TCH: The World Bank’s Pandemic Bonds

???????????Insurance works courtesy of the law of large numbers. Every year, some number of homes burn down and some proportion of cars crash and these events may be unpredictable to any one insured person. To an insurer though, the occurrence of such mishaps across their portfolios is actually quite predictable. However, insurers also have a lot of exposure to rarer events that can pose extremely high risk and cannot be so easily handled by calculating averages. Windstorms, floods, earthquakes, and wildfires can all leave behind a wide swath of destruction and do not occur with a predictable regularity, at least over short spans of time.

???????????These events, or at least the most severe among them, are not experienced often enough for the law of large numbers to provide its usual assistance. To insure against these risks, insurers have recently turned to catastrophe bonds. These bonds transfer the risk posed by severe hurricanes, earthquakes, and other perils away from insurers, providing them some relief when disaster strikes. In one instance, catastrophe bonds were issued linked to the occurrence of a pandemic and this lone issuance turned out to be quite timely and unexpectedly illustrative.

Catastrophe Bonds

???????????Property and casualty insurers are always exposed to difficult to predict and very costly ‘tail events’, those as extreme as they are rare. Category 5 hurricanes are good examples of tail events afflicting insurers, especially in the Unites States. In 1992, Hurricane Andrew caused $15.5 billion in losses to insured properties. These losses led to the insolvency of at least sixteen insurers, highlighting the sector’s vulnerability to infrequent but severe natural disasters. One of the outcomes of the disaster was the development of a new tool for spreading insurance risk, catastrophe bonds.

???????????Catastrophe bonds are devices for transferring insurance risk from the bond issuer, typically an insurance or reinsurance company, to investors in the capital markets. At a high level, they are quite simple. Investors purchase the bonds with cash which is then held in a dedicated account and invested in very-safe securities. In the event a particular peril specified in the bond documents materializes, the assets in the dedicated account are liquidated and the proceeds are disbursed to the insurance company issuing the bond. Investors in turn would receive only some, or even none, of their money back. The insurance company meanwhile is given more resources with which to make good on insurance claims submitted by customers.

???????????Until that happens though, the investors receive interest payments funded through any return earned on the securities in the special account on top of any premiums paid by the bond issuer for this insurance. If a covered peril never materializes before the bond’s maturity, then investors receive all of their investment back. Hannover Re issued what is believed to have been the first such catastrophe bond in 1994.

Triggers

???????????Catastrophe bonds have provided insurance against several types of risks over the years, most notably the risks of natural disasters like windstorms, earthquakes, flooding, and wildfires. The liquidation of a catastrophe bond’s collateral account for the benefit of the insurance company can be triggered by the output of a formula determined at the bond’s inception. The most common form of trigger is an ‘indemnity’ trigger where the collateral account is liquidated to the extent necessary to cover the actual loss experienced by the insurer issuing the bond.

???????????By contrast, ‘industry loss’ triggers are based on the losses experienced by the entire industry rather than a specific firm. Meanwhile, ‘modelled loss’ triggers are dependent on the output of a catastrophe model rather than actual insurance losses. Still other triggers are ‘parametric’ and depend simply on the occurrence of an observable event, for example, wind speeds of over 130 miles per hour over the U.S. state of Florida. These triggers are not even based on actual or projected insurance losses at all. While each trigger design has its advantages and disadvantages, the fundamental purpose is to put insurance companies in a better position to handle the risks posed by tail events at a reasonable cost. ?

Pandemic Emergency Financing Facility

???????????Though not at all as common as catastrophe bonds covering wind or earthquake risks, catastrophe bonds have also been issued covering pandemic risk. These were issued by the World Bank in 2017. The Pandemic Emergency Financing Facility (PEF) was established by the World Bank a year earlier to secure funding for countries at risk of such a health disaster. The creation of the PEF was inspired by the risks brought to the attention of world leaders and international organizations after a 2014 Ebola outbreak in West Africa.

???????????The PEF sought to address the lack of ‘surge financing’, the large amounts of money needed to confront disasters, available to poorer nations. Aid from wealthy donors to developing countries provides some regular support but facing disasters requires more money than governments and others are often inclined to give on short notice. Poorer countries may also have difficulty borrowing, especially in the midst of a pandemic. The PEF was also born at a moment when private sector involvement in public policy initiatives was being increasingly sought by the World Bank and its backers.

???????????The PEF was essentially an insurance fund in two parts. First, there was an ‘insurance window’ funded by the issuance of catastrophe bonds linked to the occurrence of a pandemic. Should the bonds trigger, the insurance window would have the cash needed to fund poorer pandemic-ravaged countries. Second, a ‘cash window’ was established by donor countries to fund disease containment in cases where the catastrophe bonds funding the insurance window did not trigger. The goal was that the insurance window would provide the required surge financing elusive in the past.

Pandemic Bonds

???????????Catastrophe bonds issued under the insurance window of the PEF covered six families of viruses deemed most likely to cause a future pandemic. These were orthomyxoviruses (causing influenza), coronaviridae (causing SARS, MERS, and later, Covid-19), filoviridae (responsible for Ebola) and other zoonotic diseases (Crimean Congo, Rift Valley, and Lassa fevers). The insurance window was funded by two classes of bonds and insurance-linked derivative ‘swap’ contracts sold in 2017. The Class A bonds and swaps raised $225 million and $50 million respectively and the riskier Class B bonds and swaps raised $95 million and $55 million respectively, for a total of $425 million lined up to cover pandemic risk.

???????????The bonds, issued by the World Bank’s International Bank for Reconstruction and Development, paid interest linked to the London Interbank Offered Rate (LIBOR). Insurance typically comes at a cost higher than government borrowing since repayment is not guaranteed. As a result, interest costs were high, LIBOR + 6.5% for the Class A bonds and LIBOR + 11.1% for the Class B bonds. Interest to investors was paid for by donor countries, specifically the governments of Japan, Germany, and Australia.

???????????The Class A bonds covered influenza and coronavirus, though it covered coronavirus risk only for up to 16.67% of the Class A bond and swap amount. The Class B bonds covered a broader range of diseases; they also covered coronavirus like the Class A bonds but also added coverage against filovirus, Crimean Congo Hemorrhagic Fever, Rift Valley Fever, and Lassa Fever. The bonds were issued on June 28, 2017 and were due to mature on July 15, 2020, subject to a possible one-year extension. Such optional extension periods are common in catastrophe bonds and provides time to estimate losses and pay out amounts due in case covered perils arise late in the life of the bond.

???????????The World Bank’s pandemic bonds were structured with the aid of the reinsurance companies Swiss Re and Munich Re and the firm AIR Worldwide was hired to provide ongoing reporting on the status of the bonds’ triggers. Investor demand was plentiful and the issuance was 200% oversubscribed, allowing the bonds to be priced at a rate lower than initial expectations and this after the transaction was enlarged. The bonds were largely sold to European investors and traditional catastrophe bond investors bought 61.7% of the Class A bonds and 35.5% of the Class B bonds. American investors and pension funds and other asset managers bought most of the rest.

Early Test

???????????Despite the investor demand, the pandemic bonds were exposed to the high risk of a trigger event almost from the start. An Ebola outbreak in the Democratic Republic of Congo in 2018 was its first major test. As a filovirus, Ebola was a covered peril under the pandemic bonds, but only exposed the Class B bonds as they were not a covered peril under the Class A bonds. In any case, even the Class B bonds continued performing without losses to investors. The 2018 Ebola outbreak did not meet all of the criteria set out in the bond documents to trigger a payment. It specifically fell short with respect to the criteria pertaining to the growth rate in cases, which required positive exponential growth.

???????????Critics of the PEF wondered how the 2018 return of the Ebola virus had failed to trigger a payment on the bonds that donor countries were paying so much to private investors to maintain. Sure, the bonds could trigger if the situation worsened but the best return on investment would be achieved if funds became available in advance of such a deterioration, the skeptics argued. Instead, the 100% donor funded cash window was utilized as the pandemic bonds never triggered during the outbreak and thus never provided funding to the potentially much larger insurance window. A total of $61.4 million was disbursed from the cash window alone to contain Ebola in the Democratic Republic of Congo in 2018 and 2019.

Covid-19

???????????The greatest test for bondholders came the following year when the first pandemic bonds ever issued were triggered in April 2020 during the Covid-19 pandemic. A summary of the triggers assigned to coronavirus outbreaks under the bond’s prospectus are listed below. The criteria required a minimum number of cases and deaths, an exponential growth rate, a minimum twelve weeks of outbreak, a minimum amount of geographic spread, and that at least 20% of the minimum required cases be confirmed. This is an example of a parametric trigger; the criteria had no relation to insurance losses, a structure appropriate for a program such as this which was not designed to compensate an insurance firm.

???????????A report showing that all of the conditions were satisfied to trigger losses to bondholders was published by AIR Worldwide on April 16, 2020. By then, almost 150,000 people had died of the virus globally. Critics, disappointed that the Ebola outbreak had failed to trigger payment under the bonds and swaps now accused the program of kicking in too slowly. Defenders argued that too much criticism was levelled at what was essentially a pilot, albeit an eventful one. Proponents also pointed out that though pandemic bonds may fund too late to raise the funds needed to stop an outbreak, they could provide a nonetheless timely source of longer-term funding for recovery.

???????????During the Covid-19 pandemic, the Class B bonds saw a full trigger and the less risky Class A bonds were partially triggered, losing the 16.67% of face value they were on the hook for with respect to coronaviruses. Together with the swaps, this meant the entire $150 million invested by Class B investors was forfeited and Class A investors lost 16.67% of their $275 million invested.

???????????In sum, $195.84 million was paid out to 64 of the poorest countries affected by the pandemic. For comparison, $107.2 million in premiums were paid to investors by donor countries for the coverage. The paltry payout, compared to the scale of funds needed to provide meaningful aid to pandemic-ravaged nations and compared to the premiums paid provided fuel for further criticism of the program, even after it had provided a net financial benefit to the World Bank and the beneficiary countries.

Future

???????????Whether as a result of the skeptics or otherwise, the World Bank opted not to continue the PEF program after the first issuance of pandemic bonds matured on July 15, 2020. Had it chosen to issue new bonds, the interest rate demanded by investors likely would have been higher anyway. It is common for returns demanded by catastrophe bond investors to increase after the occurrence of major disasters, whether hurricanes or earthquakes; the same would likely prove true of pandemics.

???????????Investors had also come to value the diversifying qualities of catastrophe bonds, helping them garner investor interest. However, the events of 2020 revealed that financial markets can become correlated to public health disasters, perhaps more so than they tend to with respect to hurricanes and earthquakes. As a result of all this, the terms of a second pandemic bond offering would likely have needed to be even more investor-friendly. Regardless, some hope that the PEF program could still prove to be a prototype for future pandemic bond issuances.

Lesson

???????????Economics and finance cannot benefit from the wonderful outcomes of the scientific method as much as natural sciences routinely do. Controlled experiments are not always helpful and experimentation with the real world is usually off limits; the economy is simply regarded as far too important to be tampered with for the sake of a doctoral thesis or even a Nobel Prize.

???????????Computer models will simply have to do, but as in other social sciences, or even in hard sciences where human behavior is nonetheless relevant … say epidemiology itself, there is often no perfect substitute for the real world. So, when the real world does give us something novel, like a pandemic-linked catastrophe bond on the eve of a global pandemic for instance, it should not go to waste. Perhaps pandemic bonds will not catch on, but if they do, future issuers will surely learn a lot from the first such bonds ever issued.

More from the Tontine Coffee-House

???????????Read about how a fire in Hamburg encouraged the creation of the reinsurance market. Also, consider subscribing to this blog’s newsletter?here.

Further Reading

1.??????DiFiore, Peter, et al. “Catastrophe Bonds: Natural Diversification.”?Neuberger Berman 'Insights', Jan. 2021.

2.??????“Fact Sheet: Pandemic Emergency Financing Facility.”?Worldbank.org, World Bank, 27 Apr. 2020.

3.??????“Prospectus Supplement - Class A Floating Rate Catastrophe-Linked Capital at Risk Notes Due July 15, 2020 and Class B Floating Rate Catastrophe-Linked Capital at Risk Notes Due July 15, 2020.” International Bank for Reconstruction and Development, 28 June 2017.

4.??????Vossos, Tasos, and Tracy Alloway. “Why World Bank's Controversial Pandemic Bonds Didn't Function as Hoped.”?Insurance Journal, Wells Media Group, Inc., 10 Dec. 2020.

5.??????“World Bank Launches First-Ever Pandemic Bonds to Support $500 Million Pandemic Emergency Financing Facility.”?Worldbank.org, World Bank (International Bank for Reconstruction and Development), 28 June 2017.

6.??????“World Bank Pandemic Catastrophe Bond.”?Artemis.bm, Steve Evans Ltd., 27 Apr. 2020.