'Somebody Is Inside an Echo Chamber. But Who?'

Brad DeLong:

Somebody Is Inside an Echo Chamber. But Who?: Paul Krugman fears that somebody is trapped inside an echo chamber, hearing only things that confirm what they already believe...

But how can we tell which side has lost contact with the reality out there? ...

I do not think we have to decide. I think that even if we are uncertain whether the optimistic “insiders” or the pessimistic “outsiders” are correct, elementary prudent optimal-control theory tells us that we should act as if the “outsiders” are right.

But I have gotten ahead of myself:...

So how would we tell whether, right now, it is the outsiders are overstating the dangers to premature tightening, or it is the insiders who are understating the dangers to premature tightening here in the United States?

To answer this question, I think we need to consider five points–the first about our decision procedure, the second about the level of spending consistent with full employment, the third about the degree of uncertainty and variability, the fourth about the vulnerabilities of the economy to spending deviations above and below the projected current-policy path, and the fifth about the effectiveness of our optimal-control levers in different scenarios.

The first point is that if it turns out that we cannot tell–that we have to split the difference–then the considerations that rule are the asymmetries in the situation.

The second point is that no one right now has a good and convincing read on what, exactly, the level of spending consistent with full employment at the currently-projected price level is. Uncertainty is rife: if there was ever a time for considering not just the central tendency of the forecast but the risks on either side and taking optimal control appropriately valuing these risks seriously, it is right now.

The third point is that we are not just uncertain about what the proper full-employment path for demand is, we have much more than the usual amount of uncertainty about nearly all other dimensions of the structure of the economy. To suppose that any of the emergent properties that are policy multipliers can be estimated from data collected during “normal” times is to make an enormous leap of faith.

The fourth point is that downside risks to the forecast greatly exceed upside opportunities. ...

And the fifth point is that, while the Federal Reserve has powerful levers to restrict demand if spending shoots above the desired policy path, its levers to expand demand if spending falls below have been demonstrated over the past six years to be relatively weak.

Thus, if it turns out that we cannot tell–and we cannot tell–then it is not correct that we should split the difference. The considerations that rule are then the asymmetries in the situation. It is, right now, much worse to undershoot than to overshoot full-employment demand...

These asymmetries mean that, as far as policy is concerned, the “outsiders” win any tie and win any near-tie: the “insiders” should govern what policy should be only if there is not just a preponderance of the but clear and convincing evidence on their side.

Yet the Federal Reserve appears to have decided:

• that those who think that the economy is near full employment and is in a durable recovery have by far the better of the argument as to what the central tendency projected current-policy demand path is.
• that it is appropriate to make policy via certainty-equivalance.

Given the inability of the Federal Reserve to attain traction at the ZLB, its current frame of mind–which appears to be doing certainty-equivalence policy–makes no sense to me. Certainty-equivalence is appropriate only with a symmetric loss function and a symmetric ability to compensate for deviations on either side of the target. We do not have either of those.

Has there been an explanation of why the Federal Reserve’s policy is appropriate, given the uncertainties, given the asymmetry of the loss function, and given the asymmetry of the control levers, that I have missed? If so, where is it?