The Continuing Saga of Rwandan Poverty Data

via Ken Opalo, there is new analysis out of the 2014 Rwanda poverty numbers that contradicts official Government reports, finding that poverty actually rose between 2010 and 2014. Professor Filip Reyntjens made a similar argument at the time, which I disagreed with.  

This new (anonymous) analysis in the Review of African Political Economy supports the conclusion of Reyntjens, based on new analysis of the survey microdata (with commendably published stata code). The key difference seems to be that their analysis updates the poverty line based on prices reported in the survey microdata rather than using the official Consumer Price Index (CPI) measure of inflation.

What I took away from this at the time was the apparent fragility of trend data on poverty that depends on consumption aggregates and price data. I also drafting a follow-up blogpost that for whatever reason never got posted, so here it is. 

What else do we know about welfare in Rwanda? 

Given the disagreement about the right price indices to use for calculating the poverty line, it might be informative to look at other indicators of welfare that we might care about, and even better to look at other indicators from a different survey. In fact, it was looking at the first table in the EICV4 Report that made me doubt Filip's claim that poverty had actually increased. This table suggests that between 2010 and 2014,

- inequality was down,

- school attendance and literacy up,

- housing and access to electricity and clean water improved,

- health services improved, and

- household savings were up.

All strong indicators of progress. 

Inequality is down on two different measures (both unaffected by the level of the poverty line) and Food Production is Up



If Rwanda fiddled the poverty numbers, did they also fiddle the entire survey? A useful check is the results from the Demographic and Health Survey (DHS), which as Justin Sandefur and Amanda Glassman have pointed out, tends to have particularly heavy donor involvement, making them particularly difficult for governments to fudge. And the overall impression from the two surveys is strikingly similar – rapid improvements in child and maternal healthcare and health outcomes.

Health Indicators have substantially improved over the same period on the DHS Survey



It's certainly possible that the whole EICV4 was fudged and that consumption poverty increased whilst at the same time health care services and health outcomes improved. To me, it just seems kind of unlikely.