Response to Ambrosini

Ambrosini generally has excellent macro posts, so I thought I would respond to his criticism of those (like me) who seemed to take DeLong’s side regarding the desirability of a 3% inflation target.

In earlier posts I have argued for removing the concept of inflation from macroeconomics, at least from the cyclical side of macro (I understand that you need inflation estimates in fields like long term economic development.)  And I have advocated replacing inflation with NGDP growth, which I argue helps clarify what we are debating when we debate demand management.

I strongly believe that, certeris paribus, lower inflation is better–at least until NGDP growth falls close to zero.  But the problem is that right now other things are not equal.  I also believe the current severe recession is caused by the fact that NGDP recently plunged 8% below its trend line.  Unfortunately, new Keynesian economists don’t talk about boosting NGDP; instead they talk about boosting inflation, which is a side effect of AD shifting to the right.  Even worse, it is also a side effect of adverse supply shocks.  So what does it mean to aim for a higher inflation target?  If you are trying to say you want more AD, why not just say so?  Or just say you want more NGDP.

So that’s my first complaint.  My second is that the misplaced focus on policy lags has led most economists to focus on long term inflation expectations.  In my view what is really important is that the inflation expectations over the next two years are far below 2%, which is roughly the inflation rate you get under the current policy regime if we were near equilibrium.  Long term the market thinks we will return to roughly 2% inflation, or perhaps a bit more.  But of course long term we will also recover from the recession.  If we had proper nominal targeting, level targeting, we would have never had a severe recession in the first place.

My next problem is that most economists talk in terms of rates of change, not levels.  As Woodford pointed out; in a liquidity trap it is essential to do level targeting, not inflation rate targeting.

So let’s put all these points together.  I prefer NGDP targets.  But let’s say it must be the price level, because that’s how everyone else talks.  And let’s say we want a very speedy recovery.  How do you do it?

Because the price level is currently a bit below trend, the level targeting approach would require a brief period of catch-up, and then return to your normal price level trajectory, which might be plus 2% per year.  So if forced to work with the price level, I’d advocate a target trajectory that is a bit more than plus 2% for the next three years or so, and 2% thereafter.  I don’t think inflation even needs to be as high as 3%, as the SRAS is not pretty flat, but 3% is fine for a few years of catch-up.  There are also some supply-side problems like extended unemployment benefits, but the reason to shoot for a demand-side solution is that AD is the main problem, and more importantly the lack of AD is precisely what caused many of those supply-side problems that people at Chicago worry about.  I’ve never seen a shred of evidence that my fellow Chicago economists understand that political reality.  Is there a “sell by date” on my stimulus proposal?  Sure, but I really don’t know exactly when that is.  My hunch is that when we are far enough into recovery where unemployment benefits have fallen back to 26 weeks we should definitely have ended all stimulus, and indeed given how slow Congress is, we should probably end it even sooner than that.

I believe Ambrosini is a grad student.  In any case, he raises a lot of good points:

First, does the Fed have an inflation target and are expectations anchored, i.e. are they equal to the target? Yep and yep, see the graph Prof. DeLong put up.

The problem here is that in a world where wages are sticky, inflation targeting is a bad policy.  You need level targeting.   Suppose you have a period of deflation, or even falling NGDP.  Nominal wages would need to fall sharply to maintain full employment.  But for whatever reason they don’t.  We had a pay freeze at Bentley this year, which was regarded a draconian pay cut by the faculty.  But it was only 4% below our normal raise.  Given that NGDP just fell 8% below trend, we and everyone else in the economy would need a pay increase 8% below trend to maintain full employment.  That ain’t going to happen.  Workers in many government and service and high tech jobs actually got raises this year.  Workers don’t lose their jobs because their wages are sticky, they loses their jobs because other people’s wages are sticky.

Suppose you have a 2% inflation target.  Then suppose you miss your target, prices suddenly fall 10% and wages don’t.  Under inflation targeting you continue to aim for 2% inflation.  Let bygones be bygones.   This causes a long period of high unemployment for reasons that aren’t fully understood.  I believe that adverse demand shocks trigger supply-side problems, such as the unemployment benefits issue I mentioned earlier.  Or maybe there is some combination of nominal and real wage stickiness.  Or maybe workers wrongly think nominal pay cuts are an insult (note my blog title.)   In any case, the optimal solution is to try to return to your old price level trajectory.  If that was plus 2% per year, and you’ve just had 10% deflation, then you should shoot for high inflation short term, and 2% thereafter.

Then Ambrosini asks:

Second, what would the impact of a change in the target be? Suppose expectations remain anchored, increasing the target will increase expected inflation, shift the Phillips curve up and we would be right back where we started. More precisely, there would be no change in output and assuming the mechanism behind Okun’s law is invariant to this change in policy, there would be no change in unemployment.

This is why I don’t like sticky-price models; I wish Keynesians would use sticky-wage models.  What Ambrosini says here is exactly right if price stickiness is the problem, and if prices are flexible over the time period of the expectations being discussed. But it is not true if wages are sticky in the near term, and you are trying to raise near-term inflation expectations.  I am not trying to criticize Ambrosini here, as DeLong may have been thinking in terms of a standard Keynesian sticky-price model.  And he was talking about long term inflation expectations.  So Ambrosini’s criticism is entirely appropriate for that policy proposal.  But even in that case raising long term inflation expectations would almost certainly have at least some positive effect on short term inflation expectations, and to the extent that wages and prices are sticky, would reduce unemployment.  Remember, wages don’t simply adjust one for one with any price change; they adjust toward their equilibrium value.  So if you’ve just had deflation, and real wages are too high, and then you suddenly have inflation, then wages won’t rise one for one with the catch up inflation.  Rather they will tend to stay put, and passively move back to their equilibrium real wage.  Unfortunately, textbooks often confuse two issues; changes in expectations that occur before wage contracts are signed, and changes in expectations that occur after wage contracts are signed.  Only the former are neutral.  And even those aren’t neutral if you have money illusion regarding nominal pay cuts.  Unfortunately, all of this is obscured by Keynesian Phillips curve models, which don’t even come close to describing the effects of the big interwar monetary shocks, and which are the only monetary shocks in all American history that we can clearly identify.  (But sticky-wage models fit those shocks almost perfectly.)

Here’s Ambrosini’s 3rd point:

Third, why might increasing the target unanchor expectations? The Fed could be seen as acting with discretion; that they’ll up the target anytime the public or politicians start whining. Here’s a post explaining why discretion is bad. Its really hard for the Fed to get expectations reanchored.

This comment reminds me of how frustrating it is that we really don’t know what the Fed is trying to do.  Suppose Ambrosini is right at the Fed is simply trying to target long term inflation expectations.  In that case, there is no problem with monetary policy.  Janet Yellen once said “we should want to do more.”  But if this view is correct, the Fed shouldn’t want to do more.  The 8% decline in NGDP relative to the mid-2008 trend line?  No problem, we aren’t targeting NGDP we are targeting 10 year inflation forecasts.  But does anyone really believe that?  Does anyone think the Fed is just as happy with NGDP falling 8% below trend as they would have been if NGDP had kept growth at something like 3% to 5% over the past 18 months?  In the real world central banks are flexible inflation targeters.  They look at both inflation and real growth.  The Fed cut rates sharply in late 2007 and 2008, and obviously it wasn’t because they were worried that long term inflation expectations were getting too low.

Would Bernanke dare go to Congress and say “yes, a bit higher inflation could dramatically boost real GDP, but we only care about 10 year inflation expectations so it is up to you guys to use fiscal policy if for some reason are upset about 10% unemployment.”  Obviously not, and I am not trying to put words into Ambrosini’s mouth here.  What I am trying to suggest is that the language we use to discuss monetary policy, especially the language of inflation targeting, merely obscures the issues and we end up talking past each other.  Of course none of this is Ambrosini’s fault.  I did seem to endorse DeLong’s exact proposal, whereas I actually favor something that is close, but differs in small but significant ways.  I basically favor targeting some sort of nominal trajectory that should have been the Fed’s target all along, starting from mid-2008.

One final plea to my fellow economists.  If we could discuss alternative NGDP target paths, level targeting, then it would be much clearer what we are actually debating.

PS.  I just noticed that Ambrosini has another post directed at me.  (And he used to like my stuff!)  I should clarify that I wouldn’t object to the BOJ’s stable CPI policy of 2002-08, if they stuck with it in recessions.  But they don’t.  Instead the Japanese fiscal authorities take over with massive stimulus spending that causes long term fiscal problems, and covers the beautiful Japanese countryside with concrete.

PPS.  A commenter noted that my previous post was long-winded.  (What else is new?)  I realize that many of my recent posts are getting repetitive.   But I feel like the previous post was the first in many weeks where I got back a bit of the creative groove I had going earlier in the year.  If I had time to write it, you have time to read it.


My recent posts remind me of the Truman Capote line:

That’s not writing, that’s typing.



13 Responses to “Response to Ambrosini”

  1. Gravatar of bill woolsey bill woolsey
    19. December 2009 at 12:31

    Sticky price models are fine if you are talking about nominal expenditure. (At least I think so.)

  2. Gravatar of Anonymous_ Anonymous_
    19. December 2009 at 12:39

    > “I believe Ambrosini is a grad student.”

    Why should that matter? 🙂

  3. Gravatar of Scott Sumner Scott Sumner
    19. December 2009 at 12:42

    Bill, Yes, that actually occurred to me as I was writing the post. But I still think it is easier to grasp the point I was making about expectations using a sticky wage model. Then you can talk about new price expectations forming after wages are negotiated, and hence a rise in inflation expectations doesn’t immediately shift the SRAS curve. But yes, in a qualitative sense I could have made the same argument in a sticky price model.

  4. Gravatar of bill woolsey bill woolsey
    19. December 2009 at 12:46

    It is market clearing assumption that leads to all of the confusion.

  5. Gravatar of Don the libertarian Democrat Don the libertarian Democrat
    19. December 2009 at 14:00

    The point is to alter the expectations of real human beings. The Models and Methods are simply ways to try and do that. Tautologies can be useful, but only as starting points. The main problem with using Inflation seems to be that it’s not a great weapon for accomplishing what we want. It’s serviceable, but we could use something better. I like Scott’s views on this, but I’m open to others.

    Right now, we need to use the vocabulary that is in play. As Burke said:

    “Before men can transact any affair, they must have a common language to speak otherwise all is cross-purpose and confusion.”

    This also relates to libertarianism and pragmatism. There is no conceivable situation in the US in which the government won’t act as a lender of last resort in a financial crisis. Bagehot made a similar point about the B of E. You can either call for its abolition, which won’t happen, or try and influence the actual policy that it follows.

    A libertarian is someone who has a bias towards less government. If it’s some kind of creed, I’d rather be called a Burkean Whig, however anachronistic that sounds.

  6. Gravatar of jean jean
    19. December 2009 at 15:29

    I think that people refuse pay cuts because:
    _they don’t manage to see by how much prices have been cut.
    _accepting wage cuts would be a bad signal towards the employer (a variant of efficiency wage theory). Because if everybody is illusioned by money, you have interest to act partly as if you were illusioned too.
    _they may feel unfair that they have a pay cut but no mortgage payment cut.
    _they have the law on their side. Why not taking advantage of it?

    In the seventies in France, wages had been tied to inflation. The french government made this kind of contracts void to pace inflation. So I am not sure that wage indexing would be a good idea. Because it would protect from nominal schocks but not from real (=supply) schocks. (A solution would be to tie wages to NGDP slightly undershooted).

  7. Gravatar of ssumner ssumner
    19. December 2009 at 17:07

    Bill, Again I think you can interpret those models in two ways:

    1. Market clearing
    2. Market clearing if actual inflation equals expected.

    Is that right? Either way I have a problem with these models.

    Don, Good points. I read something that made me think more highly of Burke than I had before. He was more of a reformer than many people realize, he just wasn’t a revolutionary.

    jean, Those are good points. You said:

    “I think that people refuse pay cuts because:
    _they don’t manage to see by how much prices have been cut.”

    Even worse, there is a “who goes first” problem.” If other workers haven’t cut their wages, then prices will have fallen much less than NGDP. But in that case why should I cut my wage?

  8. Gravatar of ssumner ssumner
    19. December 2009 at 18:24

    Anonymous. I have no idea why I said that, I guess I was implying that it was a pretty impressive blog for a grad student. I hope people take it that way.

  9. Gravatar of OGT OGT
    20. December 2009 at 09:17

    I just ran into an interest article on the sticky wage and debt problem by Charles Goodhart, London Banking and Finance professor, suggesting a lack of trust and asymetry of information is key in both:

    Economists often have a hard time understanding the reason why both wages and many financial contracts – such as bank loans and bank deposits – are fixed for certain periods of time in nominal terms. Would it not be better, it is often suggested, if all such financial assets could adjust in response to changes in the (market) values of the underlying assets, and if nominal wages could adjust flexibly in relation to underlying changes in demand and profitability. Surely this would eliminate the possibility of financial crises and serious unemployment. Indeed, this would be a better way of operating if we had complete and efficient (financial) markets, and everybody behaved with absolute honesty, so that everyone could have complete trust and confidence in everyone else.

    Unfortunately these happy conditions do not pertain. Instead the borrower and the employer have far better information on the state of their business than the lender or worker. Under these circumstances, if the lender or the worker relied on the self-certified account of their business by the borrower/employer, what the lender/worker would invariably be told was that business was bad, and that the lender/worker would have to suffer a reduction in their payment as a result. In order to keep the borrower/employer honest, the most obvious and sensible procedure is to negotiate a nominal fixed price contract which can be revised after a period, which involves a penalty on the borrower/employer if the contract is not honoured. The penalty for the borrower is bankruptcy, and the penalty for the employer is that, if he cannot pay the nominal wage, he has to fire workers, and therefore not produce as much as before.

  10. Gravatar of Doc Merlin Doc Merlin
    20. December 2009 at 10:31

    I agree OGT about the lack of trust. However, lack of trust can’t be fixed by monetary policy (although it can be made worse).

  11. Gravatar of ssumner ssumner
    20. December 2009 at 15:54

    OGT, I don’t follow that argument at all. I thought the argument was that wages should be fixed to some measure of AD, not the conditions of each firm. And that information is publically available from the government.

    On the other hand I agree with Doc that a stable monetary policy is the best solution.

  12. Gravatar of Philo Philo
    21. December 2009 at 08:26

    Don the Libertarian Democrat writes: “You can either call for its [the B of E’s, and, by analogy, the Fed’s] abolition, which won’t happen, or try and influence the actual policy that it follows.” Don evidently thinks this is an exclusive disjunction: you can do one or the other, but not both. But why isn’t it an inclusive disjunction? There’s no contradiction, or even “tension,” between saying (a) the Fed should be abolished (or, more modestly, stripped of its lender-of-last-resort function to “too-big-to-fail” institutions) and (b) the Fed, not having been abolished, should for now pursue an easier-money policy.

    Is the worry that if you say (a) most people will think you’re crazy, so they won’t listen when you say (b)? I doubt the connection (if, indeed, that is what is being alleged); besides, most people aren’t listening to proponents of (b)–to me, to Don, even to Scott Sumner–*whatever else* they are saying. Don may “try and influence the actual policy,” but that “won’t happen,” either. He gains nothing by imposing silence about (a) on himself.

  13. Gravatar of ssumner ssumner
    21. December 2009 at 13:04

    Philo, I think that is a fair point. You can do both. As a pragmatic libertarian I often say things like “Ideally we get rid of FDIC and TBTF, but that won’t happen, so we have a few common sense regulations like minimum down payments, but don’t try to micro manage banks from Washington.”

Leave a Reply