Seed Projects are intended to foster application-oriented basic research by offering a pragmatic opportunity to explore and develop visionary ideas of high social relevance in an early stage towards a concept for further research, to be eventually funded by regular funding mechanisms and bodies.
This funding opportunity for postdoctoral projects of up to one year is financed with a maximum budget of 100,000 CHF per project. The funds for Seed Projects are provided by donations from the Industry Partners of the Risk Center.
Each Seed Project is supervised by two RC Professors, ideally from two different departments. The goal of the projects is to connect modern methods and approaches to traditional risk domains in order to gain new insights into risk in markets and systems, and to develop ideas for new research projects. Each Seed Project creates impact in their respective fields.
2016 Seed Projects
How a complex network is connected crucially determines its dynamics and funtion. Percolation, the transition from small-scale to large-scale connectedness by gradual addition of links, occurs during variable growth in a large variety of natural, technological and social networked systems. By employing and extending recently developed methods in Explosive Percolation (EP) we will study the effect of repeated small interventions in complex networked systems, design to delay the onset of large-scale connectedness.
RC Professors: Hans Jürgen Hermann and Didier Sornette - Members
PostDoc: Jan Nagler - PostDocs
The purpose of the seed project is that in Switzerland the increased number of initiatives - i.e. proposals that can be submitted to voting by the citizens - is often perceived as an impediment to the proper functioning of direct democracy. To potentially reduce the number of initiatives and to improve the quality of collective decision making without limiting the right to initiate legislation, we suggest to develop a new two-round voting procedure. We will conduct an analysis of the welfare benefits and risks of the above-mentioned voting procedure. We will also identify information security risks and analyze ways to mitigate them when implementing assessment voting.
RC Professors: David Basin and Hans Gersbach - Members
PostDoc: Philippe Muller - PostDocs
Uncertainty quantification (UQ) is a research field that has recently emerged at the boundary between statistics, applied mathematics and engineering. UQ techniques have been successfully employed in a wide set of applications in the past few years. At the foundation of uncertainty quantification techniques lies the definition of an accurate probabilistic model of the system uncertainties. Therefore, with the increase in availability of large data volumes (e.g. from monitoring, historical recordings, quantitative simulations, etc.), the need of constructing detailed probabilistic representations of real data is becoming a central topic in the field. The dependence structure between the parameters plays a crucial role for describing accurately complex data-based probabilistic models, especially when the estimation of low probability events is considered. However, it is often either over-simplified or simply ignored. Copula theory and inference form a natural framework for representing this dependence in big-data-type applications. In this project our aim is to develop a suitable set of techniques and software tools based on high dimensional copulas and copula composition (e.g. vine copulas and associated inference techniques) to enable the usage of state-of-the art uncertainty quantification algorithms (with a focus on surrogate models and model resampling techniques) on real high-dimensional [big-]data streams.
RC Professors: Bruno Sudret and Paul Embrechts - Members
PostDoc: Emiliano Torre - PostDocs
The purpose of the seed project is to understand the systemic dimension of privacy risks emerging from online interaction. Today, digital traces generated by millions of users allow to infer private attributes of individuals that may not even use online media. This poses a considerable risk to privacy which needs to be conceptualized, quantified and measured. Hence, as a first step, we will explore ways to construct shadow profiles, i.e. aggregated information about individuals that was not provided by them, but inferred from the information disclosed by other users and their interactions. This allows us to reveal the conditions under which privacy risks can occur, to eventually mitigate them.
RC Professors: Frank Schweitzer and David Basin - Members
PostDoc: David Garcia Becerra - PostDocs
2015 Seed Projects
The goal of this project is to understand systemic risk in collaboration networks. We want to predict how the failure of one or a few agents can hamper the performance of the network. We intend to define a quantitative risk measure (resilience), that can be validated and tested on a set of real collaboration networks. We want to provide a quantitative definition of network performance for different instances of real collaboration networks and intend to develop an agent-based model that allows for the fine-tuning of some key microscopic parameters.
Democratic governments tend to accumulate excessive debt. We propose a new rule - the “Catenarian Fiscal Discipline” - which allows a fiscally-disciplined incumbent to limit the next officeholder’s debt-making. This way, today’s fiscal discipline can generate the fiscal discipline of the future. Such a rule would require that we broaden our notion of representative democracy, of course, but not as much as it seems, as current governments already have various ways of limiting their elected successors’ scope of action.
Economic policy conclusions are sensitive to the parameterization of economic models. In this project we will apply to economic models the formal, so-called, global sensitivity analysis that has been developed in the engineering literature. These methods are approximation methods tailored to the needs of checking the robustness of conclusions given some uncertainty over the parameterization. The techniques are similar to the approximation methods that are used in economics for the approximation of decision functions given the states of a model.
Modeling of Business Interruption (BI) is still in a relative state of infancy lacking transparency that would allow corporation managers and insurance agents to understand and control risk exposure in Supply Chain Networks (SCNs) of large corporations. The purpose of this research proposal is to develop a transparent, holistic risk management framework for BI due to natural hazards (such as earthquakes, hurricanes, and floods). Based on recently released post-disaster recovery functions of building inventory and infrastructures, the proposed study will use a stochastic approach to model different SCN structures and assess how functionality of a business is affected after a natural disaster as it tries to resurrect its facilities, obtain raw materials and continue distributing its products upstream.
2014 Seed Projects
Complex networks arise in many fields of science, economy, and engineering. Electricity grids, social networks, financial networks and evacuation plans are examples of complex networks. An important aspect in the study of risk in complex networks is that of cascades. The goal of the project is to device algorithmic methods for protecting complex networks against such risks. Concretely, given a limited budget, the decision maker has to devise a plan to reinforce some parts of the network in order to prevent a cascade. The project aims to formally define relevant optimization problems, device algorithmic resolution methods for them, and study their practical applicability to real life problems.
The aim of this project is to develop a prototype framework for modeling and evaluation of the effectiveness of structural and financial seismic risk mitigation measures in Switzerland. The intended use of this framework is to investigate which structural and financial risk mitigation measures are appropriate for Switzerland and how to calibrate their mix to achieve the highest possible seismic risk mitigation at the lowest cost.
We will develop experimental contexts where test-bed asset markets can be created, rigorously studied, and different interventions can be systematically evaluated. These experiments will be done with real decision makers and pecuniary performance based incentives. A major goal of this project is to isolate casual mechanisms that contribute to the formation and growth of financial bubbles.
This project’s goal is a risk assessment of traffic performance which includes large-scale traffic gridlocks. For this purpose, traditional models based on aggregates are available and will be consulted in this project. However, driven by the improvement of transport modeling through disaggregate models with the individual as the basic modeling unit, this project will incorporate consistent gridlock modeling in such models.