Federal Minimum Wage

Raise the federal wage floor from 7.25 to 17 dollars an hour in steps to 2030, index it to the median wage, and end the separate lower floors for tipped, young and disabled workers.

AI evaluation · not yet reviewed by a human

This evaluation was produced and sourced by an AI model; a human review is still pending. Figures and conclusions may still change. The review log is at the foot of the page.How review works →

The federal minimum wage has been 7.25 dollars an hour since July 2009 and applies in full only where a state has set nothing higher, which is the case in twenty states. The bill before Congress raises it in annual steps to 17 dollars in 2030 and then ties it to the median hourly wage, so that it moves without a further vote. It also phases out the three separate lower floors — 2.13 dollars for tipped workers, and the certificates that allow subminimum pay for young workers and workers with disabilities. Nothing in it changes who counts as an employee or which hours are paid. This evaluation looks five years ahead, at the years once the floor is fully in place.

Balance

Better for the future · 0.60 previous scale

Balance on the previous scale. The Bilanz 2.0 simulation is not yet available for this evaluation. The category comes from the share of the debate on the pro side (r).

For 29 · 60 % Against 20 · 40 %

Size class: large Scale of this evaluation: Normalised Impact — unitless, calibrated to this topic. For comparison: one point here is worth roughly 10 billion euro per year. This ledger turns on one judgement: that a dollar means more to a household near the bottom of the income distribution than to the customers and owners who hand it over. Treat the two sides as equal and the measure comes out slightly negative; at the difference used here it comes out slightly ahead, and the two sides stay close enough that the result should be read as a balance rather than a verdict. The second number that moves it is how many jobs disappear — the budget office's central estimate is 700,000, and its own range runs from none to 1.4 million. How we score →

Arguments for

Arguments against

8 arguments evaluated · Scoring v1.3 Δ absolute +9

Arguments — For

4 arguments · top 3 shown

Pay that reaches the household

24of 100

About 22.2 million workers, one in seven of everyone earning a wage, would be paid more, on average about 3,200 dollars a year each. Most of them are in the bottom two fifths of the income distribution, where the same dollar buys necessities rather than savings.

Value 5 · Household budgetsImpact 6.5Plausibility 7.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money moving from employers to the workers they employ, and money is priced at the middle of the scale wherever it appears. What makes this argument worth more than the one that pays for it is not the value class but where the money lands: a wage rise concentrated on households in the bottom two fifths does more than the same sum spread across customers and owners. That difference is carried in the Impact, not here, which is why the wage bill appears twice in this evaluation — once on this side at the recipients' weight and once against it at the payers'. Nothing is counted here for dignity, standing or the feeling of being paid properly, which are real and are not a separate good that can be added without counting the money twice. The part of the wage rise that public budgets claw back through lower benefit payments is taken out below and appears as its own argument. The value is the middle of the scale, and the whole weight of this argument sits in where the money lands rather than in what kind of good it is.

The Economic Policy Institute puts the wage bill at 70 billion dollars a year once the floor reaches 17 dollars, spread across 22.2 million workers, an average of about 3,200 dollars a year for those working year-round ^[1]. Two corrections come off that. Some of it is never paid, because minimum wage law is imperfectly enforced and violations concentrate in exactly the industries this reaches; 12 percent is taken off, in a range from 5 to 20 percent. Some of it is taken back in hours, since employers facing a higher hourly cost schedule fewer hours for the same work; 10 percent is taken off, in a range from nothing to 25 percent, which is where the disagreement between the American studies sits. That leaves 55.4 billion dollars, or 47.8 billion euro at 1.16 dollars to the euro. A quarter of it does not stay with the household, because higher earnings withdraw nutrition assistance, housing subsidy and health coverage as they rise; that quarter is booked separately as a gain to public budgets. The remaining 35.8 billion euro reaches households whose place in the income distribution gives a euro a weight of 1.8, in a range from 1.4 to 2.3. The Impact is the largest in this debate, and it is large because 22.2 million people is a seventh of everyone who earns a wage.

▸ Show calculation ▾ Hide calculation

	Additional wages once the floor reaches 17 dollars [1]	22.2 million workers, about 3,200 dollars a year each	70 billion dollars a year
×	Never paid, because the floor is imperfectly enforced Setting, range 5 to 20 percent: wage violations concentrate in the industries this reaches	− 12 %	61.6 billion dollars a year
×	Taken back in hours by employers Setting, range 0 to 25 percent: this is where the American studies disagree with one another [4]	− 10 %	55.4 billion dollars a year
÷	In euro exchange rate used throughout this evaluation	1.16 dollars to the euro	47.8 billion euro a year
×	Share that stays with the household the remaining quarter is withdrawn as benefits and is booked as its own argument, so that it is not counted twice	75 %	35.84 billion euro a year
×	Weight of a euro at these incomes Setting, range 1.4 to 2.3: most affected workers are in the bottom two fifths, some in the middle fifth	1.8	64.52 billion euro a year
÷	Normalised Impact scale of this evaluation	10 billion euro a point	6.45

Score 6.45 Impact × 5 Value × 7.5 Plausibility ÷ 10 = 24 of 100

The size of the wage bill is among the better-identified quantities in labour economics, and the disagreement is about employment rather than about the raises themselves. The counterfactual is the wage structure as it would stand with only state floors, which is what the bunching estimator of Cengiz, Dube, Lindner and Zipperer constructs from 138 state minimum wage increases: it counts the jobs that disappear from below the new floor and the jobs that appear just above it, and finds the two almost exactly offset ^[4]. That design is quasi-experimental and it is the one that carries this number. The confounder that matters is that states raise their minimum wages when their labour markets are already strong, which would make any raise look cheaper than it is; the estimator handles it by comparing each state against its own wage distribution rather than against other states. Reverse causation runs the same way and is answered by the same construction. What remains uncertain is the extrapolation: 17 dollars would be a higher bite relative to the median wage in Mississippi or Alabama than anything in the American record, so the compliance and hours corrections above are wider than the study's own error bars. It is held level with the two arguments that book the same sum on the other side of the ledger, because a wage bill that moves for the people who receive it moves for the people who pay it, and it cannot be more certain in one direction than in the other. Read back: in about seven of ten comparable cases a raise of roughly this size arrives. The Plausibility is high but not at the ceiling: the mechanism is measured, and the size of the step beyond the measured range is not.

evidence basis: Converging studies · P ceiling 8 identification: Quasi-experimental · rung ceiling 8

Counterfactual: the wage structure under state floors alone, reconstructed from each state's own wage distribution. Design: quasi-experimental — bunching estimator over 138 state minimum wage increases, counting missing jobs below the new floor against excess jobs just above it [4]. Confounder: states legislate raises when their labour markets are strong, which would understate the cost; handled by the within-state counterfactual. Direction: reverse causation runs through the same channel and is answered by the same construction. Ceiling: quasi-experimental 8.0 binds below the converging-studies ceiling of 9.0. The step beyond the measured range is carried in the compliance and hours bands, not in P. Held at 7.0 to match con-1 and con-2, which book the same sum on the paying side.

Public budgets pay out less

3.3of 100

Nutrition assistance, housing subsidy and health coverage all withdraw as earnings rise. About a quarter of the wage increase therefore never reaches the household at all — it stays in federal and state budgets instead.

Value 5 · Public financesImpact 1.2Plausibility 5.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is public money that does not have to be paid out, priced at the middle of the scale like any other money. It is the mirror image of the deduction made in the argument above: the same quarter of the wage rise, seen from the budget that no longer transfers it. Booking it here rather than leaving it inside the wage argument is what keeps the two ends of a transfer from being counted at the same weight, because a euro in a public budget counts once and a euro in a low-income household counts more. Nothing is counted for the administrative saving of processing fewer claims, which is small beside the transfers themselves. Whether the saving is spent, returned in tax or used to reduce borrowing makes no difference to the weight. The value is the middle of the scale, the level this site uses for public money whatever it is later spent on.

The wage increase reaching workers is 47.8 billion euro a year before the benefit system responds. Nutrition assistance withdraws at 30 cents on the dollar, the earned income credit phases out at 16 to 21 cents over its phase-out range, housing assistance takes 30 cents of additional income in rent, and health coverage subsidies taper with income as well. Not all of these apply to any one household and many affected workers claim none of them, so the combined effective withdrawal used here is 25 percent, in a range from 15 to 35 percent. That is 11.95 billion euro a year staying in federal and state budgets. Public money carries the standard weight of 1.0, which is what makes this argument smaller than the household argument built on the same sum. The figure is not a forecast of the budget effect of the bill as a whole, which would also have to count payroll and income tax on the higher wages and the loss of tax on the profits that fall. The Impact is a quarter of the wage bill, and it is the reason the household argument above is smaller than the headline figure.

▸ Show calculation ▾ Hide calculation

	Wage increase reaching workers before benefits respond [1]	from the argument above	47.8 billion euro a year
×	Withdrawn as benefits taper with earnings Setting, range 15 to 35 percent: statutory taper rates of nutrition, housing, credit and coverage programmes, weighted by how many affected workers stand inside a phase-out range	25 %	11.95 billion euro a year
×	Weight of a euro in a public budget the standard weight this site uses for public money	1.0	11.95 billion euro a year
÷	Normalised Impact scale of this evaluation	10 billion euro a point	1.2

Score 1.2 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 3.3 of 100

The withdrawal rates are written into the statutes and the schedules, so the direction is not in doubt; what is estimated is how many affected workers are actually inside a phase-out range. The counterfactual is the same benefit schedules with wages as they would otherwise be. The chain is named — higher earnings, benefit taper, lower outlay — and it has no behavioural link at the point that carries the quantity, except take-up: benefits that are never claimed cannot be withdrawn, and take-up of nutrition assistance runs around 80 percent while the housing programmes are rationed rather than entitlements. The confounder that would lower the figure is exactly that rationing, since a household on a housing waiting list loses nothing when its wage rises; it is named and not resolved. Reverse causation does not arise. The 25 percent is a construction from published schedules rather than a measured average effective rate, which is why the band is wide. The Plausibility is a little above the middle: the schedules are law and the share of workers standing inside them is assumed. Read back: in about five or six of ten comparable cases the benefit system takes back a share of this order.

evidence basis: Projection · P ceiling 6 identification: Mechanistic · rung ceiling 6

Counterfactual: the same benefit schedules with wages as they would otherwise be. Design: mechanistic — the withdrawal rates are statutory, but the share of affected workers standing inside a phase-out range is not measured. Confounder: rationed programmes such as housing assistance, where a household not receiving the benefit loses nothing as its wage rises; named and unresolved. Direction: no reverse causation. Ceiling: projection 6.0 binds, and mechanistic gives the same.

Fewer suicides

1.0of 100

Suicide among American adults without a college education moves with the minimum wage. The measured effect is a 3.4 to 5.9 percent fall in the rate for each additional dollar, and the step proposed here is far larger than a dollar.

Value 9.5 · LifeImpact 0.2Plausibility 6.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is people who are alive at the end of the period who would not otherwise have been, which is the good this site prices highest. It is placed half a step below the top because what the studies measure is a suicide rate in a population under financial pressure, and part of what moves that rate is distress rather than death alone. The money that relieves the pressure is counted in the first argument and is not counted again here; what is counted is a different consequence of the same payment, in the way that fewer road deaths and lower fuel bills are two separate consequences of one speed limit. Nothing is added for the families around them: the derivation counts the deaths themselves and leaves the households they belong to out, which makes the argument smaller than it should be rather than larger. The number of people involved is small beside the number who get a raise, which is why this argument is a fortieth the size of the first. The value sits just below the maximum: the stream is life, reached through relief of financial pressure rather than through medicine.

Kaufman and colleagues estimate that a one-dollar increase in the minimum wage is followed by a 3.4 to 5.9 percent fall in the suicide rate among adults aged 18 to 64 with a high school education or less, and that the effect is larger when unemployment is high ^[6]. About 49,300 Americans died by suicide in 2023, of whom roughly 20,900 were working-age adults in that education group ^[10]. Not all of them live where a federal floor would bind: about 45 percent do, in a range from 30 to 60 percent, since thirty states already set higher floors. The step from 7.25 to 17 dollars is far outside the range the studies cover, so the elasticity is applied at full strength only to the first three dollars and at half strength beyond, giving a 12 percent reduction in a range from 4 to 25 percent. That is about 1,129 deaths a year. The figure is deliberately built on the low end of the published effect and it still comes out larger than most single road-safety measures. The Impact is small beside the money in this debate and it is the argument with the highest value per unit, which is why it survives at a fortieth of the size.

▸ Show calculation ▾ Hide calculation

	Working-age adults with at most a high school education among American suicides [10]	about 49,300 deaths a year, roughly 42 percent in this group	20,900 deaths a year
×	Share living where a federal floor of 17 dollars would bind Setting, range 30 to 60 percent: thirty states already set a higher floor, and the group is concentrated in the states that do not [9]	45 %	9,405 deaths a year
×	Reduction in the rate Setting, range 4 to 25 percent: the measured 3.4 to 5.9 percent per dollar applied at full strength to the first three dollars and at half strength beyond, because the step is far outside the range studied [6]	12 %	1,129 deaths a year
×	Value of a life the value of a statistical life used across this site, 35 life years at 40,000 euro	1.4 million euro each	1.58 billion euro a year
÷	Normalised Impact scale of this evaluation	10 billion euro a point	0.16

Score 0.16 Impact × 9.5 Value × 6.5 Plausibility ÷ 10 = 1.0 of 100

The counterfactual is the same states in the same years without the minimum wage increase, which is what the study's difference-in-differences design over state-by-year variation constructs. The design is quasi-experimental, with pre-trends reported and an interaction with the state unemployment rate that behaves as the mechanism predicts — the effect is larger where the labour market is weak — which is the kind of internal check that makes a spurious result harder to sustain. The confounder that matters is that states raising minimum wages also expand health coverage, nutrition assistance and behavioural health funding in the same legislative sessions, and the paper controls for state and year but cannot separate a simultaneous package; that is unresolved and is the main reason for the discount. Reverse causation is implausible in this direction: suicide rates do not drive minimum wage legislation on the relevant timescale. The larger weakness is not identification but extrapolation, and that is carried in the halved elasticity above rather than here. The Plausibility is above the middle: one well-designed body of work supports the direction and size, and a simultaneous policy package cannot be ruled out. Read back: in about six or seven of ten comparable cases a fall in suicides of this order follows.

evidence basis: Study · P ceiling 7 identification: Quasi-experimental · rung ceiling 8

Counterfactual: the same states in the same years without the increase. Design: quasi-experimental — difference-in-differences on state-by-year minimum wage variation with reported pre-trends, plus an interaction with the state unemployment rate (Kaufman and colleagues, 2020) [6]. Confounder: states raising minimum wages also expand coverage and behavioural health funding in the same sessions; named, controlled only through state and year fixed effects, unresolved. Direction: reverse causation implausible — suicide rates do not drive wage legislation. Ceiling: set at 7.0, below the single-study ceiling, for the simultaneous-package concern; the extrapolation beyond the measured range is carried in the halved elasticity, not in P.

Fewer people quitting

0.2of 100

Low-wage jobs turn over once or twice a year, and every departure costs the employer a search, an induction and a period of lower output. Higher floors measurably slow that churn, and the saving lands with the same firms that pay the wage.

Value 5 · Company budgetsImpact 0.1Plausibility 6

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money and hours that firms no longer spend on replacing people: advertising a post, interviewing, inducting, and carrying the lower output of someone new. It is priced at the middle of the scale, like all money, and it sits on the employers' side of the ledger, partly offsetting what they pay out in the first cost argument. Nothing is counted here for what a stable job is worth to the worker, which is a different good and is not measured by any of the sources used. Nothing is counted for the customers of a business with experienced staff either, since that would be the same output measured twice. The saving is real rather than a transfer: the hours spent on hiring are consumed and do not reappear anywhere. The value is the middle of the scale, the level this site uses for money inside a company.

Dube, Lester and Reich find that a ten percent minimum wage increase reduces the separation rate of affected workers by about two percent, using pairs of counties on either side of a state line ^[7]. The 22.2 million affected workers separate from their jobs at something like 60 percent a year, which is 13.3 million separations. The average increase for affected workers is about 2.29 dollars an hour on a base near 12.50, an increase of 18 percent, giving a 3.8 percent fall in separations, or roughly 506,000 fewer a year. Replacing a worker in an hourly job costs on the order of 1,500 dollars once advertising, interviewing, induction and lower early output are included, in a range from 800 to 5,000. That is 759 million dollars, or 650 million euro at 1.16 to the euro. Company money carries the standard weight of 1.0. The Impact is the smallest on this side, which is the honest size of a saving that offsets a fiftieth of what the same firms pay out.

▸ Show calculation ▾ Hide calculation

	Separations a year among affected workers [1]	22.2 million workers at about 60 percent a year	13.3 million separations
×	Fall in the separation rate about two percent per ten percent of wage increase, applied to the 18 percent average increase for affected workers [7]	3.8 %	506,000 separations a year
×	Cost of replacing one hourly worker Setting, range 800 to 5,000 dollars: advertising, interviewing, induction and lower output in the first weeks	1,500 dollars	759 million dollars a year
÷	In euro, at the standard weight for company money exchange rate used throughout this evaluation	1.16 dollars to the euro, weight 1.0	654 million euro a year
÷	Normalised Impact scale of this evaluation	10 billion euro a point	0.07

Score 0.07 Impact × 5 Value × 6 Plausibility ÷ 10 = 0.2 of 100

The counterfactual is the same local labour markets without the state increase, which the border-pair design constructs by comparing adjacent counties that share a labour market but not a state law. That design is quasi-experimental and it is the one that carries this number. The confounder that matters is that a county on the high-wage side of a line may draw workers across it, so that lower separations reflect a bigger applicant pool rather than more satisfied workers; the paper addresses it by showing that hires fall alongside separations, which is what a stickier match looks like and not what a bigger pool looks like. Reverse causation does not arise, since a county cannot legislate its neighbour's wage. What is not measured is the replacement cost, which is a commercial estimate rather than a finding, and it is the wider of the two uncertainties here. The Plausibility is at the middle: the reduction in churn is measured on the right kind of design, and the price put on each avoided departure is not. Read back: in about six of ten comparable cases churn falls far enough to save something of this order.

evidence basis: Study · P ceiling 7 identification: Quasi-experimental · rung ceiling 8

Counterfactual: adjacent counties across a state line that share a labour market but not a minimum wage. Design: quasi-experimental — contiguous border-county pairs (Dube, Lester and Reich) [7]. Confounder: commuting across the line producing a larger applicant pool rather than stickier matches; addressed by the finding that hires fall alongside separations. Direction: no reverse causation, a county cannot legislate its neighbour's wage. Ceiling: set at 7.0 for the unmeasured replacement cost, which is a commercial estimate rather than a finding.

Arguments — Against

4 arguments · top 3 shown

Customers pay most of it

14of 100

The best evidence on who bears a large minimum wage increase says roughly three quarters of it comes back in prices. The goods concerned are restaurant meals, retail and care, which households buy across the income range.

Value 5 · Household budgetsImpact 3.9Plausibility 7

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money leaving households in higher prices, priced at the middle of the scale exactly as the wage gain is. This is the other end of the transfer: without it the wage argument above would count a gift rather than a redistribution, and the whole point of the ledger is that both ends appear. The weight differs, not the value class — a euro spent on a restaurant meal or a haircut comes from across the income distribution and slightly below the middle of it, because the goods most exposed to minimum wage costs are ones that lower-income households buy more of in proportion to their means. Nothing is counted here for the inconvenience of paying more, which is the money itself under another name. Nothing is counted for consumers who stop buying, which would be a further loss and is not estimated. The value is the middle of the scale, and this argument exists so that the wage rise above is not booked as if nobody paid for it.

Harasztosi and Lindner studied a Hungarian minimum wage increase of roughly 60 percent, the only episode of comparable size with firm-level data on both sides, and found that about three quarters of the cost was passed to consumers in prices while about a quarter came out of profits ^[5]. Applied to the 47.8 billion euro wage bill, that is 35.8 billion euro a year in higher prices. Households buying the affected goods — prepared food, retail, personal services, home care — sit across the income distribution but weighted a little below the middle, giving a weight of 1.1, in a range from 0.9 to 1.4. That is 39.4 billion euro a year. The result depends on the split rather than on the total: at a pass-through of half rather than three quarters, this argument falls by a third and the profit argument rises. What is not counted here is that some customers respond by buying less, which lowers the wage bill in the first place and is already inside the hours correction above. The Impact is the second largest in this debate and it is the same sum as the first, differing only in whose hands it is measured in.

▸ Show calculation ▾ Hide calculation

	Wage bill employers have to find [1]	from the first argument	47.8 billion euro a year
×	Share passed on in prices measured on a Hungarian increase of comparable size, the only episode with firm-level data on prices, employment and profits at once [5]	75 %	35.84 billion euro a year
×	Weight of a euro for the households buying these goods Setting, range 0.9 to 1.4: prepared food, retail, personal services and home care are bought across the income range and weigh a little more below the middle of it	1.1	39.43 billion euro a year
÷	Normalised Impact scale of this evaluation	10 billion euro a point	3.94

Score 3.94 Impact × 5 Value × 7 Plausibility ÷ 10 = 14 of 100

The counterfactual is the same firms without the increase, which the Hungarian study constructs by comparing firms with different shares of workers below the new floor. The design is quasi-experimental, it uses administrative firm-level data on prices, employment and profits at once, and it is the only study of an increase on this scale that observes all three — which is precisely why it is used here rather than an American study of a smaller step. The confounder that matters is sectoral: firms with many low-wage workers are concentrated in tradable and non-tradable sectors with different ability to raise prices, and a shift in the sector mix could produce the same measured pass-through without the mechanism; the paper addresses it by splitting the sample along exactly that line and finding the pass-through in the non-tradable sector where it should be. Reverse causation does not arise, since firms did not set the Hungarian minimum wage. The transfer to the United States is the real weakness — different market structure, different concentration, different labour share — and it is taken as a discount below the ceiling rather than a band. The Plausibility is well above the middle: the pass-through is measured directly on a comparable step, in a different country. Read back: in about seven of ten comparable cases most of the cost reaches customers in prices.

evidence basis: Study · P ceiling 7 identification: Quasi-experimental · rung ceiling 8

Counterfactual: the same firms without the increase, constructed from firms' differing exposure to the new floor. Design: quasi-experimental — firm-level administrative data on prices, employment and profits around a 60 percent Hungarian increase (Harasztosi and Lindner) [5]. Confounder: sector mix, since exposed firms differ in their ability to raise prices; addressed by splitting tradable from non-tradable and finding the pass-through where the mechanism predicts. Direction: no reverse causation. Ceiling: quasi-experimental 8.0 less one point for the transfer from Hungary to the United States gives 7.0.

Jobs that do not exist

3.4of 100

The budget office's central estimate is 700,000 fewer jobs once the floor reaches 17 dollars, with a range from none to 1.4 million. The people who lose them are the least skilled workers in the poorest states, which is the same group the measure is meant to help.

Value 5 · Household budgetsImpact 1.4Plausibility 5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is earnings that a household does not receive, priced at the middle of the scale like the earnings it does receive in the first argument. It is not given a higher class for being about employment: the loss being measured here is money, and the separate harm of being out of work is the next argument. What raises this above its face value is the same thing that raises the wage gain — the people concerned are near the bottom of the distribution, so a euro counts for more. Nothing is counted for the employer who does not fill the post, since the wage not paid is already outside the wage bill above. Nothing is counted for the output not produced, which would be the same job counted a second time. The value is the middle of the scale, and the weighting rather than the class is what makes this the largest cost on this side.

The Congressional Budget Office puts employment 700,000 lower once a 17-dollar floor is fully in place, with a range from roughly none to 1.4 million ^[2]. The jobs concerned pay near the old floor for something like 1,200 hours a year, about 16,000 euro. That is 11.2 billion euro of earnings a year. Not all of it is lost: some of those workers find other jobs, some receive unemployment insurance or nutrition assistance, and all of them get their hours back, which has a value of its own. Fifty-five percent of the gross earnings is taken as the net loss, in a range from 35 to 75 percent. Weighting by where these households sit gives 2.2, higher than the 1.8 used for the wage gain because job loss concentrates at the very bottom of the affected group. That is 13.6 billion euro a year. The Impact is a fifth of the wage gain it accompanies, which is the ordinary proportion when a policy raises the price of the least productive hour.

▸ Show calculation ▾ Hide calculation

	Jobs that do not exist once the floor is in place [2]	central estimate, range none to 1.4 million	700,000 jobs
×	Annual earnings of one such job	about 1,200 hours a year	11.2 billion euro a year
×	Share actually lost Setting, range 35 to 75 percent: other work, unemployment insurance and assistance replace part of it, and the hours come back	55 %	6.16 billion euro a year
×	Weight of a euro at these incomes higher than for the wage gain, because job loss concentrates at the bottom of the affected group	2.2	13.55 billion euro a year
÷	Normalised Impact scale of this evaluation	10 billion euro a point	1.36

Score 1.36 Impact × 5 Value × 5 Plausibility ÷ 10 = 3.4 of 100

This is the number the field genuinely disagrees about, and the disagreement is not about the design of any one study but about which studies generalise. The counterfactual is employment as it would be under state floors alone, which the budget office builds from the published elasticities rather than from a design of its own. Cengiz and colleagues find essentially no employment effect across 138 American increases; Jardim and colleagues find substantial hours losses in Seattle; Neumark and Shirley's survey of the whole literature finds a majority of negative estimates. The confounder that separates them is which comparison group stands in for a raising state — other states, other counties, or the same state's own wage distribution — and it is not resolved by any of them. Reverse causation is present in the raw correlation, since states raise wages in good times, and is what the competing designs are trying to remove. Because the finding is genuinely contested rather than merely uncertain, the plausibility is held at the middle of the scale whatever the individual designs would allow. The Plausibility is exactly at the middle: that some jobs go is not seriously disputed at this size of step, and how many is the open question of the field. Read back: in about five of ten comparable cases jobs go on something like this scale, which is how the budget office's own range reads — from none at all to twice the central figure.

evidence basis: Projection · P ceiling 5 identification: Mechanistic · rung ceiling 6

Counterfactual: employment under state floors alone, assembled by the budget office from published elasticities rather than from its own design [2]. Design: mechanistic — the projection aggregates studies whose designs disagree; no single identified estimate carries the quantity. Confounder: the choice of comparison group for a raising state, which is what separates the competing designs and is unresolved. Direction: reverse causation is present in the raw correlation, since states legislate in good times, and is exactly what the designs are built to remove. Ceiling: the contested state of the finding caps P at 5.0, below the projection ceiling of 6.0.

Profits absorb the rest

1.6of 100

The quarter that is not passed on in prices comes out of profits. Those profits are held far up the income distribution, which is why this quarter of the same sum weighs a tenth of what the wage gain does.

Value 5 · Company budgetsImpact 0.5Plausibility 6.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money that firms keep less of, priced at the middle of the scale like the rest of the transfer. It is separated from the price argument because the two land on different people: a price rise is paid by customers across the income range, a profit fall by owners, and American business equity is held overwhelmingly in the top tenth of the wealth distribution. Booking them together at one weight would hide the fact that this quarter of the transfer costs less in welfare terms than the other three quarters. Nothing is counted for firms that close, which is a different consequence and appears in the employment argument. Nothing is counted for reduced investment, which would be double counting: the profit is the investment. The value is the middle of the scale, and what makes this quarter cheap is who holds it, not what it is.

The same Hungarian evidence puts roughly a quarter of the cost on profits ^[5]. A quarter of 47.8 billion euro is 11.95 billion euro a year. Corporate equity in the United States is concentrated: the top ten percent of households hold about 87 percent of it and the top one percent about half. Weighting the loss by where it lands gives 0.4, the weight this site uses for the top tenth, in a range from 0.2 to 0.7 — the lower end if the losses fall on publicly traded firms, the upper end if they fall on small owner-operated businesses whose owners are nearer the middle. That gives 4.78 billion euro a year. The small end of the range matters more than it looks: minimum wage exposure is concentrated in franchised food service and small retail, where the owner is often not in the top tenth at all. The Impact is an eighth of the price argument built on the same money, which is what the weighting is for.

▸ Show calculation ▾ Hide calculation

	Wage bill employers have to find [1]	from the first argument	47.8 billion euro a year
×	Share that comes out of profits the remainder of the pass-through measured on the Hungarian increase [5]	25 %	11.95 billion euro a year
×	Weight of a euro where business equity is held Setting, range 0.2 to 0.7: the top tenth of households holds about 87 percent of corporate equity, but exposed firms are often small and owner-operated	0.4	4.78 billion euro a year
÷	Normalised Impact scale of this evaluation	10 billion euro a point	0.48

Score 0.48 Impact × 5 Value × 6.5 Plausibility ÷ 10 = 1.6 of 100

The counterfactual and the design are the same as for the price argument, and so is the source: the Hungarian data observe profits directly rather than inferring them ^[5]. The confounder is different, though. Which firms absorb rather than pass on depends on competition, and a market with more concentrated buyers pushes more onto profits than the Hungarian average; American food service is more franchised and more concentrated than the Hungarian sample, which argues for a larger profit share than used here. That is named and unresolved and is why the weight band runs up to 0.7. Reverse causation does not arise. The distribution of who owns the affected firms is a measured fact from the Survey of Consumer Finances rather than an assumption, but the match between that distribution and the specific firms exposed to a minimum wage is not. Who exactly holds the profits is a question of how much this costs rather than of whether the money moves, and it is already carried in the band from 0.2 to 0.7 above; it is not deducted a second time here. The plausibility therefore stands level with the price argument, which rests on the same measurement and on the same sum leaving the same hands. Read back: in about seven of ten comparable cases a quarter of the cost comes out of profits. The Plausibility is well above the middle: the split is measured directly, and where the profits are held is carried in the band rather than here.

evidence basis: Study · P ceiling 7 identification: Quasi-experimental · rung ceiling 8

Counterfactual: the same firms without the increase. Design: quasi-experimental — firm-level administrative data observing profits directly (Harasztosi and Lindner) [5]. Confounder: American exposed sectors are more franchised and more concentrated than the Hungarian sample, which would push more onto profits than assumed; named, unresolved, and carried in the weight band up to 0.7. Direction: no reverse causation. Ceiling: quasi-experimental 8.0 less one point for the context transfer gives 7.0. Held at 7.0, level with con-1 and pro-1, which book the same sum: the ownership question is a size question and sits in the weight band.

What losing the job does

0.7of 100

Losing work damages health beyond the lost income — measurably, and for years. The effect is best documented for long-tenure workers in mass layoffs, which is not this group, so the figure used here is a small fraction of theirs.

Value 9.5 · Life and healthImpact 0.1Plausibility 5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is health: the raised mortality, the drinking, the depression and the strain on a household that follow being put out of work. It is placed just below the top, in the class this site uses for life and health, because part of what is measured is death and most of it is illness that can be recovered from. It is separate from the earnings loss above in the way that the money and the harm of being defrauded are separate, and the test is the same — a household that chooses to stop working does not experience this. Nothing is counted for the family members around the person, and nothing for the wider community effects that the displacement literature also reports. What is priced is the displaced worker's own condition. The value sits just below the maximum: the stream is health, and health that most people recover.

Sullivan and von Wachter, using administrative earnings and death records, find that workers displaced in mass layoffs face a mortality hazard 50 to 100 percent higher in the year after and 10 to 15 percent higher two decades later, amounting to one to one and a half years of life expectancy each ^[8]. That population is not this one. Their workers had long tenure, firm-specific skills and no ready alternative; a worker who does not get a minimum wage job in 2030 is mostly young, mostly short-tenure, and mostly employed again within months. The figure used here is 0.05 quality-adjusted years each, in a range from 0.02 to 0.12 — about two and a half weeks, roughly a fiftieth of what the displacement literature measures. Across 700,000 people that is 35,000 quality-adjusted years, or 1.4 billion euro. The band runs further up than down on purpose: if the jobs lost are concentrated among older workers in the poorest states rather than among teenagers, the upper end applies. How far this group stands from the one that has actually been studied is not in the band at all — that distance is what holds the Plausibility at the middle of the scale. The Impact is a tenth of the earnings loss it accompanies, and the ratio would be much higher if the displaced group looked like the one that has actually been studied.

▸ Show calculation ▾ Hide calculation

	People who do not get one of these jobs [2]	central estimate of the budget office	700,000 people
×	Quality-adjusted years lost each Setting, range 0.02 to 0.12: the lower end if the displaced are mostly young and re-employed within months, the upper end if a larger share are older workers in the poorest states; the full distance to the one to one and a half years the mass layoff literature measures is not in this band but in the Plausibility [8]	0.05	35,000 quality-adjusted years
×	Value of the years the value of a healthy life year used across this site	40,000 euro each	1.4 billion euro a year
÷	Normalised Impact scale of this evaluation	10 billion euro a point	0.14

Score 0.14 Impact × 9.5 Value × 5 Plausibility ÷ 10 = 0.7 of 100

The counterfactual is workers in the same firms and years who were not displaced, which the study constructs from mass layoff events that removed a large share of a firm's workforce at once and therefore did not select on the individual. The design is quasi-experimental and it is among the cleanest in this literature, using linked administrative earnings and death records rather than survey reports. The confounder that matters is that firms shedding a large share of their workforce may be failing for reasons that also affect the health of everyone who worked there, which the event definition limits but does not remove; that is named and unresolved. Reverse causation is handled by the design, since an individual's health does not cause a firm-wide layoff. The weakness here is not the study but the distance between its population and this one, and that distance is what holds the plausibility at the middle of the scale rather than the quantity, which is already cut to a fiftieth. Read back: in about five of ten comparable cases the health harm behind a lost job appears at something like this size. The Plausibility is at the middle: the harm is well established for a group of workers who are not the ones this measure would displace.

evidence basis: Study · P ceiling 6 identification: Quasi-experimental · rung ceiling 8

Counterfactual: workers in the same firms and years who were not displaced. Design: quasi-experimental — mass layoff events in linked administrative earnings and death records (Sullivan and von Wachter) [8]. Confounder: firms shedding a large share of their workforce may be failing for reasons that independently affect worker health; limited by the event definition, unresolved. Direction: reverse causation handled by the design, since individual health does not cause a firm-wide layoff. Ceiling: quasi-experimental 8.0 less two points for the transfer from long-tenure displaced workers to marginal low-wage job seekers gives 6.0.

Summary

This is the largest measure on the English side of the site and it comes out barely on the positive side of the line. About 22.2 million people would be paid more, an average of 3,200 dollars a year, and the money lands where a dollar does the most work — that is the whole of the case for it, and it is a strong one. Against it stands the same sum, because somebody hands it over: roughly three quarters in higher prices for restaurant meals, retail and care, a quarter out of profits, and on top of that the budget office's 700,000 jobs that do not come into existence. The two sides are close enough that the result should be read as a near-tie rather than a verdict, and the thing that decides it is not a fact about the labour market but a judgement about whether a dollar means more at the bottom than in the middle. The employment estimate is the second hinge: at the top of the budget office's own range, 1.4 million jobs, the ledger tips the other way.

Outlook — effect over time

Better for the future · 0.60 previous scale

Sources

Economic Policy Institute: The impact of the Raise the Wage Act of 2025. epi.org
Congressional Budget Office: The Effects on Employment and Family Income of Increasing the Federal Minimum Wage. cbo.gov
Congress.gov: H.R. 2743, Raise the Wage Act of 2025. congress.gov
Cengiz, Dube, Lindner and Zipperer, Quarterly Journal of Economics: The Effect of Minimum Wages on Low-Wage Jobs. academic.oup.com
Harasztosi and Lindner, American Economic Review: Who Pays for the Minimum Wage?. aeaweb.org
Kaufman, Salas-Hernández, Komro and Livingston, Journal of Epidemiology and Community Health: Effects of increased minimum wages by unemployment rate on suicide in the USA. jech.bmj.com
Dube, Lester and Reich, Journal of Labor Economics: Minimum Wage Shocks, Employment Flows and Labor Market Frictions. journals.uchicago.edu
Sullivan and von Wachter, Quarterly Journal of Economics: Job Displacement and Mortality: An Analysis Using Administrative Data. academic.oup.com
Bureau of Labor Statistics: Characteristics of minimum wage workers. bls.gov
Centers for Disease Control and Prevention: Suicide data and statistics. cdc.gov

Last reviewed by Claude Opus 5 · September 6, 2026 · 2× AI, not yet reviewed by a human

September 6, 2026AI reviewClaude Opus 5re-scored
v2: Lohnbetrag als Paar verdrahtet (pro-1 ⇄ con-1 und pro-1 ⇄ con-2, jedes Bein mit dem Gewicht seiner eigenen Seite), dafuer P angeglichen (pro-1 7,5 → 7,0, con-2 6,5 → 7,0 — con-2 doppelte den Eigentuemer-Zweifel, der schon in der Gewichtsspanne steckt). i_spanne und normalisierung an allen 8, zwei Eintrittsgruppen, massstab_hinweis ohne r. Kategorie bleibt Besser (P(D>0) 0,83).
September 6, 2026AI reviewClaude Opus 5First evaluation
First evaluation: wage bill, pass-through split and employment effect derived from the budget office and the published minimum wage literature.

Evaluations are produced with AI support and reviewed on a schedule for new developments; human passes are marked separately.How we review →