pandemic risk management: resources contingency planning

39
Pandemic Risk Management: Resources Contingency Planning and Allocation Alfred Chong, Runhuan Feng, Linfeng Zhang University of Illinois at Urbana-Champaign CAS 2020 Annual Meeting October 23, 2020 Based on a working paper by Xiaowei Chen (Nankai), Alfred Chong (UIUC), Runhuan Feng (UIUC), and Linfeng Zhang (UIUC).

Upload: others

Post on 14-Jan-2022

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Pandemic Risk Management: Resources Contingency Planning

Pandemic Risk Management: ResourcesContingency Planning and Allocation

Alfred Chong, Runhuan Feng, Linfeng Zhang

University of Illinois at Urbana-Champaign

CAS 2020 Annual Meeting

October 23, 2020

Based on a working paper by Xiaowei Chen (Nankai), Alfred Chong (UIUC),

Runhuan Feng (UIUC), and Linfeng Zhang (UIUC).

Page 2: Pandemic Risk Management: Resources Contingency Planning

Illinois Risk Lab

Page 3: Pandemic Risk Management: Resources Contingency Planning

Long history of pandemics

Repeated pandemics taught us that epidemic risk is inevitable.

Page 4: Pandemic Risk Management: Resources Contingency Planning

An example of COVID-19 coverage (JustInCase, Japan)

Page 5: Pandemic Risk Management: Resources Contingency Planning

Compartmental models

S – susceptible, I – infectious, R – removed

S′(t) = −βI(t)S(t)/N,I ′(t) = βI(t)S(t)/N − αI(t),R′(t) = αI(t),

where S(0) = S0, I(0) = I0 and R(0) = 0.

The total number of individuals remains constant, N = S(t)+ I(t)+R(t).

An average susceptible makes an average number β of adequate contactsw. others per unit time. (Law of mass action)

Fatality/recovery rate of the specific disease, α.

Page 6: Pandemic Risk Management: Resources Contingency Planning

Basic reproduction number

Average number of new infections from a single infection

Rt =β

α

S(t)

N.

Average time between contacts, Tc = 1/β.

Average time until removal, Tr = 1/α.

Average number of contacts by an infected person with others beforeremoval, Tr/Tc = β/α.

(Do not confuse Rt with the size of removed class R(t).

Page 7: Pandemic Risk Management: Resources Contingency Planning

Importance of basic reproduction number

Average number of new infections from a single infection

Rt =β

α

S(t)

N.

If Rt > 1, the epidemic will break out.

If Rt < 1, the epidemic will die out.

Effect of public health policies (non-pharmaceutical interventions)

Quarantine, social distancing, mandatory face mask: lowertransmission rate β;

Vaccination: lower susceptible S(t);

Page 8: Pandemic Risk Management: Resources Contingency Planning

Infectious disease insurance

Figure: Transmission and Insurance Dynamics

Page 9: Pandemic Risk Management: Resources Contingency Planning

Insurance reserve

Consider an insurance policy that

provides 1 monetary unit of compensation per time unit for eachinfected policyholder for the entire period of treatment; (intended tocover medical costs)

collects premium at the rate of π per time unit from each susceptiblepolicyholder at a fixed rate per time unit until the pandemic ends orthe policyholder is infected; (monthly premium in practice)

Premium incomes

P (t) = π

∫ t

0

s(u) du, s(u) = S(u)/N.

Benefit outgoes

B(t) =

∫ t

0

i(u) du, s(u) = I(u)/N.

Insurance reserveV (t) = P (t)−B(t).

Page 10: Pandemic Risk Management: Resources Contingency Planning

Reserve for a typical term life contract

Source: Dickson et al. (2013) Actuarial Mathematics for Life Contingent Risks. Cambridge University Press.

Page 11: Pandemic Risk Management: Resources Contingency Planning

Four shapes of reserves

(We use various premium rate, not necessarily net level premium)

Page 12: Pandemic Risk Management: Resources Contingency Planning

Importance of basic reproduction number in reserving

Shape of V (t) Premium πIncreasing and concave [ 1

R∞− 1,∞)

Increasing and concave-then-convex [ 1Rtm− 1, 1

R∞− 1)

Non-monotonic and concave-then-convex [ 1R0− 1, 1

Rtm− 1)

Non-monotonic and convex [−∞, 1R0− 1)

Since S(t) is a decreasing function, then R0 > Rtm > R∞.The exact expression of Rtm is provided in Feng and Garrido (2011).

Page 13: Pandemic Risk Management: Resources Contingency Planning

Bubonic plague in 1665-1666

A classic example of the SIR model fitted to data from the bubonic plague inEyam near Sheffield, England.

Date Susceptible Infective

Initial 254 7

July 3-4 235 14.5

July 19 201 22

August 3-4 153.5 29

September 3-4 108 8

September 19 97 8

October 20 83 0

s0 = 254/261 = 0.97318, s∞ = 83/261 = 0.31801 andi0 = 7/261 = 0.02682;

From clinical observations, an infected person stays infectious for anaverage of 11 days, α = 1/0.3667 = 2.73;

β/α ≈ ln(s0/s∞)/(1− s∞), which implies β = 4.4773.

Design a policy that pays 1, 000 per month to all infected. The minimummonthly premium to keep positive reserves is 114.58 for each susceptible.Each survivor receives a reward of 49.44 at the end.

Page 14: Pandemic Risk Management: Resources Contingency Planning

Contingency planning

Emerging viral pandemics “can place extraordinary and sustaineddemands on public health and health systems and on providers ofessential community services.”

Page 15: Pandemic Risk Management: Resources Contingency Planning

Strategic National Stockpile (SNS)

United States’ national repository of antibiotics, vaccines, chemicalantidotes, antitoxins, and other critical medical supplies.

Page 16: Pandemic Risk Management: Resources Contingency Planning

US underprepared for COVID-19

Failure of Congress to appropriate funding for SNS and to authorizeactions to replenish stockpiles

Supply-chain changes such as just-in-time manufacturing andglobalization

Lack of a coordinated Federal/State plan to deploy existing suppliesrapidly to locations of great need.

Page 17: Pandemic Risk Management: Resources Contingency Planning
Page 18: Pandemic Risk Management: Resources Contingency Planning

Evolution of epidemic in an SEIR model

0 50 100 150 200 250 300

Days since the first infection

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

10,000

Nu

mb

er

of

pe

op

le

Susceptible (S)

Exposed (E)

Mild infected (I1)

Hospitalized infected (I2)

ICU infected (I3)

Recovered (R)

Deceased (D)

Page 19: Pandemic Risk Management: Resources Contingency Planning

Prediction of healthcare demand

An assessment of needs for personal protective equipment (PPE) set(respirator, goggle, face shield) by ECDC

Page 20: Pandemic Risk Management: Resources Contingency Planning

Recall the SIR model (S – susceptible, I – infectious)

S′(t) =− βS(t)I(t)/N,I ′(t) = βS(t)I(t)/N − αI(t).

The demand for PPE can be estimated by

X(t) = θSβS(t)I(t)/N + θII(t).

(Better estimate requires a refined compartmental model such as SEIRmodels)

Page 21: Pandemic Risk Management: Resources Contingency Planning

Resources allegiance

A central authority acts in the interest of a union to manage andallocate supply among different regions.

Six US northeastern states formed a coalition in April 2020 topurchase COVID-19 medical equipment to avoid price biddingcompetition.

US states at different phases of the pandemic:

Page 22: Pandemic Risk Management: Resources Contingency Planning

Risk aggregation and capital allocation

X(1) X(2) · · · X(n)

K(1) K(2) · · · K(n)

· · ·

S

K

· · ·

Individual loss

Aggregate loss

Aggregate capital

Allocated capital

Page 23: Pandemic Risk Management: Resources Contingency Planning

Durable resources stockpiling and allocation

· · ·

K(1) K(2) · · ·

· · ·

· · ·

Regional demand

Aggregate demand

Aggregate supply

Allocated supply

Page 24: Pandemic Risk Management: Resources Contingency Planning

Durable resources (ventilator, ICU bed, hospital bed, etc.)Shortage stockpiling

Page 25: Pandemic Risk Management: Resources Contingency Planning

Durable resources (ventilator, ICU bed, hospital bed, etc.)Oversupply stockpiling

Page 26: Pandemic Risk Management: Resources Contingency Planning

Durable resources (ventilator, ICU bed, hospital bed, etc.)Optimal stockpiling

Page 27: Pandemic Risk Management: Resources Contingency Planning

Durable resources (ventilator, ICU bed, hospital bed, etc.)Optimal allocation at the FIRST peak

First peak

Aggregate supply

Onset Onset

Page 28: Pandemic Risk Management: Resources Contingency Planning

Durable resources (ventilator, ICU bed, hospital bed, etc.)Optimal allocation at the SECOND peak

Post peak

Aggregate supply

Second peak Onset

Page 29: Pandemic Risk Management: Resources Contingency Planning

Durable resources (ventilator, ICU bed, hospital bed, etc.)Optimal allocation at the THIRD peak

Post peak

Aggregate supply

Post peak Third Peak

Page 30: Pandemic Risk Management: Resources Contingency Planning

Durable resources (ventilator, ICU bed, hospital bed, etc.)Ventilator example

Page 31: Pandemic Risk Management: Resources Contingency Planning

Single-use resources stockpil., distribution, and allocation

· · ·

K(1) K(2) · · ·

· · ·

· · ·

Regional demand

Aggregate demand

Storage and aggregate supply

Allocated supply

Page 32: Pandemic Risk Management: Resources Contingency Planning

Single-use resources (testing kit, PPE, etc.)Stockpiling and early distribution

Aggregate demand

Storage Aggregate supply

Page 33: Pandemic Risk Management: Resources Contingency Planning

Single-use resources (testing kit, PPE, etc.)Stockpiling and late distribution

Aggregate demand

Storage Aggregate supply

Page 34: Pandemic Risk Management: Resources Contingency Planning

Single-use resources (testing kit, PPE, etc.)Optimal stockpiling and distribution

Aggregate demand

Storage Aggregate supply

Page 35: Pandemic Risk Management: Resources Contingency Planning

Single-use resources (testing kit, PPE, etc.)Optimal allocation

Page 36: Pandemic Risk Management: Resources Contingency Planning

Single-use resources (testing kit, PPE, etc.)Testing kit example

Page 37: Pandemic Risk Management: Resources Contingency Planning

COVID Plan website

covidplan.io

Page 38: Pandemic Risk Management: Resources Contingency Planning

References

X. Chen, W.F. Chong, R. Feng, L. Zhang (2020). Pandemic riskmanagement: resources contingency planning and allocation.Preprint. https://www.researchgate.net/publication/

341480083_Pandemic_risk_management_resources_

contingency_planning_and_allocation

W.F. Chong, R. Feng, L. Jin. (2020). Holistic principle for riskaggregation and capital allocation. Submitted. https:

//papers.ssrn.com/sol3/papers.cfm?abstract_id=3544525

J. Dhaene, A. Tsanakas, E. A. Valdez, S. Vanduffel (2012). Optimalcapital allocation principles. The Journal of Risk and Insurance,79(1), 1–28.

R. Feng and J. Garrido (2011). Actuarial applications ofepidemiological models. North American Actuarial Journal, 15(1),112–136.

Page 39: Pandemic Risk Management: Resources Contingency Planning

Thank you!

COVID Plan websitecovidplan.io

Contact us:[email protected]

[email protected]

[email protected]

[email protected]