Ethos and the welfare state.

The OECD Secretariat released recently (November 2014) a revealing summary (pdf) of public social expenditure by OECD countries. The database the study is based on is available online. (Private social expenditure--i.e. private charity--is not covered by this post.) Social expenditure being defined as:

Social expenditure comprises cash benefits, direct in-kind provision of goods and services, and tax breaks with social purposes. ... To be considered "social", programmes have to involve either redistribution of resources across households or compulsory participation. Social benefits are classified as public when general government (that is central, state, and local governments, including social security funds) controls the relevant financial flows.

I was struck by the graph adjacent, covering specifically paid-in-cash benefits, which indicates that Anglo, Dutch and Scandinavian welfare states are strongly downwardly redistributive (i.e. the bottom income quintile receives a larger share of social expenditure than the top income quintile; and the former can be reasonably assumed to pay less tax than the latter) while the Mediterranean states are strongly upwardly distributive (i.e. the top income quintile receives a larger share of social expenditure than the bottom income quintile: one cannot say upwardly re-distributive, because that requires look at share of tax revenues). And, regarding present debates over fiscal austerity in the Eurozone: cutting public expenditure in states with downwardly redistributive social expenditure is likely to mean something rather different than doing so in states with upwardly distributive social expenditure.

There seemed to be some patterns in the data, so I downloaded said social expenditure data, added in data on economic freedom and (via Wikipedia's useful lists) on religious adherence as % of population and was able to generate various, somewhat striking, correlations.

Using the bottom quintile's share of social expenditure less the top quintile's share of social expenditure (in % points of total such expenditure) as an indicator of how downwardly distributive social expenditure was, there was no significant correlation (0.18) between the total level of social expenditure (as a share of GDP) and how downwardly distributive total social expenditure was. So, the level of public social expenditure as a share of GDP tells us literally nothing about how focused on helping low income folk such expenditure is.

Regulation fairy stories

There was quite a strong positive correlation (0.59) between the level of economic freedom and how downwardly distributive social expenditure was. Now, if you believe in the state-as-regulation-good-fairy story (the state typically regulates to improve overall social and economic outcomes), this may be a surprise.

If, however, one accepts that a significant amount of regulation is to favour selected groups and that, generally speaking, the higher the level of regulation the more this can be expected to be so, then this result will be unsurprising. (Not least because, as mechanisms of transparency and accountability are not infinitely elastic, so the more they have to cover the weaker they can be expected to operate.) Especially as the better connected, resourced and organised an interest group is, the more it is likely to be able to bend regulatory policy in its favour.

So, taking economic freedom to be an indicator of "neoliberalism", then the more neoliberal (other things being equal) your economic regime, the more downwardly distributive public social expenditure it tends to be. Shocking only if you accept the "bad fairies" theory of neoliberalism: which so many academics do; but, then, much of what academics write about neoliberalism is crap.

The deserving poor

There was quite a strong positive correlation (0.60) between the Protestant share of population and how downwardly distributive social expenditure was and a stronger positive correlation (0.64) with the no-religion share of population.  (It was clear from the sources that, depending on context, people would nominate both a religious identity and as being of no religion: I took that to mean they were culturally Protestant, Catholic, etc.) So actual and cultural Protestants, and folk with no religious belief, apparently tend to believe in the deserving poor: i.e. that welfare expenditure should be downwardly re-distributive.

This was capturing something specific, because the correlations between between the Protestant share of population and the level of economic freedom (0.47) and between the no-religion share of population and the level of economic freedom (0.36), though positive, were not as strong. 

Protestantism I would characterise as "naked before God" religion, since one has direct access to the basic religious authority (Scripture) and is entitled to make one's own judgement about it. Folk with no religion can be expected to generally also believe in a strong sense of individual moral sovereignty.

So, my tentative hypothesis would be that a confidence in one's own moral judgements (and the sense of moral sovereignty that flows from that) apparently encourages social expenditure to be downwardly redistributive: what perhaps might be called a strong sense of the deserving poor. Perhaps because it encourages considering people by fairly direct, and directly identifiable, notions of worthiness (in this case, lack of income).

Preserving rank

A very different result was gained if the Catholic+Orthodox+Muslim share of population were added together, because then there was a strongly negative correlation (-0.70) between said share of population and how downwardly distributive social expenditure was.

Again, something specific is going on, as the correlation between the Catholic+Orthodox+Muslim share of population and economic freedom, though negative, was not as high (-0.48). In keeping with the level of social expenditure not being a key factor, there was no significant correlation (-0.08) between the Catholic+Orthodox+Muslim share of population and public social expenditure as a share of GDP.

Catholicism, Orthodoxy and Islam I would characterise as "priests and clerics give detailed instructions" forms of religion. They involve both hierarchical notions of moral authority and complex moral maps--since it is in the interest of gatekeepers of righteousness to promote moral complexity, as it inflates their role. You probably don't need a priest or cleric to tell you that murder is bad; you probably do need them to tell you whether you need to wash your hair every time after you have sex or how to expiate specific sins.  

The combination of moral complexity and moral hierarchy apparently leads to public social expenditure which reflects, even reinforces, existing social rankings. Thereby leading to much less policy weight being given to such a direct characteristic as (low) level of income. Remembering that complexity of any sort is a great way to obscure who is receiving what.

Ethos matters

So, ethos appears to matter, given that there is such a vast difference between the apparent connection between Protestantism (religious or cultural) (0.60) and no-religion (0.64) on one side, and Catholic+Orthodox+Muslim share of population on the other (-0.70), and how downwardly distributive public social expenditure tends to be.

One possible mechanism via which religious roots of cultural perspectives could matter is different perspectives on time. The work of psychologist Philip Zimbardo and others on time perspectives (pdf), suggests that higher self-trusting, future-oriented Protestants might be more likely to think that the state should concentrate on those who need help. Conversely, lower self-trusting Catholics, Orthodox and Muslims who are more past or present oriented may think the state should do more for everyone regardless of current situation.

The former will lead to more downwardly distributive social expenditure, the latter much less so. Especially as, once it is accepted everyone should receive, the better organised and connected are much better place to, well, so receive.

Whatever the actual mechanism by which the observed effects happen, the data does clearly suggest that policy, over time, reflects the choices of the voters. Choices that appear, in turn, to significantly reflect what moral ethos is dominant among voters.

[Cross-posted at Skepticlawyer.]