Skip to content
 

State incomes over time: the rich stay richer, the poor stay poorer (but less so than before)

When looking at voting and state income (red states and blue states, etc.), we realized that, although many of the geographic patterns of voting in the U.S. are relatively recent (for example, New England used to be strongly Republican, now is solidly Democratic), the rankings of states as rich and poor has been steady for quite awhile. The rich states now are, by and large, the states that had factories and large cities 50 or 100 years ago. Here are some graphs (using Census data that Justin Phillips gave me):

stateincometrends.png

Apparently this isn’t just a U.S. phenomenon. Diego Comin, WIlliam Easterly, and Erick Gong find that the wealth of nations was determined in 1000 B.C. Their paper is interesting and has some cool maps, but the graphical presentation could be improved a bit. I’ll offer some suggestions on the chance that they’ll have an opportunity to revise:

Figures 1,2,3: Put all the maps next to each other (in a vertical row, there’s room once you remove Antartica!) so the comparison is clearer. Also, make the latitude/longitude lines much much lighter. They’re overwhelming the plote entirely. Finally, I can’t understand what relevance the current political borders (e.g., between the U.S. and Mexico) have in 1500 A.D.

Figures 4-7: Too many countries to be readable. Break each into a few (e.g., 4 plots), breaking up by geographical regions. Then there will be space, you can spell out the country names (use caps and lower-case, not all caps) and remove the dots (just center each country name on its point). Also show numbers for per capita income, not logs (6,7,8,etc, which are uninterpretable). Displaying on log scale is fine but use interpretable labels.

Also, show a parallel-coordinate plot (like my first two plots above)!

Get rid of table 3. These are binary variables. You can just say in the text, “45% of the 113 data points in 1000 B.C. had …, 73% in the year 0, …”

Tables 4 and 5: These should be plotted as time series.

Tables 7-12. You should graph these too. I know you won’t want to, but see here for some inspiration. You’ve done the work, why not make the paper special?

Anyway, lest I seem to critical . . . I like the paper which is why I’m taking the time to make these suggestions.