Debunking Richard Wilkinson
This was written to go somewhere else, so parts of it may sound weird, but here it is.
Last month TED Talks put out a video of Richard Wilkinson, Professor Emeritus of Social Epidemiology at the University of Nottingham, lecturing that income inequality causes social ills such as crime and physical and mental illness, even when controlling for average income. In other words, at any given income level, you are more likely to suffer violence or disease or other social ills if there are people with strikingly different incomes in your country. This video was referenced on The Motley Fool in an article calling for a higher inheritance tax on the “top 0.1%” -designed to prevent the accumulation of wealth and stem income inequality. For reference, please watch the video. I will debunk the assertions Professor Wilkinson makes. I do this not out of doubt of Wilkinson’s sincerity, but because I do not doubt his sincerity, and he is nevertheless wrong, meaning that many sincere persons can make the same mistakes he made, especially if they follow him into them.
There are minor problems with the lecture and the video (the notable absence of Singapore from many charts was suspicious), but I would like to focus on some major problems that require critical thought. First, Professor Wilkinson does not prove causality in his main point. For those unfamiliar with causality, this means that he identifies a problem, identifies a possible cause of that problem and shows that they correlate to one another statistically (where you find one, you are much more likely to find the other). While he does mention causality doubts at the end, and purports to put them to rest, he merely eliminates a few of the most obvious alternative causes (“average” income being one, and via secondhand sources, education and others), and he proceeds to apply his own worldview to the data as proof of causality “people know the causal links quite well in some of these outcomes.” This is hardly convincing, since he is trying to stretch what we already know (to wit, that envy, stress, and shame can have negative side effects) to fit around a lot more problems than we currently attribute to them (high crime rates, high infant mortality, so forth), and a much larger share of the problems we already see as related. To say in defense of this that “people know the causal links quite well” is no more convincing than to advocate the ban of caffeine by saying that “people know the causal links” between caffeine and sleeplessness, and sleeplessness causes poor work performance, or even lethal automobile accidents, so we must ban it now. It is the very first causal link that requires substantive causality, and lacks the weight that his lecture tries to lend to it. It remains a tenuous assertion dressed in so many numbers and charts.
As for the numbers and charts, it is very problematic that the type of averages that are used are not identified, and are likely means, rather than medians. If you want to evaluate whether income inequality, versus absolute income, is more important for well-being, you need to make sure that the ills are found in people with higher absolute incomes in areas with higher income inequality; while this is briefly touched on with one set of data, it is not fully addressed. For example, imagine you have two groups of ten persons each, and the two groups have the following income distributions (in the statistical sense, not in the sense of a central planner actually distributing things):
Clearly, there is shocking income inequality in the group on the left. Incomes are more equal in the group on the right, even though average income per person is lower ($57,500 per person versus $1,027,000 per person). If we find data showing that there are 2, 3, or 4 times as many crimes, preventable illnesses, and New Jersey-based reality shows in the group on the left, we may think that proves that income inequality causes real social harm, regardless of average income. In reality, this could be the result of everyone at $40,000 or less being 2, 3, or 4 times as likely to commit crimes or suffer preventable illnesses or become a fan of dreadful television. One way to fix (or at least mitigate) this oversight is to use median income instead of mean (sometimes called “average”) income. The median income is that above which half of the incomes in the group fall, while half fall below. So the median income for the first group is $32,500, and for the latter is $57,500, significantly better. So while this doesn’t prove that income inequality is insignificant in causing crime and illness, it does reveal another possible factor: there might simply be more total poor people in countries with higher average incomes. If this is the case, the prescription of taxing the rich to give money to the poor would presumably reduce crime and illness overall (at least in the short term), but if another means were found of increasing the incomes of the poor, it would still be effective at alleviating social ills even if it astronomically increased the incomes of the rich, contrary to the claims of Prof. Wilkinson.
Finally, the idea that income inequality can cause these social ills borders on impossibility, given that income inequality persists over time not among individuals (in the U.S.), but among statistical brackets with shifting membership. In case Democrats are tempted to think, after reading the first conclusion, that the 50% income mobility between 1996 (Clinton) and 2005 (Bush) is entirely downward, the second conclusion points out that 50% of those in the bottom quintile (20%) moved upward. Almost everyone in the U.S. has been or will have been in both the top 20% and the bottom 20% at some point in his life. So in most cases, income inequality measures not different classes, but different ages of a huge middle class. The argument may be made that these old data from 2005 do not give a complete, accurate picture of the struggles faced by the middle class and the poor since the 2006 real estate crash and the 2008 stock market crash, but presumably those who were changing their income bracket between 1996 and 2005 did not have time to imbue themselves with the social ills of intergenerational income stratification in six short years. In other words, if people and households frequently get richer and poorer relative to the rest of society, income inequality does not effectively exist, and if it has existed in the last five-and-a-half years (as it might have), it could not have imposed upon an entire generation the futility of their efforts to improve their lot in life, unless they are widely ignorant of the fact of their parents' being, or having been, in the top 20% of income earners. Even the top 1%, the villains of the Occupy Wall Street protests, has more than 50% turnover from one generation to the next. With so much fluctuation in the fortunes of individuals and families over the years, and income inequality existing more between one's present and one's past than between one and one's countryman, how can income inequality cause so much strife?