After posting my talk Why the Labour Theory of Value is Right on Youtube a series of comments were made about it on Facebook by Matthew Harris. Below I give the exchange which took place on Facebook. As it was getting too long for posting in that format I have prepared this text. The blue text is that put up by Harris, the red is my replies.

Matthew Harris

There are many different interpretations on the Labour Theory of Value (LTV) and while its common to associate it with Prices, as Cockshott does, this is not necessarily correct. Many will point out that Labour value and *exchange value* are not the same thing, which is supposed to be one of the main contradictions of capitalism. They state that Labour produces surplus value in the form of goods and services, and that this is wholly different to the anarchy of exchange prices.. This presents a small problem for “scientific socialism”, either 1) LTV explains prices which are a very important piece of the capitalist order, or 2) LTV does not explain price and it does not have a complete view of capitalism. If it is associated with price, such that average labour time equals average price of the commodity, then all Marx has stated is that a product’s price will be close to its marginal cost, which is only true in perfect competition. This is demonstrably false in any corporate merger. When a merger occurs, nothing changes in production but the value of the combined company is worth more than the sum of their parts. Additionally, if you ask corporations how they price their output, labor costs are often not even part of the equation anymore. They simply use the equation banks do for predicting the future value of a loan and swap the variables to instead get present value (the discounting equation). This also means Supply/Demand curves are wrong. I like going to Subway and seeing a ham sandwich and veggie sandwich at the same price because one obviously has a higher cost (the ham included) but its never part of the price.

Cockshott is using Macroeconomic data to justify what is a Microeconomic theory. The issue with this is that you can have vastly different corporate sizes within the same industry, such as McDonalds vs a small family restaurant. You wouldn’t claim McDonald’s super-profits to come from ricardian rent the same way you would Oil companies, but it exists all the same. Averages often hide these kinds of inequality which are very important to capitalism, after all the goal of many small businesses is to eventually become big and not have their prices close to labour costs. Cockshott also fudges his data at 14:50, if you look at the table it states the correlation is only tight for only 47 industries in the USA, 33 in Germany, 37 in France, and he used UK data which correlates with 101 Industries, the most out of all of them. All in all though, Paul Cockshott is saying that economic competition is irrelevant to price when in fact its the driving force. Additionally, if prices are so tightly correlated to labour costs, wage increases would always be inflationary and this is not the case. My sources are Nitzan’s “Capital as Power” and Myatt’s “The Economic Anti-Textbook”.

Cockshott

Matthew Harris seems to think that there is fudging of the data because the number of industries given for different countries are not the same. That is because different countries use different classifications and degrees of aggregation in their published tables. There is no data selection or fudging going on. The methodology is in the open literature, was orriginally developed at the New School in New York, and has been independently replicated by many Marxist economists since then for different countries.

Mathew Harris  

Thank you for the reply. You still never responded to the use of macroeconomic data to justify a microeconomics theory. That’s a huge methodological gap and the data presented can show no causality.

Cockshott

The difference between micro economics and macro economics is anachronistic when applied to classical political economy, it is a terminology developed in Bourgeois economics post Keynes and refers to on the one hand Marshallian theory ( micro economics ) and Keynesian theory (macro economics). Marxist theory is neither of these. Capital is devoted to explaining the processs of class exploitation under capitalism. As such it deploys an analysis of the value creation process and specifies that value creation will be proportional to to direct and indirect labour added. Strictly speaking all that is required to demonstrate this is to look at the labour content versus the monetary output values of three sectors, elements of constant capital, wage goods and capitalist consumption goods. This is how the reproduction schemes, an early type of input output table, in vol 2 work. The research by marxist economists since the 1980s has gone well beyond 3 sector models, including models with up to 100 sectors, easily enough to establish the marxist thesis. Data do not show causality, theories give proposed causal mechanisms, and if they are scientific theories, they make falsifiable predictions. If the predictions are born out, then the proposed causal mechanism gains credibility. Marxist theory is subject to falsification, Marshallian supply and demand theory is unfalsifiable and thus is pseudo science.

Harris

That the terminology was developed by bourgeois does not scientifically disprove its usefulness as a catagorical distinction. It is not merely a separation between marshal and Keynes, its about aggregate and disaggregate which is very important in all sciences whether it be physics or biology. There are significant differences between quantum physics and it’s larger counterparts, just as there are significant differences between microbiology and studies on ecosystems. And while you spent a long enough time saying data does not show causality, that theories do, once again you provided no theory which closes the gap, which is what was being asked of you. The part about supply/demand being unfalsifiable is true, but sidestepping the issue. Your theory is based on averages, which do not actually exist in life. Firms and wages within the same industry can have wide variation in size. An average only masks this inequality and inequality is the driving force of capitalism. Oil companies are not an outlier based on Ricardian rent through the Capital as Power interpretation of value. Anyway, why is LTV a sectoral analysis? Why didn’t you take a look at the level of individual firms? Scale is a very important thing to consider when it comes to a theories applicability. If prices don’t match up with labor time on a micro level, how do they manage to do this on a sectoral basis? Adding ham to a veggie sandwich has no cost to the customer, but it had a labor cost. LTV interpretations of price can’t explain anything about marketing tactics likes sales and coupons, which do effect price.

If Ricardian rent is considered valid, what’s the effect of conglomeration on price and thus labor value? General Electric is not merely a utility company, they are also invested in journalism. The Japanese economy has many large corporations that also have banking services as part of what they offer. If rent is at all part of the corporate portfolio, the price of non-rent goods and services can be subsidized. I suppose that also brings up the issue of government subsidies which are considered a kind of rent and common. This would theoretically distort prices from their Labor values, but your data does not show this. Why does everything just seem to work out on a sectoral level when there is quite a bit going against it on a firm by firm basis?

 

Paul Cockshott 

Harris says that the distinction between microeconomics and macroeconomics is not just the difference between Marshallian and Keynesian theory, but the difference between aggregate and disaggregated objects of study.

The reason why I raised Marshall is that the entire topic of ‘microeconomics’ as taught in colleges derives from Marshallian supply and demand analysis. Harris conceeds that this is unfalsifiable and not a coherent theory, but it is this theoretical structure that constitutes microeconomics as a subject of investigation. It is not much help to bring in analogies with the quantum theory here, as Harris does.

Harris makes the analogy of quantum theory with microeconomics and also implies that both of these deal with disaggregated phenomena. Both the analogy and the implication are wrong.

Quantum theory is a scientific theory which is testable and falsifiable and which has huge empirical support. Marshallian theory is unfalsifiable since it is a hidden variable theory with more variables than observables. At any given point in time all you observe is the quantity of a commodity sold in the past week, month etc, and the average price at which it was sold. In other words you have two observables – both of which contra Harris are aggregates: aggregate number of Volkswagen Polo’s sold in last month, average price they were sold for. Micro economics explains these TWO observables by inventing at least FOUR hidden variables : the intercept of the supply function, the gradient of the supply function, the intercept of the demand function, the gradient of the demand function. That is for the simplest possible linear supply and demand functions. If we assume that the functions are curves, then in principle you need a full Fourier expansion of the curves, giving a lot more hidden coefficients.

Since there is no independent way of determining the parameters of the curves, the curves themselves are not observable, any combination of prices over time is allowed. One just invents unobserved shifts in the supply and demand functions to explain it.

Unlike microeconomics, quantum theorists have been loath to introduce hidden variables, and since the landmark work of Bell (Bell, John S. “On the problem of hidden variables in quantum mechanics.” Reviews of Modern Physics 38.3 (1966): 447.) hidden variable theories have been rejected.

Let us then turn to the issue of aggregation versus disaggregation. The quantum theory is explicitly a theory about aggregates. If you take a classic double slit experiment, the impact points at which individual photons are detected is, according to the quantum theory, unpredictable. All that the theory can predict is the mean number of photons that are likely to impact in slices at different distances from the median line. The quantum theory is thus exactly what Harris objects to, a theory about averages and aggregates. The big change in physics, starting with Boltzman and developing through the quantum theory, was that it became a theory of stochastic processes. Although in a stochastic process individual events are unpredictable, the statistical properties  of the distributions of events are predictable.

The classical economists always implicitly took a stochastic approach, they were concerned with what they sometimes spoke of as the centers of gravity of prices and the laws which governed these. An actual stochastic formalisation of the labour theory of value had to await the work of Farjoun and Machover (Farjoun, Emmanuel, and Moshe Machover. Laws of Chaos: A probabilistic approach to political economy. Verso, 1983.). They explicitly model their theory on stochastic physics. I gave a quick summary of it in my video. Harris slights this as a ‘theory about averages’ and he says that ‘averages do not exist’. Let me examine that:

  1. Farjoun and Machover’s work is not just about averages, it is about statistical distributions characterised by both means and standard deviations, It is by making predictions about the mean and standard deviation of ψ the price to integrated wage probability density function, that they are able to deduce that prices will be highly correlated with labour values. The important point is that they explicitly model the standard deviations. Once you do this, you get strong predictions from the theory.
  2. Harris says that averages do not exist. Well if he means that you can not validly reduce a distribution to its mean, then he is quite right, but the Farjoun and Machover theory on which I relied does not do this. Its predictions come from arguments about the degree of dispersion of the economic variables.

So contra Harris, it is the statistical marxists like Farjoun, Machover, Wright etc that are actually basing themselves on the methods of modern physics. In contrast orthodox micro economics is like the aeolian theory of the winds. Its four hidden variables were the four wind gods Why does the wind today blow from the East, because Aeolus commands the Eurus to blow from the East. If yesterday it blew from the West, it was because Aeolus commands  Zephyrus of the West to do his duty etc.

Just as any combination of sale and price can be explained by the actions of the invisible supply and demand curves, any particular wind direction can be explained by how hard the different wind gods are puffing.

image1

Since invisible gods are hard to visualise, iconography helps. Look at an icon for long enough and you imagine that what it portrays is real. Get economics students to look at pictures of supply and demand curves for long enough and they think these are real.

Minor points

Harris asks why I and other Marxists only use data for industrial sectors rather than firm level data. The point is that you can not compute total labour content with firm level data. You need to know the labour content of the inputs used by the firm, which are not available. In a socialist economy the planning agency would have the capacity to calculate this, but in capitalist countries the state only publishes sector level I/O statistics. It would be great to have the level of detailed information to do what Harris asks, but we do not have it.

 

Addendum – some more data

The graphs below illustrate how poorly some other ‘natural’ inputs predict market prices. Using  the same algorithm as used to compute labour values one can compute integrated electricity, iron, computing coefficients for each UK industry. It is clear from the plots that these alternative value bases have very poor explanatory power as compared to the labour theory of value.

moneytoaltbaseslog

moneytoaltbaseslin