Shadow Socialism

Alexander Douglas
6 min readNov 10, 2022


Some socialists are “market” socialists, others are “planning” socialists. “Planning” usually means central planning, though not always. Robin Hahnel and others have for a while been advocating a system of democratic economic planning, though I think it would be fairest to say that this is somewhere in between a planning and a market system. The rough idea is that different social units — individuals, workers’ councils, neighbourhood “consumption councils”, federations, etc. — submit information about what they would like to consume and what work they are willing to do, then receive back a set of indicative prices based on the inputs offered and outputs demanded. In light of these prices (which they most likely won’t accept) they revise their offers — e.g. increasing the work supplied or reducing the goods demanded — and, feeding in these new offers, receive back a new set of indicative prices. The process iterates until some critical mass of units have accepted the resulting prices.

A more unambiguous type of planning is the one proposed by Paul Cockshott and Alin Cottrell. This version supposes that some planning authority — ideally democratic — uses a computer to build an input-output table for the entire economy, then plans the economy through linear programming. Although the computational task seems daunting — for a modern economy, the table would have to include millions of rows and columns — Cockshott, a computer scientist, explains how modern machines are easily up to the task.

The planning works as follows. All the values in the I-O table are in terms of labour-time. The table includes technical coefficients determining the amount of work needed to produce each final consumption good, including the work needed to produce all intermediate goods required for final production — building machinery, extracting raw materials, etc. — given current techniques. A computer can then solve a linear programming problem to allocate work so as to produce the optimal vector of output goods using the minimal amount of work. The techniques of Leonid Kantorovich can be used to add constraints besides the availability of labour, such as ecological sustainability.

The optimal vector of output goods can be determined in part by a market for consumer goods, but one run on very different principles to capitalist consumer markets. Each worker receives a labour-token for each hour worked (these can be graded up or down to reflect different intensities of work, also premiums could be paid on work for which there is great social need but no willing supply). Workers can then spend these on consumer goods. Excess demand will bid prices up, and excess supply will bid them down. This provides information about what the optimal composition of output should be — namely, the supply of each consumption good should be such that its market price in labour-tokens equals the total hours of work it took to produce. The planning system will increase production of goods in excess demand and reduce that of goods in excess supply, aiming thus at the optimal vector. Meanwhile, the desired output of other goods — public and investment goods — will be subject to collective, democratic decision. Average prices of consumption goods will exceed their labour-values to the degree needed to cover the cost of these other goods.

There are many objections to implementing this sort of planning system in practice. Hahnel points out that the technical coefficients in the I-O table reflect current production techniques; thus the planning system has no mechanism for deciding among other possible production techniques, which might be better. Cockshott and Cottrell, however, believe that innovation could nevertheless be fostered through research and development departments and educational institutions. Once a new technology is developed and adopted, the relevant technical coefficients would be updated in the I-O table.

Then there’s the standard objection against central planning, concerning various principal-agent problems regarding incentives. For example, information about the amount of work needed to produce things, determining the technical coefficients, can only be given by producers, but producers have an incentive to overstate this and enjoy an easier workload. Cockshott and Cottrell reply that this strategy would show up to the planning system in a discrepancy between market prices and reported labour values. The computer system could be programmed to flag up such discrepancies and trigger an investigation into the relevant producers (careful investigation would be needed, since not every such discrepancy would signal misreporting). But we can question the cost, practicality, and political consequences of such a system of surveillance.

All the same, I see no reason not to build the relevant I-O table and make it widely available. Such a table would allow political leaders to know what was practically possible in a much more concrete way than currently. Right now, political debates concern whether “there is enough money” to carry out some plan or other. This is misleading for two main reasons (I’m ignoring MMT-type considerations here; let’s just say that the amount of “money” available to a currency-issuing government is how much it can spend without missing some price-stability target).

First, the availability of money is a poor guide to the availability of resources. You can divide your total stock of money by the current price of x to find the number of units of x you can afford. But that doesn’t tell you that it’s actually possible to produce that many units of x. Even if it is, the economic shifts involved in producing those units might affect your income, and thus your stock of money.

Second, the technique of working out what the state can afford in terms of money takes prices as given. If a pharmaceutical company with a patent on some life-saving medicine triples the price, the state can afford less of it. But to conclude from this that less medicine is therefore available builds current property rights into the calculation of available resources. That rules out socialism as a logical possibility. Calculating the resources available for various tasks in terms of money takes prices as given, and taking prices as given takes property rights as given. Perhaps, as Leo XIII suggested, God doesn’t want us to abolish property rights, but surely He can’t object to us knowing what would be technically possible if we modified them.

One thing it would be very interesting to know is: how much labour time (and other scarce resources) would it take to provide decent housing, healthy food, comprehensive medicine, and basic education to the entire population? Knowing this, we could divide the resulting figure by an estimate of the available labour force and learn something pretty interesting about political possibilities.

We could also learn things like: to what extent could we provide more basic necessities for people in need by diverting production from luxury goods? If you read a standard economics textbook, you’re likely to find something like this: It might seem like a good idea for the government to tax sales of yachts, to get revenue to build children’s hospitals, but economics shows that the practical effect might be deadweight loss: rich people could just stop buying yachts, then the government wouldn’t get the revenue for the hospitals. This looks straight past the material reality into monetary abstraction. If yacht sales come down, so, presumably, will yacht production, and thus productive capacity will be freed up to build something else, e.g. equipment for children’s hospitals. It might not be technically possible to divert resources between those two productive tasks, but the point is that a comprehensive input-output table could tell us what sorts of diversions are technically possible.

Of course, an allocation of resources can be technically possible without there being any way to get to it, humans being how they are. Incentives matter and all that. But that’s no reason not to have a decent go at working out what we could do if enough of us wanted to. Implementing socialism is a tall order, but simulating it might have interesting political effects on capitalism. Shadow socialism could probably achieve a lot without being in government.



Alexander Douglas

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