# Warmth and unemployment reverse direction by country

> Do warmer cities have higher unemployment? Across 13,220 cities in 34 countries, the relationship reverses direction, making the pooled 0.06 misleading.

Published 2026-08-06 · Maksim Streltcov · https://immigrationdb.com/research/warmer-cities-unemployment/

Across 13,220 city rows, warmer places do not share one unemployment pattern. Among 28 countries with at least 20 observations, Spearman is negative in 14 and positive in 14, from -0.74 in Latvia (n=27) to +0.60 in Italy (n=812). The pooled +0.06 is not an average country effect; the useful answer is country-specific.

## What we measured
## What the data shows
## Where it breaks down
## What this means if you are moving
## Limits
## Method
## Sources

## Questions

**Do warmer cities have higher unemployment?**

Not consistently. Across 13,220 city rows, the pooled Spearman coefficient is +0.06, but the 28 reportable country coefficients split evenly: 14 positive and 14 negative. They range from -0.74 in Latvia to +0.60 in Italy, so the relationship must be read within each country.

**Why is the pooled unemployment coefficient misleading?**

The pooled +0.06 is not an average of country coefficients. It re-ranks cities after 34 national labour-market systems and different observation periods are merged, while the United States supplies 42.8% of all rows. It therefore reflects sample composition as well as within-country association.

**Where is the strongest positive relationship?**

Italy has the strongest positive country coefficient: +0.60 across 812 cities. Its Pearson coefficient is +0.57, so the direction is not created by the use of ranks. This remains a descriptive association within the Italian sample and does not show that warmer weather produces unemployment.

**Where is the strongest negative relationship?**

Latvia has the strongest negative coefficient at -0.74 across 27 cities, followed by Poland at -0.47 across 222. Latvia is only seven observations above the reporting floor, so its exact magnitude should be read more cautiously than Poland's larger-sample result.

**Does sunshine show the same labour-market pattern as temperature?**

Only partly. Full-sample temperature and sunshine coefficients share a sign in 15 of 20 countries. When temperature is recalculated on exactly the sunshine-complete rows, agreement falls to 13 of 20. Italy remains aligned and the United Kingdom remains opposite, so sunshine is a robustness check rather than confirmation.