Tag Archives: data analytics

How Analytics Can Tame ‘Social Inflation’

“Social inflation” is considered one of the major emerging risks that the insurance industry must face. While people may misconstrue the term as relating to the rising impact of social media on online behavior of netizens, it has actually to do with increasingly hostile legal environment that insurance carriers are facing today. This manifests in the form of much larger verdicts, liberal treatment of claims by boards, more aggressive plaintiff bars, etc. This article explains the trend and describes measures that carriers can deploy to keep a check on increasing legal expenditure.

Here are some signs of the phenomenon:

  • A major P&C insurer anticipates $40 million in quarterly legal costs for property losses alone
  • There is a 94% increase in assignment of benefit (AOB) lawsuits in the state of Florida in the last five years
  • An increased probability of “nuclear verdicts” (> $10 million) is a real trend. In 2018 alone, the top 100 verdicts ranged from $22 million to $4.6 billion

Here are some factors driving the trend:

  • Litigation Funding: Propelled by easy capital availability, a new class of plaintiff attorney funding model has emerged in the last 10 years. Essentially, this model provides funding for legal expenses to plaintiff attorneys in exchange for a portion of the judgment or settlement. The model is good in the sense that it levels the playing field against large, well-funded corporations, but the unintended consequences (for insurance companies) is an exponential increase in attorney representation and pursuit of aggressive legal strategies. Approximately $9 billion has been committed to this “industry.”
  • Rising anger against big corporations: The very premise of the big corporation versus the individual scenario is driven by anger. The perception among the consumers and the jurors is that a corporation has only one goal: profit. Stagnation of incomes/wages is another contributing factor to this mindset. Big businesses in America like to talk about the good jobs they provide, but median salaries in the U.S. have been flat for decades. This is not because of a failure of workers to become more productive; there were gains in productivity, but they did not go to workers. Gains mostly flowed to the organizations and their shareholders, including executives who received sizable stock-based compensation. Hourly compensation for workers remained practically flat.
  • Large verdicts being driven by general social pessimism and jury sentiments: New and interesting patterns are being observed in jury behavior, especially in personal injury and liability claims. Emotion and trust play a big role in how a jury rules. Some of the key reasons behind these large verdicts are: jurors’ distrust in big corporations and their lawyers; jurors paying less attention to lengthy testimonies and complex explanations; impact of social media on how millennial jurors view the court system; impact of emotional stories on how a jury thinks; and the “what if it were me?” attitude influencing how jurors approach justice. 

Leaving aside the financial burden from social inflation, which is quite significant on its own, increased litigation also affects carriers in other ways. There can be inaccuracies in reserving, which is based on counsel’s estimates of litigation spending. There can be negative press and poor customer engagement/satisfaction. And claims can increase in complexity, delaying settlement. .

Many insurers are either in the early stages of dealing with social inflation or are not moving as fast as they’d like on the problem. The most common reasons include:

  • Claim handling practices are often inadequately data-driven
  • There is limited ability to foresee litigation
  • There is lack of trust in analytical models
  • The company takes a semi-reactive approach for claim settlement and negotiations
  • Assignments are inefficient, either to claim handlers or attorneys

What can insurers do to manage this growing challenge?

See also: The Data Journey Into the New Normal

Enter machine learning and artificial intelligence

Use of analytics from first notification of loss (FNOL) until the claim is paid is now a norm rather than a competitive edge. All large carriers invest heavily in use cases such as fraud detection, severity-based claim assignment, automatic loss estimation, recovery optimization, etc. However, given the complex nature of how litigated claims are handled, only a few top U.S. carriers are able to weave these capabilities effectively into business processes. Insurance carriers that successfully use analytics to drive business process change in claims litigation will stay ahead of this massive threat.

There are two unmistakable trends that carriers need to leverage:

  1. Aggressive use of information sitting in claims and policy systems (structured attributes, adjuster notes) to develop signals around plaintiff attorney behavior. These signals then need to be deployed within claims operations to encourage early case assessment and litigation prevention.
  2. Use of increasingly clean and comprehensive sources of external litigation information (from state courts, where most insurance litigation lies) to inform your litigation strategy. This includes past verdicts information by venues, judges, attorney firms, case types, etc. A thoughtful use of this information can help claims adjusters and defense attorneys devise the litigation strategy to avoid worst outcomes. There are multiple firms providing research tools that are based on these. Our recommendation, however, is for carriers to ingest the information, merge it with internal claims data and develop models and tools providing a comprehensive view

How carriers can leverage this trend

Carriers can use predictive and descriptive analytics solutions in claims management to mitigate the problem. At a high level, such solutions can lower ballooning legal costs by avoiding litigation and optimizing litigation strategy. Both these approaches call for incorporating advanced analytics processes and models at the key steps of the claim settlement process. This includes:

  • Advanced analytics and machine learning models to predict and avoid litigation and post-suit strategy formulation
  • Natural language processing (NLP) to leverage unstructured data and extract additional insights
  • External data ingestion to augment internal data and enhance model accuracy
  • Enhanced data management capabilities

Any insurance claims raised can result into any of the three possible outcomes: claims without litigation, claims that involve litigation without trial and claims that involve litigation with trial. With the help of the aforementioned technologies, carriers can engineer the following mechanisms in the claims management process (Figure 1) to optimize the legal expenditure for a given product:

  1. A – Prediction of litigation likelihood 
  2. B1 – Defense counsel selection
  3. B2 – Law firm benchmarking 
  4. C1 – Attorney insights 
  5. C2 – Prediction of settlement failure
  6. D – Legal activity duration, legal expense prediction
  7. E – Case-level insights 

See also: Growing Risks of Social Inflation


Claims data within insurance companies is being increasingly seen as a key asset, not a byproduct of the claims process. However, the path to using internal and external sources of data to drive business outcomes is long and arduous. It is becoming increasingly important for carriers to incorporate insurance analytics processes geared toward optimizing legal spending. To achieve this, insurers require a combination of capabilities to these engagements, i.e. ability to handle big data, ability to develop advanced analytics solutions and knowledge of “what, why and how to deploy” in claims business processes.

New Analytics for Small Commercial

Analytics can be a great equalizer in every industry. It’s why 90% of respondents to a McKinsey survey call their analytics investment “medium to high” and another 30% referred to the investment as “very significant” proof that the surveyed understand the value that analytics possesses.

Those investors—especially the commercial insurers—understand the value of analytics and get their money’s worth. In addition to improving sales targets and reducing churn, analytics can increase profitability when it comes to underwriting and selecting risk.

Still, the full potential of analytics goes beyond the insights it provides insurers. When merged with modern technology, data and analytics can fuel efficiency, accuracy and productivity. When used within the decision engine to drive automation, for example, data and analytics can help insurers expedite processes and improve customer experiences, even without human intervention.

Automated reports and actions provide insurers new ways to optimize their day-to-day operations. However, the marriage of automation and analytics is especially vital for the small commercial market as they contend with higher volumes of policy quoting and writing. Using predictive models, automation can reduce the amount of human effort it takes to sell and service policies for small businesses.

Analytics and automation present opportunities to optimize every facet of growing market share for small commercial insurers if properly applied. The sooner that insurers embrace the two, the better off they—and their customers—will be.

Analytics and Automation Can Deliver

When it comes to risk assessment for small businesses, insurers are usually hampered with limited or even misleading information. Unfortunately, this can result in a gap between a risk-appropriate rate and the quoted premium. Thanks to automation and analytics, however, that sort of disparity can be a thing of the past.

See also: What Predictive Analytics Is Reshaping  

While there are many ways analytics and automation can be used to improve the small commercial insurance industry, there are three particular areas where major improvements have been demonstrated. For insurers that are on the fence about committing to analytics and automation, here’s where their influences will likely be most visible:

1. Simplified Applications

By automating customer quoting and underwriting, insurers can phase out the process of collecting troves of information on an application. With reliable third-party data sources, automation can fill in many of the blanks present on typical applications. Insurers will then only need to ask for what’s relevant for the predictive model to assess the risk and provide direction on pricing.

In the same vein, the automation of processes and decisions empowers insurers to use straight-through processing for new applications—quoting and binding policies entirely through an e-commerce experience, without involving staff or consuming staff time. Typically, this is a far more streamlined process for both the insured and insurer, and delivers improved customer experiences.

2. Expedited Claims Processing

Small businesses are acutely sensitive to how long it takes insurers to pay claims and how good (or bad) their experiences are. Analytics helps insurers triage claims while suggesting different processing options.

According to a LexisNexis study, the availability of this data helps shorten processing cycle times by up to 15%. For example, through IoT (internet of things) devices, an insurer can detect water heater leaks and other high-risk problems in real time, enabling the insurer to anticipate potential claims and possibly even prevent them.

Of course, being fast is only part of the equation—the process must also be accurate. Thankfully, automation and analytics improve processes by catching overlooked data points. When sophisticated analytics are applied against a large sample of detailed claims data, the resulting insights can, for example, highlight the best way to get an injured employee back on his or her feet and offer a customized plan to do so.

See also: What’s Beyond Robotic Process Automation  

3. Improved Risk Identification

By using reliable third-party data, such as information available through a data consortium, insurers can more quickly and accurately identify risk—even if it’s in a sector where they have little or no experience—and ensure that risk-appropriate pricing is quoted. Analytics thus becomes a valuable growth engine for insurers to confidently expand into different business lines and regions.  In an environment where 40% of the smallest organizations have no business insurance whatsoever, insurers that embrace modern technology could reap significant rewards. By combining analytics with automation, the small business insurance market could be transformed—which would be welcome news for both insurers and their customers.

3 Technologies That Transform Insurance

The combination of artificial intelligence (AI), robotic processing automation and predictive data analytics is fundamentally redefining how businesses operate, how consumers engage with brands and, indeed, how we go about our daily lives. The field of insurance is no exception. Outlined here are three ways smart technology is affecting insurance, with a focus on identifying lessons learned and defining specific keys to success.

Back Office Robotic Process Automation

The impact of rules-based robotic process automation (RPA) on insurance operations has been well-documented. RPA tools are driving efficiency and productivity gains in generic back-office functions such as F&A and HR, and insurers are tackling processes related to claims administration and account management.

One key challenge is scalability. In many cases, concept initiatives have failed to gain traction, resulting in isolated pockets of automation that yield limited benefit. In others, overly ambitious enterprise-wide projects struggle with boil-the-ocean syndrome. A well-defined center of excellence (CoE) model that develops and documents best practices and then propagates them across different business units has proven effective.

Another critical lesson has been the importance of CIO involvement. This was lacking in many early RPA projects. For one thing, because RPA tools focus on process and business functions rather than programming skills, CIOs often weren’t interested. Business unit heads, moreover, feared that CIO involvement would lead to bureaucratic logjams and derail aggressive adoption schedules. Practice has shown, however, that CIO oversight is essential, to avoid both general shadow IT problems as well as specific interoperability, stability and security issues related to RPA functionality.

See also: Using Technology to Enhance Your Agency  

Leading early adopters have also continually pushed the envelope of automation levels. In a claims processing environment, 70% of claims may be simple and straightforward, and therefore ideally suited to an RPA application. At the other end of the spectrum, 5% to 10% of claims are complicated and unusual, and therefore require a human’s expertise and judgment to evaluate. While doable, automating these complex outlier claims isn’t cost-effective. The challenge then becomes to focus on the remaining 20% to 25% of claims. By analyzing the frequency of different types of claims, insurers can identify cases where the time and effort needed to configure a bot will yield a return.

Applying Cognitive Capabilities to RPA

RPA has delivered impressive benefits to insurance operations in terms of cost reduction, accuracy and auditability. That said, the tools are limited to the specific if/then rules they’re configured to follow. If a bot encounters a scenario that doesn’t align with what it’s been taught, it gets stuck.

More advanced cognitive systems apply pattern recognition to analyze unstructured data to identify key words and phrases in context. This promises to take insurance automation to the next level. While an RPA bot can extract a specific piece of data such as a policy number from a specific form, it can’t interpret underwriting rules or aberrations from a form on which data is unstructured and organized differently.

A cognitive application, meanwhile, can scan documents of various types and formats and apply machine logic and learning to identify relevant data in spite of discrepancies in how the data is structured or presented. This allows people to focus on policy/claim exceptions rather than formatting issues. More specifically, by injecting cognitive applications into operational workflows at key “intelligent gates,” insurers can more easily identify aberrations in unstructured data and highlight the policies and claims that require further human involvement.

IoT, AI and Insurance Underwriting

The combination of Internet of Things (IoT) and artificial intelligence will have perhaps the most transformational impact on insurance. By deploying networks of smart, connected IoT sensors, insurers can collect and analyze volumes of data at the point of critical business activity. Leveraging the pattern recognition and predictive analytics powers of AI, meanwhile, creates insights that insurers can use to refine actuarial tables and improve the rules of underwriting.

Consider these examples:

  • Sensors in vehicles ranging from commercial trucks to passenger cars monitor and document speed and driver behavior. Insurers can analyze data to calculate accident probabilities of safe vs. risky drivers over time. Based on those calculations, premiums could be adjusted. Smart sensors and cameras can also detect drowsy drivers or erratic behavior, triggering alarms.
  • Smart home technology that monitors suspicious activity and automatically shuts off water pumps in the event of a burst pipe can lead to lower homeowner policy costs, particularly for premium coverage such as insuring valuable artwork from theft and damage.
  • Pharmacies that store and transport medicines can deploy temperature and humidity monitors to ensure that supplies stay within required guidelines. Reducing the risk of tainted medicine reaching consumers could reduce liability risk.
  • Smart video analytics can determine wear and tear of roofs, oil pipe damage from foliage and animal migration and levels of water and soil contamination. Such insights enable corrective action before catastrophes strike and reduce the level of unforeseen risk for underwriters.
  • By monitoring pressure or fluid flow in an oil pipeline, sensors can trigger shut-off valves if limits are exceeded, thereby preventing costly environmental damage.

Innovative insurers are exploring how to deploy these capabilities into policy formulation. For instance, customers who adopt the technology could qualify for discounts. (Given the privacy issues surrounding driver monitoring, the voluntary aspect would seemingly be critical for auto insurance policies.) Another option: Insurers team up with technology partners to offer smart sensor services, thereby helping policyholders while creating revenue streams.

See also: Smart Home = Smart Insurer!  

The combination of sensor array and intelligent technology is refining underwriting and claims payment. Insurers can tune actuarial tables and pricing models to cover potential losses before they occur, as well as avoid incidents by advising policy owners to take corrective actions. In other circumstances, sensors can take action on their own.

Ultimately, these capabilities will enable insurers to fundamentally redefine their operational and customer engagement models.

Could Risk Analysis Win the Lottery?

I started writing yet another article trying to convince risk managers to grow their quant competencies, to integrate risk analysis into decision-making processes and to use ranges instead of single-point planning, but then I thought, why bother? Why not show how risk analysis helps make better risk-based decisions instead?

After all, this is what Nassim Taleb teaches us. Skin in the game.

So I sent a message to the Russian risk management community asking who wants to join me to build a risk model for a typical life decision? Thirteen people responded, including some of the best risk managers in the country, and we set out to work.

We decided to solve an age-old problem – win the lottery. With help from Vose Software ModelRisk we set out to make history. (Not really: It’s been done before. Still fun, though).

Here is some context:

  • Lotteries are an excellent field for risk analysis because the probabilities and range of consequences are known
  • In Russia, as in most countries, lotteries are strictly regulated.
    There is a rule: When a large amount accumulates, several times a year it is divided among all the winners. This is called roll-down.
  • If no one takes the jackpot before or during the roll-down, then the whole super prize is divided among all other winners
  • So the probability of winning is the same as usual, but the winnings for each combination can be significantly higher if no one wins the jackpot.

We set out to test our risk management skills in a game of chance.

June 8, 2019

Whatsup group created. Started collecting data from past games. Some of the best risk managers in the country joined the team, 15 in total: head of risk of a sovereign fund, head of risk of one of the largest mining companies, head of corporate finance from an oil and gas company, risk manager from a huge oil and gas company, head of risk of one of the largest telecoms, infosecurity professionals from Monolith and many others.

June 9, 2019

Placing small bets to do some empirical testing.

June 10, 2019

First draft model is ready…

June 11, 2019 

Created red team and blue team to simultaneously model potential strategies using two different approaches: bottom up and top down. Second model is created…

June 12, 2019

Testing if the lottery is fair, just in case we can game the system without much math. Yes, some numbers are more frequent than others, and there appears to be some correlation between different ball sets but not sufficient to produce a betting strategy. The conclusion – the lottery appears to be fair, so we will need to model various strategies.

June 13, 2019

Constantly updating red and blue models as we investigate and find more information about prize calculation, payment, tax implications and so on. The team is now genuinely excited. Running numerous simulations using free ModelRisk.

June 14, 2019

Did nothing, because all have to do actual work.

June 15, 2019

After running multiple simulations, we selected a low-risk, good-return strategy. Dozens more simulations later, here are the preliminary results, using very conservative assumptions:

  • probability of loss 9.8%; worst-case scenario, we lose 60% of the money invested
  • probability of winning 90%; 80% of the time, winnings would be between 50% and 100% of the amount invested, after taxes (this means there is a high possibility to double invested cash at relatively low risk)
  • potential upside significantly higher than downside

Red and blue team models produced comparable results.

June 16, 2019

Started fundraising.

If we manage to collect more than the required budget, we decide to make two bets: one risk management bet (risk management strategy) and one speculative bet with much higher upside and as a result greater downside (risky strategy).

Full budget collected within just a few hours. Actually collected almost double the necessary amount and, as agreed, separated 50% of the funds into the second investment pool. Separate team set out to develop the risky strategy. While I was an active investor in the risk management strategy, I decided to play a role of a passive investor in the risky strategy and only invested 16% into the risky.

June 17, 2019

Continued to develop the model, improving estimates every time. Soon, we felt the financial risks were understood by the team members, and we needed to take care of other matters before the big day.

First, took care of legal and taxation risks. Drafted a legal agreement clearly stating the risks associated with the strategy, the distribution of funds and the responsibilities of team members. Each member signed. Agreed to have an independent treasurer.

Then started to deal with operational risks. Apparently transferring large sums of money, making large transactions and placing big bets is not plain vanilla and required multiple approvals, phone calls and even a Skype interview. Five team members in parallel were going through the approvals in case we needed multiple accounts to execute the strategy.

Probably the biggest risk was the ability of the lottery website to allow us to buy the tickets at the speed and volume necessary for our low-risk strategy. This turned out to be a huge issue, and we found an ingenious solution. The information security team at Monolith did something amazing to solve the problem, and I mean it, amazing. I have never seen anything like this. It’s a secret, unfortunately, because, you guessed it, we are going to use it again.

The strategy that the lottery company recommended for large bets is actually much riskier than the one we selected. How do we know that? Because we ran thousands of simulations and compared the results.

June 18, 2019

The lottery company changed the game rules slightly. Ironically, this slightly improved our 90% confidence interval and reduced the probability of loss. So, thank you, I guess.

More testing and final preparation. The list of lottery tickets waiting to be executed.

In the true sense of skin in the game, team members who worked on the actual model put up at least double the money of other team members.

June 19, 2019

8am. We were just about to make risk management history. A lot of money to be invested based on the model that we developed and had full trust in. I felt genuinely excited: Can proper risk management lead to better decisions? I am sure other team members were excited, too.

By about lunch time, the strategy was executed. We bought all the tickets. Now we just had to wait for the 10pm game. Don’t know about the others, but I couldn’t do any work all day. I couldn’t even sit still, let alone think clearly. Endorphins, dopamine, serotonin and more.

At 9:30pm, we did a team broadcast, showing the lottery game as well as our accounts to monitor the winnings, both for excitement purposes and as full disclosure.

Then came the winning numbers. Two team members actually managed to plug them into the model and calculate the expected winnings. We had the approximation before the lottery company did.

You guessed it: We won. Our actual return was close to 189% on the money invested after taxes (or 89% profit; remember, our estimate was 50% to 100% profit, so well within our model). We almost doubled our initial investment. Not bad for risk management. (Good luck solving this puzzle with a heat map.)

June 20, 2019

More excitement, model back-testing and lessons learned — and, perhaps the most difficult part, explaining to non-quant risk management friends why, no, this was not luck; it was great decision making.

In fact, our final result was close to P50. We were actually unlucky, both because we didn’t get some of the high-ticket combinations and, more importantly because five other people did, significantly reducing our prize pool.

Let me repeat that: We were unlucky and still almost doubled our money.

June 21, 2019

Job well done!

Changing Nature of Definition of Risk

As the foothold of innovation across industries grows stronger by the day, insurers are witnessing the advent of tech-based economies, and with them a fundamental shift in the very definition of risk. Every advancement stands to revolutionize how property, businesses and employees will be insured. Consider automated cars and workplace automation tools, such as Amazon warehouse robots, or the emergence of shared ownership business models, like Lyft and AirBnB. Traditional risk calculation models need to evolve to keep up with rapid change.

How shall insurers prepare for this shift? According to Valen Analytics’ 2019 Outlook Report, a key part of the answer lies in the need to weave data and predictive analytics into the fabric of their business strategies. The report, which employs third-party and proprietary data to identify key trends, revealed:

Insurers Are Heavily Relying on Advanced Use of Data and Analytics to Fuel Growth

Valen’s Underwriting Analytics study found that 77% of insurers are incorporating predictive analytics into their underwriting strategy. This marked an increase compared with the steady 60% of insurers during the past three years, demonstrating a clear emphasis by the industry on data-driven decisions.

While many factors have fueled the demand for sophisticated data and analytics solutions, one stands out. Insurers have a growing desire to reap a share of the underserved small commercial market, which represents over $100 billion of direct written premiums. Data analytics tools enable insurers to reduce the number of application questions, verify necessary information and ascertain risk much more quickly and accurately. This is particularly important in creating effective business models that align with the needs of small business owners.

The rise in insurers looking to employ advanced data analytics techniques has also resulted in the growth of data aggregation services and consortiums. With new primary customer data sources emerging, insurers have access to better insights on consumer risk and behavior. This has contributed to insurers’ appreciation of the predictive horsepower that large pools of data offer. In fact, Valen’s proprietary research found that the synthetic variables appended with consortium data are as much as 13 times more predictive than policy-only data. Synthetic variables are built from computations of more than one variable, made possible by leveraging large and diverse datasets.

See also: Understanding New Generations of Data  

Regulation and Innovation Must Go Hand-in-Hand

With a rise in advanced predictive analytics and robotic process automation in insurance, regulators are paying close attention to the industry. To ensure this oversight doesn’t stifle innovation, it is important that insurers build and document their analytics initiatives so they can be explained and understood by regulators. Being collaborative and responsive will help ensure that regulators can discern the small percentage of use cases that need to be reviewed for consumer fairness protection. In doing so, insurers have the opportunity to take the industry to Insurance 2.0 — the next phase in technology adoption and innovation.

Talent and Infrastructure Challenges

While insurers are looking to integrate data and predictive analytics into their business strategies, what will truly determine their success is their ability to hire and nurture the right talent. Unfortunately, the industry continues to suffer from a lack of the talent needed to support fast-paced innovation. Seventy-three percent of insurers surveyed indicate moderate to extreme difficulty in finding data and analytics talent, and the reasons haven’t changed over the years. While geographic location of the job is the primary reason cited by the survey respondents, more and more prospects are either looking for better compensation packages, are simply not interested in an insurance career or opt for opportunities in tech startups or data-driven companies in other fields.

Another roadblock for insurers is their dated IT infrastructures, which cause massive backlogs. While most insurers suffer backlogs of two years or more, others cannot identify how long their IT backlogs are.

See also: Insurance and Fourth Industrial Revolution  

Both of these problems go hand in hand. Clearly, there is a need to foster an innovation mindset, and, to do so, the industry needs a mix of new thinking and engaging work culture. Insurers should follow the footsteps of leading tech companies and cultivate a culture that appeals to high-level talent. By making small changes, such as embracing diversity and a remote workforce, insurers can make themselves attractive to the talent they need. This will build a workforce capable of overcoming IT infrastructural issues.

In short, to maintain a competitive advantage, insurers must not only put data and analytics at the forefront of their businesses, but also make strategic decisions on how best to employ them to enhance all aspects of their businesses, from customer service and information handling to risk calculations and claims processing.