As usual, I want to talk about crude empiricism vs microfoundations. I will argue that we should look at what has been tried in the past and whether it seems to have worked. This means I think we should not just keep our heads to down and to the grind-stone working on microeconomics until we have done it right and only then dare to attempt to advise policy makers. I have nothing against empirical microeocomic research, but I think this question is whether there are macroeconomists who have something useful to say even before the empirical microeconomics research program is completed.
I want to consider an example from another field of research -- aspirin. For all of human history, it has been reported that the roots of willows contain an analgesic. I do mean this literally -- it was reported by Herodotus. The pain killer was purified and named salycilic acid. There were three problems -- its mechanism of action (microfoundation) was completely mysterious, it caused ulcers, and it tasted horrible. The third problem was crucial, but it was solved roughly a century ago by the same reaction which gave us heroin Acetylation. Now we have access to Aceto Salyclic Acid AKA aspirin. Decades later, the mechanism was understood. Aspirin blocks synthesis of a number of molecules collectively called prostaglandins.
Now the research which explained aspirin's effect was worthwhile (well worth the real Nobel prize that was awarded). But there was no reason to refuse to take aspirin and suffer when its side effects were known but its mechanism was mysterious.
The micro (in this case nano) reasearch was useful. It made it possible to develop other non steroid anti-inflamatories. Importantly, it was understood that there were two related moelecules one of which caused inflamation, feaver and tenderness and the other of which caused the stomach and small intestine to releqase mucouse to protect themselves. This scientific research, followed by a massive controlled trial, lead to the availability of Vioxx and over 50,000 deaths.
I think the example is relevant for those who wonder if macroeconomists should stick with tried and true models (not of couse the academic literature dominatinng tried and false models) for a good long while.
Of course the old Keynesian models have their empirical failures. Such as ... uhm well not the stagflation of the 70s which was consistent with the expectations augmented Phillips curve used by the old Keynesians throughout the 1960s (starting with Samuelson and Solow 1960) but, well there was the failure to consider the hypothesis that there is Ricardian equivalence which is implied by almost all mroe modern models and supported by no data of which I am aware at all.
Well I am sure that there must have been something badly wrong with those models which were abandoned before I took my first economics course.
But seriously, I know of no difference between using those models and using aspirin in the 1960s.