Two posts on the measurement of inequality this week, so you’ll need to activate the brain cells. First up Oxfam researcher Franziska Mager summarizes a paper co-authored with Deborah Hardoon for a panel at the recent Development Studies Association conference on the power and politics behind the statistics. A version of this post appeared on Oxfam’s shiny new real geek blog.
Inequality is a touchy issue, so for different actors, measurement is vital. We saw that when Oxfam published its Davos statistic on extreme wealth in February.
There is no one ‘right’ way to measure anything. That’s what measurement is: one way to quantify out of many. But there can of course be a wrong way – what we would say in statistics was inaccurate, with a bias and led us to a false conclusion, or something excessively imprecise. Plenty of statisticians and fact checkers work to identify when measures are wrong.
Right v Wrong is one important distinction. Another is ‘who is measuring what and why?’ Inequality of opportunity and inequalities based on forms of discrimination such as gender have long been an issue of concern in international development. Inequality of economic outcomes, however, has not traditionally been a priority, notwithstanding a concern for poverty reduction in absolute terms. This is due to an unwavering acceptance of the need for growth to fuel development and a belief in meritocracy. Globally, the extreme poverty line is fixed at $ 1.90 a day, which frames thinking around meeting an absolute level of basic needs and doesn’t consider the unequal spread of incomes.
Economic inequality can’t by captured by a single metric. For starters, income has different meanings, like market incomes and those after redistributive policies (taxes, benefits). They can lie far apart – Sweden is a good example for high market income inequality but very low income inequality after redistributive policies. Not just that, within any definition of income, there are multiple ways to cut your data. You can calculate the Gini coefficient; use the Palma index or an analogous ratio relating different income groups to each other. The point is: different measures place different emphasis on different parts of the income distribution. Oxfam for instance advocate for measures that contrast the tails of the distribution better than the Gini.
Beyond incomes, the extent of inequality within an economy is also expressed by wages, data on which can again be cut differently. In one such cut, wages tell us about the dispersion of earnings between those at the top and those at the bottom in a company or country – like CEO pay and minimum wages. But we can also analyse how much income is generated by wages compared to capital (e.g. returns on shares) – the latter by definition being an income source of the wealthy. Wages contain vital information about how an economy rewards workers and the link between labour and prosperity.
Analysing the wealth distribution in turn tells us who owns what assets. Wealth inequality is even more extreme than incomes and again reveals a different story about an economy – for instance one’s ability to react to shocks or exert power and influence.
Let’s look at some income examples where different measures tell different stories.
Brazil is one success story: over the past decade the incomes of the poor have been growing faster than those of the rich, poverty has been falling, and inequality by most relative measures along with it. The red bar shows the real incomes of the bottom 40% of the population. This bar represents the story of the poor getting better off, because although small, it almost doubles in size between 2004 and 2014. The blue bar is the richest 10% of income earners. Their incomes have also been increasing – but because they were high to begin with, this percent growth rate is lower than that of the poor. This leaves the green bar: the ‘money difference’ between both. In those absolute terms, the richest are pulling away from the bottom (this is true of multiple data sources).
Using the famous elephant graph, differences concealed by rates of income growth and how this plays out on the global scale are eloquently explained tomorrow in our colleague Muheed’s post.
Now let’s think about Uganda. In this case, the choice of data source makes a bigger difference than the choice of indicator. Below we compare the Gini coefficient with the Palma ratio, the Gini being an estimate of inequality across the whole of the income distribution, whilst the Palma explicitly compares the richest 10% to the bottom 40%. As you can see, no matter the indicator, the inequality trend is roughly the same.
When we did this, we were startled by how patchy data is. World Bank estimates use consumption data (what people spend money on, rather than their incomes) and are based on household surveys conducted every few years. The last survey in 2012 calculated a Gini coefficient of 42 (the red line). In contrast, the data collected by the Ugandan Bureau of statistics estimated the Gini at 39.5 in 2012/13 – in inequality terms, a big difference. There are many more examples where a data source or indicator can change an estimate. And that’s the point: inequality measures are estimates. They’re never just true.
So who emphasizes what, and why?
An example is UN Sustainable Development Goal 10: to reduce inequality within countries. The indicator proposed is the growth of the incomes of the bottom 40% compared with the average. When incomes of the bottom 40% grow father than the average, the incomes of the poorest catch up with the average and relative inequality falls. Last year the World Bank found that the bottom 40% needed to see their incomes grow by at least 2% faster than the average in order to eliminate poverty by 2030 as per data collected by the Bank.
The emphasis on the lifting of the poor sounds good, doesn’t it? But the indicator doesn’t include any measure of the gains made at the top of the distribution. The measure also ignores starting levels of inequality, not distinguishing between high and low (initial) inequality countries.
Again, no measure in and of itself is the right one. We might call attention to these details because of Oxfam preferences, and not because of objective shortcomings. Rather, all measures are entangled with specific narratives. Why are we so enslaved to the Gini index as opposed to more intuitive measures? Do we care to know about the absolute distance that separates us from the top? How do we advance the struggle to better capture the top tail of the distribution?
There is nothing false about any measurement facet. Rather, recognising what is behind indicators, both in terms of data and the intentions of the communicator is essential to understanding both the economics of inequality and the politics of it.