In a long-form essay republished by MarsBit, Lü Peng of Tencent Research Institute argues that the debate over a possible new “Engels Pause” in the age of artificial intelligence is being framed too narrowly.
His central claim is blunt: the problem is not just distribution. It is a deeper social pathology in which the economy grows larger while society contracts. Lü calls that condition “social sarcopenic obesity,” borrowing a medical term to describe an economy where GDP, profits and capital stock keep accumulating even as social organization, collective capacity and self-development weaken.
That distinction matters because, in his view, a wrong diagnosis leads to the wrong treatment. If the issue is seen only as one of income distribution, the remedy becomes redistribution, or what he likens to a blood transfusion. The real prescription, he writes, is to let society rebuild and use its own muscle again: slim down the economy’s fat while strengthening society itself.
The Engels Pause and the wage stagnation problem
The term “Engels Pause” was coined by economic historian Robert Allen, drawing on Friedrich Engels’ 1845 work The Condition of the Working Class in England. It refers to the early phase of Britain’s Industrial Revolution, when output per worker increased but wages barely moved.
Lü cites the standard figures: between 1780 and 1840, output per worker in Britain rose by about 46%, while real wages increased only 12%. Over the same stretch, the rate of profit climbed from roughly 10% in the late 18th century to above 20% by the middle of the 19th century, and the share of profit in national income expanded from around one-fifth to more than two-fifths. In that period, the gains from technical progress were absorbed mainly by capital. Workers did not receive a comparable share of the wealth they helped produce.
That pattern did not last forever. Lü notes that conditions began to reverse after the 1840s, and by the middle of the 19th century the pause had ended. Over the next half-century, real wage growth outpaced output growth.
He says the concept is being revived now because people see echoes of the same pattern in the present: algorithms and automation are generating large gains in output and profit, labor’s share has been under long-term pressure across the world, and the stability of traditional employment relations, including some white-collar jobs, is being challenged.
China, he adds, is not in exactly the same position. Labor compensation as a share of GDP bottomed out around 2007 and later recovered, but that recovery is not firmly anchored. The social changes unfolding alongside heavy AI investment are putting that foundation under stress. In Lü’s reading, the Engels Pause in China is more a risk building for the future than a fully formed condition of the present.
He also stresses that China is not watching the issue from the sidelines. The policy principle that household income growth should keep pace with economic growth, and labor compensation should keep pace with productivity, has been written into major policy documents from the 18th National Congress of the Communist Party of China to the outline of the 15th Five-Year Plan. Lü treats those “two synchronizations” as an institutional rejection of an Engels Pause. In economic-history terms, he writes, an Engels Pause is a condition of “two non-synchronizations.” The question is no longer whether the commitment exists, but how it can be made real.
Why Lü says the issue is not just about splitting the pie
A popular account treats the Engels Pause as a distribution problem. Lü does not say that is false. He says it does not reach the core.
His formulation is that industrial development and broader economic and social development have come apart. Productive forces, in this telling, are like an engine spinning faster and faster, while the social relations that should transmit that power have become a broken drive shaft. The engine runs, but the vehicle no longer moves as a whole.
He uses two analytical “keys” to explain the disconnect.
The first is the fallacy of composition. At the micro level, an individual firm may have perfectly rational reasons to cut labor costs, replace workers with machines, or outsource employment to platform structures. But if every firm does the same thing at once, the aggregate result is a structural drop in labor’s share. Consumption can no longer absorb output. Domestic demand weakens. Overcapacity and “involution” rise together.
Lü links this to Keynes’ paradox of thrift and flips a line often associated with Michał Kalecki: workers spend what they get, and capitalists get what workers spend. If workers do not get enough in the first place, capitalists eventually run into the problem of not being able to realize demand. He also brings up Henry Ford’s decision to raise the daily wage to $5. Whether that move was intended as a doctrine or not, Lü says the underlying logic still holds: if workers cannot afford to buy cars, a car factory has no future.
The second key is what he calls the illusion of aggregates. GDP growth may be real in an accounting sense, but its welfare meaning can be misleading because aggregate indicators hide changes in composition. A large share of growth can be generated by sectors favored by capital markets, while the elements that make up the social body, households, communities, labor and trust, are thinning out. Lü points to estimates showing that AI and data-center-related investment once accounted for more than 90% of U.S. growth during a certain period. The headline number looked strong. The social structure underneath may have been getting weaker.
His phrase for that is strikingly simple: weight goes up, muscle does not.
Why the old pause could heal and the new one may not
Lü spends considerable time on the historical mechanism that ended the original Engels Pause.
Following Robert Allen’s account, he writes that after 1800 technical progress accelerated, but those technologies could only become output if they were embodied in new factories, machines and railways. Britain’s existing capital stock was too small relative to what the new technologies required. Capital was scarce, so returns on capital were high. Labor was not scarce, so wages remained flat.
Over several decades, capital stock finally accumulated to a scale consistent with the new technologies. At that point, bringing additional capital into operation required hiring workers. Demand for labor rose, and wages began to move with productivity. Lü condenses that process into a self-healing chain: high profits induced reinvestment, reinvestment expanded capital stock, larger capital stock increased labor demand, and stronger labor demand lifted wages.
That chain, he argues, depended on two hidden premises. First, profits had to be reinvested in production. Second, new capital had to require labor as a complement. In the 19th century, machines and labor tended to work together. More machinery often meant more workers were needed to operate it.
He then argues that both premises are under strain in the current cycle.
On the first point, the income squeezing wages is increasingly not profit in the classical competitive sense but rent. Lü describes profit as a temporary excess return produced in competition, one that has to be sustained through new investment and capacity expansion. Rent is something else. It is a lasting stream of income that appears after an originally open market relation becomes closed off. He mentions the term “cloud rent” to describe income maintained by network effects and data barriers. Such income can persist without continual productive reinvestment and is less vulnerable to entry by competitors.
Money extracted in that way can be used for share buybacks, dividends, acquisitions or purchases of existing assets. Those uses can still produce gains on paper as asset prices rise, but they do so through revaluation of existing assets rather than the construction of new factories or the creation of new jobs. In Lü’s framework, the self-healing chain breaks at its first link.
The second break comes later in the chain. Even when capital is reinvested, modern capital does not necessarily need more workers. The reason 19th-century capital accumulation could lift wages, he writes, was that machines needed labor to operate them. In the AI era, by contrast, capital deepening in the form of computing power, models and data centers is often aimed at replacing labor. More capital can mean fewer people. That breaks the chain again, this time at the stage where expanded capital stock would once have fed through to labor demand.
That is why, in Lü’s view, no one should expect the new pause to end naturally just because capital accumulation eventually catches up with technical progress. The old pause ended through a mix of economic adjustment and social capacity building. If a new pause is to end, the social side will have to carry a larger share of the burden.
Social vitality and productive capacity
To support that claim, Lü moves to a broader theoretical frame: the relation between social vitality and social productive capacity.
He traces a line through Chinese policy language, from the 14th Party Congress in 1992, which placed “further liberating and developing social productive forces” alongside “stimulating and strengthening social vitality,” to the 2013 Third Plenum of the 18th Central Committee and the reaffirmation in the 2024 Third Plenum of the 20th Central Committee and the 15th Five-Year Plan’s call to keep strengthening development momentum and social vitality. For more than three decades, these “two forces” have been mentioned together.
Lü’s interpretation is that social productive capacity is not a simple function of capital, technology and labor inputs. It is a function of social vitality. Capital depreciates. Technology becomes outdated. What remains decisive, he writes, is a society’s willingness to work, impulse to create, readiness to have children and habit of mutual aid. Social vitality is not only the drive shaft connecting engine and body. It is a second engine.
He does not leave the point at the level of rhetoric. The essay cites AnnaLee Saxenian’s comparison of Silicon Valley and Boston’s Route 128. Both regions had strong technology and capital bases, and Route 128 even held an early edge through MIT and defense orders. Yet Silicon Valley won out. Lü says the difference lay in social organization rather than technology alone: open labor markets, weaker corporate boundaries, dense informal knowledge exchange, a high tolerance for failure and an occupational identity rooted in community instead of a single employer.
From that case, he draws two conclusions. Social muscle is a productive asset, not consumption. And social muscle can atrophy if it is not maintained. Rising inequality and spatial exclusion in the Bay Area, he adds, are now eroding the social conditions that once gave Silicon Valley its strength.
Seen this way, growth that does not feed back into social vitality is consuming its own soil. Wage stagnation in an Engels Pause does not merely take away income on the surface. It drains willingness to work, willingness to consume and willingness to reproduce. In Lü’s argument, those are the seeds of the next round of productivity.
Diagnosis: social sarcopenic obesity
That is where the essay turns to its main diagnosis.
In medicine, sarcopenic obesity describes a patient whose body weight may not fall, whose fat may even keep accumulating, while skeletal muscle mass and strength decline steadily and often quietly. One of the earliest symptoms is simple fatigue. Lü says an economy can enter an analogous state. GDP, profits and capital stock, the “fat,” continue to thicken, while the “muscle,” meaning social organization, capacity for action and self-renewing development, wastes away.
The clinical expression of this condition in society, he writes, includes low fertility, weak desire, low participation and thinning mutual aid, a broad sense of social fatigue. The simultaneous popularity of terms such as “lying flat” and “involution” is important to him for that reason. He reads them not as isolated psychological moods but as symptoms of a wider social condition.
He also takes care to define what “obesity” means in this metaphor. It does not mean growth itself. It refers to fat, not weight. In his list, fat includes inefficient capital stock, asset bubbles, overcapacity, stacked-up debt and forms of “involutionary” competition. Slimming fat is not the same as cutting weight. Development remains the first priority. The target is not growth, but a structure that adds fat without building muscle.
Lü maps the medical diagnosis of sarcopenia onto three social dimensions.
- First is muscle mass, meaning the stock of “social forces”: the density and actual functioning of community organizations, volunteer groups, collective organizations and homeowners’ committees.
- Second is muscle strength, meaning capacity for action: whether collective bargaining can happen, whether community deliberation can solve problems, and whether workers dare to file labor arbitration claims without fearing a stain on their records.
- Third is function, meaning actual output: fertility rates, the reach of family care and community mutual aid, levels of social trust and the incidence of neighborhood support.
All three, he says, are measurable and should be measured. A development zone that tracks only industrial indicators and not social indicators will not know that its muscle is disappearing.
Four forms of substitution that weaken society’s muscle
Lü then asks what causes the atrophy.
His answer is that social muscle usually is not removed outright. It weakens from disuse. In medicine, a common cause of muscle loss is lack of load-bearing over time. Use it or lose it. The same principle, he argues, applies to society.
He identifies four substitutions that are quietly taking place.
- Administrative substitution: grassroots governance becomes more administrative, producing a pattern where the government acts and the public watches, while tasks that society itself might have handled are taken over by administrative systems.
- Algorithmic substitution: forms of interaction common in acquaintance-based communities, mutual aid, visits, exchanges among neighbors, are reduced, and relations between people are replaced by relations between people and algorithms.
- Market substitution: if everything can be bought, the community no longer seems necessary.
- Narrative substitution: a distinct feature of the AI era in his account, where the expectation that “you will be replaced by machines” weakens bargaining power even before machines have actually replaced anyone.
He says researchers studying AI-related industries have found that workers may refrain from asking for higher hourly pay or better conditions because they fear being displaced by algorithms. Companies may package labor-dependent operations as “pure tech companies,” creating the illusion that labor is easily replaceable. And grand stories in which the future of labor is decided entirely by technology and capital can persuade ordinary people that their own participation no longer matters. In one of the essay’s sharper lines, Lü writes that what takes bargaining power away is not the machine itself, but the story told about the machine.
All four substitutions produce the same result: social muscle erodes because it no longer bears weight. That, in turn, flips the treatment logic. Redistribution is like intravenous nutrition. It can save a patient, but it does not rebuild muscle. Recovery requires load-bearing again. Society has to use its own muscle.
Lessons from the West and the “social shrinkage trap”
The essay presents Western history as a cautionary sequence in two steps.
The first step is the original Engels Pause. Britain did eventually emerge from it. Economic historians often explain that turnaround through capital accumulation catching up with technical change. Lü accepts that part but says it is incomplete. Capital accumulation became wage growth only because a set of social capacities was built alongside it. He lists the repeal of the Combination Acts in 1824, the Trade Union Act of 1871, the Factory Acts of 1833 and 1847, the repeal of the Corn Laws in 1846, the Reform Acts of 1867 and 1884, and the Elementary Education Act of 1870. Capital accumulation created the possibility. Social institution-building turned that possibility into higher wages.
The second step is the one he says China should watch more closely. Since the late 1970s, the West has seen a renewed decoupling of wages from productivity. This time, he argues, the issue was not that society had no muscle to begin with. The muscle was actively dismantled. In the United States, union density fell from a peak of about one-third to around one-tenth today, collective bargaining coverage narrowed, and community and mutual-aid organizations weakened. Financialization changed where profits went. More and more profits were directed toward existing assets instead of new productive capacity.
Those funds still generated gains, but they did so by inflating existing asset prices rather than building new factories or creating new jobs. Lü notes that many large U.S. companies spent most of their net profits over recent decades on buybacks and dividends. Faced with the resulting social and economic gap, Western systems relied mainly on transfer payments, using taxes and welfare to offset the damage after the fact.
He does not dismiss those measures entirely. They did relieve pain. His point is that pain relief replaced treatment. Redistribution substituted for repairing primary distribution and social organization. The loss of social muscle was concealed rather than reversed.
The long-run expression of that choice, in his account, is a broad eruption of low fertility, weak desire, “deaths of despair,” political polarization and populist waves. At some point, the shrunken muscle can no longer carry the heavy economy.
Lü adds that social shrinkage does not line up neatly with welfare generosity. The United States, with relatively thin welfare provision, has suffered heavily from “deaths of despair.” The Nordic countries, despite extensive welfare systems, have retained high union density, strong volunteer participation and high social trust because welfare there is operated to a meaningful extent through unions, cooperatives and community organizations. Society remains under load. Japan, meanwhile, represents another route: a high-income country that, during its “lost three decades,” moved from a widely recognized high-vitality society to one associated with ideas such as the “low-desire society” and “no-relation society.”
The essay cites Xi Jinping’s warning against both blind “welfare catch-up” of the sort seen in some Latin American countries and a generalized “welfarization” that can sap social vitality, adding that economic development and social security should rise together like water lifting a boat. Lü says the key word there is vitality. Welfare itself is not the illness. Using welfare as a substitute for vitality is.
He gives this pattern a name parallel to the “middle-income trap”: the social shrinkage trap. It is more hidden, he says, and appears later. The middle-income trap is about failing to rise further. The social shrinkage trap is about no longer being able to hold up what has already been built. Japan and the United States are both presented as examples.
Why Lü thinks China’s path is different
Placed against that Western sequence, China looks different in three ways, according to the essay.
First, China will not replay the West’s first step. An original-accumulation-style Engels Pause, Lü argues, requires the state to tolerate or even enable a decline in labor’s share. In China, by contrast, raising the share of household income in national income and the share of labor compensation in primary distribution has been a standing requirement since the 17th Party Congress. The “two synchronizations” were written into the report to the 18th Party Congress and the 15th Five-Year Plan outline. The call to comprehensively address “involutionary” competition was written into the 2024 Central Economic Work Conference and the same plan outline. Labor is already in the state’s objective function, he says.
Second, the channels for “blood transfusion” are already in place and growing. China has built what Lü describes as the world’s largest social security system by population coverage, has historically solved absolute poverty, and has seen people’s livelihood spending as a share of GDP continue to rise. There is still room to improve benefit levels, but the direction is already set.
Third, and because of that, China’s real task is not to prevent the West’s first step. It is to avoid repeating the second: letting transfusion replace muscle-building, so that “the money arrived” becomes “the underlying problem no longer needs attention.” Beyond that, he says, China has to attempt a third step the West never completed, building social muscle in sync with growth. He presents that as a core challenge of Chinese modernization.
The more recent policy phrase “investing in people” is folded into that argument. Lü notes that the 2025 government work report first raised the term, and the 15th Five-Year Plan outline later called for stronger investment in human resource development and the all-round development of people, putting investment in people on the same level as investment in things. He sees that as a significant shift in social policy thinking. Education, childcare, care work and retraining, once treated as consumption spending, are being reclassified as productive investment.
Still, he wants to push the idea one step further. Investing in people is not only about giving them protein. It is about giving them weight to bear. Protein without resistance training does not build muscle.
Vitality and order
Lü also confronts a recurring concern directly: if society becomes more active, does it become less orderly?
He acknowledges that the worry has a real basis. The pattern of rigidity under excessive control and disorder after sudden release is familiar. Yet he insists the medical metaphor points in the opposite direction. Muscle is not the enemy of order. It supports order. A person with weak muscles falls more easily. A society without muscle is more fragile, not more stable.
Historically, he says, genuine disorder often appears not in organized societies but in atomized ones, where social forces either fail entirely or collapse into a crowd. The populist turn in the West, in his reading, came after the decline of unions, churches and communities, not before it.
The essay quotes Xi Jinping’s line that social development needs vitality, but that vitality must take the form of orderly activity; stagnant water is no good, and turbulent undercurrents are no good either. Lü turns that into a four-quadrant picture. High order and low vitality mean rigidity. Low order and high vitality mean disorder. The desired state is high order and high vitality.
What makes China different, he argues, is the Communist Party’s role as an organizational axis linking state and society. Organizing the masses is part of the Party’s own tradition. The mass line, he writes, is fundamentally about organization. Historical references in the essay run from the revolutionary era’s parallel focus on production and daily life to current ideas such as Party-led grassroots governance, the “Fengqiao experience” in the new era and a social governance community in which everyone has responsibility, fulfills responsibility and shares in outcomes.
On that basis, Lü says social muscle-building in China is not about importing an external model of “civil society.” It is about applying the Party’s organizational tradition under new conditions. Under Party leadership, letting society bear weight again is supposed to produce not disorder with high vitality, but strength with both order and vitality.
The prescription: slim the economy, strengthen society
The essay’s final section turns the diagnosis into a program: slim the economy’s fat, strengthen society’s muscle.
Lü uses another medical analogy to explain the stakes. In sarcopenic obesity, simple dieting can be the worst possible prescription because the first thing to go is muscle, not fat. The patient gets weaker. By analogy, if policymakers focus only on slimming the economy, deleveraging, cutting overcapacity, fighting involution, without building social muscle at the same time, the fat that is removed will not automatically become muscle. Society may end up even more fatigued.
So the prescription has two parts at once. Cut fat without cutting weight; development remains the first priority. Build muscle through load-bearing; social policy has to move from basic support to investment in people, and then from investment in people to giving people real agency and responsibility. Lü compares this shift to the move in poverty alleviation from simple transfer to endogenous capacity-building. What is needed now, he says, is to extend that logic from poor populations to society as a whole, not only supporting aspiration and skills, but supporting organization.
He adds one more point. Muscle is metabolically active tissue. The more of it a body has, the higher its basal metabolic rate. Applied to society, social muscle determines the “basal metabolic rate” of domestic demand. A society with organization, trust and mutual aid will consume, have children and start businesses even without strong stimulus. An atomized society may respond to stimulus only briefly. That is why, in his view, expanding domestic demand is ultimately a social question.
One fulcrum, two synchronizations, three tiers
Lü summarizes the policy framework as “one fulcrum, two synchronizations, three tiers.”
The fulcrum: turn the two synchronizations into hard constraints
If an Engels Pause in the Chinese context means “two non-synchronizations,” then preventing it means making synchronization a hard constraint, he writes. The share of labor compensation in GDP should rise year by year and be included in official assessment. Labor’s share, employment quality and the stock of social vitality should all be included in consistency checks for macro policy orientation.
He notes that the 2023 Central Economic Work Conference already called for non-economic policies to be included in such evaluations. His proposal is to give that idea a clearer public label and interface: a “social physical examination.” Before a growth policy is launched, the state should ask whether it adds to or subtracts from society’s muscle. He says that major industrial policies, including those for advanced industrial pilot zones and AI, especially need ex ante social impact assessment. Otherwise the same pattern returns: industrial policy running at full speed while social policy is still tying its shoes.
On involution, the essay says anti-involution measures should be linked directly to higher labor compensation. In Lü’s wording, involution is the use of workers’ muscle to feed capital’s fat. Fighting it is a way of cutting fat, but the money saved must become wages and skill investment. Otherwise one form of fat is simply replaced with another.
He also adds a third synchronization to the original two: social capacity should keep pace with economic capacity. The first two concern flows, income and compensation. The third concerns stock. Without it, the first two will be hard to sustain over time.
The three tiers: basic support, investing in people, load-bearing
The three policy tiers are basic support, investing in people and load-bearing.
Basic support means social security, the transfusion. Investing in people means education, childcare, care and skills, the protein. Load-bearing means enabling households, communities, firms and workers to exercise agency, the resistance training. China, he says, is already working on the first two. The third is what comes next.
Let money follow organization
Lü uses childcare subsidies as an example. They are protein, but they can be tied to community-based mutual childcare, time banks and mutual elder care so that subsidies help incubate care organizations rather than simply transfer money. Fiscal support to communities can also be matched to actual resident participation. He points to Hubei’s “co-creation” practices as an existing example.
He argues that residential compounds should reconnect to community life, and that community leaders, community funds and resident self-management, self-service, self-education and self-supervision should become daily practice rather than remain legal phrases.
Bring labor into the residual-claim structure
At the enterprise level, the essay suggests linking skill grades more rigidly to pay through industrial worker reform so that skills gain stronger recognition and returns; institutionalizing employee stock ownership and profit-sharing in mixed-ownership reform and in private firms; implementing democratic management within enterprises; and shifting wage formation away from pure cost competition toward residual sharing.
Lü adds that a company, in legal terms, has never been only a profit machine. Treating firms as nothing more than profit-making entities restrains their own social vitality and adds to the state’s burden in public service provision.
Build muscle strength in new forms of employment
For platform work and newer employment forms, he proposes pushing collective bargaining and algorithmic bargaining within the legal framework, so that algorithm filing is matched by algorithm negotiation. He also calls for recognition of the value created by hidden labor such as data labeling and algorithm calibration. The institutional message to workers, he says, should be that the story about machines is not for machines alone to tell.
Share AI gains on the production side
The essay argues that highly mobile digital wealth is hard to regulate sufficiently through after-the-fact taxation alone. Before control over computing power, models and data becomes fully concentrated, public computing power as a new type of infrastructure should play an active role, open-source ecosystems should be supported, and under the “three rights separation” framework for data elements, workers and small and medium-sized firms should receive their share of returns. The productive resources of the intelligent era, in his view, should not be sealed off into rent.
Use AI to feed back into society’s muscle
Lü wants AI directed first toward care, education and community services, the “function” dimension in his social diagnosis. He calls for recognition of “human-machine collaboration jobs” so that AI becomes an exoskeleton for society instead of a substitute for people. After speculative froth fades, he writes, what truly remains from technology booms is not chips that depreciate within a few years, but the skills and habits a generation forms by using new tools. That, for him, is the legacy the AI boom ought to leave behind, and the real meaning of investing in people.
Build the muscle at the county level
The essay ends with counties as a key locus. Counties are described as governance units with a full set of factors and the greatest long-term stability, while also being one of the hardest fronts in achieving common prosperity. Building county-level industrial chains for common prosperity, embedding outside capital organically in local society and keeping value inside the county all amount, in Lü’s phrase, to building muscle at the county level. Corporate participation in social development should also move beyond donation toward organization. He cites the participation of 110,000 companies in the “10,000 Enterprises Helping 10,000 Villages” program as evidence that firms themselves can form part of society’s muscle.
None of these items, he concludes, are inventions from scratch. Each already has a policy interface. The task is to place them under one prescription: not to keep giving society transfusions, but to let it regrow and use muscle, making social capacity building an important part of state capacity building.
The article was originally published on the WeChat public account of Tencent Research Institute (ID: cyberlawrc) and is credited to Lü Peng. MarsBit republished the piece.