Iceland's Inflation Has a Culprit — And It Isn't the One the Central Bank Is Chasing

Iceland's headline CPI has been stuck above target for almost four years. Peer countries have moved on. Germany, the United Kingdom, the United States all have trend inflation back within shouting distance of 2%. Iceland is still printing 5.4% headline, with services at 8.4% and housing above 7%. The Seðlabanki hiked its policy rate from 0.75% to a peak of 9.25%, cut tentatively to 7.25%, then reversed course and hiked again to 7.50% in March 2026 — all while producing a working paper arguing that the problem is fundamentally about "weakly anchored inflation expectations."

That is the official diagnosis. It is wrong.

What follows is a detective story. Economic analysis does not usually benefit from whodunit framing — there are too many suspects and too few clean data points — but Iceland's case is unusually cooperative. The suspects line up, the alibis can be checked against the CPI basket, the CBI's own credit-flow data, and the HMS property registry, and the culprit turns out to be hiding in plain sight. The forensic work has already been done. What follows is the walkthrough.

Suspect 1: The króna

The first place any Icelandic inflation story begins is the currency. Iceland has a 400,000-person economy with its own floating currency, a long history of depreciation-driven inflation spirals, and a post-2008 national trauma around the króna's role in the banking collapse. Whenever inflation prints hot, the reflexive question is "what happened to the króna?"

Let's check the alibi. Here is Iceland's CPI for March 2026, decomposed:

Look at that second line. Imported goods are deflating. The basket of stuff Iceland buys from abroad — cars, electronics, clothing, foreign food, industrial inputs — is cheaper in March 2026 than it was in March 2025. Whatever is driving Iceland's headline number, the króna is not passing foreign inflation through to Icelandic shelves. The currency channel is working. World prices have cooled, and the króna has been importing that cooling for about eighteen months. If the króna were the culprit, the imported-goods line would be hot. It is the coldest line in the table.

Suspect cleared on the direct pass-through charge. But the króna is not done being interesting.

The carry trade

Iceland's policy rate is 7.50%. Here is what the rest of the developed world looks like:

Iceland is running a policy rate four times Sweden's and nearly four times the ECB's. That differential is an open invitation: borrow in euros at 2%, park in króna at 7.5%, pocket the spread. This is the carry trade — the same mechanism that, in its most extreme pre-2008 form, fueled the jöklabréf. The CBI knows it is happening. Over just a few months in mid-2025, it purchased 29.3 billion ISK in foreign currency — nearly 40% of total FX market volume in the period — to prevent carry-driven appreciation from distorting the exchange rate. Foreign reserves stood at 917 billion ISK by late 2025.

Think about what that means operationally. The CBI keeps rates high to fight inflation. The high rates attract speculative capital inflows. The inflows push the króna up. The CBI intervenes in the FX market to prevent excessive appreciation that would hurt exporters. And the inflation the CBI was trying to fight has not budged, because — as we will see — it is driven by housing, not by anything the exchange rate touches. The CBI is running two policies that partially cancel each other out: tight money to cool the economy, and FX purchases to offset the side effects of tight money on the currency. This is the impossible trinity of open-economy macroeconomics playing out in real time — independent monetary policy, free capital movement, and a managed exchange rate. Pick two. Iceland is trying to run all three and spending reserves to paper over the contradiction.

The "drain" fear

The counter-argument that keeps rates high is capital flight: "if we cut, money will leave, the króna will crash, import prices will spike, and we'll get the inflation we were trying to avoid." This is the argument that sustains emergency-level rate differentials even when the CBI suspects the rates are not reaching the actual inflation driver.

But look at which direction the capital is actually flowing. The risk right now is carry inflows, not outflows. The CBI is buying foreign currency to prevent appreciation, not selling it to defend against depreciation. Iceland does run a current account deficit — but much of it reflects Icelandic pension funds diversifying their enormous portfolios (roughly 200% of GDP) into foreign assets, not a consumption gap that requires speculative carry financing to sustain. The "drain" scenario requires you to believe Iceland is permanently one rate cut away from a currency crisis — that the króna is, in essence, hopeless.

That belief has a specific historical basis: the 2008 collapse. But the 2008 capital flight was a banking insolvency event. Money left because the three major banks owed roughly ten times GDP and could not pay. They had used the króna as a vehicle for cross-border carry, and when that carry unwound, the currency followed. Today's banking system holds assets of roughly 1.5 times GDP, is well-capitalized, and carries no significant cross-border liabilities of that kind. The institutional environment that made the króna a crisis currency has been dismantled.

Other small-currency economies confirm this is not a size problem. New Zealand (5 million people) runs a 2.25% policy rate. Norway (5.5 million) runs 4.00%. Sweden (10.5 million) runs 1.75%. None of them maintain emergency-level rate differentials to prevent capital flight. None of their currencies have collapsed. The króna's historical volatility was a banking-regulation failure, not a currency-size failure — and that banking system no longer exists.

Suspect 1 has an alibi on the pass-through charge and a good defense on the "drain" charge. The króna is not driving inflation. But the króna is the historical justification for something else — Iceland's system of CPI-indexed financial contracts, the verðtrygging. In Icelandic policy discourse, the two are often treated as inseparable: you have the króna, therefore you must have indexation. That conflation is a mistake, and it matters enough to deserve its own suspect file.

Suspect 2: Global supply shocks

The pandemic-era hypothesis everyone reached for was that supply chains were broken. Ships were anchored off Long Beach, semiconductors were unobtainable, container rates were at 10x their pre-COVID level, and inflation was "transitory but not in the sense we meant." Brunnermeier and others built frameworks around global supply-chain pressure as the driver of advanced-economy inflation. The New York Fed's Global Supply Chain Pressure Index (GSCPI) became a standard regressor.

And the hypothesis had real explanatory power — in 2021 and 2022. It does not work for 2025–2026. The GSCPI has been near zero since mid-2023. Container rates are back to pre-pandemic levels. Semiconductors are a glut. The components of Icelandic CPI most sensitive to global supply conditions — imported goods, transport — are running at −1.2% and +5.4% respectively, with transport's hottest contribution coming from domestic fuel-tax policy rather than world prices.

If Icelandic inflation were a global supply-chain story, it would have moderated with the global supply chains. It has not. Suspect 2 has an alibi for the period in question.

Suspect 3: Verðtrygging — the indexed financial system

Iceland's verðtrygging (CPI-indexation of financial contracts) is one of the country's most distinctive institutions. Roughly 65% of outstanding mortgages are indexed to CPI — the principal adjusts monthly with the price level. Some rental contracts, wage agreements, and service contracts carry similar clauses. No other Nordic or Western European country runs anything comparable.

It was built for a reason. In the late 1970s and through the 1980s, Iceland ran inflation of 25–50% per year, with peaks above 80%. At those levels, nominal long-term contracts cannot function — a 25-year mortgage denominated in nominal krónur would be meaningless within a few years. Indexation was a rational institutional response to a genuinely dysfunctional price environment. It allowed housing finance, pension savings, and long-term contracting to survive while the state worked to bring inflation under control. For the problem it was designed to solve, it worked.

It is important to be precise about what verðtrygging is and what it is not. It is a policy choice — a set of contractual conventions and legal frameworks that Iceland adopted in response to a specific historical crisis. It is not a structural necessity of having an independent currency. The króna is a fact of Iceland's monetary arrangement; verðtrygging is a decision made within that arrangement. The two are routinely conflated in Icelandic debate — "we need indexation because we have the króna" — but this conflation collapses under the simplest international comparison. New Zealand has the NZD and nominal mortgages. Norway has the NOK and nominal mortgages. Sweden has the SEK and nominal mortgages. Each of these countries has an independent currency, full monetary sovereignty, and a mortgage market that transmits central bank rate changes directly into household budgets — without an indexation layer in between. Iceland chose differently, not because the króna forced it, but because the 1980s crisis made indexation feel necessary. That crisis is forty years ago. The choice remains.

The deeper problem is that the threat changed and the instrument didn't.

Verðtrygging indexes to CPI, not to the króna directly. It was built for a world where króna depreciation drove import-price spikes, which drove CPI, which destroyed nominal contracts. In that world — the 1980s world — indexation is a stabilizer. But in the 2020s, the inflation driver is not imports (deflating at −1.2%) and not the króna (stable, with carry inflows the CBI is actively fighting off). It is housing — rents up 8.3%, imputed rent up 6.8%, together contributing roughly 1.5 percentage points to headline CPI. When housing costs push CPI up, every indexed mortgage in Iceland sees its principal grow by the same percentage. That principal growth is not a future abstraction — it is a current expansion of the outstanding credit stock. Banks' balance sheets grow. The total quantum of private debt increases not because anyone borrowed more, but because the index moved.

This is where verðtrygging shifts from having an alibi to being an accomplice. It operates through at least three channels simultaneously:

Channel 1: It neutralizes the rate tool. The CBI hikes rates to create payment shock — to make monthly mortgage costs painful enough that households cut spending. That transmission works on non-indexed mortgages, where the borrower feels the rate hike immediately in cash. On indexed mortgages, the borrower can refinance to a lower nominal rate, deferring the inflation cost into principal growth. Monthly cash flow is restored. Consumption continues. Rental bidding continues. Between 2023 and 2025, Icelandic households did exactly this: they paid down roughly 290 billion ISK of non-indexed debt and originated about 523 billion ISK of new indexed debt. The CBI pushed on the brake; verðtrygging released it.

Channel 2: It expands the credit stock. When CPI rises and indexed principal adjusts upward, the total stock of outstanding private credit grows without any new lending decision being made. This is endogenous money creation — the credit stock expands as a mechanical consequence of the price level, which contributes to further price pressure. The magnitude is debatable and the transmission indirect, but the direction is clear: verðtrygging makes the credit stock pro-cyclical with respect to inflation. In an economy already running above target, that is the wrong direction.

Channel 3: It propagates a sectoral shock economy-wide. Without indexation, a housing-cost shock stays in the housing component of CPI — painful for renters and buyers, but contained. With verðtrygging, housing-driven CPI growth flows mechanically into mortgage principal, indexed rental agreements, and indexed service contracts. The sectoral shock becomes an economy-wide inflation problem without anyone forming an "inflation expectation." The contracts do the propagation automatically.

The combined effect is a system that forces the central bank to go harder than it otherwise would. Each increment of tightening is partially absorbed by the escape hatch (channel 1) and partially offset by the credit expansion (channel 2), while the inflation being targeted is simultaneously propagated into new corners of the economy (channel 3). The CBI is not fighting inflation with one hand tied behind its back. It is fighting inflation while an institutional mechanism recycles that same inflation back into the system faster than monetary policy can drain it.

The 2020–2022 episode illustrates the amplification in reverse. When the CBI cut rates to 0.75%, indexed mortgage lending surged, housing prices rose over 30% in two years, and inflation broke out before any global supply shock arrived. The standard reading is that low rates caused the bubble. The better reading is that low rates plus verðtrygging's reflexive credit loop caused it — in a nominal mortgage economy, the transmission from low rates to CPI is indirect and slow. In an indexed economy, it is mechanical: lower rates → higher housing prices → higher CPI → indexed principal grows → more credit outstanding → further price pressure. Verðtrygging made the 2020–2022 boom worse than it needed to be, and it is now making the cleanup harder. The accomplice was at the scene of the last crime too.

There is a distributional consequence worth naming. Indexation drives a wedge between creditors and borrowers. The creditor — bank, pension fund — holds an asset whose real value is guaranteed by the index. The borrower holds a liability that grows with every CPI print. In a housing-driven inflation, the borrower is paying for a shortage they did not create through a debt burden that expands at 5–7% per year, while the lender earns a real return from the very inflation the system helps propagate. The insurance policy designed to protect households has become a mechanism for transferring the cost of a housing crisis onto them.

Chile maintains a similar CPI-indexed unit (the Unidad de Fomento) and has experienced analogous feedback dynamics. Brazil ran full indexation until the Plano Real program unwound it over three years in the 1990s — one of the most successful macroeconomic stabilizations of the late twentieth century, motivated by exactly the pro-cyclical dynamics Iceland is experiencing now.

Verðtrygging is not the culprit. It did not cause the housing shortage, the population growth, or the supply failure. But it is the accomplice — the mechanism that takes a sectoral housing problem and converts it into an economy-wide inflation that the central bank's instrument cannot reach. It was designed as insurance against a volatile króna. The króna's threat has receded; the insurance policy has become the disease.

Suspect 4: "Inflation expectations became unanchored"

This is the Seðlabanki's own theory. The CBI's March 2026 working paper, which runs a Bayesian trend-cycle decomposition of Icelandic inflation from 1985 to 2025, concludes that Icelandic trend inflation has risen from ~2.5% to ~4.5% over the recent period and that this rise is correlated with survey measures of inflation expectations. The policy inference is straightforward: restore credibility, the expectations will re-anchor, the trend will come back to target.

The problem with this theory is not the correlation — the correlation is real — but the causal direction. Survey measures of inflation expectations are adaptive. When you ask households and firms what they expect inflation to be, they mostly tell you what inflation has been recently. The research literature on this, going back to Coibion and Gorodnichenko in 2015, is clear: survey expectations in most economies track realized CPI with a lag. They are a sophisticated way of asking people "what has been happening?" and coding the answer as "what will happen."

So when the CBI says "the trend is high because expectations are unanchored," the reader should ask: why are expectations unanchored? The honest answer is: because realized inflation keeps coming in above target, households and firms notice, and they adjust their stated expectations accordingly. Calling this "expectations deanchoring" and treating it as a cause is like diagnosing a patient's fever as the illness. The fever is a symptom. The question is what's causing it.

The CBI's paper is competent descriptive work — the trend-cycle decomposition is well-estimated, the international comparison is real, the stylized fact that Icelandic trend has risen while peer trends have not is a genuine puzzle. But it does not identify the underlying driver. It labels the persistent component "expectations" and moves on. The label does not do the causal work; it performs the causal work while the reader isn't looking.

There is a deeper identification problem. As the examination of verðtrygging showed, the dominant persistence mechanism in Icelandic inflation is contractual, not psychological. When housing costs rise, CPI adjusts, and indexed mortgages, rental agreements, and service contracts adjust mechanically — no expectation formation required. The CBI's Bayesian trend-cycle decomposition picks up this automatic persistence as a slow-moving "trend" component and calls it "deanchored expectations." But the same trend is observationally equivalent to unresolved cost pass-through in a contractually-indexed economy with a real-resource bottleneck. The econometric framework cannot distinguish between the two. The label is a modeling choice, not an empirical finding.

Suspect 4 is a symptom masquerading as a cause. Not the culprit.

Suspect 5: Wage-push from a tight labor market

Now we're getting somewhere. Iceland has had an extraordinarily tight labor market over 2020–2025. Unemployment among Icelandic nationals has been below 3% almost continuously. The construction sector has been running at capacity for years. Tourism has boomed beyond any pre-pandemic baseline. Wages have moved accordingly:

These are nominal numbers and some of the growth is simply catching up with inflation, but even in real terms wages are up — the overall launavísitala is up about 37% in real terms over the decade, far outpacing the per-worker income data. This is what a tight labor market looks like, and it is the obvious candidate for wage-push inflation in the orthodox textbook.

But here the story gets tricky, because wage-push as a cause of inflation requires a direction of causation. Did wages rise first and push up prices, or did prices rise first and wages chase? In Iceland, you cannot easily tell from the aggregate data. The wage settlements (kjarasamningar) explicitly reference inflation expectations and are partly indexed in various ways. And the sectoral ranking above is revealing in a way that undermines the classical story: the fastest wage growth is in the most foreign-labor-exposed sector (hospitality), while the slowest is in the sector with almost no foreign labor at all (finance). That is not what "labor imports suppress wages" predicts.

The tight labor market is part of the inflation story. But it is more symptom than cause — wages are rising because the economy is running hot, and running hot means there is demand chasing limited real resources. That brings us back to the question: what real resources are limited? Labor itself, clearly, because foreign workers had to be imported by the tens of thousands to fill the jobs. But if labor were the primary bottleneck you would expect rising wages to ease as foreign workers arrived, and that did not happen. The wages kept rising. So labor was not the bottleneck. It was a bottleneck that imported workers mostly solved — at the cost of putting those workers somewhere.

Somewhere like, for example, rented accommodation. This is where the plot thickens.

Suspect 5 has a motive but it's circumstantial. They're a person of interest. We're going to come back to them.

Suspect 6: Cheap foreign labor suppressing wages and dragging down real incomes

Here is the classical populist framing: immigrants came in, accepted low wages, dragged down the median, and the real-income squeeze that households are feeling is the result. It fits the intuition of most observers. It is also almost entirely wrong on the Icelandic data.

The foreign-born share of Iceland's population has risen from 12.3% at the start of 2020 to 17.9% by the end of 2025 — the fastest demographic shift in Iceland's post-war history. The foreign share of employment is even higher, at about 24% in early 2026. This is a massive labor-supply shock by any standard. Classical economics would predict this should have depressed wages, especially in the sectors absorbing the most foreign workers.

It didn't. Over 2015–2025:

• Wages for Icelandic-background workers grew +80.3% (nominal)

• Wages for immigrant workers grew +87.6% (nominal)

• Overall mean wage grew +77.3% (nominal)

Every subgroup grew faster than the aggregate. This is possible only because the employment mix shifted toward the lower-wage group. It is Simpson's paradox in its rawest form. The compositional drag from the population shift is roughly 3 percentage points over a decade, or about 0.3 pp/year — real, but not remotely large enough to explain a decade of apparent wage stagnation.

More damaging to the populist framing: look at the income distribution. Over 2015–2024, real labor-income growth by percentile:

The bottom of the distribution is growing fastest, not slowest. The top is basically stagnant in real terms. This is the distributional fingerprint of a genuinely tight labor market pulling wages up from the bottom, not of a labor-import flood thickening the bottom decile. If the populist framing were right, d10 and d20 would have grown slowly and the median would have lagged. In fact the median lagged d10–d30, and d90 and p99 are effectively flat.

The gap between Icelanders and immigrants has narrowed, not widened: immigrants earned about 81% of Icelandic-background wages in 2015 and about 84% in 2025. This is "newcomers catching up," not "newcomers dragging down." The wage-suppression-from-immigration story fails every empirical check.

Suspect 6 is alibi'd out. The uncomfortable implication is that the economy is treating foreign workers well on wages. They are still paid less than Icelanders, but the gap is closing and bottom-decile wages are pulling everyone up. The squeeze that households are feeling is real, but it is not coming from their paycheck.

So where is it coming from?

The weird clue

Look at the CPI decomposition again:

• Imported goods: −1.2%

• Services: +8.4%

• Housing: +7.0%

Imports are deflating. Everything non-tradable is hot. The hottest thing in the basket is not a traded good, not a commodity, not anything you ship on a container: it is rent. Paid rent is up 8.3% YoY, imputed rent 6.8%. And rent has accelerated while house prices have cooled — Iceland's house price index is up only about 2.4% YoY, and has actually been falling in real terms since early 2025.

This is unusual. In most inflation episodes house prices lead and rents follow. In Iceland, the two have decoupled. Prices are cooling (the CBI's rate hikes worked on that margin). Rents are not cooling. Why?

Because prices and rents are bid by different people using different money. House prices are bid by people with credit — the marginal buyer is borrowing, and their bid is sensitive to the nominal interest rate. Rents are bid by people who need a place to live and do not have a mortgage alternative. Their bid is sensitive to how much cash they can commit and how many of them there are.

When population grows rapidly — say, by about 27,000 foreign residents between 2020 and 2025, concentrated in working-age adults, concentrated in the capital area, concentrated in industries that pay wages good enough to compete at the rental market but not yet good enough to buy — you get more people bidding against each other for a stock of rental units that hasn't grown proportionally.

This is the clue we have been building toward. The inflation is concentrated in the one part of the basket that is bid by bodies competing for roofs, and that part of the basket has decoupled from prices, which are bid by credit-financed buyers. The rate tool bites the credit side of the market. It doesn't bite the rental side.

Now let's line up the rest of the evidence.

The culprit: housing demand from population growth, colliding with a supply failure on all sides

The bodies

Iceland's population grew by 60,724 people between January 2015 and January 2024 — an 18.8% increase in nine years. The foreign-citizen share rose from 7.0% in 2015 to 17.9% by the end of 2025. Most of the growth has been concentrated in the greater Reykjavík area, where the rental market is.

These numbers are not controversial. Hagstofan publishes them quarterly. What is under-appreciated is how concentrated the shock is in time and place. Roughly 27,000 of the 60,000 new residents arrived between 2020 and 2025 — a period when Iceland's population is around 380,000, so this is about 7% population growth in five years, concentrated in a metropolitan area of about 250,000. By any peer-country standard this is a very large demographic shock.

The walls

How many new housing units did Iceland build to absorb these people? The official Hagstofan construction statistics stopped being updated in 2021, but the HMS property registry contains the build year of every transacted dwelling, and from that you can reconstruct an approximate completions series. It yields a lower bound of about 14,756 new residential units built between 2015 and 2024.

Against 60,724 new residents, at a flat assumption of 2.5 people per household, that is a gap of about 9,500 units — roughly 40% below what demographic growth required. If you assume immigrants live at higher household density (3+ people per unit), the gap narrows to 20–30%. Either way, there is a clear shortfall, and it gets worse when you look at where the units were built. Completions in the greater Reykjavík area — where population growth has been concentrated — averaged about 4.3 new residents per new unit. In the rest of Iceland, the ratio was 3.8. The supply response was actually smaller in the region where demand was highest.

This is a classic supply-side bottleneck. Rent in the capital area has risen about 26% in the last 2.75 years (HMS leiguvisitala from May 2023 to February 2026). House prices over the same window rose about 17%. The difference — rents outpacing prices by ~9 percentage points — is the fingerprint of rental demand rising faster than the broader housing market can absorb it.

The construction that didn't happen

Rate hikes work on housing. In 2021–2022, new-build sales in Iceland peaked at about 1,890 transactions/year. In 2022, as the CBI pushed its policy rate from 2.0% toward 9.25%, new-build sales collapsed by 46% in a single year, to 1,020. Construction-sector credit contracted. Developers faced higher financing costs and slower sales. The FSR's 2026/1 edition describes new-build inventory time in the capital area at roughly 18 months, with more than half of new builds selling below asking price.

The rate tool worked here. It cut house-price growth, it cooled new-build sales, it punished developers carrying financing costs. The problem is that it also cut the supply response — fewer projects got financed, fewer workers were hired in construction, and the pipeline of future completions thinned out. The supply-side punishment is not symmetric with the demand-side punishment because the population kept growing. People kept arriving, they kept needing somewhere to live, and the rate tool did nothing to slow their arrival or cut their rental bids. It only cut the response on the other side of the ledger.

The fiscal retrenchment that no one is talking about

The Icelandic state'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 of the series. During the same period, the population grew 19%. At the municipal level in Reykjavík, the real per-resident operating budget of the Welfare Division rose by 21.4% over 2018-2024 while spending on Schools and Leisure rose only 3.3% per resident (essentially flat, and almost certainly declining per child given that the school-age cohort grew faster than total population because of immigration patterns). Reykjavík also built only a minority share of the capital area's new housing during the crisis years — its share of capital-area new completions fell from 47% in 2019 to 32% in 2024 and 28% in 2025 — even as the city contains roughly 60% of the capital-area population. The city was simultaneously absorbing the bulk of the population shock, building a minority share of the housing response, and expanding its welfare transfers.

The headline aggregate hides a meaningful exception. Stofnframlög — the state's capital subsidies to non-profit housing societies (principally Bjarg) — were raised substantially, reaching roughly 7.3 billion ISK in 2024, projected to deliver around 600 new non-market rental units. That is the only line inside COFOG 06 moving in the right direction, and it is roughly the correct instrument for the problem. The issue is that 600 units per year is about one-eighth of what the demographic arithmetic demands, so the state is simultaneously doing the right thing on the one sub-line that matters and letting the rest of the category — housing benefits, water and sewer, street lighting, municipal housing budgets — shrink enough that the total real per-capita figure hits the lowest level recorded.

This is the least-discussed part of the story and the most damning. At the moment when population was growing fastest, rents were rising fastest, and the private construction sector was being squeezed by high rates, the public sector's aggregate real per-capita investment in housing and community amenities fell by a quarter at the national level and by nearly a third at the municipal level — even with the meaningful but under-scaled stofnframlög increase. The money that was not spent on housing went to transfers: COFOG 10 social protection grew by 33% per capita nationally; Reykjavík's welfare division grew by 21% per resident. Iceland chose to hand out payments to the households getting squeezed rather than to fix the supply-side condition causing the squeeze.

There is no monetary-policy solution to this choice. Interest rates cannot build houses. Rate cuts would stimulate private construction financing, but they would also fuel the demand side and would not touch the municipal planning bottleneck or the zoning constraints in Reykjavík. The fiscal choice was made independently of the monetary stance, and neither side has corrected it.

The mechanism

Put the pieces together:

1. Iceland's economy has been running hot since 2021, driven by a post-COVID tourism boom and a construction pipeline that needed to be built out.

2. The labor-intensity of these growth sectors meant the labor force expanded rapidly — about 27,000 net new foreign residents arrived between 2020 and 2025.

3. These new residents concentrated in the capital area and, because they are mostly working-age adults without Icelandic credit history, concentrated in the rental market.

4. Rental demand rose faster than rental supply could respond, because (a) private construction was constrained by planning, land availability, and eventually by rate hikes, and (b) public housing investment was being actively cut at both national and municipal levels.

5. Rents rose. Paid rent is running at +8.3% YoY; the rent-equivalence component of CPI (which Hagstofan switched to in June 2024) is at +6.8%. Between them they are contributing roughly 1.5 percentage points to headline CPI by themselves.

6. Services CPI, which has rent baked into it structurally and also responds to labor costs in a tight market, is running at +8.4% — above the headline, meaning it is pulling the headline up.

7. Iceland's contractually-indexed financial system (verðtryggð mortgages, verðtryggð rental contracts, some wage agreements) automatically propagates the housing CPI into the rest of the economy. When housing pushes CPI up, indexed contract terms adjust mechanically without requiring anyone to form expectations about anything.

8. The CBI responds by raising rates. Rates successfully cool house prices (which they have done) and new-build sales (which collapsed 46% in 2022). They do not cool rents, because rents are bid by renters with cash, not by buyers with credit.

9. Icelandic households who had been on non-indexed mortgages, facing nominal payment shock from the rate hikes, refinance into indexed mortgages to restore their monthly cash flow. Between 2023 and 2025 they paid down about 290 billion ISK of non-indexed debt and took on about 523 billion ISK of new indexed debt. Consumption is preserved. Housing demand is preserved. The rate hikes are registered in future principal growth rather than in current budgets. The payment-shock channel the CBI was counting on did not fire.

10. The rate tool's effective bite is therefore asymmetric: it kills the supply response (new construction) while leaving demand (rental bidding, consumption) mostly intact. Every 100 basis points of tightening probably widens the housing imbalance more than it narrows it, because the supply side is more interest-sensitive than the demand side in an indexed economy with a rapid population inflow.

11. The CBI, observing that inflation is not moderating, interprets the problem as a credibility failure and issues a working paper arguing that expectations need to be re-anchored. The proposed solution is more firmness. But there is no firmness on the policy rate that can build a housing unit, change a zoning bylaw, or stop foreign workers from needing apartments. The tool is not connected to the machinery that matters.

The verdict

Iceland's persistent inflation is not a currency problem. The króna has been importing world disinflation for eighteen months. It is not a global-supply-chain problem. The GSCPI is near zero. It is not, contrary to the official framing, an inflation-expectations problem — expectations are adapting to observed inflation, not leading it. It is not a wage-push problem in the classical sense, because the sectoral wage data show the wrong fingerprint: wages are rising fastest where foreign labor is most concentrated, which is the opposite of what the labor-suppression story predicts. It is not a cheap-foreign-labor distributional story, because the bottom of the wage distribution is growing faster than the top and the gap between Icelandic and immigrant wages has been narrowing, not widening.

It is a housing demand shock from rapid population growth colliding with a housing supply failure on three fronts simultaneously: private construction squeezed by rate policy, Reykjavík municipality building a minority share of capital-area new completions during the crisis years (28-32% of capital-area new housing in 2024-2025, against ~60% of its population), and national housing-amenities spending (COFOG 06) down 26% per capita in real terms from 2015 to 2024. Within that headline aggregate, the one sub-line that matters most for non-market rental — capital subsidies (stofnframlög) to non-profit housing societies like Bjarg — was actually raised to about 7.3 billion ISK in 2024, on track to deliver roughly 600 new units. That is a real policy response to the housing crisis, and it is the only one in the COFOG 06 category pointing in the right direction. It is also roughly an order of magnitude too small relative to the 60,000-plus population increase and the accumulated supply gap. The contractually-indexed Icelandic financial system then propagates the resulting rental inflation mechanically into the rest of the CPI basket, and the CBI's rate tool cannot reach the part of the housing market where the pressure lives.

In the language of macroeconomic diagnosis: this is a real-resource bottleneck masquerading as a monetary phenomenon. The resource in short supply is walls in Reykjavík. The resource cannot be printed; it has to be built. Interest rates cannot build it. Fiscal policy can build it but chose to spend the money on transfers instead. Planning policy can unlock it but chose to maintain the existing zoning constraints. Immigration policy cannot directly moderate the arrival rate because Iceland's EEA commitments lock in freedom of movement for the European workers who make up the bulk of the inflow; the only indirect lever is to cool the labor-intensive sectors creating the pull, which is a fiscal and sectoral choice, not an immigration one.

The Seðlabanki has been fighting a crime it did not commit, with a tool that is not calibrated for the actual offense. Every basis point of tightening further squeezes the supply-side response to the demographic demand shock while leaving the demand-side bids largely untouched. The inflation will not moderate on the Seðlabanki's schedule. It will moderate when one of three things happens: enough homes are built to catch up with the population (a 3–5 year project that nobody is currently financing), enough foreign residents leave to ease rental demand (which the FSR's 2026/1 edition describes as a "stabilizer" but which is really the human cost of using monetary policy as a demographic tool), or the indexation system is reformed to break the mechanical feedback from housing to the rest of the basket.

It is worth being explicit about what the Seðlabanki cannot do on the demand side directly. Iceland is a member of the EEA, and the EEA Agreement includes freedom of movement for workers. Iceland cannot restrict the ability of EEA citizens to live and work in Iceland, which is the channel through which the bulk of the recent population growth has arrived. The Icelandic state has no legal instrument to "slow arrivals" in the way this is sometimes loosely discussed. The only way to reduce the labor-import pull is indirectly — by cooling the labor-intensive sectors (tourism, construction, hospitality) that are generating the demand for workers in the first place. That cooling is a fiscal and sectoral policy choice, not an immigration policy choice. Treating it as an immigration problem misreads the legal structure the country operates inside.

The most likely path, given current policy trajectories, is the second one: a slow, painful, unspoken reliance on foreign-worker emigration as the equilibrating mechanism. This is not a monetary success story. It is a coordination failure between fiscal, planning, and sectoral policy, with the central bank left holding an instrument that cannot reach the problem.

What the Seðlabanki should actually do

The implication of this reading is uncomfortable and worth stating plainly. When a central bank's only instrument cannot reach the mechanism generating inflation, the correct response is not to use the instrument harder. It is to stop using it, and to say out loud why. The Seðlabanki's policy rate is not calibrated to build walls in Reykjavík, and raising it further will not build them faster. Continuing to tighten on a schedule driven by headline CPI signals to the rest of the policy system that the central bank believes it can solve the problem — which keeps fiscal, planning, and housing policy in passive mode waiting for the rates to work.

The alternative is for the central bank to hold, communicate publicly that the remaining inflation gap is a housing-supply gap, decline to tighten further even if the headline does not come down on schedule, and pass the policy ball to the institutions that can actually act. This is uncomfortable because it requires the central bank to deliberately accept some above-target inflation in the short run rather than pretend that continued tightening is getting closer to target. But a central bank that exhausts its tool on a problem the tool cannot solve is not being prudent — it is making a coordination error that delays the necessary fiscal and planning response. Sometimes the most disciplined thing a central bank can do, given its mandate, is to declare the mandate exceeds its capacity in this particular case and wait for the right policy instrument to be deployed by the right institution. The Icelandic situation is a case study in exactly when that call should be made.

Three things that would actually work

1. A public construction program. The fiscal space exists. Iceland's general government has been running deficits of roughly 2–4% of GDP since the COVID shock, but the underlying position is comfortable: debt-to-GDP is in the low sixties and declining from its 2020 peak, it is lower than in most OECD peers, and Iceland's sovereign credit is investment grade and on an improving trajectory. A direct-build program of 3,000–5,000 units per year in the greater Reykjavík area for three years is roughly 1% of GDP annually — a marginal addition to the existing deficit path, not a stretch of fiscal capacity. The 2024 increase in stofnframlög to 7.3 billion ISK is the template; it needs to be scaled by an order of magnitude. Nothing else on offer comes close to this in cost-effectiveness.

2. Planning reform in Reykjavík. The single largest non-monetary bottleneck in the capital area housing market is zoning and permit processing. The private supply response is constrained by how fast the municipality will approve projects. This is a political and administrative problem that has nothing to do with the Seðlabanki and has been treated as intractable. It is not intractable. It is just unpopular with existing property owners.

3. Ban new CPI-indexed mortgage origination. The verðtryggð system is the transmission channel by which housing-cost inflation becomes wage and contract inflation, and — as the examination of Suspect 3 showed — the escape hatch that neutralizes the CBI's rate tool. The remedy is not gradual: ban new indexed mortgage origination by legislation. Existing indexed mortgages are grandfathered — no retroactive disruption to current borrowers or to the pension funds that hold the corresponding bonds. But every new mortgage is nominal. At the current policy rate of 7.50%, nominal mortgages are unaffordable. Mortgage origination freezes. Housing demand drops. Prices fall. And falling house prices pull the housing component of CPI down with them — which is exactly the mechanism the CBI has been trying and failing to activate for four years. With housing CPI falling, headline inflation moves toward target, and the governor is forced to cut rates — not as a concession but as a response to the market conditions the de-indexation created. Lower nominal rates make new nominal mortgages affordable again. The system rebalances at a lower rate level with a mortgage market that actually transmits monetary policy. The coordination failure breaks because the legislature acts first, the market responds second, and the central bank is compelled to follow third. The pension fund sector — roughly 200% of GDP in assets, much of it matched against indexed liabilities — manages a gradual runoff of its legacy indexed book as existing mortgages amortize, without the shock of a retroactive conversion. This is the only structural reform that simultaneously repairs the rate tool, breaks the CPI feedback loop, and forces the rate normalization that the CBI will not undertake voluntarily.

None of these three are central-bank moves. They require coordination between the Ministry of Finance, Reykjavík municipality, and the legislature. The absence of that coordination is the Icelandic policy failure that deserves the attention currently being given to the CBI's rate policy. The rate policy is an honest attempt by one institution to use the only tool it has. What is missing is everything the CBI cannot do by itself — and everything the CBI cannot do by itself is what is needed.

The inflation has a culprit. It is not the currency. It is the 9,500 missing housing units in Reykjavík, and every institution that decided in the last decade not to build them.

Appendix: The Seðlabanki's own evidence

A reader familiar with the Central Bank of Iceland's public communications in 2025 and 2026 will notice that the reading offered in this piece is not the one the bank itself has been giving. The CBI's narrative — repeated across Peningamál 2026/1, Fjármálastöðugleiki 2026/1, MPC statements, and Working Paper #100 — is that stubborn inflation is the product of a damaged anchor of inflation expectations ("löskuð kjölfesta verðbólguvæntinga") combined with lingering wage-push and service-sector indexation. In this telling, the right policy is for the central bank to demonstrate firmness until expectations re-anchor, at which point inflation will follow.

It is worth being explicit about the relationship between that narrative and the reading in this piece, because the central bank's own publications contain most of the pieces of the housing-driven story — they just stop short of connecting them.

What the CBI says it thinks. In the governor's summary of Peningamál 2026/1 (4 February 2026), the Monetary Policy Committee held the policy rate at 7.25% (it hiked to 7.50% six weeks later, on 18 March) and wrote: "Svo virðist sem löskuð kjölfesta verðbólguvæntinga og kostnaðarhækkanir á undanförnum misserum hafi leitt til þess að verðbólga hefur verið þrálátari en ella" — roughly, "it appears that a damaged anchor of inflation expectations and cost increases in recent periods have led to inflation being more stubborn than it would otherwise have been." Fjármálastöðugleiki 2026/1 makes the same claim and adds an amplifier: "útbreidd verðtrygging þjónustusamninga og skortur á kjölfestu verðbólguvæntinga" — widespread indexation of service contracts together with lack of anchored expectations. Those two sentences, repeated near-verbatim across the bank's main publications, are the public thesis.

What the CBI's own Financial Stability Report says about the 2021–2022 housing price surge. In the same edition of FSR 2026/1, in the section on the housing market, the bank writes: "Hækkun íbúðaverðs árin 2021 og 2022 var að miklu leyti drifin áfram af lækkun vaxta og aðflutningi fólks, samhliða vexti í vinnuaflsfrekum atvinnugreinum" — "the rise in housing prices in 2021 and 2022 was largely driven by rate cuts, the inflow of people, and the growth of labor-intensive industries." That is the causal chain in this piece, stated in the central bank's own publication. Rate cuts plus migration plus labor-intensive sector growth. The FSR goes on to note that foreign citizens in Iceland are disproportionately renters, that slower migration would reduce housing demand, and that the rental market is sensitive to migration flows. The demand-side demographic mechanism is acknowledged.

What the governor said in Parliament on 21 October 2025. In testimony to the Alþingi's Committee on Economic Affairs and Commerce, Governor Ásgeir Jónsson said plainly: "Það er ákveðinn þröskuldur í kringum 4 prósentin... Verðbólga án húsnæðisliðar er um 2,5% en um 4% þegar húsnæðisliðurinn er tekinn með." — "There is a certain threshold around 4 percent... Inflation excluding the housing component is about 2.5%; it is about 4% when the housing component is included." That is the governor, in public testimony, saying that the entire gap between realized and target inflation is the housing component. He also cited the Grindavík eruption displacements as a shock to housing demand. That testimony is not cited in the subsequent Peningamál or the FSR.

What the FSR sees about the rate tool and housing supply. The FSR 2026/1 devotes several pages to the construction sector under rate pressure: over half of new builds are now selling below asking price, new-build inventory time in the capital area has reached roughly 18 months, Stage-2 ("higher-risk") construction loans are growing rapidly, and bank exposure to the sector has risen from about 11% of corporate loans in mid-2021 to roughly 19% by early 2026. The report describes these as elevated financial stability risks emerging from rate policy biting on the supply side. It does not describe them as evidence that the rate tool is asymmetric — that it cuts housing supply faster than it cuts housing demand — but that is what the pattern is.

What the FSR sees about household resilience and the verðtryggð refinancing. FSR 2026/1 reports that the indexed (verðtryggð) share of outstanding mortgages rose by more than 4 percentage points year-on-year to 65% by January 2026, while household debt-to-income ratios fell across every income decile and real wages rose 2.5% year-on-year. The bank interprets this as household resilience. The alternative reading — that households used indexation refinancing to escape the payment-shock channel the rate tool was supposed to work through, leaving consumption and rental bidding intact — is the other side of the same data. The FSR does not entertain it.

What this means. Taken together, the CBI's own publications contain: (a) the causal chain from rate cuts and labor migration to 2021–2022 housing price pressure, (b) the rental-market exposure of foreign citizens, (c) the rate tool biting hardest on the construction supply response, (d) the household indexation refinancing that insulated consumption from the rate channel, and (e) the governor's public testimony that the entire inflation gap is a housing gap. The reading offered in this piece assembles these into a single causal chain: demographic demand shock into a constrained housing stock, with rate policy asymmetrically cutting supply, and indexed contracts mechanically propagating shelter inflation into the rest of the CPI basket. The CBI's official narrative — the "löskuð kjölfesta" thesis — treats the same set of facts as structural amplifiers on top of a fundamentally expectational problem, and directs policy toward firmness of the rate instrument.

The honest summary is this: the Seðlabanki has the evidence for the housing-driven reading in its own publications. It just does not draw the line between the dots. Neither does the official narrative in Working Paper #100, which, using a Bayesian trend-cycle decomposition (UCSV-AR), labels a persistent low-frequency component of realized inflation as "damaged expectations anchor." That label is a choice, not a finding: the same trend is observationally equivalent to unresolved cost pass-through from a real-resource bottleneck, and the bank's econometric work does not — and cannot — distinguish between the two interpretations.

Where this piece and the CBI agree: service-contract indexation is a structural amplifier, construction under rate pressure is in real distress, and the migration-plus-labor-intensive-growth pattern of 2021–2022 was the driver of the housing surge. Where they disagree: whether the subsequent 2023–2026 stubbornness is an expectations problem the rate tool can eventually fix, or a real-resource problem that fiscal and planning policy has to fix and the rate tool is making worse. The disagreement matters because it changes what Iceland is supposed to do next. If it is an expectations problem, more firmness on rates is the prescription. If it is a real-resource problem, rates should come down and the building should start.

The data in this piece suggests it is a real-resource problem. The Seðlabanki's own publications contain most of the evidence for that reading. The central bank is the only party in the policy conversation that has not yet said so out loud.