In recent years, with the explosion of artificial intelligence, a long-dormant concept in economic history has become popular again—the 'Engels' Pause'.
Frey revisited it in "The Technology Trap", Acemoglu and Johnson mentioned it again in "Power and Progress", and many economists have since worried: AI might make history repeat itself, with wealth created by machines pouring in while labor compensation stagnates.
But frankly speaking, most of the current discussions about the 'new Engels' Pause' remain at the distribution level: the pie has gotten bigger, but it hasn't been shared well. This understanding is not deep enough. The essence of the Engels' Pause is not distributional imbalance, but a deeper pathology—the economy is growing fat while society is atrophying. In medical terms, it's 'social sarcopenic obesity'.
And more alarming than the pathology is the misdiagnosis: if it's seen merely as a distribution problem, the prescription will only be redistribution, which is like a blood transfusion; the real prescription is to make society regrow and use its muscles—economic weight loss, social muscle gain.
The Engels' Pause: An Era of Sixty Years of Stagnant Wages
The 'Engels' Pause' was named by economic historian Robert Allen, after Friedrich Engels' 1845 publication "The Condition of the Working Class in England". It refers to a period in the early British Industrial Revolution: from 1780 to 1840, British output per worker grew by about 46%, while real wages only rose by 12%; the profit rate increased from about 10% at the end of the 18th century to over 20% by the mid-19th century, and the share of profits in national income expanded from about one-fifth to over two-fifths. The gains from technological progress were almost entirely absorbed by capital, with workers sharing little of the wealth they created. It wasn't until after the 1840s that the situation reversed: by the mid-19th century, when the pause ended, the growth in real wages over the next half-century actually surpassed the growth in output.
This history is being brought up again today because people see a similar picture: algorithms and automation are creating enormous output and profits, labor's share of income has been declining globally over the long term, and traditional labor relations and the 'stability' of some white-collar jobs are being challenged. China's situation is different—labor's share of GDP rebounded after hitting a low around 2007—but the foundation of this rebound is not solid, and the social changes accompanying the AI investment boom are testing it. In other words, for China, the Engels' Pause is more of a 'future tense' than a 'present tense', a risk that is accumulating.
It's worth noting that China is not a mere bystander in this discussion. 'Synchronizing the growth of residents' income with economic growth, and synchronizing the increase in labor remuneration with the improvement of labor productivity'—proposed at the 18th Party Congress and written again into the main goals of the 15th Five-Year Plan—these 'two synchronizations' are themselves an institutional negation of the Engels' Pause. In the language of economic history, the Engels' Pause is precisely 'two desynchronizations'. The problem, therefore, is not whether to make the commitment, but what will turn the commitment into reality. To answer that, we must first diagnose the illness correctly.
Essence: Not a Problem of Dividing the Pie, But a Broken Driveshaft
The popular explanation frames the Engels' Pause as a distribution problem. This framing is not wrong, but it doesn't touch the essence. Its essence is: industrial development has become disconnected from economic and social development—the engine of productive forces is spinning faster and faster, but the driveshaft of social relations is broken, and the power cannot be transmitted to every corner of the vehicle's body. To understand this disconnection, there are two keys.
The first key is the 'fallacy of composition': micro-rationality synthesizes into macro-irrationality. Each enterprise reducing labor compensation, replacing labor with machines, and outsourcing employment to platforms is a rational decision when viewed individually; but when all enterprises do this simultaneously, the composition results in a systemic collapse of labor's share: consumption cannot absorb output, insufficient domestic demand, leading to simultaneous overcapacity and involution. This is what Keynes called the 'paradox of thrift', and the reverse of Kalecki's famous saying—'workers spend what they get, capitalists get what they spend'—when workers don't get it, capitalists eventually can't spend it either. A century ago, Ford raising the daily wage to five dollars is often interpreted as 'if workers can't afford cars, the auto factories have no future'—whether that was Ford's original intention or not, the logic is correct.
The second key is the 'aggregate illusion': aggregation conceals composition. The growth of GDP is real in aggregate terms, but its welfare implications are questionable: aggregate indicators conceal the atrophy of constituent elements. A large part of the growth is contributed by sectors favored by capital—estimates show that AI and data center-related investments once contributed over 90% of a certain period's economic growth in the US—while the families, communities, labor, and trust that constitute the social fabric are thinning in compositional terms. The more dazzling the growth, the more likely it is a case of 'failed muscle gain': weight is increasing, but muscle isn't growing.
There's another point that differs from the old pause. To explain it, we must first look at how the old pause ended. Allen's explanation is this: technological progress accelerated after 1800, but new technologies, to become output, had to be housed in new factories, machines, and railways, and the capital stock in Britain at the time couldn't keep up with the scale required by the new technologies—capital was insufficient. Capital was scarce, so returns were high, and profit rates doubled; labor wasn't scarce, so wages stagnated. After decades, the capital stock finally accumulated to a scale matching the new technology. At that point, new capital, to start operating, had to hire people, demand for labor increased, and wages finally began to rise with productivity. This is a 'self-healing' chain: high profits induce reinvestment, reinvestment expands capital stock, expanding capital stock raises labor demand, and higher labor demand raises wages. For this to hold, there were two implicit prerequisites: first, profits must be reinvested in production; second, the new capital must require labor to complement it—19th-century machines complemented labor; the more machines, the more workers needed.
The new round of pause has problems on both these prerequisites. First, what's eating into wages is increasingly not profits but 'rents'. Profits are temporary excess income in competition, which must be constantly reinvested and production expanded to be maintained, and expansion requires hiring; 'rents', however, are sustained income from the closure of originally open market relationships—some call it 'cloud rent'—self-sustained by network effects and data barriers, not requiring reinvestment to continue, nor eroded by competitors' entry. Rents can be used long-term for stock buybacks, dividend payouts, mergers and acquisitions of peers, and purchasing various assets—this money isn't idle, asset prices rise, paper wealth increases, but the appreciation comes from the repricing of existing assets, not from new factories or new jobs. The self-healing chain breaks at the first link. Second, even if capital is reinvested, today's capital may not necessarily need people. The reason capital accumulation in the 19th century could raise wages was because machines needed workers to operate them; capital deepening in the AI era—computing power, models, data centers—aims precisely at replacing labor; the more capital, the fewer people may be needed. The self-healing chain also breaks at the third link.
This is why we cannot expect the new pause to end naturally like the old one, by 'capital accumulation catching up with technological progress'. The end of the old pause was the result of the joint action of the economy's own forces and social capacity building; if the new pause is to end, the force from the social side must bear a greater share. This is also why this article focuses on society rather than distribution.
Social Productivity is a Function of Social Vitality
To truly understand this pathology, a more fundamental theoretical perspective is needed: the relationship between social vitality and social productivity.
From the 14th Party Congress report in 1992, which first juxtaposed 'further liberating and developing social productive forces' with 'stimulating and enhancing social vitality', to the 'three liberations' at the Third Plenary Session of the 18th Central Committee in 2013, and to the reiteration at the Third Plenary Session of the 20th Central Committee in 2024 and the 'sustained enhancement of development momentum and social vitality' in the 15th Five-Year Plan, the 'two forces' have been mentioned together for over thirty years. Why juxtapose them? My understanding is: social productive forces have never been a simple function of factors like capital, technology, and labor; they are a function of social vitality. Capital depreciates, technology becomes obsolete, but only a society's willingness to work, impulse to create, resolve to procreate, and habits of mutual aid are the inexhaustible source of productivity. Social vitality is not just the driveshaft connecting the engine to the vehicle's body; it is itself the second engine.
This is not metaphysics; there is evidence. Saxenian compared Silicon Valley and Boston's Route 128: the two places were comparable in technology and capital, with Route 128 initially having an advantage relying on MIT and military contracts, yet Silicon Valley ultimately prevailed. The difference wasn't in technology, but in society—open labor markets, weakened corporate boundaries, dense informal knowledge exchange, high tolerance for failure, professional identity rooted in the community rather than individual firms. Silicon Valley's competitiveness was essentially a form of social competitiveness. Today, the worsening inequality and spatial exclusion in the Bay Area are eroding those very social conditions that created Silicon Valley. This illustrates two things: social muscle is a productive asset, not consumption; social muscle can atrophy and requires active maintenance.
From this perspective, growth that does not feed back into vitality consumes the very soil itself. The reason the Engels' Pause is a 'pause' and not merely 'injustice' is precisely because wage stagnation takes away, superficially, income, but in reality, vitality—the willingness to work, to consume, to procreate again—and these are precisely the seeds of the next round of productivity. Therefore, social sarcopenic obesity is not only unjust, but also unsustainable: one can temporarily rely on overdrafting vitality to pile up growth, but cannot forever rely on atrophied muscles to support an obese economy.
Diagnosis: Social Sarcopenic Obesity
Aiming the lens at the social side, the pathology can be diagnosed. There is a medical condition called 'sarcopenic obesity': the patient's weight may not decrease, fat may even accumulate, but skeletal muscle mass and strength progressively decline, and it's extremely insidious, often with only one symptom initially—fatigue. An economy falling into the same state is precisely like this: GDP, profits, capital stock—these 'fats' continue to thicken, while society's organizational stock, capacity for action, and self-development momentum—these 'muscles'—continuously erode. Clinically, it manifests as low fertility, low desire, low participation, weak mutual aid—a pervasive social fatigue. 'Lying flat' and 'involution' have become buzzwords simultaneously because they are no longer just individual psychological phenomena, but a societal mindset; social fatigue is the clinical manifestation of this societal mindset.
It's necessary to clarify one sentence here: 'obesity' refers to fat, not weight. What is fat? It's inefficient capital stock, asset bubbles, overcapacity, accumulated debt, and 'involutionary' competition, not growth itself. Reducing fat is not equivalent to reducing weight; development remains the top priority. This article is not opposing growing the body, but growing only fat and no muscle.
Medically, diagnosing sarcopenia looks at three dimensions: muscle mass, muscle strength, and physical function. Mapping this onto society, there are similarly three indicators. First, muscle mass, i.e., the stock of 'social strength': the density and actual functioning of community organizations, volunteer service groups, collective organizations, homeowners' associations, etc. Second, muscle strength, i.e., capacity for action: whether collective bargaining can be achieved, whether community deliberations can solve problems, whether workers dare to apply for labor arbitration without worrying about a 'stain' on their resumes. Third, function, i.e., actual output: fertility rates, coverage of family care and community mutual aid, social trust, incidence of neighborhood mutual aid. A society simultaneously degrading in all three dimensions is a society losing muscle. These three indicators are not rhetoric; they can be measured and should be measured—a pilot zone that only assesses industrial indicators and not social indicators will not know its muscles are wasting away.
Particular attention must be paid to the pathological mechanism. The most common cause of muscle atrophy is not amputation, but long-term disuse—medically called disuse atrophy, 'use it or lose it'. Social muscle is the same: it is not taken away, but not used. Four types of substitution are quietly occurring. First, administrative substitution: grassroots governance is increasingly administrative, 'the government does, the masses watch', social functions are taken over by administration. Second, algorithmic substitution: mutual aid, visits, neighborly exchanges—interactions from the past 'acquaintance society'—are decreasing, relationships between people are replaced by relationships between people and algorithms. Third, market substitution: whatever can be bought is bought, so the community is no longer needed. Fourth, and the newest, is narrative substitution unique to the AI era: the expectation that 'you will be replaced by machines' replaces people's willingness to bargain before machines actually do. Scholars studying the AI industry find that workers, fearing replacement by algorithms, dare not raise reasonable demands for higher hourly wages or better conditions; companies package labor-dependent businesses as 'pure tech companies', creating the illusion that labor is 'replaceable'; the grand narrative that 'the future of labor is determined by technology and capital' makes ordinary people believe they have no need to participate. What takes away bargaining power is not machines, but the stories told about machines.
The consequence of the four substitutions is the same: social muscle degenerates day by day due to long-term disuse. This determines that the treatment logic must be reversed: redistribution is an infusion of nutrients, it can save lives but cannot regrow muscle; the only rehabilitation method is to reload, to make society use its muscles again.
Lessons from the West and the Social Atrophy Trap
The history of the West provides a negative case study.
The first step was the Engels' Pause itself: during the decades of primitive accumulation, labor's share was systematically suppressed. Britain later emerged from the pause. Economic historians explain it as capital accumulation catching up—high profits induced high savings, and by the mid-19th century, capital stock finally matched technological progress, and wages began to grow with productivity. But why could capital accumulation translate into wages, rather than continuing into profits? This was inseparable from a whole set of social capacity building: the repeal of the Combination Acts in 1824, the Trade Union Act of 1871 allowing workers to organize legally, the Factory Acts of 1833 and 1847 limiting working hours, the repeal of the Corn Laws in 1846 lowering rentiers' rents, the Reform Acts of 1867 and 1884 expanding suffrage, the Elementary Education Act of 1870 popularizing education. Capital accumulation provided the possibility, social capacity building turned that possibility into wages.
The second step's lesson is even more worthy of China's vigilance. Since the late 1970s, a new round of decoupling between wages and productivity has appeared in the West. This time, the reason wasn't that society lacked muscle, but that muscle was actively dismantled: taking the US as an example, union density fell from a peak of one-third to about one-tenth today, collective bargaining coverage shrank, community and mutual aid organizations declined, and financialization changed the direction of profits: more and more profits were used to purchase existing assets rather than build new productive capacity. This money wasn't idle; it increased existing asset prices, not new factories and jobs; many large US companies have used most of their net profits over the past decades for buybacks and dividends, following this very path. Facing the gap, the West's chosen method of bridging it was mainly transfer payments: primary distribution deteriorates, and taxation and welfare are used to cushion it after the fact. This isn't without effect—it did stop the pain—but painkillers stop people from treating the disease. Redistribution replaced the repair of primary distribution and social organization; the erosion of social muscle was masked, not reversed. Now, we see the total eruption of low fertility, low desire, 'diseases of despair', political polarization, and populist waves—atrophied muscles can finally no longer support the obese economy, and society expresses its sense of fatigue with its votes.
Interestingly, atrophy doesn't have a necessary relationship with welfare levels. The US, with the thinnest welfare, has the heaviest 'diseases of despair'; the Nordic countries, with the thickest welfare, maintain the highest union density, volunteer participation, and social trust—because their welfare is to a considerable extent operated through unions, cooperatives, and community organizations; society has been under load all along. Japan, as a high-income country, has taken another path of atrophy during its 'lost three decades': from a once-acknowledged 'high vitality' society to the birthplace of the 'low desire society' and 'relationless society'. General Secretary Xi Jinping has stated that we must avoid blindly 'rushing into welfare' and falling into the 'middle-income trap' like some Latin American countries, and also avoid implementing 'excessive welfare' leading to insufficient social vitality like some Nordic countries; economic development and social security are related like water level and boat size—shallow water rows small boats, deep water sails large ships. The key point of this statement is precisely the word 'vitality': welfare itself is not the disease; using welfare to replace vitality is the disease.
This trap deserves a name parallel to the 'middle-income trap': the social atrophy trap. It is more insidious than the middle-income trap and comes later: the middle-income trap is stuck 'unable to rise', the social atrophy trap is stuck 'unable to sustain'; the former is a trap of the development stage, the latter can occur at any income stage. Crossing the middle-income trap does not mean crossing the social atrophy trap—Japan and the US are examples. For China, this is a mirror we must look into: what we need to avoid is not stagnation of growth itself, but losing social muscle during growth.
Why China is Different
Looking at the West's two steps within China's coordinates yields three judgments.
First, China will not replay the West's first step. The primitive accumulation-style Engels' Pause presupposes the state's acquiescence or even connivance in the decline of labor's share. In China, 'increasing the share of residents' income in national income distribution, increasing the share of labor remuneration in primary distribution' has been a consistent requirement since the 17th Party Congress, the 'two synchronizations' are written into the 18th Party Congress report and the 15th Five-Year Plan outline, and 'comprehensively addressing 'involutionary' competition' is written into the 2024 Central Economic Work Conference and the 15th Five-Year Plan outline. Labor is already in the state's objective function.
Second, China's 'blood transfusion' pipelines are already in place, and the dosage is still increasing. We have built the world's largest social security system covering the most population, historically resolved absolute poverty, and the proportion of people's livelihood expenditure in GDP continues to rise. Although the level of security still has room for improvement, the direction is already clear.
Third, precisely because of this, China's real task is not to prevent the West's first step—primitive accumulation-style suppression of labor's share, which has been excluded from the state's objective function—but to prevent repeating the West's second step—using blood transfusion to replace muscle gain, letting 'the money is in place' become 'no need to handle matters'; and to carve out a third step never taken by the West—growing muscle synchronously with growth. This is the real challenge of Chinese-style modernization.
The proposal of 'investing in people' indicates the state has realized this. The 2025 Government Work Report first proposed 'investing in people', and the 15th Five-Year Plan outline further required 'strengthening investment in human resources development and people's all-round development', placing investment in people and investment in things on equal footing. This is a major shift in the social policy paradigm: education, childcare, care, retraining—in the old paradigm, these were consumption expenditures; in the new paradigm, they are productive investments. But this article wants to push 'investing in people' one step further: investing in people is not just giving them protein, but also making them bear load. Eating protein without exercise doesn't grow muscles.
Vitality and Order: Muscle Gain is Not Adding Chaos
At this point, we must directly address a concern: will society become chaotic if it becomes more active?
This concern is not unfounded; 'unified, it stagnates; decentralized, it becomes chaotic' has indeed occurred repeatedly. But it is precisely the medical metaphor that gives the clearest answer: muscle has never been the enemy of order, but its support. People with muscle atrophy are more prone to falling; a society without muscle is more fragile. Historically, what truly became 'chaotic' was often not an organized society, but an atomized society—where social forces either fail or become a 'mob', precisely the consequence of lacking intermediate organizations. The West's populist waves occurred precisely after the decline of intermediate organizations like unions, churches, and communities, not before.
General Secretary Xi Jinping has spoken very clearly: 'Social development needs to be full of vitality, but this vitality must also be orderly activity. Stagnant water is no good, undercurrents surging is also no good.' So the question has never been whether to have vitality, but what kind of vitality. If we create a four-quadrant matrix of order and vitality, 'high order—low vitality' is rigidity, 'low order—high vitality' is disorder, what we want is 'high order—high vitality'.
Where China differs from the West is that we have the Party, this organizational axis connecting the state and society. Organizing the masses is originally the Party's tradition; the core of the mass line is organizing the masses; during the revolutionary years, 'one hand grasps production, the other grasps life', factories had deputy directors for life, classes had life commissars; today's Party building leading grassroots governance, the new-era 'Fengqiao Experience', 'social governance community where everyone has responsibility, everyone fulfills their duty, and everyone enjoys the benefits', still uses the language of organization. Social muscle gain is not about introducing some foreign 'civil society', but the application of the Party's organizational tradition in a new stage. Making society reload under the leadership of the Party yields not the chaos of 'low order—high vitality', but the strength of 'high order—high vitality'.
Prescription: Economic Weight Loss, Social Muscle Gain
The cost of misdiagnosis is prescribing the wrong medicine. Medically, the most contraindicated prescription for sarcopenic obesity is precisely simple dieting: reducing calories causes muscle, not fat, to be lost first, making the patient weaker. By the same logic, if we only do 'economic weight loss'—deleveraging, reducing overcapacity, countering involution—without simultaneously doing 'social muscle gain', the fat saved will not automatically turn into muscle, and society may become even more fatigued. Therefore, the prescription must be to do two things simultaneously: lose fat without losing weight, development remains the top priority; gain muscle through loading, social policy must move from 'providing a safety net' to 'investing in people', and from 'investing in people' to making people bear load. Poverty alleviation has already completed a transformation from transfusion to hematopoiesis, 'fostering will and wisdom before alleviating poverty, stimulating endogenous motivation'; what needs to be done today is to extend this logic from the impoverished population to the entire society—not just fostering will and wisdom, but also fostering organization.
There is another medical fact: muscle is metabolically active tissue; the greater the muscle mass, the higher the basal metabolic rate, and the body consumes more energy even at rest. What social muscle determines is precisely the 'basal metabolic rate' of domestic demand: an organized, trusting, mutually aiding society will spontaneously consume, procreate, and start businesses even without stimulus policies; an atomized society can only maintain a heartbeat temporarily even with much stimulation. This is why expanding domestic demand is ultimately a social problem.
The specific prescription can be summarized as 'one fulcrum, two synchronizations, three steps'.
One fulcrum is to turn the 'two synchronizations' from an indicative target into the objective function of macroeconomic policy. Since the Chinese definition of the Engels' Pause is 'two desynchronizations', then the way to prevent it is to make 'synchronization' a hard constraint: labor's share of GDP should increase year by year and be incorporated into assessments; labor share, employment quality, and social vitality stock should be incorporated into the consistency assessment of macroeconomic policy orientation—the 2023 Central Economic Work Conference already proposed incorporating non-economic policies into this assessment; here we need to give it a colloquial name and a clear interface, calling it a 'social physical exam': before growth policies are introduced, they must pass an exam—do they increase or decrease social muscle? Major industrial policies like future industry pilot zones and AI should especially undergo ex-ante social impact assessments; otherwise, the situation of 'industrial policy running at full speed, social policy still tying its shoelaces' will recur. Countering involution should also be linked to increasing labor remuneration: the essence of involution is feeding capital's fat with workers' muscles; countering involution is fat loss, and the money saved from fat loss must be converted into wages and skill investment; otherwise, it just swaps one type of fat for another.
Beyond the two synchronizations, a third synchronization should be added: synchronization of social capacity with economic capacity. This is an extension this article makes to the 'two synchronizations'—income synchronization and remuneration synchronization manage flows, social capacity synchronization manages stock; without the third synchronization, the first two are hard to maintain long-term.
The three steps are three tiers of social policy: safety net, investing in people, and loading. The safety net is social security, a blood transfusion, which China is already working hard on; investing in people is education, childcare, care, skills, protein, which China is doing; loading is enabling families, communities, enterprises, and workers to further play a subjective role, resistance training, which is the next step. Loading can focus on several things.
Money follows organization. Childcare subsidies are typical protein; they can be bundled with community mutual childcare, time banks, mutual aid elderly care, making subsidies one of the nutrients for incubating care organizations; fiscal investment in communities can be matched according to residents' actual participation rates, something Hubei's 'co-creation' is already doing. Residential complexes should return to being communities; community talents, community funds, residents' 'self-management, self-service, self-education, self-supervision' should become daily practice, not just legal provisions.
Bring labor into the residual claimant structure. Through the reform of industrial workforce construction, rigidly link skill levels with compensation, giving skills more recognition and reward; promote the institutionalization of employee stock ownership and profit sharing in mixed-ownership reform and private enterprises; implement democratic enterprise management at the firm level; transform wage determination from a cost competition logic to a residual sharing logic. Companies in law have never been merely profit-making machines; understanding enterprises solely as profit-making legal persons both restricts their social vitality and increases the state's burden for public services.
Muscle building for new forms of employment. Promote collective bargaining and algorithmic consultation within the legal framework in platform industries, making 'algorithm filing' and 'algorithm consultation' parallel; recognize the value of invisible labor like data annotation and algorithm calibration; use systems to tell workers that the story of machines is not told unilaterally by machines.
Production-side sharing of AI dividends. Highly mobile digital wealth is difficult to fully regulate through ex-post taxation; before the control of computing power, models, and data is thoroughly monopolized, leverage the positive role of public computing power as new infrastructure, support open-source ecosystems, implement the benefits for workers and SMEs under the 'separation of three rights' for data elements, preventing the means of production in the AI era from being enclosed into rents.
Make AI feed back into social muscle. Guide AI to be prioritized for use in 'function' dimensions like care, education, and community services, recognize 'human-machine collaboration positions', making AI society's exoskeleton, not a replacement. What truly settles after the tech bubble bursts is never computing chips—they depreciate in a few years—but the skills and habits formed by a generation using new tools. This is the legacy the AI boom should leave, and the true meaning of 'investing in people'.
Grow muscle at the county level. Counties are governance units with complete elements and the most stability, and they are also the hard nut of common prosperity. Forge county-wide common prosperity industrial chains, let external capital organically embed into local societies, keep value within counties, which is growing muscle at the county level; the way enterprises participate in social construction should also move further from donations to organization—the participation of 110,000 enterprises in 'Ten Thousand Enterprises Helping Ten Thousand Villages' shows enterprises can be part of social muscle.
None of these things are new inventions; they each have existing policy interfaces. What needs to be strengthened in the future is putting them into the same prescription: not giving society blood transfusions, but making society regrow and use its muscles, making social capacity building an important part of national capacity building.
This article is from the WeChat public account "Tencent Research Institute" (ID: cyberlawrc), author: Lü Peng.
