Inching toward NGDP targeting

An increasing number of important people recognize that NGDP level targeting is superior to inflation targeting. Nonetheless, the Fed is unlikely to suddenly switch to NGDP targeting. Rather, the change will occur gradually over time, in stages.

Let’s think about a policy of price level targeting. In what situations would that move us closer to NGDP level targeting, and when would it be inappropriate?

1. Suppose a supply shock suddenly hits the economy, such as a surge in imported oil prices caused by a shutoff of oil from an important exporter. This shock might push inflation higher and output lower. The effect on NGDP would be small.

If the Fed were targeting the price level, and the oil shock pushed inflation up to 4%, then they would be required to aim for 0% inflation next year, or 1% inflation over the next two years.

But if the oil shock has little or no impact on NGDP, that reversion to the price level trend line would actually be inappropriate. So price level targeting does not do well when there is a supply shock to the economy.

2. Suppose a demand shock suddenly pushes inflation much higher or much lower. In that case, NGDP will also rise or fall relative to trend, indeed by even more than inflation. Now a reversion back to the trend line would be appropriate. If the economy temporarily overheats, then profits will initially rise sharply. Wages are sticky and don’t move much in the short run. Bringing prices down will help to push profits back down to a more normal level, and insure macroeconomic stability.

An even more important example occurs when there is a sudden fall in inflation due to a negative demand shock. In this case, it helps to run inflation at levels above normal in a catch-up period, as this will also help to restore NGDP to its trend line and shorten the recession. A negative demand shock is perhaps the clearest case where price level targeting helps to stabilize the economy by mimicking the effects of NGDP level targeting.

When Ben Bernanke recently proposed a modified form of price level targeting, he suggested that whenever the economy falls to the zero bound during a recession, the Fed should commit to price level targeting until the previous trend line is restored. Note that recessions that push interest rates to zero are almost always caused by negative demand shocks.

Why is this important? Because Bernanke’s proposal is the single most likely outcome of the Fed’s June conference to re-evaluate its policy strategy. It is also the form of price level targeting that most closely approximates the best aspects of NGDP targeting. It is a policy that implicitly says we should not always do price level targeting, rather do it when it would be most likely to produce NGDP level targeting-like results.

The market part of market monetarism will also happen in stages. The Fed is already paying more and more attention to market indicators and less to Phillips Curve-based models of inflation (which haven’t worked.) Next step is to change the fed funds target daily, in increments of 1/100%, with up and down moves equally likely. That’s how markets behave. An NGDP futures market guardrail is still a few decades away.

In their heart, Fed officials must know by now that NGDP targeting would have been better than inflation targeting. But we will get there in baby steps, which is probably how it should be. Monetary policy is too important to make a sudden lurch into the unknown.



75 Responses to “Inching toward NGDP targeting”

  1. Gravatar of dtoh dtoh
    2. May 2019 at 16:15

    You’ve said the level of inflation does not matter when setting an NGDP target. If that’s the case there is no need for an NGDP target at all. Just set an RGDPLT target at trend RGDP. It will produce EXACTLY the same result as an NGDPLT target.

  2. Gravatar of ssumner ssumner
    2. May 2019 at 16:43

    dtoh, The Fed can only target nominal variables. They can’t target a real variable without leaving NGDP indeterminate. And unstable NGDP is a huge problem.

  3. Gravatar of John Hall John Hall
    2. May 2019 at 17:27

    ” Because Bernanke’s proposal is the single most likely outcome of the Fed’s June conference to re-evaluate its policy strategy.”

    My understanding was that the average inflation targeting regime was the most likely outcome.

  4. Gravatar of John Hall John Hall
    2. May 2019 at 17:32

    @dtoh Another big problem is who knows what the trend rate of RGDP should be. It’s not stable over long periods of time. This is why my preferred NGDP regime would be one where the Fed sets a trend rate of NGDP and then adjusts the trend rate slightly every year or so based on what long-run inflation expectations are (say a 5yr/5yr forward rate). If long-run inflation expectations are higher (lower) than 2%, then bring down (up) the trend rate of NGDP growth by some amount. This way if the trend rate of RGDP slows down or speeds up over time, you still have long-run inflation expectations anchored.

  5. Gravatar of dtoh dtoh
    2. May 2019 at 19:25

    If you insist that the amount of inflation built into the NGDP target doesn’t matter, then you can build in zero inflation and set the NGDP target to whatever you think is trend NGDP. In which case, there is no difference between an NGDP target and an RGDP target.

    Not saying I agree with this… I do not, but it is a logically inescapable result if you insist that any target rate for NGDP is equally optimal.

  6. Gravatar of dtoh dtoh
    2. May 2019 at 19:30

    @John Hall

    What makes you think 2% is the right rate for long run inflation expectations? How do you know it’s not 3% or 1% or 0.5% or 3.7%?

  7. Gravatar of Rein Rein
    3. May 2019 at 00:27

    For NGDP targeting, you use the useful analogy of steering a ship. What do you think would be a good analogy to explain the ambiguous interpretation of inflation and interest rates singals?

  8. Gravatar of Michael Sandifer Michael Sandifer
    3. May 2019 at 03:04


    The Fed is publicly exploring ways to bring about a faster electronic payment system to the US. If one day settlment, if not instantaneous settlment, becomes the norm, and GDP-related spending is reported to the Fed daily, do you think your guardrails approach is then needed?

  9. Gravatar of John Hall John Hall
    3. May 2019 at 03:33


    From a theoretical perspective, I don’t think there’s anything particularly special about 0%, 1%, 2%, or 3%, when combined with a NGDP target (because if there is a negative aggregate demand shock, then it enforces faster NGDP growth and offsets some of the issue of sticky wages that is a problem for lower inflation targets). Anchoring long-term inflation expectations is more important. So doesn’t matter what you pick, just pick something.

    However, from a practical perspective, the Fed already has 2% as its inflation target, so there would be quite a bit of challenge in transitioning from a 2% target to a 1% or 3% target. It might be something that you could accomplish gradually over a long period of time, but not something you would want to do over just a year or something.

  10. Gravatar of dtoh dtoh
    3. May 2019 at 06:29

    @michael sandifer

    The Fed does need a different payment system to get better GDP spending data. There is a ton of data out there (e.g. private forecasters use satellite imagery to count cars in Walmart parking lots.) The problem is organizational.

  11. Gravatar of dtoh dtoh
    3. May 2019 at 06:42

    @John Hall

    Sticky wages is the ONLY issue. If wages were not sticky, monetary policy would be both unnecessary and ineffective. An NGDP target that incorporates an expectation of zero inflation is not credible because there is no mechanism by which to drive real wage deflation if there is not a positive delta between real and nominal wages.

  12. Gravatar of dtoh dtoh
    3. May 2019 at 06:49

    @john hall

    Also, what is the practical limitation of changing the inflation target (other than mainstream economists having to admit they don’t have a clue.)

    Unless you believe incorrectly (as Scott does) that the relationship between real output and the NGDP target is a step function (i.e. any change in the NGDP target only impacts the price level and has no impact on real output,) then there is in fact an optimal rate of inflation and 2% is purely an arbitrary number and not necessarily optimal.

  13. Gravatar of dtoh dtoh
    3. May 2019 at 06:58

    @michael sandifer

    typo – Should read “The Fed does NOT need a different payment system..

  14. Gravatar of rayward rayward
    3. May 2019 at 07:12

    Question: Does Emi Nakamura’s research and her being awarded the Clark Medal bode well for NGDP targeting? I would think it does.

  15. Gravatar of Peter Mannino Peter Mannino
    3. May 2019 at 07:33

    Arin Dube (of minimum wage fame) likes nominal wage targeting! Is he inching toward market monetarism?

  16. Gravatar of Benjamin Cole Benjamin Cole
    3. May 2019 at 08:03

    I genuflect to NGDPLT.

    But suppose the NGDPLT is 4%. And then suppose we have 4% inflation for 10 years but no real growth.

    Then what?

  17. Gravatar of Michael Sandifer Michael Sandifer
    3. May 2019 at 08:11


    You’re correct that the data is out there already, but for daily adjustments, direct electronic data cannot be beat. If the Fed got electronic payment data once-per-week, they’d be better off than they are now.

  18. Gravatar of dtoh dtoh
    3. May 2019 at 08:21


    Not sure we would be better off. For over a decade, the Fed has had the policy tools and the ammunition, yet still they have been below one or both of their targets the entire time. I’m not sure how more data will help them if they are unwilling to use their policy tools.

    If the Fed were held accountable (as Scott recommends) for meeting their goals, then better data would help them fine tune a little, but right now, it’s not going to move the needle.

  19. Gravatar of Ewan Maclean Ewan Maclean
    3. May 2019 at 09:33

    3. May 2019 at 06:42

    If you could, humour me, a layman, and tell me what it means to say “a positive delta between real and nominal wages”. You mean there is some correlation between changes in each? And what do you mean by “no mechanism by which to drive real wage deflation”. Are inflation and real wage growth not both affected by the cycle? And is it not so whether there is an inflation target or a price level target or whatever? Or is real wage growth entirely driven by secular variables? I’m struggling (can you tell?). And, while we’re at it, what would an optimal level of inflation be? If my questions are gibberish, ignore them.

  20. Gravatar of ssumner ssumner
    3. May 2019 at 09:52

    John, It’s basically the same thing.

    dtoh, You said:

    “In which case, there is no difference between an NGDP target and an RGDP target.”

    That’s wrong and it’s an important error. If you target RGDP at say 3%, and the natural rate of growth in RGDP is actually say 2.8%, then the price level will explode toward infinity. In contrast, if you target NGDP at 3.0%, and the natural rate of RGDP turns out to be 2.8%, you merely get 0.2% inflation.

    Rein, Some compare interest rate targets to a car accelerator. More pressure on the accelerator often makes the car go faster, but on the other hand people often apply more pressure when going slow up a hill, and less when flying down a hill.

    Michael, Probably.

    Thanks Peter.

  21. Gravatar of dtoh dtoh
    3. May 2019 at 10:18


    That’s incorrect because again you’re assuming a step function relationship whereby once we have reached the natural RGDP growth rate, any and all nominal growth will translate entirely into an increase in the price level with no change in real output.

    If that were the case: a) there would be zero benefit in having any inflation once you’re at the natural GDP rate, and b) under current conditions, the Fed should obviously tighten as they could bring inflation down to zero without having any impact whatsoever on real growth.

  22. Gravatar of dtoh dtoh
    3. May 2019 at 10:32


    And BTW, I did not say [that your logic implied] that an RGDPLT was the same as an NGDPLT if you set the the RGDPLT above the natural growth rate. I said it was the same [under your logic] if you set it at [equal to] the natural growth rate. (Presumably though policy makers would lower the target if they found out they had set it too high.)

    Since you’ve responded that it wouldn’t work in the case when the target is set too high, is it safe to assume (since you haven’t rebutted it) that you agree and that your position is in fact that an RGDPLT target set equal to the natural rate would have exactly the same effect as an NGDPLT.

  23. Gravatar of Michael Sandifer Michael Sandifer
    4. May 2019 at 03:38

    The advantage of improving the electronic payment system is that it improves the electronic payment system. It will improve productivity, in addition to making NGDP information available to the Fed daily. My guess is that this is more feasible than the NGDP futures approach, in terms of implementation.

  24. Gravatar of Matthias Görgens Matthias Görgens
    4. May 2019 at 05:59

    @dtoh, show a bit more creativity and be bold! Why stop at zero inflation and not go for ‘Less Than Zero’? (Our host wrote a new foreword to the recent rerelease of that classic book.)

  25. Gravatar of Saturday assorted links – Marginal REVOLUTION Saturday assorted links - Marginal REVOLUTION
    4. May 2019 at 08:45

    […] 4. How to get to ngdp targeting. […]

  26. Gravatar of dtoh dtoh
    4. May 2019 at 09:19

    I pay everything by Fed wire already. What needs to be improved? Or do we just need to get more people to use it?

  27. Gravatar of dtoh dtoh
    4. May 2019 at 09:30


    I don’t think we should have zero inflation. I’m just using it that as an argument to demonstrate some of the logical flaws in Scott’s reasoning.

    IMHO, as long as you have sticky wages/prices, you absolutely need inflation to allow the labor markets to adjust to changed equilibria….. and not just in when you have a demand shock but even at full employment in order to facilitate wage adjustment between different economic sectors and between firms.

    That said, IMHO wage stickiness is a learned behavior and you can wean firms and individuals off of it. The only problem is that to do so would take a long time and involve a lot of pain.

  28. Gravatar of ssumner ssumner
    4. May 2019 at 09:36

    dtoh. First of all, it’s not even worth talking about setting a RGDP target exactly equal to the natural rate, as the probability of that happening is zero. So that’s a nonstarter. And if you are off by even one basis point it’s a disaster.

    Second, while there is an effect of higher nominal growth on RGDP in the short run, there is no effect in the long run. So trying to target RGDP or any real variable in the long run leaves the price level and all other nominal variables indeterminate.


  29. Gravatar of dtoh dtoh
    4. May 2019 at 09:42

    So set it 20 basis above or 50 basis above. And what’s to stop the policy makers from adjusting if they are off.

    Also if the actual NGDP target did not matter, why did you previously recommend a target of 4.8% and now recommend a target of 4.0%?

    Finally I don’t understand why “MONETARY POLICY CANNOT TARGET REAL VARIABLES.” Isn’t simply a matter of buying fewer assets if you are over target and more if you are below target? Or am I missing something?

  30. Gravatar of Regulateflorida Regulateflorida
    4. May 2019 at 11:56

    Bravo, Scott

  31. Gravatar of Michael Sandifer Michael Sandifer
    4. May 2019 at 13:23


    Most electronic payments still go through the ACH or credit networks. ACH, or Automated Clearing House is run by a cartel of major banks. Payments through that network take 1-3 business days to complete.

    Banks could use Fedwire, if they wanted, but they choose to reserve that for larger transactions by charging $25-45-per-transaction, except for preferred account holders. Also, many banks don’t allow for self-service outgoing wire transfers.

  32. Gravatar of dtoh dtoh
    4. May 2019 at 14:19

    So that was kind of my point. It’s not really a problem with the system, it seems more like an anti-trust problem to me.

  33. Gravatar of Carl Carl
    4. May 2019 at 16:01

    I think Scott already answered your question: “there is no effect on rgdp in the long run.”

    If you set a target above the natural rate of rgdp, no amount of monetary stimulus will get you to your goal. You will simply be creating inflation.

  34. Gravatar of MJ MJ
    4. May 2019 at 16:18

    If the Fed is aiming at the best system long-term, shouldn’t it working toward targeting total nominal labor compensation rather than NGDP ?

  35. Gravatar of dtoh dtoh
    4. May 2019 at 19:33

    Yes. Scott did answer my question. The reason for my follow up questions is so that Scott will see why his answer is wrong.

  36. Gravatar of Michael Sandifer Michael Sandifer
    5. May 2019 at 05:22


    Yes, it is an antitrust matter, but it seems a perfect example of government regulations helping create and maintain a cartel. That’s the broad view and deregulation is at least a large part of the ultimate solution.

    But, the Fed can perhaps address the payments situation narrowly.

  37. Gravatar of dtoh dtoh
    5. May 2019 at 05:37


    Totally agree.

  38. Gravatar of Michael Sandifer Michael Sandifer
    5. May 2019 at 05:37


    I wouldn’t be concerned about inflation exploding toward infinity, just because the long-run aggregate supply curve is vertical. The problem is that trying to target RGDP would lead to considerably more instability.

    For one thing, RGDP potential fluctuates. It’s a moving Target and merely getting it right on average isn’t good enough. It’s also best estimated ex-post, and even then there are disagreements.

    Getting close isn’t good enough, because undershooting will be at least as bad as undershooting an NGDP target, and perhaps worse, with less room for error.

    But then, overshooting the real target is much worse, because the short-run aggregate supply curve will allow for unsustainable hiring, for example, until wages adjust and then you get a recession, as output returns to long-run potential.

    So, it’s a really bad idea, unless you hate the current occupant of the White House.

  39. Gravatar of dtoh dtoh
    5. May 2019 at 07:11


    I completely agree. Targeting RGDP would be a really bad idea, but… if you believe (as Scott purports to do) that an NGDP target that generates 0.1% inflation is equally optimal to an NGDP target that generates 2.0%, then it’s illogical to believe that an NGDP target is superior to a RGDP target.

    And.. the costs that you cite of undershooting or overshooting the target are true, but only because because different levels of inflation are not equally optimal.

    It’s a mistake to believe that inflation is only beneficial when there is a generalized shock (or when aggregate unemployment is above the natural rate.) The economy is not monolithic. Inflation facilitates adjustment of real wages not just for the economy in aggregate but also between sectors, geographies and between individual firms. This increases overall efficiency and raises growth.

  40. Gravatar of ssumner ssumner
    5. May 2019 at 08:07

    dtoh, Adjusting the target means you have no target. Or you implicitly have a NGDP target. The whole point of a target is to not adjust it.

    You said:

    “Also if the actual NGDP target did not matter, why did you previously recommend a target of 4.8% and now recommend a target of 4.0%?”

    How many times do I need to answer this question? A thousand? I picked 4% for political reasons, so that Fed officials will think it consistent with their 2% inflation goal. A figure like 4.8% would also be fine with me.

    The problem with targeting real variables is that the LRAS is vertical.

  41. Gravatar of dtoh dtoh
    5. May 2019 at 08:18

    So just to clarify you think a 2% inflation goal is no better or worse than 1% inflation goal or 0.5% inflation goal.

    Also I’ll argue (at the appropriate time) that LRAS is not vertical and even if it was vertical the NPV of the utility from output varies because the SRAS is definitely not vertical. Also, we may be way underestimating what potential output might be in a stable interest rate and NGDP environment.

    Again, I’m not arguing in favor or an RGDP target. I’m just arguing that all NGDP target levels are not equal.

  42. Gravatar of Michael Sandifer Michael Sandifer
    5. May 2019 at 08:39


    I suspect that, as a result of the anchoring of expectations since hitting the ZLB, there could be another leg up or two for the economy, even now.

    Conventional theory tells us we’re now near full employment for reasons Scott has pointed out well. Headline unemployment has been flat for several months, inflation started to pick up a bit at times, and RGDP is running above levels believed sustainable.

    But, I think common expectations about RGDP potential, and/or the Fed’s expected unwillingness to give potential GDP a full runway may result in behavior at various levels of RGDP/NGDP growth that mimic reaching potential RDGP/full employment.

    If I’m right, there’s a test of faith for the Fed. They have to run higher inflation for a while, or at least demonstrate more symmetry in their targeting, to have any hope of getting back to potential output soon.

    Scott is correct to point out that the downside of me being wrong could be a recession under inflation targeting, and I agree. That’s why I favor a 5% NGDP target, we can get an immediate boost from a bit steeper future nominal growth path, but keep nominal spending from going to high for too long.

    It’ll be interesting to see if this pseudo-multiple equilibria hypothesis is correct, if we can get a good test of it. Scott’s conventional view is more parsimonious, and probably more likely correct.

    I still think though that sustainable productivity growth will be higher than most expect, as is often the case many years after a depression.

    In fact, I’ll say our current depression has lasted since about 2000-2001, in that we never fully recovered from the so-called tech crash and post 9-11 slowdown.

  43. Gravatar of dtoh dtoh
    5. May 2019 at 12:22


    Think you are right. Tyler Cowen’s secular stagnation is a crock.


    1. U6 underestimates unemployment. You need to look at LFPR.
    2. A more stable monetary policy environment will reduce risk and engender growth.
    3. The effect of the tax cut is underestimated for a) political reasons, and b) because people don’t understand the impact of asymmetric investment returns on the effective tax rate.
    4. Technological progress is accelerating and there’s a ton of existing technological gains that have yet been implemented.

  44. Gravatar of Michael Sandifer Michael Sandifer
    5. May 2019 at 19:54


    I basically agree with all that, except the effects of the tax cuts have been offset somewhat by the trade skirmishes. There seems to be a strong case that the tax cuts caused a significant rightward shift in the supply curve

  45. Gravatar of Scott Sumner Scott Sumner
    6. May 2019 at 11:16

    dtoh, I agree that not all NGDP targets are exactly equal. My point is that any reasonable target is fine with me (and better than no target), and the exact number has little impact on output. Now of course extreme cases are different, in either direction. No one knows the optimal number, and it will be lower in a regime that taxes capital income.

    Someone who thinks the taxation of capital income is a big problem should favor a lower NGDP target.

  46. Gravatar of dtoh dtoh
    6. May 2019 at 15:46

    So it seems to me, you should favor a higher NGDP target for the following reasons.

    1. The inflation / real growth trade off is less steep than you think because inflation facilitates adjustment (= more efficiency = higher output) not just when there is an aggregate demand shortfall but also at trend growth when adjustment is required at the micro-level of a economic sectors or regions or even at the level of individual firms.

    2. A little less growth is a lot worse than a little more inflation. And the people who are most hurt by low growth tend to be the less fortunate (those with a higher marginal utility of income.)

    3. There’s a fair amount of uncertainty right now about the natural real growth rate…. why run the risk of choking off potential incremental output.

    4. You run the risk of being perceived as favoring low growth because of your feelings about the current President. And BTW… you should lay off the anti-Trump screeds. You ruin your chance of being appointed to the Fed. Regardless of your personal views and the rather low probability of it happening, it would be good for aggregate utility if you were appointed to the Fed. So stop being selfish and at least give yourself a chance.

    5. Remind me again of the inflation / capital income argument. Every time I read your argument, I always agree with it, but then I forget what it was.

    7. May 2019 at 04:57

    The increase in the GDP growth rate due to the tax reform is supposed to be temporary, as it should boost real GDP for a few quarters or so, after which GDP growth should return to baseline, with the permanent advantage being baseline growth from a larger GDP base.

    So, this is the tell. If the higher average real GDP growth rates continue, requiring enough runway from the Fed of course, then likely the permanent increase in productivity is at least partially cyclical. That said, continued deregulation via executive action could also be a real factor.

  48. Gravatar of Michael Sandifer Michael Sandifer
    7. May 2019 at 05:05

    This also requires, of course, reaching a deal with China to avoid permanent tariffs.

  49. Gravatar of dtoh dtoh
    7. May 2019 at 06:49

    Why do you think the effect of the tax cut would be temporary?

  50. Gravatar of Michael Sandifer Michael Sandifer
    7. May 2019 at 08:53


    The rightward supply-side shift occurs only once, but the positive effect continues because there’s a higher GDP base from which the reversion to trend growth rate proceeds.

    It’s like increasing the principal during one year for an investment, rather than the permanent rate of return. The addition to principal only adds to the portfolio once, but the portfolio then grows at the same rate on a higher base.

  51. Gravatar of Michael Sandifer Michael Sandifer
    7. May 2019 at 09:03

    Another way of saying it is, a positive supply shock increases the GDP stock, which appears as a flow temporarily, but then subsequently trend growth resumes from a larger stock.

  52. Gravatar of Michael Sandifer Michael Sandifer
    7. May 2019 at 09:16

    By the way, the stock market dip this week indicates that the threat of further Chinese tariffs has subtracted almost .1% from GDP growth expectations, ceteris paribus.

    That is, inverse P/E for S&P 500(~4.5%) times the change in the S&P index(~2.2%). Inverse P/E is a proxy for the discount rate.

  53. Gravatar of dtoh dtoh
    7. May 2019 at 09:31

    I think what you are missing is that the lower tax rates (essentially on capital) result in increased investment, an increased capital base, higher productivity, higher output, etc. And to the extent the tax cuts are not temporary, the higher investment persists leading to a higher permanent rate of growth.

  54. Gravatar of dtoh dtoh
    7. May 2019 at 09:59

    To be clear, I should have said that a lower tax rate on capital leads not to just increased investment, but also a higher rate of investment (i.e. investment increases relative to consumption) which leads to higher growth.

  55. Gravatar of Michael Sandifer Michael Sandifer
    7. May 2019 at 12:14


    I think you’re double-counting the positive effects of positive supply shocks. There is the initial boost as the economy adjusts, leading to positive developments like increasd capital investment. After that though, there are no more shocks to growth coming from that one reform.

    But, because the GDP potential increased on a one time basis, the baseline growth rate represents a higher growth path for GDP going forward.

    (value of positive supply shock + baseline GDP)(1 + r)^t,

    where “r” is the baseline average GDP growth rate and “t” is time.

    In an advanced economy in monetary equilibrium, the rate of productivity growth depends on innovation and investment, with rate of innovation determining “r”(flow) and increased investment due to one time positive supply shocks increasing GDP once(stock).

  56. Gravatar of Michael Sandifer Michael Sandifer
    7. May 2019 at 18:57

    Another way of explaining it is that innovation necessarily involves novel outcomes, hence positive supply shocks. They happen continuously. The tax cuts and related expansion of aggregate supply is a one time event.

  57. Gravatar of dtoh dtoh
    7. May 2019 at 19:58


    Respectfully I think you have it wrong. If you reduce taxes on capital then the share of GDP that goes to investment will increase. This impact is permanent (or at least permanent for the duration of the tex cut.) A permanent increase in investment leads to a permanent increase in the growth rate.

  58. Gravatar of dtoh dtoh
    7. May 2019 at 20:01


    ……An economy that invests 20% of GDP will grow much faster than an economy that only invests 10%. Tax cuts on capital/income permanently increase the share of GDP that is invested and reduces the share that is consumed. Therefore growth is increased on a permanent basis.

  59. Gravatar of Michael Sandifer Michael Sandifer
    7. May 2019 at 20:40


    Okay, so investment increases 10% of GDP due to a one-time policy change. How does that permanently increase the growth rate? What it does, is increases expected real GDP potential until the adjustment is made and real potential increases. Then, that’s it. The economy continues growing at its baseline rate, ceteris paribus, but from a larger base and hence on a higher trend growth path.

    Show me your math on how a one-time positive supply shock permanently increases the growth rate, rather than the base from which the trend growth rate continues, because I can’t see how it would work.

  60. Gravatar of dtoh dtoh
    7. May 2019 at 21:06

    It’s simple. If I have 1

  61. Gravatar of dtoh dtoh
    7. May 2019 at 21:22

    It’s simple. If I have 10 machines and buy 1 machine this year by output goes up 10% next year. If I buy 2 machines my output goes up 20%.

    Assuming starting output of 100, I get 120 output in Y2 by investing in 2 machines. If capital tax rates stay low, I buy another 2 machines in Y2 (or actually 2 x 1.2 machines) and my output in year two goes up to 144.

    Under your scenario, you assume I’ll buy 2 machines in in Y1 and then revert to only buying 1 machine in Y2 and thereafter. But.. that’s not the case, if the government leaves me more money each year, then each year Yn I’ll buy 2 (well actually 2 x 1.2^n) machines.

    If you assume Riccardian equivalence, there’s no long term shift from government spending to private spending, but because the tax burden is reduced from returns on capital there is a shift in private spending from consumption to investment.

  62. Gravatar of Michael Sandifer Michael Sandifer
    8. May 2019 at 10:43


    Correct me if I’m wrong, but what you meant to express was this:

    10 machines(1 + r)^t = # of machines in units of time “t”, where “r” = .20.

    Or, 100(1 + r)^t = # of machines in units of time “t”, where “r” = .02.

    “r” would then reflect a permanent increase in output growth.

    Yes, modeling your idea is easy, but I don’t see how it reflects the reality. Empirically, it certainly doesn’t and it doesn’t make a priori sense to me.

    I start with RGDP as a stock. The one time cut in taxes increases investment, and after adjustment, the RGDP stock increases over some period “t”. So the composition within the increased RGDP stock shifts toward more investment.

    Then, the following period “t”, the RGDP stock grows again, but there’s no reallocation this time toward investment to grow the stock. The stock just grows due to efficiency gains in producing output.

    All the rate of growth in RGDP is here is a series of stocks, divided arbitrarily by within period “t”. Shifts in aggregate supply only increase the stock once, over some period. But, innovation occurs within each period “t”, and so is a rate over time.

  63. Gravatar of Michael Sandifer Michael Sandifer
    8. May 2019 at 15:17


    Actually, after re-reading some of your replies, I think I see where the problem is. You’re assuming compounding of the capital investment, which would ordinarily apply, but won’t in the case of the current US situation. I should have read more carefully, and made fewer assumptions.

    First, here are implicit limits to productivity growth facilitated by increased capital investment, even when such growth is sustainable. There are diminishing returns.

    But, why should we expect no lasting benefit from the Trump tax cuts? Because of how they’re financed. That leaves the cuts boosting capital investment on a one-time basis, not compounded.

  64. Gravatar of Michael Sandifer Michael Sandifer
    8. May 2019 at 15:17

    It’s just a one-time boost to the capital stock.

  65. Gravatar of dtoh dtoh
    8. May 2019 at 16:42

    I’m not following you. Why would you think the rate of investment would rise in the first year and then fall back thereafter to levels preceding the tax cut?

    To use your RGDP stock argument, in each subsequent year that the tax cut remains in effect, a higher percentage of that stock is invested because of the more favorable tax treatment to returns on capital. The compounding effect remains. I can’t imagine why you think it would not, but perhaps I’ve overlooked something.

    Scott makes the argument of diminishing returns also, but I find that wholly unconvincing. I speak only from personal observation and experience (which is not completely insignificant from a statistical perspective,) but with zero further technological advancement, you could easily double the capital stock and output with zero diminishment of returns. And in fact at the current rate of technical advancement, you would actually see vastly increasing improvements in return.

  66. Gravatar of Michael Sandifer Michael Sandifer
    8. May 2019 at 18:56


    This boils down, I think, to your statement above about Ricardian equivalance. I disagree with your statement. The tax cuts were financed with increased deficit spending. Hence, there should only be a one-time increase in the capital stock, with a higher natural interest rate choking off future additional investment.

  67. Gravatar of Michael Sandifer Michael Sandifer
    8. May 2019 at 18:57

    Oh, and the boost to investment could last for more than a year. That’s why I use the more generic term “period” above. But, it probably won’t be much more than a year.

  68. Gravatar of dtoh dtoh
    8. May 2019 at 20:06


    So a couple of things. If higher rates choked off additional investment, why wouldn’t it happen on day one.

    But in fact it doesn’t happen. We don’t really care about rates, we only care about the volume of credit/investment. Faster growth equals a higher natural rate, but it also causes a shift in the IS curve so firms and individuals will borrow the same amount or more even though rates are higher. And at the same time you’ll get an overall expansion of credit.

    From the perspective of a firm, I don’t really care that much if rates go up. (It has some effect but it’s not that big.) Generally interest expense is not that big an item, and if my rates go up, so do all my competitors and it gets passed on to customers in the form of higher prices. As a firm the only thing I care about is how much free cash I have to invest. If you cut my tax rates, it has a huge impact on how much I can invest.

    That said, I think the fact that the depreciation provisions and the treatment of pass-through income is temporary will somewhat limit the long term effectiveness of the cuts.

  69. Gravatar of ssumner ssumner
    9. May 2019 at 09:59

    dtoh, You said:

    “You run the risk of being perceived as favoring low growth because of your feelings about the current President. And BTW… you should lay off the anti-Trump screeds. You ruin your chance of being appointed to the Fed. Regardless of your personal views and the rather low probability of it happening, it would be good for aggregate utility if you were appointed to the Fed. So stop being selfish and at least give yourself a chance.”

    I disagree with everything in your comment, including the LRAS curve. But this is especially silly. First of all, I have no interest in working for the Fed. Not sure why people don’t take me at my word. If I did then obviously I would have been more diplomatic over here. Second, people have been doubting me from day one, telling me I’m doing things all wrong. I keep telling people that the only way to change policy is change the zeitgeist, not change personnel. But if you want to obsess about personnel, there’s probably a chance that someone like Beckworth might be nominated, or at least someone who is sympathetic to MM ideas. What were the odds of that happening back in 2009, before David and I started blogging. Conservatives are edging toward MM. So who says ideas don’t matter? In fact, I’ve been consistently right and my doubters have been consistently wrong.

    Third, I switched from 5% to 4% even before Trump became President, so it has nothing to do with Trump. Even a dove like Christy Romer came in at 4.5%. I picked a number that will be acceptable to hawks who obsess about 2% inflation. But your argument is not even logical, even if I did want to sabotage Trump. Obviously nothing I say here will have the slightest effect on the Fed, who are going to stick with 2% inflation for the next few years, at least. Furthermore, I’m probably the only Trump hater in the entire world who has done a blog post stating clearly that if there is a recession it almost certainly will not be Trump’s fault. If my blogging is motivated my Trump hate, why would I preclude my ability to blame Trump for a recession, if it were to occur. Why did I disagree with Krugman’s claim that the corporate tax cut isn’t boosting growth? In contrast, if there is a recession then you will be forced to blame Trump. You blamed Bernanke for 2008, so logically you’d have to blame Powell, who was picked by . . . Trump.

    For God’s sake, people need to use some logic!

    It’s actually the Trumpistas that are so blinded by politics that they think the world revolves around Trump, and everything that happens reflects on Trump. At least 97% of what happens with the economy has NOTHING TO DO WITH WHO IS PRESIDENT.

    If my plumber is doing a bad job I don’t advocate 4% growth to sabotage my plumber.

  70. Gravatar of ssumner ssumner
    9. May 2019 at 10:02

    dtoh, Higher NGDP growth means higher tax rates on capital income, because the tax applies to nominal income. You believe taxes are capital are especially bad for growth, more so than anyone else I know. So you should be especially concerned about higher trend NGDP growth. Most conservative supply siders favor low inflation, for that reason.

  71. Gravatar of dtoh dtoh
    9. May 2019 at 11:30

    I know you have no interest in being on the Fed… but I’m saying you should sacrifice your personal interest because it would be beneficial to society if you were on the Fed. A lot of people who do important things do them NOT because they want to, but because no one else can or will step up.

    I took you on your word on an earlier post that the cut in your recommended target rate for NGDP had nothing to do with Trump. What I was asking in my comment was the reason for the change if the target rate does not matter. If 3% inflation is no worse than 2% inflation… why kowtow the hawks instead of just telling them they’re wrong.

    The President doesn’t make a difference? Would we have had the tax cut with Hillary. Assuming the effect is a one time bump of 1%. That’s 5 trillion dollars over the next twenty years. I’d say that makes a difference.

    Still not sure about you’re argument on NGDP and taxes. Are you talking about bracket creep? Or just nominally higher taxes but at the same rate. In any case, with a 0% tax rate it wouldn’t make any difference.

  72. Gravatar of Michael Sandifer Michael Sandifer
    10. May 2019 at 06:16


    A combination of lagged effects from and initial uncertainty as to the increase in deficits allows for an initial positive effect from the tax cuts.

    There’s decreased temporal discounting and uncertainty during period 1, effectively killing any compound effects.

  73. Gravatar of dtoh dtoh
    10. May 2019 at 11:22


    I’m not sure I’m following. When you talk about lagged effects and initial uncertainty are you referring to Ricardian equivalence.

    If so, Ricardian equivalence does not matter.

    Think about a revenue neutral change to the tax code, where taxes on capital were reduced and taxes on consumption were increased. Then think about the recent tax changes as consisting of two separate components. One is an overall cut in the amount of tax collected. The other is a shifting of the tax burden from capital to consumption.

    The latter causes a permanent increase in the ratio of investment to consumption, raises the growth rate and is not impacted by Ricardian effects.

  74. Gravatar of ssumner ssumner
    10. May 2019 at 16:03

    dtoh, If there’s not much difference between 2% and 3%, and it’s not even clear which is better, then why should I waste time fighting for one over the other?

    Yes, I’m selfish. If I weren’t I’d give $10,000 each to 50 rural families in India. But your claim that me not wanting to be on the Fed represents selfishness seems kind of weird. There are plenty of people just as qualified as me, indeed more so.

    The tax argument has nothing to do with bracket creep, it’s because the Feds tax nominal returns, not real returns. Make the nominal return on investments higher and the effective tax rate will also be higher. With a 2% real rate and a 50% tax rate, then 0% inflation means 1% after tax real return, and 2% inflation means a 0% after tax return. A 2% inflation rate pushes the effective tax rate to 100%.

    Hillary would have cut the corporate rate to 25%, in exchange for some goodies that she wanted from the GOP.

  75. Gravatar of dtoh dtoh
    10. May 2019 at 16:49


    IME – Everyone is selfish and everyone is generous. There are probably people more qualified, but think how much attention NGDP targeting would get if you were nominated.

    I got you on inflation and taxes. I don’t know why I keep forgetting this even though you’ve talked about this several times. However, a couple of thoughts….

    1. Isn’t inflation taken into account when setting tax rates. E.g. a lot of states limit property tax increases be gearing them to inflation.

    2. Is it a bad thing to tax people who use capital inefficiently more highly?

    3. Actually I’m not against all taxes on capital. For example, I’m strongly in favor of highly progressive fixed asset taxes on property used for personal consumption (expensive cars, airplanes, big houses, etc.) I’m just in favor of limiting taxes on capital used for production.

    On 2% versus 3%, I would agree, but…..

    The difference in the impact on output between 2 and 3 is smaller then the difference between and 1 and 2.

    If we don’t know trend growth, we don’t know whether we’re talking about 2 versus 3 or 1 versus 2.

    It would be useful to know the utility and/or output trade off going from no inflation to 3 or 4% inflation. We know there’s an important impact at zero otherwise we wouldn’t worry about sticky wages/prices and monetary policy would be a non-issue. We know at high levels, additional inflation has no impact on output. Don’t we want to know what the curve looks like where there is an actual trade off?

    Can I assume that you would be opposed to an NGDP target that assumed no inflation? What about 0.1% inflation? 0.5%? 1.0%?

Leave a Reply