Costs of discretion

In the previous post, I argued upping the inflation target would have no impact on unemployment and it would risk unanchoring inflation expectations. If expectations became unanchored we would be trading off a permanent increase in expectations for a temporary decrease in unemployment.

Assume we gave half of those currently unemployed a job, output would rise by about 5%. Without retargeting, I’m guessing it’ll take 5 years for unemployment to get below 5% (unemployment has a two-year half life). So we’re getting about 13% of GDP by retargeting.

The costs of 10% of inflation are estimated between 1% and 5% of GDP per year. Woodford estimates that under a discretionary policy, inflation would be about 10% (I’m looking at figure 7.1 in his textbook). Discretion isn’t really permanent, its just really hard for the Fed to get expectations reanchored. Suppose x is the number of years of unanchoredness and so its the number of years we’d have 10% inflation.

How big can x be and still make retargeting cost effective? Using a low-end estimate of cost of inflation of 1%, x is 17 years. Using a high estimate of 5%, x is 2 years.

18 thoughts on “Costs of discretion”

  1. I do not fully trust your numbers because you basically describe a structurally-stable environment in which you consider changing the behavior of the central bank: we had a shock of some sort and now we’re doing transition dynamics back to the steady state and we’re thinking whether changing one of the policies would give us a better transition.

    I can understand you wanting to make that case but there’s also an alternative scenario under which we’re looking at a different policy environment and a different financial environment and therefore a different steady state. Just wanted to mention that.

    I would also quibble with the assumption that until now we had an (explicit or implicit) inflation target at 2 or 3%. The US Fed will accommodate stuff. A central bank that follows a target only when it’s convenient is not following a target. We have yet to see the Fed sticking to its guns through a non-trivial drop in output. In that sense, maybe the ECB has better street cred.

    While I risk being off-topic again, let me also point to the fiscal benefits of higher inflation now. VoxEU had a couple of pieces on that.

    Finally, the Fed could announce a path for inflation and make it happen at whatever cost it must bear, if it think it knows that trade-offs it’s facing, expectations-wise. They could do a hump-shapped inflation transition and if they do enjoy credibility now, that would be OK, no?

  2. AC, the name of my blog is “Ambrosini Critique” which is a play on the Lucas critique. I’m aware of your criticism. However, its not enough to state a possibility. You have to say what direction the effect will have and whether it will be empirically relevant. Otherwise, you could always scream “that parameter isn’t structural” to refute anything in economics, because we’re never talking about “deep structure” in economics.

    That said, I wrote a blog post. You shouldn’t trust my numbers.

    I think its interesting to contemplate why your other comments are off-topic. The reason given for increasing the inflation target would be to decrease unemployment. As I said in the OP, if the Fed enjoys credibility, if expectations are well anchored, then this will have no effect on unemployment. There may be another reason, though, why increasing the target may be a good idea. That other reason: it would screw foreigners who hold dollar denominated debt and our policy makers only care about Americans.

  3. Yes, I agree, if you want to “exploit” the Phillips curve you do it by inducing unexpected inflation, once, then you say you screwed up and resign. But will Bernanke do the right thing(tm)?

    But that’s not the kind of Phillips curve the proponents of more inflation have in mind, I suppose.

  4. P.S. The number I most worry about is the unemployment half-life. Maybe Mulligan is right and part of the current labor wedge is structural and not cyclical and so the steady state level of employment will be lower even in the long run.

  5. I see, Mulligan’s effect would make the unemployment rate have a faster half-life as people drop out of the labor force (after their UI expires, I imagine). This would have the effect of reducing the value of the retargeting policy, reducing my x’s.

  6. He has a list of new distortions (minimum wage hike, > 100% marginal tax rate on income of under-water homeowners, etc.) that theoretically increase the level of steady state unemployment. Actually, some of the stuff he published suggests he’s interpreting recent data as a transition to a lower BGP, with lower hours, no cyclical stuff.

    I don’t know if I believe it, but in part there should be some stuff to it.

  7. Upping the inflation target without doing any QE would likely do nothing.

    Upping the inflation target and doing more QE would permanently reduce the probability of hitting the zero lower bound again, for one, and reduce the probability we stay stuck much longer on the zero lower bound like Japan has been for the past 20 years.

    And we’re below the Fed’s own inflation target.

  8. Quit targeting inflation. Target nominal expenditure.

    A 3% growth path for nominal expenditure will result in a zero percent expected inflation rate.

    However, it would be consistent with raising nominal expenditure 11% over the next year because it is that far below trend.

    If productive capacity fell 8% of so below trend, then this would be inflationary in the short run.

    If this was a permanent drop in productive capacity, then it would imply a permanent increase of the price level–something like 8%. But there would not be a permanent increase in the inflation rate.

    Nominal expenditure targeting anchors long term inflation expectations just as effectively as inflation targeting.

  9. Inflation at 10%? C’mon. While we’re on the topic, as inflation is still under 2%, discretion is breaking from the implicit 2% rule and slowing down monetary easing.

    The change from 2-3% inflation represents a negligible direct change in welfare. 0% is not the optimal level of inflation: see Summers 1991. http://www.jstor.org/stable/1992697

  10. gabe, I’m not arguing whether or not 2, 3, 5, 20, 100, -10 or x% is the right inflation target. Maybe the target should be higher to make the zero lower bound less likely so we would avoid Krugman’s antics. I’m saying NOW is a bad time to change the target.

    If expectations are anchored then changing the target would have zero effect on unemployment. I know you argued otherwise, but I don’t know of a model that bares out your argument explicitly. I’d like to see a model, but let me turn it around on you. Suppose the Fed changed the target to 3%, how much would unemployment fall because of this change?

  11. The argument better fleshed out- at a zero lower bound, the natural rate of interest that would balance supply and demand of savings and investment, is negative. This means that nominal rate cuts are ineffective at a zero lower bound, especially given deflation. This means low investment, as well as other problems related to deflation and nominal rigidities. Thus even with anchored inflation, output can be affected.

    I thought this paper had a good model: The Zero Interest-Rate Bound and Optimal Monetary Policy, Eggertsson and Woodford 2003 http://www.newyorkfed.org/research/economists/eggertsson/BrookingsPaper.pdf

    They use a price-level targeting framework, which I think would probably be even better than inflation targeting (Nominal GDP targeting would probably be better to both though). EW find that increasing the price level target when the zero-lower bound is hit is superior than a fixed price-level target, which is similar to upping the inflation target. (see p. 185)

    Notice that this is still a rule, it’s just a state contingent rule. Is this different than discretion? This is somewhat rhetorical, but I wouldn’t consider a state-contingent rule to be discretionary.

    To answer the unemployment question: if the Fed is totally credible before and after the change, then expected inflation should rise by 1% and real interest rates should fall by at least 1% (more if long rates fall due to expectations of zero short rates for longer). How much would that lower unemployment though? Probably relatively a lot now, relatively a little at full employment. One could look at EW’s output gaps and use Okun’s law, but these estimates would probably be wildly optimistic. Are there any papers that examine changes in real interest rates and its impact on unemployment?

  12. That’s THE paper on this topic. Their key point is that managing expectations should be the only goal of the Central Bank. They say QE is one policy instrument for managing expectations and then they go on to, as you say, describe other policy instruments. The argument here is what is the best policy instrument and specifically is upping the inflation target a good way to manage expectations.

    I say no and its basically for the reasons I just gave in comments in the previous post. I think changing policy now would be more likely to unanchor expectations now because increasing the inflation target can look like a discretionary increase in money supply or cow-towing to political pressure. Perhaps a price level target makes more sense or a higher inflation target or even a state-contingent inflation target… I just think implementing this policy now is a bad idea. We do not know how expectations are formed and because of that we should be really conservative when we take actions that may change or upset expectations.

    We’ve already started QE and expectations seem to be settled down and for inflation (and not deflation). Why introduce a new policy now and risk messing with expectations?

  13. RE unemployment: your assuming that the policy we have now isn’t managing expectations correctly. EW’s comparison between their policy instrument and “steady inflation targeting” (i.e. only using interest rate policy) isn’t the right comparison for today’s policy environment.

  14. I’m trying to make sense of what you’re saying here but I just can’t.

    First you say: “If expectations are anchored then changing the target would have zero effect on unemployment. I know you argued otherwise, but I don’t know of a model that bares out your argument explicitly. I’d like to see a model…”

    Then, when sraffa links to Eggertson and Woodford you respond: “That’s THE paper on this topic. … The argument here is what is the best policy instrument and specifically is upping the inflation target a good way to manage expectations.”

    Your changing the topic, the question is this: does the Eggertson/Woodford paper provide a model in which changing the target would have an effect on output?

    The answer is yes.

    So the next question is: since you’ve clearly seen the Eggerston/Woodford paper why did you say “I don’t know of a model that bares out your argument explicitly”? You do know of a model, the one in the EW paper.

  15. Adam, did you see my next comment? The analysis in EW proceeds as-if the alternative policy is strict interest rate targeting. This is not what we have today.

    (I guess I should have qualified what I said… I don’t know of a model *with assumptions that apply to today’s environment* that gets a *substantial* effect from changing the target. The second qualifier I add because the standard model has a very small effect.)

  16. push- Would a change to a Sumnerian NGDP target in September 2008 have represented discretion? Just something to think about, esepcially since a 2% real growth rate implies 3% inflation.

    If a small change in policy (to a 3% inflation target) can unanchor expectations, then could any change in monetary policy keep expectations anchored? Basically, it seems like your position would imply that the Fed has to keep monetary policy the same as pre-2008 or risk “discretion” and high inflation.

    Doesn’t QE represent discretion? It is a change in Fed policy due to the crisis.

    Basically I think this concern over discretion is unfounded. If unemployment doesn’t decrease, it will be less likely that the Fed can withstand political pressures aligned against inflation-fighting as the costs in terms of further unemployment will be high. This also reduces the Fed’s credibility, maybe more than a small change in inflation target.

  17. If we were targeting levels before the recession, then we wouldn’t need to change the target. That’s the point of level targets in the first place.

    QE is weird — and so using it risks unanchoring expectations — but that policy has already been done. I’m saying a new policy NOW, when everyone is bitching about the Fed, would be a bad idea.

    My view of the political economy in this recession is that its not the unemployment rate that gets people to call their congresspeople and hold protests. Its the hire rate. If people are finding jobs, they’ll be happy. The unemployment rate is, mechanically, a lagging indicator even of the labor market.

  18. I realized today that 5% nominal GDP growth, subtracting pop growth rate (~2%) and eral growth rate (~2%) would mean an inflation rate of 1-2%, not 3%. I thought it was too good to be true….

    QE didn’t unanchor expectations, at least not yet, so would a different policy change do the same? It’s probably moot anyway, since Bernanke isn’t going to change policy anyway.

Comments are closed.