Deinonychus antirrhopus

Inequality and Economic Growth...Again

In the comments to this post Kevin made this comment:

5. I said "doesn't appear to be any correlation," you said "researchers are not sure of the relationship." As you know, it is correct to say that there doesn't appear to be a correlation (the null hypothesis) unless there's good evidence that there *is* a correlation. Both of the papers you cited are pretty clear that the evidence is ambiguous, which in turn means that there's no good evidence for a positive correlation.

The problem is Kevin's interpretation of the articles is wrong. The earliest work by such researchers as Alesina and Roderik that the first paper reviews indicates that there is a negative relationship between growth and income inequality. That is the higher inequality the lower the rate of growth. Then later work by the such Robert Barro found a positive relationship--i.e., higher inequality resulted in higer econocmic growth.

Now, one could say that this supports Kevin's claim. There is no correlation. There are two problems with this.

First is methodological. Just because nobody has found the connection between inequality and growth does not mean there isn't one. There might not be one, but the lack of evidence for an effect is not very good evidence that there is no effect.¹

The second comes from the second paper I have linked too. This paper extends some of the previous research by generalizing the utility function so that the previous research are no special cases. The result is that the link between inequality and growth is ambigiuous. Now Kevin can still say, "See, I'm still right." But that is only due to a superficial reading of the paper. Reading the entire paper one finds that yes the result is ambiguous, but that the authors identify when inequality and growth have a negative relationship and when there is a positive relationship.

Hence, if you know which case you are in, you will know whether or not inequality is good for growth, bad for growth or has no effect (i.e., there is literally no effect). The no effect case is actually a singlton and is highly unlikely.

This is the problem when amatuers start writing about economics. There is a tendency to focus on the univariate approach, inability to read and fully understand the literature, and superficial data analysis.²

Sadly lots of people read Kevin's site and they take his writings on these topics as being sound.
_____
¹This is why statisticians are so carefull when they do empirical research. They never say, "We accept the null hypothesis." Instead, the phrasing is, "We fail to reject the null hypothesis." It is giving implicit approval to the notion that the absence of evidence is not evidence of absences.
²In my opinion economics is in one regard harder than things like physics. Watching a commet or a planet through a telescope wont do much to change the behavior of the commet or planet. However, when you start watching people their behavior can and usually does change. Economists study people, and in studying them they could be inducing people to change their behavior. It makes for a much harder nut to crack in some regards.