| Regions/ Countries | Private % without the largest PHE sector | Private % without the two largest PHE sectors | Private % with all countries included | Private 2010 | Total 2010 | |
| Global | 32.9 | 56,722,374 | 172,546,175 | |||
| Africa (Sub-Saharan) | 16.4 | 16.4 | 17.8 | 930,016 | 5,218,120 | |
| Share of global | 1.6% | 3.0% | ||||
| Uganda | 54.6 | 103,327 | 189,396 | |||
| Ethiopia | 16.7 | 96,546 | 577,594 | |||
| Arab States | 16.4 | 14.4 | 17.4 | 1,423,630 | 8,201,861 | |
| Share of global | 2.5% | 4.8% | ||||
| Egypt | 19.9 | 436,208 | 2,192,452 | |||
| Iraq | 39.5 | 194,608 | 492,507 | |||
| Asia | 35.9 | 48.3 | 42.1 | 32,267,911 | 76,568,246 | |
| Share of global | 56.9% | 44.4% | ||||
| India | 58.3 | 12,443,748 | 21,350,427 | |||
| China | 19.6 | 4,664,531 | 23,856,345 | |||
| Commonwealth (British, Developed) | 8.3 | 11.9 | 10.1 | 318,033 | 3,162,889 | |
| Share of global | 0.6% | 1.8% | ||||
| Canada | 11.7 | 190,000 | 1,620,169 | |||
| Australia | 7.5 | 96,334 | 1,276,488 | |||
| Europe | 14.9 | 13.4 | 14.9 | 5,526,851 | 37,177,470 | |
| Share of global | 9.7% | 21.5% | ||||
| Russian Federation | 14.7 | 1,323,348 | 8,984,977 | |||
| Poland | 32.9 | 705,998 | 2,148,676 | |||
| Latin America and the Caribbean | 38.6 | 40.0 | 48.8 | 10,638,863 | 21,789,880 | |
| Share of global | 18.8% | 12.6% | ||||
| Brazil | 72.7 | 4,764,498 | 6,552,707 | |||
| Mexico | 32.3 | 918,555 |
2,847,376 |
|||
As much as the website and derivative commentary treats regions as units of analysis, the website also notes vulnerabilities of regions as coherent entities http://prophe.org/en/global-data/global-data-files/guide-to-the-prophe-dataset/. Here, as summarized in the above table, we investigate one salient question: How much do a region’s very large countries (the largest and then the two largest) skew a region’s average private share? Clearly, there is much more potential to skew a region’s private average than to skew the global private average. A first, straightforward question is therefore whether and where single countries are predominant in their regions’ private enrollment. A second is whether they have much impact on regional averages; giants whose private share is close to regional private share would largely reinforce regional tendencies whereas giants that differ sharply from regional cousins in private share would have the starkest impact on regional averages. For both questions we calculate for six regions (the US omitted for being a single-nation region) with all data shown in a single table. (Omitting the systems largest in total enrollment, rather than the largest in private enrollment, does not much affect our large country/regional analysis.)
With respect to the first question, single country enrollment predominance is limited. Canada does hold 59.7% of the Developed British Commonwealth’s very small regional private enrollment; more importantly, Brazil is next in regional predominance with 44.8% of Latin America’s private enrollment (30.1% of total enrollment). Probing further we find that even individual regions’ largest two private sectors capture only so much of regional private enrollment. Thus the absence of greater numerical predominance by large countries already limits their potential to skew greatly their region’s average private share.
Our data allow us to go past potential impact to actual impact—how much each region’s private share changes when we remove the region’s largest and then largest two private sectors. The results are mixed, the impact only very small in most regions, moderate in others. In 4 regions averages change by only 2% or less when we remove the largest private sector and they do not change much more even when excluding the two largest private sectors. Moreover, these are the 4 regions with the lowest average private shares; thus, no region is “deceptively” propped up to significant dual-sector status by the presence of a single country’s aberrantly large private sector. Nonetheless, as it is the other 2 regions (Asia and Latin America) that hold easily the largest private enrollment, they also merit a closer look. Removal of the largest private sector would lower the Asian and Latin American averages by slightly over 6% and 10% respectively. India and Brazil each move their regional averages because they are both large and have private shares far above their regions’ average. Omitting the second largest Latin American private sector (Mexico’s) has little further effect on the regional average. But additionally omitting China does significantly affect the Asian regional average. Roughly equal to India in total enrollment, the Chinese private share (19.6%) is well below Asia’s (42.1%) and even further below India’s 58.3%. Hence, whereas omitting India alone lowers the regional average (to 35.9%), omitting China alone would raise it even more (to 52.4%). Omitting China after omitting India produces a 12-point swing (to 48.3%).
