What migration system does an ageing Netherlands need?

In one passage of the 2026 King’s Speech, the Netherlands needs fewer arrivals. In another, it needs to attract people from abroad and persuade them to stay. Asylum appears through the problems of inflow, reception, and return, alongside a continuing commitment to protection. International students and specialized knowledge workers appear later as part of the economy’s future. The country is short of capacity and short of people, sometimes in the same speech. The Troonrede, delivered on September 15, contains both stories.

The apparent contradiction has a defensible explanation. The government is discussing different people, purposes, and legal powers. A person seeking protection is making a claim under a different system from an employer seeking permission to recruit a specialist. Neither decision can sensibly be made by subtracting one arrival from the other.

The harder inconsistency arises when a government promises control over migration as a whole while leaving the relationship between these systems unexplained. A country can want particular skills, owe protection to particular people, and struggle to house both. It still needs a method for deciding what to do where it has a choice, and for meeting its obligations where it does not.

The system developed here combines relatively simple skilled-work admission with a narrow route for essential skills that domestic adjustment cannot supply in time. Recruitment should sit alongside higher wages, better working conditions, training, productivity, and service reform. Housing capacity should influence the pace of discretionary recruitment in the short to medium term, while housing policy reduces the scarcity that makes those choices so painful. Workers need enough independence to leave bad employers without their residence, income, and housing collapsing around them.

This is a disclosed judgment about which imperfections to accept, not a result the evidence mechanically selects. More detailed strategic selection could identify valuable workers whom simple rules miss; a lower-migration strategy could relieve capacity pressures more decisively. Both deserve serious consideration. The hybrid gives substantial weight to administrative feasibility, service continuity, worker autonomy, and resistance to organized lobbying, and therefore accepts some loss of targeting precision.

The government’s own policy is already broader than a choice between recruiting abroad and relying entirely on domestic adjustment. Its July Talentstrategie, or Talent Strategy, combines education, participation, and productivity with international recruitment. The Prinsjesdag education announcement says the plans and budgets will be developed further. A strategy, however, has yet to answer the questions that an admission decision forces into the open: which shortage counts, who verifies it, who pays, and what happens when circumstances change?

The scale of migration makes those distinctions consequential. The Centraal Bureau voor de Statistiek, the Dutch statistics office usually called CBS, records 308,357 immigrants and 214,821 emigrants in 2025. The resulting net migration was 93,536. Arrivals were therefore much larger than the net addition from migration; neither number is the same as the number of residence permits issued. Emigration here includes the relevant administrative corrections. CBS population accounting also records births and deaths, which are separate components of population change.

The motive statistics require another distinction. Among non-EU/EFTA nationals in the provisional 2025 series, CBS counted 18,715 labour migrants, 42,760 family migrants, 34,510 asylum migrants including asylum family reunification, 21,995 study migrants, and 28,410 people under temporary protection. These are registered immigrants classified statistically, not interchangeable counts of applications or legal entitlements. The non-EU/EFTA series cannot be combined with the latest EU/EFTA purpose series, still covering 2024, to produce a supposedly comprehensive 2025 chart.

Nor does asylum’s numerical place determine its importance. Protection, reception, family separation, and the authority to remove someone are consequential even when another migration category is larger. The useful implication of these figures is narrower: ordinary third-country work admission governs only part of the movement that eventually appears in a national total. Making that part more selective would leave substantial responsibilities elsewhere.

An annual number can express a preference about population growth. It cannot explain how to reconcile a right to reside, an essential service lacking staff, and a municipality lacking homes. Those decisions require a system.

What ageing actually changes

An ageing country need not be a shrinking one. Its population can grow while the number of older people rises relative to the population conventionally described as working age. It can also have more people of working age without a proportionate increase in paid hours. Confusing these measures allows almost any demographic statistic to become an argument for almost any migration policy.

The current CBS central projection illustrates the distinction. It puts the Dutch population at about 19.64 million in 2050, with 44.4 people aged 65 or over for every 100 aged 20–64. By 2070, the corresponding figures are about 20.56 million and 48.4. This already assumes continued positive net migration, including approximately 57,500 in 2050. Continued immigration and substantial ageing coexist in the same projection. These are conditional model results, with uncertainty, rather than population targets. CBS’s 2025–2070 projection does not describe a future that policy has already secured.

The age ratio is useful, but its name can mislead. A person aged 64 is not necessarily employed, and a person aged 66 is not necessarily economically inactive. Nor is an older person simply a cost: unpaid care, voluntary activity, accumulated assets, and paid work fall outside a crude division into dependants and providers. For public finances and service delivery, the relevant questions include how many people work, their hours and earnings, what they produce, and what services the population needs.

Migration can change several parts of that calculation. Recruiting an employed adult can add skills and paid hours without waiting for a child to grow up. Whether it adds much usable capacity depends on the job, qualifications, language, recognition requirements, and the resources needed to make the work possible. A care worker cannot create a fully staffed service alone if supervision, facilities, or funding are missing. Adding people and adding functioning capacity are related operations, but the connection has to be made.

The demographic effect is nevertheless substantial. In a separate CBS scenario study published in 2024, the population aged 20–64 in 2050 ranges from 9.849 million in the low-migration scenario to 11.602 million in the labour-migration scenario; the study’s forecast scenario gives 10.834 million. Older age groups expand across the scenarios. The scenarios vary in more than migration and use an older baseline than the current projection, so their difference is not a clean estimate of migration’s isolated effect. They show plausible futures with materially different potential workforces and continued ageing.

Two familiar claims therefore need qualification. Migration cannot remove the ageing of existing generations, and immigrants who remain also grow older. Keeping an age ratio constant through repeated recruitment would create further population and capacity consequences that themselves need evaluation. Yet the fact that immigration cannot end ageing says little about whether it can make adaptation easier. A policy need not abolish a problem to be useful.

The relevant objective should be an economy and public services capable of supporting an older population. Several responses are available. More productive work can sustain more consumption and tax revenue with fewer hours. Better access to paid work can increase participation. People who want longer working hours may be able to provide them if scheduling and care constraints change. Training can redirect effort toward scarce capabilities. Some services can be reorganized; some activities may have to become more expensive or smaller.

Each option changes who bears the burden. Longer working lives have different implications for someone in physically demanding work and someone with control over their schedule. More paid hours can reduce time available for unpaid care. Productivity gains require investment and successful adoption, while a service that depends on personal attention cannot be assumed to become less labour-intensive merely because other industries do. The Rijksinstituut voor Volksgezondheid en Milieu, the public-health institute RIVM, describes both technological possibilities and implementation limits in its care and social-domain outlook. It provides no basis for budgeting on technology eliminating the need for people.

There is also no obligation to preserve a particular rate of aggregate economic growth. A smaller economy could coexist with better housing access or a more acceptable use of space. Conversely, slower growth can make it harder to finance services that residents want. The question is what happens to welfare per person, including public services and the distribution of costs, rather than whether a larger population produces a larger total GDP.

This matters when reading demographic advice. The Staatscommissie Demografische Ontwikkelingen 2050, the commission asked to assess the country’s demographic future, favored moderate growth toward roughly 19–20 million inhabitants in 2050. Its Gematigde groei report weighs broad welfare and presents guiding ranges. It does not discover a scientifically optimal population. Agreeing with its concern about capacity does not turn a preferred range into a legal instrument for achieving it.

For migration design, the demographic evidence establishes a reason to retain recruitment as an option. It leaves open how extensively to use it. That next decision depends partly on the powers the Netherlands possesses.

Which migration can the Netherlands select?

Consider two people applying for the same vacancy. An eligible EU worker may rely on free movement; a worker from outside the EU may need an employer-backed residence and work route. The vacancy is identical. Dutch authority over admission is not.

Under the EU free-movement framework, eligible EU workers cannot be subjected to an ordinary Dutch occupational admission quota. Residence conditions, safeguards against abuse, and lawful restrictions exist, but they do not create a general power to admit only workers whom Dutch officials consider economically desirable. EU family and residence rights also vary with circumstances. A national selection proposal that skips this distinction can promise much more control than it could deliver.

The Netherlands has considerable influence over the conditions in which mobile workers are recruited. Labour standards, housing regulation, inspections, taxation, training, and the organization of production all affect the demand for labour and the treatment of people supplying it. These are substantive migration-related choices even when they do not operate at the border. A government can change an economic model without deciding whether each EU worker may enter it.

For third-country work, selection authority is greater, although the route matters. The Wet arbeid vreemdelingen, the law governing employment of foreign nationals, supplies part of the work-permit architecture. The Immigratie- en Naturalisatiedienst, or IND, decides residence applications; the Uitvoeringsinstituut Werknemersverzekeringen, or UWV, has labour-market assessment functions. The national kennismigrant route for highly skilled migrants uses recognized employers and salary conditions. The EU Blue Card has its own European legal framework and should not be treated as another name for the national scheme. IND’s highly skilled migrant guidance sets out the operational requirements.

This is the main territory for the policy choice developed below: which economically selected third-country workers should have an accessible route, under what conditions, and with what independence after arrival? Even here, national design must respect EU rules and the rights attached to each status. A policy objective is not a substitute for route-specific legislation.

Study and research have another purpose. Their value includes education and knowledge, alongside later labour-market contribution. Dutch implementation of the EU study and research framework governs relevant third-country routes; EU students are within free movement. Treating every student as a preselected future employee would distort the activity being regulated. The Wet internationalisering in balans remained before the Tweede Kamer on September 17, 2026, with debate and voting unfinished. It was not an enacted replacement for existing rules.

International protection is different again. Entitlement follows the applicable protection criteria. It cannot depend on whether a person’s occupation appears on a shortage list or whether a fiscal model predicts a positive contribution. Reception, procedures, integration, and lawful return are policy responsibilities; protection itself is not a reward for labour-market usefulness. Current IND guidance on the Pact changes also distinguishes applications made before June 12, 2026, from the new procedure. A single commencement date does not make every pending case legally identical.

Family migration cuts across these categories. A partner can join a worker, a protected person, or an EU citizen under different rules, and may subsequently work. The EU Family Reunification Directive contains enforceable rights as well as conditions and national discretion within its scope. It does not turn all family applications into an economic selection exercise. Nor does a statistical family label tell us whether the person will remain outside the workforce.

The asylum-family position is particularly unsettled. In guidance updated on August 31, IND says it is applying the new conditions to certain applications lodged before June 12, 2026, but is pausing other affected cases while a court tests whether that retroactive application is legally defensible. That is an operational statement about a defined group of applications, not proof that all new restrictions are lawful or that all family cases have stopped. IND’s account of the pause leaves the transition dispute unresolved. No final ruling has been identified that would justify treating the dispute as settled.

European human-rights law also constrains how family separation is imposed. In M.A. v. Denmark, the European Court of Human Rights found an Article 8 violation in the particular Danish three-year waiting arrangement, including its inadequate scope for individual assessment. The judgment does not make every waiting period unlawful. It does make clear why a government cannot evaluate family rules solely as devices for reducing a count.

Temporary protection adds another distinct legal basis, relevant to people displaced from Ukraine. Whatever their employment outcomes, they are not admitted through ordinary economic selection. Including protected residents in housing and workforce planning is sensible; treating their entitlement as something to award only to useful occupations is a different proposition and an inappropriate one.

These distinctions set the boundary for all three systems considered here. They concern choices within discretionary economic admission and complementary domestic policy. None offers lawful comprehensive control over every arrival. A government should explain that boundary before attaching a number to its ambition.

A vacancy is not yet a migration case

A vacancy records an employer’s wish to hire under particular conditions. It does not establish that those conditions are reasonable, that the job must continue to exist in its present form, or that recruiting abroad is the best response. Equally, a vacancy that persists after better pay and serious training efforts cannot be dismissed by observing that employers would prefer cheaper workers. The policy task is to identify the mechanism producing the shortage.

The 2025 interdepartmental review of labour migration, known as the IBO Arbeidsmigratie, treats demand for migrant labour as shaped by economic institutions and business models. Employer demand is not a fixed national requirement waiting to be measured accurately enough.

Start with pay and conditions. If jobs remain vacant because the wages do not compensate for unpredictable schedules, arduous work, or expensive travel, recruitment may allow the employer to preserve those terms. Better pay could attract applicants, reduce departures, or make investment in labour-saving equipment worthwhile. But the response might instead be higher prices, lower profits, reduced output, or closure. The existence of a wage adjustment does not tell us which outcome follows or whether society should prefer it.

Public services introduce an additional decision-maker. An organization may be unable to offer better terms because its funding does not allow them. Describing the resulting difficulty as a shortage can obscure a budget choice; insisting that pay rise without changing the budget can obscure the same choice from the opposite direction. If more staff are hired at higher wages, somebody must finance the service. If the budget stays fixed, improving wages may leave less money for other capacity. Migration cannot resolve that arithmetic on its own.

Training provides a different test. A shortage may reflect a genuine delay between expanding training and producing qualified workers, or years of inadequate training effort. Both can leave a service short today. An adjustment requirement should demand evidence of funded places, completion, supervisory capacity, and time to practice. Recruitment can also change domestic entry and retention. A US study of foreign nurse recruitment found such responses in that setting, a mechanism warning rather than a Dutch effect estimate. An essential-skills route should therefore monitor realized net staffing and hours, domestic training completion, occupational retention, and service output rather than treat recruited headcount as proof of additional capacity.

Recognition and organization can be as important as recruitment. If qualified people already present cannot use their skills, the problem may lie in language support, credential recognition, or access to supervised practice. If employees want more hours but cannot obtain reliable schedules, the missing resource is not necessarily another arrival. Such possibilities need investigation rather than assumption. A government should be equally skeptical of an employer claiming that only migration can help and a minister claiming that a large unused workforce can be activated at negligible cost.

The scale of the alternatives is worth taking seriously. In a February 2024 illustrative exercise, De Nederlandsche Bank, the central bank, asked what would be required to maintain roughly 1.5% annual economic growth in 2026–2040 if only one adjustment lever did the work. The exercise required approximately 110,000 additional labour-force participants annually, rather than the roughly 10,000 expected in that exercise; 1.4% annual productivity growth, rather than the then-recent trend around 0.5%; 88% participation among people aged 15–74, rather than about 75%; or five more hours per worker each week by 2040. These are stylized alternatives, not September 2026 forecasts, targets, or an estimate of the required number of immigrants. DNB’s exercise makes a demanding counterfactual visible.

The honest comparison is not migration versus adjustment. It is one portfolio of adjustment against another. One can reject DNB’s growth objective without making the alternatives effortless. A closed vacancy can become a robot. It can also become a longer waiting list, an unbuilt home, a higher tax bill, a closed business, production abroad, or work moving into informal channels. Which response occurs depends on the sector, the institutions around it, and the time available. A lower-migration policy should therefore state which adjustments it expects, when they become available, and what happens if they disappoint.

The causal evidence also resists a universal story. Michael Clemens, Ethan Lewis, and Hannah Postel studied the exclusion of Mexican Bracero workers from US agriculture. They did not find the anticipated gains in domestic farm wages and employment; technology and crop choices were important adjustment channels. The study challenges the assumption that excluding foreign workers automatically reallocates their jobs to residents on better terms. Historical American agriculture supplies evidence about a mechanism, not a numerical forecast for Dutch care or construction.

Evidence from Switzerland points to a different mechanism. Andreas Beerli and colleagues studied expanded access for European cross-border workers. Skill-intensive firms increased productivity, innovation, and size, and highly educated resident workers gained opportunities and wages. Their findings show how access to foreign skills can expand labour demand rather than simply divide a fixed stock of jobs among more workers. Cross-border commuting, however, omits much of the settlement and housing burden relevant to Dutch immigration. Importing the estimated gain while leaving out that difference would be convenient rather than informative.

Together, these studies establish why the occupation and institutional setting matter. Scarcer labour can induce useful technological change, but the gain need not take the form of higher pay for displaced competitors. Greater access to workers can support innovation, but not every employer expansion creates the same benefit. The appropriate question is what productive activity becomes possible, or what adjustment becomes unnecessary, because the person is recruited.

Fiscal claims require the same discipline. The 2025 review by the Centraal Planbureau (CPB), the Netherlands Bureau for Economic Policy Analysis, characterizes the Dutch evidence on labour migration’s budget effect as neutral to mildly positive, with differences related to earnings, employment, age, family circumstances, and duration. That is a synthesis of labour-migration evidence, including related Dutch accounting work. It cannot be promoted into a fiscal endorsement of all migration categories.

The IZA lifetime-accounting study by Jan van de Beek, Joop Hartog, Gerrit Kreffer, and Hans Roodenburg is useful for a different warning. Under its assumptions, labour migration produces positive balances while several other motive groups produce negative balances. That is not a clean independent finding that overturns the preceding CPB synthesis: the Dutch evidence bases overlap, and the populations and questions differ. The exercise uses profiles anchored to the immigrant population present in 2016 and depends on future earnings, household behavior, discounting, and expenditure allocation. It is lifetime accounting, not a causal estimate of admitting one additional person in 2026.

For admission policy, the missing comparison is what would happen without recruitment. Would another worker fill the position, would investment change, or would output disappear? Would a partner work, would the household remain, and what services would it need? Lifetime accounting can illuminate one set of fiscal consequences; marginal causal evaluation asks another question. The less satisfying but more defensible conclusion is that fiscal effects vary by route, age, earnings, employment, family circumstances, duration, assumptions, and accounting method. No single fiscal value belongs to “a migrant.”

The practical implication is a presumption of diagnosis before exception. A persistent essential shortage can justify recruitment even when domestic reform remains necessary. A business that cannot retain workers on acceptable terms should not acquire an immigration entitlement merely by reporting that its vacancies are hard to fill.

Count households, rights, and capacity

The case for useful recruitment becomes less comfortable when the worker needs a home. An additional person can help sustain a service and still face a housing market in which the supply response is too slow. Neither contribution nor need cancels the other.

The official 2026 statistical housing shortage is 384,000 dwellings, or 4.6% of the housing stock. The measure subtracts roughly 43,000 available dwellings from estimated demand of 427,000. The ministry describes it as an indicator of housing-market tension; it is not a count of homes missing solely because of migration, and its planning framework does not imply that a measured shortage of zero is necessary. The official calculation establishes substantial present scarcity, rather than a permanent maximum population.

Housing demand must be counted through households. An individual joining an existing household, several adults sharing accommodation, and a family seeking its own dwelling create different demands. Those arrangements can also change over time. The methodological point in CBS’s household analysis of immigration is essential: persons entering the population cannot simply be translated one for one into required homes. A recruitment forecast needs plausible household trajectories, including departures and family formation.

That qualification should improve the analysis rather than dissolve the problem. When additional households compete for a stock that cannot expand quickly, the costs may appear as higher rents, longer searches, overcrowding, or delayed household formation. Their distribution depends on income, tenure, and location. A well-paid recruit’s ability to secure a home does not establish that the arrival has no effect on other people seeking one. Nor does an employer’s economic gain establish that the municipality receiving the household has the resources to meet its needs.

Some housing demand would grow without migration. Ageing, smaller households, and household formation among existing residents also matter. Supply depends on construction capacity and the availability of land, infrastructure, and workable planning decisions. The Planbureau voor de Leefomgeving, the public agency for environmental and spatial analysis known as PBL, examines these wider choices in Keuzes voor de leefomgeving 2025. Assigning the whole shortage to migrants would confuse one source of demand with the functioning of the entire housing system.

It would be equally mistaken to declare migration irrelevant because construction should have been faster. A supply failure remains a real constraint for someone looking for accommodation now. New homes, grid connections, schools, transport capacity, and municipal services do not appear immediately because an economist has correctly identified supply as endogenous. A discretionary recruitment decision can therefore be sensible over a long horizon and poorly timed in a particular place. The timing is part of its welfare effect.

My judgment is that near-term housing and infrastructure capacity should influence the pace and composition of economically selectable migration. This requires looking at actual accommodation, household needs, and realistic delivery schedules, rather than treating an announced development as if the keys were already available. Where recruitment would meet an urgent service need, government should make the competing costs visible and decide who finances the response. It should not claim that the need for workers has made the housing consequences disappear.

That judgment applies within legal discretion. Housing scarcity does not create a general authority to economically select eligible EU movers or refuse people entitled to protection. Planning must include those populations because their presence cannot be managed like an optional recruitment program. A government that plans only for the residents it finds convenient to predict is leaving its municipalities to absorb the difference.

The long-term judgment is different. The second mistake is to treat the existing housing shortage as if nature had revealed the Netherlands’ permanent carrying capacity. Housing supply is itself a policy variable. Existing residents also form households, need different homes as they age, and move toward work and family. Chronic failure to accommodate those changes is a housing-policy failure with consequences well beyond immigration.

This does not make building instantaneous or costless, or make space and infrastructure cease to matter. Policy should instead aim to reduce the scarcity that currently makes additional demand so difficult to absorb. Migration decisions would then be made increasingly on demographic, economic, rights, distributional, and environmental grounds, with continuing spatial planning. A temporary brake justified by actual delivery constraints should not silently become an ideology of permanent incapacity.

Recruitment can also help expand capacity. Construction skills, technical capabilities, and care staff may improve the services and infrastructure available to residents. The relevant comparison must consider that contribution alongside demand. It cannot presume that every additional worker is either a net burden or their own complete solution. A recruit who could help build housing still needs somewhere to live before the new housing exists.

Duration makes this harder. PBL’s study of registered EU labour migrants initially settling in rural areas found that roughly half of the selected 2011–2017 cohorts remained registered in the Netherlands six years later. That is not a retention rate for all migrants. It is sufficient reason to question a housing strategy that assumes labour recruitment creates only a succession of short stays. Temporary accommodation and a temporary employment contract do not guarantee a temporary demographic outcome.

Employer-provided housing appears to offer an immediate solution. It may supply accommodation where none would otherwise be available. It can also give the employer influence over a worker’s home as well as their wages, particularly when recruitment and transport are controlled through the same arrangement. The Nederlandse Arbeidsinspectie, the Dutch Labour Inspectorate, documents these mechanisms in a 2026 study of 55 selected completed case files. Its cases illuminate exploitation and serious disadvantage; they do not estimate prevalence among all employers or migrant workers.

Current Dutch law is not a blank slate. The Wet goed verhuurderschap requires separate employment and rental agreements for the statute’s defined category of arbeidsmigrant. Government guidance explains that job loss does not by itself end the rental agreement. But the definition is narrower than all migrant workers, and separate documents do not guarantee a different landlord or durable housing after job loss. The government is developing additional rental-contract reform and has announced an intended 2028 change to minimum-wage rent deductions. Those are proposals, not protections already in force.

The policy objective is practical independence. Leaving a bad job becomes harder if it threatens home, income, and residence at once. Where an employer or recruiter controls accommodation, government policy should prefer genuinely independent tenancy and require meaningful transition protection after job loss. The exact legal vehicle is not established here. Some protections can be conditions of corridor participation; a broader guaranteed post-employment occupancy right may require primary housing legislation. The evidence does not establish one specific Book 7 amendment as the necessary route.

The same reasoning applies to work permission. Article 11 of the EU’s recast Single Permit Directive requires employer mobility within its scope, while permitting Member States to use specified notifications, continuing-eligibility checks, and a temporary suspension of an employer change. It also protects specified periods of unemployment. Dutch transposition remained incomplete on September 17, 2026: implementation bill 36958 had not completed debate and voting. Current IND guidance describes three- and, in specified cases, six-month search periods, while identified circular text still contains an older three-month formulation. Operational guidance, EU requirements, and the route-specific Dutch legal position therefore still need reconciliation.

A reform should make the policy objective explicit: the system should minimize the worker’s dependence on one employer within the legal possibilities of each route and should legislate clearer continuity where present law does not provide it. That is not a promise of a universal six-month right or unrestricted occupation switching. A GVVA employer change may still be subject to a labour-market check and other conditions, and website guidance cannot by itself settle every source-hierarchy question. The uncertainty matters because a worker deciding whether to leave cannot organize their life around a policy aspiration.

Family life belongs in the same calculation. Retaining a person means retaining someone with relationships and plans, not merely preserving the original employment contract. Partners may work, children may need education, and households may choose to leave. That is not leakage. It is what happens when policy involves people. The economic rationale for predictable family and settlement rules is plausible, but the evidence does not establish a precise Dutch retention effect. Rights have an independent basis even where the economic effect is uncertain.

That changes how the system should count success. Admissions are observable quickly. Stable employment, progression, independent housing, and retention take longer to assess. A recruitment program that fills a vacancy briefly and leaves a household insecure may look successful in its first administrative report, but it has not yet demonstrated that it works.

Why a perfect selection system is probably the wrong goal

Once all these effects are acknowledged, an attractive proposal suggests itself: admit the people whose overall contribution is highest. Consider productivity, essential services, housing, fiscal effects, family circumstances, and the interests of countries of origin. Replace a crude salary threshold with a proper measure of social value.

As an objective for evaluating a system, this is an improvement over counting vacancies. As a test for deciding individual applications, it places extraordinary demands on officials. They would need to know how long someone will stay, what their household will become, which domestic alternative would otherwise occur, and how future public benefits compare with costs in a particular location. Some information can be estimated. Much cannot be observed at the moment permission is requested. That is not an argument for ignoring broad welfare. It is an argument about where to apply it.

The Adviesraad Migratie, the public advisory council on migration, confronts both sides of this problem in Afgewogen arbeidsmigratie. It develops a practical framework looking at the job, the migrant, and the company, while acknowledging that not every broad-welfare effect can be observed or weighted and that political choices remain. The preference here for simpler rules therefore goes beyond the council’s conclusion.

Even a perfectly informed official would face legitimate disagreement. How much output should be forgone to reduce housing pressure? How much public money should preserve a service? How heavily should benefits to migrants themselves count? A spreadsheet cannot decide those weights. Giving each criterion points merely places the political judgment inside the formula, where it becomes harder to inspect.

Occupational lists reduce the individual information problem but create a classification problem. A broad title may contain both scarce and readily available skills. A precise title may become obsolete as technology changes. Employers have reason to describe jobs in terms that qualify, while sectors have reason to present their own needs as nationally essential. These are predictable design risks, not allegations that every submission is dishonest. The state should use employer information while recognizing why it cannot be the sole basis for a decision.

There is a real argument for doing the difficult work anyway. Some shortages are sufficiently durable and some service failures sufficiently costly that rough targeting could outperform an easier rule. The relevant comparison is between imperfect institutions. A salary threshold also embeds judgments and can be engineered around; its simplicity does not make it neutral. A publicly funded occupation may have large social value without generating a salary that clears a high gate.

The case for simplicity is therefore conditional. Transparent criteria can make decisions quicker to understand, easier to contest, and less dependent on discretionary classifications. That benefit must outweigh the valuable distinctions the rule omits. It is sensible to retain some exceptional selection where the omission would be particularly damaging, provided the exception does not gradually become the main system.

Employer obligations face a similar temptation to promise precision that has not been earned. A generic charge per migrant might be described as making employers pay the social cost. But the underlying cost must first be identified. Permit administration, unpaid wages, local infrastructure, and training underinvestment are different problems. Some already have fees or legal duties; others call for broader taxation or service funding. Economic incidence does not respect the label printed on the invoice. The UK Migration Advisory Committee’s analysis of a skills charge identifies possible shifts to workers, resident wages, and consumers. It does not supply a Dutch rate.

A defensible charge would identify an incremental cost, explain why that activity is the appropriate tax base, specify the use of the revenue, and assess who ultimately pays. Without those steps, a new levy could obscure the same distributional problem it claims to solve. Enforcement against underpayment should not be replaced by allowing firms to purchase an appearance of social responsibility.

Administrative effort also has an opportunity cost. Staff evaluating an elaborate application are not simultaneously processing other cases or inspecting a sponsor. This does not prove that new tests are unaffordable. It means that each proposed test should come with observable evidence, a responsible decision-maker, a review procedure, and resources. If a criterion cannot be explained at that level, its presence in an elegant policy model is premature.

Broad welfare should therefore discipline the choice of rules and the evaluation of results. It should also determine which omissions are unacceptable. It cannot responsibly be advertised as a complete measurement of each person’s future social worth.

Three systems worth choosing between

Legal compliance and rights constrain every workable model. Within that boundary, the relevant criteria include welfare per person, the cause of labour scarcity, productivity, domestic alternatives, fiscal sustainability, and physical capacity. Distribution, worker autonomy, retention, and origin-country effects determine whose welfare is being counted. Administration, adaptability, resistance to capture, and public accountability determine whether the proposed rules could survive contact with actual decisions.

Several coherent systems can satisfy those criteria. They differ in the weights assigned to them and the risks government is willing to accept.

A strategic selection model - let’s call it Model A - would make deliberate recruitment a larger part of the response. Streamlined skilled routes would continue, while a more extensive system identified scarce capabilities and essential lower-paid occupations. An independent labour-market panel could advise government, which would publish designations and its reasons. Pay, conditions, training, and capacity would enter the assessment. Publicly valuable work would not be excluded simply because its remuneration failed to reflect its importance.

This is Model A’s strongest claim: information can be incomplete and still useful. Government need not calculate every lifetime consequence to recognize a persistent capability gap, a long training lead time, or a service that cannot be maintained without additional staff. Where the stakes are high, the cost of more sophisticated assessment could be justified. A choice to avoid administrative complexity can itself be expensive if it repeatedly excludes people whose work would relieve serious constraints.

To remain credible, Model A would make recruitment part of a sector plan rather than a substitute for one. Employers would demonstrate genuine vacancies, compliant pay and recruitment, and contributions to training. Household-sensitive regional planning would accompany designations. Family rights and predictable settlement would be preserved, while workers could change employers within the law and would not lose status merely because an occupation later left the list. Independent accommodation and effective complaints would protect the worker from becoming the means by which a staffing plan is enforced.

The beneficiaries would include selected firms, services, and their users. Costs would fall partly on public administration and on places absorbing additional demand; employers excluded from favored categories could face disadvantages. High targeting ambition would require staff capable of challenging sector evidence, handling review, and revising categories. Its strongest failure mode is not excessive migration in the abstract - it is institutional capture disguised as precision. Transparency and expiration rules can reduce that risk, but they cannot remove the information burden that creates it.

An adjustment-first lower migration model - Model B - would place greater weight on present capacity and on changing the economy’s dependence on recruitment. It would narrow discretionary third-country work routes, retain access where complementarity is especially compelling, and use only exceptional transition or emergency recruitment for essential lower-paid work. It would invest political and financial effort in wages, job quality, participation, training, automation, and managed contraction of activities whose continued scale was no longer considered worthwhile.

The serious argument for this model begins with a distinction between a firm’s viability and society’s welfare. A business can be profitable while depending on scarce housing, weak bargaining power, or public expenditure that its accounts do not fully reveal. Making labour harder to obtain could encourage capital investment and better employment, or reduce activities whose output is worth less than the pressures they generate. Lower aggregate GDP would not by itself demonstrate failure. A resident finding a home, or a worker obtaining a more sustainable job, does not receive welfare as a share of the national GDP headline.

Model B would also avoid building a comprehensive occupational scoring authority. It would use existing admission powers, stronger standards, and staged tightening, while concentrating integration effort on people already entitled to remain. Housing construction would continue, but slow near-term delivery would weigh strongly against additional discretionary admissions. Study recruitment would be aligned more tightly with educational capacity and plausible retention, rather than institutional expansion alone. Lawful family rights, existing status, and worker mobility would remain protected.

Model B is internally honest only if it accepts the consequences when adjustment does not supply enough labour. Higher wages can require higher consumer prices or larger public budgets. Training takes time; automation may not fit the task; some activities may close or move. Where services cannot respond, users may face rationing. Those costs are not merely complaints from employers deprived of cheap workers. They are losses borne by residents, including people with little ability to buy alternatives.

Workers with bargaining power could gain, while others might lose their jobs as production contracts. Municipalities dependent on shrinking activities could lose revenue or employment before new investment arrives. Tighter direct admission might also encourage subcontracting, posting, or informal work, so enforcement would remain demanding. A workable Model B needs transition support, realistic timetables, and escape clauses. It would reduce reliance on a controllable part of migration; it could not guarantee an aggregate national ceiling.

A simple-gate hybrid model - Model C - would preserve understandable skilled routes and a small, temporary essential-skills corridor. The normal skilled gates would use observable salary, qualification, and sponsor-compliance requirements appropriate to each route. The corridor would open only where a persistent, important shortage survived a serious assessment of domestic alternatives and where implementation and housing were addressed. Broader welfare would guide the system’s design and review, rather than become a score attached to every application.

Model C takes its main discipline from Model B: a vacancy does not automatically qualify for an exception. It takes a narrower insurance policy from Model A: an essential service need not wait for domestic adjustment to succeed when the evidence shows that it will arrive too late. Employer compliance, independent housing, worker portability, and predictable family and settlement rules would do more of the work that a comprehensive selection apparatus otherwise tries to accomplish at admission.

The hybrid has no privileged claim to fiscal success. Its result would depend on who arrives, household employment, duration, and the alternatives displaced. If Model B actually produced fewer net additional households under comparable departure, substitution, and supply assumptions, it would tend to reduce direct discretionary housing demand more than Model C. That is an intended direction, not a measured outcome. Model C also accepts that simple gates will miss some useful lower-paid capabilities. High-paid employers retain an accessible route while some lower-paid services face a demanding exception test. That distributional choice needs defending.

The comparison is therefore between three kinds of error. Model A risks obsolete or captured classifications. Model B risks withholding recruitment while desired services and productive capabilities deteriorate. Model C risks excluding valuable people through simple gates while its supposedly narrow corridor expands under pressure. The corridor is narrower than Model A’s apparatus, but it still creates boundaries that can be lobbied, a danger of stale evidence, the likelihood of classification disputes, and additional review costs. Simplicity reduces the surface area of politics, but it does not abolish politics.

Under this essay’s weighting, Model C is the architecture carried forward. That weighting gives substantial importance to service continuity and recruitment flexibility, distrusts a permanent apparatus that tries to measure the complete social value of a widening range of occupations, and gives worker autonomy independent weight. These objectives can conflict in practice. They are normative weights applied to evidence, not an empirical proof that Model C is optimal.

A different legitimate weighting could produce a different choice. Greater concern about immediate capacity relief could support Model B, with its larger adjustment bill. Greater confidence in public targeting, or greater concern about the capabilities salary thresholds miss, could support Model A. The evidence clarifies the trade-offs and failure modes. It does not vote.

The hybrid in practice, and the price of choosing it

Model C begins with a limited promise: make useful recruitment possible under clear rules without turning labour scarcity into worker dependence. IND would continue to decide residence and sponsor questions. UWV would supply labour-market evidence and perform Wav assessments, with any new duties created explicitly. The Labour Inspectorate would supply compliance evidence; municipalities and housing authorities would supply capacity evidence. Responsible ministers would make the broader statutory designation decision and publish their reasons. Labour-market analysis is not the same task as deciding how much service continuity or housing pressure should count.

For highly skilled routes, normal gates should remain observable: salary actually paid, qualifications where relevant, and sponsor compliance. The national route and Blue Card would retain their legal distinctions. Salary is a practical proxy, not a declaration of social value. Threshold engineering should be answered through verification and sponsor consequences rather than an ever finer taxonomy of desirable jobs.

The essential-skills corridor would need a legal mechanism, not merely a good shortage test. Model C would create it by statute inside the existing third-country work-admission architecture. Designation would not itself grant a residence permit. For a designated function, the corridor would replace the ordinary priority-supply step, where that test would otherwise apply, with its own designation and evidence requirements. Individual residence eligibility, lawful employment terms, employer compliance, and other route-specific requirements would remain. Outside the skilled routes and the corridor, ordinary third-country paid employment would continue through the normal GVVA or TWV machinery and its labour-market test. Creating the exception and its continuity rules would require primary legislation, not administrative guidance.

A sector, public-service authority, social partner, or ministry seeking designation should make a public case showing persistent scarcity, service or transition consequences, and the realistic domestic response. It should document wages and conditions, turnover, recruitment, training, housing effects, and why substitutes cannot meet the need in time. Employer submissions should be tested against evidence from workers, service users, training bodies, UWV, inspectorates, and public authorities.

The five broad elements can remain, but they are not boxes that mechanically produce a permit. Observable evidence covers persistence, vacancies, wages, turnover, recruitment, training, and housing delivery. Analysis asks whether domestic adjustment is credible on the relevant timetable. Ministers then make the normative judgment about whether service or transition consequences justify an exception. UWV can inform the first two layers without being asked to calculate broad welfare.

The domestic-adjustment test should be demanding without becoming a ritual delay. An authority should distinguish measures that have failed from measures that were never adequately funded, and measures that could work eventually from those available in time. A documented training lead time could justify temporary recruitment alongside an expanded training program. An employer refusing realistic improvements in conditions should face a different result. The monitoring obligation should then test whether recruitment produced additional capacity: realized net staffing and hours, domestic training completion, occupational retention, and service output matter more than the number of permits issued.

To keep the corridor small, government should designate defined capabilities or functions for a stated purpose, not entire sectors permanently. Each designation should specify duration, evidence, implementation resources, housing assumptions, and renewal conditions. The evidence establishes no defensible national quota or universal numerical shortage threshold.

The statute should make contestability routine. Published criteria, disclosed evidence, reasoned decisions, consistent treatment, and a defined review mechanism guard against capture and arbitrary exclusion. The exact objection or judicial-review route depends on the legal form and belongs in legislative drafting. Model C has to survive scrutiny, not evade it.

Designations should expire unless affirmatively renewed. A two-year review cycle, annual monitoring, and an independent system evaluation every three years are governance proposals, not empirical optima. New corridor admissions should stop when a designation expires. Existing workers should keep the residence and settlement protections on which they relied; the statute therefore needs explicit grandfathering and transition rules.

Those rules should answer what happens when a worker changes employer or function. While a designation remains active, an eligible corridor worker should be able to move to another compliant employer within the designated function through the simplest procedure EU and Dutch law allow. If the worker moves to a non-designated function, ordinary route eligibility should apply. If the designation expires, existing workers should retain a time-limited path to change employer within the previously designated capability during their current permit, while new recruitment closes. If a criterion does not have answers to those questions, it is not yet a policy - it is an aspiration with a dropdown menu.

Worker portability is therefore more than a kindness added to an economic program. It is a labour-market institution. Current law supplies part of the framework, but not a universal answer: Article 11 of the recast Single Permit Directive permits employer-change conditions and protects specified periods of unemployment, while Dutch transposition remained incomplete on September 17, 2026. Model C should minimize dependence on one employer within each route’s legal limits and legislate clearer status continuity where present law does not provide it. It should not promise unrestricted occupation switching or a universal six-month search period.

Employers may object that mobility lets a competitor benefit from recruitment expenditure. That incentive problem is real, but it does not justify retaining workers through threats to residence or housing. Legitimate recruitment costs can be allocated transparently without coercive debts or mobility restrictions.

Housing should be addressed before expanding the essential corridor through regional plans based on household demand and credible delivery. Those plans should distinguish accommodation available now from construction that depends on future approvals, grid connections, utilities, or staffing. They should identify who pays for the local response. Near-term capacity can therefore constrain corridor expansion without turning the 2026 housing shortage into a permanent national ceiling.

Existing Dutch law separates employment and rental agreements for a defined labour-migrant category, but its scope is narrower than all corridor workers, and separate contracts do not guarantee residential independence. Model C should prefer third-party or otherwise independent tenancy. Where an employer supplies housing, it should require a separate lawful lease, complaints outside the recruitment chain, and meaningful transition protection. Some measures can be corridor conditions; broader occupancy rights may require primary housing legislation.

Employer obligations should be concrete: genuine pay, sponsor compliance, lawful recruitment, controls on fees charged to workers where the applicable rules support them, accessible complaints, and consequences for serious repeat violations. The Wet toelating terbeschikkingstelling van arbeidskrachten, or Wtta, establishes an authorization system for labour-supply businesses. Its commencement is phased: preparatory provisions began in 2026, core application machinery follows in 2027, and the central prohibition and enforcement stage follows in 2028. The law is therefore part of the enforcement architecture, not evidence that agency abuses were already solved in 2026.

Model C should connect sponsor and agency information so corporate rearrangement does not erase serious misconduct. Disqualification needs clear standards and review, while affected workers need a route out. Enforcement that first removes the worker’s security can deter the complaints on which enforcement depends. Inspection and case-handling capacity are part of the model’s cost.

Any employer charge would need its own demonstrated cost base and incidence assessment. A local training obligation, an administrative fee, and a contribution toward specific infrastructure cannot be justified by the same vague assertion that migration costs money. Formal liability does not determine who bears the final economic burden. Revenue should be attached to the problem it is meant to address. Model C does not need a generic migrant-employer surcharge to look complete.

Family and settlement rules should be predictable enough for households to plan, and capacity forecasts should include partners, children, partner employment, retention, and departures. A long-term skills policy should offer a credible path toward durable residence within the applicable law while keeping family rights independent of fiscal attractiveness. Predictability may aid retention, but the evidence supplies no precise Dutch effect.

International students should enter this system as students. Educational capacity and the value of the education should come first, with strategic fields and later employment informing recruitment and support rather than defining the entire purpose. Nuffic’s graduate-cohort study found that 25.3% of international graduates remained resident five years after graduation, and roughly 80% of those stayers were employed. The denominator is graduates in the studied cohorts, not all international entrants, and five-year residence is not proof of permanent settlement.

The useful policy response would preserve a workable bridge from study to employment, support language and professional recognition where needed, and judge outcomes over time. Recruitment into a program should not be defended by assuming that everyone stays or condemned because some graduates leave. The education may have value in either case. Housing and teaching capacity still constrain how many students an institution can responsibly accommodate.

For refugees and other people entitled to remain, the priority is enabling participation without attaching economic conditions to protection. Language, qualification recognition, childcare, and continuity between reception and municipal support should help people use their skills. Swiss research by Jens Hainmueller, Dominik Hangartner, and Duncan Lawrence found that longer asylum waiting times reduced later employment. The causal estimate belongs to that setting; it should not be transferred as a Dutch magnitude. It does warn against treating prolonged waiting as economically inconsequential.

For eligible EU movers, Model C’s admission gates would have no ordinary selecting role. The relevant instruments remain labour standards, housing, taxation, training, and enforcement. That limited reach is why the surrounding economic reforms matter: selection over only part of recruitment cannot compensate for weak standards across the labour market.

The EU Talent Pool is a voluntary matching framework, not a residence or work right. The Netherlands abstained when the Council adopted it in March 2026. A Dutch letter before adoption left participation to a later decision; no later primary decision establishing Dutch participation had been identified before publication. A matching service cannot substitute for housing, usable qualifications, fair work, or credible family prospects.

Active recruitment must also consider origin countries. The WHO safeguards framework distinguishes organized recruitment from health workers’ freedom to migrate. Dutch-organized recruitment from vulnerable health systems should respect those safeguards and, where appropriate, use bilateral arrangements that support additional training capacity.

The government should also publish the evidence by which the whole system will be judged. Employment and earnings matter, but so do household housing, service continuity, realized net staffing and hours, domestic training completion, occupational retention, worker mobility, complaints, processing delays, and actual service output. Independent evaluation should compare outcomes with plausible alternatives, not claim success whenever admitted workers find jobs. Sector submissions and relevant lobbying contacts should be disclosed, and Parliament should be able to see why an exception survives.

There is a specific test of whether the hybrid remains worth carrying forward. It should look worse if corridor designations expand persistently, lobbying dominates the evidence, classifications repeatedly miss the problem, valuable lower-paid capabilities are excluded, domestic training falls rather than rises, recruitment fails to increase net service capacity, administrative delay becomes excessive, worker mobility remains practically constrained, or housing costs are left unmanaged. It should also count attempts to recreate corridor-style exemptions elsewhere in the ordinary system to avoid the corridor’s safeguards. A design that cannot reveal its own failure is not meaningfully simpler.

Its price is deliberate incompleteness. Some useful people will not qualify. Some firms will contract, and some services will remain under pressure while adjustment proceeds. Taxpayers and consumers may pay more for better conditions and domestic capacity. Continued skilled recruitment will still add housing demand, and protecting workers’ freedom to change jobs will prevent government from guaranteeing that every recruit remains where the original plan placed them. Those are real sacrifices.

Poorly managed trade-offs can fall heavily on housing seekers, patients, municipalities, workers tied to employers, firms that cannot adjust quickly, and taxpayers or consumers financing transition. No source shows that one of these groups must bear every alternative cost. The point is to make costs visible rather than displace them by accident. Under the weighting declared here, Model C asks government to justify recruitment, enforce its conditions, preserve domestic adjustment, and change course when evidence changes. Its claim is not that migration solves ageing, but that selective recruitment can coexist with rights, housing constraints, administrative limits, and worker autonomy.