Economies of Scale, Economies of Speed, and Push Distribution

Download: MOLOCH: Mass-Production Industry as a Statist Construct.

I. The Origins of Sloanist Mass Production

A Fork in the Road
A Wrong Turn
The Role of the State in Tipping the Balance

II. The Institutional Imperatives of Sloanism

Economies of Scale, Economies of Speed, and Push Distribution
Microeconomic Institutional Forms for Providing Stability
Mass Consumption to Absorb Surplus
Political Capitalism
State Action to Absorb Surplus: Imperialism
State Action to Absorb Surplus: Creation of New Industries

III. Conclusion

Alfred Chandler, like Galbraith, was thoroughly sold on the greater efficiencies of the large corporation. He argued that the modern multi-unit enterprise arose when administrative coordination “permitted” greater efficiencies. [45]

By linking the administration of producing units with buying and distributing units, costs for information on markets and sources of supply were reduced. Of much greater significance, the internalization of many units permitted the flow of goods from one unit to another to be administratively coordinated. More effective scheduling of flows achieved a more intensive use of facilities and personnel employed in the processes of production and so increased productivity and reduced costs. [46]

Organizationally, output was expanded through improved design of manufacturing or processing plants and by innovations in managerial practices and procedures required to synchronize flaws and supervise the work force. Increases in productivity also depend on the skills and abilities of the managers and the workers and the continuing improvement of their skills over time. Each of these factors or any combination of them helped to increase the speed and volume of the flow, or what some processors call the “throughput,” of materials within a single plant or works…. [47]

Integration of mass production with mass distribution afforded an opportunity for manufacturers to lower costs and increase productivity through more effective administration of the processes of production and distribution and coordination of the flow of goods through them. Yet the first industrialists to integrate the two basic sets of processes did not do so to exploit such economies. They did so because existing marketers were unable to sell and distribute products in the volume they were produced. [48]

The mass-production factory achieved “economies of speed” from “greatly increasing the daily use of equipment and personnel.” [49] (Of course, Chandler starts by assuming the greater inherent efficiency of capital-intensive modes of production, which then require “economies of speed” to reduce unit costs from the expensive capital assets).

What Chandler meant by “economies of speed” was entirely different from lean production’s understanding of flow. Chandler’s meaning is suggested by his celebration of the new corporate managers who “developed techniques to purchase, store, and move huge stocks of raw and semifinished materials. In order to maintain a more certain flow of goods, they often operated fleets of railroad cars and transportation equipment.” [50] In other words, both the standard Sloanist model of enormous buffer stocks of unfinished goods, and warehouses full of finished goods awaiting orders — and the faux “lean” model in which inventory is swept under the rug and moved into warehouses on wheels and in container-ships.

(The reader may be puzzled or even annoyed by my repeated use of the term “Sloanism.” I got it from the insightful commentary of Eric Husman at GrimReader blog, in which he treats the production and accounting methods of General Motors as paradigmatic of 20th century American mass-production industry, and contrasts them with the lean methods popularly identified with Taichi Ohno’s Toyota production system.)

“Sloanism” refers, in particular, to the management accounting system identified with General Motors. It was first developed by Brown at DuPont, and brought to GM when DuPont acquired a controlling share of the company and put Alfred Sloan in charge. Brown’s management accounting system, whose perverse incentives are dissected in detail by William Waddell and Norman Bodek in Rebirth of American Industry, became the basis of the Generally Accepted Accounting Principles (GAAP) that prevail throughout American corporate management.

In Sloanist management accounting, inventory is counted as an asset “with the same liquidity as cash.” Regardless of whether a current output is needed to fill an order, the producing department sends it to inventory and is credited for it. Under the practice of “overhead absorption,” all production costs are fully incorporated into the price of goods “sold” to inventory, at which point they count as an asset on the balance sheet.

With inventory declared to be an asset with the same liquidity as cash, it did not really matter whether the next ‘cost center,’ department, plant, or division actually needed the output right away in order to consummate one of these paper sales. The producing department put the output into inventory and took credit. [51]

…Expenses go down…, while inventory goes up, simply by moving a skid full of material a few operations down the stream. In fact, expenses can go down and ROI can improve even when the plant pays an overtime premium to work on material that is not needed; or if the plant uses defective material in production and a large percentage of the output from production must be scrapped. [52]

In other words, by the Sloanist accounting principles predominant in American industry, the expenditure of money on inputs is by definition the creation of value. As Waddell described it at his blog,

companies can make a bunch of stuff, assign huge buckets of fixed overhead to it and move those overheads over to the balance sheet, making themselves look more profitable.

It’s a system summed up perfectly by Paul Goodman’s notion of the culture of cost-plus. And as Waddell points out, the GDP as a metric depends on the same GAAP assumptions as American industry: it counts expenditure on inputs, by definition, as the creation of wealth. [53] The American corporate economy is governed by a set of metrics much like that of the Soviet planned economy. A given “output” represents an economic value equal to the inputs it consumes, regardless of whether anyone actually wants the output, whether they work, or whether they could have been produced with a fraction of the inputs.

American factories frequently have warehouses filled with millions of dollars worth of obsolete inventory, which is still there “to avoid having to reduce profits this quarter by writing it off.” When the corporation finally does have to adjust to reality, the result is costly write-downs of inventory.

It did not take much of a mathematician to figure out that, if all you really care about is the cost of performing one operation to a part, and you were allowed to make money by doing that single operation as cheaply as possible and then calling the partially complete product an asset, it would be cheaper to make them a bunch at a time.

It stood to reason that spreading set-up costs over many parts was cheaper than having to set-up for just a few even if it meant making more parts than you needed for a long time. It also made sense, if you could make enough parts all at once, to just make them cheaply, and then sort out the bad ones later.

Across the board, batches became the norm because the direct cost of batches was cheap and they could be immediately turned into money — at least as far as Mr. DuPont was concerned — by classifying them as work-in-process inventory. [54]

Under the Sloan system, if a machine can be run at a certain speed, it must be run at that speed to maximize efficiency. And the only way to increase efficiency is to increase the speed at which individual machines can be run. [55] The Sloan system focuses, exclusively, on labor savings “perceived to be attainable only through faster machines. Never mind that faster machines build inventory faster, as well.” [56]

The lean approach has its own “economies of speed,” but they are the direct opposite of the Sloanist approach. The Sloanist approach focuses on maximizing economies of speed in terms of the unit cost of a particular machine, without regard to the inventories of unfinished goods that must accumulate as buffer stocks as a result, and all the other enormous eddies in the flow of production. As the authors of Natural Capitalism put it, it attempts to optimize each step of the production process in isolation, “thereby pessimizing the entire system.” A machine can reduce the labor cost of one step by running at enormous speeds, and yet be out of sync with the overall process. [57] Waddell and Bodek give the example of Ernie Breech, sent from GM to “save” Ford, demanding a plant manager tell him the cost of manufacturing the steering wheel so he could calculate ROI for that step of the process. The plant manager was at a loss trying to figure out what Breech wanted: did he think steering wheel production was a bottleneck in production flow, or what? But for Breech, if the unit cost of that machine and the direct cost of the labor working it were low enough compared to the “value” of the steering wheels “sold” to inventory, that was all that mattered. Under the Sloan accounting system, producing a steering wheel — even in isolation, and regardless of what was done with it or whether there was an order for the car it was a part of — was a money-making proposition. “Credit for that work — it looks like a payment on the manufacturing budget — is given for performing that simple task because it moves money from expenses to assets. [58]

The lean approach, in contrast, gears production flow to orders, and then sizes individual machines and steps in the production process to the volume of overall flow. Under lean thinking, it’s better to have a less specialized machine with a lower rate of output, in order to avoid an individual step out of proportion to the overall production flow. This is what the Toyota Production System calls takt: pacing the output of each stage of production to meet the needs of the next stage, and pacing the overall flow of all the stages in accordance with current orders. [59] In a Sloan factory, the management would select machinery to produce the entire production run “as fast as they humanly could, then sort out the pieces and put things together later.” [60]

To quote the authors of Natural Capitalism again: “The essence of the lean approach is that in almost all modern manufacturing,”

the combined and often synergistic benefits of the lower capital investment, greater flexibility, often higher reliability, lower inventory cost, and lower shipping cost of much smaller and more localized production equipment will far outweigh any modest decreases in its narrowly defined “efficiency” per process step. It’s more efficient overall, in resources and time and money, to scale production properly, using flexible machines that can quickly shift between products. By doing so, all the different processing steps can be carred out immediately adjacent to one another with the product kept in continuous flow. The goal is to have no stops, no delays, no backflows, no inventories, no expediting, no bottlenecks, no buffer stocks, and no muda [waste]. [61]

The contrast is illustrated by a couple of examples from Natural Capitalism: an overly “efficient” grinding machine at Pratt & Whitney, and a cola bottling machine likewise oversized in relation to its task:

The world’s largest maker of jet engines for aircraft had paid $80 million for a “monument” — state-of-the-art German robotic grinders to make turbine blades. The grinders were wonderfully fast, but their complex computer controls required about as many technicians as the old manual production system had required machinists. Moreover, the fast grinders required supporting processes that were costly and polluting. Since the fast grinders were meant to produce big, uniform batches of product, but Pratt & Whitney needed agile production of small, diverse batches, the twelve fancy grinders were replaced with eight simple ones costing one-fourth as much. Grinding time increased from 3 to 75 minutes, but the throughput time for the entire process decreased from 10 days to 75 minutes because the nasty supporting processes were eliminated. Viewed from the whole-system perspective of the complete production process, not just the grinding step, the big machines had been so fast that they slowed down the process too much, and so automated that they required too many workers. The revised production system, using a high-wage traditional workforce and simple machines, produced $1 billion of annual value in a single room easily surveyable from a doorway. It cost half as much, worked 100 times faster, cut changeover time from 8 hours to 100 seconds, and would have repaid its conversion costs in a year even if the sophisticated grinders were simply scrapped. [62]

In the cola industry, the problem is “the mismatch between a very small-scale operation — drinking a can of cola — and a very large-scale one, producing it.” The most “efficient” large-scale bottling machine creates enormous batches that are out of scale with the distribution system, and result in higher unit costs overall than would modest-sized local machines that could immediately scale production to demand-pull. The reason is the excess inventories that glut the system, and the “pervasive costs and losses of handling, transport, and storage between all the elephantine parts of the production process.” As a result, “the giant cola-canning machine may well cost more per delivered can than a small, slow, unsophisticated machine that produces the cans of cola locally and immediately on receiving an order from the retailer.” [63]

In a genuine lean factory, managers are hounded in daily meetings about meeting the numbers for inventory reduction and reduction of cycle time, in the same way that they’re hounded on a daily basis to reduce direct labor hours and increase ROI in a Sloanist factory. James Womack et al, in The Machine That Changed the World, recount an amusing anecdote about a delegation of lean production students from Corporate America touring a Toyota plant. Reading a question on their survey form as to how many days of inventory were in the plant, the Toyota manager politely asked whether the translator could have meant minutes of inventory. [64]

As Mumford put it, “Measured by effective work, that is, human effort transformed into direct subsistence or into durable works of art and technics, the relative gains of the new industry were pitifully small.” [65] The amount of wasted resources and crystallized labor embodied in the enormous warehouses of Sloanist factories and the enormous stocks of goods in process, the mushrooming cost of marketing, the “warehouses on wheels,” and the mountains of discarded goods in the landfills that could have been repaired for a tiny fraction of the cost of replacing them, easily outweigh the savings in unit costs from mass production itself. The cost savings from mass production are more than offset by the costs of mass distribution.

Chandler’s model of production resulted in the adoption of increasingly specialized, asset-specific production machinery:

The large industrial enterprise continued to flourish when it used capital-intensive, energy-consuming, continuous or large-batch production technology to produce for mass markets. [66]

The ratio of capital to labor, materials to labor, energy to labor, and managers to labor for each unit of output became higher. Such high-volume industries soon became capital-intensive, energy-intensive, and manager-intensive. [67]

Of course this view is fundamentally wrong-headed. To regard a particular machine as “more efficient” based on its unit costs taken in isolation is sheer idiocy. If the costs of idle capacity are so great as to elevate unit costs above those of less specialized machinery, at the levels of spontaneous demand occurring without push marketing, and if the market area required for full utilization of capacity results in distribution costs greater than the unit cost savings from specialized machinery, then the expensive product-specific machinery is, in fact, less efficient.

Galbraith and Chandler wrote as though the adoption of the machinery were enough to automatically increase efficiency, in and of itself, regardless of how much money had to be spent elsewhere to “save” that money.

But if we approach things from the opposite direction, we can see that flexible manufacturing with easily redeployable assets makes it feasible to shift quickly from product to product in the face of changing demand, and thus eliminates the imperative of controlling the market. As Barry Stein said,

if firms could respond to local conditions, they would not need to control them. If they must control markets, then it is a reflection of their lack of ability to be adequately responsive. [68]

…Consumer needs, if they are to be supplied efficiently, call increasingly for organizations that are more flexibly arranged and in more direct contact with those customers. The essence of planning, under conditions of increasing uncertainty, is to seek better ways for those who have the needs to influence or control the productive apparatus more effectively, not less.

Under conditions of rapid environmental change, implementing such planning is possible only if the “distance” between those supplied and the locus of decision-making on the part of those producing is reduced…. But it can be shown easily in information theory that the feedback — information linking the environment and the organization attempting to service that environment — necessarily becomes less accurate or less complete as the rate of change of data increases, or as the number of steps in the information transfer process continues.

Stein suggested that Galbraith’s solution was to suppress the turbulence: “to control the changes, in kind and extent, that the society will undergo.” [69] But far better, he argues, would be “a value shift that integrates the organization and the environment it serves.”

This problem is to be solved not by the hope of better planning on a large scale…, but by the better integration of productive enterprises with the elements of society needing that production.

Under conditions of rapid change in an affluent and complex society, the only means available for meeting differentiated and fluid needs is an array of producing units small enough to be in close contact with their customers, flexible enough to produce for their demands, and able to do so in a relatively short time…. It is a contradiction in terms to speak of the necessity for units large enough to control their environment, but producing products which in fact no one may want! [70]

As to the problem of planning — large firms are said to be needed here because the requirements of sophisticated technology and increasingly specialized knowledge call for long lead times to develop, design, and produce products. Firms must therefore have enough control over the market to assure that the demand needed to justify that time-consuming and costly investment will exist. This argument rests on a foundation of sand; first, because the needs of society should precede, not follow, decisions about what to produce, and second, because the data do not substantiate the need for large production organizations except in rare and unusual instances, like space flight. On the contrary, planning for social needs requires organizations and decision-making capabilities in which the feedback and interplay between productive enterprises and the market in question is accurate and timely — conditions more consistent with smaller organizations than large ones. [71]

In short, mass production requires supply-push distribution to guarantee a market before production takes place.

Although Galbraith and Chandler commonly justified the corporation’s power over the market in terms of its social benefits, they had things exactly backward. The “technostructure” can survive because it is enabled to be less responsive to consumer demand. An oligopoly firm in a cartelized industry, in which massive, inefficient bureaucratic corporations share the same bureaucratic culture, is protected from competition. The “innovations” Chandler so prizes are able to succeed because they are determined by the organization for its own purposes, and the organization has the power to impose top-down “change” on a cartelized market, with little regard to consumer preferences, instead of responding flexibly to them. The large corporate organization is not more efficient at accomplishing goals received from outside; it is more efficient at accomplishing goals it sets for itself for its own purposes, and then using its power to adapt the rest of society to those goals.

So to turn to our original point, the apostles of mass production have all, at least tacitly, identified the superior efficiency of the large corporation with its control over the external environment. Sloanist mass production subordinates the consumer, and the rest of outside society, to the institutional needs of the corporation.

Chandler himself admitted as much, in discussing what he called a strategy of “productive expansion.” Big business added new outlets that permitted it to make “more complete use” of its “centralized services and facilities.” [72] In other words, “efficiency” is defined by the existence of “centralized facilities,” as such; efficiency is then promoted by finding ways to make people buy the stuff the centralized facilities can produce running at full capacity. These theories amount, in practice, to a circular argument that oligopoly capitalism is “successful” because it is most efficient at achieving the ends of oligopoly capitalism. Chandler’s version of “successful development” is a roaring success indeed, if we start with the assumption that society should be reengineered to desire what the technostructure wants to produce.


45. Chandler, The Visible Hand, p. 6.

46. Ibid., pp. 6-7.

47. Ibid., p. 241.

48. Ibid., p. 287.

49. Ibid., p. 244.

50. Ibid., p. 412.

51. Waddell and Bodek, p. 75.

52. Ibid., p. 140.

53. William Waddell, “The Irrelevance of the Economists,” Evolving Excellence, May 6, 2009

54. Waddell and Bodek, p. 98.

55. Ibid., p. 122.

56. Ibid., p. 119.

57. Paul Hawken, Amory Lovins, and L. Hunter Lovins, Natural Capitalism: Creating the Next Industrial Revolution (Boston, New York, London: Little, Brown, and Company, 1999), pp. 129-30.

58. Waddell and Bodek, pp. 89, 92.

59. Ibid., pp. 122-123.

60. Ibid., p. 39.

61. Hawken et al, pp. 129-130.

62. Ibid., pp. 128-129.

63. Ibid., p. 129.

64. James P. Womack, Daniel T. Jones, Daniel Roos, The Machine That Changed the World (New York: MacMillan, 1990), p. 80.

65. Mumford, Technics and Civilization, p. 196.

66. Ibid., p. 347.

67. Ibid., p. 241.

68. Barry Stein, Size, Efficiency, and Community Enterprise (Cambridge: Center for Community Economic Development, 1974), p. 41.

69. Ibid., p. 43.

70. Ibid., p. 44.

71. Ibid., p. 58.

72. Chandler, The Visible Hand, p. 487.

Free Markets & Capitalism?
Markets Not Capitalism
Organization Theory
Conscience of an Anarchist