Whilst a unifying definition may be debatable, demand surge can be thought of as a socio-economic product that follows many natural catastrophes where the collective demand for labour, construction materials and equipment significantly exceed the locally available supply. Mr James Knight of Aon Benfield Analytics explains.
Demand surge is insurer independent and can be a significant contributor to overall insured loss. It is also the largest component of non-modelled loss and, as a result, demand surge is of particular interest to property insurers, the public sector and governments that finance reconstruction following a major catastrophe.
Directly after an event, labour costs can temporarily increase as additional resource is brought in to assist the local shortfall in supply. In some instances, this additional labour resource can be less efficient and require training and certification to complete satisfactory repairs.
Increased labour and materials’ costs
Highly-skilled labour often receives a premium as temporary demand for their services increase which further drives up costs. Opportunistic contractors with a full schedule also tend to quote higher fees and end up getting the additional work due to the high demand, potentially leading to job banking and project delays. Any dislocated labour required following an event may also be associated with additional costs such as transport and accommodation.
The cost of construction materials can also increase rapidly as local supply becomes exhausted. A shortfall in local material leaves a choice to either use more expensive materials for construction, wait for new materials to be brought in from other neighbouring regions or import materials internationally. Imported materials come with potentially increased costs from currency fluctuations and additional transportation.
Factors that drive demand surge
The main contributing variables to demand surge respond in different ways and are dependent on local conditions. There are however a number of key common variables that drive demand surge:
Event size
The size of an event represents the level of economic loss. Event size is a factor of both the severity of an event, having influence on the level of potential damage, coupled with the value of assets exposed to an event. Size also indicates the total amount of repair work that is required and therefore the level of additional labour and materials needed to meet demand. The likely duration of the reconstruction phase (length of increased demand) is also indicated by the event size.
Event location
Isolated areas that have limited access to labour and materials at the time of an event are more vulnerable to rapid increases in demand. Although the total amount of repair work is likely to be much less in remote areas compared to larger urban centres, the supply gap and associated costs of bringing in additional labour and materials into these isolated regions will be comparatively higher.
Economy
The state of the local economy at the time of the event (in particular the construction industry) gives a good idea of industry capacity. This economic capacity indicates the amount of labour and materials that are locally available and highlights the significance of any shortfall in local supply. Capacity also indicates the willingness of international labour resources to move into an effected region following an event. Currency fluctuations have further influence on the costs of importing materials into a region to meet the temporary increase in demand.
Insurance policy conditions are another important factor influencing demand surge. At the time of the Canterbury earthquakes, most New Zealand insurers offered full replacement residential policies. With an understanding that this would impact significantly on post-loss amplification, insurers established Project Management Offices (PMOs) to manage inflationary factors relating to repairs for damage.
Other factors influencing demand surge include claims management and loss adjusting practice as well as the specific type and variety of damage experienced, particularly when insurance policies are capped by sum insured values.
Historical experience
Major surges in the demand for reconstruction and repair following natural catastrophes have been seen throughout the history of insurance where local supply is unable to match demand.
In the US, Hurricane Katrina made landfall at a time when construction materials were already locally in high demand. The supply chain was severely disrupted as a result of major damage to key manufacturing facilities, unable to distribute required materials when needed most. A shortage of materials combined with a shortage of skilled labour in the area contributed to total reconstruction costs 10-15% above pre-event price levels (Hallegate, 2008).
Another example of demand surge was seen after the Northridge (California) earthquake in January 1994 where, although the cost of materials remained static, labour costs from a less competitive environment and a larger, less efficient workforce inflated total insured losses by up to 20% over pre-event levels (Kuzak and Larsen, 2005).
In Australia, the reconstruction of Darwin following Cyclone Tracy in December 1974 saw insured losses settled at almost double the reconstruction cost that was anticipated before the event. After Cyclone Larry in March 2006, increased costs “of at least 50%” were estimated by many insurers (ASIC, 2007) and demand surge was also observed after the Canberra bushfires in January 2003 and the Sydney Hailstorm in April 1999 (Olsen and Porter, 2010). Although significant demand surge has been observed following these events, underinsurance is another important additional factor driving post-loss amplification.
In New Zealand, we are yet to understand the full impact of demand surge following the two major Canterbury earthquakes in September 2010 and February 2011 as the rebuild continues. Many insurers quickly realised that in order to rebuild a resilient Canterbury, a number of management measures around the repair programme needed to be introduced to ensure reconstruction and repair was completed not just in a timely manner but also remain economically viable.
It is also likely that Canterbury has seen some degree of betterment as building codes and standards are followed. Consistent repair strategies and the establishment of PMOs are some examples of the measures that have been taken in New Zealand to minimise demand surge.
Impact from demand surge still small presently
Although the above examples show demand surge as a significant contributor to insured losses, many other major natural catastrophes show little impact from demand surge on overall insured losses.
Some examples of major catastrophes where demand surge has been minimal include the Loma Prieta (US) and Newcastle (Australia) earthquakes, both in 1989, where local supply matched the temporary demand increase (Olsen and Porter, 2010).
From history, it appears that demand surge is minimised when the construction industry is allowed to remain competitive throughout the reconstruction phase and additional demand for both labour and materials can be met largely by the local supply.
Catastrophe modelling implications
As we have experienced in the past, demand surge has long been recognised as one of the most significant components of non-modelled loss amplification for natural catastrophes and yet many models fail to adequately consider demand surge.
In New Zealand, demand surge is considered by some catastrophe models but have been developed around conditions and experience in the US, using a simple scaling factor applied to ground-up losses. As demand surge is strongly location and/or event specific, a single scaling factor is not ideal and does not reflect either the size of the event or any local variation in supply and demand. A local analysis of all key contributing variables that drive demand surge, together with a complete understanding of the impact these variables have, appears to be the most appropriate way to consider demand surge in any insured loss estimation.
Catastrophe models may benefit from not considering demand surge at all as they are updated so infrequently that any attempt to consider such dynamic variables would introduce further uncertainty.
As local markets and the construction industry are cyclical in nature, it would be difficult for any catastrophe model to adequately account for this magnitude and variation of change. Demand surge is therefore best considered outside of catastrophe models using dynamic financial analysis modelling platforms such as ReMetrica®.
Demand surge model for Australia developed
Aon has found that local indices including building approvals, labour force statistics and the producer price index are suitable proxies for local demand, supply and price respectively in Australia.
Using key statistics relating to each variable combined with internal estimates of market loss, a demand surge model for Australia has been developed that can be applied to any past or future event and is specifically tailored to the local region. The model is also regularly updated to accurately reflect current conditions across Australia. The resultant measurement offers a robust estimate of demand surge as a percentage which varies spatially by the local supply available at the time of an event. The Aon Demand Surge model is currently used by our Australian clients with one client stating the model was “the most rigorous approach they’d seen to what is a difficult problem”.
Mr James Knight is a Catastrophe Research Analyst at Aon Benfield Analytics.
Aon Benfield is the 2014 winner for Reinsurance Broker of the Year at the 18th Asia Insurance Industry Awards.