With monetary exchange, and both recessions and booms.
This model really does work in a 3D Edgeworth Box. (I messed it up a bit in my last post, with 3 types of agents.) The contract curve is the diagonal line in the AC plane, and the endowment point is halfway along the B axis. But agents can only move in the AB plane and the BC plane. They can't move directly in the AC plane, which prevents some Pareto Improving trades at non market-clearing prices. (Did I say that right?)
It's a one-period model of a symmetric pure exchange economy with 3 goods (A, B, and C) and 2 types of agents. The (thousand) A-type agents each have an endowment of 200 A's and 100 B's. The (thousand) C-type agents each have an endowment of 200 C's and 100 B's. They all have preferences U=log(A)+log(B)+log(C).
Let good B be the numeraire so Pb=1 (not that this matters in what follows).
Competitive market-clearing equilibrium is (obviously): Pa=Pc=1, and each agent consumes 100 of each of the 3 goods.
What happens if Pa=Pc=2, and prices are sticky and won't fall to the market-clearing level?
If this were a barter economy, it would make no difference at all, because the relative price Pa/Pc is still at the market-clearing level, and there is no trade in good B anyway. Agents would still swap 100 A's for 100 C's, and each would still consume 100 of each good.
But suppose direct exchange of goods A and C is impossible. Because goods A and C cannot be transported, and must be consumed at the endowment location. And agents are anonymous, so promises are not credible. Only good B can be transported, and so good B is used as the medium of exchange. There is an AB market in one location, where A is traded for B; and a CB market in another location, where C is traded for B. There is no AC market.
There is an excess supply of A and C. The A-types can buy as much C as they want, but can't sell as much A as they want. So they choose the point where their (Marginal Utility of C/Marginal utility of B) = Pc = 2. They consume 150 A's, 100 B's, and 50 C's. The C-types can buy as much A as they want, but can't sell as much C as they want. So they choose the point where their (Marginal Utility of A/Marginal utility of B) = Pa = 2. They consume 50 A's, 100 B's, and 150 C's.
That is what a recession looks like. There is too little trade in A and C. The A-types would like to offer the C-types a deal: "I will buy 50 more of your C's for 100 B's if you agree to buy 50 more of my A's for 100 B's". And the C-types would like to accept that offer. But they can't make the deal stick, because individuals are anonymous, so they only accept B (cash) on the nail. The unemployed electrician and unemployed plumber would barter their services if barter were easy, but barter is usually very very hard in the real world. (Because the real world has a division of labour much more complex than my tiny model, and barter deals that require unanimity would typically involve hundreds of individuals all agreeing to the same deal, which is hard for all sorts of public choice reasons.)
What do booms look like?
If prices were stuck too low, like Pa=Pc=0.5, so there is excess demand for A and C, we also get too little trade in A and C, and we get exactly the same [edit: an even worse, see Max's comment below] consumption vector as in my previous example where Pa=Pc=2.
But you don't normally see this result in macro models. That's because (New Keynesian) macroeconomists usually assume monopoly power, so Pa and Pc are above the competitive equilibrium even in "normal" times, so that an increase in the endowment of B (an increase in the money supply) with sticky prices will cause the economy to move towards the competitive equilibrium, and trade in A and C increases. Which is what a boom looks like. A "boom" is just a smaller recession than normal.