Authorities insurance coverage for enterprise interruption losses from pandemics: An analysis of its feasibility and attainable frameworks – Klein – – Danger Administration and Insurance coverage Overview

1 INTRODUCTION

The COVID‐19 pandemic has caused business closures and restrictions, supply chain disruptions, decreased demand for some firms’ goods and services, and financial catastrophe for many terminated workers and small businesses, resulting in widespread economic losses. Firms have been generally unsuccessful to date in recovering their COVID‐19 losses through their property business income interruption (BI) insurance policies.1 This is primarily because losses from COVID‐19 have not resulted from actual physical damage as required by these policies, as most court decisions have concluded as of this writing. Standard insurance economic theory suggests, and most insurers contend, that it is infeasible for them to provide BI coverage for pandemics.2 Consequently, there is strong interest in creating a government insurance program (that could include participation by private insurers) that would cover businesses’ expenses including wages, and possibly replace lost net income arising from future pandemics.

One proposal for such a program is the Pandemic Risk Insurance Act of 2020 (PRIA), introduced in the Congress as H.R. 7011, to create a public–private insurance program, labeled the Pandemic Risk Reinsurance Program (PRRP), (similar to the program established by Terrorism Risk Insurance Act (TRIA), that would provide BI insurance for pandemics with participating private insurers retaining 5% of covered losses above a deductible.3 We review the structural features of the program contemplated by PRIA and consider whether a public–private pandemic risk insurance program might borrow from the structures of other existing catastrophe insurance programs such as the National Flood Insurance Program (NFIP) or the federal crop insurance program. Any proposal for BI pandemic insurance, with or without the participation of private insurers, raises fundamental issues regarding whether such insurance is feasible and would increase social welfare.

We conclude that the current PRIA draft is seriously flawed in its design to achieve the intended goals for financial protection and financing. Some flaws are easily corrected because the bill has incorrect assumptions about BI insurance law—specifically insurance coverage, regulations, and insurer finances. Payroll protection is discretionary now for BI insurance. PRIA assumes it is included. If the goal is to provide income protection for employees, then any pandemic BI insurance plan must require coverage for payroll, making it more comprehensive and more expensive. If however, the goal instead is to preserve a business only, so that when the pandemic is contained or ends the business can resume and hire back workers, then the program fails to solve the problem of protecting most people. This would require a new (and improved) Payroll Protection Program and supplemental unemployment insurance. This technical question underlies the conceptual economic gap in whether PRIA is a flawed finance mechanism or a flawed public welfare program. Another correctable flaw in PRIA is its failure to separate claims handling under the program from state claims statutes and bad faith laws.

Other flaws are harder to overcome based on a misunderstanding of insurance economics and general economics. One fundamental structural problem with PRIA is that, while the amount of risk that would be assumed by private insurers under the bill might seem small, pandemic BI risk is sui generis different from other catastrophic risks in that it is countrywide and worldwide. In contrast, floods, hurricanes, earthquakes, and (so far) crop failures are widespread but not countrywide (excepting small countries). Hence, pooling and diversification of BI pandemic risk is much more difficult than it is for other catastrophic risks.

Further, catastrophes are of short duration, typically hours or days (though their effects have longer duration), whereas pandemics last months or years (Smith, 2017, p. 488). Additionally, we show that even what would seem to be relatively low risk sharing by private insurers in a government program could be problematic for many in terms of the potential strain on their surplus. Moreover, insurers’ planning for an event that might be 50–100 years away creates unique capital accumulation, risk financing, and pricing problems that current tax law and insurance regulatory practices fail to address but this can be corrected.4

Pandemic business income protection raises the broader economics and policy question of whether such protection is a public good, owed to all, or a private good with substantial positive externalities. This question is linked with the question of whom or what this insurance should protect. To put it more bluntly, should BI pandemic insurance protect labor, or capital, or both? The answer to that question informs whether private insurance for pandemic BI expenses and losses on a broad scale is at all feasible and the role that the government should play in providing or financing it, if any. We acknowledge that private insurance for some portion of the economic losses from a pandemic may be possible.

If the goal is to provide income protection for employees, then any pandemic BI insurance plan must require coverage for payroll, making it much more comprehensive and expensive. If the goal of the program is to preserve a business only, then it would fail to protect most people. Further, if BI pandemic insurance has significant positive externalities beyond a firm and its employees, does this commend some form of government participation and financing to achieve a socially optimal supply of such insurance?

If such protection is not a public good, and instead is a personal protection scheme available to those who buy it (like life insurance or property BI insurance) to replace earnings and cover expenses if calamity strikes an individual or firm, then the problem is not public protection but public financing. This is a problem that arises with other types of insurance for catastrophes such as crop failures and floods and even terrorism, which sufficient federal money can solve, no matter how badly a program is designed. However, this problem is much greater for pandemic income losses that can considerably exceed the losses from other catastrophes, as demonstrated by COVID‐19.

Answering these questions reveals the core policy problems and directs the form of financing and risk‐bearing required. We conclude that the fundamental problems that need to be addressed are the scope and magnitude of pandemic income losses and the significant positive externalities stemming from insuring these losses. This leads us to consider whether these problems are more solvable if we think of BI pandemic insurance as a public good. The endemic problem of free riders arises with the provision of public goods and certain private goods, where people who do not buy the good expect to benefit from others buying the good. A business may skip buying this protection because it expects other businesses to buy it and then continue their transactions with the first firm. Additionally, a business that does buy this protection may find the free riders are gone and no longer able do business with it—this would still result in business failures and economic collapse. Businesses have additional incentives to free‐ride because the government is unlikely to allow economic collapse; it will likely provide financial aid regardless of businesses that chose not to buy this insurance.

Our analysis leads us to conclude that it is infeasible for private companies to provide this insurance on a broad scale and that private BI pandemic insurance will fall far short in protecting workers and the broader economy.5 Hence, we consider what a government BI pandemic insurance program that addresses these serious conceptual problems might look like and the challenges it would still face. This entails a critique of key structural features of PRIA and a discussion of alternative frameworks that could be preferable to PRIA. Discussing the challenges that any government program for BI pandemic insurance would face raises a number of issues and would require difficult choices. Hence, the essential question for policymakers is whether the best possible program would be in the public interest and increase social welfare. Further, even if a normative argument could be made in support of such a program, political considerations will likely influence the design of any program in ways that would make it less efficient and possibly less equitable.

Our paper is structured as follows. In the next section of our paper, we discuss the economic consequences of COVID‐19. We then review the basic features of BI coverage and the contract provisions that apply to coverage for losses stemming from contagious diseases or pandemics. This is followed by a discussion of the predicates for a pandemic risk insurance program. We then examine the nature of the risk associated with BI pandemic losses and discuss whether BI pandemic insurance has the characteristics of a public good or has substantial positive externalities. We follow with a critique of PRIA and consider alternative frameworks configured in a manner similar to the federal flood insurance or federal crop insurance programs. In this discussion, we consider important policy issues and consider how they might be resolved to make to make a program more efficient and equitable. Further, we discuss the rationale for federally subsidized premiums and how they might be designed. We conclude with a summary of our opinions.

2 THE INSURED AND UNINSURED ECONOMIC CONSEQUENCES OF COVID‐19 LOSSES

The COVID‐19 pandemic has had devastating health, financial, economic, and welfare effects in many areas of the United States and across the globe, and these effects are likely to continue. The medical expenses of treatment for COVID‐19 are covered by health or workers’ compensation insurance, for persons who have this insurance, and in some other cases paid by U.S. government assistance programs such as Medicaid.6 Long‐term debility among some survivors should trigger disability insurance payments for those persons who have such coverage through their employment or personal disability policies, or workers’ compensation where it applies. Many businesses have had severe income losses stemming from COVID‐19 due to government stay‐at‐home orders and other orders directing nonessential business to close temporarily or restrict their activities to reduce the spread of infection, among other developments such as supply chain disruptions.

Despite recent near misses of pandemics—Zika in 2015, MERS in 2012, H1N1 flu in 2009, SARS in 2003 (Abadi et al., 2020)—businesses generally have not purchased insurance for income losses and property remediation resulting from pandemics (Ziady, 2020). Income losses from pandemics have been available in some custom event cancellation policies, such as the one Wimbledon bought for $2 million, as it has for 17 years; this year it received $142 million for its cancellation due to COVID‐19 (Insurance Journal, 2020). The University of Illinois bought a custom policy against a drop in Chinese student enrollment if caused by “a predetermined list of causes, one of which is a pandemic” (Whitford, 2020). Marsh, the insurance brokerage, since 2018, has offered a pandemic risk insurance policy backed by MunichRe, called PathogenRX, with no buyers for the product (Lerner, 2020) “partly because the insurance was viewed as expensive given the risk” (Ziady, 2020).

One recent study found that businesses generally were unprepared for any unexpected global event; four out of five businesses that had a business continuity plan did not factor in unexpected global events such as pandemics, with the United States having “some preparation” for 64% of those surveyed, and “significant preparation” for 14% of those surveyed (Telstra, 2020). This simple survey suggests that many firms that may claim to have expected their BI insurance to cover events such as pandemics likely did not seem to consider this exposure before the current pandemic occurring.

The lack of specific pandemic risk insurance in place for business has led many firms to look at their existing BI insurance policies for possible coverage.7 Some states are also considering legislation that would retroactively require insurers to provide such coverage through their existing property BI policies.8 Many plaintiff attorneys have urged their clients and prospective clients to make claims asserting there is physical damage to their properties, or coverage under the civil authority provisions of their BI policies, due to COVID‐19. Insurers generally contend that, in most situations, their policies do not cover these losses because there is no physical damage to or loss of property, and the named perils do not include viruses, and in many policies, there is an exclusion for losses caused by viruses and bacteria, and the criteria for civil authority coverage is not triggered.9

Regardless of the opposing contentions, it is probably not debatable that insurers did not intend for their BI policies to specifically cover the peril of pandemics, and insureds (perhaps with rare exceptions) did not specifically seek or confirm at renewal or application whether their BI insurance would cover losses due to a pandemic in addition to other customary perils.10 Most court decisions to date (but not all) have ruled against policyholder claims for BI coverage due to the lack of physical damage to property either directly or to other property to trigger civil authority coverage.11

This problem has prompted proposals for some form of private or public insurance to provide BI coverage for economic losses stemming from a pandemic. One such proposal, introduced in the Congress on May 26, 2019, is the PRIA that would establish a program that would function in a manner similar to the program established by TRIA. Presumably, one attraction of private and/or government insurance for pandemic BI losses is that it would provide insured businesses with greater certainty that they will receive compensation rather than rely on the discretion of the federal government to provide aid and having to navigate the process to obtain it. Additionally, there is the attraction of using an insurance mechanism to place at least a portion of the burden of pandemic income losses (which may include the wages of furloughed workers) on private insurers or a government insurer, or a combination of both.

3 WHAT IS BUSINESS INTERRUPTION INSURANCE?

Though our review here focuses on the concept and possible frameworks for BI pandemic risk insurance, a brief review of property business income interruption insurance (also called business interruption insurance) as currently structured is appropriate to inform an examination of BI pandemic insurance proposals. This review reveals the correctable technical problems and the serious conceptual problems associated with PRIA and other possible frameworks for this coverage.

Manufacturers, retailers, lessors of property, and any business that needs a specific location to conduct its operations will suffer an insurable loss if the premises are damaged or destroyed, for which property insurance exists, and they will suffer an income loss while the property is unusable, for which a separate commercial policy called business income interruption insurance exists. A leading treatise on commercial insurance states: “Business income coverage is designed to replace the income that would have been earned by the business during the time when repairs are being made. For business income coverage to apply, the loss of income must be the result of physical damage from a covered cause of loss to property at the insured’s premises” (IRMI, 2020). “Because business interruption insurance coverage historically derives from the use of property, coverage is not normally afforded unless coverage under the property damage contract is afforded” (Borghesi, 1993, pp. 1150–1151).

The current Insurance Services Office (ISO) form for business income interruption insurance (CP 00 30 10 12) provides this coverage grant:

We will pay for the actual loss of Business Income you sustain due to the necessary “suspension” of your “operations” during the “period of restoration.” The “suspension” must be caused by direct physical loss of or damage to property at premises which are described in the Declarations and for which a Business Income Limit of Insurance is shown in the Declarations. The loss or damage must be caused by or result from a Covered Cause of Loss.

This insurance requires that a commercial Causes of Loss form be attached, typically either the broad form that specifies the perils insured, or the special form that is open perils with exclusions. Insurers may use their own custom forms that can provide broader, narrower, or other coverage than the ISO form.

The insurance requires, as a predicate, “direct physical loss of or damage to property.” As one author states: “In general, Business Income coverage … is tied to damage or destruction to property used, owned, leased, or operated by the policyholder in its business, property necessary for such operations (such as lobbies, stairwells, or elevators), or property at the ‘premises’ or within a certain distance (typically 100, 500, or 1000 feet). In the absence of such damage, there is no Business income coverage …” (Lewis et al., 2020, pp. 3‐36‐3‐36.1). Whether contamination constitutes physical damage has a long litigation history involving various contaminants but is an issue that is not relevant to this paper.

BI insurance comes with several additional coverages. These include coverage for extra expenses and losses due to the action of a civil authority. The latter comes into play when there is damage by a covered peril to a property other than the insured’s premises (typically, within one mile of the premises, though this can be modified) and a civil authority prohibits access to the insured’s premises. This coverage begins 72 h after the time of the civil authority action and continues for up to 4 consecutive weeks (this period can be modified). Businesses also can purchase coverage for income losses from “dependent properties.”12 Restaurants can add form CP 15 05 Food Contamination, that provides for lost income due to a board of health shut‐down order due to “food contamination.”

After the SARS outbreak in 2002–2003, many insurers added specific policy language that stated that virus‐caused interruption is not covered to remove any ambiguity as to whether viruses cause physical damage and thus trigger coverage (Exclusion of Loss Due to Virus or Bacteria, CP 01 30 07 06).

Further, property BI insurance typically does not cover losses due to the partial closure of the premises of a business, that is, if access to a building is limited but not prohibited, then any associated income losses are not covered. The relevance of this exclusion will become more apparent below where we discuss the fact that some businesses in a number of states have been subject to restrictions on how they operate but are allowed to serve their customers in ways that reduce the risk of contagion.

Many firms have sued their insurers seeking coverage for their income losses due to COVID‐19 and some states are considering making BI coverage for COVID‐19 losses retroactive. This retroactive application could create severe financial problems for the affected insurers.13 ,14 because had not priced and collected premiums for pandemic‐caused losses. Additionally, when courts or regulators retroactively expand the definition of the losses insurers are required to pay, this creates greater uncertainty for insurers. When insurers face greater uncertainty regarding the interpretation of their contract provisions they will raise their rates or withdraw from markets.

4 REQUIREMENTS PREDICATE TO CREATING PANDEMIC BUSINESS INCOME INSURANCE

If pandemic BI insurance is to be widely available, what purpose should it serve? This requires specifying who is to be protected, meaning what such insurance is supposed to pay for (the income and expenses to be covered), and specifying what triggers such coverage. Otherwise, it will be little more than a complicated pre‐event scheme for scattershot economic support. Failing to specify this coverage and its purpose afflicts PRIA, and underlies the externality problem we address later. The former problem can be solved with improvements to the bill or addressed in alternative proposals; the externalities stemming from this coverage is a fundamental conceptual problem not easily solved.

4.1 Should payroll be covered?

Property BI coverage pays to replace ongoing expenses and net income losses. Premiums depend on what the estimated expenses and income are. What constitute ongoing expenses are somewhat difficult to predict when applying for BI coverage. For existing BI policies, these expenses can include management salaries and overhead that continue, some insurance expenses that continue, and mortgage payments that probably continue (unless paid off by the property insurance). Expenses that probably cease are utilities and rent when a building is destroyed, but may continue for less severe damage. Whether to insure payroll is optional under standard BI insurance policies. A firm may either cover it if it wants to pay and retain its workforce, or not cover it if it intends to release its employees.

A firm that excludes payroll will pay lower premiums, but will likely have to find new workers when its operations resume. Therefore, it is crucial that any proposal for BI pandemic insurance specify whether the replacement income is to pay for a firm’s ongoing expenses and net income loss, or whether it is to pay the wages of the employees as the recently enacted Paycheck Protection Program does, or both. If payroll coverage is optional, it will be necessary to provide worker protection in some other way.

4.2 What are the triggers for coverage that cause a pandemic income loss?

A second consideration is what triggers the insurance. Put another way, what is the occurrence that sets in motion a covered loss? For ordinary property losses caused by any of the named perils, or nonexcluded perils using a special Causes of Loss form, the occurrence of such perils causing loss of or physical damage to property initiates the coverage. This is true even when the civil authority coverage provision is initiated due to physical damage due to a covered peril to nearby property that results in prohibited access to the insured’s property. Pandemics (viruses and bacteria) do not cause physical damage; they cause bodily injury and death due to the presence of the virus or other pathogen.15 A public health order restricting movement is intended to prevent viral (pathogenic) spread. What government orders would be necessary to cause a BI loss that could be insured?

A business can incur losses due to a pandemic in several ways. First, in response to COVID‐19, many state and local governments have issued stay‐at‐home orders that only allow citizens to be outside for limited purposes such as to obtain food, medical care or medicine, exercise, financial needs, and some other exceptions.16 Some orders require businesses to close, except for those considered “essential,” or restrict their operations. These restrictions have decreased businesses’ sales and may have increased their expenses. For example, public orders may allow restaurants and bars to operate, but only provide takeout and delivery service or operate with reduced occupancies. Even businesses allowed to operate without restrictions can suffer pandemic‐related losses if other firms they depend on are closed, their supply chains disrupted, and/or their sales fall because their customers have lost income or wealth or simply choose to stay home. For any proposal for BI pandemic insurance, it is necessary to define what triggers a pandemic related loss that is covered. The more expansive the trigger, the more problematic insurance becomes for reasons we discuss below.

The Congressional Research Service (CRS) examined these differences and concluded that the federal Public Health Services Act gives the Secretary of Health and Human Services authority to declare a “public health emergency” and to assist the states in quarantines, in addition to other powers not relevant to this paper (Swendiman & Jones, 2009, pp. 1–2). The President can declare a disaster under the Stafford Act and assist the states and local governments with emergency assistance, although “there may be uncertainty regarding whether a flu pandemic, or any outbreak of infectious disease, would be eligible for major disaster assistance under the Stafford Act.” (Swendiman & Jones, 2009, p. 3).17 The essence of their analysis is that while the federal government has the authority to issue orders that prevent disease transmission, the primary authority for quarantine and isolation exists at the state level as an exercise of the state’s police power.18

The result is that for a pandemic‐caused income loss to be insurable, the triggers must be clearly specified, not only for technical reasons but because they are the actual causes of an income loss when the orders restrict business activity. The first requirement is a declaration of public health emergency. The second requirement is a governmental order—by a state or local authority, not the federal government unless a new provides for that—prohibiting access to a business or requiring a business to be closed for a period, because the restriction is what reduces the businesses’ income and the employees’ wages. Without this state or local restriction on operations, a business would not be restricted and thus might not incur any loss. If the goal is income replacement due to income loss, then something more than a public health declaration is needed to result in a business income loss. Thus, a pandemic that does not have a quarantine order of some type is likely to cause much lower BI losses.19 Therefore, at the least this second trigger is necessary to tie a payment to something that is likely to cause an actual income loss.

5 THE BROAD REQUIREMENTS OF INSURABLE RISKS, MEASURED AGAINST PANDEMIC INCOME LOSSES

Here we begin our analysis of why creating insurance for BI pandemic losses poses many challenges. This involves both insurance economics, and insurance regulatory and financial constraints. It may be possible to address some of these challenges, but others may not be solvable or require difficult choices with considerable tradeoffs. The nature of BI pandemic risk makes private insurance on a broad scale infeasible and government insurance challenging due to the characteristics necessary for insurable risks. First, BI losses due to a pandemic will be catastrophic, highly correlated, and not geographically diversified. Second, determining and measuring the pandemic BI losses of a given business could be difficult. This would not necessarily be the case for a business that was forced to close. It would be more challenging for businesses that suffer losses due to restrictions on their operations or other reasons. Third, how a business responds to restrictions on its operations and other problems affects the amount of its losses. For example, a business that makes strong efforts to serve its customers in innovative ways will incur lower losses, all other things equal. This difficulty of fixing the net income loss, also related to mitigation of damages, could be solved with a parametric trigger that we discuss in Section 8, but this would have some drawbacks.

This leads to a broader problem that a pandemic not only becomes a new peril (if so imposed) to a business, it serves as a peril to the economy, causing widespread unemployment, hospitalization and death, and possible economic collapse. Thus, the nature of a pandemic peril is that it is an equal and lingering peril to the economic capacity of its customers to trade with that particular business. The business may reopen after some period (e.g., 6 months, 2 years, etc.), but the customer base may be reduced by a 10% or a higher death rate of the population, and the collapse of other businesses. Even if the mortality rate associated with a given pandemic is low, disability caused by infections and fear of contagion may cause many consumers to decrease their purchases of certain goods and services.

Here we discuss why we believe that relying solely on private insurers and reinsurers to provide coverage for business interruption losses on a broad scale is infeasible. We begin our discussion with a review of the conditions for an “ideally” insurable risk from the perspective of private insurers. We then consider those conditions violated by BI pandemic risk. Additionally, we discuss other reasons why it would be difficult for many private insurers to provide coverage for pandemic BI losses. We note that certain violations of these conditions also would create problems for government insurance mechanisms for BI pandemic risk.

5.1 Pandemics do not meet the requirements for an ideally insurable risk

The insurance literature outlines the characteristics of or requirements for an “ideally” insurable risk from the perspective of private insurance companies (Redja & McNamara, 2016). Some of these requirements also apply to government insurance programs. These characteristics are:

• There must be a large number of exposure units.

• The loss must be accidental and unintentional.

• The loss must be determinable and measurable.

• The loss should not be catastrophic.

• The chance of loss must calculable.

• The premium must be economically feasible.

We contend that are either not met by BI risk arising from pandemics or at least engender problems with insuring this risk.

First, the condition that a loss must not be catastrophic allows insurers to use diversification to manage the claims they would be required to pay. Private insurers do provide coverage for natural catastrophes (e.g., earthquakes, floods, hurricanes, etc.) but they can use geographic diversification, reinsurance, and financial instruments (e.g., catastrophe bonds) to manage the claims arising from these events. Earthquakes, floods, and hurricanes cause widespread damage, but do have boundaries where the damage ends, and earthquakes and hurricanes do not cause damage across an entire continent let alone the whole planet. Volcanic eruptions cause local extensive property damage and death, and the largest ones have worldwide cataclysmic effects on sunlight and food supplies and cause famine and disease far away (Oppenheimer, 2011, pp. 295–310). A volcanic eruption is a covered peril on the broad form causes of loss form, and by endorsement to the special causes of loss form.20 A business income loss due to volcanic eruption would only be covered if the volcano’s debris or explosion actually caused physical damage to the insured’s property, like any other cause of damage, or damaged nearby property that would trigger coverage for a civil authority order prohibiting access.

As Skipper and Kwon explain:

The independence property is important because it affects how well insurers can diversity the systematic risk of their insurance pools. Ordinarily insurers are able to diversify risk by forming large pools of exposure units that are statistically independent of one another thereby lowering the average risk per exposure unit in the pool. However, when the exposure units in insurance pools are all subject to the possibility of suffering losses due to a single catastrophic event, they are no longer statistically independent. In this case, the exposure units are interdependent or correlated. When risks are correlated, diversification across exposure units is severely weakened and the resulting negative effect on insurance markets can be dramatic (Skipper & Kwon, 2007, p. 489).

Private insurers in the United States also cover terrorism risk but they do so with a federal reinsurance backstop (TRIA), and again the possibility of terrorists causing widespread property damage across the United States is (so far) unimaginable.21 The September 11, 2001 attacks on New York directly affected “only” a few buildings, and resulted in closure of parts of lower Manhattan, not the entire island or the city, though the economic impact was throughout the region and the country.22

BI pandemic risk is catastrophic―pandemic income losses will be severe and highly correlated―but unlike for other catastrophic perils, geographic diversification for this risk is infeasible given that many areas in the country (and the world) can be affected at one time.23 Further, the BI losses from one pandemic could be several times larger than the property losses from a severe earthquake or hurricane. For example, CoreLogic estimated the total property losses that would result from a hypothetical 7.0 earthquake that occurred along the Hayward fault line in the San Francisco East Bay Area of California (CoreLogic, 2018). It estimated that the total property damage from such an event would be $170 billion of which only about $30 billion would be insured.

Currently, the total economic losses that will result from COVID‐19 can only be estimated. One study has estimated that losses in the U.S. gross domestic product (GDP) could range from $16 billion to $1.8 trillion under seven different scenarios (McKibbin & Fernando, 2020). Makridis and Hartley (2020) estimated the U.S. GDP losses due to COVID‐19 mitigation measures alone (e.g., partial shutdowns, social distancing, etc.) were $2.14 trillion for February and March of 2020; this figure is very close to $2 trillion cost of the CARES Act that was the first legislative initiative to deal with the economic losses of COVID‐19.24 BI losses from a COVID‐19, or any imaginable pandemic, could considerably exceed the potential property losses from other types of severe catastrophes.25

The private insurance and reinsurance industries currently do not have the financial capacity to cover the BI losses that would arise from a pandemic because insurers did not factor BI losses for pandemics into their rates or capital needs; the combined surplus of U.S. property‐casualty insurers was approximately $770 billion as of the end of the first quarter of 2020.26 To acquire such a capacity, private insurers and reinsurers would have to collect very large amounts of premiums and set aside funds over many years. This is impractical for reasons we discuss below.

Additionally, there is the issue of the timing risk associated with pandemics; timing risk has capital accumulation and liquidity implications for insurers and reinsurers. We will assume that pandemics are low‐probability, high‐consequence events like natural catastrophes. If pandemics occur infrequently and their occurrence is spread out over time, then it becomes easier for private insurers (or a government insurer) to build up financial reserves during periods when there are no pandemics, but harder to keep the necessary capital segregated for only pandemic claims.

What is the interval or range of intervals for pandemics? Based on studies of previous pandemics, the potential (though not probable) frequency of a pandemic is about 3–5 years (Klein & Weston, 2020b). Saunders‐Hastings and Krewski (2016) state “Scholars tend to give a fairly consistent estimated interval of 10–50 years between influenza pandemics. This is a very broad window, suggesting that pandemics occur with an irregularity that prevents accurate estimates of their frequency. Huynh et al., 2013, pp. 239–240), examined mortality from pandemics, and note the difficulty of estimating the frequency of pandemics: “In contrast to seasonal influenza, epidemics that occur annually, influenza pandemics are rare and unpredictable events, which have occurred irregularly throughout history. Since 1590, there have been between 11 and 14 influenza pandemics with as little as 2 years separating some outbreaks and as many as 6 years between others.”27 Huynh et al. (2013) conclude “there is a 3–4% chance of an influenza pandemic occurrence in any given year.”

Natural disasters have largely followed a 10‐year cycle in the United States.28 The next pandemic could occur within the next few years, before insurers would be able to accumulate reserves sufficient to pay their claims obligations without some form of federal backstop, or in another 100 or so years. This means that it is necessary to segregate funds for such events and the capital and implicit gains not counted against current premium rates and in calculating loss ratios. Otherwise, the very long‐term nature of reserving for this risk will be curtailed. There are also liquidity problems to deal with. We discuss these aspects of capital accumulation later.

Certain other conditions for an ideally insurable risk would be pertinent for private or government insurance for BI pandemic losses. One such condition is that losses must be accidental and unintentional, that is, not under the control of the insured. This requirement ensures that loss‐causing events are random, which allows the law of large numbers to operate in insurance pooling arrangements. Additionally, moral hazard increases if an insured can influence the amount of the insured losses they incur.

The BI losses associated with a pandemic are accidental and unintentional. However, the actions of individual firms can influence the amount of losses they will incur due to a pandemic. Some businesses can employ measures to continue to serve their customers that will reduce their economic losses due to a pandemic (e.g., providing curbside or delivery service; see Klein & Weston, 2020a). If the owner of a business knows that her BI pandemic losses will be covered by insurance, she will have diminished incentives to reduce these losses through such measures. This is little different from a property loss where the insured might either seek to continue its business operations at a reduced capacity, perhaps drawing on the extra expense coverage of BI, or take the insurance money and wait until all the property damage is fixed and operations can resume. Still, if government imposed restrictions on businesses’ activities are major causes of losses due to a pandemic, then the discretion of their owners in finding ways to operate at reduced capacity remains an issue. Further, governments also influence the likelihood and scope of such events as well as make decisions regarding closure and restrictions on different kinds of businesses, further challenging the requirement that losses be accidental and unintentional.

A third requirement is that a covered loss should be determinable and measurable. This means that the loss should be definite as to cause, time, place, and amount. The purpose of this requirement is to enable an insurer to determine if a loss is covered and how much it should pay if it is covered. This is always a challenge in property BI losses even with the best good faith estimates and projections. For a pandemic‐caused BI loss, there are external economic factors affecting the loss and recovery, thus making the calculation and resulting indemnity harder. This is a problem that could be addressed, however, albeit with some tradeoffs. The development of parametric insurance that pays a fixed amount of indemnity when triggers of coverage are met eliminates the calculation problem, but this creates “basis risk” in that with a parametric trigger, an insured could receive payments considerably less or more than their actual losses.

The fourth requirement is the condition that the chance of loss should be calculable. This means that an insurer should be able to estimate the frequency and severity of future losses with some degree of accuracy. This requirement helps to ensure that an insurer will collect sufficient premiums over time to cover its claims payments, expenses, and cost of capital, as well as purchase reinsurance. Insurers do encounter some difficulty in estimating the frequency and severity of future losses for some risks that are insured. For example, insurers use sophisticated models to estimate the risks of natural disasters and resulting property damages. However, they have no way of knowing how accurate the estimates of their models truly are due to parameter uncertainty. Further, the speed of development of therapeutics and vaccines is likely to accelerate after COVID‐19 and there will be investments in office design and air purification, thus likely reducing expected losses. Consequently, the mortality, debility, and economic losses due to the next pandemic could range widely making pricing and financing for such an event very challenging.29

Estimating pandemic risk is not entirely out of line for insurers. Life insurers model the mortality risk associated with pandemics (see, e.g., Toole, 2007) and health insurers model the morbidity risk arising from pandemics (see, e.g., Toole, 2010). There is also capital markets support for life and health expenses due to pandemics provided through an excess mortality bond for extreme health care costs due to a pandemic (Smith, 2017, pp. 489–490). If these models can be adapted to the property side of insurance, and then add the separate challenge of modeling actual business losses from a pandemic, separating losses due to closure and losses after the pandemic ends due to subsequent economic collapse caused by high mortality and other effects, then some idea of the risk and required premiums could be estimated. Perhaps the model used to develop the SwissRe business interruption cover against volcanic eruption can underlie this pandemic business interruption cover. We suggest that such models be developed first (if feasible) before expecting private insurers or a government insurer to offer coverage for BI losses from pandemics.

5.2 Capital accumulation, liquidity, and regulatory issues

Let us assume that the criteria for an insurable risk can be overcome by some fearless insurers, at least for part of the exposure. Let us also assume that the next pandemic is sufficiently far away to allow insurers to build reserves to cover BI losses arising from it.30 Such dedicated, massive capital for this loss has its own tax and regulatory problems and even problems stemming from bad faith insurance lawsuits involving other types of insurance. These are complex problems to solve because of state regulation and taxation, requiring changes to insurance regulations and tax law. We explain these problems and offer partial solutions.

There is no provision in federal or state laws that would enable an insurer to establish dedicated pandemic BI reserves that would receive favorable tax or regulatory treatment. Under current treatment, capital and reserves segregated for pandemics would be an addition to an insurer’s surplus and taxed accordingly. Additionally, if an insurer were to accumulate large amounts of additional capital for BI pandemic losses, state regulators looking at overall profitability and surplus may well pressure it to lower its rates on other insurance lines, not to mention the demands of stockholders complaining about tied‐up capital. Hartwig et al. (2020) calculate that the amount of capital needed for pandemic income insurance is 750% to 950% of the expected loss, compared to 220% for other catastrophic losses. This suggests that the tail risk for pandemic income losses is much greater than it is for other catastrophic losses. The authors also look at the pooling effects of such losses of time, which indicates a 125%–325% capital to loss ratio for each insurer, though the authors note serious problems with such pooling arrangements. Here, some sophisticated finance mechanisms could provide solutions, such as pandemic catastrophe bonds, and special purpose vehicles for holding pandemic capital accumulations that would receive more favorable tax and regulatory treatment.31

A related concern is such a large accumulated reserve (or surplus) would be used in all sorts of bad faith insurance cases to set punitive damages, because one factor in setting the punitive damage amount is the insurer’s net worth.32 Thus, this requires segregating the reserves for pandemics the rest of the company’s surplus so that they are not used in setting bad faith awards and thus reducing the funds available to pay for future BI pandemic claims.

A third problem arises if the terms of and rates for pandemic BI insurance are subject to state regulation. This may not have been a problem to date as few insurers have offered BI pandemic coverage. However, looking forward, the terms and pricing of BI pandemic insurance products may attract greater regulatory attention. Some states may be inclined to compel insurers to offer very “generous” coverage terms and/or suppress their rates for this coverage. It is reasonable to expect that coverage terms and rates would be politically contentious and this would increase pressure on regulators to adopt policies that favor insureds over insurers.33 The states may strongly resist any federal legislation that exempts BI pandemic insurance from their regulation.

6 IS BI PANDEMIC INSURANCE A PUBLIC GOOD?

Our discussion of whether private pandemic BI insurance of some sort is feasible reveals a more significant core policy question of whether this pandemic income replacement is a private or public good or a combination of both, plus some related secondary questions. PRIA skips over this question in reaching for a flawed public‐private funding mechanism. We previously pointed out the flaw on whether this program is to protect wage earners or only capital. This technical drafting flaw reveals the core economic flaw, and the need to clarify the goals of any such program. We contend that a pandemic BI insurance program that provides broad economic support to businesses and workers should be analyzed within a public good framework.

If the limited goal of pandemic BI program is a personal protection scheme available to those who buy it (like life insurance or disability insurance), with a fixed amount of compensation or of duration, then such income replacement then the program is a private good. This is because there will a limited number of buyers only and only firms who pay for the insurance protection will receive it. If, however, BI pandemic insurance aims to provide basic replacement income for all persons and businesses due to the economic fallout from a pandemic, then it could be argued that it has the attributes of a public good that must be provided to all. For income protection against economic collapse due to a pandemic, the public goods aspect is even stronger. Regardless of the scope of coverage (business income, payroll, or something broader), the nature of BI pandemic risk makes private insurance infeasible—although this does not preclude some valiant and capable insurers from offering a private version as a supplemental coverage (as an excess insurance policy or difference in condition type of policy). A public goods conception leads to a publicly financed option administered by insurers, like flood and crop insurance that could achieve the objectives associated with a public good employing the administrative resources of private insurers in servicing policies.

The standard definition of a public good is that such a good is both nonrivalrous and nonexcludable (Kalhoff, 2011; see also Samuelson, 1954 and Oakland, 1987). A good is nonrivalrous when one consumer’s use of it or the utility they derive from it does not diminish another consumer’s use or utility from the good. A good is nonexcludable when it is not possible to exclude anyone from using it. Economists often cite lighthouses and national defense as examples of public goods that have both characteristics. In contrast, a private good is one where its consumption is rivalrous (one person’s consumption of the good reduces its supply for others) and is excludable (people can be excluded from using it if they are unable or unwilling to pay for it). Lighthouses and national defense aside, most scholars acknowledge that there are gradations between public and private goods. A good may have one of the characteristics of a public good but not the other, or may have one or both characteristics in part but not in whole. Kalhoff (2011) calls these variations of pure public goods “club goods” and “common pool resources.”

The inability to exclude persons or firms from using or benefiting from a good creates a “free rider” problem. If no one can be excluded from using a good, then people or firms cannot be forced to help pay for it, and they benefit from the good regardless. Some people may be willing to pay for certain public goods voluntarily, but many may choose not to if they believe that the good will be provided regardless.34 This can lead to the underproduction of such goods by private markets alone and why public financing is used for certain public goods.

One could argue that BI pandemic insurance is a private good in the sense that one firm’s consumption of this coverage can reduce its availability for other firms, assuming that there are limits on its supply. Further, insurers do not provide this coverage to people unwilling or unable to pay for it. However, if the policy goal is that every business or individual has insurance protection against a pandemic‐caused income loss, then everyone must be compelled to buy this (or receive permission to self‐insure). So far, insurance mandates have only been successful for liability exposures from motor vehicles and workers’ compensation; the individual mandate provision in the Affordable Care Act was very unpopular and eliminated by the Congress in 2017.

BI pandemic insurance has one attribute of a public good to some degree. This argument hinges on the extent to which a firm’s purchase of this coverage provides broader benefits to others from which they cannot be excluded. Businesses that have replacement income for some time will have money to restart their businesses when the pandemic pauses or eventually ceases. Workers and consumers also could benefit from this insurance to the extent that it enables firms to remain in business and continue to pay their employees and provide goods and services to their customers. Yet the public side is only partly satisfied when only a few businesses have such pandemic income insurance because that leaves the business‐to‐business and business‐to‐consumer markets with less demand for insured businesses that remain open, and leaves the insured business with a diminished supply of inputs that their income can buy.

Looked at another way, there are two markets to consider for pandemic income problems. One is the insurance market that may treat a pandemic as an insurable peril and provide the income replacement. The other is the operating (or commercial) market because if very few other firms have pandemic income insurance then the insured business will fail for lack of inputs and customers to buy its products. This gives a different perspective on what pandemic income insurance is supposed to do―provide replacement income, or preserve aggregate spending for economic survival. Thus, a market for pandemic BI insurance, whether private or the public–private versions now under discussion, is caught in the standard free‐rider problem because if enough other businesses buy such insurance, then a firm will have less reason to buy it because enough other businesses will have money to do business with it.

We note that government management of a pandemic also has the attributes of a public good. This is relevant to the topic of this paper as how governments respond to a pandemic affects BI losses due to the pandemic. Anomaly (2011) contends that public health (as health and prevention of disease) is a public good, though Dees (2018) would expand public health to the reduction of a low‐pathogen environment of air and water. We leave aside the distinction—and a fraught public policy question in the United States—between personal health as a private consumption item of privilege or public health as a matter of community health and individual right. Indeed, public health measures combined with a government BI pandemic insurance program as well as other types of assistance constitute government pandemic risk management writ broadly.35

As Kaul (2001) notes, “what makes goods public or private is not how they are provided (i.e., whether they are produced by the state, provided by the market, or resulting from public‐private partnerships). Rather, the essential question is whether a good has significant public benefits.” This, we shall see, is how private insurers can be agents and servicers for a public‐financed option, like the flood and crop insurance programs. Sometimes the distinction is a policy choice, for example, with television, which can be open broadcast that anyone can receive with a television, or by subscription to unscramble the signal (Kaul, 2001). Holcombe (2000) argues that the distinction of public or private goes to jointness in consumption (essentially meaning nonrivalrous) so that “once a good is produced for one person, additional consumers can consume the good without reducing the consumption of any existing consumer.” Yet this also means that the price should be zero, because if the point is to make the good available to all, then consumers unwilling to pay the price are excluded (Holcombe, 2000).

Public goods need not be publicly financed (Kalhoff, 2011). Thus, whether the government or private insurers provide pandemic income replacement, or some combination of both, is not a crucial decision (assuming there is sufficient private capital, and other problems we discussed in Section 5.2 are solved). However, it will affect whether private suppliers provide the good here to firms that can choose to buy it.

Even if we accept that pandemic income replacement can be provided on a national or international basis by the private insurance market, alone or supported by public money, we must ask whether the goals of income replacement, or income sustenance for workers, can be achieved by leaving the choice of buying such insurance to employers, with sufficient uptake by businesses. If the purchase is a choice, many businesses, already reluctant to spend yet more money on insurance, might well decline, in the same way that many businesses and people decline to buy flood or earthquake insurance and count on the prospect of government support and subsidized loans postdisaster. It would take a remarkably stern government to decline to offer such aid to ease massive suffering because businesses could have bought insurance but did not. This pushes open the free rider problem, and again forces the question of whether such income protection is for the business itself (and its owner) or for the employees, or both.

The scope of BI pandemic insurance has implications for who benefits from it and its financing. If this insurance covers a firm’s income, payroll, and other expenses, then it will be provide broader benefits to the economy then if payroll is excluded. These broader benefits beyond the insured firm increase with other provisions that expand the coverage (e.g., coverage for dependent properties). The broader the benefits, the less private financing will result in a socially optimal amount of coverage produced.

BI pandemic insurance, regardless of who provides it, would have substantial positive externalities.36 Those businesses that do not buy private or public–private pandemic income replacement insurance will not incur the cost of such insurance, but may well get much of its benefits—making them free riders. If they can hang on a while, they will have customers and suppliers to transact with because those other firms bought insurance. Arguably, any form of private insurance currently offered provides benefits to those beyond the insured. However, it also can be argued that the positive externalities stemming from BI pandemic insurance would be substantially greater than they are for most other insurance products.37 Hartwig et al. (2020) state to this point:

One potential external benefit is that it reduces the likelihood that business relationships among contracting parties or between the business and its employees will be damaged or disrupted. Also, by helping to maintain economic activity, even at a lower than normal level, compensation to entities that experience a loss can reduce the magnitude and duration of a recessionary economic period. A disadvantage of purely private insurance mechanism is that these positive externalities would not be taken into consideration in the pricing and distribution of the insurance. Potential solutions include government subsidies for private insurance or even government provision of the insurance at reduced rates.

The other dimension of public goods relevant here is public welfare, as workers are more likely to be discharged if a business does not include payroll expense in their purchase of pandemic income replacement insurance. A reasonable presumption is that a business that excludes payroll expense on its standard business income interruption insurance will likewise exclude payroll expense on pandemic BI insurance. This would leave many workers unprotected and thus falling back on unemployment benefits. If this happens, then the businesses, and more specifically the owners, will have replacement income, while the workers are shown the door. Because unemployment insurance is considered uninsurable by private markets and best handled as a social insurance program by the government, this suggests that optional pandemic income replacement insurance will do little to protect the income of workers and leave that burden on government.

This seems an unsatisfactory outcome for the goal of a private or public–private partnership of pandemic insurance if one of the goals is to protect the wages of workers and the broader economy. As Anomaly (2015) says, “government action is occasionally the only feasible or cost‐effective way of bringing about an outcome which each person sees as beneficial—or would see as beneficial under idealized epistemic conditions—but which they lack the power to bring about unilaterally.” Unless we are able to require that every person be covered for their wage losses due to a pandemic, either provided by their employer or purchased on an insurance market, we cannot expect a private market system to fully protect the social welfare. Thus, we return to the fundamental question of who and what this insurance will protect, and in turn, to what extent public financing will be required to achieve an optimal supply of it.

As we noted above, the effectiveness of public health systems and measures can affect income losses for businesses and workers due to a pandemic. Potentially, the more robust and effective these systems and measures are, the lower the economic losses from a pandemic can be over the long term. Consequently, underfunded and fragmented public health systems, as we see in the United States, can increase pandemic BI risk and the cost to insure it.38 Further, governments may feel that they have greater latitude in imposing restrictions if firms and workers are adequately protected for their income losses. This underscores the argument that efficient government pandemic risk management, encompassing loss control and risk transfer, requires coordinated strategy and measures.

7 EVALUATION OF THE PRIA

Here we discuss what a federal pandemic BI insurance program might look like and the implementation challenges it would face.39 We begin with a brief summary of TRIA followed by our evaluation of PRIA, as the latter employs the framework of the former. Our critique is confined to key structural features of PRIA and does not include provisions that could readily be fixed.40 We conclude that the approach taken by the PRIA proposal does not make sense, as private insurers would only be able to assume a very small portion of the risk among other problems with this approach. This leads us to consider alternative frameworks in Section 8 that could be structured in a manner similar to the federal flood insurance program or federal crop insurance program.

7.1 Terrorism risk insurance act

TRIA was enacted in 2002 in response to reinsurers’ unwillingness to provide coverage for terrorist events following the 9/11 attacks. This, in turn, prompted primary insurers to exclude coverage for terrorism in their policies. The program essentially provides a federal reinsurance backstop for private insurers who provide coverage for property damage and liability arising from a terrorist attack. The program comes into effect when the Secretary of the Treasury certifies that an act of terrorism falls under the definition of terrorism under TRIA and triggers an event dollar threshold.

All insurers offering commercial lines covered under TRIA are required to offer terrorism coverage but a firm is not required to buy it. TRIA covers most commercial lines with a number of exclusions including commercial auto and professional liability insurance, among others. The types of losses covered under terrorism insurance include commercial property, business interruption, and general liability.

The program is triggered when industry total losses from a certified terrorist act reach $200 million. Each individual insurer retains a deductible of 20% of their direct premiums earned for commercial insurance and 20% of its losses above its deductible. The aggregate industry retention (deductibles and copayments) is capped at $46 billion in 2020.41 All losses in a program year for both insurers and the federal government are capped at $100 billion.

The government relies on ex post financing to recover the payments it makes to insurers under the program. TRIA provides for two types of recoupment—mandatory and discretionary—for the federal share of losses under the program. Mandatory recoupment applies when industry losses fall below the aggregate industry retention level. Under this recoupment, insurers are required to impose and remit a premium surcharge on all policies over a specified time‐period in TRIA‐eligible lines until total industry payments reach 140% of any mandatory recoupment amount. Under discretionary recoupment, the Treasury may require insurers to impose and remit additional premium surcharges for amounts paid by it for losses above the aggregate industry retention level.42

7.2 Key provisions of PRIA

The current draft of PRIA has a number of provisions similar to TRIA—which would be fine if pandemics were like terrorist events limited to time and location. Like TRIA, PRIA would be triggered in the event of a “public health emergency” for an outbreak of an infectious disease or pandemic for which an emergency is declared, on or after January 21, 2021, under the Public Health Service Act and that is certified by the Secretary of Health and Human Services as a public health emergency.

The bill appears intends to provide coverage for BI losses arising from a pandemic but its language is unclear as to what would be covered. In one section it defines BI insurance narrowly (requiring reimbursement only for losses due a complete shutdown of business. However, another section of the bill states that participating insurers would be required to “make available business interruption insurance coverage for insured losses that does not differ materially from the terms, amounts, and other coverage limitations applicable to losses arising from events other than public health emergencies.” This implies that businesses could be covered for losses even if they were not completely shut down. Hence, the losses that would be specifically covered under PRIA are unclear. This ambiguity can be fixed but how it is fixed involves tough choices and tradeoffs.

The lines of business covered under PRIA are essentially the same ones covered by TRIA with the same excluded lines and coverages. As presently constructed, an insurer’s participation in the program would be voluntary.

Insurers would be responsible for a deductible equal to 5% of their direct premiums earned for specified property‐casualty insurance lines for the previous calendar year. Additionally, insurers would retain 5% of their losses above their deductibles. The program would not be required to make any payments to insurers until total industry losses exceeded $250 million. Consequently, participating private insurers would be fully responsible for covering the initial $250 million of losses for which they would receive no reimbursement by the program. Total payments by the program would cease when aggregate insured losses exceed $750 billion.

Interestingly, PRIA lacks specifics on the funding of the federal share of losses. The bill states that the federal reinsurance payments and administrative costs of the program will be appropriated out of funds in the Treasury not otherwise appropriated.43 Unlike TRIA, PRIA does not provide a postevent mechanism to recover the reinsurance payments made and other costs of the program. Further, PRIA does not provide a pre‐event financing mechanism for the federal government’s costs using some form of premiums collected by insurers and remitted to the treasury. We presume that general revenues (i.e., taxpayers) would fund the PRRP’s costs. Participating insurers would determine their premium rates for the coverage they provide.44

7.3 Evaluation of PRIA

Mindful that our critique is of a bill subject to revisions, we begin with issues concerning contract design, followed by issues regarding how insurance should work. We then discuss the potential surplus strain PRIA would create for participating insurers. These are issues that also could arise with alternative frameworks depending on their features

7.3.1 Insurance contract concerns and fundamentals

Some of the issues for insurance contract coverage and triggers for coverage result from poor drafting and could be solved with technical corrections, but these corrections require some tough choices. First, what would constitute a covered loss is ambiguous in terms of what would trigger coverage. If a business could only file a claim for losses if it was forced to completely shut down, this could be relatively straightforward for an insurer to determine. However, the current draft of the bill requires only a federal declaration of a public health emergency to trigger coverage, which is not the same as ordering businesses closed or restricted, because states and municipalities issue such orders. These two different jurisdictional triggers need to be connected―declaration of a public health emergency and stay‐at‐home orders (or similar activity restrictions). The catch here is that this could make the program subject to decisions at the state and local level without a coordinated national strategy.

Second, the bill assumes a pandemic‐caused loss is the same as the perils covered by property BI insurance, but this is not the case. The bill needs to create a separate cause of loss for pandemic‐payouts only, in line with terrorism as a separate cause of loss that triggers coverage under TRIA‐backed insurance policies. Third, if the bill intends that BI pandemic insurance provide income to workers, then it must specify this requirement. Fourth, the bill appears to leave the regulation of claims handling to the states but this is problematic given the variation in state regulations as well as the law governing bad faith disputes, and then any overlay of new claims handling regulations. This can be corrected by mandating claims handling under new federal regulations that preempt state laws and regulations. Even under federal oversight and rules, claims disputes can arise as happened with flood insurance claims after SuperStorm Sandy in 2012.

We now turn to our concerns that PRIA fails fundamental rules on what is insurance and how it works. Within this category is the problem of moral hazard of allowing insureds to shutter their businesses without trying to minimize losses. The current pandemic has resulted in some businesses providing alternative services, albeit reduced from prepandemic levels. Some businesses have innovated to provide curbside pick‐up and home delivery. Having insurance could reduce the incentives for a business to find ways to serve its customers still allowed. Further, if a business could file a claim for losses arising from restrictions on its activities or for other losses that could be attributed to a pandemic, this would substantially expand the scope of what is covered.

Another concern is rate setting. It would be reasonable to expect that private insurers would be subject to political pressure with respect to their rates for this coverage. Further, if the PRIA bill is amended to authorize the Treasury to set and charge premiums for the government’s share of losses under the program, these premiums may be subject to political pressures for rates below what should be actuarially fair.45 This has happened with the federal flood insurance program. Further, studies of state insurance rate regulation indicate that some states have suppressed rates for certain types of insurance (e.g., homeowners insurance in states subject to hurricanes) for which the cost of coverage is politically contentious (Born et al., 2018).46

Further, to the extent that insurers would bear risk under the PRRP, they would need to set aside funds earned from premium to cover their potential claims. Regulation and tax laws would need to be changed so that these funds would not be counted in insurers’ surplus when evaluating lines of insurance for adequacy and rate setting, and used against them in pricing or bad faith claims.

There is also the issue of making participation by insurers voluntary. While this spares insurers from participating in a program that they deem problematic, it raises a question as to how many insurers would elect to participate. If few insurers participate, then some (perhaps many) businesses may find it difficult to obtain coverage. It is possible that less financially sound and responsible insurers may offer this coverage but then some of these insurers may fail to meet their claims obligations if a covered event occurs, unless these insurers bought sufficient reinsurance for this specific exposure. The reason that this could occur is that insurers concerned about their financial risk may be less likely to participate in the program because their retained losses could drive them into insolvency. On the other hand, insurers that are less concerned about their financial risk may be more willing to participate as they would be able to collect substantial premiums that would be siphoned off to their owners until a pandemic occurs, which could take many years.47 ,48

Finally, an issue arises with respect to how many firms would purchase this coverage, which also bumps against the public goods concern. As with terrorism insurance, a business would be able to choose whether it buys PRRP‐backed pandemic BI insurance or not and, presumably, the amount of coverage it would buy based on the coinsurance provision that it selects.49 ,50 Firms’ appetite for BI pandemic coverage is a matter of speculation, but we can consider how many firms currently purchase property BI coverage. One study found that only 40% of small firms have BI insurance (Blosfield, 2020). Another study looking at small and medium enterprises that suffered damages due to Superstorm Sandy found that 30% of the firms surveyed had BI insurance and only 11.9% had flood insurance (Collier et al., 2019).

Affordability is, of course, an issue, and spending money to protect against remote events is a hard decision that businesses must always weigh with cost‐benefit and cost of capital concerns (see, e.g., Ratliff, 2020). If covered businesses would only be required to pay a small premium relative to the coverage they would receive for pandemic losses, then the take‐up rate for this coverage could be high. If the demand for BI pandemic insurance is high, this could cause concerns among participating companies that would prefer to limit their risk.51

On the other hand, if PRIA is amended to provide for pre‐event financing through actuarially fair premiums charged by the Treasury, the cost of these premiums could discourage many businesses from purchasing the coverage. If the take‐up rate proved to be low, then the perceived benefits of the program would be reduced. A low take‐up rate would also raise issues concerning whether a business could still obtain aid (outside the program) from the federal government for its pandemic related BI losses if it had not purchased insurance. Allowing businesses to obtain such aid if they did not buy insurance would raise equity concerns as well as reduce firms’ incentives to purchase the coverage going forward.52 On the other hand, denying aid to businesses that did not purchase insurance could lead to a political firestorm.53

7.3.2 Potential surplus strains under PRIA

Noting the issues discussed above that might discourage many insurers from participating in the program, there are also issues as to how many insurers would be able to bear or feel comfortable with the potential amount of losses they would retain under the program. We performed an initial analysis of what participating insurers’ deductibles and pro rata shares of losses could be under different scenarios. More specifically, we estimated companies’ retained losses (their deductibles and pro rata shares of losses above their deductibles) in relation to their surplus. We used three different scenarios for industry losses: $250 billion, $500 billion, and $750 billion. To determine a company’s pro rata share of industry losses, we used its proportional share of the total direct premiums earned in covered lines for all companies included in our analysis. We employed two different panels of companies for our estimations: (1) the 100 largest writers of the covered lines under the Act; and (2) the 100 largest writers of commercial multi‐peril and fire insurance.54 Remember, PRIA requires of insurers first a 5% deductible, and then 5% of their direct premiums earned before federal coverage pays.

For the first panel of companies, in 2019, their combined direct premiums earned (DPE) were $132.7 billion and their combined surplus was $271.3 billion. Collectively, their deductibles as 5% of their DPE would be $6.6 billion. To illustrate our calculations, consider the case of an insurer with $800 million in direct premiums earned and $425 million in surplus, and total industry losses of $500 billion. The industry’s retained losses above their deductibles would be 5% of $500 billion minus $6.6 billion, or $24.7 billion. As this company’s share of all insurers’ DPE would be 0.6%, it would be responsible for 0.6% of $24.7 billion ($148 million) and its deductible would be $40 million. Hence, its share of the losses under its policies would be $188 million, which would represent 44.2% of its surplus.

We employed the same methodology for our analysis of the second panel of companies. In 2019, the combined DPE for these companies for commercial multiperil and property insurance was $99.4 billion and the sum of their deductibles was $4.9 billion.

The results of our analysis are shown in Tables 1 and 2. Each table shows the minimum, maximum, mean, and median values, as well as the quintile breaks for the three industry loss scenarios. As can be seen in Table 1 (the 100 largest writers of covered lines), the potential surplus strain becomes a greater problem for more companies the higher are industry losses. For example, the median value increases from 16.2% for industry losses of $250 billion to 37.6% for industry losses of $750 billion. The fourth quintile break increases from 74.2% for industry losses of $250 billion to 172.9% for industry losses of $750 billion. We see a similar pattern in Table 2 (the 100 largest writers of commercial multiperil and fire insurance), with the difference being that the surplus strain would be greater for more companies.

Table 1.
Statistics on companies’ retained losses as a percent of surplus 100 largest writers of PRIA covered Linesg’s

• Source: SNL Financial and authors’ calculations.

Table 2.
Statistics on companies’ retained losses as a percent of surplus 100 largest writers of commercial multiperil and fire insurance

• Source: SNL Financial and authors’ calculations.

Our analysis indicates that even with a relatively low percentage deductible and pro rata share, the potential losses for many companies relative to their surplus could be problematic. If companies’ deductibles and pro rata shares were increased, even more companies would face severe financial risk if they participated in the program. We note that the actual losses for any company could be significantly different (lower or higher) than what we have calculated. Clearly, participating companies would face a great deal of uncertainty with respect to what their losses would be and how they would affect their financial condition. This would likely drive the industry to a giant risk sharing pool, creating a new level of legal problems now in antitrust law.

Determining the provisions that would govern participating insurers’ share of losses under this program presents a conundrum. If private insurers’ share of losses would be small, this raises the question of what would be the point of having insurers bear any risk understanding that even low deductibles and pro rata shares may be problematic for many companies. If the industry’s share of losses would be large, then very few if any responsible companies would likely want to participate.

8 ALTERNATIVE FRAMEWORKS FOR A PANDEMIC INCOME REPLACEMENT PROGRAM

Here we consider alternative frameworks for government BI pandemic insurance that would differ from PRIA in several important ways. These programs differ in terms of their key features that warrant discussion. One key feature is whether participating insurers would bear any risk. Some frameworks have no risk bearing by insurers, as is the case with the federal flood insurance program. Other frameworks use a reinsurance mechanism in which insurers would bear some risk, similar to how federal crop insurance works. We believe that these alternative frameworks would avoid some of the problems with PRIA and offer more workable solutions, although they would still face challenges that would require difficult choices.

These key features are:

• 1.

  Is there risk sharing between insurers and the federal government?

• 2.

  Is insurer participation voluntary or mandatory?

• 3.

  Is a reinsurance mechanism employed?

• 4.

  What is covered (profit, expenses, payroll)?

• 5.

  Are payments made on an indemnity basis or based on a parametric formula?

• 6.

  Are program costs fully funded by premiums, with or without federal subsidies?

• 7.

  Are rates adjusted for risk or uniform?

• 8.

  Does the program contain a risk mitigation component?

Table 3 compares these programs as well as PRIA with respect to these key features. There are other technical aspects of these alternative schemes such as policy limits, and aggregate covered loss caps that warrant mention but which we do not discuss. We caution that these proposals are in various stages of development so they may currently lack specifics on certain program elements.

Table 3.
Comparison of key features of proposed programs

8.1 Frameworks with no risk sharing

8.1.1 Overview of NFIP

It is helpful here to briefly review how the NFIP works to consider its application to a government BI pandemic insurance program. In the NFIP, the federal government is responsible for paying all of the covered losses for the flood policies issued on its behalf. Under the Write‐Your‐Own (WYO) program, private insurers act as servicing carriers for the NFIP. These servicing carriers issue policies under NFIP rules and also adjust and pay claims. The NFIP reimburses its servicing carriers for the claims they pay as well as their expenses according to a predetermined formula. Additionally, the NFIP determines the coverages provided as well as sets the premium rates.

The NFIP has been beset by a number of problems, but the insurability of flood risk is not one of them.55 The issues with the NFIP that are contentious are the low take up rate for flood insurance, insufficient risk mitigation incentives, inadequate and subsidized premiums, and high program debt, among others.56 These issues could be resolved, albeit with some tough choices. However, reforming the NFIP has proven to be politically difficult.

8.1.2 Business continuity protection program (BCPP)

The BCPP proposed by the major insurance industry trade associations could be a feasible alternative to PRIA.57 The principal goal of the program is to help keep businesses afloat during pandemics. In this sense, the program would not work the same as property BI insurance, which among other things, reimburses a business for lost profits. Under the BCPP, businesses could purchase revenue replacement assistance for up to 80% of their payroll and other expenses. The desired level of protection could be for up to 3 months. Payroll coverage would exclude highly compensated employees, employee benefits, and operating expenses. A business could also purchase excess coverage for more than 80% of their expenses and payroll and for periods of longer than three months.

All insurers would be required to offer BCPP coverage to all policyholders who purchase covered lines of insurance, but businesses would not be required to buy it. Insurers’ participation in offering excess coverage would be voluntary. Insurers would retain 10% of excess losses with the federal government responsible for the remaining 90%. The program would be fully funded by premiums paid by insureds. However, the premiums charged would be not be adjusted for risk—businesses regardless of their industry or location, would pay the same percentage of their revenues. The program also would have a risk mitigation component in that insured businesses would need to attest to their compliance with CDC, OSHA, and other public health requirements.

An important element of the BCPP is its use of the North American Industry Classification System (NAICS) to determine whether a business would be eligible for reimbursement and how much aid it would receive. Businesses in classifications subject to full closure would receive 100% of the reimbursement specified in their policies. Businesses in classifications subject to partial closure would receive something less than 100% of the assistance specified in their policies. These percentages for partial assistance could be adjusted based on the specific industry and how much it is restricted by government orders.

8.2 Frameworks with risk sharing

8.2.1 Federal crop insurance overview

Another potential model for a federal backstop of pandemic BI losses is the federal crop insurance program (FCIP) administered by the Risk Management Agency (RMA) in the U.S. Department of Agriculture (USDA).58 Private insurance companies and their agents sell and service crop insurance policies to agricultural producers, backed by reinsurance provided by the Federal Crop Insurance Corporation (FCIC). The RMA sets the rates and terms of all policies that the FCIC reinsures. An insurer cannot issue a product reinsured by the FCIC but not approved by the RMA.

Participating insurers are required to assign each of their policies to one of two reinsurance funds specific to each state—the Assigned Risk Fund and the Commercial Fund. For policies allocated to the Assigned Risk Fund—which are typically higher risk policies—insurers retain a 20% share of the premiums collected and underwriting losses or gains. For the Commercial Fund, insurers must retain at least 35% of premiums and underwriting losses or gains; an insurer can choose to increase its pro rata share above 35% in 5% increments.

Like the NFIP, the federal crop insurance program has many flaws and is highly subsidized by taxpayers (see, e.g., Congressional Budget Office, 2017). As discussed by Klein and Krohm (2008), crop insurance products and rates are highly politicized. Agricultural interests lobby the Congress to keep rates low and approve some products that benefit certain producers but also are poorly designed and destined to be money losers.59

8.2.2 Business continuity coalition (BCC) proposal

The BCC coalition represents a broad range of business insurance policyholders. Its proposal would create a Federal Pandemic Insurance Corporation (FPIC) that would provide reinsurance for a Business Expense Insurance Program (BIP) and other covered lines, including event cancellation.60 TRIA‐eligible insurers would be required to offer BI pandemic coverage similar to a BIP policy directly or through an affiliate, or through an intermediary under a policy issued by a nonaffiliated company that could participate in a Business Expense Insurance (BEI) Pool. Similar to how federal crop insurance works, insurers writing this coverage directly or through an affiliate, would have two reinsurance options: (1) 5% insurer retention; and (2) 10% insurer retention. BEI pool subscribing insurers would purchase surplus notes to fund the pool. The pool would be provided 80% quota share plus stop‐loss reinsurance by the FPIC for its BI pandemic coverage.

If a triggering event occurs, BIP policyholders will receive a pre‐determined payment based on a percentage of their 3‐months fixed operating expenses (including payroll) for the previous year (similar to the approach used by the BCPP). Small and medium‐sized enterprises could purchase coverage for up to 80% of their 3‐month operating expenses, decreasing to 50% replacement for large firms. Applicants would be required to certify that they would only use payments for allowed purposes (e.g., retain employees and pay operating expenses). Participating insurers could voluntarily choose to offer additional and excess coverage products with parametric or indemnity triggers, backed by FPIC reinsurance similar to that for BIP policies.

This program would be fully funded with premiums charged based on federally subsidized rates. The BCC proposal is unclear as to whether premiums would be risk‐adjusted. Further, there is no risk mitigation component in this proposal as it is currently drafted.

8.2.3 Chubb proposal

The Chubb Group has been working on a plan that also would offer two levels of coverage, but these levels would only be distinguished by the size of the insureds. (McLaughlin, 2020). Part one is the BIP for small businesses (defined as businesses with 500 or fewer employees), which appears to be the same program provided in the BCC proposal. Hence, it would cover the same items (payroll and other operating expenses) with claims paid using a parametric formula. However, there would be no excess coverage provided that would be backstopped by the federal government.

Part two of the Chubb proposal creates a federal reinsurance facility, Pandemic Re (“Pan Re”), for businesses with more than 500 employees. As explained by McLaughlin (2020), Pan Re would accept risks at commercial terms at a risk‐adjusted price. Private insurers selling such coverage would write BI pandemic insurance policies at market terms and retain some portion of the risk, reinsuring the rest to Pan Re. For the first 5 years, private insurers would cede 95% of the risk to Pan Re, retaining 5%, with a maximum industry aggregate limit of $15 billion. Thereafter, private insurers would increase their retention gradually over time and decrease their cession to Pan Re from 95% to 90%, with a maximum aggregate limit of $30 billion by year 10 of the program. Insureds could elect 1–3 months of coverage, with a maximum limit of $50 million per policy. As with part one, there would be federal reinsurance for excess coverage. Claims for this part of the program would be settled on an indemnity basis, using insurers’ regular claims adjustment process.

Like the BCC program, this program would be fully funded with premiums charged based on federally subsidized rates. Unlike the BCC proposal, the Chubb program’s premiums would be risk‐adjusted. Additionally, there is no risk mitigation component in this proposal as it is currently drafted.

Further, Chubb’s program would provide for aggregate caps on the total amount of losses covered. Part one would have an aggregate risk limit of $750 billion. Part two would have an aggregate risk limit of $400 billion.

8.2.4 Zurich proposal

The third industry concept proposal, developed by the Zurich Insurance Group, also utilizes a reinsurance mechanism (Zurich, 2020). This program would create three federally backed insurance pools, which would differ in terms of the amount of risk retained by ceding insurers. In one pool, insurers would retain no risk and the federal share would be 100% of losses. In the second pool, the insurer share would be 5% and the federal share would be 95%. In the third pool, the insurer share would be 10% and the federal share would be 90%. Insurers would determine which risks to place on a policy‐by‐policy basis and the pool in which these risks would be placed. All property insurers would be required to offer this coverage.

Under this program, a firm could insure up to 80% of its operating expenses (including payroll) over 3 months capped at $20 million per month for employers with 500 or more employees. Eligible operating expenses would be the same as those covered under the BCC proposal with a parametric formula used to determine claims payments. There would be no aggregate caps in the Zurich program. Claims would be paid on an indemnity basis.

The Zurich program also would be fully funded through risk‐adjusted premiums based on an index that would vary by industry and region. Rates also would be federally subsidized. The rate‐on‐line (ROL) would be 2% for business with fewer than 500 employees and 3% for firms with 500 or more employees. Additionally, there would be a 0.5% preferred risk discount for qualifying mitigation programs. Further, insurers would provide risk mitigation consultation and services (e.g., resilience planning).

8.3 Discussion of key program features

Here we discuss the pros and cons of the key program features highlighted at the beginning of this section. We begin with issue of whether a program should operate like the NFIP or use a reinsurance mechanism like the FCIP. One advantage of the NFIP approach is that participating insurers would bear no risk so there would be no concerns with respect to surplus strains or the possibility that an insurer would be unable to cover its retention. Additionally, the administration of this kind of program could be relatively simple in the sense that insurers would only act as servicing carriers and there would not be the transaction costs associated with reinsurance arrangements. However, the administering agency would still need to supervise servicing carrier activities such as claims adjustment. This supervision would be more intensive if claims were paid on an indemnity basis.

Presumably, one primary attraction of a reinsurance mechanism is that it would entail participating insurers bearing some risk to the extent that this is feasible. This could be an efficient way of accessing capital through the reinsurance sector and the broader financial sector; participating insurers could use private reinsurance and various financial instruments such as pandemic cat bonds to further diversify their risk. Additionally, having participating insurers bear some risk would give them some “skin in the game,” increasing their incentives to carefully underwrite policies and adjust claims. On the down side, transactions costs could be higher with a reinsurance approach. There is also the question of how many insurers would choose to participate if they would be required to bear some risk. Further, the administering agency would need to vet and monitor the solvency of participating insurers as the RMA does. These issues could be partially addressed through a 100% federal reinsurance option.

The next issue is whether insurer participation should be voluntary or compulsory through a mandatory offer requirement. Some may believe that mandatory offer requirement is desirable, as businesses would not have to worry about finding an insurer willing to cover them. The concern with a mandatory offer requirement combined with some insurer retention of losses is the potential surplus strain this would impose on some companies. Some insurers would likely be better equipped to underwrite and bear some BI pandemic risk than others. Beyond the challenges this could create for some carriers, there are also associated responsibilities for the administering agency and state insurance regulators. Mandatory offer combined with a 100% federal reinsurance option could be an acceptable compromise in this regard.

Then there is the issue with the operating expenses covered and whether lost profits would be covered. All of the proposed programs would appear to cover most if not all operating expenses. Payroll coverage would be optional under PRIA, which is a concern for reasons we have articulated. Coverage of lost profits could be desirable to some businesses, but we note that even the BCC proposal would not cover profits. This coverage could be optional or not provided at all. Ensuring that businesses would use payroll coverage as intended could impose some additional burden on participating insurers.

The basis for paying claims—indemnity or parametric formula—presents important tradeoffs. One advantage of parametric coverage is that it would eliminate the moral hazard problem arising from firms’ discretion in finding ways to serve their customers under restrictions. Additionally, this approach would make claims adjustment relatively straight forward and reduce the likelihood of any disputes. Another attraction of parametric coverage cited by industry representatives is that it could expedite payments to insured businesses. The problem with this approach is that when a pandemic occurred, some businesses would likely receive more than their actual losses and other business would receive less than what they needed.61 Hence, the tradeoff between moral hazard and basis risk would need to be considered. Further, with parametric coverage, there is the challenge of ensuring that firms pay their workers if payroll is insured.

With indemnity coverage, an insured is only entitled to reimbursement for losses they actually incur and not more.62 Most insurance policies cover losses on an indemnity basis. In theory, this type of coverage could help ensure that businesses get what they in need terms of reimbursement. However, indemnity coverage engenders the problems that parametric coverage avoids. Specifically, there is the moral hazard concern. Additionally, the claims adjustment process becomes much more complicated and subject to disputes as well as bad faith claims. Modeling and pricing the risk also could be more challenging if losses are settled on an indemnity basis.

Fully funding with premiums appears not to be an issue in that only PRIA currently lacks this. There could be an issue with risk‐adjusted premiums. The sponsors of the BCPP plan present several arguments for uniform pricing. The essence of these arguments is as follows. A risk‐based model would require firms most likely to be vectors of viral transmission such as restaurants or hotels to pay higher rates. In turn, this would reduce the take‐up rate for these firms and make it more likely that they would resist closure orders. Hence, uniform pricing would better align public health interests with the incentives of those firms most exposed to the risk by increasing their take‐up rate.

These arguments warrant careful consideration but there are counter arguments. One concern with uniform pricing is that some firms may view this as inequitable. Another concern is that it could lead to greater adverse selection. This would create problems for insurers required to bear some risk but not allowed to charge risk‐adjusted rates. Additionally, uniform pricing could preclude incorporating risk mitigation incentives in pricing. Another way to address the concerns with risk‐adjusted pricing could be federal premium subsidies; we leave a discussion of subsidies to Section 9.

Finally, it would seem desirable to include a risk mitigation element in any program for reasons that we have articulated. The issue here is how such an element would be designed and enforced. To some extent, this is a technical matter but there could be concerns if risk mitigation measures are viewed as too onerous or impractical for some firms. The experience of the NFIP and the FCIP in this regard could be of some value, noting that mitigating BI pandemic risk is a different kind of animal. Moreover, risk mitigation for pandemics can be challenging as the characteristics of viruses can vary and the science continues to evolve.

8.4 Discussion of other program provisions

As we have discussed above, an important issue for any program covering BI losses from a pandemic is exactly what losses would be covered. This is more of an issue with indemnity coverage than parametric coverage. For indemnity policies, a relatively narrow definition of covered losses would make claims adjustment less difficult and would limit the scope of losses for which a program would be responsible. This would also make estimation of the risk and pricing less challenging. For example, insured BI pandemic losses could be confined to those arising from a complete shutdown of business due to government orders and/or the need to decontaminate the premises of a building.63 If the definition of covered losses is broader than this, then the problems discussed above become an issue.

Even with a narrow definition of covered losses or with a parametric approach for determining the payments to insured businesses, developing accurate premium rates for this program would be challenging for the reasons discussed above. It is also likely that rate setting would be politicized as we have seen with federal flood and crop insurance. Program administrators would likely be pressured by the Congress to charge lower rates than what would be adequate, noting that what would constitute adequate rates would be difficult to determine and contentious.

Assuming that its rates were adequate, it could take a number of years for a program to accumulate sufficient reserves to pay its claims, with the tax and regulatory problems we discussed earlier. Clearly, this would not be a problem if the next pandemic occurs many years from now. It would likely be a problem if the next pandemic occurs within the next 5–10 years. Under this scenario, the program would likely need to borrow from the Treasury (as the NFIP has done) to be able to meet all of its claims obligations. Loans from the Treasury could be repaid from future premiums (possibly with the help of a premium surcharge) but if a second pandemic occurred within a relatively short time‐period, the program could be in debt until many years had passed without a pandemic.

One way to at least partially address this problem would be having a BI pandemic insurance program use financial instruments (e.g., pandemic bonds) to help cover its costs. The cost of such instruments could be funded through the premiums charged by the program. Since 2018, the NFIP has used catastrophe bonds as well as reinsurance (beginning in 2017) to help fund its claims payments.

Further, as with the PRIA, there is the question of how many businesses would buy BI pandemic insurance from any of these alternative programs. We would expect that the likelihood that a business would purchase the coverage would depend on its owner’s perception of their risk and the cost of coverage, among other factors.64 One of these other factors would be business owner’s expectation that they could obtain government aid for their pandemic related losses in lieu of insurance.65 As with PRIA, government aid for businesses that did not buy BI pandemic insurance raises equity issues and could affect the take up rate for this coverage going forward.66 One way to achieve to achieve higher take up rates could be some form of a subsidized rate structure.

There are a number of other program features that are more technical in nature, and not a matter of policy per se. We mention only one of these features here—is the specification of program triggers. Ideally, program triggers would be designed so that reimbursements would be made only if there was a government order that would require businesses to close or restrict their activities.

8.5 The role of and need for government subsidies

This leads us to consider another fundamental issue associated with a government program for BI pandemic insurance. Should such a program intentionally charge subsidized rates and if so, how substantial should the subsidies be and who should pay for them? We believe that any program should have some level of prefunding through premiums charged to its insureds, possibly supplemented by postevent premium surcharges to help cover funding gaps. The primary rationale for subsidized rates would be the substantial positive externalities that would likely stem from such a program. Other arguments for subsidized rates are that they would reduce the cost of the program for its insureds and increase the number of firms that would buy this coverage. Additionally, subsidies could be targeted to provide the greatest assistance to small firms with limited resources and/or firms for which virus transmission is a big problem (e.g., bars and restaurants).

Subsidies are a form of support given to producers and/or consumers of a good or a service that helps to reduce its cost. Governments increase the use of certain goods or services through subsidies. Subsidies are typically provided to promote certain economic policies or social goods that are deemed to be in the public interest. They can be used to correct market failures or offset externalities.

Subsidies can be divided into four types: (1) production subsidies; (2) consumption subsidies; (3) export subsidies; and (4) employment subsidies. A certain type of insurance could be supported through subsidies given to insurers, which would be a form of a production subsidy. Alternatively, subsidies could be given to individuals or firms to purchase insurance, which would be a form of consumption subsidy. The mechanics of these two approaches would differ but the ultimate effect would be to lower the cost of insurance and increase its use by individuals or firms. There are various ways that the government could provides subsidies directly to consumers of insurance, such as vouchers or rates that are below the actuarial cost of the coverage provided. Tax credits for buying this insurance might also be used, but such credits could obscure the actual costs of these subsidies. Even with subsidies, a business might still conclude that the cost of coverage is too high for a remote risk for which the government might still intervene to forestall economic collapse.

It is helpful here to consider a study sponsored by the National Academies of Sciences (NAS) that examined various approaches to providing assistance to some individuals to help them afford risk‐based premiums under the NFIP using targeted assistance rather than generally subsidized rates (NAS, 2015). This study addresses the question of how to enable more households to afford to buy flood insurance, particularly households that pay high premiums relative to their income. One contribution of the NAS study is its consideration of the methods that could be employed to determine which households would receive subsidies and the amounts of the subsidies. The study considered but did not recommend who would pay for these subsidies.

In our opinion, if premiums for BI pandemic insurance were subsidized, it would be preferable to fund them through general revenues and borrowing rather than through inter‐program subsidies. Inter‐program cross subsidies between insureds in an insurance program are problematic when buying the insurance is voluntary as they lead to greater adverse selection.67 There are also equity issues associated with inter‐program subsidies. Further, if one of the primary reasons for having subsidies is the positive externalities stemming from BI pandemic insurance, then this argues for funding them through general revenues and borrowing. It would be preferable to employ dedicated tax streams for this purpose to create transparency as to their purposes and benefits.

In essence, the proposition here is that it would be better for insureds to pay something for the coverage from which they directly benefit with taxpayers also helping to pay for their coverage given the benefits for the economy. We note that taxpayers have fully funded the economic aid that firms, workers, and taxpayers have received due to COVID‐19. For future pandemics, if subsidized rates helped to increase the take‐up rate for government BI pandemic insurance, this should reduce the need for and cost of taxpayer‐funded economic aid to firms, workers, and others.68

In sum, premium subsidies raise a number of questions with respect to their magnitude, allocation, and financing, among others. Still, it would be preferable that any subsidies be explicit and carefully designed, in contrast to what is the case now for flood and crop insurance. There are no obvious answers to these questions but their consideration should be informed by analysis of the benefits that would be received by the program’s insureds relative to those that would be received by others.

9 CONCLUDING THOUGHTS

To summarize, if the limited goal is to provide businesses one more insurance policy to buy that will compensate them for a different peril―pandemic‐caused income losses, without regard to worker income protection―the solution may be a public‐private insurance program that is backstopped by the federal government and administered by insurers. If the core goal of a pandemic BI program is to support economic activity, and even more importantly prevent economic collapse and the destitution of workers and small business owners, then any program should be oriented towards a public finance solution that could include some funding by its insureds, even if administered by private insurers. PRIA is unlikely to protect the economy and workers’ incomes, due to its poor design, such as not requiring payroll protection. It also fails to address the free rider problem of uninsured businesses benefiting from the coverage provided to insured firms. The expectation of receiving government assistance without insurance would likely decrease the demand for this coverage, which would further exacerbate the free rider problem. This means that the government will again have to provide fundamental and universal economic relief in another pandemic funded by taxpayers.

We have outlined what we believe to be desirable features of a BI pandemic insurance program, understanding that such features will not fully resolve the problems it would face. These features include pre‐event financing through insurance premiums, carefully constructed program triggers and coverage terms, coverage of both lost income and payroll, the determination of rates and coverage provisions by the administering agency, the preemption of state market regulation, and the inclusion of risk mitigation component. Additionally, the use of taxpayer‐financed premium subsidies should be strongly considered, as the take up rate for this coverage is likely to be low without subsidies.

There are other features of these programs that raise important policy issues. These features include whether insurers should bear any risk through a reinsurance mechanism, making insurer participation voluntary or mandatory, indemnity versus parametric claims payments, and uniform versus risk‐adjusted rates.

Despite these challenges and issues, advocates of such a program might argue that it would still have several advantages relative to the status quo. These advantages might include reducing the uncertainty that many firms face with respect to funding their pandemic related losses, providing for greater economic stability, and creating a mechanism for pre‐event financing to decrease the need for taxpayer funding, among others.

While we agree that planning for rare and extreme losses such as pandemics is desirable, both as to public health responses and their economic and fiscal impacts, we have several concerns about the problems that would be faced by any government insurance program for BI pandemic risk. Any program would be challenging to administer, subject to political interference, raise equity issues, and could obligate the government to make payments to businesses that would not be adequately funded by its premium revenues and require loans from the Treasury, among others. It is understandable that different stakeholders would weigh these pro and cons differently.

Hence, the Congress faces difficult choices in considering whether there should be a government program for BI pandemic losses and how to structure such a program. Any program will be far from perfect. In this paper, we have articulated what we believe the features of the best possible program should be. If the Congress is committed to creating a BI pandemic insurance program, it should consider the issues and tradeoffs we have outlined. Further, while some may feel a sense of urgency in creating a program, we advise that the Congress take whatever time is necessary to develop the best possible program. Its deliberations should be informed by thorough analyses, including actuarial analysis of a program’s financing and pricing and study of businesses’ appetite for purchasing this coverage based on how it would be structured and what it would cost.69

REFERENCES

• Abadi, M., Cooper, H., & Techman‐Fullard, M. (2020). How the coronavirus compares to SARS, Swing Flu, Zika, and other epidemics, Business Insider. https://www.businessinsider.com/coronavirus-compared-to-sars-swine-flu-mers-zika-2020-3?op=1
• Anomaly, J. (2011). Public health and public goods. Public Health Ethics, 4(3), 251– 259.
• Anomaly, J. (2015). Public goods and government action. Politics, Philosophy & Economics, 14(2), 109– 128.
• Auerbach, A. J., & Feldstein, M. F. (1987). Handbook of public economics. North‐Holland.
• Bisco, J. M., Fier, S. J., & Pooser, D. (2020). Business interruption insurance and COVID‐19: Coverage and issues and public policy implications. Journal of Insurance Regulation, 39(5), 1– 39.
• Blosfield, E. (2020). The debate between business owners, insurers over business interruption claims continues. Insurance Journal, June, 15, 2020. https://www.insurancejournal.com/magazines/mag-features/2020/06/15/572050.htm
• Borghesi, D. A. (1993). Business interruption insurance—A business perspective. Nova law review, 17, 1147– 1165.
• Born, P. H., Bradley Karl, J., & Klein, R. W. (2018). Does state rate regulation matter? In An Assessment of the Effects of More Stringent Regulation on Insurers’ Performance in Homeowners Insurance, presented at the American Risk and Insurance Association Annual Meeting, Chicago, IL, August 2018.
• Born, P. H., & Klein, R. W. (2019). Arguments on a hybrid privatization of the U.S. flood insurance program: A debate driven by issues of sustainability. Review of Business, 39(2), 36– 66.
• Browne, M. J., & Hoyt, R. E. (2000). The demand for flood insurance: Empirical evidence. Journal of Risk and Uncertainty, 20(3), 291– 306.
• Collier, B. L., Andrew, F., & Haughwout, A. F. (2019). Firms’ management of infrequent shocks. Journal of Money, Credit and Banking, 52(6), 1329– 1359.
• Congressional Budget Office.(2017). Options to reduce the budgetary costs of the Federal Crop Insurance Program, December, Washington, DC.
• Congressional Research Service. (2019). Terrorism risk insurance: Overview and issue analysis for the 116th congress (report R45707), December 27, 2019, Washington, DC.
• Congressional Research Service. (2020). Federal authority to lift or modify state and local COVID‐10 “Stay‐at‐Home” orders: Frequently asked questions, May 15, 2020, Washington, DC.
• CoreLogic. (2018). Financial implications of the HayWired scenario, Special Report, Irvine, CA. https://www.corelogic.com/downloadable-docs/corelogic_haywired-scenario_r5.pdf
• Dees, R. H. (2018). Public health and normative public goods. Public Health Ethics, 11(1), 20– 26.
• Department of Justice. (2020). Department of justice enforcement actions related to COVID‐19, Office of the. Deputy Attorney General, March 24, 2020, Washington, DC. https://www.politico.com/f/?id=00000171-128a-d911-aff1-becb9b530000
• Doyle, J. (2020). Insuring against a pandemic: challenges for policyholders and insurers, testimony before the subcommittee on housing, community development and insurance. Committee on Financial Services, U.S. House of Representatives, November 19, 2020. https://financialservices.house.gov/uploadedfiles/hhrg-116-ba04-wstate-doylej-20201119.pdf
• Economist. (2011). The birth of the black death: Medical archeology, October 12, 2011.
• Economist. (2014). Smallpox, Whoops, July 12, 2014.
• Economist. (2020). A peculiarity of spanish flu may shed light on COVID‐19, April 25, 2020.
• Goldsmith, C. (2018). Pandemic: How climate, the environment, and superbugs increase the risk. Lerner Publishing Group.
• Grace, M. F., & Klein, R. W. (2009). A perfect storm: Hurricanes, insurance markets and regulation. Risk Management and Insurance Review, 12(1), 81– 124.
• Hartwig, R., Niehaus, G., & Qiu, J. (2020). Insurance for economic losses caused by pandemic. Geneva Risk and Insurance Review, 45, 134– 170.
• He, Q., & Faure, M. (2018). Regulation by catastrophe insurance: A comparative study. Connecticut Insurance Law Journal, 24(2), 189– 244.
• Herndon, E. L. (2007). The mold exclusion, Property Casualty 360, September 5, 2007.
• Holcombe, R. (2000). Public good theory and public policy. Journal of Value Inquiry, 34, 273– 286.
• Huynh, A., Bruhn, A., & Browne, B. (2013). A review of catastrophic risks for life insurers. Risk Management and Insurance Review, 16(2), 233– 266.
• Insurance Journal Journal. (2020). Wimbledon shows how pandemic insurance could become vital for sports, other events. https://www.insurancejournal.com/news/international/2020/04/13/564598.htm
• Interlandi, J. (2020). Why we’re losing the battle with Covid‐19. New York Times Magazine, July 14, 2020. https://www.nytimes.com/2020/07/14/magazine/covid-19-public-health-texas.html
• IRMI. (2020). Commercial property insurance. International Risk Management Institute, Dallas, TX.
• Kalhoff, A. (2011). Why democracy needs public goods. Lexington Books.
• Kaul, I. (2001). Global public goods: What role for civil society? Nonprofit and Voluntary Sector Quarterly, 30(3), 588– 602.
• Klein, R. W., & Krohm, G. (2008). Federal crop insurance: The need for reform. Journal of Insurance Regulation, 26(3), 23– 63.
• Klein, R. W., & Weston, H. (2020a). Feasibility questions about government‐sponsored insurance for business interruption losses from pandemics. Journal of Insurance Regulation, 39(7), 1– 19.
• Klein, R. W., & Weston, H. (2020b). Government insurance for business interruption losses from pandemics: An evaluation of its feasibility and possible frameworks. Working paper, Georgia State University.
• Kousky, C., Michel‐Kerjan, E., & Raschky, P. (2018). Does federal disaster assistance crowd out flood insurance? Journal of Environmental Economics and Management, 87, 150– 164.
• Kuhlmann, B. (2020). Insuring against a pandemic: challenges for policyholders and insurers, testimony before the subcommittee on housing, community development and insurance, Committee on Financial Services, U.S. House of Representatives, November 19, 2020. https://financialservices.house.gov/uploadedfiles/hhrg-116-ba04-wstate-kuhlmannb-20201119.pdf
• Kunreuther, H. C., & Michel‐Kerjan, E. O. (2009). At war with the weather: managing large‐scale risks in a new era of catastrophes. MIT Press.
• Kunreuther, H. C., & Michel‐Kerjan, E. O. (2019). The future of risk management. University of Pennsylvania Press.
• Lerner, M. (2020). Marsh gears up to expand capacity for pandemic insurance product. Business Insurance. April, 1, 2020. https://www.businessinsurance.com/article/20200401/NEWS06/912333820/Marsh-looks-to-expand-capacity-for-PathogenRX-pandemic-insurance#
• Lewis, R. P., Nicholas, M., & Insua, N. M. (2020). Business income insurance disputes ( 2nd ed). Wolters Kluwer.
• Liu, E. C. (2008). Would an influenza pandemic qualify as a major disaster under the Stafford act? Congressional Research Service, October 20, 2008, Washington, DC.
• Makridis, C. A., & Hartley, J. S. (2020). The cost of Covid‐19: A rough estimate of the 2020 US GDP impact, policy brief. Mercatus Center, George Mason University.
• McKibbin, W., & Fernando, R. (2020). The global economic impacts of COVID‐19: Seven scenarios. Brookings Institution. https://www.brookings.edu/research/the-global-macroeconomic-impacts-of-covid-19-seven-scenarios/
• McLaughlin, M. M. (2020). Insuring against a pandemic: challenges for policyholders and insurers, testimony before the subcommittee on housing, community development and insurance. Committee on Financial Services, U.S. House of Representatives, November 19, 2020. https://financialservices.house.gov/uploadedfiles/hhrg-116-ba04-wstate-mclaughlinm-20201119.pdf
• Moss, D. A. (2002). When all else fails: Government as the ultimate risk manager. Harvard University Press.
• National Academies of Sciences. (2015). Affordability of national flood insurance program premiums. National Academies Press.
• Oakland, W. H. (1987). Theory of public goods. In A. J. Auerbach, & M. S. Feldstein (Eds.), Handbook of public economics (Vol. 2, pp. 485– 535). North‐Holland.
• Oppenheimer, C. (2011). Eruptions that shook the world. Cambridge University Press.
• Priest, G. (1996). The government, the market, and the problem of catastrophic loss. Journal of Risk and Uncertainty, 12, 219– 237.
• Ratliff, E. (2020). We can protect the economy from pandemics. Why didn’t we? Wired, June 16, 2020. https://www.wired.com/story/nathan-wolfe-global-economic-fallout-pandemic-insurance/
• Redja, G. E., & McNamara, M. J. (2016). Principles of risk management and insurance. Pearson.
• Richter, A., & Wilson, T. (2020). Covid‐19: Implications for insurer risk management and the insurability of pandemic risk. Geneva Risk and Insurance Review, 45, 171– 199.
• Samuelson, P. A. (1954). The pure theory of public expenditure. Review of Economics and Statistics, 36(4), 387– 389.
• Saunders‐Hastings, P. R., & Krewski, D. (2016). Reviewing the history of pandemics influenza: Understanding patterns of emergence and transmission. Pathogens, 5(4), 66.
• Skipper, H., & Kwon, W. J. (2007). Risk management and insurance: perspectives in a global economy. Wiley‐Blackwell.
• Smith, D. (2017). The palgrave handbook of unconventional risk transfer. Palgrave Macmillan. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7121215/?report=reader#Sec2title
• Swendiman, K. S., & Elsea, J. K. (2007). Federal and state quarantine and isolation authority. Congressional Research Service, RL33201, Washington, DC.
• Swendiman, K. S., & Jones, N. L. (2009). The 2009 influenza pandemic: selected legal issues. Congressional Research Service, R40560, Washington, DC.
• Tesltra (2020). Business continuity, flexible working and adaptive infrastructure: five actions for when the economy reopens following COVID‐19. White paper. https://forms.telstraglobal.com/BusinessContinuityReport
• Toole, J. (2007). Potential impact of pandemic influenza on the U.S. life insurance industry. Research report, Society of Actuaries, May 2007.
• Toole, J. (2010). Spreadsheet documentation for health company pandemic modeling tool. Research report, Society of Actuaries, June 2010.
• Wessel, D. (2011). Tallying the toll on the economy from 9/11. Wall Street Journal, September 1, 2011. https://www.wsj.com/articles/SB10001424053111903895904576542352036340880
• Whitford, E. (2020). Are Any Colleges Insured Against Coronavirus Fallout? Inside Higher Ed, March, 9, 2020. https://www.insidehighered.com/news/2020/03/09/insurance-coverage-scarce-coronavirus-threatens-college-finances
• World Health Organization. (2009). World now at the start of 2009 influenza pandemic, June 11, 2009. https://www.who.int/mediacentre/news/statements/2009/h1n1_pandemic_phase6_20090611/en/
• World Health Organization. (2010). H1N1 in post‐pandemic period, August, 10, 2010. https://www.who.int/mediacentre/news/statements/2010/h1n1_vpc_20100810/en/
• Ziady, H. (2020). It’s too late for this pandemic. But everyone wants it against the next One, CNN Business, March, 19, 2020. https://www.cnn.com/2020/03/19/business/pandemic-insurance-coronavirus/index.html
• Zurich North America. (2020). Zurich’s draft concept for facilitating pandemic protection, December 7, 2020. https://www.zurichna.com/-/media/project/zwp/zna/docs/kh/pandemic/zurich-pandemic-risk-concept-powerpoint-final.pdf?la=en&hash=3E3F7670DC6B7E0523EA1D700082D7DD