Monetary economics, as any other academic field, requires basic agreement on a limited but vital set of definitions in order to be able to debate in a meaningful way. For central banks, two important terms are the distinction between monetary policy targets and monetary policy instruments. At the risk of stating the obvious, the targets of a central bank is generally a fairly easily verifiable variable that is supposed to be influenced by the central bank in the hope of achieving good economic outcomes. In most cases this target currently is the inflation rate, but other might also be nominal GDP, a monetary aggregate or even an exchange rate. The influencing of this target involves adjustments to the instruments of monetary policy, which include things like open market operations, the discount rate and reserve requirements.
In the choice of the target it is also of vital importance to specify whether we are talking about a growth rate target or a (growing) level target. Our current system works using growth rates: no matter what happened in the past, the central bank tries to get inflation rate back to its target of generally 2%. This is a sort of let-bygones-be-bygones policy. A growing level target, such as the one championed by Scott Sumners with NGDP level targeting, on the other hand, tries to get the target variable back on track to reach its pre-crisis trend level, meaning in the case of a depression would require considerably stronger monetary easing than a mere growth rate target. A mix of sloppyness and lack of clarity on my side has sometimes led me to muddle the two together in the past. “Basically, it is a central bank’s role to stabilize NGDP growth, and it should do whatever it takes to do so” is clearly a bit ambiguous, so let me try and set some of that straight by shortly going into why this difference is important
McCallum (.pdf) for one, in a survey of the literature on the topic, seems to favor a growth rate target. He stresses that, again by essentially ignoring the past, such a target essentially introduces a kind of random walk into the time series, and theoretically allows for the target variable to wander arbitrarily far away from any predetermined path. But then again, it wasn’t the point to begin with to follow any predetermined path. A growing levels target, on the other hand, by often requiring a more activist monetary policy in response to shocks, would tend to increase cyclical swings in demand, potentially increasing the variability (note: not the variance! I am not quite sure how accurate the term “variability” is, but I’m sticking to McCallum’s definition here) of variables such as real GDP in the process. In his words, “variability in output or other real aggregative variables is probably more costly in terms of human welfare than is an equal amount of variability in the price level about a constant or slowly-growing path”. Also, if what the economy just experienced turned out to be a permanent shock, stubbornly clinging to a predetermined growing level path would clearly be counterproductive.
The argument concerning the permanence of shocks is well-taken and important. As he notes, “although it is not entirely clear that fully permanent shocks are predominant, most time-series analysis seems to suggest the effects of shocks are typically quite long lasting – indeed, are virtually indistinguishable from permanent”. As far as I can tell this depends not so much on the nature of the shocks analyzed but rather on the policy response to that shock. If a central bank has a growth rate target, and thus does not try to force the economy back to its pre-crisis growth path, the obvious result would be to make a shock permanent even if it actually was not. How permanent a shock is would therefore seem to depend in practice, more often than not, on the adequacy of policy response to that shock rather than on the shock itself. Also, just because an economy experienced a “permanent” shock in one area doesn’t mean that with adequate policy another area of the economy can take up the slack. The bursting of a housing bubble might well be a permanent shock to the housing market, yet there is no reason it has to be a permanent shock to the economy as a whole.
And what the variability argument is concerned, I would also not agree that a lower but more stable level of economic activity, as we are currently experiencing, would be more preferable to a few years of increased variability after which we can then return to operating at full capacity. Sure, variability is a issue, but so is demand shortfall. When increased variability comes from an upswing it would seem like the lesser of the two evils. Further, the mere announcement that the level and not just the growth rate would be targeted implicitly states, as already mentioned, that a central bank is aiming to be more activist, which in an of itself might reduce the magnitude of downswings and thus variability through its effect on expectations. In any case it would seem sensible to, as generally with monetary policy, err on the side of doing too much rather than doing too little, and as such a growing levels target seems to be the superior choice.