Welfare programs and work incentives

A cash grant pays income that does not depend on work. A benefit reduction rate $t$ withdraws $t$ dollars of benefit for each dollar earned, an implicit tax that cuts a recipient’s net wage from $w$ to $(1 - t)w$. Grants and clawbacks reduce work. An earned income tax credit pays benefits only on earnings, so it raises labour force participation, although it can reduce the hours of people who already work.

The intuition

A program pays $400 a week and withdraws 50 cents of benefit for each dollar earned. For a recipient who can earn $15 an hour, an hour of work raises income by $7.50, not $15. The program makes the recipient richer, which favours leisure (income effect), and halves the reward for work, which also favours leisure (substitution effect). The benefit runs out at the break-even point, where earnings reach $G/t$ = $800 a week, or 53.3 hours.

The figures below use one worker with utility $U = C \times L$, 110 hours a week and no other income unless stated. With no program she works 55 hours at any wage: with these preferences and no non-labour income, the income and substitution effects of a raise cancel.

A take-it-or-leave-it cash grant

Suppose a grant $G$ is paid only to people who do not work at all. The budget set is the original line plus a single point, $(T, V + G)$, above the endowment point. The person compares the indifference curve through that point with the best one reachable by working, and takes the grant if it is higher. Equivalently, the grant raises the reservation wage: the wage at which the best bundle with work lies on the indifference curve through the grant point. For this worker, working gives $U = (wT/2)(T/2) = wT^2/4$ and the grant gives $U = G \times T$, so she takes the grant whenever $w < 4G/T$.

Weekly grant (G)

$440

Market wage (w)

$15.00

Reservation wage with the grant

$16.00

Hours worked

0.0

Consumption

$440

Open in Graph Lab

Figure 1. At a wage of $15 the worker reaches $U_P$ by working 55 hours. A grant of $440 a week paid only to non-workers puts the point $G$ on a higher indifference curve, so she stops working. The grant raises her reservation wage to $4G/T$ = $16, the slope of the dashed budget line that just reaches $U_G$.

People who still prefer to work receive nothing, so a pure out-of-work grant does not change their hours. It moves people only at the extensive margin, and only out of work. Few programs make payment conditional on not working at all.

Benefit reduction rates

Most programs instead reduce the grant as earnings rise. With a guarantee $G$ and reduction rate $t$, consumption is

$$C = wh + V + \max\left(0,\ G - t\,wh\right).$$

On the program the budget line starts at $(T, V + G)$ with slope $-(1 - t)w$. It meets the original budget line at the break-even hours $h_{BE} = G/(tw)$, and beyond that point the original line applies. The kink points inward, so the budget set is not convex: a worker chooses the best point on the program segment or the best point off it, and compares the two.

For a worker who ends up on the program segment, both effects reduce hours:

• Income effect. The grant raises income at every level of hours. With leisure a normal good, she works less.
• Substitution effect. The net wage falls from $w$ to $(1 - t)w$, so leisure is cheaper. She works less.

Benefit reduction rate (t)

50%

Weekly grant (G)

$400

Hours with no program

55.0

Hours with welfare

28.3

Benefit received

$188

Net wage while on welfare, (1 − t)w

$7.50

Open in Graph Lab

Figure 2. A $400 weekly grant with a 50 percent benefit reduction rate. The program segment starts at the grant with slope $-(1 - t)w = -7.50$ and meets the original line at the break-even point, 53.3 hours. The worker moves from 55 hours at $P_0$ to 28.3 hours at $P_1$. At reduction rates of 70 percent or more she prefers to leave the program and work 55 hours.

A higher reduction rate strengthens the substitution effect for recipients but lowers the break-even point, so fewer people qualify and some leave the program to work full time. At $t = 1$ every dollar earned is taken back: the program segment is flat and a recipient gains nothing from working until earnings exceed the grant.

The guarantee, the reduction rate and the break-even income are linked by $h_{BE} \times w = G/t$. A generous guarantee with a low reduction rate reaches families far up the income distribution, which is costly and exposes more people to the program’s work disincentive; a low break-even income requires a high reduction rate, which taxes recipients’ work heavily. No design avoids this trade-off.

The negative income tax

A negative income tax pays everyone a guarantee $G$ that is withdrawn at rate $t$ as income rises, delivered through the income tax system. Its budget constraint is Figure 2’s, applied to the whole population. The four US negative income tax experiments found modest reductions in work: husbands reduced labour supply by about the equivalent of two weeks of full-time work, wives and single female heads of household by about three weeks, with consistent results across experiments.¹ Canada ran its own guaranteed annual income field experiment, MINCOME, from 1974 to 1979.²

Earned income credits and the Canada Workers Benefit

An earned income tax credit is a refundable credit computed on earnings. Its schedule has three ranges, each with its own net wage:

• Phase-in: the credit is a share $s$ of earnings, a wage subsidy. Net wage $(1 + s)w$.
• Plateau: the credit is at its maximum. Net wage $w$.
• Phase-out: the credit falls by $r$ per dollar of income. Net wage $(1 - r)w$.

At zero hours the credit is zero, so a non-worker gets no income effect, and the phase-in raises the return to the first hours of work. Participation can only rise. For people already working the effect depends on the range: ambiguous in the phase-in (substitution effect up, income effect down), negative in the plateau (income effect only) and negative in the phase-out (income and substitution effects both down).

Canada’s version is the Canada Workers Benefit (CWB). For 2025, the basic benefit for a single person in most provinces is 27 percent of working income above $3,000, up to $1,633, reached at $9,048 of working income. It is reduced by 15 percent of adjusted net income above $26,855 and ends at $37,742. For families the maximum is $2,813, the reduction starts at $30,639 and the benefit ends at $49,393; the lower-earning spouse can exclude up to $16,386 of working income when the reduction is calculated.³ A disability supplement adds up to $843, and up to half of the expected benefit is paid in advance in three instalments.⁴ Alberta, Quebec and Nunavut have reconfigured versions.⁵ In Alberta the credit is 21 percent of working income above $2,760, up to $1,750 for a single person, reduced at 15 percent above $26,159.⁶

Working income

$20.0k

CWB, most provinces

$1,633

CWB, Alberta

$1,750

Change in CWB from the next $100 earned

$0.00

Open in Graph Lab

Figure 3. The basic CWB for a single person with no dependants and no other income, 2025. In most provinces it phases in at 27 percent above $3,000 of working income, reaches $1,633, stays flat to $26,855 and falls by 15 cents per dollar to zero at $37,742. Alberta's version phases in at 21 percent to a higher maximum of $1,750 and ends at $37,826.

The US earned income tax credit depends on family size: its phase-in rates are 34, 40 and 45 percent for one, two and three or more qualifying children, and its phase-out rates are 15.98 and 21.06 percent.⁷

Taste for leisure (b)

1.6

Credit phase-in rate (s)

0.40

Hours, no program

15.6

Hours, cash grant

0.0

Hours, earned income credit

20.0

Open in Graph Lab

Figure 4. Two programs with the same $120 weekly maximum, for a worker earning $15 an hour with $650 a week of other income. A grant paid only to non-workers ends her 15.6 hours of work; an earned income credit raises her hours to 20, the end of its phase-in range. Workers with a weaker taste for leisure, who work long hours in the phase-out range, cut back under the credit; those with a stronger taste start working because of it.

Key idea

A welfare grant reduced as earnings rise lowers hours through both an income effect and a substitution effect. An earned income credit pays nothing at zero hours, so it raises participation; its plateau and phase-out reduce the hours of people already working.

Worked example

Welfare pays $G = 400$ a week with $t = 0.5$. The worker has $U = C \times L$, $T = 110$, no other income and a wage of $15.

1. Write the budget constraint. On the program, $C = 400 + 7.5h$ for $h \le 400/(0.5 \times 15) = 53.33$. Off it, $C = 15h$. Break-even earnings are $400/0.5 = 800$.

2. No program. $L = T/2 = 55$, so $h = 55$, $C = 825$ and $U = 825 \times 55 = 45{,}375$.

3. Best point on the program segment. This is a budget line with virtual income 400 and net wage 7.5: $L = (7.5 \times 110 + 400)/(2 \times 7.5) = 81.67$, so $h = 28.33$, inside the program range. Then $C = 400 + 7.5 \times 28.33 = 612.50$ and $U = 612.50 \times 81.67 = 50{,}021$.

4. Compare. The best point off the program is $h = 55$ with $U = 45{,}375$. Since $50{,}021 > 45{,}375$, she stays on the program, works 28.3 hours and receives $400 - 0.5 \times 15 \times 28.33 = 187.50$ in benefit.

5. Decompose the 26.7-hour fall. At the original wage, the bundle on her new indifference curve satisfies $C/L = 15$, so $15L^2 = 50{,}021$, $L = 57.75$ and $h = 52.25$. The income effect is $55 \to 52.25$ (2.75 hours). The substitution effect of the implicit 50 percent tax is $52.25 \to 28.33$ (23.9 hours). Both reduce hours.

Exercise: the CWB at $32,000

A single person in most provinces with $32,000 of working income and no other income: the phase-in gives $0.27 \times 29{,}000 = 7{,}830$, capped at $1,633. The reduction is $0.15 \times (32{,}000 - 26{,}855) = 771.75$, so the benefit is $861.25. Each extra dollar earned costs 15 cents of benefit.

Evidence

• Single mothers. The 1986 US tax reform expanded the earned income tax credit. Comparing single women with and without children, Eissa and Liebman find that single mothers’ participation rose by up to 2.8 percentage points relative to other single women, with no change in the hours of those already working.⁸ Meyer and Rosenbaum attribute a large share of the sharp rise in single mothers’ employment from 1984 to 1996 to the credit, with smaller shares to welfare cuts and other reforms.⁹
• Married couples. The US credit depends on family income, so many second earners face its phase-out on their own earnings. Eissa and Hoynes find that the expansions raised married men’s participation only slightly and lowered married women’s by over a full percentage point.¹⁰
• Canada’s Self-Sufficiency Project. In this randomized welfare experiment, the treatment group was offered a subsidy for full-time work, paid for up to three years to those who started full-time work within 12 months. Work rose; the effect peaked about 15 months after random assignment and then faded, with no lasting effect on wages or on welfare use.¹¹

Common mistakes

Traps on exams

• Forgetting the income effect of a clawed-back grant. The grant raises income and the clawback lowers the net wage; both reduce hours.
• Saying an earned income credit raises everyone’s hours. It raises participation. Hours fall in the plateau and phase-out ranges.
• Shifting the whole budget line for a take-it-or-leave-it grant. The grant adds one point above the endowment point; the rest of the budget line is unchanged.
• Confusing the break-even point with the reservation wage. Break-even hours are where the benefit runs out, $G/(tw)$; they say nothing by themselves about whether the person works.

On the exam

For a clawback, label the grant point $(T, V + G)$, the program segment’s slope $-(1 - t)w$, the break-even point and the original segment. Show $P_0$ without the program and $P_1$ on the program, then name the income and substitution effects. For a credit, label the three ranges with their slopes before discussing behaviour.

Check yourself

1.

A welfare program pays a weekly grant and reduces it by 50 cents for each dollar the recipient earns. Leisure is a normal good. Why does the program reduce the hours of a recipient who would otherwise work in the program's range?

1The grant raises income, so the recipient buys more leisure, and the clawback lowers the net wage, which makes leisure cheaper; both effects reduce hours2Only the clawback matters, because the grant is a lump sum that does not affect choices3The grant reduces hours, but the clawback increases them because the recipient must work more to reach the same income4The program does not change hours, because the income and substitution effects cancel

Check answer

2.

Does an earned income tax credit increase labour force participation?

1Yes. At zero hours the credit is zero, so there is no income effect, and in the phase-in range it raises the return to working2No. The credit raises income, so the income effect keeps people out of the labour force3Only if the phase-out rate is higher than the phase-in rate4It is ambiguous, because the income and substitution effects work in opposite directions for non-workers

Check answer

3.

Under the program in the previous question (a $400 weekly grant, a 50 percent benefit reduction rate, a wage of $15), at how many hours a week does Sam's benefit fall to zero?

Your answer (hours per week)

Check answer

Practise more questions on this topic

Sources

1. Robins, P. K. (1985). A comparison of the labor supply findings from the four negative income tax experiments. Journal of Human Resources, 20(4), 567 to 582. IDEAS record. ↩

2. Hum, D. and Simpson, W. (1993). Economic response to a guaranteed annual income: Experience from Canada and the United States. Journal of Labor Economics, 11(1), S263 to S296. IDEAS record. Forget, E. L. (2011). The town with no poverty: The health effects of a Canadian guaranteed annual income field experiment. Canadian Public Policy, 37(3), 283 to 305. doi:10.3138/cpp.37.3.283. ↩

3. Canada Revenue Agency, Line 45300: How much you can get, and Schedule 6, Canada Workers Benefit, 2025, which gives the 27 percent rate, the $3,000 base, the 15 percent reduction rate and the $16,386 secondary earner exemption. ↩

4. Canada Revenue Agency, Advanced Canada workers benefit: how much you can get. ↩

5. Canada Revenue Agency, Budget 2021: Canada Workers Benefit questions and answers. ↩

6. Canada Revenue Agency, Schedule 6 for residents of Alberta, 2025, and Line 45300: Who is eligible. ↩

7. 26 U.S. Code § 32(b), Earned income tax credit: credit and phaseout percentages. ↩

8. Eissa, N. and Liebman, J. B. (1996). Labor supply response to the earned income tax credit. Quarterly Journal of Economics, 111(2), 605 to 637. NBER working paper 5158. ↩

9. Meyer, B. D. and Rosenbaum, D. T. (2001). Welfare, the earned income tax credit, and the labor supply of single mothers. Quarterly Journal of Economics, 116(3), 1063 to 1114. NBER working paper 7363. ↩

10. Eissa, N. and Hoynes, H. W. (2004). Taxes and the labor market participation of married couples: The earned income tax credit. Journal of Public Economics, 88(9 to 10), 1931 to 1958. NBER working paper 6856. ↩

11. Card, D. and Hyslop, D. R. (2005). Estimating the effects of a time-limited earnings subsidy for welfare-leavers. Econometrica, 73(6), 1723 to 1770. IDEAS record. ↩