- UID
- 150385
- 热情
- 109
- 人气
- 313
- 主题
- 5
- 帖子
- 312
- 精华
- 0
- 积分
- 378
- 分享
- 0
- 记录
- 0
- 相册
- 0
- 好友
- 2
- 日志
- 0
- 在线时间
- 856 小时
- 注册时间
- 2008-8-2
- 阅读权限
- 20
- 最后登录
- 2015-9-15
 
升级    59.33% - UID
- 150385
- 热情
- 109
- 人气
- 313
- 主题
- 5
- 帖子
- 312
- 精华
- 0
- 积分
- 378
- 阅读权限
- 20
- 注册时间
- 2008-8-2
|
为了表示支持我发一个以前做过的speech吧：）
How does pricing work in general insurance?
Good evening everyone. Can we have a show of hands how many people have got general insurance (e.g. car insurance, home and contents insurance)? How many of you have seen large increases in premiums recently? (How much?) How many of you understand why? The fact is general insurance pricing is a bit of a mystery to the public. Tonight I hope to clarify that mystery by talking you through how it works.
First, because all great analysis starts from data, I’m going to start by explaining the data being used for pricing. There are two categories – internal data and external data. Internal data consists of claims data and exposure data. As you can imagine, whenever you make a claim, a lot of details are recorded into the insurance system. Some examples include when does the incident happen, when was the claim reported, when was it paid, and how much was paid. Exposure refers to information such as when does the cover start, when does it end, what are the rating factors associated with them. By rating factor I mean anything that can be used to determine your premium, e.g. age, gender, location. Apart from these, we also need overhead expense data and reinsurance cost data. This is the internal data.
External data is then obtained to supplement the internal data. Nowadays there are a lot of data you can buy from third party companies or government organizations. Statistics New Zealand has census data. The police have crimes data. Some private companies have database that contains all addresses across New Zealand. By linking all these information together, it is possible to greatly enhance our understanding of the riskiness of each customer.
Given this data, we then need some kind of statistical models to allow us to predict the likelihood of each type of claim, and the average size of the claim. The industry standard is to use a model called generalized linear model, or GLM. Through its magic power we are able to work out, for each customer, all the components of the technical cost. These components are claim costs, overhead expenses, reinsurance cost, commission and profit loading.
You might think that after all this great work, this is the premium that you get charged. That is wrong. The company has to take into account what the competitors are charging as well. There is no use charging a $1000 premium when everyone else in the market is charging $300. To allow for this, competitor’s information is gathered. In Australia, there are some companies that use automated robots to go into various insurers’ website and obtain thousands of quotes from there on a regular basis, and based on these quotes it is then possible to reverse engineer the pricing algorithms used by these insurers. Such information is extremely useful for decision making.
These steps would be enough for many companies, but in sophisticated markets like the UK or Australia, demand modeling and optimization have been gaining momentum. A demand model predicts how likely an existing customer is going to renew the policy, given the rating factor information, the amount of premium changes at renewal, and the how competitive the new premium is in the market. It also models how likely a given quote is going to convert to an actual new business. Once an insurer has all these models, it can then plug them into some kind of optimization software. What it does is that it answers questions like, “what kind of price should we charge so that we can meet the company’s growth and profit objectives over, say, the next 5 years”. Without demand modeling and optimization, decision makers are actually optimizing using their own judgment, and they typically say, “I don’t want to see any premium increases by more than 10% because I think anything more than 10% means our customers will leave”. These models just put a bit more science into the process.
After a lot of discussions, back and forth, with the business managers, a decision is then made on what premium rates should be charged. Theses new rates are then implemented into the systems, and get rolled out over 12 months as each policy renews.
Of course, any emerging claims experience and sales volumes must be monitored closely. Any warning signs, like a dramatic drop in new business number, must be investigated and understood. This can potentially lead to quick actions to revise the premium rates again if required.
Finally I need to point out that so far I’ve only talked about the premium that goes to the insurer. There are also Earthquake Commission Levies, Fire Service Levies and GST, which go straight to various government agencies. The Earthquake Commission Levies have just gone up from $50 for a typical house to $150 as the result of the Canterbury earthquakes.
So here you have it, a 6 minute explanation of how pricing works in general insurance. Any questions are welcome. I hope the next time you receive your renewal notice from your insurer, you should look a lot less confused than before! |
|
发表于 2012-2-27 23:07:51
这个贴要仔细看啊！！
所以很佩服LZ MM
