In the recent book Inequality: A New Zealand Crisis, Jonathan Boston has a really useful chapter. It probably should have started the book as a way of laying some intellectual groundwork, instead of being Chapter 5. It raises the same point as Matt Nolan does at TVHE, quoting Amartya Sen, but with more detail. Here’s Boston:
As highlighted in this chapter, there is almost universal acceptance that equality matters. Yet there is no consensus on what kind of equality should be championed.
And here’s Nolan:
Everyone, especially those who are more extreme in any given “political dimension” cares about equality of something – and the underlying reason why there are trade-offs stems from (as Sen discusses in the book) the heterogeneity of individuals!
Boston tries to cope with the different equality targets in two ways. First, he says that some equalities are more important than others, arguing that there is an inequality of equalities. With this, I believe he steps right back from his initial understanding — that reasonable people may reasonably disagree. Secondly, he suggests a certain pragmatism, or specific egalitarianism. While this seems attractive, it papers over the real conflict amongst theories of egalitarianisms with a poorly defined set of concrete goals.
In doing this, Boston shows how hard it really is to have evidence-based policies. When it comes time to make a decision, we have to take values and preferences into account. A particular set of evidence can be made to suit a range of policies, once you introduce different values. I’ve not read Peter Gluckman’s new report on evidence-based policies (pdf), but the reporting I’ve seen suggests that it is insufficiently attuned to this problem.
There is a second problem less well understood. When we make judgements about inequality, we are using mental models of social systems to create counterfactuals. For example, if one says, ‘they wouldn’t be poor if they weren’t lazy’, there is a mental model of society that underpins the judgement. That model has a weighting on ‘effort’, as well as weightings on other things like ‘education’, ‘social network’, ‘ethnicity’, ‘gender’, and more. The ‘effort’ weighting is sufficiently large to counteract any negatives from the other factors. We could, through interviews and surveys, estimate the parameters that people apply to those factors.
We all have these models. They are all wrong. I say that with conviction because ‘all models are wrong.’ They are always partial — they have missing variables — and the parameters are estimated from a sample of observations rather than the population. So, in my interpretation of my experience, it may be that ‘effort’ is sufficient to make a person not-poor. That doesn’t make it so, either for my experience (which suffers from observer error) or for the wider world.
In economic analysis, counterfactuals are hard to construct. You have to decide which factors are important and how important and over what time. If I’m being cynical, I might say that the counterfactual is the most important part of any cost-benefit analysis, and it is literally something we just make up. With data and evidence, mind you. But we make it up just the same.
How much harder is it, then, to understand the counterfactuals that people create from poorly-specified mental models of complex social systems?
More evidence would help, so it’s good to see Gluckman encouraging the government to find and use more. The findings will rarely be conclusive, however, as evidenced by Boston’s discussion of equality.