Geographic information systems, or GIS, is not exactly a new concept, but there are plenty of missed opportunities to use spatial data to fill data gaps or to provide non-biased results (due to the presence of spatial autocorrelation and heterogeneity) on a variety of research topics.
But first—what exactly is geospatial data?
Geospatial data identifies the geographic location of features or boundaries associated with research observations. This kind of data is usually stored as coordinates, which means it can be easily mapped. It is becoming more common for large public data sets (e.g. IHS, DHS, and census) to include the location of their observations.
On March 29, 2017, IFPRI Research Analyst, Dr. Anderson Gondwe, presented a seminar on “Patterns of migration and employment in Malawi: spatial data analysis,” which shared results from a study that applied spatial panel data comprised of small geographical areas created from national censuses, to examine patterns of employment and migration in Malawi.
More specifically, the study used census data from 1987, 1998, and 2008. Given changes in the number of districts over time, 24 districts were matched and a panel was created: 24 districts and 178 traditional authorities. This allows comparability of the census data with other data for spatial analysis, and also allows the feasibility of integrating census data with other data for similar spatial analysis by matching small-area geographical information system (GIS) data codes that are consistent over time. This study specifically looked at how long-term changes in age structure affect the labor force participation rates, in addition to the effects of land reform policy on migration and employment. The analysis was segregated by gender to explore long term occupational mobility for women who are traditionally stuck in low paying self-employment activities.
Spatial descriptive analysis using the 2008 census show clusters of people employed, unemployed, and inactive across the country. Many parts of the central region show clusters of high rates of employment, followed by the northern region. In terms of education attainment, many parts of the northern region have people with complete primary and secondary education.
The study defines migration as movement of people from one region or district to another. It found that people tend to move between districts, but stay in the same region. There are small proportions of migration regionally, with a very small percentages of people moving across districts.
Another interesting aspect of the study evaluated the impact of land reform policy on spatial migration and employment. Results showed that the policy, which enabled land-poor households to voluntarily purchase plots from fallow estates and resettle there, had significant positive effects on migration and employment patterns in Malawi.
More broadly, the results conclude that spatial dependencies do exist between areas with respect to outcome variables. In other words, spatially close areas are more similar than distant areas. Furthermore, the study found some gender effects in terms of agricultural and government employment, and also that rural-urban migration is not as massive and widespread as people may think. By contrasting with results from non-spatial models, the researcher found that approaches which fail to take into account spatial dependencies tend to overstate the results.
The full seminar presentation can be viewed below.
