Introduction
In 1978, an American researcher Linda Gail Arrigo sat across from RCA’s personnel manager at the company’s Taoyuan plant. The manager described how RCA had called on the KMT party-state’s martial law authority to suppress a factory strike, and how the company had deliberately structured its wage package to push unwanted workers into quitting voluntarily. Arrigo documented these conditions in her field notebooks, which the Taiwan Garrison Command secret police soon seized. Arrigo eventually demanded her notebooks back. But the workers whose conditions she had documented, the tens of thousands of young women handling chlorinated solvents without adequate protection on twelve-hour shifts, would wait decades before anyone recorded their story. By then, thousands had developed breast, liver, cervical, colorectal, and ovarian cancers (Teng 2007).
These events belong to a history that Taiwan’s semiconductor historiography has consistently told as two separate stories. One takes place on the shop floor and in the dormitory. Semiconductor firms recruited young Taiwanese women from rural areas and placed them into export-oriented electronics production lines. They rewarded these young women for their speed, dexterity, and obedience. The other occurs in conference rooms and breakfast shops. Technocrats, engineers, and corporate interlocutors plotted Taiwan’s leap from low-tech export manufacturing to high-tech semiconductor production, particularly through the 1976 Radio Corporation of America (RCA) Project (ITRI 2023). Within Taiwan’s semiconductor industry historiography, the second dimension of state strategy and technology transfer has too often eclipsed the first.
Contemporary discourse reinforces this imbalance. Government and policy narratives routinely foreground Taiwan’s dominance in leading-edge manufacturing and the strategic significance of its firms. After all, Taiwan produces over 90% of the world’s most advanced chips and over 60% of all semiconductor foundry revenue (USITA 2025). Popular works like Chip War likewise privilege top-down technological development and capital investment (Miller 2022). Many Taiwanese regard the industry as 保國聖山, “a sacred mountain that protects the country” (Hsiao 2025). Such strategic framing can make labour harm appear politically inconvenient or even risky to acknowledge.
There is nothing inherently wrong with describing Taiwan’s semiconductor industry as strategically critical. But when such narratives become the default historiography, they risk reproducing a familiar erasure. The marginalised women and migrant workers who made the “strategic” sector possible disappear, and the authoritarian governance that oppressed them becomes invisible.
The scene that opens this paper illustrates what that erasure concealed. In the late 1960s, the authoritarian Kuomintang (KMT) regime invited RCA to build some of Taiwan’s first semiconductor assembly, testing, and packaging (ATP) facilities. The young women whom Arrigo encountered at RCA’s Taoyuan plant worked twelve-hour shifts handling chlorinated solvents like trichloroethylene, tetrachloroethylene, and trichloroethane without adequate ventilation or protective equipment. They lived in company dormitories enclosed behind checkpoints, often guarded by KMT military police. Their enterprise unions had no real independence. Decades later, thousands of these women developed cancers. The litigation that followed would expose not only the toxic contamination but also a systematic pattern of corporate evasion, manufactured scientific doubt, and regulatory complicity that the martial law regime had made possible.
This paper argues that these are not two separate stories about technological triumph and labour exploitation. They are a single history. The same institutional architecture that enabled the celebrated 1976 RCA technology transfer also governed the shop floors where women assembled transistors under coercive conditions. The same martial law apparatus that promised multinational firms “political stability” suppressed the worker organising that might have revealed toxic hazards in real time. The nineteen engineers who trained at RCA’s U.S. facilities and returned to build Taiwan’s first integrated circuit demonstration factory did not work in isolation from the labour regime that recruited, housed, disciplined, and ultimately poisoned the women on RCA’s production lines. They worked atop it.
Yet existing accounts consistently separate these dimensions. The technology transfer literature treats labour as a factor endowment. Labour history treats semiconductor upgrading as a separate institutional story. Taiwan’s transitional justice framework largely sidelines state-enabled corporate violence against factory workers. Why has Taiwan’s semiconductor historiography reproduced this separation? What did the KMT’s martial law apparatus do inside export electronics factories to shape recruitment, dormitory life, wage discipline, and occupational risk? And if the labour regime was integral rather than incidental to Taiwan’s technological ascent, what does that imply for the vast public investments now justified through similar “national interest” framings?
Historiography and Methods
This divide reveals itself in existing scholarship. Linda Gail Arrigo’s 1985 analysis of women workers in Taiwan’s multinational electronics factories remains the most important contemporaneous account of how martial law shaped factory life (Arrigo 1985). Based on late 1970s fieldwork, Arrigo linked piece-rate incentives, dormitory surveillance, and KMT military police to the broader structure of export-oriented industrial policy. But Arrigo was writing in real time. She could not trace the regime’s toxic consequences, its legal afterlife, or its relationship to the semiconductor upgrading story that had barely begun. Paul Jobin’s recent work addresses what Arrigo could not: the decades-long trauma of RCA’s former workers, the epistemic violence of contested causation, and the psychological toll of protracted litigation (Jobin 2025). Jobin explicitly connects RCA’s offshoring to Taiwan’s later semiconductor dominance. Yet his analysis, grounded in medical sociology, does not reconstruct the institutional machinery that made toxic exposure systematic rather than incidental. Together, Arrigo and Jobin supply this paper’s two essential poles. What neither does, and what no existing work has attempted, is to show how the same martial law apparatus that enabled celebrated technology transfer also foreclosed the worker voices who might have identified occupational hazards in real time.
Three additional bodies of scholarship fill out the analytical frame. Ming-sho Ho’s work on Leninist party-state control and labour federation politics documents how the KMT embedded its organisational apparatus within factories (Ho 2006), though it does not extend this analysis to occupational health or semiconductor upgrading. Contemporary World Bank reports reveal the de jure incentives offered to foreign investors while naturalising “political stability” and “cheap and industrious labour” as development inputs (IBRD et al. 1967). Taiwan’s transitional justice scholarship, focused largely on White Terror killings and imprisonments (Executive Yuan 2024), has not recognised how authoritarian infrastructure functioned as labour control within export factories. The RCA case illustrates why labour regimes deserve recognition as a distinct site of transitional justice reckoning.
Theoretically, the paper draws on: developmental state theory to frame how the KMT courted and constrained industry (Wade 2003; Amsden 1992), global value chain theory to examine how lead firms’ standards shaped local labour practices (Gereffi 2018), labour process theory to interpret piece-rate discipline and manufacturing consent (Braverman 1974; Burawoy 1982), feminist political economy to foreground gendered recruitment and dormitory governance (Arrigo 1985; Hsiung 1996; Kung 1995), and postcolonial critical theory to interrogate how institutional frameworks exploit workers and their bodies (Spivak 1988; Ricaurte 2022). These frameworks guide the primary source analysis.
The paper’s methodology relies extensively on primary sources that exist only because the regime’s concealment was imperfect. A post-martial law Taipei District Court verdict concerning RCA Taiwan documents toxic solvent use, groundwater pollution, and differential access to safe drinking water (Sung and Chen 2015). Thirty-five semi-structured interviews from research trips to Taiwan, South Korea, and the Philippines in 2024 and 2025 provide direct recollections from former RCA workers, migrant labourers in contemporary ATP supply chains, lawyers, labour organisers, NGO supporters, and scholars. Collected oral history records further document factory labour norms in 1960s-1970s industrial Taiwan, highlighting long shifts, forced overtime, workplace deaths, and the absence of recourse (沈德盛 2023).
A final point concerns scale. Numerous multinational and domestic actors shaped Taiwan’s ATP labour regime, but RCA is analytically central for two reasons. It was both a foundational company in Taiwan’s semiconductor origin story and a source of toxic contamination whose legal afterlife produced unusually rich documentation. That documentation retroactively illuminates workplace practices otherwise concealed by corporate and martial law secrecy. RCA thus serves simultaneously as a case of authoritarian-era multinational offshoring and as a window into the broader characteristics of early ATP labour regimes.
Building the Institutional Preconditions for Electronics Export
Martial law fundamentally conditioned Taiwan’s Cold War labour regime. During the White Terror period from 1947 to 1987, the KMT kept Taiwan under an official state of emergency, curtailing political rights, stifling opposition, and suspending many civil liberties (國家人權博物館 2017). These included the regime’s political killings, imprisonment, and torture of an estimated 140,000-200,000 dissidents, and broader persecution of ethnic Taiwanese intellectual elites (Cheung et al. 2022). The KMT’s rule extended into the workplace, where the authoritarian government effectively banned independent labour organising and treated worker dissent as a communist security threat (Jamison 2022). Even after martial law was lifted in 1987, the idea of “labour-capital harmony” continued to pervade Taiwanese politics (Huang 2001). But the labour regime that RCA and similar foreign firms would exploit did not emerge from martial law alone. It was built through a sequence of interlocking institutional developments: economic reforms designed to attract foreign capital, spatially bounded export zones that concentrated and governed production, a legal architecture that fragmented worker power, and a coercive security apparatus that backstopped the entire system.
Constructing the Investment Architecture
Taiwan’s ATP labour regime did not come from “culture” or market pressures alone (Deyo 2003). It arose from a martial law-era institutional environment built to facilitate export industrialisation in the late 1950s and 1960s. Incentivised by promises of further U.S. aid and ambitions of integrating Taiwan into global manufacturing networks, Taiwan pursued market liberalisation. By August 1960, KMT officials finalised a “Plan for Accelerated Development” that reduced tax burdens and removed industrial expansion constraints (Irwin 2021). These policy changes, paired with currency devaluation, significantly increased export-focused foreign and domestic investment.
The KMT’s reform package was comprehensive and deliberately sequenced (Wade 2003). Exchange-rate adjustments, tariff liberalisation, and a 1960 investment statute offering tax incentives and foreign exchange entitlements collectively constructed what contemporary economists recognised as an explicitly FDI-oriented policy architecture. As Yu Ching Jao observed at the time, these measures were designed for the “encouragement of foreign investment” (Jao 1976), a framing that treated foreign capital not as a byproduct of growth but as growth’s prerequisite. Export processing zones (EPZs) extended this logic spatially, bundling cheap land, infrastructure, and reduced bureaucracy into physically bounded zones optimised for multinational entry.
International development institutions reinforced this framing. A 1967 World Bank appraisal of the China Development Corporation catalogued Taiwan’s investment incentives, including tax holidays, customs duty deferrals, and duty-free import provisions. It attributed the island’s rapid growth to “political stability, appropriate Government policies, heavy American aid and the diligence of the economically active population” (IBRD et al. 1967). That language normalised repression as stability and managed low wages as diligence. By rendering authoritarian labour control legible as a technocratic input, such appraisals lent institutional legitimacy to a development model whose profitability rested on foreclosing worker agency.
Two features of export electronics made this particularly consequential for labour regimes. First, ATP depended on large numbers of low-paid workers rapidly performing repetitive tasks, requiring not merely low wages but wage discipline. This required a workforce whose compensation could be held below productivity gains without triggering collective action. Second, the EPZ framework of processing imported materials for re-export meant that Taiwanese labour was the primary site of value creation within the zone. Yet the governance structure ensured workers captured the smallest share of that value. Cold War geopolitics, economic reform, and labour control thus converged in a single developmental logic that made Taiwan particularly attractive to foreign electronics firms due to its authoritarian labour structure.
Export orientation evolved from a purely economic idea into an administrative form. Later Taiwanese semiconductor narratives often celebrate research and training organisations (Hsiao 2025). The earlier story of how the KMT built export-oriented industrial spaces and politically managed labour within them is equally important.
Export Processing Zones and the Spatialisation of “Industrial Peace”
EPZs were one of the most consequential spatial tools within Taiwan’s export strategy. KT Li, Taiwan’s “godfather of technology” (Saxon 2001), visited the free port city of Trieste in 1962 (Irwin 2021). Soon after, Li opened one of the world’s first EPZs in Kaohsiung in March 1965 (NYT 1967). The Kaohsiung EPZ offered firms tax and duty exemptions alongside purpose-built infrastructure for electronics, precision machinery, and related sectors (外交部 2022).
The Kaohsiung EPZ and its successors in Taoyuan, Taichung, and elsewhere by the early 1970s are typically narrated as investment promotion instruments (Haggard and Zheng 2013). They were also labour governance instruments, concentrating over 100,000 women at peak employment within bounded spaces where the state could regulate customs flows and human mobility (Kuo 2010). Some workers were underage girls who borrowed other people’s ID cards to gain entry (王諺玫 2009), likely hazardous child labour given the hours and chemical exposure (勞動部 2024). EPZ gates, dormitories, and checkpoints enclosed a workforce whose mobility could be monitored. Police classified EPZ employees as “special surveillance objects” (Chang 2023), collapsing industrial management into security enforcement. Workers were pushed into extensive overtime and chronic solvent exposure with little recourse (Jobin 2010). The EPZ administration’s dual mandate compounded the problem: the same authorities that promoted foreign investment also adjudicated labour disputes (Taiwan Panor. 1980), systematically favouring corporate interests over worker protections. Taiwan’s Control Yuan later criticised EPZs for courting investment while running their own labour inspection centres (高涌誠 and 王幼玲 2024).
While global value chain scholarship highlights how lead firms govern production networks by setting supplier standards (Gereffi 2018), the KMT state itself helped Taiwan’s EPZs institutionalise those requirements. The KMT state did not just invite multinationals, but it embedded their operational logic into local governance. The state was deeply responsive to and integrated with corporate priorities, yet it maintained just enough autonomy to enforce industrial priorities (Evans 1995). Taiwan’s Ministry of Economic Affairs and EPZ authorities embodied this duality, being highly responsive to multinational investment and production requirements yet also steering firms toward export targets and technology transfer goals. While institutions like the World Bank narrated EPZs as technical policy innovations (IBRD et al. 1967), they also functioned as authoritarian governance laboratories. Within these zones, the authoritarian state could refine labour control mechanisms while maintaining a facade of modernisation.
Corporatist Law and the Fragmentation of Worker Power
EPZs provided the spatial framework for Taiwan’s labour regime, but the legal system provided its structural scaffolding. While labour unions formally existed, they were legally constrained in a corporatist manner that fragmented worker power while preserving a veneer of representation. Taiwan’s Labor Union Law limited unions to individual plants and stipulated mandatory membership (S.-J. Chen et al. 2003). The law also required factories to have at least 30 workers for unionisation and divided legal unions into categories that limited sectoral bargaining powers (勞動部 2022). Industry-wide or independent unions, along with the right to strike, were banned outright. From 1947 until 1987, no legal strikes occurred in Taiwan, and KMT authorities swiftly crushed or defused any collective worker actions (Winn 1987). As a Taiwanese commentator put it, “Totalitarian rulers do not adhere to [the right to strike]. They regard strikes as rebellions to be suppressed. Taiwan’s former regime was no exception” (Huang 2001). Even within permitted single-factory unions, management or KMT loyalists often occupied leadership roles, earning these organisations the moniker 閹雞工會, “castrated chicken unions” (Huang 2001), for their lack of real bargaining power. During the martial law era, unions served as “auxiliary institutions of industrial peace… [and] administrative arms of the government for the implementation of national policies.” Some even helped the KMT mobilise workers to win sham elections (S.-J. Chen et al. 2003).
Administrative mechanisms further constrained workers’ voices. When ministries disagreed over worker protections, the Ministry of Economic Affairs’ export promotion priorities often took precedence over labour reform. The Ministry of Interior sometimes made superficial gestures for “internal reform and protection of workers,” suggesting that women workers should get mandatory breastfeeding time in 1978 (Arrigo 1985). But in practice, these policies would hardly help Taiwan’s largely unmarried young factory women. The KMT even set up a special organ within the Ministry of Economic Affairs to arbitrate foreign corporate disputes separately from public litigation within the Ministry of Justice, making labour disputes less visible and publicly contestable (Arrigo 1985). Single enterprise unionism thus fragmented worker power, while administrative channels diverted grievances from public arenas where they might have gained traction.
Leninist Discipline on the Shop Floor
To implement their labour control regime, the KMT ironically built a Leninist party-state apparatus in workplaces (Ho 2007). Despite their purported opposition to communism, the KMT’s Bureau of Labor and the official Chinese Federation of Labor embedded KMT party cell representatives and secret police within large factories (Ho 2006). These secret police officers from the Taiwan Garrison Command served as 安全室主任, “chief of security” (Hsin-Hsiang Chen 2025). Such extrajudicial processes surveilled and quelled unrest often before worker organising efforts became public. Workers were expected to resolve grievances through state-approved channels, if at all. For example, a worker might complain to a company welfare officer or at a labour-management meeting, but these mechanisms had no real power independent of management. Any attempt to organise outside the official structure risked severe punishment. The KMT brutally killed all who tried to form independent committees or protest. After all, on February 28, 1947 alone, the KMT killed up to 28,000 Taiwanese over a government tobacco monopoly dispute (Forsythe 2015). This repression extended to dissenting foreign academics and journalists as well, some of whom the KMT even murdered extraterritorially (J. Pan 2021).
The state’s enforcement of occupational safety and health regulations followed the same logic of performative governance. Martial law allowed foreign firms like RCA to draw on a remarkably compliant workforce. The legal architecture constrained workers de jure. The enforcement and security apparatus ensured their de facto subordination (Minns and Tierney 2004). Even after Taiwan enacted its first weak industrial labour code, the government “did not strongly require employers to comply,” so even state-owned enterprises often failed to correct violations (何明修 2021).
Only crisis compelled the KMT to legislate. In 1972, roughly a dozen workers at a Philco plant in Tamsui, which assembled electronic components for Ford, died agonising deaths from liver diseases caused by inhaling industrial poisons (國立歷史博物館 2020). The incident compelled the KMT to enact the 1974 Labour Safety and Health Act (林宜平 2011), Taiwan’s first major occupational safety statute. Yet penalties were so minimal that employers learned it was cheaper to ignore the law than to comply (Winn 1987). As RCA’s own inspection record would later demonstrate, meaningful enforcement never followed.
In Taiwan’s export electronics sector, “industrial peace” meant far more than the absence of strikes. It rested on interlocking mechanisms of spatial enclosure, legal fragmentation, administrative diversion, coercive surveillance, and performative regulation that together made refusal costly and collective action nearly impossible. Taiwan’s industrial relations encompassed more than simply employer-worker relations. They were state-mediated relations designed to reduce publicity, delay resolution, and constrain the arenas in which workers could speak.
“If the Prosecutor is Too Annoying, We’ll Just Change Him:” Factory Life in the 1960s
Oral histories from Taiwan’s broader industrial milieu confirm that workers felt these institutional weaknesses on the factory floor. In an interview record, Shen Desheng recalls that in the 1960s, factory work often ran on a two-shift system from 8 a.m. to 8 p.m. and 8 p.m. to 8 a.m. Base wages were low, making overtime pay indispensable, and factories provided inadequate industrial safety protections. Given these circumstances, Shen recounts a fatal elevator accident that occurred under “normal” conditions. Then still a young manager, Shen personally held the bleeding worker and transported him to Mackay Hospital in a three-wheeled cargo cart. No ambulance came, and no emergency protocol existed. The company did not initially regard itself as responsible, and Shen recalls that “there was not yet labour consciousness and labour rights” at the time. Next, when a prosecutor arrived to investigate the safety failure, the managers said 檢察官如果太囉唆，就換一位, “If the prosecutor is too annoying, we’ll just change him” (沈德盛 2023). Capital was insulated not merely by weak enforcement or toothless statutes but by the capacity to override the judicial process itself.
If Shen’s account illustrates how individual firms avoided regulatory scrutiny, Arrigo’s account of a sit-down strike at General Instruments Taiwan reveals how the state enforced that impunity at scale. After the factory manager failed to issue a customary Autumn Festival bonus, thousands of Taiwanese workers sat idle and refused to work (Arrigo 1985). The American manager soon called the KMT military police to suppress the strike. Under martial law, leading labour disturbances was a capital crime. This episode illustrates that for multinationals, Taiwan’s promise of “political stability” meant not only minimal protests, but also access to a state apparatus capable of quickly extinguishing emergent dissent.
These accounts are not specifically about RCA, but they establish the operative conditions in which RCA built its ATP labour regime. By the late 1960s, Taiwan’s factory life was already characterised by long hours, minimal safety protections, and managerial authority reinforced by social hierarchy and state coercion. The authoritarian regime had scaffolded investment incentives, concentrated export zones, and cultivated the “stable” political environment that constrained collective bargaining. RCA is significant not only because it used these conditions, but because it consolidated them into labour structures that outlasted both the company and martial law.
U.S. Trade Policy, Geopolitics, and RCA’s Turn to Taiwan
RCA’s Taiwan expansion in the late 1960s emerged from U.S. trade policy, Cold War geopolitics, and domestic labour instability. These forces helped craft a transnational production network in which companies offloaded coercive labour discipline to authoritarian governments.
The U.S. tariff code’s offshore assembly provisions were critical to this system. By allowing firms to export components for overseas assembly and pay reimport duties only on foreign value added, these provisions did not merely accommodate global production networks. They actively constructed them as economically rational (Feenstra et al. 2000). By 1970, as U.S. Tariff Commission data confirm, substantial shares of American semiconductor imports by value had already entered under offshore assembly provisions (Sutton et al. 1971), evidence that production fragmentation was not a corporate improvisation but a legally structured outcome. Taiwan had become one of the principal destinations within this architecture, receiving U.S. components for assembly and shipping finished goods back under preferential duty treatment. Offshoring, in other words, was not simply something firms chose to do in response to wage differentials. It was embedded in trade law, statistical categories, and formal policy channels that made the dispersion of production across sovereign borders a codified feature of American industrial strategy.
Such U.S. policy architecture complicates narratives that reduce offshoring simply to cheap foreign labour. Cost differentials mattered, but they operated within legal and tariff infrastructures that made global assembly legible and profitable. Successive rounds of GATT negotiations reinforced these dynamics. The Kennedy Round in the 1960s and the Tokyo Round in the 1970s lowered trade barriers on finished electronics (Kageenu 2022), further incentivising U.S. firms to shift labour-intensive assembly to locations where they could arbitrage wage differentials against tariff reductions.
Domestic conditions sharpened these incentives. U.S. electronics manufacturing wages rose nearly ten percent per year through the 1970s, (DeSilver 2018) and the politics of labour recruitment grew increasingly contentious. American chip firms had experimented with recruiting feminised and racialised workforces that they thought were inherently suited to delicate assembly. But this framing fell apart when workers organised. The American Indian Movement’s occupation and shutdown of Fairchild’s Navajo reservation plant illustrated how U.S. production sites were vulnerable to political disruption (Nakamura 2014). Thus, electronics companies relocated their facilities offshore to places with “political stability,” lower wages, and less militant workers. For RCA specifically, this began first by hiring Eastern European immigrants in Camden, New Jersey, followed by Scotch-Irish Americans in rural Indiana, then African Americans in Memphis (Cowie 1999). Only after RCA had exhausted “desirable” sources of labour in the United States did it offshore to places like Mexico and Taiwan (Nakamura 2014).
Underwritten by authoritarian governance, Taiwan offered such stability. The U.S. government actively encouraged offshoring to Taiwan. As the KMT regime was representative of “Free China,” significant American aid already went to Taiwan. Private American investment would further serve U.S. Cold War priorities by stabilising Taiwan’s anti-communist regime through export-led growth (CFR 2025). Therefore, American aid officials and Taiwanese planners collaborated to establish export processing zones and attract U.S. firms. Declassified U.S. cables confirm Washington’s interest in Taiwan’s EPZ experiment as a replicable model (US Embassy ROC 1973). For firms like RCA, Taiwan’s appeal lay not only in low wages or export incentives, but also in a U.S.-backed authoritarian KMT regime that reduced labour disruption risk.
RCA’s Labour Regime on the Shop Floor
Authoritarian labour laws and export-oriented institutions established the preconditions for Taiwan’s electronics labour regime, but RCA and similar multinationals made that regime concrete on the shop floor. Their practices operated across three registers of control: spatial and demographic governance over recruitment, housing, and movement; economic and organisational discipline through wage structures and corporatist unionism; and epistemic control through performative regulation and suppressed accountability. RCA’s Taiwan operations are especially revealing because of the company’s foundational role in Taiwan’s semiconductor story and the unusually rich evidentiary record left by its later litigation.
Dormitories, Checkpoints, and the Gendered Workforce
Managers controlled almost all aspects of workers’ lives, spanning recruitment, training, housing, incentives, and time management. Employers consciously used martial law to secure control over the factory workforce. Arrigo details how unions, media, military police, and personnel office hierarchies facilitated such control (Arrigo 1985).
Gender was a key element of managerial recruitment. RCA specifically sought young women, valued across the export electronics sector for their supposed dexterity and docility (Nakamura 2014; Jobin 2025). However, Taiwan’s particular configuration was shaped by migration, dormitories, and authoritarian labour governance, evolving into an ATP labour regime with distinct disciplinary mechanisms.
Gendered recruitment intersected with family and rural-urban migration patterns. Patriarchal family structure often permeated factory women’s lives. Although young factory women typically worked temporarily before marriage, their earnings played a crucial role in sustaining family livelihoods and financing the education of younger siblings (Arrigo 1985). Such temporality reduced the likelihood that these factory women would organise and bolstered managerial power over a constantly rotating workforce. Many young women who got jobs at one of RCA’s “model factories” were often privileged 外省人, “waishengren/mainlanders,” who fled with the KMT and considered themselves lucky (Former RCA Worker 2026, 2). KMT-affiliated youth programmes had conditioned them to follow authority. If anyone stepped out of line, covert agents, informants, or police would quickly harass or blacklist her (Y. Chen et al. 2019).
Dormitories extended managerial control beyond the factory confines. Managers misleadingly narrated dormitory housing as a benevolent employee “benefit.” Arrigo initially describes the RCA dormitories as “model” facilities with airy halls, clean bathrooms, and house mothers (Arrigo 1985). But this welfare “perk” disguised constant surveillance, curfew enforcement, and social control. The dormitories were “enclosed in a compound next to the factory, with a single gate and checkpoint [often guarded by KMT military police] where all must sign in and out” (Former RCA Worker 2026, 2). Given their low wages, many women workers had scarce alternative urban housing options. Panoptical dormitories thus became factory extensions, allowing managers to secure punctuality, normalise long shifts, constrain collective organising, and continue supervising workers. Managers justified dormitories through paternalistic rhetoric of “care” and protection of women’s virtue (Kung 1995). The “meager dormitory environment and lack of privacy” became a severe strain over time (Arrigo 1985).
State and corporate actors also presented the factory-dormitory complex as an extension of the family and school, summed up in the slogan 以廠為家，以廠為校, “take the factory as your home, the factory as your school” (陳信行 2006). This ideology treated factory women as those in need of discipline and uplift, legitimising their tight regulation in company housing (Hsiung 1996). Materially, these measures served to limit factory women’s autonomy. The KMT’s ideology of moral order dovetailed neatly with employers’ need for a disciplined, trouble-free labour force.
Figure 1. A Taiwan factory dormitory during Arrigo’s late-1970s fieldwork (Arrigo 2025). Bunk beds at close quarters with no partitions and a single entry point made every occupant visible to house mothers and security officers.
But the “model” dormitory benefits were temporary. Following the 1974 oil crisis, multinationals shifted toward cheaper, more provisional dormitory arrangements (Arrigo 1985). Having already established a framework of residential control and surveillance, managers cut back on the expensive benefits while retaining the disciplinary infrastructure.
Wage Design as Coercion
Managers did not merely rely on walls. They also converted wage structures into instruments of compliance and punishment. Managers often attached substantial “bonuses” to base wages, which workers forfeited for “offenses” like being five minutes late, taking sick leave, or refusing overtime (Arrigo 1985). Because overtime pay was calculated on the no-bonus base wage alone, the effective overtime rate was far lower than the multiplier implied. Further piece-rate pay frameworks, along with speed and quality output targets, allowed RCA managers to exploit workers without overtly appearing coercive. This is where labour process theory is particularly applicable (Braverman 1974). Gamified production, quotas, and performance pay bonuses enabled workers to discipline themselves.
An RCA personnel manager even explicitly admitted to Arrigo that he had invoked the KMT party-state’s martial law authority to suppress a factory strike (Arrigo 2025). The manager also proudly described how RCA deliberately designed its wage package to coerce overtime compliance. If workers refused overtime, the resulting pay reduction was structured to “piss the workers off” enough that they would quit voluntarily. This psychological manipulation served a precise fiscal purpose. By inducing voluntary departures rather than issuing dismissals, RCA avoided paying severance and could arbitrarily lower its labour costs by approximately six percent (Arrigo 2025). The manager’s candour transforms the structural account of state-corporate alignment into a documented management strategy in which KMT martial law enforcement and wage design operated as complementary instruments to optimise labour cost.
Wage structures that individualised punishment and reward also fragmented solidarity, particularly when combined with dormitory governance that isolated workers from family and subjected them to surveillance. With collective action foreclosed, workers’ primary coping strategies tended toward keeping bonuses, individual bargaining, quitting, or other informal tactics rather than collective disruption. When individual grievances did surface, often regarding excessive overtime or harsh dorm rules, they usually led to minor concessions or dismissal. Workers perceived as troublemakers were fired (賴建寰 2012). Such heavy-handed governance caused union and strike rates to remain very low even after they became legal following Taiwan’s democratic transition (Congju 2019).
Rather than serve workers, RCA’s enterprise union furthered this corporatist control. One particularly telling anecdote concerns RCA workers discovering that their managers deducted union dues from their already low wages. Several women subsequently submitted their resignations, prompting an American manager to respond, “You want to organise a union, not me” (Hsin-Hsiang Chen 2025). The exchange demonstrates that unions could categorically exist but remain opaque and imposed upon workers, revealing the precise condition of enterprise-level unionism under authoritarian corporatist frameworks. In interviews decades later, former RCA workers recall that during the 1970s, they could not speak up and that there were “not really” unions or support structures available (Former RCA Worker 2024). One worker noted that she had to express discontent through RCA’s internal hierarchy “or face severe punishment” (Former RCA Worker 2024).
Performative Regulation and Knowledge Suppression
Beyond helping suppress labour movements, the KMT’s regulatory enforcement regime also served foreign capital. When RCA was later found to have repeatedly violated the Labour Safety and Health Act, the general manager was convicted but fined only NT$6,000 (Hsin-Hsiang Chen 2025), roughly US$150 (UPenn 2026). The fine was so insignificant relative to the cost of compliance that it amounted not to deterrence, but to tacit permission.
The inspection record at RCA’s Taoyuan plant confirms this pattern. Between 1975 and 1991, government agencies conducted nine inspections and documented 53 violations (Sung and Chen 2015), including failures to install local ventilation, label toxic organic solvents, and provide legally mandated health examinations. Yet 16 years of documented noncompliance produced no meaningful remediation. The violations were recorded and left unaddressed, creating an appearance of oversight while production continued. RCA’s management knew its groundwater was contaminated, purchased distilled water for executives but not workers, and repeatedly failed to remediate violations after formal notice. Such choices only made economic sense where noncompliance carried no real consequences.
The same logic of performative oversight extended to how the KMT suppressed external documentation. The U.S. Embassy told researchers like Arrigo to stop investigating labour conditions and engaging with the tangwai (anti-KMT opposition) (Arrigo 2025). Despite holding U.S. citizenship, Arrigo had her field notes on RCA factory conditions seized by the Taiwan Garrison Command secret police. Agents reviewed and stamped every page before returning the materials.
Figure 2. Arrigo’s field notebook, seized and stamped page by page by Taiwan Garrison Command, with seal visible at bottom left (Arrigo 2025). Handwritten entries record company names including RCA, General Instruments, and Texas Instruments, wage figures, and references to “double accounting.”
The suppression of labour discourse extended beyond individual researchers to academic institutions. In 1978, Arrigo helped organise a labour conference at Tunghai University where RCA’s personnel manager and other industry figures described worker discontent and managerial strategy (Arrigo 2025). The department chairman intended to publish the proceedings, one of the few contemporaneous records of multinational labour practices under martial law. The publication never appeared. The government soon installed former Central Police University president Meiko Wong as Tunghai University president (Tunghai Univ. 2016), and the proceedings were buried. Arrigo was then “shut out” by the U.S. Embassy that had already urged her to stop the research. The resulting archival silence was not neutral. It was produced by the same state-corporate alignment that operated on the shop floor.
Feminisation, dormitory governance, enterprise unionism, and performative regulation enabled Taiwan’s electronics export sector to maintain a stable labour supply, constrain bargaining, normalise intense work rhythms, and externalise occupational risk. RCA’s Taiwan presence benefited from, and depended on, this hierarchy of discipline and risk distribution. The U.S. tariff regime, Cold War security posture, and Taiwan’s managed labour force served as complementary infrastructures. The United States provided the legal pathway for fragmented production. Taiwan provided a politically disciplined and gendered workforce. And both states collaborated in ensuring that the conditions under which that workforce laboured remained difficult to study, publish, or contest.
Figure 3. Women assembling television components at an RCA Taiwan factory (Teng 2007). No gloves, masks, or ventilation hoods are visible. Originally published in Taiwan Panorama to project industrial modernity.
The RCA Project and the “Technology Transfer” Labour Regime
Yet to characterise RCA as simply an exploitative extractor would be as misleading as celebrating it solely as a modernisation vector. RCA was both a pioneering foreign investor that helped Taiwan build its industrial base and a company whose labour control mechanisms and poor occupational safety practices left lasting physical and mental scars on its former employees. This duality is constitutive of how authoritarian developmental regimes legitimised themselves. The KMT’s claim to legitimacy required economic modernisation through modern facilities, increased exports, and employment. But maximising profitability required RCA to control costs, discipline workers, and externalise risk. The canonical story of Taiwan’s semiconductor ascent, centred on the 1976 RCA technology transfer project, was built atop this labour regime rather than apart from it.
Nineteen Engineers and a Founding Myth
Taiwan’s semiconductor history is often structured around a mid-1970s shift from labour-intensive assembly to higher-value production, with the “RCA Project” as its central milestone. The canonical account describes a four-year, roughly US$10 million initiative led by Minister of Economic Affairs Sun Yun-suan to transfer RCA chip technology to Taiwan (Patterson 2015). A retrospective from the Industrial Technology Research Institute (ITRI) highlights a 1974 meeting at which seven men, including Sun and RCA Laboratories Director Pan Wen-yuan, planned Taiwan’s semiconductor development (ITRI 2023). In 1976, ITRI signed a technology transfer and licensing contract with RCA, then sent 19 Taiwanese engineers to RCA facilities in the United States for integrated circuit design and manufacturing training. On returning in 1977, they helped build Taiwan’s first IC demonstration factory, whose 70% yield surpassed RCA’s roughly 50% yield in New Jersey (ITRI 2023). The project helped produce Taiwan’s first local chip company, United Microelectronics Corporation, spun off in 1980. Yet as these engineers absorbed RCA’s expertise, they also encountered and adapted its labour control techniques, which later travelled into indigenous chip firms (Hsiao 2025).
What the Pilot Line Required
But the pilot line’s success depended on institutional conditions the canonical narrative leaves unexamined. Three specific dependencies make this concrete. First, the RCA Project’s pilot IC fabrication line required a workforce already habituated to the exacting procedural discipline of electronics production. This work required contamination protocols, microscopic component handling, time-sequenced process steps, and quality inspection routines. That workforce did not materialise spontaneously. It was produced over the preceding decade through the KMT-RCA labour regime. The EPZ-based electronics sector, organised through feminised recruitment, piece-rate incentive systems, and dormitory governance, trained tens of thousands of young women in precisely these capacities. Their daily work assembling and testing transistors, capacitors, and television components under RCA and similar multinationals constituted a vast, informal training infrastructure in standardised precision manufacturing. When the 19 ITRI engineers returned from RCA’s U.S. facilities in 1977, they did not build the demonstration factory alone. They staffed it with workers drawn from this existing labour pool, operators whose compliance with procedural standards had been shaped not by semiconductor-specific education but by years of managed factory discipline. The pilot line’s celebrated 70% yield was thus not simply a triumph of transferred engineering knowledge. It also reflected the quality of a workforce whose precision and obedience had been institutionally produced through martial law labour structures.
Second, the RCA Project required political predictability across a multi-year timeline, a condition that technology transfer’s substantial upfront costs and delayed returns make especially sensitive to disruption. RCA and the KMT state had to guarantee that training investments, intellectual property transfers, and pilot-line construction would not be jeopardised by labour conflict, political instability, or public controversy over working conditions (S.-J. Chen et al. 2003). Authoritarian labour governance directly supplied this guarantee. The interlocking mechanisms of union control, security surveillance, and strike prohibition ensured that the labour disruptions, which had plagued RCA’s domestic operations from Camden to Memphis to the Navajo reservation, were structurally precluded in Taiwan. This insulation from shop floor conflict constituted a subsidy to technology transfer that never appeared in fiscal accounts but was no less consequential for the RCA Project’s success. It is difficult to imagine RCA committing proprietary IC manufacturing processes to a country where those processes might be interrupted by the very workforce executing them.
Third, the canonical narrative implicitly treats the RCA Project as an upward trajectory away from labour-intensive production toward a high-skill national economy, as though technology transfer and labour exploitation occupy different historical phases. But semiconductor production has never shed its dependence on labour-intensive ATP. Even as the ITRI pilot line demonstrated Taiwan’s capacity for front-end wafer fabrication, the island’s electronics sector continued to expand its back-end assembly, testing, and packaging operations. The RCA Project did not eliminate the need for a controlled, low-cost ATP workforce. It added a new layer of high-value production on top of it. The same institutional architecture that governed ATP, including EPZ spatial enclosure, dormitory housing, constrained unionisation, and administrative mediation, continued to operate alongside and beneath the new semiconductor fabrication facilities. Indeed, as Taiwan spun off UMC in 1980 and later established TSMC, these firms replicated many of the managerial techniques their founders had absorbed during the RCA era. Those techniques encompassed not only IC process engineering but also the labour governance practices that had made RCA’s Taiwan operations profitable. The layered division of labour between capital-intensive fabrication and labour-intensive ATP was not a transitional phase but a structural feature of semiconductor production (Warf 2020), one that the RCA Project’s institutional descendants inherited and perpetuated.
The RCA Project sat atop the labour regime, drawing on a disciplined workforce the regime had produced, a political stability the regime enforced, and a layered production structure the regime sustained. In turn, its success legitimised the very state-capital alignment that had made ATP exploitation profitable.
Afterlives: Toxicity, Litigation, and the Persistence of Managed Labour
What Democratisation Made Contestable
Taiwan’s democratisation, along with the growth of environmental and labour NGOs, shifted what could be publicly contested. Jobin’s study of hazards and protest in Taiwan describes how rapid industrial development under authoritarian KMT rule was effectively a “slash and burn” policy, and it treats electronics hazards as emblematic of the paradox of “clean” industry (Jobin 2010). Dramatic catastrophes often affect public image much more than chronic pollution and health risks. The same study explicitly frames RCA’s Taoyuan plant pollution and worker cancers as part of the larger story of Taiwan’s electronics boom and the struggle for visibility. From this perspective, the RCA case illustrates how invisibility sustained earlier labour regimes. It also shows how coalitions of workers, activists, and experts have redefined the legal and scientific contestation of toxicity for a renewed labour struggle (Hsin-Hsing Chen 2023).
RCA’s Toxic Legacy and Contested Causation
Structural labour violence is not always immediately visible. In electronics, chemical exposure is one of the most consequential and insidious forms. Throughout its decades of operation in Taiwan, RCA used chlorinated solvents like trichloroethylene (TCE), tetrachloroethylene (PCE), trichloroethane (TCA), and methylene chloride as degreasers and cleansers on its production lines (Sung and Chen 2015). Workers absorbed these chemicals through inhalation, skin contact, and contaminated drinking water, while the solvents leached into surrounding soil and groundwater. These solvent concentrations exceeded legal limits, and the soil contamination reached extreme multiples of remediation standards. More than 80,000 workers, mostly young women, passed through RCA’s Taiwan factories during the three decades before the company shifted to water-based cleaning methods in the early 1990s (Jobin 2025). For the entirety of that period, the labour regime’s discipline, speed requirements, and dormitory containment kept production flowing without meaningful interruption or scrutiny.
Figure 4. A remediation notice at RCA’s former Taoyuan site, photographed during a January 2026 field visit (Former RCA Worker 2026). The sign designates an active groundwater pollution remediation project, ongoing more than 30 years after RCA’s closure in 1992.
RCA’s handling of these hazards followed a pattern of substitution rather than elimination (Sung and Chen 2015). TCE served as the main degreaser, later replaced by PCE, then by TCA. Each swap exchanged one toxic compound for another. The company knew its groundwater was contaminated and that activated carbon filters in shop floor drinking fountains could not remove hazardous pollutants. RCA purchased distilled water for management but not for “common workers.” The industrial risk distribution thus mapped onto occupational hierarchies: workers bore significant toxic exposure while management insulated itself. Corporate paternalism and internal messaging in the RCA Home Magazine, the company’s factory publication, celebrated welfare and modernity while obscuring such risks (Hsin-Hsiang Chen 2025). Interviews with former RCA women factory workers confirm that, at the time of their employment, they had no idea their workplace was contaminated (Former RCA Worker 2024, 1).
Evidence of contamination first surfaced through an unlikely channel. Around the end of 1986, Taiwan’s Investment Commission was notified that RCA’s parent company had changed hands through a U.S. merger, transferring ownership to General Electric (GE) (Li 2025). GE subsequently sold the operations to Thomson. Ironically, this corporate succession triggered the first pollution investigations, as environmental consulting firms conducted site contamination assessments as standard due diligence during mergers and acquisitions. Public evidence suggesting RCA had used toxic contaminants in its former factory therefore emerged at the very end of Taiwan’s martial law era, three decades too late for many factory women (Jobin 2025). RCA’s Taiwan factories closed in 1992, but the contamination they left behind was only beginning to become publicly visible.
The litigation that followed proved extraordinarily difficult. Former RCA workers filed suit in 2004, but proving disease causality decades after exposure posed immense evidentiary challenges (Jobin 2025). RCA’s Taiwan successors claimed that “we don’t have anything at all—we have no data because everything was burned down” (Hsin-Hsiang Chen 2025), and the company itself no longer existed in its original form. Anticipating liability from the emerging pollution revelations, RCA Taiwan colluded with its parent entities, Technicolor and Thomson, to remit over NT$3.2 billion, roughly US$80 million (UPenn 2026), in company assets to banks in France and other foreign jurisdictions between July 1998 and January 1999. The Taipei District Court later identified this transfer as a “fraudulent transfer of assets and malicious evasion of debt” (Sung and Chen 2015), deliberately designed to decrease the company’s capitalisation below its anticipated liabilities to the poisoned workers. The same transnational corporate architecture that facilitated offshore production was repurposed to offshore accountability. Global value chains, in this light, are not merely mechanisms for optimising labour arbitrage but structures through which corporations can disaggregate legal liability across jurisdictions.
RCA’s litigation strategy extended beyond jurisdictional complexity to manufacturing scientific doubt (Sung and Chen 2015). Its legal team retained Dr. Steve Lee, an oncological radiologist, to testify on causation. The court found him unqualified in epidemiology and toxicology, the fields relevant to chemical exposure and disease. Despite the International Agency for Research on Cancer classifying trichloroethylene as a Group 1 carcinogen, Lee claimed RCA’s operations did not cause the workers’ cancers. The judges condemned his testimony as an “attempt to confuse and mislead” the court, made with the “obvious intention to deceive.” The episode shows how corporations can weaponise credentialed expertise against sick workers. What Jobin calls “epistemic and psychological violence” operated through a specialist whose credentials lent superficial authority to claims outside his competence (Jobin 2025). Workers denied safety information and collective voice could not produce contemporaneous exposure records that would have simplified causation decades later. The epistemic violence of the courtroom thus extended the epistemic closure of the shop floor. Despite hundreds of plaintiffs, multiple lawsuits, and more than a decade of litigation (Li 2025), the former RCA workers ultimately secured compensation.
After years of hearings, the Taipei District Court ruled in favour of hundreds of RCA worker-plaintiffs and ordered compensation on April 17, 2015, in what became Taiwan’s first class-action labour lawsuit (Sung and Chen 2015). Court decisions from 2015 onward recognised multicausal links between the chemicals and diverse cancers and illnesses, culminating in a 2022 Supreme Court decision recognising the psychological stress caused by past exposure (Jobin 2025). A Taiwan High Court ruling in January 2025 ordered further compensation (BHRC 2025). The RCA case forced courts to confront corporate liability for long-latency disease amid missing records and contested epidemiology. The legal struggle became bound up with martial law transitional justice debates over what it means to prove harm with decades-long causal chains.
Figure 5. Former RCA workers outside the Taipei District Court after the April 2015 ruling (H.-H. “Indy” Liu and Chu 2015). The banner reads 勝訴, “victory,” but the placards bear photographs of deceased former workers who did not survive the litigation.
The Transitional Justice Gap
Yet the transitional justice framework within which that legal struggle unfolded was not designed to accommodate cases like RCA’s. Taiwan’s post-democratisation reckoning with the White Terror has overwhelmingly centred on discrete acts of state violence. These include the execution and imprisonment of political dissidents, the persecution of Taiwanese cultural and intellectual elites, and the ethnic dimensions of mainlander-dominated authoritarian rule. The 2018 Transitional Justice Commission focused on rehabilitating political victims, declassifying political archives, and removing authoritarian symbols (Executive Yuan 2024). These were necessary interventions. But they defined authoritarian harm primarily through a political persecution lens that renders labour suppression largely invisible.
The RCA case exposes a category of martial law violence that does not fit this template. It was not a single act of state terror but a slow, cumulative, corporate-statist co-production. The state security apparatus enabled RCA to expose workers to toxic chemicals for decades. The specific underlying mechanisms include Garrison Command secret police quietly embedded in factories, corporatist unions that neutralised worker voice, and administrative mediation channels that shielded corporate liability from public scrutiny. But the women who absorbed chlorinated solvents on RCA’s production lines were not political dissidents. They were workers whose capacity to identify, contest, or even understand occupational hazards was systematically foreclosed by the same authoritarian infrastructure that imprisoned democracy activists. Their cancers and chronic illnesses represent an embodied archive of martial law governance that conventional transitional justice categories struggle to recognise. The framework’s categorical design reproduces the epistemic exclusion Spivak identifies (Spivak 1988). The vocabulary of recognition structurally cannot accommodate cumulative, corporate-mediated, and embodied harm.
Compounding this analytical challenge, many police and military records concerning factory-level surveillance, labour informant networks, and corporate-security coordination remain classified or were destroyed during Taiwan’s political transition (Su 2025). Despite lifting martial law in 1987, the KMT only allowed direct presidential elections in 1996, giving them ample time to destroy evidence of their more insidious and less visible abuses (總統府 2026). The absence of these records is not simply an evidentiary gap, but it is itself a product of the regime’s logic of concealment. Reconstructing how the secret police operated within export electronics factories thus requires archival triangulation. This paper thus combines surviving field notes seized and stamped by the Garrison Command, court documents produced decades later through litigation, oral histories from former workers and researchers, and corporate materials that inadvertently document workplace conditions. Such fragmentary reconstruction is methodologically imperfect, but it is the only available means to render visible a form of authoritarian harm that was designed to leave no trace.
Contemporary political dynamics further constrain accountability. The KMT, which administered the martial law apparatus documented in this paper, controls Taiwan’s Legislative Yuan following its 2024 electoral gains (Dotson and Levine 2024). That institutional continuity helps explain why Taiwan’s transitional justice process has not extended to labour dimensions of authoritarian governance and why archival disclosure concerning factory-level surveillance remains limited. Cases like RCA’s, where the state co-produced with multinational capital an enduring regime of toxic exposure and labour exploitation, demand an expanded conception of transitional justice.
For historians, the RCA litigation is not merely an epilogue. It reveals the hidden premises of earlier developmental narratives. When corporate statist structures disciplined workers through dormitory enclosure and constrained voice, they could neither identify nor contest toxic risks in real time. Toxicity was not a byproduct but a structured risk externalisation, which the labour regime’s compliance architecture helped conceal. Decades later, coalitions of workers, activists, and experts carried that struggle into courts, epidemiological studies, and public campaigns, redefining the terms of contestation. Even today, under the auspices of the Taoyuan County Original RCA Corporation Employees Caring Association, former RCA factory women are fighting to clean RCA’s contaminated Taoyuan factory site and turn it into a workers’ rights museum (Former RCA Worker 2026).
RCA’s role, then, cannot be reduced to either “technology transfer” or “exploitation.” It was both an institutional partner in upgrading narratives and an exemplar of how export electronics production could be organised under authoritarian governance through feminised labour, limited voice, and risk externalisation. These two dimensions operated simultaneously, not sequentially. Separating them, as existing historiography has done, reproduces analytically the very concealment that the martial law regime enforced politically.
Modern Continuities: The Labour Regime Beyond Martial Law
If the labour regime described above was only an effect of authoritarian rule, democratisation should have dismantled it. Martial law ended in 1987, independent unions became legal, and direct presidential elections followed in 1996. Yet the regime’s logic survived in two directions: within Taiwan, where Southeast Asian migrant labourers increasingly filled assembly, testing, and packaging (ATP) roles once occupied by young Taiwanese women, and across the broader semiconductor geography, where value-chain pressures reproduced comparable labour outcomes without direct martial law. RCA was not an isolated episode. It was an early template for offloading risk onto workers with the least political leverage.
From Factory Women to Migrant Workers
The workers who now fill many of these roles come primarily from the Philippines, Vietnam, Thailand, and Indonesia (勞動部 2026), though many more remain undocumented (Beltran 2025). Interviews from 2024 show the persistence of long shifts, dormitory dependency, broker control, and constrained mobility. A Filipino migrant worker at a TSMC supplier in Hsinchu described a twelve-hour evening-shift rotation, six days per week, employer-provided dormitory housing, and monthly pay of about NT$35,000 (Migrant Worker 2024a). Her broker controlled her bank account, deducted wages, confiscated her passport, and prevented her from switching jobs (僑務委員會 2026). Another migrant worker reported comparable hours and continued dormitory-fee deductions even after moving into external housing (Migrant Worker 2024b; Niece 2024). Brokerage endures because it transfers recruitment, documentation, housing, and dispute-management costs from employers onto workers (Transparentem 2025). Democratisation did not abolish coercive labour discipline. It marketised it.
The COVID-19 pandemic exposed how dormitory governance could intensify into explicit confinement. During Taiwan’s 2021 outbreak, the KMT-controlled Miaoli County restricted migrant workers to travel only between dormitories and factories while others moved freely (Templeman 2018; Davidson 2021). Migrant worker advocates described the result as workers being “locked up” in dormitories (Van Hung 2024). When workers objected, Miaoli’s KMT magistrate asked, “You tested positive, and even died because of the virus. Why talk about human rights now?” (Chien and Qin 2021). The party that constructed Taiwan’s corporatist labour and EPZ governance apparatus still administers the inspection, mediation, and dormitory oversight machinery through which migrant workers must seek recourse (臺北市政府全球資訊網 2025).
Gendered bodily discipline also persists. At Compal Electronics, a major Apple and Google contractor, thousands of Vietnamese workers protested extreme overwork, illegal pay deductions, and inhumane rules (吳睿騏 2020). One rule reportedly fined pregnant workers US$6,000-7,000, roughly a year’s wages (Van Hung 2024; Electronics Watch 2024). Where martial-law managers justified dormitory surveillance as protection of women’s “virtue,” contemporary subcontractors directly penalise reproduction itself. In both cases, the female body is governed as productive capacity whose pregnancy, illness, or fatigue becomes a violation of the labour regime’s terms.
Migrant workers also lack the democratic leverage that might force enforcement. Non-citizen workers cannot vote or engage in election-related work in Taiwan (Everington 2023). Although they formally possess labour rights, broker-mediated employment, employer-tied visas, language barriers, and deportation risk constrain collective action (A. R. Pan 2023; Tseng 2025). Official discourse can also naturalise this instrumentalisation. In 2024, Labour Minister Hsu Ming-chun justified plans to recruit workers from northeast India by noting their “similar skin color and diets,” Christianity, and aptitude for manufacturing (Madhok and Chang 2024). As Nakamura shows in her study of Fairchild’s recruitment of Navajo women, firms and states routinely recast whichever vulnerable population is cheapest and most captive as naturally suited to electronic assembly (Nakamura 2014).
The New Southbound Policy (NSP) extends this brokerage function into foreign policy. Under industry-academia programmes, Southeast Asian students have been recruited to Taiwanese universities and then channelled into electronics and semiconductor factory shifts, while brokers extract fees and schools depend on student-worker revenue (H. Liu 2025). The 2024 U.S. Trafficking in Persons Report found that traffickers exploited relaxed NSP visa rules to subject Southeast Asian students to forced labour (AIT 2024), and Taiwan’s National Human Rights Commission concluded that weak oversight enabled disguised labour importation (高華謙 2025). The KMT once recruited rural Taiwanese women through promises of wages and mobility. The NSP recruits foreign students through promises of education. In both cases, intermediaries convert aspiration into disciplined industrial labour.
Semiconductor interests often outweigh these workers’ claims. Taiwan’s chip industry has acquired “sacred mountain” national security status (Hsiao 2025), generated more than US$165 billion in 2024 revenue, and accounted for 20.7% of GDP and 60% of exports (USITA 2025; G. Chen 2025). Migrant workers report that local labour authorities prioritise operational continuity over complaints about hours, dormitories, and brokers (Niece 2024). The justification has shifted from anti-communist stability to national security. For workers, the effect is similar.
Value Chains as Labour Discipline
The same logic has travelled beyond Taiwan through semiconductor value chains. Buyer-driven production networks allow lead firms to govern suppliers through standards, pricing pressure, and the credible threat of relocation rather than ownership (Gereffi 2018). When a few dominant buyers command advanced chip orders, specialised suppliers face quasi-monopsony. Margin pressure then flows downward into wages, hours, safety investment, and recruitment practices. No actor needs to consciously reproduce RCA-era labour control. The value chain only needs to reward the same cost-minimising practices that authoritarian institutions once enforced directly.
South Korea illustrates this continuity within a democratised political economy. Samsung and SK Hynix dominate global memory production through subcontracting regimes that mirror Taiwan’s supplier hierarchies (Y. Lee and Cha 2026). Korean chaebol firms historically suppressed independent labour organising with state support, and Samsung did not permit an enterprise union until 2019 (Seol 2024), after activists had exposed hundreds of cancers, chronic illnesses, and deaths among semiconductor workers (SHARPS 2018). Even reforms such as Korea’s 52-hour workweek cap and new subcontractor-bargaining legislation have not undone the underlying incentives (Jang et al. 2025). Smaller suppliers employing undocumented workers from China and Southeast Asia use piecework frameworks that encourage excessive hours and circumvent legal limits (Chun 2025). Lead firms need not suppress unions directly when pricing power ensures that the subcontractors employing the most vulnerable workers cannot easily tolerate them.
The Philippines reveals a spatial and administrative version of the same mechanism. Philippine export zones, modelled partly on Taiwan’s Kaohsiung EPZ, concentrate electronics ATP production under the Philippine Economic Zone Authority (UNIDO Vietnam 2015; PEZA 2020). Despite constitutional labour protections, union density in electronics remains around six percent (PSA 2016). When workers at one Metro Manila ATP plant organised, the subcontractor closed the facility and relocated nearby with a non-union workforce (Philippine Supreme Court 1999). Electronics organisers also face police visits, red-tagging, and violence, including the murder of organiser Dandy Miguel near a Nexperia ATP plant in Laguna (CTUHR 2022; Nexperia Union 2025; Human Rights Watch 2024; Stephanie 2023). Contractualisation fragments solidarity further: seven in ten Philippine electronics workers are temporary agency hires despite years at the same plant (Oliquino 2016). A 2017 factory fire prompted passage of the Occupational Safety and Health Standards Act, but inspections remain rare and fines weak (Salamat 2017; EILER 2018; CTUHR/IOHSAD 2025).
These cases do not replicate Taiwan’s martial law regime. They show how its labour outcomes can be reproduced through different institutional routes. PEZA’s investment-first mandate, subcontractor flight after unionisation, Korean supplier piecework, and Taiwan’s broker-mediated migrant labour all serve a similar structural function: they reduce the capacity of workers at the bottom of the value chain to contest risk. The foundational architecture of semiconductor profitability, first consolidated in Taiwan under authoritarian conditions, remains durable because it has been embedded in markets, contracts, and production networks.
Counter-Archives and Transnational Solidarity
These shared patterns have also generated transnational counter-movements. Taiwan’s TAVOI, which helped organise the RCA litigation, built ties with South Korea’s Supporters for Health and Rights of People in the Semiconductor Industry (SHARPS) (Former RCA Worker 2024, 1; Wang 2025). SHARPS emerged after Hwang Yu-mi’s 2007 leukaemia death at a Samsung fabrication plant and campaigned for more than a decade over occupational cancers and fatalities (SHARPS 2025). In the Philippines, the Institute for Occupational Health and Safety Development (IOHSAD), the Metal Workers Alliance of the Philippines, and the GoodElectronics Network have documented chemical exposure and built worker-led safety committees (CTUHR/IOHSAD 2025).
GoodElectronics has become the main transnational platform linking these struggles. In 2025, it moved its global secretariat to a Philippine-based coalition, shifting the network’s centre of gravity toward the Global South (GoodElectronics 2025a). That year, GoodElectronics, IOHSAD, and the Asian Network for the Rights of Occupational and Environmental Victims convened a Manila session where organisers from Korea, Taiwan, Thailand, Indonesia, and the Philippines exchanged testimony on chemical poisoning and strategised around ILO Convention 170 ratification (GoodElectronics 2025b).
Artistic work has extended these counter-archives beyond labour circles. At the Manila session, organisers screened Eunhee Lee’s Colorless, Odorless (2024), which combines leaked clean-room footage, archives, and testimonies from Korean and Taiwanese semiconductor workers to make invisible hazards visible (GoodElectronics 2025b; E. Lee 2024). The film has since circulated through institutions in Barcelona, Taipei, Copenhagen, Kuala Lumpur, and Beijing.
Such coalitional and cultural work embodies TAVOI’s insistence that “our bodies are our only evidence” (GoodElectronics 2025b; Ricaurte 2022). These networks do more than protest. They build alternative archives, scientific literacies, and public narratives that challenge the industry’s image as clean, efficient, and progressive. RCA thus leaves a double legacy: it helped legitimate Taiwan’s semiconductor upgrading, but it also anchors a counter-history of industrial hazard and labour exploitation that continues across East and Southeast Asian supply chains.
Conclusion
This paper has argued that RCA and the KMT developmental state co-produced Taiwan’s electronics labour regime, and that this regime was not incidental to Taiwan’s semiconductor ascent but constitutive of it. The same institutional preconditions that enabled technology transfer also enabled labour exploitation. The canonical story of the 1976 RCA Project as a triumph of technocrats and engineers obscures the gendered, coerced, and chemically poisoned workforce upon which that project depended.
The implications extend beyond historiography. More than US$500 billion in semiconductor subsidies announced since 2020 rest on the premise that chip production serves the national interest (Ravi 2022). Yet national-interest framings repeatedly subordinate worker welfare to production targets. Public subsidy programmes still do not sufficiently condition funding on enforceable occupational safety standards or organising protections, and chemical exposure rules cover only a fraction of substances used in fabrication (Rodrigo 2024). When governments invoke “national security” to shield chip firms from scrutiny, they echo the KMT’s earlier use of anti-communist stability to suppress labour contestation (OECD 2025).
For Taiwan, this history also raises unresolved transitional justice questions. The official framework has focused primarily on political persecution, leaving state-sanctioned labour exploitation at the margins. Recognising the poisoned women on RCA’s lines as victims of authoritarian governance would not diminish political persecution cases. It would reveal how martial law operated not only through spectacular repression, but also through the ordinary reproduction of factory discipline that made export growth profitable.
Finally, this paper challenges how societies narrate technological achievement. Taiwan’s semiconductor achievement and its labour costs are not competing narratives. They are one integrated history. The engineers who built TSMC and the women who assembled RCA’s transistors inhabited the same political economy. Acknowledging the latter does not diminish the former. It shows what technological achievement required, and at whose expense.
That acknowledgement is urgent because the evidence is disappearing. Former RCA workers are now in their sixties and seventies; some died during litigation. Researchers’ seized notes remain largely unpublished, surveillance records remain classified or destroyed, and migrant worker organisers document contemporary reproduction without secure preservation. Counter-archives built by TAVOI, SHARPS, GoodElectronics, and allied organisers show that workers already know what dominant narratives obscure. The task is to ensure that their evidence is not, once again, seized, stamped, and filed away.
Acknowledgements
The author thanks the former RCA workers, labour organisers, legal advocates, scholars, archivists, and mentors who made this research possible.
Funding
This research received support from the Yale Brady-Johnson Grand Strategy Program, the Yale Blue Center for Global Strategic Assessment, the Yale Council on East Asian Studies, the Yale Council on Southeast Asian Studies, the Yale Jackson School of Global Affairs, the Yale History Department, Grace Hopper College, the Yale Institution for Social and Policy Studies, and the Obama Foundation.
CRediT author statement
Ethan Chiu: Conceptualisation, data curation, formal analysis, investigation, methodology, writing – original draft, and writing – review and editing.
Declaration of interest
The author reports no competing interests to declare.
Ethics and consent statement
This research involving human participants was reviewed by Yale University’s Institutional Review Board and determined exempt under IRB Protocol No. 2000037477. Interview participants were adults who voluntarily agreed to participate after receiving study information about purpose, risks, confidentiality protections, and the right to withdraw. Participants gave informed consent before interviews. Interview material is anonymised or described by role where identification could compromise privacy, confidentiality, or safety. No directly identifying interview material is published without consent.
Data availability statement
Most interview data supporting this article are not publicly available because disclosure could compromise participant privacy and confidentiality. Publicly available sources are cited in the references.
References
Featured/Headline Image Caption and Citation: Havana, Photo by Pedro Szekely | Image sourced from Flickr | CC License, no changes made
Braverman, Harry. 1974. Labor and Monopoly Capital. Monthly Review Press.
Gereffi, Gary. 2018. Global Value Chains and Development. Cambridge University Press.
Hsiao, Chu-Chen (CNEX Studio Corporation). 2025. “A Chip Odyssey.” https://www.cnex.tw/achipodyssey.
Kung, Lydia. 1995. Factory Women in Taiwan. Columbia University Press.
Miller, Chris. 2022. Chip War: The Fight for the World’s Most Critical Technology. Scribner.
Ravi, Sarah. 2022. Global Semiconductor Incentives. Semiconductor Industry Association.
Wade, Robert. 2003. Governing the Market. Princeton University Press.
國立歷史博物館. 2020. “飛歌電子廠爆發女工中毒死亡事件.” 臺灣女人, October 28. https://women.nmth.gov.tw/?p=19999.
王諺玫 (PeoPo). 2009. “「她們的故事」 紀錄女工歷史.” January 23. https://www.peopo.org/news/28101.
Author
You must be logged in to post a comment.