Tuesday, March 19, 2013

John Taylor's austerity model


John Taylor, possibly more than any other economist who writes in the press, likes to simplify things for public consumption. Personally I think this is kind of a shame, since A) any WSJ reader who understands economic models will not understand what Taylor is talking about until they actually go read his research, and B) any WSJ reader who doesn't understand economic models probably already believes that "austerity = good", and doesn't really care if there's a DSGE model backing up the assertion.

Fortunately, your friendly neighborhood Noah is here to read and explain where Taylor is getting his arguments.

In today's column on austerity, Taylor writes:
This week the House of Representatives will vote on its Budget Committee plan, which would bring federal finances into balance by 2023. The plan would do so by gradually slowing the growth in federal spending without raising taxes. 
Still, the plan has been denounced by naysayers who assert that it would harm the economic recovery and that, at the least, any spending reductions should be put off until later. This thinking is just as wrong now as it was in the 1970s... 
According to our research, the spending restraint and balanced-budget parts of the House Budget Committee plan would boost the economy immediately. With the Budget Committee's proposed tax reform included, the immediate impact would be even larger... 
Our assessment is based on a modern macroeconomic model (developed with Volker Wieland of the University of Frankfurt and Maik Wolters of the University of Kiel)[.]
A modern macroeconomic model! Watch out, kid, you could put someone's eye out with that!

(Random note before we get started: The "1970s" reference is pure conservative herp-derpery. People weren't trying to fight stagflation with fiscal policy in the 70s; deficits were quite low. "The 70s" is just a word that conservative writers throw into their pieces so that conservative old men who read the article will nod their snowy heads in sage agreement and mumble "Yes, the 70s. Carter. Stagflation. Mmm-hmm!")

OK, but I digress. Why does Taylor think austerity will produce growth? The article doesn't tell us a lot about the aforementioned "modern macroeconomic model," so let's go to the source. Here it is

Taylor's paper basically uses a Smets-Wouters type DSGE model, i.e. the most popular and successful DSGE model in existence at the moment. It's a New Keynesian model, which means that demand shocks exist and have an effect, through sticky prices, and hence monetary policy matters. So it's not a "freshwater" or RBC-type model (though Taylor et al. do also show that their results hold with an RBC model, somewhat unsurprisingly). 

In other words, Taylor's model is not the type of model that many (including myself) have criticized for being too easily biased toward conservative policy conclusions. Instead, it is a very mainstream type of model, of the kind usually used to justify the Fed's role in managing aggregate demand. Of course, Taylor uses it for something quite different.

So anyway, I'm not going to copy the equations from the paper, because that would just bore you, and you can just read it yourself. However, I'll try to summarize. The upshot of the paper, as Taylor said in his WSJ piece, is that a certain kind of austerity plan would be good for the economy in both the long-run and the short-run.

Here are the basic things you should know about where Taylor gets his results:

1. This is NOT the "Treasury View." 
The "Treasury View" is the simplistic and wrong idea that any dollar spent on "stimulus" has to be one dollar not spent by the private sector, and hence government spending is incapable of raising output. Taylor is not espousing this view. In fact, as we'll see, in Taylor's model changes in government spending most definitely can affect output.

2. This is NOT the "Confidence Fairy."
The "Confidence Fairy" is the idea that uncertainty over future government policy restrains investment, and usually involves the notion that austerity decreases policy uncertainty. However, Taylor's model doesn't have policy uncertainty in it.

 3. The economic boost of austerity comes entirely from tax cuts.
Taylor's model has distortionary taxation in it, and the distortions are large. Hence, spending cuts mean lower taxes and hence less distortion, both now and in the future. In other words, the GDP boost in Taylor's model doesn't come from reducing the deficit, it comes from cutting taxes. The government's long-run budget constraint, along with forward-looking expectations (the same force that powers "Ricardian Equivalence" in other models), means that spending today --> taxes tomorrow --> distortions tomorrow --> distortions today because of forward-looking expectations.

4. The "Sumner Critique" is in this model. 
Normally, New Keynesian models of this type focus on finding the optimal monetary policy. But since Taylor is looking at fiscal policy, he assumes that monetary policy is set according to an unchanging rule.  OK, I'll put in ONE equation. Here's the rule:
This is a backward-looking Taylor rule with interest rate smoothing. Unfortunately, in this working paper I can't find what values Taylor uses for the coefficients on output and inflation. But he does put in some coefficients, that's for sure.

Remember, this is a New Keynesian model with sticky prices, so aggregate demand does matter. But Taylor assumes that the Fed is already doing pretty much everything a New Keynesian would have the Fed do.  So in Taylor's model, the Fed cancels out most aggregate demand shocks, including positive demand shocks from stimulus. So it's hardly surprising that Taylor is not going to see government spending doing much good for the economy; he's assumed that 1) the Fed is perfectly capable of managing aggregate demand, and 2) the Fed will tighten to partially counteract stimulus. This is just a softer version of the common "Sumner Critique" of fiscal policy, though of course Taylor probably didn't get it from Sumner.

5. There is no Zero Lower Bound.
Note that in the monetary policy rule written above, there is nothing that says that interest rates can't go negative. In other words, Taylor assumes what most New Keynesian models assume, which is that there is no Zero Lower Bound (or that we're always far from it). Any of you who are familiar with the New Keynesian DSGE literature will recognize that Taylor's result is a very common result in this literature: Away from the ZLB, fiscal policy is not very effective. The "New Old Keynesians" such as Paul Krugman and Gauti Eggertsson, who advocate fiscal stimulus, explicitly make reference to the ZLB as the reason stimulus works. Taylor just ignores that idea in this paper.

6. In Taylor's model, if you cut government purchases, it throws the economy into a recession.
Taylor's suggested austerity plan makes big spending cuts, but the cuts are almost entirely cuts in transfers (like entitlement payments or welfare) rather than in government purchases (like infrastructure spending). Here's a picture of Taylor's austerity plan:

Now as you should remember from Econ 102, government purchases make much more effective stimulus than transfers, because government doesn't buy the same things that people would buy if you just mailed them checks (but people still receive the checks). So any Keynesian would expect cuts in transfers to be much more benign than cuts in government purchases. In fact, the difference between purchases and transfers is a consistent theme in Taylor's work, which makes me think he's more Keynesian than he makes himself out to be. But anyway, let's take a look at what Taylor's model says would happen if we implemented austerity through reductions in government purchases instead of cuts in transfers:


Wow! In Taylor's model, implementing austerity by cutting government purchases would throw the economy into a deep recession. The reason he gets short-run benefits from spending cuts has everything to do with the fact that it's almost all transfers being cut.

6. In sum, Taylor's result is a standard New Keynesian result.
Upshot: If you have no Zero Lower Bound, and if the Fed partially counteracts the demand-side effects of fiscal policy, and if people have forward-looking expectations, and if you don't cut government purchases much, and if taxes are very distortionary, then austerity works. This is not really a new result, but it rarely gets shown so explicitly, so it's good that John Taylor and his co-authors went ahead and did it.

That said, the result basically ignores the real Keynesian critique that has emerged since 2008, which is that the Zero Lower Bound matters a lot. It also probably assumes that taxes are a bit more distortionary than they really are. And it also probably overestimates the Republicans' real willingness to cut transfers (entitlements are the "third rail", after all), and underestimates their willingness to cut government purchases. In real life, spending cuts usually fall on the things that are politically most easy to cut, but are economically most valuable in both the short and long runs - infrastructure and research. Finally, Taylor's plan ignores distributional concerns, but that's pretty much par for the course.

(Oh, also, neither Taylor's model nor the RBC or Keynesian alternatives includes nonrival government capital (public goods). But that's not the point I'm trying to make here. Oh, and also, modern macroeconomic models - and any other macroeconomic models currently in existence - don't actually come close to describing reality. But that's also not the point I'm trying to make here.)

So John Taylor is not committing some major fallacy. He's just using a standard mainstream New Keynesian DSGE model to stump for the Republicans.


Update: Miles Kimball says that Taylor's result is basically all about the monetary policy reaction function (i.e. the equation I posted above). Furthermore, Kimball notes that Taylor has supported a tighter monetary policy, in direct opposition to the kind of Fed reaction function he assumes in the paper. In other words, if Taylor got the monetary policy he wants, then his own DSGE model would go *poof* and suddenly austerity of the transfer-cutting type would become much more harmful.

Update 2: A commenter points out that in Taylor's simulations, real interest rates never fall below zero. That means there's something odd about the parametrization, since real short rates are negative right now. But I don't know where the oddness comes from.

65 comments:

  1. What exactly is your algorithm for choosing photos on top?

    OK wait I have one question. I get that New Keynesians don't believe in a zero lower bound. But you qualify that with "or that we're always far from it". If it exists, isn't it possible to hit it?

    Is it like absolute zero?

    ReplyDelete
    Replies
    1. What exactly is your algorithm for choosing photos on top?

      Trade secret, I'm afraid. But note that you can use Google reverse image search to find out what each picture is... ;-)

      Delete
    2. You don't recognize John Taylor from his 80s photos, when he was associated with the collective Duran Duran?

      Delete
    3. Some of us were Duran Duran fans, ya know. That was a funny association! :)

      Delete
  2. "So John Taylor is not committing some major fallacy."

    Of course he is. He is ignoring the zero lower bound. That is as major as it comes.

    ReplyDelete
    Replies
    1. Anonymous8:10 AM

      Noah

      In, a nutshell, this is the problem with economists. You let people lie with statistics.

      Taylor has, in effect, put on Hamlet without either the Dismal Dane or most of the cast.

      That's not Shakespeare

      Delete
    2. Well, this is a problem with everyone. Lawyers, politicians, religious leaders, climatologists, people in general often make assumptions that are convenient to the point they are trying to argue. The difference is that economists are forced to write these assumptions down in mathematical form, which allows people who speak math, like Noah, to expose them.

      Delete
  3. I don't want to change the subject, but did John Taylor get that jacket from Michael Jackson?

    ReplyDelete
    Replies
    1. Everyone in the 80s had the same wardrobe person.

      Delete
  4. This post was very helpful to a not macro, economist :-) Thanks!

    ReplyDelete
  5. Simon8:15 AM

    Epic post. Hope that you do more walkthroughs of academic papers in the future for us poor souls who are unable to speak the language of mathematics.

    ReplyDelete
  6. Haven't read the model, but from this nice writeup it's even more amusing that Taylor is so anti-QE and other zero lower bounds monetary policy options.

    ReplyDelete
  7. What do you mean by 'nonrival' public goods? I'm positive Keynes made the construction of public goods a big part of his thinking. And am I the only person who remembers Keynes writing that the single best metric to measure stimulus against is employment: That if whatever you're doing isn't resulting in substantive employment creation, then try something else like hiring directly?

    ReplyDelete
  8. Anonymous9:45 AM

    Damn good post. Thanks.

    ReplyDelete
  9. One would think if there was no zero lower bound for interest rates - this Bernanke FED would have already restored full employment. And you spot on with - 'And it also probably overestimates the Republicans' real willingness to cut transfers (entitlements are the "third rail", after all), and underestimates their willingness to cut government purchases.' Such a reverse Robin Hood stance would be absolute political suicide.

    ReplyDelete
  10. Great post. Konczal brings up a good additional point, is this paper and model internally consistent with Taylor's monetary policy commentary?

    And while he's not committing a fallacy, he's stacking the deck in pretty obvious ways with his poorly grounded Ricardian assumptions and politically ridiculous spending cut assumptions.

    Lastly, from this write up I don't see why the cuts need to happen for short term growth, if one accepts the Sumner critique couldn't the Fed just goose demand without cutting Grandma's medicare?

    ReplyDelete
  11. Anonymous10:28 AM

    Noah:
    The distortionary impact of taxes is a recurrent theme in right-wing harping. Would you please point me at some (preferably publicly available, not NBER of JSTORE) references on what kind of distortionary effects different taxes have and how large they are?
    I'd be grateful for this, and even more so for if you would do summary post yourself. I really need this for a coherent view of the world; I can't be alone.
    TIA.

    ReplyDelete
    Replies
    1. binky bear4:35 PM

      JSTOR now allows non-academic affiliates to view three articles for a period of two weeks.

      Delete
  12. Anonymous10:50 AM

    Best post I've seen yet, Noah. It's really helpful for you to distill this paper into its base assumptions for those of us who don't have the time or requisite background to go wade through the document itself.

    But I think we can all wrap our heads around the fact that Mr. Taylor's model doesn't apply to the real world if it is ignoring the ZLB. Using it as the backup for for a pro Ryan-budget WSJ article seems disingenuous at best.

    I think that all macro papers (especially those used for political posturing!) should come with this type of summary. But then I guess that would defeat the purpose of the paper in the first place. Thanks again.

    ReplyDelete
  13. I second the motion for a column referring to the sources upon which conservatives base their tax views. I can't find anything that isn't either designed to produce the desired answer or is just bad economics. Is their ANY well-designed, carefully sourced, dispassionate paper [JSTOR, wherever] that provides the type of conclusive evidence that could justify conservatives' obsession with tax policies?

    ReplyDelete
  14. I'd question the assumption about purchases always being more stimulatory than transfers in a depressed economy. My intuition is that taking (or borrowing at negative real rates) from wealthy groups with little reason to spend or invest in a depressed economy, and giving to low-income groups with a propensity to spend to fulfil basic needs, e.g. education, housing, energy consumption, business startups could be just as stimulatory (or perhaps even moreso — depends on the quality of the government investment) than direct government purchases. Do we have much empirical evidence about this question?

    Also slightly off topic, but where the hell did the idea come from that austerity decreases policy uncertainty? I don't see any correlation there. I can see the WSJ logic — more spending will equate the implication of higher taxes down the line, leading to "wealth creators" (i.e. WSJ-speak for people with capital) holding off on purchases and investments in fear of future tax hikes (although I don't think there is any empirical evidence for this), but what if austerity does (like Greece, Spain and Italy) actually lead to bigger deficits (De Grauwe's finding that austerity has been self-defeating)? In that case, the austerity itself has led to a greater future necessity for tax hikes. And on the other hand, if a stimulus successfully re-energises an economy and pays for itself through new economic activity, then won't that make "wealth creators" more likely to reinvest?

    The Chicagoans have done some decent empirical work on the effects of economic uncertainty (I wrote about it here: http://azizonomics.com/2013/01/23/camerons-eu-policy-uncertainty/) as measured through word-usage in the media, but the case they seem to be trying to confirm (i.e. austerity = good) doesn't seem to have anything to do with their data, unless they can show a clear-cut association between austerity/stimulus and certainty/uncertainty. That association doesn't seem to be in the data at all...

    ReplyDelete
    Replies
    1. I have a couple comments on your post. First, why label them Chicagoans when two of the three teach at Stanford.

      Second, there study is about uncertainity impinging on near term investment decisions. It has its core in that corporate finance theory would suggest that uncertainty about the outcomes should make capital spending (fixed costs in the near-term) less attractive. That says nothing really about the austerity/stimulus debate, which is about whether in the long run its better to have expanding debt throught larger deficits or more limited debt expansion but less near term growth. I think they might well think a committed, certain stimulus plan or a committed, certain plan for austerity would be better than uncertainty about which path we are on. You could speculate, but I have not seen it from them, that they also believe that austerity leads to more certainty in the long-run because its less reliant on the willingness of the capital markets to lend. I certainly agree with the latter point of view, combined with aggressive and committed monetary policy.

      Delete
  15. Sumner critique is such a BS. How about automatic stabilizers that kick in without anybody doing anything to erase whatever stimulatory/contractinary actions the Fed undertakes? Here is my critique: the monetary policy doesn't work because it is offset by fiscal policy. Sumner critique was shown as a flop in all of the countries that did austerity, how did that work out?

    ReplyDelete
    Replies
    1. You misunderstand the Sumner critique. He argues that IF the central bank targets NGDP or Inflation AND uses everything in its power to achieve those targets (in some countries these are required by law ), that they will make fiscal policy neutral from a demand creation perspective because the greater the fiscal stimulus, the less they will do. He critique has not shown to be a flop because:
      1. No central bank actually targets NGDP.
      2. Those that target inflation have missed their targets to the downside (meaning they should have been doing more stimulus)whether the fiscal situation was in stimulus or austerity mode.
      3. Lastly, he acknowledges that Euro monetary policy as not been near accomodative enough for the countries that are going through austerity because it focuses on inflation on a Euro wide basis.

      Lastly, we all understand auto-stabilizers and why they are the compassionately right thing to have in place. However, if monetary policy can stop NGDP from falling they would be employed less (not never since the composition of NGDP growth is not likely to be even) and cost less.

      Delete
    2. "No central bank actually targets NGDP."

      Israel under Fisher?

      Delete
    3. Pemakin,

      All this assumes the CB can steer NGDP anywhere it chooses. When asked about specific actions Sumner quotes mechanisms that are then shown to him to work the other way round to what he imagines, then he simply falls quiet and does another post on NGDP targeting.

      Again, if fiscal policy has automatic stabilizers it is hard to decide what is offsetting what. You could as well argue that monetary policy is always ineffective when fiscal policy is active enough. It is like saying a cancer drug is ineffective because if it works I can always shoot the patient. Without actually having a gun r bullets or giving a plausible mechanism of how I would actually be able to kill the person. Smoke and mirrors.

      Delete
  16. This is sort of addressed by the Miles Kimball update. But I think it is worth mentioning that Taylor basically wants to have his cake and eat it too. He rails against expansionary monetary policy here: http://online.wsj.com/article/SB10001424127887323375204578267943236658414.html

    But from the article today:

    "Nor does the model account for beneficial changes in monetary policy that could accompany enactment of the budget plan. Lower deficits and national debt would reduce pressure on the Federal Reserve to continue buying long-term Treasury bonds."

    What does that mean? Does he think the Federal Reserve is holding off hyperinflation? As Miles mentioned his model seems to suggest that reducing deficits would require more expansionary monetary policy. According to the model reducing budget deficits should free up the Fed to buy MORE government bonds (or make policy more expansionary another way, i guess)

    He is advocating for tighter monetary policy, while at the same time advocating for tighter fiscal policy based on the assumption that the Fed will loosen its monetary policy. Which, maybe they would, but only if they ignore him!

    Not to rant too long, but the average person reading the WSJ who likes Taylor will think that we should have tighter fiscal AND monetary policy simultaneously. That person will think that because it is the plain english reading of what Taylor writes. That Taylor's model says the opposite--that monetary policy should, nay MUST! offset fiscal policy--suggests a pretty serious degree of intellectual dishonesty.

    Rant aside, this is one of your best posts.

    - South

    ReplyDelete
  17. Noah,

    aftear reading Taylor's paper I think your criticism is misguided. First, Taylor's monetary policy rule refers to the real interest rate, which can in fact be negative (it is right now). Second, if you look at Figure 6, the real interest rate does not fall below zero at any point during the entire period. With positive inflation, this implies a nominal interest rate above zero. So I fail to see how the zero lower bound invalidates the predictions of the model.

    ReplyDelete
    Replies
    1. ??? Even the real interest rate has a lower bound, even if it's below 0, unless you have inflation go up to infinity or something.

      Also, even if the lack of a lower bound doesn't knock out Taylor's model, his non-reality-based assumptions on what the Fed would do, where spending cuts would happen, and the distortionary effects of tax policy at current rates in our current world with our current politics are so off the mark that it's hard not to see it as a purely political hack piece.

      Delete
    2. As far as I know inflation does not have an upper bound. And no, it does not have to go to infinity.

      "non-reality-based assumptions on what the Fed would do"
      Which assumptions are you talking about in particular?

      "the distortionary effects of tax policy at current rates in our current world with our current politics are so off the mark"

      Some references substantiating your claim would be nice here. Or are we supposed to just trust you on this?

      Delete
    3. Costas, I'm pretty sure you're wrong. Here's how Taylor introduces the monetary policy rule:

      "The monetary authority sets the interest rate according to the following Taylor-type rule with interest rate smoothing, where the nominal interest rate responds to deviations of CPI inflation from the inflation target and output growth from steady state output growth"

      (emphasis mine)

      And there's no ZLB there.

      In his simulations he does show real interest rates, but the fact that they don't go below zero just shows that something's odd about his parametrization. Since real rates are now negative, I'd say this should be a bigger concern than it is.

      Delete
    4. Noah, yes, I did not express myself properly. Here is another attempt:

      1) In the model what matters is the real interest rate. This is controlled by changing the nominal interest rate (assuming sticky inflation). However, the question is whether the Fed can influence the inflation rate via other means. The question is whether one believes that at the ZLB monetary policy is ineffective, or that measures like QE are having an impact. If so, one can lower the real rate even if the nominal rate is zero, by raising the inflation rate. There is no agreement on this issue, but all I am saying is that it is a possibility. This is a minor point anyway.

      2) Typically, the Taylor rule involves parameter values that are multiples of 2. So my guess is that he is using as the neutral nominal rate R^4=4% (2% inflation plus 2% real interest rate), phi_pi = 1/2, phi_gy = 1/2. I am not sure what value he he is using for the smoothing parameter phi_R, but in any case, with these parameters the rule predicts a positive nominal rate if one assumes that the targeted inflation is equal to 2% and the targeted growth or GDP is 5%. Basically, the current nominal rate is below what a Taylor rule would suggest.

      3) My main point is that there is no drop in the real interest rate in the simulation. This implies that it is not monetary policy that is absorbing the shock, so I am not sure where Kimbal gets his result. If one reads the sensitivity analysis it is clear that the results are driven by the effect that the reduction in the tax rate is having on employment, and on consumption through the increase in expected disposable income. This is why the results are not as good when the cuts are made in government purchases or if the cuts are met by a reduction in the tax on capital income rather than labor income. Taylor's model clearly favors tax cuts on labor income.

      Now, one can argue that monetary policy should not adhere to the Taylor rule (Greenspan has argued that), that people are not that forward looking, that the political realities are such that cuts will affect purchases also, and so on. These are valid points. However, once again, I do not see how the problem with Taylor's model is that the results are driven by him violating the ZLB in the model!

      Delete
    5. So CA, are you a student? The impression I get is that you don't know much about our current real world very well.

      Anyway, on to your points:

      1. Yes in theory, inflation does not have an upper bound. In the real world (specifically, the US after what would have been a depression with a debt-deflation spiral if not for massive unorthodox political and monetary maneuvers), inflation is very hard to move upwards. Certainly much harder than cutting rates when they are far from zero, so any model that doesn't take in to account the possibility of a lower bound (soft or hard) in rates (real or nominal) is probably not going to model our current reality very well.

      2. Assumptions on Fed actions: I'll concede here and let others tackle this.

      3. Distortionary effects of taxes: I would like to see research to empirically support the distortions that Taylor assumes, considering that effective taxes as a percentage of GDP are as low as they have been in a very long time. Considering that they are a key part of his argument, I would expect him or his defenders to have the onus of coming up with that research, rather than those of us who doubt those distortions.

      4. I notice that you completely ignored a key point I mad: assumptions on where spending cuts would happen. Are you conceding that Taylor is unrealistic about what the GOP would cut if they could pass their budget? Yet if you do, how can you support Taylor's overall argument that the GOP budget would help the economy? After all, does not his own model show that if the GOP gets the cuts that they purportedly want the way that they are likely to want it, the economy would go in to a deep recession?

      Delete
    6. CA:

      The question is whether one believes that at the ZLB monetary policy is ineffective, or that measures like QE are having an impact.

      Isn't this exactly what I meant when I wrote "there is no ZLB" in Taylor's model? Yep.

      Typically, the Taylor rule involves parameter values that are multiples of 2. So my guess is that he is using as the neutral nominal rate R^4=4%

      OK but there is no reason not to give those parameters in the paper...he probably just forgot.

      My main point is that there is no drop in the real interest rate in the simulation. This implies that it is not monetary policy that is absorbing the shock, so I am not sure where Kimbal gets his result.

      Ah, but that could just be down to expectation effects.

      Delete
    7. Dohsan,

      I am an almost 40 years-old-non-virgin, father of two, citizen of two countries and permanent resident of a third...I can keep going on, but no, I am not a student.

      1. "inflation is very hard to move upwards".
      Hard and impossible are not the same. Anyway, again, this is a minor point and not significant in the discussion regarding Taylor's results.

      2. First, you tell me how distortive Taylor is assuming taxes to be, and how distortive you think they are. For the record, for the whole thing to work they need to be distortive to any extent. See also my response to Noah below. Taylor bases his parameter values regarding the elasticity of labor supply on commonly accepted microeconomic studies. Where do you base yours? And, by the way, when he talks in his model about tax cuts on LABOR income he does not necessarly mean high-income earners!

      4. From what I understand Taylor simply put out an idea, substantianted by a scientific paper. I have no clue what Taylor thinks the GOP will do if they get into power, but I assume he wants to push his idea to both parties. I personally supported Obama and have made repeated donations to the DNC congressional campaign committee. But Taylor has earned enough of my respect as an economist that when he puts a proposal out there I need to consider it seriously and evaluate it fairly.

      Delete
    8. Noah,

      on the first point, fair enough, my bad! Still friends?

      On the second point, these values are in his 1993 paper, where he introduced the rule. Maybe he thinks people who are reading it should be familiar with this by now since it is pretty standard, but yes, it would be better if he mentioned them in the paper.

      In terms of the dynamics, this is what I understand from my brief encounter with the paper (I hope I am not wrong).
      People pay taxes that reduce their disposable income. They also receive transfer payments that add to their income. People decide how much of that income, including the transfers, to consume and how much to save given their life-time income. Now suppose that the govt gradually cuts transfer payments thereby reducing the taxes that people expect to pay during their life-time. Since every dollar less received in transfer payments is a dollar less paid in taxes the two should be a wash, having no impact on income or spending. They are not because of the reduced distortion in the labor market. A drop in the labor income tax rate raises the after-tax wage that workers receive and lowers the pre-tax wage that firms pay. In response households are willing to work more and firms are willing to hire more, so overall employment rises as tax-rates go down. Consequently, potential income rises over time. However, actual income rises more than potential because people, to smooth consumption, increase consumption now in anticipation of the future incease in income, so spending (AD) rises faster than AS causing output to over-respond in the short run (the hump in the curve). This is why the Fed does not lower the interest rate, but in fact raises it temporarily. Things work differently when the government cuts government purchases instead, as government spending is used to buy goods and services directly. The increase in current consumption spending in anticipation of the future income increase is not high enough to cover the drop in government spending, so AD falls. With it so does output, which recovers over time.

      Delete
    9. CA: This is not quite my understanding of how it works. Since there are sticky wages and prices, Ricardian Equivalence will not hold in this model even in the absence of distortionary taxation. To see this, take a standard sticky-price New Keynesian model, hold the nominal interest rate constant, and change the timing of transfers; I believe that there will be a nonzero impulse response of GDP and other variables. But put in an optimal Taylor-type rule and these go away, restoring Ricardian Equivalence...I think.

      Delete
    10. Noah,

      why would RE not hold? In the absence of liquidity constraints, what is the intuition behind why sticky prices matter?

      In the model, Taylor assumes that 27% of households are in fact constrained (can smooth consumption only by holding cash). However, the remaining 73% has access to domestic and foreign capital markets. Whether this number is reasonable I cannot say. My guess is that, given the current credit condition, he is underestimating the fraction of liquidity-constrained households.

      Delete
    11. why would RE not hold? In the absence of liquidity constraints, what is the intuition behind why sticky prices matter?

      I don't actually remember why this is the case; I think the argument is that debt financing acts like expansionary monetary policy.

      Also, you're right that in this model, the liquidity-constrained consumers are going to make a difference.

      Delete
  18. Wow! Terrific post. Was directed here from a Krugman post. Now, I'm your newest fan/follower.

    ReplyDelete
  19. There's a lot of ranting about Republican tax policy here but there are two options really: 1) raise income taxes to pay for gov't spending/borrowing or 2) reduce gov't spending/borrowing. I want to be clear that you suggest federal spending is acceptable (or even beneficial) so increasing taxes is a better option? I guess the other argument is that deficits don't really matter as much as some say they do. Still, there has to be some logical extreme to the taxation argument - i.e. you can't tax 100% of income so what is the 'perfect' tax rate based on available models?

    ReplyDelete
    Replies
    1. Redwood Rhiadra3:30 PM

      Research by Peter Diamond and Emmanuel Saez indicates that the optimal tax rate for the top income bracket is roughly 70%.

      Delete
    2. Tom:

      I just wanted to add that while it is true that there are only 2 options in a world with zero growth and zero inflation (and where policies never have an effect on either growth or inflation), in the real world, in a depression economy, deficits actually can be reduced by government spending. Plus, read Krugman on why deficits don't really matter as much as some say they do.

      Delete
    3. Thanks for the replies - Is the 70% rate due to redistributive nature of tax policy, i.e. the top earners would end up sitting on that income if it wasn't taxed and spent by the gov't? I'm sure that top marginal rate A) is tied to an assumption about gov't spending levels / involvement in the economy and B) assumes gov't investments would yield more optimal outcome for the economy than allowing individuals to make those decisions.

      Delete
    4. Tom:

      Diamond & Saez's 70% is more of a novel result than a political reality. The heart of your post was more "what can we actually do now though?!" And I agree with that sentiment.

      To further your & Dohsan's point, there are actually 4 ways to fix the deficit
      1)cut expenses (spending)
      2)raise revenues (taxes)
      3)slow the rate expenses grow (entitlement cost control)
      4)increase the rate revenues grow (GDP growth & inflation)

      So you're actually correct when you say that 1) and 2) are a crapshoot -- neither's a good option! 3) and 4) are actually the easiest and most effective ways to fix the deficit right now.

      Think about it, if we:
      3)allowed Medicare & Medicaid the ability to negotiate drug prices
      4)provided fiscal stimulus in government purchases towards infrastructure

      There would effectively be no deficit problem. That's really what Noah, Krugman, and everybody else here has been saying for 5 years now.

      Delete
  20. CA,

    "With positive inflation, this implies a nominal interest rate above zero."

    Oh-kay. Meanwhile back in our world we *did* hit the ZLB. Four years and counting now.

    "So I fail to see how the zero lower bound invalidates the predictions of the model."

    You're right! The existence of the ZLB in *our* world invalidates no predictions of models of other worlds in which the ZLB is never hit. Taylor's model is really interesting for people who live in *those* worlds. So long as he is not giving advice to us, I guess it's all honky dory.

    ReplyDelete
    Replies
    1. "Oh-kay. Meanwhile back in our world we *did* hit the ZLB. Four years and counting now."

      Come on, use your brains man! Zero lower bound means that the nominal interest rate cannor fall BELOW zero. In the model the nominal interest rate, assuming positive inflation, does in fact not fall below zero. So it does not violate the bound, which is what Noah was claiming. You would embarass yourself less if your not-so-well-thought-out point was not delivered with sarcasm!

      Delete
  21. Anonymous4:16 PM

    I haven't bothered with these sort of DSGE models since grad school. This was a good read for those of us with the background to fully understand everything said (and implied) who don't keep up with the literature and debates in this field as closely as we should. Too much commentary is written too much for the layperson but I really have no desire to read a long dry technical paper in it's entirety either. It's hard to find analysis written for the kind of person who knows what the Smets-Wouters type DSGE model is (and doesn't need much more explanation than that) but doesn't actively keep up with this branch of the literature or the debates that surround it.

    ReplyDelete
  22. Anonymous4:52 PM

    Taylor's Sep 2012 paper grossly misrepresents the actual fiscal situation in 2007, which he uses as a kind of base year for arguing against the need for higher tax rates. He says, "Because the U.S. federal budget was close to balance before the crisis, (the federal deficit was only 1.3 percent of GDP in 2007) this strategy would mitigate the size of any tax rate increase. Hence, relative to the 2012 policy baseline, long-run tax rates would be lower under this alternative strategy." I'm assuming he is referring to calendar year 2007 because the fiscal year deficit was actually 1.2% of GDP according to OMB tables. But in any event, his claim is dishonest because it was only the UNIFIED budget deficit was near balance, and only because there was a surplus of 1.3% in off-budget accounts (i.e., Social Security, etc.) for FY2007. The on-budget deficit was 2.5% for FY2007 at the peak of the business cycle, and it's the on-budget deficit that should be relevant for his analysis since he wants to cut back on transfer programs. In other words, Taylor wants to count the surplus receipts from the transfer programs, but then wants to renege on paying those bonds when the Trustees go to the Treasury for redemption. This is slimy analysis at its worst. The man should not be allowed to practice economics.

    ReplyDelete
  23. Anonymous4:57 PM

    "This week the House of Representatives will vote on its Budget Committee plan, which would bring federal finances into balance by 2023. The plan would do so by gradually slowing the growth in federal spending without raising taxes."

    Ryan's plan calls for tax receipts to increase from the present amount of ~15.4% of GDP to 18.2% of GDP.

    Supposedly we're all going to hand over a smaller percent of our income to the Federal Gov't (no tax increases!), but in aggregate, we're all going to hand over a larger percent of our income to the Federal Gov't. So we each pay less, but when you add it up, it's vastly more. Reminds me of Lake Wogebon where 100% of the people are above average!

    What I'd want to see is how Taylor puts that in his model. Ryan's plan states that the percent of total income that's paid in taxes will increase. In a representative agent model, that would require that the percent of income paid in taxes for that agent would also have to increase. To actually meet Ryan's revenue figures, Taylor would have to model an increase in the tax rate, not a decrease.

    Of course, in the real world, there are many "agents." Ryan's plan can lower tax rates on some as long as the amount paid in taxes increases quite a bit for others (namely the so-called "47%" not to mention those with low, but non-zero effective tax rates). Overall though, Taylor (nor Ryan) has any right to claim there is no tax increase when, mathematically, there must be if you're going to have tax receipts increase by 3% of GDP. Even a "growing pie" argument won't work since Ryan specified not a specific dollar amount, but a percentage. You can't say a smaller slice from a bigger pie can be larger than a large slice from a small pie because he's specifying a specific percent of the pie...a percent larger than what we have now.

    ReplyDelete
  24. John Maddux5:12 PM

    I came to this post trying to understand what the hell Krugman was going on about today. This is a fantastic explanation, and I'm adding your blog to my regular reading list. Thanks!

    ReplyDelete
  25. Noah: a different take: http://worthwhile.typepad.com/worthwhile_canadian_initi/2013/03/fiscal-policy-with-old-and-new-keynesian-is-curves.html

    ReplyDelete
    Replies
    1. I think that this is right, and that Taylor could easily take his model and show that short-term govt spending doesn't really make a big quantitative difference to the effects of the long-run fiscal consolidation. But of course, that's not the rhetorical point he's trying to make.

      Delete
  26. Anonymous8:01 PM

    This is a very nice exposition of Taylor's model.

    I don't think Miles Kimball's critique is quite right though, at least from a technical point of view, because the budget change and the tighter monetary policy Taylor would like to see aren't a single policy change. It's just a question of your counterfactual really. If we're considering the effect of the budget in Taylor's world, then the correct counterfactual is an increase in interest rates and no change in fiscal policy, rather than unchanged monetary and unchanged fiscal policy.

    He argues that monetary policy should be a lot tighter at the moment, which I think in the model means a change in the monetary rule. It's not clear to me that this model offers a sensible way to assess such a change, but in any case Taylor thinks that in the real world it would be expansionary. We would then be away from the zero-lower bound, so the Fed would have room to offset changes in fiscal policy (although they wouldn't necessarily have ENOUGH room). Thus in his model the Fed would react to consolidation as described, and it could be expansionary.

    So while I don't agree with Taylor's description of the world or policy prescriptions, I think it is still a coherent world view.

    ReplyDelete
  27. CA,

    "Come on, use your brains man!"

    Alright, here goes...

    "So it does not violate the bound, which is what Noah was claiming."

    a) Where? He said: "Taylor assumes what most New Keynesian models assume, which is that there is no Zero Lower Bound (or that we're always far from it)". That is exactly correct. Where did he say that Taylor lets the nominal rate go negative?

    The point that Noah is making is that Taylor assumes the CB is following a *Taylor* rule which they can only do if there is no Zero Lower Bound or "we're always far from it". That is a pretty specious modelling choice given the amount of Taylor Ruling going on right now, right? He engineered his world so he could just pretend the ZLB was irrelevant because the nominal natural rate just happens never to go below zero. You think he didn't notice that all the predictions of his model blow up in his face if the natural rate happened to be at the same level as in the real world? Did he just randomly choose nonsense parameters? Even with all his hyper-distortionary taxes, he couldn't make it work with a zero bound! So he shifted the natural rate up.

    b) Anyways, I was merely responding to you saying this in your first comment: "So I fail to see how the zero lower bound invalidates the predictions of the model."

    *That* statement is a total red herring. Taylor is very busy using his models to make predictions about *our world*, not about some obscure hypothetical world. But it just doesn't apply. There are a lot of modelling choices that reasonable people can disagree about. But choosing the wrong interest rate, and a central bank behavior that's impossible to execute? Seriously?

    ReplyDelete
    Replies
    1. K, please see my response to Noah in a thread above.

      Delete
  28. CA,

    Well if Taylor's result survives the liquidity trap, he *should* have shown us. And why would he have gone out of his way to avoid the ZLB in that case? Why is he not trumpeting an article titled "Austerity as an Escape From the Liquidity Trap"? *That* would have been worth shouting about!

    ReplyDelete
    Replies
    1. K,

      "And why would he have gone out of his way to avoid the ZLB"

      Because it is not relevant to his results. Why make the model more complex by adding a detail that is not important to your results? Read my reply to Noah above about where I see his results coming from.

      Delete
  29. More to the point, Taylor's model implicitly assumes that if real interest rates ever did drop below zero, people wouldn't buy T-bills. But they do.

    You have to wonder what this says about mathematical economic models in general. The assumption appears to be that people who buy T-bills seek to maximize their utility function -- whereas Kahneman & Tvserky's results suggest that people actually seek to disproportionately minimize their perceived future losses, even if it means increasing their real present losses.


    ReplyDelete