It’s what they don’t say that is revealing

I suppose by now I should be used to Paul Krugman responding to arguments I never made:

Here the savings curve shifts up from S1 to S2; the upward shift of the curve at a given level of GDP is what Wren-Lewis and I mean by the “increase in savings”. In the end, of course, savings must equal investment, which is why GDP must fall from Y1 to Y2. But Wren-Lewis and I haven’t forgotten that S=I; we’re just using the perfectly ordinary method and language of comparative statics.

And Sumner thinks we have made some kind of basic error, and ties himself into knots trying to set us straight.

Here’s the argument I actually made:

Those readers who agree with Brad DeLong’s assertion that Krugman is never wrong must be scratching their heads.  He would never endorse such a simple error.  Perhaps investment was implicitly assumed fixed; after all, it is sometimes treated as being autonomous in the Keynesian model.  So maybe C fell by $20 million and investment was unchanged.  Yeah, that could happen, but in that case private after-tax income fell by only $20 million and there was no consumption smoothing at all.  Checkmate.

I think Wren-Lewis and Krugman both made the same mistake as those angry Austrians that complained I was trying to goose up “spending”.  They momentarily forgot that saving is not money down a rat hole, but rather represents spending on capital goods.  They were so overconfident that anti-Keynesian arguments are bogus that they momentarily let down their guard, and employed a bogus argument themselves.

Of course it’s very possible that investment would not fall, it’s very possible that AD would rise by the amount of the spending on the bridge.  But you can’t get from here to there by employing a consumption smoothing argument against Cochrane.  After all, consumption smoothing is a part of the classical model, and the multiplier is zero in many versions of the classical model.  (Not all, there are classical models where wasteful expenditure makes the public poorer, and hence they work harder-Larry Sjaastad did a model like that way back in the 1970s.)

Now ask yourself if Krugman did justice to my argument.  Do you notice that I mentioned that investment might be fixed, just as he showed me in his “correction” of my error?  So I clearly understand that, it’s not the issue.  I never denied that the basic Keynesian model implies that an attempt to save less will simply raise income and insure a balanced budget multiplier of one, indeed I bent over backwards to say that would have been a possibility in the third paragraph quoted above.  But that’s not the argument Wren-Lewis made and Krugman endorsed.  Their argument was that John Cochrane was wrong because of consumption smoothing.  But consumption smoothing plays no role in the Keynesian balanced budget multiplier because after-tax income doesn’t change.  For some reason Krugman still doesn’t get this point, perhaps because he just skimmed my post and assumed I was making some sort of elementary error.  After all, Krugman has already told everyone that all the Chicago school economists are ignorant of Keynesian economics 101, so why should I be any different?

If Krugman were right that consumption smoothing somehow refuted Cochrane’s argument, then it would be impossible for consumers to react to a $100 million dollar fall in after-tax income as follows:

Spending on consumer goods falls by $20 million.

Spending on new homes falls by $80 million.  Or spending on inventory accumulation falls by $80 million.

Now I’m not saying that would happen, but if it did it would validate Cochrane’s claim and yet would incorporate consumption smoothing.  So consumption smoothing can’t be the issue; it plays no role in whether Cochrane is right or wrong.  Those who think it does are implicitly treating a fall in consumption as equivalent to a fall in spending.  That’s why I assumed Wren-Lewis was thinking that way; it was the only logical explanation for him to even mention consumption smoothing.  But if consumption smoothing is actually going on, then a fall in consumption is not even close to being equal to the fall in total spending.  Thus they are forced to fall back on the argument that the basic Keynesian model is true because we assume it’s true.  Fine, then say so.  Don’t say Cochrane is wrong because he ignores consumption smoothing.   Cochrane isn’t wrong (if he’s wrong) because of consumption smoothing, but rather because velocity might change.  If the MPC was 0.99 and there was no consumption smoothing at all, he’d be equally wrong using the simple Keynesian model.  The problem isn’t that the Keynesian model shows Cochrane is wrong, but rather that it shows he’s wrong even w/o consumption smoothing.

Krugman ends his post with this classy remark:

It’s really very sad.

By the way, the phrase “consumption smoothing” appears five times in my post, three times in italics for God’s sake.  Here’s a question for all you Krugman fans.  I made it very clear in my post that I wasn’t attacking the Keynesian model, just Wren-Lewis’s use of the “consumption smoothing” argument.  That was the only issue at stake.  Now take a look at Krugman’s post and count the number of times he uses the phrase “consumption smoothing” in response to my post devoted to consumption smoothing and almost nothing else.  What does that tell you?  Then go to Mr. Wren-Lewis’s newest post on the subject and count the number of times he mentions “consumption smoothing.”  Notice it’s exactly equal to the number of times Mr. Krugman mentions consumption smoothing.  What do you think that tells us about whether I was right about consumption smoothing?

I have a simple question for Mr. Krugman.  Does he or does he not think that “consumption smoothing” is a valid argument against Cochrane’s claim?  If so, can he provided a specific reason why consumption smoothing leads to positive balanced budget multiplier?

Update: It just occurred to me that Krugman doesn’t provide a single quotation from the post he was criticizing.  Perhaps because he couldn’t find one that didn’t mention CONSUMPTION SMOOTHING.


Tags:

 
 
 

85 Responses to “It’s what they don’t say that is revealing”

  1. Gravatar of Karl Smith Karl Smith
    15. January 2012 at 19:57

    And, how many times did you use the term redhead or auburn in this post?

    That tells me I was right.

  2. Gravatar of RN RN
    15. January 2012 at 19:57

    Krugman toasted you.

    As I read your argument I too scratched my head trying to puzzle why it felt wrong.

    Paul, as always, nails it.

    You’re out of your league.

  3. Gravatar of edeast edeast
    15. January 2012 at 20:06

    Concern troll self-deleted concerned comment.

  4. Gravatar of D R D R
    15. January 2012 at 20:21

    Argh.

    When Cochrane makes an argument involving assumptions which one can accept or reject, we can argue over whether his model is right or wrong.

    But Cochrane argued that he had proven a thing, and we have no way to follow his logic so long as his assumptions remain totally obscure.

    For now, we can only provide examples which lead to different conclusions.

    What on earth more can anyone expect?

  5. Gravatar of Mark A. Sadowski Mark A. Sadowski
    15. January 2012 at 20:30

    Scott,
    May I say you are not typical of Chicago economists (that’s because I was there).

    As you said:

    “After all, Krugman has already told everyone that all the Chicago school economists are ignorant of Keynesian economics 101, so why should I be any different?”

    Krugman was using you as a fake foil.

    I know not all Chicago economists think that way becuase *I was there*.

    But he finds you a useful adversary even if that shouldn’t be the case.

    Beating you on the head does not serve our final need which is increasing AD.

    Thus I’m not sure why Krugman has indulged in this exercise.

  6. Gravatar of math math
    15. January 2012 at 20:31

    Here is an argument for “consumption smoothing.”

    Assume, government investment has caused the growth by 20% in the year 1.

    That investment will be paid for through taxes over period of 4 years. And these taxes will cause the decrease in growth by 20/4=5% over each year.

    If all taxes were collected in year 1, the move would have been neutral: increase by 20% is equal to decrease by 20%.

    But because increase happens in year 1 but decreases happen over period of 4 years, there is a growth (since government investment works during all 4 years and possible even longer).

  7. Gravatar of Brito Brito
    15. January 2012 at 20:42

    I haven’t really been following this discussion, I only commented on one of the posts because you gave an explanation as to why savings equals investment, which I disagree with, but I do agree that ultimately S = I, so it’s trivial.

    So I don’t have time to read through all the posts. However, it seems to me that you can use consumption smoothing to argue for a balanced budget multiplier, wikipedia actually has the algebraic proof: http://en.wikipedia.org/wiki/Balanced_budget#Balanced_budget_multiplier

    It also says: “Since only part of the money taken away from households would have actually been used in the economy, the change in consumption expenditure will be smaller than the change in taxes. Therefore the money which would have been saved by households is instead injected into the economy, itself becoming part of the multiplier process.”

    In other words, some of the incidence of tax falls on savings, as there is a core component of consumption (c0) that cannot be reduced or is inelastic to tax. To me that is the same thing as consumption smoothing.

  8. Gravatar of Anthony T Anthony T
    15. January 2012 at 20:43

    Tsk tsk. So sad.

  9. Gravatar of Jim Jim
    15. January 2012 at 21:04

    I thought the crux of stimulus was that taxes would rise *later* and so the increase in AD is not permanent but a matter of shifting later spending to current spending that helps alleviate a slump by allowing faster adjustment out of debt overhangs on constrained consumers… Not sure what this current argument is about.

  10. Gravatar of Noah Smith Noah Smith
    15. January 2012 at 21:35

    Not to be a rude jerk, but…I still do not get why people are spending effort on this whole argument.

  11. Gravatar of TylerG TylerG
    15. January 2012 at 21:48

    It’s depressing that Wren-Lewis is treating consumption and spending as the same thing in his consumption smoothing criticism. It’s depressing that there are some people (see commenter RN) who views these exchanges as more akin of a football game -*Go Krugman Go!*- than meaningful intellectual exchanges. It’s depressing that Krugman et al. treat legitimate criticisms as either ‘ignorant’, ‘evil’, ‘very sad’, etc. It’s depressing that rather than asking his viewership to think critically, Paul Krugman happily reassures the ranks of economically illiterate hyper-partisans of their simplistic worldviews.

    I’m depressed.

  12. Gravatar of D R D R
    15. January 2012 at 22:23

    Someone on the internet is wrong!

    More seriously, though. I am actually trying to understand Sumner’s objections. I think that in failing to understand where he was going, I may have wasted a lot of time trying to elicit irrelevant details. But it has been very helpful to me when he has actually answered directly.

  13. Gravatar of Andy Harless Andy Harless
    15. January 2012 at 22:31

    If you do it with rational expectations with a representative agent, there is no consumption smoothing on the equilibrium path, but rational expectations equilibria depend on what happens off the equilibrium path. The existence of the equilibrium depends on the fact that the representative agent is a consumption smoother. To see this, imagine that the representative agent were not a consumption smoother (e.g. assume she is liquidity constrained). In that case the equilibrium is different, and the effect of the tax increase on total output does offset the effect of the government spending increase, and in that case disposable income falls. But if you make the agent a consumption smoother, this result disappears.

    It’s more straightforward to just use either adaptive expectations or heterogeneous agents (only some of which receive income from the new government spending), in which case actual consumption smoothing does occur either at certain times or by certain agents.

    It strikes me as just a cheap debating trick to say that Wren-Lewis is wrong because no actual (aggregate, equilibrium) consumption smoothing occurs in his version of events. His result still depends on the fact that agents are consumption smoothers.

    Now Wren-Lewis is implicitly assuming that conditions are such that the multiplier is positive, which depends on (1) price or wage rigidity and (2) a monetary policy reaction function that does not fully offset the effect of fiscal policy on nominal income (e.g. a binding zero constraint on the interest rate and a central bank that is unwilling to use unconventional policies on a sufficient scale). You can challenge these implicit assumptions, or you can argue that Cochrane is implicitly denying one or both of these assumptions (though they don’t seem to me to be inconsistent with the words that Cochrane used, except with respect to his conclusion). So Wren-Lewis’ premises may be wrong (or not), but I still don’t see any mistake in his argument.

  14. Gravatar of fausto412 fausto412
    15. January 2012 at 22:33

    Reading your original post, you do make whats seems like pretty good statements and in my view Krugman isn’t distorting your argument, what you show here as your argument is an incomplete picture of what you really wrote.

    You pass judgement and then you analyze. You only show here the analysis part but not the part where you say they were both wrong. That doesn’t change the fact that you passed judgement and wrote:
    “I guess I shouldn’t have been surprised. For three years conservatives have been coming over here and sneering that I was just trying to force people to spend more. Each time I patiently pointed out that I want more spending, but not more consumption.

    Nevertheless, the last few days have been a real eye-opener. The comment sections contained a torrent of attacks against my novel S=I claim. Eventually people started quoted textbooks by Krugman and Mankiw, which also claimed it was an identity. But there was still a lot of grumbling, as there always is when you try to throw tautologies at people in an argument. If it’s just a tautology, what difference can it make? In this post I’d like to explain why it’s essential to think of saving as spending on capital goods, not as setting money aside. And I’ll also show that even Nobel Prize winners can make elementary errors by forgetting this identity. I’m going to rehash the Wren-Lewis/Krugman mistake, because I don’t think I explained it effectively in this post. Here’s how Simon Wren-Lewis criticized Bob Lucas and John Cochrane’s claim that when the government builds a bridge using tax money, the balanced budget multiplier might be zero:

    Both make the same simple error. If you spend X at time t to build a bridge, aggregate demand increases by X at time t. If you raise taxes by X at time t, consumers will smooth this effect over time, so their spending at time t will fall by much less than X. Put the two together and aggregate demand rises.

    But surely very clever people cannot make simple errors of this kind?

    Then Paul Krugman cited the Wren-Lewis quotation with approval:

    Yes indeed. A lot of people have been clutching their pearls over my (and Brad’s) simple statement that Lucas/Cochrane made simple, fail-an-undergrad-quiz-level, errors. To say that they did what they did is a “rant.” But the truth is that they did “” and have refused to admit those errors, which is worse.

    Undergraduate quiz? He’s going to bitterly regret that snide remark.

    First recall that C + I + G = AD = GDP = gross income in a closed economy. Because the problem involves a tax-financed increase in G, we can assume that any changes in after-tax income and C + I are identical. Checkmate in four.

    Suppose that because of consumption smoothing, any reduction in after-tax income causes C to fall by 20% of the fall in after-tax income. Then by definition saving must fall by 80% of the decline in after-tax income. So far nothing controversial; just basic national income accounting. Checkmate in three.

    Now let’s suppose the tax-financed bridge cost $100 million. If taxes reduced disposable income by $100 million, then Wren-Lewis is arguing that consumption would only fall by $20 million; the rest of the fall in after-tax income would show up as less saving. I agree. Checkmate in two.

    But Wren-Lewis seems to forget that saving is the same thing as spending on capital goods. Thus the public might spend $20 million less on consumer goods and $80 million less on new houses. In that case private aggregate demand falls by exactly the same amount as G increases, even though we saw exactly the sort of consumption smoothing that Wren-Lewis assumed. Checkmate in one.”

    I think it helps you lazy readers if you quote your full post and not the section that lets you off the hook. Some of us click and read the source because that is how you build credibility.

    “Update: It just occurred to me that Krugman doesn’t provide a single quotation from the post he was criticizing. Perhaps because he couldn’t find one that didn’t mention CONSUMPTION SMOOTHING.” He gave us the link to go see what you said.

    Like I stated above, your post says a lot of things. Perhaps someone should proof read your post to make sure people get out of it what you intended to get out. Krugman has Wells helping him with his writing.

  15. Gravatar of D R D R
    15. January 2012 at 23:00

    Andy,

    It is worse than just a cheap trick, because Wren-Lewis is still correct even if there is actual consumption smoothing at every stage– including in the overall balanced budget case.

    Start with Wren-Lewis’ spending multiplier of 1.0,

    C+20
    I-20
    G+100
    T+0
    Y+100
    Yd+100

    Then add in the tax hike, and we can still wind up with actual consumption smoothing and a positive multiplier.

    C-10
    I-40
    G+100
    T+100
    Y+50
    Yd-50

    This implies that for the tax hike alone, consumption is still smoothed.

    C-30
    I-20
    G+0
    T+100
    Y-50
    Yd-150

    Now, maybe Wren-Lewis doesn’t believe that the tax multiplier is only 0.5 as I assumed here. But my assumption of 0.5 is just as valid as Sumner’s assumption of 1.0.

  16. Gravatar of SB SB
    15. January 2012 at 23:01

    TylerG believe it or not there are some Krugman fans that try to understand the points of the other side. It seems like this might almost be a semantic argument between monetary and fiscal stimulus. Krugman is a calling a government loan to build a bridge fiscal stimulus. Is Cochrane saying once a loan is involved it should really be called a monetary stimulus? If you agree that getting a loan to build a bridge can be called fiscal stimulus then I am not sure how you can argue there is always no multiplier. You get to use the bridge now but you are paying for it later. There is a multiplier if the bridge provides any value. Even if you build a useless bridge to nowhere the people building the bridge have money now and it’s possible inflation alone would provide a multiplier. To the layman it seems the bridge loan has to increase demand. Possibly someone can explain in layman’s terms why this is incorrect. [And if you seriously believe being worried about taxes tomorrow affects spending today then my prediction is you are not married… or just have no idea how the majority of Americans’ finances work.]

  17. Gravatar of Liberal Roman Liberal Roman
    15. January 2012 at 23:32

    It’s amazing how many problems S=I causes for both Keyenesians and Austrians.

    Bryan Caplan recently wrote a post about a common Austrian gripe: that increasing GDP doesn’t mean much because increasing GDP could just be increased government spending. He points out that building the proverbial “bridge to nowhere” increases G and hence increases GDP.

    But if he really believed in his very own arguments, he would say that spending on a bridge has no effect on GDP since it just transfers resources that would be used otherwise into building a bridge. This is of course assuming a central bank running a proper monetary policy which does not incentiveze hoarding cash.

  18. Gravatar of TravisA TravisA
    16. January 2012 at 00:11

    @Andy: I think it is Krugman and Wren-Lewis who are engaging in cheap debating tricks. Wren-Lewis presented an argument that Cochrane is obviously, categorically wrong in all states of the world except when spending is permanent. But Wren-Lewis’ argument is actually saying that the BBM is 1, but he simplifies and disguises it to such an extent that it seems that Cochrane is making a simple math error. But if Wren-Lewis had said, “Cochrane is an idiot because as everyone knows, the BBM is 1”, that wouldn’t have had the rhetorical punch of his actual post and would have been much easier to dismiss.

    Scott is just calling out Wren-Lewis on this form of argumentation.

  19. Gravatar of BW BW
    16. January 2012 at 00:17

    Scott, you’re wrong. You’re committing the classic mistake of confusing an accounting identity for a description of human behavior. Wren-Lewis was trying to tell you that in his Q&A, and you didn’t get the hint there. Maybe it needs to be more explicit.

    [As an aside: Krugman’s argument does not depend on I being fixed. He drew a graph with flat I, but the same story holds if I has a non-zero slope. Try it. You’ll see. The fact that you focus on this aspect of the graph as some sort of vindication of your argument shows that you missed the point.]

    Here’s the mistake you make: you assert in your original post that “If taxes reduced disposable income by $100 million, then Wren-Lewis is arguing that consumption would only fall by $20 million; the rest of the fall in after-tax income would show up as less saving. . . But Wren-Lewis seems to forget that saving is the same thing as spending on capital goods.” Note that I quoted you, since you seem sensitive about such things.

    Your mistake is to assume that saving is the same thing as spending on capital goods. That’s only an accounting identity, which is to say that it doesn’t operate instanteously. Something else has to happen for spending on capital goods to come into equipoise with savings. That something, in a recession, is a rise in GDP. The bridge project intermediates unused, idle cash into productive investment, which raises GDP, which raises savings. As savings are diverted into the bridge, capital spending may fall because there is less capital to finance it — but these processes are taking place at the same time. The new equilibrium will be less than an 80% reduction in savings, less than an 80% reduction in capital spending, and higher GDP. Note that I have not relied on any assumption that I is fixed.

    Consumption smoothing plays a role in this argument because it is what allows the tax increase to play out in the savings channel — where the mediation of unused savings into increased economic activity can increase GDP — instead of the consumption channel, where Cochrane assumes it will.

    By assuming that savings and capital spending are both indepedent of GDP, you’re replicating the same mistake as Cochrane, just in the money market instead of the consumption market. You’re in essence asserting a priori that GDP cannot rise without an assumption of fixed I, when in fact Krugman and Wren-Lewis are demonstrating that GDP can in fact rise without any assumptions about I at all. All one needs to assume is that an economy can be undercapacity because money is unused.

    Why Krugman and others are having trouble understanding what you are talking about is you are 1) purporting to abandon the simplistic Cochrane assumption, but really 2) relocating that assumption to a different spot in the model and then 3) as a result, you think consumption smoothing is no longer relevant because you happen to be looking at a different part of the model. But consumption smoothing is still there — it’s the concept that is forcing you out of Cochrane-land and into your mistaken S=I land.

    I think you should give it up. Each post of yours is becoming successive more obscure and more detail-obsessive (really, you’re complaining that Krugman doesn’t use the WORDS consumption smoothing, as if the concept can’t be conveyed in different language?), which is a sign that you are getting yourself tied up in knots defending something that you know isn’t quite right.

    Now, I will give you this: you are getting yourself tied up because, unlike Cochrane, you are trying to wrestle with what Wren-Lewis and Krugman are talking about. That’s commendable. But the problem is that you still aren’t grasping essential aspects of the model.

    P.S. Krugman’s coup de grace, “very sad” was unnecessary and needlessly condescending and I don’t blame you for being a bit irritated by it.

  20. Gravatar of D R D R
    16. January 2012 at 00:31

    Travis. You have it backwards. I will dismiss Cochrane’s proof as follows:

    1) Cochrane says tax-financed government expenditures result in no increase in overall spending.

    2) I say taxing folks $100 million and spending it on a bridge has no effect on consumption or investment, so GDP rises by $100 million.

    Please tell me where my assumption contradicts Cochrane’s. You cannot, because he did not actually address this case in his “theorem.”

  21. Gravatar of Roger Roger
    16. January 2012 at 00:43

    Your agents(households) have double personality:

    1)G +100
    2)Oh, I feel poor because of future taxes!
    3)I’m going to lower consumption C -20,(increase saving S=+20)
    4) And you know I’m going to decrease investment! I=-80
    5) Oh… Wait… S=I! In fact I wanted to increase Savings!

    Tax increase in the future means that consumers will consume less and save less, not AFTER income has ajusted but BEFORE it does…

  22. Gravatar of Robert Robert
    16. January 2012 at 01:33

    A thought experiment: if the government ran all businesses and banks, wouldn’t there still be a GDP? Wouldn’t we still be able to eat, get paid, save and spend? In fact, a lot of major businesses in China are entirely run by the government, but the Chinese seem to be doing alright.

  23. Gravatar of Artturi B Artturi B
    16. January 2012 at 01:47

    @D R:

    That was not the argument that Wren-Lewis made. Hade he made that argument there wouldn’t have been any problem.

    Wren-Lewis made the argument that because of consumption smoothing Cochrane has to be wrong, when in fact he can be right or wrong with consumption smoothing. Thus consumption smoothing can’t be the sufficient condition for Cochrane beign wrong.

  24. Gravatar of Amien Johaadien Amien Johaadien
    16. January 2012 at 02:17

    Firstly thanks for making a contribution to this “discussion” which adds something rather than just pumping more hot air into this debate.

    I actually can see where you’re coming from in what you say, and I can see that there is merit in your arguments.

    Its interesting though, that you seem to be arguing that Cochrane is wrong, although not for the reason Krugman et al provide (eg consumption smoothing isn’t the reason why fiscal policy has a non-zero multiplier), and in your original argument you go on to describe a few situations where fiscal stimulus doesn’t work.

    However you also make some strange arguments. You say two things, which both rely on the same incorrect assumption:

    “First recall that C + I + G = AD = GDP = gross income in a closed economy. Because the problem involves a tax-financed increase in G, we can assume that any changes in after-tax income and C + I are identical. Checkmate in four.”

    “Suppose that because of consumption smoothing, any reduction in after-tax income causes C to fall by 20% of the fall in after-tax income. Then by definition saving must fall by 80% of the decline in after-tax income. So far nothing controversial; just basic national income accounting. Checkmate in three.”

    These arguments, while technically correct, make the strange assumption that *GDP remains fixed*! Obviously, GDP is not fixed. This is why G can increase by 100 while C only falls by 20 and I stays fixed. You seem to touch on this in the next paragraph:

    “So maybe C fell by $20 million and investment was unchanged. Yeah, that could happen, but in that case private after-tax income fell by only $20 million and there was no consumption smoothing at all. Checkmate.”

    This argument left me scratching my head. Consumption smoothing must have happened, otherwise GDP would remain fixed. (As you say, I would have to fall by 80, and GDP at t+1 would be the same as GDP at t.) Am I misunderstanding the meaning of consumption smoothing? If C faces a shock of -100 (from the increase in G), but it only falls 20, then consumption has been smoothed?

    You did however make a good point. Krugman didn’t use the most direct explanation possible, (which would be the your argument combined with my argument about non-static GDP levels). Instead he made a jump to the obvious conclusion we reach, that consumption smoothing is required for the fiscal multiplier to be non-zero (either positive or negative, if you imagine that high G induces consumers to save more than they would owe).

    Finally, thanks for blogging. You’re part of a very welcome revival of the great tradition of discussion and (sometimes heated) argument in our society, and its a well worthy cause!

  25. Gravatar of Martin Martin
    16. January 2012 at 02:34

    Scott,

    If I may summarize what I’ve seen so far:

    Lucas & Cochrane: A therefore C.
    Krugman & Wren-Lewis: Whaat? No. It’s C’. Your A =/= our A’. Stupid, stupid for not saying C’!
    Sumner: Euuh, Lucas & Cochrane, said A not A’ and A does not result in C’.
    Krugman: We’re not wrong. Sumner can’t do / does not understand comparative statics. See C’ follows from A’.

    It seems to me that it is not about going from A to C or from A’ to C’, but about asserting the truth of C’ for Krugman & Wren-Lewis.

  26. Gravatar of Kevin Donoghue Kevin Donoghue
    16. January 2012 at 02:35

    Krugman: “Well, I read Scott Sumner’s claim that Simon Wren-Lewis and I are making basic errors, and at first I couldn’t figure out at all what his problem is. Then it struck me….”

    Okay Scott, so you don’t agree with his diagnosis. And actually I think Andy Harless has done a better job of figuring you out. But let’s face it, you don’t make it easy. It’s very hard indeed to make sense of your argument.

  27. Gravatar of Martin Martin
    16. January 2012 at 02:43

    Also, I could be uncharitable and say that Krugman and Wren-Lewis do not understand what assumption drives the result in their models. Something one could think from the switch in emphasis from consumption smoothing to fixed investment. But I would not want to accuse them of not understanding their own models and hold them up as examples of a dark age in macro.

  28. Gravatar of Robert Robert
    16. January 2012 at 02:57

    Isn’t this all just a misunderstanding over plans versus outcomes?

  29. Gravatar of progrowth liberal progrowth liberal
    16. January 2012 at 03:36

    Why not just say it! You are arguing complete crowding-out as if we were at full employment today. And even then – please say it – the mechanism requires a rise in real interest rates as I aruge here:

    econospeak.blogspot.com/2012/01/scott-sumner-v-paul-krugman-on-simple.html

    And I thought this point had already been settled in the econoblogsphere world!

  30. Gravatar of Walt Walt
    16. January 2012 at 03:50

    The best I can figure out here is that Scott is taking this stand out of some sort of tribal loyalty. Krugman has his grudges — he hates “the Chicago school”, by which he means a somewhat idiosyncratic subset of what people would identify as the Chicago school — and since Scott feels lumped in with the group he feels he needs to speak up. It’s disappointing, but understandable. It would help if Krugman was more specific in his attacks, rather launching into attacks on vague categories like “macroeconomists” or “the Chicago school”.

  31. Gravatar of Adam (ak) Adam (ak)
    16. January 2012 at 03:50

    I think that D R got it right and prof Sumner got it wrong however prof Krugman did not address the main issue. I am not an economist but an electronic engineer who has read some Kalecki if that matters.

    I want to attack the main “claim that when the government builds a bridge using tax money, the balanced budget multiplier might be zero”. I am not interested in other points of the debate.

    The obvious precondition for a balanced stimulus to work is that the economy must have greater than or equal to zero spare productive capacities even after an increase in aggregate demand caused by the stimulus otherwise inflation will rise.

    I would argue that the balanced spending multiplier has to be close to 1 (most likely greater than 1 but this needs to be shown separately).

    I am merely going to restate the arguments used already by D R :

    “1) Cochrane says tax-financed government expenditures result in no increase in overall spending.

    2) I say taxing folks $100 million and spending it on a bridge has no effect on consumption or investment, so GDP rises by $100 million.”

    The key error in prof Sumner’s reasoning has been made in my opinion very early, in steps 1 and 2:

    “First recall that C + I + G = AD = GDP = gross income in a closed economy. Because the problem involves a tax-financed increase in G, we can assume that any changes in after-tax income and C + I are identical. Checkmate in four.
    Suppose that because of consumption smoothing, any reduction in after-tax income causes C to fall by 20% of the fall in after-tax income. Then by definition saving must fall by 80% of the decline in after-tax income. So far nothing controversial; just basic national income accounting. Checkmate in three.”

    Does the government need to increase taxes BEFORE spending as a precondition of increased spending or does the government spend more money first and then drain increased taxation flow AFTER that increased volume of money flows through the productive economy from firms to households as wages and profits?

    Prof Sumner’s proof is incorrect because it has been assumed that taxes drain the flow of funds available for private sector consumption spending and saving before that flow is used to purchase products and then is increased again by government spending.

    The private sector can respond in obvious ways to the reduction of revenue caused by an increase in taxation. This is the key argument in steps 3 and 4. However that reduction will never occur because points 1 and 2 are not correct.

    I am not talking about increasing budget deficit (a flow) as the stimulus is supposed to be balanced. Let’s acknowledge that the Treasury can borrow or create money (later that money will be repaid or destroyed). How does the stimulus actually work? The government decides to keep spending extra Delta_G dollars per period for example on infrastructure projects. That flow is added to pre-existing flow of aggregate demand (C + I + G) and we get AD1 = C + I + G + Delta_G. Aggregate supply is obviously equal to aggregate demand. The economy has to produce more thus increasing revenue of the private sector. The government can then drain Delta_T = Delta_G more dollars per period thus keeping the budget deficit unchanged. AD is thus increased by Delta_G and the multiplier is 1. We don’t need to worry about any consumption smoothing or rational expectations at this point.

    Let me repeat again. The government does not need to say that we’re increasing taxes for year 2011 in order to spend more in 2012. The government spends more in 2012 and will collect higher taxes over that period of time and at the end of the new period. Assuming that it succeeds, there will be no increase in the deficit but there may be an one-off increase in the quantity of money in circulation.

    I can draw a simple diagram showing an extra flow added flowing as Delta_G from the government to the firms, then as an increased revenue to households where it is drained as Delta_T – superimposed with the usual circular flow of funds but I obviously have no time to do that tonight.

    There is absolutely no reason to assume that the private agents will respond to an increase of their revenue by Delta_G followed by an increase in taxes by Delta_T by decreasing spending. The assumption made in the “proof” that the revenue is decreased first is absolutely wrong if we imagine how the monetary circuit works. It is very plausible that private sector will respond to increase in aggregate demand in an opposite way, by an increase in investment, thus making the multiplier greater than 1. But even if this is not the case, at least new infrastructure will be built and spare resources (unemployed people) will be hired.

    Obviously there can be some redistribution of the income due to changes in taxation but this is a separate issue.

    If one wants to understand better what I have written – just let’s introduce the time line in the circular flow of funds and goods to discover the causality – exactly like Michal Kalecki did. In a monetary economy spending (including investment) comes first and can be an exogenous variable (if financed by credit or money creation) – only then a revenue in any form can be reaped. Profits and wages are endogenous.

    Yes I know that what I wrote is a heresy for a neoclassical economist as everything is supposed to be in an equilibrium at once but this is what Keynes and Kalecki discovered (Karl Marx was the first to mention these issues). What I wrote invalidates the way of thinking taught on neoclassical Econ 101 courses but this is what I learned on circuit theory when I studied Electronics 101.

    NB loanable funds theory also has to go for exactly the same reason but let’s leave this topic for now.

    Reference:
    Michal Kalecki
    Theory of Economic Dynamics, 1952
    pages 45 onwards.

    I can give more references if required.

  32. Gravatar of acarraro acarraro
    16. January 2012 at 04:12

    What about this toy model?

    I live for three periods and I can work to produce apples or rest and get leisure. Apples last forever. I cannot work in the last period, but I value consumption equally. Let’s say that the initial optimal behaviour is to produce 60 apples in the first 2 periods and consume 40 each period.

    In period 2 the state comes and takes 10 apples from me. Given that I value consumption equally I will move some consuption from the future to today. So I consume 40, 35, 35. The goverment eats the 10 apples. So the consumption profile is 40, 55, 35.
    But my production profile has not changed (it’s still 60, 60, 0). So saving has changed to balance things out…

    But an optimizing agent would likely re-calculate the leisure-work trade off. If apples and leisure are complements, I would actually increase apple production to get to a new optimal equilibrium. So leisure-work smoothing gives a multiplier… Is that reasonable?

    It all rests on the complements assumption obviously…

    I think taxes are bad is mostly based on the leisure-apples are substitutes assumption (where taxation increases the price of apples and so you work less and buy more leisure). It doesn’t really work in my lump sum taxes world though…

    That’s actually an interesting twist I think. It implies that a tax based on past behaviour would have a very different effect on one based on future work, since I cannot really change my behaviour to dodge it…

  33. Gravatar of Gray, Germany Gray, Germany
    16. January 2012 at 04:12

    Afaics you’re shifting the goalposts here. Actually, despite what you pretend now, your story started this way:
    “In this post I’d like to explain why it’s essential to think of saving as spending on capital goods, not as setting money aside. And I’ll also show that even Nobel Prize winners can make elementary errors by forgetting this identity.”

    And Krugman quite convincingly showed that the didn’t forget “this identity”, even though it may have looked that way for those who don’t know about comparative statics.

    But now you argue it wasn’t your point at all to accuse Krugman of forgetting about S=I, and instead you wanted to make a minor point about consumption smoothing that ultimately makes no difference to the Wren/Lewis/Krugman argument at all (if I understand that correctly). Sorry, but that smells foul. IF that was your intention you should have said so at the start of your first story, and not make it sound as if you had convicted Prof K of making a very basic mistake.

    So, sorry, but if we readers shall believe your second story, we can only conclude that you totally screwed the first one up. You have fallen into your own hole and really should stop digging now.

  34. Gravatar of Daniel Kuehn Daniel Kuehn
    16. January 2012 at 04:13

    You made several arguments and I think he was responding to all that frustration you had convinced he wasn’t aware that S=I.

  35. Gravatar of RonT RonT
    16. January 2012 at 06:01

    Both sides would do well to learn Stock-Flow-Consistent models of MMT which analytically track any flows in the economy. You cannot miss a cent. No need to bend oneself in knots not to miss a flow into some other sector like we see everybody doing here.

    Have a look here: full employment full time, debt payments under control. True Checkmate.
    http://www.levyinstitute.org/files/download.php?file=wp_494.pdf

    SFC models helped predict the 2008 crisis too.
    http://www.voxeu.org/index.php?q=node/4035

  36. Gravatar of Bill Woolsey Bill Woolsey
    16. January 2012 at 06:28

    Consumption smoothing is relevant because otherwise Ricardian Equivalence implies that debt financed government spending has no impact on the natural interest rate.

    In other words, with consumption smoothing, debt financed government spending reduces _desired_ national saving despite Ricardian equivalence. It takes a higher real interest rate to make realized national saving (evaluated at full employment output and income) equal to investment–spending on capital goods.

    With debt financed tax cuts, on the other hand, Ricardian Equivalance suggests no increase in the natural interest rate because there is no decrease in desired national saving. Desired private saving rises an amount equal to the added budget deficit. With consumption smoothing, it is certainly true that the right time to increase saving is when taxes are cut.

    Of course, you all down to balanced budget multipliers, so Ricardian Equivalance doesn’t even matter.

    With consumption smoothing, then a temporary tax hike to fund a goverment project results in less private saving and a higher natural interest rate.

    Without consumption smoothing, then the argument is weaker. Of course, if saving is a fixed proportion of disposable income (like really simple Keynesian models,) then higher taxes reduce saving and raise the natural interest rate.

    Anyway, Cochrane’s sin was to appeal to Ricardian equivalance in passing, while failing to explain that there is a difference between the impact of debt financed government spending and debt financed tax cuts.

    In the end, Cochrane does explain how debt financed government spending could raise nominal expenditure on output. In my view, he was more or less correct, focusing on the monetary conditions.

    I am not sure that discussing the identity between realized saving and investment helps much.

    It would be like discussing the impact of a price ceiling by notng that the amount firms sell is always equal to the amount households buy. What do you mean “shortage?”

    Or, more directly related to my concerns, it is like responding to concerns about an excess supply of money–the quantity of money being greater than the demand to hold money, by explaining that all money is always held by someone. What surplus?

  37. Gravatar of Claudia Claudia
    16. January 2012 at 06:31

    I have not been following this latest grumpy review of econ concepts via blogs…I will add a few points about consumption smoothing. First, the main idea of “consumption smoothing” is that household don’t adjust their consumption much *temporary* changes in their financial resources, that is they *smooth* through these blips. (BTW, because of durable goods, like cars, consumption can stay fairly constant even with noticable changes in expenditures.) For example, if a household’s take home pay drops due to a loss of overtime, they would borrow or draw down savings to keep their consumption fairly constant (smoothing) *if* they expect to get their overtime back soon. (If they see the loss of overtime as a permanemt downshift in their earnings potential…such as, the first sign of impending layoff..they should cut back on consumption and not smooth. I would argue that many of the shocks to households lately have been more permanent than transitory.) Alternatively, if households get a one time stimulus payment from the government they won’t raise their consumption much since that bump to income is temporary. I would have thought consumption smoothing would work against not for the multiplier from temporary fiscal policy. But having read this one post, I have no desire to read the full back and forth. In addition to a lack of civility, one of my pet peeves on the blogs is economists throwing around economics jargon to make themselves sound intelligent. I dont care about word counts. It’s been helpful to me to see concepts rehashed, but I kind of know what is being talking about. I thought the blogs were supposed to open economics up to non economists. Not trying to pick on this blog…it’s widespread and I do it too.

  38. Gravatar of ssumner ssumner
    16. January 2012 at 06:47

    Karl, I admit it.

    RN, That’s right, whatever Paul says must be right.

    DR, Consumption smoothing has nothing to do with whether Cochrane is wrong. As far as I know all zero multiplier models feature consumption smoothing.

    Mark, I don’t see why Krugman finds it so hard to just admit he was wrong about consumption smoothing. It’s not that big a deal–I’ve been wrong about lots of things.

    Math, Sure growth might have occurred, but so what?

    Brito, The regular multiplier does depend on the MPC, but positively. The balanced budget multiplier is what we are discussing, and it’s 1.0 in the Keynesian model, regardless of the MPC.

    Jim and Noah Smith, The argument is about the argument. Wren-Lewis and Krugman didn’t say Cochrane was stupid because he was wrong, they said he was stupid because he used a poor argument. Obviously Cochrane might be right about the multiplier being zero (I thing it is), but I agree with them that he used a fallacious argument.

    I’m showing they are just as guilty of using a fallacious argument. People who live in glass houses shouldn’t throw stones.

    I’m not saying their conclusion (positive multiplier) was wrong–it might be right.

    TylerG, Thanks. I’m so happy I have commenters like you who actually understand the argument, not people who just blindly accept anything Krugman says.

    Andy, I do take your argument seriously, but I still think you are wrong, for several reasons. First, I plan a new post showing that all zero multiplier models I know of feature consumption smoothing, so it is not a valid criticism of Cochrane. Second, as far as I know the balanced budget multiplier is 1.0 regardless of whether the MPC is 0.0001 or 0.99999. I don’t think any reasonable person would say both of those numbers are “consumption smoothing.” Consumption smoothing is the sort of thing that occurs in a permanent income model, isn’t it? I don’t see how consumption smoothing gets you anywhere, even in the Keynesian model. But I’ll defer to your expertise on that issue. However I’m quite sure that consumption smoothing in no way refutes Cochrane’s argument.

    DR, You simply assumed the multiplier was 0.5, and then assumed consumption smoothing. Obviously that’s possible. But you didn’t rely on consumption smoothing to get your change in income, did you? You could have had an MPC of .90. You are basically redoing Wren-Lewis’s argument as follows:

    “I’m going to simply assume Cochrane is wrong, and then see if I can assume he was wrong in a model that features consumption smoothing.”

    No reasonable person would interpret Wren-Lewis’s actual argument that way.

    more to come . . .

  39. Gravatar of ssumner ssumner
    16. January 2012 at 07:30

    fausto42, Perhaps Krugman should read posts a bit more carefully before trashing them.

    SB, The whole debate is about the balanced budget multiplier.

    Thanks Liberal Roman.

    BW, You said;

    “Wren-Lewis was trying to tell you that in his Q&A, and you didn’t get the hint there. Maybe it needs to be more explicit.”

    It would be easier if he didn’t start off saying saving doesn’t equal investment, and then right after say saving does equal investment. Seriously, I do understand everything he says. I get the Keynesian assumption about attempts to save causing income to fall rather than investment to rise. Indeed I think I understood that as a college freshman. That’s not the issue being discussed. We are discussing consumption smoothing.

    You said;

    “Your mistake is to assume that saving is the same thing as spending on capital goods. That’s only an accounting identity, which is to say that it doesn’t operate instanteously.”

    If something is an identity, it does happen instantly. I think what you are trying to say is that initially the form “investment” takes is capital accumulation in the form of bigger inventories.

    DR, You said;

    “2) I say taxing folks $100 million and spending it on a bridge has no effect on consumption or investment, so GDP rises by $100 million.”

    If only Wren-Lewis had said that! We wouldn’t have this fruitless debate.

    Roger, This whole debate has nothing to do with future taxes, we are discussing the balanced budget multiplier.

    Arturri B, Exactly.

    Amien, You said;

    “These arguments, while technically correct, make the strange assumption that *GDP remains fixed*!”

    Good point, let me explain why. I argued Wren-Lewis was wrong either way. The most sensible way of reading his criticism is that he claimed Cochrane’s claim was inconsistent with consumption smoothing. I.e. even if after tax income did fall by 100, spending would only fall by 20, so Cochrane would be wrong even on his own assumptions. That’s what I thought Wren-Lewis meant. But I covered my bases by pointing out that even if that’s not what he meant, the consumption smoothing argument had no bearing on whether the Keynesian model was correct.

    I don’t think Krugman assumed consumption smoothing was necessary for the Keynesian result, indeed in the balanced budget multiplier of one, there is no consumption smoothing because after-tax income doesn’t change at all. Even worse, the size of the balanced budget multiplier is the same whether the MPC is 0.1 or 0.9.

    Martin, I’m weak in symbolic logic, but that may be right.

    Kevin, I’ll have a new post replying to Andy, who has raised the only argument that I find half way sensible.

    Martin, I think Krugman and Wren-Lewis do understand their models, which is why they didn’t mention consumption smoothing once in their replies to me.

    Cochrane also understands these models, if you read the long paper Krugman criticized a few years ago. He’s just sloppy in his arguments.

    Robert, No.

    Progrowth liberal, You do know that I favor stimulus, don’t you?

    Walt, I didn’t defend Cochrane, I criticized him.

    Adam, First you say you aren’t going to deal with my argument, but rather the broader BBM equals one claim. Fine. Then you attack me as if I’m claiming the BBM is not one. But I didn’t make that claim.

    acarraro, I’m not a fan of toy models.

    Gray, If Krugman didn’t make that confusion, why doesn’t he tell us what he was thinking when he endorsed the phony consumption smoothing argument? What was his rationale, if not confusion? He never addresses that point. I took the most likely interpretation of the argument, that they were claiming that Cochrane was wrong because consumption smoothing would occur. But Cochrane would only seem to be wrong due to consumption smoothing if you forgot about saving equaling investment.

    Daniel, Obviously he knows S=I, but he made an argument that only makes sense if it isn’t.

    RonT, That’s off topic.

    Claudia, Yes, a lower MPC makes the regular Keynesian multiplier smaller. It has no effect on the balanced budget multiplier.

  40. Gravatar of Joe2 Joe2
    16. January 2012 at 07:36

    I can’t believe Krugman is actually saying that.

    He’s so intellectually dishonest that even the Times first public editor Dan Okrent spotted it and warned people about him.

  41. Gravatar of Major_Freedom Major_Freedom
    16. January 2012 at 07:37

    Sumner:

    “So consumption smoothing can’t be the issue; it plays no role in whether Cochrane is right or wrong. Those who think it does are implicitly treating a fall in consumption as equivalent to a fall in spending.”

    A million times this. This false equivalence, made in this case by Wren-Lewis, is as pernicious in economics as a poisoned water supply is in a city.

    Not only is it wrong in itself, but it also leads to all sorts of confusions. Just witness Krugman failing to understand Sumner’s argument about consumption smoothing. Even though Krugman SEES the argument right before his eyes, he can’t understand it because his mind is warped by believing the false equivalence.

    Not consuming is saving, and since not consuming is allegedly not spending, it means saving is not spending too, and since not spending in the aggregate generates an aggregate economic correction, we are supposed to fear and hate savings, and thus promote and encourage consumption.

    Nick Rowe makes a good case for banning the concept “saving” from economic discourse entirely:

    http://worthwhile.typepad.com/worthwhile_canadian_initi/2012/01/why-saving-should-be-banned.html

  42. Gravatar of Sesh Sesh
    16. January 2012 at 08:02

    Your argument started with:

    “Because the problem involves a tax-financed increase in G, we can assume that any changes in after-tax income and C + I are identical. ”

    As far as I can tell, that means you have assumed that GDP is constant (i.e., you have assumed that the change in (C+I) = – change in G). You then go on to prove – in four steps – that, therefore, GDP is constant.

    That’s all very nice, but why did you need four steps to prove something you had already assumed to be true? Consumption smoothing seems irrelevant to this tautology.

    The simple fact that none of the equations you wrote had any time-dependence might have alerted you to the fact that you could not possibly be refuting Wren-Lewis’s argument.

  43. Gravatar of Benjamin Cole Benjamin Cole
    16. January 2012 at 08:02

    Sumner is right by my lights, but I would prefer we bring Krugman on board by building a bridge to his admission that tight money can suffocate fiscal policy (even if one does not believe fiscal policy works, or if monetarily accommodated is just Market Monetarism in drag).

    In short, let’s make Krugman into an ally, not an opponent.

    Although if Krugman is “against” us, then we probably get allies on the American right by knee-jerk reflex.

  44. Gravatar of tom tom
    16. January 2012 at 08:05

    “But in that case isn’t Wren-Lewis claiming that both saving and new inventory accumulation have to fall by 80? In that case Cochrane is right.”

    Cochrane is wrong and Simon is right. Fall in inventories has positive effect on spending (some firms receive cash from selling off their stock and others want to spend on it so it generates spending) so if you add government spending plus decrease in consumption (less than government spending) the overall effect is positive.

  45. Gravatar of Morgan Warstler Morgan Warstler
    16. January 2012 at 08:27

    SB,

    “To the layman it seems the bridge loan has to increase demand.”

    No, in layman’s terms not having the tax money later negates the value of the bridge – BECAUSE of who is making the decision about what to spend/invest money on/in. Solyndra is ALWAYS bad. 100% of the time.

    If you don’t start with an acceptance of the Knowledge Problem you:

    1. haven’t dealt with private industry enough.
    2. haven’t dealt with govt. capital spending enough.

    Dufuss, the Internet, Tang, and Velcro always exist without gvt. spending. So do roads, police, etc.

    Gvt. stumbles into power because early in its existence without bureaucracy it is far more likely to be effective.

    Anyone who denies the multiplier does it with the correct assumption: the govt. is far less competent than private individuals.

    They call them the special olympics for a reason, they say the same thing about public employees.

  46. Gravatar of Morgan Warstler Morgan Warstler
    16. January 2012 at 08:37

    progrowth liberal,

    Scott’s going to tell you he favors fiscal stimulus as tax cuts.

    At the end of the day, Sumner was tricky he hit DeKrugman on smoothing, and kept the attack there. In debate, the opposition only has to prove the affirmative was wrong on points.

    Also in debate, when DeKrugman makes that one smallish mistake, it is better to dump the point, rather than try and carry the weak along in the fight.

  47. Gravatar of Rob Rob
    16. January 2012 at 08:45

    Scott,

    May I ask for another point of clarification on the original post (that you quote again in this post)?

    “So maybe C fell by $20 million and investment was unchanged. Yeah, that could happen, but in that case private after-tax income fell by only $20 million and there was no consumption smoothing at all. Checkmate.”

    For this to happen ($100 billion Govt spending financed by $100m tax leading to only a$20m fall in C) then logically additional money must have come from somewhere, most likely cash balances. Assuming this is the case then it would seem reasonable to conclude “people dissaved in that period because of the higher tax”. But you have excluded that possibility by the tight definition S=I.

    It seems we need two definitions. Savings as defined in the S=I that relates to savings on physical goods, and another definition of savings as just “additions to cash balances that may or may not end up as savings in the other sense”. Consumption smoothing then would relate to reductions in either kind of savings.

  48. Gravatar of BK BK
    16. January 2012 at 08:57

    Krugman’s post explains how he looks at the issue. You don’t respond to the content in his post. You wasted my time.

  49. Gravatar of ssumner ssumner
    16. January 2012 at 09:25

    Bill, Yes, in the first post I did on this I pointed out that the argument they should have made was that velocity would increase. That would have been a good way of responding to Cochrane. Instead he seemed to equate “consumption” with “spending,” which is not a good response, as it’s not true.

    Joe2, Yes, I recall that warning.

    Major Freedom, Yes, that’s what they don’t get.

    Sesh, Where the hell did I assume GDP was fixed?

    Ben, Perhaps you are right.

    Tom, Sure, if you simply assume the Keynesian model is true, it’s true. But his argument is completely unpersuasive. He suggests a fall in C is equivalent to a fall in “spending,” but it isn’t.

    Rob, You said;

    “But you have excluded that possibility by the tight definition S=I.”

    That’s not my definition, that’s Wren-Lewis’s definition. See his new post. First he denied it, then he accepted it.

    BK, Krugman’s post is an explanation of the prehistoric Keynesian cross for first-graders. What specifically do you want me to respond to? “Yep, that’s the Keynesian cross model.” I’m more interested in the fact that he refused to defend Wren-Lewis’s consumption smoothing argument. he refused to address my argument.

  50. Gravatar of tom tom
    16. January 2012 at 09:36

    Simon is arguing that increase in savings equals increase in investment production but may not equal increase in investment spending. In the same way we can argue that decrease in savings doesnt have to equal decrease in invetment spending (because of stock liquidation) so I cant see error in his reasoning.

  51. Gravatar of Bill Woolsey Bill Woolsey
    16. January 2012 at 09:38

    I made an error in my last post.

    In other words, with consumption smoothing, debt financed government spending reduces _desired_ national saving despite Ricardian equivalence. It takes a higher real interest rate to make realized national saving (evaluated at full employment output and income) equal to investment-spending on capital goods.

    It takes a higher real interest rate to make desired national saving (at potential income) equal to desired investment.

    Without consumption smoothing, a debt financed increase in government spending might result in no change in desired national saving. That would be because desired private saving rises to match the increaese in government debt. In that case, there is no impact on the natural interest rate.

    Now, I am interested in how government debt impacts the natural interest rate and the allocation of private expenditure and production.

    In my view, nominal expenditure on output should be controlled by monetary policy, and so, budget deficits are about public finance. How does it impact relative prices and the allocation of resources.

    In my view, this isn’t just about aggregate demand.

  52. Gravatar of Jason Jason
    16. January 2012 at 09:44

    In fairness, your original post does not mention consumption smoothing at all, but rather the S = I identity:

    —-[quote] Wren-Lewis seems to be the one making a simple logical error (which is common among Keynesians.) He equates “spending” with “consumption.” But the part of income not “spent” is saved, which means it’s spent on investment projects. Remember that S=I, indeed saving is defined as the resources put into investment projects. So the tax on consumers will reduce their ability to save and invest.

    http://www.themoneyillusion.com/?p=12596

    Then you make the identity the title of your next post:

    —-[quote] It’s tough to argue against an identity

    http://www.themoneyillusion.com/?p=12636

    You then make your consumption smoothing argument.

    I’m still just a neophyte but when I read the consumption smoothing argument, but it appears when you are using it consumption smoothing applies to C alone and not S = I?

    Y = G + C + S

    Y = (not smoothed) + (smoothed) + (not smoothed) ?

  53. Gravatar of tom tom
    16. January 2012 at 09:52

    “He suggests a fall in C is equivalent to a fall in “spending,” but it isn’t.”

    I dont think that he said that. He said that increasing taxes by X at time t doesnt imply that spending will fall by X at time t. And later he posted his Q&A. I think he is right because if you increase taxes consumption will fall and savings and investment production will fall but investment spending doesnt have to fall (because of stock liquidation).
    So for me it all makes sense.

  54. Gravatar of D R D R
    16. January 2012 at 09:59

    Artturi B,

    That is more or less correct. Cochrane failed to complete his proof, so consumption smoothing is irrelevant.

    Cochrane more or less argues that if taxes rise by 100 and the government simply gives the money back to the private sector then nothing really changes. That’s perhaps overly simplistic but it’s an argument. Unfortunately, it does not carry over to a case where the government demands a bridge in return for sending the money back. At a minimum, we have produced another bridge, and this counts toward national income. Cochrane did not prove that C+I fall in response.

    I can imagine such an argument, but none was provided and so it remains mere assertion. Any counterexample disproves that part of his case.

  55. Gravatar of BK BK
    16. January 2012 at 10:12

    ssumner, I have just one year of economics at Berkeley over 30 years ago. I suppose that I am a “first grader.” This “prehistoric” tool is what Krugman uses. If it’s so useless, then here’s your opportunity to discredit everything he advocates. Bury the tool and you bury Krugman.

    Who cares about your argument in the context of our lousy economy? Krugman advocates fiscal stimulus. Nobody else seems to be advocating anything else except waiting. If fiscal stimulus will help then why are economists standing in the way? Is the potential downside of a failed stimulus so great that it shouldn’t be tried?

  56. Gravatar of Rob Rob
    16. January 2012 at 10:16

    I just read Wren-Lewis’s post. In his model increased cash balanced are part of savings and represented as part of investment as increased inventories (DS).

    Would he not be consistent to claim that in his model consumption smoothing was indeed out of this pool of savings and that initially it would lead to a decrease in DS ?

  57. Gravatar of Pat Pat
    16. January 2012 at 10:16

    To all the sane commenters on this thread: Can we please stop paying attention to Scott Sumner, now? I don’t think this is ever going to get better.

  58. Gravatar of 123 123
    16. January 2012 at 10:40

    @Bill Woolsey

    “That would be because desired private saving rises to match the increaese in government debt. In that case, there is no impact on the natural interest rate.”

    Krugman says the presence of liquidity constrained debtors is the source of the liquidity trap, in this the natural interest rate increases.

  59. Gravatar of Noah Smith Noah Smith
    16. January 2012 at 10:50

    Scott,

    You are correct that fiscal stimulus does not work via consumption smoothing.

    Can we move on to something else now?

  60. Gravatar of tom tom
    16. January 2012 at 10:52

    Scott, we know that investment spending=investment productions – change in inventories.

    And Simon said:
    “So S=I by definition. But here investment includes what is called ‘stockbuilding’ or ‘inventory accumulation'”

    So what do you mean by saying identity? Because Simon is referring to production point of view: “measured investment rises because firms accumulate inventories of the goods that consumers did not buy”

  61. Gravatar of Charlie Charlie
    16. January 2012 at 10:54

    Scott,

    This is boring. You think someone is gonna say, “alright you win?”

    Wouldn’t you rather talk about Cocharane’s post on the Fed’s balance sheet risk?

    http://johnhcochrane.blogspot.com/2012/01/worlds-biggest-hedge-fund.html

    I know I’d rather read about it :-)

  62. Gravatar of johnleemk johnleemk
    16. January 2012 at 11:25

    BK,

    You obviously don’t know anything about Scott Sumner and even Paul Krugman if you think the only stimulus options on the table are fiscal. Krugman endorsed Sumner’s proposal for monetary stimulus a few months ago.

  63. Gravatar of Morgan Warstler Morgan Warstler
    16. January 2012 at 11:48

    BK

    And Scott endorsed my fiscal stimulus of “Guaranteed Income / Auction the Unemployed”

    Those aren’t even tax cuts. That’s fiscal stimulus with the side effect of reduces wages – a win for sure.

    “Is the potential downside of a failed stimulus so great that it shouldn’t be tried?”

    Yes, this is a culture war, we want to put the power in the hands of the private, not the public sector. How could you miss this?!?

    Noah,

    “fiscal stimulus does not work via consumption smoothing.”

    I’m sorry, worthwhile points can be scored here by bashing DeKrugman over the head until he admits / recants.

    There are many more fights to come, chinking his armor is important.

  64. Gravatar of StatsGuy StatsGuy
    16. January 2012 at 12:12

    Ick, don’t get sucked into this fight… Or at least, don’t take it personally.

    I’m still trying to figure out the source of all of this cross-talk. K writes: “First I ask how much this would increase savings, holding GDP constant; then I ask how much GDP has to fall to restore the equality between savings and investment.”

    The notion here is that there is a temporary disequillibrium between S and I. I think it’s fair to think that Krugman did in fact mean this… Here’s a 2009 post:

    http://krugman.blogs.nytimes.com/2009/01/27/a-dark-age-of-macroeconomics-wonkish/

    “after a change in desired savings or investment something happens to make the accounting identity hold. And if interest rates are fixed, what happens is that GDP changes to make S and I equal.”

    I’m still not entirely sure what he means by Savings (desired resource savings?). I think the FUNDAMENTAL QUESTION HERE is whether or not it’s possible for this situation to be resolve by a GDP increase under a situation with a PERFECTLY fixed interest rate situation (and my suspicion is this might be possible if SRAS is perfectly flat).

    BTW, try keeping posts short.

  65. Gravatar of BW BW
    16. January 2012 at 12:24

    Scott, I have to give you props for responding to everyone.

    Still, I’m going to take issue with this:

    “If something is an identity, it does happen instantly. I think what you are trying to say is that initially the form “investment” takes is capital accumulation in the form of bigger inventories.”

    Everything that happens in the world requires a mechanism. For instance, to the extent that capital markets are weak-form efficient, but they are only efficient because there arbs make them that way. That’s why capital markets can’t be *perfectly* efficient: in equilibrium, the level of pure arbitrage is non-zero.

    The same is true of savings and investment. Something has to happen to bring savings and investment into alignment. And that something can’t be taken for granted.

    It’s true that if you want to *define* S=I, you can, but if you do, then that identity can’t be the basis for any empirical theory or prediction. All you have in that case is a tautology. As Kant long ago showed us, you can’t derive an a posteriori argument from an a priori one.

    Also, you have indeed have known what I wrote as a college freshman. Perhaps I should have phrased my comment a bit differently, to address only your theory (which does not incorporate that insight) as opposed to your person. But it’s not as if you are an innocent here. You are accusing brilliant economists of making basic errors.

    And yes, I know, Krugman is doing it too. So are Lucas and Cochrane. In fact, this type of disrespect has pervaded our entire civic discourse. It should stop.

  66. Gravatar of TheMoneyIllusion » The Keynesian bubble TheMoneyIllusion » The Keynesian bubble
    16. January 2012 at 12:44

    […] best counterargument I’ve seen is from Andy Harless, who suggests that consumption smoothing somehow underlies the Keynesian result.  Even if […]

  67. Gravatar of Gray, Germany Gray, Germany
    16. January 2012 at 13:32

    “I thing it is”
    That’s the problem with Sumner’s two blog stories – too many things in them, but not enough thinking about how to present a coherent argument. Somewhat counterproductive to blame Krugman of “confusion” only to confuse readers about the exact nature of the criticism, too.

  68. Gravatar of SDS SDS
    16. January 2012 at 14:04

    BW,

    Thanks for your posts. They really helped clarify the issue. I’m a neutral observer that is trying to make sense of macro these days. The fact that a passel of PhDs are arguing over Econ 101 concepts is very unnerving and is making me doubt most of what I was taught about Economics.

    What a mess.

  69. Gravatar of Joe Joe
    16. January 2012 at 15:09

    Scott has this all wrong:

    “Now let’s suppose the tax-financed bridge cost $100 million. If taxes reduced disposable income by $100 million, then Wren-Lewis is arguing that consumption would only fall by $20 million; the rest of the fall in after-tax income would show up as less saving. I agree. Checkmate in two.”

    Wren-Lewis argues that consumption is smoothed. So taxes would be reduced in the first year by a fraction of the cost of the bridge, not the exact cost of the bridge. In fact, if you take Scott’s arguments and use dC + dI = -kdT where k < 1, then you end up proving the very thing that Scott attempts to disprove.

  70. Gravatar of Thomas Zaslavsky Thomas Zaslavsky
    16. January 2012 at 19:34

    Adam (ak)
    16. January 2012 at 03:50:
    Thank you for your thorough analysis with variables, including GDP and time, taken into account. It makes more sense to me (not an economist) than all of Sumner’s and Krugman’s explanations.

    It seems to me that you demolished the very parts of Sumner’s argument that seemed most peculiar.

  71. Gravatar of Hal Horvath Hal Horvath
    16. January 2012 at 20:50

    “…saving is not money down a rat hole, but rather represents spending on capital goods.”

    Not during a credit downturn (such as now).

    Instead, a significant part of saving is spent on….speculation, such as speculation on oil, gold, foodstocks, etc….

  72. Gravatar of Miguel Miguel
    17. January 2012 at 04:59

    Scott,

    Check mate, game set and match:

    “If something is an identity, it does happen instantly. I think what you are trying to say is that initially the form “investment” takes is capital accumulation in the form of bigger inventories.”

    This statment is totally wrong and BW has the right analysis. Economies are not static stocks of variables, they include both stocks and flows. Identities are relationships that hold in the aggregate over time but need not hold at a specific moment while there is an adjustment in progress.

    My question is, are you engaging in this analysis from an ideological perspective to obfuscate and undermine the other side of the argument or from an honest misreading of basic economics?

  73. Gravatar of Doc Merlin Doc Merlin
    17. January 2012 at 06:12

    @ Miguel

    Did you just say “are you an idiot or a liar?”
    :-(
    Attack his arguments, ad hominem isn’t welcome here.

  74. Gravatar of D R D R
    17. January 2012 at 06:25

    Miguel,

    I don’t understand how S and I cannot balance instantly. If C falls by X, then either inventories increase (I rises by X) or services are not performed (Y falls by X).

    Similarly, if C rises, then either inventories decrease (fall in I) or something more was produced (increase in Y)

    Again, this doesn’t mean that savings *is* investment. But can you give an example of S and I failing to balance at any instant?

  75. Gravatar of D R D R
    17. January 2012 at 06:31

    Hal,

    In the case of an asset purchase, your actual act of saving *is* someone else’s actual act of dissaving. It’s not “down a rabbit hole” but moved around.

  76. Gravatar of Jon Jon
    17. January 2012 at 08:22

    I think you are confusing things by speaking in terms of after-tax income. If we have Y=C+I+G and G increases, what happens to Y? In a traditional Keynesian model, I is fixed, so the question is: what happens to C. Well C=a+b(Y-T), so there are 2 effects, C increases as Y increases and decreases as T increases. If T is fixed (i.e., the spending is deficit financed) C obviously increases. If T=G then there is no change in Y-T, they both increase by the change in G. So consumption doesn’t change, and Y increases overall by the increase in G. None of this depends on consumption smoothing. So, contrary to Scott’s claim, we have the tax-financed bridge increasing GDP by $100 million, with no change in consumption.

    How might consumption smoothing be relevant? Only if someone claims that the above model is incorrect, and try to argue that consumption must fall by the increase in G, because there must be a corresponding increase in T (either levied today or in the future). This was the essence of Lucas’s argument:

    “But, if we do build the bridge by taking tax money away from somebody else, and using that to pay the bridge builder — the guys who work on the bridge — then it’s just a wash.”

    And Cochrane’s on the same page:

    “Before we spend a trillion dollars or so, it’s important to understand how it’s supposed to work.  Spending supported by taxes pretty obviously won’t work:  If the government taxes A by $1 and gives the money to B, B can spend $1 more. But A spends $1 less and we are not collectively any better off.”

    Now, Lucas and Cochrane are aware that the traditional Keynesian model says that deficit-financed spending is supposed to increase output. They may or may not be aware that balanced-budget spending also increases output in that model (I would conjecture that they believe the result is due to short-sighted agents, who don’t realize they will be taxed in the future. Hence I suspect they are unaware of the balanced-budget result). In any case, they don’t like the Keynesian model, so we need to translate the argument into the sort of dynamic framework they prefer. It is for this reason consumption-smoothing is relevant, because a temporary increase in G requires a temporary increase in T (or a small increase spread over time). So only a household that did not smooth consumption would reduce spending by an amount equal to the increase in G.

    With consumption-smoothing, a temporary increase in G causes a smaller decrease in C. If T is raised today, households will try to borrow (reduce saving). However, there is no reason for investment to fall, because if Y is higher overall savings can stay the same. That is, consumption is smoothed by increasing current output (by increasing demand). In the future, output is lower but consumption is maintained because households don’t have to pay the tax (they paid it in the previous period).

  77. Gravatar of Hal Horvath Hal Horvath
    17. January 2012 at 14:11

    D.R., yes, that’s true, eventually, and…after wealth has been transferred, such as during gasoline purchases at the pump during the temporary bubble in prices (bubble being due to “speculation”, which is when many buyers that are not typical users enter the market). This is also akin to front running, and the way it works (instead of only canceling over time) is because demand is seasonal and/or psychological (rising prices induce more widespread stocking up, etc.). This isn’t really pleasant to consider when one would like the market to be the final authority (which is my tendency), but it is realistic. In a nutshell, a normal market in wheat or gasoline is made into a gambling game, and the pros fleece the typical consumer.

  78. Gravatar of D R D R
    17. January 2012 at 14:22

    Hal,

    I am sorry. I have absolutely no idea what you are trying to address here.

  79. Gravatar of ssumner ssumner
    17. January 2012 at 15:22

    Everyone, I have way too many old comments going back to previous posts to answer them all. Let’s move the action to the most recent post, where people finally seem to be accepting that I am right. I will read all comments, but won’t respond to stuff like whether S really equals I anymore. Read any intro text if you want to know why. If I miss an important question of yours, bring it to the new post after reading the comments that begin to accept my argument.

    Bill, Agree about public finance.

    Jason, Yup, saving isn’t smoothed.

    BK, See my new post, Krugman’s position is roughly like that Italian cruise ship.

    Rob, Inventories yes, but only in the very short run. Then fixed investment.

    Pat, Check out the comment section of the new post, people are abandoning Krugman and Wren-Lewis.

    Noah, You said;

    “You are correct that fiscal stimulus does not work via consumption smoothing.

    Can we move on to something else now?”

    Yup, just as soon as Krugman admits he was wrong and apologizes for the snarky post directed against a little Bentley Professor. :)

    Tom, Saving occurs when capital stock is built (including inventories.)

    Charlie, You said;

    “This is boring”

    I’ve never had so much fun.

    BW, I agree it should stop–I’m trying to show Krugman how it feels–he’s the most famous offender. I was initially very polite to him and then he started mocking me. Take that into account please.

    Jon, You said;

    “So, contrary to Scott’s claim, we have the tax-financed bridge increasing GDP by $100 million, with no change in consumption.”

    No, that’s contrary to Wren-Lewis’s claim.

  80. Gravatar of Miguel Miguel
    18. January 2012 at 02:43

    If S=I we would not have the current problem of excess savings. If we blindly expect all savings to be invested then S=I would mean we would have excess investment at the same time which should translate to robust growth which isn’t the case in the current liquidity trap with underutilised productive assets in the economy and dampened investment. Something else has to shift to preserve the equality and shifts don’t happen with a magic wave of a wand or snap of the fingers. Paul Krugman can answer why S=I doesn’t automatically hold at every instantaneous moment:

    So what are people like Wren-Lewis and me doing when we ask how savings or investment would change from some given shock? We’re not forgetting that in the end savings must equal investment; we’re just doing the first step in a comparative statics exercise. Suppose that I want to ask how, say, a fall in housing wealth affects the economy. First I ask how much this would increase savings, holding GDP constant; then I ask how much GDP has to fall to restore the equality between savings and investment. The picture “” which is something everyone used to learn in Econ 101 “” looks like this:

    Accounting identities are like binding constraints, the economic system reverts to the identities over time but when there are shocks in the economy there is a period of adjustment and another variable has to shift to preserve the identity. It doesn’t just magically happen like Sumner claims.

  81. Gravatar of Miguel Miguel
    18. January 2012 at 03:02

    Also, Doc Merlin and anyone else who is interested. I am not calling anyone names or insulting anyone. I am trying to understand why some notable economists are mixing an accounting identity with a causal relationship. It can be because some highly trained economists are rigidly looking at problems through a particular lens or framework that they steadfastly believe to be true (this does not make them stupid) or that they are doing it on purpose because it fits their ideological view of the world.

  82. Gravatar of Jon Jon
    18. January 2012 at 03:19

    Scott,

    I know you said you would not be responding to comments on this post, but for the record:

    You quoted me saying that, in the simple Keynesian model, a tax-financed bridge would increase GDP by the value of the bridge. I said this contradicted you, you responded that it contradicted Wren-Lewis, by which I assume you are referring to his point about consumption smoothing.

    Now, your earlier post

    http://www.themoneyillusion.com/?p=12636

    you said there was a $100 million tax-financed bridge. So we have G and T increasing by $100m. Then you said consumption falls by $20m, and either investment falls by $80m (with Y unaffected) or I stays the same and Y increases by $80m (so after-tax income falls by $20m). Thus my results contradict yours in the sense that you have consumption falling, whereas I have it constant. Also, you have GDP increasing by $80m whereas I have it increasing by $100m. This doesn’t prove anything, other than playing around with accounting identities without thinking about the underlying economics can lead to some odd conclusions.

    As for the consumption-smoothing argument, I think I explained that later in my comment. Essentially, consumption-smoothing is important because in the previous model, households are not forward-looking, which could invalidate the results. In particular, if the increase in G were permanent, Lucas & Cochrane would be correct, there were be a fall in C offsetting the rise in G, with no overall change in GDP. So, then we appeal to consumption-smoothing, and note that if the increase in G is temporary is doesn’t affect a household’s lifetime wealth very much. So, C falls by less than G increases, and there’s a decrease in *desired* saving at that level of income. However, if output is demand-determined this leads to an increase in current output, which increases desired saving, so consumption can be maintained without actually reducing saving (or investment).

  83. Gravatar of Miguel Miguel
    18. January 2012 at 07:57

    And to finish this off I turn to Noah Smith:

    Principle 4: Argument by accounting identity almost never works.

    Example: “But your theory is wrong, because Y = C + I + G!”

    Suggested Retort: “If my theory violates an accounting identity, wouldn’t people have noticed that before? Wouldn’t this fact be common knowledge?”

    Reason You’re Right: Accounting identities are mostly just definitions. Very rarely do definitions tell us anything useful about the behavior of variables in the real world. The only exception is when you have a very good understanding of the behavior of all but one of the variables in an accounting identity, in which case the accounting identity acts like a budget constraint. But that is a very rare situation indeed.

  84. Gravatar of Miguel Miguel
    19. January 2012 at 08:30

    I wanted to get to the bottom of this S=I nonsense and have finally bottomed it out: so in case anyone is still interested….

    S=I requires that investment is both new investment expenditure and accumulated inventories so that when savings rise in a bust, unsold goods pile up as inventories and are counted as increased “investment”. This is not the same as “spending on capital goods”..rather it can mean “consumer goods that no one wants to buy stored in a warehouse until demand picks up”. What’s the difference? The first is an uptick in economic activity and increased productivity, the second is a fall in economic activity and decreased productivity…yet the logic I see in this blog suggests both of these are equivalent increases in investment!

  85. Gravatar of ssumner ssumner
    19. January 2012 at 13:57

    Jon, That’s the argument WL should have made, but didn’t. The way you make excuses for him is very similar to the way I make excuses for Cochrane (who made a similar error.)

    Miguel. I contacted Noah and he said he doesn’t think I made that error.

Leave a Reply