What data do we offer?

The global Multidimensional Poverty Index (MPI) is an international measure of acute multidimensional poverty, which captures overlapping deprivations in health, education, and living standards. The global MPI complements traditional monetary poverty measures such as the World Bank’s $2.15 a day international poverty line. See What is the global MPI? for more details about the measure itself, such as the individual deprivation indicators, indicator weights and poverty cut-off.

Technically, the global MPI relies on the Alkire-Foster (AF) method and it is estimated using comprehensive high-quality household survey datasets as provided, for instance, by the Demographic Health Survey (DHS) programme and the Multiple Indicator Cluster Surveys (MICS) of UNICEF. Related estimates are available for more than 100 developing countries.

The estimates of the global MPI are released on an annual basis and provided in two flavours. The most recent estimate provides an estimate for each country at a single point of time. In contrast, the results based on data which has been harmonized over time provide estimates for each country at several points of time. While the number of countries covered in the estimates based on data harmonized over time is usually smaller due to data availability, such estimates can provide critical insight into how poverty changes over time (as otherwise deprivation indicators or subnational regions may be incomparable).

The annual updates usually add new estimates for those countries for which suitable datasets have been released over the last year. Updates are provided for both flavours (the most recent estimates and the estimates based on data harmonised over time).

As the annual global MPI releases provide results for all countries, a significant share of estimates is identical to the release of previous years (i.e. whenever the underlying micro data did not change).

One important feature of the AF method is that it provides an entire set of related metrics permitting a coherent analysis of different aspects of multidimensional poverty. In this sense it is best understood as a framework. These metrics include the MPI itself (which is sometimes also called the adjusted headcount ratio), the headcount ratio (or incidence), the intensity, as well as censored and uncensored headcount ratios and dimensional contributions (See Chapter 5, Alkire, S., Foster, J.E., Seth, S., Santos, M.E., Roche, J.M. and Ballon, P. (2015). Multidimensional Poverty Measurement and Analysis: A Counting Approach. Oxford: Oxford University Press). We provide estimates for all of the previously mentioned metrics for each country. 

Another advantage of the AF method is that changes in estimates at the national level can be meaningfully related to changes for subpopulations (e.g. subnational regions). Where the underlying survey data permits, estimates are also provided for subnational regions, area (urban / rural) and age groups.

In addition to all of these estimates, the results also include results for complementary poverty measures such as severe poverty or destitution as well as auxiliary statistics (e.g., sample drop or standard errors) which help to assess quality and precision of the provided estimates. 

Products showing our data

Different products show the data of our latest release (featuring both flavours, that is, the most recent estimates and the estimates based on data harmonised over time) including the following. 

  • The data tables, essentially spreadsheets, are particular useful for screening and inspecting our estimates. Accordingly, these data tables also contain additional external data such as the Human Development Index (HDI), monetary poverty estimates of The World Bank, and number of poor estimates based on population data of United Nations Population Division (UNPD).
  • The data bank is particular useful for interactively exploring key metrics at different geographic levels. The databank features the spatial spread of multidimensional poverty, its intensity and headcount using the most recent estimates. The application presents comparisons at global and country and subnational level.
  • The results files are of particular interest for researchers who wish to use the estimates of the global MPI as an input for their own analyses. These files are provided in dta and csv format.
  • The country briefings contain selected estimates for each country. Shown estimates refer to the latest available dataset and, if available also to changes over time for harmonised datasets. The mostly graphical analysis is of particular interest for national policymakers.
  • Finally, different reports and briefings usually present and discuss selected findings based on the estimates of the particular release. 

All of the above mentioned products are directly linked on the page of a particular global MPI release. Moreover, the result files (provided in csv and dta format) are also deposited at the Oxford University Research Archive (ORA) together with the relevant documentation as detailed below.

Note that releases of previous years may not feature each and every product. For example, the comprehensive result files are only published since 2023. Moreover, the estimates presented for a country in these products are identical for different release years if no underlying micro data has been released in the meantime. Currently, please also note that the data bank is only available with data from the latest release.

How we document our data

The documentation of our data spreads over several types of documents which complement each other and provide different levels of detail. 

  • For each annual release, we publish one or more methodological notes. More recent releases entail several notes to provide appropriate space and recognition for each subset of the related work. Methodological notes can be found under Publications.  
  • The paper on the database of the harmonised level estimates of the global MPI by Suppa & Kanagaratnam (2023) documents key aspects of our data which are common to all releases. Its focus is to provide some background on the used methods, the underlying micro datasets, selected quality checks, and critical policies (e.g., on missing indicators). This paper also provides details on the structure of the results files.
  • The technical files published with each release comprise in particular the script files (country-specific Stata do-files) for the data extraction and deprivation indicator construction. 

All files of the documentation may be accessed in several ways. First, they are directly linked on the site of a particular global MPI release. Moreover, all files are also included or linked in the respective ORA package of a given release year.

How to replicate our data

Our results files may be replicated in three steps.     

  • First, download the underlying household survey Microdata from the website of survey providers as specified in our documentation. For example, a survey identified as ‘DHS’ must be accessed on the DHS website. 
  • Second, our published Script files (Stata do-files) allow for data extraction, cleaning and deprivation indicator construction that are executed using Stata software.
  • The results are estimated from the cleaned microdata using the MPI toolbox (-mpitb-), a user-written open access Stata package available at the Statistical Software Component (SSC) and under https://gitlab.com/nsuppa/mpitb.