This article first appeared as “Warum es keine Armutsstatistik braucht” in Schweizer Monatshefte, Issue 979, July/August 2010
In the early 1960s, Chicago economist Milton Friedman travelled to Hong Kong to meet Sir John Cowperthwaite, the colony’s Financial Secretary. The economic development of Hong Kong had aroused Friedman's curiosity, but, to his surprise, he was unable to get hold of any detailed economic statistics. So he asked Sir John about the peculiar lack of data. Cowperthwaite, a classical liberal economist, explained that he deliberately refrained from having detailed economic data gathered. He argued the bureaucrats of the colonial administration should not even be tempted to intervene in the economy.
Cowperthwaite’s approach may appear a bit eccentric, but it seems it was not to Hong Kong’s detriment. According to estimates, the city’s income per capita has increased tenfold in real terms over the past half century. With that in mind, perhaps it is not the end of the world if Switzerland does not produce official poverty statistics – even though Caritas Schweiz seems to be in great pains about it. The charity’s Sozialalmanach 2010 states: "If the World Bank manages to do it on a global scale since the 1990s, then surely twenty years later, Switzerland must finally manage to do it too! [...] Up to the present day, there are no official poverty statistics in Switzerland."
The editor of the Sozialalmanach, Christin Kehrli argues that conventional international poverty measures should be applied in Switzerland as well. A seemingly reasonable enough request. Of course it would be desirable to have an accurate picture of the extent of poverty, its time trend, geographic concentration and the risk factors associated with it. But the indicators Kehrli mentions produce more confusion than clarity. The author identifies three different approaches to poverty measurement. Firstly, relative poverty: people are poor if their equivalised income falls below 60%, or 50%, of the national median. Secondly‚ subjective poverty: people are poor if they classify themselves as poor. And finally, multiple deprivation: people are poor if they lack essential goods and services from a pre-specified consumption basket.
As it happens, there is a paper from the University of York, which applies these three indicators to British data. The result is surprising. All three indicators yield similar poverty rates – but there is hardly any overlap between the three groups. To oversimplify: those on a low income did not consider themselves poor; those who considered themselves poor were not lacking anything essential and those who lacked essentials did not have low incomes. A major oops-moment, one would have thought...
The most commonly used indicator is the relative one. Setting the poverty line at 60% of median income, 15.2% of the Swiss are poor, according to the OECD. Apparently, this is the indicator which Kehrli favours: "Poverty is an indicator of exclusion and inequality in a society. [...] Poverty is therefore a relative phenomenon; it refers to a reference group, usually the people of a particular country."
In a sense, Kehrli has a point. Poverty means different things in different places. Surely all Swiss citizens would agree that a damp-free, centrally heated home with an indoor bathroom, electricity and hot water are absolutely basic needs. But in, say, the former Soviet republics, not all people will take these things for granted yet. A poverty indicator must reflect such differences in local perceptions; otherwise, it would be doomed to irrelevance.
But does this mean that poverty lines should be pegged to national median incomes? Why should the national level be the correct reference, as opposed to the cantonal or the municipal one? The problem is, if poverty lines were pegged to cantonal median incomes, then poverty would skyrocket in Geneva and Zürich and hit rock bottom in Fribourg and Appenzell-Innerrhoden. Absurd? Maybe, but then so would the OECD figures be. Taking this logic a step further, one could ask why the reference territory should be identical to a political territory at all. It could also be selected by a completely