Across Europe and elsewhere – though notably not in America – standard methods of measuring poverty is employed in contemporary official statistics. These include Eurostat’s poverty statistics, but also national statistical publications like the UK’s Households Below Average Income and the Armutsberichte in Germany. Rather than the cross-national comparative dimension of official statistics that Anne and I wrote about in a previous post, during a stay in London in July 2015 I’ve been researching the history of such poverty measures within British national official statistics from 1975 to 1995.
Archival material1 and interviews show how the contemporary form of official poverty statistics developed out of existing data, conventions and activities within British official statistics. At the end of this post I’ll try to sketch some connections between the conventions of poverty measurement and the wider economic and political changes which took place rapidly in Britain from 1980 (the “inequality turn”), and only later and rather differently in Germany.
The poor, in the most widely used contemporary measure, are people who live in a household, whose total income, adjusted for household size, is less than some threshold – most often 60% – of the median of all household incomes in the country2. The official statistics themselves often have qualified titles, like “relative low income”, “income poverty” or, especially in Germany, “poverty risk” (Armutsgefährdung). They are, however, widely used and discussed simply as poverty rates, and no other standardised statistics exist which pretend to be measures of poverty.
I’ll touch on just two features of this form of poverty measurement. It is, of course, a measure based on income only (about which there is itself more to be said), and on the income of all households, regardless of sources of income or the economic activity of the members of households. Older income statistics, however, largely treated the non-working (e.g. pensioners, the unemployed) and the working as separate domains of statistical measurement. On the one hand this included, for the non-working (nicht Erwerbstätig), counts of social security claimants and estimates of the take-up of social security benefits or the “poverty trap”, and on the other (the Erwerbstätig), measurement of employment, wage growth and its relation to inflation. Although neither of these distinct fields of official statistical activity has disappeared, contemporary poverty measurement based on the whole-population income distribution has taken over and changed part of their function.
Another lynchpin of contemporary poverty measurement is the use of equivalence scales, by which household net incomes are adjusted according to the number of people in the household living off that income. In Britain the first domestic equivalence scale, the “McClements Scale” was developed in the late 1970s by the eponymous government statistician working in the Department for Health and Social Security3. The use of equivalence scales is a much more abstract and extensible approach to the analysis of household incomes in comparison to the older use of fixed “household types” in income analysis. Such fixed “household types” are best exemplifed perhaps by the Laufende Wirtschaftsrechnung, which, until its redesign in the late 1990s provided estimates of households’ income and expenditure only for three precise household types (Haushaltstypen), roughly “two-child worker-families”, “two-child middle-class families” and “low-income social-security recipients”. The model of the social in social statistics is rarely more obvious.
I have some work left to do on the German case, but it seems clear that the contemporary form of poverty measurement appeared in German official statistics quite considerably later – in the mid-2000s, with the Armutsberichte – rather than in the early-mid 1990s, as in Britain, and more overtly under the influence of European statistical standardisation. To close, a couple of hypotheses: poverty statistics are deemed “social statistics”, and such were relatively under-developed in the central statistics of the Federal Republic, whereas Britain had in the 1970s one of the most capable central statistical offices. The division of policy responsibility and statistical labour between Bund, Länder and institutions “outside” government like the DIW is and was also rather different. British statisticians were already in the 1960s engaged in income-distribution and inequality analysis, whilst in German official statistics other principles of stratification, such as Stellung im Beruf and, from reunification in 1990, East/West, received priority in published tables.
One of the most interesting things in the archives is the extent to which official statisticians were, within government, wrestling with the analysis of the changing distribution of income in Britain from 1980 under Thatcher. This point was a marked “inequality turn”, as the prevailing trend of the 1970s, falling or steady inequality, changed direction to a rapid increase in income inequality and poverty. Elements of these analyses from 1985, as well as a very public dispute over poverty statistics from 1986 (see illustration), fed directly into the establishment of new poverty measurement conventions of the type considered here. In Germany there was no such marked “inequality turn” at this point. The economic structures that underpinned, for example, the older separate statistical treatment of working and non-working households – the prevalence of the Normalarbeitsverhältnis covered by Lohntarif, high replacement value of social security payments – were only later, slowly and partially eroded in Germany.
- As an aside: for a sociologist interested in the state, but new to archival work, there is a very seductive quality to national archives, with ribbon-bound buff folders, typewritten memoranda and annotations. It feels like one really is, at last, observing The State. In the course of various research I’ve interviewed a good number of civil servants, and their professional caution and circumspection is one distinguishing feature of such conversations. In contrast, the internal correspondence gives the effect of showing raison d’État at work – although, of course, circumspection is just as much as part of the “state effect”. I’d welcome any good literature suggestions on this, the problems of studying the state from different sources.
- This and other standard European poverty measures are described and justified in Social Indicators: The EU and Social Inclusion (2002) by Tony Atkinson and colleagues. The definition of course conceals a host of complexities of definitions (of incomes, of households, of the reference population) and differences between countries in how actually to go about collecting the required data for the construction of such a poverty rate.
- First published in 1977 as McClements, “Equivalence scales for children” in Journal for Public Economics.