Error! Reference source not found.
Appendix D:Managing the Cost of Damage to Road Infrastructure Caused by Natural Disaster - National Pool Approach - August 2012Disaster
APPENDIX D | NDRRA PHASE 2 REPORT
Australian Government Actuary
Error! Reference source not found.
Copyright Notice
ISBN: 978-1-922096-15-9 (print)
ISBN: 978-1-922096-16-6 (online)
With the exception of the Commonwealth Coat of Arms, this work is licensed under a Creative Commons Attribution 3.0 Australia licence (CC BY 3.0) (
This work must be attributed as: “Commonwealth of Australia, Department of Finance and Deregulation, Review of the Insurance Arrangements of State and Territory Governments under the Natural Disaster Relief and Recovery Arrangements Determination 2011”
Australian Government Actuary
Error! Reference source not found.
Managing the cost of damage to road infrastructure caused by natural disaster - national pool approach
Australian Government Actuary
Error! Reference source not found.
TABLE OF CONTENTS
1Introduction
2Scope of task
3Preliminaries
4Cost sharing – equitable contributions
5Funding
6Membership basis – compulsory or voluntary?
7Governance
8Conclusion
Australian Government Actuary
1.Introduction
1.1The Queensland floods of 2011 resulted in expenditure of hundreds of millions of dollars related to the repair of damaged road infrastructure. The direct costs associated with this event were shared between Queensland and the Commonwealth.
1.2At the time, Queensland did not have commercial insurance cover for its road infrastructure. We believe that Queensland has approached the commercial insurance market subsequently. Although it has purchased commercial insurance over elements of essential public infrastructure, we believe that it has not been able to secure insurance over its road network.
1.3Swiss Re has confirmed that it can be difficult to insure road infrastructure for damage from natural disasters in the commercial insurance market.
1.4It would be risky to assume, therefore, that commercial insurance will present a sustainable solution to the problem of damage to road infrastructure caused by natural disaster. Rather, it would be prudent to consider what other options might be available to manage the risk associated with the cost of damage to road infrastructure caused by natural disaster, on the assumption that commercial insurance is unlikely to be a sustainable solution.
2.Scope of task
1.5AGA has been asked to prepare a brief paper that considers a consolidated pool approach, whereby the Commonwealth, together with states and territories, would operate a ‘road insurance pool’.
1.6AGA has been asked to consider the following issues:
•How a pool could operate
•Compulsory or optional membership
•Governance
•Funding
•Equitable contributions
•Pros and cons
1.7The intention is that this paper is at a high level. It discusses principles, but not items of detail.
1.8This paper does not consider any other options. This should not be taken to imply that other options are not available or should not be considered.
1.9AGA’s starting point was to look at the possibility of a designing a road pool model which does not result in cost shifts from the Commonwealth to states and territories or vice versa but which, at the same time, deliversappropriate risk management signals to the owners of road assets.
1.10This is not intended to imply that AGA favours or recommends a cost-neutral model. AGA is indifferent on this. The only point that we would make is that there does not seem any reason for an increase in expected Commonwealth costs, relative to existing NDRRA arrangements.
3.Preliminaries
1.11Many design issues need to be considered but, first, it is helpful to try to identify a set of principles which should underpin the design of any model.
1.12Useful principles are likely to include:
•practicality
•robustness
•reasonableness
•efficiency
1.13Practicality refers to ease of implementation and administration.
1.14Robustness refers to vulnerability to manipulation. It also refers to the extent to which the arrangement provides sound behavioural incentives. A system which can be manipulated by either states and territories or the Commonwealth to their own benefit is not robust. On the other hand, a system which naturally results in sound risk management behaviours by all contributing governments is more likely to be robust.
1.15Reasonableness refers to the relative contributions required of each party.
1.16Efficiency refers to the extent to which available resources are used productively. For example, setting aside and quarantining large reserves to fund contingent liabilities which only have a small probability of being crystallised might not be efficient. This is akin to having a ‘lazy’ balance sheet in a business context.
1.17Candidate models could helpfully be assessed against these principles.
1.18Under a pool model, all jurisdictions would share in the direct costs associated with road damage arising in any given jurisdiction due to natural disaster.
1.19A pool model can be thought of, at least in this regard, as an extension of the current NDRRA arrangements where the only explicit cost sharing is between the Commonwealth and the jurisdiction that the damage occurs in.
1.20Key design issues for a pool include:
•the basis of the cost-sharing/contribution arrangements
•the funding model
4.Cost sharing – equitable contributions
1.21It is helpful to consider two different ways that costs can be shared between parties. These can be described as:
• proportional
• non-proportional
Proportional
1.22Under a pure proportional cost-sharing arrangement, each party would be responsible for a pre-determined proportion of the costs of road damage in a jurisdiction, regardless of where the damage occurred.
1.23In respect of a possible pool model for road damage, the factors that might be taken into account in determining proportions include:
•revenue base of each jurisdiction
•relative exposure to natural hazard risk of each jurisdiction
1.24A graphical depiction is presented below. This depiction has the Commonwealth meeting the majority of the cost of any natural disaster because of its relatively large revenue base. The remainder of the costs are shared between the other jurisdictions, having regard to both the revenue base and relative exposure to natural disaster risk of each jurisdiction.
1.25In principle, it would be possible to calibrate a proportional cost-sharing arrangement in a way that resulted in the expected costs for each jurisdiction being the same as under the existing NDRRA arrangements. However, this would be difficult to do in practice.
Non-proportional
1.26Under a pure non-proportional cost-sharing arrangement, the costs would be shared between participants on the basis of ‘layers of cost’, rather than proportions. For example, the jurisdiction in which the damage occurred might meet the first $x of the repair cost, with at least part of the excess, if any, being met by other jurisdictions. $x is referred to as the first layer of cost.
1.27 In practice, a non-proportional model for this purpose is likely to involve both layers and proportions. For example, if the first $x is met by the jurisdiction in which the damage occurs, the next $y might be shared between jurisdictions in fixed proportions. $y is referred to as the first excess layer of cost.
1.28One of a number of possible non-proportional structures would see the ‘first layer’ of cost being met by the jurisdiction in which the damage occurs, the ‘first excess layer’ being shared between jurisdictions (other than the Commonwealth) in fixed proportions, and any additional cost being shared between the Commonwealth andthe owner jurisdiction.
1.29A graphical depiction of how such an arrangement might operate is provided below.
1.30In respect of such a non-proportional pool model for road damage, the factors that might be taken into account in determining layers and proportions might include:
•jurisdiction in which the damage occurred
•revenue base of each jurisdiction
•relative exposure to natural hazard risk of each jurisdiction
1.31If the owner jurisdiction were required to fund a relatively large first layer of costs, an increased emphasis on risk mitigation would be encouraged. The objective would be that, over the long run, costs would reduce.
1.32In principle, it would be possible to calibrate this non-proportional cost-sharing arrangement in way that resulted in the expected costs for each jurisdiction, including the Commonwealth, being the same initially as under the existing NDRRA arrangements. This would be the case even if the owner jurisdiction were required to fund a relatively large first layer of costs. However, it is important to note, again, that this would be difficult to do in practice.
Comparison of proportional and non-proportional models
1.33Under a pure proportional model, each jurisdiction is involved from the ‘ground up’. This means that the cost of a natural disaster is shared between all jurisdictions, regardless of how low the cost is (how small the disaster is).
1.34As well as this, each jurisdiction’s exposure is, in effect, open-ended in respect of any natural disaster event under a pure proportional model.
1.35Under a non-proportional model the ‘owner jurisdiction’ covers first layer costs in full. Other jurisdictions do not participate unless the relevant threshold has been exceeded and the total cost exceeds the first layer.Under the non-proportional arrangements described here, the Commonwealth would not participate in any claim unless the total costs exceeded the first excess layer.
1.36Under the non-proportional model described here, the Commonwealth’s exposure is uncapped in respect of any natural disaster event. The owner jurisdiction also has an open-ended exposure. Other states and territories would share the costs of the first excess layer. Their exposure is therefore capped at this level.
Per-event or per-year basis?
1.37Either model could be implemented on a per-event basis or a per-year basis. For example, on a per-event basis, either model would see the costs of each event shared according to the governing formula. On a per-year basis, the costs associated with natural disaster events that occurred during the year would be shared according to the governing formula.
1.38A per-year basis would be better linked to financial reporting cycles than a per-event basis. Further, a per-year basismight avoid a number of definitional difficulties that can be associated with a per-event basis.These difficulties can manifest themselves in a non-proportional arrangement (rather than a proportional arrangement). For example, it is not always apparent when one disaster stops and another starts. It can also be difficult to know whether damage has been caused by, say, a cyclone or by a flood happening around the same time.Under a non-proportional model, the owner jurisdiction must meet the first layer of cost. If this is applied on a per-event basis, then either of the two examples described above could lead to disputes around whether the owner jurisdiction was liable for one or two first layers of cost in some cases. Using a per-year basis largely avoids these issues.
1.39To operate on a per-year basis, it would be important, however, to ensure that the underlying costs (to be shared) were assessed on an ‘incurred’ basis and not a ‘paid’ basis. That is, a per-year basis would relate to the cost of damage arising due to natural disasters that occurred during a given year. It would not be based on the level of repair payments that were made during a particular year.
Assessment
1.40The table below assesses the two models against the criteria suggested above.
5.
Proportional / Non-proportionalPracticality / Reasonably easy to describe. However, all jurisdictions are involved in all claims, including small ones.
Some practicality issues when multiple jurisdictions suffer damage in the same event. / The ‘owner’ jurisdiction is responsible for small claims. Other jurisdictions participate on larger and very large claims. The Commonwealth only participates on very large claims.
Some practicality issues when multiple jurisdictions suffer damage in the same event.
Robustness / Based on the principle of ‘solidarity’ so can be vulnerable to moral hazard. Specifically, this model does not send strong risk management price signals in respect of small claims. / If a suitably high threshold is set, this model can deliver sound risk management behavioural signals to jurisdictions. This is because ‘owner jurisdictions’ are responsible for first layer costs.
Reasonableness / Will need to be calibrated carefully / Will need to be calibrated carefully
Efficiency / Depends more on funding model – not related to pool concept. Although, in principle diversifying risk enhances efficiency / Depends more on funding model – not related to pool concept. Although, in principle diversifying risk enhances efficiency
1.41This simple analysis suggests that a non-proportional model is to be preferred.
1.42The main reason is that the non-proportional model is more robust because it is better aligned with sound risk management behaviours. As well, it is more practical because cost-sharing only applies when the cost exceeds a defined threshold, rather than from the ground up as in the proportional model.
1.43It is noteworthy in this regard that commercial catastrophe reinsurance programs operate on a non-proportional basis.
1.44The Australian Government scheme to deal with the insurance of commercial property against terrorism risk also operates on that basis. The commercial insurance industry picks up the first layer of costs following a terrorism event. A pool of funds is available to fund costs in excess of this first layer. Contributions are made to the pool by all participating insurers according to a pre-determined formula. An insurer’s contributions are, in effect, a function of the insurer’s commercial property insurance premium revenue and its relative exposure to terrorism risk. The Commonwealth is responsible for very high layer costs. The terrorism scheme also accesses the commercial insurance market to provide further protection. The possible role for commercial insurance in the road pool context is discussed further below.
Risk management
1.45Under the non-proportional model described here, first layer costs are met in full by the ‘owner’ jurisdiction.
1.46The level of the first layer should be set so that it represented meaningful expenditure relative to the revenue base of the jurisdiction and the value of its road assets. The level would be set with a view to encouraging jurisdictions to undertake effective risk management practices (through this price signal).
1.47Notably, first layer costs are likely to be associated with relatively small natural disasters (as well as large disasters). In turn, relatively small natural disasters are likely to occur more frequently than larger disasters. In principle, therefore, by requiring jurisdictions to meet first layer cost in full, a price signal is sent which is intended to encourage jurisdictions to consider risk management possibilities for those roads which are most exposed to natural disaster risk. By way of example, jurisdictions might be encouraged to think carefully about whether to construct roads which will be exposed to frequent flooding. Where a jurisdiction judges that it is necessary to construct a road in a particular location despite the flood risk, it would then be encouraged to think carefully about how that road is constructed.
6.Funding
1.48There is a question around whether a reserve (dedicated pre-funded pool of money) should be created or whether costs should be met on a PAYG basis.
1.49The discussion below follows from the earlier analysis which suggested that a non-proportional model might be preferred to a proportional model.
1.50Under the type of non-proportional pool model described in this paper:
•first layer costs would be picked up by the owning jurisdiction
•first excess layer costs would be shared between non-Commonwealth jurisdictions (including the owner jurisdiction)
•the Commonwealth would share costs in excess of the first excess layer with the owning jurisdiction
1.51Under this structure, ‘owner’ jurisdictions meet first layer costs. It would be up to individual jurisdictions to decide whether or not to set aside reserves for this purpose.
1.52The next layer of costs would be shared between all the non-Commonwealth jurisdictions. Under this structure, there is an effective cap on the second layer exposure. The size of this second layer, of course, depends on how the model is calibrated. However, it could easily be several hundreds of millions of dollars.
1.53There would be a sound argument for funding this layer. Contributions would be from all participating jurisdictions in agreed proportions. A pool of several hundred million dollars could not reasonably be funded overnight. Rather, it is more likely that the pool would be built up over a period, perhaps of several years.
1.54If the reserve was drawn down following a sufficiently large disaster, then it would need to be topped up, again with contributions from all jurisdictions in agreed proportions.Again, the pool would not need to be replenished immediately. Rather it could be replenished over a period of years, if required. This is obviously a matter of design detail which is not considered further.
1.55The Commonwealth’s exposure is uncapped under this structure. There would be little merit in the Commonwealth pre-funding its exposure.
Commercial insurance
1.56Even though there is little merit in the Commonwealth pre-funding its exposure, there would be a possible case for the Commonwealth seeking to partially reinsure its exposure in the commercial market.
1.57The Australian Government’s terrorism insurance scheme (mentioned earlier) was developed because of a withdrawal of commercial terrorism insurance capacity. The Commonwealth (through the Australian Reinsurance Pool Corporation [ARPC]) has sought, in recent years, to enhance the protection offered under the scheme by purchasing commercial terrorism reinsurance in the commercial market. ARPC has, for the last few years, purchased about $2bn in cover, excess of around $300m. Thus, the first $300m of losses on any event would be retained by the ARPC with the commercial cover only responding if losses from a single event exceeded this amount.
1.58The motivation for seeking commercial reinsurance has been, in part, to try to stimulate, to the extent possible, the commercial market for terrorism insurance. In fact, the ARPC provides a degree of efficiency in this process by ‘bundling up’ all of the terrorism risk into one place. It is arguable that the offer of a large chunk of terrorism business to the commercial market is ‘more attractive’ than an offer of many small chunks of business (which would be the case absent the ARPC). Similarly, by maintaining a reasonable first layer to its own account (around $300m), the ARPC manages to ensure a level of price efficiency (high layers of cover cost less than low layers of cover, all else being equal).