摘要：美國關於貿易逆差和關稅的爭論已經達到白熱化程度。一方面，白宮認為美國貿易逆差是其他國家不公平競爭的表現，並承諾透過積極的關稅政策來復興製造業並提供高薪就業機會。另一方面，批評人士聲稱這些關稅只會導致價格上漲。在這場激烈的爭論中，人們常常忽略的是，關稅不僅影響製
21世紀初，來自中國的進口競爭減少了美國製造業的就業崗位，但創造了更多的服務業就業機會。
2025 年 4 月 30 日 • 經濟政策研究簡報第 431 期
作者：Nicholas Bloom、Kyle Handley、André Kurmann 和 Philip A. Luck
分享
美國關於貿易逆差和關稅的爭論已經達到白熱化程度。一方面，白宮認為美國貿易逆差是其他國家不公平競爭的表現，並承諾透過積極的關稅政策來復興製造業並提供高薪就業機會。另一方面，批評人士聲稱這些關稅只會導致價格上漲。在這場激烈的爭論中，人們常常忽略的是，關稅不僅影響製造業，也影響服務業的就業——大多數美國人都從事這些行業。
為了理解這些影響，有必要回顧近代史上最重大的貿易動盪：中國在21世紀初崛起為製造業強國。這種所謂的「中國衝擊」被廣泛認為是美國製造業就業下降的重要原因。然而，在整個21世紀，新興和擴張型企業在製造業創造的就業機會仍然相當可觀，而且由於美國企業和消費者受益於中國更廉價的製成品，就業大幅轉向服務業。
我們的研究調查了製造業和服務業就業成長的相反趨勢之間的關聯程度。我們的研究結果表明，受中國進口競爭影響較大的當地勞動力市場製造業就業機會流失較多。但這些損失被更強勁的服務業就業成長所抵消，服務業就業成長主要來自公司內部的就業重新分配。重要的是，這種重新分配的程度因地區而異。受過大學教育的工人比例較高的地區（包括西海岸大部分地區和大城市）實現了成功轉型，服務業就業的成長速度超過了製造業就業機會的減少速度。受過大學教育的勞動力比例較低且製造業依賴程度較高的地區（包括中西部和南部的大部分地區）的服務業增長有限，難以彌補製造業就業機會的流失。
我們的研究結果表明，中國衝擊總體上創造了就業機會，而不是扼殺了就業機會。同時，這場衝擊不僅在工人之間，而且在各個地區都產生了贏家和輸家，因為工作從工業中心地帶轉移到了沿海地區和大城市，從而導致了美國就業分佈格局的變化。這一經歷為今天的關稅辯論提供了重要的教訓：儘管貿易壁壘可能會帶來一些製造業崗位，但它們不僅會提高價格，還可能破壞全球貿易所促進的高薪服務業就業的大幅增長，特別是在高教育和高生產力地區。換句話說，白宮倡導的激進關稅政策可能會創造一些贏家，但總體上可能會讓美國遭受損失。
我們的分析是基於美國人口普查局縱向商業資料庫中所有美國公司及其機構的詳細管理資料。該資料庫使我們能夠詳細說明當地就業崗位如何從製造業轉向服務業。我們的研究結果表明，企業透過關閉製造工廠、擴大服務業就業（儘管並不總是在同一個當地勞動力市場）或將工廠的主要活動從製造轉向服務業，例如管理、研發和批發。從製造業轉向服務業的企業佔了1997年至2007年「中國衝擊」對製造業就業造成的整體負面影響的40%。這意味著，在更匯總的公共數據中出現的製造業就業機會流失，可以部分歸因於現在被算作服務業而非製造業就業的就業機會。轉換產業的機構往往會減少就業機會，但不會摧毀所有的工作。
中國衝擊對就業的影響與整體趨勢不同。在對美國總體數據中創造和破壞就業職位的來源進行分析後，我們的研究結果顯示，1997年至2007年間製造業就業崗位總體下降的三分之二是由於企業因各種原因（無論是否與中國有關）而倒閉。相較之下，企業倒閉僅佔中國衝擊造成的製造業就業機會流失的25%，其餘大部分則是由於企業規模縮小以及工廠從製造業轉向服務業。 1997 年至 2007 年間，創造服務業就業機會的現有公司幾乎佔了服務業整體就業成長的全部。但在中國衝擊造成的服務業就業成長中，新成立的公司和機構提供的就業機會成長僅佔 25%，其餘部分來自現有公司。
我們的研究結果提供了更多證據，顯示中國衝擊並不是 21 世紀製造業工廠關閉增多的唯一因素。相反，研究結果表明，製造業就業流失和服務業就業機會成長並不符合常見敘述中的一些更廣泛的模式和軼事。此外，研究結果表明，中國衝擊促進了新服務機構的進入，這與不使用工廠、專注於行銷、研究、設計和物流的商品生產商崛起的證據一致。
我們的分析還表明，中國進口競爭對當地勞動力市場的影響因雇主特徵而異。擁有製造工廠的企業中服務業崗位的增加是製造業崗位減少的主要原因。這顯示服務業就業機會的成長來自於設計、製造、服務或行銷國外更便宜產品等多種因素的結合。相較之下，服務業就業的成長中只有約四分之一來自減少製造業就業機會的公司旗下的機構。此外，服務業就業機會成長的約四分之三來自擁有 1,000 多名員工的公司旗下的機構，這表明大部分就業成長來自較大的跨國公司。服務業就業的增加幾乎全部來自於支付業內高薪的機構。
最後，為因應中國進口競爭，企業內部從製造業向服務業的重新分配主要發生在西海岸和大城市，這些地區受過大學教育的工人比例較高。從製造業轉向貿易支援服務業的機構幾乎全部集中在這些高等教育地區，佔中國衝擊對製造業就業負面影響的一半。其餘則由於公司關閉製造工廠並擴建或開設新的服務機構。由此產生的服務業就業機會的增加超過了製造業就業機會的減少，從而增加了總就業人數。相較之下，在中西部和南部地區（這些地區受過大學教育的工人比例較低且最初對製造業的依賴程度較高），中國衝擊導致的製造業就業崗位流失主要是由於企業倒閉以及工廠在經歷失業後僅小幅擴大服務業崗位。總而言之，該國受教育程度較高的地區的公司減少了製造業的就業崗位，但管理、設計和行銷就業的成長彌補了這一不足。在該國受教育程度較低的地區，非製造業就業的抵銷性成長則少得多。
最終，中國衝擊導緻美國製造業就業機會大幅下降，加劇了製造業就業機會流失而服務業就業成長的趨勢。但服務業和製造業的對比代表著同一枚硬幣的兩面。受中國衝擊影響，失業率較高的部分原因是那些繼續營運但將主要業務從製造業轉向服務業的企業。這顯示一些製造業職位並沒有消失，而是最終變成了服務業職位。這一因素，加上其他新舊服務公司創造的就業機會，意味著該國部分地區的總就業人數增加。最後，中國衝擊帶來的服務業就業成長在各地區分佈並不均勻。受過大學教育的人口比例較小且更依賴製造業的地區並沒有像全國其他地區一樣在服務業方面獲得抵銷性成長。區域間對貿易衝擊的適應差異顯然是一個值得我們關注的重要議題。然而，目前尚不清楚關稅將在多大程度上恢復這些地區的製造業就業崗位，而且這些關稅可能會以價格上漲和整體就業成長下降的形式對全國其他地區造成高昂的成本。
註：
本研究摘要基於 Nicholas Bloom 等人的《重新審視中國衝擊：美國勞動市場的工作重新分配與產業轉換》，美國國家經濟研究局工作論文第 14 號。 33098，2024 年 11 月。
英文原文：
The China Shock Revisited: Job Reallocation and Industry Switching in US Labor Markets
Chinese import competition during the 2000s reduced US manufacturing jobs but created even more service jobs.
APRIL 30, 2025 • RESEARCH BRIEFS IN ECONOMIC POLICY NO. 431
By Nicholas Bloom, Kyle Handley, André Kurmann, & Philip A. Luck
SHARE
America’s debate about trade deficits and tariffs has reached a fever pitch. On the one hand, the White House argues that US trade deficits are a sign of unfair competition by other countries and promises a manufacturing renaissance with well-paying jobs through aggressive tariff policies. On the other hand, critics claim that these tariffs will simply lead to higher prices. What’s often overlooked in this heated exchange is that tariffs affect not just manufacturing but also service-sector jobs—where most Americans work.
To understand these effects, it is useful to revisit the most significant trade disruption in recent history: China’s emergence as a manufacturing powerhouse during the 2000s. This so-called China Shock is widely regarded as an important reason for the decline in US manufacturing employment. Yet job creation in manufacturing from new and expanding firms remained substantial throughout the 2000s, and employment shifted significantly toward services as American firms and consumers benefited from access to cheaper manufactured goods from China.
Our research investigates the extent to which the opposing trends in manufacturing and services job growth are related. Our findings reveal that local labor markets more exposed to Chinese import competition experienced larger manufacturing job losses. But these losses were offset by stronger services job growth, which mostly came from job reallocation within firms. Importantly, the extent of this reallocation varied across regions. Places with a high share of college-educated workers—including much of the West Coast and large cities—saw successful transitions, with service job gains outpacing manufacturing job losses. Places with a low share of college-educated workers and high manufacturing dependence—including much of the Midwest and the South—experienced only limited services growth to compensate for manufacturing job losses.
Our findings imply that the China Shock was, on net, a job creator and not a job killer. At the same time, the shock created winners and losers, not just across workers but also across regions, by relocating jobs from the industrial heartland to the coasts and large cities, thereby contributing to the changing geography of jobs in the United States. This experience offers a crucial lesson about today’s debate on tariffs: While trade barriers might bring back some manufacturing jobs, they may not only raise prices but also risk undermining the substantial growth in high-paying service-sector jobs that global trade has fostered, especially in high-education and high-productivity regions. In other words, the aggressive tariff policy advocated by the White House may create a few winners but will likely make America lose overall.
Our analysis is based on detailed administrative data on all US firms and their establishments from the Longitudinal Business Database of the US Census Bureau. This database allows us to provide a detailed account of how the local reallocation of jobs away from manufacturing to services occurred. Our findings show that firms shifted from manufacturing toward services by closing manufacturing plants and expanding employment in services (though not always in the same local labor market) or by switching their plants’ primary activity from manufacturing to the provision of trade-related services, such as management, research and development, and wholesale. Establishments switching from manufacturing to services accounted for 40 percent of the total negative effect of the China Shock on manufacturing employment from 1997 to 2007. This implies that what appears to be manufacturing job loss in more aggregated, public-use data can be attributed partly to jobs now counted as service-sector rather than manufacturing employment. Establishments that switched industries tended to reduce employment, but they did not destroy all jobs.
The effects of the China Shock on jobs differ from aggregate trends. After breaking down the sources of job creation and destruction in aggregate US data, our findings reveal that about two-thirds of the overall decline in manufacturing employment from 1997 to 2007 came from firms closing for any reason, China-related or otherwise. In contrast, firm closures accounted for only 25 percent of manufacturing job losses from the China Shock—the rest were mostly due to firms downsizing and plants switching from manufacturing to services. Existing firms that created service jobs account for nearly all the overall job growth in services from 1997 to 2007. But job growth from newly created firms and establishments accounts for 25 percent of the growth in service jobs attributable to the China Shock—the rest came from existing firms.
Our findings provide more evidence that the China Shock was not the only driver of increased manufacturing plant closures during the 2000s. Rather, the findings suggest that manufacturing job losses and service-sector gains do not conform to some of the broader patterns and anecdotes in common narratives. Additionally, the findings suggest that the China Shock promoted the entry of new service establishments, which is consistent with evidence of the rise of goods producers that do not use factories and that focus on marketing, research, design, and logistics.
Our analysis also reveals that the local labor market effects of Chinese import competition varied substantially by employer characteristics. The increase in service jobs at firms that own manufacturing plants accounts for most of the decline in manufacturing jobs. This suggests these gains in service-sector jobs were from some combination of designing, manufacturing, servicing, or perhaps marketing products made more cheaply abroad. In contrast, only about a quarter of the gains in service jobs came from establishments belonging to firms that reduced employment in manufacturing. Furthermore, about three-quarters of the gains in service jobs came from establishments belonging to firms with more than 1,000 employees, suggesting much of this job growth arose in larger multinational firms. Nearly all the increase in service jobs came from establishments that pay high wages relative to their industry.
Finally, reallocation within firms from manufacturing to service jobs in response to Chinese import competition occurred primarily on the West Coast and in large cities, areas with a high share of college-educated workers. Establishments switching industries from manufacturing to trade-supporting services are almost entirely confined to these high-education areas, accounting for half the negative effect of the China Shock on manufacturing jobs. The rest was due to firms shutting down manufacturing plants and expanding or opening new service establishments. The resulting increase in service jobs more than made up for the decrease in manufacturing jobs, thus increasing total employment. By contrast, in the Midwest and the South—areas with a lower share of college-educated workers and high initial manufacturing dependence—manufacturing job losses due to the China Shock were primarily driven by firms closing and plants only modestly expanding service jobs after experiencing job losses. In summary, firms in higher-educated parts of the country reduced manufacturing employment but more than made up for it with growing management, design, and marketing employment. In lower-educated parts of the country, there was far less offsetting growth in nonmanufacturing employment.
Ultimately, the China Shock led to a significant decline in manufacturing jobs in the United States, amplifying trends in manufacturing job losses while the service sector made gains. But the contrast between service and manufacturing represents two sides of the same coin. Part of the higher rate of job losses attributable to the China Shock was from establishments that continued operations but switched their primary activity from manufacturing to services. This suggests that some manufacturing jobs did not disappear but eventually became service jobs. This factor, combined with job creation at other new and existing service firms, meant that total employment increased in some parts of the country. Lastly, the gains in service-sector employment growth from the China Shock were not evenly distributed across regions. Regions with a small share of the college-educated population and more dependent on manufacturing did not experience offsetting gains in services like the rest of the country. This regional disparity in adaptation to trade shocks is clearly an important issue that deserves our attention. However, it is unclear to what extent tariffs will bring back manufacturing jobs in these regions, and these tariffs may exert a high cost on the rest of the country in the form of higher prices and lower overall job growth.
Note
This research brief is based on Nicholas Bloom et al., “The China Shock Revisited: Job Reallocation and Industry Switching in US Labor Markets,” National Bureau of Economic Research Working Paper no. 33098, November 2024.
来源：非常道