Abstracts and slides

Stefano Battiston (ETHZ) DebtRank : Too Central to Fail ? Financial Networks, the FED and Systemic Risk


Systemic risk, meant as the risk of distress of a large portion of the financial system, depends on the network of financial exposures among institutions. However, there is no widely accepted methodology to determine the systemically important nodes in a network. To fill this gap, we introduce, DebtRank, a novel measure of systemic impact inspired by feedback-centrality. As an application, we analyse a new and unique dataset on the USD 1.2 trillion FED emergency loans program to global financial institutions during 2008-2010. We find that a group of 22 institutions, which received most of the funds, form a strongly connected graph where each of the nodes becomes systemically important at the peak of the crisis. Moreover, a systemic default could have been triggered even by small dispersed shocks. The results suggest that the debate on too-big-to-fail institutions should include the even more serious issue of too-central-to-fail.

Marc Bocquet (ENPC) Towards using data assimilation in macro-economic dynamical models


The goal of this talk is to report very preliminary results on the use of data assimilation methods applied to a macro-economic model. We will first give a brief account on data assimilation methods, including very recent ones, that could be helpful towards this goal. Several examples will be used to sketch what data assimilation can currently achieve in the geosciences.
A macro-economic non-equilibrium dynamical model (NEDyM) has been shown to exhibit endogenous business cycles (see talk by P. Dumas). NEDyM is used in the present work for illustration purposes. This work’s far-reaching objectives are to : (i) adjust some of the model’s key parameters, relying on data assimilation and using real data ; (ii) perform an economic forecast based on the model thus adjusted ; and (iii) perform a re-analysis of past economic history.
Although the model is of small dimensionality, it is highly non-linear, and, moreover, may exhibit chaotic behaviour. This imposes some technical but crucial constraints on the data assimilation methods to be used.
We will present very preliminary results in perfect model conditions, including some experiments on the controllability of the system. Some of the difficulties one has to overcome to achieve our ambitious goals will be discussed.
This talk presents joint work with P. Dumas, M. Ghil, and A. Groth.

Erik Chavez (Imperial College) Modelling of climate change impacts

Weather variability affects economic activity significantly. Crops yields, energy consumption or traffic flows are shaped by weather variability. Government estimates show that approximately a third of the US economy (4 trillion USD) is affected by weather, while over three fourths of several sub-Saharan countries’ economies are sensitive to it due to their reliance on agricultural activity. Several policy assessments have been carried out to determine optimum mitigation and adaptation paths that address the challenges raised by changing climates globally. Nevertheless, the disconnect between economic impact assessments and underlying physical processes is widely acknowledged. In order to address the latter, a cascading climate-to-economy weather index-based impact modelling approach is introduced. The approach is applied to the rural sector in North and South China. Several risk management applications of this modelling approach, such as macro financial risk transfer instrument building, are presented.

Patrice Dumas (CIRED) Non-equilibrium dynamic model and impacts


Extreme event consequences are dependent on the dynamics of reconstruction, as reconstruction diverts ressources and facilities are not operational until they are repared. A nonlinear business cycle model, NEDyM, is used to model the dynamics of reconstruction, being able to represent the preexisting disequilibria and endogenous business cycles. The model also takes into account a limitation in investment that may be redirected to reconstruction. An interesting result is that extreme event happening during recessions may lead to less severe consequences, since the reconstruction will boost economic growth in such situations. Conversely, extreme event reconstruction could enhance preexisting tensions in a period of expansion. This framework has also been used to show possible effects of a series of extreme events and allow modelling embedded technical change in reconstruction.
A second series of works study the limitations on investment redirection to reconstruction at more detailed scales, trying to understand and quantify the associated processes. Three approaches are presented, one looking at labor mobility, the other at inter-industry constraints with the ARIO model, while the last study examines how a disruption of businesses network may worsen extreme events consequences.

Johannes Emmerling (FEEM) Geoengineering and uncertainty


The potential of geoengineering as an alternative or complementary option to mitigation and adaptation has received great interest in recent years. Nevertheless, uncertainties about the effectiveness, costs, and potential detrimental effects are substantial. In this paper, we study the effect of Solar Radiation Management (SRM) becoming available in the future on the optimal abatement path. Our focus is the uncertainty about the effectiveness of SRM and the interaction with uncertain climate change parameters. Our analytical results suggest that optimal abatement is decreasing significantly only if SRM is very likely to be effective. Using a stochastic version of the WITCH Integrated Assessment model, this result is confirmed even abstracting from negative side-effects from geoengineering and for low degrees of risk aversion.

Michael Ghil (ENS & UCLA) Boolean delay equations and damage propagation on networks


Boolean Delay Equations (BDEs) are semi-discrete dynamical models with Boolean-valued variables that evolve in continuous time. Systems of BDEs can be classified into conservative or dissipative, in a manner that parallels the classification of ordinary or partial differential equations. Solutions to certain conservative BDEs exhibit growth of complexity in time ; such BDEs can be seen therefore as metaphors for biological evolution or human history. Dissipative BDEs are structurally stable and exhibit multiple equilibria and limit cycles, as well as more complex, fractal solution sets, such as Devil’s staircases and "fractal sunbursts." 
BDE systems have been used as highly idealized models of climate change on several time scales, as well as in earthquake modeling and prediction, and in genetics. BDEs with an infinite number of variables on a regular spatial grid have been called "partial BDEs" and we discuss connections with other types of discrete dynamical systems, including cellular automata and Boolean networks.
We present here recent BDE work on damage propagation in networks, with an emphasis on production-chain models. This formalism turns out to be well adapted to investigating propagation of an initial damage due to a climatic or other natural disaster. We concentrate on two different network structures, periodic and random, respectively ; their study allows one to understand the effects of multiple, concurrent production paths, and the role played by the network topology in damage propagation. Applications to the recent network modeling of climate variability are also discussed.
This talk presents joint work with B. Coluzzi and G. Weisbuch (ENS), D. P. Dee (ECMWF), S. Hallegatte (CIRED & World Bank), V. I Keilis-Borok (UCLA), A.P. Mullhaupt (SUNY Stony Brook & Wall Street), P. Pestiaux (Total), A. Saunders (LAUSD), and I. Zaliapin (UNR).
[1] Coluzzi, B., M. Ghil, S. Hallegatte, and G. Weisbuch, 2011 : Boolean delay equations on networks in economics and the geosciences, Intl. J. Bif. Chaos, 21 (12), 3511–3548, doi : 10.1142/S0218127411030702.
[2] Ghil, M., I. Zaliapin, and B. Coluzzi, 2008 : Boolean delay equations : A simple way of looking at complex systems, Physica D, 237, 2967–2986, doi : 10.1016/j.physd.2008.07.006.

Andreas Groth (ENS) Singular spectrum analysis and synchronization


We live in a dynamic world and there are numerous difficulties in trying to study the coupled behavior of the socio-economic system and the climate system. Each of the systems is highly complex and nonlinear, and it possess variability on a wide range of time and space scales. We address the problem of extracting spatio-temporal information about recurrent behavior from short and noisy data by means of Singular Spectrum Analysis (SSA) In this talk, we show that the SSA approach, which requires no detailed knowledge of individual subsystems nor a suitable phase definition for each of them, provides an attractive alternative to commonly used phase-based approaches in the detection of cluster synchronization in large systems. In an example on macroeconomic indicators, the capability of SSA is shown to disentangle and simplify the interpretation of the complex business-cycle dynamics in a multi-country setting. Though, there is still no accordance about the characterization of comovements, the existence of supranational (e.g. European, G7) cycles, and the determinants of economic synchronization, we will contrast the prevailing theory of random shocks with that of endogenous dynamics.

Klaus Hasselmann (Max Planck Institute for Meteorology, Hamburg) Climate change and the financial crisis


The financial crisis has displaced climate change from the political agenda. Yet the two are intimately connected : overcoming the present recession requires major government stimulus for investments. And investments in renewable energy and sustainable technology provide an ideal development target. However, to support the green transformation, scientists need to develop new, easily understandable coupled climate-socioeconomic models that include the financial system, together with its inherent actor-driven instabilities. A simple system-dynamic, actor-based model is presented that illustrates the present conflict between austerity and Keynesian approaches to overcoming the Euro crisis, with implications for climate policy.

Jürgen Kurths (PIK & Humboldt University) Network of networks

Download part I
Download part II

Olivier Talagrand (ENS) Data assimilation : Basics and meteorology


Assimilation of observations, which originated from the need of defining initial conditions for numerical weather forecasts, is the process by which observations of the state of the atmospheric flow is combined together with a numerical model of the dynamics of the flow. The numerical dimension of the problem (the state vector of the model can have dimension as large as 109), together with the complexity of the underlying dynamics (not to mention the need for the forecast to be ready in time), make assimilation a particularly challenging problem, from the point of view of both theory and practical implementation.
Assimilation of observations is best expressed as a problem in Bayesian estimation. Namely, determine the probability distribution for the state of the system under observation, conditioned to the available information. The main two present types of assimilation algorithms, Variational Assimilation (VA) on the one hand, and Ensemble Kalman Filter (EnKF) on the other, are both empirical extensions, to weakly nonlinear situations, of Best Linear Unbiased Estimation (BLUE), which achieves Bayesian estimation in linear and Gaussian cases. Variational Assimilation globally adjusts a model solution to observations distributed over a period of time. Ensemble Kalman Filter evolves in time an ensemble of estimates of the system state, meant to span at any time the uncertainty of the state of the flow. It updates those estimates, following a basically Gaussian scheme, as new observations become available.
Particle Filters, like EnKF, evolve an ensemble of state estimates, but use an exactly Bayesian updating scheme. However, their cost has so far been too high for use for large dimension systems. Their performance, and possible future evolution, are discussed.
Assimilation is closely linked to the instabilities in the observed systems, and essentially consists in a sense in monitoring and controlling those instabilities. The links between assimilation and instabilities are discussed, and two specific algorithmic procedures are presented in this context : Quasi-Static Variational Assimilation (QSVA) and Assimilation in the Unstable Subspace (AUS).
Assimilation, in the form in which it has been developed for numerical weather prediction, has progressively extended to various applications, first in geophysics, and then to other fields. A few examples are briefly discussed.

Pietro Terna ( University of Torino) Agent-based models for exploring economic complexity


Thinking to agent-based models as artifacts, useful to explore economic complexity, means to introduce three concepts : (i) on the technical side, the agent-based methodology ; (ii) in the social science perspective, the idea of building artifacts also in the social domain ; (iii) in a more general view, the idea of complexity. With an impressive image, this way of researching involves the application of the Galileo’s method in social sciences. Remembering the theoretical roots in cybernetics and, more recently, in complexity science, we need also technical roots, with the capability of building models that could be accepted by a wide audience, comparing classical model and new ones, or proposing as well hybrid structures. Finally, we need to have a close look to agent-based models and their structures, with simple and complicated methodologies and to pay a lot of attention to the abilities of the agents, mainly in mimicking the human capability of learning and adapting.

Bert de Vries (University of Utrecht) Framework and prospects for integrated assessment modelling of global change


In modelling global change, we have to bridge the micro and macro, the natural sciences and the social sciences, and science and policy. This requires new modelling methods and paradigms. I will give a brief overview of some novel and promising approaches, that may become important building blocks for future global change (meta)models.



■ Adresse postale
École Normale supérieure
45 rue d’Ulm
F-75230 Paris cedex 05

■ Direction
Alessandra Giannini, Marc Fleurbaey

■  Responsable de l’axe agro-écologie et de l’axe transdisciplinarité

Corinne Robert : corinne[at]robert

■  Responsable pédagogique

2022-24 : Marine Fauché : marine.fauche[at]

2020-22 : Gaëlle Ronsin : gaelle.ronsin[at]

■ Administration
 Kadija Aouni : kadija.aouni