Suppose you believe that the US economy needs increased Aggregate Demand, and needs looser monetary policy to accomplish that, and needs asset purchases by the Fed to accomplish that. (I believe those  things, but am not going to argue them here). What sort of assets should the Fed buy?

I want to divide all assets into two classes: "pro-cyclical" assets and "counter-cyclical" assets. I think the Fed should buy pro-cyclical assets. I'm going to define them in a minute.

I believe there exists a long-run equilibrium path for the US economy, with higher actual and expected inflation, actual and expected GDP growth and employment, and an associated monetary policy that is consistent with that long-run equilibrium. But the US economy is currently not on that path. The problem is, how to get onto that path?

As the recent "Kocherlakota controversy" reminds us (see Rajiv Sethi's excellent summary), we can't (in this case) just find the monetary policy that is consistent with the desired long-run equilibrium, and implement it. Because even though the policy is consistent with the long-run equilibrium, implementing that policy may not actually cause us to go to that long run equilibrium. There's the problem of stability of equilibrium. But to my mind, that controversy is now just beginning to become fruitful.

"Tinkerbell" is a useful personification of self-fulfilling beliefs. If the economy were on its desired long-run equilibrium path, the Federal Funds Rate would be higher than it is now. But the only way we could raise the FFR and get to that path would be if Tinkerbell simultaneously told everyone that we were now on that path, and if everyone believed her absolutely. Tinkerbell gets the economy to jump to the new Rational Expectations equilibrium time-path, and the Fed's raising the FFR is just part of that jump. But if some people never heard Tinkerbell, or didn't believe her 100%, then she would fail. All the forces of inertia in expected inflation would work against Tinkerbell, by raising the real interest rate, lowering AD, lowering real output, employment, and inflation.

Tinkerbell is powerful, but not superhuman. We need to make sure the causal chain from policy to outcomes works with her, not against her. The Fed needs to bring her onside and build her credibility. That way people would see that things were moving in the direction of Tinkerbell's prophecy, which would increase her credibility, which would move the economy still further in the right direction, which would increase her credibility still further, and so on. People would learn that Tinkerbell is right.

Asset prices are forward-looking. They depend on people's beliefs about the future. If people believed Tinkerbell, and thought the US economy was moving back onto its desired long-run equilibrium path, some asset prices would rise, and some would fall. I call the former "pro-cyclical" assets, and the latter "counter-cyclical" assets. I want the Fed to buy pro-cyclical assets, not counter-cyclical assets.

If the Fed buys an asset, the direct effect of the purchase will be to raise the price of that asset. The increased price of that asset, plus the increase in the money supply used to purchase that asset, will have a direct effect on the economy. But there's also an indirect effect, via Tinkerbell's credibility. An increased price of a counter-cyclical asset will lower Tinkerbell's credibility, which will also have an effect on the economy.

Suppose the Fed buys a counter-cyclical asset. If the price rises, people may interpret that rise as a sign that monetary policy is having the desired effect. Or they may interpret it as a sign the economy is getting weaker. Depending on how people interpret the rise in price of the counter-cyclical asset, and the relative strengths of the direct causal effect and the Tinkerbell effect, the net effect on the economy is ambiguous. Also, if people thought that monetary policy was having the desired effect, and was not impotent, any increased optimism about the future path of the economy would tend to lower the price of the counter-cyclical asset, which would tend to make monetary policy look less effective, and snuff out that optimism.

Suppose the Fed buys a pro-cyclical asset. If the price rises, people will interpret that as a sign that monetary policy is having the desired effect. Or they may interpret it as a sign the economy is getting stronger. Both effects work in the same, desired, direction. Also, if people thought that monetary policy was having the desired effect, and was not impotent, any increased optimism about the future path of the economy would tend to raise the price of the pro-cyclical asset still further, which would tend to make monetary policy look more effective, and reinforce that optimism.

I leave it as an exercise for the reader to decide which assets are pro-cyclical and which are counter-cyclical. But Federal government bonds are almost certainly counter-cyclical under present circumstances. Nominal interest rates (both short and long) will rise if the economy moves towards recovery. Of course, there are also practical considerations about which assets the Fed should buy. The Fed is probably not ready to enter the housing market as a landlord, collecting rent cheques instead of bond coupons.