In 2015, for the first time ever, global extreme poverty will fall below 10 %, according to the World Bank in a triumphant press release of three weeks ago. But the Bank remains cautious about its in 2013 defined objective : to eradicate extreme poverty by 2030, or to have it at around 3 %.
This is obviously good news. The United Nations just adopted its ‘Sustainable Development Goals’ as a follow up to the Millennium Development Goals and objective number one, the halving of extreme poverty between 1990 and 2015 has been met. The World Bank’s Global Monitoring Report 2014/15 estimated extreme poverty in 2011 at 14,5 %, expecting it to lower to 11,5 % in 2015.
Global poverty, then, is diminishing. Some may remember that at the start of this century a percentage of around 20 was mentioned for 2010. In the past year, many scholarly articles were published saying that new measurements would further diminish extreme poverty, others estimating it to remain stable and still others expecting it to rise.
It is sometimes difficult to believe and it can be useful to try and follow the thread, look at how debates are developing and put the poverty measures into their right context. It is useless to try and prove the figures are ‘false’, since that would imply other figures are ‘right’ and that thesis particularly has to be rejected.
The history of poverty according to the World Bank
The World Bank statistics have always been very fragile. When it published its first ‘poverty report’ in 1990, it estimated extreme poverty in 1985 to concern 633 million people in developing countries. In 1980, ten years earlier, it spoke of around 800 million extremely poor people.
But in fact, it had no statistics at all. It used estimates and extrapolations. In 1990 it had data on consumption and income of households in barely 22 countries. Only in 1985 did it start to make its own surveys and when it put poverty on the agenda in 1990, it had its proper data for only four countries. In 2000 these statistics had to be reassessed and African extreme poverty rose from 39,1 to 49,7 % in 1996. In that same year the Bank estimated the extremely poor to be 1,190 billion or 24,5 % of the population in developing countries. For 1987, it than estimates extreme poverty at 28,31 % or 1,183 billion people. For these measurement the World Bank used its own poverty lines calculated on the basis of data from the ‘International Comparison Project’ of 1993.
At the turn of the century and after the adoption of the MDGs at the UN, there was quite some interest for correct poverty statistics. UNCTAD (the UN Conference on Trade and Development) made its own statistics, not using household surveys but looking at the private expenditures in the national accounts. Its results were significantly higher than those of the World Bank, around 50 %, although it is difficult to compare since UNCTAD only looked at the Least Developed Countries.
The only continent where poverty measures are more or less consistent is Latin America. ECLAC (UN Economic Commission for Latin America and the Caribbean) started in the 1970s and uses poverty lines that are far more realistic than the 1 $ a day of the World Bank. The results, then, are higher: the World Bank estimates extreme poverty in the region in 2010 at 4,8 %, whereas ECLAC arrives at 12,1 %.
Extreme poverty in developing countries according to the World Bank
Publication Reference Year Estimation
1980 1980 800 million
1990 1985 633 million (lower poverty line 275$/year)
1,116 billion (higher poverty line 370 $/year)
2003 1981 1,482 billion (lower poverty line 394$/year
1984 1,277 billion (same as above)
2001 1,089 billion (same as above)
In 2007 the ‘International Comparison Programme’ published its new data and the World Bank recalculates its poverty data. It finds out the Chinese economy has been overestimated by 40 %, and the Indian economy by 25 %! It means China all of a sudden has 65 million more people in extreme poverty. Coupled to the reassessment of the poverty line (from 1 to 1.25 $/day), it means China has now 130 million more extremely poor people! In 2005, according to the World Bank, developing countries had 1,399 billion extremely poor people, or 25,7 % of their population. Sub-Saharan Africa has 609 million extremely poor people or 50,4 % of its population. East Asia 336,9 million or 17,9 % and South Asia 595 million or 40,3 %. But the Bank repeats that extreme poverty is diminishing, because it now gives a figure for 1981 almost 2,5 times as high as it said in the 1980s: 1,904 billion.
In the preparation process for 2015 and the end term for the MDGs, the World Bank once again adjusts its statistics for 2010 and 2011, with projections for 2015, 2020 and 2030. In 2010, according to the Bank, 19,1 % of the population of developing countries was poor, or 1,125 billion people.
But in 2014, price data as well were adjusted, leading, once again, to serious variations in the poverty statistics. However, and at first sight, in the other direction! The total percentage lowers from nearly 20 % to 8,9 %, even if in Sub-Saharan Africa extreme poverty rises to 51 %.
In May 2015 two World Bank researchers react to these new figures. Calculations have to be made differently, they claim, and if you follow our reasoning, the figures will not change.
But not more than three weeks ago, the Financial Times wrote that poverty statistics were going to rise with the new price data.
The World Bank now finally has published its new statistics, and once again, it used the change in price data to adapt its poverty line. From now, extremely poor are people living on less than 1,90 $ a day. And with those new data, extreme poverty is diminishing. In 2012, global extreme poverty is only 12,8 %, in Sub-Saharan Africa it is 42,6 %, in South Asia 18,8 % and in East Asia 7,2 %.
Who is poor then?
Needless to say that the painful situation in which extremely poor people have to live, does not change with the zigzagging statistics. But the changing reasoning is sometimes difficult to follow. What figures are correct and which are not? Are statistics being manipulated? Who can tell? One has to assume researchers work seriously, the problem is they are faced with difficult methodological choices, on the one hand, and many missing data, on the other hand. Inevitably, many statistics cannot be more than highly scholarly guesswork.
A first point which has to be made and which has less to do with research than with a political choice is the change of the reference: today, percentages are expressed in terms of global population, and not of developing countries anymore. It makes quite a difference.
A second problem, repeatedly mentioned, is the calculation of the purchase power parities. These are nothing else than conversion rates in order to make national numbers internationally comparable. The error margin can be significantly high. They are used to calculate the poverty lines, but it all depends on which criteria are used and if a distinction is made between rural and urban regions, to give just one example.
A third serious problem are the missing data. UNDP (UN Development Programme) repeatedly pointed to the fact that for most of the indicators of the Millennium Development Goals, we have no data at all, let alone reliable data, in spite of all colourful graphs one can find. Half of the 48 indicators, according to a recent publication of The Economist, are based on generalisations instead of censuses. A ‘data revolution’ is now promised, because the problem is particularly serious in Sub-Saharan Africa. It all starts with the civic registrar. Only four African countries have a complete and efficient system, according to the World Bank. Only 45 % of the countries have organised household surveys for the past three years. And only 33 % of the countries had two poverty-related surveys in the past ten years. In fact, says Morten Jerven, there is very little we know for sure about the gross domestic product and even growth in Africa. It is all linked to a lack of capacity and resources, but it is precisely in that way that space for manipulation is created.
Another problem related to the measurements is that national figures often seriously differ from the international ones. This is linked to politics and even ideology. National governments need poverty statistics for political objectives, often linked to their legitimacy. This will determine the choice of the methodology, in order to have a better or worse result. And of course, multinational organisations also use poverty statistics in order to boost their legitimacy, but the ‘adjustment’ happens at another level. The result is divergent and thus unreliable statistics.
Measurements can also be influenced by the choice of household surveys: does one look at consumption or at income? Income indicates a kind of potential, whereas consumption is the output. But they can seriously differ. Income can, theoretically, be zero, whereas consumption cannot. Very often both types of surveys are organised, in the same way as for inequality statistics, and it can seriously bias the results in such a way they become non comparable.
Finally, results may also differ in function of the way in which questions are put. If one asks a household to note during seven days in a logbook the consumption of rice, the result will be lower than when one asks to note the consumption every seven days. Differences can be as high as 30 %, only because short term memory is less positive than long term memory!
These are only some of the problems arising with poverty statistics, just to show how difficult it is to give reliable data and how methodological choices can have political reasons and consequences.
This does not mean the poverty statistics of the World Bank are meaningless. One should not take them as absolute truth, and the big variations show perfectly well why not. They continue to be an indicator of the trends in poverty and this trend shows a reduction of extreme poverty. Of course, this tells us nothing on the evolution of poverty, because escaping extreme poverty does not mean one becomes middle class! On the contrary, what has happened in many countries, and what is now happening in Western Europe, is a squeezing of middle classes, with a small group joining the upper class of the rich, and the big majority falling down, till just above the poverty line and a risk of falling into poverty as soon as a small risk occurs. This is why inequality is a much more serious political problem than poverty, but poverty statistics do not tell us anything on inequality.
The World Bank has recently created an international commission in order to examine its poverty statistics and to make new proposals. This is a very good initiative because the credibility of the international institutions is at stake.
Not only poverty statistics are failing. Statistics on inequality are not reliable either. Figures on international trade are flawed, because import should in theory be identical to export, and this is far from being the case. We tend to believe that our policies are based on scientific criteria and objective data, but it is in fact easy to see that there is a big gap between beliefs and reality.
In 1973, president McNamara of the World Bank said that extreme poverty would be eradicated by the year 2000. In 2013, the World bank defined its new strategy and wants to ‘eradicate’ extreme poverty by 2030. But the Bank already said this is very ambitious, which means this will very probably not be achieved.
There can be no question about the usefulness of poverty reduction, but when one sees how very difficult it is to measure poverty and how methodology is miles apart from the reality of poor people, one can wonder how serious these statistics have to be taken. It is not even necessary to think of manipulation. There are two worlds, the world of the poor and the world of the poverty measurements, that are far apart.
Another question is if these statistics help in order to fight poverty? This answer is not so clear. The World Bank insists on ‘targeting’ social policies and in that case it is clear one has to know who the poor are. But is this the best policy? A universal social protection to which even the non poor contribute and benefit from can give better results for the whole of society.
Maybe more attention has to be given to the kind of policies we need, the neoliberal ‘poverty reduction’ of the World Bank, the social protection floor of the ILO, or the even better universal western-type social protection, or even the social commons proposals I recently made. Leave the statistics aside, and let us focus on what people want and need. This is a political choice.
 World Bank, Global Monitoring Report 2014/2015. Ending Poverty and Sharing Prosperity, Washington, The World Bank, 2015.
 Chen , S. and Ravallion, M., How did the world’s poorest fare in the 1990s?, World Bank, 2000, table 2.
 Chen, S., and Ravallion, M., op. cit., p. 5.
 UNCTAD, The Least Developed Countries Report 2002. Escaping the Poverty Trap., UNCTAD, Geneva, 2002, p. 57.
 World Bank, 2015, op. cit., p. 19.
 CEPAL, Panorama Social 2012, Santiago de Chile, United Nations, 2012, p. 14.
 Chen, S. and Ravallion, M., The Developing World Is Poorer Than We Thought, But no Less Successful in the Fight Against Poverty, PRWP 4703, The World Bank, August 2008.
 World Bank, 2015, op. cit.
 Dykstra, S., Global Absolute Poverty Fell by Almost Half on Tuesday, Center for Global Development, 2014.
 Jolliffe, D. and Prydz, E.B., Global Poverty Goals and Prices, PRWP 7256, The World Bank, May 2015.
 Donnan, S., Planet’s Poor Set to Swell as World Bank moves Poverty Line, 23 September 2015.
 Ferreira, F.H.G. et al., A Global Count of the Extreme Poor in 2012. Data issues, Methodology and Initial Results, PRWP 7432, October 2015.
 Deaton, A. and Aten, B., Trying to Understand the PPPs in ICP 2011: Why are the Results so Different?, NBER Working Paper 20244, June 2014.
 Anderson, B., Quantifying the Challenges facing the Data Revolution in Africa, Development Initiatives, 1 September 2015.
 Jerven, M., Why Economists get it Wrong, Zed Books, 2015.
 Oxford Policy Management, Measuring Post-2015 Development Performance, Briefing Note, 2015.
 Backiny-Yetna, P. et al., The Impact of Household Food Consumption Data Collection Methods on Poverty and Inequality Measures in Niger, PRWP 7090, The World Bank, November 2014.