Impact on Portfolio Optimization?

The National Association of Insurance Commissioners (NAIC) presented a proposal of new risk-based capital (RBC) charges for C1 investment risk in 2015. This proposal also introduced additional granularity of fixed income credit rating reporting, from six to twenty categories. Although the current proposal focuses on life insurers, NAIC has stated that the proposed structure of twenty rating categories would also apply to health, and property and casualty insurers. The numeric values of the respective C1 factors might vary by industry segments.

These proposed capital charges are developed based on the historical default probability and loss recovery experiences of corporate bonds; however, they will apply to other fixed income securities including municipal bonds, structured securities¹ and private placements. In addition to these base C1 factors, there will be portfolio adjustments to reflect company-specific portfolio characteristics to help ensure that the statistical safety level (i.e., confidence level) for the C1 component is met. Our case study focuses on the base C1 factor without applying company-specific portfolio adjustments.

This issue of Perspectives highlights the differences between the current and proposed C1 factors. The portfolio optimization case study then utilizes the U.S. life industry data to illustrate key differences between optimized portfolios under current and proposed C1 factors.

Our analysis includes these key takeaways:

1. Proposed C1 factors reflect the underlying default risk more appropriately than current C1 factors and might affect insurers’ asset allocations.
2. Portfolio optimization needs to evaluate the “risk-adjusted returns” of various asset classes along with their respective C1 charges. Use of marked-to-market metrics (Value-at-Risk or VaR) might yield different optimization outcomes.
3. Portfolio optimization studies indicate that the proposed C1 factors would result in further duration extension to achieve similar income returns, due to distinct C1 factors at more granular credit rating levels.
4. Under the proposed C1, portfolio optimization with duration constraints may favor structured securities as these tend to have high credit qualities and short durations.

NAIC RBC C1 Capital Charges: Current vs. Proposed 

Table 1 shows how the current C1 and proposed C1 charges have expanded from six to twenty rating categories. The current Baa3 C1 charge (0.96%) is 3.25 times the Aaa C1 charge (0.30%), while under the proposed C1 factors that multiple increases to more than seven times (1.45% vs. 0.21%). The proposed C1 factors distinguish the underlying default risk at more granular rating levels.

Chart 1 demonstrates the percentage differences between current versus proposed C1 charges. The single “A” category shows the most increases, while several lower credit rating categories reflect reduced charges. Without additional analysis, these varying levels of relative changes across rating categories might suggest benefits that may be derived from replacing single “A” securities with those of lower credit quality.

Portfolio Optimization Case Study - Initialization
A portfolio optimization framework evaluates return and risk tradeoffs among different asset classes and identifies portfolio configurations that are optimal (or more efficient) in terms of selected return and risk metrics.

Under the NAIC statutory accounting framework, life insurers typically focus on enhancing book yields (income return) while targeting certain capital ratios or liquidity scores. The risk tolerance metrics used vary by company, depending on the enterprise objectives and stakeholders’ expectations. In this issue of Perspectives, our portfolio optimization is configured to maximize the book yield at given levels of volatility while maintaining similar levels of NAIC RBC capital charges. The goal of our optimization is to identify key directional differences between the optimized allocations, based on current versus proposed C1 charges.

For this portfolio optimization review, we use U.S. life industry 2015 year-end reported statutory financials, investment holdings, and generic product and liability assumptions for an Enterprise Based Asset Allocation™ (EBAA™).² The EBAA™ starts with a breakdown of the return on equity (ROE) of a life insurance enterprise:

Table 2 highlights key components and contributions of ROE for the U.S. life industry. The investment and product leverage are based on 2015 year-end reported industry balance sheet financials. Total return of liabilities assumes a representative life and annuity business mix, with appropriate return and volatility assumptions. The return on assets reflects both the income return of fixed income securities and total return of equity-like assets in the investment portfolio outlined in Table 3 (see below).

Table 3 summarizes the asset classes that are included in the EBAA process. Given that the focus of our optimization review is to evaluate the impact of proposed C1 factors on the fixed income portfolio allocation, we exclude cash and short-term holdings, contract loans, real estate and derivatives from the life industry’s invested assets. Moreover, allocations to commercial mortgage loans (12.1%), equity (1.2%), and alternative investments (5.2%) are maintained at current levels throughout the optimization process.

Portfolio Optimization Case Study – Results and Implications
With the initial life industry portfolio established, the following EBAA optimization review takes several sequential steps:

1. Optimize the portfolio to maximize the book yield (income return) while maintaining the initial C1 charges
2. Establish the optimal asset allocations under current and proposed C1 capital charges separately
3. Evaluate the impact of duration constraints on the optimization results
4. Identify key directional differences between the optimized allocations based on current and proposed C1 charges

Chart 2 compares two efficient frontiers, both maximizing the income return while maintaining the initial level of C1 charges. The blue efficient frontier uses current C1 factors, while the green uses the proposed C1 factors. At first glance, the blue efficient frontier “trumps” the green efficient frontier, as points on the blue curve have better risk-adjusted returns than points on the green curve. But, all might not be what it initially appears.

Table 4 provides the key return and risk metrics of the current portfolio (orange dot) and the blue and green dots (portfolios) along the two efficient frontiers in Chart 2. The blue dot represents the portfolio on the efficient frontier that maximizes book yield (income return) at the current C1 level ($6,638). Similarly, the green dot represents the portfolio on the efficient frontier that maximizes book yield (income return) at the proposed C1 level ($7,028). The blue dot portfolio offers a higher book yield (5.33%) compared to the green dot portfolio (5.15%). However, when evaluated under an economic, marked-to-market framework where VaR is used as the risk metric, the blue dot portfolio’s VaR (63%) is significantly higher than the green dot portfolio’s (46.9%). We need to establish a common metric, either return or risk, to achieve meaningful comparisons. Table 5 displays an approach for these comparisons (see below).

The orange and green dot portfolios in Table 5 are the same as those in Table 4, except with additional sector and credit rating distributions. The blue dot portfolio in Table 5 represents a different point along the blue efficient frontier that provides the identical book yield (5.15%) as that of the green dot portfolio. Both the blue and green dot portfolios are from efficient frontiers and therefore are more “optimal” than the orange dot current portfolio (see Chart 3).

Optimized under current C1, blue dot portfolio’s enhanced risk-adjusted return is achieved through credit rotation or arbitrage (swapping AAA and AA with A, as all currently have the same C1 capital charges) and duration extension (from 6.73 to 7.92). The green dot represents optimization under the proposed C1 and exhibits similar directional reconfigurations in terms of credit, sector and duration; however, the degrees of these rotations differ from the blue dot.

Although the blue dot and green dot achieve the same book yield, they have different risk profiles. The green dot has a better average credit quality (A vs. A-), but longer duration (8.48 vs. 7.92); it also has higher economic tail risk (VaR of 46.9% vs. 33.3% from the blue dot). Next, we focus on constraining durations. 

Life insurers traditionally target their asset duration at certain levels based on their liability profile. Here, we impose duration constraints on the earlier developed optimizations and the resulting efficient frontiers are shown in Chart 4. The blue and green dots in Chart 4 correspond to those in Chart 2, but are constrained by the initial duration level (6.73). As expected, the additional duration constraint reduced the maximum achievable book yield under both current and proposed C1: 5.33% to 5.16% under current C1 and 5.15% to 5.05% under proposed C1.

The duration constraint significantly alters the optimal asset allocation. From a credit standpoint, among AAA, AA and A rating categories (current NAIC 1 category), the green dot, relative to the blue dot, favors AAA and AA over A; and BBB allocation is actually reduced. This credit rotation is contrary to the common rationale suggested by Chart 1, which implies that single A’s will be replaced by lower-rated fixed income securities. Thus, the relative risk-adjusted return matters, not just the changes in relative capital charges. From an asset sector perspective, structured securities are favored under proposed C1 as they tend to have high credit qualities and short durations.

Concluding Remarks
The NAIC’s Life RBC proposal presents new C1 factors for fixed income securities and also expands the credit rating reporting from six to twenty categories. The proposed structure of twenty rating categories will apply to health, and property and casualty insurers, although the numeric C1 factors might vary by industry segments.

The proposed C1 factors are likely to incentivize life insurers to reconfigure their investment portfolio. To achieve a similar book yield from the fixed income portfolio will require extending the duration under the proposed C1 optimization. This is because the proposed C1 charges remove the credit arbitrage incentives that exist in the current RBC framework.

When duration is constrained, optimization under the proposed C1 framework will favor higher (AAA and AA) over lower (A or BBB) credit quality. Thus, under the new RBC framework, structured securities, which tend to have high credit quality and short duration, could be the winners.

We welcome your feedback and comments. Please contact us if you would like to know more about the implications that current and proposed RBC C1 charges will have for the life insurance industry and, more specifically, to your business.

End Notes

¹ Structured securities will follow a two-step process. Initially, NAIC will stay with the current modeling process, but map the breakpoint price to twenty factors rather than the current six factors. The second step will be to review the entire process for establishing appropriate capital requirements for structured securities.

² Refer to NEAM’s June 2016 Perspectives – Life Insurer Asset Optimization: A Top-Down Enterprise Approach