Globally, more than 130 million people are estimated to be in food crisis. These humanitarian disasters are associated with severe impacts on livelihoods that can reverse years of development gains. The existing outlooks of crisis-affected populations rely on expert assessment of evidence and are limited in their temporal frequency and ability to look beyond several months. This paper presents a statistical foresting approach to predict the outbreak of food crises with sufficient lead time for preventive action. Different use cases are explored related to possible alternative targeting policies and the levels at which finance is typically unlocked. The results indicate that, particularly at longer forecasting horizons, the statistical predictions compare favorably to expert-based outlooks. The paper concludes that statistical models demonstrate good ability to detect future outbreaks of food crises and that using statistical forecasting approaches may help increase lead time for action.
Recent advances in food insecurity classification have made analytical approaches to predict and inform response to food crises possible. This paper develops a predictive, statistical framework to identify drivers of food insecurity risk with simulation capabilities for scenario analyses, risk assessment and forecasting purposes. It utilizes a panel vector-autoregression to model food insecurity distributions of 15 Sub-Saharan African countries between October 2009 and February 2019. Statistical variable selection methods are employed to identify the most important agronomic, weather, conflict and economic variables. The paper finds that food insecurity dynamics are asymmetric and past-dependent, with low insecurity states more likely to transition to high insecurity states than vice versa. Conflict variables are more relevant for dynamics in highly critical stages, while agronomic and weather variables are more important for less critical states. Food prices are predictive for all cases. A Bayesian extension is introduced to incorporate expert opinions through the use of priors, which lead to significant improvements in model performance.
In this paper we explore the reasons for the trend reversal in the development of household market income inequality in Germany in the second half of the 2000s.
14 December 2017 in Epidemiology and Psychiatric Sciences
This paper revisits the credit spread puzzle in bank CDS spreads from the perspective of information contagion. The puzzle, first detected in corporate bonds, consists of two stylized facts: Structural determinants of credit risk not only have low explanatory power but also fail to capture common factors in the residuals (Collin-Dufresne et al., 2001). For the case of banks, we hypothesize that the puzzle exists because of omitted network effects. We therefore extend the structural models to account for information spillovers based on bank business model similarities. To capture this channel, we propose and construct a new intuitive measure for portfolio overlap using the complete asset holdings of the largest banks in the Eurozone. Incorporating the network information into the structural model for bank credit spreads increases explanatory power and removes a systemic common factor as well as a North-South common factor from the residuals. Furthermore, neglecting the network likely overstates the importance of structural determinants.