Quantifying human suffering for humanitarian logistics: deprivation cost versus deprivation level

Optimization models for humanitarian logistics increasingly incorporate human suffering as an optimization criterion. Two specific metrics have been proposed: deprivation costs and deprivation levels. However, it is unknown how these two relate and how the choice of a metric affects results. Furthermore, little is known about how strongly both are driven by disaster experience and individual characteristics. As such, it is unclear whether the common practice of using a single function to estimate deprivation costs or levels for people is valid. We address these gaps. We conduct online surveys and a field study in China to collect data about deprivation costs and levels for three relief items from 1105 respondents. Regression analyses are used to analyze the data and derive deprivation cost and deprivation level functions for future optimization models. Our first key finding is that the two metrics are related but not linearly. Instead, deprivation costs increase exponentially with the deprivation level. As we illustrate using a case study, relief item distribution decisions that are optimal according to one metric can therefore be strongly suboptimal according to the other. Second, individual characteristics and disaster experience substantially affect deprivation costs but hardly affect deprivation levels. This suggests that deprivation cost functions are more susceptible to hypothetical bias and cannot trivially be generalized, constraining their applicability in optimization models. Our results induce new questions about the respective validity and biases of the two metrics and lay the groundwork for future research in this domain.