What a Person Is Worth

Volume V — Labour, Value, Flourishing

THE PLATE

A woman seated on the floor by a tall window writing in a notebook, smiling at what she has written. — Plate V.01The Two Ledgers.Two ledgers, kept honestly, will not agree. That is not a fault in the bookkeeping. It is the first true thing either of them tells you.

THE LETTER

There is a question this volume has to ask before it can ask anything else, and it is the one question in economics that people lower their voices to say aloud. What is a person worth?

You have met the two usual answers and neither of them survives contact with a morning. The first is that a person is worth what they can be paid for — which anyone who has cared for a dying parent knows to be false, and knows it in the body, not the argument. The second is that a person is beyond price — which is true, and is also the sentence that gets safety rails left off a stairwell, because a thing that cannot be priced cannot be put in a budget, and a thing that is not in the budget does not get built.

So this chapter does something narrower and more useful than settling the question. It takes you to the five rooms where the question is actually asked — the wage negotiation, the regulator's impact assessment, the national accounts, the underwriter's desk, the courtroom — and it shows you what each of those rooms is really asking, what number it produces, and what that number is for.

They produce different figures. Wildly different: on the arithmetic below, two of them sit a factor of eight apart for the same person on the same morning. That is not a scandal and it is not a measurement error. They are not five estimates of one quantity. They are five prices for five decisions, and the whole of the discomfort people feel in this conversation comes from treating them as though they were rivals for a single crown.

You will finish this chapter able to do four things. Say which of the five numbers a given argument is using. Compute each of them for a real case. Notice the moment someone slides from one to another mid-sentence, which is the most common dishonesty in the whole field. And — this is the part that matters on a Monday — write down a price before the decision rather than after it, because the decision is going to contain one either way.

— The Editors

DISCOVERY

What is already working

Begin where the practice is at its best, because it is better than its reputation and almost nobody has been shown the good version.

The Clean Air Act, valued and defended. The United States Environmental Protection Agency's second prospective study of the 1990 amendments projected benefits of roughly 2,000 USD bn against costs of about 65 USD bn over the period to 2020 — a ratio near 30.77 : 1. The striking part is not the ratio. It is the composition: something on the order of 85.0 per cent of the monetised benefit is avoided premature death, which means the largest cost-justified public-health intervention in the modern record stands almost entirely on a number for a life. Take the number away and the regulation does not become more humane. It becomes unfundable.

A threshold published on purpose. Since 1999 the National Institute for Health and Care Excellence has appraised treatments against a stated cost per quality-adjusted life year. Publishing it did something no unpublished rule can do: it made the rule arguable. Because the threshold was on paper, Karl Claxton and colleagues could go and measure what the health service actually displaces when it spends at the margin, and report a figure of about 12,936 GBP per QALY — below the threshold, and therefore evidence that the threshold was letting through treatments that cost more health than they bought. That finding was only possible because somebody had written a number down. An implicit threshold cannot be researched, cannot be appealed, and cannot be improved.

A fund that people chose over a court. After September 2001 the United States created a compensation fund rather than leaving 2,880 death claims to litigation. Kenneth Feinberg, as Special Master, computed awards on economic-loss principles: average death award about 2,080,000 USD, average injury award about 400,000 USD, across 2,680 injury claims. Around 97 per cent of eligible families took the fund instead of suing. The instrument worked — it delivered, finally and fast, what tort would have delivered slowly or not at all. And the same Special Master has since said in public that he would now pay every family the same amount. Hold both halves of that. It is the most honest sentence anyone in this field has said, and it is the hinge of this chapter: the formula was correct for the question how much income did this household lose, and wrong for the question the families were actually asking.

A Treasury that priced a feeling. In 2021 HM Treasury issued supplementary Green Book guidance valuing a wellbeing-adjusted life year — a one-point change in life satisfaction held for a year — at about 13,000 GBP. Whatever you make of the figure, notice what it does structurally: it puts loneliness, noise, green space and job security into the same appraisal arithmetic as a bypass. Before it existed, a scheme whose entire benefit was that people felt better had no column to be written in, and things with no column do not get built.

And the plainest case: pay itself. When Doruk Cengiz, Arindrajit Dube, Attila Lindner and Ben Zipperer counted jobs directly across a set of United States minimum-wage increases, they found the jobs paying below the new minimum disappearing and an almost exactly offsetting number appearing just above it, with total employment statistically indistinguishable from before. Read that as a measurement of the labour market's joints: if pay can move without employment moving, pay was not pinned to marginal product in the first place. There was slack, and the slack was usable.

Five cases, one pattern. In each, somebody made a valuation explicit, and the explicitness is what produced the good: the rule could be justified, researched, appealed, extended or overturned. None of the five is a claim about what a human being is. Each is a claim about what a particular decision should cost.

THE ARITHMETIC

Five numbers, and the questions they actually answer

One — marginal product, and how far pay follows it.

The textbook says a firm hires until the marginal revenue product of the last worker equals the wage. It is a good model and it answers a real question: what does one more hour of this work add to this firm's revenue? That is the hiring-at-the-margin question, and no other number answers it.

It is not, however, a description of what happens. Where a firm faces an upward- sloping labour supply curve — where paying one worker more means paying everyone more — the wage settles at a markdown, and the markdown has a closed form:

        w = MRP · e / (1 + e)

  e = the elasticity of labour supply to the individual firm

  e = 1.5    wage is 60.0 % of marginal revenue product
  e = 3.0    wage is 75.0 %
  e = 5.0    wage is 83.3 %
  e = 10.0   wage is 90.9 %

The competitive case is e → ∞, where the wage is the whole product. Published firm-level elasticities cluster in the low single digits, so the honest reading of the table is that a large share of what a worker produces does not reach the worker, and the size of that share is a property of the market's plumbing, not of the worker.

The corroborating evidence runs the same direction from three independent angles. Rent-sharing studies put the pass-through of firm value added to pay at an elasticity between 0.05 and 0.15 — a firm that raises value added per worker by 10 per cent passes 0.5 to 1.5 per cent of it into wages. The long series: between 1979 and 2019, United States net productivity grew 59.7 per cent while the typical worker's hourly compensation grew 15.8 per cent, a ratio of 3.78 x. That gap's magnitude is genuinely contested — use total compensation rather than wages, and the same deflator on both sides, and it narrows considerably — but no serious reading of the series makes it zero, and the direction has never been in dispute.

So the first number is real, useful, and much weaker as a description of pay than its reputation suggests. It answers what is one more hour worth to this firm. It has never answered what should this person receive, and it was never built to.

Two — the value of a statistical life.

Here is the number people find obscene, and here is why it is not.

A regulator writing a rule needs to know how much the public should spend to reduce a small risk spread across a large population. The number it uses in the United States — the Department of Transportation's departmental guidance figure — is 13,200,000 USD. Now do the division the headline never does. A risk step of 1 in 10,000 at that value is worth:

  13,200,000 USD × 0.0001  =  1,320 USD per person

Ten thousand people each paying 1,320 USD to remove a hazard that would have killed one of them, unidentified, spend one VSL between them. Not one of them was priced. The figure runs backwards just as well: workers who accept about 900 USD a year to take a job carrying an extra annual fatal risk of 1 in 10,000 are revealing, by their own choice, an implied value of 9,000,000 USD. The number is built out of what people charge each other for risk, and the person cancels out of the division entirely.

Note also that one government holds more than one such figure. The EPA's central value of 7,400,000 USD in 2006 money is 11.5 million in 2024 money on CPI-U, and the transport figure stands 1.15 x above it. Two agencies, one government, two prices — because they regulate different risks under different statutes, and the numbers are instruments of those statutes rather than facts about people.

Three — a year of life, priced by two different budgets.

Annuitise the transport figure over 40 remaining years at a 3 per cent real discount rate. The annuity factor is 23.115, so:

  13,200,000 / 23.115  =  571,063 USD per statistical life-year

Set that beside the health service's numbers. NICE's upper threshold of 30,000 GBP is 38,100 USD at a working rate of 1.27, which the transport figure exceeds by 14.99 x. Against the measured displacement figure of 12,936 GBP — 16,429 USD — the multiple is 34.76 x.

A factor of fifteen to thirty-five for a year of the same life, inside the same kind of state. That looks like an outrage and it is actually a category difference. The transport figure asks what will a population pay to avoid a risk? The health figure asks what does this budget displace when it is spent? The first is a willingness to pay. The second is the shadow price of a fixed constraint. Only the second can be lowered by being given more money — and noticing that is worth more than any amount of indignation about the gap.

Four — human capital.

Take median-ish earnings of 60,000 USD, 40 working years, real earnings growth of 1 per cent, a 3 per cent real discount rate. The growing-annuity factor is 27.18, giving:

  present value of lifetime earnings  =  1,630,736 USD

The transport figure is 8.09 x that. Same person, same morning, two numbers an order of magnitude apart — and both correct, because one is the discounted stream of what the person will be paid and the other is the price of a risk reduction. The human-capital number is what the national accounts use, what education investment appraisal uses, and what a wrongful-death computation starts from. It is also why the World Bank's wealth accounting finds human capital at roughly 64.0 per cent of global wealth — the largest asset class on earth, and the only one no company is permitted to put on its balance sheet, since IAS 38 forbids capitalising an internally generated workforce. A firm may capitalise a purchased customer list. It may not capitalise the crew that took nine years to train.

Five — insurance and tort.

An underwriter sizes a contract at a multiple of earned income: 600,000 USD at ten times, 900,000 USD at fifteen, 1,200,000 USD at twenty. That number is not an opinion about the insured. It is the size of the hole in a household's income statement, and it is deliberately blind to everything else about the person, which is exactly what makes it fair as an underwriting rule and useless as a statement about a life.

Tort valuation starts from the same stream and adds the law's own constraints, which is where the arithmetic turns instructive. California's cap on non-economic damages was set at 250,000 USD in 1975 and left unindexed. Deflate it and the cap is worth 45,951 USD in the money it was legislated in — 18.4 per cent of its original value. To hold that value in 2024 it would have to read 1,457,714 USD. A number that is not indexed is a number somebody is lowering every year without ever having to vote for it.

The honest negative: the price you refuse to name is still charged

Now the part that decides how you should act, and the reason the squeamish position is not the safe one.

Refusing to put a number on a life does not protect anyone. It relocates the number to whoever sets the budget, where nobody has to defend it. And the implicit price is, with dismal regularity, far below the explicit one.

Take the clearest case in the record. In the refurbishment of Grenfell Tower, the substitution of the cladding panels for a cheaper product produced a recorded saving of 293,368 GBP. Seventy-two people died in the fire whose spread the public inquiry attributed principally to that cladding. The division is elementary and nobody performed it in advance:

  293,368 GBP / 72  =  4,075 GBP per life

That is 5,175 USD at our working rate. The same government's transport department writes 13,200,000 USD into its own appraisals — a figure 2,551 x higher than the one that decision implicitly used. Nobody in that procurement chose 4,075 GBP. Nobody wrote it anywhere. It was the arithmetic left behind by a decision that declined to state a price, and that is precisely the mechanism: the refusal did not prevent a valuation, it prevented the valuation from being seen and challenged.

The pattern repeats in far less dramatic places. Orley Ashenfelter and Michael Greenstone read the American states' decisions to raise highway speed limits as a purchase of time with lives, and recovered an implied value of about 1,540,000 USD in 1997 money — 3,009,956 USD today, a figure the same federal government's own guidance exceeds by 4.39 x. The 1973 Ford memorandum that became a byword for corporate callousness used the road-safety agency's own figure of the day, 200,000 USD, which is 1,500,957 USD in today's money. The arithmetic in that memo was not the scandal. The price was — one ninth of what the same agency would use now — and because no one was required to publish the price beside the decision, nobody had to defend it until it was far too late.

Where the method loses, stated as plainly as the cases above.

It loses the moment a population price is applied to an identified group. The EPA's 2003 analysis that discounted the value of lives over seventy by 37.0 per cent — the "senior death discount" — was arithmetically consistent with annuitising and morally indefensible, and it was withdrawn. The same failure recurs internationally: transferring a VSL between countries at an income elasticity of 1.00 means a country with one thirtieth the income carries a value of 440,000 USD, which licenses a cheaper guard rail there for a reason that has nothing to do with anyone living there. A number built from a population's own willingness to pay stops being defensible the second it is used to tell a particular person what they may have.

And this volume's own instrument has a matching weakness that must be said here rather than buried. Wellbeing valuation — the WELLBY at 13,000 GBP — rests on self-reported life satisfaction, and people adapt. Measured satisfaction after severe injury recovers far more than anyone expects it to, so a strictly wellbeing-based rule will systematically underweight preventing disability compared with a rule built on health states. That is a real defect, it is documented, and it means the fifth number cannot simply replace the other four. Every one of these instruments is a lens with a blind spot at its own centre, which is the argument for holding five of them rather than for choosing one.

The cut

Here is the thing that makes all of the above bearable, and it is one sentence.

Every one of these figures is the price of a decision, not the measurement of a person — and the entire feeling of obscenity that attaches to this conversation comes from mistaking the first for the second. The VSL does not contain a person; divide willingness to pay by risk and the individual cancels out of the equation like a common factor. Human capital does not contain a person; it contains a payroll forecast. The underwriter's multiple contains a household's income statement. The court's award contains a rule that had to be administrable across ten thousand cases. Read as prices for decisions, they are modest, checkable, improvable working tools. Read as verdicts on a human being, every single one of them is an insult — and the mistake that turns the tool into the insult is made by the reader, not by the arithmetic.

DREAM

What becomes ordinary

In the organisation that has understood this, the five numbers are all present, all named, and never confused with one another.

The board has passed a resolution stating the price the company uses for a statistical life, and it is not lower than the price its regulator uses. It is on one page of the risk appendix. Every capital paper that touches safety carries a line stating its cost per statistical life saved, and any paper whose implied price falls below the published one goes up rather than through. Nobody argues about whether lives can be priced, because that argument was settled by the resolution and the resolution is three years old.

Pay is discussed with the markdown in the open. When somebody asks what a role is worth, three answers are given and labelled: what the market clears at, what the role's marginal revenue product appears to be, and what the firm has decided to pay against that. The gap between the second and the third is a decision the firm owns and states, rather than a fact of nature it shelters behind. Naming the markdown does not raise every wage on its own. It makes the wage a choice, and choices can be argued with.

The training ledger exists. The accounts still follow IAS 38 — nothing here asks anyone to misstate a financial statement — but beside the statutory numbers there is a memorandum schedule showing the cost of building the current workforce's capability and the replacement cost of losing it. It is reviewed when the headcount paper is reviewed. It is the only reason anyone in the meeting can tell the difference between a saving and a liquidation.

And the health and wellbeing spend is appraised the way a bypass is appraised, in the same arithmetic, with the same discount rate, competing on the merits. People stop having to argue that it matters. They argue about the size of the coefficient, which is a much better argument to be having and one that evidence can settle.

Nothing in that description requires a change in law, in accounting standards or in anyone's convictions. It requires four documents that do not currently exist and one board resolution that takes ten minutes.

DESIGN

The structure that gets you there

First: separate the five questions on paper, once, and keep the page.

The question	The instrument	What it decides
What does one more hour add to revenue?	Marginal revenue product	Whether to hire, and the ceiling on the wage
What should we spend to cut a small risk across many?	Value of a statistical life	Whether the rule, the guard, the retrofit is worth it
What does this budget displace when it is spent?	Cost per QALY or per WELLBY	What a fixed pot buys, and what it therefore forgoes
What income disappears if this person is gone?	Human life value, human capital	The size of a policy, an award, an education investment
What transfer settles this dispute finally?	Tort and compensation rules	What a court or a fund pays, administrably

Most bad arguments in this territory are a slide between two rows of that table inside one sentence. Keeping the page open is most of the discipline.

Second: publish your price before you need it.

The single structural move in this chapter is that the price is set in advance, in the abstract, when no particular decision is riding on it. A price named before the case is a policy. The same price named after the case is a defence. Set it by board resolution, set it no lower than the relevant regulator's published figure, review it annually with inflation, and put the date on it.

Third: require the implied price to be computed, not the conclusion.

Do not ask capital papers to assert that a proposal is safe. Ask them for one line: cost per statistical life saved. A guard-rail programme costing 2,400,000 USD that avoids 6.0 statistical deaths over twenty years implies 400,000 USD per life — 33.00 x below the published price, which makes the decision trivial and makes it fast. The value of the discipline is that it makes the easy cases obvious, which is where most of the delay in safety spending actually lives.

Fourth: state which number you are using, every time, in the sentence itself. Not in a footnote. The sentence. "On a human-capital basis this is worth X." "At our published risk price this clears." Six words of labelling prevents the whole category error.

Fifth: put the person's own valuation in the loop where it exists. The VSL is derived from what people themselves charge for risk. Where you can ask the people actually bearing the risk what they would want spent — and workforces will tell you, precisely and quickly, if you ask about their own exposure rather than about safety in general — you are not softening the arithmetic. You are sourcing it correctly.

DESTINY

How it holds when nobody is pushing

Three things keep a published price alive, and one thing kills it.

It holds when it is in the paper template. A line in a capital appraisal form outlives every champion. If the form asks for cost per statistical life saved, the number gets computed in perpetuity by people who have never heard of this chapter.

It holds when it is indexed. The California cap is the cautionary case: an unindexed figure loses its meaning silently, and at 18.4 per cent of its original value it is no longer the policy anyone voted for. Write the escalation into the resolution and the number cannot be repealed by inaction.

It holds when the five questions stay separate. The collapse mode is consolidation — somebody asks for "the one number for a life" to simplify a deck, and the moment there is one number it will be used for all five decisions, three of which it is wrong for.

And here is what kills it. It dies when the published price is used against a named individual: to set one person's severance, to decline one patient's treatment, to argue that this particular claimant is worth less. The instant that happens the whole apparatus is correctly rejected by everyone watching, and it will not be revived for a decade. The price is for populations and for decisions taken before the people are known. Write that constraint into the resolution in the same breath as the number, because the resolution will outlive everyone who understood why.

DELIGHT

What it feels like

There is a specific relief in the meeting where the number finally appears. Not the relief of winning. The relief of being able to stop performing.

Before it, every safety conversation is a small moral audition: everyone must demonstrate that they care, and nobody can say the quantitative thing without sounding like a monster, so the discussion takes an hour and decides nothing. Afterwards, somebody says "this implies 400,000 USD per statistical life against our published figure" and the room says yes, and moves on, and the guard rail goes in six weeks earlier than it would have. The number did not make the room colder. It let the room stop proving it was warm.

And there is a second pleasure, quieter and longer. Once you can hold the five figures apart, the obscene feeling goes out of the subject entirely, and what remains is oddly tender. All five are attempts, by people who mostly never met each other, to take seriously in arithmetic something that everyone already knows in the body. They are clumsy. They are the best anyone has. You will find that you stop flinching at them, and start improving them, and that the improving is good work — the kind you would do on a Saturday.

OPERATIONALIZE THIS

At the level of finance

The instrument: a board-adopted life-price resolution, financed by a safety facility repaid out of experience-rating credit.

The philosophy above costs nothing and changes nothing until it is attached to a cash flow. Here is the cash flow, and it is one your insurance broker already understands.

The structure. Two documents. The first is a one-page board resolution setting the company's published price for a statistical life, its index, its review date, and the constraint that it may never be applied to a named individual. The second is a revolving facility — treasury to operating unit, internal first — that funds safety capital expenditure and is repaid out of the reduction in insurance cost that follows.

Why there is a real cash flow here. Workers' compensation and employers' liability premium is manual premium multiplied by an experience modifier that moves with claims history. The modifier is the repayment mechanism, and it is already audited by a third party, which is why this facility does not need a verification regime built from scratch.

  insured payroll                     40,000,000 USD
  manual rate                              0.025 of payroll
  manual premium                       1,000,000 USD / year
  experience modifier 1.05  ->  0.80
  annual premium saving                  250,000 USD
  facility size                          900,000 USD
  simple payback                            3.60 years
  return on facility                        27.8 %

At a 27.8 per cent return this is not a safety proposal competing against commercial projects for scarce capital. It is one of the better-returning uses of capital in the business, and it should be presented in exactly those terms to a treasurer who has never been asked to fund a guard rail before.

The balance-sheet treatment. Safety capital expenditure capitalises normally and depreciates over the asset life. The facility sits as an internal loan, with the premium saving hypothecated to service it. The one treatment worth arguing for with your auditors is the disclosure: a memorandum note stating the published life price and the cost per statistical life saved of the programmes funded. It changes no recognised number and it is the disclosure that makes the policy hard to quietly abandon.

The counterparty. Internally, treasury. Externally — once two internal cycles are complete — the employers' liability carrier itself, or a captive cell, both of which will price a documented reduction in claims frequency because that is their entire business. A carrier that can see your implied life price and your cost per statistical life saved is being shown better underwriting information than it usually gets, and it prices accordingly.

The number that decides it. Two inequalities, both on the front page:

   annual premium saving                          cost of the programme
  ------------------------  >  WACC        and    ---------------------  <  published life price
       facility size                              statistical lives saved

The first says the money is worth having. The second says the decision is worth taking. A proposal that clears both needs no rhetoric. A proposal that clears the second and fails the first is still right — it is a decision you take and fund another way, and saying so out loud is how the resolution proves it means something.

The first ninety days.

Day	Action	Artifact
1–15	Draft the life-price resolution; take the regulator's figure as the floor	The one-page resolution
16–30	Add the cost-per-statistical-life-saved line to the capital appraisal template	The amended template
31–45	Retrospectively compute the implied price of three decisions already taken	Three implied prices
46–60	Pull the experience modifier history and the manual premium from the broker	The premium base case
61–75	Size the facility; get the single signature; deploy on the clearest case	Facility memo
76–90	Publish the memorandum note; book the first modifier review	The disclosure

The exercise on days 31 to 45 is the one that changes minds, and it costs a morning. Compute what three decisions your organisation has already taken implicitly priced a life at. One of them will be low enough that nobody in the room will want to defend it — and nobody will have to, because the whole point of the resolution is that from this date forward the price is chosen rather than left behind.

APPRECIATIVE QUESTIONS

Twelve, for a room

Discovery — what is already working

Think of a decision here that protected somebody at real cost, and that everyone still agrees was right. What did the people who made it know, and how did they get it approved?
Where in this organisation does someone already compute a price for a risk — insurance, engineering, legal, clinical — and what have they learned that the rest of us have never been told?
When has naming a number out loud made a conversation here kinder rather than colder? What made that possible?

Dream — what becomes possible

If our published price for a statistical life were on one page and three years old, what would that free us to decide quickly that currently takes months?
Imagine the training ledger existed and was reviewed beside headcount. What would we stop doing within a year?
If every wage conversation named all three figures — what the market clears at, what the role produces, what we have chosen — what would improve first?

Design — what we build

What is the one line we could add to the capital appraisal template this quarter, and whose signature does that need?
Which three decisions already taken should we go back and compute the implied price of — and who would find that interesting rather than threatening?
What constraint should we write beside our number to make sure it is never used against a named individual?

Destiny — how it holds

What would keep this alive after everyone in this room has moved on, and which of those things can we build in the next ninety days?
How will we know if the number has quietly stopped being indexed — and who would notice first?
If somebody five years from now uses our published price for a decision it was never meant for, what did we fail to write down today?

WORKS CITED

Ashenfelter, O. and Greenstone, M. (2004). "Using Mandated Speed Limits to Measure the Value of a Statistical Life." Journal of Political Economy, 112(S1), S226–S267.

Brickman, P., Coates, D. and Janoff-Bulman, R. (1978). "Lottery Winners and Accident Victims: Is Happiness Relative?" Journal of Personality and Social Psychology, 36(8), 917–927.

Card, D., Cardoso, A. R., Heining, J. and Kline, P. (2018). "Firms and Labor Market Inequality: Evidence and Some Theory." Journal of Labor Economics, 36(S1), S13–S70.

Cengiz, D., Dube, A., Lindner, A. and Zipperer, B. (2019). "The Effect of Minimum Wages on Low-Wage Jobs." Quarterly Journal of Economics, 134(3), 1405–1454.

Claxton, K., Martin, S., Soares, M., Rice, N., Spackman, E., Hinde, S., Devlin, N., Smith, P. C. and Sculpher, M. (2015). "Methods for the Estimation of the National Institute for Health and Care Excellence Cost-Effectiveness Threshold." Health Technology Assessment, 19(14).

Dolan, P. and Kahneman, D. (2008). "Interpretations of Utility and Their Implications for the Valuation of Health." Economic Journal, 118(525), 215–234.

Feinberg, K. R. (2004). Final Report of the Special Master for the September 11th Victim Compensation Fund of 2001. US Department of Justice.

Feldstein, M. (2008). "Did Wages Reflect Growth in Productivity?" Journal of Policy Modeling, 30(4), 591–594.

Grenfell Tower Inquiry (2019, 2024). Phase 1 Report and Phase 2 Report. London: HMSO.

HM Treasury (2021). Wellbeing Guidance for Appraisal: Supplementary Green Book Guidance.

International Accounting Standards Board. IAS 38 Intangible Assets.

Jorgenson, D. W. and Fraumeni, B. M. (1989). "The Accumulation of Human and Non-Human Capital, 1948–1984." In Lipsey, R. E. and Tice, H. S. (eds), The Measurement of Saving, Investment and Wealth. University of Chicago Press.

Manning, A. (2003). Monopsony in Motion: Imperfect Competition in Labor Markets. Princeton University Press.

Mishel, L. and Bivens, J. (2021). Identifying the Policy Levers Generating Wage Suppression and Wage Inequality. Economic Policy Institute.

National Institute for Health and Care Excellence (2022). NICE Health Technology Evaluations: The Manual.

Robinson, J. (1933). The Economics of Imperfect Competition. Macmillan.

Sokolova, A. and Sorensen, T. (2021). "Monopsony in Labor Markets: A Meta-Analysis." ILR Review, 74(1), 27–55.

US Department of Transportation. Departmental Guidance on Valuation of a Statistical Life in Economic Analysis. Successive memoranda.

US Environmental Protection Agency (2011). The Benefits and Costs of the Clean Air Act from 1990 to 2020: Second Prospective Study.

US Environmental Protection Agency. Guidelines for Preparing Economic Analyses.

Viscusi, W. K. and Masterman, C. J. (2017). "Income Elasticities and Global Values of a Statistical Life." Journal of Benefit-Cost Analysis, 8(2), 226–250.

World Bank (2021). The Changing Wealth of Nations 2021: Managing Assets for the Future.

Note on figures. Every figure above is computed in lib/verify/V_01.py with its inputs, its units and its source printed beside it. CPI-U annual averages are the conversion basis for all money carried between years; the working exchange rate is stated rather than assumed. Where a published figure is contested — the size of the productivity-pay gap, the level of the NICE threshold, the elasticity of labour supply to the firm — the contest is stated in the text rather than resolved silently.