The activity of the Finance of Energy Markets Laboratory is organised into three main themes: (1) consumers, (2) production, (3) the financial environment. These three research themes are backed up by two cross-cutting methodological axes: stochastic numerical methods and statistics. The research activities are steered collegially by a committee composed of René Aid, Olivier Féron, Delphine Lautier, Peter Tankov, Nizar Touzi and Bertrand Villeneuve. It should be noted that while this structuring of the research axes (which has proved to be very suitable, particularly in terms of readability) remains unchanged, the content of the axes will be subject to evolution.
For the period 2022-2026, a particular effort will be made to analyse the functioning of markets and their regulation, in particular in the context of the evolution of European markets characterised by the generalisation of 'energy only' markets. In particular, we will seek to analyse the dynamics of investments in decarbonised energies - by nature, very capital-intensive - in a financial environment marked by significant price fluctuations. Another aspect of this issue is the evaluation of the cost of capital from a risk management perspective, and the inclusion of environmental impacts in this evaluation, through the typologies of GHG emissions accounting (taxonomies, scopes).
1 - Consumers
This research theme aims to understand the interaction between the centralised production system and the customers (private individual/industry/local producer) and to provide new methods for steering/coordinating the latter.
1. Pricing: pricing and customer behaviour taking into account the information asymmetry between supplier and customer; social pricing; taking into account competition between suppliers.
2. Consumption management: incentive mechanisms for consumption management for a better reliability of the electricity system.
3. Development/management of local production: assistance in the analysis of tariff regulations; optimal subsidy policy for technological transition.
2 - Generation
The objective of this research theme is to provide decision support tools in the long term, for investment and disinvestment choices, and in the short term, for the optimisation of existing production capacities.
1. In the long term, the aim is to develop production/consumption models that take into account the specific uncertainties for an energy company caused by market changes: among other things, the development of renewable energies, the introduction of capacity markets, the arrival of new competitors, the introduction of auction systems, etc. These models could be developed by integrating behavioural phenomena such as the differentiated weight assigned to gains and losses.
2. In the shorter term, the aim is to set up methods for optimising adapted production and distribution assets. These methods will have to take into account the characteristics of the new systems as well as the appearance of new services for network management: modelling of local photovoltaic/wind production (for micro-grids), management of multi-energy and micro-grid systems, management of several storage facilities, joint price/alias models, stress test scenarios, etc.
3 - The financial environment
This research theme aims to better understand and apprehend the functioning of energy markets, its interactions with its environment and financial risks.
1. Functioning of energy and related markets: study of price formation in energy markets; role of fundamentals (production, storage) in the value of assets; study of market power and its impact on price formation; study of new or maturing markets such as the CO2 market and the capacity market; study of markets with high frequency data such as the intraday market.
2. Interaction of energy markets with their environment: study of transmissions or contagions between paper and physical markets for the same commodity, between different commodity markets, and finally between commodities and financial markets (financialisation of commodities); study of market regulation, link between environmental policy and competition policy
3. Long-term risk, in particular for the portfolio dedicated to the decommissioning of nuclear power plants: adapted diffusion models and application of robustness techniques; portfolio management and allocation strategy adapted to the long term; quantification of uncertainties adapted to long-term models.
Cross-cutting methodological axes
The general objective is to provide models that focus on fundamental phenomena in order to understand the mechanisms at work and to study the sensitivity of the results obtained to the parameters of the models. This prioritisation of problems is part of the research work, in order to avoid models that are hyper-realistic in intent but incomprehensible and unsolvable in reality. Resolutions can be exact or numerical. In some cases, the methodological innovation lies in the numerical part, which makes it possible to inform the optimal decisions of the actors. In other cases, statistical techniques are essential for estimating and analysing the impact of model inputs and outputs and the interactions between different effects.
1 - Numerical methods :
The problems addressed are often optimisation problems in an uncertain universe and may take into account several actors. In this context, the solutions are typically characterised by non-linear partial differential equations.
Some examples of techniques developed: uncertainty quantification methods (e.g. for models with partially observable variables); branching techniques; variance reduction techniques; parallelization and high-dimensional management; "data driven" algorithms (with self-calibration); development of numerical method libraries (free software) and test banks (application benchmarks to quantify the progress of numerical tools and to allow for the reproducibility of tests); deep learning methods for control problems (the latest advances allowing for the treatment of much higher-dimensional problems)
2 - Statistical methods :
The objectives of this axis are to estimate and analyse the impact of model inputs and outputs and the interactions between different effects. More precisely, this concerns problems of estimation, taking into account modelling errors and information extraction. Examples of techniques developed include: dependence of extreme events on explanatory variables; high frequency data processing; model errors and uncertainty propagation; simulation of rare events and stress test scenarios; models with unobservable variables. It also includes statistical learning methods, including deep learning (neural networks).