1. Price and/or wage stability can be destabilizing; Keynes (sometimes) presented stickiness as a policy recommendation rather than a necessary fact.
2. Financial fragility, a’la Hyman Minsky.
3. Noise traders matter for macroeconomics.
4. Moving Keynesian economics away from the consumption function and IS-LM.
5. Behavioral and psychological imperfections are relevant for macroeconomics.
6. They resurrected interest in the idea of an endogenous money supply.
I’m not entirely on board but overall that is a strong list. These ideas have now been incorporated into the mainstream, but only recently, typically in the 1990s and sometimes by Brad himself.
The post Keynesians went wrong, however, in several ways:
1. Too many of them obsessed over incomes policies. On "public choice" more generally they were naive.
2. Too many viewed Thatcherism, the war against inflation, and other developments as instruments of class warfare, designed to redistribute income from poor to rich.
3. They held too many debates over "what Keynes really meant" and cared too much about distinguishing themselves from mainstream Keynesians, whom they often treated as their worst enemy.
4. They ascribed too much power to the liquidity premium on money, and thought it was easy for an economy to end up in a corner where no one wants to produce new investment assets.
5. They failed to model the complexities of system-wide wage and price stickiness, and the differences between real and nominal stickiness, a’la Romer (ReStud 1990) or Caplin and Spulber. For this reason at some point they stopped coming up with insights on stickiness and fell far behind the rest of the profession.
6. They failed to develop an empirically progressive research program that kept pace with mainstream macro.
7. They failed to realize that the future of Keynesian economics lies in the elements of Milton Friedman — nominal stickiness is real and the action is with monetary policy — that were most Keynesian.
Overall post Keynesianism has been a far more useful heterodox thread than Sraffa and neo-Ricardianism.