| dc.description.abstract | This thesis examines human mobility patterns in developing countries. It leverages different types of mobile phone data enabling the observation of individuals' movements and locations at spatio-temporal scales which are notoriously difficult to attain with traditional survey data. These movements are investigated in various contexts, including spatial inequalities and frictions, labor reallocation dynamics in the face of climate variability, and the interplay between different modalities of short-term mobility as mechanisms to access urban markets. Chapter 1 uses smartphone app location data from three African countries over a one-year period to study a type of mobility that has previously been difficult to capture. The data allow to characterize patterns of what might be characterized as "visits", which are neither migration movements nor commuting trips. This type of high-frequency mobility is found to be ubiquitous in the three countries studied: smartphone users are frequently away from home and often travel long distances. The granular nature of the data allows to obtain insights into the specific destinations where people are observed. These include locations associated with shops and markets, government offices, and places offering a range of goods, services, and recreational venues. Big cities seem to be particularly important destinations, perhaps reflecting the range of amenities that they offer to visitors. These mobility patterns offer novel insights into spatial frictions and the geographic patterns of economic activity. Chapter 2 provides a suite of methodological tools to derive temporary migration statistics from mobile phone data. First, the chapter delves into well-known challenges surrounding representativeness of non-traditional data sources. Second, an enhanced temporary migration detection method based on a clustering technique is presented. Third, the chapter addresses challenges in creating time-disaggregated temporary migration statistics derived from individual trajectories. Applying this methodological framework to three years of Call Detail Records (CDR) from Senegal, the results unveil a remarkable level of temporary movements and bring novel understandings about the magnitude, timing, spatial distribution and orientation of temporary migration trends in Senegal. Chapter 3 investigates the impact of climate variability on the short-term spatial reallocation of individuals in a developing country setting. Centering on Senegal as a case study, the study scrutinizes the relationship between precipitation patterns during the rainy season (spanning June to October) and the temporary migration decisions for the remaining agricultural calendar. A uniquely granular temporary migration matrix is constructed drawing on data methods presented in Chapter 2, and is combined with satellite-based local precipitation measures. Estimations of dyadic regressions reveal that poorer rainfall conditions at a rural origin during the rainy season impede temporary migration during harvest but act as a push factor over the following off-season. These effects are more pronounced in rural locations exhibiting lower standards of living. On the other hand, the quality of the rainy season at a destination is found to be positively related to the level of attractiveness of that destination to temporary migrants. Finally, Chapter 4 builds upon the novel insights on human mobility provided in Chapter 1, and examines the interplay between two predominant forms of transient mobility toward cities in developing countries: visits and temporary migration. Drawing from the Senegalese mobile phone data utilized in previous chapters, this study taps into the unique amalgamation of high frequencies of observation with extended observation spans offered by CDR for capturing individuals' mobility across various temporal scales. The observed patterns of visits corroborate the main findings of Chapter 1 and expand them thanks to a larger rural coverage and longer periods of observation. Then, the relation between users' decisions to visit and temporarily migrate to cities is investigated via regression analyses. The results indicate that those who engage in urban temporary migrations exhibit 17.5 additional days of visits compared to their counterparts, which is almost entirely driven by supplementary visits made to their migration destination. Furthermore, the timing of visits to migration destination is non-random; these display discernible patterns around the departure and return dates of temporary migration episodes. The mobility patterns are consistent with anticipatory and follow-up behaviors, wherein individuals willingly incur the costs of pre-migration visits to gain information about the destination, and often make subsequent visits in the weeks following their return. Finally, the simultaneous observation of both visits and temporary migration choices allows to shed light on cost differentials between these two mobility modalities. Findings from gravity regressions suggest that the inherent fixed costs of temporary migration to cities exceed those related to visits. | en |
