when support comes too late / stoff.devlog

the swiss systems idea still feels important to me, partly because it is political and partly because it is personal.

i went through the iv system myself. it took years, cost energy, cost health, and ended with a lawyer getting involved before a pension was finally granted retroactively. even then, the result was only around 1690 chf per month, which is not enough to live on without supplementary benefits anyway.

that experience left me with a question that has not gone away:

if the eventual outcome is long-term support plus top-ups anyway, why does the path toward that outcome so often involve delay, pressure, repeated checks, institutional hopping, and avoidable uncertainty first?

this is not an anti-integration argument. that would be too simple, and it would also be false in my own case. some integration support i received earlier was genuinely valuable. my apprenticeship and bms mattered. even now, the point of finally having a pension is not to stop there forever. it is that the pension creates enough breathing room to focus on health properly and maybe study later, which would not even be thinkable without the earlier integration path.

so the point is not that integration is bad and pension is good. the point is that the transitions around integration can become full of friction, delay, and administrative nonsense. the help itself can be useful. the path through the system can still be badly sequenced.

the core tension

the swiss iv is built on a principle that makes sense in theory: integration before pension. the federal council’s 2024 report on occupational integration in the iv makes that logic explicit: a pension is only examined once a person’s integration potential has been exhausted.

in principle, that is understandable. if people can be stabilized and reintegrated into working life, that is better than treating long-term support as the first answer.

the problem begins when this principle hardens into a sequence that is too rigid for unstable cases.

for some people, especially those with serious psychological instability or long periods of uncertainty, integration is not the first meaningful step. stability is. treatment access is. predictable income is. relief from constant administrative pressure is. without those things, integration can stop being help and start becoming a performance.

that is the difference i keep returning to:

integration may still be the right long-term direction.
but stabilization may be the precondition for getting there at all.

where the current path feels backwards

what made the system feel wrong to me was not only the final amount. it was the order of things.

while iv was unresolved, i still had to deal with rav routines and job-search logic because income had to come from somewhere. from the outside that can maybe look like activation. from the inside it felt more like being kept administratively active while institutions sorted out responsibility behind the scenes.

that is the part i keep struggling with. the system often seems to act as if delay itself is neutral, as if it costs nothing to wait, appeal, deteriorate, and only then arrive at the support that was needed much earlier anyway.

but delay is not neutral. it can cost:

treatment continuity
housing security
trust in institutions
future work capacity
years of energy that could have gone into recovery

the waste is not only financial. it is also time, health, and human attention.

what public data already supports

this idea starts from lived experience, but it is not standing alone as pure feeling anymore. some public facts already support taking the problem seriously.

according to the 2024 iv statistics, around 227,300 iv pensions were paid in switzerland, and the iv financed vocational integration measures for 57,700 people in 2024. this is not a marginal system. it is trying to do both things at scale: integration and long-term support.

the economic reality after a pension decision is also often tight. according to the 2024 supplementary benefits statistics, 49.2% of iv pensioners were also receiving el. that does not prove my case, but it does support the broader point that a large share of people on iv cannot live on the pension alone.

the federal council’s 2024 report on occupational integration also shows a real shift in timing. for the iv registration cohorts it examined, the share of people receiving an iv pension within four years fell from 26% to 17%. the same report says that, after the 2008 reform, the number of people with earnings four years after registration increased, but so did the number receiving social assistance.

that combination matters. fewer earlier pensions can look like success from inside the iv. but if part of the effect is that more people are shifted into social assistance, private support, or long uncertainty, then the success may be narrower than it first appears.

the same report also shows a data problem. four years after iv registration, some people fall into a residual category with neither earnings, unemployment benefits, nor an iv or ahv pension. that group can include people on social assistance, people still in measures, and people financing themselves privately. in other words, some of the most precarious trajectories are not easy to read cleanly from one headline number.

that matters a lot. a system can look efficient if it only measures what left one institution. it can look much less efficient if it also measures what got displaced into another institution, into family support, into private savings depletion, or into untreated instability.

the system may be optimizing the wrong layer

one of the things that keeps bothering me is that the system may be administratively rational while still being socially irrational.

from inside one institution, a delayed pension decision can look disciplined. it can look like careful assessment, pressure toward reintegration, and avoidance of premature long-term dependency.

but from the person’s actual path through the system, the same delay can look very different. it can mean months or years of unstable income, fragmented treatment, repeated retelling of the same story, and pressure to stay legible to offices whose categories do not really fit the situation.

that is why this may partly be a measurement problem.

if the iv mainly measures whether people received a pension, did not receive a pension, or had earnings again after a few years, that is only one slice of the story. it does not automatically show:

how much stress the path created
how much treatment was delayed
how much cost was shifted into social assistance
how much family or partner support silently absorbed the gap
how much long-term work capacity was lost during the waiting period

a system can improve one metric while worsening the actual trajectory.

that is the deeper suspicion here: not necessarily that support is too low in principle, but that support is sequenced and measured too narrowly.

a young-adult problem too

this question matters especially for younger adults.

in february 2026, the federal council laid out the direction of the next iv reform and again emphasized integration for people aged 18 to 25, including a new integration benefit for that age group. that does not prove the system wants “no pensions before 25”, and i do not want to overstate it. but it does show clearly where the policy emphasis lies.

the question is not whether young adults should be written off early. they should not. the question is whether the system knows how to distinguish between:

someone who is ready for the next integration step
and someone who first needs stable income, treatment access, and continuity before integration has a realistic chance

if that distinction is weak, then integration first can become less of a path forward and more of a pressure loop.

not every case needs the same order

this is also not an argument for one universal rule.

some people probably do benefit from fast integration pressure. some really are close enough to the labour market that a quick push, a coaching measure, or a vocational intervention helps them avoid long-term dependence. it would be wrong to flatten all cases into one story.

that is why the real question is not whether integration is good or bad. it is whether the system is good enough at sorting cases into the right sequence.

the subgroup i keep thinking about is narrower:

people with serious instability
people whose treatment situation is not yet stable
people whose work capacity is unclear or fluctuating
people who are technically moving through institutions but not realistically ready for the next step
people who may still have long-term potential, but not under current pressure

for that subgroup, the current order may be backwards.

the issue is not motivation. the issue is timing.

the real missing piece: an existence floor

the more i think about this, the less i land on abstract ideology and the more i land on something simpler:

an existence floor.

not universal. not permanent for everyone. not the replacement of every insurance system. just a stable minimum floor that activates when a person falls into one of these transition zones:

income collapse
serious medical instability
iv clarification still pending
rav logic still running, but not realistically fitting the person
unclear institutional responsibility

the point would be simple:

survival first, classification second.

or more sharply:

decouple survival from classification.

switzerland already has fragments of this floor. social assistance is supposed to stop complete collapse. el exists because ahv and iv are often not enough on their own. unemployment insurance covers temporary income loss. but from the outside these are not experienced as one coherent floor. they are experienced as different doors, different criteria, different rituals, and different waiting times.

the floor exists in pieces. access to it is still fragmented.

what friction actually looks like

administrative friction sounds abstract until you think about what it means in practice.

it can mean:

repeating the same medical and personal story to multiple offices
being forced into job-search rituals that no one really believes in
surviving the period between one institution ending and another taking responsibility
needing legal help not to gain something new, but to secure what later gets recognized anyway
being evaluated again and again while the practical situation remains unchanged

none of this is spectacular on its own. that may be why it is so easy to normalize.

but when these layers stack over years, they become part of the actual burden of illness.

the system is then no longer responding to instability. it is helping reproduce it.

a better front end, not a whole new state

the strongest version of the reform idea is not abolish everything. it is much narrower.

keep the specialized systems.
change the user experience.

the model would look something like this:

illness / income collapse / instability
    -> single access point
    -> immediate existence floor
    -> one case lead
    -> parallel assessments
       - IV
       - RAV / unemployment insurance
       - social assistance
       - EL
    -> internal settlement between institutions
    -> long-term outcome / reassessment

the citizen sees one path:

one place to go
one initial process
one stable payment floor
one case lead

the state keeps its internal complexity:

iv still assesses disability-related entitlement
rav still handles labour-market logic where it fits
social assistance still exists
el still exists
reimbursement and legal categories are still sorted out

but that complexity stays in the background. the person does not have to perform the right kind of claimant for the right office just to survive.

that is what feels wrong in the current system: the specialization leaks outward.

current system vs. proposed logic

current path
income / health crisis
    -> which office?
       -> IV pending
       -> RAV routines
       -> social assistance fallback
    -> waiting / appeals / pressure
    -> possible pension + EL later

proposed path
income / health crisis
    -> single access point
    -> immediate floor
    -> parallel assessment
    -> internal settlement
    -> targeted long-term outcome

the real innovation here is not huge. it is basic systems design:

do not make the user carry institutional fragmentation.

a better question than “should the state pay earlier?”

i do not think the best question is whether the state should simply pay earlier.

that framing is too blunt. it makes the debate sound like generosity versus strictness.

the better question is:

when does earlier stabilization reduce downstream cost and increase the realism of later integration?

that is a systems question, not just a moral one.

if an earlier floor does not change the final support outcome, but does reduce:

legal conflict
duplicated administration
housing instability
treatment delay
institutional hopping
psychological overload

then it may still be the more efficient path.

not because it is softer, but because it is less wasteful.

what this would and would not claim

this idea does not claim that an existence floor would fix everything.

it would not:

cure illness
eliminate labour-market stigma
guarantee successful integration
remove the need for assessment
make every case cheaper

but it could plausibly:

reduce stress during instability
preserve treatment continuity
reduce pointless administrative hopping
reduce duplicated case handling
make integration measures more realistic where they are actually useful

that is already enough to investigate seriously.

in my own case, i do not need to prove that earlier stability would have made me healthy immediately. maybe it would not have. maybe the final diagnosis and long-term support outcome would have been the same. but that does not make the path neutral. if the path creates years of extra stress, legal conflict, duplicated checking, and treatment delay while ending in the same support constellation anyway, then the path itself deserves scrutiny.

what a serious existence floor would need

if this idea were ever taken seriously, it could not just be “pay everyone and hope for the best.”

a real existence floor would have to be designed carefully.

it would need:

fast activation
so people do not fall into a gap while institutions argue
simple criteria
enough to prevent obvious misuse, but not so much that the bridge becomes another bureaucratic maze
continuity
no abrupt cliff edge while classification is still unresolved
supplements
housing, disability-related costs, and family situations would still need separate treatment
reassessment
not permanent by default, but reviewed in a way that does not recreate the same instability
one responsible interface
so the client does not become the courier between offices

in other words, this would not replace institutional expertise. it would change when and how that expertise becomes the client’s problem.

the strongest version of the argument

the best version of this idea is not:

switzerland should just give pensions faster.

and it is not:

integration is bad.

it is this:

for some subgroup of cases, especially unstable or psychologically fragile ones, the system may be sequencing support badly. stabilization is sometimes the precondition for meaningful integration, not its enemy. if the state measures success too narrowly inside each institution, it may miss the cost it creates across institutions, private life, treatment continuity, and long-term work capacity.

that is the point worth testing.

the counterargument that deserves respect

there is also a real counterargument here.

any system that lowers friction risks paying some people through a less exact channel than ideal. it may create short-term extra costs. it may reduce pressure in cases where pressure actually would have helped. it may also be politically hard to defend if people believe it weakens work incentives or blurs important distinctions between unemployment, disability, and poverty.

that is why i do not think this idea should be defended with slogans.

it should be defended, if at all, through evidence:

what kinds of cases are we talking about?
what are their current trajectories?
what are the real public costs across institutions?
what are the hidden private costs?
where does the current sequencing help?
where does it merely delay?

without that, the idea stays intuitive but soft.

with that, it could become something more serious.

what still needs to be proven

there are still real gaps.

i do not yet have good public answers to questions like:

how often support is first denied and later granted anyway
how often legal escalation becomes necessary just to secure a later-recognized outcome
how often people are pushed through rav or social assistance mainly because iv is slow
how much hidden cost sits outside official system data in savings depletion, partner support, family support, or health deterioration

so this is not a conclusion yet. it is a research direction.

the strongest research question underneath it might be:

for which subgroup of cases would earlier stabilization reduce human damage and institutional waste better than prolonged classification, activation, and delay?

that feels much sharper, and much more honest, than a generic complaint about the system.

what i would want to investigate properly

if i ever turn this into actual work, i think the project would need four layers.

first, a systems map.
not just iv, but the interfaces between iv, rav, unemployment insurance, social assistance, el, and treatment systems.

second, a case layer.
not only my own path, but a small number of documented trajectories showing where people get stuck, shifted, or stabilized.

third, a data layer.
public statistics, reform reports, timing patterns, and whatever can be learned about transitions rather than only end states.

fourth, a policy layer.
not “how do we abolish the whole system,” but “what front-end redesign would reduce friction without destroying the useful parts?”

that is the level where this becomes more than a post.

why this still matters to me

what still sticks is that the system does contain real help. integration can matter. apprenticeships can matter. educational paths can matter. pensions can matter. treatment can matter.

the question is whether the path between those things has been designed intelligently.

right now, too often, it seems designed around classification first and stabilization later.

i keep thinking it should be the other way around.

support exists.
but access to stable support is fragmented, delayed, and too often tied to fitting the right institutional box fast enough.

that is not only an administrative problem. it is a health problem. a trust problem. a time problem. a recovery problem.

and maybe, in the end, a systems-design problem more than anything else.

if i take this further, that is probably where i would start.

when support comes too late