Seeing Like A Neoliberal, Part 4: Statistics, States and Seductive Stories

This is the last post in my series critiquing the idea of ‘progress’ as depicted in statistical charts, so we must now ask the natural question: is it all part of a conspiracy to keep us in little boxes?

As the observant among you might have guessed, the theme of these posts was inspired by James Scott’s masterpiece Seeing Like A State, where he critically assesses the ‘view from above’ and how states throughout history have tended to pursue specific visions at the expense of more holistic and humanistic considerations. In this post I am going to clarify why I think the view which celebrates progress — call it neoliberalism, New Optimism , or whatever else — is accurately characterised by Scott’s framework.

Social Statistics are Political

Statistics are simplifications which inform the stories we tell ourselves about the world, and a simple change in perspective can drastically alter our understanding of how the world looks. As social scientists and those who draw on social science, we are inevitably constrained by this top-down view, particularly when we are shown statistics which cover too broad a range of outcomes, time periods and places for us to scrutinise their origins. These origins are rarely, and perhaps cannot be, politically neutral.

In my second post I discussed whether improvement in certain outcomes is always evidence of progress. I suggested that Cuban health statistics, while they appear impressive at first glance, are actually more of a mixed bag when you understand their purpose in legitimising the Cuban government, which does things like randomly has doctors visit peoples’ houses to check up on their lifestyles. Someone in the comments questioned my assertion that Cuba has better health statistics than the US, suggesting the Cubans would prefer to pick the sources that paint their health system in a better light, while other sources might be more pessimistic.

Although the commenter seemed to think they were contradicting me, I’m actually thankful to them for making this point salient. It does not take a genius to realise that because Cuban health statistics make good propaganda, their government will be likely to (1) use optimistic or even inflated estimates; (2) pursue policies that enhance health regardless of their other, sometimes coercive effects; (3) nonetheless, have genuinely impressive health achievements given Cuba’s national income. All three of these points are consistent with each other, and they encapsulate perfectly what I wanted to say about social statistics in this series.

This is especially true given that I’ve focused on poverty reduction, since many poverty-relevant statistics are gathered by large, transnational organisations which are insulated from democratic scrutiny and accountability. Indeed, there is a certain level of irony in people who deride central planning, and who presumably quite like thinkers such as James Scott and Friedrich Hayek, taking the word of institutions like the IMF, World Bank and (to a lesser extent) the UN at face value.

These institutions have, in recent decades, shaped themselves around the goals of reducing poverty, mortality and malnutrition (the UN’s Millennium Development Goals take a slightly broader perspective but are still goal-oriented). And (with apologies to Pseudoerasmus) I have found Jason Hickel extremely useful here, as he has tracked these institutions over time and found that they have repeatedly moved the goalposts to make themselves look better.

Firstly, they tend to low ball poverty: people cite the $1.25/day measure a lot without stopping and considering quite how little it is — those living at this level still face substantial hardship and health risks. Researchers have suggested $4–5/day is more credible, with some even going as high as $10/day (still only $3,650/year, well below any relatable living standard for those of us in rich countries).

Secondly, the original aim was to reduce the absolute number of people in poverty, but this has stayed roughly constant at the $1.25/day line and has actually increased at higher poverty lines:

Thirdly— as I mentioned in my first post — the poverty line has been repeatedly recalculated and reevaluated based on PPP, which can have the dramatic effect of ‘lifting’ millions out of poverty even though nothing has actually happened. Remarkably, each iteration by the World Bank seems to both reduce the number of people in poverty and make the decline over time steeper. Fourthly, they have moved the starting date back to 1990, which includes a period from before they actually established the goals, counting most of China’s poverty reduction — which has precious little to do with these institutions.

Some of these changes may seem justified individually, but together they paint a picture of picking and choosing whichever measures make it seem like the goals have been achieved ,when they clearly would have been missed had these institutions stuck to their original criteria. And as Hickel points out, the most egregious example comes not from the poverty line but from hunger:

In 2009 the FAO reported that the food crisis — building on the global financial crisis — had pushed 150 million additional people into hunger. It was
a catastrophic rise: the number of the hungry was up to 1023 million, a 21% increase from 1990…It seemed a disaster. But then in 2012 the FAO suddenly began telling the exact opposite story. With only three years to go before the expiry of the MDGs, it announced an ‘improved’ methodology for counting hunger, and the revised numbers delivered a rosy tale at last.

The new line was in many ways completely inadequate: among other things, it assumed people in poverty are completely sedentary, and defined ‘hunger’ only as being malnourished for a year or more (try fasting for a morning and see if you agree with this definition).

There are so many degrees of freedom here that it becomes necessary to subject these statistics to extensive scrutiny before accepting them. What I in my third post called the ‘trend-bias’ is perhaps a specific instance of a more general phenomenon: picking a statistical property — whether it be a downward trend, particular threshold, an average, or whatever else — such that it depicts a seemingly beneficial outcome.

One such example which shocked me was Muheed Jamaldeen’s recalibration of Branko Milanovic’s famed elephant graph, which shows the global middle class as the major benefactors of modern globalisation and which paints a picture of declining global inequality. I had previously accepted this story, but Jamaldeen has pointed out that it is based on relative income increases: if a poor person earns $4000 a year and experiences an increase of $1000, this is a 25% rise; but if a rich person’s income starts at $40,000 and increases by $5000, this is only a 12.5% rise. Simply because the former’s starting point is lower, they will look relatively better off despite their gain being objectively lower.

Jamaldeen pointed out that if you instead of look at absolute income gains, the elephant graph looks more like a hockey stick, illustrating that most of the gains of globalisation have gone to the richest in the world (and also the richest within that group):

This is a drastically different (and I think, more relevant) picture than the one being told by the IMF and World Bank.

Not Taking States at Their Word

It is reasonable to be a priori skeptical of the achievements states and state-like institutions attribute to themselves, not just because they may manipulate them but because they might (deliberately or otherwise) serve to distract us from immoral or undesirable things going on underneath. Cuba’s health statistics — even if they are to be believed — do not suffice to justify their government, and neither do rosy poverty statistics justify the existence of global economic institutions or the system that they support. As Scott’s book illustrates, even the worst states have considerable resources and manpower behind them and are therefore able to marshal some achievements, regardless of how they treat their citizens or how things look from a different perspective.

In other words, the morality of a state or system does not depend on the gains it makes in specific outcomes. When Bruce Giley wrote ‘The Case for Colonialism’ he at least in part based it on supposed material gains from colonisation, but this is irrelevant to those of us who value the self-determination of colonised countries as an end in itself. Even slavery likely produced material gains, and many slaves would have been materially better off if they participated in the system — but slavery is obviously wrong, and no graph is going to change that.

A few outstanding, uncomfortable facts of the global system — ones which are plausibly positively related to overall material gains — are: the enclosure of large tracts of land, with all of the displacement of natives and wildlife that engenders; numerous transfers from poor to rich countries, such as through patent laws; implicit and explicit state subsidies to major Western companies, especially banks; and yes, the continued existence of slavery, which is well embedded into global supply chains.

Three Rules for Progress Skeptics

There are therefore a few rules to take away from this series, through which we can treat statistical indicators of progress skeptically without dismissing them entirely. Firstly, ask for ‘robustness checks’. What do the statistics look like when we change our assumptions about missing values? Or move between proportions and absolutes? What about if we change a threshold, or exclude certain countries? (If you play with the World Bank poverty data you can see they do generally pass the latter two checks, but only for proportions).

Secondly, look for indicators that are less dominant on the agenda of the institutions doing the producing, reporting and policymaking: inequality, mental health and the environment are important contemporary examples. Moreover, finding out how the people behind the statistics perceive their situation can reveal things that the statistics do not— such as safety and health risks from industrialisation, or that the misery of relative poverty is the same across countries with hugely divergent income levels.

Thirdly, ask why some things have improved. Is it because of policy, industrialisation, social movements? Or is it because of bulldozing over other things (with the Amazon rainforest being a literal example)? This can tell us whether or not progress has truly been made, and if so whether this is because of, despite or incidentally to the proclamations of states and state-like institutions.

For an example of how to do all three of these things well (certainly better than I have), check out Jeremy Lent’s review of Steven Pinker’s latest book. I myself am swearing off meta-statistical posts for a good while, and will get back to complaining about economics in due course.