Editor’s Note: This article is derived from a presentation Rowen Bell gave at the SOA 2017 Health Meeting as part of a panel discussion called “Risks, Losses and Insolvencies – Oh My!!”
In recent years, the pace of change in the U.S. health insurance industry has been frantic, making it difficult for health insurers and the actuaries who support them to focus on anything other than the novel problems and situations they are confronting currently. In this kind of environment, it can be difficult to make the time to step back and reflect on the longer-term trends affecting the health insurance industry.
This article is an effort to do just that, with respect to examining how the nature of the risks that U.S. health insurers face has evolved over recent decades. This walk-through history points me to a particular conclusion that is not widely understood by regulators or industry participants, and for that it merits further consideration. Namely: The most widely accepted tool within the health insurance industry for quantifying risk—the National Association of Insurance Commissioners’ (NAIC’s) Health Risk-Based Capital (HRBC) formula—is, in many respects, an artifact of an era that no longer exists, quantifying a type of risk that isn’t present in many health insurance markets today. As such, HRBC may be overdue for a significant modification to reflect the nature of the risks facing health insurers today.
I entered the health insurance industry in the summer of 1996, when President Clinton’s efforts to more radically restructure the health insurance market had recently failed. At that time, a health insurer’s business was dominated by the employer group market to a far greater extent than is true today. Government markets as we now know them either did not exist or were in nascent form. The over-65 individual market was predominantly Medicare Supplement, and enrollment in 1999 in what today is called Medicare Advantage was less than one-third of what it is now. The under-65 individual market was a medically underwritten market with guaranteed renewability but no guaranteed issue.
While the product mix was narrower than it is today, the health insurance market of the late 1990s was much more diverse in other ways. There were more players in the market, as there had not yet been significant consolidation among Blue plans and health maintenance organizations (HMOs). By modern standards, publicly traded insurers held relatively low market share. Also, there was greater variety with respect to the degree of managed care and network models. An important corollary here is that the health insurance market had relatively low barriers to entry, with internal provider network development not yet as critical to success as it is today.
The financial regulatory environment in which health insurers operated in the 1990s was also somewhat immature and inconsistent. Statutory accounting had not yet been codified into a set of well-publicized standards: the modern Statements of Statutory Accounting Principles (SSAPs). State regulators would sometimes permit health insurers to apply accounting practices of uncertain merit—for example, allowing an HMO to count potentially uncollectible advances it had made to financially challenged hospitals as part of statutory surplus. Risk-based capital standards for HMOs and health insurers were under development by decade’s end, but they had not yet been implemented.
In this business environment, what were the main risks that health insurers faced? Two decades ago, we talked a lot about two risk concepts—the “underwriting cycle” and “cumulative anti-selection”—but they feel somewhat quaint from a modern perspective.
The underwriting cycle concept had its genesis from the observation that, for decades and decades, the aggregate underwriting gain/loss of the country’s Blue plans had exhibited a cyclical pattern: Periods of gains were followed inevitably by periods of losses, with each full cycle lasting about six years. The existence of the underwriting cycle had an attractive economic explanation. In a market with relatively low barriers to entry, gains exhibited by existing market participants attracted new entrants to the market. In the absence of strong financial discipline among market participants, competition naturally led to underpricing and underwriting losses; this led to market exits that allowed the remaining participants to strengthen pricing and restore profitability, resetting the cycle.
Cumulative anti-selection can arise in situations where health insurers are compelled to renew their customers’ contracts but cannot fully reprice at renewal to reflect each customer’s risk. In such a situation, customers continually select against the insurer when deciding whether to renew the coverage—customers with better health status either procure alternate coverage or self-insure. This dynamic can create a self-perpetuating spiral, whereby the insurer’s efforts to restore a block to profitability by increasing premiums only exacerbates the anti-selective lapsation. To the extent the insurer cannot readily exit an underperforming block of business, the insurer is exposed to the risk of unavoidable future losses from an anti-selection spiral.
Both risks can be mitigated if the insurer has strong actuarial and health care economics functions, allowing it to readily understand in real time how its business actually has been performing and how upcoming changes in provider contracts and risk arrangements are likely to impact financial performance. However, back in the 1990s, many of the participants in the health insurance market had underinvested in not only the actuarial and health care economics functions, per se, but also in the underlying data infrastructure that those functions would need.
On one level, the first decade of the 21st century was a time when the forces impacting the health insurance market were relatively minor. The regulatory framework of the commercial group and under-65 individual markets was more or less left alone. Change came to the Medicare market, however, in fits and starts. Medicare Advantage enrollment doubled over the second half of the decade, with the reforms enacted in the Medicare Modernization Act (MMA) making the program more attractive to insurers. The MMA also created a new over-65 health insurance market for stand-alone Medicare prescription drug plans. The increasing importance of government contracting to health insurers was underscored by the significant expansion that the Medicaid managed care market experienced during the 2000s. Although, back then there was far less overlap than is true today among the insurers participating in the Medicaid market and those participating in commercial markets.
However, on another level, there were developments that took place during this decade that had a profound impact on the overall business environment in which health insurers operate—specifically, developments in financial regulation. Early in the decade, the NAIC completed its multiyear codification of statutory accounting project, creating a uniform nationwide baseline for how insurers report their financial condition to regulators. The new technical accounting guidance dealt with the admissibility of provider advances and other health care-specific assets, and it reinforced guidance regarding the need for insurers to hold premium deficiency reserves (PDRs) with respect to blocks of business where future losses were expected. Codification also contained a requirement that insurers disclose the impact on surplus and income of any permitted or prescribed accounting practices its regulator issued. My sense is that this new transparency created an environment in which regulators became less willing to permit accounting practices that deviated from the NAIC standards.
A second key development in financial regulation in the early 2000s was the finalization of risk-based capital standards for HMOs and other entities filing the NAIC’s then-new Health Annual Statement (orange blank), coupled with a regulatory tweak that migrated many health-centric insurers from the Life, Accident and Health (blue) and Property and Casualty (yellow) blanks over to the orange blank. With these developments, the vast majority of the nation’s health insurers were computing and reporting a common capital metric: NAIC HRBC.
The combination of more rigorous statutory accounting standards and a novel but widely accepted capital metric represented an overdue injection of financial discipline into the health insurance industry. This manifested itself in two important ways during the 2000s.
First, the greater focus on financial soundness created pressures that led to increased consolidation within the health insurance industry, as companies with weak HRBC positions but viable businesses were absorbed by players with more capital. The consolidation wave led to an increase in the overall market share held by publicly traded health insurers, which had a two-fold competitive advantage: They could issue stock as currency to acquire competitors, and they had a higher degree of financial leverage than a competitor whose top-level parent was subject to statutory accounting and HRBC. One side effect of consolidation was that, as it occurred, the average level of actuarial sophistication of the remaining players increased.
Second, the underwriting cycle went away. Instead, underwriting gains remained strong throughout the decade. My pet theory is that this was primarily due to increased financial discipline: In a world where insurers, regulators and other parties are paying keen attention to HRBC, it became far less attractive for an insurer to prioritize increasing market share over profitability. Other factors that contributied to the dissipation of the underwriting cycle included diminished competition (fewer players, plus greater actuarial sophistication of the remaining players) and the increasing importance over time of provider networks (increasing the barriers to entry). Given this context, the 2000s was a decade in which the health insurance industry as a whole thrived. The financial crisis of 2007–2008 was arguably the greatest source of strain, but it was not fatal to any health insurers. With that storm weathered, one might have been forgiven for assuming that the health insurance industry was settling into an equilibrium of stability.
And then, the Affordable Care Act (ACA) happened.
The predominant theme of the 2010s was massive regulatory transformation of the health insurance market, coupled with vast amounts of regulatory uncertainty—uncertainty that continues to play out as I write these words in mid-2020, with the Supreme Court soon to hear arguments in a case suggesting that the passage of the Tax Cut and Jobs Act, in reducing the tax for noncompliance with the individual mandate to $0, has made the entire ACA unconstitutional.
Clearly, the under-65 commercial market has been transformed the most: What was formerly a market with medical underwriting and guaranteed renewability has been turned upside down. Guaranteed issue replaced the insurer’s former ability to select its customers, and an annual decision cycle involving both insurers and regulators replaced guaranteed renewability by the customer. However, one shouldn’t overlook the ways in which the ACA altered the risk profile of the commercial group market, eliminating lifetime benefit caps and imposing premium rebate requirements enforced via minimum loss ratio standards. And, of course, the significant expansion of Medicaid eligibility led to new opportunities for health insurers to participate in Medicaid managed care markets.
Health insurers faced a plethora of novel risks during the 2010s. Companies participating in the ACA exchange markets were confronted with problems unlike those historically faced by health insurers. These issues included:
- A lack of data about the morbidity of the population to be insured
- Extreme uncertainty about how much enrollment each insurer would attract
- Uncertainty about how the new risk adjustment program, in which premium dollars were shifted among competitors in a zero-sum fashion after the end of the year based on a risk model, would ultimately impact loss ratios
- A significantly lengthened pricing cycle, in which premiums for the third year of the program needed to be set before the full financial results of the first year (post-risk adjustment) were known
Those were the known unknowns, to use Rumsfeldian language.1 And they were daunting enough. But what really caught the industry off-guard were the unknown unknowns, the risks that materialized but could not realistically have been anticipated. Put differently, there were many significant changes to the rules of the game long after the game had started. Two crucial changes were:
- Regulatory decisions to allow certain customers to retain so-called “grandmothered” policies, after insurers had already set their pricing under the assumption that those customers would enter the exchange’s single risk pool
- Congressional action that forced the ACA’s risk corridor program to be administered on a budget-neutral basis, contrary to its original intent, with the Supreme Court’s 2020 ruling in Maine Community v. U.S. coming far too late to undo the damage
Health RBC, Then and Now
As we’ve seen, the health insurance market and the risks its participants face have evolved dramatically over the past quarter century. On the other hand, we are now roughly two decades into the HRBC era, and yet today’s HRBC formula has not changed in a meaningful way from the original formula.
Let’s review some of the main assumptions underlying NAIC HRBC:
- Risk exposures are divided into asset (H-1), underwriting (H-2), credit (H-3) and business (H-4) risks. The RBC formula applies a covariance adjustment (square root of the sum of the squares), which equates to an assertion that each of these risk categories is considered to be completely uncorrelated with each other, and that underwriting risk is perfectly correlated across different lines of health insurance business. For a typical health insurer, the H-2 component dominates the calculation of required capital.
- The underwriting risk formula treats all types of “comprehensive medical” insurance—large group, small group, grandfathered individual, ACA exchange individual, Medicare Advantage, Medicaid—the same. That is, each of those lines of business is treated as generating the same amount of underwriting risk per dollar of incurred claims.
- The actuarial modeling that underlies the HRBC underwriting risk factor was performed by an American Academy of Actuaries (the Academy) work group in the 1990s, and it was based on a ruin theory model with a five-year time horizon. The model was trying to solve for the amount of surplus that an insurer would need to hold to insulate itself, with 99 percent confidence, from the amount of losses that its existing portfolio of contracts could generate as they were renewed and repriced over the subsequent five years. Note the similarity between this view of risk and the concept of cumulative anti-selection previously discussed. The risk that the model was designed to protect against was the risk that, over time, the insurer could find itself in a pricing spiral from which it could not recover—due to some combination of inattention, bad luck and anti-selective lapsation.
This overall approach to measuring a health insurer’s capital requirements made a lot of sense given the business environment of the 1990s. However, viewing it through a more modern lens reveals some key theoretical flaws.
First and foremost, the assumption that different lines of medical coverage have the same underlying level of underwriting risk feels seriously outdated. So, too, is the implicit assumption HRBC makes that there is no diversification benefit available from being active in multiple distinct lines of business. In today’s market, the drivers of downside risk among, say, the Medicaid, ACA exchanges and large group lines of business are so distinct that it seems unreasonable to assume that all lines would necessarily turn south at the same time. Even a “black swan” event like the COVID-19 crisis, which is still very new as I am writing these words, is affecting separate lines of health insurance business differently, not only in magnitude but even in direction (e.g., it may be a boon to dental insurers).
An important related observation is that the Academy HRBC model’s approach to measuring ruin over a five-year time horizon makes far less sense in a world where insurers explicitly make decisions on whether to continue participating in major markets, such as Medicare Advantage and ACA exchanges, on an annual basis. Medicaid is similar, although the contract renewal period may be longer than a single year.
In today’s market, insurers aren’t really exposed to one of the risks that preoccupied health actuaries in the 1990s: namely, the risk that the insurer cannot exit a poorly performing block that is spiraling out of control. The flip side of that coin is that today’s insurers face a risk not contemplated in the HRBC model: the risk of losing a massive amount of money in a single year, due to a combination of unintended underpricing and being unable to stem the flow of enrollment into the underpriced product.
The health insurance industry has changed dramatically in the past quarter century. Unfortunately, the generally accepted tool for quantifying a health insurer’s capital requirements—NAIC HRBC—has not. Plus ça change, plus c’est la même chose?2
I am not arguing that health insurers’ capital requirements are too high, nor am I arguing that they are too low. I don’t have a point of view on that question. What I am arguing, however, is that the HRBC metric was derived from an actuarial model of a type of health insurance market that, for the most part, no longer exists and has been replaced by new markets with different risk exposures.
The political uncertainty around health insurance regulation is still so great that one can’t assume we’ve reached anything resembling the steady state. COVID-19 has piled even more uncertainty on top of that. As such, the time is perhaps not yet ripe to undertake a wholesale revision of capital requirements for health insurers. Nevertheless, it is time to start having that conversation.
The views expressed in this article are those of the author and do not necessarily reflect the views of Ernst & Young LLP.
- 1. In a 2002 press conference, Donald Rumsfeld, then the U.S. Secretary of Defense, made remarks that have since become famous: “There are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns—the ones we don’t know we don’t know.” ↩
- 2. Attributed to the French critic Jean-Baptiste Alphonse Karr, this phrase translates as, “The more things change, the more they are the same.” ↩
Copyright © 2020 by the Society of Actuaries, Schaumburg, Illinois.