logo image

This article was originally published in Insurance Day, click here to view the original article.

Livestock insurance represents a significant part of global agriculture premium. Traditional indemnity insurance products are available, complemented with less common products like parametric index insurance. Managing livestock insurance is a complex business, as livestock mortality is a recurring event.

China is one of the biggest players as the world’s largest livestock producer. In China, livestock insurance premium represents about 25 percent of total Chinese agriculture premium, making livestock risk management a major concern for the insurance industry.

Like China, many countries have improved their contingency plans and established controlled diseases centers to reduce mortality event impacts. In China, major recent disease epidemics include Foot and Mouth Disease (FMD), Porcine Reproductive and Respiratory Syndrome (PRRS), Swine Fever, Avian Influenza and Newcastle disease. Noticeable epidemics since 1995 include SARS in 2003 (poultry), major FMD and PRRS in 2006 and 2007 (pigs), and FMD in 2009 and 2010.


A complete probabilistic solution can significantly extend the view of tail risk, for reinsurance purchase, as well as in terms of loss cost estimation for primary insurance underwriting. Beyond providing a view on the probabilities of severe outbreaks beyond the historical records, these solutions can also give insights into “what-if” scenarios on past events.

But building meaningful modeling solutions for livestock is complex for many reasons:

  • First, livestock covers a large range of types of exposure and vulnerabilities, from cattle, swine, sheep and poultry, to aquaculture. And these types of exposure are often subdivided into further details in policies, such breeding sows and hogs.
  • There is a large variety of perils affecting livestock: from disease, epidemics, natural disaster, and accident/fire. But these perils contribute differently to the loss distribution: while accidents/fires and diseases contribute mostly to the high frequency losses due to their recurrent nature, epidemic (highly contagious) and natural disasters (mainly drought and freezing conditions) contribute to the low frequency-high severity losses.
  • Size of operation is also an important factor: for the same livestock type, the loss extent varies according to the size of the farming operation, the location, and, in case of epidemic diseases, the extent of how the disease is addressed. It is expected that the frequency of outbreaks of disease and accidents/fires will likely be smaller for large operations than for small operations because of better sanitary conditions and resources. However, the spread of epidemic diseases is more severe in farms with a larger number of livestock and is assumed to propagate faster in function of livestock density. Due to a higher value for imported livestock, risk management standards tend to be better than for lower value domestic breeds.

Models can provide a unique perspective on livestock risk, by individual perils, based on simulating a yearly deviation from the general trends from several relevant indicators. For example, to model mortality for epidemic diseases, livestock density is considered as an indicator of the potential spread of the diseases in case of an outbreak, and vaccination and contingency plans were used to assess the level of risk management.

Data collection and quality is key: statistical yearbooks provide data on the number of heads of livestock for all the major livestock categories, but these are often incomplete and need to be assessed and compared to livestock accounts from other sources. Insurers need to make informed use of data. Annual changes of livestock counts do not directly translate into losses, as mortality occurs even if livestock numbers increase, but these data can be very useful to estimate the volatility of high frequency events.

It is also necessary to account for mitigation effort, e.g. considering vaccination and quarantine plans as well as procedures to address outbreaks of epidemic livestock diseases. Data on moved livestock within a province and the precise location of slaughter houses and major livestock markets, which are important to assess the risk of epidemic diseases, are not easily available and RMS engages with locally-based researchers on the reliability of livestock data and key factors that drive the spread of epidemic livestock diseases in a number of provinces.

Today’s business decisions are largely based on judgement of experienced individuals backed with past insurance loss data, and the input of new technologies such as remote sensing. Adopting a new, more comprehensive and scientific underwriting approach is essential to the growth and profitability of livestock business in emerging economies.

This article was originally published in Insurance Day, click here to view the original article.

You May Also Like
November 16, 2015
Are (Re)insurers Really Able To Plan For That Rainy Day?

Many (re)insurers may be taken aback by the level of claims arising from floods in the French Riviera on October 3, 2015. The reason? A large proportion of the affected homes and businesses they insure in the area are nowhere near a river or floodplain, so many models failed to identify the possibility of their inundation by rainfall and flash floods. Effective flood modeling must begin with precipitation (rain/snowfall), since river-gauge-based modeling of inland flood risk lacks the ability to cope with extreme peaks of precipitation intensity. Further, a credible flood model must incorporate risk factors as well as the hazard: the nature of the ground, such as its saturation level due to antecedent conditions, and the extent of flood defenses. Failing to provide such critical factor can cause risk to be dramatically miscalculated. A not so sunny Côte d’Azur This was clearly apparent to the RMS event reconnaissance team who visited the affected areas of southern France immediately after the floods. “High-water marks for fluvial flooding from the rivers Brague and Riou de l’Argentiere were at levels over two meters, but flash floodwaters reached heights in excess of one meter in areas well away from the rivers and their floodplains,” reported the team. This caused significant damage to many more ground-floor properties than would have been expected, including structural damage to foundations and scouring caused by fast-floating debris. Damage to vehicles parked in underground carparks was extensive, as many filled with rainwater. Vehicles struck by more than 0.5 meters of water were written off, all as a result of an event that was not modeled by many insurers. The Nice floods show clearly how European flood modeling must be taken to a new level. It is essential that modelers capture the entire temporal precipitation process that leads to floods. Antecedent conditions—primarily the capacity of the soil to absorb water must be considered, since a little additional rainfall may trigger saturation, causing “saturation excess overland flow” (or runoff). This in turn can lead to losses such as those assessed by our event reconnaissance team in Nice. Our modeling team believes that to achieve this new level of understanding, models must be based on continuous hydrological simulations, with a fine time-step discretization; the models must simulate the intensity of rainfall over time and place, at a high level of granularity. We’ve been able to see that models that are not based on continuous precipitation modeling could miss up to 50% of losses that would occur off flood plains, leading to serious underestimation of technical pricing for primary and reinsurance contracts. What’s in a model? When building a flood model, starting from precipitation is fundamental to the reproduction, and therefore the modeling, of realistic spatial correlation patterns between river basins, cities, and other areas of concentrated risks, which are driven by positive relationships between precipitation fields. Such modeling of rainfall may also identify the potential for damage from fluvial events. But credible defenses must also be included in the model. The small, poorly defended river Brague burst its banks due to rainfall, demolishing small structures in the town of Biot. Only a rainfall-based model that considers established defenses can capture this type of damage. Simulated precipitation forms the foundation of RMS inland flood models, which enables representation of both fluvial and pluvial flood risk. Since flood losses are often driven by events outside major river flood plains, such an approach, coupled with an advanced defense model, is the only way to garner a satisfactory view of risk. Visits by our event reconnaissance teams further allow RMS to integrate the latest flood data into models, for example as point validation for hazard and vulnerability. Sluggish growth in European insurance markets presents a challenge for many (re)insurers. Broad underwriting of flood risk presents an opportunity, but demands appropriate modeling solutions. RMS flood products provide just that, by ensuring that the potential for significant loss is well understood, and managed appropriately.…

November 11, 2015
European Windstorm: Such A Peculiarly Uncertain Risk for Solvency II
Laurent Marescot
Laurent Marescot
Senior Director, Model Product Strategy, RMS

Based in Zurich, Laurent initially joined RMS in 2008 as part of the Zurich account management team, servicing the European (re)insurance and ILS market. He then moved to the model product management group, leading the technical product management team for European climatic perils, such as windstorm, severe convective storm and flood. Since 2014, he has joined the model product strategy group for Europe model product line.

Prior to RMS, Laurent worked 3 years at the Swiss Federal Institute of Technology Zurich (ETHZ) as a Research Associate and Lecturer, managing multidisciplinary natural hazard research projects. Laurent still lectures regularly on geophysics and catastrophe modeling at universities, and gives seminars and invited talks in international meetings. He is a Lecturer and Scientific Collaborator at the University of Fribourg (Switzerland). Laurent co-authored numerous industry publications, reviewed scientific articles and proceeding papers. He holds an MSc in Geology from the University of Lausanne and a PhD in Geophysics from the University of Lausanne and the University of Nantes.

cta image

Need Help Managing Your Portfolio?

close button
Video Title

Thank You

You’ll be contacted by an RMS specialist shortly.

RMS.com uses cookies to improve your experience and analyze site usage. Read Cookie Policy or click I understand.