Question 1 of 8

Can Latvia stop its workforce from shrinking?

About 16 min read

The short answer

No, but it can slow the decline. The cheapest ways to do that are immigration and a return programme. If nothing changes, about 1.63 million people live in Latvia in 2040, 12 % fewer than today. A balanced policy, all four instruments together from 2028, gives 3.5 % more people in 2040. It gives a little less in extra workers, because newcomers find jobs over several years.

The newcomers cost before they pay. With the balanced policy, government debt in 2040 is 2 points of GDP higher. A larger population needs schools and doctors before all of it is in work. By the mid-2040s the extra workers' taxes have paid that back. Per extra worker, immigration and return cost least. Youth retention costs about three times as much. Family benefits add children, not workers, before 2040.

Why it matters

Between 2005 and 2025 Latvia's population fell from 2.25 million to 1.86 million, a drop of 17 %. The number of people in work fell from 1,040 thousand to 865 thousand. Two causes: emigration, above all after the 2008 crisis, and births below deaths since the early nineties.

The next fifteen years are written into today's age structure. Few children were born in the nineties. So there are now few people of child-bearing age, and births stay low even if every family chooses as it does today. Meanwhile the large post-war generations retire. The base path, that is, what happens if none of the instruments on this page is used, leads to 1.63 million people in 2040 and about 756 thousand in work, 12 % fewer than today.

The uncertainty band on the chart is narrow, from 1.62 to 1.64 million in 2040. It shows how far the path moves when the two most important assumptions are varied across their plausible ranges. Here they are two labour-market assumptions, about how fast firms fill the posts that retirements leave. They move the number of people very little. They move unemployment a great deal, and the same band is drawn on the unemployment chart further down.

Why this matters beyond the number itself. Fewer workers mean a smaller tax base to pay for pensions, doctors and schools, while the number of pensioners falls much more slowly. It is the common background of this site's questions on pensions, public finances and the regions. In almost every one of them the shrinking workforce is a condition the other instruments cannot change.

Resident population

People living in Latvia at the start of the year.

Recorded data
Base path
Uncertainty band

The grey band shows how far the base path moves when the two most important assumptions are varied across their plausible ranges.

The band is calculated to 2041; after that the line runs on without it.

Show the dataDownload CSV

Year	Recorded data	Base path	Uncertainty band – low	Uncertainty band – high
2005	2.25	2.25
2006	2.23	2.23
2007	2.21	2.21
2008	2.19	2.20
2009	2.16	2.18
2010	2.12	2.14
2011	2.07	2.09
2012	2.04	2.06
2013	2.02	2.03
2014	2.00	2.01
2015	1.99	1.99
2016	1.97	1.97
2017	1.95	1.95
2018	1.93	1.94
2019	1.92	1.92
2020	1.91	1.91
2021	1.89	1.89
2022	1.88	1.87
2023	1.90	1.88
2024	1.88	1.87
2025	1.86	1.85	1.85	1.85
2026		1.83	1.83	1.83
2027		1.81	1.81	1.81
2028		1.80	1.80	1.80
2029		1.78	1.78	1.78
2030		1.76	1.76	1.76
2031		1.75	1.74	1.75
2032		1.73	1.73	1.74
2033		1.72	1.71	1.72
2034		1.71	1.70	1.71
2035		1.69	1.68	1.70
2036		1.68	1.67	1.68
2037		1.67	1.66	1.67
2038		1.65	1.64	1.66
2039		1.64	1.63	1.65
2040		1.63	1.62	1.64
2041		1.62	1.61	1.63
2042		1.61	1.61	1.61
2043		1.60	1.60	1.60
2044		1.59	1.59	1.59
2045		1.58	1.58	1.58

On the chart the dots are the recorded data and the line is the model's base path. Up to 2025 they agree to within a fraction of a percent: the model gives 1.85 million for the start of 2025, the record was 1.86 million. That is the test. If the model can repeat the past, there is a reason to listen to what it says about the future.

What drives it

The population has inertia. It changes slowly, and three flows change it: births, deaths and migration. Each is held by a loop of its own.

The wage loop, which strengthens itself. Fewer people of working age mean slower growth and fewer opportunities. The wider the gap between wages in Latvia and in Western Europe, the more people leave, and the fewer remain. The loop also runs the other way: as wages close in on the European average, fewer people leave and more come. That is what happens on the base path. Net migration, that is, arrivals minus departures, turns from a loss of about 5,800 people a year in 2025 to a small gain, about 900 a year, by 2040.

The wage loop. A plus means "more of one, more of the other", a minus means the opposite. Two minuses in a row cancel out, so the loop as a whole strengthens itself: decline breeds decline, but growth also breeds growth.

Who arrives, and how. The model follows three streams of arrivals. Latvians who return. Citizens of other European Union countries. People from outside the Union, who come in two ways: to settle, or because an employer invites them to fill a post it cannot fill at home. That last stream, employer-sponsored immigration, answers the vacancies the shrinking workforce leaves. It is the quiet reason net migration turns positive on the base path. New arrivals earn less in their first years; the model carries that integration gap and lets it close over time. The Ukrainian refugees of 2022 are a separate group who settle after 2027.

The generation loop, which also strengthens itself, but slowly. Few births today mean few parents thirty years on and few births again. This loop is why births on the base path keep falling, from 12,700 in 2035 to 12,400 in 2040, even though every family chooses as it does now.

The labour-market loop, which corrects itself. When there are suddenly more people wanting work, firms do not take them on at once. Wage growth slows, unemployment rises for a while, and the new jobs appear over the years. So every instrument that adds people wanting work raises unemployment a little in the first years, and part of the gain arrives late. On the base path unemployment is 5.9 % in 2040, below today's 6.9 %, because retirements open more posts than the shrinking workforce can fill.

Over the next fifteen years the generation loop dominates: it is written into the number of people already born, and no instrument changes it. The wage loop is the one the instruments can push, and the labour-market loop sets how fast that works.

What can be done

Four instruments the government and parliament can actually use. For each: what it delivers in 2040 against the base path, what it costs the public budget, and how fast it acts. The cost to the budget is the instrument's own effect on the budget balance, added up over 2028–2045 in percentage points of gross domestic product (GDP, that is, everything the country produces in a year), before the EU fiscal rule responds. The difference in government debt in 2040 is shown beside it. Under the EU fiscal rules a package that spends is mostly paid for by cuts the rule forces elsewhere, so the debt figure is smaller than the cost.

Opening immigration

What it is
the conditions for work and study immigration are opened, so that arriving is 2.5 times easier than today. Most of the extra arrivals come to posts an employer has already offered.

What it delivers
3.1 % more people and 2.1 % more in work in 2040, about 16 thousand workers. The economy is 1.0 % larger. The pensioner poverty rate is 1.5 percentage points lower. With the return programme, the cheapest of the four per extra worker.

What it costs
about 4.1 points of GDP over 2028–2045 before the rule responds. Government debt in 2040 is 1.4 points higher: the arrivals' services come first, their taxes later.

How fast it acts
arrivals rise at once. The difference shows in the population from 2032 and in employment from 2034.
A return programme

What it is
support for people who want to come back, one and a half times today's programme: relocation grants, fast recognition of qualifications, a school place for the children.

What it delivers
1.7 % more people and 1.3 % more in work in 2040. The economy is 0.7 % larger. The pensioner poverty rate is 0.8 points lower. About as cheap per worker as immigration.

What it costs
about 2.5 points of GDP over 2028–2045 before the rule responds. Debt in 2040 is 1.0 points higher.

How fast it acts
visible in the population from 2033. Most people return in the programme's first years.
Youth retention

What it is
a set of measures that cuts the chance of a young person leaving by about a quarter: housing, support for the first job, places to study in the regions.

What it delivers
1.4 % more people and 1.3 % more in work in 2040. The economy is 0.7 % larger. The pensioner poverty rate is 0.8 points lower.

What it costs
about 5.2 points of GDP over 2028–2045 before the rule responds. Debt in 2040 is 1.2 points higher. Dearer than immigration and return per worker, because every young person kept is supported for years.

How fast it acts
visible in the population from 2033 and in employment from 2035.
Family benefits

What it is
family benefits are raised by 40 %.

What it delivers
7 % more births, about 810 children a year. The child poverty rate is 1.7 points lower. The population is 0.5 % larger in 2040. No change in employment before 2040.

What it costs
about 9.8 points of GDP over 2028–2045 before the rule responds, the dearest of the four. Debt in 2040 is 1.1 points higher.

How fast it acts
visible in births from 2030. Not visible in the labour market before 2040, because those children are still in school.

What the packages do

Five packages, all introduced from 2028 and phased in over three years. Four are one instrument each at full strength. The fifth is a balanced policy: all four at about half strength.

Immigration. work and study immigration 2.5 times easier than today.
Return. a return programme one and a half times today's.
Retention. the chance of a young person leaving cut by about a quarter.
Family benefits. benefits 40 % higher.
Balanced policy. all four instruments, each at about half strength.

People. Against the overall decline the differences are small, but clearly ordered. A balanced policy gives 1.69 million people in 2040 against 1.63 million on the base path, 3.5 % or about 56 thousand more. Immigration alone gives 3.1 % more, return 1.7 %, retention 1.4 %, family benefits 0.5 %. Family benefits are the one instrument with a visible effect on births: 7 % more in 2040 than on the base path.

Resident population

People living in Latvia at the start of the year.

Recorded data
Base path
Immigration
Return
Retention
Family benefits
Balanced policy

Show the dataDownload CSV

Year	Recorded data	Base path	Immigration	Return	Retention	Family benefits	Balanced policy
2015	1.99	1.99
2016	1.97	1.97
2017	1.95	1.95
2018	1.93	1.94
2019	1.92	1.92
2020	1.91	1.91
2021	1.89	1.89
2022	1.88	1.87
2023	1.90	1.88
2024	1.88	1.87
2025	1.86	1.85	1.85	1.85	1.85	1.85	1.85
2026		1.83	1.83	1.83	1.83	1.83	1.83
2027		1.81	1.81	1.81	1.81	1.81	1.81
2028		1.80	1.80	1.80	1.80	1.80	1.80
2029		1.78	1.78	1.78	1.78	1.78	1.78
2030		1.76	1.77	1.76	1.76	1.76	1.77
2031		1.75	1.75	1.75	1.75	1.75	1.76
2032		1.73	1.75	1.74	1.74	1.73	1.75
2033		1.72	1.74	1.73	1.73	1.72	1.74
2034		1.71	1.73	1.72	1.72	1.71	1.73
2035		1.69	1.72	1.71	1.71	1.70	1.72
2036		1.68	1.71	1.70	1.70	1.68	1.72
2037		1.67	1.70	1.69	1.69	1.67	1.71
2038		1.65	1.70	1.68	1.68	1.66	1.70
2039		1.64	1.69	1.67	1.67	1.65	1.70
2040		1.63	1.68	1.66	1.66	1.64	1.69
2041		1.62	1.68	1.65	1.65	1.63	1.68
2042		1.61	1.67	1.64	1.64	1.62	1.68
2043		1.60	1.67	1.64	1.63	1.61	1.67
2044		1.59	1.66	1.63	1.62	1.60	1.67
2045		1.58	1.66	1.62	1.62	1.59	1.66

In employment the picture is the same, but delayed: arrivals and returners do not all find work at once. A balanced policy gives 2.4 % more people in work in 2040, about 18 thousand; immigration alone 2.1 %, about 16 thousand. Family benefits leave employment in 2040 unchanged, because children born after 2028 are still in school then.

Employment

People aged 15 to 74 in work.

Recorded data
Base path
Immigration
Return
Retention
Family benefits
Balanced policy

Show the dataDownload CSV

Year	Recorded data	Base path	Immigration	Return	Retention	Family benefits	Balanced policy
2015	969	967
2016	962	959
2017	947	946
2018	942	939
2019	933	931
2020	897	900
2021	878	876
2022	872	868
2023	878	872
2024	872	869
2025	865	862	862	862	862	862	862
2026		858	858	858	858	858	858
2027		851	851	851	851	851	851
2028		840	840	840	840	840	840
2029		829	829	829	829	829	829
2030		822	822	822	822	822	822
2031		814	814	814	814	814	814
2032		805	807	806	807	806	807
2033		798	800	799	800	798	801
2034		791	796	794	794	791	797
2035		785	792	789	789	785	793
2036		779	787	784	784	779	789
2037		772	783	779	779	772	785
2038		767	779	775	774	767	781
2039		761	776	770	770	761	778
2040		756	772	766	766	756	774
2041		752	769	763	762	751	771
2042		747	766	758	758	747	768
2043		742	763	754	754	742	765
2044		737	760	751	750	737	762
2045		733	757	747	746	733	760

The gain. More workers are not the goal in themselves. The goal is what they produce and what their contributions sustain. The economy, that is, everything the country produces in a year, is measured in today's money (2025 prices). With a balanced policy it is 0.8 % larger in 2040; with immigration alone 1.0 %. That is less than the rise in employment. Income per head does not rise at all: in 2040 it is 1.8 % (immigration) to 2.3 % (balanced policy) below the base path. Newcomers produce less in their first years. A relatively larger workforce therefore slows the growth of the average wage. The gain is not a richer average resident. It is a larger economy and more contributors to the system that pays pensions.

Size of the economy (GDP)

The value of everything produced in Latvia in a year, in constant prices.

Recorded data
Base path
Immigration
Return
Retention
Family benefits
Balanced policy

Show the dataDownload CSV

Year	Recorded data	Base path	Immigration	Return	Retention	Family benefits	Balanced policy
2015	36.3	33.9
2016	37.2	35.2
2017	38.5	36.5
2018	40.2	37.9
2019	40.4	39.5
2020	39.0	40.2
2021	41.7	40.5
2022	42.5	41.7
2023	42.1	42.5
2024	42.1	43.5
2025	43.0	44.2	44.2	44.2	44.2	44.2	44.2
2026		44.3	44.3	44.3	44.3	44.3	44.3
2027		44.4	44.4	44.4	44.4	44.4	44.4
2028		45.0	45.0	45.0	45.0	45.0	45.0
2029		45.8	45.8	45.8	45.8	45.8	45.8
2030		46.5	46.5	46.5	46.5	46.5	46.5
2031		47.3	47.3	47.3	47.3	47.3	47.3
2032		47.9	47.9	47.9	47.9	48.0	48.0
2033		48.5	48.5	48.5	48.6	48.6	48.6
2034		49.0	48.9	49.0	49.1	49.2	49.0
2035		49.6	49.5	49.5	49.6	49.6	49.5
2036		50.1	50.1	50.1	50.1	50.1	50.1
2037		51.1	51.2	51.1	51.1	50.9	51.0
2038		52.1	52.3	52.2	52.2	51.8	52.2
2039		53.3	53.6	53.5	53.4	53.0	53.4
2040		54.3	54.8	54.6	54.6	54.1	54.7

In the pension system that shows at once. The pensioner poverty rate is the share of people aged 65 and over whose income is below the poverty threshold. On the base path it rises from 42 % in 2025 to 53 % in 2040; why is explained on the old-age income page. More workers mean more contributions and higher pensions. A balanced policy cuts that share by 2.0 percentage points in 2040, immigration alone by 1.5. It is a small step against the rise the base path shows, but a step in the right direction, and it comes without changing the pension system.

Pensioners at risk of poverty

Share of people aged 65 and over whose income is below 60 % of the national median.

Recorded data
Base path
Immigration
Return
Retention
Family benefits
Balanced policy

Show the dataDownload CSV

Year	Recorded data	Base path	Immigration	Return	Retention	Family benefits	Balanced policy
2015	34.6	46.7
2016	38.1	43.1
2017	39.9	46.3
2018	45.7	46.2
2019	47.9	44.3
2020	40.9	43.6
2021	44.6	44.4
2022	40.5	45.1
2023	40.1	46.1
2024	41.4	43.6
2025	41.7	42.2	42.2	42.2	42.2	42.2	42.2
2026		42.9	42.9	42.9	42.9	42.9	42.9
2027		43.4	43.4	43.4	43.4	43.4	43.4
2028		44.1	44.1	44.1	44.1	44.1	44.1
2029		45.7	45.7	45.7	45.7	45.7	45.7
2030		46.9	46.9	46.9	46.9	46.9	46.9
2031		47.6	47.6	47.6	47.6	47.6	47.6
2032		51.3	51.1	51.2	51.2	51.3	51.1
2033		50.0	49.3	49.7	49.6	50.1	49.3
2034		49.1	47.8	48.3	48.5	49.3	47.8
2035		48.3	46.6	47.3	47.7	48.6	46.7
2036		48.5	47.1	47.6	47.9	48.6	47.0
2037		50.2	48.6	49.1	49.5	50.0	48.4
2038		50.9	49.2	49.8	50.1	50.7	48.9
2039		51.5	50.0	50.6	50.7	51.0	49.5
2040		53.0	51.5	52.1	52.2	52.4	50.9
2041		54.3	52.7	53.4	53.6	53.7	52.3
2042		55.2	53.5	54.3	54.6	54.7	53.2
2043		57.1	54.3	55.8	56.1	56.3	54.0
2044		57.5	55.4	56.3	56.8	57.3	55.2
2045		57.9	55.9	56.9	57.4	57.9	56.0

The side effect. The unemployment rate is the share of people out of work among all those working or looking for work. It is where the labour-market loop shows. Instruments that add people wanting work raise it slightly: immigration by 0.4 percentage points in 2040, a balanced policy by 0.5. The grey band on this chart is wide, from 4.8 to 7.9 % around the base path's 5.9. It is the same two labour-market assumptions as on the population chart, and here is where they bite. Every package's effect sits inside that band.

Unemployment rate

Share of people who want to work and are looking for a job.

Recorded data
Base path
Immigration
Return
Retention
Family benefits
Balanced policy
Uncertainty band

The grey band shows how far the base path moves when the two most important assumptions are varied across their plausible ranges.

Show the dataDownload CSV

Year	Recorded data	Base path	Uncertainty band – low	Uncertainty band – high	Immigration	Return	Retention	Family benefits	Balanced policy
2015	9.9	9.9
2016	9.7	9.7
2017	8.7	8.7
2018	7.4	7.4
2019	6.3	6.3
2020	8.1	8.1
2021	7.6	7.6
2022	6.9	6.9
2023	6.5	6.5
2024	6.9	6.9
2025	6.9	6.9	6.9	6.9	6.9	6.9	6.9	6.9	6.9
2026		6.4	6.1	6.7	6.4	6.4	6.4	6.4	6.4
2027		5.7	5.3	6.3	5.7	5.7	5.7	5.7	5.7
2028		5.5	4.9	6.5	5.5	5.5	5.5	5.5	5.5
2029		5.6	4.8	7.2	5.6	5.6	5.6	5.6	5.6
2030		5.3	4.7	7.2	5.4	5.4	5.4	5.3	5.4
2031		5.5	4.8	7.2	5.6	5.5	5.5	5.4	5.6
2032		5.6	5.0	7.3	5.8	5.8	5.7	5.6	5.9
2033		5.8	5.1	7.4	6.2	6.1	6.0	5.8	6.2
2034		6.0	5.0	7.6	6.3	6.2	6.1	5.9	6.4
2035		5.9	4.9	7.7	6.3	6.2	6.1	5.9	6.4
2036		6.0	4.9	7.9	6.3	6.2	6.2	6.0	6.4
2037		6.0	4.9	8.0	6.4	6.3	6.2	6.0	6.5
2038		6.0	4.8	8.1	6.3	6.2	6.2	6.0	6.4
2039		5.9	4.8	8.0	6.3	6.2	6.1	6.0	6.4
2040		5.9	4.8	7.9	6.3	6.2	6.1	6.0	6.4

The price. On the base path government debt rises from 47 % of GDP in 2025 to 61 % in 2040; why it rises is explained on the public finances page. Every instrument costs the budget before the rule responds: immigration about 4.1 points of GDP over 2028–2045, return 2.5, retention 5.2, family benefits 9.8, a balanced policy 11.0. Two things cost money here: the programmes themselves, and the fact that a larger population needs schools, doctors and benefits before all of it is in work. In 2040 debt is higher under every package: by 1.4 points with immigration, 1.0 with return, 1.2 with retention, 1.1 with family benefits and 2.0 with a balanced policy. The rule forces cuts elsewhere to pay for part of it. The rest the extra workers pay back as their taxes grow, and by the mid-2040s the debt difference has turned negative for every instrument except family benefits.

Government debt

What the state owes, as a share of what the country produces in a year.

Recorded data
Base path
Immigration
Return
Retention
Family benefits
Balanced policy

Show the dataDownload CSV

Year	Recorded data	Base path	Immigration	Return	Retention	Family benefits	Balanced policy
2015	38.3	40.9
2016	41.7	39.9
2017	40.3	40.0
2018	38.3	38.4
2019	37.9	36.8
2020	44.0	39.2
2021	45.9	41.9
2022	44.4	43.5
2023	44.4	42.3
2024	46.2	43.5
2025	46.9	44.5	44.5	44.5	44.5	44.5	44.5
2026		45.4	45.4	45.4	45.4	45.4	45.4
2027		47.7	47.7	47.7	47.7	47.7	47.7
2028		49.8	49.8	49.8	49.8	49.8	49.8
2029		50.9	50.9	50.9	50.9	50.9	50.9
2030		52.0	52.0	52.0	52.1	52.2	52.2
2031		53.4	53.4	53.4	53.7	54.0	53.9
2032		54.1	54.2	54.2	54.6	55.3	55.1
2033		54.9	55.2	55.1	55.6	56.5	56.4
2034		55.8	56.5	56.3	56.7	57.6	57.8
2035		57.2	58.6	58.2	58.4	58.9	59.8
2036		58.7	60.5	60.0	60.0	60.1	61.5
2037		60.3	62.4	61.8	61.7	61.5	63.3
2038		61.1	63.2	62.7	62.6	62.2	64.0
2039		61.7	63.5	63.0	63.0	62.8	64.2
2040		61.2	62.6	62.2	62.4	62.3	63.2

Assessment

Package	Population 2040	Employment 2040	Economy 2040	Pensioner poverty 2040, percentage points	Budget cost 2028–2045, points of GDP	Government debt 2040, points	Who gains, who pays	Confidence
Immigration	+3.1 %	+2.1 %	+1.0 %	−1.5	−4.1	+1.4	firms, the tax base and pensioners gain; the budget pays first, and local job-seekers for a while	medium
Return	+1.7 %	+1.3 %	+0.7 %	−0.8	−2.5	+1.0	returning families and the regions gain; the budget pays	medium
Retention	+1.4 %	+1.3 %	+0.7 %	−0.8	−5.2	+1.2	young people and the regions gain; the budget pays, dearly per person	low
Family benefits	+0.5 %	−0.1 %	−0.3 %	−0.6	−9.8	+1.1	families with children gain and child poverty falls; the budget pays; the labour market gains nothing before 2040	medium
Balanced policy	+3.5 %	+2.4 %	+0.8 %	−2.0	−11.0	+2.0	all of the above gain; the budget pays about as much as for family benefits	medium

Population, employment and economy: the difference from the base path in percent. Pensioner poverty and debt: the difference in percentage points. Budget cost: the package's own effect on the budget balance, added up over 2028–2045, before the EU rule's response; a negative figure is a cost. Confidence: how sure the direction and size of the result are, given the assumptions it rests on.

The table says four things. First, no package changes the picture: even with a balanced policy the population in 2040 is about 9 % smaller than today. The question is not whether to stop the decline but by how much to slow it, and at what price. Second, the order by cost is firm. Immigration and return give almost as many people as retention for less than half its cost. That order does not change however the assumptions are varied. Third, extra workers give a larger economy and more pension contributions, not a richer average resident. Fourth, family benefits answer a different question. They raise births and cut child poverty, but add nothing to the workforce before 2040. That is not an argument against them. It is an argument for not putting them in the same row as the workforce instruments.

For a decision-maker a simple sequence follows. First the cheap instruments that act quickly: immigration conditions and the return programme. Then retention where it is aimed at particular regions. And family policy on its own merits, not as a workforce measure.

What to keep in mind

A labour inflow costs debt first and pays it back later. The debt column in the table is 2040. By 2045 the sign has turned: government debt is 0.7 points of GDP lower than on the base path with immigration and 1.9 points lower with a balanced policy, because the extra workers' taxes outgrow the services their families use. The years after 2040 rest on more assumptions than the years before, which is why the page prints 2040.

The model does not employ the whole inflow. By 2050 about three-quarters of an extra labour inflow is in work in the model; the rest shows as unemployment. The European Commission's Ageing Report assumes, by convention, that all of it works. The gap is not the labour market: it is the time the capital stock takes to grow with the workforce, and the demand side of the economy. Read every employment figure on this page with that in mind.

The model carries two readings of structural unemployment, and they disagree. The labour-market module puts unemployment at 5.9 % in 2040. The economy module carries a structural rate, that is, the rate the economy returns to when the ups and downs are taken out, of 6.7 %, a point above it. The two are being reconciled; until they are, the base path's unemployment is best read as "about six per cent", and the band on the unemployment chart says how far it can move.

What the demand-led channel is worth. If employers could not invite workers to posts they cannot fill, the population in 2040 would be 1.3 % smaller and about 6.3 thousand fewer people would be in work. That channel is in the base path because the register shows it; its size is a judgement.

The model is not exact about the past. For population and employment it matches the 2025 record to within a fraction of a percent. For births it gives a 2025 figure 8 % above the record (12,900 against 11,900), so the base path starts from slightly too many births. The recorded pensioner poverty rate before 2016 does not match the model either: in the crisis years median income fell and the poverty threshold with it, while pensions held, so the recorded rate dropped below 20 % in 2010 to 2013. The model does not replay the crisis. From 2017 on the two agree, and the calculation starts from 2025, where the fit is good.

The cost of the programmes per person is an assumption, not a measurement. What one extra arrival, one returner or one young person kept costs the state budget has not been measured in any Latvian programme. The retention cost is the least certain of the three, which is why its confidence in the table is low. Integration is modelled as a wage gap that closes over time, not as language courses and recognition of qualifications; if integration is slower in reality, the employment gain arrives later.

The uncertainty band covers two assumptions, not all of them. They are the share of the posts that departures leave which a firm advertises again, and the time it takes to open the posts it still wants. Neither has a Latvian record behind it. They move the population little and unemployment a lot. Other assumptions, for instance about fertility, mortality or how strongly emigration answers the wage gap, are not in the band.

The 2028 start is an assumption. Every package starts in 2028 and reaches full strength in 2030. A later start shifts everything to the right; the result in 2040 is then smaller, but the relations between the instruments are the same.

Can Latvia stop its workforce from shrinking?

The short answer

Why it matters

What drives it

What can be done

Opening immigration

A return programme

Youth retention

Family benefits

What the packages do

Assessment

What to keep in mind