Real wages continue to climb despite weaker wage growth, albeit much more slowly than before. Headline inflation measured 5.3% in July, and real wages have therefore risen by 0.4% in the past twelve months. In comparison, real wage growth measured 0.7% in 2024 and 3.7% in 2025. In our macroeconomic forecast, we assume that wage growth will average 6.2% this year, 5.7% in 2027, and 4.9% in 2028, and that real wages will rise by just over a percentage point per year throughout the forecast horizon.

Growth in immigration loses pace

According to SI, the number of foreign nationals with a legal address in Iceland increased by 3.1% YoY in Q2/2026, to 71,130, or 17.9% of the country’s total population, at the end of the quarter. Growth in the number of foreign nationals has slowed almost unimpeded since the start of 2023, when it measured 20.1%, and is now at its weakest since the pandemic. During the pre-pandemic upswing, it peaked at 24.5% at year-end 2017.

Figures from Registers Iceland show a more dramatic shift: 83,719 foreign nationals had a registered address in Iceland on 1 August, a decline of 220, or 0.3%, since 1 December 2025.

Growth in the immigrant population has long correlated with the state of the labour market, as foreigners are less likely to move to Iceland when fewer jobs are available. Slower immigration is therefore one of many signs of a cooling labour market. For example, registered unemployment measured 4.0% nationwide in July, but underlying that figure is a wide gap by nationality: 7.9% of foreign nationals were unemployed, as opposed to 2.5% of Icelandic nationals. During the first seven months of the year, the jobless rate averaged 9.3% among foreign nationals and 2.7% among Icelanders, and unemployment among foreigners has increased YoY each month since autumn 2023.

As a consequence, fewer people migrate to Iceland, and the labour force is growing more slowly than before, at a time when a slack has developed in the labour market. These two factors pull against one another: a smaller labour supply curtails the slack in the market, while a smaller population dampens aggregate demand.

The two measures of unemployment

The two measures of unemployment have long told the same story, although they do not measure exactly the same thing. SI’s labour force survey (LFS) is a random sample survey, and the results fluctuate more from month to month than the Directorate of Labour’s (DoL) registered unemployment rate, which is based on the number of jobless persons registered with the Directorate. The two are defined differently as well. In order to be included in the DoL’s unemployment register, individuals must satisfy set requirements to be eligible for benefits. On the other hand, the LFS captures all those who were out of work during the reference week, have been looking for work in the past four weeks, and could begin work within two weeks, irrespective of whether they are entitled to benefits or not. Furthermore, the DoL register includes part-time unemployment, whereas the LFS does not.

When the two measures diverge, there is usually an explanation for it. During the pandemic, registered unemployment was far above the LFS-based jobless rate, with the difference between the two peaking at nearly 10 percentage points in spring 2020. This was due largely to the part-time unemployment benefits programme, which 37,000 employees had utilised by May of that year. Under that programme, workers maintained an employment relationship with their employer, but they reduced their working hours and received benefits to compensate for the reduction. As a result, they were considered employed for LFS purposes even though they were on the DoL’s unemployment register.

At the beginning of 2026, the two measures parted ways again, but in the opposite direction. The LFS began to show far higher unemployment than DoL figures did, with the difference peaking in May. That said, the greatest incongruity between measures tends to occur in May, when the LFS typically peaks, while DoL figures peak in mid-winter. In recent years, the widest gap between the two has often been in May. What makes 2026 unusual is not that the two measures should diverge, but that they should do so for such a protracted period of time. For five consecutive months – from January through May – the LFS-based unemployment rate was well in excess of registered unemployment figures; indeed, it is the first such episode in data going back to 2003 that the two measures of joblessness have diverged so sharply and for such a long interval. While this pattern has been seen before – in 2006, 2013, and 2014, for instance – on those occasions it did not persist this long.

In June 2026, the gap narrowed and the two metrics correlated more closely, with the LFS giving a jobless rate of 3.9% and the DoL 4.0%. This is due to seasonal factors: students get jobs over the summer, but they are considered unemployed according to the LFS if they are seeking work, even though they are generally not entitled to unemployment benefits.

The deviation in the first half of the year is harder to explain. According to information from SI, there has been no change in methodology, sampling, or data collection for the LFS, so the deviation is most likely the result of several factors combined. It should be borne in mind that the LFS is a random sample survey, and the confidence bands in any given month are wide. In January, for instance, SI estimated the number of unemployed persons at 19,700, with a confidence interval of 4,900, so each monthly measurement tells only part of the story.

Wage agreement review ahead – in a changed labour market

A large share of wage agreements in the Icelandic labour market are up for review in September, owing to the failure of the premises underlying the contracts. The clause laying down these premises stipulates that contracts will be subject to review unless twelve-month inflation either measures below 4.7% in August 2026 or averages 4.4% YoY over the six-month period from March through August 2026. Twelve-month inflation has topped 5% each month in 2026 to date, and our forecast assumes that it will measure 5.6% in August, so it is safe to predict that the review clause will indeed be triggered. SI will publish inflation figures for the month on 27 August.

The long-term wage agreements that shape wage developments were signed in March 2024 and do not expire until the end of January 2028. At the time they were signed, Q1/2024 unemployment measured 4.1% according to the LFS and 3.8% according to the DoL, and the wage index had risen 7.4% YoY in a tight labour market featuring significant wage drift. Now, however, annual wage growth has eased to 5.7%, while unemployment has climbed steadily. This tapering stems in large part from the public sector labour market. When public sector wage agreements were finalised in the wake of the so-called stability agreements, the public sector wage index spiked, with YoY pay rises peaking at 13.1% in spring 2025, far outpacing those in the private sector. That spike has now reversed, and the two markets are again moving broadly at the same pace.

In H1/2026, unemployment according to the LFS overtook annual wage growth, but unlike in most if not all neighbouring countries, wage rises in Iceland typically outpace the rise in unemployment. This indicates how hot the domestic labour market has been running for quite some time. But it is wise to avoid overinterpreting the intersections between the two variables, as one of them measures YoY changes and the other indicates a share of the labour force. Even so, the timing this year is interesting. Such intersections generally occur in spring when LFS measures reach their peak, but this year it also happened in January and February. This is rarer than it might appear. Since 2015, it has only happened once during the winter – in February 2020 – after a steady rise in unemployment. As this illustrates, the labour market tension prevailing when contracts were signed in March 2024 has disappeared, and the contracts will be reviewed in an environment utterly different from the one in which they were negotiated.

At the same time, the total number of wage-earners is virtually unchanged, although different segments of the market show differing trends. In the private sector, job numbers have fallen steadily in the recent term. They were down 1.9% YoY in May, whereas public sector job numbers were up 4.9% at the same time. Public sector hiring is determined by budgetary allocations rather than demand in the marketplace, so the cooling in the labour market is limited to one half of it. Added to this is the above-mentioned slowdown in labour immigration. Thus the slack in the labour market shows not only in a drop in job numbers, but also in the reduced number of foreign nationals who come to Iceland to work. This pushes overall unemployment lower than it would be otherwise and is actually yet another manifestation of the flexibility of the Icelandic labour market.

Consensus more likely than contract cancellation

Inflation will doubtless trigger the contract review clause, but the slack in the labour market will affect the review process. Nevertheless, we think it most likely that the social partners will find some common ground and avoid terminating the wage agreements, and that they will negotiate a modest additional pay hike for 2027, either with a 3.5% general pay increase or a higher floor for those with the lowest wages (in the current contracts it is 23,750 kr.), or a combination of the two. In our macroeconomic forecast from June, we assume that pay rises in 2027 will be slightly larger than is provided for in the current wage agreements, but considerably below the ten-year average of 7.6%.