Unions and collective bargaining

A union is an organization of workers that bargains collectively with employers over wages, benefits and working conditions. In the monopoly union model the union sets the wage and the firm picks employment on its labour demand curve, so a higher wage costs jobs. In the efficient contract model the two sides bargain over both the wage and employment, reach the competitive level of employment and use the wage only to divide the surplus.

Unions in Canada

Two measures are used. The union coverage rate is the share of employees who are union members or are covered by a collective agreement. The unionization rate counts members only. In 2025, 30.6 percent of Canadian employees were covered and 28.5 percent were members.

Union coverage and unionization rates, Canada

Employees aged 15 and over, annual averages

• Union coverage rate
• Unionization rate

Coverage exceeds membership by about two percentage points: covered non-members. Both rates fell by two to three points between 1997 and 2025.Source: Statistics Canada, Table 14-10-0129-01, Union status by geography. Retrieved 2026-10-23.

Show the data and notes

Coverage rate: employees who are union members and/or covered by a collective agreement, as a share of all employees. Unionization rate: union members only. Labour Force Survey, both genders, 10 provinces. Statistics Canada cautions that 2006 changes to LFS processing may have shifted levels; compare across 2006 with care.

Period	Union coverage rate	Unionization rate
1997	33.7%	30.9%
1998	33.0%	30.5%
1999	32.3%	29.9%
2000	32.3%	30.1%
2001	32.3%	30.2%
2002	32.2%	30.1%
2003	32.2%	30.1%
2004	31.7%	29.8%
2005	32.0%	29.8%
2006	31.6%	29.2%
2007	31.3%	29.1%
2008	31.0%	28.8%
2009	31.5%	29.4%
2010	31.4%	29.3%
2011	30.8%	28.8%
2012	30.9%	28.9%
2013	30.9%	28.8%
2014	30.2%	28.1%
2015	30.4%	28.2%
2016	30.0%	28.0%
2017	30.0%	28.0%
2018	29.6%	27.6%
2019	29.8%	27.8%
2020	31.0%	29.1%
2021	30.7%	28.9%
2022	30.4%	28.2%
2023	30.5%	28.4%
2024	30.2%	28.3%
2025	30.6%	28.5%

Coverage is concentrated in the public sector: 77.5 percent of public-sector employees were covered in 2025, against 15.1 percent in the private sector, down from 21.3 percent in 1997. Manufacturing coverage fell from 36.4 to 21.6 percent over the same years.¹ The longer decline was among men: the share of men in unionized jobs fell from 42 percent in 1981 to 29 percent in 2012, while women’s rate stayed near 30 percent.²

Canada is far more unionized than the United States. In 2025, 11.2 percent of US wage and salary workers were represented by a union and 10.0 percent were members; the private-sector membership rate was 5.9 percent, against 13.4 percent in Canada.³¹

Why workers join

Membership is the outcome of a demand for and a supply of union jobs.

• Demand by workers. A worker wants a union job when its package (a higher wage, better benefits, job security, a grievance procedure) is worth more than the dues and the non-union alternative. Demand rises with the union wage gain, which is largest for less-skilled workers (see union wage effects).
• Supply of union jobs. Organizing is costly. It is cheaper in large workplaces and the public sector, and harder where employers resist and the law makes certification difficult.

The law matters. Under the Canada Labour Code the Canada Industrial Relations Board certifies a union once it is satisfied that a majority of the unit wants it, and must hold a vote only when 35 to 50 percent of employees are members, so a majority of signed members can win certification without a vote (card check).⁴ Alberta requires evidence of support from at least 40 percent of the unit and then a mandatory representation vote.⁵ Across nine Canadian jurisdictions, mandatory votes lowered certification success by about 9 percentage points relative to card check.⁶ When British Columbia switched from card check to votes in 1984, success rates fell by an average of 19 points and rose again when card check returned in 1993; employer opposition was about twice as effective under votes.⁷

A union that wins certification represents every worker in the unit, which invites free riding. The Rand formula, from Justice Ivan Rand’s 1946 arbitration award that settled the 1945 Ford strike in Windsor, requires every covered worker to pay dues whether or not they join.⁸

The monopoly union model

Assumptions: the firm sells its output in a competitive market, so its labour demand curve is $VMP_E$. Without a union it would pay the competitive wage $w^*$, the wage workers can get elsewhere, and hire $E^*$. The union cares about both the wage and employment, so its indifference curves in $(E, w)$ space slope down and are convex. The union sets the wage; the firm keeps the right to choose employment and always hires on its demand curve.

The union therefore chooses a point on the demand curve, as a monopolist chooses a point on its product demand curve. Its best point $M$, with wage $w_U$ and employment $E_U$, is where an indifference curve is tangent to the demand curve.

Union wage (w_U)

$26.00

Employment E_U

40

Jobs lost (80 - E_U)

40

Union rent (w_U - w*) × E_U

$160 an hour

Deadweight loss

$80 an hour

Open in Graph Lab

Figure 1. Demand $w = 30 - 0.1E$ with a competitive wage of $22. A union that maximizes total rent $(w - w^*)E$ reaches its highest indifference curve at $M$: a wage of $26 and 40 workers instead of 80. Any other wage puts the union on a lower indifference curve; the shaded triangle is the deadweight loss of the 40 jobs not filled.

A useful special case: the union maximizes its members’ rent, $(w - w^*)E$, with $w$ read off the demand curve $w(E)$. The first-order condition is

$$\underbrace{w(E) + E\,w'(E)}_{\text{marginal revenue of a job}} = w^*,$$

so the union behaves like a monopolist with marginal cost $w^*$. With linear demand $w = a - bE$ this gives $a - 2bE = w^*$, so $E_U = \frac{a - w^*}{2b} = \frac{E^*}{2}$ and $w_U = \frac{a + w^*}{2}$: employment halves and the wage is halfway between $w^*$ and the demand intercept.

The jobs between $E_U$ and $E^*$ produce output worth more than $w^*$, the value of workers’ time elsewhere. Losing them is a deadweight loss. Displaced workers move to the non-union sector and push wages down there, an effect covered in union wage effects.

Efficient contracts

The monopoly outcome leaves gains on the table. An isoprofit curve is the set of $(E, w)$ pairs that give the same profit $\pi = R(E) - wE$, where $R(E)$ is revenue and $R'(E) = VMP_E$. Holding profit constant, $R'(E)\,dE - w\,dE - E\,dw = 0$, so

$$\left.\frac{dw}{dE}\right|_{\pi \text{ constant}} = \frac{VMP_E - w}{E}.$$

Left of the demand curve $VMP_E > w$, so the isoprofit curve rises; right of it, it falls. Each isoprofit curve is hump-shaped with its peak on the demand curve, and lower curves mean higher profit.

At $M$ the firm’s isoprofit curve is flat (it peaks there) while the union’s indifference curve slopes down. The two curves cross, so a lens of contracts with more employment and a lower wage makes both parties better off.

Open in Graph Lab

Figure 2. At the monopoly-union point $M$ the union's indifference curve $U_M$ is tangent to demand while the isoprofit curve $\pi_M$ peaks. Contracts in the shaded lens beat $M$ for both sides; on the contract curve at 80 workers, any wage from $24 to $25 does.

An efficient contract is one where neither side can gain without the other losing: an isoprofit curve tangent to a union indifference curve. The set of tangencies is the contract curve. With a rent-maximizing union, $(w - w^*)E$ is constant along an indifference curve, so its slope is $-(w - w^*)/E$ and tangency requires

$$\frac{VMP_E - w}{E} = -\frac{w - w^*}{E} \quad\Longrightarrow\quad VMP_E = w^*.$$

Employment is set where the value of the marginal product equals the competitive wage, $E^*$, whatever the bargained wage. The contract curve is vertical: a strongly efficient contract. The wage only splits the surplus, and its level depends on bargaining power.

Bargained wage (w_B)

$24.50

Employment at B

80

Union rent at B (at M: $160)

$200 an hour

Profit at B (at M: $80)

$120 an hour

Joint surplus at B (at M: $240)

$320 an hour

Open in Graph Lab

Figure 3. Each efficient contract $B$ is a tangency between an isoprofit curve and a union indifference curve, and all of them lie at 80 workers. Raising the bargained wage from $22 to $26 moves the whole $320 surplus from the firm to the union without changing employment.

Key idea

In the monopoly union model the firm stays on its demand curve and jobs are lost. In an efficient contract the firm is off its demand curve, employment is at the competitive level, and the union wage is a transfer from profits.

Two implications follow. First, at $E^*$ the bargained wage exceeds $VMP_E$, so the firm would cut jobs if it could; efficient contracts need enforceable employment rules such as staffing levels or crew sizes. Second, an efficient contract creates no deadweight loss. Abowd found that unexpected increases in union labour costs reduced shareholder wealth about dollar for dollar, as strongly efficient bargains predict.⁹ The efficient-bargaining model is due to McDonald and Solow.¹⁰

Strikes

If both sides could predict a strike’s settlement, they could sign that agreement at once and split the wages and profits the strike would destroy. A strike is then a mistake both sides should avoid: the Hicks paradox.¹¹ One resolution is private information. The firm knows its profitability and the union does not; a firm’s willingness to sit through a strike signals that it cannot pay more.

The Ashenfelter-Johnson model adds politics.¹² Union leaders know what the firm can pay, but members expect more. A strike lowers members’ minimum acceptable settlement along a resistance curve, and the firm chooses the strike length that maximizes profit.

Cost of one strike week to the firm (in points of wage increase)

0.50

Strike length chosen by the firm

3.7 weeks

Settlement

5.0%

Open in Graph Lab

Figure 4. The firm picks the point $S$ on the union resistance curve that touches its lowest isoprofit line. When a strike week costs the firm the equivalent of 0.5 points of wage increase, it accepts a 3.7-week strike to settle at 5.0 percent instead of 8; when a strike week costs 1.25 points or more, it settles at once.

Card tested an asymmetric-information model on a large sample of collective agreements. Strikes were less likely when regional unemployment was high and more likely when industry selling prices were high, but negotiated wages showed no systematic relationship with strike outcomes, contrary to the model’s signalling premise.¹³

Canadian law limits when strikes can happen. In Alberta a strike is lawful only when no collective agreement is in force, after a majority strike vote and with notice.⁵ The Supreme Court of Canada held in 2015 that freedom of association in s. 2(d) of the Charter protects a right to strike.¹⁴ Since June 20, 2025, federally regulated employers may not use replacement workers to do the work of striking or locked-out employees.¹⁵

Worked example

Labour demand is $w = 30 - 0.1E$, so revenue is $R(E) = 30E - 0.05E^2$ (the area under the demand curve). Profit is revenue minus the wage bill; other costs are fixed. The competitive wage is $w^* = 22$ and the union maximizes rent $(w - 22)E$.

1. Competitive benchmark. $30 - 0.1E = 22$ gives $E^* = 80$.

2. Monopoly union. Rent is $(30 - 0.1E - 22)E = 8E - 0.1E^2$. Setting the derivative to zero, $8 - 0.2E = 0$, gives $E_U = 40$ and $w_U = 30 - 0.1(40) = 26$. The union’s rent is $(26 - 22) \times 40 = 160$ dollars an hour.

3. Profit at the monopoly point. $R(40) = 1{,}200 - 80 = 1{,}120$, minus the wage bill $26 \times 40 = 1{,}040$, leaves $\pi_M = 80$.

4. Deadweight loss. $\tfrac{1}{2}(26 - 22)(80 - 40) = 80$ dollars an hour.

5. Efficient contracts. Employment returns to 80 and joint surplus is $R(80) - 22 \times 80 = 2{,}080 - 1{,}760 = 320$, which is $160 + 80$ plus the recovered deadweight loss of 80.

6. Wages that beat $M$ for both sides. The union needs $(w - 22) \times 80 \ge 160$, so $w \ge 24$. The firm needs $2{,}080 - 80w \ge 80$, so $w \le 25$. Any wage between $24 and $25 with 80 workers makes both better off than $M$.

Exercise: demand $w = 40 - 0.2E$ and $w^* = 24$

$E^* = 80$. The rent-maximizing union picks $E_U = 40$ and $w_U = 32$, with rent $8 \times 40 = 320$. Revenue is $R(E) = 40E - 0.1E^2$, so $\pi_M = 1{,}440 - 1{,}280 = 160$ and the deadweight loss is $\tfrac{1}{2}(8)(40) = 160$. At 80 workers, $R(80) = 2{,}560$. The union needs $w \ge 24 + 320/80 = 28$; the firm needs $2{,}560 - 80w \ge 160$, so $w \le 30$.

Common mistakes

Traps on exams

• Putting the monopoly union off the demand curve. In the monopoly model the firm always picks employment on its demand curve. Only efficient contracts lie off it.
• Drawing isoprofit curves like indifference curves. They are hump-shaped, peak where they cross the demand curve, and lower curves mean more profit.
• Saying an efficient contract means the union gains nothing. The wage is above $w^*$; the union gains through the wage while employment stays at $E^*$.
• Mixing up coverage and membership. Coverage includes non-members covered by a collective agreement, so it is always at least the unionization rate.

On the exam

For efficient contracts, draw the demand curve, mark $M$ on it with a tangent union indifference curve, and draw the isoprofit curve through $M$ peaking at $M$. Shade the lens, then draw the vertical contract curve at $E^*$ through it.

Check yourself

1.

In the monopoly union model, the union sets a wage $w_U$ above the competitive wage $w^*$. Which describes the outcome?

1The firm hires on its labour demand curve, so employment falls below the competitive level2The firm keeps employment at the competitive level and pays the higher wage out of profits3The firm hires where the union wage meets the labour supply curve, so employment rises4Employment is unchanged because the union also sets the number of jobs

Check answer

2.

Where does a firm's isoprofit curve reach its peak in a graph with employment on the horizontal axis and the wage on the vertical axis?

1Where it crosses the labour demand curve, because there $VMP_E = w$2Where it crosses the competitive wage $w^*$3Where it is tangent to the union's indifference curve4It has no peak; isoprofit curves slope down everywhere

Check answer

3.

A union maximizes total rent $(w - w^*)E$ and bargains with the firm over both the wage and employment. The contract curve is vertical at the competitive employment level $E^*$. What does this imply?

1Employment equals the competitive level, and the bargained wage only divides the surplus between the union and the firm2Employment equals the monopoly-union level, and the wage equals the competitive wage3The firm pays $w^*$ and the union gains only through more jobs4The firm stays on its labour demand curve whatever the wage

Check answer

Practise more questions on this topic

Sources

1. Statistics Canada, Table 14-10-0132-01, Union status by industry, annual Labour Force Survey estimates. ↩ ↩²

2. Galarneau, D. and T. Sohn (2013). Long-term trends in unionization. Insights on Canadian Society, Statistics Canada catalogue no. 75-006-X. ↩

3. US Bureau of Labor Statistics, Union Members — 2025, news release, February 18, 2026. ↩

4. Canada Labour Code, R.S.C. 1985, c. L-2, sections 28 and 29. ↩

5. Alberta Labour Relations Code, R.S.A. 2000, c. L-1, sections 33, 34 and 73. ↩ ↩²

6. Johnson, S. (2002). Card check or mandatory representation vote? How the type of union recognition procedure affects union certification success. Economic Journal, 112(479), 344 to 361. IDEAS. ↩

7. Riddell, C. (2004). Union certification success under voting versus card-check procedures: Evidence from British Columbia, 1978 to 1998. ILR Review, 57(4), 493 to 517. doi:10.1177/001979390405700402. ↩

8. Dion, G. and D. Panneton, Rand Formula, The Canadian Encyclopedia. ↩

9. Abowd, J. M. (1989). The effect of wage bargains on the stock market value of the firm. American Economic Review, 79(4), 774 to 800. IDEAS. ↩

10. McDonald, I. M. and R. M. Solow (1981). Wage bargaining and employment. American Economic Review, 71(5), 896 to 908. IDEAS. ↩

11. Hicks, J. R. (1932). The Theory of Wages. London: Macmillan. ↩

12. Ashenfelter, O. and G. E. Johnson (1969). Bargaining theory, trade unions, and industrial strike activity. American Economic Review, 59(1), 35 to 49. IDEAS. ↩

13. Card, D. (1990). Strikes and wages: A test of an asymmetric information model. Quarterly Journal of Economics, 105(3), 625 to 659. NBER working paper 2550. ↩

14. Saskatchewan Federation of Labour v. Saskatchewan, 2015 SCC 4. ↩

15. Employment and Social Development Canada, Protecting Canadian labour: Replacement workers legislation now in force, June 20, 2025. ↩