This article addresses the complexity within life and annuity risk through the lens of systems science. Whereas most scientific disciplines drill downward into ever more minute details, systems science looks upward to discover how different systems interact to understand the big picture and explain the commonalities among different systems.
To describe why complexity is an emerging risk, I first will provide a brief history lesson, so we can compare today’s insurance business paradigm to past paradigms. Second, I will describe the competing systems and the challenges they present. Lastly, I will discuss how to make organizations resilient to change by making their people and processes more adaptable.
History of the World, Part III
CIPR Study, State of the Life Insurance Industry: Implications of Industry Trends has a wonderful and detailed account of the history of the insurance industry. I categorize the history into three waves of insurance regulation, which were brought on by major market dislocations:
- Pre-Great Depression
- Post-Great Depression
- Post-2008 Great Recession
The first wave established the foundation of the regulation that exists today. The second wave response to the Great Depression saw a large spike of new regulations that had a significant impact on banks and the insurance industry. During the start of the second wave, insurance companies were not allowed to hold corporate stock, liabilities were diversified by adding more policies and assets were safe and held to maturity. This shielded insurance companies from the brunt of the Great Depression, which allowed them to act as a source of liquidity.1 We currently are living through the third wave, which is represented in the timeline in Figure 1.
Figure 1: Third Wave of Insurance Regulation
Financial Services Modernization Act
Y2K and introduction to the internet with Internet Explorer
Long-term capital management collapses
Financial Services Modernization Act of 1999
C3 – Phase II
Actuarial Guideline 43
Valuation Manual 20
Valuation Manual 21
Long-duration targeted improvements
The genesis of the third wave started in the early 1990s but was not ushered in until 2008 with the Great Recession. We have many forces currently at play that are different this time, such as:
- Massive computing capabilities beginning with high-performance and cloud computing after the introduction of the internet
- Demutualization, mergers and acquisitions, which cause large organizations to form under the Financial Services Acts of 1991 and 1999
- Demand for retirement products fueled by stiff competition, causing an arms race, which saw guarantees tied to or replicating equity and interest rate markets to squeeze out performance and differentiate products on price and complexity
- Challenging market dynamics due to low interest rates for many years, inflation, COVID-19 and the war in Ukraine and sanctions on Russia
- Several new regulations, such as Actuarial Guideline (AG) 43, Valuation Manual (VM) 20, VM 21, own risk and solvency assessment (ORSA) and Solvency II
- New accounting regimes, such as Financial Accounting Standard (FAS) 133, FAS 157, long-duration targeted improvements (LDTI) and International Financial Reporting Standard (IFRS) 17
All these issues lead to an extraordinarily complex insurance industry with many dynamics and challenges that feed off each other.
The challenges can be broken into two categories, which are nowhere near mutually exclusive or independent:
- Fast-moving external environment
- Slow-moving internal insurance organization
The external environment challenges are related to technology, the market, regulation, accounting and competition. The internal insurance organization is related to technology and organizational complexity.
Fast-Moving External Environment
To paraphrase Modelling in Life Insurance: A Management Perspective, products with a separate account with guarantees, equity indexed products and interest sensitive products require forecasting economic and demographic data, which cannot be diversified by selling more policies. The guarantees assume perfect markets and hedging opportunities for even the audacious financial contracts. Coupled with the recent fair-value regulations and accounting standards, such as Solvency II, this combination may cause a harmonizing positive feedback loop, which intensifies the market deterioration in time of crisis.
The complex world in which we live is not linear because linear systems do not contain feedback loops. Feedback loops come from complex systems with nonlinear relationships and power laws. Power laws tend to compound and are far more likely to result in extreme events.2 Nonlinear systems have long memories and state changes, which may or may not be reversible. The hearts of our models are Gaussian, Poisson and Exponential distributions and Brownian motion. These models assume that processes are memoryless, independent and linear.3 These distributions keep our models stable, especially over the long time horizons we need to project. But, keep in mind, this is a simplification of reality.
Slow-Moving Internal Insurance Organization
You can think of an organization as multiple complex networked systems commingling together: the human systems and informational systems. The humans enter the data and computation instruction into the information systems that help them organize and complete tasks. Communication and information are the lifeblood of decision-making.
Inevitably, the informational systems only evolve at the rate at which people can communicate and implement their structures and functions across the organization. The speed of implementation is bounded by the rate at which people can comprehend and learn what needs to be implemented. Unfortunately, learning is an embedded, slow and complex process that cannot be circumvented. The transfer rate of information is the throttle, and coordination among people and teams is the friction to how quickly organizations can respond to their environment. This becomes ever more apparent as the size and complexity of the business increases.
The Consequence of Interdependent Systems
Technology and the markets are evolving faster and faster every day, but organizations and people change at a much slower pace. Critical, complex processes that run over sufficiently different time scales can have hidden negative consequences for the interdependent systems. Complex systems constantly adjust themselves via feedback loops, but when interdependent components operate with feedback loops of different temporal scales, the slower adapting system can become unstable.4 This is like Lucy in the chocolate factory!
You only need to look to the collapses in 1998 surrounding long-term capital management (LTCM),5 Great American Life Insurance Company in 2000, and Bear Sterns and AIG in 2008 to understand the implications of oversimplifying reality. It is easy to criticize these companies after the fact because hindsight bias makes their mistakes look obvious. When running a large, complex organization before its downfall, there are many trade-offs and judgments to be made, such as budget, resources, time, risk tolerances and so on. If nothing bad happens, confirmation and desirability biases will reinforce the decision process that everything is fine and the risk management methods are appropriate. When the market moves against a company or industry, it can move quickly and catch them off guard.
How do we speed up business operations to better match the speed of the external environment? It is all about organizations’ ability to adapt.
Given that no one can change the speed of the external environment, we must improve the rate at which organizations can change and make decisions. Hence, we must find ways for the people and information systems to adapt.
Adaptive Human Systems
Conway’s law states that organizations design systems that follow their communication structures. It was established from a practical perspective that there is propensity to start with the preexisting corporate structures for simplicity. This normally leads to lackluster systems that break where the human communication is dysfunctional, which causes them to be brittle and hard to modify.
Brendan Burns, co-founder of the open-source system Kubernetes, argues that if you build better, more adaptive information systems, then you will build better, more scalable adaptive teams. His presentation, Why You Should Care About Microservices from Microsoft’s 2020 .NET Conference: Focus on Microservices, is a terrific explanation for the interplay among people and information systems. To build more adaptable teams, you must mimic the behavior of adaptable systems. His specification boils down to building small teams that have clear boundaries, cohesive sets of responsibilities, loose coupling with other teams, contracts for expectations of deliverables and policies on defining when and how deliverables can change. To better adapt, teams must optimize their ability to collaborate while reducing their required coordination, so velocity in projects can be increased.
Adaptive Information Systems (AIS)
There are two major pieces to adaptive systems: the processes for using the system and the system itself. In this article, I focus on the system itself, but for a deep dive into the processes, read DevOps—The Path To Actuarial Modernization And Consolidation.
The five components of AIS, or what I call “the model,” are:
- Calculator—what we traditionally think of as the model
- Transformer performs the data transformation from the primary source to the calculator and from the calculator’s output to the final results
- Conductor orchestrates all the transformations and model runs from beginning to end
- Controller tracks all the changes that occur to the first three components
- Librarian organizes all the input and output data, so it can be organized, searched, reconciled and compared to other sources of data and results
All these systems must communicate and work together in a harmonious fashion to reduce operational complexity and better adapt to the outside environment. Just like Lucy in the chocolate factory, as the speed increases, it becomes more difficult to keep up. Without the conductor, controller and librarian tracking process, changes and data, material errors likely will creep into the results and impede the speed of information and important decisions.
We are living in an ever more complex world. To understand the risks, we must separate ourselves from the view of economic shocks coming from external, independent events and instead embrace a systems science approach and realize that the entire economy is a massive industrial machine with many interlocking gears that all affect each other through positive and negative feedback loops. When interacting systems move at different temporal speeds, they have a propensity to become unstable, such as the slow-moving insurance organization interacting with the fast-moving economic environment.
Unlike before the Great Recession, accounting and regulations are becoming market consistent. Unless proper risk management is in place, market noise can flow through the books of the insurance organization. Defensive risk management can intensify market deterioration in a time of crisis through feedback loops. Proper risk management, which will stave off instability, requires resilient and adaptive organizations.
Insurance organizations are a commingling of intertwined information systems. It is important to create systems that facilitate an efficient flow of information through the organization. If teams follow patterns of scalable software architecture and disobey Conway’s law, then it is possible to build scalable and adaptable teams to keep pace with the external environment.
For teams to meet maximum efficiency, they need to work with adaptable information systems. These systems have five major components: calculator, transformer, controller, conductor and librarian. The components must work together to provide robust change management, and organizations must learn to cope with the velocity of business and the complexity of the outside environment.
Statements of fact and opinions expressed herein are those of the individual authors and are not necessarily those of the Society of Actuaries or the respective authors’ employers.
- 1. Laurent, Jean-Paul, et al. 2016. Modelling in Life Insurance: A Management Perspective. Springer International Publishing Switzerland. ↩
- 2. Fieguth, Paul. 2021. Introduction to Complex Systems: Society, Ecology, and Nonlinear Dynamics. Springer. ↩
- 3. Ibid. ↩
- 4. Mobus and Kalton. 2015. Principles of Systems Science. Springer. ↩
- 5. Supra note 2. ↩
Copyright © 2022 by the Society of Actuaries, Schaumburg, Illinois.