The change to Statec’s methodology has altered the number of people considered to be living in poverty. These indicators should be treated with great caution… as they do not cover 100% of households. (Photo: Shutterstock)

The change to Statec’s methodology has altered the number of people considered to be living in poverty. These indicators should be treated with great caution… as they do not cover 100% of households. (Photo: Shutterstock)

The at-risk-of-poverty rate fell from 18.1% in 2024 to 14,8% in 2025 in Luxembourg. However, this decline does not reflect a corresponding increase in the population’s wealth. A study by Statec shows that it is mainly due to a new method of measuring income.

Depending on the method used, nearly 38,400 people fall below or above the poverty line. Yet they do not actually become any richer or any poorer. It is the income that the statistician assigns to them, and the threshold against which it is compared, that change.

Why has the poverty rate fallen to 14,8%?

Statec has changed the way it measures income in the European Survey on Income and Living Conditions (EU-SILC). Until 2024, the calculation relied more heavily on the responses provided by households in the questionnaire. Since 2025, these self-reports have been supplemented on a large scale by administrative data from the General Inspectorate of Social Security (IGSS). This enables Statec to identify wages, pensions, benefits or replacement income that were not included in the questionnaire.

This is not income that has been concealed from the tax authorities or social security. It is accurately recorded in the administrative records. The household may have omitted it from the survey, declared it under a different heading, or reported it for a different period or amount.

Once this income is taken into account, some households appear to be better off than their survey responses alone would suggest. Some of them fall back above the at-risk-of-poverty threshold. The official rate thus falls from 18.1% in 2024 to 14,8% in 2025.

Did these households actually earn more money?

Not necessarily. A household may have received exactly the same income, but this is now better recorded by Statec. The 3.3-point fall cannot therefore be interpreted as evidence that the population’s material circumstances have improved by the same proportion. The change lies primarily in the measurement method.

Statec states this explicitly: the inclusion of administrative data creates a break in the statistical series and prevents a direct comparison between the 2024 and 2025 rounds.

Why does the study also put the figure at 17.7%?

The researchers sought to measure the effect of the new method. They started with the official figure of 14,8%, then excluded certain income identified in administrative records but apparently missing from tax returns. However, they retained amounts that may simply have been declared under the wrong heading. For example, a benefit payment might have been classified as wages rather than as replacement income.

Following these adjustments, the rate rises to 17,7%. The 2.9-point difference from the official figure of 14,8% gives an indication of the impact of the new method of measuring income.

Is the actual rate therefore 17,7%?

No. Statec does not present this result as a new official rate. It is a simulation designed to test how the result is affected by the method used. This simulation does not fully replicate the old system. It cannot recreate all the validations, corrections and imputations previously applied to the questionnaires. Nor does it account for differences in time periods, confusion between gross and net amounts, or income that is not recorded in administrative records.

The figure of 17,7% therefore answers the following question: what would the 2025 result be if certain administrative revenues, which appear to be missing from the questionnaires, were excluded? It does not provide a definitive answer to this question: what would the official rate for 2025 have been if the old method had been retained?

So, which figure should we go with in the end?

To describe the situation in 2025, the official rate of 14.8 per cent is the best estimate available. It is based on a more comprehensive understanding of the income actually paid out. To measure the change between 2024 and 2025, there are no perfectly comparable figures. The figures of 18,1% for 2024 and 14,8% for 2025 were obtained using two different methods.

The study allows us only to conclude that the bulk of the reported decline is linked to the way income is measured. Based on the 17,7% simulation, the gap with 2024 would be just 0.4 percentage points, compared with 3.3 percentage points in the official series.

Does the new weighting also explain the decline?

Very little. Statec has also changed the way respondents are weighted to ensure they are representative of the population as a whole. Using the new income figures but the old weighting, the rate would have been 14,4%, compared with 14,8% using the new weighting. In the simulation, it rises from 17,4% under the old weighting to 17,7% under the new one.

The effect of the weighting therefore amounts to around 0.3 to 0.4 points. The main discrepancy does indeed stem from the income attributed to households.

How many people change category?

The simulation produces two opposing effects. By removing certain administrative revenues, 29,199 people – or 4,3% of the population – move from the non-poor category into the poor category. At the same time, 9,165 people, or 1,4% of the population, move from the “poor” category to the “non-poor” category.

In total, 38,364 people therefore change their statistical classification depending on the method used. The number of people falling into poverty is three times higher than the number moving out of poverty.

Why do some people escape poverty even though their income is falling?

Because the poverty line is also shifting. The rate used in Luxembourg measures the proportion of the population with less than 60% of the median income. When income is removed from the simulation, the median income falls. The poverty line falls with it.

Some people still have an income lower than that used in the official calculation, yet find themselves above the new threshold. Their exit from poverty is therefore purely statistical. It does not reflect any improvement in their living conditions.

Are these 38,364 people all those whose income has been adjusted?

No. They apply only to people whose difference in measurement is large enough to alter their ranking in relation to the threshold. Other people may have had their statistical income adjusted without crossing this threshold. Conversely, several people affected may belong to the same household.

The study does not specify how many households saw their income actually adjusted as a result of the integration of administrative data. It states that 84,6% of eligible individuals could be linked to this data, and 89,8% of those registered with the Luxembourg social security system. However, being linked to a database does not necessarily mean that the declared income has been adjusted.

Is the new rate absolutely accurate?

No. It is probably more accurate, but it still has its limitations. The IGSS data do not cover all forms of income. Some income from assets, capital or transfers is still based on self-declaration. Not all respondents are registered in Luxembourg, and not all eligible individuals could be matched to the records.

Furthermore, the at-risk-of-poverty rate remains a relative indicator. It measures income in relation to the median income. It does not directly reflect wealth, essential expenditure, the cost of housing or material deprivation.

The new indicator is therefore likely to be more accurate. However, as the figures for previous years have not been recalculated using this indicator, Luxembourg is not yet able to accurately measure the actual trend in poverty.