Understanding Risk Accumulations in Taiwan’s Science Parks
Ben ReynoldsDecember 19, 2016
“The 6.4 magnitude Tainan earthquake in February 2016 resulted in a sizeable insured loss from the high-tech industrial risks and reminded the insurance industry of the potential threat from the risk accumulated in science parks.” (A.M. Best Special Report, Sept 2016)
Reading the sentence above you might be forgiven for wondering why science parks would give insurers and reinsurers any particular cause for concern. But consider this statistic: although Taiwan’s three major science and industrial parks occupy only 0.1% of the island’s total land mass, they represent 16% of Taiwan’s overall manufacturing – they are hugely significant, both economically and with regards to the insured exposure in Taiwan.
For example, the Hsinchu Science Park (HSP), known for semiconductor production, employs more than 150,000 people and contributes over $32 billion in revenues – approximately 6% of national GDP. By one estimate HSP represents over $319 billion in total insured values. In addition, some of the latest high tech areas within HSP, such as advanced “clean rooms,” present additional challenges due to their vulnerability to ground shaking or power interruption. The importance of this risk was observed in February’s Tainan earthquake where some significant losses to high-tech industrial risks were caused by damage to the equipment and the related business interruption due to power outage.
Improving data quality for advanced and detailed modeling is an important way to manage these risks, concludes the A.M. Best report quoted above. This is so as to accurately assess the potential loss impact on insurers’ books. RMS has already been analysing earthquake risk in Taiwan for 12 years – long before this year’s Mw 6.4 event – and in that time our view of seismic risk in Taiwan has not changed, since our model benefits from spectral response-based hazard and damage functions, that even include local liquefaction and landslide susceptibilities.
The 1999 Chi-Chi Earthquake (known in Taiwan as the 921 Earthquake) was the key event in building the RMS® Taiwan Earthquake Model in terms of the quake’s seismicity, ground motion, soil secondary effects and building response. Since then there have been no significant events to justify a re-calibration of the components of the model. In fact, the damages observed in this year’s event were broadly in line with RMS’ expectations and validated the robustness of the current model.
But although A.M. Best views the Taiwan insurance industry as prudently managed with relatively high catastrophe management capability, there are still lessons to be learnt from the 2016 event, and RMS has solutions which offer additional insight into understanding the risk posed by these business parks in Taiwan.
Concentration of Exposure into Science Parks
The RMS® Asia Industrial Clusters Catalogs were released in 2014 to identify hotspots of exposure, and profile their risk. The locations and geographic extent of the science parks within Taiwan are detailed to help understand risk accumulations for industrial lines and develop more robust risk management strategies.
High Fragility of the Semiconductor Industry
For coding of Industrial Plants, the RMS® Industrial Facilities Model (IFM) captures the unique nature of different industrial risks, as a high percentage of property value is often associated with machinery and equipment (M&E) and stock. This advanced vulnerability model supports the earthquake model to define the damageability of a comprehensive set of industrial facilities more accurately, and calculate the financial risk to these specific types of facilities, including building, contents, and business interruption (BI) loss estimates. The IFM differentiates the risks for different types of business within the science parks, and highlights the higher fragility of semiconductor plants compared to other industrial units, as shown below.
The huge damage from the 1999 Chi Chi earthquake has not halted the rapid development of Taiwan’s science parks in this seismically active area – indeed the island’s third biggest science park has since been built there. But this year’s comparatively small Mw 6.4 event further highlighted the substantial exposures concentrated within this sector, reminding the industry of the potential for significant losses without sound accumulation management practices, informed by the best modeling insights.
You May Also Like
December 08, 2017
Iran-Iraq Earthquake: Understanding Both the Causes and Remedies to Build Resilience
In a year that has so far seen a below-average seven M7+ earthquakes globally, the Mw 7.3 earthquake that struck the mountainous Iran-Iraq border region late evening local time on Sunday November 12, ranks as the most destructive and deadly earthquake of 2017. It was the eighth strongest to hit the region since 1900, and the strongest to hit within 150 miles (250 kilometers) in the last century. The epicenter was located 20 miles (32 kilometers) south of the city of Halabja in Iraq. Reports state that over 500 people were killed and over 8,000 injured, according to the United States Geological Survey (USGS).
Figure 1: Map showing USGS PGA footprint and USGS Moment Tensor at the epicenter for the eventIn Iraq, at least nine people were killed and 535 injured, with the most severe shaking affecting a smaller, more sparsely populated area. The worst-affected settlements are reported to be the city of Khanaqin and the town of Darbandikhan, both in Sulaymaniyah Governate.
Iran suffered the biggest impact — Iran’s National Disaster Management Organization (NDMO) stated that 123 settlements have been damaged. An estimated 11,000 rural dwellings were destroyed, alongside thousands of homes in taller multifamily dwellings in the town and cities. The total estimate of damaged buildings ranges up to 30,000 and leaves tens of thousands of people homeless.
Many towns and cities in the Iranian province of Kermanshah, close to the Iraq border, were affected. The town of Sarpol-e Zahab, with a population of 45,000 and some 20 miles (32 kilometers) from the border saw about 90 percent of buildings destroyed. The hospital, which is Kermanshah Province’s primary hospital, is reported to be severely damaged. More than 140 fatalities are reported in this town alone.
Across Kermanshah Province, damage was also reported in the province capital Kermanshah with a population of around 850,000, and ranged from Eslamabad-e-Gharb in the south of the province and Paveh in the north. The earthquake affected the city of Ilam in Ilam Province south of Kermanshah, and Khorramabad and Borujerd in Lorestan Province, which are about 170 miles (280 kilometers) from the epicenter. In Borujerd alone, 354 buildings are reported damaged.
A “Slip Over a Larger Fault Area”
The Zagros Mountains, spanning from southeastern Turkey and through the Iranian western border with Iraq, contains the 1,100 mile-long (1,800 kilometer) Zagros fold and thrust belt which marks the boundary between the Arabia and Eurasia plates, and is a known zone of active crustal deformation and seismicity.
This earthquake happened due to oblique-thrust faulting where the plates converge. The size and mechanism of the quake mean that it is better thought of as a “slip over a larger fault area” than a point event, according to the USGS. Events of this magnitude may have a length of about 40 miles (65 kilometers), and damage will therefore occur at some distance from the designated epicenter.
This region is sparsely populated, otherwise the number of deaths may have been in the tens of thousands, similar to the 2003 earthquake that hit the ancient city of Bam in southeast Iran.
Although the earthquake risk is well recognized, there is not enough consideration of the impacts that large earthquakes could have on the buildings in this region, which is one reason the event lead to so many collapsed buildings and deaths.
Damage has been particularly severe to state-subsidized buildings constructed under the Mehr housing program. The program, started by President Ahmadinejad in 2007, had the aim to build two million affordable homes within five years. By 2014, some 1.18 million residences had been built, with developers offering free government land to build affordable housing units for first-time owners. Reports suggest that 90 percent of the Mehr housing units in Kermanshah Province were damaged. It is unclear whether construction regulations were followed, and many of the failed or partially destroyed apartment blocks appear to be made with unreinforced masonry infill walls that have crumpled.
Improving Risk Insight; Improving Resilience
RMS is currently developing earthquake models for this region to aid global reinsurers in estimating the earthquake risk, and to help development of reinsurance programs to cover such devastating events. These models need to evaluate the potential correlation of risk posed by “slip over a larger fault area” leading to correlated damages along the strike orientation of the fault plane. If earthquakes are modeled based on epicenters or point events, footprints fail to capture the correlation appropriately.
The RMS approach of using fault rupture planes to release seismic energy allow appropriate calculation of the spatial correlations of hazard and loss, which is why in our event descriptions we do not consider earthquake epicenters.
Similar to many other earthquake-prone regions, the core principles of building resilience need to be robustly pursued, such as risk avoidance by stricter building codes; risk control through education programs to reduce potential impacts; risk acceptance by budgeting for costs of future events, and potential risk transfer by the development of insurance programs. As we build new models of earthquake risk for the region we will quantify the seismic risk and the potential losses any future events may have, and we hope these enable better organization to develop risk strategies to build a more resilient future.…
Those were the words of the then Italian Prime Minister, Matteo Renzi, in the aftermath of two earthquakes on the same day, October 26, 2016. As a statement of indomitable defiance at a scene of devastation it suited the political and public mood well. But the simple fact is there is work to do, because Italy is not as strong as it could be in its resilience to earthquakes.
There’s a long history of powerful seismic activity in the central Apennines: only recently we’ve seen L’Aquila (2009, Mw6.3), Amatrice (August 2016, Mw6.0), two earthquakes in the area near Visso (October 2016, Mw 5.4 and 5.9) and Norcia (October 2016, Mw6.5). These have resulted in hundreds of fatalities, mainly attributed to widespread collapse of old buildings, emphasizing that earthquakes don’t kill people – buildings do. Whilst Italy’s Civil Protection Department provides emergency management and support after earthquakes, there is too little insurance help for the financial resiliency of the communities most affected by all these events. While the oft-repeated call for earthquake insurance to be compulsory continues to be politically unobtainable, one way it could be spread more widely is through effective modeling. And RMS expertise can help with this, allowing the market to better understand the risk and so build resilience.
Examining High Building Fragility
The two most significant factors for earthquake risk in Italy are (i) construction materials and (ii) the age of the buildings. The majority of the damaged and destroyed buildings were made from unreinforced masonry, and built prior to the introduction of the most recent seismic design and building codes, making them particularly susceptible. With the RMS® Europe Earthquake model capturing both the variations in construction types and age, as well as other vulnerability factors, (re)insurers can accurately reflect the response of different structures to earthquakes. This allows the models to be used to evaluate the cost benefits of retrofitting buildings. RMS has worked with the Italian National Institute for Geophysics and Volcanology (INGV) to see how such analyses could be used to optimize the allocation of public funds for strengthening older buildings, thereby reducing future damage and costs.
Seismic Risk Assessment
The high-risk zone of the central Apennines is described well by probabilistic seismic hazard assessment (PSHA) maps, which show the highest risks in that region resulting from the movement of tectonic blocks that produce the extensional, ‘normal’ faulting observed. The maps also show earthquake risk throughout the rest of Italy. RMS worked with researchers from INGV to develop our view of risk in 2007, based on the latest available databases at that time, including active faults and earthquake catalogs. The resulting hazard model produces a countrywide view of seismic hazard that has not been outdated by newer studies, such as the 2009 INGV Seismic Hazard Map and the 2013 European Seismic Hazard Map published by the SHARE consortium, as shown below:
The Route to Increased Resiliency
Increasing earthquake resiliency in Italy should also involve further development of the private insurance market. The seismic risk in Italy is relatively high for western Europe, whilst the insurance penetration is low, even outside the central Apennines. For example, in 2012, there were two large earthquakes in the Emilia-Romagna region of the Po valley, where there are higher concentrations of industrial and commercial risks. Although the type of faults and risks vary by region, such as the potential impact of liquefaction, the RMS model captures such variations in risk and can be used for the development of risk-based pricing and products for the expansion of the insurance market throughout the country.
Whilst Italy’s seismic events in October caused casualties on a lesser scale than might have been, the extent of the damage highlights once again the prevalence of earthquake risk. It is only a matter of time before the next disaster strikes, either in the Central Apennines or elsewhere. When that happens, the same questions will be asked about how Italy could be made more resilient. But if, by then, the country’s building stock is being made less susceptible and the private insurance market is growing markedly, then Italy will be able to say, with justification, it is becoming stronger than any earthquake.…
As a Senior Manager within Model Product Management, Ben focuses on Asia and Europe earthquake models. He joined RMS in 2011, and was previously an Account Manager and a Model Solution Specialist, providing client support for the complete suite of RMS models. He holds a Masters (M.A.) from Cambridge University, and a Doctorate from Oxford University. Ben formerly worked as a Lecturer in Earth Sciences at ETH Zürich.