16 Nov 2013

A Policy Proposal That All Economic Schools of Thought Should Prefer to the Status Quo

Austrian School, DeLong, Economics, Krugman, Market Monetarism, Nick Rowe, Scott Sumner 40 Comments

[UPDATES in the middle and then at the end.]

Once again, Nick Rowe has argued that according to the very model that Paul Krugman claims to be using–and I mean here, the full-blown model with intertemporal optimization, not the cruder IS/LM model that Krugman thinks gives surprisingly good policy recommendations, using the full-blown model as the benchmark–it is actually NOT true that the government needs to increase spending today, in order to restore aggregate demand and hence full employment.

On the contrary, Nick argues, the crucial variable in Krugman’s formal model is G(t)/G(t+1), which must increase in order to help the economy today. In other words, what’s important to get people today to spend more, is that the expected growth rate in government spending goes down.

[UPDATE: In a follow-up post, Nick spells out the situation much more clearly. You can see that I’m not misreading him.]

So, notice that a sudden surge today in government spending (which will subside once the economy is fixed) will achieve that goal.

But, Nick’s crucial point is that another way to make that ratio increase, is to today cut the expected level of government spending in the future. That’s a different way to get the growth rate in government spending (gauged today) to decrease.

Can New Keynesians chime in? Is this true?

If so, then we should all be able to agree on a plan for the government to slash spending in, say, 2016 by, say, 10 percent compared to the current baseline level. That will–assuming Nick is correct–put a lot of people back to work today. The Fed will be able to boost NGDP growth even as it scales back its asset purchases. And of course, the size of government will (in a few years) be much lower than it is now.

So Keynesians, market monetarists, and Austrians should all prefer this policy to the status quo. Agreed?

Now let’s get Krugman, DeLong, Sumner, et al. to get on board. The first two guys in particular are the ones who really care about helping the unemployed, and really want to ground policy proposals in formal models.

UPDATE #2: From Brad DeLong’s response to John Cochrane, I’m guessing the way DeLong gets out of this one, is to say he does not believe in New Keynesian models. OK fair enough. But do self-described New Keynesians believe their models? After all, this is a School of Thought, isn’t it? It would be weird if no Austrian economists believed the implications of Austrian business cycle theory, right?

40 Responses to “A Policy Proposal That All Economic Schools of Thought Should Prefer to the Status Quo”

  1. Kevin Donoghue says:

    “…what’s important to get people today to spend more, is that the expected growth rate in government spending goes down.”

    Conversely, to curb spending in an inflationary environment, the expected growth rate in government spending goes up.

    Still keen on the idea?

    • Bob Murphy says:

      Kevin Donoghue wrote:

      Conversely, to curb spending in an inflationary environment, the expected growth rate in government spending goes up.

      Still keen on the idea?

      I would indeed be caught in a pickle, if in the past I had said I believed in New Keynesian models to justify a certain policy recommendation.

      I really hope you have something more substantial to say about this. From your remark, can I conclude that you agree with Nick’s basic point?

      • Kevin Donoghue says:

        No, I don’t agree with him, as my comments on his blog will show. The problem is that solving a New Keynesian model is much harder than Nick is willing to acknowledge. (See Jordi Gali’s book for the gory details.) Nick tries to do it all by juggling with the Euler equation. It simply doesn’t work. That’s just one element in a complicated model.

        There’s a growing literature on optimal fiscal policy rules. I’m not familiar with it, but I’ll be amazed if Nick’s rule ever shows up there.

        • Daniel Kuehn says:

          If this is really all just coming from some musings on the Euler equation in an NK model, then my post no Cochrane is relevant to Nick as well I think.

    • Tel says:

      Conversely, to curb spending in an inflationary environment, the expected growth rate in government spending goes up.

      That does make logical sense if government also raises taxes to fund the spending. Since very few governments can resist the idea that other people are spending their money without government getting dibs on it… I think we can say this part of the system will happen automatically.

    • Matt Tanous says:

      “Conversely, to curb spending in an inflationary environment, the expected growth rate in government spending goes up.”

      Such an environment would never exist in a system where governments are continuously cutting their rate of growth or even shrinking their spending over time. It could only exist if one goes for random bursts of stimulus spending to solve economic woes.

      One might argue that it would if the New Keynesians were right, but even so, the fact remains that we are not arguing that “the New Keynesian model is such-and-such, so therefore…” We are saying “your model says you could get the same results from doing what we propose, so why are you against it….?”

  2. Kevin Donoghue says:

    Your Update #2 is basically right. DeLong’s line is that the NK model was developed as a response to Lucas & Co., to show that even if we grant them (for the sake of argument) Ricardian Equivalence, Rational Expectations etc., all it takes is a litle bit of nominal rigidity to generate somewhat Keynesian results. Obviously that doesn’t mean we have to sincerely sign up to that stuff.

    • Nick Rowe says:

      Kevin: see my updated comment on WCI. If we add hand-to-mouth agents to a NK model (half agens HTM and the other half Ricardian), then we get the following conclusion (I did the math, and I *think* I got it right):

      To raise the natural rate of interest, either cut Gdot or INCREASE Tdot.

      Weird, huh? And VERY different from Old Keynesian recommendations at the ZLB.

  3. Daniel Kuehn says:

    I’m not quite sure what’s going on in that post – I’d have to see it all laid out. I quite distinctly remember – a year and a half ago now – talking about the prospect of fiscal policy (as K & D conceive it) in an NK model. The question is whether monetary policy makes more sense or the monetary policy reaction function, not whether fiscal policy worked. So… I don’t know. Nick’s smart but so if my prof. I don’t know.

    On Cochrane – it was a good post, but here are some thoughts I had had: http://factsandotherstubbornthings.blogspot.com/2013/11/four-thoughts-on-cochrane-on-new-and.html

    • Martin says:

      So Daniel, what you are telling us is in your post is that the models are only true in so far as they get the results we want, otherwise they’re false. For as I read you, the models are merely an illustration and only an illustration for the points the builder of the model wants to make.

      What’s the point of it then? It’s not even a check for consistency or for whether your conclusion follows from your premises, because anything else is ignored.

      I mean if you believe something different from what one model says, such as that people don’t optimize perfectly for psychological reasons and so on (in aggregate) and that people are credit constraint, then why not write a different model that says those things? At least the model as an illustration then matches the content of what is said.

      Note that this is not a point on the realism of the model assumptions, for you can write perfectly good ad hoc model that displays those features, but is just as unrealistic as any other model. No, this is a point about intellectual honesty: how else are people going to discuss with one another if anyone can just turn around and say “I don’t believe what I wrote down to illustrate my point”?

      • Daniel Kuehn says:

        I wouldn’t say that, no.

        But it’s important to know how models are true and how they are false when we draw conclusions from them. See Kevin’s point about about why NK models came about in the first place.

        Things like the Old Keynesian model are good for sketching out the general problem with Say’s Law, determinants of macroeconomic equilibrium, etc. Things like New Keynesian models are good at illustrating how intertemporal behavior and rational expectations tempers those findings. Things like behavioral economics and models with debt and credit rationing (there ARE these models out there – I’m not sure why you’re acting like they’re not being made) show what may happen when RE breaks down somewhat.

        Models are for illustrating mechanisms at work in the economy. It’s fine to reference any particular model, just make sure you’re doing it in the right way.

        Cite IS-LM to talk about how output adjusts to demand shocks, not to show how consumption smoothing impacts the macroeconomy.

        Use plain vanilla NK models to show how consumption smoothing affects things, not to understand how the macroeconomy responds to credit constraints.

        • Martin says:

          See this is what I do not like. A model’s conclusion only follows from the assumptions if the author of the model wants the conclusion to follow from the assumptions. Other conclusions are apparently ignored.

          How are we supposed to have any sensible discussion within macro if not only the assumptions can be chosen at will, but also the conclusions?

          You say for example that:

          “Use plain vanilla NK models to show how consumption smoothing affects things,”

          but when one of the implications of consumption smoothing is raised then you say that that implication is not true because of credit constraints:

          “not to understand how the macroeconomy responds to credit constraints.”

          So how do you know that one conclusion is correct and another conclusion is to be ignored? Except for that the builder of the model tells you that this is so?

          It seems to me that the whole model building endeavor is rather superfluous if you can pick and choose your conclusions.

          How serious should you take any model in macro if other conclusions are ignored? Macro then is merely a body of conclusions rather than a body of arguments.

          • Daniel Kuehn says:

            I think in the real world there is more consumption smoothing than there is in the old Keynesian models. I compare the NK to the OK models to talk about that.

            Never have I committed to the view that perfect consumption smoothing is the real world. So don’t start a sentence with “…but when one of the implications….”

            If ANY implication is dependent SOLELY on the perfectly rational expectations and consumption smoothing I would not accept that implication uncritically.

            I may compare that implication with the other extreme – no consumption smoothing – and note that the real world is somewhere in between.

          • Daniel Kuehn says:

            In other words, you’re trying to pretend that I pick and choose the implications I want. It’s absolutely not what’s going on. You’re spinning it that way because you don’t like the broader view of the operation of the macroeconomy that I adhere to.

            If you don’t like that, fine, but don’t argue against it by accusing me of inconsistency unless you’ve actually got evidence of inconsistency.

            Construct an actual argument Martin.

            • Martin says:

              Daniel, I am not a macro guy, it is not even close to my field. I have no horse in that race and I don’t care. I think Samuelson, Keynes, Friedman, etc are all great and Sargent is today my favorite. What I like and don’t like has nothing to do with how you think the macro economy works.

              What I do not like however is that you can apparently pick and choose one model for one conclusion and another model for another conclusion and exclude whatever conclusion in either if it does not chime with what you are trying to say.

              I am willing to change my mind when the assumptions that go into the model tell me something new about reality. The impression I have of models as illustration however is that what the model says does not matter unless it chimes with what whoever build the model is trying to say.

              When I read Nick or Cochrane, there is apparently some weird stuff going on in the NK model and some counter-intuitive policy conclusions follow from the Euler equation. When I read you it is as if those conclusions do not matter, because they do not chime with what the model is trying to say.

              Why should I take anyone serious who will pick and choose the conclusion he or she wants?

              • Yancey Ward says:

                Martin, you have missed nothing. There are no conclusions from the models that the proponents are incapable of ignoring based on what they want to believe. It really is no different from a religion trying to interpret the biblical texts.

              • Daniel Kuehn says:

                I don’t pick and choose. I’m sick of going on these comment threads and having to deal with accusations like this rather than what you actually think of the macroeconomics.

              • Martin says:

                Daniel, as I said I don’t think anything about how the macro economy works. I’d just like that the people who do think something about how the macro economy works to be consistent and derive their conclusions from the premises of the model they use rather than the other way round and ignore those conclusions that do not jive with what they want the model to say.

        • Major_Freedom says:

          “Things like the Old Keynesian model are good for sketching out the general problem with Say’s Law”

          You mean the Old Keynesian models that are a result of a misunderstanding of Say’s Law.

          The fact that you believe there are indeed “general problems” with it shows that you also misunderstand it.

          • Daniel Kuehn says:

            If you think it’s unfair to Say then we can call it something else.

            I think if you read Say it’s not unfair to him at all (or at least to some things he’s said). But that history of thought question is not the important one.

            • Matt Tanous says:

              If you read Say, and then read Keynes’ “refutation” of Say, the only thing you will be is confused at what it is Keynes thinks Say said.

              But then, the Old Keynesian models are based on the following assumption:

              “Saving is always equal to investment; if it seems like it isn’t, this is just optical illusion. Except for this chapter – in this chapter, we’ll assume everything I said before was utter nonsense. And in the next chapter, we’ll assume this chapter was all nonsense. Eventually, you’ll be so confused, you’ll just accept whatever I say. We could eliminate scarcity in a generation if only those evil capitalists weren’t so keen on having high incomes instead of super abundance. Saving is evil.”

              Personally, I’m all for Hazlitt’s beer multiplier. Have to buy more beer. That’ll keep the economy going.

            • Major_Freedom says:

              Not unfair? Say made it clear that his “law” is grounded on the assumption that each productive line is expanded only so far that no partial over-production and no partial under-production is taking place.

              When Keynes “critiqued” Say, he totally ignored and even contradicted that crucial assumption.

      • Nick Rowe says:

        Martin: “I mean if you believe something different from what one model says, such as that people don’t optimize perfectly for psychological reasons and so on (in aggregate) and that people are credit constraint, then why not write a different model that says those things?”

        Sensible suggestion. Funnily enough, I just did that. I assumed that half the people were rational optimisers, and the other half just always consumed all their current disposable income.

        And the policy recommendation was even weirder: to escape the ZLB the government should cut the growth rate of spending and/or increase the growth rate of taxes.

        It gets even further away from Old Keynesian policy conclusions.

        • Martin says:

          Nick, I just saw it in one of the comments on your post. It’s quite funny. I should ask someone who knows more about this than me to check whether what you do is right or work through the math myself this weekend. I do not see any real objections to what you did in the comments so far.

          Have you considered sending an email to Jordi Gali otherwise?

    • Nick Rowe says:

      Daniel: read my second post. It’s simpler and clearer.

  4. Yancey Ward says:

    “We are all not New Keynesians now.”

  5. Nick Rowe says:


    Start with the Austrian theory of the (natural) rate of interest.

    Delete investment from the model, to keep it simple.

    Add government spending to the model, but assume government spending has no effect on the marginal utility of consumption (to keep it simple).

    So C+G=Y

    Switch from discrete time to continuous time.

    Now suppose you want to increase the natural rate of interest to avoid the Zero Lower Bound. What should you do with government spending?

    You now understand my second post.

    • Major_Freedom says:


      If you start with a natural rate of interest, then you can’t eliminate investment. It presupposes both consumption and investment.

      If investment were zero, then time preference would be maximized. Natural interest rates would actually be undefined (or infinite, depending on the derivation method).

      You wouldn’t be able to “increase the natural interest rate” with your assumption of zero investment. It would be the highest it can possibly get.

  6. Tel says:

    If we use calculus and integrate that New Keynesian picture over time, we can figure out what the New Keynesian picture looks like for the level of G. G would be at a peak at exactly that point between the end of one boom and the beginning of the next recession. And G would be at a trough at exactly that point between the end of one recession and the beginning of the next boom. G would be rising most steeply at the peak of the boom, and falling most steeply at the trough of the recession. The sine-wave for the level of G would be exactly a quarter period out of phase with the original sine wave. I bet nobody has ever drawn that picture for their students either.

    That would appear to require that government can see in advance that a boom or recession is coming… or possibly the government behaviour is what causes the booms and recessions.

    At any rate the peak of G and the trough of G are the critical policy decision-making points. Thus if we have had maybe 3 years since the peak of the boom and G has been rising during those 3 years, some government dude has to say, “Time to turn G around here, and start it falling in preparation for the bust that’s coming in about another 3 years.” The theory and the dynamics may be sound, but I just can’t imagine such a thing ever happening. Government dudes just don’t work that way.

    Nick argues, the crucial variable in Krugman’s formal model is G(t)/G(t+1)

    It would be really nice if economists could follow established notation conventions. I know everyone is sick of me banging on, but using established notation would not only help me, it would also help other people by making it easier to read your work, and you would be helping yourselves by being able to quickly cross reference against well known results.


    That’s the standardized formula for exponential growth, have I posted this link before? Nick Rowe’s formula should look like this:

    G(t) = α . G(t-1)

    The meaning of G(t) is government spending now, while G(t-1) is government spending in the past. We make decisions now based on what has happened in the past. We make estimates about the future, but the future is not here yet. Please, don’t use G(t+1) unless you are talking about predictions.

    You will notice that if you plot an exponential growth formula with the vertical axis set to a log scale, you end up with a straight line. That’s very convenient. Everyone sets their vertical axis to a log scale in economics, right?

    log(α) = d/dt log(G)

    That is to say the first derivative w.r.t time of the log of G as a function of time, is the slope of the graph.

    Growth => log(α) > 0 (graph of G slopes upwards on semi-log scale)

    Decay => log(α) < 0 (graph of G slopes downwards on semi-log scale)

    All nice and sane. Please keep it that way.

    • Nick Rowe says:


      1. Switch to continuous time, so you can forget about all that stuff.

      2. Try to pretend you are a Keynesian, who thinks the government actually knows something.

      • Tel says:

        The advantage of discrete time is you can do it in a spreadsheet. Everyone loves spreadsheets.

        Reformulation as log-linear growth still works the same in continuous time, and I think it is still worth doing when you are primarily dealing with exponential growth and decay type situations. If you like continuous time you might want to convert into s-transform which will definitely be sufficient for an exponential growth model but also linear control theory is done that way. However, not many people know about it, so then you will have a big, “Huh?” factor.

        There are plenty of numerical simulators out there, being Canadian you could talk to Maplesoft, or if you don’t believe that economics is about spending money, you could give Steve Keen’s “Minsky” a go for free:


        I tried it a while back, but it crashed a lot. I think they are still actively working on cleaning it up. Steve’s concept of making it free is that more people will download it, so if you want to demonstrate a model you can do that without making life difficult for the readers.

  7. Gamble says:

    Man this socialism stuff is complicated, I prefer economies as they tend to take care of themselves.

    • Keshav Srinivasan says:

      You don’t think we had any financial crises under free banking?

      • Bob Roddis says:

        No. Actually, no Austrian has ever even thought of that before.


      • Gamble says:

        An economy is not the same as “free banking.” An economy is free people, freely trading. Interventionism is something altogether different.

        Sure and economy will have ups and downs and maybe even a few corrections but nothing like we have experienced. Also, maybe ledger entry’s should not held with such a tight fist? Meaning when you have money in the market, you should not consider the value as real and yours until after you get out. Meaning that if you are in the market or in business, you have to realize you are in a state of flux and you cant always keep everything. It is not over till it is over.

        Instead we have system of cry babies who have privatized their gains and socialized their losses.

      • Matt Tanous says:

        We had free banking? Was that when the colonies were chartering banks? Or was it the Bank of England? Oh, I know,! It was under the National Banking System in the US, right?!

        Psh…. as if banking has ever been free of government interference. You MIGHT get some of that if Bitcoin becomes the general money in use… maybe. As long as government doesn’t find a way to get involved in that too.

  8. skylien says:

    Slightly off-topic, but since we are already at all different schools, I would be interested in how those different schools view/interpret those charts:


    From an Austrian perspective I guess those charts indicate that this recovery is no recovery. Though whatever the others say about the employment chart, I guess they will claim that it means nothing to divide the NYSE index by the Fed’s balance sheet, right?

    • skylien says:

      Thanks in advance for any answer!

    • Gamble says:

      I have never fully grasped the term “recovery.”

      Recover to what? A previous, more ideal state? Who said the previous state was ideal?

      I think recovery is just another catch phrase used to manipulate.

      • Cody S says:

        A “Recovery” is an economics journalism term meaning “Time when left-wing politicians were “*in control.”

        * “In Control” is an economics journalism term meaning “Responsible for only the bad things that happen” during right-wing rule, and “Responsible for only the good things that happen” during left-wing rule.

Leave a Reply