Small 2005 States Forum 2005 Annual Meetings World Bank Group/International Monetary Fund Washington, DC September 24, 2005 www.worldbank.org/smallstates The Role of Insurance in Managing Recovery from Disaster in Small States Session II Presentation by Sir John Vereker Sir John Vereker is Governor and Commander in Chief of Bermuda. He was Permanent Secretary of the United Kingdom s Department of International Development from 1994-2002. This paper is written in a personal capacity.
THE ROLE OF INSURANCE IN MANAGING RECOVERY FROM DISASTER IN SMALL STATES 1. Small states are regularly subject to natural disasters. These range from the relatively routine and manageable (such as local flooding after heavy rain) to the catastrophic (such as volcanic eruption, or earthquake-triggered tsunami), capable of destroying whole communities. The most common disaster is, however, a hurricane. This paper looks at the role that insurance mechanisms may play in helping small states to recover from natural disasters on the basis of the experience of Bermuda, a British Overseas Territory, in managing the economic consequences of Hurricane Fabian. Some comparisons are drawn with the impact of Hurricane Ivan on Cayman, another British Overseas Territory. The two Territories concerned are of particular interest because assets in both are extensively insured, and because neither receives concessional flows or debt relief. Bermuda, 2003: Hurricane Fabian 2. On 5 September 2003, a Category 4 hurricane struck Bermuda with wind speeds of over 130 miles per hour; unrecorded gusts were much higher. The storm was large and hurricane force winds lasted about six hours. Storm surge brought flooding 6-10 feet above normal tide levels. Fabian was probably Bermuda s most severe storm in a century. Four lives were lost. Although property damage was extensive, most buildings remained habitable. Asset Loss 3. Various estimates have been made of physical asset damage and loss. An immediate insurance industry estimate based on catastrophe modeling put the cost to local and international insurers at between $300m and $350m 1. The Bermudian insurance company that carried about half of the local risk estimated the total loss to local insurers at about $125m 2. It is likely that the damage to infrastructure in Bermuda was less than shown in the normal models because of the high standard of construction of houses, which are built to withstand wind speeds of over 100mph. But both the two main hotels were extensively damaged and the insurance loss for them, carried outside Bermuda, was probably over $100m 3. There are no figures for uninsured private sector losses but they were probably low: most Bermudians carry property insurance. 4. The public sector was much less severely affected than the private sector. Damage to public infrastructure, including to the causeway connecting the airport and St George s to the rest of the island, sea walls and a cruise ship dock, was estimated by the Government of Bermuda to amount to $46m, of which about $30m was insured 4. 1 Press Notice, AIR Worldwide Corp., Sept 2003 2 Bermuda Fire & Marine, Special Report, Jan 2004 3 Industry estimate 4 2003 Economic Review, Ministry of Finance, Bermuda, Feb 2004
5. Taking all this evidence into account, a reasonable estimate of total asset loss would be: Insured private infrastructure $300m Uninsured private infrastructure $50m Insured public infrastructure $30m Uninsured public infrastructure $20m Economic Impact TOTAL ASSET LOSS $400m % GDP c10% 6. The two main pillars of Bermuda s economy, financial services and tourism, recovered quickly. Financial services were largely unaffected apart from a brief loss of electricity and the distraction of employees needing to repair their houses. Bermuda s main tourist season runs from April to November; although about 1200 beds, 20% of the hotel capacity, were lost, the impact of a hurricane at the beginning of September was limited to about two months of tourist receipts. Business travel resumed earlier. A year after the hurricane both sectors were growing robustly: employment in the financial services industry rose by 8.3% in 2004/05 and in the hotel sector by 3.3%. Unsurprisingly, construction sector employment increased by 8.6% 5. 7. Real GDP was forecast in February 2003 to grow in the range 1.0-1.5% in 2003/04 6. In practice it was estimated to have been 2.1% 7. That reflects Bermuda s integration into the global economy global economic growth was 2.5% in 2003 as much as the effect of post-hurricane economic activity. But the hurricane, despite its severity and the damage caused, certainly does not appear to have had an adverse effect on economic growth even in the year in which it struck. Growth in 2004/05 was forecast in February 2004 to be in the range 2.0-2.5% 8 ; it is now estimated to have been nearly 3% 9. Again, there is no discernable hurricane effect. 8. The Government s revenue estimate for 2003/04 of $650m was exceeded by some $16m or about 2.5% 10, largely due to increased payroll tax and customs duties. The Government s total spending estimate of $727m was exceeded by some $9m, so the expected budget deficit narrowed from $78m to $70m 11. Recurrent spending was about $18m higher than the original estimate, while capital spending fell from a projected $103m to $93m. 9. Bermuda has a statutory ceiling of $250m on public debt, currently about 6.25% of GDP. Actual debt is only about 4% of GDP, but the Government has announced its intention to raise the statutory limit to $375m (about 9% of GDP) in order to accommodate some long-standing 5 Budget Statement, Government of Bermuda, Feb 2005 6 Budget Statement, Government of Bermuda, Feb 2003 7 The 2003 Government of Bermuda s Economic Review, Feb 2004, estimated growth in 2003 at 2.5%; that was revised downwards to 2.1% in the Budget Statement of Feb 2005 8 2003 Economic Review, Government of Bermuda, Feb 2004 9 Budget Statement, Government of Bermuda, Feb 2004 10 2003 Economic Review, Government of Bermuda, Feb 2004 11 Ibid, as in errata slip 2
requirements for capital projects 12. Hurricane Fabian had little impact on indebtedness. Net borrowing in 2003/04 was only $18m. In 2004/05 revenue was also higher, and spending lower, than forecast; the borrowing requirement was again less than $20m 13. 10. In short, Bermuda survived as asset loss amounting to some 10% of GDP with economic growth, fiscal discipline and borrowing discipline all intact. Cayman, 2004: Hurricane Ivan 11. On September 11 2004, a Category 4 hurricane (Ivan) struck Grand Cayman, having devastated Grenada on September 7-9. The eye of the hurricane passed 21 SW of Grand Cayman with winds of over 150mph. The storm was very large, and slow moving: hurricane force winds lasted over 24 hours. The storm surge peaked at 10ft. Two lives were lost, and about 17% of the population was displaced. Property damage was extensive: unlike Bermuda, Grand Cayman is flat and 75% of the damage resulted from storm surge. Some 10,000 of the 42,000 population of Grand Cayman left the island in the following three weeks. The per capita damage was extraordinarily high around $75,000 a head. Residents and contractors took over almost all of the unaffected tourist accommodation for several months. Asset Loss 12. There are few reliable estimates of physical damage and loss. The main source is a UNDP/ECLAC paper of December 2004 that was prepared at the request of the Cayman Islands Government. This data is not directly comparable to that for Bermuda after its hurricane a year earlier. The estimates of the cost of loss and damage in that paper are: Public Infrastructure $ 150m Private Infrastructure $3,200 TOTAL ASSET LOSS $3,350m 14 %GDP c180% 13. It is not known how much of this was insured and how much uninsured, but the insurance industry calculated later that the insured damage amounted to only $1.4bn 15, of which 85% was paid by May 2005. So, if all these figures are correct, over half of the private loss must have been uninsured. Many homeowners were insured only for the amount of their outstanding mortgages. 12 Budget Statement, Government of Bermuda, Feb 2005 13 Ibid, Table IX 14 The Impact of Hurricane Ivan, ECLAC, Dec 2004 15 Report to Cabinet, Cayman Islands Monetary Authority, May 2005 3
Economic Impact 14. Hurricane Ivan had a much more significant economic impact on Cayman than Hurricane Fabian had on Bermuda. The damage to physical infrastructure was at a level of severity that interrupted productive economic activity for substantial periods of time. Financial services recovered relatively quickly, but tourism was badly affected, especially the more lucrative overnight (as opposed to cruise ship) business. There was a significant medium term loss of tourist receipts, estimated at $86m in 2004, which will continue into 2005 16. The construction industry was unable to cope with the post-ivan demand and inflationary pressures soon emerged in the rental and construction sectors. 15. As a result of the hurricane the economy contracted in 2004: real GDP growth was expected to fall from a forecast +3.1% to 2.2% 17. 16. At the end of 2004, it was estimated that the overall central government fiscal balance had moved from a forecast $29m surplus to an estimated $46m deficit, with revenue falling 5%, current expenditure rising 9% and capital expenditure rising by 75% 18. 17. Cayman s government debt, like that of Bermuda, is managed conservatively; the Public Management and Finance Law requires debt service to be less than 10% of recurrent revenue. New borrowing in 2004/05 had been forecast at $21m 19 ; in the event, after the hurricane, it was $75m 20. 18. In short, Cayman s extensive asset loss, exceeding one year s GDP, much of it uninsured, led to a contraction of the economy, a sharp deterioration in the fiscal balance and an increased indebtedness, but the underlying economy was strong enough to sustain these shocks and to build a strong recovery. The Role of Insurance Schemes 19. There are, broadly speaking, two ways in which insurance can cover natural disaster: payments can be triggered by damage to specified assets and related business activity; or payments can be triggered by natural phenomena of a specified level of severity. In the cases of both Bermuda and Cayman, reconstruction of physical infrastructure was aided by substantial payments from conventional insurance of the first kind, i.e. linked to damage to specific assets. Such payments are, typically, concentrated in the private sector (90% or more in each case); made within a year of claim; and disbursed without conditions. 20. The general relevance of the experience of Bermuda and Cayman is clearly limited by the fact that they are both strong economies with low levels of debt and high GNI per capita. They were much better placed than many small states to withstand the impact on their economies of 16 Ibid 17 Ibid 18 Ibid 19 CI Strategic Plan for FY ending 30 June 2005, March 2004 20 This was the new level approved by the UK; actual borrowing may been lower. 4
high levels of asset loss. But their experience does illustrate the significance of insurance payments in recovery from natural disaster, as well as some of the other constraints, notably local capacity. They also benefited from the enforcement of robust planning regulations and building standards. 21. Less prosperous countries tend to rely on external assistance to help their recovery from natural disaster. Immediate humanitarian assistance to save lives and relieve suffering tends to be disbursed quickly. Assistance for rehabilitation and reconstruction tends to take much longer. The rationale behind the proposed Natural Disaster Insurance Scheme for small states is that post-disaster reconstruction is often constrained by the complexity of donor approval processes, at a time when the decision making capacity of the state concerned is already under strain. An insurance scheme, under which payments were triggered by parametric criteria, so that no negotiation was required, could provide for more rapid disbursement of funds for reconstruction. It could provide breathing space for countries without sufficient budget capacity to withstand the immediate fiscal shock. 22. Some schemes are already in place. One private sector scheme the Commonwealth Disaster Management Agency Ltd (CDMA) focuses on credit support insurance to enable small states to continue to service external debt after a natural disaster. CDMA developed this scheme in response to a report discussed in the Forum in 2000 21. Rainfall-indexed catastrophe bonds are available in Tunisia and Morocco, which pay out automatically to farmers when rainfall is low 22. But in many small states there are virtually no risk transfer mechanisms in place. Government assets are often not insured and there are no national recovery funds. 23. It is true that post-disaster economies, like post-chaos economies, tend to grow faster than normal, for well-established reasons. Private flows increase in response to insured losses. Official flows may increase if there is humanitarian aid or debt relief related to the disaster. Public finance benefits on the current account from increased tax take on imports and reduced opportunities to spend. On the other hand, uninsured public infrastructure has to be replaced, putting pressure on the capital account. Borrowing usually increases; and the construction industry usually overheats, with demand running well ahead of capacity. After a severe or catastrophic disaster, labour shortages may compound the problems. 24. In the case of Bermuda, it is evident that there was very little impact on growth or public finance. That was chiefly because economic activity other than tourism was interrupted only for a few days, and because the main losses were insured. Capacity constraints quickly emerged in the construction industry and in the labour market more generally. In the weeks after Hurricane Fabian, truck drivers were demanding $80 an hour (about the normal price of a highly skilled mechanic) merely to remove debris. Capacity, rather than borrowing power, limited the pace of repair and replacement of public infrastructure. 25. In the case of Cayman, there was a significant but short-term impact on GNP, notwithstanding the extent of insurance, because the severity of the storm resulted in extensive 21 Small States: Meeting Challenges in the Global Economy, Commonwealth Secretariat/World Bank Joint Task Force Report, April 2000 22 Details in World Bank Policy Research Paper 2577, April 2001 5
loss of both public infrastructure and private housing. Had both Bermuda and Cayman been covered by the proposed natural disaster insurance scheme, the parametric criteria might have triggered a higher payment to Cayman, but the storms were of comparable size. As it was, the level of insurance claims experienced in Cayman caused difficulties for some insurers; one was forced into liquidation, two had to recapitalize and one was partly taken over by the Government. Increased premiums mean that even fewer individuals and businesses could then afford adequate insurance coverage. Labour shortages were such that an outside contractor was brought in from the US to help with the clean up. The economy is however now recovering strongly 3.7% in 2004/05 23 - aided by rapid disbursement against insurance claims, robust growth in financial services and resurgent property development. 26. Both Territories have well established and robust mechanisms for budgetary control. In the case of Bermuda a separate budget line was established, and controlled by the Ministry of Finance, for spending directly related to the recovery and reconstruction costs. That enabled Government Departments to plan spending beyond their original appropriations without going through time consuming process, and, given the robustness of revenues, without fiscal irresponsibility. 27. In the case of both Bermuda and Cayman, an insurance payment linked to predetermined natural events of a certain severity would have enabled otherwise uninsured public infrastructure to be repaired or replaced without recourse to increased borrowing. But it would not necessarily have been best for such a payment to be made in a lump sum immediately after the event. Decision-making capacity (e.g. should the causeway be repaired or replaced by a bridge; what were the respective priorities of road repair, school repair and dock repair; how much short term foreign labour should be imported?) and construction capacity were both already fully stretched. A sudden additional inflow could have undermined budget discipline, priority setting and monetary stability. Points for Discussion 28. A number of issues arise from this analysis, which the forum may find it helpful to discuss: (1) What are the limiting factors in post-disaster recovery? Those who have experienced the aftermath of natural disaster will know that direction and management are often one of the early constraints. A disaster, especially a catastrophic one, tends to erode the decisionmaking ability of parts of the government: key figures may be lost, or may become inaccessible; unfamiliar pressures, including outside media pressure, may be too great to handle; the decisions needed may be of an order which is unfamiliar to small states. Management capacity will be stretched to the limit, especially in the public sector. Furthermore, the replacement or repair of damaged infrastructure will rapidly fill up the capacity of the local construction industry. It can respond to some extent by taking on more unskilled labour, but the constraints of available capital equipment, building supplies and skilled labour will soon be felt. Outside contractors may need to be brought in. 23 Strategic Policy Statement, Cayman Island Government, August 2005 6
In the short term, these factors may be more significant than the financial constraints. As illustrated above, enhanced receipts of taxes and import duties mitigate immediate pressures on the recurrent budget. Budget discipline is unlikely to constrain spending on immediate post-disaster recovery. On the capital side, however, uninsured public infrastructure will need to be repaired or replaced, and in the longer term budgetary issues will become acute. Small states will, typically, need to increase borrowing, within the limits of prudent debt management. (2) Would a natural disaster insurance scheme for small states help? This paper does not address the issues of the design and affordability of such a scheme. But it does illustrate the potential significance of insurance payments for small states needing to recover from a disaster. The current high level of hurricane activity, since 1995, probably marks the start of a long-term cycle that may last 25-40 years 24 ; and hurricane damage in a small state easily reaches a significant proportion of GDP. Many large states already depend heavily on insurance 25. Many small states whose physical assets are not insured do not have economies strong enough, or large enough, to cope with a major disaster. A mechanism that triggers payments according to parametric criteria so that scarce administrative capacity is not taken up in complex claim negotiations has clear advantages. Furthermore insurers tend to enforce mitigation measures, such as compliance with building codes, in order to minimize risks, so the existence of an insurance scheme can also reduce losses. The lesson from both Bermuda and Cayman is that all small states would be well advised to consider to what extent their assets, both public and private, are already insured; what options are available to enhance that to an acceptable level in the light of the risk; and what coverage they can afford. (3) What other actions should a small state undertake as a part of its risk comprehensive risk management framework? Although the purchase of insurance protection against the financial and structural impacts of natural disasters may be a prudent and necessary part of a comprehensive risk management framework what other steps should be taken within such a framework in order to mitigate the impact of such a disaster? For example, what can be done to improve building codes, or enforce them if already in place? Should key utilities such as power lines be placed underground? Can mutual support agreements be reached with neighboring states to provide the human and material support needed to rebuild key infrastructure such as power lines, telephone line and cellular towers, roads, etc. Over time this would allow the country to reduce the required amount of disaster insurance coverage, and associated premium costs, as the infrastructure and associated business activities become more robust and less at risk of loss from a natural disaster. Hamilton August 2005 24 Hurricane Research Division of the National Hurricane Centre, NOAA, 2005 25 See, for instance, Catastrophe Risk: US and European Approaches, GAO, Feb 2005 7