Loading school data…

Does a more expensive school produce better results?

The relationship between municipal school spending per pupil and actual school results.

Use this for

Get the main pattern quickly, find a local angle and jump straight to the charts that carry the story.

Data & format

Statistics story · 3 visualisations · copy reviewed 18 October 2026 · Skolverket, Kolada and SCB.

Cost per compulsory school pupil differs about twofold between Swedish municipalities, from around SEK 120,000 to 238,000 (Kolada 2025). Does higher cost go together with better results? No positive relationship appears. Municipalities with higher cost per pupil have slightly lower merit scores on average, but the pattern almost disappears once municipality size or income level is taken into account.

Cost per pupil vs merit score

The first chart plots cost per pupil against merit score (r = −0.33 for 290 municipalities). If higher cost went together with higher results, the dots would form an upward slope. They do not.

Municipalities such as Danderyd, Lidingö, Lomma and Nacka have low costs and high merit scores, while several small municipalities with high costs are below the median. The colour shows teacher certification. Cost per pupil is higher in small, sparsely populated municipalities, and it is also affected by school size and pupils' needs. A high cost per pupil is therefore not the same as more teaching per pupil.

No positive link between cost and results. Colour shows teacher certification.

Open Cost vs Results fullscreen

Socioeconomic background — the hidden variable

A large part of the variation in school results co-varies with municipalities' socioeconomic profiles. Median income and education level co-vary considerably more strongly with Kolada's merit score, the measure in the first chart (r ≈ 0.70 and 0.75), than cost per pupil does. Municipalities with high median income and many residents with post-secondary education have higher results on average, which makes it difficult to isolate what school funding means.

The chart opens on the share of residents with a foreign background. Choose median income or post-secondary education in the menu to see the stronger correlations; the colour shows the independent-school share.

Each bubble a municipality. Size = number of schools. Colour = independent school share. Choose the measure in the menu.

Open Socioeconomic correlation fullscreen

SALSA: results compared with expectations

The SALSA model adjusts for parts of pupils' backgrounds and shows which schools and municipalities have results above or below expectations. Green municipalities on the map have results above what the model predicts, red below. The municipal value is an unweighted mean of school deviations, and every municipality counts equally, so most municipalities, especially the small ones, fall below zero.

Municipalities with results above expectations exist at every cost level, and cost per pupil does not co-vary with the SALSA deviation (r ≈ 0). The map does not show why: it cannot separate how money is used from other differences the model does not measure.

Green: results above expectations. Red: below. Adjusted for the background factors included in the SALSA model.

Open SALSA map by municipality fullscreen

Higher cost per pupil does not go together with higher results. The ten municipalities with the lowest cost all have merit scores above the median, and the municipalities with the highest cost are mainly small and sparsely populated. Socioeconomic background co-varies considerably more strongly with results than cost level does.

This is a comparison between municipalities at one point in time. It does not show what raising or lowering school funding would lead to in an individual municipality.

What the data does not show

The copy was reviewed on against cost per pupil, merit scores and teacher certification from Kolada 2025, median income for 2024, merit scores in chart 20 mainly for the 2025/26 school year and SALSA for 2024/25. The charts load the latest published data and may show a newer period than the copy.