ComradeOm
23rd July 2007, 12:01
One feature of the USSR that I always feel is neglected, in favour of political critiques and sectarian point scoring, is the command economy that it operated on. This remains the only alternative to the market to be practised on a large-scale and for much of its history was remarkably effective. Unfortunately amongst all the discussion of "abolishing hierarchy" and "vanguards" this crucial example of socialist economic governance is usually ignored. When it is criticised its often done so using typical liberal critiques.
I mention this now as I recently came across an extremely interesting post on another forum. I'm sure that the author, who is certainly no socialist, wouldn't mind me reproducing it here. Bear in mind that this is not an essay but a simple off-the-cuff response to a question on comparing the free market and command economy.
-----
In principle, central planning should be as efficient, if not more efficient, than a free market system. It doesn't have the drawbacks of built-in periodic crises ('business cycle'), with all that attendant unemployment or inflation, it doesn't have the inherent inefficiencies of private monopolies and their artificial scarcities nor is it tied into particular specializations which may stunt long-run growth.
A central planner can (in theory) achieve an economy that is more efficient, more just and with faster growth than the free market. The question is whether it does. And here, the answer is usually no.
In practice, Soviet Russia achieved some degree of success on all three fronts (no unemployment, strong growth), but it also achieved some overwhelming and devastating failures (severe misallocations, famines, etc.)
The big monkey wrench in central planning was (and is) inadequate information and politicization.
To give an example, let's say you want to increase the nation's agricultural productivity by introducing tractors to all farms. In a free market system, buying a tractor may not be profitable for a private farmer - maybe grain prices are not good enough to overcome the purchase price of a tractor, maybe the scale of his production is too small to justify it, maybe the financing is too expensive, maybe there are cheaper labor-based alternatives, etc. At the same time, the price of tractor may be too high for the farmer because tractor producers are facing expensive steel prices, or have problems attracting skilled labor to their tractor factories, or just don't feel good about the size of the market and didn't invest in expanding their tractor plants, etc.
Unless the market conditions are aligned rightly, the farmer will not buy a tractor. And, unless the government intervenes with some sort of lures and incentives (e.g. depreciation allowances, manipulating interest rates), that's the end of the story. Farms will not have tractors. Agricultural productivity will remain low.
In a central planning system, you don't need to wait for conditions to be right. The central planner can just cut through the knot and order the tractor factories to produce X tractors, give them the necessary steel, allocate the necessary labor, undertake the necessary plant expansion, and then transport the finished tractors to all the farms. Farms will have tractors. Agricultural productivity will now rise.
The problem is coordinating all that. Do you know how much steel you have available and what are the other demands upon it (ship-building, car manufacturing, etc.)? Are you allocating it between those industries properly? Is the labor that you're allocating enough/too much? Is the plant expansion to large/too small? And then to the farms, how many tractors per farm? Did you provide tractors but forgot to provide sufficient fuel to run them? (How much fuel? What other industries need the fuel?, etc.)
As you can see, the informational needs and computational elements add up quickly to a monstrous calculation problem. And you can make tremendous mistakes. You may end up with far more tractors than you need, or have fuel to run, or too few, etc. Lots of waste can happen.
Happily, during the 1940s & 1950s, Gosplan got better and better at it, and started producing some tremendous computational models of the entire Russian economy and new mathematical tools to calculate the most efficient way to allocate the resources. They were so good at it, that the same models were adopted in the US to help government economists estimate the progress of the US economy. They even got a couple of Nobel prizes for it.
The problem, naturally, was that the whole process quickly became politicized. So even if the planners figured it was most efficient to allocate steel to produce 100 tractors this month and the rest to ships and railways, that meant nothing if Comrade Stalin was on an agro-kick and wanted to ratchet the production of tractors to 500, even if that contradicted his ultimate target. The end result is that tractors ended up rusting in the fields while there wasn't sufficient railways and ships to transport the resulting grain. The cities starve for bread while grain rots in the barn.
The input information was also distorted and politicized. Anxious to get his steel quota, manager of Tractor Factory No. 3 exaggerates his need for steel. So does the manager of Shipyard No.5. Since the allocation is based on this erroneous reported information, not even the best computational models will get it right. Again, misallocations.
Then tack on the fact that the manager of Tractor Factory No. 3 has better connections in the party than the manager of the Shipyard No. 5, and the planning models stop to matter at all. We all know who will be getting the steel.
A sad footnote to all this, is that even the best and most efficient computation models used by Gosplan professionals were under constant attack for not being 'politically correct' and had to be adjusted, twisted, buried under piles of nonsense and doo-doo, to make them palatable to the luminaries of the CPSU and their manager friends. By the 1970s, they practically gave up being serious about it altogether and just started asking the party to decide for them.
And, oh, to add one final point: there was one manager who was always f***d over in all this, the one with absolutely no connections or voice in the party, no input into the planning process at all: the manager of Consumer Factory No.1 - the average Soviet citizen.
That is the principal drawback of the central planning. Too disorganized, with no 'manager' or advocate to represent them in the hierarchy, the consumer's 'needs' were always put at the back of the priorities.
That was the perennial problem of Soviet Russia from the very first day to the very last: everybody was employed, yes; everybody received a decent wage, yes; everybody had access to generous healthcare, retirement pensions, yes. But all that meant nothing if there was nothing to buy. This 'wage-goods' gap was the gnawing constant of Soviet life.
The need to break the red managers' control over allocation decisions, to give consumers a voice in the process, was the high principle behind Gorbachev's 'perestroika'. It ended disastrously.
----
When I inquired into the details of these Soviets plans:
----
I haven't actually seen the models they used, and I doubt they were explicitly published (although they are probably accessible now). But there were many papers relating the techniques used and developed. Look up "planometrics", "soviet standard methodology", and the like, and you're bound to get a lot. If you like pain, the Soviet journal Matekon was the main publishing arm of the planners.
You mentioned operations research. Naturally, you should know that Leonid Kantorovich, one the brightest lights in the Soviet firmament, practically founded that field (well, before Danzig & Co. anyway) and deployed all the linear programming methods he developed into Soviet planning in the 1960s. You may want to check his work out. The final version of the mathematical programming method, known as the "Soviet Standard Methodology" (SSM) was officially released in 1977 and you can probably find that somewhere.
Before that, the most straightforward was the use of norms and material balances and, from the 1950s, input-output models and, under the guidance of Novozhilov, investment project valuation methods. Janos Kornai developed a lot of the methods for more dynamic planning in 60s Hungary and has published extensively.
Observers like Alec Nove, Abram Bergson, Michael Ellman and, again Kornai, give a lot of details of the actual nitty-gritty of the Soviet planning process. They've got some rather ugly (but authoritative) books & articles on the subject.
As for the idea of socialist efficiency, the classic here is Oskar Lange, and the follow-up work by Lerner, Koopmans & co. Lange himself had an ingenious 'solution' to the central planning problem: have Gosplan just yell out 'prices' and order managers to minimize costs. i.e. just replicate the market economy without the capitalists. Naturally, Lange's 'solution' did not go down well with Marxist-Leninists who had doctrinaire ideas about 'value' and 'price'*, but such 'market socialism' models were experimented to a degree in 'liberal' socialist countries, like 1960s Hungary and 1980s Russia.
(* - Doctrine was a problem Kantorovich always faced. He didn't dare confess his 'shadow prices' put the lie to Marxian economic theory, so he had to do a lot of relabeling of terms and pad it with generous quotes from Capital to obfuscate it.)
----
And on the managing the information with the available computational power of today:
----
Yeah, lack of sheer computing power was a monstrous hurdle, so they never actually developed a single big model covering absolutely everything but usually broke it down into layers. There'd be the top "long -run' Five Year plan, with very aggregated suggestions, dependent more on mildly multi-sectoral growth models, and then the semi-aggregated yearly figured out by Gosplan and then the very disaggregated local plans of Gossnab who did the nitty-gritty.
The most monstruous was the yearly plan. According the Nove, it was broken down into quarters & months, and involved projecting inputs and outputs for 48,000 'products' (each of these 'products' actually represents an aggregate of 250 or so sub-products that was to be left for the sub-plans). For a modern computer, that'll take all of five seconds to figure out. Back then, it was a monstrosity. According to one planner's quip, a fully disaggregated yearly model, with every entry checked and balanced, would take 30,000 years to be ready.
In the 60s, when Kantorovich was trying to first deploy linear programming on a modest scale (for Soyuzglavmetal, that is metal allocation), they ended up with a model of about one million unknowns and 30,000 constraints. Took six years to collect the data.
As for it being 'possible' with modern computing power, heck it is not only possible, economists do it all the time. In a bizarre twist of fate, since an 'efficient' socialist solution is not worse than a free market solution, the easiest (and most common) way of modeling a market economy on a computer is actually to assume it is socialist first, finding the 'optimum' like a good socialist planner would, and then translating the solution back into market terms. Many economic forecasts these days are generated that way.
I mention this now as I recently came across an extremely interesting post on another forum. I'm sure that the author, who is certainly no socialist, wouldn't mind me reproducing it here. Bear in mind that this is not an essay but a simple off-the-cuff response to a question on comparing the free market and command economy.
-----
In principle, central planning should be as efficient, if not more efficient, than a free market system. It doesn't have the drawbacks of built-in periodic crises ('business cycle'), with all that attendant unemployment or inflation, it doesn't have the inherent inefficiencies of private monopolies and their artificial scarcities nor is it tied into particular specializations which may stunt long-run growth.
A central planner can (in theory) achieve an economy that is more efficient, more just and with faster growth than the free market. The question is whether it does. And here, the answer is usually no.
In practice, Soviet Russia achieved some degree of success on all three fronts (no unemployment, strong growth), but it also achieved some overwhelming and devastating failures (severe misallocations, famines, etc.)
The big monkey wrench in central planning was (and is) inadequate information and politicization.
To give an example, let's say you want to increase the nation's agricultural productivity by introducing tractors to all farms. In a free market system, buying a tractor may not be profitable for a private farmer - maybe grain prices are not good enough to overcome the purchase price of a tractor, maybe the scale of his production is too small to justify it, maybe the financing is too expensive, maybe there are cheaper labor-based alternatives, etc. At the same time, the price of tractor may be too high for the farmer because tractor producers are facing expensive steel prices, or have problems attracting skilled labor to their tractor factories, or just don't feel good about the size of the market and didn't invest in expanding their tractor plants, etc.
Unless the market conditions are aligned rightly, the farmer will not buy a tractor. And, unless the government intervenes with some sort of lures and incentives (e.g. depreciation allowances, manipulating interest rates), that's the end of the story. Farms will not have tractors. Agricultural productivity will remain low.
In a central planning system, you don't need to wait for conditions to be right. The central planner can just cut through the knot and order the tractor factories to produce X tractors, give them the necessary steel, allocate the necessary labor, undertake the necessary plant expansion, and then transport the finished tractors to all the farms. Farms will have tractors. Agricultural productivity will now rise.
The problem is coordinating all that. Do you know how much steel you have available and what are the other demands upon it (ship-building, car manufacturing, etc.)? Are you allocating it between those industries properly? Is the labor that you're allocating enough/too much? Is the plant expansion to large/too small? And then to the farms, how many tractors per farm? Did you provide tractors but forgot to provide sufficient fuel to run them? (How much fuel? What other industries need the fuel?, etc.)
As you can see, the informational needs and computational elements add up quickly to a monstrous calculation problem. And you can make tremendous mistakes. You may end up with far more tractors than you need, or have fuel to run, or too few, etc. Lots of waste can happen.
Happily, during the 1940s & 1950s, Gosplan got better and better at it, and started producing some tremendous computational models of the entire Russian economy and new mathematical tools to calculate the most efficient way to allocate the resources. They were so good at it, that the same models were adopted in the US to help government economists estimate the progress of the US economy. They even got a couple of Nobel prizes for it.
The problem, naturally, was that the whole process quickly became politicized. So even if the planners figured it was most efficient to allocate steel to produce 100 tractors this month and the rest to ships and railways, that meant nothing if Comrade Stalin was on an agro-kick and wanted to ratchet the production of tractors to 500, even if that contradicted his ultimate target. The end result is that tractors ended up rusting in the fields while there wasn't sufficient railways and ships to transport the resulting grain. The cities starve for bread while grain rots in the barn.
The input information was also distorted and politicized. Anxious to get his steel quota, manager of Tractor Factory No. 3 exaggerates his need for steel. So does the manager of Shipyard No.5. Since the allocation is based on this erroneous reported information, not even the best computational models will get it right. Again, misallocations.
Then tack on the fact that the manager of Tractor Factory No. 3 has better connections in the party than the manager of the Shipyard No. 5, and the planning models stop to matter at all. We all know who will be getting the steel.
A sad footnote to all this, is that even the best and most efficient computation models used by Gosplan professionals were under constant attack for not being 'politically correct' and had to be adjusted, twisted, buried under piles of nonsense and doo-doo, to make them palatable to the luminaries of the CPSU and their manager friends. By the 1970s, they practically gave up being serious about it altogether and just started asking the party to decide for them.
And, oh, to add one final point: there was one manager who was always f***d over in all this, the one with absolutely no connections or voice in the party, no input into the planning process at all: the manager of Consumer Factory No.1 - the average Soviet citizen.
That is the principal drawback of the central planning. Too disorganized, with no 'manager' or advocate to represent them in the hierarchy, the consumer's 'needs' were always put at the back of the priorities.
That was the perennial problem of Soviet Russia from the very first day to the very last: everybody was employed, yes; everybody received a decent wage, yes; everybody had access to generous healthcare, retirement pensions, yes. But all that meant nothing if there was nothing to buy. This 'wage-goods' gap was the gnawing constant of Soviet life.
The need to break the red managers' control over allocation decisions, to give consumers a voice in the process, was the high principle behind Gorbachev's 'perestroika'. It ended disastrously.
----
When I inquired into the details of these Soviets plans:
----
I haven't actually seen the models they used, and I doubt they were explicitly published (although they are probably accessible now). But there were many papers relating the techniques used and developed. Look up "planometrics", "soviet standard methodology", and the like, and you're bound to get a lot. If you like pain, the Soviet journal Matekon was the main publishing arm of the planners.
You mentioned operations research. Naturally, you should know that Leonid Kantorovich, one the brightest lights in the Soviet firmament, practically founded that field (well, before Danzig & Co. anyway) and deployed all the linear programming methods he developed into Soviet planning in the 1960s. You may want to check his work out. The final version of the mathematical programming method, known as the "Soviet Standard Methodology" (SSM) was officially released in 1977 and you can probably find that somewhere.
Before that, the most straightforward was the use of norms and material balances and, from the 1950s, input-output models and, under the guidance of Novozhilov, investment project valuation methods. Janos Kornai developed a lot of the methods for more dynamic planning in 60s Hungary and has published extensively.
Observers like Alec Nove, Abram Bergson, Michael Ellman and, again Kornai, give a lot of details of the actual nitty-gritty of the Soviet planning process. They've got some rather ugly (but authoritative) books & articles on the subject.
As for the idea of socialist efficiency, the classic here is Oskar Lange, and the follow-up work by Lerner, Koopmans & co. Lange himself had an ingenious 'solution' to the central planning problem: have Gosplan just yell out 'prices' and order managers to minimize costs. i.e. just replicate the market economy without the capitalists. Naturally, Lange's 'solution' did not go down well with Marxist-Leninists who had doctrinaire ideas about 'value' and 'price'*, but such 'market socialism' models were experimented to a degree in 'liberal' socialist countries, like 1960s Hungary and 1980s Russia.
(* - Doctrine was a problem Kantorovich always faced. He didn't dare confess his 'shadow prices' put the lie to Marxian economic theory, so he had to do a lot of relabeling of terms and pad it with generous quotes from Capital to obfuscate it.)
----
And on the managing the information with the available computational power of today:
----
Yeah, lack of sheer computing power was a monstrous hurdle, so they never actually developed a single big model covering absolutely everything but usually broke it down into layers. There'd be the top "long -run' Five Year plan, with very aggregated suggestions, dependent more on mildly multi-sectoral growth models, and then the semi-aggregated yearly figured out by Gosplan and then the very disaggregated local plans of Gossnab who did the nitty-gritty.
The most monstruous was the yearly plan. According the Nove, it was broken down into quarters & months, and involved projecting inputs and outputs for 48,000 'products' (each of these 'products' actually represents an aggregate of 250 or so sub-products that was to be left for the sub-plans). For a modern computer, that'll take all of five seconds to figure out. Back then, it was a monstrosity. According to one planner's quip, a fully disaggregated yearly model, with every entry checked and balanced, would take 30,000 years to be ready.
In the 60s, when Kantorovich was trying to first deploy linear programming on a modest scale (for Soyuzglavmetal, that is metal allocation), they ended up with a model of about one million unknowns and 30,000 constraints. Took six years to collect the data.
As for it being 'possible' with modern computing power, heck it is not only possible, economists do it all the time. In a bizarre twist of fate, since an 'efficient' socialist solution is not worse than a free market solution, the easiest (and most common) way of modeling a market economy on a computer is actually to assume it is socialist first, finding the 'optimum' like a good socialist planner would, and then translating the solution back into market terms. Many economic forecasts these days are generated that way.