Making property/casualty underwriting investments that pay off

Underwriting is at the core of the insurance industry. The processes of selecting and pricing risk and the additional operational processes necessary to deliver a policy and provide ongoing services are essential to the overall profitability of a carrier. Over the last few years, carriers have been heavily engaged in replacing core policy admin systems and increasing the automation of their underwriting processes.

Automation of underwriting processes carries the promise of improved results, but can come at a significant cost — both the hard costs (purchasing technology, implementing technology, and changing processes) and the soft costs. Change can be hard on both underwriting staff inside a carrier and on the agents who receive the output of the underwriting process.

So when does it make sense to invest in automation — or, put another way, are there pieces of the underwriting process that when automated are more likely to result in improved results? We thought it would be interesting to investigate these questions to provide guidance to carriers that are trying to prioritize their efforts.

Our goal was to understand the actual state of underwriting automation in the insurance industry. Are carriers living up to the hype in the media that implies that virtually every carrier out there has automated every step of the process? Or is the progress slower? Are carriers with older systems at a disadvantage against those who have replaced their systems with modern solutions? Do high levels of automation actually result in better financial results?

The process of underwriting was broken into 26 logical components of work. For each component, three levels were defined — ranging from little automation used to significant levels of automation. Carriers can use this report as a self-diagnostic tool by comparing their scores to the benchmarks that follow in this report. To understand what top carriers are doing in this area, Celent conducted a survey around this topic looking to answer these key research questions.

  1. What are the different components of underwriting that can be automated?
  2. Where are carriers utilizing automation in underwriting?
  3. Are high levels of automation in underwriting correlated with improved metrics?

Our key findings were:

  • Average levels of automation vary dramatically by line of business, even within the same company.
  • Personal lines carriers are more likely to be applying high level of automation in the front end processes related to automated quote, issue, and renewals — including automated communications with policyholders.
  • Commercial lines carriers tend to apply higher levels of automation for the back end including workflow, product management, rating, and reporting/analytics.
  • Workers compensation and specialty carriers tend to have slightly lower levels of automation in all aspects of underwriting but can achieve significantly better results when applying automation to processes related to analytics and service.
  • Carriers with newer systems are using high levels of automation in more of the processes. Those who have had their systems for over 15 years have had a lot of time to customize their solutions and have slightly more highly automated processes than those whose systems are between 10 and 15 years old.
  • Personal lines carriers are the most likely to benefit from high levels of automation, especially automation related to process efficiency and underwriting insights.
  • Commercial and specialty carriers benefited most from high levels of automation in processes related to underwriting insights. Generally, the best combined ratios were found in those carriers with a medium level of automation — processes that were supported by technology, but had some level of human intervention as well.
  • Workers comp carriers are most likely to benefit from high levels of automation in processes related to driving underwriting insights.

Here’s a link to the report.  You can download it if you’re a customer. If you’re not a client, ping me and we can chat.

Your Natural Best Friend will certainly know that you are sad. But will your customer service chat bot know?

AI and machine learning things are moving right along. A few months ago, in a Celent report, I predicted the emergence of a “Natural Best Friend,” a term combining “natural language” and “best friends forever.” However, there is nothing organic about the Natural Best Friend; it is completely a product of technology. The Natural Best Friend will at some point pass the Turing Test (interacting with a person in a way that is indistinguishable from how another person would interact). Natural Best Friends will become sources of not only trusted information and advice, but also of companionship, friendship, and perhaps even some form of wisdom and intimacy. The use of the Natural Best Friend has obvious applications in throughout the entire insurance life cycle: from underwriting to service to claims. Even the possible characteristics of companionship, friendship, wisdom, and intimacy may be of use to insurers. Consider insurers’ brands, built over decades, which stand for trust, reliability, and succor. Once it becomes socially normal to have a personal relationship with the Natural Best Friend, insurers’ (and many other service industries’) sales and service processes will change dramatically. IBM has just announced it is developing customer service software that can interpret the customer’s emotional state by the content and pattern of the customer’s chat messages. Somewhere in the future, the software may be able to analyze a customer’s voice to determine the emotional playing field. Here’s a link to the WSJ story (warning: this might be behind a paywall). The family tree that will produce a baby boom of Natural Best Friends now has a new branch.

A new and innovative way to issue life insurance? Is that possible?

Hartford Life just introduced Issue First, a new way to provide immediate life insurance protection. Get this . . . the policy is issued before underwriting. The upper limit on face value and age is $2 million and 66, respectively, so they are not just targeting small policies or young insurers. The Hartford estimates that on average it takes 48 days to issue a permanent policy; that’s almost two months! But with Issue First, the policy is issued in as little as five days if the answers to eight medical questions meet the eligibility requirements. The Hartford’s agents must be loving this. Five days. That’s a huge reduction in time for the prospective policyholder. For The Hartford it means fewer withdrawn applications. And, from what they have found in a historical review on non-Issue First cases that 95% of the time, a final Issue First rate would have been the same or better than originally illustrated. For the policyholder it means immediate life insurance coverage without a higher price tag. At the completion of underwriting, the policyholder can accept the final rate, which may be the same, higher or lower than the initial illustration, or exercise a free-look and receive a full return of premiums. So why is this so new and innovative? Well, first I don’t know that anything like First Issue has been done before. It’s changing the way that insurance has been issued for decades. Second, the Hartford analyzed the wealth of data they had to determine that the risk of this process was worth undertaking to speed up the policy issuance process and to grow their business. Lastly, they are changing a process that they see is unfriendly to prospective policyholders even if it means that they are disrupting the way they have always done business. Disruption. Innovation. Words that are not normally attached to insurance. Celent recently hosted a Creative Disruption Workshop in Boston where this topic was discussed. See the video: http://vimeo.com/31409934 Although innovation and disruption are not typically associated with insurance, there were several examples presented where both have happened. For example, Progressive changed the way car insurance is priced which in turn has had a lasting effect on the industry. Telematics and ‘pay as you drive’ is changing how car insurance is underwritten and priced. Forward looking ideas like replacing a call center with a Watson like system were suggested as potential future disruptions. Hartford Life’s Issue First can be considered another such example. Can you think of other examples? I’d like to hear of them. And if your example has an IT project associated with it and you think it is worthy of an award, why not nominate your insurer and the project for Celent’s 2012 Model Insurer Awards. Nominations are being accepted now at https://oliverwymangroup.wufoo.com/forms/celent-model-insurer-2012-selfnomination-form/.

Key European GI Policy Admin Report published

The European Insurance team has been working hard over the Spring and Summer to produce one of our key reports Policy Administration Systems for General Insurers in Europe 2011. It’s a topic of great interest to insurers wanting to replace their core underwriting system, and vendors wanting to have a view of the competitive landscape. This report uses Celent’s ABCD vendor view, which is a standard representation of a vendor marketplace designed to show at a glance the relative positions of each vendor in four categories: Advanced technology, Breadth of functionality, Customer base, and Depth of client services. The report also has the first four PAS systems XCelent Awards for Technology, Functionality, Customer Base, and Service. Since the first report in 2005, activity level has remained high among both insurers and policy administration system vendors. In the two years from January 2009 to January 2011, over 130 insurers had licensed a new policy administration system. Since 2007, the UK market has seen seven new entrants primarily from the United States. This adds to an already crowded space. And of these vendors, most (50%) are small with less than 10 clients and under $10 million in annual revenue. So the vendor market remains fragmented and challenging for the insurer buyer to navigate. Recent acquisition announcements of Accenture/Duck Creek and Sapiens and IDIT are not surprising. We can expect further consolidation in a tough market. Look out for the upcoming European PAS deal trends report which will explore this trend in more detail.

Can mid-market expansion happen with the newest generation of new business and underwriting products?

Will automated underwriting systems finally replace the underwriter for term life insurance products? Could these systems be the catalyst that finally allows life insurers to penetrate the ever elusive middle market? I think both answers are “maybe.”

Today’s new business and underwriting systems offer broad functionality for the producer and the underwriter. Reflexive electronic applications, instantaneous third party data integration, and risk related configurable underwriting rules all can help the producer to potentially issue a larger number of policies instantly.

The newest generation of new business and underwriting systems offer electronic applications that give early insight into how much medical underwriting is needed on a case. Using insurer defined rules that assess risk and reflexive questioning, the system determines what, if any, medical underwriting requirements are needed. Because the system evaluates the risk, the application may never be seen by an underwriter. That’s because for many term products up to a certain face value, the insurer’s rules may rely solely on instantly accessed third party databases. According to Celent’s research, this currently happens less than 25% of the time, but the capability is there to expand the use of these systems to meet the needs of the middle market and at the same time lower the cost of issuing lower face value policies.

The recession has meant there is a larger percentage of the population on the lower end of the income spectrum than ever before. This has important implications for life insurers since more people may look to buy lower face value policies. These applicants demand lower premiums and faster issue times. To streamline the application to issue process and to meet these demands insurers can look to today’s new business and underwriting systems with their straight through processing (STP) functionality and their potential advantages in reducing the cost and time associated with insurance sales and underwriting for all policy types.

The newest generation of new business and underwriting system can also give an insurer deeper insight into its own underwriting rules. The data analytic capability of the system monitors system behavior so insurers can make immediate risk related rules adjustments. If the access to data helps push the level of reliance on the underwriting engine by increasing instant issue policies, the new business and underwriting system could pay for itself through operational savings. Although overly simplistic, using data from Celent’s discussions with life insurers that actively use a new business and underwriting system, a 10% increase in simplified issue policies (assumes the switch of approximately 525 policies from medical underwriting to instant issue with savings of $.40/$1000 per new business face amount) could result in savings of approximately $100,000 per year. That’s a potential break even in four to five years for some vendor systems available today.

According to a Society of Actuaries 2010 study, automated underwriting has been fairly successful for life insurers who use automated underwriting for simplified issue and non-medical underwriting, and to a slightly lesser degree for flagging pieces of information in the underwriting process for review by an underwriter. If an insurer can get past the cultural issues related to using automated underwriting systems, and as long as it does not try to replicate all medical underwriting with an automated system, true benefits will be seen. So, underwriters will still be needed for term products, but to a lesser degree, and the growing middle market has the potential to be penetrated further than ever before.

For further reading on the new business and automated underwriting systems available in the North American market today, read Celent’s North American Life Insurance New Business and Underwriting Systems, 2010: Life, Health, and Annuities ABCD Vendor View.

Social Networking, Meet Underwriting

Our esteemed social networking guru, Craig Beattie, circulated a blog posting that he found at http://itmanagement.earthweb.com/features/article.php/12297_3905931_1/Pre-crime-Comes-to-the-HR-Dept.htm

It describes an internet company, Social Intelligence that monitors social networks to help companies with hiring decisions. Their data mining tool collects information from the major sites looking for behavior-based information about job applicants and summarizes what is found in a report. It uses only publically shared data and includes a review by humans to eliminate any “false positives”. There is also a service for continual monitoring of existing employees. According to the blog posting, the company makes the point that with the emergence of social networks, shareholders will expect companies to use such services to evaluate new and existing hires and reduce the liability of the company from lawsuits, damage to reputation, etc.

Celent has not reviewed this company or its solution. However, in discussing what this approach might mean from an insurance perspective, several questions arose. Will such monitoring be considered a mainstream risk management technique one day? Would an insured using such a tool be rated a lower risk than one that does not? Should the shareholders of an insurance company reasonably expect the underwriting process to include the monitoring of social networking sites, especially for the general liability, disability and workers compensation lines of business? In the past, such data mining has been blocked by regulators based on privacy issues, but if all this information is willingly made public will those objections still be valid? Social networking, meet underwriting.

Underwriting Using Social Networking Tools

My colleagues, Catherine Stagg-Macey and Craig Beattie, released a research report today titled Leveraging Social Networks: An In-Depth View for Insurers. They predict that “insurers will increasingly use public shared data to inform pricing decisions and aid in fraud detection.” They cite several situations where data held on social networks has already been used in court cases in the US. So, how could social networking be used in underwriting and where might we see it emerge first?

Consider a low frequency, high severity line of business such as high limit Personal Umbrella. Improving selection on $5 million liability policies can have a significant impact on results.

How would these connections be established? Will the application process include a link to a person’s social networking page? Will insurers offer incentives such as rate decreases or coverage extensions to incentivize potential insureds to link their personal data to insurers? (I am sure that state insurance commissioners will want to weigh in on the legality/acceptability of this!)

Even if an insurer cannot gain access directly to someone’s page, the publically available information might provide useful underwriting information. For example, if someone checks the “no” box next to the “Do you skydive?” question on the application, but they are a “fan” of a skydiving equipment company, this will likely cause an underwriter to ask a follow up question. Or, alternatively, this may result in an automatic decline by a rules engine applying eligibility rules.

Finally, even if there is no direct information available via social networking pages, it will be straightforward to construct relationship networks for an applicant and at least identify if they are linked to anyone for whom an insurer has in depth information about. To continue with the skydiving example, if several skydivers are linked to a prospective insured, it should create underwriting concern. Additionally, expect to see information vendors provide products which scan social networks for data which can be used to inform the underwriting process. Expect to see social networking reports alongside CLUE and MVR data.

Up to now, a good deal of discussion around social networking has been about how to use these tools for marketing and branding. Thanks for Catherine and Craig for extending the discussion into these other important areas of the insurance process.