All Value-Form, No Value-Substance: Comments on Moseley’s New Book, Part 11

by Andrew Kliman

Here is the eleventh installment of “All Value-Form, No Value-Substance,” the series of comments I’m writing on Fred Moseley’s new book, Money and Totality: A Macro-Monetary Interpretation of Marx’s Logic in Capital and the End of the “Transformation Problem.” It responds to part of Moseley’s reply to the tenth installment.

Please see the “Miscellaneous” section on the homepage of With Sober Senses for links to previous installments.


  1. Comment on Kliman’s Part 11

    Well this is progress I guess. Kliman is accusing me of adopting TSSI rather than being “physicalist”! Actually I am neither.

    I have argued on many occasions that Marx’s prices of production are *long-run equilibrium prices* that change *only if* the productivity of labor or the real wage changes. I think there is very strong textual evidence from all the drafts of Capital to support this long-run interpretation of Marx’s prices of production. I have summarized the textual evidence in Chapters 9 and 11 of my book and in this 1999 working paper on Academia:

    In the general case (i.e. with positive SL and positive profit), long-run equilibrium prices are prices for which *input prices = output prices*, not because input prices and output prices are determined simultaneously (as in Sraffian theory), but as a *necessary feature of long-run equilibrium prices with equal rates of profit.* If input prices were ≠ output prices, then the rate of profit in the next period would not be equal unless input prices and output prices change in the next period (and in the period after that, etc.) in order to continue to equalize the rate of profit in each period. And if input prices and output prices continue to change in subsequent periods (even though productivity and the real wage remain constant), then these prices are not long-run equilibrium prices. If output prices are to be long-run equilibrium prices, input prices must = output prices.

    If there is surplus labor in a capitalist economy, then according to my interpretation of Marx’s theory, the rate of profit will be positive and rates of profit of individual industries will tend to be equalized and market prices will tend toward long-run equilibrium prices (prices of production) for which input prices = output prices.

    In the very exceptional case of an actually existing fully automated capitalist economy, long-run equilibrium prices do not exist and such an economy could exist only by chance and for a very short period of time. In this exceptional case, if there were physical surpluses of some inputs (as in Kliman’s examples), then the prices of these inputs would ≠ their prices as outputs, as in the TSSI. The output prices of these goods would be less than their input prices, so that profit would be zero. But these output prices would not be long-run equilibrium prices; if this economy lasted for another period (not likely since it is not viable), then output prices would fall again and thus could not be long-run equilibrium prices. This economy has no long-run!

    Another reason why a fully automated capitalist economy is very unlikely to ever actually exist in the real world is that the trend toward automation (labor-saving technological change) would take place over an extended period of time. But according to Marx’s theory of the rate of profit, continued labor-saving technological change would cause the rate of profit to fall which would eventually cause a depression long before the economy became fully automated. That is what Mandel argued.

    And a fully automated capitalist economy should not be confused with the Great Depression. Profit was = 0 in 1932 because the economy had collapsed, not because surplus labor = 0 in a long-run sense, as the recovery of profit in the subsequent years demonstrated. Capitalism has a “recovery mechanism” from depressions – mainly the devaluation of capital through widespread bankruptcies. There would be no “recovery mechanism” in the case of a fully automated capitalist economy. If such a fully automated capitalist economy somehow managed to exist, it would not last very long.

    In any case, this very exceptional case of an actually existing fully automated capitalist economy does not affect the general case in which there is surplus labor and a positive profit, and in which the rates of profit of individual industries tend to be equalized and market prices tend toward long-run equilibrium prices which have the necessary feature that input prices = output prices, unlike the TSSI.

    I am leaving tomorrow for a trip to Brazil for a month (talks in Belo Horizonte, Sao Paolo, and Rio, plus some R&R), so this will be my last comment for a while (except for one more explained below). I would be happy to continue the discussion after I return. I think we have made some progress in clarifying the issues. My comments on Parts 9, 10, and 11 best reflect my current understanding.

    I will also post below a reply to Michael S’s comment on Kliman’s Part 10.

  2. Reply to Michael S’s comment on Kliman’s Part 10:

    Michael, thanks for your comment. I am glad that you think that the discussion has been worthwhile. I think so too.

    Your final comment was:
    “Third, it is no doubt tedious for Andrew to demonstrate again and again, example by example, where Fred makes an error. But, it’s nevertheless insightful because it highlights the issue from different angles.”

    Andrew’s arguments and numerical examples “demonstrating” that my interpretation of Marx’s theory of the rate of profit is the same as his “physicalist” rate of profit are in fact based on *circular reasoning*. His argument *assumes* my Marxian rate of profit and uses my rate of profit to derive his I/O coefficients; and then he uses these I/O coefficients to derive his “physicalist” rate of profit. Andrew’s I/O coefficients are not the standard physical coefficients (as in Sraffian theory), but are instead *derived from my monetary quantities and my monetary rate of profit*.

    *There is only one rate of profit that is consistent with his coefficients – and that is my Marxian monetary rate of profit from which his coefficients are derived!*

    Please read again my comment on Part 9 and especially my summary in the Conclusion of Andrew’s logic of derivation in his numerical examples – from my Marxian monetary rate of profit to his “physicalist” rate of profit.

    Below is a quote from the Conclusion of my comments on Part 9:

    I have come to realize more clearly as a result of this discussion that *because Kliman’s I/O coefficients are derived from my monetary quantities* (e.g. a1 = C1 / P1), his I/O coefficients result in a “physicalist” rate of profit that is the same as my rate of profit derived from monetary quantities and the LTV. However, as the above comparisons with Sraffa’s theory show, this conclusion does not imply that my Marxian monetary rate of profit is determined by actual physical quantities (as in Sraffian theory), but instead implies that Kliman’s “physicalist” rate of profit is *not* determined by actual physical quantities, but is instead determined by the *monetary quantities* from which his I/O coefficients are derived.

    Here is the key point: when Kliman calculates the I/O coefficients from given monetary quantities, my Marxian monetary rate of profit (already determined) is *presumed* in these calculations. In particular, my Marxian monetary rate of profit is *presumed* in the calculation of P1 and P2, which are then used to calculate Kliman’s I/O coefficients, and these derived I/O coefficients are then used to calculate his “physicalist” rate of profit.

    For example, in his case of technological change in Part 1, the rate of profit is *presumed* in the determination of P1 and a1 in the following way:

    before technological change:
    *my Marxian monetary rate of profit = 50%*
    P1 = (C1 + V1)(1 + r) = (10 + 2)(1 + 0.5) = 18
    a1 = C1 / P1 = 10 / 18 = 0.56
    *Kliman’s “physicalist” rate of profit = 50%*

    after technological change:
    *my Marxian monetary rate of profit = 25%*
    P1 = (C1 + V1)(1 + r) = (10 + 2)(1 + 0.25) = 15
    a1 = C1 / P1 = 10 / 15 = 0.67
    *Kliman’s “physicalist” rate of profit = 25%*

    Kliman’s “physicalist” rate of profit is calculated from a1 and the other I/O coefficients that are derived from my Marxian monetary rate of profit the same way, and it obviously follows that *the CALCULATED rate of profit* (Kliman’s “physicalist” rate of profit) *will be equal to the PRESUMED rate of profit* (my Marxian monetary rate of profit): 50% before technological change and 25% after technological change. That is just a matter of *circular arithmetic*. But it does not imply that my Marxian monetary rate of profit is determined by actual physical quantities because no actual physical quantities are included in the argument.

    I wish I had realized this crucial circularity in Kliman’s arguments earlier in this discussion. [end of quote]

  3. My interest is in re-empowering Mars’ economic theory as the one that explains capitalism. But for this,
    unfortunately, “The reclamation of Capital from the myth of inconsistency is … an absolutely necessary
    and vital precondition …” as Andrew K. stated in his “Reclaiming Marx Capital” book.
    Now, only very recently I became aware of this extensive debate here stimulated by Fred M.’s new book.
    ‘Fascinating’ is a good word for it and as a positive side effect it made me read Andrew K. book in its
    major parts more intensively.
    But now the debate seems having reached a certain ending point, as Fred M. has found in Andrew’s
    calculations an error of ‘circular arithmetic’
    – see 6th Oct 2016.
    However, the question remains, what is the cause for such an error?

    It has to do with how to analyse non-stationary temporal systems.

    a) Non-stationary models
    Analysis requires reduction of complexity to get simpler models of the real system. Systems consist of
    processes of potentially diverging cycle period. First simplification may be assuming a fixed period.
    For characterization of non-stationarity, it’s wise to distinguish at least between short-, medium-,
    and long-term aspects. Short-term aspects last shorter than the modelled system period.
    So far, such a system model looks like this (let’s assume a period of one year, a system with one
    process A, having one amount-Parameter ‘Par a’ that is result/output of one process cycle and input for
    the next cycle):

    2020 2021 2022 2023 2024
    |- 1. cycle of A —|- 2. cycle of A —|- 3. cycle of A —|- 4. cycle of A —| …
    |50in – Par a out50|50in – Par a out60|60in – Par a out60|60in – Par a out70| …
    | ^ext. input | ^ext. input | ^ext. input | ^ext. input |

    This is the model of a long-term non-stationary growth system, but medium-term changes are
    so slow that there is only every 2nd year a change of Par a. Besides Par a there is an external
    system input. In real systems this is often supply of energy that may stem from an abstraction layer
    underneath, where e.g. in case of a battery chemical energy is transformed into electrical energy.

    In order to observe real systems and their models, namely for checking to what extend they deviate,
    snapshots are made at regular periods, where the Parameters of all processes are collected (in our
    example this would be only one Parameter picked up say 1. November).
    All these picked up values are results of the cyclic functioning of the processes in before. They are
    simply a simultaneous collection of data. Never in my over 30 year lasting experience as system
    analyst a I became aware of any trial to artificially create a functional relationship within
    such a data collection – with one exception, the one of Sraffa and friends.

    I consider it as a historical merit of Andrew K. et al of having erected a stop sign against Sraffanian
    dominance by insisting on: though there is a layering of abstract labour, value price and production
    price, it’s still one system. It is non-stationary temporal, processes are circulation of capital and
    interconnection from period to period is through prices/money and not physical units. External input
    is a substance that keeps the system alife and enables growth: abstract labour.

    b) Stationary models of non-stationary systems

    When medium-term changes in relation to modelled system period are slow (as in our example),
    a stationary model can be applied looking like this:

    |- one cycle of A –|
    *|50in – Par a out50| –> back-link to *
    | ^ext. input |
    ^ snapshot

    Besides input and output having same amount, what does this change? There is still a circulating
    process, there is a before and a after, temporal or at least sequential and the system depends
    crutcially on the external input. Only the data collection at snapshot incident is simultaneous
    in this model.

    Andrew K., however has the conviction that input/output equality inevitably leads to a) simultaneism
    and b) to physicalism (see page 76f of his reclamation book). This being incompatible with labour
    value theory and makes it redundant.

    a) by pouring the stationary sequential model above into a big pot of simultaneous evaluations
    Andrew K. misses something. All the collected snapshot data are results of the circulation process
    of capital together with labour exploitation going into this process in before. And this living creation
    before and dead result after is lost once ‘sequential’ and ‘simultaneous’ are no longer differentiated.
    This ‘being a result of capitalist exploitation process’ is also true for the physical quantities.
    Only those physical quantities go into the data collection that have contributed by being commodities
    to profit generation. Other do simply not appear. So in the data pattern of all the physical quantities
    profit rate and prices are encrypted like in a hologram. From this hologram together with an appropriate
    equation system profit rate (or at least something alike) and prices could be calculated back. But doing
    this would be quite strange, as the correctly collected profit rate and prices are available in the data

    b) even (as Straffa & friends do) one constructs an exactly-one-solution artificial equation system
    between mere result-data, it is not inevitable to get from simultaneism to physicalism. Instead of
    calculating price/profit rate = f(physical quantities) he/they could have calculated physical quantities =
    f(profit rate/prices), simply by changing coefficients into variables and vice versa. Of course they
    didn’t do so because their interest was not to explain profit. Their interest is to hide the origin of
    profit, therefore they go the strange way.
    Everybody knows that mere results do not explain where they are coming from. But by creating
    some magic equations around them one can create the impression of having an explanation.
    And the coronation is when just this figure ist coming out as resulting solution that was the cause
    of the real results observed. Such pseudo explanation makes the real one seemingly redundant,
    clear. So it’s not the simultaneism that creates the problem, it’s the inability of understanding the
    Sraffanian trick and of criticizing this theory as bourgeois affirmative ideology.

    Now consider the logical pattern of Sraffa’s approach:
    a1) physical quantities are determined by profit rate + prices. This happens through real system and is
    hidden to nearly everybody
    b1) Physicalists take resulting physical quantities as originating givens and calculate profit rate

    And compare it was Andrew K. did from his conviction that Fred M. must be physicalist because he
    is using input/ouput equality:
    a2) physical units are tacitly calculated from Fred’s profit rate (+ other monetary figures).
    b2) Andrew uses physicalist’s approach for accusing Fred of being physicalist.

    Fred M. did identify the error, probably because a2) is in a formula and not as hidden as in a1)
    There is more to be said about this. E.g. there are good reasons for Andrew K.’s conviction, what
    is wrong nevertheless? what happens when medium-term is much shorter? Too much for a comment.

  4. No Herbert Panzer, you’ve misunderstood. Moseley’s “circular arithmetic” charge is a red herring. I’ll discuss this more fully in an upcoming reply to him, but the key point is this. In order to compare his rate of profit to that of the physicalists, we have to deal with the SAME economy. And if we are dealing with the same economy–including the same physical quantities (whether posited as starting points or deduced from his “macro-monetary” numbers)–Moseley’s rate of profit is the SAME as the rate of profit of all other physicalists.

    That is what I’ve maintained all along and that is what he now concedes. The difference between them is therefore all value-form, no value-substance.

    He goes on to try to argue (completely unsuccessfully, as I’ll show) that his physical quantities differ from those of all other physicalists. In plain English, what that means is this: that if he is dealing with one economy while they are dealing with a DIFFERENT economy, they get different rates of profit.

    Wow. Major revelation.

  5. when a system is stationary (or at least you have created a stationary model from it),
    all it’s parameters are fixed, typically in the form b = f(a). However, this is not everything
    known about the system. Assume an oscillator. One has to continuously inject
    energy to get some amplitude out. So, amplitude = f(energy). It will remain clear
    that the energy is the cause and the amplitude is the consequence. The existence
    of function f is not what creates simultaneism. Simultaneism is: taking a relationship
    R(b,a) and forgetting about everything else, namely that it is about a process and
    something that drives it. Because then you can ‘theorize’ a = f*(b), i.e. let’s create
    energy from amplitude, what in our case of economical theories means physicalism.
    This is what I have called Sraffa’s trick and where I miss criticism.
    Taking this f* out of any real context allows theorizing anything, however if
    f* is just the inverse, then it is no surprise at all that if ‘a’ is the same in f and f*
    also ‘b’ is the same (as you say Fred M. now concedes). Nevertheless: b = f(a)
    explains what a certain amount of substance ‘a’ creates, while a = f*(b) is simply
    stupid (or in best case would answer the question: how much of substance ‘a’ do
    I need to create an amplitude of ‘b’).
    Talking about same or different system:
    in b = f(a) ‘b’ is constrained by the possible settings of ‘a’. In a = f*(b) without
    wanting anything else to know, the setting of ‘b’ is free from these constraints.
    Now these Sraffanian guys belief they can give ‘b’ (the physical quantities)
    any value and from their f* some profit rate is coming out (whereas e.g.
    in full automation case the real result would be 0).
    This is indeed something else. However it is not talking about different
    economies but about one economy and a hypothetical construct in the brain of Mr. Sraffa
    that takes assumptions that in any capitalism do and will not exist. For me
    Fred M. is only confronting a real result with Sraffians’ belief to show
    indeed they are dreaming.

  6. Hi Fred,

    thanks for your reply to my comment above.
    (And sorry now for my late answer.)

    I’m happy that we can (at least) agree upon:
    “I am glad that you think that the discussion has been worthwhile. I think so too.”

    So, if followed your advice and carefully read (or reread) this comment and also your comment on part 9.

    But, to be honest, your arguments — especially the “circular reasoning” or “circular arithmetic” — simply did not convince me (quite the contrary).

    Here’s are the reasons:

    Andrew has shown that it’s possible to start with your macro-monetary quantities (let’s forget about a simple scaling factor) and also the macro-monetary rate of profit based on them, and then — if (and only if) these data is combined with your stipulation that input price (of production) = output price (of production), i.e. simultaneous valuation or pricing — one is able to calculate the I/O coefficients and relative prices.
    We may call this the “back-calculation”.
    So, your start point is quite different to the Sraffians. They start with physical quantities, I/O coefficients and then — thanks to simultaneous valuation or pricing — they are able to calculate relative prices and a (physical) rate of profit. And, finally, by using some sort of essentially arbitrary scaling factor they end with their macro-monetary quantities.

    Now, Andrew has shown that we can start with your macro-monetary quantities and back-calculate and then we can “forward-calculate” — again, thanks to simultaneous valuation or pricing — and end up with your macro-monetary quantities.

    You call this “circular reasoning” or “circular arithmetic” and this gives the impression (at least to me) that this possibility of a “back-calculation” and “forward-calculation” is meaningless, tautological, trivial or obvious. But that’s wrong.

    I think this means (and proofs) *a lot*.
    What this means is that a model of macro-monetary quantities is one-to-one-transformable into a physical model (again, let’s forget about a simple scaling factor). Moreover, since one can then derive the gist of the macro-monetary model from the physical stuff means that the macro-monetary quantities are a mere veil that hide the physical stuff behind them and obfuscate the underlying physicalist logic.
    And that’s because — thanks to simultaneous valuation or pricing — the macro-monetary model is hard-wired in one-to-one relationship (again, let’s forget about simple scaling factors) with a physicalist model.

    The way is which we start the calculation is essential irrelevant, the fact that we can go both directions is important.

    I think that this point also plays a big role in the discussions regarding a full-automated economy.
    Because what Andrew tried to show (and I think he succeeded) is that if profit is zero in your macro-monetary model, then — because this model is one-to-one hard-wired with a physical model (thanks to simultaneous valuation or pricing) — there cannot be a physical surplus. If there would be a physical surplus then your macro-monetary profit cannot be zero. (Because of the hard-wiring.)

    Now, I have considered your comment to part 9 — especially the conclusion that you quoted.
    This example just shows that it’s possible to go the “whole way” (or, if you prefer, the “whole circle”), i.e. to start with the macro-monetary quantities, deduces the physical stuff behind it, and then finally to calculate the (gist of) the macro-monetary model (kind of a control account).
    And, again, yes, this means a lot. It means that both models, the macro-monetary model and the physical stuff, are just two sides of the same coin, are essential interchangeable and that they follow the same physicalist logic.

    For example, in the part of your comment on part 9 that you quote, we have a technological change.

    Now, first, the word “change” is ambiguous since it can mean technological progress or regress.
    In case of a technological progress the I/O coefficients decrease (less input is required to produce one unit of output) and in case of a regress those coefficients increase (more input is required to produce one unit of output). In all physicalist interpretations of Marx’s theory technological progress can not depress the rate of profit. There the rate of profit can only fall because of technological regress (if we regard the real wage as constant). That’s were physicalism is obviously at odds with Marx’s law of the tendential fall in the rate of profit because of technological *progress*.

    Second, let’s take a look why the macro-monetary rate of profit has fallen in the quoted example.
    The answer is it has fallen because the economy has more productive before the “technological change” — before the change a1 was 0.56, but afterwards it increased to 0.67! So we have a technological regress.

    This again means something, i.e. it exemplifies that the macro-monetary model combined with simultaneous valuation or pricing follows the same physicalist logic as the other physicalist models/interpretations.
    We should not forget about this example when we consider the “the all-important question of the effect of labor-saving technological change on the rate of profit” (comment on part 4@22nd Jul 2016 9:41 am).

    So, I hope I could clarify (a bit) why your arguments did not convince me.

    I’m sure that the your macro-monetary interpretation should not be a physicalist one and should not in case of a technological progress/regress follow the same logic as a physicalist one, but as far as I can tell it (unfortunately because of simultaneous valuation/pricing) nevertheless does.

    I have not yet read your “summing up” document – I’ll do so in the next days.

    So long,

  7. Dear Michael Schmid,

    I’m your predecessor in this comment list, have written a review of ‘Money and Totality’ in ‘Marx-Engels Jahrbuch 2015/16’, have a track record in (telecom) system analysis, modelling and design, and, while you were writing your comment to Fred M., was mainly busy with pushing my granddaughter on the swing in our garden.
    From start I applied constant push energy and after a while the swinging movement did not become stronger, i.e. the system obviously was in a steady or equilibrium state. When repeating the same with more push momentum Em, the steady state movement was also stronger, i.e. more swinging energy Es was in the system. In short: Es = f(Em).
    On a swing, energy oscillates between potential form Ep and kinetic form Ek. In steady state we get Ep + Ek = Es. Now, for simplicity reasons, let’s just consider a swing deflection where Ep = Ek. Then we get the following equilibrium equations: Ep = mgh = Es/2 and Ek = mv²/2 = Es/2. With m = ‘mass of granddaughter + swing’ given, we can easily calculate h and v from Es. And, as in steady state – where input energy equals output energy – model parameters are hard-wired, as you stated, Michael, one also easily can make the “back-calculation”. In our swing system there are even two options: Es = f(h) = 2mgh or Es = f(v) = mv².

    Though in both, Es = f(Em) and e.g. Es = f(v) Es appears as a ‘function of something’, there is a significant difference:
    a) Es = f(Em) shows, how a certain periodical cause Em leads to an effect Es; whereas f(v) is only another expression of ‘energy’, i.e. energy expressed as a product of m and v².
    b) Es = f(Em) comprises the process how over time Es is created. Es = f(v) is only about results, with the cause of their genesis being wiped out.
    c) In Es = f(Em) Em is a true free variable. It’s choice fixes all the other parameters. In Es = f(v), v is not a free variable at all. It’s likewise fixed from Em, as all the other parameters.

    So, equilibrium equations are good for nailing down equilibrium conditions. They are useless for cause-effect explanations. Even worse. Owing to their formal similarity with functions of ‘explanatory’-type they open the door for quite some scientific misuse, when poor modelling is applied.

    Let’s make a model of the swing system. What shall be its scope? i) at least one complete steady state swing? or ii) only the deflection moment chosen above? In i) Es = f(Em) is in. In ii) it’s out, making this a model with very limited explanatory power.
    From the fact that steady state modelling leads to simultaneous equilibrium equations, Andrew K. has drawn the wrong conclusion that steady state modelling as such is devil’s work: simultaneism, Therefore he is convinced: i) is impossible. Steady state postulate and ii) are equivalent. So, if someone (let’s call this person Fred) stipulates input=output, for Andrew in order to criticize this guy only ii) needs to be considered. And here Andrew silently first calculates physical units and afterwards loudly recalculates Fred’s macro-monetary quantities (obviously for Andrew the way in which he starts the calculation is very relevant). Both calculation directions take place in a world of frozen results only, i.e. have no explanatory cause-effect relevance. Therefore none is a veil for the other. What is a veil, however, is that with wrongly restricting the consideration to ii) Fred’s derivation of macro-monetary values is excluded from consideration.

    Still a word to ‘full automation’. It think it’s true, that within Fred’s model when applying Andrew’s calculations and the profit is 0 then there is no physical surplus.
    This is however not true in Sraffa’s hypothetical model.´The fact that (back to swing model) the v in f(v) is not a free variable is not known to him. I.e. he is free to select any values of v (or say physical units) he wants, regardless whether the possibly of a real economy, i.e.
    system the fits to his model, exists or not. So indeed in his model physical surplus would be possible, but only there (see also my comment before yours).

    So long,

Leave a Reply

Your email address will not be published.