How does the MPI respond to changes over time?

The MPI can be used to study changes in multidimensional poverty patterns over time providing that the data at different points in time and MPI estimates are strictly harmonized. If estimates are not harmonized, readers are advised to carefully interpret the changes over time for a particular country because different indicators could be missing from the surveys.

This year, we are releasing changes over time estimates for 80 countries with suitable surveys of which 28 countries have data for three time periods. These estimates are presented in Table 2 of the 2021 MPI publication. This table presents harmonized estimates and it allows for a strict comparison across time within a country. The harmonization process aligns the indicator definitions, meaning that, where necessary, this process re-creates the indicators in the global MPI so that they are using exactly the same information and deprivation cutoffs in both years. For instance, if a survey in one year collects nutrition information from children under 5 years and for women aged 15–49 years and in the other year it collects nutrition information from children under 5 years only, the harmonized computations will only use information from children under 5 in both years, the common denominator, so that any change in the nutrition indicator is due to a change in the nutritional status of children rather than the inclusion of a group of women who may tend to be more (or less) undernourished.

Users are advised to distinguish between the unharmonized estimates in table 1 and the harmonized estimates in table 2. The first ones represent the best possible MPI estimate for a particular year and survey and include all the information that is available, while the second ones allow for a strict comparison across time (within a country) by removing the components that were not collected in the previous point in time. Unharmonized and harmonized estimates may be identical in some cases, but they may differ in many others.