Chapter 08 of 13

The culprit: housing demand from population growth, colliding with a supply response that built the wrong thing

The bodies

Between 2010 and 2024, Iceland added roughly 75,000 residents to a population that started at 318,000 — a 24% increase, the fastest in the developed world over the period. Two-thirds were foreign citizens (Forsætisráðuneytið, March 2026). The foreign-citizen share rose from about 7% in 2015 to nearly 18% by the end of 2025; the foreign share of employment is higher still, at roughly 24%. The Forsætisráðuneytið's own growth accounting is blunt: 60% of Iceland's real GDP growth over the last 15 years is attributable to increased labor supply. Without the population growth, average annual GDP growth would have been 1.1%, not 2.9%.

These workers did not arrive by accident. Iceland's EEA membership guarantees freedom of movement for European workers, and the economy pulled them in. Tourism tripled its share of exports from 19% to 32%. Construction ran at capacity. The labor-intensive service sectors — hospitality, cleaning, food service, fish processing, warehousing — could not fill shifts with Icelandic workers alone. The vacancies were real, the flights were cheap, and the EEA meant no visa was needed. Efling, the general workers' union that represents many of these sectors, grew into one of the largest unions in the country on the back of this expansion.

The EEA framework was designed for a continental economy. When Polish or Romanian workers move to Germany — 84 million people, deep housing stock, a mature institutional rental sector — the demand shock is diffuse. When the same workers move to Iceland — 400,000 people, negligible housing buffer, no institutional rental sector to speak of — the proportional impact is orders of magnitude larger. The framework guarantees the same freedom of movement to both, but the absorption capacity is radically different. This is not an argument against free movement. It is an observation that Iceland's domestic housing policy needed to account for it, and did not.

The populist framing says the new workers drove down wages. That is not an inflation story — wage suppression is deflationary — and it is not what happened. Over 2015–2025, wages for immigrant workers grew +87.6% (nominal), outpacing Icelandic-background workers at +80.3%. The bottom of the income distribution grew fastest in real terms (d10: +32.5%, d20: +40.8%) while the top was essentially flat (d90: +11.0%, p99: +0.9%). The immigrant-Icelandic wage gap narrowed, not widened. This is a tight labor market pulling up from the bottom — the opposite of the suppression story.

But the workers' own union identifies the actual channel. Efling, under Sólveig Anna Jónsdóttir, has been explicit that the catch-up wage demands of the last several kjarasamningar are not about headline inflation in the abstract — they are about capital-area rent consuming their members' take-home pay. The workers are not saying their wages are too low. They are saying their housing costs are too high.

That is the channel. Every one of those 75,000 new residents needed somewhere to live. They are overwhelmingly renters — foreign workers in Iceland rarely have the credit history, savings, or verðtryggð-mortgage access to buy. They concentrate in the capital area, where the jobs are. Their rental bids compete with each other and with Icelandic renters for a housing stock that did not grow to match.

HMS's own numbers put a shape on the pressure. The non-market rental waiting list for working-age low-income tenants — the specific demographic the labor-absorption boom is pulling in — more than doubled between 2022 and 2025, from 1,923 to 3,871 households. Every other waitlist category (elderly, student, established social rental) stayed flat or shrank. The surge is in one category, and it is the one that matches the flow.

The question becomes: what happened on the supply side?

The walls

How many new housing units did Iceland build to absorb these people? About 26,700 between 2015 and 2024. On the cruder measure — headcount per household at 2.5 — the aggregate volume roughly matches the 60,000 new residents. But the cruder measure is misleading, and three sharper measures all point the same direction.

First measure: adults per dwelling. HMS uses this instead of headcount because housing is bid by adults, not households. Over 2015–2024, Iceland's adult population grew 22% while the completed dwelling stock grew 19%. That three-point gap is what HMS quantifies as the íbúðaskuld, the accumulated housing debt: 10,000–15,000 units, most of it piled up in 2017–2019 when demographic acceleration ran ahead of the pipeline. 2025 drew the capital-area backlog down by about 500 units. At that rate the clearance takes decades.

Second measure: absorption. Completions are not occupancy. The Central Bank's FSR 2026/1 reports that new-build apartments in the capital area now take about 18 months to clear at current sales velocity, against roughly six months for older stock. More than half of new builds sold below asking price in February 2026. New builds dropped from 20% to 16% of capital-area sales in 2025 even as they made up a larger share of listings. Units enter the completions tally. They don't enter households.

Third measure: the size mismatch. Iceland needs roughly 4,000–5,000 new units a year for the next 25 years; about 3,000 are being built. The shortage is sharpest in apartments under 60 m² — the segment that actually matches single workers, young couples, and widowed seniors downsizing. The current development pipeline for sub-60 m² units runs at around 450 units against a requirement of roughly 1,200 per year. The market is not producing what the demand is bidding on.

Fourth measure: the scale gap. Construction-cost estimates from Hannarr ehf (compiled by Margeirsson, 2022) run the arithmetic on the same 85 m² apartment built two different ways on an identical plot. A two-storey "sprawl" development yields 88 units at 42.9 million ISK per unit. A five-storey dense development on the same 10,000 m² plot yields 221 units at 34.1 million ISK per unit — a 21% lower cost per unit and 2.5 times the unit count. The scale economy is real and available. It is not being captured because individual private developers building 20–40 units at a time cannot absorb the land cost or the upfront financing that unlocks it. Only balance-sheet investors with long-duration liabilities — pension funds, insurance companies, large non-profit housing societies — can. And in the Icelandic model, they are not the ones building.

The four measures describe a single story. Private developers face the same incentive structure — high land cost, fixed permitting overhead, expensive financing — and they all converge on the same product: a 100–110 m² owner-occupied apartment priced for the upper half of the income distribution. Everyone chases the margin. Nobody builds the small rental stock. Nobody captures the density dividend. The surplus from that convergence sits in 18-month inventory while rents on existing stock get bid up by the workers and seniors the new product does not reach. Rent in the capital area has risen about 26% in the last 2.75 years (HMS leiguvísitala, May 2023 to February 2026); house prices over the same window rose about 17%. Rents outpacing prices by nine points is the fingerprint of that divergence — unsold new-build inventory stacking up on one side of the ledger, and renters bidding up existing rentals on the other.

The wrong product

Look at what the private sector actually produced. Of the 3,912 new-build apartments sold in the capital area since 2022, only 52 — 1.3% — sold for under 40 million ISK. The median sale price is 73.9 million, the median size 91 m². Units under 50 m² make up just 2.5% of new construction; in the pre-2000 stock that has transacted recently the same size class is 6.4%. Even the old buildings — built in an era of lower costs and smaller household expectations — had more than double the share of small-format housing developers are building today.

Iceland's dozens of small developers all optimise at the unit level, and they all reach the same answer: a 70–110 m² apartment priced for the top half of the income distribution. Nobody is building the 45 m² studio the arriving Efling member can afford, because nobody can carry small rental stock on a developer balance sheet. The uncoordinated market converged on the unit the credit-financed buyer wanted, not the one the renter needed.

The orthodox defense of this pattern is filtering: a household buys the new 74-million-króna apartment, vacates a mid-market unit, whose buyer vacates a smaller one, whose buyer vacates a rental, and so on down the ladder until the renter at the bottom gets a choice they didn't have before. Build at the top, the argument goes, and the bottom eventually clears.

It was never going to clear in Iceland. Over the 33 months from May 2023 to February 2026 the HMS rent index rose 24.5% while the house-price index rose 15.4%. Rents are running nine points ahead of prices. If top-of-market supply were relieving bottom-of-market pressure, the two indices would be converging, or at least tracking. They are diverging — and the reason is mechanical. The filter depends on existing households trading up into the new stock, but Iceland's incremental demand is not trade-up demand. It is migration-driven demand, concentrated in the rental market, with no trade-up chain to begin with. A new 110 m² penthouse in Fossvogur does not free a 45 m² studio in Hlíðar for an Efling member who showed up in 2023. The ladder is missing a rung exactly where the new arrivals actually land.

When the rate hikes hit, developers did not discount. They held their 2021–2022 list prices and watched unsold units pile up on their balance sheets, waiting out what they read as a temporary dip in credit-financed demand.

The shape of that wait shows up best against the full housing stock. Iceland has about 144,000 dwellings — almost all owner-occupied or rented privately, turning over slowly as households move. Above that vast, near-static base sit two much smaller inventories: the non-profit rental stock (Bjarg, Búseti, Brynja, Blær, Félagsbústaðir, Félagsstofnun stúdenta and regional operators) and the developer-held pipeline of new builds that hasn't transacted to a household or a non-profit operator. Each year 3,000–3,800 new completions drop in from above; they either flow down into the private stock via kaupskra first-sales, feed the non-profit band, or pile up at the top.

Two things stand out. The non-profit band barely moves: from ~6,200 at end-2019 to ~8,900 by end-2024 — roughly 450 units a year against 3,000–3,800 annual completions. The part of the system that builds explicitly for the renters on waiting lists absorbed about 13% of the country's new construction, and in 2024 even that share collapsed: Bjarg alone went from 212 new units in 2023 to 47 in 2024, and no other operator picked up the slack.

The red band is what we can count — units physically sitting on developer balance sheets. Anchored to BYGG hf's audited inventory line, which rose from 6.5 to 17.3 billion ISK between 2021 and 2024 (roughly 88 → 234 units at the 74-million capital-area median), then scaled across the six biggest pure-play residential developers, it grew from ~400 at end-2019 to ~2,250 by early 2026. HMS's January 2026 count independently puts 1,611 actively listed unsold new-builds in the capital area (up from 1,116 a year earlier) — consistent with this series once you add non-capital listings. This is the hard, physically verifiable warehouse.

The gray band above it is everything else that doesn't reconcile — cumulative HMS completions since 2019 minus kaupskra first-sales minus direct non-profit deliveries minus the audited developer-book series. By early 2026 it sits at ~6,900 and contains, in unknown proportions: 2023–2025 cohorts still working through a normal 24-to-36-month sell-through; institutional buyers whose purchases appear in kaupskra as corporate transfers rather than first-sales; build-to-let holding companies beyond the top-six sample; and owner-retained self-build completions. We know these units exist (the accounting has to balance) but can't cleanly attribute them. The Fjármálastöðugleikaskýrsla confirms the direction: new-build time-on-market at ~18 months, more than half of sales closing below asking, construction-sector credit growing roughly 28% in real terms year-on-year. So the "9,000 not cleared" number decomposes into ~2,250 we can verify on audited dev books and ~6,900 the data can't explain unit by unit but can only bracket.

The forward picture is worse. HMS's own April 2025 status report gives the cumulative view: 4,051 units financed since mid-2016, 2,934 already delivered, 78% in the capital area, 41% earmarked for working-age low-income tenants — the exact demographic whose waiting list doubled 2022–2025. For 2025 HMS has 7.3 bn ISK budgeted; through mid-April 393 applications had arrived and only 36 had been approved. Approvals typically lag completions by about three years, so the 2025 approval cliff is baked into 2027 and 2028 deliveries.

Zoom in to one developer. BYGG hf — the largest pure-play residential developer in Iceland — grew its inventory of unsold units from 6.5 billion ISK in 2021 to 17.3 billion by 2024, a 168% increase, while holding operating margins at 17–18% through 2023. Financial expenses roughly tripled, from 302 to 855 million ISK, and management did not discount to move inventory. Jáverk, a mixed contractor-developer, went further: its annual reports show rising interest expenses being capitalized directly into work-in-progress rather than expensed through the P&L. Jáverk's 2023 annual report told investors rising rates would have "negative effects on the profitability of our own projects, but have not yet affected the progress of our projects." In 2024 Jáverk reported zero sales of its own projects.

BYGG's 2024 results are where the wait breaks. Operating margin collapsed from 18% to 8.6% while inventory grew another 41%. The FSR's below-ask sales rate crossed 50% in the same window; new-build inventory time sat at 18 months. This is what capitulation on the developer side of a rate-hike cycle looks like, and it arrived three years after the CBI started hiking, not in the hike year itself. What looks like a delayed supply response is the rate-tool pass-through finally catching up — not through list prices, but through the balance sheet.

None of this reduced the demand that was actually driving housing inflation. The rate tool bit exactly where theory said it should — on credit-financed buyers of 74-million-króna apartments — and what it produced was a developer balance-sheet problem, not a cheaper apartment for a renter. Biting the buyers of the wrong product does not cause the right product to appear. The price-side cooling looks like policy working; read the rent index and you can see what's actually happening. The renter chasing a job in the capital is still bidding up the same thirty-year-old stock against the same migration inflow. The policy rate cooled the thing that was never the problem.

The forward pipeline

Volume caught up in 2025. The capital area delivered 1,991 new housing units against a municipal estimate range of 1,563–2,884, and HMS calls 2025 the first year since housing-plan records began in which new need was met nationally and the accumulated íbúðaskuld was drawn down slightly. That is the lagging indicator, and it is the recovery story the fiscal ministry has been emphasising.

The leading indicators point the other way. HMS publishes a measure that compares the stock of building-ready plots at the start of each year to that year's estimated housing need. The capital region sits at 58% for 2025 — the lowest of any densely populated region in Iceland. The 1,991 completions came from projects allocated 3–5 years earlier, during the 2020–2022 low-rate window. What the 58% says is that the capital area met its 2025 need by draining a pipeline that is now half-empty.

A second, independent signal corroborates the thinning. Planitor indexes every meeting minute from Reykjavík's planning committees; filtered to unique new-build residential applications deduplicated by address, the series runs at 85–90 projects per year through 2016–2018 and then collapses: 28 in 2022, 17 in 2023 — the year the policy rate sat above 9% — and only partial recovery since (29 in 2025). Meeting-minute counts are a rough proxy, but the HMS plot inventory and the Planitor application flow point at the same thing from different directions: the front of the pipeline is thinner than the back.

HMS's monthly bulletin is already showing the drought arriving. H1 2025 completions came in at 1,483, below H1 2024 (1,599) and H1 2023 (1,624). The pipeline lag from first application to occupiable unit runs 3–5 years. What entered planning in 2022–2023 is what will be completed in 2025–2028, and what entered planning in 2022–2023 is nearly nothing. 2025 is the peak of the recovery, not the trough of the decline.

The fiscal composition failure

Suspect 6 cleared the level charge against fiscal policy but kept the configuration charge open. This is where the configuration lives.

Iceland's functional spending on housing and community amenities (COFOG 06 in the national accounts) fell by 26% in real per-capita terms between 2015 and 2024 — the lowest level on record. Population grew 19% over the same window. At the municipal level in Reykjavík, the real per-resident operating budget of the Welfare Division rose 21.4% while Schools and Leisure rose only 3.3%. The city was absorbing the population shock and expanding welfare transfers while the aggregate housing-supply line shrank.

The stofnframlög line — capital subsidies to non-profit housing associations — is the only COFOG 06 sub-component moving in the right direction. The state raised it to roughly 7.3 billion ISK in 2024 against a target of 600 new non-market rental units per year. Actual delivery has averaged roughly 460. 2025 landed at 197 units, the worst year on record, arriving at the same moment the low-income rental waiting list doubled from 1,923 to 3,871 over three years. The national non-market queue now stands at 6,864 households, 85% of it concentrated in the capital area.

At the moment when population was growing fastest, rents were rising fastest, and private construction was being squeezed by high rates, the public sector's aggregate real per-capita investment in housing and community amenities fell by a quarter nationally. The money that was not spent on housing went to transfers: COFOG 10 social protection grew by 33% per capita nationally, and Reykjavík's welfare division by 21% per resident. Iceland chose to hand out payments to the households getting squeezed rather than fix the supply-side condition causing the squeeze — and no monetary instrument corrects that choice.