In his book Antifragile Nassim Nicholas Taleb breaks all things into three categories: \u00a0the fragile, the robust and the anti-fragile.<\/p>\n
Take for example Malaria. \u00a0Malaria has killed and weakened humans on a massive scale for millennia. Malaria is caused by the\u00a0Plasmodium parasite which is transmitted when the female Anopheles mosquito bites a human being. \u00a0In West Africa, where the mosquito is most prevalent the population has evolved a significant genetic resistance to Malaria in the form of a single copy of the gene that causes Sickle Cell anemia. People with one copy are much less likely to get Malaria and if they get it, suffer much less damage from it than those with no copies. \u00a0But there is a cost to this Malaria defense: \u00a0children who get two copies of the Sickle Cell gene get Sickle Cell anemia and die. \u00a0Evolution made this trade-off over hundreds or thousands of generations of West Africans because the extra children produced by those with one gene far exceeded the number of children lost. \u00a0So in this example, people lacking a single Sickle Cell gene are fragile with respect to Malaria, they can’t adapt or react. \u00a0People with one sickle Cell gene are Robust with respect to Malaria . \u00a0The entire population of West Africa, however is Anti-fragile: \u00a0it has been able able over a long period of time to “recognize” and “adapt” to the threat.<\/p>\n
Now compare Malaria and the Anopheles mosquito to Demodux Follicurum or the ‘Fat Bottomed Skin Mite’ that lives in our hair follicles. The first time I saw one in a microscope I couldn’t stop scratching. \u00a0Yet they’re there and so far as we can tell, there is no biological adaptation we have to keep them off. \u00a0Why? Well probably because they don’t seem to be doing much, if any harm, indeed they seem to be eating our dead skin. \u00a0But evolution doesn’t “think” so how did it “discover” that we could “tolerate” the skin mites on our skin but not the Plasmodium in our blood? \u00a0It tried different “interventions” for example (and I’m just making this up) evolution could have tried to produce a human with fewer or no follicles and since skin mites live in our pores and shelter in them during showers and so on, having a lack of hair follicles got rid of most or all of the mites. \u00a0Unfortunately the cost of the ‘intervention’ far exceeded the benefits that the poor, follicle-less human incurred so the tested intervention\/mutation did not “succeed” and died out.<\/p>\n
Notice what happened in these two examples (and I’m getting to the point for insurance, I promise): in both cases there was an ugly looking parasite with six legs and a frightening looking mouth or needle like proboscis yet our blind watchmaker evolution “sorted” between the two and “decided” that the Plasmodium was a terrible disease that needed to be managed even at the cost of dead Sickle Cell victims while it decided not to mess with the Mites. \u00a0It differentiated between the two by the difference in their cost to human reproduction. High cost? High cost response is justified. \u00a0Low or no cost? \u00a0Ignore it.<\/p>\n
The same sort of calculus goes into human technological innovation. \u00a0Malaria? \u00a0By applying a series of techniques and technologies we’ve been able to reduce the number of Anopheles \u00a0mosquitoes where I live in Houston enough to make it nearly impossible for the Plasmodium parasite to be transmitted from one human to another. And we judge that the occasional ‘unusual’ case is the acceptable price we pay for the level of investment we are prepared to make. And I suppose we could get rid of all skin mites by regularly bathing in pesticides but we’ve judged the benefit to not exceed the very real costs. \u00a0But now comes the Zika virus and our Malaria calculus goes out the window: \u00a0different mosquito, different parasite. \u00a0We may find an human adaptation for this new vector but it may turn out to not be very effective or too costly relative to the alternative: \u00a0which is to avoid most of the risk by in this case keeping pregnant women inside and away from mosquito bites. But rest assured evolutionary adaptation is working on the problem.<\/p>\n
You’ll notice that all of these adaptations have three things in common: \u00a01. They identify the vector – a mosquito with Plasmodium, a skin mite, \u00a02. They calculate the cost of the vector and 3. They throw a series of possible solutions at the vector (either randomly through evolution, or through technological trial and error) until they find an intervention or bundle of interventions (drain swamps, spray streets, screen doors) that both “works” where the costs of intervening are exceeded by the benefits of the intervention.<\/p>\n
So what does this have to do with Insurance? \u00a0Simple, Insurers, particularly auto insurers are absolutely riddled with fraud and data defect parasites. \u00a0It is estimated that fifteen to twenty percent of auto insurance premiums are lost to fraud – either up front deceptions to get better rates or back end claims fraud to get unjustified payments. The fraud burden in auto insurance is similar to the prehistoric Malaria burden. But auto insurers aren’t fragile, they’ve coped with this burden \u00a0by the simple expedient of building the total loss experience – including the embedded fraud – into their rating systems. \u00a0This means that a 15 to 20% fraud “wall” is built into every policy. Therefore auto insurers can be said to have a “robust” defense against fraud. Which is great so long as all insurance companies have a similar fraud parasite burden. One of the things that anthropologists and linguists tell us about the settlement of Africa is that the language and genetics indicate that the original population of West Africa spread over time to all of sub Saharan Africa engulfing other groups in its way. \u00a0One of the hypothesized reasons for this massive population shift is that West Africans didn’t have nearly as much malaria so they had more kids and thus expanding populations looking for new land, their warriors were larger, healthier and more numerous and after they had killed off competing warriors were able to marry local women and transmit the Sickle cell gene to their offspring. \u00a0Thus the adaptation that gave them robustness with respect to Malaria gave them advantage over their less adaptive neighbors, leading to the complete eradication of their languages and cultures.<\/p>\n
The other tribes were fragile, they couldn’t adapt to the new reality of Malaria resistant West Africans fast enough to survive. \u00a0That’s the risk auto insurers in particular are taking now. But you say “yes, we understand but we’re working on technologies to prevent this fraud so we’ve got that covered! ” But are do you really? \u00a0There are a lot of fraud ‘solutions’ out there that purport to address “the” fraud problem. \u00a0But by and large they’re solutions that have identified the problem at one point in time and built tools to identify that single problem. For example, there is a famous chart produced by Verisk that purports to tell us where underwriting fraud comes from. \u00a0It’s very interesting and I’m sure has some validity but it’s not really telling any insurance company much of anything: \u00a0first of all, it’s composite of different insurance companies with different rating methods and environments, second of all it’s got very broad categories and doesn’t have anything to say about back end claims fraud and thirdly it’s a point in time estimate that’s getting on 4 \u00a0years old. Yet vendors are pitching ‘solutions’ that ‘solve’ this static fraud problem. \u00a0Most of them work by taking data and dumping it at the Underwriting “factory door” saying: \u00a0“match this to that and you’ve solved your fraud problem”.<\/p>\n
Were it so simple. \u00a0As we learned from our friend the Mosquito adaptive systems need three elements to identify and eliminate a parasite whether he be Plasmodium or John Plasmodi. First you need to find the vectors (or what at VeracityID we call ‘defects’): \u00a0not just the nine categories but the specific defects: \u00a0the Illinois defects and the on line channel defects, the Senior Citizen def
\nects. Second you need to understand just how destructive each geographic, segment, channel vector is: individually and by category \u00a0by matching the claims loss to each individual case. \u00a0Thirdly you need to have a process to identify and attack each vector that tries to enter your process. \u00a0Fourthly you need to be able to measure the impact of each intervention in terms of loss mitigation so you can evaluate, revise and try again.<\/p>\n
And you need to do this over and over because we’re dealing with human innovation, not long term evolution. \u00a0Get good at stopping one type of defect and humans adapt and pursue a new kind. \u00a0Relying on four year old aggregated, undifferentiated estimates of loss doesn’t get one very far. But if you can create this kind of adaptive system and apply it rigorously year after year, before long you’ll be in the postion that the West African Bantu found themselves in: \u00a0you’ll be bigger, stronger, more profitable and able to overwhelm the sickly parasite ridden competitors who have done nothing or done a half thing which is what “dump the data at the Underwriting factory door” really is.<\/p>\n
<\/p>\n","protected":false},"excerpt":{"rendered":"
In his book Antifragile Nassim Nicholas Taleb breaks all things into three categories: \u00a0the fragile, the robust and the anti-fragile. Take for […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[15],"tags":[],"class_list":["post-1316","post","type-post","status-publish","format-standard","hentry","category-blog"],"yoast_head":"\n