The Labour Theory of Value is Correct and Important

Alexander Douglas
Genus Specious
Published in
9 min readOct 9, 2023
A poster showing workers in front of a plant with the words “LABOR CREATES ALL WEALTH” above it.

I’m returning to the Labour Theory of Value (LTV) after writing a series of posts about it here, here, here, here, and here.

Writers on Marx often claim that the LTV isn’t necessary to understand his story about capitalism, the nature of exploitation, etc. A recent example is found in this blog post by Professor Liam Kofi Bright at the London School of Economics. I think people say this because the LTV is supposed to have been refuted, and they don’t want to damn the rest of Marx’s theory alongside it.

I believe that (i) the LTV is indispensable to Marx’s story and (ii) this shouldn’t worry those who believe his story, because the theory is correct. The refutation rests on a false assumption.

What is the Labour Theory of Value?

By the LTV here I mean the theory that the exchange value of a commodity — any good or service — is determined by the amount of average labour time that goes into making it and making any inputs into its production (for example, extracting the raw materials that it is made from, or building the tools and machinery used in making it). Call this its “labour content”.

Exchange value is something like a long-term average of price, not the price at any given moment. Bright gives an example: “the price of shovels goes up when a snow-storm is anticipated even though the anticipation of a snow storm has done nothing to change the socially necessary labour time involved in producing those shovels”. The LTV predicts only that the price of snow shovels will, in the long run, gravitate around a value determined by the labour content, not that it won’t fluctuate in the short-term due to changes in demand.

Why Is It Needed?

The LTV is very important for Marx’s story of exploitation, I think. G.A. Cohen argued that it isn’t, but his argument is invalid in my opinion (see here and here).

Cohen and other “Analytic Marxists” essentially argue that we can reconcile much of Marx’s analysis of capitalism with the neoclassical theory of value. I disagree, because I believe that the notion of exploitation is fundamental to Marx’s analysis, and exploitation must mean here something simple and non-normative, e.g. that workers receive less in wages than the value of their labour.

This is easy to calculate on the LTV. Suppose that I, a capitalist, pay a group of workers wages worth 1000 hours of average labour-time and supply 200 hours’ worth of fixed capital (tools, raw materials, etc.). If I sell the resulting products for 1400 hours’ worth of money, then I have made a profit of 200/1400 (around 14%). But then the commodity must, by the LTV, have a labour content of 1400 hours. Since I only spent 1200 hours’ worth on costs, where did the extra 200 hours come from?

The price I paid for the fixed capital must, on average, be proportionate to its labour value, since this is true of all commodities according to the LTV. So I must have underpaid for the workers’ labour. They must have supplied 1200 hours of labour, even though I only paid them wages worth 1000 hours.

But isn’t labour also a commodity? Shouldn’t it, therefore, also be subject to the LTV? Here Marx’s answer is complex: the workers do not sell their labour to the capitalist; they sell their labour power. Labour power is the capacity to work for as many hours as physically possible given the “input” of a certain wage. Since my workers could work for 1200 hours while consuming wages worth 1000 hours, buying their labour power allowed me to extract value from them. This is fundamentally how capitalism works, on Marx’s theory.

It raises many questions. For one thing: why don’t workers sell their labour instead of their labour power? Everyone else in the economy demands equivalent value in exchange for what they give up: if the workers give up 1200 hours, shouldn’t they demand 1200 hours? A naive Marxist will reply that the workers under capitalism are too oppressed to make any demands, but the question is why. You might have thought, after all, that competition among capitalists would bid up wages for a limited supply of labour.

Marx’s answer to this involves a complex story about the distribution of power under capitalism, the dynamics of the population of workers and employment cycles, the theory of crisis, etc. The problem with this is that, in my opinion, the more complex a story is, the more limited its application. If the conditions of capitalism require a specific arrangement involving the legal power of workers, a certain population dynamic, a particular pattern of business cycles, etc., then this muddies the definition of capitalism, and working out whether Marx’s analysis applies to a given situation becomes a very vague matter.

No Exploitation Without LTV

What is clear, however, is that the LTV is crucial to the story. Cohen suggests, by contrast, that we can define exploitation using a neoclassical (or, he says, any other) theory of value. Just say that the workers produce the whole of the product but are paid only some of its value in wages.

But, in the first place, it isn’t clear that the workers are in any way disadvantaged by this, if we aren’t sure that labour is the sole source of value. The capitalists can say that while the workers produced the product, the capitalists supplied the capital necessary for production and are thus entitled to some of its value. The landlords can likewise say that they, the landlords, supplied the land on which the production took place, so they are also entitled to some of the value. The value of the product, on a simple textbook model, will be a function of the factors that went into it — land, labour, capital — so splitting the proceeds among the workers, capitalist, and landlord might well count as giving back everyone the value they supplied.

Secondly, on a neoclassical theory it can be argued that the workers are paid the full value of the product, even if e.g. the product sells for £1200 net of non-labour costs and the workers are paid £1000. That is because the workers were paid before or during the period of production, but the product was only sold at the end of the period. The rate of time-discounting enters in here, and if it is 16.6% over the given period then the workers can be said to have received the full (present discounted) value of their product. The capitalist reaps the rewards of patience, and there is no exploitation.

John Roemer also attempts to square Marx with a neoclassical theory of value, and ends up concluding that Marxists shouldn’t be interested in exploitation.

Why Is LTV Correct?

The neoclassical theory, however, is largely devoid of empirical support. The most sophisticated version of the neoclassical theory of capital involves intertemporal general equilibrium models, whose disconnection from any empirical evidence was demonstrated long ago by Dan Hausman.

The LTV, by contrast, is very well-supported empirically, as has been shown by Anwar Shaikh, Paul Cockshott and Allin Cottrell, Lefteris Tsoulfidis and Thanasis Maniatis, Dave Zachariah, and many others.

The key to confirming it empirically is to treat it as what it is: a theory of a statistical equilibrium. Prices can be all over the place, but they will on average tend to be proportional to labour content, although any given price can be much higher or lower than this. Such a statistical correlation calls out for a causal mechanism to explain it, and a mechanism is suggested by the mathematicians Emmanuel Farjoun and Moshé Machover in Laws of Chaos and by a team of engineers and physicists in Classical Econophysics (ch.9).

What they do is use probabilistic methods to study value as an emergent statistical property of a very random system operating under certain constraints, which looks much more like a real economy than the stable solutions of neoclassical economics (human choices, for example, are better modelled by random noise than they are by rational optimisation).

The truth of the LTV as a matter of statistical averages was also recognised by Joan Robinson, who wrote, in Economic Philosophy (1962) that in manufacturing industry “differences in prices are more or less proportional to labour cost, and on the other hand they are not exactly proportional”, for a variety of reasons. “What”, she asked, “is all the fuss about?” The LTV holds up very well as a general predictor of relative prices, so why has there been so much controversy about it?

The Transformation Non-Problem

Marx’s theory of exploitation is sometimes rejected on account of something called the “transformation problem”, which arises from the LTV along with the assumption that competition should drive different industries to have the same rate of profit even though they use differing proportions of labour. The idea is that if different rates of profit are available in different industries, capitalists will move their investments out of the lower-profit ones and into the higher-profit ones until all profit rates are roughly equal. This will lead to prices no longer matching labour content.

Farjoun and Machover have explained that while scholars took the LTV to be the culprit in the transformation problem, there is in fact no good reason to accept the assumption of a uniform rate of profit, equalized through competition. As they put it in a later work: “the very mechanism of equalization — competition and capital reallocation — also creates irreducible motion of firms away from a state in which all profit rates equal the average”. It is true that self-interested capitalists will move out of lower-profit industries into higher-profit industries, but as a result prices and quantities will be constantly changing across the economy and profits will end up forming a fairly random distribution around a certain average.

Exploitation and the LTV

Where this leaves exploitation is interesting. Farjoun and Machover do not take on Marx’s theory of “labour power”, with all its heavy commitments. But this then raises the question of why, if prices generally correspond to labour content, the same shouldn’t go for labour. Why isn’t the price of an hour’s labour, on average, a wage worth one hour’s labour?

If it were, the average profit rate would be zero. By the time the capitalist had paid all the workers for their hours of labour and all the sellers of all non-labour inputs for the hours of labour embodied in those inputs, she would have paid the full value of the product sold and have nothing left for profit. Again, all of this is on average. Some capitalists would earn positive profits and others would lose money, but it would all average out to zero, and capitalism would be unviable in the long run.

To see why labour is not paid proportionately to its value, on average, we can compare it to another commodity in this probabilistic theory of prices. The price of an ordinary commodity-type will form a normal distribution around an average determined by labour content. Some commodities of that type will be sold for much more than their value, others will be sold for much less, but the bulk will be sold for something pretty close to their value.

For labour, however, there is another constraint. Labour cannot be sold for too little, otherwise workers will starve. This means that the distribution around value of labour prices (wages), will have a positive skew: the curve will be squashed up against the lower bound. As this post very helpfully explains, in a distribution with a positive skew, the median value (in this case, the hourly wage that most people are paid) will be lower than the mean (the actual labour-value of labour).

This is an interesting result, because it suggests that it is a fundamental feature of capitalism, quite broadly defined, that it involves exploitation in the sense that workers are systematically paid less than the value of their labour. Such a consequence is simply the statistical outcome of a system in which prices are determined in exchange and a significant proportion of the population need to sell their labour in order to survive.

Some workers, of course, are paid more than their value. If they were paid less — that is, if worker incomes were more equal—then the distribution would be squashed less hard against its lower bound and the median would move closer to the mean. More workers would be paid something close to the actual value of their labour.

If you want to know whether you’re part of the problem, Farjoun, Machover, and Zachariah estimate (ch.6) that the proportion of workers earning more than the value of their labour in wages is something like 10%. So that’s you, if you’re in something like the top 10% of earners. In the UK this is anyone on a yearly salary of around £65,000 or upwards. You could reduce the rate of exploitation by taking a pay cut. But eliminating exploitation would require a completely different system.



Alexander Douglas
Genus Specious

Lecturer in Philosophy, University of St. Andrews — personal website: