Measuring global poverty levels is hard, so hard the actual numbers might be of limited use. According to Angus Deaton, the godfather of poverty measurement, "it seems impossible to make statements about changes in world poverty when the ground underneath one's feet is changing in this way."  I thought it would be helpful to illuminate why.  

Technically speaking, creating a measurement of global poverty is a daunting endeavor.  It hinges on three key things: 1) household surveys 2) a single global poverty line, which enables aggregation across countries, and 3) measurements of global purchasing power parity, which is necessary to convert to a single currency in a way that measures how dollars translate into welfare.  Each of these have significant methodological challenges.

First there are challenges with the collection of micro-level data.  We measure poverty by looking at consumption, typically defining absolute poverty as the consumption level required to get enough calories with a buffer for shelter, clothes, etc. (Note: we use consumption instead of income because income is a misleading measure of welfare when families live off their own agricultural production).  This information is collected at the country level through household surveys.  Poverty counts are highly sensitive to survey design, which is extremely heterogeneous across countries.
 
Second is the creation of a global poverty line.  National poverty lines are determined locally, which then need to be converted into a single poverty line.  The way this is done is by taking a subset of poor countries (i.e. we don't want to include relative poverty like the US), converting to a single currency in purchasing power parity terms, and taking an average.  Counterintuitive things happen when the countries on that list change.  India's growth led to its exclusion from the list, which raised average used for the global poverty line, which led to an increase in India's poverty count for the global aggregate.    

And then there's determining PPP conversions, which is the really hard part.  How to determine the basket to measure what $10 gets you if you're a poor person in Ecuador vs. India?  At the same time that we want to measure apples to apples, the reality is that people eat oranges in some countries and apples in others.  Insisting on measuring apples to apples leads to problems if apples are expensive imported goods that poor people don't eat in some countries.  Comparing apples to oranges introduces subjectivity, which has its own problems.  The International Comparison Program (ICP) which does this work is always making improvements, but adjustments can create shifts in global poverty measurements.

In 2005, 500 million people were added to the poverty count.  This had little to do with changes in welfare, and a lot to do with updates to PPP and global poverty line measurements.  If you want to really geek out on this, Angus Deaton's paper "Price Indexes, Inequality, and The Measurement of World Poverty" gets into it in (unfortunately rather painful) detail.

There are also conceptual issues with global poverty measurements, which I'll turn to in my next post.