How to Improve Stress Testing

A major survey finds that less than 20% more effort would yield 80% more value.

|
In spring 2016, PwC investigated the current state and future direction of stress testing. We surveyed 55 insurers operating in the US about their stress testing framework and the specific stresses that they test. We also engaged in more detailed dialogue with a number of insurers in the US and globally, as well as with some North American insurance regulators. Our principal conclusion is that stress testing, though well established, would benefit significantly from a modest amount of additional effort. Borrowing terminology from the Pareto principle, we think less than 20 percent more effort would yield 80 percent more value. A brief history Thanks to the requirements of the Dodd Frank Act of 2010, we expect that stress testing is the most widely recognized and understood risk management tool. The basic concept is relatively simple and most people in business and government readily accept the notion that if a specified future unfolded –say a repeat of the last economic crisis –it would be good to know ahead of time if banks would remain financially viable. After its initial introduction, stress testing continues to maintain a high level of attention via the ongoing publication of results from the Federal Reserve Board’s Comprehensive Capital Analysis and Review (CCAR) which the media, financial commentators, and the banks themselves eagerly anticipate. It is easy to see how stress testing concepts in the Dodd Frank Act could apply to insurers. And, indeed the insurance industry (more specifically, its actuaries) has widely used stress testing and scenario analysis for decades. More recently, 2013 was especially noteworthy for insurance stress testing, with publications on the subject by the North American CRO Council, the CRO Forum and the International Actuarial Association. From a regulatory perspective, the National Association of Insurance Commissioner’s (NAIC’s) Own Risk and Solvency Assessment (ORSA) calls for a prospective solvency assessment to ascertain that the insurer has the necessary available capital to meet current and projected risk capital requirements under both normal and stressed environments. In Canada, the Office of the Superintendent of Financial Institutions (OSFI) has provided clear direction on stress testing governance and methodology in its 2009 publication on Sound Business and Financial Practices (Guideline E-18). It also is noteworthy that, in Europe, despite all of the attention lavished on Solvency II and internal capital models, the European Insurance and Occupational Pensions Authority (EIOPA) launched a Europe-wide stress test for the insurance sector in May 2016. Equally as important as the regulatory initiatives are the business applications and benefits of stress testing. As we address in more detail below, survey results show that insurers make good use of this risk management tool and are looking to expand its application even further. A little more effort We see three areas where only a little more effort can yield substantial benefit: 1) a clearer definition of stress testing, 2) more thoughtful stress construction, and 3) a more robust stress testing platform. As a start, it will be useful to clarify what we mean by stress testing. As we use the term here, we mean a projection of income statements, balance sheets and –most importantly –projected available and required capital over a multiyear business planning timeframe (including new business over the planning timeframe). Typically the test is done for the entire enterprise and includes a base case and a number of stressed future states. This definition of stress testing is consistent with how both insurance (ORSA Guidance Manual) and banking (FRB CCAR) regulators use the term. It contrasts with risk-specific stress testing. Risk-specific stress testing typically looks at a single risk, often only for the part of the enterprise susceptible to that risk. And, it frequently assesses the impact over a range of stochastically determined scenarios. Distinguishing between stress testing and risk-specific stress testing needs little effort but can help companies avoid considerable confusion as they enhance and apply stress testing capabilities. Only with clear definitions can an insurer evaluate whether or not it has deployed the tool effectively. A vague notion of stress testing taking place somewhere in the organization typically means that there is unawareness of potential gaps in the enterprise risk management (ERM) framework. See also: The Key Role for Stress Tests in ERM Another area where we believe a little more attention would pay major benefits is the development of comprehensive stress scenarios. When describing future states, insurers have many factors to consider in order to articulate the risks that can impact their business. As an indication of the range of these factors, the section of our survey that addressed stresses had 32 questions, many with sub-parts, each covering a different risk. However, rather than starting with an effort to combine all of these risks, stress testing benefits from starting instead with a narrative that articulates a potential future and then addresses how that future would impact the insurer through various risk factors. For example, a stress narrative could be based on a prognosis of an ongoing steady decline in the price of oil and other commodities, then a postulation of the resulting impact on economic growth, interest rates, equity valuation, employment rate, etc. The narrative then could move to an analysis of the impact of these factors on the insurer’s risks, leading to a projection of how the company’s income statement, balance sheet, available and required capital would fare if this future, in fact, unfolded. Lastly, we note that despite the considerable attention and utilization of stress testing as a management tool, it appears that, for many insurers, the infrastructure that produces results is ad hoc and likely inefficient. Our survey indicates that only 10% of respondents have built a bespoke platform for stress testing. 78% of them use spreadsheets alone or spreadsheets combined with actuarial/projection software. In terms of how long it takes to conduct stress tests, 42% of respondents indicate the process takes between one and two months. A further 35% report that it takes more than two months, and sometimes longer than three months. While systems infrastructure updates do not normally result in major improvements from little effort, many insurers, particularly in the life sector, have already embarked on a process of modernization. As they are looking to address their risk, actuarial, and financial reporting needs in a comprehensive manner, we recommend that stress testing capabilities receive high priority. With a modest amount of extra effort, insurers should be able to incorporate significant enhancement to their stress testing platform as part of this modernization. This in turn will yield the benefit of more timely, accurate and insightful stress testing results. A lot more value Insurers already use their stress testing for many purposes. Survey results show that respondents currently utilize their stress testing for an average of almost five different uses. Additionally, respondents indicated they each had plans to add almost four new uses in the future. More than half of the respondents reported using their stress testing work for strategic planning, calibrating their risk tolerances and limits, assisting with dividend, share-repurchase and similar capital planning, and regulatory impact assessments. These are critical business decisions and further highlight the value of stress testing. Furthermore, stress testing usage has had a positive impact at a significant majority of respondents’ companies. 36% reported instances where key decisions have been made very differently compared to the process prior to stress testing. An additional 29% reported that the results of stress testing has a measureable influence on decision making, though no specific decisions were cited.   See also: New Approach to Risk and Infrastructure? More benefits We see a few additional areas where better articulated stress testing processes and procedures could result in significant benefits.
  • Recognize that stress testing is a separate tool in the risk manager’s tool kit–Frequently, publications and discussions on insurance stress testing describe it as something that supplements other risk management tools. We believe that relegating stress testing to supplementary status undervalues its benefits and contribution. It would be more productive to recognize stress testing for what it truly is: a separate tool with different strengths and applicability compared to VAR-based economic capital.
Some risks –for example, liquidity risk –can be addressed only via a stress test. Adding more required capital does not effectively address the problem; liquidity risk needs to be addressed by developing a preplanned course of action, including accessing prearranged liquid funds. Likewise, reputational risk –and in particular the reputational impact of a cyber event –is better addressed via the stress test tool than via the economic capital route (and the potential addition of more required capital). Similarly, for some risks where economic capital looks like a satisfactory tool, it can give misleading information. Often pertinent risks only reveal themselves fully via stress testing. New business is a good example. Economic capital can include one or more years of new business, typically by assuming new business premium, claims, expenses, etc. are a replica of previous years’ values. But this fails to provide a platform to study how external factors could impact the insurer’s fundamental business model, leading to little or no sales of any new business that resembles prior years’ business. Lastly, we note that most other measures, especially traditional economic capital, concern themselves primarily with very extreme, “in the tail” events. Stress testing is useful not only for high impact, low probability events. More likely events warrant attention –in fact, they may warrant more attention because they often represent more tangible and practical problems that management needs to address immediately.
  • Use stress testing to “war game” management action and prepare in advance for risk crises –In our survey, we asked insurers if stress testing incorporates management actions. In other words, as stress events unfold, presumably management would take some form of corrective action in response, and that corrective action would impact future financial results. Almost half said they do not incorporate management actions. We believe this is a significant oversight.
Stress testing provides a ready platform to prepare in advance for risk crises. Insurers can use the tool to test different responses and select the one that yields the most effective resolution. They then can put in place a contingency plan and pre-event corrections appropriate to the event. Here again stress testing can provide a different perspective than economic capital and similar measures. Economic capital works well as a tool to quantify the impact of taking certain types of action in the present. For example, it can help determine the reduction in required capital if a particular reinsurance treaty were implemented. On the other hand, faced with a multifactor, multiyear stress event (perhaps including changes in interest rates, inflation and equity values, with increases in unemployment and deteriorating buying patterns), stress testing would be a more effective tool in judging if and when to reconfigure the asset portfolio, alter products and prices, and the cost and manner of reconfiguring staffing models. It is worth noting that, in our discussions with regulators about the merits of including the impact of management actions, their expectations are that, yes, insurers should include them. They recognize the benefit that stress testing can provide as an opportunity for planning ahead. However, they indicated that it would be appropriate to show the stressed result both before and after the application of management action. Showing both results can help promote thoughtfully developed post-management action results, not just a broad assumption that management will take appropriate actions.
  • Take advantage of the board’s and senior management’s broad business insights to construct more insightful stress narratives –Our survey shows that most boards receive the results of the stress test either directly or via the risk committee of the board. However, only 11% report asking either the board or board risk committee to approve the stresses the company uses. We believe this represents a missed opportunity to gain board members’ insights and benefit from their engagement in the stress testing process. Not all directors will necessarily have detailed knowledge of the range of potential outcomes of all of the risks that can impact an insurer or the potential stochastic distributions of those risks, but directors typically are experienced and knowledgeable, often with a high level of business and economic acumen. Utilizing their individual and collective skills to contribute ideas on the types of stresses that merit study seems like a good fit for their role and an effective complement to managements’ efforts.
  • Stress testing represents a potential avenue for global capital consistency –As a final potential benefit, we note again that stress testing seems to have a role in all major insurance and other financial services regulatory regimes. At the same time, the global insurance industry is challenged by the task of agreeing to a capital adequacy ratio, presumably based on an economic capital VAR-like foundation. A simpler capital formulation coupled with a robust stress testing regime may hold more promise for a globally agreeable approach.
See also: Key Regulatory Issues in 2016 (Part 2) A bright future Based on survey results and various discussions we had with insurers and other stakeholders, stress testing is universally accepted as a useful tool. We suspect that this is a consequence of its being directly related to the common business practice of preparing a financial plan. Including a few more future states or stresses and incorporating a measure of required and available capital in the financial plan are not major steps. Accordingly, the transition from planning to stress testing should be easy to accommodate. We note how sharply this contrasts with the introduction of economic capital, especially in the US insurance industry. Though its usage is growing, economic capital is not a uniformly accepted regulatory and business tool even after two decades. On the other hand, stress testing is already actively and universally used as a management and regulatory tool. With a little more effort, we believe it can yield very substantial benefits for all.

Henry Essert

Profile picture for user HenryEssert

Henry Essert

Henry Essert serves as managing director at PWC in New York. He spent the bulk of his career working for Marsh & McLennan. He served as the managing director from 1988-2000 and as president and CEO, MMC Enterprise Risk Consulting, from 2000-2003. Essert also has experience working with Ernst & Young, as well as MetLife.

MORE FROM THIS AUTHOR

Read More