Phenomenal Cosmic Powers
or, how the Gini Index has become our singular moral compass and why that’s really not a good thing.
“The distribution of wealth is one of today’s most widely discussed and controversial issues”
This is how Piketty’s ‘Capital in the 21st Century’ starts, and if the distribution of wealth is the issue, then the Gini Index is the measure.
The UN, WHO and World Bank (and most national governments) use the Gini Index as their primary measure of inequality. In addition, in the award-winning and highly influential book, The Spirit Level, the index is used to assess inequality against a range of social ills (Wilson and Pickett, 2009: 32).
You may not have even heard of it before. Still, the Gini index has become a powerful tool to describe and direct policy on inequality.
With all the decisions and tracking pinned to the Gini Index and the readiness of its use, there must continue to be a caution: here be dragons. The index is just a model and, George Box famously warned us, “all models are wrong”. Yet it is “widely used as [the primary] proxy measure for distributive injustice in both academic and popular discussion” (Everett and Everett, 2015: 188). Still, as we shall see, we’re asking a lot of one number. As the real Genie said — we’ve put “Phenomenal Cosmic Powers in an Itty Bitty Living Space”.
The belief in those ‘phenomenal’ powers can lead us astray, especially when the index has major misgivings. So let’s go through those and see if we can humble the Gini somewhat. Specifically, we will find that the index has five major pitfalls we should always look out for:
- It is, as all models, a vast simplification in which we lose vital context
- The same index value can describe very different distributions
- It treats sections of the distribution with different weight
- Its use can lead morally neutral circumstances to be seen as moral
- There is little consensus on what is a good and bad index value
So first, let’s recap quickly. What is the Gini Index, and what can and can’t our little cosmic friend do?
Named after its inventor, the Italian statistician, fascist and eugenist, Corrado Gini. The index summarises the Lorenz Curve (shown below) to a simple ratio on a scale of 0 to 1. This ratio describes the divergence of an actual distribution (the solid line) from a theoretically perfect one (the dashed line). When the index returns a 1, then one person owns everything, and when it returns a 0, everyone owns the exact same.
To illustrate, assume a society of 100,000 people with a total income of $10m. We start counting them one by one, adding up their income as they pass us. Receiving an equal amount, each new person we counted (x-axis) would increase the cumulative income (y-axis) by the same amount, or y = x. In our example, each person would have $100 or 0.001% of total income [($10m/100,000)/$10,000,000] giving us the dashed ‘Line of equality’ below.
In reality, we know this isn’t what happens, but it gives us a baseline to measure the divergence of the actual distribution. To create the actual distribution of income (the solid ‘Lorenz curve’, shown above), we rank individuals from lowest to highest-paid and start our counting and summing again. Due to the ranking, each subsequent person either keeps the curve straight (no change in income) or curls the curve upwards (an increase in income). The steeper the curl, the more the last few people in line are hoarding.
To go back to our example, our first person (the poorest) may only receive $2 and the last (the richest) may receive $2,000. Both of which are lower and higher, respectively, than our theoretically equal society where everyone received $100 equally.
When we plot these theoretical and actual scenarios, the Lorenz curve lets us visually and quickly represent what percentage of the population (x-axis) owns what percentage of the income (y-axis).
Here is where we go back to our Italian Fascist, Corrado Gini, and his Gini Index. Corrado noticed that if we divided the area between the line of ‘Perfect equality’ and the ‘Lorenz curve’ (shown as ‘A’) and divided that by the whole area under the line of ‘Perfect equality’ (shown as A + B). We would have a simple proportion known as the Gini Index (Sitthiyot and Holasut, 2020: 2).
In a way, treating us as units, the Lorenz Curve stacks us from poorest to richest on top of each other. For example, shown below, we can derive a simple Lorenz curve from UK income deciles stacked on top of each other.
We can start to see the Lorenz Curve as graphically representing another image we all know and the Gini and summarising it.
There, all caught up. Now that we’re on the same page, a quick point on its popularity as a measure.
“The liberty of the strong, whether their strength is physical or economic, must be restrained.” — Isaiah Berlin
Everyone knows inequality (except apparently Economists) is a current and immediate problem. Policies, protests, conferences, books, papers, headlines, and bars are filled with the topic: politicians partying while people can’t pay rent; banks bailed-out in crisis but people left in theirs; tax evasion by the wealthy; the “1%” on one side and the starving, poverty-stricken and oppressed on the other.
It is within this maelstrom of complexity and discussion where the Gini Index derives so much power. In one number between 0 and 1, everyone can speak the same language. This singular focus is a relatively new phenomenon. As Noble Prize winner, Angus Deaton, and Princeton University Professor, Anne Case, point out:
“In the early days of the 21st century, [the Gini Index] was merely one of several measures of statistical dispersion that specialist researchers would use, among other things, to gauge the gaps between incomes. But after the financial crisis this single number suddenly loomed large in blockbuster state-of-society books, such as The Spirit Level, and in seminars in the Obama White House, where it was used to show that America’s problem with social mobility was matched by its problem with inequality.”
And so it has been ever since. There have been notable amendments, newly proposed statistics (Sitthiyot and Holasut, 2020), and stratifications (Piketty and Goldhammer, 2014). However, none have managed to move the needle away from the all-seeing Gini.
For all the advantages of having one metric, which is easily understandable and comparable, we must know the limitations. We must guard against giving too much power and to an ‘itty bitty space’.
The Pitfalls (or, The 5 Squirtels)
Simple at Heart
“All Models are Wrong, Some Models are Useful” — George Box
The first pitfall is the weakness and strength of all statistic models. They, and other models, are wrong by creation. Their creation relies on allowing errors to produce simplification and, therefore, usability. For all an obvious point, it’s worth re-noting when we see one statistic portrayed as the holy grail.
We shouldn’t expect anything more. This is not a failure of statistics but of humans “to expect one summary measure to capture the features of an entire distribution” (Gastwirth, 2017: 9). Yet, since the index influences policy and discourse, we must keep this in mind and interpret it with care.
As we have seen, using the Lorenz Curve, the Gini Index is held up as summarising the whole distribution against perfect equality. But ‘summarising’ is the opposite of ‘elaboration’, the price of which is we lose a lot of information.
Yet we still act as if the Gini index can tell us “all that a distribution can tell us about the inequality between the bottom and the middle of the hierarchy as well as between the middle and the top or between the top and the very top.” (Piketty and Goldhammer, 2014: 266). This is plainly not true. It’s an average.
Forgetting this risks “following the white rabbit” of our model where we claim the index is showing us something deeper than it can. This matters, for if we want to explore exactly how much the 1% actually have, or if the poorest have made gains, the index cannot by itself tell us this. Within the index, the 1% is as well hidden as the most destitute.
Remember, the index is only used to measure wealth and no other aspect of inequality. We all know that there are many features of inequality other than income: education, health and happiness (Luptáčik and Nežinský, 2020: 562). None of which are held up as high as income as they are not as easily quantifiable. Yet, there are situations in which, if contextualised with other features, we may not have such a problem with inequality of wealth. For example, if everyone was fed and free, would it matter so much if some had more than others?
In a similar but more narrow vein, the index collapses down different forms of wealth, such as employment, rent, capital etc. This produces yet another problem for us. As pointed out by Piketty and Goldhammer: “Gini coefficients and other synthetic indices tend to confuse inequality in regard to labour with inequality in regard to capital, even though the economic mechanisms at work, as well as the normative justifications of inequality, are very different in the two cases.” (Piketty and Goldhammer, 2014: 266). This, in turn, makes it “so that it is impossible to distinguish clearly among the multiple dimensions of [income] inequality and the various mechanisms at work.” (Ibid: 243).
In short the index is a model in which we pick one aspect of inequality, wealth, and then aggregate several types of wealth into one distribution before collapsing it all down into one number. Why does this matter? Well it matters because for all we can summarise inequality so we can talk about it using the index, we need to act not talk. The index sterilises any pointing of a finger or direction of policy as we lose the ability to say who, what and why. Using the index you cannot say “in the last 5 years the 1% has taken an extra 15% of all rental income” you can only say “wealth has become more concentrated, on average” which leaves us in inertia.
Same-Same but Different
“Data are just summaries of thousands of stories — tell a few of those stories to help make the data meaningful.” — Chip and Dan Heath
Some commentators highlight that the index gives the same or similar figures for different distributions. They point out that the Lorenz curve, from which the index is derived, can look very different but give the same overall index. De Maio (2007: 850) states that knowing the index can produce “differing patterns of income distribution, but nevertheless resulting in very similar Gini coefficient” is, in fact, its main weakness.
Sitthiyot and Holasut (2020: 3) give an excellent example of this, worth repeating here in full:
“whenever two or more countries share the same value of the Gini index… income inequality among them could be very different if taking into consideration the information on the income share held by the richest and that held by the poorest. For example, based on the data from the World Bank, in 2015, Greece and Thailand have the same Gini index (0.360) but the ratio of the income share held by the richest 10% to the income share held by the poorest 10% in Greece is 13.8 while that in Thailand is 8.9. In addition… in 2015, the United Kingdom and Israel also share the same Gini index (0.360) but the ratio of the income share of the top 10% to the income share of the bottom 10% in the United Kingdom equals 4.2, whereas that in Israel equals 5.8. That countries share the same Gini index but differ in the income gap between the richest and the poorest indicates that the Gini index alone cannot tell the difference in income inequality among countries.”
What does knowing that Israel and the U.K. have the same index tell us? Certainly not that the two countries have the same income distribution or that income shares concentrate in the same places (i.e. the top 20%, 10% or 0.01%).
This problem finds its roots in simplification, but the actual problem itself is due to comparison of that simplification. Also known as Anscombe’s quartet and more modernly as ‘Datasaurus’, shown below. Quite colourful patterns can have the same mean, deviations and correlations. This is also true of inequality summary statistics.
As well as not being able to tell us the difference in income distribution across countries, we lose sight of the minimum and maximum incomes using the index. For example, in two societies “where [in one] annual income ranges between $1000 and $100,000… [, and the other] income ranges between $1,000,000 and $100,000,000… the Gini coefficient would be identical for these two societies since it measures only relative differences in income or wealth” (Everett and Everett, 2015: 189).
OK fine, I hear you say, we can’t directly compare but surely having a lower index mean a more equal distribution? Unfortunately, not necessarily:
“The fact that the Gini coefficient is lower does not imply that the Lorenz curve is everywhere higher [or closer to equality]: the curves may intersect. The Gini coefficient can [therefore]… be described in terms of the mean difference.” (Atkinson and Bourguignon, 2015: xiii).
Take the below of which all four panels represents different income distributions. The top and bottoms have the same index values and the left values are lower (more equal) than the right. We can clearly see from this that in the top-right the bottom 30% of society receive around 15% of the total income. Whereas in the bottom-left (with a lower Gini value) they receive 8% or nearly half the total wealth of the other distribution with a higher index value.
We must therefore take care in our interpretation of the same index value and an index that is lower (Sitthiyot and Holasut, 2020: 2). Specifically, a lower index could be averaging comparatively lower and higher parts of the distribution to produce a value we think of as “more equal” when it is in fact anything but.
We need to ‘tell a few more stories’. Academics accomplish this using the many other statistics which are available to us. Palma (2011) and Dorling (2014), for instance, use the “ratio the share of total income of a fixed percent, for example, 10 or 20, at the upper end of the distribution to the share of income of a fixed percent, for example, 40 or 20" (Gastwirth, 2017: 2). This would guard us against some of the loss of contextual data in each respective distribution due to its simplification.
There is, of course, another way to “tell a few of those stories”, and that is, obviously, to actually tell people’s stories. In a finance lecture, I was once told that “a profit and loss statement cannot tell you if there is a rat in the kitchen”, and the same applies here. What good is knowing if there is more equality if there are rampant social ills? We could find the same index across countries, but this could mean very little in relation to their actual distribution. Although the index allows us to compare, and this is a good thing, it becomes sterile without this comparison being framed in context. Leading us to pitfalls such as thinking a lower index is always better.
Where does the Index Concentrate
“The eye of wealth is elevated towards higher stations, and seldom descends to examine the actions of those who are placed below the level of its notice.” — Samuel Johnson
We can only assess what we focus on. The same is true of statistics. So we need to know where the Gini Index focuses to use and interpret it accordingly. It has been challenged routinely for being “least sensitive to transfers among the very rich and very poor.” (Gastwirth, 2017: 1). This causes problems when our discussion centres so much of the very wealthy.
The index is subject to the gravitational pull of certain factors, much like the mean average is subject to outliers. Gastwirth contends there are two types of gravitational pulls on the index due to modern income distributions: fat, populous middles, and long, less populous tails. Transfer effects differ within this fat middle or between the tails (ibid: 7). Within these types, transfers can preserve or change the relative order of the donor and recipient.
This can lead to some counterintuitive results. When the Gini index value changes the effect will vary “whether it preserves or changes the order or ranks of the households, the difference in the ranks of the donor and recipient of a transfer, and who receives the additional income.” (ibid: 2)
For example, small transfers to the lower tail of the distribution decrease the index, which is intuitive. But suppose the same transfer goes from the middle to the top. Then, even if the transfer changes an individual rank, it is given less weight than if it were given to either tail. Given the increased discussion of ‘the 1%’, we can see how this fact could make the index inappropriate to assess this phenomenon.
We can see this below. We take the UK 2021 decile data and manipulate it to transfer 10% between deciles to record the changes to the index. In the last column to the right we see what would happen to the index if each decile individually sent the top 10% of society, 10% of their respective incomes. We would expect that the index would continually increase (i.e. that wealth was becoming more concentrated) but that’s not what we see.
Up to deciles 1–4 our index is increasing (to 0.3420) but once we hit our 5th decile the index starts decreasing and keeps decreasing until we reach the actual index for the UK in 2021 (0.3339).
As Gatswirth notes, the index “cannot capture all the changes in the income distribution that economists or policy makers are interested in” (ibid: 9). This is not only a limitation of the index it is a solid argument to assess the underlying data and not just assume that a change in the index has (or has not) produced more overall equality.
What we should take very seriously is that the index will not always signal changes which we would assume increase or decrease inequality:
“The transfers of primary interest in economics obey the Pigou–Dalton criteria (Thon and Wallace 2004), which states that transfers from a poorer to a richer household increase inequality, while transfers from a richer household to a poorer one… decreases inequality. Although [as shown]… in a few illustrative examples this principle will not hold.” (ibid: 9)
The Index as a Moral Compass
“Statistics are human beings with the tears wiped off.” — Paul Brodeur
Max Lorenz knew that his work on inequality could not by itself make value judgements. He states in his 1905 work, “There may be wide difference of opinion as to the significance of a very unequal distribution of wealth, but there can be no doubt as to the importance of knowing whether the present distribution is becoming more”.
Yet, the index’s use swings between mathematically unequal and morally unfair very quickly, and it is to this subject we now turn. Before we start, we won’t be saying we cannot form value judgements from statistical facts (by our very nature, we cannot escape this leap, facts have meaning). Instead, we show how much the index can be explained by factors other than those which lead to them being ‘morally unfair’.
By its very nature, a population contains people with different characteristics. These characteristics produce a very skewed view of the world when aggregated in summaries such as the index. We must also note that some of these characteristics are fluid in time whereas the index is a ‘snapshot’.
Take age, for instance. Some of our population will be working whereas, some will be retired. If we accept that as we near retirement, we are at the height of our earning potential, taking a lower income in retirement. We can end up in a situation where the index would give us high inequality figures that reflect this fact more than an ‘unfair’ distribution. Everett and Everett (2015) highlight this by modelling an equal society in which everybody earns, promotes, saves and lives equally with no inheritance:
“First, looking at income, we see that salary increases from $50,000 at 21 years of age exponentially to $129,813 at the retirement age of 65, at which point an income of $41,852 from annuities kicks in until death at the age of 80. This produces a Gini of .208 for income (using the usual, static OECD calculation). When we look at wealth instead of income, the inequality is considerably greater (as it is in all actual Western societies), with a Gini of .400, resulting from older workers’ accumulation of compounded salary increases and interest on regular savings over time.” (Everett and Everett, 2015: 192)
This fact does not mean that an index of 0.4 is fair in the real world. On the contrary, it shows us that an index of 0.4 could originate from a morally neutral position based on one primary characteristic of all societies, i.e. age.
We can look at this another way. Suppose you were this society’s technocratic government and wanted to reduce income inequality instead of dealing with any fundamental cause. In that case, you could force people to retire five years later, at 70 instead of 65. In this case, “we would see the income Gini drop from .208 to .165, largely because there would be fewer ‘poor’ retirees with relatively small incomes” (Everett and Everett, 2015: 193).
The above shows us the Lorenz curves for the U.K. population in 2021 but this time split by non-retired and retired individuals. As we can see, the Lorenz income curve for the retired group is lower across the population than for the non-retired. The Gini Index is also lower (3.52%).
In this way, the abstraction of the index could force us to jump to conclusions that are not warranted. For example, increases in inequality due to malicious income concentration instead of an ageing population.
Piketty and Goldhammer (2014: 267) warn us that the index produces “an abstract and sterile view of inequality, which makes it difficult for people to grasp their position in the contemporary hierarchy (always a useful exercise, particularly when one belongs to the upper centiles of the distribution and tends to forget it, as is often the case with economists).” But we can also be sent on wild goose chases when we use the index, as we do, as a moral compass. We can end up being led towards a ‘datasaurus’ instead of where we should be being led, which is towards concentrations of wealth or people who need support.
In this way, by themselves, at least “Gini coefficients ought not to be used as proxy measures of distributive injustice.” (Everett and Everett, 2015: 202). Instead, we must create policy and our frame of reference on who holds what in a far more granular and targeted way to create the necessary weight of argument. This is where the ‘1%’ metric (the share of income held by the top 1% of earners) has played a vital role. This metric gives us a window into the Gini Index’s hidden tail and allows this to lie at the feet of a definitive group. Compare this to finding a 0.7 index and we can see what Piketty means by being ‘abstracted and sterilised’ into inaction.
What Does Good Look Like?
“If all the statisticians in the world were laid head to toe, they wouldn’t be able to reach a conclusion” — Anon
Just what is a ‘normal’ index value? What value is a practical aim? 0 (perfect equality), or would 0.15 be OK? Does history inform us of what we have reasonably been able to achieve? When do we have to start worrying, 0.4 or 0.85?
We should, as social scientists, have something to say about what good looks like. Attempts to answer this question in reference to the Gini index almost exclusively rely on history to establish the ‘top’ of the index, i.e. how far have we managed to push equality in past known egalitarian societies. As an example, Luebker (2010: 4) concentrates on the close-to-egalitarian city of Teotihuacan, 100 B.C. to A.D. 600 (which Graeber and Wengrow recently relied heavily on in ‘A Dawn of Everything’):
“The city is particularly interesting from a social perspective for what may have been one of the earliest examples of state-subsidized housing in the ancient New World… Our results… yield a Gini value of .12 for the households of Teotihuacan, indicating a lack of concentration of wealth in the ancient city. The high-status houses were only slightly larger than the average apartment compound, and the city lacks a huge royal palace. As for low- status residences, these structures — and their area per household — are much smaller than apartment compounds, but the difference is not sufficient to generate a high level of wealth concentration.” (Smith et al., 2014: 319–20)
More recent examples exist which show the modern limits:
“It comes as no surprise that no [current] country has a Gini coefficient close to zero or one, the theoretical limits of the measure. To find genuinely low levels of inequality, one has to resort to historical examples from the Slovak Republic (0.19), Sweden (0.20), the Czech Republic (0.21) and Finland (0.21) that all date back to the 1980s and early 1990s (inequality in these countries has since increased).”
These examples are informative. Firstly, we should repeat, that very similar index values (0.12 for Teotihuacan and 0.19 for the Slovak Republic) can be generated for very different societies and should be interpreted with care and within a broader analysis. Nevertheless, these examples give us a semblance of what the ‘top’ looks like for normal variation within the index and gets us some way to manage our expectations.
As for what normal looks like, there is a rough and broad agreement on the index’s normal modern ranges. Wilson and Pickett (2009: 32) state the “most common values tend to be between 0.3 and 0.5”. Piketty and Goldhammer (2014: 266) tell us that “the Gini coefficient varies from roughly 0.2 to 0.4 in the distributions of labor income observed in actual societies, from 0.6 to 0.9 for observed distributions of capital ownership, and from 0.3 to 0.5 for total income inequality.”
None of this (historic or modern) gives us a real view of what we should be aiming for though, only what was or is. Helpfully, Luebkar (2010: 4) tries to to give us an answer through applicable thresholds with examples.
This is not a perfect framework. For example, according to anthropology, “egalitarian communities have Gini coefficients of less than 0.40” (Smith, 1991: 389). Also, we have seen that most countries range between 0.3 and 0.5 or 0.2 to 0.4, dependent on the source. Does this mean that most modern nations are egalitarian? You’d be hard to find anyone who would back this claim. Nevertheless, Luebkar gives us a framework which we can use to direct further analysis.
Concluding Remarks
“The emphasis should be on eliminating poverty” — Feldstein
We’ve tried here to examine if the Gini Index is, in reality, a “Phenomenal Cosmic Power in an Itty Bitty Living Space” as the real Genie says. Instead, we’ve found (probably unsurprisingly) that it has been imbued with powers way beyond its limitations.
The index is not any more right nor useful than any other statistic. However, it was adopted as it had advantages over its predecessors (class and skillset income tables of absolute values). This was because of its ability to aggregate, standardised, and be bounded (Sitthiyot and Holasut, 2020: 3), making it extremely useful for comparisons.
It can be applied to different facets of inequality to show the distribution “irrespective of the form the wealth takes: landownership, financial assets, net worth, or animal ownership.” (Smith, 1991: 370). It can also be applied to different steps in income distribution, such as pre-tax and post-tax income. This gives us the ability to assess the changes taxes and benefits make to the distribution. These are its main advantages.
As we have shown though, there are multiple pitfalls in using or interpreting the index. Pitfalls that we will undoubtedly fall into without context or a broader analysis. This was, in fact, Corrado Gini’s original intention. Gini never gave priority to his models. Instead, “He confronted reality by trying to understand the logic behind the events and only after he used those statistical and mathematical tools strictly necessary to make a model” (Giorgi, 2011: 15). Gini (for all his failures) knew that his area of study was not mathematics. It was the situation. His models could not point us to the right feet to lay blame. They could only support an argument.
Modern economists have partly remembered this fact but replaced the ‘human context’ (the actual situation) with other models. According to Frank Cowell, an economist at the London School of Economics, “One is to look beyond the Gini as a single statistic. The other is to consider whether it might be useful to use a model of the upper tail of the distribution, so you get a clearer picture.” This advice points to other models, of which “there are well over 50” (Sitthiyot and Holasut, 2020: 2). So which do we use? And would this provide the practical direction given what we’ve seen about amoral determinants?
The first question has multiple answers. These answers are based on the analysis target and the analyst’s preference. This will differ depending on the target: poverty, class differences, labour versus capital, or the 1% etc. remembering “the way one tries to measure inequality is never neutral.” (Piketty and Goldhammer, 2014: 270).
Some have proposed composite indexes incorporating the Gini index and the income share of the top and bottom 10% (Sitthiyot and Holasut, 2020: 3). Others have pushed “Government agencies [to] publish the Gini index along with the mean, median, and quintiles [and] include the deciles and 95th and 99th percentiles of the distribution and the shares (or average income) of each decile and the top 5 and 1 percent” (Gastwirth, 2017: 10; Piketty and Goldhammer, 2014: 269). Some point to decile ratios (De Maio, 2007: 850). Piketty has also recently argued that we should revert to how we used to measure income inequality in “social tables” during the enlightenment (ibid: 269).
These give us vastly more context, as shown below. Though, in my personal opinion, there is any replacing simply showing the Lorenz curve along with the Gini summary statistics. Failing that, we need to at least show these other summaries to contextualise our average.
What is surprising is that in “trying to tell more of those stories”, nobody seems to consider the stories of the homelessness, food bank users or individuals in despair. No amount of statistics can show this.
To answer the second question, even if we devised new composite indexes, published more granular data or went back to a historical method of measuring inequality, would this give us clear policy direction? This is the ‘Feldstein question’. Feldstein steps back from the index’s abstraction (ironically following its creator’s lead) and asks what we are trying to achieve. His answer inverts the thinking in which indices would point us:
“To the extent that distributional concerns motivate the design of social insurance, the emphasis should be on eliminating poverty and not on the overall distribution of income or the general extent of inequality. Like most economists, I accept the Pareto principle that an economy is better off if someone gains and no one loses. This is true even if the gainer has above-average income, causing a Gini coefficient measure of income distribution to shift to greater inequality.” (Feldstein, 2005: 12)
And surely this is true. Less inequality is not our aim (especially if that’s directed by an index which can show greater equality while wealth in concentrating). Our aim should be to help those that need it. To curtail vast concentrations of wealth not because they are, in and of itself, but because others need support. We cannot make a statistic our moral compass in this.
This is where we leave it. We have shown that the Gini Index is a useful tool but vastly imbued with powers unwarranted to its station. Its creator knew the index produced a “simplification of reality” (Giorgi, 2011: 15) and needed to be used appropriately. Piketty recently took up this charge, knowing the index “gives a simplistic, overly optimistic, and difficult-to-interpret picture of what is really going on” (Piketty and Goldhammer, 2014: 367).
The key here, I feel, is that using statistics (even when measuring inequality) should be post-politics, not pre. We should define what our political aims are first. Statistics cannot determine these for us. If we aim to eliminate poverty, then let’s do that. If we desire to curtail the use of income to produce undemocratic power structures, let’s change that. The index and a host of others can be used to highlight the problem and verify what we define as the problem. Still, they cannot define the problem, and more so, they cannot show us what to do. We probably already know.
All code for the above graphics can be found, here: https://github.com/NearAndDistant/ruk/tree/main/projects/lorenz_gini
Bibliography
⁃ Anand, S. and Segal, P. (2008) ‘What Do We Know about Global Income Inequality?’, Journal of Economic Literature, 46(1), pp. 57–94.
⁃ De Maio, F.G. (2007) ‘Income inequality measures’, Journal of Epidemiology and Community Health (1979-), 61(10), pp. 849–852.
⁃ Everett, T.J. and Everett, B.M. (2015) ‘Justice and Gini coefficients’, Politics, Philosophy & Economics, 14(2), pp. 187–208.
⁃ Feldstein, M. (2005) ‘Rethinking Social Insurance’, The American Economic Review, 95, pp. 1–24.
⁃ Gastwirth, J.L. (2017) ‘Is the Gini Index of Inequality Overly Sensitive to Changes in the Middle of the Income Distribution?’, Statistics and Public Policy, 4(1), pp. 1–11.
⁃ Giorgi, G.M. (2011) ‘Corrado Gini: the man and the scientist’, METRON, 69(1), pp. 1–28.
⁃ Lamb, E. (no date) Ask Gini: How to Measure Inequality, Scientific American.
⁃ Lerman, I. (1984) ‘A Note of the Calculation and interpretation of the Gini Index’, p. 6.
⁃ Lorenz, M.O. (1905) ‘Methods of Measuring the Concentration of Wealth’, Publications of the American Statistical Association, 9(70), pp. 209– 219.
⁃ Luebker, M. (2010) ‘TRAVAIL Policy Brief №3’, p. 8.
⁃ Luptáčik, M. and Nežinský, E. (2020) ‘Measuring income inequalities beyond the Gini coefficient’, Central European Journal of Operations Research, 28(2), pp. 561–578.
⁃ Nolan, B. (2001) ‘Inequality: The Price of Prosperity?’, Studies: An Irish Quarterly Review, 90(357), pp. 29–37.
⁃ Pyatt, G. (1976) ‘On the Interpretation and Disaggregation of Gini Coefficients’, The Economic Journal, 86(342), pp. 243–255.
⁃ Sitthiyot, T. and Holasut, K. (2020) ‘A simple method for measuring inequality’, Palgrave Communications, 6(1), p. 112.
⁃ Smith, C.L. (1991) ‘Measures and Meaning in Comparisons of Wealth Equlality’, Social Indicators Research, 24(4), pp. 367–392.
⁃ Smith, M.E. et al. (2014) ‘Quantitative Measures of Wealth Inequality in Ancient Central Mexican Communities’, Advances in Archaeological Practice, 2(4), pp. 311–323.
