Multidimensional Poverty Index: simulation and optimization

Ms. Lucia Ferrone and Ms. Margherita Squarcina from University of Florence and UNICEF presented their study that aims to reveal the potential impacts of COVID-19 on child multidimensional poverty in the MENA region. 

Results show that multidimensional poverty is higher among younger children, with the highest impact prevailing in the long term. Rural parts were more affected, compared to urban parts. The highest deprivations induced by COVID-19 were manifested in health in the short-run.

The presentation was followed by a 5-minute discussion moderated by Mr. Vladimir Hlasny.

Ms. Sama El Hage Sleiman from ESCWA built on the previous presentation by Ms. Ferrone and Ms. Squarcina and offered an extensive explanation of the Monte Carlo simulation exercise carried out by ESCWA for the case of Lebanon. Simulations allow for updating the available data and being responsive to member states amid the fast-changing realities. The method also enables predicting the effect of policies and the effect of COVID-19 or economic shocks as the case in Lebanon.

Results indicate that the health dimension was the lead contributor to MPI in Lebanon, followed by general services and employment. In 2021, MPI was expected to double and poverty headcount considerably increased in 2021 to 81%.

The second part of the session was followed by a 5-minute discussion moderated by Mr. Vladimir Hlasny.

Mr. Touhami AbdelKhalek and Ms. Dorothee Boccanfuso from the Université Mohammed VI Polytechnique discussed the implementation of new targeting approaches to detect the impact of social protection programs on multidimensional poverty in Morocco. Using the Revised Arab framework, they implemented two different approaches. The first follows random selection among deprived households on each of the indicators considered, and the second is a more objective approach using a probit model to identify households whose status changes from deprived to non-deprived.

Findings suggest that social protection reforms may lead to a decrease in deprivations from the targeted indicators (intensity decreases), but leaving the incidence of multidimensional poverty (H) unchanged. To enhance the model, interdependence among indicators should be accounted for in order to refine the assessment of the impact of welfare reform, which may be underestimated.

The end of session 2 included discussions and some questions and answers.

Mr. Paul Makdissi from the University of Ottawa proposed a different approach for nowcasting multidimensional poverty in the occupied Palestinian territory using the national MPI framework with a higher weight allocated to the monetary resources dimension.

Results show that after COVID-19, education becomes more important to the MPI in Palestine in 2021, but in all cases monetary resources dimension remains the lead contributor to MPI. Moreover, MPI increases after 2018, where the most substantial increase occurs in 2021.

Following the presentation by Mr. Makdissi, the floor was opened to questions and discussions for a 5-minute time frame, and a 10-minute break followed.

In their study, Mr. Ricardo Nogales and Mr. Nicolai Suppa (presenters), together with Ms. Sabina Alkire and Natalie Nairi Quinn, from OPHI differentiated their work by capturing the effect of COVID-19 on multidimensional poverty in the global level rather than in regional or national contexts. Two MPI indicators were simulated: nutrition and school attendance.

The results call for policy action to mitigate the potential global impact of the pandemic. Methodologically, the implemented approach may be extended to other development indicators.

The 5-minute discussion following the presentation was moderated by Mr. Vladimir Hlasny.

The final presentation was led by Mr. Hassan Hamie from ESCWA and Mr. Majd Olleik from the Technical University of Vienna. They described an innovative optimization model that aims at obtaining reasonable budgeting estimates required for reaching a certain MPI target in a certain year.

The case of Lebanon is adopted to provide a practical exercise using a real dataset (survey 2019). The national MPI framework for Lebanon including six dimensions is used.  Assuming a desired reduction of the MPI by 20% and setting hypothetical costs per flip for each activated indicator, three models are identified, the difference between the models is on the targeting approach and state capabilities.

The results indicate that the geographic cell model is the most practical, realistic, and robust. It also provides solid solutions despite random societal response.

The presentation concluded with the statement of the limitations of the model and the possible future extensions of the work.

Ms. Racha Ramadan provided formal comments on the presentation. A lengthier discussion with participants ensued focusing on the limitations of the model and the significance of implementing the model in the future for policymaking.