2 questions for IT fans

This post is aimed at economists who know more about formal monetary models than I do.  I am interested in why so many economists seem to favor inflation targeting, whereas price level targeting seems preferable to me.  I am especially interested in whether the models used to compare these two policy regimes incorporate pertinent facts about the real world.

Models often contain “shocks”, which temporarily throw the economy off course.  One question I have is:

1.  Does the size of shocks in these models depend on the type of monetary regime?

Let me explain with an example.  We are back in June 2008, about to be hit by a severe banking crisis.  In my view the severity of the banking crisis depends to a great extent on whether the central bank is doing growth rate or level targeting.  Under level targeting, investors will expect the economy to bounce back strongly after any severe crisis, and hence asset prices will not fall anywhere near as sharply in the short run.  And with asset prices being more stable, the financial crisis (i.e. the “shock”) will be much milder.

Under level targeting of prices, investors would have expected the price level (PCE) to be about 22% higher in the 10 years after June 2008.  We are now almost 8 years past June 2008, and the price level has rising by 8%.  It looks like it will be about 11% or 12% higher after 10 years, not the 22% expected under level targeting.  Thus under level targeting, investors would have expected a far more expansionary monetary policy over the past 8 years.  That would have made the 2008 financial crisis much milder.  For a recent example of the role of expectations, look at how Brazilian asset prices have been recovering under expectations of impeachment.  Brazil is a mess, but investors are already looking beyond the current inept government.

2.  Is the governing board of the central bank (in these models) split between hawks and doves under inflation targeting, under price level growth rate targeting, or under both?

There is only one correct answer.  Under IT, central banks are split between hawks and doves, and under price level targeting they are not.  Indeed under level targeting the terms “hawk” and “dove” have no meaning.  The only difference is the aggressiveness in which they want to return to the trend line.

Because there is no hawk/dove split under price level targeting, markets have less uncertainty about where the price level will be in the future.  In contrast, under IT you may have some people viewing 2% inflation as a target, and others viewing it as a ceiling.  Or they may differ in terms of whether they’d rather risk erring on the side of too much inflation, or too little.  In contrast, under price level targeting the long run inflation rate is essentially identical regardless of whether the central bank views the target as symmetrical or as a ceiling.

To summarize, I’m asking those who know the literature on IT whether existing models take relevant real world considerations into account.  If not, perhaps someone should create more realistic models.



19 Responses to “2 questions for IT fans”

  1. Gravatar of David R. Henderson David R. Henderson
    27. April 2016 at 06:52

    I’m confused. You started with an issue of inflation targeting and price level targeting, and then in your question 2, didn’t ask about that, but introduced growth targeting in place of price targeting.

  2. Gravatar of John Hall John Hall
    27. April 2016 at 08:09

    Asymmetric policy response to inflation targets is so damn important. If only we could convince central bankers that a 2% target isn’t the same thing as a 2% ceiling…

  3. Gravatar of Benoit Essiambre Benoit Essiambre
    27. April 2016 at 08:46

    It’s a poor argument IMO but If I recall correctly, the Bank of Canada papers on the subject concluded PLT was not worth it given the risks from having to tighten too much after a price overshoot.

  4. Gravatar of ssumner ssumner
    27. April 2016 at 10:16

    David, Yikes, a senior moment. I fixed it.

    Benoit, They may be right, which is an argument for NGDP level targeting.

  5. Gravatar of Doug M Doug M
    27. April 2016 at 10:45

    Regarding the 2% inflation target. Most of the Fed watchers I know don’t believe that the Fed really has a 2% target. Most think that the Fed has an upper bound of 2%, and that the Fed is actually targeting something at least a half point below that.

    The TIPS market is certainly priced for a 1.5% level target.

  6. Gravatar of Anthony McNease Anthony McNease
    27. April 2016 at 13:06

    My sense is that IT has reached a point similar to religious dogma. Milton came around and convinced everyone that IT was better, and it was. It worked very well for decades. Price level targeting and NGDP targeting come around and from what I can gather reading the experts the response reminds me of the iconic scene in Spinal Tap:

    “I’ve got the best amps in the world. They go all the way to 11.”

    “Why don’t you just get bigger amps?”

    “But…..these go to 11.”

  7. Gravatar of LK Beland LK Beland
    27. April 2016 at 13:18

    Benoit Essiambre

    “the Bank of Canada papers on the subject concluded PLT was not worth it given the risks from having to tighten too much after a price overshoot.”

    I think that this links to the following part of Scott Sumner’s post: “Indeed under level targeting the terms “hawk” and “dove” have no meaning. The only difference is the aggressiveness in which they want to return to the trend line.”

    For the reason highlighted by the BoC, the central bank may want to correct overshooting/undershooting over a couple of years.

  8. Gravatar of Benjamin Cole Benjamin Cole
    27. April 2016 at 15:08

    Well I prefer NGDPLT. And shoot high. And send in the choppers, loaded for bear. Why ding-dong around?

  9. Gravatar of Benjamin Cole Benjamin Cole
    27. April 2016 at 15:50

    OT but I can’t stop myself: “Ted Cruz just named Carly Fiorina as his VP pick in a desperate campaign shake-up bid.”–Vox

    This strikes me as a declaration of defeat. The wheels coming off. Waterloo. Thomas Eagleton. Cleveland Browns. SpongeBob Squarepants’ Weenie Bar. Barney Fife.

    Although, according to reports, Fiorina displayed her singing skills in accepting the veep slot.

  10. Gravatar of ssumner ssumner
    27. April 2016 at 16:33

    Doug, I actually do believe they have a 2% target; they are imply a bit clumsy, or more than a bit.

  11. Gravatar of George Selgin George Selgin
    27. April 2016 at 17:43

    Scott, in my experience most models that support IT as an optimal policy either (1) rely on loss functions into which the (absolute value of the) inflation rate itself, or its variability, enters directly; or (2) abstract from supply (and especially productivity) innovations, and so treat only the cases in which IT and NGDPT are equivalent.

  12. Gravatar of H_WASSHOI (Maekawa Miku-nyan lover) H_WASSHOI (Maekawa Miku-nyan lover)
    27. April 2016 at 17:50

    When I saw the definition of price index, I immediately realized those(IT models) have big problems.

  13. Gravatar of Beniamin Cole Beniamin Cole
    27. April 2016 at 18:16

    Fiorina=Dan Quayle=Sarah Palin=Geraldine Ferraro=Thomas Eagleton=Fat Lady Singing?

    She weighs 349 pounds and has a voice like a fog horn….

  14. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    28. April 2016 at 04:56

    That December Fed decision to double the interest rate it pays banks to withhold money from circulation, reaps what it sowed;


    ‘Gross domestic product, a broad measure of economic output, advanced at a 0.5% seasonally adjusted annualized rate in the first quarter, the Commerce Department said Thursday. That was the worst performance in two years.’


    Nonresidential fixed investment, a measure of business spending, fell 5.9%, the biggest drop since the waning days of the recession. Spending on structures and equipment both sank. The energy industry has been especially constrained amid low commodity prices and investment in mining shafts and wells was a major drag in the first quarter.

    Trade and inventories also subtracted from growth at the start of the year.

    The decline in net exports reflects a strong dollar and spotty demand from overseas.

  15. Gravatar of flow5 flow5
    28. April 2016 at 08:19

    How stupid. No need for the distinction. Roc’s in M*Vt = roc’s in inflation. No need for any adjustments.

    Bankrupt U Bernanke should never have opened up reciprocal currency swap agreements prior to 2008. There were already too many dollars in foreign hands driving down the dollar’s exchange rate. Then in 2008 we had a reversal – “flight-to-safety”.

    – Michel de Nostredame

  16. Gravatar of ssumner ssumner
    28. April 2016 at 10:24

    George, I agree, but as I indicated here I think the problems go much deeper.

  17. Gravatar of jonathan jonathan
    29. April 2016 at 04:01

    I’m not familiar with direct comparisons of IT and ILT, but I would be very surprised if they included (1) and (2).

    (1) is really about endogenous amplification processes under policy regimes. There’s plenty of work on that in (e.g.) the literature on financial crises, and some recent work trying to incorporate these ideas into larger-scale models; but I don’t think has proceeded to the point of doing the comparison you ask about.

    I highly doubt anyone has tried to model (2).

    I think skepticism about level targeting is about the credibility of promised future high inflation. Given how allergic to inflation central bankers seem to be, would we really believe that they would purposely produce a period of high inflation after a recession, at great cost to the economy, simply to live up to their level target?

    For instance, suppose the central bank is facing a 5% price level shortfall, due to several years of below-target inflation. Is it really plausible that the central bank will tolerate 5 years of 3% inflation (say), for no other reason than to achieve its level target?

    (This is analogous to a stated policy by the treasury not to bailout systemically important banks. This is sufficiently inefficient ex-post, that the ex-ante promise is not credible.)

  18. Gravatar of jonathan jonathan
    29. April 2016 at 04:06

    p.s.: completely side note, but has anyone suggested buying and selling treasuries and TIPS at various maturities to target 2% expected inflation at every time horizon? Of course, there are liquidity issues in these markets so that the breakeven price may not be an unbiased forecast of the mean inflation rate, but it does seem like a nice market monetarist way to implement IT.

    More broadly, I’ve always thought of your (Scott’s) views as logically breaking into three points:

    (1) We should use market expectations to determine whether we are hitting our targets, and adjust until we hit those targets; preferably we should tie our instruments directly to the markets that we are targeting (e.g. create a NGDP futures market, and then buy and sell futures).

    (2) We should target NGDP rather than prices.

    (3) We should target levels rather than rates of change.

    But there’s no reason that (1) is contingent on (2) and (3). You could have a market monetarist inflation targeting central bank — it just needs an actual market forecast of inflation to target, and then commit to doing what it takes to hit that target, preferably by buying and selling inflation futures directly.

  19. Gravatar of ssumner ssumner
    29. April 2016 at 16:16

    Jonathan, Just to be clear, I think a NGDP level target would be better. But yes, if the central banks decided to adopt price level targeting, I believe they would implement it, because it would lead to better outcomes than growth rate targeting. People would no longer think in terms of inflation rate, they’d think in terms of price levels, are we above or below the target–that’s all that would matter. There’d be no reason for newspapers to even report inflation, it would be like doing a weather forecast in some place where the weather never changes.

    All that would matter was if prices were above or below the target line.

    On your second point, I agree that those three policies are separable.

Leave a Reply