The Relevance of Models in Risk Management:
Models, driven by computer science, applied mathematics, and financial theory, promise to create effective risk management systems. However, questions arise about the distorted and uncontrollable nature of these models, especially at the macroeconomic and microeconomic levels. The models are promoted by regulators and universities, leading to complexity and difficulty in comprehension.

Two Key Speakers:
Nassim Nicholas Taleb: Former trader, current professor, and author, known for his critique of quantifying low probabilities. Jean-Philippe Bouchaud: Professor, founder of a quantitative investment fund, and researcher in the physics of stock markets and economic crises.

The Law of Large Numbers:
Nassim Nicholas Taleb highlights the difference between the law of large numbers taught in classrooms and its application in the real world. In finance, a law like Pareto 80-20 requires significantly more data to achieve the same level of accuracy compared to a Gaussian distribution.

Two Traditions of Risk:
Nassim Nicholas Taleb draws attention to the historical context of risk analysis, emphasizing two traditions: The probability-based approach, influenced by thinkers like Abraham de Boivre, often fails in real-world applications. The scholastic tradition, represented by Pierre de Jean-Olivier, provides a rigorous definition of risk and emphasizes the role of contracts in mitigating risk.

Attitudes Toward Risk:
There are two main attitudes toward risk: The statistician’s approach aims to provide a numerical figure for risk, which may not be entirely accurate. The merchant’s approach acknowledges that risk is inherent in life and focuses on managing it through contracts and other strategies.

Probabilistic Edges:
Probabilists often provide edges or bounds rather than precise numbers. This approach acknowledges the inherent uncertainty and complexity of risk assessment.

Insurance Industry’s Approach to Risk:
Insurers focus on managing risk by defining clear contract terms and conditions. They limit their liability and exposure to losses under specific circumstances.

Exposure Management in Options Trading:
Options traders aim to minimize exposure to losses by using contracts that limit their downside risk. This strategy involves buying contracts that change their exposure to certain risks.

Conflation of Risk and Exposure:
People often conflate risk and exposure, leading to misconceptions about the true nature of risk. Statistical risk and exposure are distinct concepts; risk assessment should focus on understanding exposure rather than predicting uncertain events.

Fragility as a Measure of Risk:
Fragility refers to the sensitivity of a system or asset to certain risks. It is easier to measure and understand fragility than to predict external risks.

Non-linearity and Fragility:
Fragile systems exhibit non-linear reactions to risks. This non-linearity can be used to detect fragility and assess the potential impact of shocks.

Scale Economies and Fragility:
Large organizations may experience scale economies, but they also become more vulnerable to shocks and fragility. This is because larger organizations have more complex structures and interdependencies.

Stress Testing for Fragility:
Stress testing can be used to detect fragility by observing the acceleration of losses under different scenarios. A series of stress tests (S3) can provide a more accurate assessment of fragility compared to a single stress test.

Non-Linearity and Survival:
Non-linearity is crucial for survival in life. Systems that remain intact, such as our bodies, exhibit non-linear risk profiles. Small perturbations may have disproportionately large effects, while larger changes might have limited impact.

Convexity and Resilience:
Convexity, characterized by accelerating losses or gains, is linked to resilience. Banks with apparent risks may instill fear, leading to risk mitigation and stability.

Time and Variability:
Those who benefit from variability also benefit from time. Time acts as an exponent in financial equations, indicating its importance in risk assessment.

The Lindy Effect and Fragility:
The Lindy effect suggests that non-perishable items that have survived for a long time have an increased life expectancy. Technologies, however, are constantly evolving and being replaced by new ones, making them fragile to time.

Fragility and Planning:
Fragility is characterized by a system’s sensitivity to shocks and its survival properties. Systems that have endured for thousands of years have demonstrated resilience against time.

Planning for the Future:
Instead of envisioning futuristic additions, planning for the future can involve identifying and removing fragile elements from the present. This approach acknowledges the unpredictable nature of technological advancements.

Conclusion:
Non-linearity, convexity, and the interplay between time and variability are essential factors in understanding resilience and fragility. Removing fragile elements from the present can provide a more realistic perspective on the future, accounting for the rapid pace of technological change.

Imperfections of First-Generation Models:
Initial models in finance and economics assumed randomness and uncertainty could be summarized by the central limit theorem, leading to Brownian motion and Gaussian statistics. This resulted in the Black-Scholes theory and the concept of perfect replication, which distorted intuition regarding risk control and risk realization. The world according to Black-Scholes is considered a “bad world” due to its deviating behaviors and poor representation of the risky world.

Emergence of Second-Generation Models:
Second-generation models emerged in response to limitations of first-generation models, attempting to integrate more realism. These models include Estone, Garch, local volatility, Levy model, multifractal model, and models for correlations. Calibrating these models to data doesn’t necessarily lead to understanding or quality; perfect fit can be misleading and distract from essential factors. Rare events that have never been observed in data can still exist and cause significant impact if overlooked.

The Need for Third-Generation Models:
Third-generation models are necessary to address the shortcomings of first- and second-generation models. These models should focus on non-predictable elements and the inherent properties of systems, rather than statistical properties. Agent models are one type of third-generation model that incorporates individual behavior and interactions to understand collective outcomes.

Example of a Second-Generation Model:
The multifractal model of Mandelbrot, Calve-Fischer, or Bakrimuzi is a mathematically elegant second-generation model. It provides a more refined representation of randomness and uncertainty compared to first-generation models.

Unreliable Models:
Financial models, including multi-fractal models, often fail to accurately predict market behavior. These models lack a solid foundation and cannot be justified from basic principles. Model calibration based on observed data can be misleading and lead to miscalibrated models. The risk of Black Swan events, unpredictable and extreme occurrences, is significant.

Efficient Market Theory:
Efficient market theory assumes that markets are efficient and provide reliable information about the true value of assets. This theory has faced challenges, as evidenced by market bubbles and crises. Fama, a Nobel Prize recipient, believes that markets are efficient and anticipate recessions. Schiller, another Nobel Prize recipient, argues that market prices are not reliable and can lead to economic crises.

Identifying Mechanisms:
A third generation of models is needed to understand financial market dynamics. These models should focus on identifying the mechanisms underlying agent behavior. Agents’ actions impact market prices, creating destabilizing feedback loops. Black-Scholes model contains destabilizing feedback loops in its strategy. Regulation can also have unintended destabilising effects.

Universality of Statistical Regularities:
Statistical regularities observed in financial markets suggest underlying principles. The law of distribution of price variations shows universality across different markets. This universality indicates the potential for modeling from more fundamental principles.

Unpredictability and Complexity:
Markets are inherently subject to what people do and their impact on prices. Retroactions through impact are crucial and contribute to market complexity. Non-linearity and heterogeneity among agents can lead to unexpected outcomes. Understanding feedback loops is essential for anticipating potential crises.

Key Insights from the Presentation:

Modeling Complex Systems:
The financial markets, like physical systems, exhibit statistical regularities that can be studied scientifically.

Limitations of Pure Mathematical Models:
Complex agent-based models often defy traditional mathematical analysis.

The Value of Computer Simulations:
Simulations can reveal phenomena that are beyond the grasp of unaided human analysis.

Surprising Emergent Phenomena:
Agent-based models can produce unexpected collective behaviors not initially apparent in the model’s design.

Understanding Correlation Disruptions:
Identifying mechanisms that disrupt correlations, such as massive liquidations, can help predict risk in times of crisis.

Snowball Effects and Trust Networks:
Trust networks can exhibit two coexisting states: high trust and low trust.

Black Swan Events and Model Predictions:
Models can predict effects before they occur, even if they cannot pinpoint the exact timing.

The Need for Predictive Theories:
Developing theories that predict unexpected outcomes is crucial for addressing systemic risk.

The Underutilized Potential of Agent-Based Models:
Agent-based models are underutilized in risk and systemic risk analysis.

Imagination and Scenario Exploration:
Digital simulations can foster imagination and help understand unlikely scenarios.

Balancing Simplicity and Precision in Modeling:
In random environments, a good model strikes a balance between simplicity and precision, avoiding overfitting and preserving intuition.

The Goal of Modeling:
A good model allows us to envision possibilities that have not yet occurred.

Conclusion:
The presenter emphasizes the importance of focusing on mechanisms and data in financial modeling, utilizing agent-based models to explore complex interactions, and embracing imagination and scenario exploration to better understand and anticipate systemic risks.

Models and Event Recalibration:
A good model is one that remains unchanged after an event, indicating its quality. In finance, models are constantly recalibrated, demonstrating their perpetual surprise at events. This implies a lack of convergence, suggesting that models are inadequate.

Black Swans and the Focus on Major Events:
Black swans, unpredictable and rare events with significant impact, should be the primary focus for decision-makers. Daily fluctuations and small events are less consequential than major, infrequent occurrences.

Pathways to Big Events and System Instability:
Dissection of financial markets reveals unstable loops of retroaction, contributing to big events. Understanding the mechanisms behind these events is crucial for preventing or mitigating their impact.

Imagination, Optimization, and Decision-Making:
Increased imagination and ecological thinking can help us avoid over-optimization and fragility in systems. Optimization efforts should be balanced with resilience and adaptability. Decision-makers should consider both scientific knowledge and practical considerations in their strategies.

Risk Analysis and Modeling Challenges:
Nassim Nicholas Taleb criticizes the RAND Corporation’s approach to risk analysis and modeling, calling it “Pakot selling” and emphasizing the futility of trying to predict specific events. In contrast, Taleb suggests focusing on the fragility of countries and systems to make more robust decisions.

The Catholic Church as a Model of Non-Fragility:
Taleb presents the Catholic Church as an example of a non-fragile organization due to its decentralized structure, operational diversification, and ability to absorb small shocks. He attributes the Church’s longevity (2,400 years) to its adherence to the principle of subsidiarity, where decisions are made at the lowest possible level.

The Soviet Union as an Example of Fragility:
Taleb characterizes the Soviet Union as a fragile system due to its centralized decision-making, lack of volatility, and inability to adapt to unexpected changes. He highlights the importance of system diversity and the ability to learn from errors and accidents.

Systems That Know How to Work with Shocks:
Taleb identifies transportation systems as examples of systems that can handle uncertainty and errors effectively. He points out that accidents in these systems often lead to improved safety measures and reduced risks.

The Role of Heuristics and Regulation:
Taleb emphasizes the value of heuristics in decision-making, arguing that good models should not undermine these heuristics. He criticizes the shift from heuristics to pseudoscientific methods like Value at Risk in finance, attributing this change to regulators who lack the necessary expertise. Taleb contrasts the effective regulation in air transport with the failed regulation in finance, highlighting the need for regulators with a deep understanding of the specific fields they oversee.

Asymmetrical Consequences:
Regulators often centralize systems, making them more susceptible to failure. The concept of “skin in the game” is essential for mitigating risk and encouraging responsible decision-making. In ancient times, laws like Hammurabi’s code enforced the principle of suffering consequences for one’s actions.

Importance of Personal Responsibility:
Individuals who have the potential to cause harm should also bear the consequences of their actions. The idea of “skin in the game” acts as a filter, deterring reckless behavior and promoting accountability.

Examples of Skin in the Game:
In Brazil, helicopter repairmen were required to take random flights in the aircraft they repaired, reducing accidents. Roman engineers were forced to sleep under the bridges they built, ensuring the quality of their work.

Application in Finance:
Financial risks arise from two sources: those caused by randomness and those due to inadequate risk assessment. Regulators often fail to address the latter, leading to systemic failures.

Conclusion:
Skin in the game is a crucial concept that promotes personal responsibility and discourages reckless behavior. Regulators should prioritize subsidiarity and encourage individual accountability to prevent systemic failures.

Incentives and Risk:
In the capitalist system, incentives often lead people to prioritize personal gain over risk management. This can lead to decisions that benefit individuals in the short term but harm the long-term viability of organizations.

Hedge Funds and Skin in the Game:
Hedge funds are an exception to the lack of skin in the game in finance. Hedge fund owners typically have a significant portion of their own wealth invested in their funds, which aligns their incentives with those of their clients.

Scandal Game:
The “Scandal Game” refers to situations where individuals make decisions that increase short-term gains while transferring the associated risks to others. This can have severe moral and societal repercussions.

Tension between Short-Term and Long-Term Research:
There is a tension between short-term research focused on immediate solutions and long-term research that aims for fundamental understanding. Both types of research are important, and neglecting one at the expense of the other can lead to problems.

Historical Models and Data-Driven Approaches:
Economic models based on historical data can be unreliable, as they may not capture rare, unpredictable events with significant impacts. Data-driven approaches and experiments can provide more accurate insights.

The Problem of Regulation:
Current regulations, such as those in insurance, can incentivize companies to act in ways that limit volatility and increase homogeneity, which can have negative consequences for the overall system.

Skin in the Game for Consultants and Economists:
Consultants and economists often receive payment for their advice, regardless of whether that advice is accurate or leads to positive outcomes. This lack of skin in the game can lead to poor decision-making and a lack of accountability.

The Importance of Penalties for Errors:
In many fields, individuals who make errors do not face significant consequences, which can lead to a lack of accountability and a perpetuation of poor practices. Penalties for errors can help to align incentives and improve decision-making.

Historical Examples of Skin in the Game:
Throughout history, societies have used various methods, such as punishments for cowardice or financial penalties, to ensure that individuals have skin in the game and face consequences for their actions.

The Dangers of Informational Modernity:
In the modern world, it is easier for individuals to speak and make decisions without taking personal risks. This can lead to a lack of accountability and an increased risk of poor decision-making.

Reconsidering the Concept of Black Signs:
Nassim Nicholas Taleb questions the objective existence of black signs and emphasizes their epistemic nature. Black signs depend on an individual’s knowledge and model of the world, and can vary based on their understanding.

Understanding Black Signs and Their Impact:
Taleb suggests that we should not attempt to understand black signs too deeply, but rather focus on building systems that are less reliant on predicting them.

The Relationship between Robustness and Black Signs:
The more optimized and dependent a system is on external events, the more it relies on understanding black signs. Conversely, robust systems are less dependent on understanding black signs and can withstand their impact.

Identifying System Properties:
Taleb emphasizes the importance of identifying system properties that reduce dependence on black signs. This involves focusing on the “sign of a black sign” (SI), which provides information about the potential impact of black signs on a system.

Skill in the Game and Black Signs:
“Skill in the Game” allows individuals to recognize and mitigate the impact of black signs, effectively removing them as black signs in their eyes. It addresses hidden risks that tend to emerge simultaneously and systematically, often associated with people hiding risks in their wallets or when companies go public.

Public Companies and Their Demise:
As soon as a company goes public, it is doomed to decline due to the delegation of decision-making to photogenic actors and the introduction of policies focused on volatility reduction and risk minimization. The longevity of family-managed companies, which maintain traditional practices, is significantly higher compared to public companies. Public companies often take on excessive debt and make policy decisions aimed at eliminating volatility and lowering risks, which can lead to their eventual downfall.

The Role of Photogenic Actors and Skin in the Game:
The appointment of photogenic individuals as company leaders, often lacking entrepreneurial skills, triggers a countdown to bankruptcy. Skin in the game, where decision-makers have a personal stake in the company’s success, eliminates the problem of misaligned incentives and reckless decision-making.

The Overreliance on Financial Mathematics:
Financial mathematics, particularly the Black-Scholes model, has been widely adopted in finance despite its limitations and failure to predict extreme events. The current regulatory framework in Europe substitutes good practices and mathematical theories with a primitive version of financial theory, potentially leading to adverse consequences.

Insurance and the Management of Extreme Risks:
Insurance can be divided into two categories: Centails insurance for distributed risks and Fatel insurance for extreme risks. Insurers have traditionally managed extreme risks effectively through techniques such as reinsurance, where the focus is on minimizing risk exposure after catastrophic events. The replacement of these traditional methods with value-at-risk methods based on flawed mathematical models can lead to increased financial instability.

Invisible Risks and the Maximization of Hidden Dangers:
In certain environments, invisible risks tend to increase. This phenomenon is attributed to the tendency of regulations to focus on visible risks, leading to the neglect of hidden dangers that may pose a greater threat to the financial system.

Overview:
This summary presents key insights from a discussion among experts on the challenges of assessing and predicting financial risks. The speakers emphasized the limitations of current models and proposed a research agenda to develop more robust approaches.

Key Points:

1. Regulation and Its Impact:
Regulations often lead to unintended consequences, as people optimize their behavior to maximize what is not covered by the regulations.

2. Questioning the Nobel Prize:
The award of the Nobel Prize to Fama after the 2008 crisis raises questions about the validity of the efficient market hypothesis and the reliance on models that do not adequately capture extreme events.

3. The Importance of Theoretical Background:
The education and theoretical background of individuals play a crucial role in risk management, as biases and prejudices can influence the development of regulations.

4. Challenges of Modeling Extreme Events:
Current models often exclude the possibility of extreme events, leading to a false sense of security. There is a need to develop models that can capture the potential for improbable but highly impactful events.

5. The Agent Model Approach:
Agent-based models offer a promising approach to understanding complex systems, as they allow researchers to observe emergent collective phenomena that are not apparent from individual behaviors.

6. The Need for a Real Science of Economics:
Despite the importance of the economic system, there is a lack of scientific rigor in its study compared to other disciplines. More resources and efforts should be dedicated to developing a comprehensive understanding of the economic system.

7. Classifying Outliers and Invalidation:
Outliers can be classified based on their impact on models. Some models are invalidated by extreme events, while others are more robust. The goal is to identify the characteristics that invalidate models and develop more resilient approaches.

8. Networks and Contagion:
The study of networks can provide insights into how contagion spreads, helping to understand the mechanisms behind extreme events.

9. Identifying Black Marks:
Developing a systematic approach to identify black marks, or indicators of potential risks, is crucial for early detection and mitigation. These indicators should be integrated into decision-making processes to improve risk management.

10. Robustness and Heuristic Methods:
Focusing on robustness, rather than specific events, can lead to more resilient risk management strategies. Heuristic methods, while not strictly scientific, can provide valuable insights and practical solutions.

11. Rethinking the Notion of Models:
There is a need to challenge the traditional view of models as predictive tools and instead focus on their ability to capture plausible scenarios and guide decision-making.

12. Embracing Turbulence and Complexity:
The recent turbulence in financial markets highlights the need to understand the complex interactions between technologies, actors, and regulations.

13. Shifting from Heuristics to Models in Insurance:
Similar to asset management, the insurance industry has undergone a shift from heuristics to models, leading to challenges in regulation, risk management, and asset allocation.

Conclusion:
The speakers emphasized the need for a fundamental shift in the approach to financial risk assessment and prediction. They proposed a research program focused on identifying black marks, developing robust models, and embracing complexity to better understand and manage financial risks.

Decontextualization and Ignoring Consequences:
In today’s insurance, as in yesterday’s titration, there has been a tendency to decontextualize and trust theories, ignoring the consequences of this evolution in terms of risk perception and responsibility.

Sociological, Psychological, and Human Factors:
The sociological, psychological, and human sides have been neglected, leading to the failure of rational mechanics that relied on established hypotheses.

Multidisciplinary Approach:
To move forward, a multidisciplinary approach is needed, combining epistemological and sociological dimensions with the skills of ENSAE and Sciences Po.

Various Issues:
There are various issues to address, including technical, management, business, council and prevention, political, regulation, responsibility, and governance.

Three Areas of Impact:
The research will have three areas of impact: tomorrow’s insurance, strategic allocation of assets, and prudential regulation, with an operational dimension.

Website, Newsletter, and Working Papers:
The research team has a website, sends out a newsletter, and publishes working papers.

Monthly Seminars:
Monthly seminars will be organized starting in September, and participation is welcome.

Inspiration and Different Frame:
The research team is inspired by the presented elements and sees them as a source of inspiration for setting a different frame for insurance and risk perception.