A Valuation Perspective
Valuation has been evolving right alongside new technology. December 2019/January 2020Valuation has always lagged behind product innovation. As insurers come up with new and innovative product designs, features and underwriting methods, the regulations and accounting methodologies are always trying to catch up—it is only natural that the regulatory bodies are not able to predict every new idea. But valuation has been evolving. It’s a slow evolution, but the future seems clear. Increasingly, valuing insurance liabilities is becoming a process for developing a current market risk-adjusted view of the business. The recent changes for Generally Accepted Accounting Principles (GAAP) Long-Duration Targeted Improvements, International Financial Reporting Standards (IFRS) 17 and principles-based approaches to U.S. statutory valuation are examples of this.
More recent regulatory changes have focused on the stochastic analysis of market risks, using various methods to assess capital markets and interest rate risks. It is likely that a stochastic view of risk will continue to be the future direction of this evolution. As tools and technology advance, stochastic analysis of other risks will also become more commonplace and routine. For example, risks around mortality, lapses, partial withdrawals and other policyholder behavior will be more easily calculated using similar tools and methods. This will go beyond the simple sensitivity testing often performed today to full nested stochastic models that test each of the key risks, both independently and interdependently.
Additionally, companies are starting to use predictive analytics, machine learning and artificial intelligence (AI) to come up with better models for setting assumptions, underwriting business and analyzing risk. The next evolution of this process will be to create machine learning or AI processes that replicate or mimic these types of analytics on-the-fly to model the dynamic and reactive nature of the changing course of business within stochastic projections.
The use of machine learning and AI will need to be well understood, including the risks that:
- These new methods may not perform as expected.
- Users of these methods may not understand the results and inner workings as thoroughly as they do with more traditional methods.
This will be especially challenging for regulators that will need to develop a framework for understanding the myriad new methodologies being developed across the multiple companies they oversee.
These technological advances will only accelerate in the future, and it will be all but impossible for highly specific regulations to keep up with the ever-changing ways in which insurance is being written and distributed. To cope with this, a principles-based approach will almost certainly be necessary to adapt quickly enough to the changing landscape. This means that actuarial judgment and expertise will become even more important.
The job of the actuary will evolve from explaining the appropriateness of current processes, procedures and methods to also educating regulators and users of valuation results on the new methods and technologies. It will be necessary to help them understand how these technologies are utilized, managed and validated.
Copyright © 2019 by the Society of Actuaries, Chicago, Illinois.