The Amazing Thing About American Inequality: How Equal The Country Is
Census has just released the figures on income inequality in the United States. Fascinating reading it makes too. Well, OK, fascinating for the data geeks like myself perhaps. For there’s an interesting little misunderstanding about these figures. There’s an awful lot of people using them to tell us how amazingly unequal the United States is. When in fact the figures show that the country is nothing out of the ordinary for an advanced industrialised nation. The problem is that people are looking at the US figures as released and comparing them to those of other countries. Something that you really cannot do given the way that the various sets of figures are collected.
Here’s an example at the New York Times: The Gini index value for the United States in 2011 was 0.475, higher than it was in 2010 at 0.469. The index rose in 20 states last year (including New York); there was no statistically significant change in the rest of the states and the District of Columbia (which, at 0.534, has a higher index value than any state). The Gini index value for New York State was 0.503, which means the state’s household incomes are about as equally distributed as those in Costa Rica, at least according to the most recent international data available. The figures they use for a comparison are here. Looking at those you might think, well, if the US is at 0.475, Sweden is at 0.23 (yes, the number of 23.0 for Sweden is the same as 0.23 in this sense) then given that a lower number indicates less inequality then Sweden is a less unequal place than the US. You would of course be correct in your assumption: but not because of these numbers. For the US number is before taxes and before benefits. The Swedish number is after all taxes and all benefits. So, the US number is what we call “market income”, or before all the things we do to shift money around from rich to poor and the Swedish number (in, fact, the numbers for all other countries) are after all the things we do to reduce inequality. You can see this here at Wikipedia. There they have both sets of numbers. Inequality by market income, before taxes and benefits, plus inequality after taxes and transfers. The US, by market income standards, is less unequal than Italy, Germany, France, about the same as Finland, a bit more than the UK or Sweden. That really isn’t the story we normally get told. Nor is it the usual story that’s being told now about these new figures. Just to emphasise this point, about how everyone usually reports entirely different figures for the US and everyone else. Here’s the Census report again. That’s the official release on inequality in the United States. Here’s a couple of sources of information about inequality in Europe. They are reporting entirely different things. The US is reporting market inequality, before the effects of taxes and benefits, the Europeans are reporting the inequality after the effect of taxes and benefits. There are two more points we might make about all of this. From the NYT again: Of all American states, New York again has the most unequal income distribution, according to a new report from the Census Bureau. Wyoming has the most equitably distributed income. Well, yes, that’s something that we would pretty much expect actually. The larger your data set the more variance you expect to have in your data set. We would rather expect to have greater income inequality between the 20 million odd in NY State than we would in the 500,000 in Wyoming. Just to change the example, think of temperature. We know that temperature varies, both through the year and year by year. The more years we measure temperatures the more extreme values we would expect to see. The same would be true of rainfall, hurricane frequency and so on. Much of the science of statistics is concerned with unraveling these effects: are our extreme values because there is some change in the underlying trend or is it just that we’ve got a larger data set? Extremely important in climate change research for example: that we’ve got extreme summer temperatures in one year might just be that we’ve been measuring summer temperatures for a couple of hundred years. That extreme has to come some time. When we see that actually recent summers are on average hotter than earlier summers then we might start to think about trends, not just the variance that comes from more data. The same would be true of income variance as a population rises. we’re not surprised that there’s more variance in NY State than Wyoming. True, we’re also not surprised because NY State has both Buffalo and Wall Street in it, so we’ve got both just more data and other good reasons to expect greater income inequality. Which brings us to the 300 million people in the US. Is it really fair to be comparing income inequality among 300 million people with inequality among the 9 million of Sweden? Quite possibly a more interesting comparison would be between the 300 million of the US and the 500 million of the European Union. Or the smaller number in the EU 15, thus leaving out the ex-communist states with their own special problems. Not that it matters all that much as the two numbers for the Gini are the same: 0.3*. Note again that this is post tax and post benefit. On this measure the US is at 0.38. So, yes, the US is indeed more unequal than Europe. But by a lot smaller margin than people generally recognise: or than by he numbers that are generally bandied about. Which brings us to the second point. Even here the US number is (marginally) over-stated. For even in the post-tax and post-benefit numbers the US is still an outlier in the statistical methods used. In looking at inequality, poverty, in the US we include the cash that poor people are given to alleviate their poverty. But we do not include the things that people are given in kind: the Medicaid, SNAP, Section 8 and so on. It’s possible (I’m not sure I’m afraid) that we don’t include the EITC either. We certainly don’t in the poverty statistics but might in the inequality. All of the other countries do include the effects of such policies. Largely because they don’t offer benefits in kind they just give the poor more money and tell them to buy it themselves. This obviously turns up in figures of how much money the poor have. Is the US a grossly unequal country? That’s a political or moral question which I cannot really help you with. But the figures that Census have released are about market incomes. By which standards the US is less unequal than Germany and France for example. So it isn’t, by this measure, a markedly unequal country. When we look at inequality after taxes and benefits then yes, the US is more unequal than many others. But if we compare to the EU, rather than an individual country in it, the difference is a lot less than normally realised. 0.3* to 0.38 or so. * I would add that I don’t believe that Gini figure for the EU for one moment, it appears far, far too low to me. But I’ll have to do more investigation to confirm or refute my prejudice on that. I have a feeling that it’s an arithmetical average of the individual country Ginis, not a calculation of the Gini over the EU population.** **In fact, now that I’ve thought about it a little more that EU figure is clearly and obviously nonsense. It absolutely is not inequality among the population of Europe. For we have an EU 15 number of 0.3 and an EU 27 number of 0.3. And we really simply cannot believe that adding the A8 (the ex communist countries) to the EU does not increase inequality across the total population. That would be absurd.