Central Planning and Heuristic Pricing

 *Note: I'm reposting this and maybe a few other posts from an earlier defunct blog of mine for 'archival' reasons.

My interests have led me to begin reading various works in the area of socialist economic theory (and to some extent economic practice), and this has spurred a great deal of thinking on my part about the topic. As such, I will be writing a series of posts on related subjects I've recently been reading or thinking about. Today, we will be revisiting the socialist calculation debate that emerged in the 1920s; more specifically, the most robust argument made by proponents of central planning and the public ownership of factors of production: Fred Taylor's proposed 'trial and error' approach to pricing goods and services (Taylor, 1929). Some socialists have argued that Taylor's proposed mechanism was never really refuted by the Mises or Hayek during the original discourse. Jossa and Cuomo (1997), for example, argue that Hayek mistakenly believed he had refuted Taylor's method with his famed 'knowledge problem.' However, Hayek, they contend, did not understand Taylor's method. Hayek failed to note that the prohibitively massive and expensive demand data collection project was not necessary.

 

 

 

For this post, I will first describe Taylor's heuristic market socialist approach to pricing, and then offer my criticisms of it. But first, some background. The central question behind this debate is: can the state set the prices and production levels of each good produced so as maximize efficiency? Or, put slightly differently, can it do so as efficiently as a free market, in which the means of production are privately owned, can? Ludwig von Mises and Friedrich Hayek argued (and most economists have found their argument persuasive; Brad DeLong, a Keynesian not at all sympathetic to Austrian economics, declared, "even liberal Keynesian social democrats acknowledge that the Austrians won victory in their intellectual debate with the central planners long ago."; see references for link) that the central planner would have to solve the system of general equilibrium equations for all goods being produced, which itself would be untenable; of course, one might use computational methods to approximate it, but Hayek argued that one would still have to collect information on the demand for each and every good or service to plug into the general equilibrium model, and that the subsidiarity of knowledge about demand rendered such a data collection enterprise by a central planner inevitably more costly than merely letting the local market participants determine prices. For socialists contemporary with von Mises and Hayek, this was quite problematic.

 

 

 

However, an alternative method of pricing was proposed by Fred Taylor. This method would not require one to solve an exorbitantly complex system of equations and collect massive amounts of data almost perpetually on demand for a myriad of goods. Instead, Taylor argued for a 'trial and error' approach that largely recapitulates the heuristic processes by which von Mises argued markets achieve optimal prices (and therefore optimal allocation of resources). This method would avoid both the need to simultaneously solve the general equilibrium equations and to collect data on demand for all goods and services in the economy almost constantly. In other words, it differentiated itself from other proposed centrally planned systems by managing to be plausible.

 

 

 

Taylor's Pricing Approach

So, how would Taylor's mechanism for achieving an optimal set of prices work (or at least my potentially erroneous understanding of it). Let's suppose that, in our hypothetical economy, there are N  factor goods, produced by the state, for which a central planner dictates the production levels and prices of each factor good (a factor good is a good produced for the purpose of producing a final good to be consumed; Taylor uses sewing machines as an example of a factor good necessary to produce textiles). Each factory producing a consumer good must in turn each charge a price equal to the sum of the prices of the factor goods necessary to produce the consumer good (or rather, weighted sum, according to the quantities of the input factors needed to produce it). Each factory is ordered to increase production to the point where either there are no more willing buyers or where marginal cost equals marginal revenue (i.e., the point of profit maximization in standard microeconomics), with the price fixed by the central planner. This ensures the optimal allocation of resources for the production of N goods given a set of prices. The issue is: how to find the optimal prices?

 

 

 

To achieve this aim, the central planner starts by simply assigning a price (with the specification that it be at least high enough to cover the cost of production) to each factor good. This first guess of a price would, per Taylor, reflect the best estimate the central planner can come up with a priori, but we can just as well assume that it is essentially random, and this only changes the alacrity with which the system approaches equilibrium. Each citizen would be allotted by the state an income, and would then be allowed to spend it as he pleases to purchase goods from the assorted state-owned firms (I will use the term 'firm' to refer to the state-owned producers and distributors of goods; in other words, any entity that sells goods to either consumers or another entity) at the assigned price. From here, Taylor argues, it is simple to determine whether the price of a factor good is too high or too low: if too high, the firm producing it will accrue a surplus (in other words, the rate at which the good is produced will exceed the rate at which it is sold and consumed); conversely, if the price is too low, the good will sell out faster than inventories can be supplied. The central planner may therefore simply adjust the price upward for any good that is sold out (or for which inventories are declining) and adjust downward the price of any good for which surpluses are accruing. This process is repeated until the inventory for each good is constant.

 

 

 

There are two key innovations to this system: 1) it does not require solving the general equilibrium equations because prices are not determined instantaneously and analytically but rather algorithmically; and 2) information on demand for each good need not be collected and centralized; instead, all the central planner must know at any given time is whether the producer of each good is accruing a surplus or a deficit in the good they are producing. This second innovation is why, according to defenders of market socialism, Hayek's knowledge problem does not apply here, and that merely allowing consumers to purchase what they will provides enough information about whether demand exceeds supply or supply exceeds demand. Moreover, in such a system, prices would only need to be managed for primary factor goods, since, if we accept Taylor's premises, a surplus or dearth of a consumer good will ultimately be passed up the chain to primary factor goods, and if the primary factor goods are optimally priced, then all derivative goods will be too, as, recall, the optimal price of a consumer good is the marginal cost of production, which is equal to cost of purchasing the necessary factors of production.

 

 

 

Criticisms

1. The Discrete Classification of Goods and Granularity of Variation

 

Taylor's pricing system requires a central planner to designate a discrete, finite number of unique goods (factor or consumer goods) to produce, whereas in fact no such discrete set exist; rather, goods vary continuously in quality and other characteristics. This classification (treating goods as a finite, discrete, and static set) inevitably leads to inefficiencies. In reality, there is not a discrete, finite set of goods and services; rather, a good may vary essentially continuously in quality (not merely in the sense of being better or worse, but in characteristics like size, shape, material, etc.). As long as the set of goods is fixed in some predetermined way, the potential for variability will be suppressed, to the detriment of consumer welfare. But even given the shape, size, color, and every other characteristic of a hypothetical widget, the price of that distinct good also varies continuously across space and time. It is from this point that it is clear that Hayek's knowledge problem does indeed apply here. Turning the estimation of demand for a particular good into a binary, algorithmic determination only slightly eases the difficulty faced by the producer. One might argue that, across the dimension of time, iterations of Taylor's process may be done frequently enough to reasonably approximate optimal resource allocation and leave little left on the table, but this would be woefully insufficient: prices of consumer goods vary drastically even within a city, and factor goods do as well, let alone between regions. Moreover, the nature of the variation itself may change over time, so even the discovery of a particular pattern of variation across geography is of only temporary value, and potentially obsolete by the time it can be put into practice.

 

 

 

One might argue the production of factor goods may be greatly centralized, though transportation costs are unlikely to be negligible. However, production and sale of consumer goods must necessarily be disbursed, and though, given a particular distribution of sellers, Taylor's approach may optimize prices, where to sell each good in what quantity requires a knowledge of local demand for each good or service throughout our hypothetical economy, knowledge which, as Hayek articulated, is prohibitively difficult to accumulate centrally. If the location at which a good is produced is sub-optimal, then optimizing the price given the sub-optimal location will still yield a gross inefficiency. The problem goes beyond geographical variability though. Goods vary in a host of characteristics. For example, the optimal material out of which to manufacture a textile product, or the appropriate sizes, or preferred colors, etc. The problem of what price to charge for a generic sweater may solvable by Taylor's method, but not what kind of sweater to manufacture at a given location and time, or where and when is the optimal time (time of day, season, etc.) and place to sell a given kind of sweater. If firms are sub-optimally distributed, constrained in what they can produce by the arbitrary definition of their task by the central planner, and further constrained in the availability of capital goods by the lack of variation in the latter imposed by the equally arbitrary definition of the task assigned to their producers, then Taylor's algorithm may be able to bridge only a fraction of the gap in efficiency between the starting point and that of an efficient market economy.

 

 

 

We might ask, how, within this centrally planned heuristic framework, might we give the production system greater flexibility so as to remedy these shortcomings. What is necessary is a mechanism for experimentation: for producers of consumer goods to vary local prices. For example, suppose our hypothetical country has multiple provinces and each consumer good has a firm selling it in each province. If price varies from one province to another, then, to reach the actual optimum, each provincial seller must be allowed to vary his price from the one indirectly set by the central planner (by setting the factor prices) to determine if the price in his province is below or above the 'national equilibrium' price determined by Taylor's process, so the provincial seller can transmit this information to the producers of the factor goods, who should in turn be able to respond by engaging in price discrimination between the various provincial sellers in order to maximize the efficient allocation of resources. In other words, finding the optimal 'national' price is not sufficient to reach the optimal allocation of resources as long as prices varies within the country at a more granular level. Moreover, provincial firms must be free to vary prices between localities within their respective provinces. So, we find ourselves back with Hayek, where, even with the optimal national price determined, the determination of local prices is still best deferred to the most local market participant - the entity responsible for selling the consumer good at that level. One might mention that, for some goods, prices will be fairly constant from one region to another, but even the determination of the granularity of price variation itself requires local experimentation.

 

 

 

Not only must our system still devolve decision making on local prices onto local sellers to meet Hayek's still very much relevant objections, it must also allow for experimentation at all levels in the variation of the qualities of goods and services, not just to tailor goods to the specific needs or preferences of national citizens, but to allow for the simultaneous production of many different subtypes of each particular good to meet the varied preferences among citizens within the country. Producers and distributors of consumer goods (and arguably factor goods if transportation costs are not negligible) must also be given the freedom to vary the specific location of production and/or distribution, as, even if the price set by the central authority is 'nationally optimal', demand for a good is almost certainly not homogeneous throughout the country. And because there is essentially no limit to the granularity with which prices and quality preferences can vary, our system must allow for the unlimited establishment of new production/distribution franchises for the system to effectively probe for the appropriate level of granularity of variation in price and quality. Our system must also, of course, allow for the timely closing of franchises at which cost exceeds revenue. So, either firms must be free to open up new franchises, or citizens must be free to organize their own firms. But this introduces the new problem of how startup costs are to be met. The central planner might allot temporary funding until the new firm or franchise (remember, existing firms are not, in a socialist system, allowed to accrue surplus wealth, and so will be unable to absorb startup costs for new franchises) either finds itself earning revenue equal to cost, or is decidedly unable to do so. Of course, the time frame for making such a decision must vary from one good to another, as startup costs vary from one good to another. How to we determine the appropriate time-frame for running these production experiments? How do we decide whether a firm is likely worth starting up to begin with? For if our central planner takes all comers, it may become extremely expensive, as there would be no incentive for anyone to avoid frivolously starting an unnecessary firm. The issue of incentives will be dealt with in a future post, but suffice it to say, we consistently find ourselves in a position of needing to defer to more local market participants to most efficiently experiment with local price, quality, and geographical distribution so as to bring the economy closer to its actual optimal point.

 

 

 

2. The Pricing of Factor Goods

 

Another major flaw in Taylor's heuristic pricing system is his assumption that the value of a factor good is determined by the prices of the consumer goods derived from it. Superficially his claim may make sense, and is often repeated by market socialists in criticizing von Mises's claim that the private ownership of the means of production (and the accretion of profit therefrom) is necessary for a market for factor goods to exist and therefore for factor goods to be optimally priced. However, Taylor's rejection of von Mises's claim is incorrect. The value of a factor good (or, to be in strict accordance with marginal utility theory of value, we might say, how much producers of consumer goods will tend to value a factor good) is not merely determined by present prices and demand (that is, current expected revenue) but rather by the expected revenue (itself determined by expected demand and expected cost) going forward well into the future. This is a very key point: an efficient market is forward looking. Resources must not merely be allocated across the many (as I argued above, virtually infinite) different varieties of goods and services, but also efficiently allocated across time. For example, we positively do not want our grain farms to accrue no surpluses over time, as demand for grain, and cost of production for that matter, tend to be cyclical. The price which, in Taylor's model, 'optimizes' grain production, in the long run, leads to underproduction in the future. We should want a surplus to be accrued to be stored in granaries for when either demand or cost of production increases (such as when winter comes, or when a famine occurs).

 

 

 

The other side of this coin is that demand or cost of production may, for some reason, be expected to decline in the future (either in a cyclical pattern or secular trend), meaning the immediately optimal price of the good, as determined by the heuristic approach, will lead to overproduction. This is obviously the case for durable goods whose utility is spread across a long time-frame: if we might produce a car next year much more cheaply than this year, and consumers' time-preference utility function extends out as far out as, say, 10 years, and the car will last at least 10 years, the loss from the car not being available for the next year may be superseded by the gains from producing it much more cheaply and having it available for the remaining 9 years, even completely discounting the consumers' use of the car beyond the 10th year. However, even the price and production level of non-durable goods must be optimized over time if startup or expansion costs are not negligible (and they rarely are). Consider perishable food crops: if the cost of planting and cultivating the crop (i.e., the preparatory costs of ultimate production) are significant - and especially if there's a significant time delay before amount of goods we decide to produce will actually make it to market - then even if the good is very perishable, we must take into account how much demand there will be for this good in, say, six months, in addition to how much there is today; for if demand goes up in six months, then our decision to allocate only enough land and labor to produce the optimal amount today will mean we will be underproducing in six months; once again, it may be worth it to allocate 'too much' land, labor, etc. to the production of this good today in order to produce the optimal amount in six months. In other words, having idle resources today may actually be productive in the long run as such idle resources may find efficient use in the future, and disassembling such idle resources for some other purpose today may optimize production today but, in the long run, yield a sub-optimal state (see W.H. Hutt's The Theory of Idle Resources for a more detailed analysis of idle resources).

 

 

 

So, it is not sufficient that we merely optimize prices of factor goods based on the immediate demand for consumer goods. Once again, we might ask, how might we alter the heuristic pricing system so that we can actually optimally price factor goods, accounting not just for current demand and costs, but future demand and costs as well? Well, for durable goods, we may do this simply by allowing citizens to buy up as much of the good as they please, allowing them to store the good ('hoard' it, as some would say), so that they may sell it in the future on the private market should demand (or cost of production) go up. Basically, we would be permitting citizens to engaging in arbitraging across time. A couple issues, however, would likely be present: it is highly doubtful that a socialist society would be particularly amenable to the practices of 'hording' and 'price gouging' and let them serve their purposes. Additionally, allowing consumers to store durable goods in bulk would arguably be in violation of the principle of forbidding private ownership of the means of production. Our hypothetical hoarders would own warehouses and granaries and, should their investments prove wise, would earn what we might call profit. It would be ironic that the only citizens earning a profit in our socialist society would be those engaged in mere speculation (to be sure, speculation is, contrary to conventional wisdom, often quite productive, as it can serve, again, as the equivalent of arbitrage across time). There is also a slippery slope here: if we accept that it is beneficial to allow speculative purchase if not production, why is it not also beneficial to allow private citizens to buy up land and allot it to the production of perishable goods on the expectation that the demand for them will increase in the future, so that they might earn a speculative profit? To allow them to do so would be just as beneficial to society, and yet would clearly be forbidden by the public ownership of primary factors of production.

 

 

 

So, if we maintain public ownership of primary factors, how can the central planner efficiently allocate resources across time? How can we estimate whether a given price for a given good will be too low or too high in the future? Once again, for consumer goods, we might do best to defer this decision-making to the agents closest to consumption, as they will be best able to predict whether demand for a particular good (or very particular subtype of a good) at a particular location or among a particular consumer demographic. On the factor side of things, it is also reasonable that we give those involved in managing the production of the factor good the freedom to speculate on future costs of production: those who grow soybeans for a living in a particular location will tend to have the best idea of, say, how weather patterns affect the cost of production of soybeans. Another necessary feature, however, is multiplicity of experimentation methods. The usefulness of private investment markets is derived in large part from the fact that countless different firms and individual investors are constantly experimenting in a wide array of methods for speculating on future demand and costs. This allows the economy to capitalize on the Darwinian process in which those most prescient in their anticipation of future demand and costs are most likely to succeed and reinvest the surplus earnings. Speculation about future prices is, then, not a binary choice like deciding how to alter current prices in Taylor's method; rather, there are an infinite number of possible patterns that may occur across time. But markets are not merely Darwinian: they are also Lamarckian. Unsuccessful investors (or new, 'naive' investors) learn from successful ones and imitate them (or their methods or algorithms) and often vary them slightly to see if they can even further improve upon them. Allowing each firm to engage in speculation (i.e., allowing its managers to factor their own estimates of future demand and production cost into their pricing) is the only way to maximize the capture of local information about likely future trends. We might in theory allow only a fraction of our firms for each good to experiment with prices by speculating about future changes in demand and costs, and this would meet the requirement of having multiple 'competing' approaches to speculation, from which the most successful may be periodically ordered to be replicated among the other firms. However, we once again run into the problem of granularity: demand and cost trends are often heterogeneous across the country, so we cannot merely pick the most successful pricing mechanism among 100 experimenting firms and apply it everywhere; we might allow 100 firms in each province to experiment, but what about intra-province variation in demand and cost trends? Once again, as we change are system, we approach the point where each firm is permitted to experiment. Worries about the cost redundant experimentation may be answered by the fact that no firm is required to develop its own method for anticipating changes in demand and cost; it may simply imitate that of another firm that is nearby or highly similar. And who is in the best position to ascertain whether one firm's expectations can be reasonably extrapolated to another? The managers of the latter firm (and perhaps the former one as well), of course. They are the ones who will most thoroughly and immediately observe the similarities between their firm's circumstances and those of one they are considering imitating, and observe when or where such similarities cease to apply, and being able to respond in the most timely manner. Thus, devolving the task of factoring future costs and demand upon each individual firm itself is the most efficient way to allocate resources across time for both factor goods and consumer goods as well, as it preserves both the Darwinian and Lamarckian properties of the market.

 

 

 

3. Pricing Labor

 

Labor is potentially the most important factor good, and this fact often proves an inconvenience for socialist theory, as the efficient pricing of labor often contradicts with socialist aspirations toward equality of outcome. Taylor's heuristic pricing algorithm is no exception. Indeed, the problem insufficient granularity in pricing of factor inputs is most apparent with labor because the heterogeneity of the value of labor is so self-evident. Taylor does not mention applying his heuristic algorithm to determine the price of labor generally (that is, the universal wage for all laborers), presumably for the obvious reason that, since the amount earned in income (in other words, labor price times quantity) necessarily equals revenue, one universal wage differs from another only nominally, not in real purchasing power. Therefore, the pricing labor is purely a relative issue: how much should each type of laborer be paid relative to each other type? (note that I am assuming monetary neutrality for the sake of simplicity)

 

 

 

Incidentally, though Taylor does not suggest it, if we define a finite set of occupations (like with other factor goods), we can just as easily apply Taylor's algorithm here as elsewhere. Occupations vary in difficulty and skill. If we accept the rather trivial assumption that people are at all motivated by material compensation (i.e., that ceteris paribus, a person will tend to prefer one hour of easy work to one hour of hard work if they pay the same wage), if one occupation is under-priced relative to another (suppose, for example, they pay the same wage, but the former is more difficult, dangerous, etc.) then workers will flee the under-priced occupation for the over-priced one. The surplus or dearth of workers seeking each occupation can therefore inform the gradual adjustment of wages in each occupation toward the optimal relative pricing of the many species of labor, which is conducive to the optimal allocation of resources and the maximization of social welfare. Indeed, since the utility of optimally pricing and allocating other factors of production largely hinges on whether these other factors are being mixed with the right kind of labor, the efficient pricing of labor is likely more important than the efficient pricing all other factors combined. This is not so much a criticism of centrally coordinated heuristic pricing in theory as a challenge to the selective willingness to apply it. All of the same economic arguments behind applying Taylor's pricing algorithm to other factors are equally applicable to labor, and its application to labor is at least as conducive to improving general welfare. At this point, however, socialists tend to break with economic reasoning in favor of political axioms. Most socialists, I expect, would not permit the heuristic search for optimal labor pricing because, to put it bluntly, the sound economizing of labor (a scarce resource as much as anything) contradicts their political/moral valuation of material equality for its own sake, and they are generally willing to sacrifice the former the sake of the latter.

 

 

 

It is also worth noting that the above arguments for permitting sellers (laborers) to experiment with setting their own prices, and the purchasers of factor goods (firms) be permitted to vary the prices of their products and the quantity of the input factor purchased (and consequently the amount of goods produced) in accord with labor prices are every bit as applicable to labor as to other factor goods, as labor varies in character, in production cost (training and education), and in optimal price (with geography and time) just as other factors do. Therefore, if our goal is still the optimal allocation of resources for the maximization of general welfare, we should also devolve decision-making on the cultivation, employment, and pricing of labor, in all its forms, to each local buyer and seller of labor to effectively account for all the axes on which labor may vary, rather than compel compliance with centrally determined wages.

 

 

 

4. The Problem of Centrally Coordinated Demand and Cost Assessment

 

Finally, I will argue that the more centrally planned alternative the 'devolutionary' solutions to the problems inherent in a centrally planned, heuristic pricing system is unavoidable inefficient. We might imagine that the central planner could establish a Bureau of Demand and Cost Assessment, which would have branches all across the country assessing variation in demand across dimensions of geography, consumer demographic, product characteristics, and time. Of course, positing such a bureau to avoid the devolution of decision-making about pricing, production, and ultimately resource allocation, and maintain centrality, leads us right back into Hayek's knowledge problem, demonstrating that, contrary to what some market socialists have claimed, Hayek's point about the decentralized distribution of knowledge prevails as a devastating impediment to centralization. I will add that the sheer granularity of variation necessitates that the cost of running a Bureau of Demand and Cost Assessment increases exponentially the more one expands its operations to approach efficiently granular decision-making. We might have one Bureau branch in each province directing provincial firms in matters based on research about province trends, but then variation within each province would be ignored. We might have a branch in each city and town, but anyone can readily observe that prices and availability for goods tend to vary (even among stores of the same chain) even within a city. And with each added level of subsidiarity, the cost of running the Bureau increases in proportion to the number of provinces, the number of cities, the number of neighborhoods. Eventually, to truly optimize pricing and production, we approach the point where each firm has a concomitant Bureau branch analyzing how demand and costs will change over time for that firm. At that point, however, it would be easily argued we have an entirely redundant bureaucracy consuming a massive amount of resources. Everything the Bureau branch is observing is also being observed by those managing production themselves, so why not simply merge the task of producing with the task of determine what to produce, where and how to produce it, and how to price it?

 

Conclusion

Taylor's method for optimizing prices is meant to be a robust form of market socialism that allows central planners to exploit the information provided by the market (the relative demand for goods) rather than attempting to fully suppress it. Taylor and other market socialists break with Marx in that they accept that economic categories such as prices, profit, rent, etc. cannot be practically dispensed with as long as there is scarcity. Instead, Taylor and other advocates of market socialism seek use market derived information - here, the observed relative demand for final consumer goods - to inform centrally coordinated pricing orchestrated heuristically rather than analytically.

 

 

 

Two major themes recur in my criticism of centrally coordinated trial and error pricing. First, though heuristic pricing based on relative demand is superior to central pricing approaches that ignore organic demand for the various goods produced, a damning flaw in central pricing remains equally applicable here: the inevitable variation in both demand and cost of production within any defined 'class' of good with potential variation in the characteristics of the good, and variation in demand and cost across other dimensions like geography, consumer demographic, and - as all of these factors are dynamic rather than static - across time. Taylor's system of assigning prices to a finite set of of classes of goods, the production of which is distributed (geographically, temporally, and across other dimensions) in a fixed manner, is inevitably sub-optimal in a way that can be readily improved simply by devolving decision-making onto ever more local, rather than central, market participants. The optimal level of granularity in price and quality variation is ultimately achieved by permitting individual firms to make decisions in a manner unmoored from the edicts of the central planner.

 

 

 

The second theme is the necessity of constant experimentation. There is no single, optimal national price for a given good, nor is there really a nationally optimal way to produce a given good. What is optimal varies, as mentioned above, across geography, and what is optimal even locally at the moment won't be optimal in the future. To find a more granular optimum, and to continue to rediscover it most efficiently as it moves, firms must always and everywhere be free to experiment by varying products, prices, and location. The freedom to conduct such constant experimentation requires permitting firms to ignore the central planner, and to the extent that we allow such experimentation, we render the central planner redundant.

 

 

 

In the course of my criticism, I have offered putative solutions that might mitigate the shortcomings of centrally coordinated pricing. But each solution that improves the efficiency and dynamism of the pricing system is itself a clear deviation from central pricing toward a free market economy. Inasmuch as we alter the central pricing system to make it more efficient, we march ever closer to the very system the central pricing system itself was conceived in order to avoid.

 

 

 

Lastly, I will note that there are other issues with the centrally planned approach to pricing that I have not mentioned here that relate to issues with incentives, such as who bears the cost of failed experimental underpricing and who keeps the gains from successful experimental overpricing. I have instead decided to save those for a future post, dealing exclusively with incentive-related issues.

 

 

 

 

 

References:

 

1. Taylor, Fred. "The Guidance of Production in a Socialist State"

 

2. von Mises, Ludwig. "Socialism: An Economic and Sociological Analysis."

 

3. Hayek, Friedrich. "The Use of Knowledge in Society."

 

4. Jossa and Cuomo. "The Economic Theory of Socialism and the Labour-managed Firm."

 

5. Hutt, William H. "The Theory of Idle Resources."

 

6. DeLong, Bradford. "James Scott and Friedrich Hayek." http://delong.typepad.com/sdj/2007/10/james-scott-and.html