Based on Peter Rutland, The Myth of the Plan (Open Court, 1985), and Alec Nove, The Soviet Economic System (George Allen & Unwin, 1980)
From its inception until the death of Stalin, the Soviet Union was in constant turmoil — the civil war, famine and forced collectivization, the Great Terror and the Gulag, the Great Patriotic War, the arduous task of postwar reconstruction. The years that followed Stalin’s death were more orderly, and in some ways, they were years of great achievement. The Soviet Union launched the world’s first orbital satellite in 1957 and the first manned orbital satellite in 1961. It developed nuclear weapons and rivalled the United States for global dominance. Its growing influence led a number of countries, India among them, to believe that central planning was the key to development. But the Soviet Union’s economic success was largely an illusion: the country that matched the United States ICBM for ICBM did not reliably provide food and housing to its people.
The figure below shows the Soviet Union’s per capita GDP in the years between Stalin’s death and the country’s collapse. It also shows the per capita GDPs of the United States and of the first two “Asian tigers.” The figure is log-linear, so rates of growth are proportional to slopes — a steeper curve implies faster growth.
The Soviet Union might have been expected to grow faster than the United States. It substantially increased its industrial labour force by moving people out of the countryside and into the cities, and by making employment the social norm among women. It had complete control over aggregate investment, and consistently prioritized investment over consumption. Simply replacing old technologies with more modern ones would have produced bursts of “catch-up growth.” But the faster growth did not materialize: the per capita GDPs of the Soviet Union and the United States both grew by 116% between 1953 and 1991. For the Soviet Union, merely keeping pace with the United States was tantamount to failure. Its per capita GDP in 1991 was only 60% of what American per capita GDP had been in 1953, almost four decades earlier.
The comparison with Japan and South Korea is stark. These countries were poorer than the Soviet Union in 1953, but both countries embraced market economics and grew rapidly. Both countries were significantly more affluent than the Soviet Union by the time of its collapse.
The premise of communism was that capitalism was failing, and that public ownership of the means of production would result in a fairer and stronger economy. Under communism the economy would be organized for the common good, not driven by the whims of the market. Stalinist central planning was supposed to achieve these ends, but it was an unwieldy system. Khrushchev introduced major reforms to the system, but they were reversed a decade later. The “Kosygin reforms” of 1965 were extensive but largely ineffective. Smaller reforms and reorganizations, each a response to a perceived problem in the system, were frequent. Even so, the fundamentals of Stalin’s system remained in place until shortly before the country’s collapse.
Central planning was put to the test during the period 1953-1991, and the judgment must be that it was far less successful than the system that it sought to replace. The purpose of this post is to explain how Soviet central planning worked and why it failed. Its shortcomings were all variations on a single theme: the system required planners to assemble and manipulate far more information than they could possibly handle, even after the advent of input-output analysis and electronic computers.
Both industry and agriculture were centrally planned, and both sectors were inefficient. This post discusses only industrial planning, as it was the more elaborate of the two systems.
The Basics of Central Planning
Soviet planners constructed two kinds of plans. The longer term plans — commonly known as Five Year Plans, although some plans were of longer or shorter duration — were used to allocate investment across sectors. For this purpose, goods could be aggregated into a few hundred categories. The annual plans, on the other hand, contained instructions for every industrial enterprise — which goods it would produce and how much of each, and what material supplies it would receive. These plans required a more complete specification of the goods.
It is not enough to speak of farm machinery. There must be a distinction made between different kinds of combine-harvesters, harrow, plough, cultivators, milking machines and so on…From a few hundred aggregated designations one rapidly runs into hundreds of thousands, and in fact the full list of identifiably different commodities…runs into the millions.1
The total number of goods, “disaggregated down to specific types of ball-bearings, designs of cloth, size of brown shoes, and so on,” was actually about 12 million.2
The planners’ main tool was the material balance, which set out how units of a good would be obtained and how they would be used. Units could be domestically produced, imported, or drawn from inventory. They could be used as an input, a capital good, a consumption good, an export, or an addition to inventories. The quantity obtained and the quantity used had to “balance,” that is, they had to be equal.
Each material balance was linked to the instructions given to individual enterprises and had to be consistent with them. Domestic production of each good was the sum of the amounts that the individual enterprises were ordered to produce. The amount to be used as an input was the sum of the amounts assigned to the individual enterprises, and the same was true of the amount to be used as a capital good.
The material balances were also linked to each other. An increase in the production of a good affected the material balance for that good, but it also affected the material balances of the goods used in its production, and the material balances of the goods used in their production, and so on. The planners could not hope to trace back all of the repercussions of a single change. Alec Nove gives this example:
Suppose it is decided to expand the output of the oil industry in Siberia. This decision calls for certain material inputs: machinery and equipment, road-making machines and materials, large-diameter pipe, perhaps a new railway line, houses for the workers, and so on. But all of these impose “indirect” requirements on the rest of the economy: equipment, pipe and railway-lines require steel, which in turn requires iron ore, coking coal, probably additional investments which in turn require building materials, machines (which require more steel), fuel (including still more oil), means of transport, which in turn require…and so on. Needless to say the process of planning is not confined to the consequences of a decision to increase the output of oil…Errors and omissions inevitably occur, and amendments have constantly to be made, as bottlenecks emerge. The process is never completed.3
Although there were 12 million identifiable goods, an annual plan involved the construction of material balances for fewer than 30,000 goods.4 Millions of goods were omitted entirely from the balances, and millions of goods appeared only as part of an aggregate.
Aggregation was itself a problem: the goods that were produced within an aggregate were not necessarily the goods that consumers or other producers wanted. If the aggregate were men’s footwear, for example, there could be a shortage of work boots and an excess of casual shoes. The planners often compounded the aggregation problem by setting the enterprises’ production targets in physical units. The managers’ responsibility was to meet their targets, and they often “gamed” their orders to do so. If the target for plate glass was set in square meters, they produced windows so thin that they shattered in the first storm. If the target for steel sheets was set in tons, they produced sheets so thick that their “customers” had to machine them down before using them. If the target for truck transport was set in ton-kilometers, the truckers drove roundabout routes or moved goods unnecessarily. These decisions were counter-productive. They were the result of an erosion of the concept of “responsibility” that the planning system itself had engendered:
It has ceased to signify responsibility for independent decisions, and has come to denote a much narrower concept of legal responsibility for the carrying out of the orders of higher authorities.5
How an Annual Plan was Made
The plan was the responsibility of Gosplan, the state planning commission, and Gossnab, the state supply commission. Gosplan formulated the plan, while Gossnab matched input producers with the enterprises that required their inputs.
The informational requirements of planning meant that Gosplan could not act alone. The individual enterprises participated by passing proposals upward and by commenting on draft orders. The enterprises were organized into a number of ministries based upon the nature of their product, and these ministries acted as intermediaries between Gosplan and the enterprises. Here is an outline of the process:
The whole annual plan formulation procedure begins when enterprises communicate their input requirements. This procedure begins half-way through the previous year. They make at the same time their estimates of what they will produce next year, though without yet knowing what their production plan will be. Their requests, and draft production plans, are aggregated by the ministries…Then in Gosplan the resultant magnitudes are contrasted with the resources available in the light of policy guidelines and priorities, and amendments made, as material balances can hardly be expected to balance on the first round. Administrative and statistical iterative procedures eventually lead to the emergence of a set of output and material supply plans which are supposed to match. Thus the enterprise manager should be told to produce a batch of goods (by his ministry) which corresponds to the supply allocations, and these may or may not approximate to his original requests.6
An important part of planning was setting the relationship between inputs and outputs. Gosplan took into consideration the enterprises’ supply requests, but was aware that managers often overstated their requirements in order to give themselves a buffer. Gosplan placed more emphasis on “norms,” estimates that were largely inferred from past experience. Although the enterprises discussed technical issues with their ministries, very little technical information was passed upwards to Gosplan, which would have been overwhelmed by it.
The Role of the Enterprise Managers
It is sometimes imagined that Gosplan handed down a detailed plan to its subordinates, who then executed it with military precision. The complexity of planning precluded any possibility of such a system. According to Peter Rutland,
The plan has to be repeatedly revised, and even then will still be broken at the end of the year. In some years around one in four of the industrial ministries have failed to meet their principal targets, and regularly up to one half of all the contracts between enterprises for the supply and delivery of goods are broken.7
In the course of the plan year a typical enterprise could find some six to ten major revisions in its plan.8
Alec Nove reports this survey result: in a sample of 95 enterprises, the average annual number of production plan amendments was 1554.9 Eugene Zaleski writes:
[The existence of] a central national plan, coherent and perfect, to be subdivided and implemented at all levels, is only a myth. What actually exists…is an endless number of plans, constantly evolving, that are coordinated after they have been put into operation.10
In view of the changing and often ephemeral nature of the plans, management emerges as the only constant in the system…It seems more nearly correct to call the economy “centrally managed” rather than centrally planned.11
In some abstraction of the planning process, managers simply obey the orders that are handed down to them. In practice, though, soviet managers were not just dutiful caretakers. They had to be agile, able to adjust and to improvise when new orders replaced old orders. They also had to be adept negotiators, able to make deals with the enterprises that they supplied and the enterprises that supplied them. They might also be required to negotiate with their ministry, if the ministry had some latitude in the way that it implemented new orders.
The managers, aware of the system’s failings, tried to protect themselves and their enterprises. An enterprise’s production plan was not completely disaggregated, so a manager sometimes adjusted the product mix so that he could more easily reach his targets. The plan did not exactly specify the quality of goods, so managers often produced low-quality goods to economize on material inputs or labour usage. The proposal submitted by an enterprise manager at the beginning of the planning process had some influence over the enterprise’s eventual output targets and supply entitlements, so the manager would shade his supply requirements upward and his production capacity downward.
Of course, the planners were aware that the managers “gamed the system,” so they made their own adjustments, confirming the managers’ belief that they had to protect their own interests.
[The] superior authority…assumes that the enterprise will conceal [information] and that it is not telling the whole truth. This helps to explain the apparently irrational behaviour of planners who seem to allocate more than there is to allocate: there must be something hidden, they reason, and pressure will compel it to emerge…But if this is so, it is very dangerous indeed for management to tell the whole truth about its productive potential, for it will not be believed and the plan will be set higher than is feasible. For the same reason, it will seem dangerous to overfulfil the plan by too much.12
The planners’ uncertainty about the behaviour of the managers also led to planning “on the achieved basis” — what was done in the past could be done again.
Conscious of the inadequacy of their knowledge of productive potential, of hoarded stocks of materials, and of the degree of tautness of past and present plans, the ministries and Gosplan tend to rely heavily on the one set of data in which they have confidence: past performance…It is grounded in necessity: one can only plan on the basis of what one knows.13
Planning on the achieved basis led to unwanted rigidities: the plan remained static for far too long when confronted with new opportunities or new hazards. For example, the Soviet Union was slow in introducing new technologies and slow in responding to environmental decay.
Managers “won” by not losing — by not missing their targets. Their focus was always on the downside risks, so they built slack into the system wherever they could. The “Kosygin reforms” of 1965 attempted to improve managerial incentives, but the problem of incentives was still unresolved when the Soviet Union collapsed in 1991.
More than 70 percent of Soviet output consisted of goods that went into the production of other goods.14 One enterprise’s failure to meet its target caused other enterprises to miss their targets, and their failures led to further failures. For example, a shortage of coking coal could lead to a shortage of steel, which in turn could lead to a shortage of stoves or refrigerators or industrial machinery.
Supply shortages were endemic to the Soviet system. Soviet managers typically spent about half of their time searching for the supplies that they needed to keep their factories operating, and failure to obtain supplies frequently idled factories.15
Gossnab issued to each enterprise a naryad, an order for supplies, but the naryad was not a guarantee that the enterprise would receive satisfactory supplies. It was more like a hunting license16 or a starting gun.
Because of the tremendous pressure for goods, there is no guarantee that the supplies designated in the naryad will actually materialize…Even if a supplier may be wary of failing to deliver the requisite quantity of goods specified in his plan, the areas of product mix and product quality offer an infinite variety of opportunities for short changing the customer enterprise. The purchasing manager — even with a legal naryad in his hands — has to rely upon threats, guile and the offering of return “favours” in order to secure his supplies.17
The persistence of supply problems led to the appearance of go-betweens who operated at the margins of the system, often without a supporting naryad.
The tolkach, literally “pusher,” is a breed of unofficial supply agent, whose job is to agitate, nag, beg, borrow, sometimes bribe, so that the necessary materials, components and equipment arrive.18
Managers tried to isolate their enterprises from these supply uncertainties. They built up inventories of raw materials as protection against future shortages. They also reduced their reliance on outside suppliers by manufacturing many of their own components, even to the point of manufacturing their own screws. In other instances the managers built up the capacity to produce their own components if necessary, but continued to use outside suppliers whenever they could.
One study found that 38% of the workers in a Soviet engineering factory are designated as repair staff, as opposed to 11% in an American factory. This difference partly reflects the low quality of the inputs received and of the equipment used, but it also reflects the desire to have the facilities to produce components from scratch when supplies fail.19
The managers’ unwillingness to rely on outside suppliers meant that the Soviet Union did not fully exploit the gains from specialization. There was also an ongoing loss in having supplies hidden away in every corner of the economy.
Consumption and Labour
The central planners did not attempt to control Soviet citizens in their roles as consumers or workers. Workers were paid in cash and were free to spend their incomes as they chose. The prices of consumer goods were set by the state, but shortages were common, for two reasons. First, the government’s emphasis on growing the economy meant that the planners favoured capital goods over consumer goods. Second, consumer goods were the last stage of a production process in which shortages abounded, so there were unplanned shortages of these goods as well. For consumers, queues and empty shelves were everyday events.
These shortages gave rise to the second economy.
Since the center would not supply what people needed, they struggled to do so themselves, developing in the process a huge repertoire of strategies for obtaining consumer goods and services. These strategies, called the “second” or “informal” economy, spanned a wide range from the quasi-legal to the definitely illegal. In most socialist countries it was not illegal to moonlight for extra pay — by doing carpentry, say — but people doing so often stole materials or illegally used tools from their workplace; or they might manipulate state goods to sell on the side. Clerks in stores might earn favours or extra money, for example, by saving scarce goods to sell to special customers, who tipped them or did some important favour in return. Also part of the second economy was the so-called “private plot” of collective farm peasants, who held it legally and in theory could do what they wanted with it — grow food for their own table or to sell in the market at state-controlled prices. But although the plot itself was legal, people obtained high outputs from it not just by virtue of hard work but also by stealing from the collective farm: fertilizer and herbicides, fodder for their pigs or cows, work time for their own weeding or harvesting, tractor time and fuel for plowing their plot, and so on. The second economy, then, which provisioned a large part of consumer needs, was parasitic upon the state economy and inseparable from it.20
The second economy might have served a useful purpose. It limited the people’s dissatisfaction with the planned economy by giving them a way to circumvent it. As well, since the prices were higher than in the planned economy (they equated supply and demand), the second economy absorbed money that would otherwise have gone unspent, leaving the workers to wonder exactly what it was that they were working for.
The state attempted to assign each individual worker to a job during the period 1940-56. It then abandoned this policy, so that workers became largely free to choose their place of residence and their place of employment.
Wage rates were set by top-level officials on the advice of a committee of experts. There was an attempt to follow the Marxist principle that workers should be paid in accordance with the amount of “socially necessary” labour that they provide. The committee tried to describe every job and determine its appropriate wage, but the actual wage often deviated from the committee’s prescribed wage. Labour shortages were widespread, especially in remote areas, and labour turnover was high. The enterprise managers, forced to compete for labour and always under pressure to meet their targets, found ways of providing their workers with extra compensation. One survey found that “one-quarter of all wage payments in one ministry were in violation of legal requirements.”21
Linear Programming and the Electronic Computer Couldn’t Save Central Planning
The development of linear programming by Leonid Kantorovich, the simplex solution method by George Danzig, and input-output analysis by Wassily Leontief, all in the 1930s and 1940s, provided a solid technical base for central planning. The manipulation of very large data sets became possible only a little later, when the first programmable electronic computers appeared. The Soviet Union’s mathematical economists explored the uses of these innovations, and by the 1960s, they had come to believe that central planning was potentially the most efficient way of allocating resources. The novelist Francis Spufford has captured the optimism of these times:
The electrons flowing through the vacuum tubes represent digits; and tonight the digits the BESM is processing represent potatoes. Not, of course, potatoes as they are in themselves, the actual tubers, so often frost-damaged or green with age or warty with sprouting tubercles — but potatoes abstracted, potatoes considered as information, travelling into Moscow from 348 delivery units to 215 consuming organizations. The BESM is applying [Kantorovich’s] mathematics to the task of optimizing potato delivery for the Moscow Regional Planning Agency. Seventy-five thousand different variables are involved, subject to 563 different constraints: this problem is out of reach of fingers and slide rules. But thanks to computers, thanks to BESM’s inhuman patience at iterating approximate answers over and over again, it is a problem that can be solved.
The BESM is using [Kantorovich’s] shadow prices to do what a market in potatoes would do in a capitalist country — only better. When a market is matching supply and demand, it is the actual movement of the potatoes themselves from place to place, the actual sale of the potatoes at ever shifting prices, which negotiates a solution, by trial and error. In the computer, the effect of a possible solution can be assessed without the wasteful real-world to-ing and fro-ing; and because the computer works at the speed of flying electrons, rather than the speed of a trundling vegetable truck, it can explore the whole of the mathematical space of possible solutions, and be sure to find the very best solution there is, instead of settling for the good enough solution that would be all there was time for, in a working day with potatoes to deliver…
The market’s clock speed is laughable. It computes at the rate of a babushka in a headscarf, laboriously breaking a two-rouble note for change and muttering the numbers under her breath. Its stock arrives one sack or basket at a time, clutched on a peasant lap. It calculates its prices on cardboard, with a stub of pencil. No wonder that Oscar Lange over in Warsaw gleefully calls the marketplace a “primitive pre-electronic computer.”22
The impact of these innovations fell short of expectations. Input-output analysis was useful to the planners, and computers reduced their workload, but material balances remained their primary tool. What went wrong?
The main issue was that the dimensionality of the problem was too great for the computers of the time. There were 12 million goods, to be delivered to perhaps a thousand locations, so there were about 12 billion combinations of good and location — and each of these combinations was a variable to be solved for. The computers were fine at allocating potatoes. Allocating everything was another matter.
Having the computational power to construct a complete plan would have eliminated the problem of omitted or aggregated goods, but it wouldn’t have prevented shortages from occurring. Some of the shortages were caused by the use of historical norms to determine input entitlements. Norms could not be based purely on technical factors. The resource cost of laying a kilometer of railway depends upon whether the land is firm or boggy, whether there is rock to be blasted or gullies to be filled. The principal cost of refining oil is the cost of heating it from ambient temperature, so it is as variable as the weather. The output of a workshop depends upon the skills of the workers, which in turn depends upon the manager’s recruitment skills. But basing the norms on past results was also problematic. The strategic behaviour of the managers and the workers distorted historical norms. As well, historical norms excluded the impact of technological progress, so this impact had to be roughly estimated. The problem of norms was one of data quality, not data quantity, so it wasn’t a problem that could be solved by more computational power.
Another problem that a complete plan would not solve was the Soviet Union’s slowness in introducing new technologies and its frequent reliance on the reverse engineering of technologies developed elsewhere. Successful innovation is often a “bottom up” phenomenon. People who are intimately familiar with a process recognize both a problem and a possible solution. Given the opportunity and the resources, they can sometimes make their solution work — although sometimes their efforts lead nowhere. Central planners, by contrast, have no detailed knowledge of any of the available technologies, and cannot know whether there are innovations that are worth pursuing. Furthermore, the planners’ demand that subordinates meet their targets with limited resources discourages the subordinates from experimenting with production techniques that might not prove successful. As Rutland observes,
Backing innovation is an intrinsically decentralized process. It involves placing a high degree of trust in subordinates; accepting that the results will be uneven; and a willingness to see risk as an intrinsic part of economic life. The Soviet system is highly inimical to all of these tendencies.23
Mathematics and computing had the potential to improve the kind of central planning that the Soviet Union didn’t actually have: the planning in which a complete plan filters downward to be implemented by obedient managers. As Zeleski observed, the Soviet Union was actually centrally managed rather than centrally planned. One plan replaced another as problems emerged, and managers were expected to deal with the fallout.
Assuming that sufficient computational power was available, would a “complete plan, obedient managers” scheme perform better than the “incomplete plan, active managers” scheme actually used in the Soviet Union? There are two reasons why it might not. The first reason is that the former scheme imagines that all relevant information is possessed by the planners (or more accurately, by the planners’ supercomputer). Hayek argued that a lot of economically valuable information is transitory knowledge of time and place. With such information constantly appearing and disappearing, the former scheme allows two possibilities: the information is ignored, or the information is passed upward and the entire plan is revised time and again. The first option ignores the value of the information, and the second option forces the managers to constantly change course. It is not clear that either possibility is better than having the local manager deal with local matters on their own initiative, as he did under the latter scheme. Also, the former scheme imagines that all valuable knowledge can be reduced to technical relationships — to lines of code — but a lot of important knowledge is inductive. Managers change what they do as they gain experience. They can’t always explain the choices that they make but they get better results all the same — they exercise better judgment. The former scheme minimizes the opportunity for the exercise of judgment; the latter scheme relies upon it. The second reason is that the former scheme treats subordinates as if they were impartial distributors of information. The cadres during China’s central planning era were something like these mythical obedient managers: Yang Jisheng has described them as slaves facing upward and dictators facing downward.24 But the cadres recognized that their success hinged on their ability to please their superiors. They tailored their reports accordingly, withholding bad news and exaggerating good news. Yang believes that their actions contributed to the Chinese famine of 1958-62, in which upwards of 30 million people died. Any system that treats people as if they were information-disseminating robots is going to run into difficulties.
The great fallacy of central planning is that it is possible to design a social system without taking into account the hopes, fears, and ambitions of the people governed by it. These things drive much of economic behaviour. The Soviet planning system was endlessly disrupted by them precisely because it paid so little attention to them. The market system, by contrast, places these things at the center of the system. “It is not from the benevolence of the butcher, the brewer, or the baker that we expect our dinner, but from their regard to their own self-interest.”
- Nove, The Soviet Economic System, p. 34. ↩
- Alec Nove, The Economics of Feasible Socialism (Allen & Unwin, 1987), p. 33. ↩
- Nove, The Soviet Economic System, pp. 39-40. ↩
- Rutland, The Myth of the Plan, p. 116. ↩
- Rutland, The Myth of the Plan, p. 147. ↩
- Nove, The Soviet Economic System, pp. 44-5. ↩
- Rutland, The Myth of the Plan, p. 119. ↩
- Rutland, The Myth of the Plan, p. 118. ↩
- Nove, The Soviet Economic System, p. 107. ↩
- Excerpt from Eugene Zaleski, Stalinist Planning for Economic Growth. Reproduced in Martin McCauley, Stalin and Stalinism (Routledge, 2019), p. 168. ↩
- Excerpt from Eugene Zaleski, Stalinist Planning for Economic Growth. Reproduced in Martin McCauley, Stalin and Stalinism (Routledge, 2019), p. 168. ↩
- Nove, The Soviet Economic System, p. 106. ↩
- Nove, The Soviet Economic System, p. 108. ↩
- Nove, The Soviet Economic System, p. 35. ↩
- Rutland, The Myth of the Plan, p. 128. ↩
- This comparison is from Gregory Grossman. Cited by Rutland, The Myth of the Plan, p. 130. ↩
- Rutland, The Myth of the Plan, pp. 129-30. ↩
- Nove, The Soviet Economic System, p. 103. ↩
- Rutland, The Myth of the Plan, p. 130. ↩
- Katherine Verdery, What Was Socialism, and What Comes Next? (Princeton, 1996), p. 27. ↩
- Rutland, The Myth of the Plan, pp. 151-2. ↩
- Francis Spufford, Red Plenty (Graywolf Press, 2012), pp. 115-6. ↩
- Rutland, The Myth of the Plan, p. 147. ↩
- Attributed to Yang Jisheng by Edward Friedman and Roderick MacFarquahar, in their introduction (p. 6) to Yang’s Tombstone: the Great Chinese Famine, 1958-1962 (Farrar, Straus and Giroux, 2012). ↩