the multi-dimesional poverty index and per-capita gdp

WillyNilly wrote:i performed a brief and incomplete analysis of the relationship between the multi-dimensional poverty index (mpi) as defined by amartya sen, and the per-capita gdp (pcgdp) of twenty poor countries.

the pcgdp has served us well for decades as an indicator of standard of living. no one thinks that it's perfect, but not surprisingly, it is highly correlated with general standard of living. its wide availability is its greatest strength.

the question i was trying to answer, even in a minor way is: is the mpi highly correlated with pcgdp?

the larger question is: is the pcgdp useful enough for identifying the poorest countries of the world, when poverty is defined by a composite score based on such factors as nutritional deficiency, infant mortality, life expectancy, illiteracy, etc., as amartya sen, the poverty expert, and the u.n.d.p. would define it?

i used mpi data to which rashmun had provided a link a few days ago - rashmun, could you post that link again here? - and international monetary fund's data on pcgdp (in ppp terms or purchasing power parity terms). i selected twenty of the many poor countries in sen's list, including the ones with the lowest mpi scores, i.e. the poorest (niger, ethiopia), the least poor (argentina, brazil), and those that fell in between (india, nigeria), over the entire interval of mpis.

my quick-and-dirty analysis of the twenty countries' data shows the correlation coefficient between pcgdp and mpi to be -0.73.

the negative value is expected, because as pcgdp increases, mpi (poverty) decreases. perfectly negative correlation between pcgdp and mpi would yield a correlation coefficient of -1.

from this limited analysis, i conclude that poverty is well-correlated with per-capita gdp at a macro level, but the degree of correlation may not satisfy those who want to make a special study of poverty, its causes and consequences.

i must end this post now, but may have more to say on this later.