The Growing Importance & Evolution Of Measuring Investment Skill For Insurers – Commentary

The asset management business has been around for over 100 years, and global equity markets and bond markets are both approaching $100 trillion in value1 Insurance companies manage almost $30 trillion of assets globally.2 They look continually to hire individuals with investment skill to join their internal teams, and for companies possessing this quality to become part of an outsourcing relationship. Numerous consultants help asset owners find highly-credentialed individuals and advisors. However, despite all this focus, measuring investment skill is elusive, especially for highly constrained mandates such as insurance company portfolios.

This article highlights aspects of understanding and finding investment skill, both within the investment industry at large, and with the additional challenges introduced with a multi-faceted business intermediary such as an insurance company.

What is Investment Skill?
It is essential to start with a clear definition that avoids misunderstandings on what investment skill is and what it is not.

Investment skill is different from, but dependent on, investment expertise (e.g., a high level of knowledge3). Skill is the ability to use one’s knowledge effectively and readily in execution or performance3. In portfolio management, we quantify skill through excess return. The investment business relies on professionals having a high level of domain knowledge in several areas. All play an important role. A subset of these professionals are in analyst and portfolio management positions that depend on skill as the core competency, and even fewer demonstrate this rare ingredient consistently. A more subtle but equally important test of investment skill is associated with model design. Technical knowledge is needed to build quantitative approaches, or to “model” an investment framework. But design decisions that properly recognize where, and where not, discretion applies are critical to investment success. As noted below, artificial intelligence is fundamentally altering the landscape of where this discretion resides and how it is deployed.

A Difficult Question….How Should Skill be Measured In the Investment Advisory Business?
In simple terms, skill should be measured by evaluating decisions on risk and return within precisely defined boundaries referred to as the opportunity set. Even as the investment industry has matured, it has a long and understandably difficult history in measuring investment performance. Challenges include selection of the right time horizon, lack of “peer group” comparability, benchmark composition, capturing the impact of portfolio scale, various definitions for risk and return; use of leverage, security valuation, fee and expense structure, and changes in staff.

Why Isn’t Investment Skill Measurement a Prominent Business Standard in the Insurance Industry?
It’s hard enough measuring investment skill in predominantly unconstrained (pre-tax total return versus a benchmark) portfolios. Compounding the question’s difficulty with the myriad constraints needed to safely run an insurer makes finding a fair answer even more difficult. Studies based on publicly reported data are very helpful indicators of the absolute and relative financial contribution of the investment function to the firm’s economic growth. However, because these results are often driven by idiosyncratic business needs, many of which are not disclosed, drawing reliable conclusions on investment skill in general, and peer group comparability in particular, is questionable.

More specifically, three reasons explain why skill-measures are not as prevalent in the insurance industry:
• Business Requirements: Skill measurement becomes clouded when business needs (operating and balance sheet) of a complex, diverse, and highly regulated enterprise shape portfolio decisions. These needs include, but are not limited to, liquidity, asset-liability management, capital efficiency, multiple financial statement regimes, various regulations, rating agencies, and conflicting stakeholder interests.
• Business Model: Some companies prefer to de-emphasize investment risk/reward for good reasons, including (1) their profile as an “underwriting company”; (2) the cost of an industrial-strength investment function; or (3) use of an outsourcing model for investments.
• Business Reporting: Standard financial reporting does not provide enough information/data for calculating investment skill in ways similar to the investment advisory business. In addition, this reporting, and taxes, create active management criteria having nothing to do with investment talent, such as gain or loss recognition.

Of course, these are generalizations in an industry that defies the validity of making them, and this topic is no exception. For example, some companies, including entities sponsored by investment firms, promote their investment skill as a competitive advantage, and measure it accordingly.

Why is it Important to Measure Investment Skill, Difficulties Notwithstanding?
Although not always measured — or measured as rigorously as underwriting skill — a well designed skill-measurement system is vital to an insurance company for all the same reasons it is a baseline for underwriting:
• Financial Results: Even modest changes in asset exposures have a substantial call on overall company performance. Measurement of strategic and tactical asset allocation decisions, including no changes, confirms this importance.
• Talent: Insurance companies are among the best environments to train and have a career in investments. As customization takes hold in wealth and institutional management more broadly, investment experience in the insurance industry is seen as especially valuable. A culture that recognizes and rewards investment skill has enabled many firms to retain exceptionally talented professionals.
• Pay-for-performance culture: Although investment expertise and insurance domain knowledge are essential, skill-based measures add to how performance is evaluated and rewarded for appropriate roles.
• Fiduciary responsibility: Investment skill for intermediaries in particular starts with superior management of asset risk.

A Nearly Unanswerable Question….How Should Investment Skill be Measured In the Insurance Industry?
Given the importance of investment performance to enterprise results, it is critically important for insurers to isolate investment skill, the difficulty of doing so notwithstanding. This requires that the investment team (1) clearly understands what is essential to the organization, (2) communicates the realities of the capital markets to their business colleagues, and (3) recognizes how these two dimensions intersect in the context of a program that measures investment skill. Management needs to address cross-currents from the start. Answering the following questions will go far in reconciling what are often considered mutually exclusive perspectives of corporate financial goals/constraints, and skill-based investment performance.

Are investment objectives and corporate goals fully aligned?
Typically, this is the case, as it should be. However, this condition doesn’t guarantee that investment skill is being tracked. When these measures are added, it is important to safeguard their alignment with the company’s investment objectives and the overall company goals.

Is the company’s investment objective focused on investment income or overall investment return?
As described below, these do not have to be contradictory measures, and companies can include both without creating ambiguity.

What is the time horizon?
Insurance companies with strong capital positions can look through moments of investment volatility. But it is important to have a specific, multiple-year look-back period to evaluate regularly. There should also be a shorter time period to recognize that some opportunities have a shorter half-life. In this way, the investment function can, and would be expected to, add value as both short- and long-term opportunities arise. We believe a combination of 1- and 3-year timeframes, or something similar, is a good starting point for skill-based evaluation. Again, this tracks the approach applied to the underwriting and pricing performance culture so embedded in the insurance industry.

Is the investment objective an absolute or relative target?
If the former, are the returns measured against a budget, the prior year or external expectations? If the latter, is it against a publicly available benchmark, or a peer group? Almost all skill-based investment measurements should be thought of as relative, providing comparisons to either (1) other options within a specific opportunity set, or (2) a carefully selected peer group. Peer group comparisons are notoriously questionable due to the lack of comparability and absence of total-return measures.

With the benefit of answers to these questions, management can develop a measurement system to capture skill and complement corporate financial measures.

Suffice it to say, it is easier to measure investment skill in unconstrained portfolios. Most insurers have some level of constraints. So skill measures need to isolate and evaluate manager discretion. Below are steps that help form the criteria. A common ingredient for all is that they can co-exist with typical financial goals such as capital efficiency, ROE, growth in book value, spread margin, and others.

1. Credit Quality Anticipation
The insurance industry has a long and impressive track record conducting fixed-income credit analysis. This investment foundation has been motivated through a hold-to-maturity assumption with origins in accounting and regulatory treatment, and in risk management tied to pricing for investment-oriented lines in particular.

Over the years and continuing today, the combination of internal decisions, changing requirements for statement valuation, and market conditions have imposed financial recognition of credit risk before final redemption. For these reasons, internal systems now work to anticipate, rather than confirm, the earliest signals of potential changes by the NRSOs that drive market valuation.

To this end, domestic insurers review their portfolios every quarter, or more frequently, for possible impairment. As part of this process, they compare price movements (duration adjusted spread movements) of their sector/ portfolio with price movements of the comparable overall credit sector of the market with the hope that their credit analysis has projected these movements in advance.

2. Clarity of Investment Objective
It is critically important to have a clear investment objective, and entirely reasonable to have more than one, as long as priorities are established and conflicts avoided. For example, some companies have investment objectives that are income-based and silent about the trade-off between income and total-return (a more skill-based measure). In these instances, we recommend adding a total-return constraint, such as maximizing income subject to total-return not falling below 80% of the performance in a benchmark or Model Portfolio (MP).

For individual asset classes, companies can also carve out a small portion to be managed for total-return exclusively, subject to common-sense restrictions on active weights and limits on holdings outside the benchmark. An assumption of the benchmark returns would be applied to pre-designated windows when gain/loss recognition would take place for financial statement or tax purposes. In this way, alignment with the overall income objective is preserved, but a skill-based measure has been added.

3. Strategic and Tactical Asset Allocation
Data confirms that turnover in insurance portfolios has been rising. This can be due to changes in long-term strategies, such as the addition of new asset classes or a revised outlook for the existing set of asset classes, or due to temporary (e.g. tactical) departures from strategic weights to exploit short-term market dislocations. The distinction between strategy and tactics should persist. However, the pace of change in how both are designed and implemented by insurers has been accelerating.

The methods for evaluating strategic and tactical performance exceed the scope of this paper, but it is safe to say that doing so is the most important aspect of an investment review. To confirm this we, along with Larry Pohlman, recently completed a study that evaluated sequential 5-year periods, starting in 1990, using three allocation strategies which produced the results in the graph below.

All three strategies started with portfolios of 70% fixed-income (5 benchmarks), and 30% equities (3 benchmarks), were rebalanced monthly, and were subject to min/max holding limits to retain diversification. The MP maintained the initial weightings. The “Best” used perfect foresight, funding the next month’s top performers from the poorest using a waterfall. The “Worst” portfolio was a perfect disaster, funding the next month’s weakest performers from the best. As shown in the graph, the range in potential outcomes for asset allocation is substantial.

4. Defining and Recognizing Passive Decisions
An important test of investment skill is establishing where, and when, to pursue active management. Management typically makes these one-time decisions in the design phase surrounding individual asset classes. Less explicit, but no less critical, is an active manager’s discretion to lock-down to a benchmark (or MP) when conditions provide few active opportunities. This option can be a specific aspect of mandate design, including time limits and fee-implications, for active-management sleeves. An Investment Committee can then review both forms of active/passive decisions (passive exposure selection and the lock-down option for active sleeves) as part of a comprehensive skill-based investment system.

5. New Markets/Decisions/Measures/Roles
The markets in which insurers now invest are changing at an unprecedented pace. Skills and associated measurements – especially attribution software — need to adapt accordingly. For example, there has been a significant transition to private markets by insurers. Companies should measure decisions regarding the valuation of the illiquidity spread throughout the capital structure, just as they evaluate the skill surrounding valuation of credit spreads.

To best leverage investment skill, the architecture of transaction authority should not confuse responsible governance with management discretion. For example, does it make sense for an Investment Committee meeting quarterly to approve changing allocations between small-cap value and growth, or for these decisions to be made, and measured, by a small-cap core manager looking at this question daily for a living?

The importance of asking the right questions about roles and authority in insurance asset management has never been higher, and insurance companies are asking these questions.

6. “Asset Class” Scope – The New Security Selection
The NYSE trades around 2,800 stocks. There are now approximately 2,200 Exchange Traded Funds (ETF’s), and an area of significant relevance to insurers – fixed income ETF’s — is poised for growth due to various positive trends including capital treatment. Along with the proliferation of index funds, published benchmarks, smart beta, and because of the relative sensitivity of overall results to top-down decisions, beta selection should be a regular aspect of performance evaluation by all investors.

7. Track Record Relevance
It has been said the only part of the investment profession more difficult than outperforming the market is selecting active managers to do so. This is especially true in the insurance industry because track records, essential to due diligence, for promising managers are often not achieved with the constraints that will be apply to a particular company. However, if business requirements can be addressed elsewhere in portfolio construction, as described in the next section, eliminating these highly-skilled managers from consideration may not be necessary. Even for advisors having a large insurance practice, it is essential to understand their ability to manage an unconstrained portfolio, attribute their excess returns, and compare this process to how the insurer expects their portfolio to be managed.

8. Deployment of Risk Controls
Whenever possible, risk factor limits, through chosen benchmarks and otherwise, should be managed separately from security selection. In this way, asset class portfolio managers can deliver the value for which they have been trained and hired (internally or externally), subject to common sense concentration and out-of-benchmark limits. The latter is especially important to accurate conclusions regarding skill. This limit avoids false conclusions regarding security selection skill (and unintended factor exposure) that in reality may represent risk premium having little relevance to the benchmark.

With this separation, the CIO exercises control (within TAA authority limits) over macro risk and other factors through benchmark selection, asset allocation, completion portfolios, and overlays, but not security selection in core sleeves. For instance, if the company’s neutral point on asset allocation is to have 50% in credit, but given spread compression, they’d like to decrease their exposure, they can move the benchmark weight to 45% or 40% and still allow the manager to be slightly overweight or underweight.

Insurance companies have realized many benefits from this approach: maximizing investment skill, clean performance evaluation, the relevance of third-party GIPS compliant composites, more active and effective risk management. Meaningful skill measurement, both top-down and bottom-up, is enhanced.

9. Return Attribution
Customized return attribution remains the best method of skill measurement. If an equity manager claims to add value by sector allocation, but the attribution shows excess returns were generated by stock selection, you might be skeptical of his or her ability to repeat this performance. Similarly, if a fixed-income manager claims to add value by credit selection, and attribution shows excess returns were generated by duration calls, you’d again worry.

Managers can generate returns that exceed their benchmark for reasons unrelated to their differentiated process. This raises questions about sustainability. Attribution that directly maps to the steps in their own investment process determines whether this performance came from luck or skill. When customized in this way, attribution also increases the likelihood that solid performance will be sustained, because the manager knows which components are working, and which need change.

Even an insurer with the most restrictive investment constraints and an income objective should consider calculating total-return on their portfolio(s), comparing it to the most appropriate benchmark, and attributing the sources of excess return (positive or negative) versus the benchmark. There is value, to both the investment and risk functions, to a vivid attribution picture irrespective of the investment objective. If excess return comes from overall duration positioning, you may be offsides in ALM and want to get back to a matched book. If excess return comes entirely from excess spread, this may point out that the company has excess credit risk – fine if that had been the intention and is compliant with policy limits.

10. No Pass for Machines
Data science is having a major impact on the insurance and investment businesses, including the private markets in which both have recently focused. Machine learning is especially relevant to the topic of this paper. Breakthroughs with neural networks enable algorithms to identify human bias and offer additional levels of investment skill analysis. Reviewing decisions over market moves and economic conditions are very revealing. They can highlight serial biases the portfolio managers were not aware they had. But research has also confirmed that machines are subject to biases as well. No doubt systems can acquire knowledge and detect subtle patterns, but can they be trained to have skill when faced with conditions having no precedent (e.g. out of sample)? Model validation in this sense is new terrain, but is a necessary form of quality control for investment groups choosing to adopt these exciting new capabilities.

How Does an Insurer Build the Right Skill Measurement System: Metrics Help, But are Not a Substitute for Expert Discussion and Effective Communication
These are easy recommendations to make and difficult to implement, especially at once. They each require a willingness to invest in both the financial commitment to design these metrics as well as the fidelity to acknowledging their results. Having said that, identifying and assuring genuine investment skill is highly accretive to investment and business results. It reduces the frequency of negative surprise and exposure to unidentified risks.

Finally, a concise but complete description of a skill-based investment performance system and its reconciliation with corporate financial objectives is critical to success and reflects very positively on an insurance company’s attention to the highest standards in all functional areas.