Rationale for choice of indicators

We identify the indicators that can help the development agencies in gaining a spatial understanding of the socio-economic conditions and stimulate effective decision making. Rather than concentrating only on a single indicator we here consider a set of indicators as it is now established that the effect of socioeconomic factors is much broader than just poverty or income (Adler and Ostrove, 1999; Gornick, 2002; Marmot, et al., 2010). Most of the indicators we identify are included amongst the National Priority Indicators of the United Nations’ Sustainable Development Goal (SDG), while some are additional indicators that may provide useful supplemental information to the decision-makers.

We select the relevant data sources for monitoring these indicators. To identify relevant and usable data sources, we chose sources that contained data that had at least the district level of granularity. It is to be noted here that downsampling from data sources that contain data at higher levels of granularity can increase sampling errors, thereby reducing the feasibility of monitoring these indicators correctly. Making such sources the focal point for decision making is risky.

We had to consider an important aspects traditional data that relates with how recent the data was, and the period after which it will be updated again. Conducting nationwide surveys and generating enough data to capture the condition of the indicators at a subnational level of granularity is costly and time-consuming, many of the traditional data sources are refreshed after considerably long periods, making the data stale in many cases.  To address this, we are conducting research on the feasibility of using state-of-the-art computational mechanisms to harness nontraditional heterogeneous data sources, such as the household level energy consumption data, traffic data, sensor data, port and road connectivity data, high volume satellite daytime imagery and nighttime lights to monitor the socio-economic conditions of Bangladesh. The results from our findings on the feasibility of Nighttime Lights were documented in the paper our recent publications and presentations.

We have considered five basic categories of indicators- living standard, health status, education, infrastructure, and environment. In living condition category, we have tried to reflect the economic status of the households living in the district. Living condition is captured by household monthly per capita income, % of people living below the extreme poverty line, access to electricity, access to safe drinking water and sanitation facility. Income and poverty both are widely used indicators for reflecting economic condition. Access to electricity has a robust impact on rural economic development and poverty reduction (Rud, 2012; Peter and Vans, 2011; Khandkar et al. 2012 and 2013; van de Walle et al. 2015). Several studies have found that access to electricity has a positive impact on educational performance, income level, and health (Parikh et al., 2015; Lenz et al., 2017). Access to safe drinking water and improved sanitation facility is also included explicitly in the Sustainable Development Goals (Goal-6). Environmental sustainability and poverty alleviation is closely associated with access to these two facilities. Empirical studies have also suggested that access to improved water and sanitation facility has a positive and significant impact on health, education, income, and wealth (Parikh et al., 2009). Therefore, both the variables are included in the living condition category.

The next category is Health. Health status represents local development and it’s an indicator of human capital. To represent this category, we have included three indicators- infant mortality rate, % of stunted growth, and birth registration. Infant and child mortality rates reflect a country’s level of socioeconomic development and quality of life. The rates are also useful in identifying promising directions for health and nutrition programs in any country (NIRTOP, 2016). Stunted growth rate is also an important indicator of nutritional status and health in populations (WHO, 2006). Birth registration rate is included as it is considered as a necessary indicator for government’s planning of national policies (UNICEF, 2020). Target (16.9) under Goal 16 of the SDGs aims to provide legal identity for all, including birth registration, by 2030.  Moreover, Target (16.9) under Goal 16 of the SDGs aims to provide legal identity for all, including birth registration, by 2030.

Similarly, education is a central part of development status as it enhances the earning potential and productivity of the household members and recognized as key determinant of household poverty (Cameron and Cameron, 2006; Rodriguez and Smith, 1994). Education is fundamental to personal empowerment as it helps to reduce inequality and increased political and community participation (Hannum and Buchmann, 2003). In this study, we have considered primary education completion rate to proxy this variable. 

Next, to proxy infrastructural development, we have used the percentage of the length of paved road of a district. High quality road facilitates rural poverty alleviation by boosting up agricultural production, reducing price distortion, promoting the local market, employment, school enrollment and expanding domestic trade (Gibson and Olivia, 2010; Wong et al. 2017; Khandaker et al., 2009 and Warr, 2005).

We have also included environmental indicators as development cannot continue if the environment and natural resources deteriorate. Two specific indicators were considered- natural hazard and reserve forest area. Vulnerability to natural disaster is one of the major constraints to alleviating poverty and facilitating economic development (Sawada, 2007; Haen and Hemrich, 2007). Poor people are the most vulnerable to sudden shocks, risks and uncertainty and they also suffer disproportionately from hazards relative to the rich (Varley, 1994; Okuyama, 2004). Disasters like flood and cycle causes destruction of country’s social and physical infrastructure which in turns changes individual’s production and consumption behavior and investment decision (Haque and Jahan, 2015). In this study multi hazard map prepared by Islam (2017) was used. This work constructed multi hazard score based on a combination of observed local hazard intensity, frequency of occurrences and vulnerability. Among the hazards it includes arsenic, earthquake, drought, flood, cycle, tornado.