Like all New Keynesians, I believe two things that are apparently contradictory:
1. We each get paid too much; 2. Because of 1, we all get paid too little.
It only makes sense if you understand fallacies of composition. What is true of each part isn't necessarily true of the whole.
It's a very Hobbesian perspective. It's not a war of Us against Them; it's a war of All against All. Those who can only think in an Us against Them perspective will have the hardest time grasping it. You have to think in a Hobbesian perspective. Or representative agent model, as we call it nowadays. The distinction between the individual and the collective good is what's at issue, not the distinction between the good of one group of individuals and another group of individuals.
I have been thinking like this since I was 18, but I only understood it vaguely back then. I understood it better when I was 30, and started building macro models with monopolistic competition.
Each individual faces a downward-sloping demand curve. Taking other individuals' prices as given, the higher I set my price the lower the quantity I can sell. That's because we are not all identical. Nobody is a perfect substitute for anybody else. So we are not in perfect competition. So each individual faces a trade-off between price and quantity.
Microeconomists are with me so far; here's where I'm in danger of losing them.
But the demand curve facing all of us is horizontal. There is no trade-off between relative price and quantity at the aggregate level. The price of everything relative to everything must be one. And because the trade-off facing each individual is different from the trade-off facing all of us, each individual will choose a point on the trade-off that is bad for all individuals.
OK, let's go back and round up the stray microeconomists (and probably a few macroeconomists as well).
Every microeconomist knows that it's relative prices that should be on the vertical axis of demand curves. Suppose one individual wanted to sell more. Unless he is a perfect substitute for other individuals, he will need to lower his price relative to other individuals' prices. But if all individuals want to sell more, and all lower their prices, the average relative price does not change. If we aggregate up over individuals' downward-sloping demand curves, we get an Aggregate Demand curve, with the average relative price on the vertical axis, that is horizontal at one.
And now I've lost some macroeconomists too. Because I've drawn an AD curve with relative price on the axis, rather than the nominal price level. But don't worry. Remember that the AD curve is an equilibrium locus. It's a set of points at which output equals output demanded. We can ask what happens to the nominal price level as we move along that set of points, and that's the normal AD curve. But we can also ask what happens to relative prices as we move along that set of points, and that's the AD curve I'm talking about here.
When we talk about being paid too much, we should always understand this in real terms. What matters is not how many dollars we get paid; it's how much we can buy with those dollars we get paid. It's the relative price, not the nominal price that matters.
Because each individual faces a downward-sloping demand curve, but all individuals together face a horizontal aggregate demand curve, each individual will choose a price that is too high and a quantity that is too low. Because one individual's selling price is another individual's buying price, all would be better off if all cut prices and increased quantity. Prisoner's dilemma. The attempt by each individual to raise his own real income above the competitive equilibrium results in a lower real income for all.
If all prices are sticky, then an expansionary monetary policy, which increases aggregate demand, increases output and makes everyone better off. That's why booms are good, because it brings the economy closer to the competitive equilibrium. And why recessions are bad, because they take the economy further away from the competitive equilibrium.
And cartels, like labour unions, just make the problem worse. Because by joining together with similar sellers into a group, the demand curve facing the group is steeper than the demand curve facing the individual, since members of the group no longer compete against each other for buyers. So there is an even bigger difference between the downward-sloping trade-off facing the group of sellers and the horizontal trade-off facing us all.
Unions are bad for the very same reason that recessions are bad.
All New Keynesian macroeconomists have understood the above for the last 20 years. Which is why all New Keynesian macroeconomists are fundamentally opposed to cartels, labour unions, minimum wage laws, etc.. OK. It's why they should be opposed to such things.