It's a loaded statement for 140 characters or less, and I was left pondering whether the Dalai Lama is correct. Certainly it's easy to accept that poverty — a lack of being able to afford basic necessities — can cause suffering. But it's not clear to me that economic inequality — a big variance in individual incomes — also causes suffering. I decided to try and test the Dalai Lama's theory.
A common way to measure economic inequality is with the Gini coefficient. A coefficient of zero indicates complete equality, whereas coefficients close to 100 indicate high levels of inequality. The CIA World Factbook provides a list of Gini coefficients by country for household income. It also provides data on GDP per capita, which is a half-decent measure for suffering (or lack thereof). We'd expect that the higher a country's GDP per capita, the less suffering its people experience.
I combined the Gini coefficient data with GDP per capita data to create a data set for 134 countries. Plotting the data, there doesn't appear to be any obvious link between a country's Gini coefficient and its GDP per capita.
However, when I run a simple regression using the data, it turns out there is a statistically significant correlation. Each one-point increase in a country's Gini coefficient can be expected to reduce its GDP per capita by $603.
I did another regression, replacing GDP per capita with life expectancy (we'd expect people with shorter life expectancies to suffer more). It again appears that more income inequality is correlated with shorter lives. I observe that a one-point increase in the Gini coefficient reduces life expectancy by almost half a year.
I can't say definitively that income inequality causes poverty. It may be other factors that are actually causing the results. For example, perhaps countries with high Gini coefficients tend to have corrupt governments, and it is having a corrupt government that is actually causing the lower GDP per capita or lower life expectancies. My analysis doesn't control for possibilities like that. I'd be more convinced of the Dalai Lama's statement if I could observe how countries' Gini coefficients and GDP per capita or life expectancy changed over time, since it would allow me to control for some of these unobservable differences between countries.
This is by no means a perfect test of the Dalai Lama's theory. Life expectancy and GDP per capita aren't perfect measures of suffering. And more importantly, the Dalai Lama emphasizes that it's inequality between countries that is particularly responsible for suffering; I tested inequality within countries. To test inequality between countries I'd need multiple years of data, which was not readily available.
But it does seem (much to my surprise) that the Dalai Lama's theory is consistent with what we observe in the real world.
Nice way to link economics to the real world. However, I think your analysis can be greatly improved by separating the country sample into two groups: high and low developed countries. A natural classification for the highly developed group would be OECD countries. The subtle change of dividing your sample will likely change your results. Hope you post the results!
ReplyDeleteThanks for the suggestion. The direction of the results still hold when I divide the sample into OECD and non-OECD countries, but a lot of the magnitude and statistical significance goes away when I break up the sample.
ReplyDeleteMoving up a point on the Gini coefficient scale leads to a $61 decrease in GDP per capita for non-OECD countries, but it's not statistically significant. For OECD countries, increasing the Gini coefficient by a point leads to an $894 decrease in GDP per capita, which is statistically significant at the 5% level.
For life expectancy, a one-point increase in the Gini coefficient for non-OECD countries leads to a reduction in life expectancy of 0.3 years, and it's statistically significant at the 1% level. For OECD nations, a one-point increase leads to a decrease in life expectancy of about a month, but it's not statistically significant.
Very interesting and thanks for posting the results.
ReplyDeleteThe next obvious question is about causality. Do richer countries - by virtue of factors such as institutional superiority - cause lower inequality or does the incidence of lower inequality cause higher average wealth.
If the line of causation runs even loosely in one direction, I suspect it would run from higher wealth to lower inequality. The reason is that the variables that generate economic growth also lead to more equality of opportunity and thus less inequality. These variables include sound government, strong property rights, good financial system, less corruption, social trust, higher education, physical capital formation, etc.
Looking at the line running from reduced inequality to higher growth just doesn't seem to jibe. After all, endogenous policies that attempt to reduce inequality tend to hinder economic progress. For example, increased redistribution by definition requires increased taxation which the economic literature clearly shows has negative impacts on growth.
What are your thoughts?
The Dalai Lama would disagree with you and would argue it's the lower inequality that causes higher average wealth.
ReplyDeletePersonally, I think your intuition makes more sense than the Dalai Lama's. But without panel data I think it would be difficult to test who is right.
I agree that we need panel data to test which way the causality line runs. If you had the data, what econometric technique would you use - Granger causality test, instrumentation, etc?
ReplyDeleteIn case any econ nerds read this blog post, I should point out a grave error made in an earlier comment. Essentially, I called GDP per capita "average wealth" but should have called it "average income." Indeed, I'm guilty of inappropriately distinguishing between a stock and flow variable. To all econ nerds, my bad!
I was thinking of using a fixed effects estimation, which would mitigate some of the unobservable differences between countries that stay constant over time. This would help control for some of the factors you mentioned that might be affecting the results: sound government, higher education, strong property rights, etc. I'd try to avoid instrumental variables if possible because finding good instruments for economic inequality might be tough.
ReplyDeleteBut the econometrics stuff you're asking is getting out of my comfort zone. I'd definitely defer to textbooks and the advice of applied economists and econometricians if I actually had the data in front of me and was planning to do a technically-sound analysis (as opposed to a really rough back-of-the-envelope analysis that I did with this post).
I read this tweet from the Dalai Lama (and read much agreement from others) and initially thought it was incorrect... but useful.
ReplyDeleteAs you note, there is a whole lot to unpack in his statement, the primary being the very definitions of "economic inequality" and "suffering". It may seem quite straightforward, but people will read different things into these words.
I'm also a little skeptical of the underlying premise that economic inequality causes these matters. It implies - at least to me - that economic equality is some sort of default, and only through the birth of the inequality is so much suffering brought. I am unsure that merely achieving economic equality (or having never deviated from it) would reduce suffering greatly.
Now, it's quite possible that "economic inequality" is a stand-in for poverty, and this can make sense - if you take poverty to be realized only through a comparison to the wealthy. Although I can agree that poverty causes suffering, the emphasis on relative wealth and relative poverty misses the target. Achieving relative equality, in and of itself, brings no joy to a specific individual.
Relative poverty is useful as a signal. It's a signal that there is likely suffering that could, somehow, realistically be alleviated. And it's a signal that we there are probably some underlying issues that need to be addressed. This seems like the useful takeaway from the Dalai Lama's tweet... and it seems like what you've started to do.
Of course, I haven't really thought this out a whole lot.
Thanks for taking this a step beyond what I would have thought. While I concluded from looking at the scatterplot there was no correlation, in fact there is a negative correlation. Cheers.
ReplyDelete