Yesterday I was really scratching my head over this post by Scott Sumner. I wasn't quite sure how to respond to it, or if I even should. On the one hand he was making some strange statements about Keynesianism, but then he also attributed that to a "naive version" of Keynesianism. And his alternative sounded an awful lot like... well... like Keynesianism to me so I wasn't sure what to do with it.
Brad DeLong and David Glasner have good responses, though (Brad's is to a related post - not the one linked above). DeLong had the same reaction that Sumner seemed to be acting like he was disagreeing when he more or less agreed - which made me feel better).
Then Nick Rowe wrote an even stranger post that also seemed to take issue with pretty elementary Keynesianism, and instead offering... more pretty elementary Keynesianism. He doesn't like the concept of saving because it obscures whether that saving is invested, hoarded, or spent on goods produced in a previous period. But it only obscures it if you're teaching it in an shockingly neglectful way!
When I taught my students Keynesian macro last semester, whenever I would derive the S=I from Y=C+I+G and then the multiplier too I would always have a money supply and liquidity preference schedule right next to it on the dry erase board. And then right next to that I would have a loanable funds market. Nick can fault me for not having an antique furniture market up there too, but I think we did pretty well for a group of freshmen. But the thing is, that's the whole argument. Keynes didn't talk about saving without talking about liquid assets. I didn't learn Keynesianism with saving but no liquid assets. And I'm sure as hell never going to get up and teach it without it. That's the whole point.
Scott agrees with Nick and says we should be teaching M and V rather than S and I. I'm not sure what the point would be of replacing it. We currently teach all of it. We talk (perhaps unfairly) about classical notions of the neutrality of money with the quantity theory, but then we also talked about the relationship between V and liquidity preference.
I don't see how this really solves the problem either. After all, we can talk about a quantity theory with current output, or we can talk about a transactions quantity theory (the only one that really makes sense) with antique furniture and liquid assets. But then if we do the transactions version that's not NGDP anymore. It's nominal transactions. You're facing the same dilemmas that you had with the Keynesian model. It doesn't solve any of this stuff - you have to talk about these issues in either case.
The other problem with just jettisoning savings and talking instead about investment, hoarding, and antique furniture buying is that these aren't distinct in the real world. One of the points that Keynes makes emphatically is that different assets have different degrees of liquidity, and that an increase in liquidity preference shifts the demand for liquidity across a whole spectrum of assets. Even what we traditionally think of as "investments" have some liquidity properties to them. Think about investing in a company's stock vs. investing in your buddy's company. There's a difference in the liquidity of those investments. Or the easiest thing to talk with students about is their own savings - think about what you have in a checking account vs. a savings account vs. a longer term savings vehicle. All of it's intermediated by the banks into some other investment, but each is progressively less liquid. But you can't point to a given dollar and say "that is hoarded" or "that is invested".
None of this is to disagree with the points that Nick and Scott are making on the economics - but I do think pedagogically they're worrying about the wrong thing - or perhaps just denouncing a straw man version of "savings" in macroeconomics when in reality it's a very useful concept.