Is it time to blow up the New Keynesian model?

This is from a paper by Gauti Eggertsson, a very distinguished New Keynesian economist:

Tax cuts can deepen a recession if the short-term nominal interest rate is zero, according to a standard New Keynesian business cycle model. An example of a contractionary tax cut is a reduction in taxes on wages. This tax cut deepens a recession because it increases deflationary pressures. Another example is a cut in capital taxes. This tax cut deepens a recession because it encourages people to save instead of spend at a time when more spending is needed.

Elsewhere he showed that in a deep depression, the NK model suggested that artificial attempts to raise wages, such as FDR’s NIRA, could be expansionary.

Several economists have recently suggested that the NK model has NeoFisherian implications. If the Fed were to raise its fed funds target, the policy shift would be inflationary.

Nick Rowe shows that the NK model implies that a slowdown in the rate of government spending growth is expansionary.

To summarize:

1.  The NK model implies that higher taxes on wages can be expansionary.

2.  The NK model implies that higher capital gains taxes can be expansionary

3.  The NK model implies that raising the aggregate wage level by government fiat can be expansionary.

4.  The NK model implies that an increase in the fed funds target can be expansionary.

5.  The NK model implies that fiscal austerity can be expansionary, if done by slowing the growth in government spending.

These are claims made by NK supporters, NK opponents, and people in the middle. And what they all have in common is that they suggest that the NK model has preposterous implications.  These claims about the world aren’t just wrong, they are borderline absurd.  I’ve studied macroeconomics all my life, both the Great Depression and the post-1970 period.  I’ve seen hundreds of natural (policy) experiments, and watched how both markets and the broader economy reacted to those natural experiments.  And if these claims are true then I might as well stop being an economist, because it would mean I understand nothing about the world, absolutely nothing.

Perhaps merely to keep my sanity, I prefer to find another model, which doesn’t yield one preposterous result after another.  Here’s the best I can do:

The Musical Chairs model:

1.  In the short run, employment fluctuations are driven by variations in the NGDP/Wage ratio.

2.  Monetary policy drives NGDP, by influencing the supply and demand for base money.

3.  Nominal wages are sticky in the short run, and hence NGDP shocks cause variations in employment in the same direction.

4.  In the long run, wages are flexible and adjust to changes in NGDP. Unemployment returns to the natural rate (currently about 5% in the US.)

This models seems to fit the stylized facts quite well:

In my model, artificial attempts to raise wages are contractionary, because they reduce NGDP/W.

In my model, higher wage taxes are contractionary, because they reduce NGDP/W (“W” is actually hourly labor costs for firms, including employer-side taxes and benefits.)

In my model, capital gains taxes have little impact on employment.

In my model, increases in the fed funds target are achieved via decreases in the base. This reduces NGDP, and thus NGDP/W, and therefore is contractionary.

In my model, a decrease in the growth rate of government spending doesn’t effect employment.

So in all five cases where the NK model has loony implications, my model has implications that are more consistent with what we know about how the world actually works.

Oh, and one other thing, the NK model has enormous academic prestige, whereas my model is ignored, or dismissed as non-rigorous.

I am quite certain that my model will not win out in the long run; it’s too crude. (Even I could do better.)  But I’m equally certain that the NK model can’t survive in its current form.  It’s time to blow it up, and start over with a completely different framework. Preferably a model that doesn’t focus on inflation, interest rates and output, but rather NGDP, nominal wages, and employment.

PS.  I have a new post at Econlog; metaphors run wild.


Tags:

 
 
 

63 Responses to “Is it time to blow up the New Keynesian model?”

  1. Gravatar of E. Harding E. Harding
    31. October 2015 at 11:23

    Hm. So payroll and capital tax increases during a recession are expansionary. No wonder 2013 wasn’t the disaster predicted by some Keynesians.

  2. Gravatar of Miguel Miguel
    31. October 2015 at 12:09

    In my model, at ZLB, a increase of G financed with monetary base is expansionary in the short run. I don’t know in long run. But in short run, it is impossible that it would not be expansionary.

  3. Gravatar of TallDave TallDave
    31. October 2015 at 12:18

    The surfeit of models in economics is amusing to us programmers, who routinely discard more models in a week than economics seems to dispose of over a century.

  4. Gravatar of Doug M Doug M
    31. October 2015 at 12:25

    I my model profitability is the primary driver of employment.
    When a firm is profitable it adds workers when it is not, it cuts workers.
    At the firm level employment follows a saw-tooth pattern. Workers are added gradually over time, workers are dismissed en-mass.
    In a competitive economy profitability is not guaranteed. It is very rare that a firm is continuously profitable.
    Never is every firm profitable nor is it ever the case that every firm is unprofitable.

  5. Gravatar of Jean Jean
    31. October 2015 at 12:30

    Of course it is time to drop Keynes – neo or otherwise. Everything he wrote was for a gold standard world. Keynes was made irrelevant by 1933 when Fischer persuaded FDR to confiscate gold. The gold standard ‘lived’ on, from 1933, in a coma with artificial life support, until Nixon closed the gold window and much of the world went to free floating currencies.
    Can we please have price level targeting now?

  6. Gravatar of David Gulley David Gulley
    31. October 2015 at 14:01

    Scott,

    At first I thought that Eggertsson was using these results to suggest the NK model isn’t all that useful at the ZLB. Alas, no. The intent was to provide rationale for more G. At no point does he question his results.

  7. Gravatar of John Handley John Handley
    31. October 2015 at 14:48

    Scott,

    I agree with you that

    “1. The NK model implies that higher taxes on wages can be expansionary.

    2. The NK model implies that higher capital gains taxes can be expansionary”

    are both ridiculous and point to the fact that NK models are bogus at the ZLB, but on four and five, you just failing to understand NK properly.

    The Neo-Fisherian result is entirely dependent on equilibrium selection and there are only multiple equilibria because fiscal policy is ambiguous as there is not a specified monetary policy rule. The whole problem could be solved if money demand was added to the model, which is extremely different from “blow[ing] up the New Keynesian model”

    As to number 5, the NK model never shows that austerity is expansionary, simply that it raises the natural rate of interest which might be able to help the economy off of the ZLB. You can read my comments on Nick Rowe’s post (and his replies) for a bit more information, but government spending has the same effect in NK models as it does in RBC models. That is, any permanent reduction in government spending causes a permanent reduction in output.

    Essentially the effect Nick is highlighting is that, by reducing potential output relative to actual output, austerity raises the natural rate of interest. Austerity is not expansionary in this scenario because it’s not actually raising output.

    If the central bank couldn’t observe the natural level of output (as they can’t in real life), austerity would have the expected contractionary affect.

    As to the musical chairs model, what you have outlined in this post is non-rigorous. You’ve given a bunch of qualities of a model that someone could build, but you haven’t given a model. Here’s what I can gather from your description: the only friction in you model is nominal wage rigidity and it has a money demand function. This makes me think of the model in this paper: http://www.columbia.edu/~mu2166/Making_Contraction/paper.pdf with the addition of money demand and a policy rule that uses the money supply as opposed to the nominal interest rate.

    Now I suppose the only task for market monetarists is to come up with a money demand function that is empirically accurate and does not create an indeterminacy at the ZLB.

  8. Gravatar of marcus nunes marcus nunes
    31. October 2015 at 16:05

    @ David G
    How can one question a “rigorously derived” result?

  9. Gravatar of benjamin cole benjamin cole
    31. October 2015 at 16:47

    I do confess one weakness for the Neofisherian model: if the Fed, by holding the Fed Funds rate near zero, beats inflation, then the unemployment target rate should also be very low, say around 1% or 2%, or even target labor shortages. Why not? We can’t have inflation as long as the Fed keeps the Fed Funds to rate low.

    I like that fantasy better than the NK fantasies.

  10. Gravatar of ssumner ssumner
    31. October 2015 at 16:55

    E. Harding. Good point, although strictly speaking the payroll tax argument applies to employer side changes, and the 2013 increase was an employee side change.

    Miguel, Does that model apply to Japan? And if so, what went wrong?

    Jean, I agree that Keynes’s model is basically a gold standard model.

    David, No, he’s a fan of the NK model.

  11. Gravatar of E. Harding E. Harding
    31. October 2015 at 17:29

    @ssumner

    -What’s the difference? Didn’t you point out tax incidence theory effectively led to employer- and employee- side taxes having the same results?

  12. Gravatar of Ray Lopez Ray Lopez
    31. October 2015 at 19:35

    The reason NK models predict a tax cut is deflationary is it makes assumptions about how governments vs people save or spend money: it’s well known rich people spend less than poor for example.

    Sumner: “I am quite certain that my model will not win out in the long run; it’s too crude. … But I’m equally certain that the NK model can’t survive in its current form. It’s time to blow it up, and start over with a completely different framework.” – how about this: the economy is non-linear, albeit cyclical, and what works in the past may not work today (sticky wages not so sticky anymore for example; velocity is not so constant anymore for another). In the long run, only technology matters (Solow model), and arguably technology is not exogenous but driven by demographics (Great Stagnation theme). So just accept this fact and stop the vain attempt to influence the economy by printing money.

    The above model is simplicity. Simplicity, sans Stubborn Scott Sumner (S3) stumping stupid shibboleths*? [hereinafter S8]

    * Nobel Prize-laureate economist Paul Samuelson applied the term “shibboleth” in works including Foundations of Economic Analysis to an idea for which “the means becomes the end, and the letter of the law takes precedence over the spirit.”

  13. Gravatar of Tom Brown Tom Brown
    31. October 2015 at 20:28

    Scott, here’s a model with pretty straightforward math. I’m a long ways from being a math genius, but even I understand this journal-style paper explaining it:
    http://econpapers.repec.org/paper/arxpapers/1510.02435.htm
    It may not suffer from the problems you identify above: you’d have to ask the author.

    I do know that it doesn’t use the usual micro foundations representative agent that I assume the NK model does. Instead you can think of it using an emergent “representative agent” if you want, but you don’t need a representative agent at all to explain the model. Plus it makes predictions that you can check (the author will send you the code upon request). Here’s one recently updated example for core PCE inflation (the model is compared to FOMC forecasts and the NY Fed DSGE model):
    http://informationtransfereconomics.blogspot.com/2015/10/core-pce-inflation-update.html?showComment=1446350474506#c7981707331453365587
    Enjoy

  14. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    1. November 2015 at 05:59

    ‘In my model, artificial attempts to raise wages are contractionary, because they reduce NGDP/W.’

    As Tacoma WA is about to discover when they vote to raise the minimum wage to either $12 or $15 this week;

    http://www.thenewstribune.com/news/local/politics-government/election/article41869509.html

    ———–quote———-
    The accountant, Margaret Clarkson, 81, said that for a tavern, that threshold amount could equate to $50,000 net, under tight management.

    The $12 an hour minimum, if passed, would affect all businesses. For [the bar owner] Peluso, who is 59, it would add about $25,000 a year to his costs. He grimaced and clutched his glasses while he explained his dilemma.

    “I am not against people making a livable wage. I am not against that,” he said. “… I grew up around here. I see how people struggle.”

    He added that success for either minimum-wage proposal could push him toward selling his bar sooner than he had planned. The lower would allow him to keep afloat a few months while he gauged its effects, he said.

    “I can’t vote for 15,” he said. “I may vote for 12, to be honest with you, if it’s done right. It’s still a lot, a lot of money.”
    ————-endquote———-

    He doesn’t seem to have considered that a raise in the minimum will affect the price another businessman will be willing to pay for his bar.

  15. Gravatar of ssumner ssumner
    1. November 2015 at 06:13

    John, Thanks for your very useful comments. But I have a few questions:

    1. You say that adding money demand to the NK model would help address the NeoFisherian problem. That seems plausible. But doesn’t that also expose a major weakness of NK models, that they rely on interest rates, rather than the quantity of money? And doesn’t the reliance on interest rates lead many NK economists to misjudge the stance of monetary policy in real time? Didn’t many NK economists believe monetary policy was expansionary during 2008? (I understand that they are now re-evaluating, and Curdia has a paper saying policy was actually tight. But I talking about in real time.) Market monetarists are not confused about whether money was easy or tight in 2008.

    2. You said:

    “You can read my comments on Nick Rowe’s post (and his replies) for a bit more information, but government spending has the same effect in NK models as it does in RBC models. That is, any permanent reduction in government spending causes a permanent reduction in output.”

    But I find that result to be equally bizarre. Why would less government spending permanently lower output? Does it permanently lower employment? If so why? Or does it permanently lower productivity? And if so, why? The only mechanism I am aware of is that government output is wasteful, and hence lower output makes workers feel richer, and hence they work less hours. But then does output, properly measured, actually fall?

    You said:

    “As to the musical chairs model, what you have outlined in this post is non-rigorous. You’ve given a bunch of qualities of a model that someone could build, but you haven’t given a model.”

    Actually I have provided a model. But this is a very common view I see among modern economists, equating the terms ‘model’ and ‘mathematical model.’ Models do not have to be expressed in the form of equations. It may well be desirable to express them in the form of equations, but it’s not necessary. Merely writing down:

    M/P = f(i, Y) doesn’t make the model more “rigorous” that saying “base demand is a function of interest rates and output.”

    You said:

    “Now I suppose the only task for market monetarists is to come up with a money demand function that is empirically accurate and does not create an indeterminacy at the ZLB.”

    Actually market monetarists oppose the old monetarist approach of trying to model money demand. We favor targeting market expectations of future NGDP. The zero bound question is interesting, although widely misunderstood by almost all macroeconomists (as I’ve argued in many other posts.) But it’s not really the essence of any NK model, or any MM model to replace it.

    But again, thanks for those comments, they were very helpful.

    E. Harding, It makes no difference in the long run, when wages are flexible. But it does in the short run if nominal wages are sticky.

    Ray attempts to discuss NK. Cute.

    Tom, I’m not quite sure what you see in Jason’s models. Can you describe what you like in simple English?

  16. Gravatar of Ray Lopez Ray Lopez
    1. November 2015 at 06:23

    The paper cited by Tom Brown is nothing more than a pedagogical teaching aid, that essentially is using a ‘representative agent’ writ large. Note how it smooths actual data (see Fig. 10), and as such does not use interactive agents, but instead statistical averages (i.e. a smoothing function) and first order differential equations with constant coefficients along the lines of f'(x) = k*x, which leads to a standard exponential function and/or Cobb-Douglas logarithmic function. In short, the paper is trivial (and over Sumner’s head)

    Paper: “The present approach can be thought of as looking at the economy from a telescope on a distant planet and treating economic agents as invisible atoms. Even if it does not lead any further than the models presented here, the infor-mation equilibrium framework may still have a pedagogical use in standardizing and simplifying the approach to Marshallian crossing diagrams”

  17. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    1. November 2015 at 06:30

    Anyone seen Charlie Calomiris this week-end? He must have a big smile on his face after this;

    http://www.federalreserve.gov/newsevents/press/bcreg/20151030a.htm

    ————quote————
    The proposed rule would apply to domestic firms identified by the Board as global systemically important banks (GSIBs) and to the U.S. operations of foreign GSIBs. These institutions would be required to meet a new long-term debt requirement and a new “total loss-absorbing capacity,” or TLAC, requirement. The requirements will bolster financial stability by improving the ability of banks covered by the rule to withstand financial stress and failure without imposing losses on taxpayers.

    To reduce the systemic impact of the failure of a GSIB, an orderly resolution process should allow a GSIB to fail, and its investors to suffer losses, while the critical operations of the firm continue to function. Requiring GSIBs to hold sufficient amounts of long-term debt, which can be converted to equity during resolution, would facilitate this by providing a source of private capital to support the firms’ critical operations during resolution.
    ————-endquote————

    Team CoCo wins!

  18. Gravatar of Benjamin Cole Benjamin Cole
    1. November 2015 at 06:55

    Pat Sullivan: I like the idea of banks issuing mandatory convertible debt, triggered by certain clauses or covenants.

    Still, usually when a publicly held business declares Chapter 11, the public shareholders are squeezed out and bondholders take over. Indeed, there is active trading in the bond debt as various private equity “vulture funds” jockey to get a controlling stake of the bonds.

    Which makes me wonder (and as some have said): Were the US bank bailouts just bailouts of bondholders? Why not let the bondholders run the banks? Or was there yet as shortage of money to keep the banks running?

  19. Gravatar of TravisV TravisV
    1. November 2015 at 07:24

    WOOOOOOOOO!!!!

    http://www.reuters.com/article/2015/10/31/ecb-draghi-idUSL8N12V04A20151031

    Draghi: “ECB to do all needed to keep inflation target on track”

  20. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    1. November 2015 at 08:44

    Benjamin, it’s all about the incentives the CoCos create. Presumably the bondholders didn’t want equity in the bank or they would have bought it in the first place. Current equity holders don’t want their ownership diluted by the creation of new shares not owned by them. Managers don’t want to anger either group, because they might get fired.

    If a bank’s financial condition is moving toward the trigger that will automatically convert the CoCos, that’s a canary in the coal mine.

    Ironically, Gramm, Leach, Bliley (1999) authorized what the Fed just announced they’re proposing. The Fed studied it back then too, concluded it was a good idea…and said, ‘Therefore we’ll study it some more.’ Had they implemented it then, there probably would have been no crisis.

  21. Gravatar of John Handley John Handley
    1. November 2015 at 08:45

    Scott,

    I won’t discount the possibility that too much focus on interest rates led to wrong predictions about monetary policy in 2008. In fact, I agree that interest rates are a bad indicator of the stance of monetary policy. I would argue, however, that focusing too much on interest rates is a problem that stems from the economists who wanted to ignore money demand when writing down their models rather than the models themselves.

    I think the typical explanation for the real effects of government spending is that the wealth effect induces agents to work more hours when they expect more taxes. This is not a reduction in productivity; agents work more, so output goes up.

    The lack of rigor in the musical chairs model extends beyond simply lacking a bunch of equations. The predictions that it can make are entirely qualitative as opposed to the quantitative predictions of the more empirical New Keynesian models like Smets-Wouters 2003, for example. On top of that, M/P = f(i, Y) is equally as non-rigorous as “base demand is a function of interest rates and output”. They both literally read the same way. If, on the other hand, you gave me an actual function, say log(M/P) = log(Y) – ai, then you would have given a lot more information about your model than the text. Now I know what shape your money demand function takes and I have the ability to estimate it.

    How do market monetarists explain the fact that central banks can seemingly increase the base ad infinitum without causing inflation to budge? (see the monetary base in the UK, the US, and Japan). Conventional models suggest that agents don’t care about the composition of their asset portfolio’s when the opportunity cost of holding money is zero, so monetary expansion is useless when the interest rate on reserves is equal to the interest rate on other assets. This generally seems to line up with low inflation in areas that have tried QE at the zero lower bound (again, see the UK, the US, and Japan).

    Thanks for the response!

  22. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    1. November 2015 at 08:57

    Here’s Calomiris in 2014;

    http://www.wsj.com/articles/charles-calomiris-a-better-solution-for-too-big-to-fail-banks-1401405342

    ———-quote———
    Start with a requirement that a megabank maintain a 10% book equity-to-asset requirement””but add to it the requirement that 10% of its assets be issued in CoCos that convert from debt into equity if the market value of equity relative to assets falls below a critical ratio, say 10%, on average for a period of 120 days. If conversion does occur, the CoCos exchange at a premium of, say, 5%. CoCo holders end up with more shares than the face value of their debt holdings.

    By using the market value of a bank’s equity as a conversion trigger, bank managers have an incentive to maintain sufficient true economic capital. Conversion means a significant amount of stock is issued, diluting the value of the equity held by the rest of the owners. To avoid this outcome, bank managers would choose to issue new stock (fewer shares than would be issued than under conversion) to offset declines in their market valuation. Managers would make that choice because dilutive conversion is more costly to existing stockholders; both the holders of newly converted shares and pre-existing shareholders would likely agree to sack management incompetent enough to allow such a conversion to happen.

    The 120-day moving average ensures that banks have plenty of time to arrange an offering in response to market perceptions of losses. Setting the CoCo trigger at 10%””far above the insolvency point of the bank””ensures that the bank still will have access to the market to issue new equity.

    Relying on the market’s perception of bank losses to encourage new equity offerings sidesteps the problem of bank managers and regulators understating or manipulating the riskiness of their assets. Market expectations about cash flows will determine the market value of a megabank’s equity, avoiding unwarranted emphasis on balance sheets to gauge bank health. And the higher the riskiness of a bank’s cash flows, the more willing the bank will be to maintain a large buffer of equity over and above the trigger threshold. The bank’s voluntarily chosen market-to-equity ratio will vary appropriately with its cash flow risk.
    ———–endquote———-

  23. Gravatar of Ray Lopez Ray Lopez
    1. November 2015 at 09:17

    Sumner: “Ray attempts to discuss NK. Cute.” – no, I am discussing a simpler framework to describe the economy, as you wished in your closing paragraph (“It’s time to blow it up, and start over with a completely different framework”)

    Lack of reading comprehension is not a bug with Sumner, it’s a feature of cognitive dissonance, to prevent him from having to change his warped mind.

  24. Gravatar of Ray Lopez Ray Lopez
    1. November 2015 at 09:21

    @Patrick R. Sullivan – Calomiris is talking his book “Fragile by Design”, and it’s not clear how this topic is relevant today, 7 years after the crash. The meltdown of the financial sector may have contributed to the recession in 2009 and 2010, but now? It’s unlikely to be the reason for the continued funk, anymore than zombie banks are the reason in Japan. More likely structural reasons are to blame. And, pace our host, printing money will not cure anything.

  25. Gravatar of Jason Jason
    1. November 2015 at 11:22

    Ray, haha, funny you should mention reading comprehension.

  26. Gravatar of ssumner ssumner
    1. November 2015 at 15:17

    Patrick, That sounds very promising.

    You said:

    “Ironically, Gramm, Leach, Bliley (1999) authorized what the Fed just announced they’re proposing. The Fed studied it back then too, concluded it was a good idea…and said, ‘Therefore we’ll study it some more.’ Had they implemented it then, there probably would have been no crisis.”

    Interesting. Do you know of a paper that makes that claim?

    Travis, That sounds very promising.

    John, Your first point about the failure of NKs to properly focus on shifts in money demand, is a good one, and one I expected you to make. But can you see how a pragmatist like me would prefer a model that didn’t seem to easily lead people astray?

    You said:

    “I think the typical explanation for the real effects of government spending is that the wealth effect induces agents to work more hours when they expect more taxes. This is not a reduction in productivity; agents work more, so output goes up.”

    Yes, I assumed it was labor supply, basically because they feel poorer. But I don’t really buy that implication of the model, I don’t think more G causes GDP to rise in the long run, as the negative supply-side effects overwhelm the mechanism you refer to.

    You said:

    “They both literally read the same way. If, on the other hand, you gave me an actual function, say log(M/P) = log(Y) – ai, then you would have given a lot more information about your model than the text. Now I know what shape your money demand function takes and I have the ability to estimate it.”

    Yes, that’s basically what I meant, but again, the MM model doesn’t require that policymakers estimate a money demand function, just that there is no zero lower bound problem with NGDP futures targeting, and no shortage of eligible assets for the central bank to buy.

    You said:

    “How do market monetarists explain the fact that central banks can seemingly increase the base ad infinitum without causing inflation to budge? (see the monetary base in the UK, the US, and Japan). Conventional models suggest that agents don’t care about the composition of their asset portfolio’s when the opportunity cost of holding money is zero, so monetary expansion is useless when the interest rate on reserves is equal to the interest rate on other assets. This generally seems to line up with low inflation in areas that have tried QE at the zero lower bound (again, see the UK, the US, and Japan).”

    I could write a book in response to this question, and indeed have 100s of blog posts on the subject. Here I’ll just make a few brief points. There is no evidence at all that the BOJ has fallen short of its inflation goals until the last few months, and I expect them to take further steps. The cause of the bloated monetary bases is tight money policies, which cause slow NGDP growth and hence very low nominal interest rates. Most economists look at this issue backwards. The question is not what QE can or cannot do at zero rates, it’s how much base money as a share of GDP does the public want to hold when NGDP growth is expected to be on target. I predict not very much (see Australia). But if I’m wrong, then policymakers need to choose between socialism and inflation (something most conservatives haven’t figured out yet.) That is, they need to choose between a balance sheet so big it goes beyond public debt, and a higher NGDP (and hence higher nominal interest rate.)

    BTW, the Fed is thinking of raising rates, so don’t assume that NGDP is lower than they want it.

    Of course there’s much more to be said on this issue.

  27. Gravatar of John Handley John Handley
    1. November 2015 at 16:37

    Scott,

    If someone were to write down a model that was exactly the same as NK except it had a money demand function and the central bank adjusted the money supply according to a rule instead of the interest rate, it would still be a New Keynesian model. The fact that economists ignore money demand is their own stylistic choice and not a reflection of the necessary characteristics of New Keynesian models. As far as I’m concerned, any model with monopolistic competition and pricing frictions counts as a New Keynesian model.

    The result is somewhat dependent upon the utility function, but the effect of a permanent increase in government spending funded by distortionary taxes causes output to increase in RBC (and NK) models.

    So you get around the zero lower bound problem by assuming it doesn’t exist? Where are your microfoundations for this? What is your theory of money demand? This seems extremely suspect in the eyes of the Lucas Critique.

    Why don’t you think monetary expansion is irrelevant when the opportunity cost of holding money is zero. Is there some other reason why agents would want to economize their money balances?

    The ineptitude of the Fed doesn’t change the facts. Inflation has been below target for some time now and the fact that the Fed somehow doesn’t care is not relevant to the apparent ineffectiveness of QE.

  28. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    1. November 2015 at 17:16

    The 1999 Fed study is here;

    http://www.federalreserve.gov/pubs/staffstudies/1990-99/ss172.pdf

    ‘Using Subordinated Debt as an Instrument of Market Discipline’

    from the conclusion;

    ‘Although we do not draw policy conclusions, our
    study makes clear that assessment of the benefits and
    costs of a policy proposal would be helped greatly
    by more research in a number of areas. For example,
    a better understanding of the marginal benefits of
    requiring banks to issue SND relative to the benefits
    of the existing SND market, along with associated
    marginal costs, would be quite useful. Such a study
    would need to examine the market discipline benefits
    of currently outstanding SND as well as the benefits
    of other medium- and long-term uninsured liabilities.
    A second area of useful research would be a close
    examination of the potential benefits of using the
    existing SND market for banking organizations, and
    the current markets for BHC equity and selected
    uninsured bank liabilities, as aids to bank supervi-
    sory surveillance activities.’

  29. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    1. November 2015 at 17:27

    And a Calomiris and Herring paper;

    ‘Why and How to Design a Contingent Convertible Debt Requirement’

    http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1815406

    Footnote 14 reads, in part;

    ‘In response to the mandate within the Gramm-Leach-Bliley Act of 1999 that required the Federal Reserve and the Treasury to study the efficacy of a sub debt requirement, a Federal Reserve Board study reviewing and extending the empirical literature broadly concluded that sub debt could play a useful role as a signal of risk. Despite that conclusion, no action was taken to require a sub debt component in capital requirements; instead the Fed concluded that more research was needed.’

  30. Gravatar of Ray Lopez Ray Lopez
    1. November 2015 at 18:55

    From Brad DeLong’s blog “Cracking the Hard Shell of the Macroeconomic Knut: “Keynesian”, “Friedmanite”, and “Wicksellian” Epistemes in Macroeconomics” – more proof that Sumner and his colleagues are all arguing over semantics: Keynesian, Friedmanite and Wicksellian frameworks are all roughly the same (I too found this out years ago, for the first two, flip sides of the same coin).

    The irony is that the economy is non-linear and money neutral, so none of these frameworks explain anything, anymore than after-the-fact stock market explanations do (i.e., ‘the stock market’s up/down/sideways due to (insert reason here) say experts’). They sound plausible but have no predictive efficacy.

    DeLong:

    Greenspan thus shifted the focus of America’s macroeconomic discussion away from the level of spending and the quantity of money to the configuration of asset prices. In some ways this is no big deal: “Keynesian”, “Friedmanite”, and “Wicksellian” frameworks are all perfectly-fine ways to think about macroeconomic policy. They are different–some ideas and some factors are much easier to express and focus on and are much more intuitive in one of the frameworks than in the others. But they are not untranslateable–I have not found any point that you can express in one framework that cannot be more-or-less adequately translated into the others.

  31. Gravatar of gofx gofx
    1. November 2015 at 19:23

    @John Handley. I don’t think the “opportunity cost of holding money” is necessarily zero at the ZLB.

  32. Gravatar of Major.Freedom Major.Freedom
    1. November 2015 at 19:25

    Ray:

    MONEY IS NOT NEUTRAL.

  33. Gravatar of James Alexander James Alexander
    1. November 2015 at 23:09

    TravisV
    Draghi has said that before, but it is wearing a bit thin. He needs to stop talking about better weapons and instead alter the target – the best weapon of all.

    John Handley
    “I think the typical explanation for the real effects of government spending is that the wealth effect induces agents to work more hours when they expect more taxes. This is not a reduction in productivity; agents work more, so output goes up.”
    Bring on that 90% marginal tax rate. That’ll get us all out of bed in the morning, or at least down to our tax advisors.

  34. Gravatar of James Alexander James Alexander
    1. November 2015 at 23:31

    John Handley
    The major central banks all forecast inflation returning to 2% within 2 years or so, with a risk of overshooting in 3 years. The expectation in the market is that central banks think money is slightly too loose. Thus, their QE is partly offset by their model. The market sees the model as wrong judged by constant under-shooting, but sees central bankers clinging to the model. It’s a condundrum, but also a fact. No wonder QE doesn’t work very well.

  35. Gravatar of Central bankers on target with their targets | Historinhas Central bankers on target with their targets | Historinhas
    2. November 2015 at 04:33

    […] has been an illuminating exchange between Scott Sumner and Nick Rowe with John Handley who has been defending bravely New Keynesian models. New […]

  36. Gravatar of Nick Rowe Nick Rowe
    2. November 2015 at 05:26

    Scott and John: Assume an RBC model, and lump-sum taxes, and that government spending gives people no utility (it’s wasteful). Then a permanent increase in G leads to a wealth effect increase in labour supply, and higher *measured* RGDP. (But *useful* RGDP falls.)

    Adding in distortionary taxes would have an offsetting effect.

  37. Gravatar of ssumner ssumner
    2. November 2015 at 06:05

    John, Good questions. You said:

    “If someone were to write down a model that was exactly the same as NK except it had a money demand function and the central bank adjusted the money supply according to a rule instead of the interest rate, it would still be a New Keynesian model. The fact that economists ignore money demand is their own stylistic choice and not a reflection of the necessary characteristics of New Keynesian models. As far as I’m concerned, any model with monopolistic competition and pricing frictions counts as a New Keynesian model.”

    Most NK models that I have seen focus on three key variables; interest rates, inflation and output. In my view all three are the wrong variable. Interest rates should be replaced with NGDP futures prices. Inflation should be replaced with NGDP growth rates, and output should be replaced with employment. The key friction is wage stickiness, not price stickiness, even though there is of course lots of price stickiness out there.

    Here’s just one example. If you rely on inflation rather than NGDP growth, you come up with crazy claims such that higher minimum wages or employer side payroll taxes are expansionary, because they boost inflation. The fallacy is that they don’t boost NGDP growth expectations, which is actually what matters for output. Higher inflation is only expansionary if it is demand side inflation, i.e. caused by higher NGDP growth. But the two policies I just mentioned are supply side inflation, they don’t boost NGDP expectations. To square the circle the NKs then start make crazy claims about the AD curve being upward sloping.

    Take another example, using interest rates leads to all sorts of indeterminacy and neoFisherian nonsense, because lower rates can mean easier or tighter money, depending on context. Is promising zero rates for 20 years a really expansionary policy? Or is it telling the market that we will replicate Japan’s past 20 years? It could be either. NGDP futures prices are a much less ambiguous indicator of the stance of monetary policy, (also less ambiguous than the money supply.)

    Output is also the wrong variable. In 2011-13 you had people like Krugman and Wren-Lewis talking about how austerity was slowing RGDP growth in Britain. But the slow growth was almost entirely due to productivity; employment was doing pretty well. So they confused a supply and a demand shock by focusing on output, not employment.

    I don’t doubt that there are more sophisticated versions of the NK model that can address my objections, I’m just telling you what I see from actual NKs using their model. I’m not going after the platonic ideal of NK models, maybe there is an ideal one out there somewhere. I’m not qualified to say.

    “The result is somewhat dependent upon the utility function, but the effect of a permanent increase in government spending funded by distortionary taxes causes output to increase in RBC (and NK) models.”

    Yes, I understand that the model says that, my point is that this implication of the model is not actually true in the real world, in my view.

    “So you get around the zero lower bound problem by assuming it doesn’t exist? Where are your microfoundations for this? What is your theory of money demand? This seems extremely suspect in the eyes of the Lucas Critique.”

    No, that’s ot what I’m assuming. I warned you that it would take an entire book. There’s lots to be said on this issue. First of all, the cost of holding cash is not zero, and in Europe they are already negative rates. But let’s assume there is a true lower bound at minus 1%. Then there is the issue of longer-term interest rates, which are generally still above zero. But you are missing the much more important issue. You are assuming that the central bank opperates by adjusting the policy rate relative to a Wicksellian natural rate that is beyond their control. I claim the central bank’s more powerful tool is changes in the Wicksellian natural rate. That was also Friedman’s view. He claimed near zero rates were a sign that money has been tight. If you do QE, then that increases inflation expectations, which increases the Wicksellian interest rate. Only if permanent? Yes, but that’s always true of expansionary policy, even when not at the zero bound. Nick Rowe has written on the silly claim people make that assuming permanent QE is not the baseline assumption.

    Now let’s look at this from the opposite perspective, what would base demand look like if NGDP growth were on target? There are two possibilities:

    1. Base demand exceeds the eligible assets that the central bank is allowed to buy. In that case you have a credibility problem at the zero bound. That’s what I meant by “inflation or socialism? You would need to either set a higher inflation target (which reduces base demand) or let the central bank buy more assets than they are currently allowed to buy (say corporate stocks and bonds.)
    That’s where the term “socialism” comes into play. So yes, there is a theoretical zero bound problem, but if you look closely it’s not so much the zero bound on interest rates, it’s the zero bound on assets left to buy.

    2. Base demand when on target. is smaller than eligible assets, in which case there is no zero bound problem. That describes the real world, in my view.

    Here’s where Keynesians often misunderstand me. From point number 1 they assume I am actually recommending that the central bank buys stocks and corporate bonds. I’m not, because they haven’t even come close to exhausting their existing tools, and I very much doubt there would ever be a need to actually go out and buy risky assets, there are plenty of risk-free assets to buy, they haven’t even scratched the surface. Even worse, they haven’t even used the negative interest on reserve tool properly, which can almost eliminate demand for excess reserves, if used properly. Again I get misunderstood, people then think I’m advocating negative IOR, whereas I don’t think even that tool would be necessary if the Fed did level targeting of NGDP futures prices. Demand for base money is simply not very high when NGDP is expected to grow along a 5% trend line. If you promise to do whatever it takes, you’ll end up having to do very little. But if you make that promise, you must be willing to do whatever it takes, if necessary.

    So why did Japan jhave so much trouble for 20 years? These questions make me want to pull my hair out. NKs paid no attention to the fact that the BOJ repeatedly raised interest rates during the long deflation, in 2000 and 2006—this was the “tell” that despite their protests, inflation was about where they wanted it. If you act like you have a zero percent inflation cap, then you’d get zero percent inflation as a cap.

    “Why don’t you think monetary expansion is irrelevant when the opportunity cost of holding money is zero. Is there some other reason why agents would want to economize their money balances?”

    Let’s turn this question around. Why do NKs keep making this claim when we have overwhelming empirical evidence that money matters at the zero bound? Every time a major central bank announces QE, asset prices react as if it’s effective. The day QE1 was announced the dollar fell 6 cents against the euro. Before Abenomics, Krugman predicted that the Japanese would be unable to depreciate the yen, because monetary policy is ineffective at the zero bound. Don’t forget that inability to depreciate is another implication of the monetary ineffectiveness proposition. But the Japanese did succeed in depreciating the yen, from 80 all the way to 120 to the dollar. That’s a lot. Then the NKs insist it wasn’t the money, it was the expectations that did it. But that’s always true, even if not at the zero bound! If interest rates are 10%, and the Fed announces it will double the monetary base, but only for 3 weeks, of course commodity prices won’t move. Why would they? Monetary policy has always been 99% about expectations. That’s a flaw in the old Monetarist model, but no one can claim its a flaw in the market monetarist model, which focuses on market expectations.

    So to answer your question as to why QE works at zero rates, it’s 99% expectations and 1% the fact that base money and 5 year T-notes are not perfect substitutes. If you have negative IOR, then it might be 10% due to currency and T-notes not being perfect substitutes. That’s true whether you are at the zero bound or not.

    You said:

    “The ineptitude of the Fed doesn’t change the facts. Inflation has been below target for some time now and the fact that the Fed somehow doesn’t care is not relevant to the apparent ineffectiveness of QE.”

    I hope by now you can see it is very relevant. QE only matters to the extent it’s expected to be permanent. I published a paper in 1993 pointing out that temporary currency injections are not inflationary, 5 years before Krugman. So I don’t think of his point as being an especially NK observation, it’s just sound economics.

  38. Gravatar of ssumner ssumner
    2. November 2015 at 06:17

    Patrick, Thanks. Within a few days I’ll do a post on that. Very interesting.

    Nick, Yes, I was taught that way back in the late 1970s, by Larry Sjaastad at Chicago. But it’s weird to claim that “output” rises in that case. It’s like, “if the government produces an extra $100 billion in worthless stuff, then output of useful stuff will only fall by $50 billion, hence output rises.” Huh?

    Are we thinking of “output” in crude physical terms, or in terms of stuff that generates utility? Or has value in the market?

    But yes, I agree that that is an implication of many plausible macro models.

    BTW, over at Econlog yesterday I tried to borrow your thermostat example, but sort of botched it in execution, as I admitted at the end. Still I hope people see the general idea. (Not my latest, the one before that.)

  39. Gravatar of Ray Lopez Ray Lopez
    2. November 2015 at 07:15

    Sumner: “Why do NKs keep making this claim when we have overwhelming empirical evidence that money matters at the zero bound? Every time a major central bank announces QE, asset prices react as if it’s effective. The day QE1 was announced the dollar fell 6 cents against the euro.”

    Wow. Sumner cites as evidence for his theory the fact that day-traders responded to QE1. Heck, day-traders sell stocks on rumors that the president has been shot; is this evidence that POTUS’s health is necessary for a bull market? Of course not. It’s just that day-traders are so conditioned by the maxim “don’t fight the Fed; follow the Fed” that they react the way they do. (BTW ‘don’t fight the Fed’ has been shown to be an ineffective way to time the market).

  40. Gravatar of Frustrated Macroeconomist Frustrated Macroeconomist
    2. November 2015 at 12:27

    Scott,

    You’re driving me a little crazy. The whole point of quantitative models is that it commits you to something clear and unambiguous: you can see the advantages when you derive possibilities from the NK model that are absurd. I can’t see how you’re so sure about your own model until you write it down: after all, I would have been quite certain about the implications of the NK model…right up until I was wrong!

  41. Gravatar of John Handley John Handley
    2. November 2015 at 17:45

    Scott,

    I agree with you about the fact that “using interest rates leads to all sorts of indeterminacy and neoFisherian nonsense.” Thinking in terms of interest rates when modeling inflation has meant that we must think in terms of off-equilibrium paths. In a Taylor Rule, the central bank is promising to generate a hyperinflation which can’t happen because of Blanchard-Kahn conditions and, under an interest rate peg, fiscal policy must be either non-ricardian or active to ban all but one equilibrium solution to inflation. This is why there needs to be a good money demand model; we can avoid the ambiguity that interest rates create because of their reliance on both monetary and fiscal policy. We seem to all agree that money causes inflation, but we don’t all agree that we need to have a good model of money demand.

    In any rational expectations model, pegging the nominal interest rate at zero will cause inflation to be low (unless you are doing something interesting with liquidity like Stephen Williamson and David Andolfatto). The only real theoretical disagreement comes when you leave fiscal policy ambiguous so any initial inflation rate is an equilibrium or when you have sticky prices that make it unclear whether or not interest rate increases will take the form of changes in inflation or changes in output (the real interest rate). This is why money is necessary.

    I have no idea where you were coming from when you wrote “[y]ou are assuming that the central bank opperates by adjusting the policy rate relative to a Wicksellian natural rate that is beyond their control.” I though I had made it pretty clear that I think interest rates are not a good policy instrument. In case I hadn’t, I don’t think interest rates are a good policy instrument. Regardless, I don’t understand how the argument you outlined has anything to do with indeterminacy at the zero lower bound. In every plausible model of money demand I’ve seen, zero interest rates mean that further monetary expansion is powerless to cause inflation because agents won’t do anything with the extra money. The empirical evidence (skyrocketing excess reserves, mainly) seems to suggest that money and bonds are perfect substitutes and the lack of inflation confirms that monetary expansion when money and bonds are perfect substitutes is useless.

    Can I have a link to the paper you published in 1993? It sounds intriguing.

  42. Gravatar of John Handley John Handley
    2. November 2015 at 17:58

    Nick and Scott,

    We don’t actually count *useful* or *utility enhancing* GDP in the real world. I’m simply saying that GDP will rise as a result of stimulus. I agree that this stimulus is not a good thing for utility unless the government spending actually has a supply side purpose. Whether or not certain government programs are beneficial to utility, though, is an extremely hypothetical debate that I don’t think would be wise to get into, especially if we’re just discussing economics.

  43. Gravatar of Jose Romeu Robazzi Jose Romeu Robazzi
    3. November 2015 at 03:17

    @John Handley
    I humbly would like to make a tiny comment on an ongoing discussion between people that know a far more economics than I do, I would like to point out my understanding of MM’s view on money demand: money demand may not have a closed form that entirely fits the data. That does not mean one cannot have a rigorous idea of its behavior. One way around this probably is to look at money demand function as a “black box” and look instead at its output. In the MM’s view, it is NGDP. Since we have a pretty good idea about what is going on with money supply, given that and a certain behavior of NGDP growth, one can imply what is going on with money demand.

  44. Gravatar of ssumner ssumner
    3. November 2015 at 12:42

    Frustrated macroeconomist, You are frustrated?!?! How about me? Several years ago in this post I sketched out what a MM model would look like, and asked skilled model builders to construct one for me. I’m not a skilled mathematical model builder.

    https://www.themoneyillusion.com/?p=14826

    No one came through. I can’t do everything, I’m already having enough trouble trying to juggle 10 balls in the air at the same time. When am I going to get some help from the rest of the world? Why don’t you build the model?

    John, Sorry if I misunderstood your view on the Wicksellian rate.

    You said:

    “In any rational expectations model, pegging the nominal interest rate at zero will cause inflation to be low”

    Yes, I agree, but it’s important to note that not every ATTEMPT to peg interest rates at zero will lead to low inflation. Some will lead to high inflation, which leads the central bank to abandon the peg, in order to avoid hyperinflation.

    Here’s my 1993 paper, unfortunately I think it’s gated:

    http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=4133216&fileId=S0022050700012420

    You said:

    “In every plausible model of money demand I’ve seen, zero interest rates mean that further monetary expansion is powerless to cause inflation because agents won’t do anything with the extra money.”

    No, that’s not true. If the monetary injection is expected to be permanent, then expected future NGDP and prices will rise. A rise in expected future NGDP and prices tends to raise current prices, and current NGDP. If nominal wages are sticky, then higher current NGDP tends to increase current output. Even Krugman’s 1998 model says permanent monetary injections are effective.

    You said:

    “The empirical evidence (skyrocketing excess reserves, mainly) seems to suggest that money and bonds are perfect substitutes and the lack of inflation confirms that monetary expansion when money and bonds are perfect substitutes is useless.”

    Sorry, but the empirical evidence doesn’t suggest that at all. Markets consistently respond positively to QE announcements. In addition, it is not correct to view recent real world QE as a highly expansionary monetary policy. It wasn’t, it’s an example of what happens AFTER you have a highly contractionary monetary policy. It was comparable to a small rate cut, that is better than nothing but falls short of the falling Wicksellian rate. The best example of ineffective QE occurred under Herbert Hoover in the spring of 1932. And the reason it was ineffective is very revealing, the gold standard constraint made the monetary injections seem temporary.

    And I still don’t understand why we need to model money demand. If we peg CPI futures prices, or better yet NGDP futures prices, then the market will implicitly forecast money demand. The market forecast is presumably the optimal forecast. What makes us think some sort of structural model of money demand could do better? (My paper that makes this argument is titled “Let 1000 Models Bloom”):

    http://www.degruyter.com/view/j/bejm.2006.6.issue-1/bejm.2006.6.1.1466/bejm.2006.6.1.1466.xml?format=INT

    FWIW, I think nominal money (base) demand is negatively related to the 5 year bond yield, and positively related to NGDP.

    On your second comment, I would point out that RGDP estimates require some sort of hedonics. How else, for example, can someone claim that PC prices are plunging? The government is basically attempting to figure out how much extra “computer utility” is being produced. Otherwise the price indices for PCs are utterly meaningless. RGDP is not number of products, it’s not tons of output, it’s either something loosely related to utility, or it’s a completely meaningless concept.

    And again, on what basis do we decide on government output? We use the cost basis. But we don’t do that for the private sector, because it’s considered unreliable. And this points to what I find so bizarre about the employment effect. If you assume the government produces something that actually makes the public better off, gives them utility, then they don’t feel poorer and they don’t work harder. Sorry, but I find that whole analysis to be very bizarre, even if “the model” tells us that GDP will rise.

  45. Gravatar of John Handley John Handley
    3. November 2015 at 18:28

    “In addition, it is not correct to view recent real world QE as a highly expansionary monetary policy. It wasn’t, it’s an example of what happens AFTER you have a highly contractionary monetary policy. It was comparable to a small rate cut, that is better than nothing but falls short of the falling Wicksellian rate.”

    If this is the case, how do you suppose monetary offset of fiscal contraction works at the zero lower bound?

    “And I still don’t understand why we need to model money demand.”

    It’s not necessarily that I think we need to model money demand, I really just want an economic explanation of how monetary policy can work at the zero lower bound. How does monetary offset work at the zero lower bound? How could the Fed achieve an NGDPLT at the zero lower bound if it tried? To be clear, I don’t want a practical example of a bunch of possible policies. Rather, I want to understand what your ‘microfoundations’ for money demand are. Why do you think agents hold money? Why do you think that policy can be effective when money and bonds are perfect substitutes (assuming they are, regardless of whether or not you think the empirical evidence suggests that)? The usefulness of mathematical models comes into play here. I can use your assumptions/microfoundations to come up with a set of equations and see what the model predicts at the zero lower bound, for example.

  46. Gravatar of John Handley John Handley
    3. November 2015 at 21:37

    Scott,

    Since you mention Krugman’s 1998 paper, I decided to skim through it again, and the model he presents basically suggests that monetary expansion at the ZLB is useless. It’s the size of the future money supply, not the permanence of the QE, that help the economy out of the liquidity trap. Krugman does assume that the liquidity trap is over in the second period of his model, though, which obfuscates the reality of the model.

    More on that here: http://ramblingsofanamateureconomist.blogspot.com/2015/11/monetary-policy-effectiveness-in.html

  47. Gravatar of Blowing Up the New Keynesian Model | A Divided World Blowing Up the New Keynesian Model | A Divided World
    4. November 2015 at 08:32

    […] I very much doubt I have pride of place. What brought this subject to mind again was the post Is it time to blow up the New Keynesian model? by Scott Sumner, a monetarist. Now Sumner and the folks he cites may or may not be hoping to […]

  48. Gravatar of Don Geddis Don Geddis
    4. November 2015 at 16:46

    @John Handley: why do you think it’s an interesting hypothetical to explore, to assume that money and bonds are perfect substitutes? (You want to discount the empirical evidence … but then, why do you want to explore this case in the first place?) It seems the obvious first step would be to say, “no, they aren’t perfect substitutes”, and then build models and predictions from there.

  49. Gravatar of ssumner ssumner
    4. November 2015 at 18:23

    John, If money and bonds are perfect substitutes, forever, then yes, obviously OMOs have no effect. In that case 30-year bond yields are zero, and T-bonds are identical to cash. Surely you aren’t claiming that’s the world we live in?

    We live in a world where the zero bound is not expected to last forever. Bond yields are positive. That means permanent monetary injections lead to higher expected inflation, and lower real interest rates. And we do in fact see lower real interest rates after QE announcements. That’s the real world.

    (Just to be clear, QE is not my preferred policy, level targeting is far superior)

    You said:

    “Since you mention Krugman’s 1998 paper, I decided to skim through it again, and the model he presents basically suggests that monetary expansion at the ZLB is useless. It’s the size of the future money supply, not the permanence of the QE, that help the economy out of the liquidity trap. Krugman does assume that the liquidity trap is over in the second period of his model, though, which obfuscates the reality of the model.”

    Both Nick Rowe and I have been highly critical of that interpretation. Throughout history, the standard monetarist thought experiment has been a one time permanent monetary injection. Krugman is basically saying monetarism no longer holds because his standard definition is a temporary monetary injection. Even at the zero bound, if a central bank injects money with the purpose of boosting prices, then presumably the market will believe it is permanent, and prices will rise. QE failed in cases like Japan 2002-06, where the central bank indicated the injections were temporary.

    Krugman is also wrong in the implication that this is some sort of special characteristic of the zero bound. If interest rates are 5% or 15%, and the Fed suddenly doubles the monetary base, while promising a month later to pull all the extra money out of circulation, it will do almost nothing. It’s not the zero bound that’s key to the 1998 paper, it’s the temporary nature of the currency injections. But why assume they are temporary? Why assume a failed monetary policy? Krugman’s response was the BOJ would not be believed if it promised inflation, until the Abe government proved him wrong.

    BTW, if fiscal policy were judged the way monetary policy is judged in his 1998 paper, it would also fail. Suppose tax cuts were offset by equal future expected tax increases.

    Here’s what Krugamn wrote in 1999, one year after his famous 1998 paper:

    “What continues to amaze me is this: Japan’s current strategy of massive, unsustainable deficit spending in the hopes that this will somehow generate a self-sustained recovery is currently regarded as the orthodox, sensible thing to do – even though it can be justified only by exotic stories about multiple equilibria, the sort of thing you would imagine only a professor could believe. Meanwhile further steps on monetary policy – the sort of thing you would advocate if you believed in a more conventional, boring model, one in which the problem is simply a question of the savings-investment balance – are rejected as dangerously radical and unbecoming of a dignified economy.

    Will somebody please explain this to me?”

    I couldn’t put it better myself. He stopped believing this when he moved to the left after 2001. But the recent data out of Japan actually strongly support his 1999 comments.

  50. Gravatar of John Handley John Handley
    4. November 2015 at 21:50

    Scott,

    For the sake of argument, let’s say a hypothetical central bank has, for some reason, committed to a lower price level in the next period that the current period. If the future price level is sufficiently low compared with the current price level, the zero lower bound will bind. At this point, the entire path of price levels is indeterminate in a Cash-in-Advance model; it doesn’t matter what you do with the path of the money supply, because there is nothing pinning down the price level. There are only two ways to force an equilibrium at this point: either lower the current money supply until the nominal interest rate exceeds zero and the cash-in-advance constraint binds again or switch to a non-ricardian/active fiscal regime until the zero lower bound no longer binds (in case you were wondering, this is the Neo-Fisherian solution). Probably the best policy is to grow the monetary base at whatever rate is consistent with its target (so as to not bathe the fiscal authority in seigniorage revenue from a massive increase in real balances) and have the fiscal authority issue as many bonds as necessary to achieve a non-zero interest rate and then switch back to a normal regime.

    “If interest rates are 5% or 15%, and the Fed suddenly doubles the monetary base, while promising a month later to pull all the extra money out of circulation, it will do almost nothing.”

    Actually, at least in Krugman’s model, interest rates will fall to zero if the central bank tries to do that. The effect of the expansion will be null because the cash-in-advance constraint no longer binds.

    “If money and bonds are perfect substitutes, forever, then yes, obviously OMOs have no effect. In that case 30-year bond yields are zero, and T-bonds are identical to cash. Surely you aren’t claiming that’s the world we live in?”

    As far as I know, the reason 30 year rates are above zero is liquidity and/or risk, not because they are not perfect substitutes with money. Perhaps 30 year bonds (adjusted for liquidity) and 3 month bonds are not perfect substitutes, then perhaps buying long term bonds can have an expansionary effect, maybe.

  51. Gravatar of Weekend Reading: Copious Contemplations | JPPress Weekend Reading: Copious Contemplations | JPPress
    6. November 2015 at 13:52

    […] The Absurd Claims Of Keynesian Economics by Scott Sumner via The Money Illusion […]

  52. Gravatar of Weekend Reading: Copious Contemplations | QuestorSystems.com Weekend Reading: Copious Contemplations | QuestorSystems.com
    6. November 2015 at 14:40

    […] The Absurd Claims Of Keynesian Economics by Scott Sumner via The Money Illusion […]

  53. Gravatar of ssumner ssumner
    6. November 2015 at 18:55

    John, You said:

    “For the sake of argument, let’s say a hypothetical central bank has, for some reason, committed to a lower price level in the next period that the current period. If the future price level is sufficiently low compared with the current price level, the zero lower bound will bind. At this point, the entire path of price levels is indeterminate in a Cash-in-Advance model; it doesn’t matter what you do with the path of the money supply, because there is nothing pinning down the price level. There are only two ways to force an equilibrium at this point: either lower the current money supply until the nominal interest rate exceeds zero and the cash-in-advance constraint binds again or switch to a non-ricardian/active fiscal regime until the zero lower bound no longer binds (in case you were wondering, this is the Neo-Fisherian solution). Probably the best policy is to grow the monetary base at whatever rate is consistent with its target (so as to not bathe the fiscal authority in seigniorage revenue from a massive increase in real balances) and have the fiscal authority issue as many bonds as necessary to achieve a non-zero interest rate and then switch back to a normal regime.”

    You are saying that if a central bank does something incredibly idiotic (promising deflation), there might be all sorts of undesirable side effects that make traditional policy tools ineffective (I’d say ineffective in theory, merely less effective in practice). Then I completely agree.

    But here’s where we disagree:

    1. I conclude they should not do incredibly idiotic things, or stop doing them if they’ve started.

    2. You seem to take the idiotic actions as a given, and look for alternative solutions such as fiscal policy. But why? Why not just stop hitting yourself on the head with a hammer, and adopt a sensible policy target? Stop promising a lower price level.

    You said:

    “Actually, at least in Krugman’s model, interest rates will fall to zero if the central bank tries to do that. The effect of the expansion will be null because the cash-in-advance constraint no longer binds.”

    Yes, but that’s missing the point. The point is that the problem is not the zero bound. In my thought experiment we were not at the zero bound when the Fed began implementing the policy, and it still failed. It doesn’t matter what the interest rate is when the Fed first adopts a new policy, it matters whether the policy is expected to be permanent.

    You said:

    “As far as I know, the reason 30 year rates are above zero is liquidity and/or risk, not because they are not perfect substitutes with money. Perhaps 30 year bonds (adjusted for liquidity) and 3 month bonds are not perfect substitutes, then perhaps buying long term bonds can have an expansionary effect, maybe.”

    Well then we have to disagree. I believe the reason long term bond yields are higher than short term bond yields is that the zero bound is not expected to last forever. Rates are expected to rise. Indeed zero rates are not expected to even last another 2 months!

    The problem here is that Keynesians insist on defining monetary policy as cash for T-bonds. That’s the wrong way to think about monetary policy. Any injection of new cash is monetary policy, regardless of what it’s swapped for. It doesn’t become fiscal policy if the cash is swapped for corporate bonds. But I’ve seen Keynesians call it fiscal policy, which seems bizarre. Monetary policy is about money, it’s not about bonds. If interest rates truly were 0% forever, then yes, the entire national debt would essentially become cash, and obviously monetary policy would no longer involve swapping money for T-bonds, T-bonds WOULD BE MONEY. New cash would be swapped for something else. But again, we don’t live in that world. Maybe the Japanese do, but even they have no trouble depreciating their currency with QE.

  54. Gravatar of Weekend Reading: Copious Contemplations – News Near You | Latest Trending News Weekend Reading: Copious Contemplations - News Near You | Latest Trending News
    7. November 2015 at 01:16

    […] The Absurd Claims Of Keynesian Economics by Scott Sumner via The Money Illusion […]

  55. Gravatar of John Handley John Handley
    7. November 2015 at 23:32

    Scott,

    “You are saying that if a central bank does something incredibly idiotic (promising deflation), there might be all sorts of undesirable side effects that make traditional policy tools ineffective (I’d say ineffective in theory, merely less effective in practice). Then I completely agree.

    But here’s where we disagree:

    1. I conclude they should not do incredibly idiotic things, or stop doing them if they’ve started.

    2. You seem to take the idiotic actions as a given, and look for alternative solutions such as fiscal policy. But why? Why not just stop hitting yourself on the head with a hammer, and adopt a sensible policy target? Stop promising a lower price level.”

    I agree that, if a central bank is following a rule that made this situation frequent, it should stop following that rule. The only point I would disagree here is that the zero lower bound may not be the central bank’s fault. The “IS curve” is Cash-in-Advance models can shift to the left because of changes in the equilibrium real interest rate, which is not something that the central bank has control over.

    Assuming the zero lower bound is actually the central bank’s fault, I do take the actions as given because they actually happened. How are we supposed to analyze real world scenarios if we refuse to model them on the grounds that they require stupid actions by a central bank?

    “The problem here is that Keynesians insist on defining monetary policy as cash for T-bonds.”

    No. That’s just plain wrong. The reason we focus on government bonds is because they are a proxy for all safe assets. They could be houses, corporate bonds, etc. Generally speaking, models suggest that all non-money assets (adjusted for liquidity) are perfect substitutes. We just don’t want to go through the pain of modelling all that. Consequentially, there are only two assets in most models.

  56. Gravatar of Don Geddis Don Geddis
    8. November 2015 at 09:19

    @John Handley: The ZLB on interest rates happened, yes. But your hypothetical was a central bank credibly promising future deflation. That isn’t at all the same thing.

  57. Gravatar of ssumner ssumner
    8. November 2015 at 18:52

    John, You said:

    “How are we supposed to analyze real world scenarios if we refuse to model them on the grounds that they require stupid actions by a central bank?”

    I think you missed my point. I agree we can contemplate that sort of policy mistake. But why rule out reversing the mistake? The solution is to raise the target (preferably NGDP, inflation if you must.) Not fiscal stimulus. Now if you rule those out then you may end up at the zero bound. But in that case I still prefer monetary stimulus over fiscal stimulus, as it is less costly. Even if the central bank has to buy risky assets (which in practice it would not have to do.)

    You said:

    “Generally speaking, models suggest that all non-money assets (adjusted for liquidity) are perfect substitutes. We just don’t want to go through the pain of modeling all that. Consequentially, there are only two assets in most models.”

    I don’t like any model that implies cash is a perfect substitute with houses, even if houses were 100% safe. Assets vary along many dimensions, risk is only one. For instance, we’ve learned that the lower bound is actually well below zero, because it’s hard to store 100s of billions of dollars in currency when just withdrawing a few thousand dollars in cash from your bank account makes you a criminal in the eyes of the US government. Indeed the former Speaker will go to jail for withdrawing $3000 at a time from his own account. Now trying legally handling $100s of billions in currency. We still haven’t found how low IOR can go, but it is certainly well below zero. Cash is toxic in large quantities, big institutions want nothing to do with it.

    And again, if the zero bound is not expected to last forever then QE can be effective (if expected to be permanent) in any model, even where cash and bonds are currently perfect substitutes. Some Keynesians object it only works through signaling, but as Nick Rowe likes to point out, all monetary policy is 99% signaling, even when rates are positive.

    And finally, one argument for NGDP targeting is that the Wicksellian real rate usually falls when RGDP growth falls, and under NGDP targeting you are then allowed an offsetting rise in inflation.

  58. Gravatar of Rob Carter Rob Carter
    10. November 2015 at 01:02

    Nick Rowe shows that the NK model implies that a slowdown in the rate of government spending growth is expansionary.
    To summarize: Rob Carter can agree but only if we don’t comprehend that Keynes himself allowed “Austerity” & “Stimulus” necessities to be sectoral and run simultaneous so the “Austerity” in Waste sectors like War preparations inequality growers like profit directed to business & FED Debt creation Fiat or otherwise must be bad. Then those savings pay the cost of “Stimulating” Sectorally for direct hire from welfare sector for Infrastructure & EPA like safety of bush fire break clearance, pollution control for water & air even plastic cleanup manual projects funded at Local & State government levels utilizing their normally only partially used mainly emergency standby engineers, staffing, machinery etc. Rather than debt growth & fiat printers. As the following means to say short term Keynes patching roofs in the rain for repairing in the post crisis times, really means looking after long-term economics when utilizing such short term remedial theories too far. Or as usually opted for by oligarchy puppets of the plutocrat bullies, bribers and lobby con men, raising debt in the rain without preparing for repayment immediately after the crisis ends. The view that next generation inherit the depreciated assets for far less than they would have to spend next inflated era of higher wage & costs, so let them pay the cheaper debt growth of today as a share of the assets they get FOC.

    I also suggest  Monetary policy drives NGDP, by influencing the supply and demand for base money. Rob says yes Inflation is the Government and their business buddies way of reducing wage rises back to the below necessity rates, specially they oppose cost of living linked wage automatic raises, or inflation auto-linked pay rises, even GDP links fail.

  59. Gravatar of Jeff Jeff
    10. November 2015 at 09:49

    John, here’s my model: Zimbabwe. All of your demands that Scott explicitly delineate a monetary transmission mechanism before you’ll allow him to say that a central bank can create inflation if it wants to fail in the face of my model. My model is hundreds of years of history, which tells us very clearly that it’s very easy to debase the currency. You don’t need a model, you just turn on the printing presses and go to town. If you have a nice measure of expected inflation, you can slow the presses down when expected inflation exceeds your target.

    I know that if I start parking cement trucks on a bridge near here that eventually the bridge will collapse. I don’t know how to write the equations to predict how many trucks it will take, but I’m pretty sure that if I park enough of them out there, maybe stack them with a crane, eventually the bridge will fall. Am I wrong?

    What you’re advocating is paralysis by analysis.

  60. Gravatar of My macroeconomic framework, circa 2015 My macroeconomic framework, circa 2015
    12. November 2015 at 21:45

    […] while ago Scott Sumner laid out at least part of his framework, I thought I should lay out some key parts of mine.  Here […]

  61. Gravatar of The Arthurian The Arthurian
    16. November 2015 at 03:28

    “1. In the short run, employment fluctuations are driven by variations in the NGDP/Wage ratio.”

    If you think of wages as a measure of the price level, you’ll see that the NGDP/Wage ratio gives you a measure of RGDP. If this is correct, then your item #1 simply restates Okun’s law.

    http://newarthurianeconomics.blogspot.com/2015/11/frameworks.html

  62. Gravatar of My Framework My Framework
    8. January 2016 at 15:31

    […] Cowen, Scott Sumner, and Arnold Kling have all given their macro frameworks. I might be a little late to the party, but […]

  63. Gravatar of bbrophy bbrophy
    13. January 2016 at 08:05

    agreed

Leave a Reply