Joint post by Marion and Eve
Poverty is often measured in terms of income: all those whose income falls below a certain threshold are considered as poor. As Benoît explained in his post (January 22nd), this poverty line is supposed to be the income threshold allowing purchasing a bundle of goods giving the minimal welfare level. Over the recent decades, research has been developed on the idea that we could directly focus on the bundle of goods people manage to get: people are considered as deprived if they cannot afford a certain number of items considered as basic needs. This concern led to the inclusion of deprivation measures in the list of Laeken indicators and to the introduction of deprived people in the target population of EU2020 aims. As detailed in the next post, one of the goals of the EU2020 is to decrease the number of poor or at risk of poverty individuals by 20 millions in the EU by 2020.
Brian Nolan and Christopher T. Whelan present in their recent book Poverty and Deprivation in Europe  the pattern of poverty in Europe when it is measured as deprivation. This book illustrates how the use of non-monetary measures of poverty could give a picture of poverty, different from income poverty measures and how it could impact the comparison between countries.
Poverty as deprivation: non-monetary indicator of poverty
A very classic measure of poverty is based on the measure of income: the total income of the household per consumption unit is compared to a certain threshold. Those below this threshold are considered as poor. The underlying idea is to measure the inability to reach a certain standard of living. But measuring poverty through income is an indirect way to focus on living conditions of people. A low level of income is linked to the inability to afford some conditions of living if this inability is attributable to inadequate resources. As a consequence, the income measure of poverty suffers severe drawbacks. First, poverty is multidimensional (for instance, Ravallion (1996)): for example, health status and housing quality are important ingredients in individual well being. Second, in the framework initiated by Amartya Sen, income does not reflect the capabilities of agents. In other words, income does not necessarily give you the capacity to reach a certain well being (having gold is not useful in the desert if there is no place where you can buy water with it).
Non-monetary measures of poverty can be used as a direct measure of poverty as they intend to look directly at the bundle of goods, and it measures what people are lacking of. The idea is the following: a list of items considered as basic items is made and some surveys such as the ECHP or the EU-SILC ask people if they are deprived over that item. In the ECHP, items included in the list are of different nature. They bring information on the affordability of basic amenities (car, colour TV, telephone, etc.), quality of the housing (keeping house warm, replacing worn out furniture, having a bath of shower, hot running water, indoor flushing toilet, shortage of space, leaky roof, etc.), quality of the environment (noise from neighbors, vandalism, pollution, etc.), social interaction (having contacts with friends of family), other items (eating meat or fish every second day; buying new clothes, etc.) and on arrears. The EU-SILC data provides information on less criteria but in the same vein. For each item, respondents are asked to specify if they do not possess the item because they could not afford it or because they chose not to possess it. This question is supposed to measure the incapacity to get the items.
Of course, those surveys provide a lot of information. The authors propose different ways to aggregate the information into more comprehensive indices: keep variables on material deprivation (and exclude social interactions) and use confirmatory factor analysis , to gather the different items into 5 dimensions in the ECHP and into 3 dimensions using the EU-SILC . Each dimension is composed of several items and an individual is considered as poor on that dimension if she is lacking at least a certain number of items, the threshold depending on the dimension. Then, what do those 3 or 5 indices tell us about poverty as deprivation that monetary indicators do not?
Mismatch between income poverty and deprivation
Let us compare non-monetary indicators to monetary indicators. In this section, we focus only on a part of the result of the authors. They construct the measure of income poverty as follows: they take the equivalized net income of the household, and consider as poor those below the 60% of the national median. Therefore, the poverty line is country specific. They also provide a country specific measure of deprivation: each item is weighted inversely to the proportion of people possessing the item in the country (they do so to take into the prevalence of the item in the country). The data comes from the EU-SILC 2006 survey.
As Figure 1 shows, the ranking of countries over the deprivation measure is quite similar to the ranking of countries using the income measure of poverty. Moreover, the proportion of income poor is not really different from the proportion of deprived people. Although the measures are country specific, the contrast between countries is large. On average, across EU member states, 16% are of risk of poverty, with only 10% in Czech Republic and the Netherlands, whereas at most 23% in Latvia. Denmark, Finland, Sweden, Germany, Hungary, Slovakia, and Slovenia show low levels of poverty over the two different measures. Ireland, the UK, Spain, Italy, Portugal, Etsonia, Lithuania, Poland, and Romania presents higher levels of poverty over the two measures of poverty.
Figure 1: realized on data from Nolan and Whelan, 2011 (Table 6.2).
However, the authors present an interesting fact: the income measure of poverty and the deprivation measure do not label the same individuals as poor. As shown in Figure 2, only a small part of the population is labeled as poor over the two indices of poverty, whereas a large fraction of the population is either income poor or deprived, but not both (between 15% and 20% in most countries). As a consequence, deprivation measures describe a different pattern of poverty compared to the income measure of poverty.
Figure 2: realized on data from Nolan and Whelan, 2011 (Table 6.2)
The authors document the mismatch. In particular, it is interesting to see that although income poverty and deprivation are highly correlated, there are individuals considered as deprived at all levels of the distribution of income. On the other hand, the mismatch tends to be particularly severe for some parts of the population: elderly and self-employed workers tend to be more often income poor but not deprived.
The authors propose some reasons that could explain the mismatch between monetary measures and non-monetary measures of poverty. Especially, income is transitory and many of the items considered are durable goods (or at least, things that are not easily changed: for example, you do not move all the months whereas income could change all the months). Moreover, income is most likely to be badly measured than the lack of items. As a consequence, the authors consider that the measures do not reflect different situations and should be seen as complementary. As the next post will explain, this vision had a big impact on the measure of poverty retained by the European Union.
So, what does deprivation measure?
The income measure of poverty suffers many drawbacks that justifies the quest for other measures. However, the authors claim that the income measure of poverty might be affected by two others drawbacks: income is not necessarily well measured in data and the choice of the threshold might turn out to be a difficult one. It is not clear to us that non-monetary measures tackle those drawbacks. Measuring deprivation over one item might be difficult. Let’s take an example: does my neighbor make noise? Clearly, someone suffering insomnia is not likely to answer the same way as someone who sleeps very well. As a consequence, the measure could be highly subjective and therefore, it does not solve the problem of comparisons between individuals and between countries. But even some less controversial item could be a bit tricky: my partner and I have a car so we are not deprived on that dimension, but only my partner uses it to go for work. Am I deprived? Moreover, the choice of the threshold might be a difficult one: are you deprived in terms of consumption when you are lacking 3 over 7 items of consumption? Or 4 over 7? The choice of the threshold in the EU indicators will be documented in the next post.
Moreover, even the choice of items could be ad hoc, as it barely takes preferences into account, especially in the aggregation of items: if the weights are the same for all items, it means that you equally decrease deprivation by giving a TV to everyone or by giving meat every other day to everyone. But if I care more about meat than about TV, those two items do not have the same impact on my well-being. Moreover, using the same weight for everybody leads to the same problem: having a TV might affect more my neighbor’s well-being than mine. This problem is exactly the problem of aggregation, to which Decancq, Fleurbaey, and Maniquet (2013)  provide an answer, by taking preferences into account. The main idea is to consider how individuals weight each item in the measure of poverty. Applied to the measure of deprivation, it would means that I am more deprived if I don’t have a TV and I really like to watch TV than if I don’t like to watch TV. So the point is not that I live in a country where everybody have a TV and I don’t but that I really like watching TV and I can’t.
 Brian Nolan and Christopher T. Whelan, Poverty and Deprivation in Europe, Oxford : Oxford University Press, 2011.
 Ravallion M (1996) « Issues in measuring and modelling poverty ». Economic Journal 106: 1328-1343.
 ECHP : European Community Household Panel (panel data 1994-2001) and EU-SILC : European Union Statistics on Income and Living Conditions (EU-SILC)
 The confirmatory factor analysis (CFA) is statistical method of analysis of data. In social sciences, it is often necessary to summary different variables into a simpler index. The CFA is a method to check if the way the variables are gathered makes sense by computing the correlation between the different factors.
 The 5 dimensions of deprivation using the ECHP are: in basic needs (such as heating, eating meat), in secondary needs (TV, dishwasher, etc.), housing facilities (shower, flushing toilet, etc.), housing deterioration (leaking roof, etc.) and environment (noise, pollution, etc.). The 3 dimensions of deprivation using the EU-SILC are: consumption, household facilities, neighborhood environment.
 “Shouldn’t we take preferences into account?”