a descrete continuous variables defined as: (gpd(potential) - gdp current) /indicator (potential) - indicator (current)

. In other words I need to find the what is the ideal value the indicator should assume for each period in order gdp to be on it's potential value. Once found then need to construct the index of cost

I will take the hp filter in order to calculate potential gdp and gdp gap. My question is it possibile to use hp filters on quality indicators or another way is more appropriate (I e likelihood) ? If so, which one?


  • $\begingroup$ Not really trying to answer the question thus the comment... You’re right to question doing signal extraction on quality variables but setting that aside— please, please consider using a band pass (Baxter King) filter instead of the HP, which has a host of well-known issues that create weird artifacts. $\endgroup$ Dec 5, 2020 at 2:53
  • $\begingroup$ Also Hamilton has questioned hp filter and proposed his test. Empirical tests show hp filter is less volatile than his. Yet, is it possible to apply filters on quality variables and if not what's the alternative? $\endgroup$
    – David K
    Dec 5, 2020 at 9:44


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.