Chapter 1: More People, More per Person
Prologue: A Larger Output and a Smaller Promise
In an imagined region, the new warehouse is open. Trucks wait at a junction designed for a quieter road, and the morning bus now takes longer to reach the station. Inside the warehouse, people are doing work that did not take place here last year. Some have moved to the region for it. Some left other local employers, who are still deciding how to replace them.
Imagine a region with expanding production, new building, and a larger population. Its statistical office estimates that output has increased after allowing for changes in prices. The increase is real within the measure being used. It is not a trick invented by an official who has forgotten about the bus. Yet the person arriving late for a shift may reasonably find the announcement incomplete. More is being produced. The route by which people obtain it has become less dependable.
The two observations can both be true.
An account of economic growth needs to explain the increase without treating every accompanying difficulty as an objection to measurement. It also needs to explain the difficulties without pretending that additional production is irrelevant. The warehouse may support worthwhile jobs and services. It may divert people from activities whose loss has not yet become visible. A road improvement might make the new work easier, but it would require materials, labor, land, and time that could be used elsewhere. Calling it an investment begins a question about those resources; it does not answer it.
What the announcement measures
GDP measures production during a period. The U.S. Bureau of Economic Analysis describes it in terms of final goods and services, excluding the double counting of intermediate inputs. Its explanations also distinguish inflation-adjusted output per person from a growing aggregate. More people can produce a larger total without increasing the amount per person. Neither measure, on its own, establishes how the benefits are distributed or how well people live.[0-1]
Our imagined statistical office cannot simply add the warehouse's sales to every sale made by its suppliers. Some transactions pay for inputs used in producing something counted later. An economy's larger turnover and its larger output are different matters. Prices also change: a greater money total might reflect dearer goods, more goods, a different mix, or several changes together. Obtaining a useful comparison requires decisions about coverage and valuation that deserve respect even when they leave questions unanswered.
The regional figure would not tell us whether each household became better off. A new arrival, a tenant facing higher rent, an owner of industrial land, and a retired person waiting for the bus occupy different positions. Nor would a higher average settle their differences. We would still need to know what happened to their incomes, costs, time, health, and practical choices. These questions do not invalidate the production account. They concern the uses to which we put it.
A further question reaches beyond the current period. Has the region acquired an ability to keep producing something useful, or is it drawing down resources that will be expensive to replace? The distinction is not always visible immediately. A business can postpone repair while keeping the present shift running. A public agency can approve expansion while leaving the next budget to find the workers who will inspect it. Some investments will prove better than expected; others will disappoint despite having been reasonable when undertaken. Looking forward means carrying uncertainty into the account, rather than giving every new building the status of a future success.
Capacity and its use
A region with empty premises and people unable to find work faces a different immediate problem from one running against shortages of space, skills, or power. In the first, additional demand might bring existing resources into use. In the second, the same spending pressure might raise prices or lengthen queues before production can expand. Long-term improvements can change either position, but they take different routes through the work that people can actually undertake.
This is why I would begin with what is constraining the next useful activity. The answer might involve a machine or a road. It might involve training, a disputed right, a missing source of finance, or a purchasing commitment no one can make credible. Several constraints can operate together. Removing the most visible one may reveal another; sometimes a better design removes several at once.
At the scale of an economy, these decisions interact. The worker recruited by the warehouse is no longer available to another employer on the previous terms. A successful business can expand the market for nearby services while competing with them for premises. A household's saving can finance future production through institutions that may also create fragile claims on it. Individual prudence does not automatically assemble into a workable collective result. Nor is there a single office able to know and direct every adjustment cheaply.
The chapters that follow move from agrarian and industrial constraints to commitments, learning, research, enterprise, trade, finance, and the frontier of new knowledge. Their subject is sustained change in what an economy can do. Some changes raise output once; others alter the possibilities for further improvement. Distinguishing them matters more than calling every improvement a revolution.
The word structural directs attention to the relationships that make these changes possible: who can commit resources, learn from results, adapt a shared rule, and carry the cost when an undertaking fails. It is not a claim that material resources have ceased to matter. A society cannot coordinate its way out of every physical limit, and more elaborate organization can create burdens of its own. The useful achievement may be a simpler process that lets people accomplish more without having to manage another dependency.
Back at the junction, the question is concrete enough. Can the region connect its new work to homes, transport, and public services without allowing each expansion to leave the next participant a worse problem? The production increase is worth understanding. So is the work required to make its promise credible to the people who will live with it.
Notes and Sources
[0-1] U.S. Bureau of Economic Analysis, “Gross Domestic Product,” https://www.bea.gov/data/gdp/gross-domestic-product; “Prototype Measures of Economic Well-Being and Growth,” https://apps.bea.gov/well-being/. The discontinued prototype (last data update April 2024) is used for measurement distinctions only. The region is hypothetical.
Chapter 1: More People, More per Person
A fuller granary can mean several different things. The harvest may have improved because rain arrived at the right time. Cultivators may have brought additional land into use, worked longer, or learned a better way to preserve fertility. There may also be more people doing the work. These changes can produce the same increase in total grain while leaving very different possibilities for the people who must live on it. Before asking why economies grow, we need to know which of these possibilities the word growth is describing.
An economy can produce more without producing more per person. It can produce more per person without making every household better off. It can raise current production while damaging the soil, exhausting a resource, or neglecting the equipment that future production will need. None of these observations makes increased output unimportant. They make its interpretation more demanding. When food is scarce, a larger harvest can be the difference between survival and death. When the question is sustained improvement across generations, the relation between production and population becomes central.
Real gross domestic product measures the volume of goods and services produced within an economy, using methods that try to separate changes in quantities from changes in prices. GDP per person divides that aggregate by population. Productivity asks another question: how much output is obtained from a given amount of labor or other inputs? More adults working can increase output per resident even if output per hour does not change. Longer working days can do the same. A higher figure therefore needs a second sentence explaining how it was achieved.
For an agrarian economy, that second sentence often concerns land. Land is not a uniform substance. Some fields are close to water and settlements; others demand drainage, clearance, transport, or protection against erosion. Bringing another field into cultivation can increase the harvest while yielding less for each additional day of work. The difference between more production and more productive work is already present in this ordinary decision. It does not require a modern factory or a complicated statistical system.
The pressure of a larger harvest
Consider a hypothetical district whose households obtain most of their food locally. The district has productive fields, rough grazing, woodland, a market, and a road that becomes difficult in wet weather. A new rotation improves the average harvest. The initial gain allows families to eat better, retain more seed, and carry a reserve into the next season. The gain also releases some work for repairing buildings and making tools. Nothing in this example requires population to increase immediately. Births, deaths, migration, and household formation move on different schedules.
Suppose, over time, more children survive and additional households settle in the district. There are more consumers and eventually more workers. The improved method continues to produce a larger total harvest than the old one would have produced. Yet a growing population can absorb the additional food. The district may sustain more people without sustaining a permanently higher average consumption. A technical improvement has mattered enormously, but its long-run result cannot be read from the first successful harvest.
The pressure becomes sharper when households divide existing holdings. One family can cultivate its smaller plot more carefully, adding work at the busiest times. Another must accept distant land whose produce is expensive to bring home. A third seeks paid work from a larger landholder. All three responses can raise total activity. They can also change bargaining power. More workers competing for seasonal employment may find that higher production does not translate into a higher wage. Landowners may receive a larger share of the gain through rents.
This is the mechanism that makes the population question more than arithmetic. Additional people have needs, but they also contribute effort, skills, inventions, and relationships. Whether those contributions raise output sufficiently depends on what they can work with and how their work is organized. A larger population confined to fixed methods and increasingly poor land faces a different situation from a larger population able to improve tools, exchange products, or develop new kinds of work. Head counts alone cannot settle the matter.
Thomas Robert Malthus made population pressure the organizing problem of his 1798 Essay. He contrasted the potential expansion of population with the slower expansion of subsistence and described feedback through wages, food prices, marriage, and mortality. His argument allowed improvement in cultivation, but treated it as vulnerable to renewed population growth. This was a model of an adjustment mechanism, presented in an intervention against contemporary hopes of indefinite improvement. It was not a comprehensive measurement of every preindustrial economy. [1-1]
The model asks a useful question: what prevents an improvement from being fully absorbed by additional claims on resources? Its usefulness does not require accepting its numerical illustrations as a law of human behavior. People decide when to form households under changing conditions. Mortality responds to disease environments as well as nourishment. Movement between regions can relieve a local shortage while creating pressure elsewhere. Institutions affect access to land, food, and work. These are parts of the mechanism, rather than disturbances that can always be set aside.
A society also need not return to precisely the same condition after every disturbance. Once a road has been improved, a crop introduced, or a skill transmitted, the next generation begins with something different. A succession of such changes can alter the relation between population and output. The question is whether the changes arrive fast enough, reach enough people, and remain useful under the new population and resource conditions. An improvement that is real but intermittent differs from a process that repeatedly expands productive possibilities.
What surviving records can show
Historical evidence makes this inquiry harder and more interesting. The distant past did not leave a continuous national account prepared according to a modern statistical manual. Researchers reconstruct quantities, prices, wages, rents, occupations, and population from records made for other purposes. A payment recorded by a landowner can tell us something important about hired work. It may tell us little directly about household production, unpaid labor, days without employment, or people excluded from the record. A long series remains an achievement even when its coverage is incomplete.
Gregory Clark's author-hosted study of English farm laborers, covering 1208–1850, constructs wage and cost-of-living estimates from manuscript and published material. Its methods show why apparent historical continuity needs investigation: locations enter the surviving record unevenly, and pay differs across places. Clark adjusts for those differences rather than treating each decade's raw average as directly comparable. He also finds that the resulting account differs from an older building-wage series. This is evidence about reconstructed male agricultural day wages in England, with thin coverage in some periods; it is not an observation of everyone's annual income. [1-2]
The distinction between a day's wage and a year's livelihood changes the interpretation. If the day rate rises while employment becomes less regular, annual purchasing power may improve less than the rate suggests. If other household members obtain work, household resources can rise without a corresponding movement in the recorded male wage. Prices matter too. A worker paid more coins may command less food if the cost of necessities rises faster. A meaningful historical comparison therefore joins the payment to the goods it could buy and the work that was available.
A different reconstruction makes the disagreement concrete. In their December 2011 output-based account, Stephen Broadberry and his coauthors assemble estimates for agriculture, industry, and services and combine them with population to estimate GDP per person. Their principal aggregate series covers England for 1270–1700 and Great Britain for 1700–1870. They find modest positive trend growth before industrialization rather than an unchanging level. Their agricultural calculations distinguish crop acreage, yields, seed, and animal feed, and they explain assumptions needed to move from surviving estates to national totals. A gap between major agricultural sources is filled using an estimated demand relationship. These are reconstructed accounts, with meaningful assumptions and interpolation, rather than continuous direct observation. [1-3]
The difference between an output reconstruction and a wage series is itself informative. One asks about production divided by population; the other observes the purchasing power of particular work. They need not move together. The comparison should encourage investigation of workdays, occupations, household resources, and distribution. It should not encourage choosing whichever series produces the simplest story. A history of growth becomes more credible when its measures can disagree without one being treated as irrelevant.
Population estimates introduce another layer. If output is reconstructed with reasonable care but the population denominator is uncertain, an estimate of output per person inherits that uncertainty. A revision to population can change the inferred pace of improvement without changing a single surviving harvest record. This does not mean historians should abandon comparison. It means a statement about several centuries should preserve the difference between a broad pattern and an exact trajectory. The data may support the existence of pressure more securely than the precise date at which pressure weakened.
These limitations should make us wary of an image of the entire preindustrial world waiting motionless for a single modern breakthrough. Improvements could occur in agriculture, transport, commercial organization, and craft production without generating the sustained rates associated with later industrial economies. Gains could be concentrated in a town, a region, an occupation, or a generation. A subsequent war or epidemic might reverse some gains and leave others intact. Slow average change can contain substantial local invention and institutional rearrangement.
Nor does a rise in survivors' wages after population loss establish that population loss was beneficial. An average calculated among those alive cannot count the welfare of those who died as an improvement. Bereavement, broken households, abandoned work, and lost knowledge remain part of the event. A scarcity of labor may strengthen the bargaining position of some workers while reducing total production. The causal observation and the moral evaluation must not be collapsed into one another. Higher output per survivor can coexist with catastrophe.
Surplus is a claim on future work
Return to the hypothetical district. Its improved harvest produces a surplus beyond immediate food consumption. Calling that surplus investable can conceal several decisions. Some grain must be reserved for seed. Some is prudent insurance against the next poor season. Some supports children, older people, or people unable to work. Some may be claimed as rent or tax. Only after these uses are considered can we ask how much can support people who are building a road, digging drainage, or learning a specialized craft rather than producing this season's food.
The district chooses to drain a group of fields before rebuilding the road. That choice puts ditch diggers to work now, while the additional crop will not arrive until a later season. They must eat from the same reserve that protects households against a poor harvest. If the district commits the whole surplus, a disappointing season could force it to abandon the works just before completion. Its members therefore retain a reserve and begin with the section most likely to improve the next harvest. Some worthwhile road repairs wait.
Who controls the grain now affects who can undertake the work. A large landholder may finance drainage on his own holdings because he expects higher rents. Smaller cultivators cannot necessarily follow, even if their plots would benefit as much: they may need every remaining sack for food and seed. A shared contribution could extend the project, but only after people agree which fields receive the first improvement and how those who pay will benefit. The technical opportunity is common; the capacity to use it is uneven. That difference can turn a larger harvest into higher rents for some households well before it becomes a better livelihood for others.
The district's woodland illustrates the difficulty. Clearing it can create additional fields and raise measured agricultural production. It also removes a source of fuel, grazing, materials, and protection against runoff. The value of the new crop may be easy to observe at market; the lost services may appear slowly and in other households' accounts. A harvest series can rise even as the community reduces a stock on which later harvests depend. The distinction between current output and wealth is therefore present before industrialization as well.
Wealth includes resources and produced assets that can support future activity. Output is a flow during a period. A household can eat by selling an asset, but the sale itself does not demonstrate that the household has learned to produce more. Likewise, reallocating ownership of a field changes claims without creating another field. A district can benefit from a better allocation if the new user cultivates more effectively, but that consequence must be established. An exchange of claims and an increase in production are separate events.
The same care applies to taxes and transfers. Moving grain from one household to another can prevent hunger, support construction, or sustain an unproductive privilege. Its effect depends on use and on what the contributing household gives up. The movement does not create grain at the moment of transfer. Yet the absence of a new physical quantity does not make distribution irrelevant to growth. A transfer that allows a skilled worker to survive a bad season may preserve productive ability. One that removes necessary seed can diminish the next harvest.
Moving the constraint
The district need not respond to every additional mouth by extending cultivation. Suppose the road is improved enough to allow dependable exchange with a neighboring region. The district can sell a product it produces relatively well and buy food that is cheaper to obtain elsewhere. Local land now places a less immediate limit on local consumption. The constraint has moved into the reliability of exchange: transport capacity, access to buyers, purchasing power, and the conditions under which supplies remain available.
This change does not abolish agriculture or the physical world. Someone still grows the imported food. A local improvement can depend on resources beyond the district's boundary. If all regions face the same crop failure, exchanging claims cannot create the missing harvest. If failures differ across places, exchange can reduce exposure to a local shock. The distinction matters because a district's success in relaxing its own land constraint is not proof that the larger system faces no constraint at all.
Trade can also change the allocation of work within the district. A craft worker who previously spent part of the year cultivating a small plot may specialize and purchase food. Specialization can improve skill and reduce time lost between tasks. It also makes the worker more dependent on a continuing market. If buyers disappear before the household has accumulated reserves, a previously productive choice becomes dangerous. The gain from deeper specialization and the need for arrangements that sustain it arise together.
Better cultivation offers another route. A new crop may use a season or soil condition that existing crops leave underused. Improved drainage can make labor more effective. Better storage can reduce losses after harvest without changing the amount initially reaped. These mechanisms deserve separate attention because they imply different requirements. Knowledge about a crop must be tested locally. Drainage may require neighboring owners to cooperate. Storage may require buildings, pest control, and agreement about access. A recommendation is only the beginning of implementation.
An innovation can also redistribute risk. A crop with a higher typical yield may perform poorly in an unusual season. Adopting it across every plot could increase expected output while making failure more concentrated. Keeping a less productive crop can then be a reasonable form of insurance. An observer who sees only the difference in ordinary yields might call the household backward. The household may be responding to the fact that it cannot borrow food after a common failure and cannot afford the worst outcome.
Once insurance, transport, or reserves improve, that decision can change. Households can accept risks whose downside was previously intolerable. Their willingness to adopt need not reflect a sudden change in temperament. It can follow from a change in the consequences of failure. This is one reason a productive technique may spread unevenly through apparently similar farms. Differences in assets and obligations alter what is feasible, even when knowledge is shared.
Children, work, and the denominator
Population is itself a sequence of age groups rather than a single undifferentiated number. More surviving children initially increase needs relative to the work adults can provide. Later, those children can add to the labor force. Eventually they become older people requiring support. Output per resident is affected by this movement even if the productivity of a working adult remains unchanged. A temporary favorable balance between workers and dependents should not be mistaken for an indefinitely renewable source of growth.
Household choices respond to much more than wages. The cost of raising children, the availability of education, women's opportunities, health conditions, and expectations about old age can all affect the terms of family life. Their influence cannot be reduced to one universal rule assigning a fixed response to an increment of income. For the argument here, it is enough to recognize that population adjusts through choices and circumstances with substantial delays. Those delays create periods when output and population move at different speeds.
Migration can change the denominator faster. If working adults leave the district, output per remaining person may rise or fall depending on whose work departs and what assets become available. Money sent home can support consumption without increasing domestic production by the same amount. Incoming workers may fill scarce skills or compete for crowded land. Their effect is shaped by housing, rights, demand, and the capacity to create complementary work. Neither departure nor arrival carries a fixed arithmetic verdict about economic success.
Distribution enters again when the district reports an average. Households with land may obtain larger rents while laborers face expensive food. A higher average can therefore coexist with worsening conditions for a substantial group. Conversely, a redistribution that improves nutrition among poorer households can make their work and learning more effective, even if the immediate aggregate is little changed. Growth analysis must retain these pathways without pretending that an average is a household budget or that every distributional question can be settled by its effect on output.
The difference matters for public choice. A district deciding whether to finance a road might compare expected gains in output with construction costs. It must also decide whose land is used, whose taxes are collected, and who obtains access to the resulting exchange. If some households bear immediate costs while gains arrive elsewhere, opposition may reflect an accurate reading of the proposal. Calling all resistance an obstacle to growth would evade the actual choice. A workable project needs terms under which affected people can live with the commitment.
From an improvement to a continuing process
A successful rotation can raise the level of output. Once every suitable field uses it, its contribution to further growth diminishes unless it keeps improving or enables other changes. During adoption, output can rise year after year as the method spreads. When adoption is complete, that transitional growth ends. The higher level remains, provided the method and its supporting resources are maintained. A continuing rise in output per person requires additional improvements or an ongoing change in how resources are combined.
This distinction helps explain why an impressive local achievement need not establish an escape from recurring pressure. A district can gain a generation of higher consumption and still face renewed difficulty later. That is no reason to dismiss the generation's gain. It is a reason to ask what its members do with the opportunity: preserve reserves, develop skills, improve exchange, reduce avoidable mortality, or establish rules that support future investment. Temporary room can be used to build a less vulnerable position.
Maintaining the gain requires work that may be almost invisible in an account centered on novelty. The improved road needs repair. Drainage channels fill. Seed quality changes. People who know how to use tools must teach others. If each generation inherits a larger set of productive assets but makes no provision for their upkeep, gross production can conceal a weakening base. Sustained growth includes replacement and renewal as well as the first act of invention.
There is also a limit to treating greater effort as the answer. Adults can work longer only up to a point, and extra work can reduce rest, care, or health. Output per person might rise while life becomes more exhausting. Better methods offer the possibility of greater consumption with less sacrifice, but even they involve choices about how to use the time saved. Some households may prefer additional leisure or more care for children. A slower increase in measured output can then reflect a valued use of improved capacity.
The central problem of early growth is consequently neither a condemnation of population nor a story of effortless human ingenuity. It is the relation between additional claims and the means of meeting them. Land, energy, knowledge, institutions, and work all enter that relation. Under some conditions a larger harvest chiefly supports a larger population. Under others it supports a continuing expansion of what each person can produce and choose. Explaining the difference requires following the adjustments between these outcomes.
The hypothetical district ends this chapter with more options than it began with, but no guarantee. Its members can cultivate better, exchange more reliably, and undertake projects that reach beyond a season. They must still feed people while those projects mature and preserve the resources that make them useful. The next change will widen the possibilities further: power that can perform work beyond the annual harvest's immediate supply of human and animal effort. It will bring a new set of dependencies with it.
Notes and Sources
[1-1] Thomas Robert Malthus, *An Essay on the Principle of Population* (1798), especially chapters I–II, https://www.gutenberg.org/files/4239/4239-h/4239-h.htm. The argument is used as an original model, not as a universal empirical finding.
[1-2] Gregory Clark, *The Long March of History: Farm Laborers' Wages in England 1208–1850*, author-hosted manuscript, https://faculty.econ.ucdavis.edu/faculty/gclark/papers/long_march_of_history.pdf, especially the introduction and wage-construction sections. The linked file's title and period differ from the separately circulated September 2005 version. The evidence concerns reconstructed day wages and prices, with uneven geographical and chronological coverage.
[1-3] Stephen Broadberry, Bruce Campbell, Alexander Klein, Mark Overton, and Bas van Leeuwen, *British Economic Growth, 1270–1870: An Output-Based Approach*, December 2011 manuscript, University of Kent Discussion Paper 1203, https://www.econstor.eu/bitstream/10419/105558/1/1203.pdf. Introduction and agricultural reconstruction sections; estimates depend on representativeness assumptions and interpolation.