 In our risk management decisions we are subject to certain behavioral patterns or biases. What I want to talk about here is insurance as a very important financial risk management tool. Let me ask you, have you bought extended warranties on some of your household goods? Yes, that's not sensible because the expected benefit from these are much too low compared to the price of it. Even with high risk aversion. So here, but still this type of overpriced protection is bought. Then do we care about this? Let's say no, we care about catastrophic losses here. Like for example from natural disasters. Now for these it seems that there is an array of anomalous behavior for these sort of correlated low probability high consequence events. Natural disasters, you see here the trend in the losses which won't surprise you. What might surprise you is in blue you see the insured share of it which is very low. What might be a reason for it? One might be that people actually expect public assistance. So in this case here we would have the Samaritans del Dilema that's inefficient risk sharing but that's individually perfectly rational behavior here. Now we don't know there's only scarce evidence whether people actually do take this expected public assistance into account. But what we know is in that uncommon behavior for example with respect to floods is that people don't buy protection. Then there's an occurrence, suddenly they buy insurance, but then after a short time they cancel the policy again. So what might explain this behavior? One is the availability or salience bias. So people tend to assess the likelihood of an event by recalling recent occurrences of it. So meaning that decision making is too strongly based on the salience of an issue. Another problem is that for financial risk management people do tend to think in what's called an investment frame. Which is of course missing the point for risk management measures. So people who think of insurance as risk management if they don't have a payout the year they will regret having bought it and then cancel the policy. This is a bit on demand here supply, terrorism risk. Before 9-11 terrorism wasn't included or excluded and suddenly after 9-11 coverage was only available at extremely high prices or not at all. Like San Francisco, Golden Gate Park couldn't get terrorism coverage at any price. What might be a reason for this is that even the supply side insurers and capital market investors they might be averse to ambiguity or uncertainty. Let's see these events when the exact probabilities are not known. Now how can for example public policy deal with it? The problem is that policy makers might also be subject to biases like this salience bias in particular when there are elections coming up. But policy tools might be for example to provide information for a better risk assessment. Here how this information is presented is crucial. So factory owners and home owners were asked on earthquake risk and they took it much more seriously when they said like over a 25 year period the chance of at least one earthquake is greater than one in five rather than it was presented saying that in any given year the chance of an earthquake is one in a hundred. It's a very thin line here between just presenting information for accurate risk assessment or framing this information in such a way that you already nudge people towards a particular decision that you do want them to take. And then the question is what's the normative basis for sort of trying to directly control these decisions. What we can do is however provide the correct regulatory framework. One would be mandates. So in Switzerland there's a debate about mandating earthquake insurance Switzerland wide. So what my question is for you here for these events these correlated low probability or ambiguous but really high consequence events where we do observe certain behavioral bias is how should we design risk sharing agreement.