What is the extent of the migration phenomenon in Moldova? What is the population density and how are rural areas affected? What are the patterns of migration, and how do they change thorough the year? Can we see how this pattern changed over the past 3, 5, or 7 years? How to make local public investment decisions sensitive to migration flows? These are the questions that MiLab is trying to answer through the Ghost Villages Project.

Official data on migration, both external and internal, is grossly underestimating the true extent of the phenomenon in Moldova. The most recent available data on population is the newly released 2014 census, which does specifically that – capturing the situation as of 2014. Indeed, we need to bear in mind that national censuses are designed to provide a snapshot every decade rather than offer high frequency granular information. Nevertheless, updated, reliable, and disaggregated data is required for policy making and targeted development interventions. Ultimately, we believe that access to this kind of data will be very valuable for decision makers at the local and central levels, in such cases as sizing and directing local infrastructure projects and for informing the territorial administrative reform which is planned to be carried out in 2018.

The ultimate objective of the project is to map the actual population density in the rural areas of Moldova, spotting the “ghost villages” – the communities or parts of communities where dwellings have been abandoned due to the increasing migration, which the country has been facing for the past several years. The mapping will be done using electricity consumption data as a proxy indicator of whether the dwelling is occupied or not. Where possible, alternative source of data will be used to triangulate the findings and provide more granular insights. The project is run together with the NBS, who will be the ultimate owner of the tool.

We are also hoping for some additional insights. For instance, what if looking at the electricity consumption patterns of rural population gives us insights about fluctuation of consumption and income level of the people? Or for instance, as preliminary data seem to confirm (spikes of energy consumption during Christmas and Easter holidays), what if energy consumption data gives us almost real-time information on circular migration flows? This clearly would give us insights on the best period to engage with migrants on decision making at local level? But what else?

As a first step, we are in the process of selecting a sample of 10 communities for testing the hypothesis that energy consumption data is a viable predictor of household inhabitance. Then, we will put up energy consumption on a map to understand the spatial dimensions of energy consumption. This will be followed by analysis of energy consumption patterns and estimation of minimum consumption threshold – a nominal value under which a dwelling would be considered unpopulated. We will then send enumerators to the pilot villages to validate the results obtained, as well as refine the estimations, if necessary. Based on the data obtained we will create a grid map of population density for the pilot communities. The last step will be scaling up the map for the entire country and integrating this data into the decision-making processes.