Branko Milanovic wrote: "...thanks to a database of household surveys put together recently by the World Bank, we can actually find out for the first time, from a single and consistent data source, who the real winners and losers of globalization are." Matt Berkley: As a non-economist, perhaps I am failing to understand your thinking. The data omit: a) analysis of what people need to spend, b) capital gains or losses, c) changes in debt levels, d) non-traded benefits such as health or education services or environmental factors, e) spending on things which may be neutral or harmful such as overuse of drugs (tobacco, alcohol), f) spending on things which may represent partial or total compensation for disbenefits such as in the case of medical expenses; g) demographic influences on population or population-segment statistics, as distinct from influences deriving from trend changes for individuals and/or families - you are inferring aggregate longitudinal trends for individuals from cross-sectional statistics. There may be a reason or reasons why you make that statement in the light of those omissions. Could you please explain what such reasons are? Secondly, could you please explain why you say there is a single and consistent data source (see below on survey methods and possible variation)? BM: Household surveys are nationally representative surveys of people’s income or consumption. MB: Strictly speaking, the surveys are not of either. They are of answers to questions about income, consumption expenditure and/or the value imputed by researchers to answers on own produce consumed. My third question is: Why do you say they are nationally representative when: a) Martin Ravallion has said that the rich and the destitute are hard to reach, and b) the survey methods may vary across time or countries in respect of recall periods, questions asked, methods of imputing value to own produce (farm-gate prices versus market prices) and so on? BM: This adjustment is done by converting incomes into so-called dollars of equal purchasing parity (PPP dollars) such that with $PPP 1 dollar, a person can purchase the same amount of goods and services in any part of the world. MB: Fourthly, I am unable to see how what you say about what has been done with these adjustments is true. Why do you say people can purchase the same amount of goods and services in different places with a PPP dollar, and by implication could do so at different times? For all years prior to 2005 the conversions were done wholly from national CPI rates. These were not adjusted for spending or consumption patterns, or prices faced by the people, at any different levels of spending/income/imputed-value-of-own-produce-consumed. In practice, I cannot purchase the same amount (or the equivalent amount) of goods and services in different countries with such a PPP dollar's worth of local currency, because I cannot buy a tiny fraction of an expensive item. Expensive items have disproportionate effects relative to the numbers of people buying them on the CPI. So it is hard for me to see how what you say is true even for purchasing power over the mix of all goods and services that the national CPI measures. Whether it is true for purchasing power over items people actually buy at different levels of spending/income etc - for example at the PPP $1.25 level - is not known. There may be some reason why you infer that the buying power of PPP dollars does not vary across countries or times according to levels of spending etc - if so, perhaps you could explain the reason. Similarly, your references to "real income" are in the context of national PPP rates for all levels of spending/income etc, at least most of the period. BM: ...the proportion of what the World Bank calls the absolute poor (people whose per capita income is less than 1.25 PPP dollars per day)... MB: The context (your mention of those slightly less poor than the poorest 5%) does relate it to the kind of level you are talking about. But your phrase "what the World Bank calls the absolute poor" is incorrect. They are in fact people the World Bank calls "extremely poor" in the context of claims that what is being measured is absolute poverty. Also, it is not really simply income, as mentioned above. The World Bank Millennium Goal monitoring team of Chen and Ravallion chose the $1.25 line you mention because they see it as representative of the lowest 15 or so national "poverty lines" in the years closest to 2005. They chose the $2 line as representing the median "poverty line" in "developing countries". The mean line would be somewhat higher. Fifthly, why are you using the terms "real income" or "absolute income gains" when it is not really income, and the inflation adjustments are not for prices faced by people at different levels, and there is no mentioned thought about needs for expenditure on rent, food or other things? BM: In 1988, a person with a median income in China was richer than only 10% of world population. Twenty years later, a person at that same position within Chinese income distribution, was richer than more than one-half of world’s population. Thus, she leapfrogged over approximately 40% of people in the world. MB: The sixth question is this: Why are you saying she became richer than that proportion of people? The statistics on countries as a whole are based on answers to questions about what people spent etc. as described above. Spending is, if one assumes for some reason that the answers were accurate and the data adequately comparable in reliability and survey methods, a measure of flow, not acquisition. So is income.