Skip to content
World in Transition

Is Money Necessary?

A Model of Coordination With and Without Currency

Model script: money-vs-moneyless-model.py Python · reproduces every number and chart on this page · what it is and how to run it

In brief

Every society must get a limited amount of work, materials and energy to the people who need what they can produce. Money is one way — goods go to whoever can pay — but not a logically required one: the problem can be solved from needs and capacities alone, and large parts of today’s economies (firms, public health systems, households, open projects) already work that way. This page compares both approaches in one transparent model, for the same people, needs and capacity, charging each for its own waste and coordination failures. Counting only goods and labour, the two come out close, with currency-based allocation ahead in about three-quarters of the 200,000 modelled scenarios by a median of under 4%. Counting who receives what — using the well-established finding that an extra unit of food, housing or care matters far more to someone with little than to someone with plenty — currency-free coordination comes out ahead in all 200,000 scenarios against a high-inequality money economy (Brazil- or Colombia-like), in 99.6% against a US-like economy, and in 88% even against the most equal economies that use money today.

The page models two regimes. Where capacity physically falls short of need, allocation by need shares the shortfall evenly. Where capacity is sufficient — as is plausible for many genuine needs once waste is removed — it meets every need with capacity to spare, while allocation by purchasing power, in the same model with the same capacity, still leaves 28–54% of people with some need unmet at 1.5 times the capacity needs require, depending on inequality.

Which comparison matters. Transition does not treat output totals — units produced, GDP-like aggregates — as a priority over human and societal wellbeing and harmony. The output-only comparison is included for completeness and honesty; the wellbeing comparison is the one that reflects Transition’s goals.

A note on certainty. Human societies are not machines; any model of them gives ranges, not certainties. What matters is whether a conclusion holds across the plausible range of assumptions. So every parameter below has a documented range, the model is run across all of them at once, and the page says plainly where the answer is robust and exactly which assumptions it hinges on where it is not.

1. Feasibility: is currency logically or technically necessary?

Short answer: no. Currency is one coordination mechanism among several; the mathematics of allocation does not require it, and non-monetary coordination already operates at very large scale inside modern economies. No entire modern society has yet run without it, so the claim is precise: money is not necessary in principle, and coordination without it already works at large scale within today’s economies.

1.1 The formal answer

Strip away the institutions and the problem is this: people have needs, and capacities (kinds of work, materials, energy, land, machine time) are finite. Choose what to make and who receives it so that needs are met as well as possible without exceeding any capacity. That is a linear or convex program:

maximise   ∑i ∑g wig(xig)
subject to   ∑g arg yg ≤ Kr  for every capacity r,    ∑i xig ≤ yg  for every good g,    x ≥ 0

xig: amount of good g reaching person i · yg: amount of g produced · arg: capacity r needed per unit of g · Kr: available capacity r · wig: a concave function of how well need i g is met (diminishing benefit as the need is filled).

No price, wage or budget appears in it. Leonid Kantorovich formulated this kind of problem in 1939 to plan production in a plywood trust, and invented linear programming to solve it (Nobel memorial prize, 1975).[1] Solving it also yields what he called “objectively determined valuations”: computed numbers showing how much an extra unit of each scarce capacity would help, which no one has to own, earn or spend. The script with this page solves such a problem with 360,206 variables (10,000 people, 60 goods, 6 capacities) in under two minutes on a laptop, meeting every need to at least 74.8% when capacity falls short. Computing an allocation is not the obstacle; gathering accurate information about needs and capacities is.

1.2 The calculation debate — and what it was actually about

The classic objection came from Ludwig von Mises (1920), who argued that without market prices for the means of production a socialist economy could not calculate rationally at all.[2] Oskar Lange (1936) replied that a planning board could reach the same allocation by trial and error, adjusting accounting values in response to shortages and surpluses.[3] Friedrich Hayek’s deeper reply (1945) still matters: the knowledge needed to allocate well is dispersed, local and often tacit, and cannot be collected into one centre; prices, he argued, summarise it.[4]

Two points about that debate are often lost.

First, the Mises–Hayek critique was aimed chiefly at central planning — a single authority collecting information and issuing instructions. The historical record bears out much of it: centrally planned economies suffered chronic shortages alongside unwanted surpluses, which János Kornai documented in detail as a systemic “economics of shortage”.[5] Those failures were real. The alternative described here is not central planning. It is polycentric coordination: local nodes (a neighbourhood, a town, a clinic network) matching local needs to local capacity with direct feedback signals — stock levels, waiting lists, usage data — and exchanging information and surpluses with neighbouring nodes and regional federations, with no central bank or central allocator (see the Financial Commons workstream and Post-Currency Foundations). Ostrom showed that overlapping, semi-autonomous centres can govern complex shared resources well where markets and states both fail.[6] Detailed decentralised proposals include Albert and Hahnel’s participatory economics[7] and Pat Devine’s “negotiated coordination”;[8] Cockshott and Cottrell argued that modern computing removes the computational objection;[9] and Evgeny Morozov argues that digital feedback infrastructure changes what non-market coordination can do.[10]

Second, keeping decisions local answers part of Hayek’s objection, not all of it. The people with local and tacit knowledge decide, no centre has to know everything, and waiting lists carry information about needs directly rather than filtered through who can pay. What still applies: aggregation (nodes must learn about surpluses and shortages elsewhere, and summaries lose information), discovery (markets reward trying new products; a needs-based system must build its own channels), and incentives (people must keep contributing without income pressure). The model gives each its own cost term.

1.3 A working demonstration of decentralised feedback

The script includes a small simulation: 12 local nodes on a ring, 20 goods, random needs each period, and capacity slightly below average need. Each node sees only its own capacity, stocks and waiting list, plus its two neighbours’ waiting lists. It meets local needs, passes leftovers to neighbours (losing 10% in transfer) and shifts work toward goods with waiting lists. No prices, no central allocator. An “oracle” that knows every node’s needs in advance and optimises the whole system met 100% of needs; the decentralised network met 92.8%. This toy model shows that simple local feedback comes close to the full-information optimum, and that the gap is a real cost — which the comparison model charges for.

1.4 Two regimes: shortage and sufficiency

The demonstration in section 1.1 deliberately poses the hard case: capacity physically below total need, so that some needs go partly unmet whatever the mechanism. That boundary condition stays in the analysis, because physical limits do not disappear with money. It is shortage of a physical kind — not enough work, material or energy — not the artificial scarcity of goods held back to protect prices. But for many genuine needs the opposite case is plausibly the relevant one: capacity at or above need. So the script runs the same construction (2,000 people, 50 goods, 4 capacities, needs drawn as in section 1.1, its own random seed) at capacity levels from 0.8 to 3 times what meeting every need requires, with every capacity scaled by the same factor, and allocates the same capacity among the same people in two ways.

  • Needs-based (no prices): the linear program of section 1.1 — first raise the share of need that everyone receives, then meet as much total need as possible.
  • Gated by purchasing power: a budget-constrained market. Each capacity has one market-clearing price, and a good’s price is the value of the capacity it uses. Incomes are log-normal, at the Gini levels of the three money economies in section 5 (0.26, 0.35 and 0.525), drawn independently of needs. Each person spends on needs first, buying the same share of each of their needs until their income runs out; income left after all their needs are met goes to uses beyond need. Prices adjust until every capacity is fully bought.

The market side is given every benefit of the doubt: full employment, everything produced is sold, no market power, no money-handling overhead, and people spend on needs before anything else. In other words L1, L2 and L4–L7 of section 3 are all set to zero, and so are the coordination costs of section 4: this isolates the allocation rule alone. One simplification cuts the other way: there is no non-market provision on the money side (public health care, food assistance), which in real money economies closes part of the gap. A reference case with equal incomes shows what unequal needs alone do under a budget rule.

Same needs, same capacity: needs-based allocation vs allocation by purchasing power
Capacity (multiple of total need)Needs-based (no prices)Purchasing power: share of people with at least one need not fully met
Every need met toCapacity spareEqual incomesLow inequality (Gini 0.26)US-like (0.35)High inequality (0.525)
Shortage
0.8×80.0%0%84.0%72.9%73.0%75.1%
Sufficiency
1.0×100%0%48.6%57.0%61.2%68.6%
1.2×100%16.7%15.4%42.9%51.4%62.6%
1.5×100%33.3%1.1%27.8%37.5%53.9%
2.0×100%50.0%0%11.5%22.9%42.3%
3.0×100%66.7%0%2.3%8.6%27.5%
Allocation by purchasing power at 1.5× capacity (needs-based: every need met in full, 33.3% of capacity spare)
Income distributionPeople with an unmet needNeed met, lowest-income 10%Total need metCapacity to uses beyond needCapacity that would close every gap
Equal incomes1.1%–99.9%33.4%0.1%
Low inequality (Gini 0.26)27.8%63.1%92.9%38.2%4.9%
US-like (Gini 0.35)37.5%43.2%86.9%42.2%8.9%
High inequality (Gini 0.525)53.9%17.4%72.6%51.8%18.4%
Two line charts against capacity from 0.8 to 3 times total need. Needs-based coordination meets 80% of every need at 0.8 times and 100% from 1 times onward. Under allocation by purchasing power, total need met rises slowly with capacity and stays below 100%: at 1.5 times capacity, 92.9% with low inequality, 86.9% US-like and 72.6% with high inequality. For the lowest-income tenth, the share of need met at 1.5 times capacity is 63.1%, 43.2% and 17.4%, and at 3 times still 95.3%, 80.6% and 34.8%.
Figure 1. Same needs, same capacity, from shortage (grey band) to sufficiency. Vermillion: needs-based coordination, no prices. Blue: allocation by purchasing power under a budget-constrained market at three levels of income inequality, with full employment and no market power. Top: total need met. Bottom: share of need met for the tenth of people with the lowest incomes.

Shortage. At 0.8× capacity the needs-based allocation meets every need to 80%: it spreads the physical shortfall evenly, so no one is fully served but no one is left far behind. (Counting “people with an unmet need” therefore says little in shortage; the lowest share met is the relevant measure.) Allocation by purchasing power meets less of total need (70.7%, 65.4% and 54.1% from low to high inequality), the lowest-income tenth receive 33.8%, 23.0% and 9.3% of their need, and 12–33% of capacity still goes to uses beyond need.

Sufficiency. From 1.0× onward the needs-based allocation meets every need in full and leaves the rest of the capacity spare (16.7% at 1.2×, 33.3% at 1.5×). Allocation by purchasing power leaves needs unmet even though the capacity to meet them exists and is in use: at 1.5×, 28–54% of people have at least one need not fully met, while 38–52% of capacity goes to uses beyond need — and closing every remaining gap would take only 4.9–18.4% of capacity. Those uses are not worthless to the people who choose them; the point is the order of priority. Even with equal incomes, a budget rule fails people whose needs are larger than average: 48.6% fall short at 1.0× and 15.4% at 1.2×. The model reproduces in miniature what the evidence in section 3 shows at scale: food insecurity alongside large food surpluses, and homelessness alongside vacant homes (L2).[46][47][48][49]

Where the result is weaker. With a lot of slack and low inequality, the money side comes close: at 3× capacity only 2.3% of people fall short in the low-inequality case and 99.4% of total need is met (US-like: 8.6% and 97.7%). With high inequality the gap persists (27.5% of people short at 3×). So under sufficiency, how much allocation by purchasing power leaves unmet depends mainly on inequality and on how far capacity exceeds need. The levels used here are illustrative; the model is not calibrated to the ratio of capacity to need for any good in any real economy.

What sufficiency changes on the currency-free side. Spare capacity absorbs coordination errors. Re-running the decentralised network of section 1.3 (new random seed) with spare capacity, the share of needs met rises from 92.6% at 0.95× mean need to 99.8% at 1.10× and 100.0% at 1.25×, against 100% for the full-information optimum in each case. Local mismatch and cross-node aggregation error (Clocal, Cagg) should therefore matter less for goods in sufficiency; the comparison model keeps their ranges unchanged, so for those goods it is, if anything, conservative. Slack is not waste in a needs-based system: it can hold buffer stocks, become shorter working hours, or go to uses people decide on openly.

Which regime applies varies by good. Many genuine needs are plausibly in or near sufficiency once waste, duplication and premature replacement (L5–L7) are removed: basic food in the United States, where 31% of the supply goes unsold or uneaten,[46] and plausibly in many other regions; housing stock in the United States, where “other vacant” homes outnumber homeless people roughly four to one (not all are habitable or located where the need is),[48][49] and plausibly in many other countries; clothing, where worldwide the number of times a garment is worn fell by 36% in fifteen years while production roughly doubled;[73] household durables, many of which are replaced while still working;[61][62] and much of education and information, which can be copied with little extra work or material once made. Other things are physically scarce and plausibly stay so for some time: specialist care, where skilled people take years to train; some materials; and land in some places. For those, the shortage case of section 1.1 is the relevant one. None of this is proven good by good; it is the reason both regimes are modelled.

1.5 Coordination without money in practice

The strongest evidence that currency is not necessary is how much provisioning already happens without it. Examples are grouped by whether they are present or past, and arose spontaneously or were deliberately organised, each with its limits.

Real-world provisioning in which money is absent from the allocation
ExampleWhat it showsLimits
Present · deliberately organised
Internal planning in giant firms (Walmart, Amazon) Inside firms, goods move by instruction and data, not purchase; much of a modern economy’s coordination happens this way.[11][12] Amazon forecasts demand for over 400 million products daily and places stock algorithmically.[13] Phillips and Rozworski call such firms planned economies larger than many countries.[14] Hierarchical; money still governs the firm’s dealings with customers and suppliers.
Public health care at the point of use (e.g. the UK NHS) “Access to NHS services is based on clinical need, not an individual’s ability to pay.”[15] Tens of millions of people a year, with no purchase at the point of care. Funded and staffed through money; waiting lists show capacity limits.
Wikipedia and open-source software (Linux) Large-scale production coordinated by peers, without markets or managerial command.[16] Tens of thousands of active editors each month;[17] over 15,600 Linux kernel developers from 1,400+ organisations.[18] Open to everyone, without purchase. Information goods are cheap to copy; most kernel developers are paid by employers.
Blood and organ donation Vital, perishable goods matched to medical need: 79 countries collect over 90% of their blood from voluntary unpaid donors, the safest donor group according to WHO.[19] A narrow class of goods with strong altruistic norms.
Food sharing and rescue (e.g. Foodsharing.de) Over 100,000 volunteers in Germany, Austria and Switzerland collect surplus food from thousands of businesses and pass it on without purchase.[20] Depends on surplus from the money economy; small relative to total food supply.
Time banks, tool libraries, libraries of things Services and equipment shared by local agreement rather than price; time banks reach people outside formal volunteering.[21] Small scale; time banks keep an hour-for-hour ledger.
Present · spontaneous
Households and unpaid care The largest non-monetary economy: 16.4 billion hours of unpaid care work every day, equivalent to 2 billion full-time workers.[22] Unequally shared (women do 76%); small groups with close ties.
Mutual aid, including pandemic mutual aid About 4,300 UK mutual-aid groups with an estimated three million participants formed within weeks in 2020, delivering food and medicine by need.[23] Many faded as the emergency eased.
Disaster response Disaster research finds panic and predation rare and spontaneous cooperation the norm, from San Francisco 1906 to New Orleans 2005.[24] Temporary; formal relief is money-funded.
Village and Indigenous reciprocity (minga, gotong royong, harambee) Collective labour and provisioning by reciprocity across Latin America, Southeast Asia and East Africa today (see Culture and Transition). Local, and embedded in economies that also use money.
Past · spontaneous
Hunter-gatherer sharing economies For most of human history, food and tools were allocated by sharing norms, with active levelling of would-be accumulators.[25][26] Small bands; simple technology.
Gift and reciprocity economies (e.g. Haudenosaunee) Durable confederations distributing goods by reciprocity and redistribution (see Documented Alternatives). Pre-industrial; interpretations vary.
Past · deliberately organised
The Inca state (mit’a labour, storehouses) An empire of millions largely without markets or money: labour obligations filled state storehouses that supplied armies, workers and famine relief.[27] Strongly hierarchical and coercive — scale without money, not egalitarian coordination.
Monastic communities For fifteen centuries the Rule of St Benedict has required that all things be held in common and distributed according to need.[28] Small, voluntary, religiously bound, governed by an abbot.
Kibbutzim For decades, Israeli kibbutzim ran internally without wages or money, allocating housing, food, care and education by need. By 2010, 75% of non-religious kibbutzim had moved to differential salaries;[29] they always traded in money.
Revolutionary Catalonia and Aragon, 1936–37 Many rural collectives abolished money internally and distributed by need or ration.[30] Contested: wartime conditions, coercion in places, labour-discipline problems;[31] cities kept money; suppressed within about a year.
Argentina’s barter networks, 2001–02 Millions of people organised thousands of exchange fairs within months of the financial collapse.[32] Used paper tokens that collapsed through over-issue and counterfeiting — a lesson about tokens, not need-based allocation.
Mondragon cooperatives (present) A worker-owned federation of some 70,000 people: democratic coordination at industrial scale.[33] Uses money and markets throughout — partial evidence only.

Together, these cases establish feasibility with confidence: allocation by need, without purchase, works reliably for vital goods, at enormous scale, for complex production and under stress. Their limits are as consistent: the largest present-day examples sit inside a money economy, and the historical ones that ran without money were small or hierarchical. The open question is not whether coordination without currency works, but how its costs at full scale compare with those of currency — which the model estimates.

2. The model

The model compares two ways of allocating the same things: N people with needs and a productive capacity K (work, materials, energy, equipment). Only the allocation mechanism differs.

  • Currency-based allocation: goods go to whoever can pay; production follows what sells; a share of the workforce handles money itself; and the market makes its own coordination errors (duplication, premature replacement, unsold and returned goods).
  • Currency-free polycentric coordination: local nodes match local needs to local capacity using direct signals (stocks, waiting lists, usage data) and exchange surpluses and summary information with neighbouring nodes and regional federations; goods are open to everyone who needs them, without purchase.

Every loss and cost is a fraction of K that fails to become wellbeing. With no losses at all, welfare would be 1 (100% of K). Each system delivers:

Wcur = (1 − L1)(1 − L2)(1 − L3)(1 − L4)(1 − L5)(1 − L6)(1 − L7)
Wfree = (1 − Ccoord)(1 − Cfed)(1 − Clocal)(1 − Cagg)(1 − Cdyn)(1 − Cinc)(1 − Cerr)
Symbols
SymbolMeaning
L1Money-handling overhead, net of what any system needs
L2Access-gating: capacity idle although needed
L3Allocation by income instead of need (wellbeing-equivalent)
L4Market power: markups and artificial scarcity
L5Competitive duplication and advertising-induced demand
L6Premature obsolescence
L7Unsold, destroyed and returned goods; failed ventures
CcoordLocal coordination labour
CfedFederation (cross-node) labour
ClocalLocal mismatch within a node
CaggCross-node aggregation error (Hayek’s problem, decentralised)
CdynSlower discovery and innovation
CincParticipation shortfall without income pressure
CerrNeed-assessment error (wellbeing-equivalent)
εCurvature of wellbeing (how fast extra units matter less)
GGini coefficient of household resources

The break-even condition

Currency-free coordination is more efficient exactly when Wfree > Wcur. Taking logarithms gives an exact condition, and a simple approximation when all terms are small:

∑k −ln(1 − Lk)  >  ∑j −ln(1 − Cj)   ≈   L1 + … + L7  >  Ccoord + Cfed + Clocal + Cagg + Cdyn + Cinc + Cerr

The output-only comparison counts goods and labour, ignoring who receives them (L1, L2, L4–L7 against the six output costs); each side is charged for its own coordination errors. The wellbeing comparison adds L3 and, symmetrically, Cerr, both measured with the same wellbeing function.

Why allocation by income loses wellbeing (L3)

Suppose everyone has the same wellbeing function u, and u is concave: each extra unit helps a little less than the one before. A standard form is u(c) = c1−ε/(1−ε), with u(c) = ln c when ε = 1. For any fixed total amount of goods C shared among N people, Jensen’s inequality gives

(1/N) ∑i u(ci)  ≤  u(C/N),    with equality only when every ci = C/N.

So, when needs are equal, allocating by need yields strictly more total wellbeing than allocating in proportion to an unequal income distribution, from the same goods. (When needs differ, the same argument applies to goods per unit of need.) The size of the gap is the Atkinson index Aε: the share of all goods that a need-matched allocation could give up and still produce the same total wellbeing as the unequal one.[34] For a log-normal distribution with Gini G it has a closed form:

σ = √2 · Φ−1((1 + G)/2),     L3 = Aε = 1 − exp(−εσ2/2)

Φ−1 is the inverse of the standard normal distribution. The same formula, with the need-assessment error s in place of σ, gives Cerr. The script checks the closed form against a distribution with a heavy (Pareto) top tail at the same Gini; the difference is two percentage points or less.

The assumption and its support. Diminishing marginal wellbeing is among the best-supported findings in the field. Life evaluation rises with the logarithm of income (ε ≈ 1), with no satiation.[35] Kahneman and Deaton found day-to-day emotional wellbeing plateaued above about $75,000;[36] Killingsworth found it kept rising log-linearly;[37] a joint “adversarial collaboration” resolved the conflict: for most people it keeps rising with log income, while for the least happy fifth it levels off.[38] Either way the curve is concave, which is all the argument needs. Estimates of ε from wellbeing data centre on 1.26 (Layard, Mayraz & Nickell);[39] a meta-analysis of 1,711 estimates finds 1.61 (95% interval 1.18–2.05).[40] The model uses 0.7–2.0, centred on 1.3, and conservatively leaves out relative-status effects, which would make inequality costlier still.

3. Where currency-based allocation loses capacity

Each channel uses published data. Most detailed figures exist for the United States, and the model is calibrated to them; where international data exist they are noted. Wherever a function would still be needed without money, it is subtracted rather than counted as waste.

L1 — Money-handling overhead: median 7.3% (90% range 5.9–9.0%)

Gross money-handling overhead, United States (share of total labour)
ComponentEvidenceRange used
Finance and insurance6.73 million of 167.8 million jobs in 2023 (4.0%);[41] the sector’s share of GDP is higher, about 7.4%[42]3.5–4.5%
Excess health-care administrationAdministration took 34.2% of US health spending in 2017, against 17.0% in Canada[43] — about 3.3% of GDP; reduced to avoid double-counting insurers1.5–3.3%
Tax compliance7.9 billion hours in 2024, equal to 3.8 million full-time workers (2.2% of jobs)[44]1.8–2.6%
Advertising1.3–2% of GDP over the past 20 years (St. Louis Fed)[45]1.3–2.0%
Other money handlingBilling and collections outside health care, payment handling, pricing, transactional legal and real-estate work (assumption)0.5–3.0%
Gross totalMedian 11.8% (90% range 10.7–13.0%)
Residual still needed without moneyRecord-keeping, risk pooling, product information and logistics accounting persist in any system (assumption)25–50% kept

After subtracting the residual, L1 is about 7%.

L2 — Access-gating: median 1.6% (90% range 1.1–2.2%)

Capacity that already exists goes unused while people need it. In 2023, 31% of the US food supply — 73.9 million tons, worth $382 billion or 1.4% of GDP — went unsold or uneaten,[46] in a year when 13.5% of US households (18 million) were food insecure.[47] The 2020 Census counted 3.36 million “other vacant” homes (not for sale, rent or seasonal use) out of 140.5 million,[48] while 771,480 people were homeless on a single night in January 2024.[49] And labour and equipment sit idle in downturns: from 1980 to 2024, US output averaged about 1.2% below the Congressional Budget Office’s estimate of potential (calculated for this page from the CBO series).[50] Only part of each is caused by lack of purchasing power (much food waste happens in homes), so the model counts 10–40% of surplus food, 10–50% of “other vacant” homes and 0.5–2% idle capacity.

L3 — Allocation by income, not need: median 23.8% (90% range 16.1–32.4%)

Using the formula above with a Gini of 0.30–0.40 and ε between 0.7 and 2.0. The US disposable-income Gini is about 0.39 (OECD);[51] consumption inequality is lower,[52][53] hence the range. This is the largest channel, and the most dependent on assumptions. It is also not unique to currency: money economies that redistribute strongly reduce it. Results are therefore shown against three kinds of money economy: high-inequality (Gini 0.50–0.55; World Bank estimates for 2024 put Brazil at 0.50 and Colombia at 0.54, and South Africa has been higher still, around 0.60–0.65 in 2005–2014 surveys),[54] where L3 roughly doubles to a median of 49%; US-like; and low-inequality (Gini 0.26, close to the most equal OECD countries), where it is roughly half as large.

L4 — Market power: median 4.2% (90% range 2.3–6.8%)

Average markups of US firms over marginal cost rose from 21% in 1980 to 61% in 2016 (De Loecker, Eeckhout & Unger).[55] Edmond, Midrigan and Xu estimate that removing markup distortions would raise welfare by the equivalent of 7.5% of consumption, about a third of it from misallocation across firms;[56] Baqaee and Farhi estimate that eliminating markup misallocation would raise aggregate productivity by about 15–20%, with substantial data imputation.[57] The model takes 1.5–8% (central 3.5%), below the larger figures because part of the markup cost concerns how much people work, handled on the other side in Cinc. Deliberate destruction of goods to protect prices (see Manufactured Scarcity) is not separately added.

L5–L7 — The market’s own coordination errors

The market is charged for the same kinds of error as the currency-free system. These are the least precisely measured channels on the currency side, so ranges are wide and central values modest.

L5 — duplication and advertising-induced demand: median 3.8% (90% range 2.1–5.8%). An estimated macroeconomic model of the US finds that advertising raises aggregate consumption and hours worked, leaving households worse off because the extra work outweighs the extra consumption.[58][59] Lacking a verified point estimate, the model assumes 0.5–5% of production serves advertising-induced wants rather than needs. Parallel product lines and infrastructure for functionally equivalent goods add 0–3%: theory shows free entry can be excessive,[60] but the best recent estimate finds the cost of inefficient entry small,[56] and duplication also buys variety and resilience — hence a low central value of 1%.

L6 — premature obsolescence: median 1.2% (90% range 0.6–1.8%). In Germany, the share of large household appliances replaced within five years because of a defect rose from 3.5% in 2004 to 8.3% in 2012–13 (German Environment Agency); the same study found that most replacements are driven by the wish for a newer device rather than failure.[61] In Austria, most replaced mobile phones still worked at least partly.[62] The European Environmental Bureau estimates that extending the life of just four product types in the EU by one year would save about 4 million tonnes of CO2 a year.[63] The model applies an achievable lifetime extension of 5–50% (central 20%) to US spending on durable goods and clothing (about 9.5% of GDP in 2023),[64] and subtracts 20–50% of the saving for the repair labour and sturdier design that longer lifetimes require. See also Designed to Fail.

L7 — unsold, destroyed and returned goods: median 1.0% (90% range 0.5–1.7%). US consumers returned $685 billion of merchandise in 2024, 13.2% of retail sales;[65] an estimated 9.5 billion pounds of returns went to landfill in 2022.[66] Before France banned the practice in 2020, about €630 million of unsold non-food goods were destroyed there each year.[67] Most returns are resold, so only processing, transport, disposal and resources sunk in failed ventures are counted (0.3–2%); food waste and idle capacity, already in L2, are excluded.

Horizontal bar chart of median losses with 90% ranges. Currency-based losses: money-handling overhead 7.3%, idle capacity 1.6%, allocation by income instead of need 23.8% (wellbeing-equivalent), market power 4.2%, duplication and advertising-induced demand 3.8%, premature obsolescence 1.2%, unsold, returned and failed output 1.0%. Currency-free costs: local coordination labour 4.5%, federation labour 1.9%, local mismatch 3.8%, cross-node aggregation error 2.7%, slower discovery 3.1%, participation shortfall 5.9%, need-assessment error 2.9% (wellbeing-equivalent).
Figure 2. Each loss channel of currency-based allocation and each cost of currency-free coordination, as a share of total productive capacity: median and 90% range across 200,000 Monte Carlo draws. Starred items are wellbeing-equivalent.

4. What coordinating without currency would cost

The currency-free system is charged six output costs and one wellbeing cost. Ranges are wide because there is no full-scale modern example to measure.

Costs charged to currency-free polycentric coordination (share of total capacity)
CostRange (low – central – high)Basis
Ccoord local coordination labour2 – 4 – 8%People’s time gathering needs data, matching, deliberating and keeping open records. Computation is cheap (section 1); this is the human cost.
Cfed federation labour0.5 – 1.5 – 4%Exchanging node-level summaries and surpluses between nodes and regions.
Clocal local mismatch1 – 3 – 8%Wrong goods or timing within a node, counted in full (the market’s own mismatch is charged separately as L7). Kept lower than under central planning by local and tacit knowledge, but not zero.
Cagg cross-node aggregation error0.5 – 2 – 6%The part of Hayek’s knowledge problem that survives decentralisation. The high end reflects the shortages and surpluses of planned economies (Kornai).[5] The toy simulation in section 1.3 fell about 7% short of a perfect-information optimum; the combined range for local and cross-node mismatch (1.5–14%) brackets that figure.
Cdyn slower discovery0 – 2 – 8%Markets reward trying new products and methods; a needs-based system must build its own channels. Partly offset because much basic innovation is already publicly funded or open source.
Cinc participation shortfall0 – 4 – 15%See below.
Cerr need-assessment errorlog-error s = 0.08 – 0.35Needs are recognised imperfectly; s = 0.35 alone is equivalent to a Gini of about 0.20.

Incentives. The closest experimental evidence comes from guaranteed-income studies, which remove part — not all — of the pressure to earn. In the largest US trial, $1,000 a month for three years reduced recipients’ work by 1.3–1.4 hours a week, about 5–6% of earnings.[68] Alaska’s universal dividend produced no fall in employment and a rise in part-time work.[69] Psychology adds that paying for activities people value can crowd out intrinsic motivation,[70][71] (see the evidence on motivation without money). None of this measures a society without money, so the range runs to 15%, more than double the guaranteed-income effect.

Not modelled on either side: the one-time costs of transition and of trading with economies that still use money; and, in the other direction, the documented costs of financial insecurity itself, such as the measurable loss of cognitive capacity under scarcity.[72]

5. Results: break-even and sensitivity

Every parameter is drawn 200,000 times from its range (fixed seed, exactly reproducible).

Share of 200,000 scenarios in which currency-free coordination comes out ahead, by comparison
ComparisonCurrency-free aheadMedian advantage90% range
Wellbeing (the comparison that reflects Transition’s goals)
vs a high-inequality money economy (Gini 0.50–0.55, Brazil- or Colombia-like)100%+82.8%+45.2% to +136.3%
vs a US-like money economy (Gini 0.30–0.40)99.6%+22.1%+7.5% to +40.5%
vs a low-inequality money economy (Gini 0.26)88.5%+7.7%−2.7% to +18.6%
Output only (included for completeness)
Goods and labour, ignoring who receives them22.9%−3.6%−11.8% to +4.2%
Four histograms of the advantage of currency-free coordination across 200,000 scenarios. Output only: centred a little below zero; currency-based ahead in 77% of scenarios. Wellbeing against a high-inequality money economy: entirely above zero, centred near +80%; currency-free ahead in 100%. Wellbeing against a US-like economy: almost entirely above zero; currency-free ahead in 99.6%. Wellbeing against a low-inequality money economy: mostly above zero; currency-free ahead in 88.5%.
Figure 3. Distribution of the net advantage of currency-free coordination across 200,000 modelled scenarios within the stated parameter ranges, for output only and for wellbeing against high-inequality, US-like and low-inequality money economies. Bars to the right of zero (vermillion) are scenarios where currency-free coordination comes out ahead; bars to the left (blue), scenarios where currency-based allocation does. 2.1% of high-inequality scenarios lie beyond +150% and are not drawn.

What the results say

The wellbeing comparisons reflect Transition’s goals, which do not put output totals ahead of human and societal wellbeing; they are reported first.

On wellbeing, currency-free coordination is robustly ahead wherever inequality is at or above US levels: in all 200,000 scenarios against a high-inequality economy, and in 99.6% against a US-like one (median advantage 22%). The reason is structural: allocation by purchasing power sends much of what is produced to people for whom it adds little, while people with unmet needs go without. In the high-inequality case that loss alone is about half of all capacity in wellbeing terms (L3 median 49%), so the size of the advantage there (median 83%) is less certain than its direction: at such Gini levels the log-normal shape assumed for the income distribution matters more. Against a US-like economy at central values, currency-free coordination stays ahead as long as the participation shortfall stays below 24.7% (over four times the guaranteed-income effect), all six output costs together stay below 33.7% (central 15.4%), and inequality exceeds a Gini of 0.11 — far below any OECD country.

Against the most equal money economies (Gini 0.26), currency-free coordination is ahead in 88.5% of scenarios, by a median of 7.7%. It falls behind where the participation shortfall and coordination costs are both high — questions only practice at growing scale can settle.

On output alone, the two systems are close, with currency-based allocation usually ahead. The output losses charged to the currency side (L1, L2, L4–L7) total about 19% of capacity in the median scenario; the costs of coordinating without money, about 23%. At central values currency-free coordination is ahead by 0.5%, but across all scenarios currency-based allocation leads in 77%, by a median of 3.6%: the model does not support a claim that removing money would, by itself, produce more goods from the same resources. This result turns above all on how much less work gets done without income pressure (rank correlation −0.66); at central values the two break even at a shortfall of 4.5%, close to the effect measured in the largest guaranteed-income trial.

Two tornado charts showing the advantage of currency-free coordination when one parameter at a time moves to its 5th or 95th percentile. Wellbeing comparison, central +27.5%: largest swings from wellbeing curvature (+18% to +39%), inequality (+21% to +35%) and participation shortfall (+31% to +17%); no single parameter reaches break-even. Output-only comparison, central +0.5%: participation shortfall swings it from +2.9% to -8.0%, followed by slower discovery, local mismatch, market power, local coordination labour, aggregation error and advertising-induced demand.
Figure 4. Sensitivity: each bar shows the result when one parameter moves to its 5th (vermillion) or 95th (blue) percentile with all others at central values. The dashed line is break-even. Top: wellbeing comparison. Bottom: output only.

Which assumptions matter most. The wellbeing result is most sensitive to the curvature of wellbeing ε (rank correlation +0.58), the participation shortfall (−0.41) and the degree of inequality (+0.41), followed by need-assessment error (−0.27) and slower discovery (−0.20). No single parameter, moved alone to the edge of its plausible range, reverses it. The output result is dominated by the participation shortfall, then slower discovery (−0.34), local mismatch (−0.29) and market power (+0.27). The data-based overhead components barely move either result: each is measured reasonably well, so its range is narrow.

Why this model does not reproduce the 15–17-hour week

This is a deliberately narrow analysis of one systemic question: is currency necessary for a modern society to function, and what are the trade-offs between currency-based and currency-free coordination of today’s pattern of production? Within that scope, once residual functions and new coordination and participation costs are counted, the hours needed to produce today’s useful output barely change (median about 40 hours a week for today’s workers, against 38.6 now; 90% range 37–44).

The site’s work-week estimate of 15–17 hours rests on factors outside this model’s scope, chiefly: goods and services made to the highest technically practical quality, designed to last and to be easily repaired and upgraded, so that production needs taper off as needs are met; the end of the runaway produce–consume–discard–replace cycle and of artificially sustained high production and consumption; necessary work shared across all able adults; and the disappearance of whole categories of work, money handling among them (see the projections page for the full list). This is a difference in scope, not a contradiction: the model holds today’s pattern of production largely fixed, so it neither tests nor confirms those factors.

6. What this does and does not show

Well supported: currency is not logically or technically necessary to allocate a modern economy’s capacity to needs; coordination without money already works reliably at large scale inside today’s economies; and allocation by purchasing power carries a wellbeing cost that exceeds the costs of polycentric coordination in all 200,000 modelled scenarios (within the stated parameter ranges) against a high-inequality economy, over 99% against a US-like economy, and 88% against the most equal money economies. Where capacity is sufficient, allocation by need meets every need with capacity to spare, while allocation by purchasing power leaves needs unmet wherever incomes are unequal relative to needs — in the model, and in the record of food insecurity alongside food surpluses and homelessness alongside vacant homes.

Not shown: that an entire modern society can run without currency (the model estimates; it does not demonstrate); that a currency-free system would produce more output (on output alone, currency-based allocation leads in about three-quarters of scenarios); or anything about the 15–17-hour week, which rests on factors outside this model’s scope. Nor does it show which goods are in sufficiency, and by how much, in any real economy: the two-regimes demonstration is illustrative, not calibrated.

Main limitations:

  • The currency-free costs and the market’s own coordination errors (L5–L7) rest on judgement informed by evidence, not full-scale measurement; both ranges are wide.
  • Calibration is mostly to US data; elsewhere overhead (especially health administration) is smaller and inequality differs.
  • The model is static: it compares steady states, not the path between them, and treats innovation as a level effect.
  • It assumes a common wellbeing function, and counts neither the value some place on market choice nor the harms of financial insecurity.
  • The polycentric simulation illustrates the mechanism; it does not calibrate it.
  • The two-regimes demonstration (section 1.4) scales every capacity by the same factor, gives the market side favourable assumptions and no public provision, and leaves out the costs of sections 3 and 4: it isolates the allocation rule.

7. Try it yourself

Nothing on this page has to be taken on trust. The whole model is one file you can download, read and run.

Download the model script money-vs-moneyless-model.py · Python · about 59 KB
What it is
A single Python script, about 1,080 lines with comments, that reproduces every number and all four charts on this page. It is plain text: open it in any text editor to read it before running it. Every parameter is listed near the top with its range and its source, or is marked as an assumption.
What you need
Python 3.9 or later, with the numpy and matplotlib packages. The scipy package is needed for the demonstrations in section 1 (the linear program, the decentralised network and the two regimes); if it is missing, the script skips them with a message and runs everything else. To install all three: 
python3 -m pip install numpy matplotlib scipy
How to run it
From the folder where you saved it, one command: 
python3 money-vs-moneyless-model.py
Add --no-charts to print the results without drawing the charts. A full run takes about two minutes on an ordinary laptop; most of that time is the 360,206-variable linear program.
What it outputs
Printed results, in order: the feasibility demonstration (section 1.1) and the decentralised-feedback simulation (1.3); the check of the L3 formula against a heavy-tailed distribution; central estimates of every channel and cost; the Monte Carlo results for all four comparisons (section 5); rank correlations, the one-at-a-time sensitivity analysis and the break-even values; the work-week consistency check; and the two-regimes demonstration (section 1.4): the needs-based allocation and the budget-constrained market at each capacity level and inequality level, and the decentralised network re-run with spare capacity. It then writes Figures 1–4 as SVG files to a new folder, money-model-charts, in the folder you ran it from. The random seed is fixed, so every run gives the same numbers.
Open to read and change
Change any range, central value or assumption you disagree with, re-run, and see whether the conclusions survive. That is what the script is for.
Licence
GNU General Public License, version 3 or later (full text). Anyone may use, study, share and modify the script, for any purpose. If you share a modified version, it must stay under the same licence, so improvements remain open to everyone.

The oligarchic capture model explains how allocation by purchasing power concentrates wealth over time — the dynamic that keeps L3 large and makes redistribution hard to sustain. Oligarchic and egalitarian attractors shows both configurations are stable; this page asks which uses the same resources better. Post-Currency Foundations sets out the coordination whose costs appear here as the C terms, and Documented Alternatives expands on section 1.5.

Prefer a shorter read without equations? The plain-language version explains the same model, numbers and limits in everyday language.

References

  1. L. V. Kantorovich, “Mathematical Methods of Organizing and Planning Production,” Management Science 6(4), 1960 (English translation of the 1939 original). ↩
  2. Ludwig von Mises, Economic Calculation in the Socialist Commonwealth (1920; English translation 1935). ↩
  3. Oskar Lange, “On the Economic Theory of Socialism: Part One,” Review of Economic Studies 4(1), 1936. ↩
  4. F. A. Hayek, “The Use of Knowledge in Society,” American Economic Review 35(4), 1945. ↩
  5. János Kornai, Economics of Shortage (North-Holland, 1980). ↩
  6. Elinor Ostrom, “Beyond Markets and States: Polycentric Governance of Complex Economic Systems,” American Economic Review 100(3), 2010. ↩
  7. Michael Albert and Robin Hahnel, The Political Economy of Participatory Economics (Princeton University Press, 1991). ↩
  8. Pat Devine, Democracy and Economic Planning: The Political Economy of a Self-Governing Society (Polity, 1988; reissued 2010). ↩
  9. W. Paul Cockshott and Allin Cottrell, Towards a New Socialism (Spokesman, 1993; full text). ↩
  10. Evgeny Morozov, “Digital Socialism? The Calculation Debate in the Age of Big Data,” New Left Review 116/117, 2019. ↩
  11. R. H. Coase, “The Nature of the Firm,” Economica 4(16), 1937. ↩
  12. Herbert A. Simon, “Organizations and Markets,” Journal of Economic Perspectives 5(2), 1991. ↩
  13. Amazon Science, “The evolution of Amazon’s inventory planning system.” ↩
  14. Leigh Phillips and Michal Rozworski, The People’s Republic of Walmart (Verso, 2019). ↩
  15. UK Department of Health and Social Care, “The NHS Constitution for England,” principle 2 (updated 2023). ↩
  16. Yochai Benkler, “Coase’s Penguin, or, Linux and The Nature of the Firm,” Yale Law Journal 112, 2002. ↩
  17. Wikipedia, “Wikipedia:Statistics” (live editor counts). ↩
  18. The Linux Foundation, “2017 Linux Kernel Report Highlights Developers’ Roles and Accelerating Pace of Change.” ↩
  19. World Health Organization, “Blood safety and availability” (fact sheet). ↩
  20. “Foodsharing.de,” Wikipedia (organisation statistics). ↩
  21. Gill Seyfang, “Time banks: rewarding community self-help in the inner city?” Community Development Journal 39(1), 2004. ↩
  22. International Labour Organization, Care Work and Care Jobs for the Future of Decent Work (2018). ↩
  23. Anne Power and Ellie Benton, “Where next for Britain’s 4,300 mutual aid groups?” LSE COVID-19 blog, May 2021. ↩
  24. Rebecca Solnit, A Paradise Built in Hell: The Extraordinary Communities That Arise in Disaster (Viking, 2009); see the review in the Journal of Moral Education 40(1), 2011. ↩
  25. James Woodburn, “Egalitarian Societies,” Man 17(3), 1982. ↩
  26. Michael Gurven, “To give and to give not: The behavioral ecology of human food transfers,” Behavioral and Brain Sciences 27(4), 2004. ↩
  27. Terence N. D’Altroy and Timothy K. Earle, “Staple Finance, Wealth Finance, and Storage in the Inka Political Economy,” Current Anthropology 26(2), 1985. ↩
  28. St. Benedict’s Rule for Monasteries, chapters 33–34 (Project Gutenberg edition). ↩
  29. Raymond Russell, Robert Hanneman and Shlomo Getz, The Renewal of the Kibbutz: From Reform to Transformation (Rutgers University Press, 2013). ↩
  30. Albert Balcells, “Collectivisations in Catalonia and the Region of Valencia during the Spanish Civil War, 1936–1939,” Catalan Historical Review 10, 2017. ↩
  31. Michael Seidman, Workers Against Work: Labor in Paris and Barcelona during the Popular Fronts (University of California Press, 1991). ↩
  32. Inter Press Service, “Argentina: Auge y caída del trueque” (November 2002, in Spanish). ↩
  33. Mondragon Corporation, “About us.” ↩
  34. A. B. Atkinson, “On the Measurement of Inequality,” Journal of Economic Theory 2(3), 1970. ↩
  35. Betsey Stevenson and Justin Wolfers, “Subjective Well-Being and Income: Is There Any Evidence of Satiation?” American Economic Review 103(3), 2013. ↩
  36. Daniel Kahneman and Angus Deaton, “High income improves evaluation of life but not emotional well-being,” PNAS 107(38), 2010. ↩
  37. Matthew A. Killingsworth, “Experienced well-being rises with income, even above $75,000 per year,” PNAS 118(4), 2021. ↩
  38. Matthew A. Killingsworth, Daniel Kahneman and Barbara Mellers, “Income and emotional well-being: A conflict resolved,” PNAS 120(10), 2023. ↩
  39. Richard Layard, Guy Mayraz and Stephen Nickell, “The marginal utility of income,” Journal of Public Economics 92(8–9), 2008. ↩
  40. Daniel Acland and David H. Greenberg, “The Elasticity of Marginal Utility of Income for Distributional Weighting and Social Discounting: A Meta-Analysis,” Journal of Benefit-Cost Analysis 14(2), 2023. ↩
  41. Michael J. Rieley and Javier Colato, “Industry and occupational employment projections overview and highlights, 2024–34,” Monthly Labor Review, US Bureau of Labor Statistics, January 2026. ↩
  42. US Bureau of Economic Analysis via FRED, “Value Added by Industry: Finance and Insurance as a Percentage of GDP” (series VAPGDPFI). ↩
  43. David U. Himmelstein, Terry Campbell and Steffie Woolhandler, “Health Care Administrative Costs in the United States and Canada, 2017,” Annals of Internal Medicine 172(2), 2020. ↩
  44. Scott Hodge and Claire Rock, “Tax Complexity Now Costs the U.S. Economy Over $546 Billion Annually,” Tax Foundation, 2024 (updated 2025). ↩
  45. Ricardo Marto and Hoang Le, “The Rise of Digital Advertising and Its Economic Implications,” Federal Reserve Bank of St. Louis, October 2024. ↩
  46. ReFED, “From Surplus to Solutions: A Q&A on ReFED’s New U.S. Food Waste Report” (2023 data). ↩
  47. Matthew P. Rabbitt et al., Household Food Security in the United States in 2023, USDA Economic Research Service, ERR-337, 2024. ↩
  48. US Census Bureau, “Housing Characteristics: 2020,” 2020 Census Brief. ↩
  49. US Department of Housing and Urban Development, The 2024 Annual Homelessness Assessment Report to Congress, Part 1. ↩
  50. US Congressional Budget Office and Bureau of Economic Analysis via FRED, “Real Potential Gross Domestic Product” (GDPPOT) and Real GDP (GDPC1); mean shortfall below potential, 1980–2026, computed for this page. ↩
  51. OECD, “Income and wealth inequalities,” Society at a Glance 2024. ↩
  52. Dirk Krueger and Fabrizio Perri, “Does Income Inequality Lead to Consumption Inequality? Evidence and Theory,” Review of Economic Studies 73(1), 2006. ↩
  53. Orazio P. Attanasio and Luigi Pistaferri, “Consumption Inequality,” Journal of Economic Perspectives 30(2), 2016. ↩
  54. World Bank, “Gini index — Brazil, Colombia, South Africa” (indicator SI.POV.GINI; Poverty and Inequality Platform). ↩
  55. Jan De Loecker, Jan Eeckhout and Gabriel Unger, “The Rise of Market Power and the Macroeconomic Implications,” Quarterly Journal of Economics 135(2), 2020. ↩
  56. Chris Edmond, Virgiliu Midrigan and Daniel Yi Xu, “How Costly Are Markups?” Journal of Political Economy 131(7), 2023. ↩
  57. David Rezza Baqaee and Emmanuel Farhi, “Productivity and Misallocation in General Equilibrium,” Quarterly Journal of Economics 135(1), 2020. ↩
  58. Benedetto Molinari and Francesco Turino, “Advertising and Aggregate Consumption: A Bayesian DSGE Assessment,” Economic Journal 128(613), 2018; summary: Royal Economic Society, “The impact of advertising on consumption.” ↩
  59. Jess Benhabib and Alberto Bisin, “Advertising, Mass Consumption and Capitalism,” working paper, New York University, 2001. ↩
  60. N. Gregory Mankiw and Michael D. Whinston, “Free Entry and Social Inefficiency,” RAND Journal of Economics 17(1), 1986. ↩
  61. German Environment Agency (Umweltbundesamt), “Lifetime of electrical appliances becoming shorter and shorter” (February 2016), reporting Prakash et al., Umweltbundesamt / Öko-Institut. ↩
  62. Harald Wieser and Nina Tröger, “Exploring the inner loops of the circular economy: Replacement, repair, and reuse of mobile phones in Austria,” Journal of Cleaner Production 172, 2018. ↩
  63. European Environmental Bureau, Coolproducts don’t cost the Earth (2019). ↩
  64. US Bureau of Economic Analysis via FRED: Personal Consumption Expenditures: Durable Goods (PCEDG) and Clothing and Footwear (DCLORC1A027NBEA). ↩
  65. Appriss Retail and Deloitte, “2024 Consumer Returns in the Retail Industry.” ↩
  66. Earth911, “Rescuing Product Returns Before They Reach the Landfill,” reporting Optoro’s 2022 estimate. ↩
  67. Ellen MacArthur Foundation, “France’s Anti-waste and Circular Economy Law.” ↩
  68. Eva Vivalt, Elizabeth Rhodes, Alexander W. Bartik, David E. Broockman and Sarah Miller, “The Employment Effects of a Guaranteed Income: Experimental Evidence from Two U.S. States,” NBER Working Paper 32719, 2024. ↩
  69. Damon Jones and Ioana Marinescu, “The Labor Market Impacts of Universal and Permanent Cash Transfers: Evidence from the Alaska Permanent Fund,” American Economic Journal: Economic Policy 14(2), 2022. ↩
  70. Uri Gneezy and Aldo Rustichini, “Pay Enough or Don’t Pay at All,” Quarterly Journal of Economics 115(3), 2000. ↩
  71. Edward L. Deci, Richard Koestner and Richard M. Ryan, “A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation,” Psychological Bulletin 125(6), 1999. ↩
  72. Anandi Mani, Sendhil Mullainathan, Eldar Shafir and Jiaying Zhao, “Poverty Impedes Cognitive Function,” Science 341(6149), 2013. ↩
  73. Ellen MacArthur Foundation, A New Textiles Economy: Redesigning Fashion’s Future (2017), “Summary of findings.” ↩

Figures were checked against the sources above in October 2026. The model script lists every parameter with its source. Corrections are welcome through the discussion links elsewhere on the site.

Site map