Predictive analytics

How Milliman predictive analytics helps clients uncover value in their data

Video transcript

Nancy Watkins (Milliman principal and consulting actuary): Predictive analytics is simply the use of lots of data from multiple sources to make predictions. And so our clients want to know who their best customers are, where they are, how likely they are to buy their products and how much profit they can expect to make off of them.

Matt Chamberlain (Milliman actuary): There’s a progression towards more accurate pricing that’s ultimately based on our understanding of what drives the risk. To the extent that there’s science that can be used to understand what the drivers of risk are, we should use that. 

Nancy Watkins: In order to get the most value out of all of the new data sources that are available, we have to evolve from just thinking about the math and the business to also thinking about the new possibilities of the data sources, how they integrate with the business dynamics that we know about. 

Peggy Brinkmann (Milliman actuary): What’s happened since I started my career is the availability of much atomic, granular data that is very well suited for predictive modeling and multivariate modeling. So actuaries have had to adapt their skillsets and toolsets to be able to manipulate and use that kind of data rather than data that’s been already presummarized for them. 

Nancy Watkins: The quality of analysis is not only dependent on having the tools to operate the analysis. It’s also heavily dependent on the quality of the framing of the problem, and the expertise of the person who is interpreting the results. 

Sheri Scott (Milliman principal and consulting actuary): And now that the data is very complex, you need to know how to analyze the data and process it and explain it in terms of business needs. It’s more important to understand what the most appropriate technique to use is than it necessarily is to just know how to manipulate big data. 

Nancy Watkins: What I really like is synthesis. I like bringing things together that each on their own might be useful but have limited use. And for me the property insurance area is the most promising with respect to that type of synthesis. We get catastrophe model output that used to be used only for reinsurance placement and for PML management. But we can use it for all kinds of other things. So I like to be the person in the middle that takes the information that each provider knows a little bit about and put it all together and make something out of it that has never been made before. 

Sheri Scott: I think that companies have to use predictive analytics to gain an advantage in the marketplace because if they don’t, they’re going to be subject to regulation that’s going to put them in this box where they just can’t compete or they can’t excel. And if you understand the predictive analytics, and the power of it, you can actually get outside that box and start being very creative and very profitable. 

Matt Chamberlain: It’s very important for insurers to demonstrate to rating agencies that they’re managing their catastrophe risk. This is something that the rating agencies have emphasized repeatedly. So I think it’s very important that insurance companies have good quality data, that they understand how the CAT models work, and that they demonstrate that they’re using them to improve their pricing and improve their underwriting. 

Peggy Brinkmann: Being able to demonstrate that I’ve got a good business plan, which involves understanding my catastrophe risk, having the right reinsurance plan in place, having good underwriting and pricing in place, so that I’m not adversely selected against. 

Nancy Watkins: There’s never a bad time to know what’s going on with your business. If you’re losing money, this knowledge can help you lose less money. And if you’re making money, this knowledge can help you make more money.