Haute Lumière
Commerce · V.07 · MMXXVI · daylight
TEN CONCEPT BRIEFS · Chapter V.07 — Income, Floor, and Dignity
One page each. A reader who reads only these ten pages has the chapter.
BRIEF 1 — A Floor Is Two Numbers, Never One
The idea. Every income floor ever proposed is described by two numbers, and almost every public argument about one quotes only the first.
G the guarantee — what you get when you have nothing
t the claw-back — how fast it is taken away as you earnG decides whether the floor changes anybody's life. t decides whether anyone can afford it, and it decides what happens to the next pound a person earns. A proposal that states G and not t has not been costed. A proposal that states both has been, and can be argued with.
The identity that governs everything downstream. Break-even — the income at which the guarantee has been entirely withdrawn — is simply
B = G / tso t × B = G. The claw-back rate times the width of the band is the guarantee, always, in every design.
Worked example. A guarantee of £5,200 a year clawed back at 50 pence in the pound is fully withdrawn at £10,400. Halve the claw-back to 25 percent and break-even doubles to £20,800 — the same guarantee, a band twice as wide, and a much larger bill.
Why it matters. It means the three things people want — a decent guarantee, a gentle claw-back, and a low cost — cannot all be had. Pick two. Every real proposal is a position on that frontier, and naming the position is the whole of an honest argument.
You already know this because you have seen a discount that tapers. The deeper the discount and the slower it phases out, the more it costs the shop. This is that, with a country.
BRIEF 2 — The Withdrawal Band, and the Rate Inside It
The idea. The effective marginal rate is what a person actually keeps from the next pound, after every tax and every withdrawal. It is not the headline rate and it is frequently much higher.
The arithmetic, and the trap inside it. Rates do not simply add, because means tests usually taper what is left after tax. For a British worker on Universal Credit:
1 − (1 − 0.20 − 0.08) × (1 − 0.55) = 67.6%Income tax at 20 percent and National Insurance at 8 percent come off first; Universal Credit then removes 55 percent of the remainder. With a student loan repaying at 9 percent on top, it is 71.6 percent.
The comparison that should end the argument. The additional rate of income tax is 45 percent. The highest effective marginal rate in Britain is 67.6 percent — 22.6 points higher — and it is paid by someone earning around fifteen thousand pounds, not by the richest person in the country.
There is a second band at the other end: between £100,000 and £125,140 the personal allowance withdraws at £1 for every £2 earned, so the marginal rate is 1.5 × 40 = 60.0 percent, or 62.0 percent with National Insurance.
Why it matters. Every objection to an income floor's withdrawal rate is an objection to the system already running, stated more loudly. The choice on offer is not band or no band. It is where the band sits and who has to stand in it.
BRIEF 3 — Gross Cost Is the Least Useful Number
The idea. The headline cost of a universal payment is almost meaningless, because most of it is paid to people who fund it themselves within the same tax year.
Worked, for the United Kingdom. A floor of £100 a week per adult and £50 per child across 54,200,000 adults and 14,100,000 children:
| Gross | £318.5 bn |
| less personal allowance abolished | £108.1 bn |
| less National Insurance threshold abolished | £28.2 bn |
| less Child Benefit and the UC standard allowance | £40 bn |
| Net fiscal cost | £142.2 bn |
Gross to net is 2.24 to one. The gross figure is 12.4 percent of GDP and the net figure is 5.6 percent — 11.6 percent of public spending rather than 25.9.
What must not be netted off. Housing, disability and carer support are not replaced here and not counted as offsets. A floor that swallows the additional-needs system leaves a disabled person poorer than before and buys its own fiercest opposition with its own money.
Why it matters. Whichever side of the argument you are on, quoting the gross number is a tell. It says you have not opened the tax side of the ledger.
You already know this because you do not describe your mortgage as costing you its full value every year. You net the payment against what it replaced.
BRIEF 4 — The Break-Even Income
The idea. In any funded floor there is an income at which what you receive exactly equals what the funding costs you. Below it you gain; above it you pay. That single number is the political arithmetic of the whole proposal.
break-even = the floor / the rise in the tax rateWorked. Funding £142.2 billion needs 12.3 points on income tax, once the abolished personal allowance has brought £465.1 billion of income newly into charge and a penny on that widened base raises £11.6 billion rather than £6.9 billion. So:
£5,200 / 0.123 = £42,227The median gross individual income is £28,000, so break-even sits at 1.51 times the median, and a clear majority of adults are net receivers.
What the number cannot tell you. It is an individual break-even, not a household one, and households pool income. A two-earner couple on £35,000 each are individually below break-even and jointly above it. Any serious proposal computes both.
Why it matters. This is the number a finance ministry computes on day one and rarely publishes. Computing it yourself changes what you are able to say in a room: not this is affordable, which is an opinion, but the majority of people in this country are net receivers up to £42,227, which is arithmetic.
BRIEF 5 — Three Instruments, Not One Argument
The idea. There are exactly three mechanisms for putting a floor under income. Everything else is a variant. They are not rivals; they do different jobs, and they are costed on the same basis here.
| Net cost | Reaches | Band rate | |
|---|---|---|---|
| Universal payment | £142.2 bn | 68,300,000 | none — flat 40.3% everywhere |
| Negative income tax at 50% | £62.8 bn payment leg | 54,200,000 adults | 58.0% below £10,400 |
| Job guarantee, one million posts | £18.7 bn | 1,000,000 | none |
Read the last column before the second. The universal payment has no band at all — its cost is a visible, high, flat entry rate of 32.3 percent, 40.3 percent with National Insurance. The negative income tax is less than a third of the gross outlay and buys that saving entirely with the band. The job guarantee is cheapest of all and does nothing whatsoever for anyone whose constraint is not the absence of a job offer.
The per-person figures, and why they are not a verdict. The floor costs £2,083 per person reached; the job guarantee costs £18,716. Nine times more — and the comparison is wrong-headed, because the job guarantee buys a wage floor, a price anchor and a public output, while the floor buys an exit option.
Why it matters. Most of the public argument is two people advocating different instruments while believing they disagree about evidence.
BRIEF 6 — What a Trial Can and Cannot Support
The idea. Income floor trials are good evidence about a narrow question and are routinely quoted as evidence about a wide one. Four limits, and they are not caveats — they are the specification.
They are small. Stockton treated 125 people against 200 controls. From the published proportions the standard error on the treated arm's endpoint is 4.38 percentage points and on the difference 5.55. A finding can be real, point the right way, and still not be precise.
They are time-limited. Two years of guaranteed money is a different object from a permanent right to it.
They are not general equilibrium. Pay 125 people in a city and nothing moves in wages, rents or prices. Pay everyone and all three are endogenous.
They often test something other than their headline. Finland's sample was drawn entirely from people already unemployed and already on benefit. It is a first-rate test of removing conditions from an existing payment; it is not a test of a universal income, and its evaluators say so.
The honest summary of what they do support, across six studies, four continents and sixty years: the money is not spent on vice — under one percent on alcohol and tobacco in Stockton — it does not produce a withdrawal from work, and the largest measured effects are in health and schooling rather than in income.
Why it matters. A policy costed off a misread trial fails in production, and the misreading is almost always the same one: treating a two-year payment to a sample as evidence about a permanent payment to everybody.
BRIEF 7 — The Duration Finding
The idea. People respond to the length of a guarantee, not only its size. This is the single most important methodological result in the literature and it came from inside the experiments themselves.
The evidence. Seattle–Denver, the largest of the four American negative income tax experiments with some 4,800 families, randomised the duration of the guarantee as well as its level: a three-year arm and a five-year arm, same payment schedule. The five-year arm reduced hours by more than the three-year arm.
What follows, immediately. Every short trial understates the labour-supply response to a permanent programme. A two-year pilot that finds no effect on work has not shown that a permanent floor would have none. It has shown that a two-year payment does not.
The design that answers it. GiveDirectly's Kenyan study runs the same monthly payment — about $0.75 PPP a day, $22.81 a month — for two years in one arm and twelve years in another, against a third arm receiving the entire two-year total of $548 in a single payment, and a control. Three different instruments, randomised, in one study.
The other answer. Alaska's dividend has been paid to every resident every year since 1982. It is permanent, universal and general equilibrium, and its measured aggregate employment effect is not distinguishable from zero, with a 1.8 percentage-point rise in part-time work.
Why it matters. When somebody cites a trial at you, the first question is not how large but for how long, and did the participants know?
BRIEF 8 — General Equilibrium, and the One Study That Has It
The idea. A payment to a few people leaves prices alone. A payment to everybody changes wages, rents and prices, and those changes can take back most of what was given.
The one measurement. Dennis Egger and colleagues studied GiveDirectly transfers at the level of whole local economies — about $1,000 per recipient household, roughly $10,500 per treated village — and measured what happened to households who received nothing. The local fiscal multiplier was about 2.4, implying some $25,200 of local income created per village, with a price response of roughly 0.1 percent. The money circulated rather than inflating.
The boundary, stated plainly. That was a village economy with idle capacity and elastic local supply of both labour and goods. A national programme faces a labour market that can bind and a housing market that frequently cannot respond at all. The 2.4 does not transfer, and anyone quoting it as though it does is quoting it wrongly.
The case that should worry you. A floor paid into a housing market with fixed supply is a transfer to landlords with extra steps. The incidence of any transfer lands on whoever owns the scarce thing, and in most rich countries that is land.
Why it matters. It converts a slogan into an engineering requirement: a floor without a supply policy on housing is a subsidy to whoever owns the scarce thing. Pair them, or do not do it.
BRIEF 9 — Indexation, Universality, Ring-Fencing
The idea. Three design choices decide whether a floor survives a decade. They are made at the beginning or they are not made at all.
Indexation, or it dies of arithmetic. A floor fixed in nominal terms falls by whatever inflation is, annually, without anyone deciding anything — which is how most guarantees in history have ended. Index to median earnings rather than prices: prices hold it constant against the past, earnings hold it constant against the society, and the second is what a floor is for.
Universality, or it dies of politics. A payment to everyone has everyone as its constituency. A payment to the poor is defended only by the poor, which is the weakest coalition available. Alaska's dividend has survived four decades of fiscal pressure because every voter receives it. That is structure, not sentiment.
A ring-fenced source, or it dies of competition. A floor funded from general revenue competes annually with health and schools and loses, because its beneficiaries are diffuse and its rivals have hospitals. A floor funded from a named source — a sovereign fund, a land value charge, a carbon dividend — is defended by that source.
The measurable version of what a floor buys. In Dauphin, Manitoba, hospitalisations fell 8.5 percent against matched comparison communities, with accidents, injuries and mental-health admissions carrying most of it.
Why it matters. Most floors do not fail. They erode, which is cheaper to do and harder to notice, and all three of these choices are defences against erosion rather than against opposition.
BRIEF 10 — The Floor Facility
The idea. A national floor needs a legislature. An employer can build one this quarter, and the thing it insures is not low pay — it is variance in pay.
Why variance. An hourly workforce on a fluctuating rota faces monthly income that moves ten to twenty percent with no warning. That variance, not the level, is what drives the payday loan, the missed rent and the resignation.
The structure. Guarantee each hourly employee a minimum net monthly pay equal to a fixed share of their own trailing three-month median. Top up any shortfall. Not a loan, not advanced against future hours, never recovered.
The pricing, worked. 2,000 people, median net monthly pay £1,850, monthly standard deviation 12.0 percent of that — £222.00 — and a floor at 90.0 percent of trailing median, £1,665.00. The expected top-up is the shortfall integral:
E[(a − X)⁺] = σ · ( z·Φ(z) + φ(z) ), z = (a − μ)/σ = −0.8333
Φ(z) = 0.2023, φ(z) = 0.2819
top-up = £222.00 × 0.1133 = £25.15 per person per month£301.84 a year each, £603,688 a year across the workforce.
The number that decides it. Turnover is 60.0 percent — 1,200 leavers at a loaded £3,000 each, £3,600,000 a year. So
£603,688 / (2,000 × £3,000) = 10.06 percentage pointsIf the guarantee cuts annual turnover by more than 10.06 points, it is free.
Why it matters. It converts a question about dignity into a retention instrument with a published break-even, and the second version gets signed.
All figures in these briefs are computed in lib/verify/V_07.py and sourced in the chapter's Works Cited.