Word Count: 1848

Introduction

Poverty is arguably the most important determinant of development and its alleviation is one of the most sought after processes in the modern world. As of 2021, approximately 700 people or 9.82% of the world live below the global poverty line. In addition, poverty is a major factor of adverse health outcomes, child mortality, population growth, social instability, and conflict. The eradication of poverty is vital for the progression of the world through the development lens and is the first target of the Sustainable Development Goals. The nature of identifiers and various measurements of poverty are still discussed among development experts in the literature, ranging from construction of unidimensional and multidimensional indices to the use of monetary/nonmonetary metrics. Resultantly, researchers have turned to quantifying poverty metrics in a digestible format that development experts can use to accurately pinpoint the location and nature of poverty within the local context of a given region. Poverty maps compile indicators into a geospatial dataset of disaggregated areas for a region. The advent of Big Data has revolutionized how data scientists can map poverty by taking advantage of mobile phone and satellite data to compile highly accurate plots. In addition, modern poverty maps are not only high in accuracy, but can be updated in real time as a result of the constant stream of data that they use as inputs. The statistical techniques used to compute outputs for the maps are advanced in nature and made possible by advanced computational software like supervised learning models that employ machine learning algorithms to train and test a set of data over thousands of iterations to define meaningful relationships between input data. Different approaches to map compilation depend on the type of methods and models data scientists use, thus altering the usefulness of the maps depending on what context they were compiled in. For example, some researchers will use data sets with similar features (e.g. night-time lights, building density, population measures) to create different maps ranging from population distributions to income disparity and consumption levels. Thus, with the recent emphasis placed on the compilation of poverty maps, the natural next question becomes “How effective are poverty maps for alleviating poverty?” If this question is answered, a cost benefit analysis can be conducted to determine whether or not poverty maps have usefulness outside of their inherent nature as identifiers. In other words, by using poverty maps for alleviation purposes rather than only as a way to identify poverty, specific routes of alleviation might be revealed ranging from targeting and distributing public investment to improving service delivery and infrastructure in areas of need.

Novel Hypothesis

To determine the effectiveness of poverty maps for alleviation purposes, a 4 step plan is proposed. First, a region that has been uplifted out of poverty will be identified. This will be made possible by using different metrics of poverty to compare the region’s current state to its previous impoverished state using various consumption and poverty metrics like before/after wealth levels, before/after income levels, before/after relative comparison to established poverty lines, among others. Ideally, the region will also have data depicting the state of different economic, social, and political conditions before and after poverty uplift. While the actual nature of poverty alleviation can be gradual in nature, the beginning and end “values” for identifiers should contrast enough to warrant analysis. Second, call detail records (CDR) and remote sensing satellite data (RS) will be collected using existing frameworks for the identification of poverty indicators that are specific to that region. To process the data, Voronoi tessellation can be used to allow the subsequent model to be built on the scale of Voronoi polygons. Each polygon can be assigned RS and CDR values representing the mean, sum or mode of the corresponding data. The data will be summarized and subsequently matched to the polygons based on the GPS located coordinates of PPI data, the latitude and longitude representing the centroid of each cluster and the home tower where the CDR data was collected. This has a dual effect of maximizing readily available data while minimizing cost. Third, hierarchical bayesian geostatistical models (BGMs) will analyze the aforementioned data and indicators to compile a poverty map of the area. This compilation will serve two primary purposes; A. evaluate the accuracy of the map when compared to the before/after poverty measures and B. provide insight about the nature of alleviation. Fourth, a simulation of poverty alleviation methods will be implemented over many iterations using data from the compiled poverty map to determine effectiveness of the maps in identifying poverty. If the alleviation methods based on the map from the simulation yield equivalent or better metrics than the current state of the region, then the poverty map can be considered an effective tool of alleviation.

Objectives, Arguments, Objections

The primary objective of this proposal is to determine the usefulness of poverty maps for poverty alleviation rather than just poverty identification. Naturally, alleviation follows identification, and poverty maps seem to be the best tool for this purpose. This is because they are unique in contextualizing poverty to a specific area with relative ease compared to other techniques that require more human resources and capital. If poverty maps can be used effectively for alleviation, a vast number of doors can be opened in the future for improving living conditions and expanding development globally. However, this alludes to the difficulty of this objective as complications arise in extending the application of poverty maps to alleviation. Currently, most of the research in the literature focuses on using poverty maps to identify poverty with little emphasis on alleviation. This can most likely be attributed to the lack of statistical data in areas with higher levels of poverty; researchers focus their efforts instead on coming up with innovative techniques to identify poverty in these areas in light of the lack of data. Furthermore, regions that experience an uplift from poverty are typically shocked into production as a result of drastic shifts in economic or social policies. While this research proposal aims to overcome these limitations, there are still some obstacles barring success. Primarily, because poverty is not absolute and must be contextualized to a specific area, input data indicators in one region that are used to compile the maps might not be predictive of other regions, especially in data-deprived regions that cannot provide similar input CDR and RS data. However, as technology progresses and societies advance, more information will become readily available to use as input data, further contributing to the accuracy of the maps. Once the effectiveness of maps for alleviation is determined, subsequent research of alleviation efficiency can be explored. In other words, the focus can shift from the basic possibility of alleviation to the nature and type as well as effectiveness of alleviation. The results of this proposal would also allow researchers to identify what aspects of poverty should be targeted (as learned from the maps) for the most efficient alleviation, further maximizing the use of resources and capital to reduce poverty. This would serve as an innovation for development as it streamlines the process of alleviation in contrast to the current strategies which are highly resource exhaustive and somewhat ineffective. Ultimately, this proposal takes advantage of the variability in poverty by accounting for specific contexts rather than a one size fits all approach.

Budget

With a $100,000 cap, the budget can be partitioned according to the steps of the research plan, with additional funds provided for researchers, logistical arrangements, and necessary equipment including software. As this proposal is driven by data, anywhere from 10-15% can be allocated to data collection. At least 10,000 will be used to obtain historical data, CDR data from cell phones and cell towers, and RS data. Some portion of that initial 10,000 for data will account for additional processes to ensure accuracy and that individuals’ privacy is preserved. About 30-40% funds can safely be allocated for research compensation and living expenses. Depending on the region for analysis, there is a possibility that researchers might have to do physical data collection involving on the ground work to account for gaps in CDR and RS input data. Another 30-40% of the budget can be allocated to equipment involving physical hardware and digital software as some labs contain hundreds of thousands of dollars of super computers and storage as well as software that must be purchased to use. While already existing data is preferable, some new data might need to be collected, thus warranting the need to invest in equipment. Most importantly, this part of the budget will pay for the intense hardware and software-dependent simulations. Finally, the last portion of the budget can pay for any miscellaneous expenses that are incurred as a result of the project. A key part of this proposal is to minimize the use of capital and human resources. Thus, minimizing costs everywhere possible is essential.

Conclusion

Poverty mapping can be achieved in numerous innovative ways as a result of the ability of supervised learning models to be trained using currently available geospatial data sets as well as other input factors and employing machine learning algorithms to find relationships between attributes for poverty indication. The 4 steps of this research proposal aim to evaluate the usefulness of poverty maps and subsequently expand them to account for alleviation. The research proposal takes advantage of easily attainable CDR and RS data as inputs for a BGM to compile the poverty maps and subsequently simulate alleviation techniques. This also has impacts outside of the initial goal of evaluating poverty alleviation from maps by providing researchers with a streamlined process to contextualize alleviation to the surrounding area just like how poverty is contextualized to the surrounding area. In turn, this allows for the analysis of alleviation itself and provides for ways of determining effectiveness and efficiency. With a conservative budget and miscellaneous expenses accounted for, this proposal serves as a unique extension of the current literature and aims to fill in the gaps through the lens of deliverable application from alleviation.

Works Cited

Steele Jessica E., Sundsøy Pål Roe, Pezzulo Carla, Alegana Victor A., Bird Tomas J., Blumenstock Joshua, Bjelland Johannes, Engø-Monsen Kenth, de Montjoye Yves-Alexandre, Iqbal Asif M., Hadiuzzaman Khandakar N., Lu Xin, Wetter Erik, Tatem Andrew J. and Bengtsson Linus 2017 “Mapping poverty using mobile phone and satellite data” J. R. Soc. Interface.142016069020160690

Pape, Utz, and Luca Parisotto. “Estimating Poverty In A Fragile Context: The High Frequency Survey In South Sudan”. 2019. World Bank, Washington, DC, doi:10.1596/1813-9450-8722. Accessed 10 Mar 2021.

Rue, Håvard et al. “Approximate Bayesian Inference For Latent Gaussian Models By Using Integrated Nested Laplace Approximations”. Journal Of The Royal Statistical Society: Series B (Statistical Methodology), vol 71, no. 2, 2009, pp. 319-392. Wiley, doi:10.1111/j.1467-9868.2008.00700.x. Accessed 26 Apr 2021.

Gelfand, Alan E., and Sudipto Banerjee. “Bayesian Modeling And Analysis Of Geostatistical Data”. Annual Review Of Statistics And Its Application, vol 4, no. 1, 2017, pp. 245-266. Annual Reviews, doi:10.1146/annurev-statistics-060116-054155.