Labour Supply and Income Support Programs

---

EC306- Wages and Employment

Justin Smith

Wilfrid Laurier University

Fall 2026

Goals of This Section

Goals of This Section

In this section we will:

• Describe the main income support programs in Canada
• Use the labour supply model to evaluate a policy, step by step
• Analyze the work incentives of demogrants, social assistance, and negative income taxes
• Analyze wage subsidies and refundable tax credits like the Canada Workers Benefit
• Examine how economists measure these effects, including with experiments

Income Support Programs in Canada

Income Support and the Social Safety Net

Key Definitions

Income support programs: cash transfers meant to guarantee a minimum standard of living to people with low income, the core of Canada’s social safety net

Means-tested: eligibility depends on income or wealth

• Some programs target low income that is permanent, others low income that is temporary
• Social insurance is a separate group: EI, CPP, disability
  • It pays out based on past contributions, not current need, and we cover it in Chapter 5
• Both are income maintenance programs: they support incomes, and they change how much people choose to work
• That is the tension in this chapter: support versus work incentives

Two reasons to care about work disincentives, and the second is the one students do not expect. The first is the standard one: lost output when labour supply falls below where it would otherwise be. The second is Atkinson’s: a program can reduce poverty and still leave people cut off, because being out of work for a long time means being out of the networks that come with work. Also flag that everything in this chapter is static. Programs surely affect schooling, careers, and family decisions over time, but those dynamics wait for later chapters.

Government Transfers in Canada, 2022

In 2022, 95% of Canadian families received at least one government transfer

Benefit	Percent of families	Average amount
All government transfers	95.1	$13,969
Federal child benefit	21.5	$6,166
Provincial child benefit	12.0	$2,349
GST credit	51.8	$692
Social assistance	8.2	$11,270
Canada Workers Benefit	11.5	$1,347
Federal refundable tax credits	52.6	$677
Old Age Security	26.4	$10,122
Guaranteed Income Supplement	10.2	$6,900
Source: Canadian Income Survey (2022). Census families; average amount among families receiving the benefit.

This is the book’s Table 4.1, rebuilt from the same public-use microdata we used in Chapter 2. Every cell matches the published table except the OAS share, 26.4% here against the book’s 26.3%, a rounding difference. Point at the spread. Refundable credits like the GST credit reach about half of all families but pay a few hundred dollars. Social assistance reaches 8% of families and pays over $11,000 to those who get it. The EI and CPP rows from earlier versions of this table are gone on purpose: those are social insurance, and they belong to Chapter 5.

Issues in Program Design

Universal vs. Targeted Programs

• Universal: easy to implement but expensive
• Targeted: benefits only those in need but costly to administer
• Canada swapped the universal UCCB for the income-tested CCB in 2016

Work Disincentives

• Some programs reduce incentive to work
• Example: welfare benefits can encourage labour force dropout for those with strong preference for leisure

Permanent vs. Transitory Income Loss

• Different situations require different program designs
• Temporary layoff → short-term support
• Long-term disability → longer support + reintegration costs

iClicker Question

Which of these is a means-tested income support program, rather than social insurance or a universal payment?

A) Ontario Works payments to a single adult with no other income
B) EI benefits for a worker laid off from a plant
C) A CPP retirement pension
D) The UCCB’s $100 a month for a child under 6
E) Workers’ compensation after a workplace injury

Answer: A. Ontario Works is social assistance: eligibility depends on current income and assets, which is what means-tested means. B, C, and E are social insurance. They pay because you contributed and an insured event happened (a layoff, retirement, an injury), not because your income is low now, and they belong to Chapter 5. D is the trap for anyone who equates income support with any government cash: the UCCB went to every family with a young child regardless of income, so it is a demogrant.

Demogrants and Social Assistance Programs

Using the Model to Evaluate a Policy

• Same tool as Chapter 3: comparative statics, comparing choices before and after a change
• Policy rarely changes what people like; it changes what they can have
  • So the action is in the potential income constraint, not the indifference curves
• Every program in this chapter gets the same three steps:
  1. Draw the constraint without the policy
  2. Draw the constraint with the policy
  3. Given preferences, compare the chosen bundles

Key Definitions

Partial equilibrium effects: how the people a program reaches respond, holding wages and everything else fixed

General equilibrium effects: the knock-on effects once the program works its way through the whole economy

We stay in partial equilibrium for the whole chapter, and it is worth saying what that leaves out. Suppose a program pulls a lot of people out of work. Employers now compete for the ones left, wages rise, and some people come back, which softens the effect we started with. That feedback matters for real policy. The first question is simpler, though: does anyone respond at all?

Four Standard Income Support Programs

We will examine four policy approaches:

1. Demogrants – Fixed payment to a demographic group, such as everyone over 65, not means-tested
2. Social assistance (welfare) – Tops up income to a threshold for those with little else, clawed back dollar-for-dollar with earnings
3. Negative Income Tax (NIT) – A guarantee clawed back at a rate < 100% as earnings rise
4. Wage Subsidies – Addition to the wage rate

Demogrant: Theory

How it works:

• Unconditional lump sum transfer \(G\)
• Shifts budget line up parallel
• Wage unchanged → pure income effect

Labour supply impact:

• If leisure is normal → optimal leisure rises
• Reduces hours worked
• Raises reservation wage → some may quit

Work Incentive Effect

Unambiguously negative

Demogrants in Canada

Old Age Security (OAS)

• Monthly benefit for people 65 and over who meet residency rules
• Not a pure demogrant: it is taxed back from high-income seniors
• Together with the Guaranteed Income Supplement, more than $76 billion in 2024

Universal Child Care Benefit (UCCB)

• Paid to families with children from 2006 to 2016, whatever their income
  • Started at $100 a month per child under 6
  • From 2015: $160 a month under 6, $60 a month aged 6–17
• Taxable, so lower-income families kept more of it: not a pure demogrant either
• Replaced in 2016 by the income-tested Canada Child Benefit (CCB)

Pure demogrants are rare. Both of Canada’s examples are taxed, so higher-income recipients keep less. They are still worth the diagram, because most programs share the demogrant’s income effect. Hold the UCCB in mind: the research on what it did to parents’ work is in the evidence section, and Figure 4.2 is exactly the prediction those studies test.

iClicker Question

The UCCB paid families $100 a month per child under 6, whatever their income and whether or not they worked. Assuming leisure is a normal good, what does the labour supply model predict for a working mother’s hours?

A) More hours, because the money helps pay for child care
B) Ambiguous, because the income and substitution effects pull in opposite directions
C) No change, because the payment does not depend on whether she works
D) Fewer hours, through an income effect only
E) Fewer hours, through a substitution effect

Answer: D. The UCCB did not change her wage, so there is no substitution effect. It only raised her income, and with leisure a normal good she works less. A is the misconception worth naming: the program’s name says child care, but an unconditional cash payment is not a child-care price subsidy, since she gets it whether or not she buys child care or works. A subsidy tied to child care lowers the cost of working and can raise hours; that is the Quebec $5/day program from Assignment 1. B is the wage-subsidy answer, C assumes an unconditional payment changes nothing, and E has the right sign for the wrong reason. This is the prediction Schirle (2015) tests in the evidence section.

Social Assistance: Overview

Key Definition

Social assistance (income assistance, or welfare): means-tested provincial support for people with little or no other income

• Perfect targeting: top income up to a threshold \(z\): grant \(G = z - (wH + Y_N)\)
• Every dollar earned costs a dollar of benefit: a 100% clawback rate
• Most recipients have a disability or another barrier to full-time work
• Benefits are modest: Ontario pays a single employable adult at most $733 a month (2025), about 10 hours a week at minimum wage
• A lone parent with one child got 57–86% of the poverty line, depending on province (2023)
• Source: Laidley & Tabbara (2024)

The reality check matters for how students read the next three diagrams. Drawn with a generous flat segment, welfare looks like an easy choice over work. Scale the benefit to what provinces actually pay and the disincentive looks much less alarming. And for many recipients the straight-line constraint is the wrong picture anyway, because their barriers to work are not about the wage. The 57 to 86 percent range is the book’s Table 4.2; Nova Scotia is at the bottom and PEI at the top.

Social Assistance: Budget Constraint

Budget line structure:

• Black line: original budget (leisure at \(L_0\))
• Red then black: with social assistance up to the threshold \(z\)

Key features:

• Grant \(G = z\) at zero hours of work
• Reduced $1 for each $1 earned
  • Equivalent to 100% tax rate on earnings
• Income stays at \(z\) until earnings reach \(z\)
• After that, earn wage \(w\) per hour worked

Social Assistance: No Effect Case

For some workers, welfare has no effect:

• Person working \(H_0\) hours before benefit
• Still maximizes utility at \(H_0\) hours after
• Generally occurs with people already working many hours

Social Assistance: Work Disincentive Case

Others face strong work disincentive:

• Person chooses zero hours after welfare benefit

This occurs when:

• Welfare benefit is generous
• Individual places high value on leisure

Work Incentive Effect

This is the problematic effect of welfare programs

iClicker Question

Social assistance guarantees $733 a month, clawed back dollar for dollar with earnings. Maya earns $600 a month working part time. If she qualifies, what does the model predict?

A) She keeps working and gets a $133 top-up, so her income rises to $733
B) Nothing changes, because she was already working
C) She works more hours, to earn her way off assistance
D) We can’t tell without knowing whether leisure is a normal good
E) She stops working entirely

Answer: E. With the program, Maya’s income is $733 whether she works zero hours or her current hours, because every dollar she earns is clawed back. Same income, more leisure, so she quits. A is the most common pick: she would get a $133 top-up if she kept working, but that top-up is exactly what makes working pointless. D sounds careful, but there is nothing to trade off. The zero-hours bundle dominates her current one, so anyone who values leisure at all chooses it. This is the work disincentive case from the previous slide with numbers attached.

Negative Income Tax: Overview

Key Definition

Negative income tax (NIT): welfare program where benefit reduced at rate < 100% of earnings

• Example: guarantee \(z\) of $10,000, reduced by 50¢ per dollar earned

Advantages

Provides income support with fewer work disincentives. Still maintains incentive to work since benefit reduces slowly.

How it relates to basic income:

• A targeted basic income guarantee (BIG) works like an NIT (or like welfare, if the clawback is 100%)
• A universal basic income (UBI) pays everyone the same amount, so it works like a demogrant

Negative Income Tax: Budget Constraint

NIT has two components:

• Guarantee \(z\) at zero hours of work
• Implicit tax rate \(t\) on earnings, until the benefit is gone at the breakeven point \(B\)

Effect on labour supply:

• Pivots budget line up vs. pure welfare
• Work hours reduced from \(H_0\) to \(H_1\)
• Person still works some hours

Work Incentive Effect

Smaller disincentive than welfare for those welfare would push to zero hours

Negative Income Tax: Income and Substitution Effects

Income effects:

• Benefit increases non-labour income → increases leisure ↑
• Tax has negative income effect → reduces leisure ↓
• Net: benefit effect dominates → leisure ↑

Substitution effect:

• Tax rate reduces effective wage
• Makes leisure relatively cheaper → reduces work hours ↓

Overall Effect

Negative effect on work hours; it also reaches earners whom welfare leaves alone

Negative Income Tax: Math

The budget constraint with NIT:

\[ I = \begin{cases} wH + z - twH, & \text{if } z - twH \geq 0 \\ wH, & \text{otherwise} \end{cases} \]

Interpretation:

• Income = earnings (\(wH\)) + net benefit \(G = z - twH\), as long as \(z - twH \geq 0\)
• Otherwise just earnings (\(wH\))
• The same thing in one line: \(I = z + (1-t)wH\), with \(G = \max(z - twH,\, 0)\)

Breakeven point \(B\): the benefit reaches zero when \(wH = \frac{z}{t}\)

iClicker Question

A negative income tax guarantees $12,000 a year and claws back 40% of earnings. Priya earns $20,000. What is her total income?

A) $12,000
B) $20,000
C) $24,000
D) $27,200
E) $32,000

Answer: C. Her benefit is 12,000 − 0.4 × 20,000 = $4,000, so her income is 20,000 + 4,000 = $24,000. Equivalently, I = z + (1 − t)wH = 12,000 + 0.6 × 20,000. E forgets the clawback. B applies a 100% clawback, which is social assistance, not an NIT. D applies the 40% to the guarantee instead of to earnings. A is the guarantee alone. If there is time, ask for the breakeven: z/t = $30,000.

Basic Income in Canada: MINCOME

Manitoba Basic Annual Income Experiment, 1975–1979

• A targeted basic income, so an NIT, offered to eligible families in two places
  • Winnipeg: families assigned to plans or to a control group
  • Dauphin: a saturation site, where every resident was eligible, with a 50% clawback
• Plans varied both the guarantee and the clawback rate
• The idea had real policy weight: the Macdonald Commission (1985) recommended a national NIT for low-income families

Be careful with Dauphin. Everyone in town was eligible, not everyone received a payment; families with enough other income got nothing, exactly as an NIT predicts. The saturation design is what makes Dauphin interesting, because it can pick up community effects like changing norms about work, but it also means there is no control group inside the town. Winnipeg is the cleaner comparison for labour supply.

MINCOME: Results

Labour supply effects:

• Hours fell, but not by much (Hum & Simpson 1991)
  • Men: 0.8–1.6%
  • Married women: 2.4–3%
• The US income maintenance experiments found much larger drops for husbands and wives (Hum & Simpson 1993)

The debate continues:

• Basic income is back on the policy agenda in Canada
• Koebel & Pohler (2019): grow the Canada Workers Benefit into a basic income run through the tax system
• Open questions: cost, financing, and federal–provincial cooperation

The model says an NIT cuts hours; MINCOME says yes, by a little. Why the Canadian estimates are so much smaller than the American ones is still an open question, and researchers are still testing how robust the Winnipeg numbers are. The Koebel and Pohler proposal is a good bridge to the next section: it bolts a guarantee onto a refundable tax credit we are about to study.

Other Ways to Reduce Work Disincentives

Policy options:

1. Reduce the guarantee \(z\)
   • But then support may be inadequate for those who cannot work
2. Move toward an NIT
   • Clawback below 100%; most provinces now let recipients keep some earnings

3. Change preferences
   • Stigma, or support for getting into work: flatter indifference curves
   • Difficult to achieve
4. Raise the wage
   • Training, job information, minimum wages, unions
   • Or subsidize the wage directly: next section

Options 3 and 4 are the two panels of Figure 4.6: flatten the indifference curve, or steepen the budget line, and the same person moves off assistance into work. Do not draw option 2 for them. Problem 2 asks them to work out what an earnings exemption does to the constraint.

Wage Subsidies and Refundable Tax Credits

Wage Subsidies: Theory

Key Definition

Wage subsidy: payment to workers that increases their wage rate

• Example: $5/hour subsidy increases $15/hour wage to $20/hour

Effect on budget constraint:

• Budget line pivots upward

Labour supply effects:

• Income effect → increases leisure ↑

• Substitution effect → decreases leisure ↓

Work Incentive Effect

Overall effect on work hours is ambiguous

Wage Subsidies: Diagram

Budget line pivots up with subsidy

• Graph shows person working more
• But effect is theoretically ambiguous
• Income and substitution effects work in opposite directions

Wage Subsidies: Policy Implications

Advantages

• Less negative work incentives than welfare or NIT
• No reduction in effective wage rate

Disadvantages

• Do not provide income support to non-workers
• High earners qualify too, so it is a very expensive way to reach those in need

In practice:

• Pure wage subsidies are rare, because of that cost
• The fix is to target: pay the subsidy only to low earners
  • Canada’s version is the Canada Workers Benefit

The textbook’s point is that a pure wage subsidy has the demogrant’s problem in a different form: everyone who works collects, including people who need nothing. Targeting it on low earnings is what makes it affordable, and that is the design of the CWB. Subsidies paid to employers, like the pandemic wage subsidy, come up later when we get to labour demand.

iClicker Question

Which program does more to draw people who are not working into the labour force?

A) A negative income tax, because it gives low-income people more money
B) A wage subsidy, because it pays only when you work, raising the return to the first hour
C) They are the same, because both raise the incomes of low earners
D) Neither, because both create income effects that reduce work
E) A negative income tax, because its clawback is below 100%

Answer: B. For someone at zero hours, an NIT pays the full guarantee without any work, an income effect that makes staying out more attractive. A wage subsidy pays nothing at zero hours and raises the return to the first hour worked, so it can only pull people in. A is the intuition that more money means more help, which is true for income but backwards for participation. E is a real advantage of an NIT over social assistance, not over a wage subsidy. Tax Credits and Participation, with Figure 4.8(b), comes back to this.

Refundable Tax Credits

Key Definitions

Non-refundable: can reduce tax owed to zero (but not below)

Refundable: can reduce tax owed below zero (results in payment)

Income-tested wage subsidies paid as refundable credits:

• Canada Workers Benefit (CWB) – Canada, since 2007 (first called the WITB)

• Earned Income Tax Credit (EITC) – US, since 1975, now one of the largest US income supports

• Similar programs: the UK’s former Working Families Tax Credit and New Zealand’s In-Work Tax Credit

• Popular across the political spectrum: targeted, pro-work, and cheap to run through the tax system, with no caseworkers

• Source: Nichols & Rothstein (2015); Blundell (2006)

The Canada Workers Benefit, 2024

• Nothing until earnings pass $3,000
• Phase-in: 27 cents per dollar earned above $3,000
• Flat range: the $1,590 maximum, reached at about $8,900 of earnings
• Phase-out: clawed back at 15% of earnings above $26,149
• Gone entirely at $36,749
• A small program next to the EITC: low benefits and low thresholds

These are the book’s example parameters, and the publisher’s own CWB example file uses the same ones. The real program varies with province and family type. Problem 4 gives them the EITC’s parameters and asks what a $3,000 entry threshold like this one would do to it, so leave that question alone here.

Tax Credits: Budget Constraint

Three segments:

1. Phase-in: Subsidy proportional to earned income, up to income \(I_L\)
2. Flat range: Maximum benefit level, between \(I_L\) and \(I_H\)
3. Phase-out: Benefit reduced with earned income above \(I_H\)

Combines elements of:

• Wage subsidy (phase-in)
• Demogrant (flat range)
• NIT (phase-out)

Tax Credits: Effect in Phase-In Range

Person in phase-in range:

• Sees increase in effective wage
• Acts like pure wage subsidy
• Income and substitution effects in opposite directions

Work Incentive Effect

Effect on hours is ambiguous

Tax Credits: Effect in Flat Range

Person initially in the flat range:

• Tax credit gives maximum benefit
• Acts like pure income effect
• If leisure is normal good → reduce work hours

Tax Credits: Effect in Phase-Out Range

Person in phase-out range:

• Similar to negative income tax
• Income effect → reduce hours ↓
• Substitution effect → reduce hours ↓

Work Incentive Effect

Effect on work is unambiguously negative

Tax Credits and Participation

Source: Benjamin et al. (2026), Figure 4.8(b)

Key Result

Effect on hours is ambiguous, but the incentive to participate rises

• For a non-worker, the phase-in raises the return to the first hours of work
• Against social assistance with a 100% clawback, the contrast is sharper
  • Without the credit: no work, collect assistance (\(E_0\))
  • With it, if the benefit is large enough: work, on a higher indifference curve (\(E_1\))

This is the answer to the clicker question from the wage subsidy section, and it is the result in this chapter most worth remembering. The intensive margin is a mess of competing effects; the extensive margin is not. Problem 6(a) asks them to compare the CWB with perfectly targeted social assistance for someone indifferent between the two, so describe this picture but do not work that version through.

iClicker Question

Ali is single, lives in Ontario, and earns $30,000 a year. Compared with no CWB, what does the CWB do to his incentive to work more hours?

A) Reduces it: the income and substitution effects both cut hours
B) Reduces it, through an income effect only
C) Ambiguous: a higher effective wage, but also an income effect
D) No effect: he earns too much to qualify
E) Increases it: the clawback raises the payoff to each extra hour

Answer: A. At $30,000 Ali is in the phase-out range, between $26,149 and $36,749. The benefit he still receives raises his income, an income effect toward less work, and the 15% clawback lowers his take-home pay on each extra hour, a substitution effect toward less work. Both point the same way. C is the phase-in answer and B is the flat-range answer, so wrong picks show who misplaced him on the schedule. E gets the sign of the clawback backwards: it lowers the payoff to an extra hour. D misreads the cutoff.

Empirical Evidence of Incentive Effects

Social Assistance in Ontario, 1980–2024

Walk the picture left to right before asking anything. The share of Ontarians on assistance climbed through the 1980s, spiked in the early 1990s, and fell hard in the second half of the decade. Benefits tracked it closely until the 1995 cuts, when the province also brought in workfare. There is a smaller rise around 2008 to 2012. Then ask the question the next slide answers: does this picture show that generous benefits cause people to go on assistance?

Correlation Is Not Causation

• Benefits and recipiency move together, but which way does causation run? We saw this problem in Chapter 2
• Reverse causation: a big caseload is a big constituency, and governments raised benefits to match
• Confounding: unemployment drives both; recessions push people onto assistance, booms pull them off
• Berg & Gabel (2015): across provinces, work requirements did the most to cut caseloads, 1986–2009

Key Definitions

Nonexperimental sources: responses to real policy changes, compared across people, places, or time

Experimental sources: programs that randomly assign who is offered the treatment

Look back at the figure for the confounding story: unemployment spikes in the early 1990s and again in 2008, and recipiency follows. There is an extra twist in the early 1990s, when tighter unemployment insurance rules pushed more of the unemployed onto welfare. The two definitions set up the rest of the section: Quebec and the child benefit studies are nonexperimental, and the Self-Sufficiency Project is experimental.

Quebec: A Discontinuity at Age 30

• Until August 1989: Quebec paid singles and couples without children far less if they were under 30
  • Age 29: $185 a month; age 30: $507 a month
• Someone a few months short of 30 is otherwise much like someone a few months past it
  • Comparing the two is a regression discontinuity design, as in Chapter 2
• Results:
  • Fortin et al. (2004): the higher benefit led to much longer spells on assistance
  • Lemieux & Milligan (2008): employment of male high-school dropouts fell about 4 percentage points past age 30 (1986)

Key Result

Two nonexperimental studies, one conclusion: higher benefits reduce work

Some of the young claimants sued under the Charter for age discrimination, and the Supreme Court upheld the rule in a split decision (Gosselin v. Quebec, 2002). A policy that was hard on young people turned out to be very good for research. Nobody can choose their birthday to qualify, so people just either side of 30 differ in almost nothing except the benefit they could get. That is what makes the jump at 30 believable as a cause.

iClicker Question

Which comparison gives the most credible estimate of how a higher welfare benefit affects work?

A) Employment in provinces with high benefits versus provinces with low benefits
B) Employment in one province before and after it raised benefits
C) Hours worked by welfare recipients versus non-recipients
D) People just under versus just over an age cutoff where benefits jump
E) Asking recipients whether the benefit made them work less

Answer: D. Just under and just over the cutoff, people are alike except for the benefit, so the jump in outcomes at the cutoff is the benefit’s effect. That is the Quebec age-30 design. A compares provinces that differ in labour markets and much else (confounding). B credits the benefit with everything else that changed over the period, like a recession. C compares people who chose to be on assistance with people who did not (selection). E relies on self-reports. A randomized experiment would be more credible still, but it is not an option here, and the SSP, coming up, is one.

Child Benefits and Work

• National Child Benefit, 1998: some provinces effectively cut the clawback for welfare recipients
  • Work incentives improved, which could explain much of the 1997–2000 caseload fall
• UCCB, 2006: compare parents of children 5 and under to parents of older children, before and after
  • A difference-in-differences design, as in Chapter 2
  • Married mothers cut both participation and hours, as the demogrant model predicts
  • From the paper itself: participation of lower-educated married mothers fell 3.2 percentage points
  • Divorced mothers worked more
• UCCB expansion and the CCB: large cuts in poverty, no sign of lower labour supply
• Source: Milligan & Stabile (2007); Schirle (2015); Koebel & Schirle (2016); Baker et al. (2023)

Child benefits have been redesigned several times, and each redesign is a natural experiment. Schirle’s result is the demogrant diagram coming true: a pure income effect, fewer hours and less participation. Koebel and Schirle are the reminder that the simple model is not the whole story, since divorced mothers moved the other way. Baker and coauthors are the most recent word, and they find the CCB cut poverty a lot without any measurable labour supply cost. Note that the 3.2-point figure comes from Schirle’s paper; the textbook reports only the direction.

The Self-Sufficiency Project

Key Definition

Self-Sufficiency Project (SSP): a randomized Canadian experiment that paid an earnings supplement to long-term assistance recipients who left for full-time work

• Who: about 5,600 single parents, mostly mothers, on assistance for at least a year
• Where and when: New Brunswick and British Columbia, 1992–1999
• Design: random assignment to a program group or a control group
• The offer: leave assistance for a job of 30+ hours a week within a year, and collect a supplement
  • Half the gap between earnings and a benchmark of $30,000 (NB) or $37,000 (BC)
  • Paid for up to three years, as long as the work stays full time
  • Roughly doubled what assistance paid

Do not draw the budget constraint for this one. Problem 5 asks them to build exactly this program’s constraint, so describe it in words and let them do the geometry. Worth saying that this is not an NIT: there is nothing at all unless you work full time, which is a very different shape.

SSP: Results

• Naive estimate: full-time work in the program group went from 6% to 30%, so 24 points
• But the control group got to 15% with no offer at all
• Experimental estimate: 30 − 15 = 15 points
• The gap closed after the offer ended
  • 9 points at 36 months
  • Gone by 54 months; assistance receipt equal by month 69

The naive number is a before-and-after comparison, and it is wrong because single parents leave assistance over time anyway. The control group shows how many would have left without the offer, which is the whole point of an experimental source. Fifteen points is still a big effect for the group with the lowest exit rates of anyone on assistance.

SSP: Why the Effect Faded

• Card & Hyslop (2005): the time limits created two incentives
  • Establishment incentive: find full-time work within 12 months to qualify
  • Entitlement incentive: once qualified, keep working to keep collecting
  • Both end when the program ends
• Riddell & Riddell (2014): the program group went back to school less
  • Less high-school completion, college or trade school, and university
  • Work-first policies can trade long-run earning power for short-run jobs

Key Result

The SSP sped up the move from assistance to work while it ran; it did not change where people ended up

• Source: Card & Hyslop (2005); Riddell & Riddell (2014)

Riddell and Riddell’s framing is worth repeating: both groups built up their earning power during the experiment, the program group on the job and the control group in the classroom, which is part of why the gap closed. Do not let them leave thinking the SSP failed. It moved people into work faster, and the control group’s behaviour is itself a finding: a slow but large exit from assistance, much of it through finishing high school. That tells us something about how people get off assistance when nobody pays them to.

Summary

Summary

Income support programs:

• Provide crucial income support to various groups
• Can create work disincentives in some cases
• Must balance support needs against incentive effects

We examined several programs through the labour supply model:

• Three steps: constraint before, constraint after, compare choices
• Demogrants, social assistance, NIT, wage subsidies, and tax credits like the CWB

Evidence:

• Experiments and natural experiments beat simple correlations
• Both benefit generosity and labour market opportunities predict who relies on assistance

References

• Atkinson, A. B. (1998). Social exclusion, poverty and unemployment. In A. B. Atkinson & J. Hills (Eds.), Exclusion, employment and opportunity (CASEpaper 4, pp. 1–20). Centre for Analysis of Social Exclusion, London School of Economics.
• Baker, M., Messacar, D., & Stabile, M. (2023). Effects of child tax benefits on poverty and labor supply: Evidence from the Canada Child Benefit and Universal Child Care Benefit. Journal of Labor Economics, 41(4), 1129–1182.
• Benjamin, D., Foley, K., Gunderson, M., Lemieux, T., Schirle, T., & Skuterud, M. (2026). Labour Market Economics, 2026 Release. McGraw Hill. Chapter 4.
• Berg, N., & Gabel, T. (2015). Did Canadian welfare reform work? The effects of new reform strategies on social assistance participation. Canadian Journal of Economics, 48(2), 494–528.
• Blundell, R. (2006). Earned income tax credit policies: Impact and optimality. Labour Economics, 13(4), 423–443.
• Card, D., & Hyslop, D. R. (2005). Estimating the effects of a time-limited earnings subsidy for welfare-leavers. Econometrica, 73(6), 1723–1770.
• Fortin, B., Lacroix, G., & Drolet, S. (2004). Welfare benefits and the duration of welfare spells: Evidence from a natural experiment in Canada. Journal of Public Economics, 88(7–8), 1495–1520.

References (cont.)

• Hum, D., & Simpson, W. (1991). Income maintenance, work effort, and the Canadian Mincome experiment. Economic Council of Canada.
• Hum, D., & Simpson, W. (1993). Economic response to a guaranteed annual income: Experience from Canada and the United States. Journal of Labor Economics, 11(1, Part 2), S263–S296.
• Koebel, K., & Pohler, D. (2019). Expanding the Canada Workers Benefit to design a guaranteed basic income. Canadian Public Policy, 45(3), 283–309.
• Koebel, K., & Schirle, T. (2016). The differential impact of universal child benefits on the labour supply of married and single mothers. Canadian Public Policy, 42(1), 49–64.
• Laidley, J., & Tabbara, M.-D. (2024). Welfare in Canada, 2023. Maytree.
• Lemieux, T., & Milligan, K. (2008). Incentive effects of social assistance: A regression discontinuity approach. Journal of Econometrics, 142(2), 807–828.

References (cont.)

• Milligan, K., & Stabile, M. (2007). The integration of child tax credits and welfare: Evidence from the Canadian National Child Benefit program. Journal of Public Economics, 91(1–2), 305–326.
• Nichols, A., & Rothstein, J. (2015). The Earned Income Tax Credit (EITC). NBER Working Paper No. 21211.
• Riddell, C., & Riddell, W. C. (2014). The pitfalls of work requirements in welfare-to-work policies: Experimental evidence on human capital accumulation in the Self-Sufficiency Project. Journal of Public Economics, 117, 39–49.
• Royal Commission on the Economic Union and Development Prospects for Canada. (1985). Report. Minister of Supply and Services Canada.
• Schirle, T. (2015). The effect of universal child benefits on labour supply. Canadian Journal of Economics, 48(2), 437–463.
• Statistics Canada. Canadian Income Survey, 2022: Public Use Microdata File.