2 Agenda Why this presentation is really interesting Introduction to predictive modelling Case studies Conclusions
3 Why this presentation is really interesting Presentations that appeal to marketing / product development actuaries Presentations that appeal to geeky actuaries This presentation
4 Bingo ANALYTICS PREDICTIVE UNDERWRITING THE CLOUD OPEN SOURCE TECHNOLOGY BIG DATA DATA MINING MACHINE LEARNING GLMs CROWDSOURCING
5 Agenda Why this presentation is really interesting Introduction to predictive modelling Case studies Conclusions
6 What are predictive analytics? Tools and technologies for analysing and understanding business performance Business Intelligence Extensive use of data with statistical and quantitative analysis Analytics A process by which current or historical facts are used to create predictions about future events or behaviour. Predictive Analytics
7 What are predictive analytics? A process, not a product Good data is vital for success A process by which current or historical facts are used to create predictions about future events or behaviour. Typically predictions are created through the use of sophisticated statistical models Focus on predicting probability of future events and behaviour
8 The process Deployment Business Understanding Evaluation Data Understanding Source: adapted from Cross Industry Standard Process for Data Mining (CRISP DM) Modelling Data Preparation
9 Underlying the models Forecasting Operational Research Simulation Optimisation Simulated Annealing Fourier Transforms Wavelets K-Means Clustering Genetic Algorithms Graph Theory Link Analysis Decision Trees Random Forest Support Vector Machines Data Mining BI Neural Networks Querying OLAP Cross-tabs Visualisation SQL Modified from a version presented by John Elder, Harmonic Analysis Linear, Logistic Regression, GLMs Time-series Analysis Bayesian Networks Monte Carlo Principle Components Reliability/Survival Analysis ANOVA MANOVA Correlation Factor Analysis Statistics
10 Why predictive analytics? internal data external data Processing power Data Technology Competition new types of data New tools
11 Agenda Why this presentation is really interesting Introduction to predictive modelling Case studies Conclusions
12 Worldwide projects* UK: Basis Setting (mortality, morbidity and lapses) Postcode pricing model Enhanced experience analysis Predictive underwriting on credit rating agency and bank data Broker Quality * Courtesy of Peter Banthorpe RGA USA: Pricing override model for group LT disability Lapse basis Predictive underwriting on Non-Life data Term Tail Lapses Mortality prediction on credit rating agency data Europe: Predictive underwriting on bancassurance data South Africa: Enhanced Experience Analysis Predictive underwriting on bank and credit card data India: Claims Fraud Prediction Australia: Predictive underwriting / cross sell on bancassurance data Asia: Predictive underwriting on bancassurance data Finer price segmentation Propensity to buy Cross sell of insurance on bank data
13 Case Study 1: Lapse Assumptions variable annuity with GMIBs Current lapse modeled based solely on plan type, duration, In the moneyness ( ITM ) Proposal: Evaluate current model based on 12 quarters of observations Method: Develop an alternative statistical model based on current variables, augmented by additional policy characteristics and macroeconomic variables Compare predictive performance of the two models Performance measured by ratio of Actual/Expected dollars lapsed in out of sample future period.
14 Predictors Positively associated with lapse Duration (adjusted for Surrender Charge Period) Negatively associated with lapse Annuitant age at issue Anniversary of issue date Total rider charge (bps) Policies sold in channel 1, 5, 6 In-the-moneyness (ITM) State unemployment rate Held in qualified plan? US inflation rate Policy size (cumulative deposits) Risk aversion (low % of equity) Policies sold in channels 3, 8 Past partial withdrawals Not significant: Gender, Surrender Charge, DB Rider, GMIB reset, past quarter S&P 500 return, 10 Yr T Note yield
15 Performance of predictors At each quarter, split data by past versus current experience Actual experience (black bars) Statistically predicted (red dots) expected lapse and confidence bands Calculated (blue dots) expected lapse based on current assumption Statistical A/E outperforms current assumption in all quarters
16 Case Study 2: Predictive Underwriting Model Client: Bancassurer in Asia with large customer pool, but low penetration in life product Goal: to predict UW decisions on its existing customers Major challenges very limited data A total of about 8k 9k full UW cases Target variable UW decision, with very low declined/rated cases, ~3.0% Many missing values due to old time, especially for sub STD Not all information collected at the time of UW Source: Peter Banthorpe RGA.
17 Modeling Approach / Key Variables GLM with binomial and logistic link function About a dozen of predictor variables that are statistically significant for prediction & readily available in client database Key predictor variables Positive means the probability of being a standard rate case increases if the value goes up; otherwise, it is Negative Name Type Note Age_At_Entry Numeric Negative; less likely to qualify for STD as age goes up Branch Categorical Proxy of geographic locations Asset Under Management Numeric Positive; more likely to qualify for STD with large AUM Customer_Segment Categorical Positive for Gold, negative for other Nationality Categorical Positive for domestic; negative for certain others Source: Peter Banthorpe RGA.
18 Model Results Lift Plots In sample results show model performance under optimal condition May over fit data 0.5% of sub STD in top 30% Validation results are a better test of model performance in real business 0.6% sub STD in the top 30% Source: Peter Banthorpe RGA.
19 Model Results Gain Curve Another way to understand model capability to differentiate STD from sub STD Best 30% of model outputs contains about 5% of total non STD Lowest 30% captures about 75% of bad risks Source: Peter Banthorpe RGA. non STD % Model Gain Curve In sample results Validation results Random Sorted Model Output
20 Case study 3: non life Predictive analytic techniques used to analyse motor insurance portfolio to: Identify predictors of claims and hence model a profitability score per customer Identify predictors of propensity to renew Allows analysis of portfolio by profitability and propensity to renew
22 Agenda Why this presentation is really interesting Introduction to predictive modelling Case studies Conclusions
23 Key messages Data driven process Broad potential applications for insurance Non Life is way ahead of us Simplified underwriting a key area of focus (but there are many more applications) Not an off the shelf solution Customised, based on specific data and specific needs No two exercises are the same flexibility of approach is key
CEIOPS-FS-08/05 S CEIOPS Preparatory Field Study for Life Insurance Firms Summary Report 1 GENERAL OBSERVATIONS AND CONCLUSIONS 1.1 Introduction CEIOPS has been asked to prepare advice for the European
Introduction to Data Mining and Knowledge Discovery Third Edition by Two Crows Corporation RELATED READINGS Data Mining 99: Technology Report, Two Crows Corporation, 1999 M. Berry and G. Linoff, Data Mining
2011 Annual Report Genworth Financial, Inc. Message from the Chairman In 2011, we continued to experience challenges posed by the uncertain economic environment and negative impacts from a weak U.S. residential
Asset Allocation with Annuities for Retirement Income Management April 2005 Paul D. Kaplan, Ph.D., CFA Vice President, Quantitative Research Morningstar, Inc. 225 West Wacker Drive Chicago, IL 60606 The
Zsuzsanna Tajti: The methodological opportunities of quantifying the retail mortgage loan s LGD in Hungary - 1 - Institute of Finance and Accounting Department of Investments and Corporate Finance Tutor:
RETHINKING THE RISK BUSINESS DATA AND ANALYTICS IN THE INSURANCE INDUSTRY Insurance companies are investing heavily in data to improve risk management and boost returns across their investment portfolios.
Determinants of Life Insurance Consumption across Countries Thorsten Beck and Ian Webb* World Bank and International Insurance Foundation The importance of life insurance companies as part of the financial
UNIVERSIDADE CATÓLICA PORTUGUESA CATÓLICA LISBON SCHOOL OF BUSINESS AND ECONOMICS Master of Science in Business Administration Using Data Mining to Predict Automobile Insurance Fraud JOÃO BERNARDO DO VALE
The Business Management Solution for Small and Midsize Enterprises Solution Overview Contents Introduction 3 SAP Business One: Key Differentiators 4 Business Benefits 5 Functionality Overview 6 Finance
Martian Chronicles: Is MARS better than Neural Networks? by Louise Francis, FCAS, MAAA Abstract: A recently developed data mining technique, Multivariate Adaptive Regression Splines (MARS) has been hailed
Life Insurance the way we see it Analytics: A Powerful Tool for the Life Insurance Industry Using analytics to acquire and retain customers Contents 1 Introduction 3 2 Analytics Support for Customer Acquisition
2015 CIMA PROFESSIONAL QUALIFICATION SYLLABUS 2015 CIMA PROFESSIONAL QUALIFICATION SYLLABUS FOR MORE INFORMATION VISIT WWW.CIMAGLOBAL.COM 2/3 CONTENTS This updated syllabus and assessment bridges the skills
NATURAL HEDGING OF LIFE AND ANNUITY MORTALITY RISKS SAMUEL H. COX AND YIJIA LIN ABSTRACT. The values of life insurance and annuity liabilities move in opposite directions in response to a change in the
Aon Benfield Insurance Risk Study Growth, profitability, and opportunity Ninth edition 2014 Risk. Reinsurance. Human Resources. Table of Contents Introduction...3 Global Premium, Profitability, and Opportunity...5
Return on Investment Analysis for E-business Projects Mark Jeffery, Northwestern University Introduction 1 The Information Paradox 2 Review of Basic Finance 4 The Time Value of Money 4 ROI, Internal Rate
The Role of Real Annuities and Indexed Bonds In An Individual Accounts Retirement Program Jeffrey R. Brown Kennedy School of Government, Harvard University and NBER Olivia S. Mitchell Wharton School, University
How to embrace Big Data A methodology to look at the new technology Contents 2 Big Data in a nutshell 3 Big data in Italy 3 Data volume is not an issue 4 Italian firms embrace Big Data 4 Big Data strategies
Regulation Impact Statement Life and General Insurance Capital Review (OBPR ID: 13560) Introduction This Regulation Impact Statement (RIS) follows the previous related preliminary assessment OBPR ID: 13560
Technical Discussion Paper B for public comment Enabling a better income in retirement 21 September 2012 National Treasury Contents Contents... 2 1. Introduction... 3 Executive Summary...3 Living annuities...4
Financial Risk Management of Guaranteed Minimum Income Benefits Embedded in Variable Annuities by Claymore James Marshall A thesis presented to the University of Waterloo in fulfillment of the thesis requirement
SCOR inform - November 2012 Preferred Risk in Life Insurance Preferred Risk in Life Insurance Authors Matthew HUGHES SCOR Global Life Americas Editor Bérangère Mainguy Tel: +33 (0)1 58 44 70 00 Fax: +33
LIFE INSURANCE SENSE AND NONSENSE for people who don't mind a little complexity by Glenn S. Daily GLENN S. DAILY www.glenndaily.com fee-only insurance consulting Copyright 1992, 1998 by Glenn S. Daily.