nominal GDP level targeting
There's a been a lot of talk in the blogosphere over the last few years about targeting GDP levels.  The naive thing to do might be to target the growth rate, but this leads to instabilities; both for reasons of stability and credibility, it's important to at least partly correct for past mistakes, so that a period of excess growth is followed by a reduction in target levels.

On some level, though, only partially adjusting is likely to be more credible than a full rigid level targeting regime.  For one thing, if we set the level optimally now (at the beginning of the regime), there may be more reason to believe that we would be tempted to "start over" in the future, resetting the target to a new path.  For another, if there is a large shock to the economy (in either direction), the temptation to reset the target will obviously be much stronger if it's rigid than if the regime itself dictates a partial adjustment.  Accordingly, it makes sense to me to set the initial target path in part on the basis of past changes in GNP; in a simplistic setting these would be regarded as "sunk", but my contention is that they interact with policy going forward to affect expectations, and therefore should factor into policy decision-making.

One particularly cute way of achieving this is to set the nominal GDP target level, two years ahead, at some multiple of the total GDP since the beginning of the republic.  If the growth rate is e.g. 5%, this would mean that shocks are accommodated, instead of resisted, after about two decades; it seems likely to me that we would want a shorter accommodation period, in which case the target level should be a discounted sum, but the charts I've produced here are based on the naive sum.


The data from FRED go back to 1947.  Making very conservative assumptions about what happened before then, we can pin down the ratio of current GDP to total accumulated GDP to between the lines.  As inflation has come down, the ratio has gradually come down over the past couple decades; it has been helped down more recently by a large recession (which you may have heard about).

In all the plots, the natural units of the vertical axis are 1/year, i.e. I'm dividing an annualized GDP by a sum.

Insofar as I've proposed a two-year ahead level target, this is probably the plot to look at.  A 5% annual increase would suggest a multiple of 0.055 or so, in which case the Fed would be encouraged to adopt a contractionary stance from current levels; in essence, it would still be fighting back at inflation from the nineties, and even to some extent the Carter years (which still get 1/8 as much weight as the most recent quarter does; all four Carter years count the same as the last 6 months).  Most likely a shorter backward-looking window would be desired.

Update: At the risk of belaboring the obvious, the Fed could adopt this long-window backward-looking plan with a ratio of 0.06 or even (on the basis of the middle of the last decade) 0.068 or something; I'm imagining the Fed wouldn't want to lock itself into that high a presumed rate of inflation.
- posted by dWj @ 1:27 PM