In my first post:
Core CPI is fine without including substitution effects, I think (BTW, core CPI definitely includes healthcare costs). Oh, I could see using a chained CPI ex-food and fuel (a 'core chained CPI?') because it will be less volatile while still keeping track of substitution effects. Yet I doubt monetary policy would be significantly different either way - the differences we're talking about are fractions of a percent. Relevant for indexation, since it accumulates, but monetary policy is such a blunt instrument anyways, they just try to keep inflation within a rough 'target band' rather than a specific number.(Note that I think this is irrelevant from a monetary policy perspective - IMO core CPI is more than adequate for guiding monetary policy.)