Date: Thu, 19 Jan 2006 13:10:09 +0000
Subject: World Bank and FAO aims and research
Dear Mr Ziegler
World Bank and FAO aims and research
I learn from the Bretton Woods Project website that you have recently made comments on failings of the World Bank.
I think you may be interested in the following. I am a private citizen who has taken an interest in the aims and methods of policies on poverty in the human species.
This email deals with errors of reasoning concerning statistics and demography:
1) the World Bank fails to adjust global estimates for food needs and
2) several Millennium Goal indicators fail to look at either birth rates or death rates.
1. The World Bank food error
Firstly, the World Bank fails to adjust for food needs as the proportion of children falls.
Its staff do not adjust its global poverty estimates according to the proportion of children.
The FAO does adjust its global hunger estimates crudely in this respect.
The global proportion of children is falling and adults need more food. The difference between the Bank and FAO methods would seem to account for at least part of the difference in reported progress between the two organisations for the Millennium Goals.
In other words, the fact that the Bank has reported satisfactory progress on poverty while the FAO has reported unsatisfactory progress on hunger (for substantially the same people who can afford little besides food) may be partly due to the fact that the Bank method spuriously generates poverty reduction as time goes by.
Clearly, one implication of the error is that other things being equal the Bank has overestimated poverty alleviation as a whole, and especially in China.
Further details are available at:
Confirmation of the World Bank error is available from the draft UN Statistics Division handbook on poverty statistics:
"[the per capita] approach is taken, for example, in calculating the widely-used $1/day and $2/day per capita poverty lines....a population experiencing a rapidly declining fertility rate....will experience an exaggerated decline in the short-term reduction in poverty when poverty is measured on a per capita basis"
(Jonathan Morduch, 2005. Concepts of poverty. Draft chapter for United Nations Statistics Division handbook on poverty measurement.
2. Demographic errors in social science
Secondly, the FAO and Bank statistics, like many Millennium Goal indicators, are influenced by other demographic changes, in addition to age structure and aggregate trends for individuals.
One aspect is that this kind of method fails to distinguish between fewer babies and more food per person. The fact that fewer babies are born (reducing the number of hungry people) is not the same as the fact that people rose out of hunger.
Another is that the statistics look better if more people die of AIDS or other causes.
This error in the theory of economics and other social sciences afflicts 21 indicators for the Millennium Goals. I have warned officials and academics of this error in relation to the aim of "poverty reduction" since 2000.
This may justifiably not be of any concern in some countries. Changes in longevity across levels of variables may often be in the same direction as the traditional conclusions by social scientists. However, the traditional way is not the correct way to approach these kinds of statistics, since the burden of proof is on the scientist to show their argument holds up.
The most obvious countries for which there may be cause for concern are where life expectancies have fallen. On this point, there are more details at:
Professor Sachs' Commission on Macroeconomics and Health did subsequently look in a limited way at this issue. It did not to my knowledge look at my suggestion to him that all international indicators commonly thought to aggregate outcomes for individuals take survival rates into account.