I SSU E 47 | JU N E 2, 20 22 DASHBOARDS FOR DEVELOPMENT: THE POWER OF GEOSPATIAL DATA AT YOUR FINGERTIPS Takaaki Masaki, Lander Bosch, Arden Finn, Moritz Meyer, Syed Zeeshan Haider, Eduard Bukin One of the most important challenges for the World Bank Group is ensuring that its resources reach those who are most in need. To this end, geospatial dashboards have become a widely used tool to inform spatial targeting. A geospatial dashboard can combine and provide unique insights into a rich array of subnational development indicators which can be leveraged to better understand spatial disparities along various thematic and sectoral dimensions (e.g., poverty, education, health). It can also be used to inform where development projects should be targeted to most effectively reach areas that hold the greatest potential to achieve the World Bank Group’s twin goals of poverty eradication and shared prosperity. This note introduces the need for and various uses of geospatial dashboards, and how they can serve as effective tools in policymaking and geographical targeting. Moreover, a step-by-step guide illustrates how to design and construct dashboards, with the various application possibilities, and offers links to World Bank resources including sample dashboards UNDERSTANDING SPATIAL DIMENSIONS OF DEVELOPMENT Do people have the same opportunities and face the same challenges across space, or do spatial disparities exist? Which geographic areas are most deprived? Where should the World Bank target resources to maximize its development impact? Fiscal and financial constraints – coupled with limited implementation capacity – make the implementation of multisectoral, universal coverage programs practically infeasible. It is thus critical that resources are spent where they hold the greatest potential to help achieve the World Bank's twin goals of ending extreme poverty and boosting shared prosperity. A first, pivotal step to achieve these aims is to look beyond the national average. It is widely acknowledged that deep spatial disparities in welfare and development outcomes exist, yet a detailed picture of these disparities is often lacking at the administrative level where investment decisions are made. Moreover, lagging regions are often deprived due to multiple, compounding factors. World Bank resources therefore need to support the targeting of areas that stand out as deprived along multiple indicators. DASHBOARDS FOR DEVELOPMENT Three types of data provide a rich and diverse set of source materials which can inform the identification and targeting of priority areas. This includes data obtained through surveys and/or censuses, big data – either remote-sensed or administrative – which are geospatial in nature, and sector-specific data derived from World Bank, client, or development partner projects. This fast-growing inventory of large-volume data, spurred by recent technological innovations, holds a lot of powerful information and evidence to shed light on spatial disparities. Having vast amounts of data is one thing. Using them to generate astute insights into spatial disparities and influence policy decisions and tailor investments by governments, the World Bank Group, and development partners is another. Simply bringing numerous indicators together in large tables or graphs does not help unravel their inherently spatial correlations and patterns. The visualization of these patterns and interrelations through user-friendly, interactive mapping is critical to identify areas that suffer from poor connectivity, low access to core public services, frequent conflict, or climate shocks, as well as higher poverty levels and other lagging development outcomes. As such, high-resolution geographic data can be employed to locate deprived regions that are in greatest need or areas where resources can be used more efficiently to maximize their development impact. While data visualization through mapping is thus highly valuable, harnessing the power of geospatial data to inform data-driven decision-making remains challenging. Much of the existing data tend to be available in silos, hard to integrate due to their different formats, and requiring significant time, resources and expert skills by data scientists and geographers for processing and interpreting. Such siloed information prohibits the uptake of geospatial data among policymakers, researchers, and World Bank staff, let alone citizens. Integrating these diverse sources of data into a single, easy-to-use geospatial dashboard enables broad audiences to explore, use and digest key messages. Such geospatial dashboards serve a threefold purpose: I) to visualize welfare indicators and development outcomes from various sources in an integrated manner to show spatial disparities at high resolutions; II) to inform the dialogue between the World Bank, clients, and development partners, and support decisionmakers in aligning the allocation of resources with spatial disparities in development outcomes; and III) to monitor changes in development outcomes over time. While all three objectives could be achieved by a static map, the key added value of a geospatial dashboard stems from its interactivity, flexibility, and ease of use. Interactive maps allow users to choose any indicator of their interest, to display it at different levels of disaggregation, and to overlay this indicator with other data to gain further insights, for instance surrounding specific project sites or settlement areas. 2 POVERTY & EQUITY NOTES DASHBOARDS FOR DEVELOPMENT As such, light analytics are directly possible in the tool, such as tracking a time series or performing bivariate correlation analytics between variables of the user’s choosing. The geospatial dashboard can also be updated dynamically as new data come in, which makes it particularly useful for monitoring fast-changing events like COVID-19 or conflicts, or longer-term trends such as poverty dynamics. This allows for a comparison of developing situations across both space and time. The World Bank Poverty and Equity Global Practice’s Data for Operational Impact (D4OI) team has spearheaded efforts to develop geospatial dashboards as an instrument to promote the use of data and evidence to inform the spatial targeting of the Bank portfolio. Beyond simple visualizations, these dashboards also offer integrated analytics, leveraging the power of diverse data sources and indicators to explore spatial patterns and develop indices relevant to policy, investment decisions, and specific projects. One of the tools offered by the team is a Project Targeting Index (PTI) dashboard, which serves both as a database for a large set of geospatial indicators as well as a user-friendly tool for constructing a composite index called the PTI, 1 allowing users to rank areas by priority based on objective criteria. The PTI can be constructed by the user based on the thematic area of interest. To establish where to allocate resources for a sanitation project, for instance, the user can construct a WASH-focused PTI, bringing together indicators on handwashing, access to improved toilet facilities, and waterborne disease. The PTI dashboard has been rolled out to more than 12 different countries, mainly in Africa, where the Country Management Unit (CMU) and project teams have used it to identify priority areas for their interventions. Moreover, it can be easily integrated with other thematic dashboards. This has been done for the Pakistan Indicator dashboard developed by the Pakistan Poverty and Equity team, which links the PTI to a diverse set of poverty and development indicators. Such integrated dashboards offer comprehensive yet digestible insights into the spatiotemporal distribution of indicators relevant to the World Bank’s Global Practices, clients, development partners and the wider public. The D4OI team has developed a framework for and prototype of both the geospatial dashboard and the PTI in a way that can be easily replicated and implemented by other teams. All source files and code are available on a public GitHub page, which teams can employ to tailor and replicate the dashboards and PTI to their needs and the context in which they operate. 1 See more details on the PTI methodology in Finn and Masaki (2020). 3 POVERTY & EQUITY NOTES DASHBOARDS FOR DEVELOPMENT METHODOLOGY AND DATA When constructing geospatial dashboards, three core aspects need to be considered: the dashboard scope and contents, the underlying data infrastructure, and its applications and functionality. Firstly, the thematic area and relevant indicators, geographic coverage and spatiotemporal resolution, and data sources determine the scope of the dashboard. A dashboard with a carefully curated selection of indicators will be easier to navigate and use than a simple bundling of numerous variables. A streamlined thematic or sectoral approach also provides focused insights into spatial disparities in the policy area of interest. Once the content and scope have been established, the format of the data, the need for user- friendliness, and the foreseen final use drive the selection of the online infrastructure underlying the geospatial dashboard. The D4OI team has opted to use R Shiny because of its user-friendliness and possibility to easily integrate analytical functions, and the shareable materials on GitHub are tailored to the development of dashboard apps using this R package. Finally, once the baseline dashboard has been developed, specific functionalities and applications beyond simple data visualization can be added. This can include the PTI, as well as bivariate, comparative mapping or data download functionality. Taking these three pillars into account, constructing a geospatial dashboard results in an eight-step process, as illustrated in Figure 1 and outlined below. Steps 1-4 cover the content, while step 5 focuses on the data infrastructure and step 6 on the applications and functionalities. Steps 7 and 8 draw these components together through the testing and publication of the geospatial dashboard. Step 1: Identify development dimensions. Each country has a different set of development priorities. The dimensions of development that go into the dashboards ultimately depend on what its users are trying to achieve. Strategic focus areas are typically drawn from a Country Partnership Framework or country-specific development strategy documents and typically include themes such as poverty, human capital, access to services, accessibility, climate, or conflict risks. Step 2: Identify relevant development indicators. A geospatial dashboard is useful only to the extent that input data are available with high geographical granularity. Thus, this second step involves an extensive stocktaking of all relevant data that are available at the subnational level for the development dimensions of interest, either derived from conventional data sources (e.g., census, household budget survey), administrative data sources (e.g., school enrolment data or subnational budgets), or other innovative forms of data (e.g., remote sensing or satellite data). In some countries – particularly in FCV contexts – no recent household 4 POVERTY & EQUITY NOTES DASHBOARDS FOR DEVELOPMENT survey or census data exist to measure socioeconomic conditions. In these data- scarce environments, publicly available geospatial data can play a particularly important role in filling such data gaps, such as the Humanitarian Data Exchange, the World Bank’s Open Data platform, or OpenStreetMap. High-resolution population maps, for example, have been estimated for a number of different countries across the globe by different initiatives, including WorldPop, Meta, Global Human Settlement Layer, and Africapolis, just to name a few. These data can complement conventional data sources in helping users to more effectively identify areas for intervention and investment. To the extent possible, these publicly available data require validation to ensure their accuracy, where possible through ground truthing. The source, primary collecting organization, and date of data collection need to be accurately documented and reflected on the geospatial dashboard, and made available for download in line with licensing agreements. Figure 1. Eight-step process of constructing a geospatial dashboard Steps 3: Determine the level of spatial aggregation. Since raw geospatial data are often available with different geographical granularity and in varying formats (e.g., Shapefiles, TIFF files, or NetCDF), it is difficult to perform systematic analysis unless they are pre-processed to construct indicators at a consistent level of 5 POVERTY & EQUITY NOTES DASHBOARDS FOR DEVELOPMENT (dis)aggregation. In many cases, the team generates subnational indicators at the lowest level of disaggregation where an official administrative boundary shapefile is available such that visualization and analysis can be performed at that level of granularity. Step 4: Process and display data. Once all data are aligned in terms of formatting and the level of spatial aggregation has been settled on, they can be processed. This includes the teasing out of relevant indicators, ensuring their coherence and comparability, for instance by aligning units of measurement and dimensions, and their display in the dashboard. Step 5: Determine which platform to use for the construction of a dashboard. Many different platforms can be used to construct a geospatial dashboard, including ArcGIS, Tableau, R Shiny, Microsoft Power BI, Google Data Studio and HubSpot. The team has built a back-end infrastructure in R Shiny, which has made the construction of a geospatial dashboard easier and scalable. This is available as an R package called ‘devPTIpack’. To launch a dashboard, a user can download this package, prepare the baseline metadata – which is essentially a brief description of data sources and indicators – as well as the baseline shapefiles. Figure 2 provides an example of the PTI dashboard for Somalia. Figure 2. Somalia PTI dashboard Step 6: Design the structure and functionalities of the dashboard. Geospatial data can be displayed on a map in many ways while other types of functionalities can be added within the dashboard. For instance, the Pakistan Indicator dashboard allows users to compare indicators through side-by-side maps and perform basic diagnostics through scatterplots (to show a relationship between different 6 POVERTY & EQUITY NOTES DASHBOARDS FOR DEVELOPMENT subnational indicators) or bar charts (to visualize the explicit ranking of subnational areas), as well as to plot summary tables. Moreover, any plotted map, graph or table can be directly downloaded and used. A GitHub repository is available, containing the source material for this dashboard and all information needed for its replication. Step 7: Quality assurance. Once a prototype of the dashboard is prepared, it will go through multiple rounds of quality assurance to address any potential bugs and identify areas to improve the user experience. This feedback can be directly obtained through user consultations, who will be able to provide experience-based feedback on the content and functionality of the dashboard. Step 8: Publication. Once all technical glitches are resolved and map clearance from the World Bank Cartography Unit 2 is received, the dashboard is ready for launch and dissemination. It is important to stress that the publication of the geospatial dashboard is not an end point. With new data continuously becoming available, and user expectations and experiences feeding back to the developers, this is a live tool that will be regularly updated. POLICY IMPLICATIONS Understanding spatial disparities for a diverse range of development outcomes is critical to ensure that resources are allocated effectively to reach those who are most in need. A plethora of geospatial development data derived from conventional data sources as well as non-traditional, emerging sources like remote sensing or satellite data can help aid this understanding. However, processing and integrating these various datasets to provide useful analytical insights for decision- making is no trivial task. It requires expert knowledge, time, and resources. The key utility of a geospatial dashboard lies in its ability to bring such geospatial data to your fingertips. User-friendly, interactive geospatial dashboards make these data readily accessible to client governments, development partners and World Bank operational teams alike. It can serve as a one-stop shop where users can come to explore various subnational development indicators, visualize them, and combine them with other sources of data to gain further insights, as well as perform some simple analytics. The dashboards help in identifying areas that are particularly vulnerable to various development challenges, ranging from poverty, human capital deficits, lack of access to services, low connectivity, climate 2 Clearance can be requested via email at mapclearance@worldbank.org. 7 POVERTY & EQUITY NOTES DASHBOARDS FOR DEVELOPMENT change impacts to conflict risks. This can feed the discussion on priority areas for investment, and shape policy narratives. REFERENCES Finn, Arden; Masaki, Takaaki. 2020. Subnational Targeting of Project Sites Using Project Targeting Index. Poverty and Equity Notes; No. 33. World Bank, Washington, DC. © World Bank. https://openknowledge.worldbank.org/handle/10986/34311 License: CC BY 3.0 IGO. Pape, Utz; Yoshida, Nobuo; Malgioglio, Silvia. 2020. Monitoring Poverty and Equity for Development Effectiveness. Poverty and Equity Notes; No. 34. World Bank, Washington, DC. © World Bank. https://openknowledge.worldbank.org/handle/10986/34705 License: CC BY 3.0 IGO. About The Authors Takaaki Masaki is an Economist in the World Bank’s Poverty and Equity Global Practice. Lander SMM Bosch is the Regional Geographer for the South Asia Region in the World Bank’s Poverty and Equity Global Practice and a World Bank Young Professional. Arden Finn is a Senior Economist and Task Team Leader of the World Bank’s Poverty and Equity program in the West Bank and Gaza. Moritz Meyer is a Senior Economist in the World Bank’s Poverty and Equity Global Practice and Task Team Leader of the Pakistan Poverty and Equity team. Syed Zeeshan Haider is a Consultant in the World Bank’s Poverty and Equity Global Practice. Eduard Bukin is a Consultant in the World Bank’s Poverty and Equity Global Practice. CONNECT WITH POVERTY & EQUITY GLOBAL PRACTICE www.worldbank.org/poverty @WBG_Poverty This note series is intended to summarize good practices and key policy findings on Poverty-related topics. The views expressed in the notes are those of the authors and do not necessarily reflect those of the World Bank, its board, or its member countries. Copies of the notes from this series are available on worldbank.org/poverty. 8 POVERTY & EQUITY NOTES