Migration and job turnover

Migration is a human capital investment: a worker moves to another region, or to another employer, when the present value of the expected gain in earnings exceeds the full cost of moving. The rule predicts that the young, the educated and the unemployed move most, that couples move less often than single people, and that some moves are reversed.

Costs now, gains later

Moving costs are paid up front: direct costs (transport, selling one home and finding another, re-licensing), forgone earnings while searching and settling in, and psychic costs of leaving family, friends and a familiar place. The gains arrive a little at a time, as a higher wage every year for the rest of a career. An electrician in Saint John who could earn $6,000 a year more in Edmonton moves only if that stream, valued today, is worth more than the lump-sum cost.

The comparison is the one used for schooling as an investment. The approach goes back to Larry Sjaastad.¹

The model

A worker earns $w_{0t}$ at home in year $t$ and could earn $w_{1t}$ at the destination. $M$ is the full cost of moving in dollars, psychic costs included, $r$ is the discount rate and $T$ the number of working years left. The net present value of migrating is

$$PV = \sum_{t=1}^{T} \frac{w_{1t} - w_{0t}}{(1+r)^t} - M .$$

The worker moves if $PV > 0$; with several destinations, to the one with the largest $PV$. Each term predicts something:

1. The wage gap, $w_{1t} - w_{0t}$. A boom at the destination pulls workers in; job losses at home push them out.
2. The moving cost, $M$. Distance, housing costs at the destination, family already there and information about jobs all work through $M$.
3. The payoff period, $T$. The same yearly gain is worth more with more years to collect it.
4. The discount rate, $r$. A lower $r$ puts more weight on distant gains.

Yearly wage gain, w₁ − w₀

$5.0k

Cost of moving, M

$50k

PV of gains, moving at 25

$85.8k

Net gain, moving at 25

$35.8k

Net gain, moving at 50

$1.9k

Oldest age at which moving pays

50.8

Open in Graph Lab

Figure 1. Present value of a $5,000 yearly wage gain collected until age 65 at a 5 percent discount rate, against a $50,000 moving cost $M$. Moving pays only below the cut-off age of about 51; a higher cost lowers the cut-off and a larger wage gain raises it.

Who moves

• The young. The present value of a given yearly gain falls with age because fewer years remain to collect it, while the cost does not fall (Figure 1). Young workers also tend to have lower costs: no house to sell, no children in school.
• The more educated. Migration rates rise with education. Educated workers search over wider areas and face national markets for their skills, which lowers the cost of finding a distant job.
• The unemployed. A worker without a job has a low $w_{0t}$, so the gain from moving is large.

Migration also raises total output. Workers leave regions where the value of their marginal product is low for regions where it is high, so labour supply falls in low-wage regions, rises in high-wage regions, and the wage gap narrows. It need not close: a gap whose present value is below the cost of moving persists.

Key idea

A worker moves when the present value of the earnings gain exceeds the full cost of moving. A larger wage gap, a longer payoff period (youth) or a lower cost (shorter distance, better information) makes moving more likely.

Family migration: tied movers and tied stayers

A couple maximizes its combined gain. Let $\Delta PV_H$ and $\Delta PV_W$ be the husband’s and the wife’s private gains, what each would gain by moving if single. The family moves when

$$\Delta PV_H + \Delta PV_W > 0 .$$

This analysis of family migration is due to Jacob Mincer.²

Husband's gain, ΔPV_H

$35k

Wife's gain, ΔPV_W

−$25k

Husband's gain

$35k

Wife's gain

−$25k

Family's gain

$10k

Open in Graph Lab

Figure 2. Couples plotted by each partner's private gain from moving. The family moves above the line where the two gains sum to zero. In the purple triangles one partner moves at a private loss (a tied mover); in the red triangles one partner gives up a private gain by staying (a tied stayer).

• A tied mover moves although their own gain is negative, because the partner’s gain is larger.
• A tied stayer would gain by moving alone but stays, because the partner would lose more.

Three predictions follow. Families move less often than single people, since one partner’s loss can cancel the other’s gain. After a family move, the tied mover’s employment and earnings often fall, at least for a while, even though family income rises. And as more couples include two partners with careers, more people become tied stayers and migration falls.

Return and repeat migration

The present-value rule predicts two kinds of return migration. Some moves were mistakes: they rested on expected wages that did not materialize, and with better information staying no longer pays. Others were planned: a few years of high pay elsewhere, then home. Studying immigrants who leave the United States, George Borjas and Bernt Bratsberg argue that both occur.³

Moving costs differ across people, so those who have moved once are disproportionately people with low costs, and the model predicts they move again. Moving more than once is repeat migration.

Job turnover and job matching

Changing employers is migration between firms. A worker quits when the present value of the better job exceeds the cost of switching: search time, lost seniority and any training specific to the old firm.

The quality of a job match, how productive a given worker is at a given firm, cannot be observed in advance; worker and firm learn it on the job.⁴ Bad matches are discovered and end early, and good matches survive. So the probability of separating falls with tenure, and wages rise with tenure partly because the matches that last are the good ones.

Young workers search through matches quickly. Robert Topel and Michael Ward found that in his first ten years in the labour market a typical young man holds seven jobs, about two-thirds of his career total, and that wage gains at job changes account for at least a third of early-career wage growth.⁵

Worked example

A worker can earn $6,000 a year more in Calgary than in Halifax. Moving costs $15,000, paid now. Gains arrive at the end of each year and the discount rate is 5 percent.

1. Two years left. $PV_{\text{gains}} = \dfrac{6000}{1.05} + \dfrac{6000}{1.05^2} = 5714.29 + 5442.18 = 11156.46$. Since $11156.46 - 15000 = -3843.54$, this worker stays, although the undiscounted gain of $12,000 is close to the cost.

2. Three years left. Adding $\dfrac{6000}{1.05^3} = 5183.03$ makes the gains worth about $16,340, more than $15,000. This worker moves.

3. Thirty years left. A constant gain $g$ for $T$ years is worth $g \times \dfrac{1 - (1+r)^{-T}}{r}$. Here $1.05^{-30} = 0.2314$, the factor is $(1 - 0.2314)/0.05 = 15.3725$, and the gains are worth $6000 \times 15.3725 \approx 92235$: a net value of about $77,235.

4. The cut-off. Moving pays when $6000 \times \dfrac{1 - 1.05^{-T}}{0.05} \geq 15000$, that is $1.05^{-T} \leq 0.875$, or $T \geq \dfrac{\ln(1/0.875)}{\ln 1.05} = 2.74$. Anyone with at least 3 working years left moves.

Second example: the same move also carries $20,000 of psychic costs, so $M = 35000$. Who moves now?

Show the solution

The condition becomes $1.05^{-T} \leq 1 - \dfrac{0.05 \times 35000}{6000} = 0.7083$, so $T \geq \dfrac{\ln(1/0.7083)}{\ln 1.05} = 7.07$. With 7 years left the gains are worth $34,718, just short of the cost; with 8 years they are worth $38,779. Only workers with at least 8 years left move: a higher cost screens out older workers first.

Evidence

Canadians move between provinces far less than they used to. The interprovincial out-migration rate fell from 18.0 per 1,000 people in 1971/72 to 7.0 per 1,000 in 2025/26 (a preliminary estimate), with a short-lived rebound in 2021/22.

Interprovincial out-migration rate, Canada

Interprovincial migrants per 1,000 population, July-to-June years

People who moved to another province or territory during each July-to-June year, per 1,000 population. The rate has fallen by more than half since the early 1970s.Source: Statistics Canada, Table 17-10-0015-01, Estimates of the components of interprovincial migration, by age and gender, annual; Table 17-10-0005-01, Population estimates on July 1. Retrieved 2026-10-23.

Show the data and notes

Each year x covers July 1 of x to June 30 of x+1 (e.g. 2024 = 2024/2025). Rate = people who moved to another province or territory during the period ÷ Canada's population on July 1 at its start (Table 17-10-0005-01, https://doi.org/10.25318/1710000501-eng) × 1,000. For Canada as a whole, out-migrants equal in-migrants. The latest period is preliminary.

Period	Out-migrants per 1,000
1971	18.0
1972	17.8
1973	19.5
1974	18.1
1975	16.2
1976	15.2
1977	15.4
1978	15.0
1979	15.3
1980	15.6
1981	14.4
1982	12.2
1983	11.0
1984	10.6
1985	11.1
1986	11.6
1987	12.2
1988	12.2
1989	13.1
1990	11.4
1991	11.3
1992	10.7
1993	10.1
1994	9.8
1995	10.0
1996	9.9
1997	10.3
1998	9.2
1999	9.4
2000	8.8
2001	9.4
2002	8.8
2003	8.3
2004	8.9
2005	8.9
2006	9.4
2007	9.2
2008	8.4
2009	7.7
2010	7.6
2011	8.2
2012	7.5
2013	7.8
2014	8.0
2015	7.8
2016	7.2
2017	7.1
2018	6.9
2019	7.6
2020	6.6
2021	9.1
2022	8.8
2023	8.3
2024	7.3
2025	7.0

Population aging explains part of the decline, since fewer people are at the peak moving ages. But Statistics Canada found that between the late 1970s and 2011 migration rates fell at virtually every age, most of all among young adults, so aging is not the whole story.⁶ The model points to smaller gains (narrower gaps between provinces), higher costs (housing in high-wage cities) and more two-career couples (tied stayers). Internal migration in the United States has also fallen since the 1980s, across demographic groups and distances, without a convincing single explanation.⁷

The pattern of who moves fits the model. In Canadian tax records for 1982 to 1995, Ross Finnie found that age, marriage and children all lowered the probability of moving between provinces, for men and women alike. Residents of smaller provinces were more likely to leave, and moving was more likely when the provincial unemployment rate was high and among people receiving unemployment insurance or social assistance, whose $w_{0t}$ was low.⁸

Common mistakes

Traps on exams

• Comparing undiscounted totals. Three years of a $6,000 gain are worth about $16,340 today, not $18,000. Discount each year’s gain before comparing it with the cost.
• Explaining age with tastes. In the model, older workers move less because the payoff period is shorter, not because they prefer to stay.
• Requiring both partners to gain. The family moves when the sum of the private gains is positive; the partner who loses is a tied mover.
• Confusing net and gross migration. A province can have zero net migration while thousands move in and out. The rate in the chart counts gross moves.

On the exam

For a present-value question, write the discounted sum term by term, state the rule ($PV > 0$, move) and name the term the scenario changes. For family migration, put the two private gains on the axes, draw the line $\Delta PV_H + \Delta PV_W = 0$, and label the tied-mover and tied-stayer regions.

Check yourself

1.

Two workers in Halifax would each earn $5,000 a year more in Calgary and face the same moving cost. One is 25 and the other is 55. According to the human capital model, why is the 25-year-old more likely to move?

1She has more years over which to collect the wage gain, so its present value is larger2Younger workers have higher discount rates, which make future gains more valuable3The yearly wage gain is larger for younger workers4Moving costs do not matter to younger workers

Check answer

2.

A husband would gain $40,000 in present value by moving to Calgary on his own; his wife would lose $15,000 by moving on her own. The family maximizes its combined present value. What happens?

Husband's gain, ΔPV_H

$35k

Wife's gain, ΔPV_W

−$25k

Husband's gain

$35k

Wife's gain

−$25k

Family's gain

$10k

1The family moves, and the wife is a tied mover2The family stays, and the husband is a tied stayer3The family moves, and the husband is a tied mover4The family stays because the move does not make both partners better off

Check answer

3.

Moving would raise a worker's earnings by $4,000 a year for 3 years, with the gains arriving at the end of each year. Moving costs $9,000 today and the worker's discount rate is 10 percent. What is the net present value of moving, to the nearest dollar?

Your answer (dollars)

Check answer

Practise more questions on this topic

Sources

1. Sjaastad, L. A. (1962). The costs and returns of human migration. Journal of Political Economy, 70(5, Part 2), 80 to 93. ↩

2. Mincer, J. (1978). Family migration decisions. Journal of Political Economy, 86(5), 749 to 773. NBER working paper 199. ↩

3. Borjas, G. J. and Bratsberg, B. (1996). Who leaves? The outmigration of the foreign-born. Review of Economics and Statistics, 78(1), 165 to 176. NBER working paper 4913. ↩

4. Jovanovic, B. (1979). Job matching and the theory of turnover. Journal of Political Economy, 87(5), 972 to 990. ↩

5. Topel, R. H. and Ward, M. P. (1992). Job mobility and the careers of young men. Quarterly Journal of Economics, 107(2), 439 to 479. NBER working paper 2649. ↩

6. Bohnert, N. (2013). Migration: Interprovincial, 2009/2010 and 2010/2011. Report on the Demographic Situation in Canada, Statistics Canada catalogue no. 91-209-X. Statistics Canada. ↩

7. Molloy, R., Smith, C. L. and Wozniak, A. (2011). Internal migration in the United States. Journal of Economic Perspectives, 25(3), 173 to 196. NBER working paper 17307. ↩

8. Finnie, R. (2000). Who moves? A panel logit model analysis of inter-provincial migration in Canada. Statistics Canada, Analytical Studies Branch Research Paper Series, no. 142. Statistics Canada. ↩