Thanks for putting this together - I always find your writing to be simultaneously very readable and also dense with novel ideas, which is rare to see.
As part of a contracting position with the Open Philanthropy Project, I have been looking into different metrics on the prevalence and success of democratic forms of governance over time. One piece of data that we were interested in but couldn’t find in the existing literature was the percent of GDP in democracies vs non-democracies over time. To answer this question, I combined data from Polity on categorizing regime type over time with the IMF’s data on nominal GDP. I ended up only going back to 1980, because before then I started to lose a lot of reliable GDP data for smaller countries. You can find my analysis here - in particular I recommend looking at the first two sheets that graph percent of GDP by regime type. I used Polity’s classifications for determining which countries counted as democracies, autocracies, and in between “anocracies." (Disclaimer that I did this independently, may have messed something up, and it did not undergo the sort of review that would be associated with anything formally released by Open Phil. )
Overall, the main takeaway was that, at least from 1980 to today, democracies hold a disproportionate percentage of income in comparison to their population. (Compare with Our World in Data’s graph on percent of world citizens living in a democracy over time here.) However, while the percent of nominal income in democracies was increasing from 1980 to ~2000, after that it has started to reverse, largely because of increased GDP in authoritarian China. If one looks only at “western democracies” as the authors do here, the trend for relative financial power of non-democratic governments becomes more pronounced. I don’t think these analysis rebut any of your core points - the fact that this graph only covers 38 years underscores that these trends could easily be historical anomalies that disappear in the long run - but hopefully it adds an additional metric to use in thinking about the future prospects of democratic government.
Hey Ben,
Thanks for putting this together - I always find your writing to be simultaneously very readable and also dense with novel ideas, which is rare to see.
As part of a contracting position with the Open Philanthropy Project, I have been looking into different metrics on the prevalence and success of democratic forms of governance over time. One piece of data that we were interested in but couldn’t find in the existing literature was the percent of GDP in democracies vs non-democracies over time. To answer this question, I combined data from Polity on categorizing regime type over time with the IMF’s data on nominal GDP. I ended up only going back to 1980, because before then I started to lose a lot of reliable GDP data for smaller countries. You can find my analysis here - in particular I recommend looking at the first two sheets that graph percent of GDP by regime type. I used Polity’s classifications for determining which countries counted as democracies, autocracies, and in between “anocracies." (Disclaimer that I did this independently, may have messed something up, and it did not undergo the sort of review that would be associated with anything formally released by Open Phil. )
Overall, the main takeaway was that, at least from 1980 to today, democracies hold a disproportionate percentage of income in comparison to their population. (Compare with Our World in Data’s graph on percent of world citizens living in a democracy over time here.) However, while the percent of nominal income in democracies was increasing from 1980 to ~2000, after that it has started to reverse, largely because of increased GDP in authoritarian China. If one looks only at “western democracies” as the authors do here, the trend for relative financial power of non-democratic governments becomes more pronounced. I don’t think these analysis rebut any of your core points - the fact that this graph only covers 38 years underscores that these trends could easily be historical anomalies that disappear in the long run - but hopefully it adds an additional metric to use in thinking about the future prospects of democratic government.