Musings on Krugman’s Musings on Inequality and Growth

Paul Krugman is skeptical about the alleged negative effect of inequality on growth

there have been a number of studies that seem to find a negative relationship, all based on some kind of international cross-section approach (some with time-series aspects too). So what is my problem? In general, I have doubts about the whole growth regression methodology, which has lots of problems in identifying causation (remember, that’s the methodology behind the Reinhart-Rogoff debt-threshold paper). Beyond that, there just isn’t a striking, simple relationship between inequality and growth; all the results depend on doing fairly elaborate data massaging, which might be right but might also be teasing out a relationship that isn’t really there.

He then presents Krugmanmetrics — a simple scatter of data from a small set of rich countries. The scatter shows almost exactly zero correlation.

My comment: Ah a post on a topic which I know something about.

First, a warning to a possibly careless reader, don’t reject the alternative.

Second, the sample is much smaller than the one usually analysed. The literature generally considers all countries for which data on inequality are available (note I didn’t say “high quality and comparable”). There is an advantage of looking at otherwise similar countries with extremely different GINIs, but the tiny sample means tiny power.

I recall statistically significan negative simple correlations between inequality and growth — it isn’t always true that the data have to be massaged.

I think it is important that the alleged stylized fact isn’t reported only by lefties — at all.

OK now how about causation. I think one hint in the data (which wouldn’t show up in the sample analysed here) is that extremely surprisingly high enrollment in primary and secondary school in poor countries is correlated with low inequality and subsequent rapid growth (look for cites of Roberto Perrotti).

Extreme cases of high enrollment, low inequality and high growth are the usual suspects — Japan and the four little tigers. Here it is interesting that dictators who delivered low inequality and high growth were not at all perceived to be leftists. Chiang Kai Shek was beloved by the (now painfully moderate) Bush Sr so much that he shouts £unleash Chiang” when serving at tennis (his son Jeb demonstrated that he doesn’t know who Chiang was — clearly presidential timber). He was despised by the international left, the international center left and the international center. Another hero of equality and growth was late dictator of South Korea Park Chung Hee, again a reactionary hated by the center and left (definitely including me, but not, it seems, most South Koreans who have elected his daughter Park Geun hye President).

This would tend to suggest that it isn’t true that anything which reduces inequality increases growth. There is a large literature seeking and failing to find an association one way or the other between high taxes and growth (remember don’t reject the alternative). I have no idea about the association between strong trade unions and growth.

Finally a whole lot of the raw correlation is due to Latin America in the 80s. They had extreme inequality and extreme suffering during the 1980s debt crisis. Was this the business cycle messing up the regression or the causal connection ?