In discussing macroeconomics' Faustian bargain, Simon asks:

By putting all our macroeconomic model building eggs in one microfounded basket, have we significantly slowed down the pace at which macroeconomists can say something helpful about the rapidly changing real world?

Let me deepen this question by pointing to five newish facts about the "real world" which any good, useful macro theory should be compatible with.

1. The unemployed are significantly less happy than those in work. This doesn't merely provide the justification for an interest in macroeconomics. It also casts grave doubt upon RBC-style theories which unemployment is voluntary. Sure, at the margin, some people might choose to be unemployed. But the fact that most unemployed are unhappy suggests that most unemployment is involuntary.

2. Price and wage stickiness is over-rated. One Bank of England study has found that:

Nearly half of firms changed their prices within three months of an increase in costs or a fall in demand

This suggests that price stickiness isn't universal. And whilst pre-recession evidence pointed to some wage stickiness, more recent evidence has cast doubt upon this. Models in which stickiness plays a central feature might not, therefore, be well-grounded empirically.

3. The failure of a handful of organizations can have massive macroeconomic consequences. The great recession originated in the collapse of a few banks. This tells us that we need models in which micro failures generate macro ones, of the sort Xavier Gabaix has proposed.

4. Supply shocks do happen. It's improbable that all productivity fluctuations are due merely to labour hoarding in the face of demand shocks. One way in which they might happen has been described by Gabaix - the failure of a few big firms shows up in macro data as a fall in productivity.

5. Interactions between agents can magnify fluctuations. We know there are expenditure cascades, which occur because consumers copy other consumers. And it's likely that animal spirits can spread in much the same way as diseases do (pdf) - because optimism and pessimism are infectious. We can talk ourselves poorer.

These facts are a challenge to both RBC and New Keynesian models. But they have something in common. They stress the heterogeneity of agents: some unemployed respond to benefit changes but most don't; some prices are sticky but others aren't; some firms are big enough to have macro effects, others aren't. This, I fear, means that the problem with conventional macro isn't so much its microfoundations per se as the assumption that these microfoundations must consist in representative agents.

Now, you'll object that any model that tries to embed these five facts will very quickly become very complex. But that's my point - the economy is a very complex system and any theory that doesn't see this is very dubious. It could be that the best way to understand it - if at all - lies in agent-based models rather than conventional DSGE ones.

However, such models are - for now - beyond the understanding of most economics students (and me!). Which poses the question: what should students be taught?