Improving the Availability and Affordability of Pandemic Risk Insurance (RAND RRA1223-1)

Summary

A 2021 RAND Corporation report by Lloyd Dixon and Jamie Morikawa, RAND publication RRA1223-1. It describes the distinguishing features of the leading proposals for pandemic risk insurance and develops a quantitative model of their potential consequences. The contents list five proposals: the Business Continuity Protection Program, the Chubb Pandemic Business Interruption Program, the Zurich Preliminary Pandemic Proposal, the Pandemic Risk Insurance Act and the Business Continuity Coalition Program. The report states that programs are evaluated by the proportion of revenue decline replaced, efficiency, affordability, risk borne by commercial insurers, expected annual government net outlays, and subsidy provided to policyholders.

Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used

Issuer: Source documents
Document type: Improving The Availability And Affordability Of Pandemic Risk Insurance Projected Performa
Case: Improving The Availability And Affordability Of Pandemic Risk Insurance Projected Performance Of Proposed Programs

Cite this document

“Improving the Availability and Affordability of Pandemic Risk Insurance (RAND RRA1223-1).” Original at doi.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/improving-the-availability-and-affordability-of-pandemic-risk-insurance-projected-performance-of-proposed-programs/, SHA-256 3aeaec788dc00bbb27c13b07762643afd88d495d5863d7297ff8ae9dea6b7994.

Full text

C O R P O R AT I O N

LLOYD DIXON, JAMIE MORIKAWA

Improving the
Availability and
Affordability of
Pandemic Risk
Insurance
Projected Performance of Proposed Programs
For more information on this publication, visit www.rand.org/t/RRA1223-1.
About RAND
The RAND Corporation is a research organization that develops solutions to public policy challenges
to help make communities throughout the world safer and more secure, healthier and more prosperous.
RAND is nonprofit, nonpartisan, and committed to the public interest. To learn more about RAND, visit
www.rand.org.
Research Integrity
Our mission to help improve policy and decisionmaking through research and analysis is enabled through
our core values of quality and objectivity and our unwavering commitment to the highest level of integrity
and ethical behavior. To help ensure our research and analysis are rigorous, objective, and nonpartisan,
we subject our research publications to a robust and exacting quality-assurance process; avoid both
the appearance and reality of financial and other conflicts of interest through staff training, project
screening, and a policy of mandatory disclosure; and pursue transparency in our research engagements
through our commitment to the open publication of our research findings and recommendations,
disclosure of the source of funding of published research, and policies to ensure intellectual independence.
For more information, visit www.rand.org/about/principles.
RAND’s publications do not necessarily reflect the opinions of its research clients and sponsors.
Published by the RAND Corporation, Santa Monica, Calif.
© 2021 RAND Corporation
is a registered trademark.
Library of Congress Cataloging-in-Publication Data is available for this publication.
ISBN: 978-1-9774-0707-8
Cover: Design by Rick Penn-Kraus. Photos: anyaberkut/Getty Images/iStockphoto (man with umbrella);
BlackJack3D/Getty Images/iStockphoto (coronavirus).
Limited Print and Electronic Distribution Rights
This document and trademark(s) contained herein are protected by law. This representation of RAND
intellectual property is provided for noncommercial use only. Unauthorized posting of this publication
online is prohibited. Permission is given to duplicate this document for personal use only, as long as it is
unaltered and complete. Permission is required from RAND to reproduce, or reuse in another form, any
of its research documents for commercial use. For information on reprint and linking permissions, please
visit www.rand.org/pubs/permissions.
About This Report

Given the closures and other restrictions on business activity during the COVID-19
pandemic, stakeholders have developed several proposals for an insurance-based pro-
gram to provide businesses with the resources needed to maintain payroll and benefits
and to cover the ongoing operating expenses necessary to survive. This report describes
the distinguishing features of the most-visible proposals and develops a quantitative
model that shows their potential consequences. Proposed programs are evaluated in
terms of the proportion of revenue decline replaced (efficacy), efficiency, affordability,
the risk borne by commercial insurers, expected annual government net outlays, and
the amount of subsidy provided to policyholders. The analysis should be useful to
members of Congress, congressional staff, insurers, and policyholders as they consider
whether to support the creation of a pandemic risk insurance program and how such a
program can be best designed.

The RAND Kenneth R. Feinberg Center for Catastrophic Risk
Management and Compensation

The Feinberg Center seeks to identify and promote laws, programs, and institutions
that reduce the adverse social and economic effects of natural and manmade catas-
trophes by improving incentives to reduce future losses; providing just compensation
to those suffering losses while appropriately allocating liability to responsible parties;
helping affected individuals, businesses, and communities to recover quickly; and
avoiding unnecessary legal, administrative, and other transaction costs.
Questions or comments about this report should be sent to the project leader,
Lloyd Dixon (dixon@rand.org). For more information about the Feinberg Center, see
www.rand.org/ccrmc or contact the director at ccrmc@rand.org.

Funding

Funding for this research was provided by the generous contributions of the RAND
Kenneth R. Feinberg Center for Catastrophic Risk Management and Compensa-

iii
iv Improving the Availability and Affordability of Pandemic Risk Insurance

tion Advisory Board and, in part, by a contribution to the center by the American
Property Casualty Insurance Association.

Acknowledgments

This report was improved by comments, information, and expertise provided by many
people. We would like to thank staff at the American Property Casualty Insurance
Association for directing us to useful data sources, discussing assumptions they have
used in projecting the impacts of their proposed Business Continuity Protection Pro-
gram, and providing insurance industry financial statistics. The Reinsurance Asso-
ciation of America provided helpful information on modeling approaches and data.
Chubb provided insight into the structure and modeling of its proposed program.
Charles Landgraf at Arnold and Porter helped us better understand and describe the
Business Continuity Coalition proposal. Davide Priebe at Guy Carpenter provided
valuable insights into the strengths and weakness of different approaches.
Constructive peer reviews of the draft report were provided by Michael Dworsky
at the RAND Corporation and Jason Schupp at the Center for Better Insurance. Very
helpful comments were also provided by AIG, Liberty Mutual, State Farm, and Trav-
elers. Paul Steinberg at RAND improved the flow and clarity of the exposition, and
Rebecca Fowler skillfully edited the document.
Contents

About This Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Figures and Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii

CHAPTER ONE
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Research Questions and Study Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Report Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

CHAPTER TWO
Overview of Proposed Programs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Proposal 1: Business Continuity Protection Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Proposal 2: Chubb Pandemic Business Interruption Program. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Proposal 3: Zurich Preliminary Pandemic Proposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Proposal 4: Pandemic Risk Insurance Act . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Proposal 5: Business Continuity Coalition Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

CHAPTER THREE
Modeling Approach and Base-Case Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Modeling Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Base-Case Parameter Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

CHAPTER FOUR
Projected Program Performance in the Base Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Affordability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Efficacy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Efficiency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Risk Borne by Commercial Insurers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

v
vi Improving the Availability and Affordability of Pandemic Risk Insurance

Expected Annual Government Outlays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Policyholder Subsidy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

CHAPTER FIVE
Sensitivity of Projections to Changes in Parameter Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Sensitivity of Results to Changes in Return Period Used by Insurers in Pricing . . . . . . . . . . . . . 60
Sensitivity of Results to Changes in Other Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Sensitivity of Results to Change in Assumption on Whether Government Pays Loss
Adjustment Expenses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

CHAPTER SIX
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

APPENDIXES
A. Overview of Business Interruption Policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
B. Insurance Payments in a Partial Shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
C. Pandemic Risk Insurance Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
D. General and Selling Expense and Loss Adjustment Expense Ratios by Line of
Business. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Figures and Tables

Figures
S.1. Annual Premium in Base-Case Projection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii
S.2. Take-Up Rate in Base-Case Projection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
S.3. Expected Annual Government Net Outlays at Base-Case Parameter Values . . . xiv
S.4. Annual Premium When Insurers Use 30-Year Return Period in Setting
Prices and Other Parameters Remain at Base-Case Values . . . . . . . . . . . . . . . . . . . . . . . . . xv
3.1. Model Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.1. Program Efficiency for Small and Medium-Sized Firms at Base-Case
Parameter Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.2. Program Efficiency for Large Firms at Base-Case Parameter Values . . . . . . . . . . . . . 50
4.3. Expected Annual Government Net Outlays per Participating Firm for
Small and Medium-Sized Firms at Base-Case Parameter Values . . . . . . . . . . . . . . . . . . 52
4.4. Expected Annual Government Net Outlays for Small and Medium-Sized
Firms at Base-Case Parameter Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.5. Expected Annual Government Net Outlays per Participating Firm for
Large Firms at Base-Case Parameter Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.6. Expected Annual Government Net Outlays for Large Firms at Base-Case
Parameter Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.7. Expected Annual Government Net Outlays for All Firms at Base-Case
Parameter Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.8. Policyholder Subsidy for Small Firms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.9. Policyholder Subsidy for Large Firms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.1. Premium for Small and Medium-Sized Firms as Pandemic Return Period
Used by Insurers Varies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.2. Premium for Large Firms as Pandemic Return Period Used by Insurers
Varies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.3. Take-Up Rate as Return Period Used by Insurers Varies . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.4. Capital Help by Insurers as Return Period Used by Insurers Varies . . . . . . . . . . . . . . . 63
5.5. Expected Annual Government Net Outlays for All Firms as the Return
Period Used by Insurers Varies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

vii
viii Improving the Availability and Affordability of Pandemic Risk Insurance

Tables
S.1. Ranking of Modeled Programs by Program Metric at Base-Case Parameter
Values. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
2.1. Features of Proposed Pandemic Risk Insurance Programs . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1. Model Inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.2. Model Outputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.3. Measures Used to Evaluate Program Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.4. Distribution of Firms in the United States with Employees by Firm Size in
2017 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.5. Base-Case Parameter Values for Modeled Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.6. Major Contagious Disease Incidents Since 1918 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.1. Program Performance at Base-Case Parameter Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2. Ranking of Modeled Programs by Program Metric at Base-Case Parameter
Values. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.1. Range of Parameter Values Used in Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.2. Sensitivity of Program Outcomes to Changes in Parameter Values . . . . . . . . . . . . . . . 65
5.3. Sensitivity of Program Outcomes to Change in Assumption on Whether
Government Pays Loss Adjustment Expenses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
C.1. Parameter Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
D.1. General and Selling Expense Ratio by Line of Business Between 2015 and
2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
D.2. Loss Adjustment Expense Ratio by Line of Business Between 2015 and 2019 . . . 83
Summary

The COVID-19 pandemic led to a substantial drop in U.S. economic activity in 2020.
Government stay-at-home orders, restrictions on business activities, and consumer con-
cerns about potential exposure to the virus caused unemployment rates to rise and put
many businesses in financial jeopardy. Businesses often purchase business interruption
coverage that covers loss of revenue due to fires and other perils, but policies typically
require that the interruption be due to physical damage at the insured property, and,
in recent years, the policies have excluded loss due to contagious diseases. As a result,
insurers have held that in most cases they are not obligated to cover the enormous busi-
ness interruption losses caused by COVID-19.
Now, insurers, insurance industry trade groups, policyholder groups, and Con-
gress have developed proposals that would enhance the role that insurance plays in
providing businesses with the resources they need to maintain payroll and benefits and
cover ongoing operating expenses during an outbreak. The most visible proposals are
the following:

• The Business Continuity Protection Program (BCPP)—proposed by the
American Property Casualty Insurance Association, the National Association of
Mutual Insurance Companies, and the Independent Insurance Agents and Bro-
kers of America—provides payroll, benefits, and expense support to the private
sector in the event of a declared public health emergency.
• Chubb’s Pandemic Business Interruption Program has two parts: the Pan-
demic Business Expense Insurance Program, which is open to small and
medium-sized firms (firms with 500 or fewer employees), and the Pandemic
Business Interruption Reinsurance program (Pandemic Re), which is open to
firms with more than 500 employees.
• The Zurich Preliminary Pandemic Proposal proposes a flexible program that
allows insurers to choose how much risk to bear.
• The Pandemic Risk Insurance Act of 2020 (PRIA), introduced by Representa-
tive Carolyn Maloney in May 2020, seeks to establish a Pandemic Risk Reinsur-
ance Program modeled on the federal Terrorism Risk Insurance Program.
• The Business Continuity Coalition (BCC) proposal would make coverage for
pandemic-related losses available in a broader range of insurance policies than

ix
x Improving the Availability and Affordability of Pandemic Risk Insurance

other proposals. The BCC is composed of insurance policyholders from across
the U.S. economy.

Objectives and Approach

This report seeks to inform the discussion of the benefits and drawbacks of the five
proposals for expanding insurance for revenue declines due to pandemic-induced busi-
ness closures or restrictions on economic activity by addressing the following research
questions:

• What are the distinguishing features of the most visible proposals for a pandemic
risk insurance program?
• How will these programs perform in terms of specific key dimensions: efficacy,
efficiency, affordability, risk borne by commercial insurers, government net out-
lays, and the extent of policyholder subsidies?

To address these questions, we first identified the most-visible proposals and
compared their key features. We then developed a high-level, quantitative model to
project the key outcomes for each of the proposals. The model is based on a number
of underlying parameters, and we reviewed economic and insurance literature and
data to develop reasonable values for each. We also consulted with experts about what
values would be reasonable. The model uses several simplifying assumptions for each
program. In some cases, these assumptions were necessary because the program pro-
posal is silent on the particular issue. In others, we made assumptions to simplify
comparisons across programs. For example, the coverages provided by many of the
proposals are similar but not identical. We standardized the coverage to allow use of a
demand function that enables us to compare take-up rates across programs. Thus, we
project outcomes for insurance programs that are similar, but not identical, to those
proposed. Harmonizing some aspects of the proposals allows us to better highlight
the strengths and weaknesses of the different approaches. For this study, we modeled
programs motivated by the BCPP, Chubb, and PRIA programs. The results of these
models give insights into the performance of the Zurich and BCC programs, which
we did not separately model. For each modeled program, we projected outcomes in a
base case and then examined the sensitivity of the results to variation of the underlying
parameters over plausible ranges. The modeled programs were evaluated in terms of
efficacy, efficiency, affordability, risk borne by commercial insurers, expected annual
government net outlays, and policyholder subsidy.
Summary xi

Results

The relative performance of the modeled programs under the base-case parameter
assumptions is summarized in Table S.1. In the base case, insurers use a five-year pan-
demic return period in pricing the risk that they bear. Such a return period would be
most relevant in the first years after an insurance program is signed into law, and the
law applies to losses due to COVID-19 and its variants.
For small and medium-sized firms, the BCPP approach performs best in terms
of affordability, efficacy, and efficiency. The PRIA approach does the best in terms of
expected annual net government outlays and policyholder subsidy. It also does best in
terms of risk borne by commercial insurers, assuming that best in this context means
the most risk borne by commercial insurers. For large firms, the BCPP approach again
performs best in terms of affordability, efficacy, and efficiency. The PRIA approach
still transfers the most risk to commercial insurers, but the Chubb approach now gen-
erates the lowest expected annual government net outlays and policyholder subsidy. A
summary of the findings on each metric follows.

Affordability
The greater efficiency of the BCPP approach and the ability of the government to
spread risk over a longer period than the private sector can results in the BCPP’s lower
premium (Figure S.1). Although the $442 premium projected for small and medium-
sized firms in the BCPP approach seems modest, it remains to be seen whether 65 per-
cent of firms would buy coverage at this rate, as assumed in the base case. Very rough
calculations suggest that premiums at this level could increase the cost of commercial
multiple peril policies on the order of 4.4 to 8.8 percent. Some firms may be reluctant

Table S.1
Ranking of Modeled Programs by Program Metric at Base-Case Parameter Values

Small and Medium-Sized Firms Large Firms

BCPP Chubb PRIA BCPP Chubb PRIA
Program Metric Approach Approach Approach Approach Approach Approach

Affordability Most Mid Least Most Least Mid

Efficacy Highest Mid Lowest Highest Lowest Mid

Efficiency Highest Mid Lowest Highest Mid Lowest

Risk borne by commercial Least Mid Most Least Mid Most
insurers

Expected annual net Highest Mid Lowest Highest Lowest Mid
government outlays

Policyholder subsidy Most Mid Least Most Least Mid

NOTE: Shading indicates the program that performs best on the metric.
xii Improving the Availability and Affordability of Pandemic Risk Insurance

Figure S.1
Annual Premium in Base-Case Projection

1,800 8

Annual premium, small and medium-sized firms ($)
1,630
1,600 BCPP 6.86

Annual premium, large firms ($ millions)
7
Chubb
1,400 PRIA
6
1,200
1,060 5
1,000
4
800
3
600
442
2
400

200 0.70 1
0.16
0 0
Small and medium-sized firms Large firms

to purchase coverage at these rates, remembering the large government-funded Pay-
check Protection Program (PPP) in the current pandemic.
The premiums for large firms are comparable to those for small firms in the
BCPP and PRIA approaches when normalized by payroll (premium per $1,000 pay-
roll). Due to the market-based rate charged for the risk borne by the government, the
premium charged by Chubb’s Pandemic Re is projected to be much higher ($6.86 mil-
lion per firm) than in the other two approaches.

Efficacy
None of the programs does a particularly good job in terms of the percentage of losses
reimbursed over a 12-month pandemic (efficacy). This is largely because of the moder-
ate to low take-up rates and the 20 percent policyholder co-pay (Figure S.2). The BCPP
approach does best because its lower premium results in the highest take-up rate. Take-
up and the percentage of losses reimbursed for the large firms in Chubb’s Pandemic Re
approach are particularly low. Large firms presumably have a better ability to weather
revenue declines than smaller firms, but the fact remains that they account for 60 per-
cent of payroll at firms with employees, and their financial health will have important
consequences for overall employment levels during a pandemic.

Efficiency
Program efficiency is the ratio of claim payment to overall program costs and driven
by general and selling expenses, loss adjustment expenses, and private-sector capital
Summary xiii

Figure S.2
Take-Up Rate in Base-Case Projection

70 BCPP
65 65 Chubb
PRIA
60

Take-up rate (%)
50
42
40
32 32
30

10
10

0
Small and medium-sized firms Large firms

costs. BCPP does best on this metric partly because it relies on parametric policies
to reduce costs and to address the practical problem of simultaneously adjusting an
enormous number of claims.1 BCPP also fares best in part because it does not require
insurers to hold capital to protect against solvency risk. The cost of such capital is the
cost of transferring risk from the public to commercial insurers. Higher capital costs
reduce efficiency but are associated with lower government outlays should an event
occur and with other potential benefits that come with greater private-sector skin in
the game. These other potential benefits include greater attention to fraudulent claims,
as well as the development of the expertise and experience needed to grow the private
sector’s ability to write pandemic risk. When deciding whether to proceed with a pan-
demic risk insurance program, program efficiency should be compared with that of
other approaches, such as the PPP. We found that the efficiency of the BCPP approach
compares favorably to the administrative costs of the PPP, but unless pandemic return
periods are short, efficiencies of the other program do not compare so favorably.

1 Payment by a parametric insurance policy is triggered by a metric or index that is easy to determine rather
than property damage or other loss suffered by the insured. In the context of pandemic risk insurance, payment
would be triggered following a government declaration of a public health emergency and restricted to firms in the
geographic area and industries identified in the closure order. Such a parametric trigger would avoid the need for
the traditional claims adjustment process associated with indemnity policies.
xiv Improving the Availability and Affordability of Pandemic Risk Insurance

Risk Borne by Commercial Insurers
The amount of risk borne by commercial insurers is highest in PRIA, but, reflecting
the difficulty of insuring pandemic risk, it is not particularly high in any of the pro-
grams. Chubb proposes that the insurer risk share double over time, but even so, it will
remain modest. One advantage of the insurer deductible in the PRIA approach is that
the share of losses borne by insurers would be higher in smaller or better-contained
pandemics. The capital insurers’ need to hedge solvency risk is not large relative to
industry surplus for the Chubb and PRIA approaches (3.8 percent of an estimated
$325 billion for commercial lines), but further analysis is required to determine how
difficult it would be for the industry to raise this amount of capital.

Expected Annual Government Net Outlays and Policyholder Subsidy
Illustrating the usual trade-off between affordability and government cost, the BCPP
approach does best on affordability but worst on expected government annual net
outlays and policyholder subsidy. Expected annual government net outlays depend on
what the pandemic return period (as opposed to the return period used by insurers in
setting premiums) turns out to be. There is a great deal of uncertainty regarding this
return period, and Figure S.3 plots expected annual government net outlays for dif-
ferent return periods. As can be seen, expected annual net outlays are lowest for the
Figure S.3
Expected Annual Government Net Outlays at Base-Case Parameter Values

250

Expected annual government outlays ($ billions)
BCPP
200
Chubb
PRIA
150

100

–50
0 10 20 30 40 50 60 70 80 90 100
Actual return period (years)
Summary xv

Chubb approach, with government generating a surplus in the base case if the return
period turns out to be approximately ten years longer than the five-year return period
insurers use in setting premiums.
Our measure of government net outlays is only the direct revenues and outlays
of the program. Indirect costs, such as an increased demand for government assistance
during an event when there is limited insurance coverage, should also be considered
when assessing the full effect of a pandemic insurance program on overall government
outlays. Such an analysis was beyond the scope of this study, but the amount of risk
borne by commercial insurers is likely a better indicator of overall net government out-
lays than direct government outlays through the pandemic insurance program.
Projected outcomes are sensitive to several underlying model parameters about
which there is considerable uncertainly. In particular, the pandemic return period that
insurers use in pricing coverage has an important impact on expected premium and
take-up in the Chubb and PRIA approaches. If the proposed programs do not apply
to COVID-19 and its variants, a longer return period than the five years used in the
base case might be a more accurate expectation of insurer behavior. When the return
period used by insurers in setting prices is 30 years, the projected premiums for small
and medium-sized firms is similar in all three programs (Figure S.4). The premium
for large firms remains substantially higher in the Chubb approach than in the other
two approaches.

Figure S.4
Annual Premium When Insurers Use 30-Year Return Period in Setting Prices and Other
Parameters Remain at Base-Case Values

600 1.4

Annual premium, small and medium-sized firms ($)
1.22
BCPP

Annual premium, large firms ($ millions)
502 1.2
500 Chubb
442 PRIA
1.0
389
400

0.8
300
0.6

200
0.4

100 0.16 0.19
0.2

0 0
Small and medium-sized firms Large firms
xvi Improving the Availability and Affordability of Pandemic Risk Insurance

As apparent from Table S.1, no program as currently proposed dominates the
others on all the performance metrics. Rather, each has advantages and disadvantages
in terms of the performance metrics of concern.

Limitations

This analysis provides information useful to assessing the advantages and disadvan-
tages of different approaches for improving the availability and affordability of pan-
demic risk insurance. However, it does not seek to evaluate and compare the full range
of policy options for addressing pandemic losses. Such a policy analysis would system-
atically compare the advantages and disadvantages of an insurance-based approach
with a government-based approach, such as the PPP. That analysis would also take a
more comprehensive view of some of the outcome measures used in this analysis. For
example, in this report, we projected government net outlays directly through the pro-
posed programs, but a more comprehensive analysis would include the indirect effects
of the program on other government outlays.
The various proposals that we have examined contain several important gaps that
remain to be filled—for example,

• whether and how the parametric trigger would address firms either subject to a
partial shutdown order or that experience substantial revenue decline but are not
subject to a closure order
• whether policies backed by PRIA would provide coverage for pandemic losses
when the underlying business interruption policies require physical damage to
the insured property
• whether parametric policies would result in some firms receiving more compensa-
tion than they need and others less (basis risk).

In this analysis, we have in effect assumed that gaps are filled, thereby resulting in
the right amount of assistance getting to the right firms, but additional work is needed
to flesh out the proposals to deal with these and other gaps and to determine the extent
to which unaddressed issues can be resolved.
Abbreviations

B billions
BCC Business Continuity Coalition
BCPP Business Continuity Protection Program
BEIP Business Expense Insurance Program (Chubb)
BI business interruption
BOP business owner policy
COVID-19 coronavirus disease 2019
G&SE general and selling expenses
GDP gross domestic product
LAE loss adjustment expenses
M millions
NAIC National Association of Insurance Commissioners
P&C property and casualty
Pandemic Re Pandemic Reinsurance Program (Chubb)
PPP Paycheck Protection Program
PRIA Pandemic Risk Insurance Act of 2020
T trillions
TRIP Terrorism Risk Insurance Program
ZPPP Zurich Preliminary Pandemic Program

xvii
CHAPTER ONE

Introduction

The COVID-19 pandemic led to a substantial drop in U.S. economic activity in the
first part of 2020. Government stay-at-home orders, restrictions on business activities,
and consumer concerns about potential exposure to the virus caused unemployment
rates to rise and put many businesses in financial jeopardy. Businesses often purchase
business interruption (BI) coverage for loss of revenue because of fires and other perils,
but policies typically require that the interruption was due to physical damage at the
insured property, and, in recent years, the policies have excluded loss due to contagious
diseases. As a result, insurers have held that in most cases they are not obligated to
cover the enormous BI closures caused by COVID-19.1
Using an insurance approach to address pandemic risk has several advantages.
These include the ability to define benefits and a benefit distribution mechanism in
advance rather than rely on a hastily crafted government program. Also, an insurance
approach can leverage the deep capabilities of the commercial insurance industry to
place policies and to adjust claims. Moreover, insurance can transfer some risk to the
private sector.2
However, pandemic risk also creates a number of challenges for the private sector.
The global reach of pandemics limits the ability of insurers to diversify and spread risk.
As in the case of terrorism, pandemic impacts are very much determined by govern-
ment decisions that make it difficult to estimate potential losses or price policies. In the
case of pandemics, these decisions involve how broadly to restrict business and social
activity.3 As a result, commercial insurers have little interest in providing coverage for

1 Appendix A provides an overview of BI policies.
2 In other settings, insurance can create incentives to reduce risk by providing premium reductions to poli-
cyholders that adopt risk mitigation measures. It is not obvious, however, whether insurance can create similar
incentives to mitigate pandemic risk. As an example, consider a mitigation measure that would reduce seating
capacity in a restaurant during an outbreak. Such a measure would increase revenue loss, justifying an increase,
not decrease, in premium.
3 For a discussion of the insurability challenges posed by pandemics, see Hartwig, Robert, Greg Niehaus, and
Joseph Qiu, “Insurance for Economic Losses Caused by Pandemics,” Geneva Risk and Insurance Review, Vol. 45,
2020.

1
2 Improving the Availability and Affordability of Pandemic Risk Insurance

pandemic-induced business closures or restrictions on business activity, as evidenced
by the policy exclusions put in place before the COVID-19 outbreak and moves by
insurers to broaden these exclusions in the wake of COVID-19.
Insurers, insurance industry trade groups, policyholder groups, and Congress
have developed proposals that would enhance the role that insurance plays in providing
businesses with the resources they need to maintain payroll and benefits and to cover
the ongoing operating expenses during an outbreak. Some of these approaches are
motivated by the federal Terrorism Risk Insurance Program (TRIP) that was adopted
following the 9/11 attacks in response to the lack of insurance coverage for terrorism
events. Other approaches are shaped by the particular features of pandemic risk.

Research Questions and Study Approach

This report seeks to inform the discussion of the benefits and drawbacks of the dif-
ferent approaches for expanding insurance for pandemic-induced business closures or
restrictions on business activity by addressing the following research questions:

• What are the distinguishing features of the most-visible proposals for a pandemic
risk insurance program?
• How will these programs perform in terms of specific key dimensions: efficacy,
efficiency, affordability, risk borne by commercial insurers, and the extent of poli-
cyholder subsidies?

Although this report will compare, to some extent, an insurance-based approach
for addressing losses due to pandemics with an approach based on government pro-
grams, such as the Paycheck Protection Program (PPP), it does not seek to evaluate and
compare the full range of policy approaches for addressing pandemic losses. Rather, it
seeks to inform the debate on how society can best address pandemic risk by compar-
ing the different insurance-related approaches and predict how they might perform
under a range of assumptions.
To address these research questions, we first identified the most-visible proposals
and compared their key features. We then developed a high-level, quantitative model
to project the key outcomes for each of the proposals. This model is based on a number
of underlying parameters, and we reviewed economic and insurance literature and data
to develop reasonable values for these parameters. We also consulted with experts in
the field about what values would be reasonable.4 For each program, we projected out-

4 The experts we consulted had considerable knowledge about the cost components of and pricing practices for
insurance policies or about the provisions of proposed pandemic risk insurance programs. Interviews followed a
semistructured, open-ended format and were conducted via Microsoft Teams between November 2020 and April
2021.
Introduction 3

comes in a base case and then examined the sensitivity of the results to variation of the
parameters over plausible ranges.

Report Organization

Chapter Two describes five pandemic insurance programs: three proposed by insurers
or insurance industry associations, one proposed by a coalition of business policyhold-
ers, and one proposed by members of Congress. Chapter Three describes the modeling
approach and the parameters used in the projections, while Chapter Four presents and
discusses the projected outcomes for the base-case set of parameter values. Chapter Five
examines the sensitivity of the results to changes in parameter values, and concluding
comments are provided in Chapter Six. The provisions of a typical BI insurance policy
are described in Appendix A. The equations that underlie the pandemic risk insur-
ance model that we developed are presented in Appendix B. Appendix C provides an
example of how insurance payments might be determined when a business is partially,
as opposed to fully, shut down. Insurance industry data on general and selling expenses
(G&SE) and loss adjustment expenses (LAE) by property and casualty (P&C) insur-
ance line are provided in Appendix D.
CHAPTER TWO

Overview of Proposed Programs

In this chapter, we compare and contrast five of the most-visible proposals for provid-
ing business firms with the support they need to cover payroll, benefits, and ongoing
operating expenses during the economic disruption caused by a pandemic. Three of
these proposals have been offered by insurers or insurance industry groups, one has
been offered by a policyholder group, and the final has been introduced as congres-
sional legislation. In each case, we describe the policy offered to businesses (the policy
features) and the features of the program provided to participating insurers (the pro-
gram features). Businesses pay a premium to insurers for the coverage, and, in some
cases, insurers pass on part of this premium to the federal government for the risk it
bears. In the context of the proposed pandemic risk insurance programs, the risk shar-
ing provided by the government is sometimes referred to as reinsurance and sometimes
as a government backstop.1 Insurers might or might not pay a premium for this coverage.

Proposal 1: Business Continuity Protection Program

The Business Continuity Protection Program (BCPP)—proposed by the American
Property Casualty Insurance Association, the National Association of Mutual Insur-
ance Companies, and the Independent Insurance Agents and Brokers of America—
provides payroll, benefits, and support for ongoing operating expenses to the private
sector in the event of a declared public health emergency.2 This revenue replacement
assistance would be available through state-regulated insurance entities that voluntarily
participate in the BCPP.

1 Reinsurance refers insurance purchased by primary insurers from other insurers (the reinsurers). Primary insur-
ers are insurers that write policies for the end customer—whether a business or an individual.
2 This description of the BCPP is based on American Property Casualty Insurance Association, Independent
Insurance Agents and Brokers of America, Inc., and National Association of Mutual Insurance Companies, Busi-
ness Continuity Protection Program, updated September 2020.

5
6 Improving the Availability and Affordability of Pandemic Risk Insurance

Policy Features
The policy offered to businesses would provide up to three months of payroll, benefits,
and ongoing operating expenses and would be available to all for-profit and nonprofit
businesses (“Policy Features” section of Table 2.1). The revenue losses could occur
anytime during the typical one-year policy period, meaning that a total of three times
the monthly payroll, benefits, and ongoing operating expenses could be paid over the
course of the policy. Purchase would be voluntary, although a strong opt-out would
be required: Firms declining coverage would have to acknowledge in writing that they
would be ineligible for benefits if a pandemic-related business closure occurs.
There is no waiting period before payments begin, and policyholders could select
the policyholder co-pay, subject to a 20 percent minimum. Payment would automati-
cally be triggered following a government declaration of a public health emergency and
restricted to firms in the geographic area and industries identified in the closure order.
Such a parametric trigger would avoid the need for the traditional claims adjustment
process associated with indemnity policies.
Several questions remain about how the trigger would work in practice. First, the
proposal does not discuss whether payments would be triggered for firms that suffer
revenue declines but are not directly subject to a shutdown order. For example, would
payments be provided to a travel agency that can operate remotely but whose business
drops because hotels are closed?
Second, the proposal also does not discuss how partial shutdowns would be han-
dled. When a business completely shuts down and its revenue falls to zero, a claim pay-
ment equal to payroll, benefits, and ongoing operating costs would allow the firm to
maintain payroll and survive until the policy limit is reached. For a policy limit equal
to three months of preshutdown payroll, benefits, and ongoing operating expenses, the
coverage would last for three months under a complete shutdown. Claim payments
would not cover preshutdown profits or the costs that the firm can readily reduce
(variable costs). In a partial shutdown (for example, restaurants that are able to oper-
ate at 25 percent seating capacity and provide takeout service), revenue would not fall
to zero. The BCPP proposal does not discuss how claim payment would be adjusted
if this were the case. If the same claim payment were made as in a full shutdown, the
firms would likely need to return some of the payment if the goal is to maintain pay-
roll, benefits, and ongoing expenses but not to reimburse foregone profits. Appendix B
describes how much would have to be returned under various scenarios. The analysis
shows that firms would have to consider preshutdown profits and their success in
reducing variable costs during the shutdown when determining how much of the para-
metric payment to return or reserve for future use. Whether and how such a process
would be implemented remains to be specified.3

3 The proposal also does not address some of the regulatory issues associated with parametric insurance policies.
Under some configurations, parametric policies might be considered financial derivatives and therefore might be
Overview of Proposed Programs 7

Program Features
Insurer participation in the program is voluntary, but participating insurers would
be required to offer the pandemic coverage to all firms that purchase specified lines
of insurance. Insurers would be reimbursed for payments by the federal government.
Thus, there is no insurer deductible or co-pay. The federal government would also
reimburse participating insurers for the costs associated with selling policies and oper-
ating the program (G&SE), as well as for LAE. The BCPP leverages the capacity of
commercial insurers to issue policies, collect premiums, and pay claims, but all loss
payments and program costs would be funded by the federal government.
The BCPP proposes that the premium paid by policyholders be set at a percent-
age of policyholder revenue that does not vary over firm size or industry. The premium
would be set “to ensure widespread take-up,” but it would be left to the government
to determine what fraction of expected program costs would be recovered through the
policyholder premium.4
As with the policy features of the BCPP proposal, some aspects of program fea-
tures are not fully fleshed out. For example, insurer participation in the program is
voluntary, and it is not clear what would happen to businesses whose insurers decided
not to participate in the program. Would such business be able to buy coverage from
other insurers, and, if so, how might the cost of the policy be affected? It might be
more expensive for an insurer to set up coverage with a new policyholder than to add
pandemic coverage on to an existing policy

Proposal 2: Chubb Pandemic Business Interruption Program

Chubb’s Pandemic Business Interruption Program has two parts. Part I, the pandemic
Business Expense Insurance Program (BEIP), is open to small and medium-sized
firms, which are firms with 500 or fewer employees. Part II, the Pandemic Business
Interruption Reinsurance program (Pandemic Re), is open to firms with more than
500 employees. We describe each program in turn.5

Chubb Business Expense Insurance Program
Policy Features
The BEIP provides coverage similar to the BCPP in many respects (as shown in
Table 2.1). Payroll and expenses are covered (the proposal is silent on benefits) for up to
three months, a simple parametric structure provides for accelerated claim payments,

subject to oversight by the Commodities Futures Trading Commission, as opposed to state insurance regulators.
Consideration of such issues is beyond the scope of this analysis.
4 American Property Casualty Insurance Association, Independent Insurance Agents and Brokers of America,
Inc., and National Association of Mutual Insurance Companies, 2020, p. 3.
5 This description of the Chubb program is based on Chubb, Pandemic Business Interruption Program, Warren,
N.J., July 8, 2020.
Table 2.1 8
Features of Proposed Pandemic Risk Insurance Programs

Improving the Availability and Affordability of Pandemic Risk Insurance
Chubb Chubb Pandemic
Business Continuity Business Expense Business Interruption Zurich Preliminary
Protection Program Insurance Program Reinsurance Program Pandemic Program Pandemic Risk Business Continuity
Attribute (BCPP) (BEIP) (Pandemic Re) (ZPPP) Insurance Act (PRIA) Coalition (BCC) Program

Policy Features

Firms eligiblea All firms Firms with ≤500 Firms with >500 Firms eligiblea All firms All firms
employees employees

Losses covered Payroll, benefits, Payroll and ongoing Payroll and ongoing Payroll, benefits, Payroll, benefits, Payroll and fixed costs
and ongoing expenses expenses and ongoing ongoing expenses, for a nondamage
expenses expenses event cancellation parametric BI policy;
same coverage that
is provided for other
perils in standard P&C
policies

Policyholder No waiting period 14-day waiting 30-day waiting Policyholder can Same as underlying None for parametric
deductible period period (or longer, as select waiting BI policy policy; same as
desired) period underlying P&C
policies for other
coverages

Policyholder Policyholder choice None specified None specified 20% Same as underlying None for parametric
co-pay with 20% minimum BI policy policy; same as
underlying P&C
policies for other
coverage

Policy limit 3 months of payroll, 3 months of payroll, 3 months of payroll, 3 months of payroll, Same as underlying 90 days of payroll
benefits, and benefits, and benefits, and benefits, and BI policy and fixed costs for
ongoing expenses ongoing expenses ongoing expenses expenses, capped at parametric BI policy,
capped at $50M per $20M per month for same as underlying
firm large firms ($60M P&C policies for other
over 3 months) coverage

Claim payment Parametric Parametric Indemnity Parametric with Indemnity Parametric and
mechanism policyholder indemnity
required to self-
certify losses
Table 2.1—Continued

Chubb Chubb Pandemic
Business Continuity Business Expense Business Interruption Zurich Preliminary
Protection Program Insurance Program Reinsurance Program Pandemic Program Pandemic Risk Business Continuity
Attribute (BCPP) (BEIP) (Pandemic Re) (ZPPP) Insurance Act (PRIA) Coalition (BCC) Program

Policyholder Voluntary with Mandatory or Voluntary Voluntary Voluntary Voluntary
purchase strong opt-out voluntary with
requirement strong opt out

Program Features

Insurer Voluntary Voluntary Voluntary Mandatory Voluntary Mandatory
participation
requirement

Insurer offer Participating Mandatory offer Voluntary offer Mandatory offer Mandatory to Mandatory to
requirement insurers required to policyholders policyholders who businesses with P&C
to offer to all with fixed property purchase business insurance
policyholders who coverage interruption
purchase covered coverage
lines of commercial
insurance

Program limit None $750B $400B No limit specified $750B No limit

Insurer $0 $0 $0 $0 5% of direct earned $0
deductible premium in specified

Overview of Proposed Programs
lines

Insurer co-pay 0% 6% up to $250B 5% up to $300B for 0%, 5%, or 10%; 5% of losses above 5%
(rising to 12% over first 5 years (rising to insurers can assign insurer deductible
20 years); 0% above 10% by year 10); 0% individual policies to up to $750B
$250B above $300B any pool

Government Premium set to None Market rate Below market None None during first
charge for ensure widespread rate; amount of 5 years; following
backstop take-up premium ceded economic recovery
to government period, government
depends on pool has authority to
chosen charge for backstop
9
Table 2.1—Continued 10

Source of Government will pay Premium covers Included in Government Included in premium Not discussed
payment for fee for insurer G&SE G&SE and LAE premium; 3% pays commission
insurer and LAE for broker 5–7% to insurers that
expenses for insurer is 25–30% of
administrative fee; premium ceded
claim handling to government
fee not specified; depending on pool
Pandemic Re chosen
administrative and
operating fee 0.25%
of written premium

Premium Based on % of Expected to be small Both insurers and Government sets Not discussed Not discussed
policyholder relative to a firm’s government are premium with target
revenue; uniform overall insurance paid an appropriate of 2% rate on line
across all firms; expenses risk-adjusted price for <500 employees
set to promote for pandemic cover and 3% rate on line
widespread take-up for v500 employees

Expected take- Not discussed 90% 30% Varies by Not discussed Not discussed
up rate industry (90% for
accommodations
and food services,
30% for utilities);
65% economy-wide

SOURCES: Author summary based on American Property Casualty Insurance Association, Independent Insurance Agents and Brokers of America, Inc., and
National Association of Mutual Insurance Companies, 2020; Chubb, 2020; Zurich, Zurich’s Draft Concept for Facilitating Pandemic Protection, Los Angeles,
Calif., December 7, 2020; U.S. House of Representatives, Pandemic Risk Insurance Act of 2020, H.R. 7011, May 26, 2020; Business Continuity Coalition,
“BCC Proposal: A Proposal from the Policyholder Community,” September 2020.
a Includes both for-profit and nonprofit firms.
Overview of Proposed Programs 11

and purchase is voluntary with a strong opt-out. Unlike the BCPP, there is a 14-day
waiting period between when business closures or restrictions are announced and cov-
erage begins, but there is no policyholder co-pay.
Program Features
Like the BCPP, insurer participation in the program is voluntary, and insurers choosing
to participate must offer coverage to their policyholders. The major difference between
the two programs is that, in Chubb’s proposal, the commercial insurers bear some
risk. In the first year of the program, insurers pay 6 percent of claims up to $250 bil-
lion in policyholder losses—a maximum of $15 billion. The federal government pays
100 percent of any additional losses up to $750 billion. The industry exposure rises to
12 percent of the first $250 billion (up to $30 billion) over 20 years.
The BEIP envisions a policyholder premium that covers the risk borne by com-
mercial insurers. Insurers do not pay a premium to the government; thus, the poli-
cyholder premium does not include a component that covers the potential payments
by the federal government. The BEIP proposal does not specify whether the govern-
ment would reimburse commercial insurers for G&SE; however, given that there is no
indication that the government would pay G&SE, it seems reasonable to assume that
Chubb envisions these costs being built into the premium—all of which is kept by
insurers. Given the strong opt-out and the subsidized premium, Chubb expects 90 per-
cent of small businesses to participate in the program.

Pandemic Re
The objective of Pandemic Re is to promote a market-oriented program for large busi-
nesses. Purchase of coverage by the roughly 20,000 businesses with more than 500
employees is voluntary, with no strong opt-out or requirement that insurers offer cov-
erage to large firms.
Policy Features
The policy again looks similar to those offered by the BCPP and Chubb’s BEIP: Payroll
and expenses are covered for three months with no policyholder co-pay. In this case,
however, there is a 30-day waiting period, or longer if desired, and a $50 million cap
on payments per policy. Importantly, traditional loss adjustment techniques are used,
which are expected to be much more costly than those needed for parametric policies.
Program Features
Insurers pay 5 percent of insured losses up to $300 billion ($15 billion maximum), and
again the government pays 100 percent of additional losses up to an overall program
cap of $400 billion (government and commercial insurer claim payments combined).
In contrast to the BCPP and the BEIP, Pandemic Re is a program “in which both
the insurance industry and the government are paid an appropriate risk-adjusted price
12 Improving the Availability and Affordability of Pandemic Risk Insurance

for pandemic cover” (italics added).6 We take this to mean that the policyholder pre-
mium should cover (1) the expected loss payments and associated LAE for which com-
mercial insurers are financially responsible, G&SE, and the costs of capital required by
commercial insurers to mitigate their solvency risk and (2) the government’s expected
loss payment and the associated LAE. This approach differs critically from the BEIP
and BCPP, which charge insurers no premium for the government backstop. A pan-
demic return period must be assumed to calculate expected loss and LAE. As will be
discussed in Chapter Three, the return period is the inverse of the annual probability
that a pandemic occurs. Chubb assumes a 30-year return period (0.033 annual prob-
ability) in its proposal.

Proposal 3: Zurich Preliminary Pandemic Proposal
Policy Features
The Zurich Preliminary Pandemic Proposal (ZPPP) is open to firms of all sizes and
offers coverage very similar to what is offered by the BCPP and Chubb programs. Like
the Chubb proposal, the ZPPP includes a limit on claim payments to large policyhold-
ers: $20 million per month or $60 million over three months for large firms, compared
with $50 million in Chubb’s Pandemic Re proposal.

Program Features
The ZPPP proposes a flexible program (see Table 2.1) that allows insurers to choose
how much risk to bear. Depending on how much risk insurers choose to bear, the pro-
gram can end up looking quite similar to the BCPP. It can also end up looking very
similar to the Chubb proposal. As will be discussed further, the ZPPP expects insurers
to retain 1 percent of the risk, resulting in a program that is similar to the BCPP. The
ZPPP is inspired by the federal crop insurance program, with which Zurich has con-
siderable experience.7
Like the BCPP and Chubb proposals, insurers participating in the ZPPP must
offer coverage to their customers with fixed property coverage. In contrast to the BCPP
and Chubb proposals, insurers are required to participate in the ZPPP.
The main distinguishing feature of the ZPPP is that insurers can select how much
risk to bear on a policy-by-policy basis. Insurers can place policies in pools in which
100 percent of the premium and risk are ceded to the government, 95 percent is ceded
to the government, or 90 percent is ceded to the government. By allocating across
pools, insurers can tailor their portfolios to their risk appetites. One advantage of this
approach is that insurers can adjust the risk they retain over time by modifying the

6 Chubb, 2020, p. 4.
7 This description of Zurich’s program is based on Zurich, 2020.
Overview of Proposed Programs 13

distribution of policies across pools. Zurich projects a very modest insurer risk appetite
initially and estimates that 99 percent of the premium and risk would be ceded to the
federal government in the first years of the program.
Like the federal crop insurance program, policyholder premiums in the ZPPP
are subsidized by the federal government.8 Zurich does not specify the amount of sub-
sidy but envisions a rate on line of 2 percent for small and medium-sized firms and
3 percent for large firms.9 This target premium can in principle be used to impute an
implied level of subsidy.10 Insurer G&SE and LAE would be covered in part through
a commission on the premium ceded to the pool. The ceding commission starts at
25 percent for the 100 percent pool and rises to 30 percent for the 90 percent pool.
The ceding commission structure provides an incentive for insurers to bear more risk.
Insurers will also use the premium that they retain to cover G&SE, LAE, and the por-
tion of loss payments for which they are responsible.
Zurich expects take-up to vary by industry, projecting in its program proposal,
for example, 90 percent and 30 percent take-up rates for the accommodation and food
services sector and utilities sector, respectively. Zurich expects take-up rate to average
65 percent across the economy as a whole.11

Proposal 4: Pandemic Risk Insurance Act

Introduced by Representative Carolyn Maloney in May 2020, the Pandemic Risk
Insurance Act of 2020 (PRIA) seeks to establish the Pandemic Risk Reinsurance Pro-
gram modeled on the federal TRIP.12 PRIA requires participating insurers to make
available in all BI policies coverage for losses from public health emergencies declared
for infectious disease. All firms are eligible for the policy, and purchase is voluntary.13
In contrast to TRIP, insurer participation in the program is voluntary.

Policy Features
Participating insurers are required to make available coverage for pandemic-related
losses that does not differ materially from that available for events other than public

8 The federal subsidy in the federal crop insurance program amounts to approximately two-thirds of the
premium.
9 Rate on line is premium as a percentage of the amount of coverage provided.
10 Zurich would implement a subsidy by allowing the government to use a longer return period in its determina-
tion of premium than would be required by the private sector.
11 The 65 percent economy-wide average is based on our conversations with analysts familiar with the Zurich
proposal.
12 U.S. House of Representatives, 2020.

13 See U.S. House of Representatives, 2020.
14 Improving the Availability and Affordability of Pandemic Risk Insurance

health emergencies. Coverage limits, waiting periods, and the type of expenses covered
can vary across BI policies, and PRIA does not specify the amount and scope of cover-
age that should be provided. BI coverage typically includes such expenses as utilities
and rent and lost profit. However, it might not cover payroll for employees that could
easily be laid off and then rehired or replaced (see Appendix A for an overview of typi-
cal BI coverage). If this were the case, BI coverage backstopped by PRIA might not be
sufficient to maintain prepandemic employment levels. In addition, BI coverage often
applies only when there has been physical damage to the insured property. Policies that
contain such physical damage requirements might not pay out when the cause of loss
is a contagious disease.

Program Features
Turning to PRIA’s risk-sharing provisions, insurers bear the first layer of loss (the
insurer deductible) and then share losses above the deductible with the federal govern-
ment. Mirroring TRIP, PRIA’s deductible for commercial insurers as a whole is 5 per-
cent of the direct earned premium in specified insurance lines of business. As will be
discussed in Chapter Three, the insurer deductible would have been roughly $12 bil-
lion in 2019. Insurers then are responsible for 5 percent of losses above the program
deductible up to $750 billion in insured losses (the maximum industry exposure is thus
roughly $49 billion).
Like the Chubb BEIP, the government does not charge a premium for the risk it
bears in the PRIA proposal. Insurers will presumably set the premium to cover their
share of loss payments and G&SE and LAE.

Proposal 5: Business Continuity Coalition Program

The Business Continuity Coalition (BCC) program provides more-expansive coverage
for pandemic-related loss than the other proposals. The BCC is composed of business
insurance policyholders from across the U.S. economy.14

Policy Features
Insurers would be required to make available coverage for “insured losses related to
covered public health emergencies” in all their P&C insurance policies.15 The BCC’s
mandatory offer requirement is broader than PRIA’s. PRIA applies only to BI insur-
ance, including event cancellation. In contrast, the BCC proposal applies to all P&C
policies, including workers’ compensation, commercial general liability, directors and

14 Business Continuity Coalition, homepage, undated.

15 Business Continuity Coalition, “Pandemic Risk Insurance Act Business Continuity Coalition Proposal:
Section-by-Section Description,” March 2021.
Overview of Proposed Programs 15

officers liability, excess insurance, and event cancellation (subject to several excep-
tions). The BCC proposal in addition requires insurers to make available to all their
P&C policyholders parametric BI coverage. (Insurers can arrange for such coverage to
be made available by an affiliate or a parametric insurance facility in which the insurer
participates.) The parametric coverage would not require the incident to cause physi-
cal damage at the insured property. This parametric nondamage BI coverage would
provide 90 days of fixed costs and payroll and would be triggered by certification of
a public health emergency and a closure order for the policyholder’s North American
Industry Classification System (NAICS) code.
Program Features
Unlike the other proposed programs, all insurers would be required to participate in
the BCC program. Insurers would bear 5 percent of the losses and the federal govern-
ment 95 percent. There would be no cap on government or insurer outlays. Insurers
would not be charged a premium for the quota share reinsurance provided by the gov-
ernment during an economic recovery period. The economic recovery period covers
the five years following congressional enactment of the program and resets if there is a
public health emergency during the five years following enactment.

Discussion

The proposed programs share many features but also differ in important ways. The
BCPP, Chubb, and Zurich proposals offer similar coverage for payroll, benefits, and
ongoing operating expenses. PRIA takes a different approach by requiring insurers
to make coverage available for pandemic-related losses that does not differ materially
from that available for perils covered by BI provisions. The BCC proposal goes fur-
ther by supporting insurance coverage for pandemic losses in other lines, such as event
cancellation and directors and officers liability, in addition to providing payroll and
operating expenses due to BI.
The proposals leverage the underwriting and claim adjusting capabilities of the
private sector but vary in the amount of risk borne by commercial insurers. Commer-
cial insurers bear no risk in the BCPP, and insurers are responsible for up to $30 billion
in loss payments in the Chubb proposal (small, medium-sized, and large firms com-
bined) initially, doubling over 20 years. Commercial insurers are potentially respon-
sible for $50 billion in losses under PRIA. However, the private insurer share is not
large relative to total insured losses in any of the proposals. An attractive feature of the
Zurich proposal is that insurers can adjust the risk they retain over time by modifying
the distribution of policies across pools.
To reduce the costs of simultaneously adjusting a large number of claims, all
but one of the proposals rely at least in part on a parametric trigger. The Chubb pro-
posal takes a hybrid approach, using a parametric trigger for the vast majority of firms
16 Improving the Availability and Affordability of Pandemic Risk Insurance

with employees that have 500 or fewer employees (discussed further in the following
chapter) and an indemnity approach for the relatively few firms with more than 500
employees. PRIA is the only proposal that does not specifically call for the use of a
parametric policy, but insurers are presumably free to write policies with parametric
triggers if they choose to do so.
Premiums are subsidized either explicitly or implicitly in all the proposals. PRIA
and Chubb’s BEIP for small and medium-sized firms provide no payments to the fed-
eral government to cover the cost of the claim payments made by the federal govern-
ment. The same is true for the BCC program during a five-year economic recovery
period that can reset. The BCPP expects the government to use a long return period in
calculating the price of the coverage it provides, and this return period may well exceed
the actual return, causing premiums to be lower than actuarially justified. Chubb’s
Pandemic Re for large firms appears to be the only program that does not rely on gov-
ernment subsidies, calling for a program “in which both the insurance industry and the
government are paid an appropriate risk-adjusted price for pandemic cover.”16
Several important gaps need to be addressed in all these proposals. Whether and
how the parametric trigger would address firms subject to partial shutdown order or
firms that experience substantial revenue decline but are not subject to a closure order
need to be worked through. PRIA might not result in broader coverage for pandemic
losses if the underlying BI policies retain a requirement that revenue declines be caused
by physical damage. In addition, standard BI coverage can include payroll and ben-
efits for all employees, but provisions vary. If a goal is to maintain employment during
a pandemic, PRIA may need to be modified to make it clear that policies would be
required to support payroll and benefits for all employees (up to a certain limit), not
just those who are difficult to replace if laid off.17 Insurer participation is voluntary in
all proposals except the BCC’s. The potential for and consequences of low insurer par-
ticipation are not addressed.
In following chapters, we model the potential effects of programs motivated by
the BCPP, Chubb, and PRIA proposals. The results for these programs give insights
into the performance of the Zurich and BCC programs; we do not separately model
the Zurich and BCC programs. The ZPPP can perform quite similarly to the BCPP
if insurers decide to retain only a very small portion of the risk or perform more like
the Chubb program if insurers retain more risk. Zurich projects that 99 percent of risk
will be transferred to the government at least initially and, if so, will look similar to the
BCPP in the first years of the program.

16 Chubb, 2020, p. 4; italics added.

17 The potential for captive insurers to undermine the performance of the PRIA should also be considered. Cap-
tive insurers, which can provide coverage to a single insured, have come under much criticism in TRIP. Given
that PRIA is modeled on TRIP, similar concerns are relevant for PRIA. For discussion of the captive insurer issue,
see Centers for Better Insurance, Terrorism Risk Insurance Act: Shifting Risk Through Terrorism Insurance Captives,
Frederick, Md., 2020.
Overview of Proposed Programs 17

The BCC’s nondamage parametric BI insurance program has much in common
with the BCPP and Chubb proposals, and we do not model it separately for the same
reasons that we do not model the ZPPP. The other component of the BCC proposal
expands coverage beyond the revenue replacement needed to maintain payroll, ben-
efits, and ongoing operating expenses. Although coverage for various additional costs
or liabilities that firms could face during a pandemic are potentially an important part
of overall strategy to increase business resilience to pandemic risk, evaluating programs
to increase the availability and affordability of such coverage is beyond the scope of
this analysis.
CHAPTER THREE

Modeling Approach and Base-Case Parameters

In this chapter, we first discuss our approach to modeling the effects of the proposed
pandemic risk insurance programs on relevant outcomes. To do so, we describe the
model inputs, the outputs, and the criteria used to evaluate program performance. We
then motivate and specify the values for the input parameters used in the base-case
simulations. As explained at the end of Chapter Two, we model the BCPP, Chubb, and
PRIA proposals. The models are based on a number of simplifying assumptions for
each program. In some cases, these assumptions were necessary because the program
proposal was silent on a particular issue. In others, we made assumptions to simplify
comparisons across programs. For example, the coverages provided by many of the
proposals are similar but not identical. We standardized the coverage to allow use of an
insurance demand function that enabled us to compare take-up rates across programs.
Thus, we project outcomes for insurance programs that are similar, but not identical,
to those proposed. Harmonizing some aspects of the proposals allows us to better high-
light the strengths and weaknesses of the different approaches.

Modeling Approach

As shown in Table 3.1, the pandemic risk insurance model takes as inputs the features
of the risk, program and policy features, take-up rate for the BCPP, and the price elas-
ticity of demand. It then predicts outputs, such as insured loss, the premium, program
expenses, and the allocation of cost and risk between the private and public sectors.
These outputs can then be used to construct the following performance metrics: the
percentage of loss insured, program efficiency, the premium, risk borne by commercial
insurers, and the share of program costs covered by premiums. In the remainder of the
chapter, we detail the model inputs, model outputs, and performance metrics. The
equations that underlie the model are presented in Appendix A.

Model Inputs
The inputs begin with a characterization of the economic loss caused by a pandemic.
Our focus is on providing insurance coverage that allows firms to maintain payroll

19
20 Improving the Availability and Affordability of Pandemic Risk Insurance

Figure 3.1
Model Overview

Inputs Outputs
• Economic loss due to • Loss
a pandemic • Take-up rate relative to BCPP
• Return period • Insured loss
• Policy features Model • General and selling expenses
• Program features • Loss adjustment expenses
• Program costs • Capital costs
• BCPP take-up rate • Premium
• Elasticity of demand • Losses incurred by insurers
• Losses incurred by government

Performance metrics
• Efficacy
• Efficiency
• Affordability
• Risk borne by commercial insurers
• Expected annual government outlays
• Policyholder subsidy

and employee benefits and to cover ongoing operating expenses during the economic
disruptions caused by a pandemic. We specify the number of firms affected by the pan-
demic and the amount of revenue that needs to be replaced in order to sustain payroll,
benefits, and ongoing operating expenses at prepandemic levels (see the “Economic
Loss from the Pandemic” section of Table 3.1). We model the outcomes of the vari-
ous proposed programs under different assumptions about the frequency with which
a pandemic occurs.1
The features of the insurance policy offered to business firms under the proposed
program are fed into the model. These include the losses covered by the policy, the
policyholder waiting period, the policyholder co-pay for losses following the waiting
period, the coverage period, and the policy limit. The policyholder deductible is speci-
fied in weeks following the declaration by public authorities that triggers the policy.
The coverage limit is specified in terms of the number of months of prepandemic pay-
roll, benefits, and ongoing operating expenses.
Model inputs also include the features of the proposed program (the “Program
Features” section of Table 3.1). The firm sizes eligible for the program are specified.
For this analysis, small and medium-sized firms (firms with fewer than 500 employees)

1 A pandemic that occurs with annual probability p has a return period of (1/p) years.
Modeling Approach and Base-Case Parameters 21

Table 3.1
Model Inputs

Input Description

Economic Loss from the Pandemic

Firms affected Number of firms affected by pandemic

Required revenue Amount of business revenue that needs to be replaced to sustain payroll,
replacement benefits, and ongoing operating expenses at prepandemic levels

Return period The probability of an event occurring is assumed to be a dichotomous
random variable with probability p. The return period is 1/p years.

Insurance Policy Features

Losses covered Revenue needed to cover payroll, benefits, and ongoing operating
expenses

Policyholder waiting period A waiting period before coverage begins

Policyholder co-pay Proportion of loss after waiting period that is not covered by the
insurance payment

Coverage duration The number of months following the waiting period during which losses
will be covered

Policy limit Maximum dollar amount that can be paid to a policyholder in a single
event

Program Features

Firms eligible Program eligibility based on firm size (as measured by number of
employees)

Program cap Maximum aggregate claim payments allowed under the program

Insurer deductible The amount of claim payments that must be paid by commercial insurers
before government claim payments begin

Insurer co-pay Proportion of claim payments above insurer deductible that is paid by
commercial insurers

Program Costs

G&SE Broker commissions and other costs of selling policies and insurer costs
of running the program; specified in dollars per policy

LAE ratio Costs of adjusting and paying claims; specified as a proportion of claim
payments

Rate of return on capital Rate of return required on the capital commercial insurers hold for
solvency risk

Diversification factor The diversification factor captures the extent that insurers can use
surplus held for other hazards to reduce the amount of capital they need
to support pandemic risk policies

Return periods used in pricing The pandemic return periods that commercial insurers and the
government use when pricing coverage; the return period used by
insurers can differ from the return period used by government

Take-Up Rate

BCPP take-up rate Proportion of eligible firms that purchase coverage in the BCPP
22 Improving the Availability and Affordability of Pandemic Risk Insurance

are distinguished from large firms (firms with 500 or more employees). The insurer
deductible and co-pay proposed in the program and the maximum aggregate claim
payments allowed under the program (program cap) are specified.
The model requires parameters that are used to calculate the costs of running the
program. G&SE per policy is specified to capture the broker and other costs associated
with placing insurance policies and the administration and operation costs of the pro-
gram. In standard insurance cost reporting, these costs are represented as a percentage
of the premium. As discussed later is this chapter, we take a different approach because
the premium does not reflect the full costs of the program in some cases. To calcu-
late costs of adjusting and paying claims, an LAE ratio is applied to insured loss. The
ratio depends on the adjustment process—lower for parametric policies and higher for
indemnity policies.
To mitigate solvency risk, insurers hold capital, and the rate of return that insurers
must pay on this capital is specified. The diversification factor captures the extent to
which insurers can use surplus held for other hazards to reduce the amount of capital
they need to support pandemic risk policies. The capital cost of a program is deter-
mined by multiplying the rate of return on capital by the amount of capital required.
The model does not consider the government’s costs of administering and oper-
ating the program (however, government claim payments are of course included). As
illustrated by TRIP, the government administrative and operating costs are not likely
to be large relative to overall program outlays and might not vary a great deal across
the programs examined. We also do not cost out the potential government expense of
financing these payments. After adjusting for inflation, the cost of federal borrowing is
currently very low (and perhaps even negative), and we ignore this cost in our analysis.
To project a premium that would be charged by a program, the pandemic return
period used by insurers in setting prices is specified. The return period used by insurers
can differ from that used by the government, and both return periods can be different
from the actual return period, which is highly uncertain.
The percentage of eligible business firms deciding to purchase the policy (take-
up rate) in the BCPP is also specified as an input. The take-up rate in the Chubb and
PRIA programs are calculated by the model to be consistent with the difference in
premiums across the programs.

Model Outputs
On the basis of these inputs, the model projects outputs relevant to the evaluation of
each of the proposed pandemic insurance programs. Table 3.2 lists the model outputs
and summarizes how they are calculated. Calculations are done separately for each of
the modeled insurance programs using parameter values specified in the next section
of this chapter. In the remainder of this section, we provide an overview of how each
of the outputs is calculated. The equations used to calculate these outputs are provided
in Appendix A.
Modeling Approach and Base-Case Parameters 23

Table 3.2
Model Outputs

Output Description

Number of participating firms Number of firms purchasing coverage; projected separately for
small and medium-sized firms (<500 employees) and large firms
(v500 employees)

Loss should an event occur Shortfall in revenue needed to sustain payroll, benefits, and
ongoing operating expenses regardless of whether the decline
is insured

Insured loss should an event occur Aggregate insurance payments based on policyholder waiting
period, insurance co-pay, policy limit, and take-up rate

Losses incurred by commercial insurers The loss borne by commercial insurers if an event occurs
should an event occur

Losses incurred by the government The loss borne by the government should an event occur
should an event occur

LAE should an event occur Cost of adjusting claims should an event occur

Annual G&SE Annual G&SE of the program; there are annual recurrences of
G&SE regardless of whether a loss occurs

Capital held by commercial insurers Amount of capital held by commercial insurers to hedge against
for solvency risk solvency risk

Cost of capital for solvency risk Amount of capital needed by commercial insurers for solvency
risk multiplied by cost of capital

Annual premium Determined by insured losses should an event occur,
distribution of insured losses between government and
commercial insurers, return period used in pricing, G&SE, LAE,
and capital costs

Take-up rate Take-up rate consistent with the difference in premiums across
programs and the take-up rate assumed for the BCPP

Number of participating firms: The number of participating firms is the
number of business firms eligible for the program multiplied by the take-up rate. The
number of participating firms is calculated separately for small and medium-sized
firms and large firms.
Loss should an event occur: This is the shortfall in revenue needed to sustain
payroll, benefits, and ongoing operating expenses caused by the pandemic for all busi-
nesses nationwide at prepandemic levels.
Insured loss should an event occur: This is the amount of loss that is reim-
bursed under the program. It is a function of both policy features (e.g., policyholder
waiting period, the policyholder co-pay, the policy limit) and program features (e.g.,
any cap on overall outlays, take-up rate).
Losses incurred by commercial insurers should an event occur: These are the
claim payments that are the financial responsibility of private insurers. Private insurers
24 Improving the Availability and Affordability of Pandemic Risk Insurance

may be tasked with the administrative responsibility of paying the full claim, but losses
incurred by commercial insurers pertain only to that portion of the claim payments for
which insurers are financially responsible. The losses incurred by commercial insurers
depend on the risk-sharing features of the proposed program.
Losses incurred by the government should an event occur: These are the
part of claim payments that are the financial responsibility of the government. The
losses incurred by the government depend on the risk-sharing features of the proposed
program.
LAE should an event occur: These expenses are the administrative costs of
adjusting and paying the claims. They are projected as a proportion of insured losses
(using the LAE ratio) and differ depending on whether a traditional indemnity policy
or parametric policy is used.
Annual G&SE: These expenses are the brokerage, commission, and other selling
costs associated with placing the policies and the insurer costs of operating the pro-
gram. Annual G&SE should be calculated by multiplying G&SE per policyholder by
the number of firms participating in the program. G&SE per firm varies by firm size.
Capital held by commercial insurers for solvency risk: The insurance indus-
try’s cost of bearing pandemic risk includes the cost of the capital needed to ensure,
with a given probability, that insurers have sufficient resources to finance claim pay-
ments.2 The planning probabilities that insurers are thought to use in making this
decision are high—99 percent or higher.3 The planning probability determines the
size of the insured loss that insurers plan for, and the amount of capital that insurers
need to hold is the planning loss less the annual premium. For this analysis, we have
assumed that whether a pandemic occurs is a dichotomous random variable and that
insurers plan for an event that is of the same magnitude as the current event. The plan-
ning loss used in our analysis is thus insurer claims plus LAE should an event occur.
When insurers cannot spread risk across other insurance lines, the amount of capital
required is the planning loss less the annual premium net of G&SE and capital cost.
The diversification factor captures the extent that insurers can use surplus held for
other types of hazards to reduce the amount of capital they need to support their pan-
demic risk book of business.
Cost of capital needed by commercial insurers for solvency risk: The cost of
capital needed by insurers for solvency risk is the amount of capital multiplied by the
cost of capital.

2 The risk that the premium or accumulated capital will not be adequate to cover claims can be referred to as
solvency risk, volatility risk (the volatility of claim payments relative to premium revenue), or timing risk (the risk
that a large loss will occur before sufficient premium has been collected).
3 With a planning probability of 99 percent, there is a 1 percent chance that the loss will exceed the planning loss.
Lloyd Dixon, Robert J. Lempert, Tom LaTourrette, and Robert T. Reville, The Federal Role in Terrorism Insurance:
Evaluating Alternatives in an Uncertain World, Santa Monica, Calif.: RAND Corporation, MG-679-CTRMP,
2007, p. 17.
Modeling Approach and Base-Case Parameters 25

Annual premium: The premium is the sum of expected annual insured loss,
expected annual LAE, annual G&SE, and the annual cost of capital required to hedge
solvency risk. Expected annual insured loss is the product of the insured losses should
an event occur multiplied by the inverse of the return period that is used in setting the
premium. We will make different assumptions about the return period used by insur-
ers and the government in setting prices. In some cases, the insured loss relevant to
pricing is only the part of overall insured loss that is the responsibility of private insur-
ers. In others, it is the total insured loss.
Take-up rate: The take-up rate in the Chubb and PRIA programs are calculated
by the model to be consistent with the difference in premiums across the programs. As
described in Appendix C, we use a constant-elasticity demand function to estimate the
take-up rates in the Chubb and PRIA programs, given the premium and take-up rate
in the BCPP model. The take-up rates for the Chubb and PRIA programs are thus best
interpreted as relative to the BCPP take-up rate.

Performance Metrics
Program performance is evaluated across six dimensions: efficacy, efficiency, afford-
ability, risk borne by commercial insurers, expected annual government net outlays,
and policyholder subsidy.
Efficacy is measured in the terms of the percentage of losses that are insured
(Table 3.3). The percentage of losses that are insured depends on the take-up rate, the
characteristics of the policy offered under the program, and the program cap. This
measure of efficacy assumes that insurance payments are delivered to the firms that
need them and are used to support payroll, benefits, and ongoing operating expenses
rather than, for example, profit. As discussed in Chapter Two, further work is needed
to flesh out the proposal details that will determine the extent to which this turns out
to be the case.
The ratio of claim payments to the overall costs of the program is a measure of
program efficiency. It measures the fraction of overall resources consumed by the pro-
gram that result in claim payments to policyholders.
Affordability is characterized by the annual premium per firm, the annual pre-
mium per $1,000 payroll, and the premium per $1,100 coverage. Average premium per
firm is most relevant to an average-sized firm and does not capture how the premium
might vary by firm size. Average premium per $1,000 payroll normalizes the premium
by a measure of firm size but assumes that the policy pricing per $1,000 payroll will
not vary by firm size. Premium per $1,000 coverage is the ratio of the premium to the
amount of coverage provided by the policy.4

4 In calculating premium per $1,000 coverage, we deduct the policyholder co-pay from the policy limit. We do
not consider the overall program cap in calculating premium per $1,000 coverage, arguing that the individual
policyholder does not know whether the cap will be binding when purchasing the coverage.
26 Improving the Availability and Affordability of Pandemic Risk Insurance

Table 3.3
Measures Used to Evaluate Program Performance

Performance Metric Significance

Efficacy

Percentage of losses that are A higher share of losses that are insured reduces the impact of a
covered pandemic on firms and employees

Percentage of losses that are A higher share of losses that are insured reduces the impact of a
insured for firms that participate pandemic on firms and employees
in the program

Efficiency

Ratio of claim payments to overall Measures the efficiency of the program in providing insurance
program costs payment to policyholders

Affordability

Annual premium per firm and per High premiums can reduce willingness to purchase policies,
$1,000 payroll increasing the economic consequences of a pandemic for firms and
employees

Premium per $1,000 coverage Premium divided by amount of coverage provided by insurers

Risk Borne by Commercial Insurers

Insurer incurred loss and LAE A measure of the role commercial insurers play in bearing
should an event occur pandemic risk

Capital required by commercial When compared with industry surplus, provides a sense of how
insurers for solvency risk difficult it will be for the insurance industry to bear the risk
envisioned under the proposed program

Percentage of overall program Measures the role that commercial insurers play in funding loss
costs borne by insurers payment and program expenses

Expected Annual Net Government Outlays

Expected annual government net Measures the fiscal burden the program places on the government
outlays

Policyholder Subsidy

One minus the ratio of premium to Measures the share of overall program costs subsidized by the
overall program costs government as opposed to being borne by policyholders

Several metrics are used to characterize the risk borne by commercial insurers:

• The loss and LAE borne by commercial insurers if an event occurs measure the
risk borne by commercial insurers.
• The capital required by commercial insurers for solvency risk reflects the cost of
requiring commercial insurers to bear some of the pandemic risk. When com-
pared with overall insurance industry surplus, the capital required provides a
Modeling Approach and Base-Case Parameters 27

sense of how difficult it would be for the insurance industry to bear the level of
risk envisioned under the program.
• The percentage of overall program costs borne by commercial insurers measures
the financial responsibility of insurers for claim payments, LAE, and program
costs relative to the government.

Expected annual government net outlays are calculated by subtracting annual
premium transmitted to the government from expected annual government outlays.
This metric measures the direct financial burden the program places on the govern-
ment. When assessing the effect of a pandemic insurance program on government net
outlays, one should compare all government outlays when the pandemic insurance
program is in place with government outlays in the absence of a pandemic insurance
program. Relevant to this comparison would be not only government outlays for insur-
ance claims but also any assistance the government might provide for uninsured losses.
For example, if the government covered all uninsured losses, the government would
ultimately cover all losses less the amount of claim payments funded by commercial
insurers. Direct government outlays through the program are a useful metric of pro-
gram performance, but a full analysis of the effect of program outlays would consider
the indirect effect of the program on government outlays outside the program. Such an
analysis is beyond the scope of this study.
Policyholder subsidy is defined as the ratio of premium to overall program costs
subtracted from 1.0 and captures the share of program costs covered by the private
sector as opposed to being subsidized by the government. As discussed previously in
this chapter, the government’s costs of operating its part of the program and costs of
financing government claim payments are not included in this measure.

Base-Case Parameter Values

We base the parameter values on the features of the proposed program (Table 2.1) and
on relevant economic and insurance industry data. The models are also based on a
number simplifying assumptions for each program. In some cases, these assumptions
were necessary because the program proposal was silent on the particular issue. In
others, we made assumptions to simplify comparisons across programs. For example,
the coverages provided by many of the proposals are similar but not identical. We stan-
dardized the coverage to allow use of a demand function that enables us to compare
take-up rates across programs and to better compare the strengths and weaknesses
of different approaches. Thus, we project outcomes for insurance programs that are
similar, but not identical, to those proposed. To streamline the following description
of the base-case parameter values, we sometimes refer to the BCPP, Chubb, and PRIA
approaches simply as BCPP, Chubb, and PRIA, respectively.
28 Improving the Availability and Affordability of Pandemic Risk Insurance

Economic Loss Parameters
We start by characterizing the overall size of the U.S. economy. We model the effects
of the proposed program on the approximately 6 million firms with employees in the
United States (Table 3.4). The U.S. Census Bureau’s Survey of U.S. Business reports
that payroll for these firms totaled $6.73 trillion in 2017. Small and medium-sized
firms (firms with less fewer 500 employees) accounted for 5.98 million of the 6 mil-
lion firms with employees and about approximately 40 percent of the payroll. Approx-
imately 20,000 firms had 500 or more employees (large firms) and accounted for
approximately 60 percent of the payroll ($199 million per firm).5
There are also many firms without employees.6 The U.S. Census shows 26.5 mil-
lion nonemployee firm establishments in 2018 (a firm can have multiple establish-
ments), with $1.29 trillion in receipts.7 We do not include these firms in our analysis.
Even though the $1.29 trillion in receipts for nonemployee firms is small relative to
the $37.4 trillion in receipts for firms with employees, providing revenue replacement
assistance for nonemployee firms may be an important part of a pandemic strategy;

Table 3.4
Distribution of Firms in the United States with Employees by Firm Size in 2017

Firms Payroll

Number of Number of Percentage of Payroll Percentage of Average Payroll
Employees Firms Total ($ billions) Total per Firm

1 to 4 3,698,086 61.7 277 4.1 74,904

5 to 9 1,009,851 16.8 255 3.8 252,512

10 to 19 631,981 10.5 338 5.0 534,826

20 to 99 544,485 9.1 928 13.8 1,704,363

100 to 499 92,358 1.5 914 13.6 9,896,273

500+ 20,139 0.3 4,013 59.7 199,265,108

Total 5,996,900 100.0 6,725 100.0 1,121,413

SOURCE: U.S. Census Bureau, 2020.
NOTE: Percentages are rounded.

5 U.S. Census Bureau, “Number of Firms, Number of Establishments, Employment, Annual Payroll, and Pre-
liminary Receipts by Enterprise Employment Size for the United States, All Industries: 2017,” Excel file, March 6,
2020.
6 Firms without employees are most concentrated in such industries as performing arts and passenger transpor-
tation but appear in practically every industry. For more information on nonemployer businesses, see U.S. Small
Business Administration, Office of Advocacy, “A Look at Nonemployer Businesses,” factsheet, August 2018.
7 U.S. Census Bureau, 2020.
Modeling Approach and Base-Case Parameters 29

however, analysis of extending the pandemic insurance programs to such firms was
beyond the scope of this study.
Data of the ratio of benefits to payroll are used to scale up payroll totals to include
employee benefits. According to the U.S. Bureau of Labor Statistics, benefits in the
private sector are 42 percent of wages, increasing the $6.73 trillion in payroll to $9.55
trillion.8
Internal Revenue Service Statistics of Income data for corporations provide a
basis for estimating ongoing operating expenses. Ongoing operating expenses cannot
be easily reduced when firm revenue declines if the firm expects to survive. Exam-
ples include rent; mortgage, insurance, and loan payments; and the costs of security
services and utilities. These data indicate that ongoing operating expenses account
for between 7 and 20 percent of revenue, depending on what expense categories are
included in the total.9 Applying the midpoint of this range to the $37.4 trillion in
revenue for firms with employees yields $5.05 trillion in annual ongoing operating
expenses (see Table 3.5).
Starting with these estimates of prepandemic economic activity, we use economic
data for the COVID-19 pandemic to provide an estimate of the payments needed to
maintain payroll and benefits and to cover the ongoing operating expenses needed
to stay in business. If the economy completely shuts down, $1.21 trillion per month
would be required to enable firms to keep payroll and benefits at prepandemic levels
and cover ongoing expenses.10 However, the economy will not completely shut down.
Many businesses will continue to operate, perhaps at reduced levels.
One measure of the economic impact of COVID-19 is the change in employ-
ment. Overall employment fell approximately 15 percent between February 2020
(pre–COVID-19) and April–May 2020.11 The federal PPP, passed as part of the Coro-
navirus Aid, Relief, and Economic Security Act (CARES Act), moderated employment
declines.12 Some researchers found that the PPP increased U.S. employment by about

8 U.S. Bureau of Labor Statistics, “Employer Costs for Employee Compensation—September 2020,” Decem-
ber 17, 2020.
9 The following expense categories are included for the lower estimate: repairs and maintenance, rent paid, taxes
and licenses, and interest paid. The upper estimate adds in the “other deductions” category. Internal Revenue
Service, Statistics of Income—2017: Corporation Income Tax Returns, Washington, D.C., 2020.
10 ($9.55 trillion in annual payments and benefits + $5.05 trillion in annual expenses)/12.

11 Michael Dalton, Elizabeth Weber Handwerker, and Mark A. Loewenstein, “Employment Changes by
Employer Size During the COVID-19 Pandemic: A Look at Current Employment Statistics Survey Microdata,”
Monthly Labor Review, U.S. Bureau of Labor Statistics, October 2020. The percentage drops in employment by
firm size reported in this article are combined with Survey of U.S. Business data on employment by firm size to
generate an estimate of the decline in overall employment (U.S. Census Bureau, 2020).
12 Pub. L. 116-136, Coronavirus Aid, Relief, and Economic Security Act, March 27, 2020.
30 Improving the Availability and Affordability of Pandemic Risk Insurance

Table 3.5
Base-Case Parameter Values for Modeled Programs

BCPP Base Chubb
Parameter Program Chubb BEIP Pandemic Re PRIA

Revenue Replacement Needed to Sustain Prepandemic Payroll, Benefits,
and Ongoing Expenses for 12 Months

Firms affected 6.00M 5.98M 20,000 6.00M

Annual payroll and $9.55T $3.82T $5.73T $9.55T
benefits

Annual ongoing $5.05T $2.02T $3.03T $5.05T
operating expenses

Revenue needed to $1.23T $0.493T $0.740T $1.23T
sustain prepandemic
payroll, benefits,
and expenses during
pandemic

Policy Features

Losses covered Payroll, benefits, Payroll, benefits, Payroll, benefits, Payroll, benefits,
and ongoing and ongoing and ongoing and ongoing
expenses expenses expenses expenses

Waiting period None None None None

Policyholder co-pay 20% 20% 20% 20%

Coverage duration 12 months 12 months 12 months 12 months

Policy limit 3 months of 3 months of $50M per firm 3 months of
payroll, benefits, payroll, benefits, payroll, benefits,
and ongoing and ongoing and ongoing
expenses expenses expenses

Program Features

Firms eligible 6.00M 5.98M 20,000 6.00M

Insurer deductible $0 $0 $0 $12B

Insurer co-pay 0% 6% up to $250B; 5% up to $300B; 5% of losses above
0% above $250B 0% above $300B insurer deductible

Limit on insured losses None $750B $400B $750B
(program cap)
Modeling Approach and Base-Case Parameters 31

Table 3.5—Continued

BCPP Base Chubb
Parameter Program Chubb BEIP Pandemic Re PRIA

Program Cost Assumptions

G&SE per policy for $100 $100 NA $100
small and medium size
firms

G&SE per policy for $10,000 NA $10,000 $10,000
large firms

LAE ratio 0.04 0.04 0.15 0.15

Cost of insurer capital NA 9% 9% 9%

Diversifiability index NA 0.5 0.5 0.5

Pandemic return period NA 5 years 5 years 5 years
used by insurers

Pandemic return period 200 years NA 5 years NA
used by the government

Take-Up Rate

BCPP take-up rate 65% NA NA NA

Price elasticity of NA –0.50 –0.50 –0.50
demand

NOTE: NA = not applicable.

2.3 million workers (or 1.4 percent of 2017 employment).13 Adding this effect to the
observed change in employment implies that COVID-19 would have reduced employ-
ment by 16.4 percent absent the PPP. Another broad measure of economic activity,
U.S. GDP, declined approximately 9 percent between the first quarter of 2020 and the
second quarter of 2020 before subsequently rebounding.14
Data on changes in firm revenue during the pandemic are not available. For this
analysis, we calculate the percentage drop in employment for each month between
March 2020 and February 2021 relative to February 2020. The result for each month
provides a measure of the proportion of February 2020 payroll that would have to be
provided that month to maintain payroll at February 2020 levels. Over the 12-month
period between March 2020 and February 2021, these changes in percentages sum
to 93 percent of February 2020 employment. Adding 1.4 percent per month for six

13 David Autor, David Cho, Leland D. Crane, Mita Goldar, Byron Lutz, Joshua Montes, William B. Peterman,
David Ratner, Daniel Villar, and Ahu Yildirmaz, “An Evaluation of the Paycheck Protection Program Using
Administrative Payroll Microdata,” unpublished manuscript, July 22, 2020.
14 U.S. Bureau of Economic Analysis, “Gross Domestic Product, Third Quarter 2020 (Advance Estimate),” web-
page, last modified October 30, 2020.
32 Improving the Availability and Affordability of Pandemic Risk Insurance

months to account for the effects of the PPP increases the total to 101.4 percent of
monthly employment. Applying this percentage to the estimate of monthly payroll,
benefits, and ongoing operating expenses means that required payments come to
approximately $1.23 trillion over a 12-month period (see Table 3.5).15

Policy Features
The policy offered to firms in each of the modeled programs covers payroll, benefits,
and ongoing operating expenses. The program proposals contain different provisions
for the policyholder co-pay. For example, the BCPP includes a 20 percent co-pay, while
a co-pay is not discussed in the Chubb proposal (see Table 2.1). To facilitate compari-
son of the different proposals, we assume a 20 percent co-pay for each program. As
discussed in Chapter Two, the coverage provided under PRIA would be the same as the
underlying BI policy, and a number of issues would need to be addressed before payroll,
benefits, and ongoing operating expenses during a shutdown would be guaranteed. We
assume that these issues are resolved and that the coverage purchased through PRIA
would provide payments needed to maintain payroll, benefits, and ongoing expenses.
Unlike standard BI policies, profits would not be covered.
The BCPP and Chubb BEIP for small and medium-sized firms set the policy
limit at three times monthly payroll, benefits, and ongoing expenses. Consistent with
the proposal, payments are capped at $50 million per firm in the Chubb’s Pandemic
Re program for large firms. As discussed in Appendix A, there is typically no dollar
limit on time-limited BI policies, and we do not include one for the PRIA program.

Program Features
The program features assumed in the simulations mirror those in the program propos-
als. The BCPP and PRIA are open to all firms, so all 6 million firms with employees
are eligible. The number of firms eligible for the two Chubb programs is based on the
number of firms in each firm-size category.
The insurer deductible in PRIA is 5 percent of the premium in specified insur-
ance lines. The direct written premium in these lines totaled $239 billion in 2019,
resulting in an industry deductible of $12 billion.16 Under PRIA, insurers are thus
responsible for the first $12 billion of insured losses. There is no insurer deductible in
the other programs. Above the deductible, insurers are responsible for 5 or 6 percent of
insured losses up to a specified threshold (the “Program Features” section of Table 3.5),
with the government covering the rest. Some programs cap the amount of insured

15 1.014*($9.55 trillion + $5.05 trillion)/12 = $1.23 trillion.

16 National Association of Insurance Commissioners data via S&P Global Intelligence, provided to RAND by
the American Property Casualty Insurance Association. The lines of business included in the total are aircraft,
allied lines, boiler and machinery, commercial multiple peril (liability and nonliability), excess workers’ compen-
sation, fire, inland marine, ocean marine, product liability, and workers’ compensation.
Modeling Approach and Base-Case Parameters 33

losses. PRIA caps insured losses at $750 billion per calendar year, and Pandemic Re
caps them at $400 billion. Consistent with TRIP, we assume in the base case that
the government will reimburse insurers for LAE on the portion of insured losses that
the government assumes.17 The sensitivity analysis will explore how much the results
change when this is not the case.

Program Cost Assumptions
General and Selling Expenses
The insurance industry generally reports G&SE as a ratio of G&SE to the premium.
The ratio was 0.176 for the P&C industry overall over the last five years (Table D.1 in
Appendix D).18 Using a G&SE ratio to calculate G&SE in this study presents chal-
lenges because we do not have a full premium to which to apply the G&SE ratio. The
premium in the different programs is subsidized to varying degrees. What is more,
different return periods are used in calculating the premium for the different pro-
grams, and these different return periods generate very different premium levels and
thus G&SE when the ratio of G&SE to the premium is constant. It seems reasonable
to expect, in contrast, that the G&SE should be insensitive to the return period that is
used to determine the premium.
A better approach in this case is to estimate the G&SE per policy. Data on G&SE
per policy are not generally available, so we piece together information from different
sources to come up with plausible estimates. Data available from the National Flood
Insurance Program (NFIP) indicate that G&SE average $199 per policy.19 The vast
majority of NFIP policies are written on residential structures. Data on the residen-
tial homeowners insurance market in California show G&SE of $222 per policy.20
These studies both suggest G&SE costs of approximately $200 for residential policies.

17 Insured loss in TRIP “includes reasonable loss adjustment expenses, incurred by an insurer in connection with
insured losses, that are allocated and identified by claim file in insurer records, including expenses incurred in
the investigation, adjustment, and defense of claims, but excluding staff salaries, overhead, and other insurer
expenses that would have been incurred notwithstanding the insured loss” (31 CFR 50.4[n]).
18 As shown in Appendix A, there was not a great deal of dispersion across lines, with the median ratio of 0.21
and 80 percent of the 30 lines falling between 0.166 and 0.370. The G&SE ratio for the line in which BI policies
are concentrated (commercial multiple peril, nonliability) is 0.242.
19 For the National Flood Insurance Program, agency commissions and selling expenses are 28.9 percent of
$3.39 billion in premium in 2019. Dividing the product by the 5.10 million National Flood Insurance Program
policies results in G&SE of $199 per policy (National Flood Insurance Program, Federal Emergency Manage-
ment Agency, The Watermark, Vol. 11, June 2020).
20 NAIC reported the G&SE for the homeowners multiple peril line as 26.2 percent of $8.07 billion in premium
in 2018, or $2.11 billion (National Association of Insurance Commissioners, Report on Profitability by Line by
State in 2018, Kansas City, Mo., 2019). Dixon, Tsang, and Fitts reported that there are approximately 9.5 million
homeowners policies in California (Lloyd Dixon, Flavia Tsang, and Gary Fitts, The Impact of Changing Wild-
fire Risk on California’s Residential Insurance Market, Sacramento: California Natural Resources Agency, August
2018). Dividing $2.11 billion by 9.5 million policies results in $222 per policy.
34 Improving the Availability and Affordability of Pandemic Risk Insurance

National Association of Insurance Commissioners (NAIC) data combined with rough
estimates of the number of business firms with insurance policies provide a rough
estimate of the G&SE per policy for commercial policies. NAIC reports $7.72 bil-
lion in G&SE for the commercial multiple peril, nonliability line in 2018, which is
the line in which the majority of BI policies are written. Approximately 3.8 million
business owner policies (BOPs) were in place in 2018.21 BOPs provide BI coverage and
are reported in the commercial multiple peril line. These two figures produce what is
likely an upper estimate on G&SE: approximately $2,030 per policy. If 5 million of the
6 million firms with employees have commercial multiple peril nonliability policies,
the average falls to approximately $1,540.
It seems reasonable to expect that the G&SE for the proposed pandemic risk
insurance programs will be lower than these estimates for commercial multiple peril
nonliability policies. The pandemic coverage will be added on to existing policy cover-
age, and the incremental cost might not be large. In addition, the streamlined nature
of the policy placement process should limit G&SE. We thus set G&SE at $100 per
policy for small and medium-sized firms in the base case. It is reasonable to expect
G&SE to be higher for large firms. Payroll per firm is approximately 450 times higher
at large firms than small and medium-sized firms. To reflect this large difference in
firm size, we assume in the base case that the G&SE amount is 100 times larger than
small and medium-sized firms—or $10,000 per firm.
We use the same G&SE per policy for all the programs. Chubb’s Pandemic Re
program and PRIA take a more traditional approach to underwriting and claim adjust-
ing than does the BCPP and Chubb BEIP. However, the pandemic insurance coverage
will typically be added to other coverage being provided, and the incremental G&SE
in Pandemic Re and PRIA will likely not be as large as the G&SE when writing the
coverage from scratch. Absent better information about the incremental G&SE for
these programs, we assume that the G&SE amount for the Chubb Pandemic Re and
PRIA programs is the same as for BCPP and BEIP.
LAE Ratio
The LAE ratio for the P&C industry as a whole was 0.179 over the last five years
(Table D.2 in Appendix D). The LAE ratio for the insurance line in which BI policies
are concentrated (commercial multiple peril, nonliability) is 0.11. The proportion of
losses from BI loss in that line is likely modest, however, and 0.11 might not be a good
estimate for BI policies. BI policies are typically indemnity policies and can require
detailed forensic investigation, generating substantial LAE. Data on the LAE for BI
indemnity policies are not available, but it stands to reason that the LAE ratio for BI
policies is higher than the 0.11 for the commercial multiple peril, nonliability line, as a
whole. On the basis of these considerations, we set the LAE ratio at 0.15 for indemnity

21 Figures provided to authors by Verisk’s Insurance Services Office via the American Property Casualty Insur-
ance Association.
Modeling Approach and Base-Case Parameters 35

policies in the base case. It remains possible, however, that insurers will develop more-
streamlined LAE for indemnity policies. It also possible that the surge in demand for
adjustors in a pandemic that affects the nation as whole pushes the LAE ratio above
the base-case value assumed (and results in substantial delays in claim payments). We
investigate the consequences of both lower and higher LAE ratios in Chapter Five.
The LAE ratio for the parametric policies envisioned in the BCPP and Chubb
BEIP proposals is expected to be much lower. Data on parametric policies that have
been used in other settings could provide the basis for estimating the LAE ratio in a
parametric pandemic risk insurance program. However, we have not been able to col-
lect such information. What is more, the experience in other settings might not be par-
ticularly informative because the number of policyholders involved in other settings is
typically small—unlike the potentially millions here. For this analysis, we set the LAE
ratio for parametric policies to 0.04, on the basis of the advice of experts we spoke with
during this study.22
Cost of Capital
In terms of the rate of return that insurers would be required to pay on capital held to
mitigate solvency risk, Kielholz found that the cost of capital for insurance companies
ranged from approximately 8 to 11 percent above the risk-free rate of return between
1992 and 1998.23 In a more recent study, Barinov et al. found that the cost of equity
capital for P&C insurers ranged from approximately 6 to 10 percent between 2009
and 2014.24 Because of these findings, we set the cost of capital to 9 percent in the base
case.
Diversifiability Index
The diversifiability index captures the extent that insurers can use surplus held for
other types of hazards to reduce the amount of capital they need to support pandemic
risk policies. The index varies between 0 and 1. A value of 0 means that that pandemic
risk has no diversification benefit to insurers and that decisions on how much capital
is required to support the coverage are made assuming that coverage for a pandemic is
a completely separate line of business. A value of 1 means that insurers believe that no
additional capital is needed to support the risk. Lacking better information, we set the

22 Table D.2 in Appendix D identifies a number on P&C insurance lines with LAE ratios in the neighborhood
of 0.04.
23 Walter Kielholz, “The Cost of Capital for Insurance Companies,” Geneva Papers on Risk and Insurance,
Vol. 25, No. 1, January 2000, p. 16. Kielholz’s analysis focused on the use and cost of equity. He pointed out that
although insurers do use debt financing, it is typically a small portion of the total capital structure (p. 5). This
observation, though, applies to the time when this article was written. We base our estimates of the cost of insurer
capital on the cost of equity capital.
24 Alexander Barinov, Jianren Xu, and Steven W. Pottier, “Estimating the Cost of Equity Capital for Insurance
Firms with Multiperiod Asset Pricing Models,” Journal of Risk and Insurance, Vol. 87, No. 1, 2020.
36 Improving the Availability and Affordability of Pandemic Risk Insurance

diversification factor at 0.5 and allow it to vary between 0.25 and 0.75 in the sensitiv-
ity analysis.
Pandemic Return Period
The pandemic return period plays a critical role in the cost and performance of the
proposed programs. The projected premium in each program requires an assumption
on the return period that insurers use in pricing the coverage or an assumption on the
return period the government uses in setting the premium—or both. In addition, the
evaluation of overall program performance requires an estimate of the actual overall
return period, which may differ from that used by insurers and the government in pric-
ing. We discuss each in turn.
There is a great deal of uncertainty about the return period for a pandemic the
size of COVID-19. Table 3.6 lists major contagious disease outbreaks since 1918 and
the number of deaths worldwide associated with each. Also reported are measures of
transmissibility and severity used by the Centers for Disease Control to characterize
events. The last event of comparable magnitude in the United States was the Spanish
flu in 1917 and 1918 (the 1918 pandemic); consequently, COVID-19 is often referred
to as a one-in-100-year event. If the 1957 Asian flu is also considered a major pan-
demic, the average return period based on these three events is 50 years. If the 1968
pandemic, which caused 1 million deaths, is also included, the average return period
declines to 33 years. Given increases in population density and global interconnected-
ness and a changing climate, major pandemics could become more frequent. However,
continued advances in vaccine technology and early detection capabilities could work
in the opposite direction.
During discussions with informed stakeholders, we heard that insurers will likely
assume a very short return period in setting premiums, at least initially. Part of the
reason is that COVID-19 and its variants will still be lurking for several years, making
resurgence more likely. A short return period may also be necessary if insurance regu-
lators are to approve insurer plans to raise equity capital or take on debt to support
the risk. In addition, a short return period may be necessary to convince investors to
provide capital at rates that have typically been paid by the industry (see the preceding
discussion of the cost of capital). Some of those with whom we conferred thought that
return periods might need to be as short as two or three years initially, lengthening if
there is no recurrence in the next few years. Chubb uses a 30-year return period in its
calculations of likely premiums in its proposed programs, but this longer return period
rests on the important assumption that the programs will not apply to SARS-CoV-2
virus or its variants.
We set the return period used by insurers in setting premiums at five years in the
base case. The return period used by insurers varies between three and 100 years in
the sensitivity analysis. The shorter return periods would be most relevant in the first
years after an insurance program is signed into the law, and the law applies to losses
Modeling Approach and Base-Case Parameters 37

Table 3.6
Major Contagious Disease Incidents Since 1918

Worldwide Measure of Measure of
Incident Year Deaths Transmissibilitya Severitya

1918 pandemic (H1N1) 1918 50Mb 5 7

Asian flu (H2N2) 1957 1.1Mb 4 4

1968 pandemic (H3N2) 1968 1.0Mb 4 3

SARS 2002 775c NA NA

H5N1 2005 282d NA NA

2007–2008 flu season 2007–2008 NA 2 3

2009 Pandemic (H1N1) 2009 0.4Mb 3 2

MERS 2012 858 e NA NA

Ebola 2014 11,325f NA NA

Zika 2015 51g NA NA

COVID-19 2020 3.4Mh 5 4–7

NOTES: Transmissibility is scored on a scale of 1 to 5 and severity is scored on a scale of 1 to 7. NA = not
available.
a Carrie Reed, Matthew Biggerstaff, Lyn Finelli, Lisa M. Koonin, Denise Beauvais, Amra Uzicanin,
Andrew Plummer, Joe Bresee, Stephen C. Redd, and Daniel B. Jernigan, “Novel Framework for Assessing
Epidemiologic Effects of Influenza Epidemics and Pandemics,” Emerging Infectious Diseases, Vol. 19,
No. 1, January 2013. Preliminary scores for COVID-19 are from André Ricardo Ribas Freitas, Marcelo
Napimoga, and Maria Rita Donalisio, “Assessing the Severity of COVID-19,” Epidemiologia e Serviços de
Saúde, Vol. 29, No. 2, April 2020.
b Centers for Disease Control and Prevention, National Center for Immunization and Respiratory
Diseases, “2009 H1N1 Pandemic (H1N1pdm09 Virus),” webpage, last reviewed June 11, 2019.
c Centers for Disease Control and Prevention, National Center for Immunization and Respiratory
Diseases, Division of Viral Diseases, “Frequently Asked Questions About SARS,” webpage, last reviewed
May 3, 2005.
d Centers for Disease Control and Prevention, National Center for Immunization and Respiratory
Diseases, “Highly Pathogenic Asian Avian Influenza A(H5N1) Virus,” webpage, last reviewed
December 12, 2018.
e These are deaths since 1982. World Health Organization, “Middle East Respiratory Syndrome
Coronavirus (MERS-CoV),” webpage, undated.
f Centers for Disease Control and Prevention, National Center for Emerging and Zoonotic Infectious
Disease, Division of High-Consequence Pathogens and Pathology, Viral Special Pathogens Branch,
“2014–2016 Ebola Outbreak in West Africa,” webpage, last reviewed March 8, 2019.
g Jaime A. Cardona-Ospina, Valentina Henao-SanMartin, Wilmer F. Acevedo-Mendoza, Katherinn
Melissa Nasner-Posso, Dayron F. Martínez-Pulgarín, Abril Restrepo-López, Valentina Valencia-Gallego,
Matthew H. Collins, and Alfonso J. Rodriguez-Morales, “Fatal Zika Virus Infection in the Americas: A
Systematic Review,” International Journal of Infectious Diseases, Vol. 88, November 2019.
h “The Covid-19 Tracker,” STAT, updated May 19, 2021.
38 Improving the Availability and Affordability of Pandemic Risk Insurance

due to COVID-19 and its variants. The longer return periods are most relevant to a
time after the current pandemic has passed or if the legislation excludes COVID-19
and its variants.
In setting a premium, the federal government might not consider only the best
available scientific estimates but would presumably also balance coverage affordabil-
ity with impact on the government budget. Budget impacts would include both the
direct cost of government-funded claim payments net of premium revenue and the
indirect reduction in postevent government assistance that would result from lower
premiums and higher take-up. Analysis of this trade-off is beyond the scope of study.25
When setting the premium for the BCPP, we assume that the government in effect
uses a 200-year return period. This return period results in a moderate premium in
the BCPP—a program in which government bears all the risk. In Chubb’s Pandemic
Re program, “both the insurance industry and the government are paid an appropriate
risk-adjusted price for pandemic cover.”26 We take this to mean that government will
follow the private sector in setting rates—and thus assume that the government uses a
five-year return period in the base case.

Take-Up Rate
Background
Relevant to the projection of take-up rate for the policies offered by the proposed pro-
grams are (1) the percentage of businesses with insurance prepandemic and (2) the per-
centage of firms with BI coverage. The experts with whom we conferred during this
study thought that nearly all but the smallest businesses have some type of insurance.
Insurance is required for loans and lines of credit, and landlords will typically require
renters to carry property and liability coverage. There is more uncertainty about insur-
ance take-up for microfirms (firms with one or two employees), and, given the large
number of such firms (see Table 3.4), it is difficult to estimate the take-up rate across
all firms.
There is also no systematic information about the percentage of firms with BI
insurance. Approximately 3.8 million BOPs were in place in 2018.27 BOPs are typi-
cally purchased by firms with fewer than 100 employees, and according to the Census
data, there are about 5.88 million firms with fewer than 100 employees.28 This sug-
gests a substantial take-up rate BI coverage, but, as discussed in Appendix A, there are

25 See Dixon et al., 2007, for examination of these direct and indirect effects in the context of TRIP.

26 Chubb, 2020, p. 4; italics added.

27 Figures provided to authors by Verisk’s Insurance Services Office via the American Property Casualty Insur-
ance Association.
28 U.S. Census Bureau, 2020.
Modeling Approach and Base-Case Parameters 39

conflicting views about whether firms can and how frequently they do opt out of the
BI coverage included in the standard BOP.29
Other data suggest that the share of small and medium-sized firms with BI cov-
erage is modest. A survey conducted by Harris Poll Online in 2015 found that only
34 percent of small U.S. businesses carried BI insurance.30 For the purposes of the poll,
small firms were defined as those with fewer than 300 employees. However, Harris
did not report the take-up rates by firm size, making the results difficult to interpret.
It may also be the case that many businesses, particularly smaller businesses, are not
fully informed about what their policies cover and might not report that they have BI
coverage when they actually do. Data from another source showed that 37 percent of
policies in commercial fire and allied lines included BI coverage in 2018.31 The size
distribution of firms with policies in this category is unknown, and it is possible that
these firms could carry other policies as well.
A potential anchoring point for the likely take-up of pandemic insurance poli-
cies is the take-up of terrorism risk insurance coverage, coverage that is subsidized by
TRIP. TRIP provides an example of what take-up can look like when the premium
is subsidized, as it is in the proposed pandemic insurance programs. The insurance
broker Marsh reported that 62 percent of its clients had property policies in 2018 that
provide terrorism coverage (so called embedded terrorism coverage).32 Marsh clients are
typically larger firms. The Federal Insurance Office reported that 79 percent of poli-
cies contained terrorism coverage in 2019 (this figure appears to include multiple peril
insurance policies into which terrorism coverage is embedded, as well as stand-alone
terrorism insurance policies).33 Both these studies consider the proportion of insurance
policies with terrorism coverage, not the proportion of firms with terrorism coverage.
One would expect the take-up rate for terrorism or pandemic insurance to be
driven at least in part by the extent to which the premium is subsidized. More firms
will likely buy coverage when the premium is low relative to the losses expected by the
firms. We have not investigated the subsidy rate in TRIP, but for completeness and
comparison with the projections in Chapter Four, we review the data available on the
cost of terrorism coverage for certified events in the United States. In cases when the
policy covers both terrorism and other perils (“embedded policies”), the Federal Insur-
ance Office found that insurers provide terrorism coverage at no additional charge in

29 Further complicating the situation is the possibility that some small firms may purchase insurance but not a
BOP and that some firms without employees could purchase a BOP.
30 Businesswire, “Most Small Business Owners at Risk for a Disaster,” August 31, 2015.

31 Figures provided to authors by Verisk’s Insurance Services Office via the American Property Casualty Insur-
ance Association.
32 Marsh, 2019 Terrorism Risk Insurance Report, New York, May 2019, p. 8.

33 Federal Insurance Office, U.S. Department of the Treasury, Report on the Effectiveness of the Terrorism Risk
Insurance Program, Washington, D.C., June 2020, p. 29.
40 Improving the Availability and Affordability of Pandemic Risk Insurance

policies accounting for 30 percent of direct earned premium. When terrorism coverage
adds to the cost of the policy, terrorism coverage accounted for 2.5 percent of the direct
earned premium in 2019.34 In terms of the price of coverage relative to the amount
offered, the premium was $0.015 per $1,000 coverage for embedded policies and $0.44
per $1,000 coverage for stand-alone policies in 2019.35
The developers of the different program proposals have varying expectations for
the resulting take-up rate. As reported in Chapter Two, Chubb projects that 90 percent
of small and medium-sized firms will participate in its program and 30 percent of large
firms. Recall that the program features imply that premiums will be higher for large
firms per $1,000 of coverage than for the small firms in the Chubb proposal. Zurich
expects take-up to vary by industry and to average 65 percent across the economy as
a whole, with premiums at $20 to $30 per $1,000 of coverage. Take-up rates are not
projected in either the BCPP or the PRIA proposal.
Base-Case Assumptions
We assume that in the BCPP the government will set premium at a level that results
in fairly widespread take-up. Following the experience with TRIP, we assume that the
BCPP take-up rate will be 65 percent at the heavily subsidized premium projected for
the BCPP.

Price Elasticity of Demand
Previous research does not provide a great deal of direction in setting the elasticity of
demand for payroll, benefits, and ongoing operating expense coverage. Michel-Kerjan
and coauthors found that the elasticity of demand for terrorism coverage for busi-
nesses with over $1 million in total insured value ranged from –0.11 to –0.25.36 How-
ever, their analysis examined the amount of terrorism coverage purchased conditional
on the purchase of terrorism coverage. In contrast, the elasticity of concern for our
analysis is the relationship between the percentage change in the number of firms that
purchase coverage to the percentage change in premiums. Their analysis also excluded
smaller firms.
Price elasticities in other settings are potentially also relevant. The price elasticity
of demand for crop insurance by farmers has been found to range from –0.30 to –0.90.
The insurance purchase decisions of smaller businesses may look more like those of

34 Federal Insurance Office, U.S. Department of the Treasury, 2020, pp. 20–21.

35 Federal Insurance Office, U.S. Department of the Treasury, 2020, p. 28.

36 Erwann Michel-Kerjan, Paul Raschky, and Howard Kunreuther, “Corporate Demand for Insurance: New
Evidence from the U.S. Terrorism and Property Markets,” Journal of Risk and Insurance, Vol. 82, No. 3, Septem-
ber 2015, p. 510. An earlier Wharton School study put the price elasticity of demand for the amount of terrorism
coverage purchased at –0.65. See Howard Kunreuther and Erwann Michel-Kerjan, TRIA and Beyond: Terrorism
Risk Financing in the U.S., Philadelphia: Wharton Risk Management and Decision Process Center, University of
Pennsylvania, August 2005, p. 178.
Modeling Approach and Base-Case Parameters 41

individuals than large businesses, so individual behavior may also be relevant. Esti-
mates of the demand for residential flood insurance vary widely, ranging from –0.62
to –4.48, and one study put the elasticity of demand for residential hurricane insurance
at –1.9.37
Previous research suggests that the business demand for insurance is inelastic,
but the appropriate value when considering pandemic insurance is highly uncertain.
Considering the range of estimates for terrorism and crop insurance, we set the elastic-
ity of demand for pandemic risk insurance at –0.50 in the base case. This means that
a 1 percent increase in price induces a 0.5 percent drop in the number of firms that
purchase pandemic business expense coverage.

Limitations

We examine the U.S. economy as a whole and do not break our analysis down by eco-
nomic sector or by insurer size. Our high-level overview of program performance is a
useful first step and can be viewed as examining average effects across economic sectors
and types of insurers. Exploring how effects might vary by economic sector would be
a productive area for further analysis.
We have filled in several gaps in the proposed programs. For example, we have
assumed that PRIA legislation is written so that the underlying BI policies cover losses
from contagious disease during a public health emergency even when there is not phys-
ical damage to the insured property. We have also assumed that parametric policies
succeed in sending the right amount of compensation to the firms that need it. It
remains to be seen whether these issues are resolved in practice.
Projections on take-up rate for the Chubb and PRIA programs should be inter-
preted as relative to the take-up rate in the BCPP. Given the extent of the subsidy in
the BCPP program, expectations of the program proposers, and review of take-up in
other settings, we assume that take-up in the BCPP would be 65 percent. The take-up
rates for the other programs are then adjusted from the 65 percent baseline based on
the relation of their premiums to that predicted for the BCPP. Whether the 65 percent
take-up rate for the BCPP is realistic will be better understood as more experience is
developed with pandemic risk insurance.
Finally, there is substantial uncertainty about the appropriate values of many of
the input parameters. For example, the pandemic return period that insurers use in set-
ting premiums is a fundamental determinant of program outcomes and will depend on
a range of factors, including whether the program covers losses due to COVID-19 and
its variants. In the next chapter, program performance is projected under a plausible
set of base-case parameters, and the analysis provides insights into key factors driving

37 Michel-Kerjan, Raschky, and Kunreuther, 2015, p. 510.
42 Improving the Availability and Affordability of Pandemic Risk Insurance

performance. However, it should be kept in mind that the base case predicts perfor-
mance for only one set of parameter values—presumably for a case that is highly plau-
sible. Attention should be paid to how results may differ as parameter values change.
In Chapter Five, we explore the sensitivity of the outcomes to changes in several key
parameters.
CHAPTER FOUR

Projected Program Performance in the Base Case

In this chapter, we present results for the modeled versions of the BCPP, Chubb, and
PRIA proposals using the base-case set of parameter values. Importantly, insurers use a
five-year pandemic return period in the base case. Such an assumption is most appro-
priate for a pandemic insurance program that applies to COVID-19 and its variants.
Projections using different return periods are presented in the following chapter. The
modeled programs are motivated by, but not identical to, the proposed programs. To
streamline the following presentation, we sometimes refer to the BCPP, Chubb, and
PRIA approaches simply as BCPP, Chubb, and PRIA, respectively.

Affordability
Affordability for Small and Medium-Sized Firms
The annual premium projected for small and medium-sized firms in the BCPP
approach is $442 (top panel of Table 4.1). This premium is due to the government
pricing coverage at levels that ensure widespread take-up, and it assumed that this
price is sufficiently low to induce 65 percent take-up. The $442 premium results from
what amounts to the government using a one-in-200-year return period in pricing the
coverage.
For comparison, the BCPP premium is low relative to the price of workers’ com-
pensation coverage but high relative to terrorism coverage backstopped by TRIP.
Once normalized by payroll, the projected premium is $0.98 per $1,000 payroll (see
Table 4.1). In contrast, workers’ compensation coverage averaged $12.10 per $1,000
in the United States in 2018.1 The Federal Insurance Office puts the cost of terrorism
coverage when embedded with other coverage at $0.015 per $1,000 coverage,2 far lower
than the $2.26 reported in Table 4.1.

1 National Academy of Social Insurance, Workers’ Compensation: Benefits, Coverage, and Costs, Washington,
D.C., November 2020.
2 Federal Insurance Office, U.S. Department of the Treasury, 2020, p. 28.

43
Table 4.1 44
Program Performance at Base-Case Parameter Values

Improving the Availability and Affordability of Pandemic Risk Insurance
BCPP Chubb PRIA
Approach Approacha Approach

Small and Small and Small and
Medium- Large All Medium-Sized Large All Medium- All
Performance Metric Sized Firms Firms Firms Firms Firms Firms Sized Firms Large Firms Firms

Affordability

Annual premium per $442 $164,000 $986 $1,070 $6.86M $6,580 $1,630 $696,000 $3,950
insured firm

Total annual premium $1.72Ba $2.13Ba $3.84B $2.67Bb $13.8Bc $16.5B $3.17B $4.52B $7.69Bb

Annual premium per $0.98 $0.80 $0.88 $2.38 $34.00 $2.45 $3.62 $3.45 $3.52
$1,000 payroll

Premium per $1,000 $2.26 $1.86 $2.02 $5.46 $137 $27.9 $8.32 $7.93 $8.08
coverage

Efficacy

Loss for all firms should an $492B $738B $1.23T $492B $738B $1.23T $492B $738B $1.23T
event occur

Insured loss for all insured $256B $384B $640B $164B $59.3B $ 242B $128B $192B $320B
firms should an event
occur

Insured loss per insured $65,800 $29.5M $164,000 $65,800 $29.5M $89,500 $65,800 $29.5M $164,000
firm should an event occur

Percentage of loss insured 80% 80% 80% 80% 80% 80% 80% 80% 80%
for firms with insurance

Take-up rate 65% 65% 65% 42% 10% 42% 32% 32% 32%

Percentage of loss insured 52% 52% 52% 33% 8% 18% 26% 26% 26%
Table 4.1—Continued

BCPP Chubb PRIA
Approach Approacha Approach

Small and Small and Small and
Medium- Large All Medium-Sized Large All Medium- All
Performance Metric Sized Firms Firms Firms Firms Firms Firms Sized Firms Large Firms Firms

Efficiency

LAE per participating firm $2,630 $1.18M $6,560 $2,630 $4.43M $6,190 $9,870 $4.43M $24,600
should an event occur

Annual G&SE per insured $100 $10,000 $133 $100 $10,000 $108 $100 $10,000 $133
firm

Annual capital cost per $0 $0 $0 $148 $64,300 $199 $233 $105,000 $582
insured firm

Risk borne by commercial

Projected Program Performance in the Base Case
insurers

Insurer-incurred loss and
LAE should an event occur

Per participating firm $0 $0 $0 $4,110 $1.70M $5,470 $6,480 $2.91M $16,200

Total program $0 $0 $0 $10.3B $3.41B $13.7B $12.6B $18.9B $31.5B

Amount of capital held by
insurers for solvency risk

Per participating firm $0 $0 $0 $1,640 $714,000 $2,220 $2,590 $1.16 M $6,460

Total program $0 $0 $0 $4.11B $1.44B $5.54B $5.04B $7.56B $12.6B

Insurer share of loss and 0% 0% 0% 6.0% 5.0% 5.7% 8.6% 8.6% 8.6%
LAE should an event occur

45
Table 4.1—Continued 46

Improving the Availability and Affordability of Pandemic Risk Insurance
BCPP Chubb PRIA
Approach Approacha Approach

Small and Small and Small and
Medium- Large All Medium-Sized Large All Medium- All
Performance Metric Sized Firms Firms Firms Firms Firms Firms Sized Firms Large Firms Firms

Government outlays and
revenue

Government-incurred loss
and LAE should an event
occur

Per participating firm $68,500 $30.7 M $171,000 $64,300 $32.3M $90,200 $69,200 $31.0 M $172,000

Total program $266B $399B $665B $1.61B $64.8B $226B $134B $202B $336B

Annual government
premium revenue

Per participating firm $342 $154,000 $853 $0 $6.51M $5,230 $0 $0 $0

Total program $1.33B $2.00B $3.33B $0 $13.1B $13.1B $0 $0 $0
a For the initial year of the Chubb program.
Projected Program Performance in the Base Case 47

Also relevant in evaluating the magnitude of this premium is the typical price of
the property and liability coverage purchased by small and medium-sized businesses.
However, systematic statistics on annual premium are not readily available. In the
absence of such data, we provide three very rough estimates. First, Insureon reports
that the average price of a BOP is approximately $1,200 annually for small businesses.3
Insureon does not report the size of the firms in its sample, so it is difficult to compare
the findings with our projections for firms with fewer than 500 employees. Because
BOPs provide both property and liability coverage, Insureon’s average price presum-
ably includes both property and liability coverage, but it is unclear whether it includes
the price of BI coverage.4
Second, NAIC data on direct written premiums for commercial multiperil poli-
cies can be used to develop a rough estimate of the price of commercial multiperil
policies. As shown in Appendix D, direct written premiums in this line (liability and
nonliability portions combined) averaged approximately $41 billion per year between
2015 and 2019. Using the same rough estimates of the number of policies written as
in the “General and Selling Expenses” section of Chapter Three (3.8 to 5.0 million)
results in a range of $8,200 to $10,800 per policy. Policies in this line are typically held
by small and medium-sized firms, but again the distribution of firm size in the com-
mercial multiple peril line is unknown.
Finally, we constructed a rough estimate of the average insurance cost for firms
with fewer than 500 employees. Using the Insureon data and input from experts with
whom we conferred during the course of this study, we set the average BOP insurance
premium (including BI coverage) at $1,200 for firms with fewer than ten employees.
We then scaled up the premium for larger firm-size categories using average payroll per
firm and then weighted the results by the number of firms in each firm-size category.
The average premium across all firms with fewer than 500 employees comes to $4,800
per firm.
Considerable uncertainty remains on the average cost of property and liability
insurance for firms with fewer than 500 employees. But if the average property and
liability insurance cost for business with fewer than 500 employees falls somewhere
between $5,000 and $10,000 per firm, the $442 premium for the BCPP would add
between 4.4 and 8.8 percent to the cost of the coverage.
The projected annual premium is considerably higher for small and medium-
sized firms in the Chubb and PRIA approaches ($1,070 and $1,630, respectively).
These higher premiums are due in important part to the five-year return period that
insurers are assumed to use in setting the premium for the portion of the risk that they

3 Insureon, “How Much Does a Business Owner’s Policy Cost?” webpage, undated-a.
4 When describing business interruption, Insureon says that the cost of adding BI coverage is $500 to $1,000
per year. See Insureon, “How Much Does Business Interruption Insurance Cost?” webpage, undated-b.
48 Improving the Availability and Affordability of Pandemic Risk Insurance

bear. As will be discussed in the following section, these higher premiums result in
much lower take-up rates.

Affordability for Large Firms
The premiums for large firms are comparable to those for small firms in the BCPP
and PRIA approaches when normalized by payroll (premium per $1,000 payroll). Due
to the market-based rate charged for the risk borne by the government, the premium
charged by Chubb’s Pandemic Re is projected to be much higher ($6.86 million per
firm and $34 per $1,000 payroll) than in the other two approaches.

Efficacy

As shown in the “Efficacy” section of Table 4.1, the assistance needed to maintain
payroll, benefits, and ongoing operating expenses during a 12-month pandemic event
is constant across modeled programs at $1.23 trillion. Small and medium-sized firms
purchasing coverage receive $65,800 on average in all three programs, and large firms
receive $29.5 million.
The percentage of firms purchasing coverage (take-up rate) varies across the pro-
grams because of difference in premium. The BCPP is the reference approach, with
take-up for both small and medium-sized and large firms assumed to reach 65 percent
at the premium charged by the program. The higher premium for Chubb’s Pandemic
Re approach drives its take-up rate down to 42 percent for small and medium-sized
firms and 10 percent for large firms. Take-up for both firm size categories is projected
to be 32 percent in PRIA under the base-case parameter assumptions. The take-up
rates for the PRIA and Chubb approaches should be evaluated relative to the BCPP.
Multiplying the fraction of losses covered given a firm purchases coverage by the
take-up rate results in the percentage of the overall $1.23 trillion in event losses cov-
ered by the program. At 18 percent, the percentage is lowest for the combined Chubb
approach, but it is not high for any of the approaches.

Efficiency

There is considerable LAE variation across the approaches, driven by the different
loss adjustment approaches (see the “Efficiency” section of Table 4.1). LAE per par-
ticipating firm is lowest for the BCPP, which relies exclusively on parametric policies
(averaging $6,560 across all participating firms); it is approximately four times higher
for PRIA, which relies solely on indemnity policies ($24,600 across all participating
firms). LAE for the combined Chubb approach falls in the middle, with the BEIP for
Projected Program Performance in the Base Case 49

small and medium-sized firms using a parametric policy and Pandemic Re using an
indemnity approach.
The amount of annual G&SE per participating firm differs by firm size but is not
assumed to vary across approaches. This annual G&SE figure is not large relative to
LAE when an event occurs, but there are annual recurrences of G&SE, while an LAE
incurrence happens only when there is a loss.
The approaches also vary considerably in the annual cost of capital per participat-
ing firm. There is no capital cost in the BCPP because insurers bear no risk, and we do
not consider any costs to the government in financing the losses it incurs. Capital costs
are highest in PRIA ($582 per participating firm), consistent with the relatively high
amount of risk borne by the commercial insurers in that approach (discussed further
later in this chapter).
A measure of overall approach efficiency is the ratio of expected annual claim
payments to total expected annual program costs. Total expected annual program
costs consist of annual G&SE, annual capital costs, and expected annual claim pay-
ments and LAE. A pandemic return period must be assumed to calculate expected
annual claim payments and expected annual program costs. And given the consider-
able uncertainty about the return period for a major pandemic, we calculate the effi-
ciency for different pandemic return periods. The pandemic return period can differ
from the return periods used by insurers and the government in setting premiums, and
we refer to it as the actual return period. Figures 4.1 and 4.2 show program efficiency
for small and medium-sized firms and large firms, respectively. As can be seen, the
BCPP is the most efficient of the three approaches because of the reliance on para-
metric policies and the absence of a capital cost. Program efficiency hovers around
95 percent at low return periods and declines gradually for both firm-size categories as
the actual return period increases because G&SE and capital costs are spread over less
frequent claim payments. The efficiency of the Chubb and PRIA approaches are lower
than that of the BCPP, first because both include a capital cost for the risk borne by the
commercial insurers and second because indemnity policies are used for large firms in
the Chubb proposal and for all firms in the PRIA proposal.
Although a full assessment of the administrative and loan processing costs of
the PPP has not been completed, initial information from the PPP provides a useful
reference point for interpreting the efficiency projections in Figures 4.1 and 4.2. The
U.S. Small Business Administration paid lenders between 1 and 5 percent in the first
PPP draw, depending on the loan amount,5 and analysis by the New York Times sug-
gests that processing fees amount to between 2.6 and 4.0 percent of the loan amount.6

5 U.S. Small Business Administration, “Second Updated Paycheck Protection Program Lender Processing Fee
Payment and 1502 Reporting Process,” Procedural Notice 5000-20091, February 8, 2021.
6 Cowley estimated that banks fees totaled between $13.7 and $20.9 billion on $525 billion in loans. Stacy
Cowley, “Despite Billions in Fees, Banks Predict Meager Profits on P.P.P. Loans,” New York Times, October 1,
2020.
50 Improving the Availability and Affordability of Pandemic Risk Insurance

Figure 4.1
Program Efficiency for Small and Medium-Sized Firms at Base-Case Parameter Values

100

Program efficiency (%)
70

30 BCPP
Chubb
20 PRIA

0
0 10 20 30 40 50 60 70 80 90 100
Actual return period (years)

Figure 4.2
Program Efficiency for Large Firms at Base-Case Parameter Values

100

Program efficiency (%)
70

30 BCPP
Chubb
20 PRIA

0
0 10 20 30 40 50 60 70 80 90 100
Actual return period (years)
Projected Program Performance in the Base Case 51

The program efficiency for the BCPP is comparable to PPP administrative costs that
have been measured to date for small and medium-sized firms when the return period
is shorter than approximately 30 years and for large firms across all return periods
examined.

Risk Borne by Commercial Insurers

The programs vary greatly in the amount of risk borne by the commercial insurers,
as measured by loss payments plus LAE incurred by commercial insurers should an
event occur. No risk is borne by the commercial insurers in the BCPP, compared
with $13.7 billion and $31.5 billion in Chubb and PRIA approaches, respectively (the
“Risk Borne by Commercial Insurers” section of Table 4.1).7 The $12 billion industry
deductible and large LAE are important contributors to the higher insurer outlays in
PRIA. Nevertheless, insurers bear a relatively small share of loss payments and LAE
should an event occur, ranging from zero in BCPP to 8.6 percent in PRIA.
The amount of capital that the private sector holds to mitigate solvency risk mir-
rors the amount of private-sector risk. The cost of this capital is the cost of requiring
insurers to bear risk. No capital is required for the BCPP, $5.54 billion is required for
the combined Chubb approach, and $12.6 billion is required for the PRIA approach.
Recall that the diversifiability index is set to 0.5 in the base case, which means that
insurers need to hold half as much capital as would be the case if they treated the pan-
demic coverage as entirely separate from insurance for other hazards. For comparison,
the surplus of the P&C industry was $867 billion at the end of 2019.8 But this surplus
covers both commercial and personal insurance lines and includes surplus reported by
Berkshire Hathaway that covers its noninsurance business. The surplus available for
commercial lines is on the order of $325 billion once these two factors are considered.9

7 Insurers participating in the BCPP, as well as in the other proposed programs, may face risk that the govern-
ment will delay or fail to reimburse insurers for claim payments and expenses. Such risk is not considered in this
analysis.
8 American Property Casualty Insurance Association, “Property Casualty Insurance Industry Suffered Largest-
Ever Drop in Surplus in the First Quarter of 2020,” July 28, 2020.
9 To make this adjustment, the ratio of Berkshire’s surplus for its insurance operations to direct written pre-
mium is assumed to be equal to that for the other P&C insurers (0.98). Surplus is then allocated to commercial
lines based on the ratio of direct written premium for commercial lines to total direct written premium. Calcula-
tions are based on data from Verisk’s Insurance Services Office via the American Property Casualty Insurance
Association.
52 Improving the Availability and Affordability of Pandemic Risk Insurance

Expected Annual Government Outlays

The last section of Table 4.1 reports government outlays for claim payments and LAE
should an event occur, as well as projected annual government premium revenue. The
government receives the premium less G&SE in the BCPP and a large share of the pre-
mium in Pandemic Re. The government receives no revenue to offset the risk it bears
in the Chubb BEIP and PRIA proposals.
Figure 4.3 shows expected annual government net outlays per participating firm
(expected loss and LAE less premium revenue) for small and medium-sized firms. The
net outlays per firm turn out to be very similar across all three approaches, even though
the approaches have very different pricing and risk-sharing approaches. Net govern-
ment outlays fall as the actual return period lengthens because expected loss and LAE
fall while the premium the government receives (if any) is not affected by the actual
return period.
The programs can be costly to the government. For example, if the actual return
period is 15 years, the expected annual net government outlays for small and medium-
sized firms in the BCPP is approximately $17 billion per year (see Figure 4.4). The lower
take-up rates for small and medium-sized firms in the Chubb and PRIA approaches
result in the lower expected annual government net outlays for those programs.
The projections of net government outlays for large firms in BCPP and PRIA
follow patterns similar to those for small and medium-sized firms. Because Pan-

Figure 4.3
Expected Annual Government Net Outlays per Participating Firm for Small and Medium-
Sized Firms at Base-Case Parameter Values

25,000

Expected annual government net outlays
BCPP
20,000 Chubb
PRIA

15,000

per participating firm ($)
10,000

5,000

0
0 10 20 30 40 50 60 70 80 90 100
Actual return period (years)
Projected Program Performance in the Base Case 53

Figure 4.4
Expected Annual Government Net Outlays for Small and Medium-Sized Firms at Base-Case
Parameter Values

100

Expected annual government net outlays
90
BCPP
80 Chubb
PRIA
70

($ billions)
40

0
0 10 20 30 40 50 60 70 80 90 100
Actual return period (years)

demic Re charges a market-based premium and insurers are assumed to use a five-year
return period in the base case, government net outlays will be negative (government
runs a surplus) if the actual return period exceeds five years (see Figures 4.5 and 4.6).
Figure 4.7 reports expected net government outlays for both firm-size categories
combined. Expected annual net government outlays can be very large when the return
period is short. For example, if the actual return period is 15 years, the expected annual
net government outlays for the BCPP is approximately $41 billion per year. For com-
parison, the PPP lent a total of $718 billion through March 21, 2021, with loans aver-
aging $87,000 across the 8,246,234 loans that were made.10 Care should be taken in
comparing the PPP figures with the expected annual outlays in Figure 4.7.

Policyholder Subsidy

As shown in Figure 4.8, the premiums for small and medium-sized firms are heavily
subsidized in all three programs for shorter actual return periods. Subsidies are highest
in the BCPP approach, followed by Chubb and PRIA. For actual return periods of 30
years of less, the subsidies for all approaches are considerable. For example, the subsidy

10 U.S. Small Business Administration, “PPP Data: Find Data and Reports Related to the Paycheck Protection
Program (PPP),” webpage, undated.
54 Improving the Availability and Affordability of Pandemic Risk Insurance

Figure 4.5
Expected Annual Government Net Outlays per Participating Firm for Large Firms at Base-
Case Parameter Values

Expected annual government net outlays
10
BCPP
8 Chubb
PRIA
6

per participating firm ($ millions)
0

–2

–4

–6

–8
0 10 20 30 40 50 60 70 80 90 100
Actual return period (years)

Figure 4.6
Expected Annual Government Net Outlays for Large Firms at Base-Case Parameter Values

140

Expected annual government net outlays
120 BCPP
Chubb
100 PRIA

($ billions)
40

–20
0 10 20 30 40 50 60 70 80 90 100
Actual return period (years)
Projected Program Performance in the Base Case 55

Figure 4.7
Expected Annual Government Net Outlays for All Firms at Base-Case Parameter Values

250

Expected annual government net outlays
BCPP
200
Chubb
PRIA
150

100

($ billions)
50

–50
0 10 20 30 40 50 60 70 80 90 100
Actual return period (years)

Figure 4.8
Policyholder Subsidy for Small Firms

100

BCPP

Percentage of program costs subsidized
80 Chubb
PRIA
60

–20

–40

–60
0 10 20 30 40 50 60 70 80 90 100
Actual return period (years)
56 Improving the Availability and Affordability of Pandemic Risk Insurance

for the BCPP at a 30-year return period is 81 percent, which means that the full cost of
the program (for which small and medium-sized firms pay $442) is $2,330.11 Subsidies
decline and ultimately disappear as the actual return period increases. For example, the
subsidy in the Chubb approach for small and medium-sized businesses disappears once
the return period reaches approximately 85 years.
For large firms, the findings on subsidies are similar to those for small and
medium-sized firms for the BCPP and PRIA approaches (see Figure 4.9). Subsidies
occur in Chubb’s Pandemic Re only if the actual return period turns out to be less than
the five years that insurers are assumed to use in the base case when setting premiums.

Discussion

The relative performance of each of the modeled programs under the base-case param-
eter assumptions is summarized in Table 4.2. For small and medium-sized firms,
BCPP performs best in terms of affordability, efficacy, and efficiency. PRIA does the
best in terms of expected annual net government outlays and policyholder subsidy. It
also does best in terms of risk borne by commercial insurers, assuming that best in this
context means the most risk borne by commercial insurers. For large firms, the BCPP
again performs best in terms of affordability, efficacy, and efficiency. PRIA still trans-

Figure 4.9
Policyholder Subsidy for Large Firms

100
0

Percentage of program costs subsidized
–100
–200
–300 BCPP
–400 Chubb
–500 PRIA
–600
–700
–800
–900
–1,000
–1,100
–1,200
–1,300
–1,400
–1,500
–1,600
0 10 20 30 40 50 60 70 80 90 100
Actual return period (years)

11 Program costs = premium/(1 – subsidy).
Projected Program Performance in the Base Case 57

Table 4.2
Ranking of Modeled Programs by Program Metric at Base-Case Parameter Values

Small and Medium-Sized Firms Large Firms

BCPP Chubb PRIA BCPP Chubb PRIA
Program Metric Approach Approach Approach Approach Approach Approach

Affordability Most Mid Least Most Least Mid

Efficacy Highest Mid Lowest Highest Lowest Mid

Efficiency Highest Mid Lowest Highest Mid Lowest

Risk borne by commercial Least Mid Most Least Mid Most
insurers

Expected annual net Highest Mid Lowest Highest Lowest Mid
government outlays

Policyholder subsidy Most Mid Least Most Least Mid

NOTE: Shading indicates the program that performs best on the metric.

fers the most risk to commercial insurers, but the Chubb approach now generates the
lowest expected annual government net outlays and policyholder subsidy.
The greater efficiency of the BCPP and the ability of the government to price cov-
erage using a much longer return period than the private sector results in the BCPP’s
lower premium. Although the $442 premium projected for small and medium-sized
firms in the BCPP seems modest, it remains to be seen whether 65 percent of firms
would buy coverage at this rate, as assumed in the base case. Very rough calculations
suggest that premiums at this level could increase the cost of commercial multiperil
policies on the order of 4.4 to 8.8 percent. Some firms may be reluctant to purchase
coverage at these rates, remembering the large government-funded PPP.
None of the programs does a particularly good job in terms of the percentage
of losses reimbursed over a six-month pandemic (efficacy). This is largely because of
the moderate to low take-up rates and the policyholder co-pay. The BCPP does best
because its lower premium results in the highest take-up rate. Take-up and the percent-
age of losses reimbursed for the large firms in Chubb’s Pandemic Re proposal are par-
ticularly low. Large firms presumably have a better ability to weather revenue declines
than smaller firms, but the fact remains that large firms account for 60 percent of
payroll at firms with employees, and their financial health will have important conse-
quences for overall employment levels during a pandemic.
Program efficiency as we have defined it is driven by G&SE, LAE, and private-
sector capital costs. The BCPP does best on this metric partly because it relies on para-
metric policies to reduce costs and to address the practical problem of simultaneously
adjusting an enormous number of claims. The BCPP also fares best in part because
it does not require insurers to hold capital to protect against solvency risk. The cost
of such capital is the cost of transferring risk from the public to commercial insurers.
58 Improving the Availability and Affordability of Pandemic Risk Insurance

Higher capital costs reduce efficiency but are associated with lower government outlays
should an event occur and with other potential benefits that come with greater private-
sector skin in the game. These other potential benefits include greater concern about
fraudulent claims and the development of the expertise and experience needed to grow
the private sector’s ability to write pandemic risk. When deciding whether to proceed
with a pandemic risk insurance program, program efficiency should be compared with
that of other approaches, such as the PPP. We found the efficiency of the BCPP com-
pares favorably with the administrative costs of the PPP, but unless pandemic return
periods are short, efficiencies of the other programs do not compare favorably.
As shown in Table 4.2, the amount of risk borne by commercial insurers is high-
est in PRIA but, reflecting the difficulty of insuring pandemic risk, is not particularly
high in any of the programs. Chubb proposes that the insurer risk share double over
time—but even so, it will remain modest. One advantage of the insurer deductible
in the PRIA approach is that the share of losses borne by insurers would be higher in
smaller or better-contained pandemics. The capital required to hedge solvency risk is
not large relative to industry surplus for the Chubb and PRIA approaches (3.8 percent
of an estimated $325 billion for commercial lines), but further analysis is required to
determine how difficult it would be for the industry to raise this amount of capital.
Illustrating the usual trade-off between affordability and government cost, the
BCPP approach does best on affordability but worst on expected government annual
net outlays and policyholder subsidy. Our measure of government net outlays consists
of only the direct revenues and outlays of the program. As discussed in Chapter Two,
indirect costs, such as an increased demand for government assistance during an event
when there is limited insurance coverage, should also be considered when assessing the
full effect of a pandemic insurance program on overall government outlays. Such an
analysis is beyond the scope of this study, but the amount of risk borne by commercial
insurers is likely a better indicator of overall net government outlays than direct gov-
ernment outlays through the pandemic insurance program.
As apparent from Table 4.2, no approach as currently proposed dominates the
others on all the performance metrics. Rather, each has advantages and disadvantages
in terms of the performance metrics examined.
CHAPTER FIVE

Sensitivity of Projections to Changes in Parameter Values

There is considerable uncertainty over the appropriate values for many of the param-
eters that underlie the projections in the base case. In this chapter, we examine the
sensitivity of the results to changes in the input parameters.

Approach

We examined the sensitivity of the results by varying the eight input parameters in
Table 5.1. Plausible ranges are selected for each parameter based on the information
collected during the study.
We assumed in the base case that insurers would use a five-year pandemic return
period in pricing the risk that they bear. This short return period would be most rel-
evant if the insurance program applies to COVID-19 and its variants. If the insurance
program excludes COVID-19 and its variants, insurers may price coverage assuming a
longer return period. Chubb, for example, excludes COVID-19 and its variants from

Table 5.1
Range of Parameter Values Used in Sensitivity Analysis

Value Used in Baseline
Parameter Projections Plausible Range

Pandemic return period used by insurers in pricing 5 years [3, 100]

G&SE for small and medium-sized firms $100 [$50, $200]

G&SE for large firms $10,000 [$5,000, $20,000]

LAE ratio for parametric policies 0.04 [0.02, 0.06]

LAE ratio for indemnity policies 0.15 [0.010, 0.200]

Demand elasticity –0.50 [–0.75, –0.25]

Diversifiability index 0.50 [0.25, 0.75]

Rate of return on capital 9% [6%, 12%]

59
60 Improving the Availability and Affordability of Pandemic Risk Insurance

its proposed program and assumes a 30-year return period in its pricing.1 Given the
key role that the return period plays in determining the premium and uncertainty over
whether Congress will require a pandemic insurance program to cover COVID-19
and its variants, we first provide a detailed analysis of the sensitivity of the outcomes to
changes in the return period.
For each of the other seven parameters in Table 5.1, we project outcomes by set-
ting the parameter at the minimum and the maximum of the plausible range and
holding other parameters at base-case values. The resulting range of projections is then
reported for four key outcomes: premium and take-up for small and medium-sized
firms, premium and take-up for large firms, amount of capital held by insurers, and
expected annual net government outlays. Probabilities are not associated with particu-
lar parameter values or the resulting outputs. Rather, this analysis illustrates the sensi-
tivity of the results to the underlying parameters.
Finally, we examine the sensitivity of the results to change in the assumption
regarding whether the government will reimburse insurers for LAE on the part of the
claim payments it funds. Motivated by TRIP’s decision to include LAE in insured loss,
we assumed in the base case that the government would cover LAE for its portion of
claim payments. The alternative examined in this chapter assumes that insurers will be
responsible for all LAE in the Chubb and PRIA approaches. Because insurers bear no
risk in the BCPP, we assume that the government continues to reimburse insurers for
all LAE in that program.

Sensitivity of Results to Changes in Return Period Used by Insurers in
Pricing

The projected annual premium for small and medium-sized firms in the Chubb and
PRIA approaches varies considerably as the return period changes (see Figure 5.1).
Note that the BCPP premium does not change as the return period varies because no
risk is borne by insurers in the BCPP, so the return period that insurers use in pricing is
immaterial. The base-case results can be read off Figure 5.1 using the five-year return
period. Some stakeholders with whom we conferred thought that insurers might use
an even shorter return period if the program applies to COVID-19, and the substan-
tially higher premium that would result from a three-year return period is shown in
Figure 5.1 (first dot plotted from the left for each program). For example, small and
medium-sized firms would pay $2,760 under PRIA if insurers use a three-year return
period versus $1,630 in the base case. If insurers use a 30-year return period, the pre-
miums in all three programs are quite close for small and medium-sized firms.

1 State insurance regulators could potentially resist the use of such longer return periods, fearing that these
longer returns could jeopardize insurer solvency.
Sensitivity of Projections to Changes in Parameter Values 61

Figure 5.1
Premium for Small and Medium-Sized Firms as Pandemic Return Period Used by Insurers
Varies

3,000

BCPP
2,500
Chubb

Annual premium per firm ($)
PRIA
2,000

1,500

1,000

500

0
0 10 20 30 40 50 60 70 80 90 100
Return period used by insurers in setting premium (years)

Because Pandemic Re uses a market-based rate for the risk borne by both insurers
and the government, the premium for large firms in Pandemic Re remains consider-
ably higher than in the BCPP and PRIA approaches over a wide range of return periods
(see Figure 5.2). Even with a return period of 30 years, the premium for large firms is
$1.22 million in the Chubb approach versus $164,000 and $190,000 in the BCPP and
PRIA approaches, respectively. Only for very long return periods do the premiums of
all three programs converge. These results show the challenges of using the Pandemic
Re approach for large firms if the program applies to COVID-19 and its variants.
The effect of return period on take-up rates mirrors the patterns in Figures 5.1
and 5.2. In the base case, take-up rates for the Chubb and PRIA approaches are much
lower than for the BCPP (see the results for the five-year return period in Figure 5.3).
However, once the return period equals 30 years, the take-up rates for small and
medium-sized firms are similar in all three approaches. The take-up rate for large firms
in the Chubb approach remains substantially below that in the BCPP for all return
periods considered.
The amount of capital that insurers hold to protect against solvency risk rises
considerably as the return period increases (see Figure 5.4). For both the Chubb and
PRIA approaches, the amount of capital approximately doubles as the return period
that insurers use in setting premiums rises from five years to 30 years. The amount of
capital increases because the annual premium across all participating firms falls as the
62 Improving the Availability and Affordability of Pandemic Risk Insurance

Figure 5.2
Premium for Large Firms as Pandemic Return Period Used by Insurers Varies

BCPP

Annual premium per firm ($ millions)
10
Chubb
PRIA
8

0
0 10 20 30 40 50 60 70 80 90 100
Return period used by insurers in setting premium (years)

Figure 5.3
Take-Up Rate as Return Period Used by Insurers Varies

BCPP (small, medium, and large firms) PRIA (small, medium, and large firms)
90 Chubb (small and large firms) Chubb (large firms)

Take-up rate (%)
50

0
0 10 20 30 40 50 60 70 80 90 100
Return period used by insurers in setting premium (years)

NOTE: The take-up rates for small and medium-sized firms and large firms are the same in the BCPP; the
same is true for PRIA.
Sensitivity of Projections to Changes in Parameter Values 63

Figure 5.4
Capital Help by Insurers as Return Period Used by Insurers Varies

Capital held by insurers ($ billions)
20

10
BCPP
Chubb
PRIA
5

0
0 10 20 30 40 50 60 70 80 90 100
Return period used by insurers in setting premium (years)

return period increases,2 increasing the gap between the loss that insurers plan for and
the annual premium collected (see Appendix C for a discussion of planning loss).
Finally, expected annual government net outlays in the Chubb and PRIA
approaches increase considerably as the return period grows (see Figure 5.5). Expected
annual government outlays in Figure 5.5 are calculated assuming an actual return
period of 30 years and cover all firm sizes. Expected annual government net outlays in
the Chubb approach remain below those for PRIA across the range of return periods
examined and approaches those for the BCPP for long return period. When the return
period used by insurers in setting premiums is sufficiently below the assumed 30-year
actual return period, the Chubb approach generates a government surplus.

Sensitivity of Results to Changes in Other Parameters

Table 5.2 summarizes the sensitivity of the results to variation in the other parameters
listed in Table 5.1. The projection when the parameter is set at the left end of the inter-
val examined is reported as the left-hand number in each bracket, and the projection
when the parameter is set at the right end of the interval examined is reported as the
right-hand number in each bracket. Outcomes that are particularly sensitive to varia-
tion of the parameter over the range examined are shaded.

2 This result is a consequence of the inelastic demand curve.
64 Improving the Availability and Affordability of Pandemic Risk Insurance

Figure 5.5
Expected Annual Government Net Outlays for All Firms as the Return Period Used by
Insurers Varies

Expected annual net government outlays
25

BCPP
10
Chubb
PRIA
5

($ billions)
0

–5

–10

–15
0 10 20 30 40 50 60 70 80 90 100
Return period used by insurers in setting premium (years)

NOTE: Assumes an actual pandemic return period of 30 years.

Overall, the outcomes are not particularly sensitive to variation of each param-
eter over its plausible range, holding other parameters at their base-case values. For
example, varying the LAE ratio for parametric policies between 0.02 and 0.06 does not
cause large changes in the outcome of the BCPP and Chubb approaches (PRIA does
not use a parametric policy). However, some outcomes are quite sensitive to some of
the parameters examined. Because of its low value in the base case, varying the G&SE
for small and medium-sized firms between $50 and $200 causes considerable propor-
tionate change to the premium in the BCPP, with consequent large effect on the take-
up rate. The demand elasticity also has a considerable impact on the outcomes for the
Chubb and PRIA approaches, and the diversifiability index has considerable impact on
the amount of capital held by insurers in the Chubb and PRIA approaches.
The results of this sensitivity analysis suggest areas in which further work to
narrow uncertainty over likely parameter values would be most useful.

Sensitivity of Results to Change in Assumption on Whether
Government Pays Loss Adjustment Expenses

The first set of rows in Table 5.3 reproduces the base-case results in which the govern-
ment reimburses insurers for the LAE on the portion of the loss it funds. The second
Sensitivity of Projections to Changes in Parameter Values 65

Table 5.2
Sensitivity of Program Outcomes to Changes in Parameter Values

Small and
Medium-Sized Firms Large Firms All Firms

Expected
Amount of Annual
Parameter Premium Capital Held Government
[parameter Premium Take-Up ($1,000 per Take-Up by Insurers Net Outlays
range] ($ per firm) Rate (%) firm) Rate (%) ($ billions) ($ billions)a

G&SE for small
firms and medium-
sized firms
[$50, $200]

BCPP [382, 542] [69, 59] [164, 164] [65, 65] [0, 0] [19.1, 17.8]

Chubb [1,260, 1,410] [38, 36] [6,860, 6,860] [10,10] [6.16, 5.90] [–4.84, –5.17]

PRIA [1,580, 1,740] [33, 32] [694, 701] [33, 32] [12.7, 12.5] [11.2, 10.9]

G&SE for large
firms [$5,000,
$20,000]

BCPP [442, 442] [65, 65] [159, 173] [65, 62] [0, 0] [18.8, 18.3]

Chubb [1,311, 1,311] [38, 38] [6,860, 6,870] [10, 10] [6.06, 6.08] [–4.96, –4.95]

PRIA [1,630, 1,630] [33, 32] [690, 708] [33, 32] [12.6, 12.6] [11.1, 11.0]

LAE ratio for
parametric policies
[0.02, 0.06]

BCPP [436, 349] [65, 65] [161, 166] [66, 65] [0, 0] [18.4, 18.8]

Chubb [1,290, 1,330] [38, 37] [6,860, 6,860] [10, 10] [6.02, 6.12] [–5.02, –4.89]

PRIA [1,630, 1,630] [32, 32] [696, 696] [32, 32] [12.6, 12.6] [11.1, 11.1]

LAE ratio for
indemnity policies
[0.10, 0.20]

BCPP [442, 442] [65, 65] [164, 164] [65, 65] [0, 0] [18.6, 18.6]

Chubb [1,311, 1,311] [38, 38] [6,570, 7,160] [10, 10] [6.04, 6.10] [–4.71, –5.19]

PRIA [1,524, 1,710] [33, 32] [659, 734] [33, 32] [12.2, 13.0] [10.9, 11.3]

Demand elasticity
[–0.25, –0.75]

BCPP [442, 442] [65, 65] [164, 164] [65, 65] [0, 0] [18.6, 18.6]

Chubb [1,311, 1,311] [50, 29] [6,860, 6,860] [25, 4] [9.73, 4.09] [–20.0, –0.260]

PRIA [1,428, 2,080] [48, 19] [606, 898] [48, 19] [16.0, 9.56] [16.4, 6.33]
66 Improving the Availability and Affordability of Pandemic Risk Insurance

Table 5.2—Continued

Small and
Medium-Sized Firms Large Firms All Firms

Diversifiability index
[0.25, 0.75]

BCPP [442, 442] [65, 65] [164, 164] [65, 65] [0, 0] [18.6, 18.6]

Chubb [1,400, 1,220] [36, 39] [6,900, 6,830] [10, 10] [8.91, 3.11] [–5.19, –4.70]

PRIA [1,780, 1,490] [31, 34] [763, 632] [31, 34] [18.4, 6.48] [10.6, 11.7]

Rate of return on
capital [0.06, 0.12]

BCPP [442, 442] [65, 65] [164, 164] [65, 65] [0, 0] [18.6, 18.6]

Chubb [1,250, 1,370] [39, 37] [6,840, 6,890] [10, 10] [6.16, 5.98] [–4.79, –5.12]

PRIA [1,530, 1,730] [34, 32] [653, 740] [34, 32] [12.8, 12.4] [11.5, 10.8]

NOTES: Outcomes particularly sensitive to variation of the specified parameter over its plausible range
are shaded.
a Assumes actual pandemic return period of 30 years.

set of rows presents the results when insurers are responsible for all LAE, regardless
of what part of overall claim payments are funded by the government. Because insur-
ers bear no risk in the BCPP, we assume that the government continues to reimburse
insurers for all LAE in that program.
Requiring insurers to bear all LAE has a substantial impact on premium and take-
up for small and medium-sized firms. For example, premium in the Chubb approach
for small and medium-sized firms rises from $1,070 to $1,650, and the take-up rate
falls from 42 percent to 34 percent. The effect on large firms is also considerable in
PRIA, although the effect is minimal in Chubb’s Pandemic Re approach because the
premium is already at a market-based rate. Capital held by insurers to protect against
solvency risk roughly doubles if the insurers bear all LAE. In contrast, government net
annual outlays fall as more of the overall program costs are passed to the private sector.
Sensitivity of Projections to Changes in Parameter Values 67

Table 5.3
Sensitivity of Program Outcomes to Change in Assumption on Whether Government Pays
Loss Adjustment Expenses

Small and
Medium-Sized Firms Large Firms All Firms

Expected
Amount of Annual
Capital Held Government
Premium ($ Take-Up Premium Take-Up by Insurers Net Outlays
Assumption per firm) Rate (%) ($ per firm) Rate (%) ($ billions) ($ billions)a

Government pays
LAE on its portion
of claim payments

BCPP $442 65% $164,000 65% $0 $18.7

Chubb $1,070 42% $6.86M 10% $5.54 –$5.59

PRIA $1,630 32% $696,000 32% $12.6 $11.1

Government does
not pay LAE on its
portion of claim
payments

BCPP — — — — — —

Chubb $1,650 34% $7.06M 10% $11.1 –7.33

PRIA $2,330 20% $1.01M 20% $20.6 $5.90
a Assumes actual pandemic return period of 30 years.
CHAPTER SIX

Conclusion

Several programs have been proposed to expand the availability of insurance for busi-
ness revenue declines due to pandemic-induced business closures or restrictions on
business activity. These programs seek to keep employee paychecks coming during
a pandemic and to cover the ongoing operating expenses needed for firms to survive
until the pandemic passes. The programs differ in several key dimensions, including
the amount of risk borne by commercial insurers, the approach to paying claims, the
extent to which the U.S. government receives a premium for the risk it bears, and the
extent of policyholder subsidies.
The programs all leverage the deep capability of the commercial insurance indus-
try to place policies and adjust claims. The programs all seek to define benefits and
the benefit distribution mechanism in advance rather than rely on programs hastily
crafted after an event occurs.
The programs were modeled and performance characterized in the following
dimensions:

• efficacy
• efficiency
• affordability
• risk borne by commercial insurers
• expected annual government outlays
• extent of policyholder subsidy.

For small and medium-sized firms, the BCPP approach performs best in terms of
affordability, efficacy, and efficiency. The PRIA approach does the best in terms of
expected annual net government outlays and policyholder subsidy. It also does best in
terms of risk borne by commercial insurers, assuming that best in this context means
the most risk borne by commercial insurers. For large firms, the BCPP again performs
best in terms of affordability, efficacy, and efficiency. PRIA still transfers the most risk
to commercial insurers, but the Chubb approach now generates the lowest expected
annual government net outlays and policyholder subsidy. Projected outcomes are sen-
sitive to a number of underlying model parameters, about which there is considerable

69
70 Improving the Availability and Affordability of Pandemic Risk Insurance

uncertainly. In particular, the pandemic return period that insurers used in pricing
coverage has an important impact on premium and take-up in the Chubb and PRIA
approaches, with the return period used in pricing depending importantly on whether
the program applies to COVID-19 and its variants.
This analysis provides information useful to assessing the advantages and disad-
vantages of different approaches for improving the availability and affordability of pan-
demic risk insurance. However, it does not seek to evaluate and compare the full range
of policy options for addressing pandemic losses. Such a policy analysis would system-
atically compare the advantages and disadvantages of an insurance-based approach
with a government-based approach, such as the PPP. It would also take a more com-
prehensive view of some of the outcome measures used in this analysis. For example,
in this report we have projected government net outlays directly through the proposed
programs, but a more comprehensive analysis would include the indirect effects of the
program on other government outlays.
The various proposals that we have examined contain several important gaps that
remain to be filled—for example,

• whether and how the parametric trigger would address firms either subject to a
partial shutdown order or that experience substantial revenue decline but are not
subject to a closure order
• whether policies backed by PRIA will provide coverage for pandemic losses when
the underlying BI policies require physical damage to the insured property
• whether parametric policies will result in some firms receiving more compensa-
tion than they need and others less (basis risk).

In this analysis, we have in effect assumed that gaps are filled in that result in the right
amount of assistance getting to the right firms, but additional work is needed to flesh
out the proposals to deal with these and other gaps and to determine the extent to
which unaddressed issues can be resolved.
APPENDIX A

Overview of Business Interruption Policies

BI insurance, also known as business income and extra expense insurance, covers net
income that would have been earned during the period of interruption plus continuing
normal operating expenses.1 The provisions of the BI policies can vary, and the follow-
ing overview attempts to characterize the policies typically held by small and medium-
sized businesses.2
BI coverage is typically tied to a property policy. It is included in the standard
BOP and is primarily designed to cover loss of business income due to suspension of
operations caused by direct physical damage to the covered property. The physical
damage must be caused by a hazard covered by the policy (a covered cause of loss).3 The
policy covers net income (profit) that would have been earned or incurred if no physi-
cal loss or damage had occurred, as well as continuing normal operating expenses. Cover-
age applies during the period required to repair the damaged property or to find a new
location, up to 12 consecutive months.
Continuing normal operating expenses include such items as rent, loan payments,
utilities, and payroll.4 However, the standard policy does not treat the payroll of all
employees the same. Payroll for officers, executives, managers, employees under con-
tract, and other specified job classes are covered for up to 12 consecutive months.
These are generally employees with hard-to-replace skills. Payroll for everyone else
is referred to as ordinary payroll, and ordinary payroll is covered for 60 days in the
standard policy. There was disagreement among the experts with whom we conferred
during the course of this study about how frequently firms that purchase property
insurance have BI coverage. Some maintained that there are limited options to opt out

1 Elsewhere in the report we use the term business revenue rather than business income and do not distinguish
between the two terms.
2 This overview is based in part on Insurance Services Office, Businessowners Coverage Form, BP 00 03 01 10,
2009.
3 Insurance Services Office, 2009.
4 In this report, ongoing operating expenses refers to that part of normal operating expenses other than payroll
and benefits.

71
72 Improving the Availability and Affordability of Pandemic Risk Insurance

of the BI coverage included in the standard BOP. Others said that small businesses in
the United States typically opt out of ordinary payroll coverage.
Continuing normal operating expenses are not explicitly defined in the standard
policy, which is one of the reasons that business income losses are more costly to adjust
than losses in many other lines of insurance.
There is typically no dollar limit on this time-based coverage. All losses (other
than ordinary payroll) in the 12-month period following the physical damage are cov-
ered even if the 12-month period extends beyond the end of the policy period. What is
more, the 12-month period resets for each loss. Thus, a firm with a nine-month closure
due to property damage at the beginning of the policy period and a 12-month closure
due to property damage on the last day of the policy period could receive payment for
income losses over 21 months. There is also no waiting period before coverage applies
(except for the civil authority coverage).
Loss of business income due to actions of a civil authority is also covered in the
standard BOP, but not nearly to the same extent as losses due to the suspension of
operations caused by physical damage to the insured property. Coverage applies when
the insured property is in the area closed because of the incident and no more than one
mile from the damaged property. Coverage typically begins 72 hours after the closure
and applies for four consecutive weeks following the closure.
APPENDIX B

Insurance Payments in a Partial Shutdown

This appendix shows how much of a parametric payout a firm would need to return
(or reserve for future use) during a partial shutdown. The analysis assumes that firms
are not reimbursed for lost profits and are provided with enough income to maintain
payroll and benefits at preshutdown levels and to cover ongoing operating expenses
postshutdown.
Monthly firm revenue preshutdown is the sum of the following components:

I0 = π0+ PBOE0 + VC0,

where

I0 = monthly income preshutdown
π0 = monthly firm profit preshutdown
PBOE 0 = monthly payroll, benefits, and ongoing operating expenses preshut-
down
VC 0 = monthly firm variable costs preshutdown.

When a shutdown occurs, the firm receives a parametric insurance payment equal
to preshutdown payroll, benefits, and ongoing operating expenses. The firm monthly
income postshutdown is thus

I1R = I1 + PBOE0,

where

I1R = postshutdown income received, including insurance payout
I1 = income postshutdown without insurance payment and 0 ≤ I1 < I0.

73
74 Improving the Availability and Affordability of Pandemic Risk Insurance

To maintain payroll, benefits, and ongoing operating expenses postshutdown, the
firm needs the following income:

I,y= PBOE,+ VC,
where

Ly = monthly income needed to maintain preshutdown payroll and benefits
and to support postshutdown ongoing expenses

VC, = monthly variable costs required to support postshutdown level of pro-
duction.

The firm makes zero profit in this situation.

Subtracting the amount of income needed from the amount of income received
postshutdown results in the amount of the insurance payment that is returned or
reserved for future use:

Amount of the insurance payment returned =

OwhenVC\ > J,
I,—VC, when 1, — PBOE, < VC. <I,
PBOE, whenVC, < I, — PBOE,.
APPENDIX C

Pandemic Risk Insurance Model

In this appendix, we provide the equations used to calculate the model outputs listed
in Table 3.2. Table C.1 defines the parameters used in the equations.

Table C.1
Parameter Definitions

Parameter Definition

event_loss Reduction in payroll, benefits, and revenue needed to cover ongoing costs
due to the pandemic

eligible_firms Number of firms eligible to participate in program

covered_costs_per_month Payroll, benefits, and ongoing operating expenses covered by the policy in
dollars per month

dindex The diversifiability index factor, which captures the extent that insurers can
use surplus held for other types of hazards to reduce the amount of capital
they need to support pandemic risk policies

months_coverage Months of costs provided by policy

pholder_deductible The amount of loss absorbed by the policyholder before insurance
payments begin

pholder_copay Proportion of loss after waiting period that is not covered by the insurance
payment

program_cap Maximum aggregate claim payments allowed under the program

insurer_deductible The amount of claim payments that must be paid by commercial insurers
before government claim payments begin

insurer_copay Proportion of claim payments above insurer deductible that is paid by
commercial insurers

ror Rate of return required on the capital that commercial insurers hold for
solvency risk

event_prob The pandemic return periods that commercial insurers or the government
use in pricing coverage; the return period used by insurers can differ from
that used by government

lae_ratio Ratio applied to claim payments to calculate LAE

takeup Proportion of eligible firms that purchase coverage

75
76 Improving the Availability and Affordability of Pandemic Risk Insurance

The number of firms participating in the program is
participating _ firms = eligible _ firms * takeup. (C.1)

The number is calculated separately for small and medium-sized and large firms.
The maximum amount of loss covered by the policy is

policy _limit = covered _costs_ per _month* months _ coverage. (C.2)

Losses that occur any time during the policy period are covered up to the policy limit.
The policy limit for Pandemic Re is the $50 million dollar limit prescribed in the
Chubb’s proposal multiplied by the number of participating firms.
The amount of loss that is insured should an event occur is
insured _ loss = (C.3)
min (takeup * (event _ loss — pholder _ deductible ) * (1 — pholder _ copay) ;
takeup * ( policy _limit — pholder _ deductible) * (1 — pholder _ copay) ,
program _cap),

The amount of insured loss incurred by insurers should an event occur is

insurer _ claims = (C.4)

insured _ loss if insured _ loss < insurer _ deductible
insurer _ deductible + insurer _ copay * (insured _ loss — insurer _ deductible)

if insured _loss > insurer _ deductible.

govt _ claims = insured _loss — insurer _ claims. (C.5)
LAE = insured _ loss* LAE _ ratio. (C.6)

$100 for small and medium-sized firms

C,
$10,000 for large firms (C.7)

G&SE -|
Pandemic Risk Insurance Model 77

The insurance industry’s cost of bearing pandemic risk includes the cost of the
capital needed to ensure, with a given probability, that insurers have sufficient resources
to finance claim payments. The planning probabilities that insurers are thought to use
in making this decision are high—99 percent or higher.! For this analysis, we have
assumed that a pandemic is a dichotomous random variable and that insurers plan for
an event that is of the same magnitude as the current event. The planning loss (poss) in
this case is thus insurer claims plus LAE should a pandemic occur plus G&SE.

ploss = insurer _ claims * ( + LAE _ ratio) (C.8)

In PRIA and Chubb’s BEIP, the insurers retain all the premium and thus

capital = (1— dindex) * ( ploss — ( premium —G&SE — ror * capital )) (C.9)

In Pandemic Re, insurers receive only a part of the premium, which will be

addressed below.

Rearranging C.9 yields
capital = ploss — premium + G&SE . (C.10)
a
where
1

Q = ——__—_ — ror

1—dindex
0 < dindex <1.

Premium is the sum of expected annual loss, expected annual LAE, G&SE, and
the costs of capital required to hedge solvency risk:

premium = event _ prob* insurer _claims* (1+ LAE, jy) -+ G8CSE + ror * capital, (C.11)
where event_prob is the event probability used by the agent (either insurers or the gov-

ernment) in setting premium. Substituting C.10 into C.11 and rearranging yields the
following for the PRIA program:

1 Dixon et al., 2007, p. 17.
78 Improving the Availability and Affordability of Pandemic Risk Insurance

premiumpery = (C.12)
event _ prob * insurer _ claims *(1+ LAE _ ratio) + "| * ploss
G&SE + -
ror
147"
a

Because no capital is required by the BCPP and all claims are funded by the
government,’ the premium calculation for the BCPP simplifies to

premiumpcpp = G&SE + event _ prob” insured _loss*(1+ LAE _ratio) (C.13)

Equation C.12 is also used to calculate the premium for Chubb’s BEIP for small
and medium-sized firms.

Insurers retain only a portion of the premium in Pandemic Re, and we assume
that insurers also retain the amount needed to cover their annual G&SE:

premium retained by insurers pp, = industry _copay* premium+ (C.14a)

(1 — industry _ copay) *G&SE |

premium ceded to govt py,p, =(1—industry _copay)*( premium —G&SE ). (C.14b)

We combine C.14a, C.10, and C.11 and recognize that the premium in Pandemic
Re is set at a market-based rate that covers the risk borne by both the government and
insurers. The premium becomes

PVEMIUM pinro = (C.15)

event _ prob* insured _ loss * (1+ LAE _ratio)+ 2" * ploss

G&SE + a

147" * copay _ industry
a

The take-up rates for the Chubb and PRIA programs are calculated using a con-
stant elasticity demand curve that has been fit to the premium and take-up rate for the
BCPP. The demand curve is of the form

2 Insurers participating in the BCPP, as well as the other proposed programs, may face risk that the govern-
ment will delay or fail to reimburse insurers for claim payments and expenses. Such risk is not considered in this
analysis.
Pandemic Risk Insurance Model 79

−k
⎛ premium ⎟⎟⎞ ,
Q d = β * ⎜⎜ (C.16)
⎜⎝ participating _ firms ⎟⎟⎠

where Qd is the number of firms participating in the program and –k is the demand
elasticity. The value for the demand elasticity in the base case is –0.50, and β is deter-
mined using the projected premium per firm and assumed take-up rate for the BCPP.
The demand equation is then used to project take-up for the Chubb and PRIA pro-
grams using the projected premiums for those programs. The premium and take-up
rates in the Chubb and PRIA programs are adjusted using the supply relationships
(C.12–C.15) and demand relationship (C.16) until the market reaches equilibrium.
APPENDIX D

General and Selling Expense and Loss Adjustment Expense
Ratios by Line of Business

Table D.1 reports the G&SE for P&C insurance in the United States by line of busi-
ness based on the annual statements submitted by insurers to the National Association
of Insurance Commissioners. Also reported is ratio of G&SE to premium written. To
smooth out annual variation in the ratio, G&SE and premium are each summed over
the five-year period spanning 2015 to 2019. Table D.2 contains similar information for
LAE, direct incurred losses, and the LAE ratio.
Table D.1
General and Selling Expense Ratio by Line of Business Between 2015 and 2019

Direct Written Premium Ratio of G&SE to
Line of Business ($1,000s) G&SE ($1,000s) Premium

Accident & health 31,814,525 7,918,434 0.249

Aircraft 8,210,399 2,138,151 0.260

Allied lines 57,707,058 16,109,499 0.279

Boiler & machinery 8,749,284 2,677,392 0.306

Burglary & theft 1,607,666 542,205 0.337

Commercial auto 179,308,672 47,045,168 0.262

Commercial multiple peril (liability) 74,977,328 23,721,032 0.316

Commercial multiple peril 129,430,443 39,595,324 0.306
(nonliability)

Credit 10,140,205 3,318,878 0.327

Earthquake 11,722,601 2,998,260 0.256

Farmowners multiple peril 21,186,593 5,801,467 0.274

Federal flood 14,409,404 3,432,840 0.238

Fidelity 6,222,379 1,965,377 0.316

Financial guaranty 5,834,175 2,299,795 0.394

81
82 Improving the Availability and Affordability of Pandemic Risk Insurance

Table D.1—Continued

Direct Written Premium Ratio of G&SE to
Line of Business ($1,000s) G&SE ($1,000s) Premium

Fire 64,866,013 17,310,988 0.267

Homeowners multiple peril 466,275,407 119,320,224 0.256

Inland marine 113,496,742 30,052,156 0.265

International 339,248 289,823 0.854

Medical professional liability 46,129,920 9,666,079 0.210

Mortgage guaranty 25,292,672 5,635,553 0.223

Multiple peril crop 49,516,999 5,422,631 0.110

Ocean marine 17,728,623 4,942,283 0.279

Other P&C 7,196,874 1,758,662 0.244

General liability (other liability and 346,047,180 89,933,360 0.260
product liability combined)

Personal auto 1,125,460,577 238,944,576 0.212

Private crop 5,198,145 1,191,984 0.229

Private flood 2,012,950 582,827 0.290

Surety 30,375,003 14,441,833 0.475

Warranty 12,945,324 2,345,340 0.181

Workers comp 275,414,629 57,867,280 0.210

All P&C lines 3,149,617,038 759,269,421 0.241

SOURCE: National Association of Insurance Commissioners data via S&P Global Intelligence, provided to
RAND by the American Property Casualty Insurance Association.
General and Selling Expense and Loss Adjustment Expense Ratios by Line of Business 83

Table D.2
Loss Adjustment Expense Ratio by Line of Business Between 2015 and 2019

Direct Incurred Loss Ratio of LAE to Direct
Line of Business ($1,000s) LAE ($1,000s) Incurred Loss

Accident & health 24,028,892 1,317,504 0.055

Aggregate write-ins 3,285,362 382,824 0.117

Aircraft 4,688,739 871,490 0.186

Allied lines 47,331,501 4,619,968 0.098

Boiler & machinery 3,533,629 244,793 0.069

Burglary & theft 452,303 67,444 0.149

Commercial multiple peril (liability) 36,575,534 17,014,514 0.465

Commercial multiple peril 77,240,461 8,464,595 0.110
(nonliability)

Commercial auto 123,397,681 22,035,696 0.179

Credit 4,441,500 356,332 0.080

Earthquake 264,489 78,853 0.298

Farmowners multiple peril 12,593,644 1,359,583 0.108

Federal flood 15,739,579 904,611 0.057

Fidelity 2,351,748 309,658 0.132

Financial guaranty 2,694,464 828,519 0.307

Fire 36,319,805 3,210,440 0.088

Homeowners multiple peril 286,960,410 41,501,726 0.145

Inland marine 56,244,592 4,198,689 0.075

International 220,661 36,985 0.168

Medical professional liability 22,485,904 12,748,962 0.567

Mortgage guaranty 3,419,471 294,421 0.086

Multiple peril crop 34,112,439 974,899 0.029

Ocean marine 9,622,185 1,305,775 0.136

General liability (other liability and 202,117,172 56,944,364 0.282
product liability combined)

Personal auto 758,526,979 127,147,169 0.168

Private crop 4,881,920 194,393 0.040

Private flood 1,431,272 64,389 0.045

Surety 4,934,434 1,511,457 0.306
84 Improving the Availability and Affordability of Pandemic Risk Insurance

Table D.2—Continued

Direct Incurred Loss Ratio of LAE to Direct
Line of Business ($1,000s) LAE ($1,000s) Incurred Loss

Warranty 7,469,610 303,029 0.041

Workers comp 138,269,326 36,351,867 0.263

All P&C lines 1,925,635,706 345,644,949 0.179

SOURCE: National Association of Insurance Commissioners data via S&P Global Intelligence, provided to
RAND by the American Property Casualty Insurance Association.
References

American Property Casualty Insurance Association, “Property Casualty Insurance Industry Suffered
Largest-Ever Drop in Surplus in the First Quarter of 2020,” July 28, 2020.
American Property Casualty Insurance Association, Independent Insurance Agents and Brokers
of America, Inc., and National Association of Mutual Insurance Companies, Business Continuity
Protection Program, updated September 2020. As of May 5, 2021:
https://home.treasury.gov/system/files/311/BCPP-Overview-9-20.pdf
Autor, David, David Cho, Leland D. Crane, Mita Goldar, Byron Lutz, Joshua Montes, William B.
Peterman, David Ratner, Daniel Villar, and Ahu Yildirmaz, “An Evaluation of the Paycheck
Protection Program Using Administrative Payroll Microdata,” unpublished manuscript, July 22,
2020. As of May 5, 2021:
https://economics.mit.edu/files/20094
Barinov, Alexander, Jianren Xu, and Steven W. Pottier, “Estimating the Cost of Equity Capital for
Insurance Firms with Multiperiod Asset Pricing Models,” Journal of Risk and Insurance, Vol. 87,
No. 1, 2020, pp. 213–245.
Business Continuity Coalition, homepage, undated. As of May 10, 2021:
https://www.businesscontinuitycoalition.com
———, “BCC Proposal: A Proposal from the Policyholder Community,” September 2020. As of
January 19, 2021:
https://home.treasury.gov/system/files/311/FACI-BCC-Proposal-9-20.pdf
———, “Pandemic Risk Insurance Act Business Continuity Coalition Proposal: Section-by-Section
Description,” March 2021. As of April 16, 2021:
https://www.businesscontinuitycoalition.com/recommended-proposal
Businesswire, “Most Small Business Owners at Risk for a Disaster,” August 31, 2015. As of April 4,
2021:
https://www.businesswire.com/news/home/20150831005924/en/
Cardona-Ospina, Jaime A., Valentina Henao-SanMartin, Wilmer F. Acevedo-Mendoza,
Katherinn Melissa Nasner-Posso, Dayron F. Martínez-Pulgarín, Abril Restrepo-López, Valentina
Valencia-Gallego, Matthew H. Collins, and Alfonso J. Rodriguez-Morales, “Fatal Zika Virus
Infection in the Americas: A Systematic Review,” International Journal of Infectious Diseases, Vol. 88,
November 2019, pp. 49–59.
Centers for Better Insurance, Terrorism Risk Insurance Act: Shifting Risk Through Terrorism Insurance
Captives, Frederick, Md., 2020. As of May 20, 2021:
https://www.betterins.org/bin/resources/TRIA/
TRIA%20-%20Terrorism%20Captives%20Report.pdf

85
86 Improving the Availability and Affordability of Pandemic Risk Insurance

Centers for Disease Control and Prevention, National Center for Emerging and Zoonotic Infectious
Disease, Division of High-Consequence Pathogens and Pathology, Viral Special Pathogens Branch,
“2014–2016 Ebola Outbreak in West Africa,” webpage, last reviewed March 8, 2019. As of April 16,
2021:
https://www.cdc.gov/vhf/ebola/history/2014-2016-outbreak/index.html
Centers for Disease Control and Prevention, National Center for Immunization and Respiratory
Diseases, “Highly Pathogenic Asian Avian Influenza A(H5N1) Virus,” webpage, last reviewed
December 12, 2018. As of April 16, 2021:
https://www.cdc.gov/flu/avianflu/h5n1-virus.htm
———, “2009 H1N1 Pandemic (H1N1pdm09 Virus),” webpage, last reviewed June 11, 2019. As of
April 16, 2021:
https://www.cdc.gov/flu/pandemic-resources/2009-h1n1-pandemic.html
Centers for Disease Control and Prevention, National Center for Immunization and Respiratory
Diseases, Division of Viral Diseases, “Frequently Asked Questions About SARS,” webpage, last
reviewed May 3, 2005. As of April 16, 2021:
https://www.cdc.gov/sars/about/faq.html
Chubb, Pandemic Business Interruption Program, Warren, N.J., July 8, 2020. As of January 14, 2021:
https://www.chubb.com/content/dam/chubb-sites/chubb-com/us-en/about-chubb/pandemic-
business-interruption-program/documents/pdf/pandemic-business-interruption-program.pdf
“The Covid-19 Tracker,” STAT, updated May 19, 2021. As of May 19, 2021:
https://www.statnews.com/feature/coronavirus/covid-19-tracker/
Cowley, Stacy, “Despite Billions in Fees, Banks Predict Meager Profits on P.P.P. Loans,” New York
Times, October 1, 2020. As of April 7, 2021:
https://www.nytimes.com/2020/10/01/business/ppp-loans-bank-profits.html
Dalton, Michael, Elizabeth Weber Handwerker, and Mark A. Loewenstein, “Employment Changes
by Employer Size During the COVID-19 Pandemic: A Look at Current Employment Statistics
Survey Microdata,” Monthly Labor Review, U.S. Bureau of Labor Statistics, October 2020.
Dixon, Lloyd, Robert J. Lempert, Tom LaTourrette, and Robert T. Reville, The Federal Role in
Terrorism Insurance: Evaluating Alternatives in an Uncertain World, Santa Monica, Calif.: RAND
Corporation, MG-679-CTRMP, 2007. As of May 5, 2021:
https://www.rand.org/pubs/monographs/MG679.html
Dixon, Lloyd, Flavia Tsang, and Gary Fitts, The Impact of Changing Wildfire Risk on California’s
Residential Insurance Market, Sacramento: California Natural Resources Agency, August 2018. As of
May 5, 2021:
https://www.rand.org/pubs/external_publications/EP67670.html
Federal Insurance Office, U.S. Department of the Treasury, Report on the Effectiveness of the
Terrorism Risk Insurance Program, Washington, D.C., June 2020.
Hartwig, Robert, Greg Niehaus, and Joseph Qiu, “Insurance for Economic Losses Caused by
Pandemics,” Geneva Risk and Insurance Review, Vol. 45, 2020, pp. 134–170.
Insurance Services Office, Businessowners Coverage Form, BP 00 03 01 10, 2009. As of April 5,
2021:
https://www.propertyinsurancecoveragelaw.com/files/file/BP%2000%2003%2001%2010.pdf
Insureon, “How Much Does a Business Owner’s Policy Cost?” webpage, undated-a. As of April 5,
2021:
https://www.insureon.com/small-business-insurance/business-owners-policy/cost
References 87

———, “How Much Does Business Interruption Insurance Cost?” webpage, undated-b. As of
April 5, 2021:
https://www.insureon.com/small-business-insurance/business-interruption/cost
Internal Revenue Service, Statistics of Income—2017: Corporation Income Tax Returns, Washington,
D.C., 2020. As of February 5, 2021:
https://www.irs.gov/statistics/soi-tax-stats-corporation-complete-report
Kielholz, Walter, “The Cost of Capital for Insurance Companies,” Geneva Papers on Risk and
Insurance, Vol. 25, No. 1, January 2000, pp. 4–24.
Kunreuther, Howard, and Erwann Michel-Kerjan, TRIA and Beyond: Terrorism Risk Financing
in the U.S., Philadelphia: Wharton Risk Management and Decision Process Center, University of
Pennsylvania, August 2005. As of February 5, 2021:
https://riskcenter.wharton.upenn.edu/wp-content/uploads/2014/07/TRIA-and-Beyond.pdf
Marsh, 2019 Terrorism Risk Insurance Report, New York, May 2019.
Michel-Kerjan, Erwann, Paul Raschky, and Howard Kunreuther, “Corporate Demand for Insurance:
New Evidence from the U.S. Terrorism and Property Markets,” Journal of Risk and Insurance,
Vol. 82, No. 3, September 2015, pp. 505–530.
National Academy of Social Insurance, Workers’ Compensation: Benefits, Coverage, and Costs,
Washington, D.C., November 2020. As of February 18, 2021:
https://www.nasi.org/sites/default/files/research/
November%202020%20-%20Benefits,%20Coverage,%20and%20Costs.pdf
National Association of Insurance Commissioners, Report on Profitability by Line by State in 2018,
Kansas City, Mo., 2019.
National Flood Insurance Program, Federal Emergency Management Agency, The Watermark,
Vol. 11, June 2020. As of January 22, 2021:
https://www.fema.gov/sites/default/files/2020-09/fema_watermark_report_06-2020.pdf
Pub. L. 116-136, Coronavirus Aid, Relief, and Economic Security Act, March 27, 2020.
Reed, Carrie, Matthew Biggerstaff, Lyn Finelli, Lisa M. Koonin, Denise Beauvais, Amra Uzicanin,
Andrew Plummer, Joe Bresee, Stephen C. Redd, and Daniel B. Jernigan, “Novel Framework for
Assessing Epidemiologic Effects of Influenza Epidemics and Pandemics,” Emerging Infectious
Diseases, Vol. 19, No. 1, January 2013, pp. 85–91.
Ribas Freitas, André Ricardo, Marcelo Napimoga, and Maria Rita Donalisio, “Assessing the Severity
of COVID-19,” Epidemiologia e Serviços de Saúde, Vol. 29, No. 2, April 2020.
U.S. Bureau of Economic Analysis, “Gross Domestic Product, Third Quarter 2020 (Advance
Estimate),” webpage, last modified October 30, 2020. As of May 5, 2021:
https://www.bea.gov/news/2020/gross-domestic-product-third-quarter-2020-advance-estimate
U.S. Bureau of Labor Statistics, “Employer Costs for Employee Compensation—September 2020,”
December 17, 2020. As of May 5, 2021:
https://www.bls.gov/news.release/archives/ecec_12172020.htm
U.S. Census Bureau, “Number of Firms, Number of Establishments, Employment, Annual Payroll,
and Preliminary Receipts by Enterprise Employment Size for the United States, All Industries: 2017,”
Excel file, March 6, 2020. As of May 5, 2021:
https://www.census.gov/data/tables/2017/econ/susb/2017-susb-annual.html
U.S. House of Representatives, Pandemic Risk Insurance Act of 2020, H.R. 7011, May 26, 2020.
88 Improving the Availability and Affordability of Pandemic Risk Insurance

U.S. Small Business Administration, “PPP Data: Find Data and Reports Related to the Paycheck
Protection Program (PPP),” webpage, undated. As of March 29, 2021:
https://www.sba.gov/funding-programs/loans/covid-19-relief-options/paycheck-protection-program/
ppp-data
———, “Second Updated Paycheck Protection Program Lender Processing Fee Payment and 1502
Reporting Process,” Procedural Notice 5000-20091, February 8, 2021. As of April 7, 2021:
https://www.sba.gov/sites/default/files/2021-02/Procedural%20Notice%205000-20091%
20-%202nd%20Updated%20PPP%20Processing%20Fee%20and%201502%20Reporting-508.pdf
U.S. Small Business Administration, Office of Advocacy, “A Look at Nonemployer Businesses,”
factsheet, August 2018. As of April 7, 2021:
https://www.sba.gov/sites/default/files/advocacy/Nonemployer-Fact-Sheet.pdf
World Health Organization, “Middle East Respiratory Syndrome Coronavirus (MERS-CoV),”
webpage, undated. As of April 16, 2021:
https://www.who.int/health-topics/middle-east-respiratory-syndrome-coronavirus-mers#tab=tab_1
Zurich, Zurich’s Draft Concept for Facilitating Pandemic Protection, Los Angeles, Calif., December 7,
2020. As of January 14, 2021:
https://www.zurichna.com/-/media/project/zwp/zna/docs/kh/pandemic/zurich-pandemic-risk-
concept-powerpoint-final.pdf?la=en&hash=3E3F7670DC6B7E0523EA1D700082D7DD
Kenneth R. Feinberg Center for Catastrophic
Risk Management and Compensation

T
he COVID-19 pandemic led to a substantial drop in U.S.
economic activity in 2020. Businesses often purchase
business interruption coverage for loss of revenue due to
fires and other perils, but insurers have held that in most
cases they are not obligated to cover the enormous losses
caused by COVID-19. Now, insurers, insurance industry trade groups,
policyholder groups, and Congress have developed proposals to expand
the availability of insurance for pandemic-induced business closures or
restrictions.

The programs differ in several key dimensions, including the amount of
risk borne by commercial insurers, the approach to paying claims, the
extent to which the U.S. government receives a premium for the risk it
bears, and the extent of policyholder subsidies. But the programs all
seek to define benefits and the benefit distribution mechanism in advance
rather than rely on programs hastily crafted after an event occurs.

The authors describe the distinguishing features of the most-visible
proposals and develop a quantitative model that projects their potential
consequences. Proposed programs are evaluated in terms of the
proportion of revenue decline replaced (efficacy), efficiency, affordability,
the risk borne by the commercial insurers, expected annual government
net outlays, and the amount of subsidy provided to policyholders. This
analysis provides information useful to assessing the advantages and
disadvantages of different approaches for improving the availability and
affordability of pandemic risk insurance.
$27.25

ISBN-10 1-9774-0707-2
ISBN-13 978-1-9774-0707-8
52725

www.rand.org 9 781977 407078

RR-A1223-1

File and source

File: improving-the-availability-and-affordability-of-pandemic-risk-insurance-projected-performance-of-proposed-programs.pdf
Size: 1,366,799 bytes
SHA-256: 3aeaec788dc00bbb27c13b07762643afd88d495d5863d7297ff8ae9dea6b7994
Our copy: improving-the-availability-and-affordability-of-pandemic-risk-insurance-projected-performance-of-proposed-programs.pdf
Original: doi.org