When Flourishing and Output Diverge

Volume V — Labour, Value, Flourishing

THE PLATE

A woman in a cream shirt dress standing in a sunlit room, looking toward the windows. — Plate V.11Two Dials, One Hand.Two true instruments can disagree, and the disagreement is data. What you do next depends entirely on knowing which of them measures a flow and which measures the stock the flow is drawn from.

THE LETTER

This chapter closes Volume V, and it is the one to read if you intend to run anything on the evidence in the ten before it.

Those ten chapters made a case. A person's capacity is a stock with a regeneration rate, not a cost line. Care is infrastructure. Health is a return on capital. Meaning is an input with a measurable coefficient. Ownership changes what a downturn does to people. We believe all of it, the arithmetic is in the modules, and you can check every figure.

And the case is not that flourishing and output always move together. They do not. Sometimes they come apart for a quarter and then rejoin. Sometimes they come apart because one of the two instruments is lying. And sometimes — this is the part a serious volume has to say out loud — they come apart permanently, in a way no design removes, and somebody has to carry the difference.

So this chapter does three things. It takes the oldest and best-argued dispute in the field, the Easterlin paradox and the Stevenson–Wolfers reply, and works out what the two camps are actually claiming, which turns out to be very nearly the same thing measured over different windows. It walks the cases where well-being rose while output fell — a recession, an island under embargo — and says what was really happening in each, because two of them are not what they are usually made to mean. And then it hands you the operating part: when your own two measures disagree, which one is the leading indicator, and under what conditions is each of them the artefact?

There is a rule at the end of it, and it is short enough to carry. It is built out of the equation Volume I opened with, and Volume V closes on it because the whole volume has been one long worked example of it.

Nothing here asks you to choose between people and performance. It asks you to know, on any given Tuesday, which of your two numbers is telling you about the flow and which is telling you about the stock — and to know what it costs, and to whom, when the honest answer is that you cannot have both.

— The Editors

DISCOVERY

What is already working: the disagreement, conducted well

Begin with the positive core, and here the positive core is not a firm. It is a forty-year argument that two sets of serious people have conducted in public, with their data attached, and which has produced a better answer than either side started with.

Easterlin's question, asked properly. In 1974 Richard Easterlin asked whether economic growth improves the human lot and did something unusual for the period: he put three findings side by side rather than choosing one. Within a country at a point in time, richer people report higher happiness — reliably, and the gradient is steep. Across countries, the relationship looked weaker than that gradient predicted. And over time within a country, average reported happiness in the United States had not risen with income. Three facts, one of which embarrassed the other two. He published all three.

He has never stopped checking it. Easterlin, McVey, Switek, Sawangfa and Zweig (2010) took the question to 37 countries — seventeen developed, nine developing, eleven in transition — and reported that over the long term, ten years and longer, the relation between the growth of happiness and the growth of income is nil. That is a clean, falsifiable, repeatedly-restated claim about a specific window, and it has stood a great deal of contact.

Stevenson and Wolfers's reply, made properly. In 2008 Betsey Stevenson and Justin Wolfers brought the Gallup World Poll — a sample spanning countries from roughly $840 a head to roughly $127,000 a head, a range of about 150 to one — and reported a robust, clean, log-linear relation between subjective well-being and income, with no evidence of a satiation point, and with the within-country and between-country slopes closely similar. In 2013 they went back specifically to look for satiation at the top and again did not find it.

Notice what makes this exemplary. They did not argue that Easterlin's numbers were wrong. They argued that a different functional form and a much wider income range give a different reading of the same underlying relation — which is a claim that can be settled by computing, and in the next movement we compute it.

The adversarial collaboration that settled a neighbouring case. In 2010 Daniel Kahneman and Angus Deaton reported that emotional well-being stopped rising with income at around $75,000 a year. In 2021 Matthew Killingsworth, with far more experience-sampling data, found no satiation at all. Instead of trading rebuttals for a decade, the two camps ran a joint reanalysis. Killingsworth, Kahneman and Mellers (2023) located the answer: the plateau is real for roughly the least happy fifth of people and absent for everyone else. Two teams, opposite published findings, one procedure, one better answer, and nobody had to lose. That is the standard this chapter is trying to meet.

And the working cases, in firms, that are not in dispute at all.

Employee satisfaction and equity returns. Alex Edmans (2011) tracked a portfolio of the "100 Best Companies to Work For in America" from 1984 to 2009 and found a four-factor alpha of about 3.5 percent a year, roughly 2.1 percent after industry adjustment. Compounded over those twenty-six years that is 2.45 times the market, or 1.72 times on the industry-adjusted figure. This is the honest positive of the whole volume and we will read it carefully in a moment, because it says something sharper than it first appears.

Happiness and measured productivity. Oswald, Proto and Sgroi (2015) ran randomised happiness shocks against a piece-rate task and measured a productivity lift of about 12 percent — a controlled experiment, with a control group, on the causal direction.

Hours, cut, with output held. Iceland's public-sector trials between 2015 and 2019 moved a large share of the state and municipal workforce from 40 hours a week to about 35.5 — an 11.3 percent cut — with service output maintained or improved. Output held on eleven percent fewer hours is, mechanically, 12.7 percent more output per hour.

Four kinds of evidence — a market anomaly, a laboratory experiment, a national field trial, a forty-year argument between honest opponents. What they have in common is that in each, somebody was willing to publish the number that did not suit them. That is the discovery, and it is the condition for everything below.

THE ARITHMETIC

What works, what does not, and who pays when they cannot both be had

First, compute the line, because almost nobody in this argument does.

Take twenty-four countries spanning roughly $840 to $127,000 of GDP per head at purchasing power parity, with their Cantril ladder scores on a nought-to-ten scale. Regress the ladder on the natural logarithm of income. Every row, both sums and every term are printed by lib/verify/V_11.py.

  slope        b  =  0.92  ladder points per natural log unit
  intercept    a  = -3.34
  R²              =  0.86

Eighty-six percent of the variance in national average well-being, across a hundred-and-fiftyfold income range, from one line through the log of income. That is Stevenson and Wolfers's finding, reproduced on a napkin. The relationship is real, it is strong, and there is no plateau in it.

Now hold health, social support and freedom constant, as the World Happiness Report's own model does, and the coefficient falls to about 0.35 — roughly 2.6 times smaller. That is not a refutation. It is a specification: most of what income buys, it buys by buying those three things.

Second, read the same slope as a price. This is where the dispute dissolves.

A slope in logs is not a rate of gain. It is a price, and the price is proportional. From b = 0.92:

  gain per DOUBLING of income        b · ln2        =  0.64  ladder points
  income multiple for ONE point      exp(1 / b)     =  2.9 x

At the controlled slope of 0.35 the same arithmetic gives 0.24 points per doubling and 17 times income for one point. Convert that into time, which is the unit an economy actually spends:

  years of real growth to buy one ladder point out of ten
    at 1 %/yr        109 yrs        (287 at the controlled slope)
    at 2 %/yr         55 yrs        (144)
    at 3 %/yr         37 yrs         (97)

Here is the cut, and it is the only one this chapter makes. Easterlin and Stevenson–Wolfers are not disagreeing about the coefficient. They are disagreeing about whether fifty-five years is a long time.

Run the line forward inside a normal policy window and watch the paradox appear on its own, out of the same slope that is supposed to refute it:

Window at 2 % real growth	Income	Predicted gain	Share of the scale
10 years	×1.22	0.18 pts	1.8 %
20 years	×1.49	0.37 pts	3.7 %
30 years	×1.81	0.55 pts	5.5 %
50 years	×2.69	0.92 pts	9.2 %

Easterlin's long run is ten years and up. The log-linear relation that is said to refute him predicts, over his own window, a gain of under two percent of the scale. He looked, correctly, and found approximately nothing, because approximately nothing is what the other camp's own model says is there.

Take the United States, the series Easterlin actually used. Real GDP per head went from roughly $27,000 in 1972 to roughly $62,000 in 2022 — 2.30 times, or 1.20 doublings. Predicted ladder gain: 0.77 points on the bivariate slope, 0.29 on the controlled one. On the three-point General Social Survey item, 0.29 ladder points is 0.058 of a scale point — 2.9 percent of that scale's entire range.

Is that detectable? Against sampling error, yes: with 1,500 respondents and a standard deviation of 0.65, the standard error of the mean is 0.0168, so t is about 3.5. And that is precisely the trap. The dispute does not live in sampling error. It lives in everything that is not in that standard error — fifty years of changes in question order, survey mode, response rate and who answers the telephone, every one of which can move a mean by three percent of a scale. Which is why running the survey again settles nothing, and why two honest camps can read the same series in opposite directions without either of them being careless.

Third, the honest positive, read honestly.

Edmans's 3.5 percent annual alpha is the strongest single number in this volume: a flourishing investment that paid on the conventional measure, in public equity, over twenty-six years, 2.45 times cumulative. Take it.

Then read what an alpha is. A persistent abnormal return means the market was mispricing the thing. Employee satisfaction paid because it was not on the conventional measure — it was value the accounts could not see, and the return was the fee for seeing it first. Two consequences follow and both are useful. One: measuring flourishing is not a cost centre, it is an information advantage. Two: that advantage shrinks as the measure spreads, which is an argument for building the measure now rather than an argument that it will pay for ever.

Fourth, the cases where well-being rose while output fell — and what was actually happening.

The recession. Christopher Ruhm (2000) found on US state panels that total mortality falls by roughly 0.5 percent for each percentage point of unemployment. US unemployment went from 4.6 percent in 2007 to 9.3 percent in 2009 — 4.7 points — which on Ruhm's coefficient implies mortality 2.35 percent lower, or about 58,000 deaths not occurring in a year of 2.47 million. That figure is an illustration of magnitude applied out of its estimation sample, and it is labelled as such in the module.

And then the composition, which is the actual finding. Over the same two years the US suicide rate rose from 11.3 to 12.0 per 100,000 — about 2,100 additional deaths. Fewer road deaths, fewer cardiac deaths, more suicides, more despair. A recession is not good for people. It is bad for people in one way and incidentally protective in another, and a single mortality line hides both halves. Anyone who cites Ruhm without the second half is not reporting a finding; they are using one.

The island. Cuba's output fell about 35 percent between 1990 and 1993. Franco and colleagues (2007) documented an average adult weight loss of 5.5 kg across the population, with diabetes mortality 51 percent lower and coronary heart disease mortality 35 percent lower. It is regularly cited as a natural experiment in the health benefits of eating less and cycling more. What is cited far less often is that the same years produced an epidemic of optic neuropathy — roughly 50,000 cases — from nutritional deficiency. Two health measures improved and people went blind. The case is in this chapter because it refuses to be a story for either camp, and because it is the clearest available demonstration that "a well-being measure rose" and "people flourished" are different sentences.

The control that keeps it honest. Russia in the 1990s: output collapsed and life expectancy, life satisfaction and mortality all moved sharply the wrong way. Falling output is not a flourishing strategy. It is occasionally, partially, accidentally protective, and that is all.

Fifth — and this is the honest negative — the divergences that are real, permanent, and paid for by somebody you can name.

Three of them, worked, because a volume about flourishing that ends without naming who bears the cost has not finished the argument.

(a) Shift work. A plant can run 168 hours a week; a single day shift runs 40. Running the asset around the clock is a 4.2× multiple on capital utilisation and there is no design that obtains it without somebody being awake at four in the morning. The International Agency for Research on Cancer classifies shift work involving circadian disruption as probably carcinogenic to humans, Group 2A — assessed in 2007 and reaffirmed in 2019. Who bears it: the night-shift worker, in health, for a premium generally in the region of a tenth to a fifth of base pay. That premium is a price, and the honest statement is that nobody has established it is the right one.

(b) Unpaid care. The ILO (2018) counted 16.4 billion hours of unpaid care work performed every day — at an eight-hour day, 2.05 billion full-time equivalents — valued at about 9 percent of global GDP, some $11 trillion at 2011 purchasing power parity. Women perform 76.2 percent of those hours: 12.5 billion hours a day, or 1.56 billion full-time equivalents. This is not a divergence that a better index closes. The work is the reason the measured economy can function, it is excluded from the measure by construction, and who bears it is stated in that one number and nowhere in any set of national accounts.

(c) Commuting. Stutzer and Frey (2008) found that people with long commutes report systematically lower well-being and are not compensated for it by higher wages or cheaper housing — an hour each way requires something like 40 percent more income to leave a person as well off. On a £45,000 salary that is £18,000 a year against 460 commuting hours, a shadow price of £39 an hour that nobody pays. Agglomeration genuinely raises output. Who bears it: the commuter, uncompensated, which is what makes it a paradox rather than a trade.

And one divergence that has already been settled, in the workers' favour, and is invisible because of how we measure. Annual hours per worker in the United States fell from roughly 3,100 in 1870 to roughly 1,760 in 2019 — 43 percent given up. At today's hourly productivity and 1870's hours, output per worker would be 1.76 times what it is. That difference is not lost output. It is flourishing already bought and paid for, at an enormous price, by people who struck for it — and GDP records exactly none of it. It is the largest measured divergence in this chapter, and it went the right way.

DREAM

What becomes ordinary

In the organisation that has absorbed this, nobody argues about whether to believe the people number or the money number, because both are on the same page and the page says which is which.

The monthly pack carries two columns. One is the flow: revenue, output, units, margin. The other is the condition of the stocks the flow is drawn from — voluntary turnover, hours actually worked against hours contracted, sickness absence, the share of people who can name how the value they create reaches them, the internal fill rate on senior roles. Nobody calls the second column soft. It is the column that says whether the first column is repeatable.

When the two columns disagree, there is a procedure and everyone knows it. The disagreement is not an argument; it is a trigger, with a threshold, that opens a defined enquiry with a name against it and a date. The enquiry's first question is fixed: which of these two is measuring a flow, and is the flow being borrowed from the stock? Most of the time the answer is discovered inside three weeks and is undramatic.

Compensation committees ask what a quarter's outperformance was drawn from. Not as a challenge — as a line in the paper, with the stock measures for the same quarter beside it. A quarter that was bought out of a stock is still a good quarter. It is simply recorded as what it is, which is a withdrawal, and the board knows the balance.

And the lenders ask. The facility carries two key performance indicators rather than one, the margin moves on both, and the credit committee has learned what everybody who has lent against a single number eventually learns: one KPI can be met at the expense of the borrower's own capacity, and a lender is exactly the party who should mind.

Nobody in this organisation believes flourishing and output always agree. They have the cases where they do not, written down, with the names of the people who carried the cost. The night shift is paid a premium arrived at by a process somebody can describe. The carers are counted. The commute is priced into the location decision at something better than zero.

None of this requires a new index, a new philosophy or a change in the law. It requires two columns instead of one, a threshold, and the willingness to say who pays.

DESIGN

The rule, and the structure that runs it

Here is the operating rule this chapter exists to deliver. It is built out of the equation Volume I opened with — S = D / (R · r), scarcity as a ratio of demand to a stock times its regeneration rate — and it is short enough to carry.

Output is a flow. Flourishing measures the condition of the stock the flow is drawn from. A flow can always be raised for a period by drawing the stock down, and no flow measure can tell you whether that is what happened.

From which the two diagnoses follow directly.

When output is up and flourishing is down, output is the suspect. You are very likely looking at a withdrawal reported as income — the same error Volume I named in the fixed asset register, arriving through the payroll instead. Test it: is the gain concentrated in overtime, deferred maintenance, unfilled vacancies, postponed training, or a temporary rise in hours? Every one of those is a drawdown wearing a margin. The stock measure is your leading indicator and the output measure is the artefact — and the artefact expires, typically inside four to six quarters, when the stock cannot fund another withdrawal.

When flourishing is up and output is down, flourishing is the suspect — unless you can name the stock being rebuilt. That last clause is the whole of it. A genuine rebuild is nameable: people are being trained, the vacancies are being filled, the maintenance backlog is falling, the debt of unused leave is being paid down. If you can name it and measure it, flourishing is leading and output will follow on a stated horizon. If you cannot name it, you are probably looking at a measurement artefact, and there are three reliable ones.

The three conditions under which the flourishing number is the artefact.

It is self-reported and not anonymous. A survey whose results reach a line manager measures the relationship with the line manager. Run it through a third party, publish the response rate, and treat a rising score with a falling response rate as a falling score.
It has no anchor and has been running long enough to adapt. Hedonic adaptation is real: Brickman, Coates and Janoff-Bulman (1978) found lottery winners and people with severe spinal injury converging far closer to baseline than anyone expected. A pure self-report series drifts back toward its own mean regardless of conditions. Pair it with at least one behavioural measure — turnover, absence, internal applications — that cannot adapt.
It is in somebody's bonus and has no denominator. Any measure that pays somebody and states no denominator will improve. Publish what the measure did not look at, in the same table.

And the honest limit on adaptation, because the adaptation literature has its own honest negative. Lucas, Clark, Georgellis and Diener (2004) followed people through unemployment in a fifteen-year panel and found that life satisfaction did not return to baseline even after re-employment. Set-point theory is wrong in the one case where being wrong matters most. Do not use adaptation as a reason to discount a fall.

The structure, in three parts.

The paired series. Two measures, permanently, at the same cadence, in the same pack, on the same page. One flow, one stock. Neither is allowed to appear alone.

The threshold. A divergence is declared when the two move in opposite directions and both moves clear a materiality gate for two consecutive periods. The gate is not decoration; the arithmetic is in the next movement but one and it is the difference between a covenant and a nuisance.

The enquiry. One named owner, three weeks, one question — what is the flow being drawn from? — and a written answer that goes in the pack whether or not it is flattering. An enquiry with no publication requirement will always conclude that nothing is wrong.

DESTINY

How it holds when nobody is pushing

Three things keep a paired measure alive, and the same three kill it when they are absent.

Both numbers have to be in the same pack, on the same page. A stock measure kept in a people-function deck and a flow measure kept in the finance pack will never be seen diverging, because nobody holds both at once. The whole mechanism is a layout decision before it is anything else.

The threshold has to be written before the first divergence. A threshold agreed after a number has moved is a negotiation about that number. Write it cold, publish it, and let it fire on you once without being adjusted. The first time it fires and nobody moves the goalposts is the moment it becomes real.

Somebody has to own the enquiry who does not own either number. If the enquiry is run by whoever owns the flow, the answer will be that the stock measure is soft. If it is run by whoever owns the stock, the answer will be that the flow was borrowed. Internal audit is the natural home and is usually free.

Now the failure modes, named plainly.

It fails when the two measures are chosen to agree. Pick a flow and a stock that are mechanically linked — revenue per head and headcount — and the pair can never diverge, so it can never inform. The pair is only useful if it can disagree.

It fails when the threshold is set on sign alone, at which point it fires constantly, everyone learns to ignore it, and the mechanism is dead within a year. The arithmetic is in Operationalize This and it is the single most practical number in the chapter.

It fails when the divergence is real, permanent and nobody will say who is carrying it. The night shift, the carers, the commute. An organisation that declares a divergence and then describes it as a trade-off without naming the party on the losing side has learned the vocabulary and none of the discipline. Naming the party is the entire content of the finding.

And it fails when someone uses this chapter to argue that output does not matter. It does. Output pays the wages, funds the training, buys the shorter week and clears the covenant. The claim here is narrower and harder: output is a flow, flows can be borrowed, and a measure that cannot tell income from a withdrawal should never be the only number in the room.

DELIGHT

What it feels like

There is a specific relief in the meeting where the two numbers disagree and nobody has to pretend otherwise.

Before the pair exists, a divergence is a political event. Somebody's number is up, somebody's number is down, and the meeting becomes an argument about whose instrument to trust — which is really an argument about who is to be believed, conducted in the vocabulary of measurement. It is exhausting and it produces nothing.

After the pair exists, the same divergence is simply Tuesday. The threshold fired, the enquiry has a name against it, the question is fixed and everybody already knows what it is. Three weeks later there is a written answer, and about half the time the answer is nothing was drawn from anything, the two series just moved. Nobody has lost. The argument has been replaced by a procedure, and procedures are restful in a way that being right is not.

And then the better pleasure, which comes later. Somebody in a part of the business you have never visited puts a stock measure next to a flow measure in a paper you did not ask for, because that is now simply what a paper looks like here. They will not have thought about where it came from. That is the moment it stopped being a policy and became the house's idea of how to count.

OPERATIONALIZE THIS

At the level of finance

The instrument: a paired-KPI facility with a divergence standstill.

Sustainability-linked loans are established and your treasury will recognise the mechanics immediately: a revolving credit facility whose margin ratchets up or down against defined key performance indicators, verified annually, with the market convention for the ratchet running at roughly 2.5 to 7.5 basis points each way under the Sustainability-Linked Loan Principles. The standard form keys the margin to one KPI, and the standard form has the exact weakness this chapter is about: one KPI can be met out of the borrower's own capacity, and the lender is the party who should mind most.

So key it to two, and add the clause that makes the pair mean something.

The mechanics.

Two KPIs, one flow and one stock. The flow is whatever the credit is already underwritten against. The stock is a behavioural, non-self-reported measure of capacity — voluntary turnover, absence, internal fill rate on senior roles. Both verified by the same external assurance provider, same period, same report.
A two-way margin ratchet of ±10 basis points, earned only when both KPIs are met. Meeting one alone earns nothing. This is the whole structural change and it costs the lender nothing to write.
The divergence standstill. If the two KPIs move in opposite directions, and both moves clear one standard deviation of their own trailing series, for two consecutive periods, the ratchet freezes at its current level and the borrower delivers a written enquiry within ninety days: what the flow was drawn from, and what is being done. No default, no acceleration, no penalty. A freeze and a paper.
Term and review. Three to five years, with the KPI definitions re-baselined once at the midpoint on a published method, never on an ad-hoc basis.
Security. Unchanged. This clause changes information, not recourse, which is why a credit committee will pass it.

Why the threshold has to be gated, with the arithmetic. Take two series that are doing nothing at all. The probability they move in opposite directions in a given period is 0.50, so a trigger on sign alone fires on two consecutive periods 25 percent of the time — a quarter of all periods, on pure noise, and the covenant is dead within a year. Now require both moves to clear one standard deviation. The one-tailed normal probability is 0.1587, so opposite-and-both- material is 2 × 0.1587² = 0.0503 per period, and twice running is 0.253 percent — roughly once in four hundred. The gate improves the false- trigger rate by about ninety-nine times, and it is one sentence of drafting. One standard deviation is a policy choice; 1.5 or 2 are equally defensible. A trigger on sign alone is not.

The balance sheet treatment. The ratchet is an adjustment to interest expense and is accounted for as such — no reclassification, no new line, nothing that needs an audit opinion. The verification fee sits in administrative expense. This instrument is deliberately boring on the balance sheet, because a structure that requires an accounting argument does not get signed in the quarter you need it.

The counterparty. Your existing relationship bank, at the next refinancing, not as a new facility. Sustainability-linked structures are already on their shelf; you are asking for a second KPI and a standstill clause, not a new product. Bring the trigger arithmetic above to that meeting — the credit officer's first objection will be that the clause will fire constantly, and the answer is 0.253 percent.

The number that decides it. On a £40 million facility, a 10 basis point ratchet is worth £40,000 a year. A 3-percentage-point fall in voluntary turnover across 600 staff is 18 people not replaced; at a £45,000 salary and a fully loaded replacement factor of 0.75 — the reported range is 0.5 to 2.0 times salary — that is £607,500 a year. Verification costs £25,000.

  retention value  £607,500  /  ratchet  £40,000     =  15.2 x
  net value  £582,500  /  verification  £25,000      =  23.3 x

The ratchet is worth a fifteenth of the behaviour it prices. The margin is not the incentive and was never going to be. What the facility actually buys is an externally audited paired series that the organisation cannot quietly stop producing, and that series returns 23 times its own verification cost in one line item. Price the measurement, not the margin — and if the finance committee asks why a lender should care about turnover, the answer is that a lender underwrites the borrower's capacity to keep producing the flow, and this is the only clause in the document that looks at it.

The first ninety days.

Day	Action	Artifact
1–15	Choose the pair: one flow, one stock, capable of disagreeing	Two definitions, one page
16–30	Pull three years of both; compute each series' standard deviation	The trailing sigmas
31–45	Set and publish the threshold, cold, before any divergence	The written threshold
46–60	Name the enquiry owner — neither number's owner	Terms of reference
61–75	Both series into the standing monthly pack, same page	The pack, changed
76–90	Take the clause to the relationship bank with the trigger arithmetic	Facility term sheet

APPRECIATIVE QUESTIONS

Twelve, for a room

Discovery — what is already working

When have our people numbers and our money numbers told us different things, and what did we learn by taking the disagreement seriously rather than resolving it?
Which quarter in the last three years produced output we are genuinely proud of and left people better off than it found them — and what made that possible?
Who here has changed their mind about a measure because of evidence, and what did it take? What can we learn from how that happened?

Dream — what becomes possible

If our monthly pack carried one stock measure beside every flow measure, what would we start to see that we cannot see today?
Imagine the meeting where our two numbers disagree and nobody has to argue about whose instrument to trust. What is happening in that room instead?
If a lender priced our capacity as well as our output, what would we be able to attempt that we cannot attempt now?

Design — what we build

Which two measures would we choose that are genuinely capable of disagreeing with each other — and what does it tell us that the obvious pair cannot?
What threshold would we be willing to publish before we knew which way it would fire on us first?
Who in this organisation could run an enquiry into a divergence without owning either number, and what would we have to give them?

Destiny — how it holds

What would have to be true for both columns to still be in the pack when everyone in this room has moved on?
Where are we carrying a divergence that is real and permanent — and can we name, out loud, who is bearing the cost of it today?
What is the first sign we would see if one of our two numbers were quietly becoming decorative, and who would notice it first?

WORKS CITED

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.

Easterlin, R. A. (1974). "Does Economic Growth Improve the Human Lot? Some Empirical Evidence." In David, P. A. and Reder, M. W. (eds), Nations and Households in Economic Growth. Academic Press.

Easterlin, R. A., McVey, L. A., Switek, M., Sawangfa, O. and Zweig, J. S. (2010). "The Happiness–Income Paradox Revisited." Proceedings of the National Academy of Sciences, 107(52), 22463–22468.

Edmans, A. (2011). "Does the Stock Market Fully Value Intangibles? Employee Satisfaction and Equity Prices." Journal of Financial Economics, 101(3), 621–640.

Franco, M., Ordúñez, P., Caballero, B., Tapia Granados, J. A., Lazo, M., Bernal, J. L., Guallar, E. and Cooper, R. S. (2007). "Impact of Energy Intake, Physical Activity, and Population-wide Weight Loss on Cardiovascular Disease and Diabetes Mortality in Cuba, 1980–2005." American Journal of Epidemiology, 166(12), 1374–1380.

Haraldsson, G. D. and Kellam, J. (2021). Going Public: Iceland's Journey to a Shorter Working Week. Autonomy and Alda.

Helliwell, J. F., Layard, R., Sachs, J. D., De Neve, J.-E., Aknin, L. B. and Wang, S. (eds) (2023). World Happiness Report 2023. Sustainable Development Solutions Network.

Huberman, M. and Minns, C. (2007). "The Times They Are Not Changin': Days and Hours of Work in Old and New Worlds, 1870–2000." Explorations in Economic History, 44(4), 538–567.

International Labour Organization (2018). Care Work and Care Jobs for the Future of Decent Work. ILO, Geneva.

Kahneman, D. and Deaton, A. (2010). "High Income Improves Evaluation of Life but Not Emotional Well-Being." Proceedings of the National Academy of Sciences, 107(38), 16489–16493.

Killingsworth, M. A. (2021). "Experienced Well-Being Rises with Income, Even Above $75,000 per Year." Proceedings of the National Academy of Sciences, 118(4), e2016976118.

Killingsworth, M. A., Kahneman, D. and Mellers, B. (2023). "Income and Emotional Well-Being: A Conflict Resolved." Proceedings of the National Academy of Sciences, 120(10), e2208661120.

Loan Market Association, Asia Pacific Loan Market Association and Loan Syndications and Trading Association (2023). Sustainability-Linked Loan Principles.

Lucas, R. E., Clark, A. E., Georgellis, Y. and Diener, E. (2004). "Unemployment Alters the Set Point for Life Satisfaction." Psychological Science, 15(1), 8–13.

Oswald, A. J., Proto, E. and Sgroi, D. (2015). "Happiness and Productivity." Journal of Labor Economics, 33(4), 789–822.

Ruhm, C. J. (2000). "Are Recessions Good for Your Health?" Quarterly Journal of Economics, 115(2), 617–650.

Schor, J. B., Fan, W., Kelly, O., Gu, G., Bezdenezhnykh, T. and Bridson-Hubbard, N. (2023). The Results Are In: The UK's Four-Day Week Pilot. Autonomy.

Stevenson, B. and Wolfers, J. (2008). "Economic Growth and Subjective Well-Being: Reassessing the Easterlin Paradox." Brookings Papers on Economic Activity, Spring 2008, 1–87.

Stevenson, B. and Wolfers, J. (2013). "Subjective Well-Being and Income: Is There Any Evidence of Satiation?" American Economic Review, 103(3), 598–604.

Stutzer, A. and Frey, B. S. (2008). "Stress That Doesn't Pay: The Commuting Paradox." Scandinavian Journal of Economics, 110(2), 339–366.

World Bank. World Development Indicators — GDP per capita, PPP, current international dollars.

Note on figures. The twenty-four-country regression, the price of a ladder point, the window table, the US series, the survey resolution arithmetic, the divergence trigger probabilities and the facility arithmetic are all computed in lib/verify/V_11.py, which prints its inputs, its intermediate terms and every reader-facing figure in the form this page shows it. Figures taken from a named paper rather than derived here — Edmans's alpha, Ruhm's elasticity, the ILO's care hours, Stutzer and Frey's compensating differential, the four-day-week pilot's self-reported results — are printed there under REPORTED, NOT DERIVED.