This is another rather incredible initiative undertaken in Vietnam which relates to Urban Information Modelling. Interestingly, this study looks at poverty levels not only on the urban scale but also on the regional and national scale. The article written about the modelling process is extremely interesting since much of the problems the developers of the program faced are similar to our own problems with formatting data.
For example, the article describes that the team originally had to account for regional differences in the data by adopting seventeen household characteristics which appeared in the earlier census data and then extrapolating them across the country. Trying to compare urban and rural populations, however, became problematic. According to the article:
“The team considered that this use of only two consumption models, one rural and one urban, for the whole country was somewhat unsatisfactory; it seems unlikely that returns to each of the rural (or urban) household characteristics would be the same across the entire country. As an alternative, the regression models included dummy variables for each of the regions to account for regional differences.” Through the inclusion of the dummy variable, however, much of the data ended up showing unrealistically large differences across the borders of two regions.
Another extrapolation made by the developers was necessary when raw data from the most recent census polls was unavailable to the public. As a result, the researchers could only purchase three-percent of similar data generated by the GSO (General Statistics Office) and extrapolate the data across the country based upon geography and income levels as a proportion of that small percentage of data. As more and more organizations grew interested in refining their data, the developers were only able allowed to purchase thirty-three percent of the data in order to show the regional differences on a much finer level.
Apart from the technique, the mapping revealed newfound estimates of poverty headcounts for provinces, districts, and communes. “The maps provide a striking visual account of the depth of poverty in the mountains to the north, the upland areas along the coast, and the Central Highlands. They show much lower levels of poverty in the Southeast (where Ho Chi Minh City is located), other lowland areas, the Mekong River Delta, and the Red River Delta.” Though to anyone familiar with the geography of Vietnam or the war in Vietnam this is not incredibly uprising, the maps suggest that there is a strong correlation between poverty and geography rather than geography and administrative boundaries (a reversal of the antiquated, wartime viewpoint).
Lastly, the map representing poverty headcounts for provinces, districts, and communes revealed the heterogeneity within provinces in many parts of the country. “Many rich provinces have several poor districts, while the reverse is true for several poor provinces. By extension, many richer districts include a large number of poor communes.” Another map which shows the poverty density across Vietnam reveals that poverty is not high in areas of greater poverty density.
Anyway, the map has an interesting narrative regarding the mapping process: https://docs.google.com/viewer?a=v&q=cache:VLjmhReZMxkJ:siteresources.worldbank.org/INTPGI/Resources/342674-1092157888460/493860-1192739384563/10412-14_p261-286.pdf+mapping+poverty+vietnam&hl=en&gl=us&pid=bl&srcid=ADGEESiSMdXvMSMXGTHgANDuetBnMBPPLy6eHUDZ303RyB4XrOhjT91490fUdy4_c7WhnLejkcOwE0udH63bBQ4zcxYPC7itr22D3FS_2DGenAW8DL2AMny_uyZDdLNCfF-QyHCQ5mAs&sig=AHIEtbQENFbE06lzvNcpBxnXKMbdZFfJ1Q