The champagne glass effect


  • Julian Champkin


It has become a classic image. It is known as the champagne glass, though its originators did not call it that. It shows, oh so clearly, the Pareto 80%–20% distribution of the wealth of the world – with perhaps a small extra message smuggled in. Julian Champkin drinks it in.

It is beautifully clear. It shows wealth and poverty, haves and have-nots. The richest 20% of the world are at the top. They own and enjoy and wallow in all that thick wide bit, of wealth, of money, of possessions: 80% of everything is theirs. The poorest 20% are at the bottom. All they have is that sad spindly narrow bit – and they have all the rest of the world weighing down on them as well. It is, in its way, a graphic of genius. Who could possibly misinterpret its message?

It first appeared, not quite in the form overleaf, more than 20 years ago in a United Nations Development Programme (UNDP) report on human development1. It has been redrawn and reproduced many times since. It is not hard to see why. It is utterly simple. It is utterly obvious. And it is utterly intuitive. It tells a story.

The story is implied in the data – but that is not really how the diagram tells it. For 99.99% of people who see it, plus or minus almost nothing, the data is irrelevant. The story is the outline. It is the shape of a glass of champagne.

We know about champagne. It is what rich people drink. And that is the power – and the problem – of this image.

Like every half-decent graphic, it does have data. It does (or can) contain numerical information (depending on whether or not axes and scales are drawn in). Most versions divide the vertical axis into quintiles (fifths to non-statisticians), but few bother to put a scale to the x-axis. The quantity measured by the x-axis can vary: the original, anonymous, authors gave it five different quantities. They put figures in the section for the rich for share of GDP; share of world trade; share of commercial lending; share of domestic savings; and share of domestic investments (see Figure 1). All of these categories come pretty close to the so-called Pareto ratio, first observed more than a century ago by the Italian economist of that name: that 80% of a nation's wealth, resources, land and riches all tend to be owned by that top 20% of its people.

Figure 1.

The graphic as it originally appeared in the UNDP report. From UNDP, Human Development Report 19921, with thanks

But for non-specialists, the casual viewer, those precisely definable quantities are over-specific. They could all be replaced, and sometimes have been, by the general term “wealth”. The numbers do not matter much either. The thing that does matter is the shape.

Champagne is about conspicuous consumption. Should a diagram about inequality of wealth remind us of it?

Because champagne is not value-free. Champagne is about wealth. More, it is about flaunting wealth. It is the symbol of conspicuous consumption. It is about hedonism. Of money spent on luxury which could be better spent on – well, on almost anything that is a little less transient than a bubble-fuelled foolishness and a head next morning. It speaks of spraying bubbly over the crowd at far too much a bottle, of luxury, of excess, of irresponsibility, of dancing while Rome burns and the hungry masses huddle outside. In anyone who has even two puritan hackles to rub together, it raises them.

If the same data had turned out the shape of a beer glass it would not have worked. The associations would have been with the solid working man's tipple: a small deserved treat at the end of a hard day. Or turn the same data on its side, with normal x- and y-axes, and it becomes simply an S-curve (Figure 2) – which carries no emotional baggage at all. But the very word “champagne” screams inequality, privilege, and the excesses that led to the French revolution. And we have not even mentioned Marie Antoinette's breasts, on which the shape of the champagne glass is alleged to have been modelled. (The story is, of course, apocryphal. Nor is it relevant to any of the above that true aficionados of champagne prefer to drink it from flute-shaped glasses, which concentrate the aromas and keep the bubbles in for longer, thus staving off the moment when the drink goes flat.)

Figure 2.

The same data represented as an S-curve, with no calls on free association of ideas

All sorts of tales can be added to it. The narrow column has to bear huge weight – so the poor are not just deprived, they also have to support all that privilege above them (which you might or might not think true, depending on what sort of economist you are: some might say that it is the rich who support the poor, through taxation and benefits. Either way it is an extra, unspoken but suggested, added on to the data.) When the glass is full it overflows. The wealth of the rich trickles out of the top of it down – perhaps – to the poor. So deeply embedded is this imagery and its undertones that even Pope Francis has used it, to elaborate his end-of-year message on inequality2: “The promise was that when the glass was full, it would overflow, benefiting the poor. But what happens instead, is that when the glass is full, it magically gets bigger: nothing ever comes out for the poor.”

The UNDP evidently realised its power. They put a stylised version of it on the cover of their report. (Figure 3).

Figure 3.

Stylised version of the graphic on the cover of the UNDP report. From UNDP, Human Development Report 1992 [Ref 1], with thanks.

Is it, then, an honest image? In the obvious sense, of course it is. No one is implying that its data is wrong. Nor is there anything hidden about it. It wears its heart on its sleeve – which may in itself be what is wrong with it.

Is it, alternatively, manipulative? You could say that it plays to subconscious emotions, in the same way that advertisements for commercial products try to do. It is not illegal for advertisers to do that. Statisticians may like to think of themselves as morally above that sort of thing, superior to advertising men, serving a purer truth. But whoever drew that graphic had to choose between half a dozen different ways of presenting their data. Should we blame them for choosing the one that has the most impact? They had their reasons. And the UNDP report had, after all, a message to convey, namely that there was too much inequality in the world.

The moral perhaps is that visualisation is an art, and art by definition comes with values added on. Any storyteller worth his salt tells his story as powerfully as he can. Those who made the champagne glass image told their story very powerfully indeed.


The trickle-down effect? Elizabeth Hachem/Digital Vision/Thinkstock


Redrawn from: Conley, D. (2008) You May Ask Yourself: An Introduction to Thinking Like a Sociologist. New York: W.W. Norton and Company, with thanks