Business Intelligence Southeastern Actuaries Fall 2012 Meeting Baltimore, MD Paul Ramirez, ASA, MAAA What is Business Intelligence? The ability to process raw data in order to make actionable, strategic decisions Data mining Performance management Predictive analytics Effective use of business intelligence can be key to gaining an advantage in a competitive environment Often achieved using decision support systems 2 1
Data Warehouse - Definition Central data repository that can integrate data from multiple sources A data warehouse with online analytical processing technology (OLAP) gives the ability to run multi-dimensional queries quickly OLAP cubes are a data structure optimized for: Slicing Dicing Drill Down Aggregation 3 Benefits of using a data warehouse Provides a single source for data analysis, rather than multiple different sources Ability to join data from multiple sources Generally easier to retrieve data from a data warehouse rather than from the original source systems Historical data is retained Complex data queries can be run quickly 4 2
Reasons not to implement a data warehouse Data warehouse implementation projects are often cited as having very high failure rates Cost of implementation/maintenance Time Cost Hardware Cost Human capital cost Benefits from data warehouse are often intangible System fatigue How will this benefit me? 5 Vended or Homegrown? Common question in most information system choices Vended Systems Ability to leverage prior experience from vendor Should be a faster path to implementation System can be a black box Homegrown System Ability to customize system to unique requirements Very difficult project to complete 6 3
Potential obstacles in implementation Requires commitment from management. Data warehouse projects require time and resources to be successful. Data quality is key! Identifying problems early is ideal How will non-granular data be loaded into the data warehouse? Validation of data is key; output must match existing reporting, or users will not trust data warehouse 7 Best practices in implementation Business units will need to be very involved in order for project to be successful Quality high-level design should be completed early in the project Clearly defined scope will define expectations for data warehouse Document the process! Create data flow diagrams, ETL mapping, etc. 8 4
Data Warehouses in Insurance Companies What types of data sources? Policy Administration System Claims Administration System Valuation Data Projection Data Accounting Systems New Business System Agent Data Metadata 5
Income Analysis Create income statement using many different attributes Group level Agent level State level Calendar year level Issue duration level Industry Calculate income at cell level Helps switch focus to profitability, rather than sales Senior management can relate to profit margins rather than A/E ratios, persistency, etc. Loss Ratio Analysis Display paid loss ratios, incurred loss ratios, expected loss ratios, A/E loss ratios by different attributes Duration Rating class Issue state Issue age Can be used for pricing, rate changes, or rate certifications 6
Experience Studies Ability to run experience studies faster Critical ability with principles-based valuation Can run experience studies at multiple attribute levels Types of studies Lapse Mortality Morbidity Claim Termination Incidence Operational Reporting Can replace other operational reporting because of ease of access and quick runtime Sales reporting Claims reporting Inforce reporting Agent/Group reporting Ideal for ad-hoc reporting 7
Why should I use a data warehouse? Inforce management using a data warehouse can result in LARGE savings Provides analysis that is impossible without the data warehouse Redefines the role of the actuarial department The ability of actuaries to analyze data By increasing operational efficiency, more time is available for analysis Questions???? 8