Capitalism produces a bizarre, paradoxical form of often persistent unemployment: the workers exist able and willing to do the work, the need for the product of the work is acute, and the resources required for the workers to do the work are also plentiful, but, somehow, it is impossible to link all those things together to actually get the workers employed and the work done, for the mutual benefit of all. Such paradoxical unemployment has been observed throughout modern history: the phrase “Great Depression” has been used many times, not only for the 1930s, but also, for example, sometimes to refer to the “Long Depression” from 1873 to 1896, or the “Second Great Depression” following the Global Financial Crash of 2008. And, of course, it is about to be seen again.
Sometimes, the problem of paradoxical unemployment is so acute that it requires a massive stimulation on the part of the state to restore something approaching full employment. In the most famous “Great Depression”, that of the 1930s, even Roosevelt’s New Deal alone was not sufficient to solve the problem for the US economy: instead it was the demands created by needing to support the war effort that finally lifted the US economy out of depression and created the prosperity of the post war years.
How can such a counter-intuitive and dysfunctional outcome arise? What causes this strange “paradoxical unemployment” and why is it such a prominent feature of capitalism? In truth, the answer has been known about for at least two centuries, as an accepted aspect of mainstream economics. It is just not talked about in polite company, and its implications are quite definitely not confronted.
The explanation lies in the meaning of capitalism itself. Capitalism doesn’t mean an economic system based on “free “ or open markets, because the ultimate dream of many capitalists is to capture markets, to create monopolies or oligopolies. Equally, socialism doesn’t necessarily imply detailed central planning, and there are many forms and variations on “market socialism”. Rather the distinguishing feature of capitalism is the control of “capital”, of the things people need to live and make a living, by the few and not the many. Such a state of affairs is far harder to defend than the fact that the market mechanism can sometimes be useful, which is why markets and capitalism are deliberately muddled together in popular discourse by the apologists for capitalism. It is that profound inequality in access to capital that makes the recurrent and persistent creation and perpetuation of paradoxical unemployment an inevitable consequence of “capitalism”.
It’s also the case that exactly how inequality in access to capital causes paradoxical unemployment has been known for at least two centuries. It is discussed in the works of Thomas Malthus, the infamous prophet of “Malthusian catastrophe”, specifically in his “Principles of Political Economy”, published in 1820, where he says:
“…no productive labour can ever be in demand with a view to profit unless the produce when obtained is of greater value than the labour which obtained it. No fresh hands can be employed in any sort of industry merely in consequence of the demand for its produce occasioned by the persons employed. No farmer will take the trouble of superintending the labour of ten additional men merely because his whole produce will then sell in the market at an advanced price just equal to what he had paid his additional labourers. There must be something in the previous state of the demand and supply of the commodity in question, or in its price, antecedent to and independent of the demand occasioned by the new labourers, in order to warrant the employment of an additional number of people in its production.” (REF 1)
Those deductions are then explored in the works of Karl Marx, where they are extensively quoted in part 3 of his “Theories of Surplus Value”, published in 1863 (REF 2). Now Malthus is usually associated with the right-wing of social and economic debate, Marx, of course, with the left. But a similar idea forms part of neoclassical economics, normally associated with the right of the debate, as the concept of the “Marginal Revenue Product of Labour”, or “MRPL” (REF 3)
The marginal revenue product of labour model is an application to labour of the more general economic principle that an enterprise will increase its output up to, but not beyond, the point where the money it makes from that extra unit of output less the cost of producing that extra unit of output is zero, or, in the jargon, the point where marginal cost = marginal revenue. That idea is itself a generalisation of the law of diminishing returns. It relies on the fact that, over what economists call the “short run”, meaning over an interval of time sufficiently short that an enterprise cannot expand it’s basic means of production, so, for example, a farmer could not buy new land or a factory owner new industrial plant, extra production will gradually produce diminishing returns, such as the farmer exhausting the soil, or the machinery in the factory wearing out too quickly.
Note that this notion also assumes that that the enterprise in question is also too small a player in its markets, in terms of its outputs, for the level of its output to effect demand and thereby price: for example that, over at least the short run, the enterprise could not depress prices by flooding the market with too much output.
The idea of the MRPL is then simply that the same argument can be applied to labour, for example, that if a restaurant employs too many cooks then, without making the kitchen bigger, which is something economists would say only happens in the “long run”, the cooks will start to fall over each over and a point will be reached when the revenue produced by hiring an extra cook will be exceeded by the cost of hiring the extra cook. (REF 4)
The upshot of all this is therefore as follows: a capitalist will only hire workers up to the point where it ceases, for the capitalist, to be profitable to do so (the point where the marginal cost of that extra worker is equal to the marginal revenue they generate). How quickly that point is reached will depend on the demand for whatever is being produced. That demand in itself will reflect general economic conditions. If there is high unemployment there will be widespread poverty and lower demand.
Here it is crucially important to note that no individual capitalist can simply, as Malthus observed, make it profitable for themselves to hire more labour by simply creating more demand through hiring more workers. The capitalist can, arithmetically, never receive more money back from the worker, through that created demand, than they pay the worker in the first place, unless the worker has other sources of income and is very reluctant to buy from anyone except their own employer. Indeed, neoclassical economics could not assert any validity for its concept of the Marginal Revenue Product of Labour if it was possible for capitalists to demand drive their own profits by hiring labour in this way (REF 5). Therefore, the possibility that unemployment can reinforce itself, that workers can be unemployed simply because they lack access to capital and can’t be employed profitably by those who control that capital, because of the lack of demand cruelly caused by the unemployment of all in their situation in the first place, follows directly from MRPL and the neoclassical economics ironically often used in defence of capitalism itself.
Of course, in what economists call the “long run” widespread unemployment would most likely force down wages, reducing the cost of labour and thereby increasing the amount of labour that can be employed until the point where the marginal revenue generated by an extra worker less the marginal cost of that extra worker is zero. Except there is no guarantee that the cost of labour needed to allow the capitalists to employ all the workers, in a way that is profitable for the capitalists, will be above subsistence levels, especially as falls in wages for those actually employed will still further reduce demand, and therefore prices, and therefore the marginal revenue that can be generated by each additional worker.
Even worse, capitalists will continually seek to maximise returns by minimising the cost of the labour they need to use. In the long run that could mean more automation, but, as soon as societies start to trade and use money to create stores of value and facilitate exchange, it has meant many other “innovations” too. Self-sufficient societies, at least those that lack advanced mass automation, need to use a considerable amount and variety of labour to meet all their needs, but money and trade open up the possibility of specialising to “work” capital in non labour intensive ways, for example, in the Highland Clearances, evicting tenant farmers to graze sheep for wool production instead.
Consider, for example, the limiting case, a theoretical very “long run”, where the owners of factors of production have eliminated all labour costs, because they can produce outputs using only robots, and even the robots can make other robots. At this point anyone who owns sufficient means and factors of production uses those means and factors of production to satisfy their own needs. Anyone who does not is not only unemployed, but starves. (REF 6)
If the capitalist cannot profitably hire workers, because of a lack of demand caused by high unemployment, then that lack of demand could be fixed by reducing unemployment. But that general reduction in unemployment is impossible because it isn’t profitable for any of the individual capitalists to hire more workers. Low demand sustains the unemployment that causes the lack of demand in the first place, in a vicious circle. The result is paradoxical unemployment, the strange combination of jobless workers able and willing to work, unfulfilled real needs and “capital” laying idle, that can only be resolved by a general stimulus to demand, usually eventually provided by the government of a state, because only state-wide institutions can exert a widespread enough effect roughly simultaneously across a broad sweep of the economy (in this sense escaping the paradoxical unemployment trap is also an example of what economists call a “coordination problem”: the capitalists could also escape the trap if they all undertook to hire at the same time). Sadly, this kind of demand stimulus often only happens in response to war.
Now let’s again envisage a situation where paradoxical unemployment could arise, so the things that people need have the workers needed to produce them, and the capital also exists that the workers need to do the work, as it is the fact that all the ingredients are in place and in potential alignment but the economy still can’t “produce” that makes paradoxical unemployment “paradoxical” in the first place. Let’s make one small difference though: now let’s assume that all the workers have access to the means and factors of production. Now everyone can use those means and factors of production to work to satisfy their needs, either by trading what they produce with others or using their output directly themselves. Provided workers retain access to capital the connection between the workers, the things they need to do their work and their own needs is maintained, and paradoxical unemployment becomes impossible.
The approaching mass wave of AI driven automation, on top of economic shocks delivered by pandemics and climate catastrophe, mean that mass deprivation can only be avoided by shifting towards a model where the many control the things they need to live and make a living. If we don’t reorganise our economy in that way, the Great Depressions of the past will pale before the disasters to come.
As we rebuild our economies after COVID we must therefore give careful consideration to all forms of economic democracy: worker-owned enterprises operating in open but well-regulated markets as well as more democratic public ownership at all levels. We must give more legal recognition and protection to cooperative arrangements and make it easier for worker-owned enterprises to raise investment without permanently alienating ownership to a capitalist class. Where state aid is given to rescue firms, provision for extending worker ownership should be part of the deal. Without giving the workers themselves a say, any protection of their jobs is likely to be transient.
Allowing the few to hoard capital for their own profit can lock workers out of access to the capital they need to be productive and earn a living. Workers eager and able to work can be unable to work simply because they lack the things they need to do their work, such as factories, machinery, warehousing, raw materials, tools, vehicles, office space, retail space, land and “real estate” in general, IT or networks of suppliers and customers. At the same time they can’t be hired by capitalists, who do own and control such things, because there is too little demand for what those workers could produce, not because there isn’t a real, desperate need for the products of those workers, but because those with the needs are poor, because, in turn, they are the very workers who don’t have the available capital to do their work in the first place. It is a catch-22: a trap constructed out of “paradoxical unemployment”.
Paradoxical unemployment arises because of two very obvious aspects of capitalism: the inequality in access to capital and the fact that capitalists work for a profit, combined with one subtler deduction, which is the notion captured above in the thoughts of Malthus and in the MRPL, that there is no reason to suppose that the mass of capitalists, each hiring labour to work their own individually controlled capital for their own profit, will necessarily produce full employment, even if all workers are eager for work and have useful skills. As these subtler generalisations are part of mainstream economics the following conclusion is unavoidable: it’s official: it’s been official since at least the days of Malthus: capitalism causes unemployment, and of the most bizarre, pointless kind. Just try not to say so in polite company. (REF 7)
Notes and Sources
REF 1:
https://oll.libertyfund.org/titles/malthus-principles-of-political-economy , specifically: https://oll.libertyfund.org/titles/malthus-principles-of-political-economy#Malthus_1462_945
REF 2:
https://www.marxists.org/archive/marx/works/1863/theories-surplus-value/ , especially https://www.marxists.org/archive/marx/works/1863/theories-surplus-value/ch19.htm , Section 12: “The Social Essence of Malthus’s Polemic Against Ricardo. Malthus’s Distortion of Sismondi’s Views on the Contradictions in Bourgeois Production”
REF 3:
See: https://courses.lumenlearning.com/boundless-economics/chapter/demand-for-labor/, noting especially that “Since the price of the output is a component of MRPL, changes will shift the demand curve for labor. If the price that a firm can charge for its output increases, for example, the MRPL will increase”).
REF 4:
But what if the capitalist just makes the kitchen bigger? Then the Marginal Revenue Product of Labour threshold goes up, as each worker is more productive, so it is profitable for the capitalist to employ more workers, and Malthus’s objection ceases to apply. Does this offer a way to escape the paradoxical unemployment trap? After all, there is lots of skilled labour available, though the reality of mass unemployment will also be depressing demand and prices.
In the above example, of the restaurant enlarging its kitchen, either it is acting alone, in which case the restaurant, continuing the “short run” assumption that it is a small player even in its local market, does not depress prices with its own additional output, so will make more money by selling more meals. But how far can we extend this process? In fact we are incorrectly allowing a “short run” assumption, the fact that the restaurant can increase its output without affecting prices, to continue into the long run, but in the long run the restaurant, if it continues to expand, will start to affect prices in its market with its output, thus lowering the MRPL threshold, and reaching a new equilibrium point where it isn’t worth hiring any more workers. On the other hand, if the restaurant does not continue to expand, it will have negligible affect on employment.
What we need to escape the paradoxical unemployment trap is something that encourages each capitalist to individually hire more labour, but that also applies to many such capitalists at the same time (to resolve the “coordination problem” inherent in the paradoxical unemployment trap), so that the prices for each enterprise’s increased output are better sustained by generally increasing demand.
Of course, perhaps the restaurant does not act alone. Perhaps an opportunity for expansion, say an innovation, gives a range of enterprises, or a whole sector of the economy, a chance to expand all at roughly the same time. In this instance, whether or not that expansion actually propels the economy out of the paradoxical unemployment trap, will depend on whether this stimulus, this injection of investment, is sufficiently large and widespread. If not, again, a new MRPL threshold will be reached at a point where there is still widespread locked-in unemployment.
To understand this point we have to consider that the spending associated with the extra output will initially produce demand amongst those associated with the range of enterprises or sector in question, especially their workers and suppliers, and the workers of those suppliers. However, that extra demand won’t all be spent on the increased output; rather it will be spent on a basket of goods and services produced by the whole economy. Therefore the extra output will also induce a fall in prices that will somewhat counteract efforts to increase the MRPL threshold. The effectiveness of increasing output as a way out of the paradoxical unemployment trap therefore depends on the sweep of the economy that finds itself able to profitably increase its output at roughly the same time, and escape is only guaranteed if that sweep is the entire economy.
If we want to, we can explore this as a toy mathematical model:
The extra output should, as we have seen, produce extra income for those involved with this range of enterprises or sectors, in terms, for example, of more workers receiving wages and increased profits. This will produce a change, an increase in demand, or a “delta” demand, or Δ demand following the convention of using the Greek letter for “delta”. As the associated Δ demand (the monetary receipts associated with producing the output) will be spent on things other than the Δ output, and holding the Δ demand of the rest of the economy as zero, the Δ demand that can be spent on that Δ output = Δ demand multiplied by the proportion of total demand that is satisfied by that part of the economy that is generating the increased output.
Now, let’s make demand for our enterprise or sector’s output = D, change in that demand as Δ D, original output as O and change in output as Δ O. The new average price for each unit of that part of the economy’s output will be related to:
(D + Δ D)/(O + Δ O)
But what will Δ demand be?
On the assumption that the increase in demand is a result of the spending associated with the Δ output, for example items such as wages paid to workers, then Δ demand will be proportional to output.
Let’s say Δ demand = r Δ output, where r is just a constant that allows us to convert from “output”, and however that is measured, to demand, however that is measured.
However, that Δ demand, though it is being generated amongst those associated with our enterprise or sector, such as its workers or suppliers (or their workers) is being spent, as we touched upon above, for goods and services produced across the economy. Therefore let’s add another variable, called P, that will represent the proportion of total demand satisfied by the sector or enterprise in question.
Now, to keep average prices for our enterprise’s or sector output constant, despite the change in output (for example, as created by our enlarged kitchen or a technological innovation affecting our entire sector) the ratio between demand and output (for our enterprises’ or sector’s output) must remain constant.
Or D + ΔD/ O + ΔO must = D/O
Relating all the demand terms to output by using our constant r gives changes the above equation to the following:
(r O + r ΔO P) /(O + ΔO) = r O/O
Therefore ((r O + r ΔO P)/(O + ΔO)) – (r O/O) = 0
Making the denominator common so we can do the subtraction:
((O (r O + r ΔO P))/(O(O + ΔO ))) – (((r O O) + (r O ΔO))/( O(O + ΔO )))
For the above equation to = 0, it is the numerator that must = 0, so:
(r O O) + (r O ΔO P) – (r O O) – (r O ΔO) = 0
Therefore (r O ΔO P) – (r O ΔO) = 0
Therefore r O ΔO P = r O ΔO
Therefore P (at the point where the change in output does not produce a fall in prices)
= (r O ΔO)/(r O ΔO) or 1.
(Furthermore, note that even including delayed effects does not significantly prevent demand associated with the increased output from being insufficient to sustain prices. Specifically, the demand eventually “returning”, from the rest of the economy, to the part of the economy, that has been able to able to increase its level of employment and output, will only be ((1 – P) P) ΔD, as the rest of the economy, even once itself “boosted” by the extra demand from the part of the economy able to hire more workers, will still only spend in turn across the whole economy.
We can take this a bit further, and consider, for our “toy” closed economy with no savings and no inflation or deflation, the total demand created by a sector of the economy’s additional output for its own output that arises simply by virtue of the money spend on producing that output. That will be P ΔD + ((1 – P) P) ΔD, or (P + P – P²) ΔD, or (2P – P²)ΔD. That means, for example, that even if the sector increasing its output is 0.8 of the total economy, only 0.96 of the money spent on producing an increment of that sector’s output will “return” to the sector to be spent on its products, meaning that, in reality, that increase in the sector’s output will, all other things being equal, reduce the prices for its products.
It is therefore clear that, rather than creating a “positive feedback loop” that can drive the whole economy out of the “trap” of paradoxical unemployment, lifting the MRPL threshold in just a sector of the economy improves the conditions for some but then induces a “negative feedback” in terms of price suppression that dampens out the MRPL threshold lifting effect).
In other words, to stand the best chance to escape the paradoxical unemployment trap by hiring more workers, and increasing output, do that across the maximum sweep of the economy possible, as, unless that sweep is the entire economy, the increase in output will reduce prices and therefore at least partially counter efforts to raise the MRPL threshold. That is why it took World War Two on top of Roosevelt’s New Deal to finally lift the USA out of the Great Depression.
REF 5:
Instead the possibility that the worker buys back from their own employer merely constitutes a small adjustment to MRPL, effectively reducing the cost of labour slightly by acting as a deduction from wages.
REF 6:
Automation can actually have a complex affect on employment. For example, if MRPL is applied using not only the conventional simplifications (that we are only considering what economists call the “short term” and that the level of the firm’s output forms a sufficiently small part of the market that it can be treated as not affecting overall demand) but also a whole set of optimistic assumptions, automation can actually result in employment increasing. That is because automation increases the productivity of labour, and therefore the revenue generated by each extra worker, and therefore raises the point at which the marginal revenue generated by each extra worker equals the extra cost of each extra worker, and so allows more workers to be profitably employed. The optimistic assumptions on which that deduction depends are, however, highly questionable. For example, it assumes that, in the new post automation regime, there are no issues with retraining the existing workers or finding workers with the necessary new skills, even though automation can completely erase certain jobs at the same time as it creates new ones.
Even more important, though, is ignoring any effect on demand. Even if the firm in question is a small player in its own markets, so there is no risk that the firm will flood its markets, it seems very likely that other firms in the same market, subject to the same market pressures and having the same technological opportunities, will, in the long run, behave the same way. Therefore, automation will result in an overall increase in output that, unless demand can expand indefinitely, could flood the market and depress prices.
If we make the contrary assumptions, specifically that most of the firms in the market will behave in the same way, and that demand is fixed, then the result of automation is entirely different. Expanded output depresses prices, reducing, for each firm, the point at which marginal revenue equals marginal cost, and so leads to overall output, in order to match demand, falling back to the volume of output at the point where automation was adopted, as there is simply no demand for the possible extra output. That volume of output can now be produced with less workers, so unemployment results. That unemployment is likely to depress wages so that, although each employed worker is now producing more and so could command a higher wage, there is also a strong possibility of downward pressure on demand and prices and even a subsequent “deflationary spiral”.
Of course, in reality, the truth is normally somewhere between those two extremes: demand, in the terms used by economists, is neither perfectly elastic or perfectly inelastic. Also automation, though it may come in waves, is likely to continue relentlessly, so the workers find themselves continuously at a disadvantage in respect to their potential employers. Even more importantly, the theoretical logical end case, where all demand for labour is eliminated, and “robots” can even make other “robots”, indicates that demand limitations need to be carefully considered when modelling the effects of automation. In that end case the needs, and even the survival, of the now entirely unwanted unemployed count for nothing, only the needs of the owners of the robots.
Again, contrast this with the effects of automation in a world where workers have access to their own capital. In this situation, even if demand is perfectly inelastic, instead of facing unemployment each worker is able to work fewer hours while still earning the same purchasing power, as any reduction in take-home earnings caused by working fewer hours is potentially compensated for by the falls in prices resulting from the general use of automation. The productivity increases produced by automation benefit the workers themselves, primarily by increasing leisure time. The end case, where all demand for labour is eliminated, results not in mass starvation, but in a potential utopia where toil has been eliminated for all, as everyone’s needs are met by the robots they all share in owning, and remaining work is of a vocational or recreational nature, done for the joy of the work itself.
REF 7:
Or you’ll find yourself on an “extremist register” for promoting anti-capitalism!
gezwinstanley 