no image

PPI update and challenges in using old PPIs

Marco Antonielli
• Nathan Associates
• United Kingdom
• 10/24/18

Hi PPI users and managers,

As you might know, the World Bank has recently updated the poverty rate figures for a range of developing countries. The new figures are generally based on 2015/16 data while the previous figures were based on 2010/11 data. Currently the PPIs are based on the 2010/11 survey waves, so my first question for the PPI managers is when could users expect updated PPIs to be published on the website?

As I'm designing surveys for projects' M&E in Ethiopia and Malawi, I am also wondering what the impact is of using in 2018 a PPI questionnaire developed to estimate likelihood of poverty in 2011. To what extent are these estimates relevant and appropriate to assess present-time poverty likelihood? My guess is that as the characteristics of poor households change overtime, so will good proxies of poverty, but I'm not quite sure what the impact might be.

To give an example of where some comparability issues may arise, take the following question from the Ethiopia PPI:

Does the household currently own any radios/radio-and-tape players/tape players?

Assuming that between 2011 and 2018 the prices of electronics dropped, owning one of these items in 2018 might not be as good a predictor of non-poverty than it was in 2011.

I would welcome any inputs about this and I'm particularly looking forward to check my understanding of PPI!

NB - the PPIs for ETH and MLW are still officially called Simple Poverty Scorecards.


Hi Marco, Thanks for your question. Responses are in order as follows: 1. You can find our current PPI development pipeline here - We prioritize countries based on the needs of our PPI Alliance, and can also accommodate requests from other organizations that are interested to sponsor a particular PPI. New national survey data of course must be available, and in some cases even though a survey was conducted in 2015/16, the data is not publicly available for some time (as was the case for Kenya whose 2015 survey data was not made public until March 2018. See updated Kenya PPI here - The PPIs we have updated over the past year are listed in bold on this page - Most of those are with 2015 or later data. 2. You are correct, it is not possible to know the extent to which a given PPI will become inaccurate when applied to data collected in years subsequent to that of the national survey on which it is based. The change in accuracy will depend on the overall change in the economic situation of each country; this change could happen very quickly or could progress more slowly. However, as we create new PPIs, we typically test how the older PPI performs in later years. For example, to do this for Kenya, we applied the 2005 Kenya PPI to the 2015 KIHBS survey data. While, as may be expected, the new 2015 PPI provides more accurate estimates of poverty in 2015 compared to the 2005 PPI, the difference between the two PPIs is small for representative samples of the National, Urban and Rural population. However, the difference varies across regions. Please see section 1E and Table 9 of the 2015 Kenya User Guide for those details. Since moving to the PPI's new methodology at IPA, we have refrained from including questions related to technology (e.g. mobile phones) that may quickly become ubiquitous or out-of-date, and thus could negatively impact the accuracy. 3. Correct, the tools for Ethiopia and Malawi are not actually PPIs. They are Simple Poverty Scorecards. If you have any further questions, please feel free to email us at
no image
Hi Julie - thanks very much for your detailed and clear responses.