Poverty Mapping Research Articles

Predicting Poverty

Abstract Poverty prediction models are used to address missing data issues in a variety of contexts such as poverty profiling, targeting with proxy-means tests, cross-survey imputations such as poverty mapping, top and bottom income studies, or vulnerability analyses. Based on the models used by this literature, this paper conducts a study by artificially corrupting data clear of missing incomes with different patterns and shares of missing incomes. It then compares the capacity of classic econometric and machine-learning models to predict poverty under different scenarios with full information on observed and unobserved incomes, and the true counterfactual poverty rate. Random forest provides more consistent and accurate predictions under most but not all scenarios.

Human bias and CNNs’ superior insights in satellite based poverty mapping

Satellite imagery is a potent tool for estimating human wealth and poverty, especially in regions lacking reliable data. This study compares a range of poverty estimation approaches from satellite images, spanning from expert-based to fully machine learning-based methodologies. Human experts ranked clusters from the Tanzania DHS survey using high-resolution satellite images. Then expert-defined features were utilized in a machine learning algorithm to estimate poverty. An explainability method was applied to assess the importance and interaction of these features in poverty prediction. Additionally, a convolutional neural network (CNN) was employed to estimate poverty from medium-resolution satellite images of the same locations. Our analysis indicates that increased human involvement in poverty estimation diminishes accuracy compared to machine learning involvement, exemplified with the case of Tanzania. Expert defined features exhibited significant overlap and poor interaction when used together in a classifier. Conversely, the CNN-based approach outperformed human experts, demonstrating superior predictive capability with medium-resolution images. These findings highlight the importance of leveraging machine learning explainability methods to identify predictive elements that may be overlooked by human experts. This study advocates for the integration of emerging technologies with traditional methodologies to optimize data collection and analysis of poverty and welfare.

Poverty mapping in the age of machine learning

Recent years have witnessed considerable methodological advances in poverty mapping, much of which has focused on the application of modern machine-learning approaches to remotely-sensed data. Poverty maps produced with these methods generally share a common validation procedure, which assesses model performance by comparing sub-national poverty estimates with survey-based, direct estimates. While unbiased, direct estimates can be imprecise measures of true poverty rates, meaning that it is unclear whether these validation procedures are informative of actual model performance. In this paper, we use a rich dataset from Mexico to provide a more rigorous assessment of the modern approach to poverty mapping by evaluating its performance against a credible ground truth. We find that the modern method under-performs relative to benchmark traditional methods, largely because of the limited predictive capacity of remotely-sensed covariates. For a given covariate set, we also find that machine learning produces more biased poverty estimates than the traditional procedures, particularly for the poorest geographic areas.

Mapping Poverty for Sustainable Development Using AI, A Review of Literature

Extreme poverty is among the challenges the United Nations seeks to eradicate by the year 2030 as outlined in its Sustainable Development Goals. However, governments and other stakeholders face challenges in accurately identifying poverty in households for evidence- based implementation of SDG programs. Current strategies are slow, inaccurate and costly to efficiently support efforts to identify poverty for sustainable development. Consequently, many strategies to map out poverty for intervention measures do not succeed which could be contributing to the global decline in the rate of reducing poverty. Artificial intelligence which has become widely available and has been used in many sectors, could be leveraged to improve poverty mapping for evidence-based interventions for sustainable development. Despite living in the era of AI, it has not been fully utilized in mapping poverty. This review seeks to explore the extent of research on the adoption of AI in mapping poverty so as to find the gap for further research. It aims to establish the extent of AI-based research on identification of poverty in respect to global distribution of research studies, methods, algorithms and sources of data which have been used in studies to identify poverty. The findings will help to identify gaps for research to help in designing evidence-based strategies for intervention measures. A systematic review was done for the period 2020 to 2024 using databases and snowballing hybrid search approach. A qualitative analysis was done on the extracted data to uncover new patterns and identify research gaps.

Machine Learning for Mapping and Forecasting Poverty in North Sumatera: A Data-Driven Approach

Discussing poverty is crucial because it affects many facets of society, including socioeconomic disparity, crime, and the inability to obtain high-quality education. One of the provinces with the highest poverty rate in Indonesia is North Sumatra. A strategy is required to gather accurate data to effectively reduce poverty. Poverty mapping and prediction were conducted in North Sumatra to get a precise spatial distribution of poverty, the operation of the poverty model, and forecasting using machine learning (ML). Poverty prediction was conducted using a random forest (RF) algorithm and poverty mapping was conducted using the K-Means algorithm. The poverty mapping showed a significant inertia value decline in the third and fourth clusters of the elbow graph. The third cluster (0.313) was superior to the fourth cluster (0.244) in the silhouette index. Thus, there were three poverty clusters - low, medium, and high - that were used in the model. The best model was created using the grid search cross-validation, while the best prediction results were created using the RF algorithm, with the following parameters: n-estimator = 50, max depth = 10, min samples split = 2, and min samples leaf = 1. The mean squared error (MSE) of the RF model's predictions was 0.002617, or satisfactory precision.

Mapping Nigeria’s Poverty with Satellite Imagery Using Deep Learning: A Comparative Study of Day and Night Light Satellite Images

Abstract: Poverty mapping plays a crucial role in understanding and addressing socioeconomic disparities within a country. Traditional methods of poverty estimation often rely on survey data, which can be time consuming, expensive, and limited in coverage. In recent years, the advent of satellite imagery and deep learning techniques has opened up new avenues for poverty mapping. This study aims to compare the effectiveness of day and night light satellite imagery in mapping poverty in Nigeria using deep learning models. In this thesis, daylight satellite data is used to directly forecast poverty by evaluating Convolutional Neural Network (CNN) models. one methods semantic segmentation are put forth and contrasted with the multistep learning method that makes use of an image of nighttime lights. We conduct our experiments with satellite pictures from Google Maps and NOAA that are publicly available throughout the day and at night. Combining the night lights data with the semantic segmentation approach to yields the best model

Combining samples in small area estimation: An application to poverty mapping in Vietnam

AbstractSmall area estimation (SAE), which combines a population census and a sample survey, is widely used to estimate poverty and welfare indicators in small areas. A common situation in practice is that a population census is conducted using both short‐ and long‐form questionnaires. The short form is used to collect basic demographic information for the whole population, while the long form is used to collect additional information, such as employment, from a random sample. This study shows that combining both short‐ and long‐form data can improve estimation efficiency. This method is applied to poverty maps for the 2014–2019 period in Vietnam.

The impact of local cost-of-living differences on relative poverty incidence: an application using retail scanner data and small area estimation models

Estimating economic poverty indicators at the local level is essential for well-targeted data-driven welfare policies. However, Italy is a country characterized by strong geographical heterogeneity represented by unequal price levels among different areas, and computing poverty indicators with a national monetary poverty threshold can be misleading. This work proposes a novel approach to estimate monetary poverty incidence at the provincial level in Italy considering the different price levels within national boundaries. To account for local price variation, Spatial Price Indices (SPIs) are computed using scanner data on retail prices. The SPIs are estimated in two ways, referring to the mean local prices and using the 20th percentile of the prices. These two kinds of SPIs are used to adjust the national poverty line when computing the poverty incidence at the provincial level using Small Area Estimation (SAE) models. Our findings suggest that adjusting the national poverty line using the SPIs to compute a monetary poverty index can modify the poverty mapping results from the map produced with the traditional national poverty line that ignores price differences.

Small Area Poverty Estimation Under Heteroskedasticity

Abstract Multilevel models with nested errors are widely used in poverty estimation. An important application in this context is estimating the distribution of poverty as defined by the distribution of income within a set of domains that cover the population of interest. Since unit-level values of income are usually heteroskedastic, the standard homoskedasticity assumptions implicit in popular multilevel models may not be appropriate and can lead to bias, particularly when used to estimate domain-specific income distributions. This article addresses this problem when the income values in the population of interest can be characterized by a two-level mixed linear model with independent and identically distributed domain effects and with independent but not identically distributed individual effects. Estimation of poverty indicators that are functionals of domain-level income distributions is also addressed, and a nonparametric bootstrap procedure is used to estimate mean squared errors and confidence intervals. The proposed methodology is compared with the well-known World Bank poverty mapping methodology for this situation, using model-based simulation experiments as well as an empirical study based on Bangladesh poverty data.

PEMETAAN KANTONG KEMISKINAN DAN POTENSI WILAYAH UNTUK PEMBERDAYAAN KELUARGA MISKIN PASCA COVID-19 DI KABUPATEN KONAWE SELATAN

Poverty mapping is carried out by examining the economic characteristics of a region and household, such as the poverty of families whose livelihoods are fishermen, farmers, miners, the informal sector, trade and services. This is relevant to the National Economic Recovery Program (PEN) as one of a series of activities to reduce the impact of Covid-19 on the economy, one of which is mapping pockets of poverty and recovery and empowerment policy strategies for the community, including poor families, both in rural and urban areas. The aim of this research is to examine the potential of the region and produce a Post-Covid-19 Poor Family Empowerment Strategy in South Konawe Regency. This research use desciptive qualitative approach. Referring to this design, this research was carried out in South Konawe Regency, using purportive sampling technique. The data collection techniques used were document study, in-depth interviews, questionnaires and observation. This research data processing and analysis uses two methods, namely quantitatively and qualitatively. The research results show that the socio-economic conditions of poor families in South Konawe Regency are reflected in several items, namely employment conditions, income level, house type, house size, ownership and use of toilets, clean water sources, lighting sources, education level, and types of side jobs. Of these items, there are only two different items that occurred in the research sample village communities both before the pandemic and after the pandemic. These items are employment conditions and income level. Regional Potential for Empowering Poor Families Post-Covid-19 is divided into traditional fishermen (fish catchers using simple technology) in Tinanggea District, plantation farmers and cultivators with very limited access to agricultural land in Benua District, informal sector workers in Andoolo District, stone miners traditional in North Moramo District and commuters (travel workers from out-of-town residents working in Kendari City) in Ranomeeto district. The Post-Covid-19 Poor Family Empowerment Strategy can be divided into several things, namely existing policies for poor families during the pandemic, developing KUBE, empowering the community through developing institutions and social capital, maximizing the use of village funds, and providing ease of obtaining KUR facilities.

Mixed Autoregressive Model for Spatial Data: A Bayesian Application to Poverty Mapping

Mixed Autoregressive Model for Spatial Data: A Bayesian Application to Poverty Mapping

Water woes and vulnerabilities: a case study of Bally-Jagachha block of Howrah District in West Bengal, India

ABSTRACT Water is one of the most contended resources on the planet. The need and the un-ubiquitous nature of the resource is a reality faced by the Bally-Jagachha block in Howrah, West Bengal. Here, the available water sources are tube-well and piped water, both having Water Quality Indices of more than 300 and 200, respectively. This indicates that drinking water is unpalatable, making people vulnerable to diseases. Additionally, as is evident through the Water Poverty map, inaccessibility is an expensive, debilitating problem, as people have to travel twice in 4–7 min to and from the water source to collect it and then again spend more to filter the same. The Ridit scores thus state that people are overcome by water woes in supply, availability, consumption, quality, demand, etc., but yet have no alternatives like rainwater harvesting (RWH), despite people’s acceptance to it. Thus this paper attempts to collate all the problems regarding water and suggest ways by which the Gram Panchayats here can safely implement RWH and resolve the situation for the future.

ANALISIS KEMISKINAN EKSTREM PROVINSI BENGKULU MENGGUNAKAN METODE GEOGRAPHICALLY WEIGHTED REGRESSION (GWR) DENGAN PEMBOBOT ADAPTIVE GAUSSIAN KERNEL DAN ADAPTIVE BI-SQUARE

Extreme poverty is a condition of inability to fulfill basic needs, namely the need for food, clean drinking water, proper sanitation, health, shelter, education, and access to information which is not only limited to income, but also access to social services (United Nations, 1996).Geographically Weighted Regression (GWR) model is used in mapping extreme poverty of all level 2 regions in Bengkulu Province using Adaptive Gaussian Kernel and Adaptive Bi-Square weights as well as finding the best GWR model and analyzing the model against extreme poverty mapping of Bengkulu Province. The data used in this study is the March 2022 Susenas data. Of the 18 variables that allegedly affect extreme poverty, only 6 variables support the assumption of spatial heterogeneity in GWR modeling. Based on the selection of the best model, it is known that the GWR model with Adaptive Bisquare Kernel weighting is a suitable model for the percentage of extreme poor people in Bengkulu Province with the smallest AIC value

Point-to-Region Co-learning for Poverty Mapping at High Resolution Using Satellite Imagery

Despite improvements in safe water and sanitation services in low-income countries, a substantial proportion of the population in Africa still does not have access to these essential services. Up-to-date fine-scale maps of low-income settlements are urgently needed by authorities to improve service provision. We aim to develop a cost-effective solution to generate fine-scale maps of these vulnerable populations using multi-source public information. The problem is challenging as ground-truth maps are available at only a limited number of cities, and the patterns are heterogeneous across cities. Recent attempts tackling the spatial heterogeneity issue focus on scenarios where true labels partially exist for each input region, which are unavailable for the present problem. We propose a dynamic point-to-region co-learning framework to learn heterogeneity patterns that cannot be reflected by point-level information and generalize deep learners to new areas with no labels. We also propose an attention-based correction layer to remove spurious signatures, and a region-gate to capture both region-invariant and variant patterns. Experiment results on real-world fine-scale data in three cities of Kenya show that the proposed approach can largely improve model performance on various base network architectures.

Using poverty maps to improve the design of household surveys: the evidence from Tunisia.

In this paper we aim to propose a new method for improving the design effect of household surveys based on a two-stage design in which the first stage clusters, or Primary Selection Units (PSUs), are stratified along administrative boundaries. Improvement of the design effect can result in more precise survey estimates (smaller standard errors and confidence intervals) or in the reduction of the necessary sample size, i.e. a reduction in the budget needed for a survey. The proposed method is based on the availability of a previously conducted poverty maps, i.e. spatial descriptions of the distribution of per capita consumption expenditures, that are finely disaggregated in small geographic units, such as cities, municipalities, districts or other administrative partitions of a country that are directly linked to PSUs. Such information is then used to select PSUs with systematic sampling by introducing further implicit stratification in the survey design, so as to maximise the improvement of the design effect. Since per capita consumption expenditures estimated at PSU level from the poverty mapping are affected by (small) standard errors, in the paper we also perform a simulation study in order to take into account this addition variability.

High-resolution rural poverty mapping in Pakistan with ensemble deep learning.

High resolution poverty mapping supports evidence-based policy and research, yet about half of all countries lack the survey data needed to generate useful poverty maps. To overcome this challenge, new non-traditional data sources and deep learning techniques are increasingly used to create small-area estimates of poverty in low- and middle-income countries (LMICs). Convolutional Neural Networks (CNN) trained on satellite imagery are emerging as one of the most popular and effective approaches. However, the spatial resolution of poverty estimates has remained relatively coarse, particularly in rural areas. To address this problem, we use a transfer learning approach to train three CNN models and use them in an ensemble to predict chronic poverty at 1 km2 scale in rural Sindh, Pakistan. The models are trained with spatially noisy georeferenced household survey containing poverty scores for 1.67 million anonymized households in Sindh Province and publicly available inputs, including daytime and nighttime satellite imagery and accessibility data. Results from both hold-out and k-fold validation exercises show that the ensemble provides the most reliable spatial predictions in both arid and non-arid regions, outperforming previous studies in key accuracy metrics. A third validation exercise, which involved ground-truthing of predictions from the ensemble model with original survey data of 7000 households further confirms the relative accuracy of the ensemble model predictions. This inexpensive and scalable approach could be used to improve poverty targeting in Pakistan and other low- and middle-income countries.

Photography in Social Work and Social Change: Theory and Applications for Practice and Research, Matthias J. Naleppa, Kristina M. Hash and Anissa T. Rogers

This book poetically describes photography as ‘painting with light’ and makes a convincing case for the fit between photographic approaches and ‘the social work profession’s focus on the importance of place, the environment, and the needs and assets that they can encompass’ (p. 120). It consists of three parts: Photography in Social Work, which considers historical, practical, theoretical and research issues; Photographic Methods and Applications, which presents five specific methods; and Technical Aspects and Future Applications. The introductory chapter helpfully presents a brief history of photography before moving on to its use in social work. It provoked new thinking for me: for instance, what photographic evidence there is of the work of the Settlements in England in the early years of photography and whether there was any photographic accompaniment to influential documents like Charles Booth’s Poverty maps of London, which themselves rely on image, if not of a photographic type. I was attracted to the authors’ ideas about social work journalism in which different media combine to tell the stories of the people and communities in contact with social work.

Analyzing Water Poverty Components Using Geospatial Tools: Resource and Environmental Constraint in Kathmandu District

This current paper examines water poverty situation in Kathmandu district in terms of water resource and environment constraints. Integrated methodology was adopted for the study. Water source sample survey was carried out using GPS tool. Study area comprised total of 13 water sample sites. Household questionnaire survey was carried out for water poverty mapping and analysis. Resource component of water poverty comprised indicators, namely seasonal variation in water availability, water supply frequency and groundwater recharge potential. The environment constraint component comprised water quality rate of soil erosion and topographic wetness index. The result show that the average calculated value of water resource limitation component is 6.31 and out of 13 studied communities, 7 communities fall below average. It is found that environmental constraint is less associated with urban housing density. It is found that environmental constraint is less associated with urban housing density. The average calculated value of environmental component is 3.43 and out of 13 studied communities, 8 communities fall below average. Spatial variability in water poverty is prominent and highest water poverty is found in urban cores. Communities with lower water poverty are found in peri-urban location near the foothills. The average calculated value of water poverty is 9.74 and out of 13 studied communities, 9 communities fall below the average. The study concluded that water poverty index and resultant map is a very effective tool to visualize the distribution of water poverty at local spatial scale and to present the complex nature of water poverty.

Multivariate Mixture Model for Small Area Estimation of Poverty Indicators

Abstract When disaggregation of national estimates in several domains or areas is required, direct survey estimators, which use only the domain-specific survey data, are usually design-unbiased even under complex survey designs (at least approximately) and require no model assumptions. Nevertheless, they are appropriate only for domains or areas with sufficiently large sample size. For example, when estimating poverty in a domain with a small sample size (small area), the volatility of a direct estimator might make that area seems like very poor in one period and very rich in the next one. Small area (or indirect) estimators have been developed in order to avoid such undesired instability. Small area estimators borrow strength from the other areas so as to improve the precision and therefore obtain much more stable estimators. However, the usual model-based assumptions, which include some kind of area homogeneity, may not hold in real applications. A more flexible model based on multivariate mixtures of normal distributions that generalises the usual nested error linear regression model is proposed for estimation of general parameters in small areas. This flexibility makes the model adaptable to more general situations, where there may be areas with a different behaviour from the other ones, making the model less restrictive (hence, more close to nonparametric) and more robust to outlying areas. An expectation-maximisation (E-M) method is designed for fitting the proposed mixture model. Under the proposed mixture model, two different new predictors of general small area indicators are proposed. A parametric bootstrap method is used to estimate the mean squared errors of the proposed predictors. Small sample properties of the new predictors and of the bootstrap procedure are analysed by simulation studies and the new methodology is illustrated with an application to poverty mapping in Palestine.

Multi-source satellite imagery and point of interest data for poverty mapping in East Java, Indonesia: Machine learning and deep learning approaches

This study proposes a novel approach to provide a more granular poverty map in terms of coverage (up to a grid level with the spatial resolution of 1.5 km) with less cost and time to update to support better poverty monitoring. Two poverty estimation model development scenarios using machine learning and deep learning were evaluated in this study. First, the model is constructed using multisource satellite imagery and geospatial point of interest location of economic infrastructures using zonal statistics feature extraction, which is the proposed approach in this study. The specific multisource satellite imagery to be used as poverty indicators are night-time light intensity (NTL) as an approach to economic activity, normalized difference vegetation index (NDVI) as an approach to detect rural areas based on vegetation, built-up index (BUI) as an approach to detect urban areas based on building distribution, normalized difference water index (NDWI) to detect land cover, land surface temperature (LST) as an approach to detecting urban areas based on surface temperature, carbon monoxide (CO), nitrogen dioxide (NO2), and sulfur dioxide (SO2) as an approach to detecting economic activities based on pollution, as well as point of interest (POI) density, and POI distance which show the accessibility of an area. Second, the model is constructed from daytime multiband and nighttime light intensity satellite imagery using the deep learning architecture of Resnet-34 transfer learning feature extraction. In each scenario, we compared the performance of different machine learning (support vector regression/SVR, decision tree regression/DTR, and random forest regression/RFR) algorithms along with deep learning (multiple layer perceptron/MLP and one dimensional-convolutional neural network/CNN-1D) algorithms. Based on the results of the model development evaluation, the CNN-1D model was chosen as the best model in the first scenario, and the Resnet-34 + MLP model was chosen as the best model in the second scenario. The two best models from the first and second scenarios are then used to predict poverty at the grid level with a spatial resolution of 1.5 km. Based on official poverty data, the estimated poverty mapping built using the CNN-1D model in the first scenario was chosen as the best map with the value of root mean squared error (RMSE) 1.95 and adjusted R2 0.84 at the district level. The results of visual identification revealed that high estimates of poverty are typically found in sparsely inhabited areas surrounded by unoccupied land, which is usually an agricultural area. Otherwise, low poverty estimates, on the other hand, are more likely to be found in highly populated regions with convenient access. Our finding is aligned with the official report of Statistics Indonesia, which claims that poverty in rural regions is higher than in urban areas.