After a recent chat with an underwriter from AON Minet, I understand that there may be another way of tackling the issue.

We had been thinking about a scheme to insure each rural resident in the Amboseli ecosystem so that s/he would covered against personal injury from wildlife rather like one may have medical coverage or insurance for a motor car.

The other way, if I understand his remarks on 'Public Liability' correctly, would be for a putative 'owner' of the wildlife (KWS, or the Amboseli Trust for Elephants, or the Oloitokitok District Commissioner's office...) to take out an umbrella policy to insure third parties in the ecosystem for damage from 'the policy holder's wildlife'. I have no views or experience to comment on which of these options would be best in terms of management, verification; the underwriter is currently examining the options.

_________
HC

This is a really interesting idea -- KWWG [Kenya Wildlife Working Group] did a big effort with the Insurance industry a year or so ago, I'll look up the stuff -- could be of interest.
Mike Norton-Griffiths

An insurance underwriter asked us a number of questions to try to get at the likely magnitude of claims for human-wildlife incidents. On 18-May-2006, we sent him the following:

Size of insurable domain: humans

From the 1999 Kenya census, there were ca. 120,000 people in Kajiado South. Is it possible to think of some kind of group coverage for a defined population for specific risks from injury or death from wildlife incidents? See table below for estimates of numbers of incidents.

Size of insurable domain: livestock

From a number of sources, there are probably some 150,000 cattle and perhaps 200,000 shoats (= sheep + goats; in aerial counts we don't distinguish; in any event, they have about the same monetary value).

Number of insurable incidents

For elephants, from data gathered from KWS and Amboseli-Tsavo Game Scouts Association over the past decade (1993-2005, elephant data; 1994-2004, predator data), we can derive the table attached to this comment.

Now these numbers are quite rough to be sure, given the lack of a sampling frame and the heterogeneity of the sources. But they certainly give an ideal of order of magnitude, probably erring on the low side, but not by much. Several points:

1. Data sources. The data come from Occurrence Book reports from incidental observations, particularly by the 80 Maasai scouts that patrol the ecosystem. There are three main reporting stations from which the data were transcribed (by AERP): Mbirikani, Oloitokitok and the Public Campsite in Amboseli. Going by the place names recorded, I'd hazard a guess to say that the coverage of the reporting probably covers 80-90% of the ca. 7,000km2 range of the Amboseli elephant population (the Kenya part of their range). On the other hand, it covers the areas where elephants at least are most likely to occupy, so, from the elephant point of view, the coverage is quite representative.
2. The numbers are not very large. There are at least three obvious reasons, in no particular order: (a) It could be that in fact that getting assaulted by wildlife, elephants in particular, is a relatively rare event, compared, for example, to becoming involved in a motor car accident. In that regard, an elephant attack is rather like airplane crash: relatively infrequent and improbable, but spectacular and potentially costly when it occurs. (b) Undercounting or non-reporting, which biases the sampling downwards because of missing or unreported events, in my view is rather unlikely. Even if the scouts do not achieve complete coverage in time and space (despite being obviously dedicated and keen), there is an incentive for victims to make certain somebody knows about the attack. If anything, one might expect over-reporting on the part of victims. (c) Falsified records. As mentioned in (b), both the Scouts and the victims should be motivated to report and perhaps even exaggerate. For KWS officials, however, who are the custodians of the wildlife and in part responsible for reducing human-wildlife conflict, there might be motivation to create an impression for headquarters of 'no worries here' by underreporting incidents. I hasten to say that we see no evidence whatsoever of this mindset in the current and recent senor staff in place.
3. Local effects. That said, we need to be aware that predator attacks can increase in frequency locally. The Mbirikani Predator Compensation Scheme reports that in 2004, there were a total of 746 predator attacks of stock, some two per day. Now, Mbirikani is some 1,200 km2, about 15% of the ecosystem. If you were to extrapolate that willy-nilly to the ecosystem, you get of the order of 5,000 attacks a year. That's almost certainly a gross exaggeration, since 13 attacks a day would be headline news, every day. You might want to take a figure of a couple of thousand as a worse-case scenario for predators (not for elephants!).
4. Change over time>. Although the data aren't good enough for a time series analysis, there seems no indication that there has been much change in volume or frequency of events over time. If anything, there is an indication that there was a loading of events in 1999, which was a particularly dry year when we expect conflicts to increase.

Size of likely property claims.

Concerning the property insurance portion (and not, of course, human injury or death), the amounts to be insured would be relatively small, compared, say, to a home theatre system or motor car. The AERP Consolation Scheme is currently leveraging donor money to pay for each verified case of a fatal elephant attack on stock outside of the park KES 15,000 for a cow and KEW 5,000 for a shoat, which we know is well above market rates. We assume that an insurance scheme would reimburse at current market rates.

Potential types of coverage

In 2006, we suggested to the broker that we should separate potential coverage from two kinds of hazard: one is attack by elephants, the other, attack by lions and other predators. Also, each of these would have two policies (or whatever you call them): one for personal injury or death, the other for damage to property, specifically, cattle, sheep or goats. We suggested not to get into insurance for non-living property or crop damage at this stage (although we may discuss later on): the area is supposed to be pastoral, with a rainfall inappropriate for rain-fed agriculture.

Claims verification.

Both the current lion and elephant schemes, and the new ALCF predator proposal on the table have mechanisms for verification upon which an insurer could build. For the elephant compensation/consolation scheme it requires on-site examination by a group of partners: the researchers involved with the particular species, the Game Scouts, KWS and the Group Ranch. Presumably the insurer would want in place its own objective investigation, equivalent to a police report on a motor car accident. There must be provisions for such in general accident policies.

_________
HC