The Indian economist and philosopher Amartya Sen has long argued that human lives are battered and diminished in all kinds of ways. Measures of multidimensional poverty seek to describe these forms of deprivation directly within the various dimensions of well-being (e.g., health, education). Thereby, multidimensional poverty measures complement conventional monetary poverty measures. Moreover, multidimensional measures also make it possible to analyse poverty understood as multiple deprivation. As different combinations of forms of deprivation may constitute poverty, examining multidimensional poverty can help to unveil the many faces of poverty.

Launching an official poverty measure would require a public debate involving various stakeholders, experts, and policymakers to agree upon broadly accepted value judgements (which are inherent to any such exercise). Nonetheless, a candidate measure may be computed with data from the European Union Statistics on Income and Living Conditions (EU-SILC).

The deprivation indicators are inspired by the EU framework for Social Inclusion and previous research. Technically, the analysed measure relies on the approach proposed by Alkire and Foster and identifies individual adults as poor. 

According to the proposed measure, a person is considered as poor if they experience more than a third of the maximum possible forms of deprivations. To this end, forms of deprivation are added up using the normative weights (also shown in the table). For example, a person who reports bad overall health, problematic housing conditions (e.g., a leaking roof) and unemployment would score  1/10 + 1/10 + 1/5 = 2/5 > 1/3 and would therefore be classified as poor.

For Spain the incidence of multidimensional poverty amounts to some 15% and barely changed over the period of observation (2016-2020). Other European countries with initially higher incidences (e.g., Portugal and Greece) or initially similar incidences (e.g., Poland and Hungary), however, experienced incidence reductions by up to 5 percentage points. 

One way to explore overlapping forms of deprivation is to look at the proportion of the population which is both deprived in a particular indicator and (multidimensionally) poor. Below we see this quantity and the proportion of the population that is deprived in a particular indicator (while not necessarily poor) for Spain in 2019.

Two observations stand out. First, the proportions of people who are both poor and deprived are substantially smaller than the proportions of people who are deprived (but not necessarily poor) for every indicator, which shows that of those people who are deprived in a particular indicator, only a subset suffers from multiple deprivation (i.e., is multidimensionally poor). Put differently, there is no single form of deprivation that mechanically implies poverty (i.e., multiple deprivation). Second, the proportion of people who are both poor and deprived in a particular indicator  varies between 2 and 10% whereas the poverty incidence is about 15% (indicated by the dashed line), which implies that there is no form of deprivation experienced by all poor people. The evidence, therefore, suggests that the forms of deprivations may cluster in different ways.

While multidimensional poverty measures are popular in both academia and practice, very little is known about related dynamics at the level of households or individuals. One way to explore these dynamics on a year-to-year basis is to compare deprivation entry and exit probabilities for (multidimensionally) poor and non-poor people. If deprivation begets further deprivation, the expectation is that poor people are more likely to enter a new form of deprivation and less likely to exit an existing form of deprivation than non-poor people. Differences for these entry and exit probabilities may be estimated for every indicator and every year-to-year change. 

For example, poor people who are not experiencing material deprivation are 12-15 percentage points more likely to enter this form of deprivation each year than non-poor people who are not experiencing material deprivation. At the same time, poor people who already suffer from material deprivation are up to 20 percentage points less likely to leave this form of deprivation than non-poor people who also suffer from it.

The presented numbers seek to summarise the relevant dynamics for a particular year-to-year change. More specifically, different factors and mechanisms may produce the observed patterns, including the business cycle and other macro-economic trends, institutional features (e.g., support for people with disabilities), the specific configurations of the health and schooling systems, and also unobservable characteristics of the respondents (e.g., personality traits, talents, etc.). For instance, the insignificant differences observed for entering low work intensity are most likely driven by the sharp decline of the unemployment rate by some 12 percentage points during that period. 

Are the observed patterns also found in other European countries? Below we see the pooled differences for entry and exit probabilities by indicator for over 20 European countries and all four year-to-year changes. Besides the individual estimates of the differences, the figure also shows the median probabilities along with lower and upper quartile probabilities, which together give a sense for the commonly observed order of magnitude.  Salient observations include that similar differences in entry and exit probabilities are observed elsewhere in Europe which are largely consistent with the idea that deprivation couples over time. Moreover, probability differences in absolute terms are, by tendency, larger for deprivation exits than for deprivation entries, which suggests that already experienced deprivations play a more important role for impeding deprivation exits than for bringing about new deprivations. 

Four insights are worth highlighting. First, there is neither a form of deprivation that immediately implies poverty, nor one that is shared by all poor people, which suggests that forms of deprivation cluster in different ways. 

Second, the presented evidence underlines the importance of coordinating programmes and measures across policy fields —in particular to overcome already experienced deprivations, but also to prevent future entries into deprivation. Fighting poverty requires, therefore, joint and well-orchestrated actions by several actors.  Multidimensional poverty measures may offer more specific guidance along the way.

Third, monitoring differences in deprivation entry and exit probabilities on an annual basis may (i) help to draw more attention to the coupling of deprivations and (ii) encourage required action (e.g., well-targeted and designed policy programmes). Notably, the principal approach is not limited to multidimensional poverty measures.

Finally, the cross-country evidence also shows differences in entry and exit probabilities to be larger in some countries than in others, which suggests that social arrangements may either encourage or discourage the coupling of deprivation. Social arrangements, broadly understood, may include the institutional setting, the legal situation, and social norms, for instance. A more careful analysis of these differences may provide further policy-relevant insights.

ALKIRE, S. y FOSTER, J. (2011): «Counting and multidimensional poverty measurement». Journal of Public Economics, 95(7-8).

SEN, A. (2000): «A decade of human development». Journal of Human Development and Capabilities.