What is the middle class, and who counts as part of it?

In 1982 Statistics Sweden (SCB) published an official classification of everyone in work in Sweden.^{7, 14} It was called the socio-economic classification, SEI for short, and in the agency's aggregated version it sorted people in work into twelve groups by occupation.^{7, 13} The criterion was how many years of education after compulsory school the occupation normally required, not what the individual person had studied.^{13, 14}

The groups were unskilled workers and skilled workers, both split into goods production and service production, lower white-collar employees at two levels, intermediate white-collar employees, higher white-collar employees, senior positions, self-employed professionals with graduate occupations, business owners and farmers.¹³ Employees were divided into blue-collar workers and white-collar employees according to which family of trade unions the occupation normally belonged to.^{13, 14}

In the middle of the classification sat group 46, intermediate white-collar employees. Three years after compulsory school were required, six years were not.¹³ That is as close to an official Swedish middle class as the statistics ever came, and the group was defined neither by income nor by self-image.

Statistics Sweden phased out SEI from 2026. The agency's own reasoning is that the classification was already out of date in the early 2000s, that the changes to the education system after the 1980s were never reflected in it, and that several occupations had come to require higher qualifications.⁷ No new Swedish classification replaced it. So Sweden had an official sorting of the population into layers for forty-four years, and it never contained the words middle class.^{7, 13}

This article goes through how the term is actually measured, what answers the different measures give for the same population and why the answers differ by a factor of ten.

Three rulers

Economists measure the middle class in three fundamentally different ways, and they answer different questions.

The first is an absolute band. You set a floor and a ceiling in dollars per day, adjusted for purchasing power, and count how many people fall in between. The measure is built for global comparisons, where the question is what a person can afford. The World Bank's platform reports the share below any given dollar line for most countries in the world.¹⁰

The second is a relative band around the median in a country. The OECD uses 75% to 200% of median income and splits the band into lower, middle and upper middle class.^{9, 16} Pew Research Center, a non-partisan American research institute, uses two thirds to 200%.^{22, 24} The measure answers where someone stands compared with their fellow citizens.

The third is an occupational or class scheme of the same kind as SEI. It answers what someone works with. SEI was phased out from 2026, but occupations can still be described with the Swedish occupational classification SSYK, and ESeG is a European socio-economic classification.⁷ Neither of them is the same class scheme as SEI.

Five choices made before the threshold

Before you even get to the percentages, five decisions have already been made, and each one moves the answer.

Which concept of income is used is the first. Statistics Sweden counts disposable income, after taxes and benefits, and publishes two versions: one with realised capital gains and one without. The difference is not negligible. With capital gains, 13.1% of the population had a low economic standard in 2024 and the Gini coefficient was 0.314.² Without capital gains the figures were 12.8% and 0.288.²

Which equivalence scale is used is the second. The household's income is divided by a number that depends on how many people have to live on it. Statistics Sweden counts a person living alone as 1.00, a cohabiting couple as 1.51, each additional adult as 0.60, the first child under 20 as 0.52 and each additional child as 0.42.^{5, 6} The OECD instead uses the square root of the household size.^{16, 23} The same family gets a different income per consumption unit depending on which scale is chosen.

The three remaining choices are which year is meant, which population is counted and whether the unit is the household or the individual. The year is especially treacherous in this subject, because the sources are out of step: Swedish register data refer to 2024, the World Bank's Swedish figures refer to income year 2023, and Eurostat's Swedish values labelled 2024 also refer to income year 2023.^{4, 10}

Same population, three answers

You can lay all three rulers on the same population and see what happens. The data are Statistics Sweden's cut-off values for each percentile of the Swedish income distribution in 2024, plus the World Bank's figures for Sweden.^{1, 10}

With an absolute band of 8.30 to 25.00 dollars per day, about 7.3% of Swedes end up in the middle class, based on the World Bank's data for 2023.¹⁰ Almost everyone else is above the ceiling. Our own calculations on Statistics Sweden's data for 2024 give about 66.1% with the OECD's lines of 75% to 200% of the median and about 74.0% with Pew's lines of two thirds to 200%.^{1, 9, 22} The two relative figures are interpolations on Statistics Sweden's consumption unit scale, not the OECD's or Pew's own estimates with their household scales.

All three figures are correct. The difference between 7% and 66% depends on which question the measure asks. It says nothing about how Swedes are doing.

One caveat belongs to the absolute figure. The World Bank publishes no official line for a global middle class. The floor of 8.30 dollars is the bank's own poverty line for upper middle income countries, while the ceiling of 25.00 dollars is our choice.¹⁹ A different ceiling gives a different figure.

What it means in kronor

The median and the thresholds per month

The median economic standard in Sweden in 2024 was 327,255 kronor a year.⁵ For a person living alone, who counts as one consumption unit, that is 27,271 kronor a month after taxes and benefits.⁵

The thresholds then fall here. Statistics Sweden's line for a low economic standard, 60% of the median, is at 16,363 kronor a month.⁵ The OECD's floor for the middle class, 75%, is at 20,453 kronor.^{5, 9} The ceiling that both the OECD and Statistics Sweden use, 200%, is at 54,543 kronor. Above that line, Statistics Sweden counts you as having a high economic standard, and 7.2% of the population were there in 2024.^{2, 3}

Below the line, and counted for a household

Being below 60% of the median is called a low economic standard in Sweden, and Statistics Sweden itself writes that it is not necessarily the same as being poor.^{3, 4} The EU calls the same measure at risk of poverty.⁴ The Ministry of Finance reports both measures for the same year in its report on income distribution. With an absolute basket measure, 3.9% of the population had a low income standard in 2024, about 400,000 people, while the relative measure gave 13.1%, just over 1.3 million.^{8, 15} Same population, same year, more than three times as many people.

For a household, the amounts are converted with the consumption unit scale. A cohabiting couple with two children under 20 equals 2.45 consumption units: 1.51 for the couple plus 0.52 for the first child and 0.42 for the second.⁶ The OECD's band for that household therefore runs from 50,111 to 133,629 kronor a month in disposable income for the whole family.^{5, 6} Same band, same year, completely different figures.

Relative measures measure position, not level

The band is measured against the home country's median

Because the relative band is measured against each country's own median, it compares positions within each country's income distribution, not living standards between countries. Eurostat reports median income in purchasing power standards, a unit adjusted for price levels, and there the problem shows straight away.

For 2024 the median was 23,693 purchasing power units in Sweden, 13,023 in Romania and 37,781 in Luxembourg.^{11, 21} The ceiling for the Romanian middle class, 200% of the Romanian median, is therefore 26,046. That is higher than Sweden's median. The floor for the Luxembourg middle class, 75% of the Luxembourg median, is 28,336, which is above the Romanian ceiling.^{11, 21} Nobody has changed income in that example. Only country.

The result is that international comparisons of the size of the middle class do not measure how many people are well off. The OECD reports Iceland at 71.9% for 2010 and South Africa at 33.5% for 2012, and that figure describes how compressed the income distribution is.^{9, 17} A country with evenly spread small incomes gets a large middle class.

For the same reason, the term middle income country does not mean that the country's inhabitants are middle class. The World Bank's country classification groups countries by gross national income per inhabitant, and it says nothing about how incomes are spread within the country.²⁰

The same lines counted in dollars

The difference between the two kinds of measure is clearest if you convert the Swedish line into dollars. Statistics Sweden's line for a low economic standard, 60% of the median, equals just over 33 dollars per day, adjusted for purchasing power.¹⁰ The World Bank's line for upper middle income countries is 8.30 dollars, and the bank's international poverty line for low income countries is 3.00 dollars.¹⁹ So the Swedish poverty line sits four times above the level that counts as the poverty line in large parts of the world. A relative measure measures the distance to the neighbours, not access to food and a roof.

With the absolute measure, the world looks like this according to the World Bank's estimates for 2024. The series are based on household surveys with estimates for the years in between. Of the world's population of 8,141,808,945 people, 10.40% lived on less than 3.00 dollars per day, 46.07% on less than 8.30 dollars and 77.18% on less than 25.00 dollars.¹⁰ The band between 8.30 and 25.00 dollars therefore held about 31.1% of the world's population, roughly 2.53 billion people.¹⁰ In Sweden the same band held about 7.3%.

The middle is a place, not a group of people

The strongest Swedish figure on the nature of the middle class comes from the Ministry of Finance's analysis of mobility. Of those who in 2015 were in income group 5 or 6, in the middle of the distribution, almost half had moved to another part of the distribution by the following year.^{8, 15} Eight years later, in 2024, more than two thirds had left the middle layer altogether, about a third downwards and just over a third upwards.¹⁵

Mobility is greatest right in the middle, and the reason is simple. Incomes are packed closely there, so it takes only a small change to switch group. At the far ends of the distribution the distances are larger: income group 10 had 3.2 times the economic standard of the median earner and eight times that of income group 1, which was at about 40% of the median.¹⁵

So the middle class as a statistical category is stable. The middle class as a collection of named people is not.

What the figures do not say

The share with a low economic standard does not move in one clear direction. It fell from 14.7% in 2021 to 13.1% in 2024, which according to the Ministry of Finance is the lowest level since 2007.¹⁵ The real economic standard of the median earner rose by 16% between 2011 and 2024.¹⁵

The OECD has measured how the middle income class changed between the mid 1980s and the mid 2010s in 17 countries.¹⁶ Sweden had the largest fall, minus 7.43 percentage points, of which 4.15 went to the lower income class and 3.28 to the higher.¹⁸ The OECD average fell from 63.73% to 61.18%.¹⁸ Two things belong with that figure: Sweden's middle income class was very large to begin with and, according to the same source, is still larger than the OECD average, and the OECD's Swedish figures are based on EU-SILC and refer to 2013 and 2015.^{16, 17, 18} So they cannot be set against the calculation on Swedish register data for 2024 as a time series, because the data source, the equivalence scale and the year all differ.

What you can take away is simple. The term middle class has no answer until someone has said which ruler is used, which year is meant and whether the measure is absolute or relative. When someone gives a figure without saying that, the question has not been answered.

What to take away

The same population gives about 7.3% middle class with an absolute band in dollars, about 66.1% with the OECD's relative band and about 74.0% with Pew's.^{1, 9, 10, 22} The median economic standard was 327,255 kronor in 2024, or 27,271 kronor a month for a person living alone.⁵ The structural point is that the middle is a place and not a group of people: of those in the middle of the distribution in 2015, more than two thirds had left the middle layer by 2024.¹⁵ Mobility is greatest right in the middle, because incomes are packed closely there.¹⁵ Follow the median and the percentile cut-offs in Statistics Sweden's income distribution statistics,^{1, 5} the share with a low economic standard in the Ministry of Finance's report on income distribution¹⁵ and the World Bank's dollar lines.¹⁹

The next episode is S5:E5 Emerging Capital Markets, about what separates a large capital market from a small one.

Education, not advice.