The World Bank remains uniquely positioned to monitor global poverty. Countries share data with the World Bank that they don’t share with anyone else. The World Bank has invested a significant amount of money into ensuring that data is internationally comparable, timely and methodologically sound. The World Bank’s PovcalNet database, which drives global estimates of poverty, was updated on 10 October 2017. The updates were good in some senses, but they also signified a step backwards for inclusive data.
What’s been added this year?
Last year, the World Bank published recommendations of the Commission on Global Poverty, also known as the Atkinson Commission. The latest PovcalNet update included some of the commission’s recommendations. For instance, the PovcalNet website now makes it very simple to toggle between three major poverty lines (the traditional line of $1.90, a line at $3.20 and another at $5.50), and poverty estimates from high-income countries are included in main global poverty estimates. These changes should be celebrated.
The latest PovcalNet updates also include estimates from the Middle East and North Africa, which were omitted in the last release due to concerns about purchasing power parity. This is certainly an improvement; however there are still challenges with the data. For instance, estimates for Syria (0.5% in extreme poverty) are based on a survey conducted in 2004, long before the protracted conflict there. Generally, though, the addition of Middle East and North Africa fills an important gap.
The World Bank also added data from Myanmar for the first time. Anyone hoping to use PovcalNet to better understand the humanitarian crisis in Myanmar will not be able to draw insights from this data –numbers aren’t disaggregated by ethnicity or race, and so it’s not clear if the Rohingya are included in the survey. Nevertheless, the addition of data from Myanmar is important.
The World Bank publishes a fairly detailed description of the updates. One such update was to switch from defining poverty by income in Haiti (the norm for Latin American and Caribbean countries) to consumption (the norm for sub-Saharan African and Asian countries). This meant that Haiti went from having 53% of its population living in extreme poverty to 24%. Consequently, Haiti went from being the 13th poorest country in the world to being 36th.
What’s lost in this data update?
The PovcalNet team provides an annual update of data for specific poverty surveys every year. These surveys typically take some time to process, so PovcalNet numbers are generally published for data collected three years ago. Many countries do not conduct surveys every year, so the World Bank has developed a method to line up surveys and estimate income levels between surveys. With the method for lining up data, it is possible to produce global estimates of poverty plus regional and national poverty estimate at the same time. Beginning in 2013, the World Bank began publishing annual estimates using these methods. With annual updates, it was possible to have reasonable estimates of trends in poverty. Starting last year, the updates from PovcalNet were analysed in the Poverty and Shared Prosperity report.
This year, PovcalNet added new surveys conducted since 2013 from about 70 countries, but it did not line up the data to provide global or regional estimates. This is somewhat surprising because the amount of effort required to publish the estimates is minimal compared with the effort put into adding the surveys. On top of this, some effort is required to find the newer data because 2013 data features more prominently.
The World Bank has announced that the publication of global and regional estimates will be reduced to every other year, as will publication of the Poverty and Shared Prosperity report. The decision to turn away from annual analysis of progress against the World Bank’s goals, and the corresponding Sustainable Development Goals, is unfortunate. The World Bank’s blog explains that year-on-year differences are usually insignificant for many countries so new data does not provide much additional information. This is true for many countries, but it is not always the case. For instance, PovcalNet reported that 6.47% of China’s population was living in extreme poverty in 2012. By 2013 that number had dropped to 1.85%. The degree of change in a single year in the world’s most populous country can hardly be dismissed as a small difference.
The World Bank also argues that many changes in the annual estimates are based on extrapolation methods that may introduce errors. However, the reliance on extrapolated data is fundamentally a problem about infrequent data collection, not a problem with the frequency of extrapolation. Without more investments in data collection, biannual PovcalNet updates will be as dependent on extrapolation as annual updates. A better solution is to significantly increase investments in developing statistical systems to produce better, more frequent data.
With very small investment the World Bank could increase the frequency of surveys and decrease the reliance on estimates. According to the latest estimates from PARIS21, only 0.3% of official development assistance supports statistics. Doubling the spending on aid statistical systems would have a substantial impact on revolutionising data for development. Despite this, the World Bank seems to decrease its funding for statistical capacity, making it likely that surveys will be less frequent. To further advance the goal of ending poverty and boosting shared prosperity, the World Bank could increase investment in poverty surveys and broader investments in national statistical systems.
The World Bank’s contribution to poverty measurement has been invaluable and they should be praised for their efforts to engage with experts to improve poverty data. We hope that they will continue to invest to improve systems in a valuable way and will not weaken their resolve to provide regular updates on progress towards their goals and the Global Goals to end poverty in all its forms everywhere.