Haute Lumière

Commerce · III.11 · MMXXVI · daylight

La Bourse  /  Volume III  /  Nº III.11

What Money Cannot Do

Volume III — Money, Energy, Information

---

THE PLATE

---

THE LETTER

This is the last chapter of a volume that has spent ten chapters taking money seriously — what it is made of, what it denominates, how it circulates, how it settles, how it is held in reserve. It would be a poor volume that ended by declaring the whole apparatus suspect, and this chapter does not.

It does something more useful and considerably harder. It draws the boundary of the price mechanism exactly where the evidence puts it — not where sentiment would like it, and not where the market's advocates insist it is not. Both of those are available positions and both are cheap. The first produces a reader who distrusts prices in general and therefore cannot use them at all. The second produces a reader who prices a thing that was holding an obligation together and is surprised when the obligation dissolves.

So we will be precise. There are three separate claims that get confused into one, and only two of them are true.

One: paying for a thing sometimes reduces its supply. True, measured, replicated, and much narrower than it is usually reported to be.

Two: a stated price for an unpriced good is sometimes a number with no referent. True, and demonstrable with arithmetic you can check in an afternoon.

Three: therefore some things should not be priced. This does not follow, and the substitution of the third claim for the first two is where most writing on this subject stops being useful. A refusal to price is not an abstention. It is an allocation rule, it has an incidence, and somebody pays it. In the middle of this chapter we will compute who.

Then we will do the constructive half, because a boundary without a set of instruments on the far side of it is just a fence. Five instruments — quantity, standards, rights, rationing, deliberation — each with a case that actually ran and a cost that was actually borne.

— The Editors

---

DISCOVERY

What is already working

Start where the evidence is strongest, and notice that the strongest evidence is about mechanisms that worked.

The Swedish blood experiment. Carl Mellström and Magnus Johannesson ran the cleanest test of Richard Titmuss's forty-year-old claim that paying for blood reduces its supply. Among women, the donation rate with no payment offered was 52.0 percent. Offered SEK 50, it fell to 30.0 percent — a fall of 22.0 percentage points, which is a relative fall of 42.3 percent and a supply multiplier of 0.577. Titmuss was right, and for the first time he was right with a control group.

Then the finding that makes it useful rather than merely striking. When the same SEK 50 could be assigned to a charity instead of pocketed, the effect disappeared. Among men, there was no statistically significant difference between arms at all. So the money was never the problem. The transaction was — specifically, the conversion of a gift into a sale, which is a change in what the act means and not a change in what it pays.

The Haifa day-care centres. Uri Gneezy and Aldo Rustichini studied ten centres over twenty weeks. Six introduced a fine of 10.00 NIS for collecting a child late — against a monthly fee of 1,400.00 NIS, so the fine was 0.714 percent of the fee. Late collections rose from roughly 8.0 a week to roughly 20.0, a multiplier of 2.50. Then the fine was withdrawn, and the rate did not come back down.

Read that last sentence again, because it is the whole finding. A price did not fail to deter. It replaced something — an unpriced obligation to the teacher who was waiting — and when the price was removed the obligation did not return, because the thing that had been destroyed was the belief that one existed. What worked, here, was the diagnosis: the researchers established that the fine had overwritten a norm by removing the fine and watching nothing happen.

The Swiss repository. Bruno Frey and Felix Oberholzer-Gee asked residents of a Swiss community whether they would accept a nuclear waste repository. 50.8 percent agreed. Asked again with compensation attached, agreement fell to 24.6 percent — down 26.2 percentage points, a relative fall of 51.6 percent. Raising the offered amount through roughly $2,200, $4,400 and $6,600 a year per person did not recover it. Civic duty and market exchange were not additive. They were rival descriptions of the same act, and the money selected the description in which refusal was the sensible move.

And now the discipline. Nicola Lacetera, Mario Macis and Robert Slonim ran incentive experiments across American Red Cross blood drives at a scale the laboratory cannot reach, and found that gift-card incentives raised turnout, with the effect rising in the value of the item — alongside real displacement from neighbouring un-incentivised drives. That result is not a refutation of Titmuss. It is a specification of him. Crowding out is a real phenomenon with a restricted domain, and a practitioner who treats "never pay" as a principle rather than as a finding about certain acts, certain framings and certain people will lose supply they needed and will not know why.

Four studies, and every one of them is a measurement of a boundary rather than a verdict on markets. That is the appreciative core of this chapter: we now know where the line is, to within a few percentage points, in several concrete domains. Forty years ago we had an argument. We have arithmetic.

---

THE ARITHMETIC

What works, what does not, and where the line sits

First: what a stated price for an unpriced good is actually measuring.

Contingent valuation asks people what they would pay for a thing that is not sold. It is the only method available for a great many public goods, and it has two measured pathologies that every user of it must be able to compute.

Scope insensitivity. In the study most often cited, respondents were asked what they would pay to prevent migratory birds dying in oil ponds. For 2,000 birds: $80. For 20,000: $78. For 200,000: $88. The good increased by a factor of 100. The price increased by a factor of 1.10. The scope elasticity is therefore:

  ln(88 / 80) / ln(200,000 / 2,000)  =  0.0207

A scope elasticity of 0.02 is not a demand curve. A demand curve responds to quantity; this responds to the existence of the question. What is being measured is the size of a feeling about birds, which is real, and which is the same size whatever number follows it.

Protest zeros. A respondent who writes zero may mean this is worth nothing to me, or may mean I refuse to answer a question posed in this currency. Those are opposite facts and they arrive in the same cell of the spreadsheet. The analyst decides which they were, and the decision is worth money. Take an aggregate the size of the Exxon Valdez passive-use claim — a median of $31.00 per household against $2,800,000,000, which implies 90,322,581 households and is arithmetically sound against a United States that had roughly 94 million of them. Now decompose a survey with that mean: 1,000 respondents, 780 positive answers averaging $39.74, 120 true zeros and 100 protests.

  protests counted as zero     780 x 39.74 / 1,000  =  $31.00
  protests dropped             780 x 39.74 /   900  =  $34.44
                                                       -------
  difference                                            11.1 %

At that population, the two codings give $2,799,747,097 and $3,110,830,108. One analyst's judgement about what a zero means is worth $311,083,011. No fieldwork changed. No respondent changed their mind.

Second: the difference between a price that is hard and a price that is absent. This distinction is the load-bearing one in the chapter, and it is usually asserted rather than tested.

A difficult shadow price converges as the method improves. The value of a statistical life is the worked example. The EPA's central figure is $7,400,000 in 2006 dollars. Deflate it with the CPI-U annual averages — 201.6 for 2006, 313.7 for 2024, a deflator of 1.556 — and you get $11,514,782, which is inside the range transport regulators reach by an entirely different route. Spread over 40 remaining life-years at 3 percent, an annuity factor of 23.115, it implies $498,157 per life-year. Contested, yes. Undefined, no. The estimates sit in a band and the band narrows.

An absent shadow price does not converge. Take a language with two hundred remaining speakers. Method A prices it at what full documentation costs: $500,000. Method B prices it at $5 per household across the same 90 million households: $451,612,903. The two defensible answers differ by a factor of 903.2, and no improvement in survey design closes that gap, because the methods are not estimating the same quantity with different error — they are estimating different quantities. That is the operational test for incommensurability, and it is a test rather than an intuition: does the interval narrow as the method improves? If it does, price it and argue about the number. If it does not, you do not have a price to argue about.

Third — the honest negative, and it inverts.

Everything above is an argument for restraint. Here is the arithmetic that makes restraint expensive, and it is the number this chapter exists to put on a page.

Take a city of 1,000,000 households that has decided, on excellent moral grounds, not to charge properly for water. The tariff is held at $0.35 per cubic metre against a cost of supply of $0.90, so every metered cubic metre carries $0.55 of subsidy. Households without a connection buy from vendors at $2.80 — 8.00 times the tariff. Connection rates rise with income, as they do everywhere the pattern has been measured: 35%, 50%, 70%, 85%, 95% across the quintiles, with connected use of 9, 11, 14, 18, 24 cubic metres a month and unconnected purchase of 4.0.

Now compute who receives the subsidy.

The richest quintile takes 40.3 percent of the subsidy and the poorest takes 5.6 percent — a ratio of 7.24. The poorest quintile pays $1.458 per cubic metre and the richest pays $0.371: the poorest pay 3.93 times per litre what the richest pay.

And then the line that should be read slowly. Q1 spends $8.38 a month on water. Q5 spends $8.54. They spend, to within 1.9 percent, the same money. Q5 receives 4.00 times as much water for it.

There is no tariff on that woman's water. The refusal to price is the tariff. It is levied in connection status rather than in currency, it is collected by vendors rather than by the utility, it is invisible in every published schedule, and it is the single most regressive charge in the city. A price would have to be designed with real malice to beat it.

This is the honest negative of the whole living-systems position, and it points the opposite way from where such negatives usually point: the move that feels protective is frequently the move that transfers most to whoever already holds the asset. Refusing to price housing favours incumbent tenants over future ones. Refusing to price road space favours the car owner over the bus rider stuck in the same queue. Refusing to price water favours the household with the meter. In each case the beneficiary is the party already inside.

---

DREAM

What becomes ordinary

In the economy that has learned this properly, the question should this be priced has stopped being a question about virtue and become a question with a procedure.

An analyst facing an unpriced good runs the convergence test first. Two independent methods, honestly applied; if the interval narrows, the good is priceable and the argument moves to the number, where arguments are cheap and productive. If the interval does not narrow — if it spans three orders of magnitude and stays there — the analyst writes no shadow price in the box, and that entry is respected rather than treated as a failure to try hard enough. A missing price and a price of zero are different facts, and the systems record them differently.

Where the good is priceable, it is priced, and the distributional consequence is computed at the same moment as the price, on the same page, by the same person. Nobody presents a tariff without an incidence table. Nobody presents a refusal without one either, because the refusal has an incidence too and everyone now knows how to compute it.

Where the good is not priceable, the allocation is made by an instrument chosen on purpose rather than by default. A cap. A standard. An inalienable right with a clearing mechanism behind it. A ration with a floor. A room of people with a budget and a week. These are not consolation prizes for markets that failed to form. They are a catalogue, and practitioners know it the way a treasurer knows the difference between a revolving facility and a term loan.

And the crowding-out literature is used the way a pharmacologist uses a dose response — as a statement about which acts, at which amounts, in which framings. An organisation that wants more of a voluntary behaviour tests a recognition that is not convertible into cash before it tests a payment, because that is what the Swedish result actually says. An organisation that wants more supply at scale, from people who were never going to act from duty, offers the gift card, because that is what the American result actually says. Neither is a betrayal of the other. They are two coefficients in the same equation.

The tone of all of this is the thing most worth having. It is unanxious. The question of what money can and cannot do has stopped being a loyalty test.

---

DESIGN

The five instruments, with what each one cost

One — quantity. Fix the quantity and let the price be discovered. The American sulphur dioxide allowance market cut covered emissions 36.0 percent between 1990 and 2004 while coal-fired generation rose 25.0 percent — an emissions intensity of 0.5120 of the starting level, a fall of 48.8 percent. The cost: roughly $1 billion a year less than the command-and-control equivalent, which is the rare instrument that arrived under budget. Use it when the physical quantity is what you care about and the marginal cost of abatement varies widely across holders.

Two — standards. Set the outcome and let everyone find their own way to it. Children's blood lead in the United States fell from a geometric mean of 15.0 µg/dL in 1976–80 to 2.8 µg/dL in 1991–94 — a fall of 81.3 percent, a ratio of 5.36. No microgram of lead was ever priced. The economic benefit has been estimated at $110,000,000,000 to $319,000,000,000 per annual birth cohort; against 4,000,000 births that is $27,500 to $79,750 per child. The cost was refinery capital, and the phase-down used tradable inter-refinery lead credits to find the cheap tonnes — a standard with a market inside it, which is the design most people mean when they say they are against markets.

Three — rights, with a clearing mechanism. Declare the good inalienable and then build the machine that clears it anyway. Organs cannot be sold in most jurisdictions; kidney paired donation was built instead, and runs on the order of 1,000 transplants a year. The arithmetic: dialysis costs about $94,000 per patient-year and a functioning graft about $38,000, a saving of $56,000 a year. The transplant year itself runs to roughly $120,000, $26,000 above dialysis, recovered in 5.6 months. Over ten graft-years that is $560,000 undiscounted, $477,691 at 3 percent, and $56,000,000 a year recurring at current volumes. A matching mechanism produced the supply the price was supposed to produce. The cost is real — a registry, surgical logistics, chains that fail when one donor withdraws.

Four — rationing. Fix the per-person entitlement. Cape Town cut daily consumption from 1,200.0 to 500.0 megalitres between February 2015 and February 2018 — 58.3 percent, 700.0 megalitres a day — under a personal limit of 50.0 litres a day, which across 4,000,000 people is a floor of 200.0 megalitres. And here is the cost, stated plainly because it is the one people omit: the reported revenue shortfall of R1,400,000,000 to R2,000,000,000, midpoint R1,700,000,000, against 255,500,000,000 litres saved in a year, is R0.00665 per litre — R6.654 per cubic metre. Rationing works and it breaks the revenue model of the utility that has to keep the pipes full afterwards. Budget for that in advance or the drought ends and the maintenance backlog begins.

Five — deliberation. Put the allocation in a room. Oregon ranked 709 condition-treatment pairs and funded down to line 587 — 82.8 percent of them, with 122 below the line, named. The honest cost: the first list was rejected by federal authorities in 1992 on disability-discrimination grounds and had to be rebuilt. Deliberation's cost is a rewrite, and the rewrite is not waste; it is the instrument working.

And the case that closes the loop. Porto Alegre put its capital budget in a room with the people who lacked a pipe. Household water connections went from 75.0 percent to 98.0 percent and sewerage from 46.0 percent to 85.0 percent between 1988 and 1997 — 92.0 percent of the water gap and 72.2 percent of the sewerage gap closed in nine years. That is the exact problem the untargeted tariff subsidy could not solve, solved by the instrument with no price in it at all.

---

DESTINY

How it holds when you stop pushing

The failure mode of this chapter is that a reader takes the word incommensurable and uses it as a door to close.

Three things make the discipline hold.

The convergence test is written down and applied by someone who is not arguing. A boundary that is redrawn per decision is not a boundary. Put the test in the analysis template — two methods, does the interval narrow, what was the ratio — so that it is answered before anyone knows which answer they want.

Every refusal gets an incidence table. This is the whole contribution of the chapter, reduced to a habit. If a body declines to price something, it computes who receives the benefit of the non-price and who pays the implicit one. The water table above took an afternoon. It is not difficult; it is simply never done, and it is never done because nobody has ever been asked for it.

The instrument catalogue is a catalogue. Five named options, each with a case and a cost, so that "we will not price it" is followed by a choice rather than by silence. Silence is the default allocation, and the default allocation favours whoever is already inside.

Now where this fails. It fails when the convergence test is run by someone who has already decided, since two methods can be chosen to agree. It fails when the incidence table is built on connection or eligibility data the institution does not actually hold, which is common, and the honest response is to say the incidence is unknown rather than to assume it is flat. It fails when a crowding-out finding from one domain is carried into another — the Swedish result is about women, small sums and a visible conversion of a gift into a sale, and it does not travel to a gift card at a Red Cross drive. And it fails, most expensively, when an organisation adopts the vocabulary of incommensurability without the arithmetic, at which point this cannot be priced becomes the most convenient sentence in the building.

---

DELIGHT

What it feels like

There is a particular relief in being handed a test instead of a stance.

Anyone who has sat in a room where this subject comes up knows the shape of the usual conversation: one person who believes everything can be priced, one who believes pricing is a kind of violence, and a long middle in which neither produces a number. The convergence test ends that conversation in about twenty minutes, and it ends it without anyone losing. You run two methods. You look at the ratio. The ratio decides, and nobody has to have been wrong about their values.

And then the second pleasure, which is stranger and better. The moment you compute the incidence of a refusal, the moral confidence in the room redistributes itself, and it does not redistribute toward the market. It redistributes toward whoever was standing at the standpipe at six in the morning, because for the first time there is a figure attached to them — a figure that was always true and was never written down. $8.38 a month, for a quarter of the water.

That is the feeling this chapter is for. Not the satisfaction of having been sceptical about prices. The considerably better one of having found the person the absence of a price was costing, and being able to say by how much.

---

OPERATIONALIZE THIS

At the level of finance

The instrument: a revolving connection facility, repaid from the tariff spread that cost-reflective pricing restores.

The structure exists to make the argument of this chapter bankable. You are not proposing that water be made expensive. You are proposing that the money currently flowing to metered households as a per-cubic-metre subsidy be converted into capital that puts meters in the houses that do not have one, repaid out of the spread the reform itself creates.

The mechanics.

• Facility size. $40,000,000, funding 100,000 connections at $400.00 of capex each. Sized to one year of works, not to the whole programme — the facility revolves, and a revolving facility that has completed one cycle is a different credit from a proposal.
• The source of repayment. Moving the tariff from $0.35 to the cost-recovery $0.90 restores $0.55 per cubic metre. Across 100,000 new connections consuming 9 cubic metres a month, that is 100,000 × 9 × 12 × $0.55 = $5,940,000 a year.
• Coverage. Against facility service at 6 percent — $2,400,000 — the repayment stream covers 2.48 times. That ratio is the front page. Below 1.5× the programme is not self-funding and should be financed as a grant and called one.
• The lifeline block is not optional. The first block of consumption is priced at or near zero for every household, funded from the same restored spread. Cost-reflective pricing without a lifeline block is the reform that makes this chapter's critics right.
• The rebate, where a lifeline block is not available. The existing subsidy budget of $74,646,000 returned as an equal per-household transfer is $74.65 a year, $6.22 a month. Under cost-reflective pricing Q1 pays $1.73 a month more in tariff and receives $6.22, a net of +$4.49 a month — +$53.86 a year. Q5 pays $12.54 more and receives the same $6.22, a net of −$6.32 a month, −$75.83 a year. The reform is progressive on the first page of the model, which is where the political argument is won or lost.
• Term. Five years, with a hard review at two. The unconnected population of 330,000 costs $132,000,000 to reach, which is 1.77 years of the current subsidy — so the programme's own arithmetic says it finishes.
• Security. The tariff stream, ring-fenced, with the connection asset on the utility's balance sheet.

The balance-sheet treatment. Capitalise the connection as network infrastructure and depreciate it over the service life of the asset, not the tenor of the facility. The subsidy it replaces was an operating expense that purchased nothing durable; the connection is an asset that generates revenue for decades. Converting a recurring operating subsidy into a depreciating asset with a revenue stream attached is the entire financial argument, and it is one your auditors will recognise immediately.

The counterparty. The municipal utility first, with the regulator's tariff determination as the condition precedent. A development lender or a domestic pension fund takes the second tranche once one cycle has revolved, because by then you are presenting a track record rather than a model.

The number that decides it. One inequality, on the front page:

                connection capex per household
   -------------------------------------------------------  <  5 years
    annual saving vs the vendor price, at constant volume

Here: $400.00 ÷ $91.20 = 4.39 years, or 52.6 months — the vendor bill of $134.40 a year against a piped bill at full cost of $43.20 for the same 4 cubic metres a month. The household is better off at a cost-reflective tariff with no subsidy whatsoever, which is the sentence that ends the meeting.

The first ninety days.

---

APPRECIATIVE QUESTIONS

Twelve, for a room

Discovery — what is already working

1. Where in this organisation does something valuable get done reliably with no payment attached to it — and what is holding it up, if it is not money?
2. Think of a time we chose not to charge for something and it clearly worked. What made that the right call, and who benefited that we did not expect?
3. Which of our allocation decisions are already made by a rule rather than a price, and which of those rules would we defend in public?

Dream — what becomes possible

1. If every decision not to price something came with a table showing who benefits, what would we discover about ourselves in the first month?
2. Imagine we had an instrument catalogue — cap, standard, right, ration, room — and everyone could name all five. What decision would get made differently next quarter?
3. What would it feel like to be able to say no shadow price exists here and have that accepted as a finding rather than heard as a refusal?

Design — what we build

1. What is the one good in our business where the convergence test would settle an argument we have been having for years?
2. Where could a recognition that cannot be converted into cash do the work we have been trying to do with a bonus?
3. If we were going to price something we currently give away, what would the lifeline block be, and who writes it?

Destiny — how it holds

1. Who in this organisation would notice first if incommensurable started being used to close conversations rather than to open them?
2. What data would we need to hold for an incidence table to be honest — and what is the smallest step toward holding it?
3. Ten years from now, what allocation are we proud of having made without a price, and what did we build so that it could hold?

---

WORKS CITED

Titmuss, R. M. (1970). The Gift Relationship: From Human Blood to Social Policy. Allen & Unwin.

Arrow, K. J. (1972). "Gifts and Exchanges." Philosophy and Public Affairs, 1(4), 343–362.

Mellström, C. and Johannesson, M. (2008). "Crowding Out in Blood Donation: Was Titmuss Right?" Journal of the European Economic Association, 6(4), 845–863.

Lacetera, N., Macis, M. and Slonim, R. (2012). "Will There Be Blood? Incentives and Displacement Effects in Pro-Social Behavior." American Economic Journal: Economic Policy, 4(1), 186–223.

Lacetera, N., Macis, M. and Slonim, R. (2013). "Economic Rewards to Motivate Blood Donations." Science, 340(6135), 927–928.

Gneezy, U. and Rustichini, A. (2000). "A Fine Is a Price." Journal of Legal Studies, 29(1), 1–17.

Gneezy, U. and Rustichini, A. (2000). "Pay Enough or Don't Pay at All." Quarterly Journal of Economics, 115(3), 791–810.

Deci, E. L., Koestner, R. and Ryan, R. M. (1999). "A Meta-Analytic Review of Experiments Examining the Effects of Extrinsic Rewards on Intrinsic Motivation." Psychological Bulletin, 125(6), 627–668.

Frey, B. S. and Oberholzer-Gee, F. (1997). "The Cost of Price Incentives: An Empirical Analysis of Motivation Crowding-Out." American Economic Review, 87(4), 746–755.

Frey, B. S. and Jegen, R. (2001). "Motivation Crowding Theory." Journal of Economic Surveys, 15(5), 589–611.

Desvousges, W. H., Johnson, F. R., Dunford, R. W., Boyle, K. J., Hudson, S. P. and Wilson, K. N. (1992). Measuring Nonuse Damages Using Contingent Valuation. Research Triangle Institute Monograph 92-1.

Kahneman, D. and Knetsch, J. L. (1992). "Valuing Public Goods: The Purchase of Moral Satisfaction." Journal of Environmental Economics and Management, 22(1), 57–70.

Arrow, K., Solow, R., Portney, P. R., Leamer, E. E., Radner, R. and Schuman, H. (1993). "Report of the NOAA Panel on Contingent Valuation." Federal Register, 58(10), 4601–4614.

Diamond, P. A. and Hausman, J. A. (1994). "Contingent Valuation: Is Some Number Better than No Number?" Journal of Economic Perspectives, 8(4), 45–64.

Carson, R. T., Mitchell, R. C., Hanemann, M., Kopp, R. J., Presser, S. and Ruud, P. A. (2003). "Contingent Valuation and Lost Passive Use: Damages from the Exxon Valdez Oil Spill." Environmental and Resource Economics, 25(3), 257–286.

Carson, R. T. (2012). "Contingent Valuation: A Practical Alternative When Prices Aren't Available." Journal of Economic Perspectives, 26(4), 27–42.

Hausman, J. (2012). "Contingent Valuation: From Dubious to Hopeless." Journal of Economic Perspectives, 26(4), 43–56.

Anderson, E. (1993). Value in Ethics and Economics. Harvard University Press.

Raz, J. (1986). The Morality of Freedom. Oxford University Press.

Sandel, M. J. (2012). What Money Can't Buy: The Moral Limits of Markets. Farrar, Straus and Giroux.

Radin, M. J. (1996). Contested Commodities. Harvard University Press.

Schmalensee, R. and Stavins, R. N. (2013). "The SO2 Allowance Trading System: The Ironic History of a Grand Policy Experiment." Journal of Economic Perspectives, 27(1), 103–122.

Grosse, S. D., Matte, T. D., Schwartz, J. and Jackson, R. J. (2002). "Economic Gains Resulting from the Reduction in Children's Exposure to Lead in the United States." Environmental Health Perspectives, 110(6), 563–569.

Roth, A. E., Sönmez, T. and Ünver, M. U. (2004). "Kidney Exchange." Quarterly Journal of Economics, 119(2), 457–488.

Roth, A. E. (2007). "Repugnance as a Constraint on Markets." Journal of Economic Perspectives, 21(3), 37–58.

Komives, K., Foster, V., Halpern, J. and Wodon, Q. (2005). Water, Electricity, and the Poor: Who Benefits from Utility Subsidies? World Bank.

United Nations Development Programme (2006). Human Development Report 2006 — Beyond Scarcity: Power, Poverty and the Global Water Crisis. Palgrave Macmillan.

Oberlander, J., Marmor, T. and Jacobs, L. (2001). "Rationing Medical Care: Rhetoric and Reality in the Oregon Health Plan." Canadian Medical Association Journal, 164(11), 1583–1587.

Baiocchi, G. (2005). Militants and Citizens: The Politics of Participatory Democracy in Porto Alegre. Stanford University Press.

Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.

Note on figures. Every number above is computed in lib/verify/III_11.py and printed by python3 lib/verify.py III.11, with each input marked PUBLISHED or MODELLED. The water incidence table is a model at city scale, calibrated to the documented pattern that piped connection rates rise steeply with income; the inputs are printed so a reader can re-calibrate them and see that the direction of the result does not move. The Cape Town revenue shortfall is reported in a range and the midpoint is stated as such. Where a published figure is approximate, the approximation is the source's and is marked.