This blog is based on an article in the Journal of International and Comparative Social Policy by Ilari Ilmakunnas. Click here to access the article.
Despite the shared understanding that it is useful to analyse poverty by means of different measures, one measure is more commonly used than others. In the EU, the at-risk-of-poverty threshold is typically set at 60% of national median disposable equivalent income. By analysing poverty trends in Europe in the mid-2010s, I show that linking the threshold with changes in average income has an impact on the picture of age-specific changes in poverty rates.
The alleviation of poverty is amongst the main goals of welfare states. For this reason, it is important to follow the development of poverty rates. Yet, the choice of poverty measure can affect what poverty trends look like. Linking the threshold with changes in average income can affect poverty rates for some population subgroups more than others. Some groups for example are less likely to experience year-to-year variation in income. Additionally, individuals in some groups have, on average, income that is closer to the at-risk-of-poverty threshold.
In this article, I explored whether the magnitude and direction of changes in at-risk-of-poverty rates vary between age groups. Age groups are an interesting case since the old-age population is less likely to experience changes in their level of income as they are less connected with the labour market. Additionally, the level of income is often close to the at-risk-of-poverty threshold in this age group. The aim of the study was to identify any general tendencies regarding the changes in the age-specific poverty rates. The study focused on short-term changes in the poverty rates in 30 European countries in the mid-2010s (changes in the periods 2013–2018, 2015–2018 and 2017–2018). The age groups analysed were children (0–17), the working-age population (18–64) and the older population (65+).
The results showed that on average, older individuals experienced greater changes in at-risk-of-poverty rates than children or working-age individuals. This is somewhat paradoxical, considering that it is also shown that the older population experienced, on average, smaller changes in income. Changes in old-age at-risk-of-poverty rates typically moved in the opposite direction compared to changes in at-risk-of-poverty rates amongst the working-age population. In other words, when the old-age at-risk-of-poverty rates increased, the rates for the working-age population typically decreased.
So why do these result matter?
First, the opposite direction of changes in old-age and working-age at-risk-of-poverty rates has the effect of suppressing changes in the total population rates. This means that changes in the total population at-risk-of-poverty rate may hide improvements in the risk of poverty for a large share of population. Second, the study also shows that the association between changes in at-risk-of-poverty rates and changes in poverty thresholds, income and employment significantly differed between old-age population and other two age groups. Increases in the employment rate and income were associated with decreases in child and working-age poverty risk. However, even an opposite pattern was found for the older population.
My study highlights one useful practice when analysing trends in income poverty. That is the use of a so-called fixed (or anchored) poverty threshold. In the study, a fixed threshold referred to a at-risk-of-poverty threshold that was adjusted for inflation for subsequent years in the period of interest. Thus, changes in median income did not affect the level of the threshold.
Unlike children or the working-age population, changes in at-risk-of-poverty rates and poverty rates calculated using the fixed threshold were negatively correlated amongst the old-age population. In other words, the two measures typically gave a different picture regarding the development of the risk of poverty among the older population. When a fixed poverty threshold was used, a negative association was found in all age groups between employment and income changes and changes in poverty rates. This means that unless a fixed poverty threshold is used, the positive income (or employment) development does not necessarily correspond with the development of old-age income poverty rates.
The study provided new information on how the choice of the poverty measure affects the picture of poverty trends for different age groups. The results show that the most widely used measure of the risk of poverty is associated with some general tendencies regarding both the magnitude and direction of changes in age-specific poverty rates. Thus it is important to use, for instance, so-called fixed poverty thresholds alongside the at-risk-of-poverty threshold. However, the use of monetary poverty measures that are more closely aligned with definitions of poverty and are more likely to follow countries’ economic performance, such as measures based on reference budgets, would be a larger improvement.
About the author
Ilari Ilmakunnas is a Senior Researcher at the Finnish Centre for Pensions.