Speedup or Closure: The Global Shop Floor of 1970s Silicon Valley

Hostname: page-component-8678677fb7-xsjpr Total loading time: 0 Render date: 2026-10-08T11:35:43.101Z Has data issue: false hasContentIssue false

Home
>Journals
>International Labor and Working-Class History
>FirstView
>Speedup or Closure: The Global Shop Floor of 1970s...

You have Access Open access

English
Français

International Labor and Working-Class History

Article contents

Published online by Cambridge University Press: 08 October 2026

Joseph Daniel Parziale [Opens in a new window]

Show author details

Joseph Daniel Parziale* Affiliation:

History, University of Illinois Chicago, Chicago, IL, USA

Email:

Article

Article contents

Save PDF (0.33 mb) View PDF [Opens in a new window]

Copy
Post
Mail

Rights & Permissions [Opens in a new window]

Extract

Between 1968 and 1978, the overseas plants of Signetics Corporation, a manufacturer of integrated circuits (ICs) based in Silicon Valley, were host to numerous job actions and genuine organizing campaigns by workers, including at least two strikes that went a long way toward stifling production. The results of these actions were, to say the least, mixed. Some succeeded in yielding concessions, particularly with respect to wages; others led directly to plant closures despite the relatively modest demands involved. Exploring these confrontations and their disparate outcomes provides a rare window onto class conflict in the semiconductor industry and allows us to test where existing explanations of deindustrialization apply to such high-tech production and where new ones might be needed.

Information

Type Special Feature

Information

International Labor and Working-Class History , First View , pp. 1 - 30

DOI: https://doi.org/10.1017/S0147547926100313 [Opens in a new window]

Creative Commons

This is an Open Access article, distributed under the terms of the Creative Commons Attribution-NoDerivatives licence (http://creativecommons.org/licenses/by-nd/4.0), which permits re-use, distribution, and reproduction in any medium, provided that no alterations are made and the original article is properly cited.

Scholars who have written about semiconductor production workers have tended to emphasize worker quiescence or highlight the failure of unions to penetrate the industry.Footnote 1 This perspective is obviously not without justification, and viewing the early chip industry as an origin point for the more or less open shop that persists in today’s Silicon Valley makes good sense. However, by focusing primarily on the United States—which mainly hosted highly automated segments of the production process since relatively early—scholars have neglected the instances where struggles have taken place, namely at the chipmakers’ foreign assembly, test, and packaging (ATP) plants.

This article focuses primarily on several case studies at the Signetics overseas plants to trace semiconductor capital’s constant search for a workforce over which it could assert a large measure of control. Consistent with scholarship on deindustrialization of the Global North in the second half of the twentieth century, I find that the semiconductor industry has both stimulated class formation in each new area it has begun production and attempted to resolve the conflicts it thereby created by shifting production to new low-wage areas with large reserve armies of labor.Footnote 2 However, I also see evidence that is suggestive of investment strategies vis-à-vis labor that are particular to the industry (or, perhaps, to certain segments of electronic components manufacture). More specifically, I argue that, owing to the exceptionally short product lifecycles involved, capital calculated that it was economical to rely on sweated labor for assembly until capitalists were confident that automated ATP processes were sufficiently flexible and reliable to justify the risk of investing the fixed capital required. And although chipmakers were like other industries in the sense that they could not simply perform overnight vanishing acts at the first sign of trouble and relocate to any corner of the globe that they pleased, the extremely light shipping weight of their products did make short-notice changes to their companies’ geospatial production patterns easier—and the threats to make such changes more plausible.

One of the key factors—perhaps the decisive factor—that determined whether capital remained in place and stuck out worker activism was the political context in which that activism took place. In one case, for example, when production workers formed a council at the Signetics plant in Setúbal, Portugal, during the left-led Carnation Revolution, management resolved that its shop-floor authority was sufficiently threatened that it undertook to rapidly wind down production, even though the company appeared to lack the capacity to service all its markets in the plant’s absence, leading to a near-term production bottleneck. In Korea, by contrast, with the state under the control of a militarist dictatorship that often intervened to suppress independent union activity, Signetics proved willing to arrive at corporatist compromises with the state-dominated union that claimed to represent workers. There, the company tolerated worker protests that went a longer way than the Portuguese actions had toward temporarily throttling production at the plant. As we will see, these protests in 1968 and, especially, one in 1977, were well-timed, but ultimately, the plant remained operational because management trusted that the Korean state could be relied upon to re-establish control if things escalated.

As an assessment of capital mobility, technological unemployment, and deindustrialization, I follow Jefferson Cowie in finding that “capital includes more than equipment, buildings, parts, and raw materials; it embodies also a complex social relationship.” In finding new areas of production to escape workers’ demands for control over the shop floor and a larger share of the proceeds those workers generate for investors, capital virtually always sows the seeds of new tensions by inscribing “an incipient class awareness” that will serve to “erode the very social control management gained” by relocating production.Footnote 3 This effect can be especially marked where exploitation is greatest, and where large segments of the workforce are exposed to the harsh discipline of industrial labor for the first time. In the case of the semiconductor ATP plants, this was indeed the case for many locales. In East and Southeast Asia, companies like Signetics tended to recruit young women from agricultural regions unaccustomed to wage labor; even in industrialized Linlithgow, Scotland, as we will see, executives saw the greatest source of potential workers in the area’s “large reservoir of housewives.”Footnote 4

Scholars such as Beverly Silver, meanwhile, have shown not only that capital flight has been a basic feature of capitalism—what David Harvey called a “spatial fix” to labor unrest and periodic crises of profitability—but that capitalists acknowledge that the fix is only ever temporary; that it must be combined with significant process changes to forestall new crises (what Silver calls the “technological or organizational fix”); and that a reliable return ultimately depends on shifting investment into new products (the “product fix”).Footnote 5 More recent variations on this theme include Matthew Myers’ examination of a 1974 strike at the East Midlands plant of a typewriter manufacturer that had recently come under U.S. ownership. There, he finds that the factory’s closure months after the strike was a direct result of the unrest there, rather than the obsolescence of manual typewriters.Footnote 6 In at least one case examined in this article, too—that of Scotland—Signetics justified the plant’s closure by pointing to its inability to produce the latest ICs, even though a recent organizing campaign by a relatively militant union is the much more plausible explanation. And the imminent threat of automated assembly processes was wielded consistently as a tool to discipline its worldwide labor force. Part of what distinguishes semiconductors from the nineteenth and twentieth century industries explored by these and other scholars is the somewhat compressed timeline in which these events took place. Unlike the rusting ruins abandoned by capital in the late-twentieth-century Global North, whose post-industrial afterlives have become a mainstay of the literature on deindustrialization, the industrial communities at issue here scarcely had any time at all to develop before capital pulled up stakes.

There is a robust historiography around labor and U.S. empire, including a pool of scholarship that took off with the new millennium dealing with the Cold War period. This scholarship has covered wide ground, including the ways in which workers’ movements formed a core constituency in national liberation movements, the role of the American Federation of Labor and Congress of Industrial Organizations (AFL-CIO) in subverting genuine, independent unions in other countries, and the implications for workers laboring in liminal spaces beyond the legal reach of either their home governments or the United States.Footnote 7 I hope to add to this scholarship, in however modest a way, by illustrating how the ability of states to police independent workers’ movements was among the key considerations for the semiconductor industry in determining where to invest capital. The case study dealing with the Signetics plant in Thailand, in particular—where Signetics management appears to have taken earnest steps toward closing up shop merely because workers were asking for a representation election in the context of moderate labor reforms—shows just how fickle these companies could be if they sensed that political changes could mean returning a measure of control to workers.

Relatedly, historians of technology have highlighted the role of the Cold War state in shaping Silicon Valley and high-tech industries such as computing more generally, from state-funded research and development to the militarized and hierarchical discourse centered around such state priorities as command and control.Footnote 8 However, they have mostly neglected a subtler Cold War connection: the dependence of the semiconductor industry—so central to the region’s development that Silicon Valley is named after the chief substrate on which ICs (as well as discrete diodes and transistors) are built—on the United States’ partnership with rightist dictatorships that could keep unions at bay in the name of fighting Communism.

I begin the article by offering an explanation of how manufacturers of semiconductor devices came to rely so heavily on offshore ATP processes. Manufacturers have insisted on tight control over the labor process from the early days of the industry—a problem that was even more acute in IC production (which Signetics was established to focus on) than in discrete transistor manufacture. A market logic that demanded packing ever more components onto an IC without a proportional increase in prices also required firms to engage in constant process changes and invest increasingly large sums of capital in their wafer fabrication facilities, such that automation in chip fabrication became, more or less, a technological necessity. However, the same logic also implied exceptionally short product lifecycles, which made automation of the most labor-intensive processes (ATP) impractical. Therefore, the semiconductor industry relied for a long time on what one executive called “jet-plane automation,” i.e., leveraging favorable tariff laws and the light weight of semiconductor devices to shift these processes to areas where chipmakers might find large reserve armies of labor that they could pay significantly less and over which they could exert significantly more control. In the following section, I move into the case studies, examining plants in Scotland, Thailand, and Portugal. In these first studies, I inquire into what happened when workers managed to gain a measure of independent control amidst political changes of varying magnitudes. I find that management responded swiftly and drastically when it sensed its prerogatives confronted by even a relatively moderate challenge. The question of control here appears to have taken precedence over considerations of maintaining a cohesive supply chain, at least in the near-term. In a third section, I look at a fourth case study, that of the Signetics Korea plant, where workers were represented by a union whose national affiliate was so closely tied to the state that the security services were said to have played a role in selecting its leadership. Nevertheless, workers leveraged management’s dependence on Signetics Korea’s capacity to engage in two major job actions—one in 1968 and another in 1977. These actions raise the question of why Signetics did not ultimately pull out of Korea after weathering the storm. I argue that available evidence, though sparse, suggests that political conditions in Korea played an important role. Finally, I wrap up with some general impressions.

“Jet-plane automation” and the “ethos of control”

Signetics was established in 1961, one of many offshoots of Fairchild Semiconductor. Fairchild, itself only operating for less than four years, had been the site of several momentous process and product innovations, namely the first silicon-based commercial transistor (germanium had been the substrate material of choice up until that time); the so-called planar process—essentially, a manufacturing innovation that insulated the electrical junctions of transistors (and diodes) under a layer of silicon dioxide grown on top of the device, which was seen as vastly improving component reliability and eliminating the need to manually wire together components in integrated devices; and, as a direct result of the planar process, the first commercially viable IC, which drew together multiple components on a single piece of silicon.Footnote 9 The promise of these “microchips,” as they came to be known, was that if engineers could work out ways to pack into them greater numbers of transistors—components that replaced vacuum tubes as electronic switches and signal amplifiers, but in a smaller package and with lower power requirements—they would have created the basis for devices capable of carrying out increasingly sophisticated logical operations in contexts such as digital computing. (Analog ICs were also possible—indeed, produced—and had numerous important potential uses, but they posed special design problems that prevented the kind of exponential growth in complexity for which digital ICs became known.)

The founders of Signetics decided that they wanted their new company to exclusively produce ICs, becoming the first firm to do so. (As Christophe Lécuyer has documented, Signetics at first determined, even more boldly, to produce custom chips rather than standard, mass-marketed ICs, although before long it had to retreat from this posture—a strategic error that compelled the company’s leaders to eventually turn to Corning Glass Works in 1962 to rescue the company with an infusion of cash in exchange for a controlling interest.)Footnote 10 While the company’s concentration of resources on advanced chips during the 1960s and 1970s allowed Signetics to market several highly innovative products, such as its widely copied NE555 timer chip,Footnote 11 it also created challenges that ultimately hampered the competitive position of the firm for years to come. Signetics had focused so intently on providing the latest technology that when new commercial markets opened up for ICs and the military funding landscape began to change with Defense Secretary Robert McNamara’s moves to eliminate lucrative cost-plus contracts, the company was unprepared to compete on pricing.Footnote 12

In 1964, Fairchild attacked Signetics’ position with its introduction of a new family of diode-transistor logic chips, comparable in capability to Signetics’ most advanced microcircuits—but at less than half the price. This harsh early lesson that technological leadership in the field hardly ensured stable returns, combined with the dizzying pace at which engineers were able to design ever more complex chips, meant that constant process changes would be needed in order to remain cost competitive with other chipmakers. A company like Motorola did not feel quite the same pressure. The semiconductor arm of that company, then best known for its two-way radios, had set up shop in the Phoenix area in the 1950s. During the IC boom of the 1960s, Motorola showed little interest in producing the latest and greatest chips. Instead, it preferred to master ways to lower production costs of its discrete devices (especially transistors), offering products that had essentially been on the market for several years, but aimed for greater reliability and sold at prices considerably lower than its competitors. This production orientation, in contrast to that of the leading chipmakers, such as Signetics, allowed for much greater stability.Footnote 13 Though it did eventually start producing diverse lines of advanced chips, Motorola used this same basic formula to maintain a leading position in the industry for decades (often, second in market share among U.S. producers only to Dallas-based Texas Instruments).Footnote 14

Signetics may have borne severe financial consequences for a few years, but the competitive pressures faced by producers of advanced microchips were not unique to that company. For one thing, the pace of change was frenetic. This was best represented by Moore’s Law, named after the Fairchild and Intel co-founder, who correctly projected in 1965 that the number of transistors that can be cost-effectively crammed onto a chip will double roughly year (though the pace of growth has slowed somewhat, first to eighteen months, and then to two years), leading to exponential growth in processing speed and number of functions.Footnote 15 However, cost-effectively is the operative term here. Integration on an ever-larger scale has necessarily meant a roughly correlating growth in the expense of the front-end processes of IC manufacture (in contrast to “back-end,” labor-intensive ATP processes), which have been capital-intensive from the outset. An important corollary to Moore’s Law, sometimes called Moore’s Second Law, projected that the cost of a wafer fab facility doubles roughly every four years.Footnote 16 As long as more or less ideal market conditions prevailed and firms had access to enough capital to re-invest in advanced production equipment, theoretically the march of Moore’s Law could continue right up until chip designers ran out of physical space to reliably map transistors. The degree to which Moore’s initial projections (which only covered ten years) have largely borne out has led to a rich dialogue in the historical literature about exactly what drives the “law.”Footnote 17 Paul Ceruzzi has argued convincingly that, whatever the exact causes, there does seem to be an “internal logic” at work that scholars should ignore at their own risk (though he was careful to avoid recourse to simple technological determinism, the bête noire of historians of technology of virtually all stripes).Footnote 18 And yet, even Silicon Valley acknowledged that the path companies took to keep up with Moore’s Law was an extremely fraught one indeed. An environment where firms seemed to catch up with one another technologically, only to engage in destructive price wars was one factor among several contributing to an unstable market, and the semiconductor industry faced severe recessions roughly every five to seven years between the late 1960s and the mid-1980s, after which a major wave of mergers and acquisitions ensued. Robert Noyce, another co-founder of both Fairchild and Intel and a leading voice in the semiconductor industry, predicted in 1977 that “further miniaturization is less likely to be limited by the laws of physics than by the laws of economics.”Footnote 19

This exceptionally competitive environment called for tight discipline over production, or what one supervising engineer called “ethos of control,” that was present from the earliest days of IC manufacture. According to Eugene Kleiner, co-founder of Fairchild Semiconductor, in the early 1960s, “we almost had one or two PhDs watching a girl do something.” Managers would “prescribe the exact steps,” and had to warn people that they “should not use common sense.” “In the manufacture of semiconductors … the discipline to follow exact prescriptions is important,” he said, “… because you may start off making a thousand and wind up with ten good devices and you don’t really know sometimes why the others are not good.”Footnote 20 Companies became fixated on improving these “yields,” their term for the proportion of salable devices produced in a batch. During these early days, yields were generally so abysmal that some technologists questioned whether ICs were even financially viable.Footnote 21 As Kleiner’s comments suggested, in this environment, any variation by an individual operator who was not closely watched could prevent engineers from tracing the source of losses.

Before too long, chipmakers did introduce process changes that dramatically improved yields. (One major contributor was the decision to equip production environments with complex ventilation systems and outfit workers with head-to-toe gear meant to prevent particulate matter from spoiling devices, i.e., the construction of modern “clean rooms.”) Nevertheless, some yield loss was inevitable, and semiconductor companies implored workers to think of every bad device that passed through the line as a threat to the company’s competitive position. Jay Last, a Fairchild co-founder, and then-executive at Amelco Semiconductor, remarked in 1965 of IC manufacture that “the cost items are nearly all labor and yield loss.”Footnote 22

The ethos of control explains the somewhat misunderstood attitude of Silicon Valley generally, and the semiconductor industry in particular, toward unions. While it is true that in the overwhelming majority of cases, management viewed the specter of union representation by even relatively conservative workers’ organizations as a threat, executives resembled counterparts in other industries in viewing unions opportunistically. If unions could be tamed to their liking, while stabilizing turnover, helping to rationalize the wage structure, and channeling worker discontent into a restrictive grievance procedure, management could see their way to tolerating the union’s presence. On the other hand, organizing efforts by unions who sought to attain any measure of shop-floor control required vanquishing at any cost. Charles E. Sporck, the longtime CEO of National Semiconductor, had begun his career as a manufacturing engineer at General Electric (GE) steam turbine and capacitor plants. Both plants were represented by the United Electrical, Radio, and Machine Workers of America (UE), a left-wing dissident union with a reputation for shop-floor militance, and Sporck on multiple occasions cited the union’s control over production as prompting his own decision to quit GE and take a job with Fairchild in California. Sporck complained that at the steam turbine plant, the

shop steward could walk in and turn the main power off and just shut the whole place down and go into discussions. I mean, it was an entirely different environment [compared to the semiconductor industry of the late twentieth century]. Now, you can imagine somebody turning the power off in a semiconductor wafer fab facility.

And Sporck clearly was not alone in viewing control over production as the key motivating factor for keeping unions out. A Federal Trade Commission report published in 1977 to the industry’s success concluded that the “ease of starting, ending, or changing new projects may … be explained in part by the fact that many of the semiconductor companies have not been unionized so that they lack the rigidities built into the work rules of many union contracts,” a view endorsed by “a number of semiconductor executives.”Footnote 23 The need to impose discipline only became more acute with the intensification of Japanese competition starting in the 1970s, and Christophe Lécuyer has argued that Intel’s success in the second half of the 1980s in improving quality control to outcompete Japanese companies was due in part to management’s ability to impose strict discipline on the line without objections from unions.Footnote 24

Still, if control over labor in front-end production remained a major priority of semiconductor companies through the 1980s and beyond, it was at least the case that the automated equipment necessary for building more complex chips would lessen companies’ dependence on front-end operatives to keep the line running. In the most labor-intensive parts of the production process, i.e., ATP, companies like Signetics could rely on no such “technological fix” to improve productivity and remain competitive.

It was theoretically possible to automate labor-intensive processes such as securing the chip to a substrate and then threading through the electrical connections (“die attach” and “wire bonding,” respectively) as early as the 1950s with the emergence of computer numerical control machining, and yet the first automatic wire bonders did not really enter widespread use until well into the 1980s. Management saw as cautionary tales such companies as Philco, which was forced out of the semiconductor business in the 1950s after investing in expensive equipment to automate production of its patented surface-barrier transistor, only to have that equipment made hopelessly obsolete after only a few short years, and several Japanese competitors automated back-end production earlier in the 1970s and as a result could not keep its products cost-competitive for years.Footnote 25 While some companies pursued automation of certain discrete steps with potentially universal (or, at least flexible) applicability, in the early 1960s, the chipmakers did not consider large-scale ATP automation as a serious possibility. David James, a Signetics executive and co-founder, laid out the rationale in a 1962 article. Assuming as the ideal goal of all manufacturing engineering, regardless of industry, “tighter controls over each step,” such that “machines can make the decisions and manipulations without human aid where possible,” James, in the pages of a trade journal, cautioned fellow manufacturing engineers with an important proviso. “Regardless of the efficiency of automation,” he wrote, “it is only feasible if the process can be fixed long enough not to obsolete expensive equipment before it has paid for itself.” In the case of semiconductor assembly, James continued, “full mechanization would be costly and of doubtful efficiency.” Nevertheless, he and his peers certainly saw the need to continue to reduce costs and keep tight control over individual steps, pointing out that “it is the only true line operation in the manufacture of integrated circuits.”Footnote 26 Compounding the problem, as we have seen, was the fact that ever larger sums of capital would need to be poured into the front end. If, in accordance with Moore’s Law, chipmakers needed to offer improved processing power without a corresponding rise in prices (or the same processing power at lower prices) at short, regular intervals, this clearly required holding down costs elsewhere in the operation.

Overwhelmingly, semiconductor manufacturers’ solution to this problem was to shift its labor-intensive processes to low-wage areas. Sporck, who was manufacturing director at Fairchild until the late 1960s, got the ball rolling by setting up a plant in Hong Kong in 1963, supplemented, around the same time by plants in South Portland, Maine, and on the Navajo Nation reservation in Shiprock, New Mexico.Footnote 27 Within a few years, it had built additional facilities in South Korea and Mexico. Fairchild’s competitors followed suit in short order.Footnote 28 It was a phenomenon that Fairchild’s Wilf Corrigan aptly called “jet-plane automation”:

The manufacturing strategy initiated by Charlie Spork [sic] at Fairchild in the early sixties could be described as jet plane automation–you flew chips from San Francisco to the Far East where, for eight cents an hour, they assembled the product and sent it back within the week. We were doing a form of arbitrage. Since we were accessing a twenty-to-one savings in labor costs, we didn’t have to bother with automation. Certainly Silicon Valley companies were not leaders in automation. Using Far East labor meant that as products evolved, we didn’t have to worry about retooling costs, a major factor when products are changing rapidly.Footnote 29

In other words, keeping the international production complex economical meant keeping sweated international labor more productive than automation.

By the mid-1970s, U.S. manufacturers of semiconductor devices were using a complex international division of labor that was pretty exceptional for its time. Numerous other industries had extensive foreign plant, but the foreign operations were typically designed to tunnel under tariff walls or facilitate distribution to export markets, or they were otherwise sufficiently close to U.S. borders to allow overland transport. While some elements of the chipmakers’ foreign operations were designed for access to foreign markets, particularly their European wafer fabrication plants, semiconductor firms also sent huge numbers of U.S.-made components abroad (typically, though not exclusively, to East and Southeast Asia), only to absorb them back into the United States for domestic sale. Even firms with a quite modest sales volume (as little as $3 million) offshored their most labor-intensive processes. Part of what made this scheme feasible was item 807.00 of the tariff schedule, which allowed firms to re-import their own goods and pay duties only on the portion of the value added abroad. According to calculations by the U.S. Department of Commerce, the semiconductor industry accounted for the highest value of duty-free imports falling under item 807.00 (see Table 1) in 1970 and 1972, and the third-highest in total value. The auto industry was first in total value by a very comfortable margin, but in addition to the industry’s much larger aggregate size, most of its re-imports came from within North America (particularly Mexico)—it would hardly have been economical to haul car parts halfway across the world, employ Asian workers to assemble them, and then put them back on shipping containers for a transoceanic return voyage. By contrast, electronic components were light enough so that chipmakers didn’t even need to worry about production lags because of long shipping times; most chose to carry them back and forth by air.Footnote 30

Table 1.

Imports of Major Commodities into the U.S. Under Tariff Item 807.00

Table 1 Long description

Commodity groups are listed with five columns: 1970 Total, 1970 Duty-free, 1970 Dutiable, 1972 Duty-free, and 1972 Total. Semiconductors and parts show 126,758 total in 1970, with 74,792 duty free and 51,966 dutiable, and 74,976 duty free in 1972 with a 1972 total of 161,560. Office machines and parts list 137,890 total in 1970, with 61,612 duty free and 76,278 dutiable, and 124,347 duty free in 1972 with a 1972 total of 251,034. Electronic memories list 42,429 total in 1970, with 28,596 duty free and 13,833 dutiable, and 41,759 duty free in 1972 with a 1972 total of 61,231. Television apparatus and parts list 57,180 total in 1970, with 28,106 duty free and 29,074 dutiable, and 56,841 duty free in 1972 with a 1972 total of 128,068. Textile products, total list 43,039 total in 1970, with 27,820 duty free and 15,219 dutiable, and 55,077 duty free in 1972 with a 1972 total of 86,774; wearing apparel and accessories list 42,266 total in 1970, with 27,512 duty free and 14,754 dutiable, and 53,658 duty free in 1972 with a 1972 total of 84,780; other textile products list 773 total in 1970, with 308 duty free and 465 dutiable, and 1,419 duty free in 1972 with a 1972 total of 1,994. Television receivers list 56,069 total in 1970, with 19,677 duty free and 36,392 dutiable, and 30,350 duty free in 1972 with a 1972 total of 144,234; radio apparatus and parts list 58,141 total in 1970, with 16,478 duty free and 41,663 dutiable, and 16,649 duty free in 1972 with a 1972 total of 80,349. Internal comb. engines, non-piston list 34,927 total in 1970, with 15,561 duty free and 19,366 dutiable, and 16,441 duty free in 1972 with a 1972 total of 152,669. Toys, dolls, and models list 38,521 total in 1970, with 15,252 duty free and 23,269 dutiable, and 11,622 duty free in 1972 with a 1972 total of 35,807; scientific instruments list 21,334 total in 1970, with 11,597 duty free and 9,737 dutiable, and 7,120 duty free in 1972 with a 1972 total of 16,731; aircraft, nonmilitary list 33,725 total in 1970, with 5,107 duty free and 28,618 dutiable, and 7,212 duty free in 1972 with a 1972 total of 43,858; sewing machines, household types list 22,180 total in 1970, with 1,663 duty free and 20,517 dutiable, and 251 duty free in 1972 with a 1972 total of 30,603.

Note: By the early 1970s, the semiconductor industry was a leader in offshoring production, particularly to distant locales.

Source: Background Materials on Articles Assembled and Fabricated Abroad, Items 806.30 and 807.00 of the Tariff Schedules of the United States: Hearing before the Subcommittee on Trade of the Committee on Ways and Means, United States House of Representatives, Ninety-fourth Congress, second session (1976), 86.

Why did chipmakers opt to spread production across multiple continents? Observers, then and now, have always emphasized low-wage labor, which was typically the reason that semiconductor capitalists themselves offered up.Footnote 31 This was clearly a key element. In the late 1960s, as the internationalization of semiconductor production was taking off beyond a few plants, wages for production workers were about ten cents on the dollar in Hong Kong and Korea compared to their U.S. counterparts, nine cents on the dollar in Singapore, and eight cents on the dollar in Taiwan. As the 1970s wore on and wages rose in high-growth Singapore and Hong Kong, semiconductor capitalists spread their bets among the even lower-wage areas of Southeast Asia, such as Malaysia (ten cents on the dollar in 1975), the Philippines (six cents), Thailand (five cents), and Indonesia (five cents) (see Table 2). Another major consideration when deciding where to invest in labor-intensive plants was infrastructure. What would be the time and expense required to link up to, say, the electrical grid and the water supply? Given that workers would often be drawn in from the rural periphery, was there sufficient housing in the targeted city or town for hundreds of workers to move there? How close could they be placed to the proposed plant, and what transportation systems were in place that they could use to travel to and from work? How easily could foreign plant managers communicate with corporate headquarters in the United States?Footnote 32

Table 2.

Relative Hourly Compensation of Electronics Production Workers, Selected Southeast Asian Countries (U.S. = 100)

Table 2 Long description

Relative hourly compensation is listed by country for electronics production workers in 1969, 1975, and 1985, with values shown as an index where the United States is 100. Hong Kong is 10 in 1969, 12 in 1975, and 16 in 1985. Indonesia has no value in 1969, then 5 in 1975 and 4 in 1985. Korea is 10 in 1969, 7 in 1975, and 14 in 1985. Malaysia has no value in 1969, then 9 in 1975 and 10 in 1985. Philippines has no value in 1969, then 6 in 1975 and 8 in 1985. Singapore is 9 in 1969, 12 in 1975, and 19 in 1985. Taiwan is 8 in 1969, 7 in 1975, and 16 in 1985. Thailand has no value in 1969, then 5 in 1975 and 5 in 1985.

Note: By the early 1970s, the semiconductor industry was a leader in offshoring production, particularly to distant locales.

Source: Allen J. Scott, The Semiconductor Industry in South-East Asia: Organization, Location and the International Division of Labor, Working Paper Series – 101 (Los Angeles: Institute of Industrial Relations, 1986), 50.

Important as wages and infrastructure have been (about which, see more below), however, semiconductor executives have understood that they cannot be examined in isolation from productivity. As the case studies below illustrate, wage demands alone did not provoke panic among Signetics executives. What did—indeed, what was generally enough to cause them to seriously explore an exit strategy—were genuine threats to its control over the shop floor. It is worth keeping in mind, too, that the instances of labor unrest that we will examine offer an impression of management extinguishing local fires, but to think only about the workers directly affected would be a mistake. The chipmakers’ quest for control was total, extending to the workers at their large wafer fabs in Northern California, and the ability to discipline labor in one place held lessons for the entire global workforce.

Threats to management control: Scotland, Thailand, and Portugal

Semiconductor management examined both physical and social infrastructures as a way of determining how they could get the most from workers at the lowest possible cost, as well as whether there was a sufficient supply of technically educated people who could be recruited to manage that mass of production workers. In this respect, managers inquired about the feasibility of drawing from a reserve of low-wage labor and the character of that labor. When Don Liddie, the longtime Signetics executive then serving as chief of international operations, was looking to place a new plant to serve the European market (initially to perform back-end functions and then expand into wafer fabrication for some of the company’s simpler chips), he rejected Northern Ireland because of its “low female unemployment rate” and “lack of a demonstrated supply and ability to attract electronic, technical and professional personnel.”Footnote 33

By contrast, Liddie ultimately settled on Linlithgow, Scotland, in part because of what he perceived to be its “good labor climate.” “Although the female unemployment level is low,” he reasoned, “so is the employment level, indicating a reservoir of housewives.” This is as clear an indication as one is likely to find that, even in this overwhelmingly white region, Signetics chose its location based on a very specific idea of the kind of operative it wanted to hire. Management’s conception of a “housewife,” with little long-term attachment to her work and not accustomed to extensive industrial employment, perhaps working to earn some extra spending money, and whom the company could justify paying lower wages by presuming the workers were married to a breadwinner husband, of course, has a very long history in industrial capitalism.Footnote 34 And as David James told Christophe Lécuyer many years later, going back to the early days of Fairchild, when all production was integrated into one Silicon Valley plant, to fill jobs for assemblers, management recruited “largely housewives.”Footnote 35 Liddie hastened to add that the “general philosophy in Eastern Central Scotland leads us to believe that we will have the opportunity to operate without a union.”Footnote 36 This last consideration turned out to be a bad bet. Within two years, Liddie’s deputy was decrying the fact that “the British people basically cannot conceive of any other way of communicating with ‘the company’ other than through a formal labor union,” and recommending that a conservative union be invited in to forestall an organizing attempt by a more militant one.Footnote 37

Liddie did not mention it at the time, but the political backdrop loomed large here, for at just the time he sent off his December 1970 memo, a widely publicized labor relations bill, backed by the new Conservative government of Edward Heath, was being introduced into Parliament. The Tories had campaigned on reining in unions and reducing industrial disputes by mandating greater state oversight of collective bargaining, requiring balloting before strikes, imposing “cooling-off periods,” and weakening union security agreements. Most controversial of all was a provision in the bill establishing a process whereby unions were required to register with the government or face sequestration of their funds if sued for “unfair industrial practices,” such as wildcat strikes. In short, the bill, which passed the following summer as the Industrial Relations Act 1971, sought to undermine the syndicalist leanings of the left wing of the British labor movement by making collective bargaining agreements strongly binding under the law and placing pressure on unions to police militance. Signetics strongly preferred no union at all, but if it was true that the mentality of the British worker precluded that option, as the company claimed, the next best thing was to have a union in place that it could reasonably control. The union that Liddie suggested inviting in, the General and Municipal Workers, was in the minority of unions who initially signaled their willingness to accept the prospective new conservative order.Footnote 38

Linlithgow combined front-end and back-end functions, and the environment there was more predictable than many of the assembly plants sited further afield, but events there nicely demonstrate a core contradiction of the semiconductor companies’ strategy of stratifying their division of labor by nationality. On the one hand, compared to the high-tech fiefdoms in Silicon Valley, Signetics was still operating in a foreign environment; it had limited control over the labor market and even less control over political developments. The company learned that the hard way when, rather than signing a sweetheart deal with labor conservatives, it was organized by the skilled worker militants of the Amalgamated Union of Engineering Workers (AUEW)—the very union that successfully led resistance to the new law by urging unions not to register with the government. If the purpose of the new law was to quiet industrial strife, it had failed miserably. Fewer than five percent of union members belonged to the minority of unions that chose to register, and unions like the AUEW were now calling strikes that targeted the state directly. To make matters worse, the Organization of Arab Petroleum Exporting Countries imposed an oil embargo on the United States and its allies, creating an inflationary spiral and forcing the government to turn to the nation’s coal miners—i.e., some of its most militant unionists—to make up the energy shortfall.Footnote 39 Heath called a snap election, which Labor won narrowly, immediately imperiling the law, which was replaced within months with a law that offered some key concessions to unions. By the spring of 1974, the AUEW had signed up the majority of the shop at Signetics, and though Liddie trotted out the standard employer playbook of questioning the integrity of that majority in its discussions with the union, Liddie was forced to admit privately to top management that “we do believe that a majority of our people want a union.” And despite the typical public rhetoric depicting the union as a meddling third party, he acknowledged that the organizing drive resulted from an “unsolicited request from some of our employees,” rather than AUEW’s courting of Signetics workers.Footnote 40

On the other hand, what left the Linlithgow workers vulnerable were the vagaries of foreign investment. The company was in the process of switching from a process of fabricating circuits on 2-inch silicon wafers to a 3-inch process. Linlithgow was equipped to fabricate 2-inch wafers, and the company justified the closure on the grounds that “adequate capacity of 3[-inch] wafer production in the corporation eliminates the need for the Linlithgow fabrication.” Was it just a coincidence, then, that the company was on the cusp of signing an agreement with the AUEW? Considering that the 3-inch process was several years in the making, that the Linlithgow operation—by the company’s own admission—had played a key role in servicing the European market, and that the facility’s labor struggles stretched back to 1970 at the latest, it makes much more sense to conclude that the presence of a union that had a reputation for relatively militant steward agitation played a decisive role.Footnote 41

Thailand was another place where workers learned how capricious management could be in the face of worker unrest, and the connection to larger political developments was even more apparent than in Scotland. As a way to show that management’s central concerns were not unique to Signetics, I will make reference to campaigns both at Signetics’ Thailand plant and that of National Semiconductor. Both companies faced organizing campaigns at around the same time, but there is no evidence to suggest that they were connected in any way, or even that they knew of each other’s efforts. Rather, both appear to have been spontaneous expressions of worker discontent in the context of a fragile and short-lived democratic opening in the aftermath of the October 1973 popular uprising that toppled the military government of Thanom Kittikachorn.

When Signetics, the better documented case, established assembly and test operations in Thailand February 1974, there were no reasons to expect trouble—at least on the surface. The popular uprising itself, led by moderately reformist student activists, was a thoroughly respectable affair that consciously excluded the radical left. King Bhumibol offered his “tacit support” to the students as way of heading off leftist agitation and attempting to manage support for political liberalization. That he “hand-picked” a royalist bureaucrat and jurist as prime minister to steward the process of constitutional reform signals elite expectations in a moment where repression seemed to be more politically costly than moderate concessions to democratic impulses that were becoming increasingly difficult to ignore.Footnote 42

But beneath the surface, there were signs of widespread radicalization and consciousness raising led, at least in part, by the burgeoning industrial working class. Workers, particularly those in strategically important industries such as textile, had grown increasingly restless over unsafe conditions, low wages, and capricious and authoritarian workplace control by management. According to industrial relations scholar Andrew Brown, between 1972 and 1975, workers engaged in “1,000 strikes, numerous rallies and demonstrations, street marches, sit-ins, factory blockades and, in one famous incident involving women workers employed in the Hara Jeans factory, the seizure of the means of production.”Footnote 43 Beginning in 1972, the government made concessions on workers’ heretofore strictly circumscribed freedom of association, culminating in a 1975 law that codified, for the first time, workers’ ability to form unions, bargain collectively, and engage in strikes. One distinct feature was that it allowed relatively small groups of workers (as few as fifteen percent of the total workforce) to represent them on some shared basis. While this provision did not have the radical, minoritarian implications that employers (notably, Signetics) claimed it did—after all, such a feature hardly seems exotic to Western liberal democracies, including the United States, many of which grant bargaining rights to subsets of workers in a plant based on craft—it allowed workers to try and gain a foothold before establishing a majority of non-exempt employees. Though the law aimed to embed unions in a corporatist system that shepherded struggle into economistic and de-politicized channels, it opened enough political space to force workplace conflict into the open in a way that it could not have been under the repressive periods of military rule. At the same time, however, a countervailing rightist movement had begun to mobilize and strengthen over the course of 1975. This essentially reactionary movement opposed liberalization as a destabilizing force that would inevitably be seized by radicals, and on that pretext sought to re-impose the military order.Footnote 44

It was in this context that, in September 1975, a veritable campaign to organize the Signetics plant in Bangkok began in earnest. An extended memo from Don Liddie, then president of Signetics International, to his at-large manager of the overseas test and assembly plants, lays out in remarkable detail management’s severe measures to undermine any effort to confer upon workers even a modest measure of plant governance, including a swift readiness to invoke the proverbial nuclear option of shutting down the plant. According to Liddie, the trouble began when several of the plant’s employees attended classes offered by the Labor Department to inform workers of their rights under the new law. Correctly observing that plant productivity was healthy and steadily improving, employees apparently began to question why they had scarcely seen the fruits of such gains. The executive openly wondered why the government had not properly surveilled these workers, and blamed government bureaucrats for putting ideas in the workers’ heads: “The problem was that our employees participating in the Labor Dept. classes was not brought to our attention since they were ‘students’ and not representatives of Signetics. Unfortunately, part of the classroom lecture centered around the labor law covering the formation of various employee committees, including unions and their certification.” Seven activists easily picked up the requisite fifteen percent to form a recognized “committee” to file grievances and represent workers in collective bargaining negotiations. Signetics probably found especially threatening the fact that the leaders were not low-grade operatives but skilled and semi-skilled technicians who were credibly threatening to expand their base. On March 24, the committee began circulating plant-wide petitions, sending management into a frenzy. Liddie soon boarded a plane to Bangkok so that he could implement a firefighting strategy in real time.Footnote 45

Management’s first impulse was to delay the establishment of either minority or plant-wide representation until they could browbeat people out of supporting the committee. Liddie and plant manager Tom Frampton insisted on overseeing an election where employees could indicate their preference of committee representation and their choice of leadership. The activists objected, pointing out that such an election contravened the new Labor Relations Act—representation had already been established—but management threatened to immediately shut down the plant if they did not get their way. “‘Stopping final test, no build up, resulting lay off’ had its effect,” Liddie reported. “The pressure on the committee was heavy and they finally told [management] they would drop their demands and resign from the committee if we would not take action against them.” However, when the election results came in on the morning of Friday, March 26, it had become abundantly clear that employees quickly rebuked management by voting for the committee by a “large majority,” electing a group of ten representatives—among them the seven original activists, and “none pro-management.” The committee eventually articulated a fairly broad range of demands. These included a structured grievance procedure, wage gains to match increases in productivity, job security, and safety features, including a fire evacuation plan. Another key demand was to ease the relentless pressure on the line that workers sense was leading to severe eyestrain, which resulted from workers peering into microscopes for hours on end to bond wires and inspect for defects. None of these demands was seriously considered. As Liddie openly admitted, “all of this activity was buying time. [Frampton] wanted to keep the committee from contacting the Labor Dept, until he had time to develop a plan.”Footnote 46

The plan was to take swift, merciless action. Management became “convinced that if the committee continued in the direction of negotiating demands, the first action Monday 3/29/76 would be to lay off 120 trainees. This would be followed by the cancellation of final test and Bangkok would not have a production buildup.” As a gunpoint “compromise,” committee representatives proposed allowing Frampton to choose five of the ten representatives and immediately agreeing to some of the committee’s most modest demands, including hiring a staff nurse to tend to workplace injuries and illnesses and allowing workers to listen to music on the line. Frampton agreed, much to the chagrin of Liddie. Management expressed an openness to listen to grievances and demands, but on the critical condition that no committee could “mediate” between management and rank-and-file workers, none of their demands be accorded any legal legitimacy, and any concessions be granted solely based on the goodwill of management. In other words, management offered the status quo ante in exchange for not firing everyone and leaving town, i.e., absolute, unquestioned company control over all workplace affairs. Liddie explained “Sunnyvale [Silicon Valley headquarters] feelings regarding unions and labor problems was [sic] put in some pretty strong words and most of the Thais understood the meaning without translation. Sunnyvale does not want any god damn union in the Bangkok plant!” However, as a one-time, good-faith measure, all exempt employees were asked to donate fifty baht (about $2.50 in 1976 dollars) to a fund for redistribution to the hourly employees. For context, there were forty-two exempt employees at Signetics Thailand at this time.Footnote 47 If everyone donated and the fund was paid out to all 660 hourlies, this one-time donation would amount to just over sixteen cents per worker—and even this sum would be borne by individual managers rather than the company. At the risk of extreme understatement, this was a small price to pay to maintain control.Footnote 48

Much less well documented, but comparably revealing nonetheless, was an organizing effort at National’s Thailand assembly plant around the same time. While it is unclear whether workers at National ever registered a committee/union, the company had actively negotiated with workers for at least several weeks. Given management’s general unwillingness to recognize independent worker representatives unless absolutely essential, it seems likely that a functional (if not necessarily stable) union existed at this time that credibly claimed to represent most non-exempt workers. The workers were demanding a “500-baht salary [monthly] increase [$25 in 1976 dollars], five percent company contribution to the retirement fund and two months’ salary as an annual bonus.” Management countered with a ninety-baht increase and a rejection of all other demands, leading to an impasse. On May 21, all 3,700 hourly workers downed tools and engaged in a strike that lasted at least four months.Footnote 49 In late September, Sporck informed shareholders that the Bangkok facility would completely close and operations would be transferred to two other East Asian plants.Footnote 50 However, less than two weeks later, a bloody coup returned to power the ideological heirs of the royalist military men who had dominated Thai politics since the late 1940s. The new government of Thanin Kraiwichian “outlawed strikes, banned union meetings, deregistered some unions and targeted those activists deemed ‘social dangers.’”Footnote 51 Documentation of National’s decision to reverse course and stick around remains elusive. What is clear is that the company remained in Thailand for at least sixteen more years.Footnote 52

One final illustration of semiconductor management’s willingness to abandon otherwise viable plants when their authority was challenged is the case of the Signetics assembly and test facility in the southern Portuguese port city of Setúbal in the waning days of the corporatist dictatorship and the heady opening months of the Carnation Revolution. Activism in the plant there—which started production in 1970Footnote 53—dated to June 1973 at the latest. Clearly, the political context is once again essential for understanding the struggle. Discontent in the country at large had been simmering for many years, as Portugal—the least developed state in Western Europe—fought desperately to hang onto its colonial possessions amid indigenous insurgencies in Angola, Guinea-Bissau, and Mozambique. The state had been pouring tremendous resources into fighting in countries where the will to resist seemed boundless and, given the pace of African decolonization since the late 1950s, a rebel victory seemed inevitable. Many military-aged men deserted or failed to register for the draft, culminating in the dramatic desertion of seven officers in 1970, who fled to Sweden, pleaded solidarity with the rebels, and urged Portuguese troops to “sabotage” and “defect” from “this … criminal war.”Footnote 54 Meanwhile, the fascistic Estado Novo regime, which had been in power since the 1930s, cracked down on dissent using the secret police—prompting further resentment. After the longtime dictator António de Oliveira Salazar suffered a cerebral hemorrhage in 1968, his successor Marcelo Caetano made moves to soften some of the regime’s worst abuses, but when that resulted in increased opposition pressure, Caetano attempted to tighten the screws once more, and that reversal succeeded above all in supplying fodder for regime critics while dividing allies.Footnote 55

Labor relations at the Portugal Signetics plant were governed by an agreement with a state-dominated union, but by the early 1970s, wildcat strike activity had been on the rise across the country; as during the revolution, feminized industries, such as textile/garments and electronics, played a particularly prominent role. For instance, strikes broke out every year in various textile factories between 1970 and 1973.Footnote 56 In June, the women on the overnight shift at Signetics engaged in a brief walkout over pay. They returned to their positions before the shift was up, but then resorted to slowing down the line. Management reportedly responded by promoting the handful of women who dissented from the job actions and promising modest raises to everyone else—conditioned on productivity increases.Footnote 57

Then, in November, workers again took action, and were in this instance, better coordinated and more consciously political. The primary issue again was pay—electronics assembly was at the bottom of industrial wage scale—but it was timed to express solidarity and take concerted action with strikers at the large Lisbon works of Southern California-based tape head manufacturer Applied Magnetics, joined eventually by local workers at the French manufacturer Standard Electric, the British electronics giant Plessey, and a plant of the Minneapolis-area computer maker Control Data Corporation located in the town of Palmela, just north of Setúbal.Footnote 58 The Signetics workers’ other demands included Saturdays off (i.e., a reduction in the work week from forty-eight hours to forty hours), four weeks’ vacation, a one-time bonus, and crucially, an end to speed-ups. Some workers circulated leaflets reading, in part:

We have been wanting increases in our salaries for a long time. Time passes and we continue receiving the misery of 50 [escudos daily, about $2 in 1973 dollars] and waiting for them to give us Saturdays. American bosses only know how to make promises and say that they can’t do anything …. Meanwhile, they force us to work harder and harder to increase production and earn more money at our expense …. We live a life of slaves in this capitalist society. As our men’s wages are not enough to support the family, we are forced to work in the factory where they pay us poverty wages much lower than those of men …. [W]e work as much as men in factories and we still have to do the work at home that the bosses don’t care about making easier by refusing us a Saturday rest …Footnote 59

These workers beseeched their comrades not to fold until there was a firm commitment from management, noting that the promised raises from June had never been delivered. Management initially refused to meet with workers, instead calling out the National Republican Guard to allow selected strikebreakers into the plant to operate at limited capacity. The women maintained pickets for five days and nights before yielding to various threats and promises, saving face yet again with a verbal commitment from management to raise pay slightly, reduce the workweek to forty-five hours, and not to retaliate against participants in the strike.Footnote 60

The final collapse of the teetering regime came on April 25, 1974, when a group of junior officers—many with links to the parties of the left—mobilized the troops under their command to challenge loyal generals’ control of the country’s key ports and lay siege to the civilian government. Most of the generals’ soldiers joined the coup plotters or melted away, refusing orders to fire on the insurgents. In a matter of hours, with scarcely a shot fired, the hated Estado Novo dictatorship was forced to hand over power. Despite repeated pleas from the insurgent officers to remain at home, a jubilant population soon crowded the streets of Portugal’s towns and cities, determined to claim a stake in the consolidation of power. Raquel Varela has argued that the level of popular mobilization was so profound that historians are still reckoning with the scale of struggles in the nation’s factories, urban neighborhoods, and countryside.Footnote 61 What followed was a tumultuous period of intrigue and agitation from both the right and the left and a series of abortive coups that need not detain us here. The important point is that, over the next twenty months, mobilizations from below—neighborhood- and city-level popular assemblies, strikes and factory occupations, land seizures, etc.—framed the struggle at Signetics, which was facing a revolt in its own plant. For, these mobilizations simultaneously forced the government to take progressive social and economic measures and created the conditions for massive capital flight that forced the government to make concessions to private investors, prompting even the Portuguese Communist Party—unlike the many sectarian ultra-left groups that flourished in this atmosphere—to try (often without success) to mobilize workers in a “battle of production” to promote development.Footnote 62

Though the documentation around what occurred following the revolution at the Signetics plant is sparse—limited to a few scattered local newspaper articles and passing mentions in the secondary literature, with few management-side records addressing the issue directly—there is enough evidence to establish that management was willing to eat the costs of closing a plant and transferring capacity once its prerogatives were challenged. At some point, the workers established a plant committee that they viewed as a collective representative along the lines of a union, and that committee appears to have been linked to one of the small parties of the revolutionary left.Footnote 63 Management, in light of the revolutionary atmosphere and the sporadic job actions of 1973, and apparently hoping to stabilize production by giving the company representatives with whom to negotiate to head off disruptive actions, reluctantly recognized the committee, though the two sides seem to have had vastly different understandings of the authority of this body. It was a move management later came to regret, as the committee, according to Liddie, “escalated from problems with the canteen and company picnic to those dealing with such things as wages and reinstatement of terminated employees.” Liddie claimed that the “trouble started when management could not meet the expectations of the employees through their committee.”Footnote 64 In other words, management was willing to recognize this committee as long as it retained no power to enforce its demands at the point of production and kept to polite requests over marginal issues. As soon as workers expressed a willingness to use the main leverage that they had—i.e., the power to withhold their labor—the committee became a mortal threat. Already by May 1—probably not incidentally, a day of nationwide radical action full of strikes and street protests, as the first free May Day since the 1920s—company executives were signaling trouble and exploring contingencies.Footnote 65

It is also important to note, in light of claims that Signetics executives made to justify the closure of the Portugal plant, that prior to the revolution, management never appeared to question the viability of the plant. As late as October 1973, after the unrest began, Liddie was reporting to company President Charles C. Harwood and other top managers that the plant was contributing positively to the strong net operating margin of the international holding company subsidiary, and management’s decision to wind down capacity in the summer of 1974 without the immediate ability to use other plants in the international pipeline to make up the shortfall caused a significant backlog and had significant customs implications.Footnote 66 Even if we take claims about the effects of the revolution on wages by Signetics management at face value—namely, that the industrial minimum wage instituted by the new government raised the wages of most operatives by a factor of two to three—the plant would have remained profitable, though substantially less so than under the old wage scale. On the other hand, a sales slump hit profitability across the company and the industry in 1974.Footnote 67

Already skittish before a popular revolution threatened its property, what seems to have pushed Signetics over the edge into determination to close the plant was not that the government was automatically and from the outset impossible to trust with its investment, but that the new political conditions were such that long-simmering grievances could no longer be repressed away. Already, in February 1974, two months before the coup, the Caetano government—clamoring for popular legitimacy—proposed a new wage settlement for the seventy-five-percent foreign-owned electronics industry that would have provided across-the-board raises “ranging from 150 to 300 percent depending upon the particular employer’s scale.” Signetics was among a group of U.S. electronics firms that visited the ambassador to demand delaying the settlement and calling for direct negotiations between the state and the companies.Footnote 68 Then, in the immediate aftermath of the coup came a wave of strikes in electronics, including one at Applied Magnetics in Lisbon that developed later in the summer into a full-fledged factory occupation.Footnote 69 This was accompanied by government promulgation of a general industrial minimum wage of 3,300 escudos (some $132) per month, which would have effectively more than doubled the base wage of assemblers, who made up the largest classification of employees at the plant.Footnote 70 And sometime that spring, as we have seen, a politicized, militant workers’ committee had sprung up in the plant to make demands on the company.

By summer, management had seen enough. In August, with most of the production workforce away on summer vacation, executives publicly announced plans to rapidly draw down Portuguese operations, in which they would reduce employment to fewer than 300 by the end of the year. The workers countered by signing onto a joint letter issued by workers at several of the electronics firms that presented a sophisticated financial analysis demonstrating the firms could afford to pay the new wages and accusing the U.S. multinationals of economic sabotage against the revolutionary government and the promise it held for the democratization of Portuguese industrial life.Footnote 71 The announcement led workers to engage in an unsuccessful defensive protest, featuring disruptive pickets that temporarily prevented management from removing machinery from the plant. The workers’ committee proposed, in lieu of drastic layoffs, “a fifty percent reduction in work with no layoffs; purging two administrators; equal salaries for all; and that the workers would forgo their Christmas bonus of an extra month’s pay.”Footnote 72 Then, the Ministry of Labor stepped in and proposed to bear forty percent of the cost of the wage cost; Signetics agreed—but continued to cut employment at the plant.Footnote 73 A takeover of Signetics by the Dutch electrical company Philips in 1975 sealed the plant’s fate. Finally in November, the company announced it would close down by early 1977. The company visited the U.S. embassy, which was growing nervous about leftist backlash against capital flight, to relay this decision, arguing that while the workforce remained reasonably productive, the plant’s output—which had diminished dramatically as a result of conscious company policy in reaction to the political situation—could no longer justify its high capital cost.Footnote 74

The instances of organizing attempts at semiconductor plants in Scotland, Thailand, and Portugal show that firms in this period proved extremely reticent to countenance challenges to management control of the shop and prepared to quickly take dramatic action, up to and including total closure of the plant, when those challenges surfaced. The effect of unionization on wage levels was certainly a consideration, but the issue ran deeper and concerned a fundamental desire for more or less unquestioned control of production, unhampered by workers’ input into the speed of the line, safety rules, or hiring and dismissal. Strengthening support for this view are the cases where firms had relationships with unions for years, provided that those unions remained under the control of an authoritarian state. As we will see, semiconductor companies were willing to engage in a coercive corporatist bargain of higher wages for steady productivity gains. To be sure, management learned the hard way that even where unions existed that appeared to be fully incorporated into an authoritarian, developmentalist state, the specter of workers meaningfully taking control of those same organizations and employing them to their own ends lay just beneath the surface. Even so, what the case of Signetics Korea (as well as the National plant in Singapore, to which I also make reference, again to illustrate that the phenomena I describe are not unique to one company) demonstrates, above all, is that silicon capital tolerated a rank-and-file that displayed periodic episodes of militancy, which I attribute to the fact that capital was confident that the state could be relied upon to bring recalcitrant workers into line.

State-controlled unions

One place where semiconductor firms engaged in long-term relationships with state-dominated unions is South Korea, and the companies’ experiences in that country nicely illustrate the potential perils for their project of labor control of recognizing even these subordinate and politically conservative institutions. As many observers have noted, the Federation of Korean Trade Unions (FKTU), which had a virtual monopoly on legitimate unionism dating back at least to the early 1960s, had a close relationship not only with the ruling party (in the Park Chung-Hee years, 1961 through 1979, the Democratic Republican Party), but also with Korean intelligence, which bankrolled the federation and helped select its leadership. The federation lacked meaningful independence and was subordinated to business interests in the regime coalition, but as a tool for conferring popular legitimacy to the ruling party, it was occasionally granted limited concessions under a system that used compulsory arbitration to impose wage settlements in lieu of strikes. The danger for the regime, and for business interests, was workers’ tendency to periodically drive a wedge into the slight opening provided by their unions, dissenting from their leadership to press genuine rank-and-file demands.Footnote 75 This happened at least twice at Signetics.

Signetics had set up shop in Seoul in the fall of 1966. An article in the house journal of the state-funded Korean Trade Promotion Corporation touted the opening of the plant as heralding an influx of investment from makers of ICs and other electronic components. Playing into the gendered and racialized depiction of docile East Asian women with small, nimble hands, the journal featured an image of a row of Signetics workers hunched over microscopes, with the following caption: “The high educational level of Korean girls, their low wages and great hand-eye dexterity make them very attractive to foreign manufacturers. These girls are conducting microscopic quality control inspections.” The article, written by Signetics Korea (Sig-Kor) President James E. Stokes, went on to tout the “patience and diligence” of these “well-disciplined” workers.Footnote 76 Signetics was apparently pleased with its investment, building up employment levels to some 460 within the first two years and pouring money into the first of several plant expansions with the intent of eventually employing several thousand workers.Footnote 77

The honeymoon, however, did not last long. By the middle of 1968, the workers were in revolt against both management and the National Metal Workers Union, with whom Signetics had just concluded a collective bargaining agreement. The thrust of the problem was that the agreement was reached in one-on-one talks with the union president with no input from workers. According to ethnographic research by George E. Ogle, then working as a Protestant missionary in South Korea and collecting data for his industrial relations scholarship, the grievances at the plant stretched back to its opening, when Signetics sent a plant manager who workers claimed was both incompetent and abusive. By early 1968, the company had replaced with someone with a softer touch. The new manager’s first instinct was to attempt to mollify the assemblers by wooing the union that had been eyeing the plant, working out an agreement on representation over dinner at a hotel. The union returned to the workers and boasted that they had reached an agreement but left details sketchy. Appealing to the company for information, the workers learned that the two parties had agreed to a fifteen percent raise over the life of the contract—half of what the union had promised. Resentful that they had been cut out of talks, workers staged a raucous demonstration in the company cafeteria in which speakers took turns denouncing the union president as a sell-out. The company stepped in to break up the gathering, identifying six “ringleaders” who they claimed were responsible for the trouble and suspending them with immediate effect. That move only exacerbated the problem. A petition made the rounds demanding that the union president resign and the six workers be reinstated. It garnered signatures from virtually the whole shop, but the company would not be moved. The talks with the union had been aboveboard, it maintained, and Signetics would not “undermine company discipline” by reinstating these workers. Several days later, management invited the union president into the cafeteria to defend his actions and assail the fired petition organizers. When the union leader attempted to circulate a new petition backing the incumbent leadership a scuffle ensued resulting in several arrests. In defiance of both the company and the union, workers held their own gathering at which they elected a new slate of officers.Footnote 78 All of this led Liddie to remark that the “labor situation [in Korea] is a lot more volatile than originally expected.”Footnote 79

Several months of low-intensity industrial warfare ensued. Workers returned to work as normal, but workers slowed down the line and engaged in sabotage, and tensions in the plant ran high. Management warned that if workers continued to resist, it would shift future investments elsewhere, but this was a bit of an empty threat. At this stage, the company lacked the capacity to make up the shortfall, and workers at the plant would have seen the construction projects underway in the Seoul plant that clearly indicated Signetics had already committed to a production buildup in Korea. In this instance, the company caved: if the government recognized the legitimacy of the union officers that the workers had selected, Signeitcs promised, the company would do likewise. The company agreed to send outstanding grievances to arbitration, and the protests quickly died down. However, the Park government, worried that foreign investors would be spooked by protests that had occurred at Signetics and Illinois-based Oak Electronics earlier that year(the latter of which followed through on threats to relocate production), passed emergency legislation severely restricting union activities at foreign-invested firms. Under the law, strikes at such companies were banned, disputes were channeled into compulsory arbitration, and obstacles were placed in the way of new organizing attempts, resulting in a drastic reduction in the number of new unions formed in the early 1970s.Footnote 80

As Hwasook Nam has demonstrated, some of the fiercest repression in post-war Korea took place during the years of the so-called “Yushin Constitution” (1972 through 1979)—essentially affirming longtime President Park Chung-Hee’s right to rule by decree—of which a key element was to force all FKTU unions that exhibited any measure of independence under the secure control of the state. A few, courageous minju (democratic) unions continued to operate, but their margin for an independent existence, at least aboveground, was exceedingly small. It was a major concern of state officials that the women constituting the bulk of production workers in export industries, who faced much greater levels of exploitation than their male counterparts at the growing industrial chaebol, such as Hyundai and Daewoo, and they responded with corresponding levels of repression, lest foreign investors read too much into, say, the periodic unrest breaking out at the country’s textile plants.Footnote 81

According to sociologist Eu-Jin Lee, semiconductor ATP plants in South Korea (as well as, by the late 1970s, a growing number of wafer fabs) were notoriously harsh places to work in the 1970s and 1980s. Managers maintained punishing production quotas, which they set based on time-and-motion studies, and Signetics in particular (along with a Control Data plant) routinely revised its quota upward. Senior management would create the quotas and pass them along to plant managers, who would then impose quotas slightly higher, just beyond what they knew to be workers’ limits, to create relentless pressure to move faster. Workers required permission to leave their stations to use the restroom, and chipmakers encouraged workers to compete with one another by establishing prizes for those who beat the quota, while threatening workers who fell behind with pay cuts and dismissal. The major U.S. electronics companies with plants in Korea, such as Signetics, Fairchild, and Control Data, all had recognized unions which would bargain periodically, primarily over wages, but did little to make working conditions more tolerable.Footnote 82

In August 1977, workers at SigKor, after years of relentless increases in line speed and little corresponding rise in real wages, once again defied their union, the government, and the company, and engaged in an unsanctioned job action. Keeping wages into closer alignment with productivity increases was one of their core demands, but the workers also raised the issue of line speed, contending that working at a relentless pace at the microscopes without breaks was leading to severe eyestrain and headaches. Though the strike is not very well documented, available sources indicate that workers engaged in a sit-in in the company cafeteria lasting multiple days, while workers in the plant slowed down production. Asking for a forty-six percent wage increase, they eventually won a twenty-three percent raise.Footnote 83 If the company yielded any concessions to the workers on line speed, that was not a publicized dimension of the settlement. But it is not hard to see why the company felt the need to placate the workers; as had been the case in the 1968 dispute, management clearly relied very heavily on the ATP operations in Korea. It had no obvious way to transfer capacity elsewhere. Thailand was set up to assemble and test only a fraction of what Korea was then processing, and a new facility in the Philippines would not be ready to handle significant capacity for another two years.Footnote 84

Despite these instances of labor unrest and the existence of a union, Signetics maintained a presence in Korea for some thirty years. At least part of the explanation for that can be found in the fact that the company felt it could rely on the union to police workers’ militancy, that the government would step in on behalf of foreign investors when things seemed to slipping out of control, and that workers—intentionally or not—chose strategic times to protest conditions and a willingness to settle for higher wages when the company proved unwilling to compromise on issues of control over production.

In Singapore, state-controlled unions’ ability to police worker militancy and restrict themselves to wage settlements in exchange for productivity increases was much more total, such that even the fiercely anti-union National Semiconductor was willing to tolerate a union in its Singapore operation. Over the course of the early and mid-1960s, the regime of Lee Kwan Yew had repressed its leftist rival, the Socialist Party, and its union arm the Singapore Association of Trade Unions, and consolidated his power into de facto one-party rule. Abandoning the left-leaning nationalism of his youth, Yew sought to attract foreign investors in a variety of industries, including electronics, where he hoped to reap the fruits of technology transfer. He combined this repression of politicized labor protest with a paternalistic welfare state and a coercive corporatism in labor relations. For example, in the early 1970s, the government passed a law creating a National Wage Council that recommended annual wage increases corresponding to economic development. The moderate faction of the union movement was allowed to continue to function, but it was brought under the tight control of the state—which could not only force parties into binding arbitration, but could unilaterally alter agreements to bring them into line with National Wage Council recommendations—and they were forbidden from encroaching on management prerogatives. The result was that there were less than fifty work stoppages between 1970 and 1977, and none between 1977 and 1985.Footnote 85

One published agreement from the early 1970s between National Semiconductor and the Singapore Industrial Labor Organization reflects this paternalistic philosophy. A clause in its recognition article provides that, “The Union recognizes the rights of the Company to control, operate, and manage its business in all respects as it thinks fit,” and that the union “shall use its best endeavors to see that all its members loyally cooperate in working for the advancement of the Company’s interest and business in all respects.” The contract also contained an extremely restrictive grievance procedure that required five steps before referral to an arbitrator, mandated the union to file within three days within the date of occurrence, and allowed a window of three days for appeals at each step. On the other hand, the agreement contained paid annual leave and hospitalization benefits (albeit both fairly modest), and allowed for annual wage increases in line with the recommendations of the National Wage Council, in addition to a one-month wage supplement for the lowest pay grades, including assemblers, bonders and die attach workers, and materials handlers.Footnote 86 Given management’s experience with high turnover in their early years in East Asia, particularly in Hong Kong, this agreement made good sense from a company perspective.Footnote 87 It promised steadily rising wages but (through the mechanism of the Wage Council) tied the bulk of those raises to productivity improvements; enticed fickle new industrial workers to stay by offering a few fringe benefits; and channeled disputes into a pro-management grievance procedure (culminating in pro-management government arbitration), thereby attenuating collective militancy.

In short, the union in National’s Singapore plant played the role that Signetics undoubtedly had hoped the FKTU-affiliated union would at SigKor. Singapore was hardly a model liberal democracy during these years, but it did not have to resort to the kinds of harsh labor repression that South Korea had prior to its liberalization starting from the late 1980s (in response to pressures, it should be said, that the labor movement played a leading role in bringing to bear).Footnote 88 In Singapore, thanks in part to its supplementary authorities, such as the National Wage Council, responsible for implementing such arrangements, the corporatist bargain was sufficiently institutionalized to keep explosions of unrest—and the radical demands they tend to generate—at bay.

Conclusion

In the first group of case studies—those of Scotland, Thailand, and Portugal—I have tried to illustrate two persistent commonalities that survive the dramatically different social and political contexts involved. Examining the history of the Portuguese plant in isolation, it would be easy to dismiss Signetics’ decision to withdraw as based on the very rational fear of a revolution involving significant working-class grassroots mobilization and carrying (at moments, at least) an undeniable radical edge. The company arguably faced the greatest threat of worker expropriation of its property only after it decided to divest—that was the context for most factory occupations at the time, and it was the basis of workers’ unsuccessful bid to prevent management from moving machinery—but that did not change the fact that real political space had clearly opened for workers to press demands for greater control over the shop floor. And yet, the other two cases demonstrate that nothing so exceptional was at work. In Scotland, a slightly more favorable political environment for the unions than had existed a few years prior allowed a relatively militant union to come sign up a clear majority of the shop before Liddie could conspire to install a controlled opposition. Management may have claimed that the plant was technologically obsolete, but it never bothered to explain why it could not be upgraded and had never signaled an inclination to wind down operations there before the AEWU came on the scene. As for Thailand, where Signetics faced an organizing drive during that country’s (ultimately short-lived) period of quite moderate reforms—among them, creating a viable basis for legal unionism—Liddie and his team appeared shaken, not so much by the fact that workers were asking for a union or even that the union was pressing demands beyond wages and hours, but that the government had the temerity to inform workers of their rights. The only apparent reason Signetics did not withdraw is because its threats to do so succeeded in browbeating the workers into submission. The two commonalities, then, are: first, that meaningful worker challenges to management control of the shop floor, rather than demands for higher wages, were what the company deemed most unacceptable; and secondly, that Signetics’ determination about whether the host state could be counted on as a partner to beat back those challenges was ultimately decisive in informing the company’s foreign investment decisions.

By way of contrast, the case of SigKor shows that management was ultimately willing to countenance displays of rank-and-file agitation that were probably more disruptive than anything the company had seen in Scotland, Thailand, or Portugal. Signetics’ reliance on Korean production capacity, and workers’ excellent timing, offered the workforce in Seoul space to extract a better wage bargain than it would have received if it remained quiescent, but this reliance does not explain why Signetics did not consider eventual moves to pull out once other planned plants, notably one in the Philippines, could lighten the load. And Signetics had certainly proved willing to forfeit sunk capital before. I find the most plausible explanation to be the company’s trust in the state to protect its interests above all else. The strike in 1968 had been the government’s stated pretext for a draconian law banning strikes and other basic union freedoms at foreign-invested firms. Eager to stress the point, Park Chung Hee made extensive moves to bring the union movement under even tighter state control during the 1970s than it had been in the decade prior. Ultimately, therefore, Signetics rewarded Korea for its vigilance by staying the course, even after another strike broke out in 1977. The brief look at National Semiconductor’s Singapore operations indicates a variation on the same theme. Because it was able to institutionalize the corporatist bargain of modest wage gains for steady productivity improvements, National effectively instrumentalized the union as force to stabilize the labor force and police worker militancy. If a kind of ideal type of workers’ organization from the perspective of semiconductor capitalists can be said to exist, the union in Singapore came as close as one is likely to find.

By the 1980s, automated bonding and die-attach machines started to fill the foreign ATP plants, leading chipmakers to consolidate and scale back their foreign operations. At Signetics, for example, employment at the Korean plant peaked in 1979, while employment in Thailand peaked in 1984. Meanwhile, a new plant in the Philippines that broke ground in 1979 lasted only four years before the company divested. In 1995, the Korean operation was wound down and sold off to local investors. Total company employment dropped more than seventy percent between 1982 and 1994 despite an increase in sales volume and profit.Footnote 89 To be sure, the role that automation actually played in that increase in productivity can easily be overstated. In these years, U.S. chipmakers increasingly began to specialize in design, while contracting out not only their back-end processes, but all production. In 2008, when chipmaker Advanced Micro Devices spun off its manufacturing business and sold out a majority stake to an Emirati investment firm, it left Intel as the last major U.S. producer of ICs to actually manufacture its own microchips. The rest rely on “foundries,” dedicated manufacturing plants, and above all, the Taiwan Semiconductor Manufacturing Company (TSMC), for their front-end manufacturing.Footnote 90

ATP plants are still a presence in many of the same places that U.S. firms had set up shop in the 1960s and 1970s, if in smaller numbers and now typically run by contractors of native capital.Footnote 91 Despite heady industry rhetoric about “lights-out manufacturing,” there is no viable path at the moment to running these plants completely free of worker intervention.Footnote 92 The productivity gains from back-end automation were real enough, but we should not allow them to obscure the workers who remain. More to the point of this article, we should not allow them to obscure a period when remaining competitive depended as much on sweated labor as on technological improvements. In the formative years of the U.S. semiconductor industry, it was exploitation, rather than automation, that allowed companies to survive and the march of Moore’s Law to proceed.

Acknowledgements

The author wishes to thank Jeff Sklansky, Clare Kim, Aaron Benanav, and two anonymous referees, whose insights and critiques on drafts of this article have considerably improved the final product.

Competing interests

The author declares none.

Article contents

Extract

Information

“Jet-plane automation” and the “ethos of control”

Threats to management control: Scotland, Thailand, and Portugal

State-controlled unions

Conclusion

Acknowledgements

Competing interests

Comments

More materials