The book's subtitle relates to the phrases ‘To learn more about European efforts’ to ‘improve their measurements of income, poverty and material well-being’ (p. 6), with which the editors express the motive behind the conference, organised by the University of Maryland and the OECD (Paris, 2009), from which the book derived. As an active participant in poverty measurement methodologies since the eighties, I understand that nowadays ‘new thinking’ on poverty measurement means multidimensional/combined methods, which complement income poverty with direct deprivation indicators (see Alkire and Foster, Reference Alkire and Foster2008, whose proposal has many followers; Boltvinik, Reference Boltvinik2013, for a description and evaluation of the Latin American experience of multidimensional poverty measurement since the eighties; and Grusky and Kanbur, Reference Grusky and Kanbur2006, as well as Kakwani and Silber (eds.), Reference Kakwani and Silber2007, for examples of current discussions). However, from the nineteen chapters in the book, only chapter 16, by Nolan and Whelan, properly broaches multidimensional/combined methods.

The book is divided in six parts, and in each of the first five the papers included are commented by American scholars. A careful reader of the book, paying close attention to these commentaries, which represent the best of the book, particularly those by Burkhauser, Bavier and Gilbert, will perceive that European poverty measurement is in an exit-less maze.

In Europe, the concept of poverty has become confusedly linked with social exclusion (a vague concept in itself) regarded as broader than poverty. Following Gilbert (chapter 18), under the rubric of social exclusion in the EU, it has been agreed to include indicators on three dimensions: (a) income poverty, (b) material deprivation and (c) other social issues as education, housing, health care, unemployment. But Marlier et al. (chapter 15, the first of four in Part V, ‘Multidimensional Measures’) reject the possibility of integrating these dimensions in a composite indicator, arguing that doing so requires value judgements to define their weights and the combination procedure. Composite indicators are disqualified as attractive only to the general public and the media. By doing this, multidimensional poverty measurement, which precisely means arriving at a composite indicator which reflects an integrated view of poverty (as can be confirmed in the above-cited references), is rejected. Measuring many dimensions, present in any social statistical handbook, is not the same as multidimensional poverty measurement. In the absence of a composite indicator, income poverty will continue reigning with its biases, limitations and contradictions, highlighted by many authors in the book, so the rejection of a composite indicator can be interpreted as a subterfuge to maintain such dominance.

Marlier et al. pretend implicitly that in measuring many dimensions they avoid value judgments, which is obviously not the case. For instance, the indicator adopted as material deprivation is based on nine items of enforced lacks (people who want the item but can't afford them). The selection of these items and the exclusion of many basic satisfiers, like the whole block of dwelling indicators included in the EHCP (European Community Household Panel Survey) is a non-grounded value judgement, as is the allocation of equal weights to all nine indicators, as pointed out by Gilbert (chapter 18). One of Marlier's co-authors is Nolan who, in his paper with Whelan, proposes a multidimensional composite indicator of poverty which identifies as poor only households which comply with two conditions (the intersection criterion of both sets): their income is below the poverty line (PL) and they have three or more enforced lacks among seven selected items of material deprivation, thus lowering poverty incidence drastically. This involves a much heavier value judgement: designing a procedure by which you minimise the inclusion error (identifying as poor someone who is not) but maximises the exclusion error (identifying as non-poor many poor households).

The following are some of the biases/limitations/contradictions of the indicators of income poverty (currently employed in the EU and the OECD), of the material deprivation indicator agreed at the EU, as well as of the combination proposed by Nolan and Whelan of both:

1. (A) The definition of the PL (set as 50/60 per cent of median household income) is presented as the application of Townsend's ideas, who defined relative poverty as the condition of exclusion experienced by individuals/households due to insufficient resources to participate in the living patterns, customs and activities (Reference Townsend1979a). Townsend (Reference Townsend1979b) added that, ‘Lacking an alternate criterion, the best assumption would be to relate sufficiency to the average rise (or fall) in real incomes.’ But it should be noted that to define the initial PL level, he never proposed to base it on mean/median observed income. In chapter 6 of his magnum opus (Reference Townsend1979a), Townsend proposed to relate the direct indicators of deprivation with household income in order to identify the level of income below which observed deprivation grows rapidly, thus revealing what he called the objective PL. The methodology adopted in the EU and the OECD is not Townsend's methodology, because as Bradshaw has said, and Besharov and Couch cite approvingly in the Introduction, ‘there never has been any scientific justification for the 60 per cent of median equivalent income threshold’. Why a percentage below 100 per cent, and not 100 per cent, or higher than 100 per cent? Why the same percentage of the median income for countries as disparate (in their living standards) as Luxembourg and Lithuania?

2. (B) As Sen (Reference Sen1981: chapter 2, and Sen, Reference Sen1983) warned, adjusting the PL based on the behaviour of the mean/median income may lead to absurd situations. Poverty increased slightly, in OECD countries as a whole, according to Förster and d'Ercole (chapter 2) between 1985 and 2005, a growth period, while it decreased slightly from 2005 to 2010, a period of crisis. It could not be more paradoxical.

3. (C) Upon adoption of different poverty PLs in proportion to the median income of each country within the EU (a borderless labour economic unit), as Burkhauser (chapter 4) analyses in detail, the incidence of poverty is lower in Poland than in the UK, to give an example. Who can find useful such information?

4. (D) Unaware of the inherent contradictions involved, the EU has been using completely relativist PLs, which change annually, defined in terms of the value of the median income of each country in each year, side by side with absolute material deprivation indicators (and also absolute indicators of the third group), which are the same across countries and through time. The results (e.g., low correlation between absolute and relative indicators) are surprising only for those who are unaware of this conceptual inconsistency. (Gilbert, chapter 18, is an inch away of perceiving this. His critiques are brilliant.) Similarly, welfare supports are allocated in the EU on the basis of absolute deprivation criteria (Bavier, chapter 11). In many cases, this is done based on normative family budgets, a method commonly rejected, in the same countries, for poverty measurement.

5. (E) The third type of indicators (employment, education, health care) are handled separately and piecemeal.

6. (F) At the end of the day, the EU neither measures poverty nor social exclusion. Income poverty, which is indeed measured, is called at-risk-of-poverty, and regarding multidimensional poverty, as in the proposal by Nolan and Whelan – also applied for many years by the Bristol Group and integrated by followers of Townsend (see Pantazis et al., Reference Pantazis, Gordon and Levitas2006) – but has not been accepted officially, is not measured. Other indicators of social exclusion (unemployment, education, housing, health services) are measured piecemeal. Thus social exclusion is, as poverty, not measured.

The book is a must for poverty researchers and policy makers.