Thursday 26 November 2015

The Paradoxical Kalecki (and part iii)

(From part ii)

The tables below may help in this discussion.

In principle, capitalists hold the frying pan by the handle: they decide (i) the payroll cut magnitude (D) and (ii), from what they get, capitalists decide how much to consume and invest (without loss of generality, in the examples following they consume as much as they invest).

Table 1 illustrates the baseline scenario, as in Kalecki's equation (1): payroll cut is nil and capitalists consume 50% of profits, investing the remainder 50%. The baseline wage share of Y of this hypothetical economy is 50% (= 100*W/Y).

Table 2 illustrates my own re-distribution example (readers can think of it as the "cut-and-spend" scenario): D ($100) is cut from W and paid together with P to capitalists (for short: capitalists, having the power to do so, re-distribute income in their favor). They now consume/invest (this is the and spend bit) more than they did in Table 1. GNP and employment do not change. Consumption is also re-distributed. Note carefully: compared to the baseline scenario, wage share of GNP fell to 40% without the economy being in recession.

Capitalists -- in the aggregate -- seem capable of handling their business. The only limit to their power to re-distribute incomes from their workers to themselves is given by the workers' patience and/or physical endurance: a "large" D may well cause a backlash, "small" ones may not.


Before proceeding, I invite readers to consider these kind comments -- by Robert King -- posted on November 25, 2015 at 1:43 PM in reply to my previous post. Having written this before receiving his feedback, I decided to re-write some parts of it to address his objections. Otherwise, the post is essentially the same.

Let me first acknowledge this: Mr. King is absolutely right in writing "You partly assume your conclusion to start: revenues are fixed at Y."  I agree with him on that and I want to emphasise this fact for all readers.

However, unlike him, I don't see any problem in that. Conclusions must follow from their premises/assumptions and it's a bad argument one where conclusions fail to do that. If my argument at least supports my conclusion, I could feel satisfied.

I can't see a problem in assuming Y fixed -- as I certainly did -- either. For one, the Levy team did not state anything to the contrary. For another, the derivation of Kalecki's equation also assumes a fixed Y. Therefore, I see little reason to worry about this.

In addition, Mr. King himself offers some premises/assumptions from which the Levy team results (namely "Therefore, cutting payrolls will not directly increase corporate profits") follows. Personally, I do not hold that against him and, if anything, thank him for that. For one, it would have saved me some guesswork (luckily, I seem to have guessed more or less correctly).

At this stage, readers might be asking themselves: If all depends on assumptions, what's the point of this? Is this just a game?

Far from it. The more closely the assumptions relate to reality, the greater our confidence (albeit not our certainty) that the conclusions also relate to reality. This, as hinted in the previous post, is the purpose of this last part: to assess the plausibility, verisimilitude of the premises and conclusions.

What follows has been edited, to address Mr. King's other observations, not explicitly addressed here, but otherwise remains pretty much as I had prepared.



It took some guesswork to illustrate the Levy publication's results. Table 3 is my best shot (judging by Mr. King's comment, apparently my guess was more or less accurate).

Why the guesswork? Wages can fall in two different but not mutually exclusive ways: by (i) reducing employment or by (ii) cutting down hourly wages (or a combination). Looking at flows of funds exclusively -- as the Levy team seems to do -- these procedures are indistinguishable and perhaps that's why they don't specify what they had in mind. However, these different causes have different implications for the "real economy". To match their result, I chose the first option: to reduce employment (a good guess, seemingly). It's worth to make this clear: This is not a necessary assumption, but is the one they seem to have assumed, thus I reflect it in Table 3 (in Table 2, I assumed an hourly wage cut, as per the previous post).

In Table 3 capitalists cut their payroll by D = $100 (as in Table 2).

In Table 2, capitalists used those savings to fund an increase in dividends: payroll cuts logically precede the increase in dividends; the former are the cause, the latter, the effect (in other words, dividends without a corresponding cash availability mean little). In other words: capitalists re-distribute the income in their favour.

In Table 3, they didn't. They cut payroll but -- apparently -- that's all they did.

In terms of the circular flow diagrams, this is how I visualise the situation:

So, what happened to the D dollars capitalists saved? Well, whatever they did with the $100 thus saved, capitalists didn't spend them (as they did in Table 2 with much better results).

Compared to the previous "cut-and-spend" scenario, from Table 2, readers can think of this newer Table 3 scenario as the "cut-and-do-nothing" scenario.

Is this "cut-and-do-nothing" scenario a logical possibility? Of course it is. Does it make any sense in real life? I suggest none whatsoever. I'll, however, leave readers to judge.

At any event, in comparison to Table 1, wages fell (to $400) and profits remain unchanged (as the Levy team would have it); additionally, GNP fell to $900. Because employment fell, physical output fell too: the economy is in recession. Not only that, the wage share of GNP fell, as well: now is 44.4%. Note carefully: both the wage share of GNP fell and the economy is in recession. In this, too, the scenario in Table 3 differs from the scenario in Table 2 [*]

(To the extent the scenario above reflects the Levy team conclusion, my guesses seemed to hit the mark. However, at the risk of being repetitive: the outcome reflected in Table 3 is not necessary; for one, it was avoided in Table 2).


But if the outcome of Table 3 is not necessary, something additional must have been added to arrive at it.

The astute reader may have guessed my own guess: the recession in Table 3 was caused by capitalists' Animal Spirits, not by the payroll cut.

What the Table 3 scenario shows is the paradox of thrift, in disguise, inadvertently superimposed over the payroll cut. It's an unstated assumption. Unless I'm badly mistaken, to conclude that "cutting payrolls will not directly increase corporate profits" -- as the Levy team did -- it takes assuming that capitalists' Animal Spirits will choose a recession and add an unnecessary leakage in the macroeconomic flow of funds. That's where the unspent and "missing" D in possession of capitalists went.

Let's pause and think for a moment. Why would capitalists, after getting a windfall of D, forget everything about it? Let's further remember the first post: doesn't equation (1) say that workers' demand is irrelevant for profits? It doesn't matter that workers' aggregate demand may fall: capitalist's profits are safe. It would take for capitalists to be irrational to believe otherwise.

But what if capitalists sometimes are irrational, as the Keynesian Animal Spirits allows?

Capitalists' self-defeating bouts of irrationality are only the least interesting half of it. The other, more interesting half has clear empirical implications. To justify the Levy team's conclusion capitalists need to act not only in a self-defeating manner, but consistently so, each and every single time, otherwise sometimes "cutting payroll will increase corporate profits". It's not enough being irrational once, but they must be so always: every single time capitalists manage to re-distribute wages in their favour, a recession must ensue.

It is in this light that this chart becomes relevant:


As the chart indicates, the wage share of Y does fall during recessions (the narrow vertical grey stripes). This is consistent with the Table 3 "cut-and-do-nothing" scenario and presumably with the Levy team conclusion: those stripes show both a recession and a fall in the wages share. (Incidentally, it may be consistent with other hypotheses, too, but the fact remains, it is consistent with their hypothesis). But it also falls outside of recessions (the much broader white areas). Their scenario is inconsistent with the falls in wage share during the much longer normal times: the wages share falls almost without interruption.

In that sense, and taking into account that the American economy is a lot more complex than my circular flow diagrams, the "cut-and-spend" hypothesis seems to fare reasonably.

But I'll leave readers to be the judges of that.


Everything may be a matter of assumptions: mine versus theirs. But their assumptions do not seem analytically or empirically defensible. Frankly, they seem extremely implausible. Therefore, I am afraid I remain critical of their conclusion.

Again -- as is my policy -- I'll invite readers to judge.


Is this an argument against Kalecki? I don't think so. If anything, my problem is with the proliferation of Keynesian "paradoxes". I find that style of argumentation problematic: is there a reliable method to distinguish a real paradox, from a questionable argument covered by the "paradox" label?

Neither is this an argument against the Levy team's methodology, in general. My problem is with one of their conclusions, only.

I had planned to add some general comments about the need to consider things beyond flows of funds. Things like re-distribution, real flows of goods and services and factors of production. But I'll leave that for a future opportunity. This post is too long as it is.

To close, I'll remind readers something modern economists often forget:
"To determine the laws which regulate this distribution, is the principal problem in Political Economy." (David Ricardo)

[*] This ignores, for simplicity's sake, the effect of the multiplier on Y: Y' - Y = - m.D (where the multiplier m is a positive real number greater than 1)


  1. not so fast magpie. how about this,
    This can be seen if you rerun your example with a raise in dividends without any wage cuts. Thus, all you prove is that increasing dividends increases profits.

  2. magpie you didn"t reply to,
    Indeed, it is the increase in dividends—a profit source—that creates additional profits, not the wage cut. This can be seen if you rerun your example with a raise in dividends without any wage cuts. Thus, all you prove is that increasing dividends increases profits.

  3. OK. Good stuff. I could not have done nearly as well. Again, more thought is necessary for a proper response.

  4. @Anonymous(es)

    You are right. Apologies for the omission. Let me fix that.

    We do like we did in post ii: exactly the way the Kalecki equation is derived. D, now, is a dividend firms paid capitalists, without a corresponding payroll cut. Firms paid those dividends out of thin air, so to speak.

    Y by income: Y = W + (P + D)

    Y by spending: Y = Cw + Cp + I

    Writing Y (by income) = Y (by spending)

    W + (P + D) = Cw + Cp + I

    Workers spend what they get (Cw = W) and subtracting from both sides:

    P + D = Cp + I

    Compare to equation (1) in post i. The RHS (Cp + I) are the same here and there, yes? But the LHS are not. In equation (1) it is P; above, P + D.

    What does that tell you about D? :-)

    Post i:

    1. "D, now, is a dividend firms paid capitalists" Therefore both P and Cp (which by the way is Ck in your tables) increase by D.

    2. I’m not sure I understand what you mean. Clearly if the LHS is different, then the RHS must also be different (in contrast to what you conclude); this is just simple algebra, no? In other words, if you are trying to show that aggregate income (Y) equals the original sum of aggregate spending (Y from part ii), you have not accomplished that. The original Y by income (W + P) has clearly increased by D. So, one can no longer consider the term “Y”. Instead, it can only be considered something different: say, Y1. Now, clearly the “Y by spending” does not equal Y1 unless the pieces of “Y by spending” are increased by some equivalent amount as D.

      Pls clarify thanks!

    3. Given that P is a positive real number, there is one way and only one way to make P + D = P.

      D = 0.

      The only way a dividend paid out of thin air can be possible is if it pays 0.

      In English: you cannot pay dividends (from the Latin dividendum: something to be divided) without something to divide. Thin air doesn't count.


    4. Kalecki equation suppose all W is consumed by workers and all P distributed by firms and consumed by capitalists.
      So you shouldn't add a variable D to the equation but increase both P and Cp+I as Anonymous said in the comment.
      your new equation is P (600) = Cp + I (600)
      or with your nomenclature P+D (500+100) = Cp + I (600)

      so Y =1100 increased

      You pay a dividend with 100 cash (not thin air) and it returns to you again in cash with 100 new revenues (Cp or I) [cause capitalists spends all their income by assumption]
      No change in financial assets but 100 more revenues, so 100 more profits (Revenues-Costs)

    5. This comment has been removed by the author.

    6. Very well, E. Let's do a little role playing to see how your argument works.

      You are the CEO of ACME, the only firm in our economy. I am the only shareholder in ACME. You decide to distribute $100 in dividends. {Incidentally, CEOs cannot emit currency, they are not the Central Bank. But let's leave that aside; let's assume that you can.]

      As I am the only shareholder, the $100 go straight into my pocket, yes?

      Now, you say that "capitalists spends all their income by assumption" (my emphasis).

      But, try as I might, I don't remember having ever signed a document forcing me to spend all my money. Like other capitalists, I, too can fall prey to ((Animal Spirits)). So, I decide not to spend the $100. I decide to spend less. Maybe I choose to spend $0, maybe $99.99. The precise amount doesn't really matter. What matters is that I spend less than $100.

      Well, Kalecki's equation failed. You distributed more money than you had to distribute. You are in trouble.

      Note that my situation as a capitalist is substantially different to the situation of the workers under your direction. They earn little and have by assumption no margin of maneuver, so to speak. To put it simply: they are living from hand to mouth. That may not be literally true, but at least one can say it may approach reality.

      Now, are you sure you want to argue Kalecki's equation by assuming that capitalists are also living from hand to mouth?


      Kalecki's aphorism[*], E, is this:

      "The workers spend what they get, and capitalists get what they spend"

      Two sentences [(1) "the workers spend what they get" (2) "the capitalists get what they spend"] joined by a conjunction (and), right? Moreover, they are very similar in their constructions, yes?

      Similar as they are, the two sentences come from very different places. "The workers spend what they get" is an assumption. "The capitalists get what they spend" is the conclusion, is Kalecki's finding.

      So, to demonstrate that Kalecki is correct ... you are asking us to assume that Kalecki is correct.

      You can see the problem there, can you not?



      Later on I might -- or I might not -- post some further comments, to draw attention to the circular flow of the economy. Unfortunately, it seems the fact that real resources exist was not appreciated.

    7. I assumed it cause that's the assumption of Kalecki profit equation.

      If we integrate marginal propensity of capitalists and workers in the model we get a situation more kaldorian (capitalists saves more than workers)

      By the way the circuit logic is that capitalists invest, which increase profit, which (by assumption) is distributed and (by assumption) entirely consumed.

      If you want to delete the assumptions you get two corresponding leakage of money from the circular flow (firms retained earnings and capitalists savings)

      So less money (cash) in the flow but still P = I+Consumption of capitalists

      It's an accountability matter, cause dividends figure as previous profit shared and not costs, you don't have to subtract it to the revenues you get from I and Cc (which you have to do for Wages -> so Cw-W=0)

      That's profit source.
      That's not cash flow source.
      Kalecki was talking about generating a profit account, not generating net financial assets in the circuit
      That's why real resources are not there (but they obviously limit the profitability of new I and influence consumption prices and quantities (Cc=const =p x Q where p and Q not constant) in the economy)

      Hope the comment is enough clear, not always easy I admit.

      PS that's the interpretation I come reading some works about this stuff. Can't guarantee it's Kalecki true understanding.
      But seems consistent to me

    8. Hi E,

      Sorry for the delay replying. For one, that's because I've been giving your comment some thought.

      (Lately I've also been having a really hard time at work and haven't had much time to think about this.)

      I suppose the best way to proceed would be for you to indicate which exactly are those studies you are basing yourself on.


      Still, let me comment on this:

      "By the way the circuit logic is that capitalists invest, which increase profit, which (by assumption) is distributed and (by assumption) entirely consumed."

      Let's parse that for comments purposes.

      "By the way the circuit logic is that

      (1) capitalists invest, which increase profit,
      (2) [increased profits] which (by assumption) is distributed and
      (3) (by assumption) entirely consumed [money entirely spent by capitalists]."

      Kalecki meant (1) as his finding. A little later you offered the mathematical expression for that:

      [A'] P = I+Consumption of capitalists

      To adopt the notation in the post above (the same used in the Wikipedia entry*), let's rewrite that as:

      [A] P = I + Cp.

      Now let's think of this in terms, as you say, of the circuit logic: it flows from money spent (I + Cp) on the RHS, to money earned (P) on the LHS. In other words, RHS is first, LHS follows, yes?

      How much money capitalists had when they started? Easy: they had money to invest, money to pay for their own consumption and money to pay as wages so that their workers themselves pay for their own consumption. They had, then, I + Cp + Cw.

      How much money is there at the end of the cycle? Easy, too: P + W.

      The two amounts -- at the start and at the end of the cycle -- are equal:

      P + W = I + Cp + Cw.

      Now, as by assumption workers spend what they earn: W = Cw.

      So, P = I + Cp (which is equation [A]).


    9. (from the previous comment)

      Let's look at this from a slightly different perspective. Capitalists invest money to have more money. You, for example, put money in a savings account to have more money at the end, no? Although that kind of financial investment is not what economists refer as "investment", the idea still illustrates my point.

      So it's critical to know what profit is. Profit is by definition the difference between the amount of money one puts into a business and the amount of money one obtains from that business.

      As we've seen, this is what capitalists put into this business: I + Cp + Cw. And this is the revenue they got from the business: P + W (W = Cw).

      Ah, but by that measure capitalists had no profit at all! (See equation [A]). They may not be losing money, but they aren't getting any richer either. The money they had at the beginning is the money they have at the end.

      So, where are the increased profits you mention in (2)?

      (Let's leave (3) out of the picture, because frankly, I can't see it making much sense: it reverses the relationship between spending and earning in (1)).

      You may object that by that measure of profit (revenues minus costs) then profit should be P - I. You can see right there the problem with Kalecki's notion of profit, right? Kalecki's profit is profit minus investment (and, on top, as if things weren't confusing enough, we are to add dividends!).

      But never mind that, because from equation [A]: P - I = Cp. That means that capitalists are consuming all their profits without having to work for their own consumption. That's rather reasonable.

      But I don't think that's good enough. Kalecki's capitalists may be accumulating physical assets, through investment (assuming, which I don't think it was assumed, that those assets are durable). But, what's the point? They, after all, are getting in each cycle the same amount of money they are putting in, no matter how much they have in assets!


      Now, maybe I'm misunderstanding Kalecki. Maybe in his writings he solves all the doubts I have. But, with due respect, those attempting to explain his theories aren't much help.


    10. "Profit is by definition the difference between the amount of money one puts into a business and the amount of money one obtains from that business."
      You are saying P=I-DIV. Ok, I agree BUT this is not happening at the same period. I appear at t0 and you earn P at end of t0, DIV are distributed the next period (or still at the end of t0 if you prefer. It should only be a little analytical difference)
      That's why my point (3) appears as inverted causation, but it's cause doesn't happen at the same period of time.
      AND.. that's capitalists or firms profits? We will return later.

      As we've seen, this is what capitalists put into this business: I + Cp + Cw. And this is the revenue they got from the business: P + W (W = Cw).
      Here it's not clear. Your first equation IS firm revenues (not capitalist costs). And second? It's not capitalists revenues (only P) nor firm costs (only W) [which, for firms, makes by the way P=[Cw+Cp+I]-W=Cp+I as Kalecki said].
      Capitalists have spent I+Cw and get Pshared=P(cause by assumption all profit distributed) [=I+Cw cause equation 1 row above] So they earn what they have spent.
      The fact is that costs of one sector usually are revenues of another sector (so Firms rev=costs other sectors and Firms costs=revenues other sectors), hope it makes more clear why our results differ.

      You may object that by that measure of profit (revenues minus costs) then profit should be P - I. You can see right there the problem with Kalecki's notion of profit, right? Kalecki's profit is profit minus investment (and, on top, as if things weren't confusing enough, we are to add dividends!).
      Kalecki profit is profit, what you saying is still more a firm cash flow (which I agree to be zero if doesn't enter another source of money in the circuit-which usually indebt itself to buy consumption goods at a price which permits firms cash flow >0)

      I would point more generally a thing: in your scheme capitalists and workers are one sector, firms another one.
      But in many your comments you seems to instead merge capitalists and firms in one sector, and workers alone.
      The best approach would be separate all of them in 3 sectors, so you would find profit of firms[R-C=(I+Cp+Cw)-W (still, I repeat, Pshared isn't a cost in accounting principle for firms)], capitalists [Pshared-I-Cp] and workers [W-Cw=0]

      If you think my last line is correct than our problem was a different sectoral composition, if you don't agree point out the exact error in that 3 equations and how you would rewrite them. Maybe doing so we can circumscribe the matter.

  5. To what degree are Cp and Cw autonomous versus income-dependent? What are their respective marginal propensities to save?

    I think we all agree it's reasonable to assume Cw is wage-dependent: pay the workers less money in aggregate, and they will spend less in aggregate. But to what degree is Cp dependent on P? If Cp is autonomous and prior to P, then King et al. are right: it is the autonomous increase in Cp that stabilizes Y; Cw and W are irrelevant.

    If, however, Cp is dependent on the prior period's P, and Cw is dependent on the prior period's W, then you're right: it is necessary to cut W in period 1 to reallocate consumption from Cw to Cp via profits.

    Marginal propensity to save also comes into it. If we assume that MPS increases with household income, and if it happens that people hold money as savings rather than investment, so I != S, then concentrating monetary flows can cause Y to fall.

    1. "To what degree are Cp and Cw autonomous versus income-dependent? What are their respective marginal propensities to save?"

      Good questions Larry. Frankly, I'm not sure I have good answers.

      "I think we all agree it's reasonable to assume Cw is wage-dependent: pay the workers less money in aggregate, and they will spend less in aggregate."

      The derivation of Kalecki's equation assumes that Cw = W. That means that there is no autonomous component and that worker's Marginal Propensity to Save = 0. That much is clear.

      "But to what degree is Cp dependent on P?"

      That's the chicken and egg problem, isn't it?

      BTW, don't forget investment (I). Capitalists' spending is both, investment (including inventory formation) and personal consumption. Both are determined by different things: one by profit expectations and the other by income.

      "If Cp is autonomous and prior to P, then King et al. are right: it is the autonomous increase in Cp that stabilizes Y; Cw and W are irrelevant."

      I'm afraid I disagree. That's one reason (not the only one) why they are wrong.

      When it comes to argue their conclusion ("cutting payrolls will not directly increase corporate profits"), the Levy team forgets all about W and Cw irrelevancy:

      "However, for the whole business sector, cutting employee compensation reduces revenue as well as expenses. Less worker pay means less personal income and, therefore, less personal spending on the goods and services sold by businesses."

      Because it's such a glaring contradiction, that was, in fact, the first inconsistency I found in this kind of argument: W and Cw are both irrelevant and all-important for profits -- at the same time, sometimes within the same paragraph. Incidentally, others, besides the Levy guys, have argued in that exact same manner.


      Ultimately, Larry, we may spend years going back and forth on this without reaching any conclusion. Trust me, I've tried hard, on and off, since encountering that argument the first time.

      The bottom line is that their explanation for a wage fall requires a recession. The US has had some 50 years almost continuous fall in the wage share of GDP. Where is the recession explaining that?