There are many things right with New Keynesian macroeconomics. It is a very good synthesis of many strands in Monetarist, Keynesian, and New Classical macroeconomics. But this post is not about what's right with New Keynesian macroeconomics.
There are also many things wrong with current New Keynesian macroeconomics. New Keynesian macroeconomists are aware of many of those things. But this post is not about all the things that are wrong. It's about some of the things that are wrong. I'm coming at this from a "Market Monetarist" (aka Quasi Monetarist) perspective.
1. New Keynesian macroeconomics makes no sense whatsoever in a barter economy. And yet nowhere is the assumption of monetary exchange made explicit. You can't see money in the model, but it must be there somewhere, or the model just wouldn't make any sense.
1a. The producers in New Keynesian models are imperfectly competitive. They choose to set price above marginal cost. If two producers have price above marginal cost, both could gain by a barter deal in which they both produce additional output and exchange it at the ratio given by their posted prices.
1b. Even if we replace imperfect with perfect competition, and assume prices are fixed, it would be impossible to have a recession with excess supply of all goods in a barter economy. Two producers who wanted to sell more but couldn't find customers would simply produce additional output and do a barter deal.
There is something fundamentally wrong with a model that implicitly assumes monetary exchange but does not make this assumption explicit, and does not even have money in the model.
2. New Keynesian models explicitly assume that money is the medium of account. But again, there is no money in the model. Prices are set in terms of a good that does not exist.
3. New Keynesian models explicitly assume that the nominal rate of interest is set by a central bank. Central banks have special power over monetary policy only because they issue irredeemable money. Yet that irredeemable money does not appear in the model.
4. New Keynesian models have only one transmission mechanism for monetary policy -- the effect of current and expected future interest rates on desired consumption and investment.
4a. We know that a permanent change in the stock of central bank money will change the long run equilibrium values of all nominal variables but will leave interest rates unchanged.
4b. We know that monetary policy will still affect nominal variables in the long run, and real variables in the short run (assuming sticky prices), even in a world with no interest rates, no borrowing or lending, or where interest rates are set by law. So the interest rate channel cannot be the only possible transmission mechanism for monetary policy.
4c. We can imagine a world in which central banks implement monetary policy in many different ways than by setting a rate of interest. For example, they can set the price of a commodity like gold, and vary that price. Or they can give away money for free, and vary the quantity they give away.
4d. Central banks that issue irredeemable money aren't even banks. A bank both borrows and lends. It has assets and liabilities. An irredeemable liability is not a liability, so central banks don't have liabilities. They don't even need assets, unless they choose to buy back some of their money in exchange for those assets. A central bank, since it is not a bank, does not have to borrow or lend, and need not have anything to do with interest rates.
5. New Keynesian macroeconomics asserts that central banks must lower current or expected future real interest rates in order to get an economy out of a recession. This assertion is, I believe, often false. The IS curve will probably slope upwards in a recession. Recovery from a recession is compatible with an increase in real interest rates.
6. This to me is the killer. New Keynesian models lead good economists, who correctly diagnose the monetary nature of the recession, at the same time to believe that monetary policy is powerless at the zero lower bound. And recommend fiscal policy instead. This is like correctly diagnosing magneto trouble, then recommending we all get out and push the car, rather than fixing the magneto. I just refuse to accept that that's the best we can do. We need to understand that monetary magneto better, and learn how to fix it. And it is my frustration with this lack of correspondence between diagnosis and policy prescription that has lead me on my three year search for something better.
I have been eclectic in my search, taking ideas that seem useful regardless of their source. Some are Monetarist. Some are Old Keynesian. And Clower is as much a Keynesian as a Monetarist. I side with Silvio Gessell against Keynes on the role of money in general gluts.
Like other "Market Monetarists" I think that monetary policy can cure what is at root a monetary problem. But we don't all agree on everything. Lars Christensen provides a good survey (pdf). This post is in part a response to Arash Molavi Vassei. Scott Sumner responds here. Josh Hendrickson here.
All equilibrium theories have a disequilibrium story on the side. If the demand curve for apples shifts right, that creates excess demand for apples at the old equilibrium price, so individual sellers can raise their prices above other sellers' prices and still sell their apples, and this process is what gets the price to the new equilibrium. In monetary economics we call this disequilibrium story the monetary policy transmission mechanism. The interest rate transmission mechanism is the New Keynesian disequilibrium story. It's not the only possible story. It's not even a very good story, as I argue above.
My MX6 developed "magneto" trouble last Summer, 100kms away from home. Replacing the alternator is a 2 hour job, and I didn't want to do it at the side of the road. So I bought a new battery at Canadian Tire, replaced the old battery when it finally died, and that got me home. I replaced the alternator the next morning. There are circumstances when a bodge job is the best you can do. But it's not really satisfactory.