The EU’s income-poverty risk fell from 17.4% in 2015 to 16.2% in 2023, according to a new study that also examined housing, deprivation and digital skills. Bulgaria, Greece, Latvia and Romania nevertheless formed its most vulnerable country group in the final period.
The peer-reviewed research, published in PLOS One on September 30, compares all 27 EU member states using Eurostat data from 2015 to 2023. Its findings describe that historical period, rather than establishing that poverty is continuing to fall in 2026.
Silvia Megyesiova and Cecilia Olexova, of the Bratislava University of Economics and Business in Slovakia, found that national economic output alone could not explain the different combinations of hardship across the bloc.
For businesses, those differences matter when assessing customers’ budgets, planning online services and deciding where training may be needed. The authors discuss these commercial implications, although their study did not measure company sales or customer payment behavior.
Romania improved but remained vulnerable
The EU’s at-risk-of-poverty rate measures the share of people whose household disposable income, adjusted for household size and composition, is below 60% of the national median after social transfers such as benefits.
It measures relative income disadvantage within each country. It does not apply an identical minimum living standard throughout Europe. The income figures also refer to an earlier period: the 2023 survey generally measured income received in 2022.
Romania recorded the largest improvement on this measure, falling from 25.4% in 2015 to 21.1% in 2023, a reduction of 4.3 percentage points. Luxembourg’s rate increased from 15.3% to 18.8% over the same period.
A separate measure showed a larger EU-wide improvement. Severe material and social deprivation fell from 9.7% to 6.8%.
Severe material and social deprivation means being unable to afford at least seven of 13 specified items or activities. These include keeping a home adequately warm, meeting an unexpected expense and replacing worn-out clothes.
Bulgaria’s deprivation rate dropped from 36.8% to 18.0%, while Romania’s fell from 34.0% to 19.8%. Both countries made substantial progress while remaining above the EU rate.
Employment did not eliminate income-poverty risk. Romania’s in-work poverty rate was 15.3% in 2023, compared with 2.8% in Finland and 3.1% in Czechia. This indicator concerns workers’ household incomes, rather than their individual wages alone.
Housing and digital skills change the comparison
The researchers initially examined 10 indicators, then retained six for the final analysis: income poverty, material and social deprivation, in-work poverty, severe housing deprivation, economic output per person adjusted for price differences, and basic digital skills.
They grouped countries with similar profiles across three periods: 2015 to 2019, the pandemic year of 2020, and 2021 to 2023. Some indicators, including digital skills, used 2023 observations in the final period because of data availability.
The final analysis produced seven groups. Bulgaria, Greece, Latvia and Romania combined relatively low economic output per person with the highest group levels of income poverty, material and social deprivation, and severe housing deprivation. Czechia and Finland shared a group with particularly favorable poverty and digital-skills indicators.
Housing affordability remained a separate concern. The housing-cost overburden indicator was excluded from the final statistical grouping because it did not vary sufficiently in line with the retained measures. The authors argue that this may reflect a distinct form of vulnerability deserving separate attention.
The official housing-cost overburden measure identifies people in households spending more than 40% of disposable income on housing, after accounting for housing allowances. It helps show how housing expenses can squeeze the money available for other purchases.
Basic digital skills were associated with lower poverty indicators across countries. That relationship does not establish that digital training caused poverty to decline, or that someone without those skills is necessarily poor.
Customer budgets and online access matter to businesses
The authors argue that constrained household budgets can mean more price-sensitive customers, less predictable demand and greater risk of payment difficulties. They also suggest that lower digital readiness can increase recruitment and internal training requirements.
These are interpretations of the country profiles. The research did not test whether businesses in the vulnerable group experienced higher bad debts or staff turnover.
For a retailer or service provider, the practical question is whether its prices, payment terms and customer support fit the market it serves. A national GDP figure cannot answer that question by itself.
Our earlier coverage of digital literacy and network access examines a related distinction: connecting people to the internet does not automatically equip them to use online services.
The national averages also leave important gaps. Household composition was not fully incorporated into this study, and the country groups cannot describe every community or customer within them.
Newer figures provide a check on the historical trend. Eurostat’s May 2026 release put EU income-poverty risk at 16.3% in the 2025 survey, based on 2024 income. It described the rate as broadly stable, reinforcing the need to distinguish the study’s earlier progress from current conditions.
