ITA LIVE Top Banner
Follow Us



Plugging the Underwriting Leaks

Michael P. Voelker | December 19, 2014

It’s an annual ritual for many homeowners. As winter approaches, they go around the house caulking gaps that let the cold air in and drive heating costs up, because leakage costs money.

Leakage in commercial underwriting costs money as well. “One aspect of leakage in the underwriting process is mispricing business—charging too little so that you lose money through a high loss ratio, or charging too much and losing the opportunity to write the business,” says Karlyn Carnahan, research director in Celent’s insurance practice.

To plug the leaks in the commercial underwriting process, insurers have turned to technology.

“Insurers have reached the limit in ways to control expenses and loss ratios through non-technology approaches, and now they are looking at technology,” says Deb Smallwood, founder at SMA. “Technology gives them discipline, consistency, and efficiency. It won’t automatically make them write more profitable business, but it will enhance the underwriting process and let underwriters focus on higher-value tasks.”

“There is a lot of activity and investment going on in using technology to improve the underwriting process,” Carnahan observes. “Obviously, we are seeing a lot of technology being applied on the personal lines side, but we’re also seeing it in small workers’ compensation and BOP, where there have been several advances in straight-through processing. Carriers are now showing increased interest in applying technology to the middle market book, both to increase efficiency and to guide and improve underwriting decisions.”  

According to a 2014 research report by SMA, commercial underwriting systems ranked third on IT project budgets. More than two-thirds of all insurers are planning to spend over 10 percent of their IT budget on commercial underwriting automation technology over the next 18 months. Among insurers that write simple accounts, 15 percent say that they will spend over 30 percent of their IT budget on underwriting, while 20 percent of insurers that write complex accounts will do likewise.

Pricing Precision

“There’s a saying that the only bad risk is an underpriced risk. We do what we can to not underprice a risk,” says Greg Tacchetti, CEO of AssureStart, a Seattle-based insurer that sells business owners policies direct to small businesses.

In developing its highly automated approach to commercial underwriting, AssureStart began with the philosophy that premium leakage is caused not by inadequate rates, but by excessive crediting in the individual risk premium modification (IRPM) process that is part of the typical underwriting evaluation.

“Generally risks are over-credited as a total group,” claims Jim Swegle, AssureStart’s chief product officer. “That’s just a dynamic of the traditional distribution model. The underwriter gets a call from the broker who applies pressure to give credits. New business applications are rarely debited.”

AssureStart developed a proprietary pricing model with Deloitte that the insurer believes mirrors the individual-risk precision that has been established in personal lines. The model helps support a fully automated platform that includes a proprietary web-based policy application system and a cloud-based deployment of MajescoMastek’s STG Suite for policy administration and billing, including automated use of ISO’s Electronic Rating Content. The pricing model analyzes a combination of information that customers enter on AssureStart’s online application with data pulled from a variety of industry information sources, such as Experian, as well as public information sources.

“There is still much more robust third-party data in the personal lines space, but in the last 10 years there has really been a proliferation of data available to the commercial underwriting process, much of which is free,” Tacchetti says. “It’s easy for us to use that data to validate the information customers put in on the website regarding their class of business to be sure it’s classified correctly from a rating point of view, and also to augment what a customer enters with additional information.” 

Use of external data sources to increase both efficiency and accuracy continues to grow in commercial lines as it has in personal lines.

“More and more carriers are prefilling apps with third-party data, which can improve the accuracy of data obtained [from the agent] and also streamline the process for an agent. Agents are also looking for the ability to upload or enter applications with the least amount of information necessary into the carrier’s system and get a rapid indication of whether an account is a ‘yes’ or a ‘no,’” says Carnahan.

AssureStart’s application asks for just four pieces of customer-provided information: zip code, description of operations, whether the business location is owned or rented, and the number of employees. Detailed geocoding allows risks to be rated in what Tacchetti refers to as “micro-territories,” and AssureStart has also applied technology to the IRPM process.

“We have automated the application of discounts based on factors such as how long you’ve been in business, your commercial credit score, claims history, and other stability factors,” Swegle explains.

Although it’s still early in the company’s history—AssureStart launched its first state in November 2013— Swegle says that the company’s results are promising thus far. “We’ve had very few claims, and profitability has actually been better than we modeled,” he said.

Automation also targets leakage by keeping AssureStart’s expense ratio low: about 20 percent, compared to the 30s for most commercial carriers. “We automate the process for small business and make money in the process. It’s a classic win-win,” Tacchetti says. 


Working with Deloitte, AssureStart invested $3 million in a data warehouse and analytic tools in order to fine-tune its underwriting and pricing models going forward. “We are very data-driven,” says Tacchetti.

Applicants that don’t pass the automated system can choose to connect with a licensed agent in the company’s call center to determine if they are eligible. However, there are no exceptions made for either eligibility or IRPM.

“That was a conscious decision on our part for two reasons,” Tacchetti says. “First, we’ve found there is a material difference in underwriting results—and not a positive difference—when credits or exceptions are applied at the desk level.  Second, it helps with regulatory compliance because there is no subjectivity in the process.”

Ineligible businesses are referred to Bolt Solutions, which has partnered with AssureStart to provide nationwide market access to small businesses. “Going forward, we’re tracking which businesses don’t fit our underwriting model to see if we can expand our appetite,” Swegle says. “The class we see most frequently that doesn’t fit is contracting. We hope to eventually have a market for that, but it’s been a challenging class because of the amount of variability within it.”

Louisiana Workers Compensation Corporation (LWCC) is undertaking a project to increase its pass-through rate from 20 percent to 50 percent of new business applications.

“Since we are the market of last resort, we have to accept almost every application that is submitted to us. When we spend time underwriting accounts that we’d prefer not to write, it takes away from working with agents and being more competitive on preferred risks,” explains Karen Mouton, manager of LWCC’s project management office.

“Freeing up underwriters’ time could allow them to visit agents in the field more often as well,” adds Kelly Pocorello, project manager.

At the heart of LWCC’s underwriting automation is CompZone, a distribution portal based on technology from Agencyport. Agents can enter application information within the CompZone portal, work renewal policies, and request endorsements. The system also accepts uploads from a half dozen supported agency management systems. A rules-based workflow determines which accounts are automatically approved and which ones are referred to underwriting.

LWCC’s project to improve the pass-through rate is a two-pronged effort, involving adding questions to the application process to reduce the need for information-based referrals and analyzing what the impact of rules changes would be based on past account experience. The insurer expects to complete the effort by the second quarter of 2015.

Tech for Larger Lines

“You can’t automate complex risks to the extent that you can automate personal auto, home, or even small commercial, because there has to be some underwriter-driven risk and financial/loss analysis that goes on. Also, in larger commercial lines, the leakage picture is more complex because underwriters have a lot of levers they can pull to price accounts,” Smallwood says.

Although straight-through processing is still limited to smaller commercial lines, SMA found in its research that insurers are paying closer attention to how they can automate at least parts of the underwriting of complex and specialty risks as well.

“When you look at the underwriting process from start to finish, there are many steps that underwriters do where underwriting judgment is not needed—ordering reports, entering information into the policy administration system,” Smallwood says. However, she is surprised with the manual processes that still exist at some carriers.

“I have met with companies where underwriters still key state, class code, payroll, and other information into standalone spreadsheets to calculate a loss ratio, or where they go to a bunch of different websites to get information, or pull data manually from NCCI. They hunt and peck and search and re-key. There’s no reason those processes can’t be done automatically and the results brought to their desktop. At least just give underwriters a link to click rather than having to search. Give them prepopulated and integrated spreadsheets. Find the processes in the underwriting workflow that don’t require judgment and automate them,” she says.

Automation in larger lines of commercial insurance is almost exclusively targeted at helping insurers be more informed and efficient in those non-judgment processes. Underwriting workstations are being used to create a one-stop shop for underwriters, consolidate information, and help underwriters visualize risks and hazards through property-specific mapping and geospatial data. In SMA’s study, most insurers identified the need for underwriting workstations as a high priority for technology investment.

“Today, most companies are already using intranet sites or portals to have underwriting guidelines, rules, and forms in a single place—somewhere that they can go to pull everything they need rather than keeping it on their desktop or even on paper where they deal with lack of consistency,” Smallwood says. Workstation technology takes that consistency to the next level by using rules or analytics to guide or automate the ordering and receipt of risk assessment data, provide analytics-guided decision support, enable collaboration among underwriters, and more.

“One use of predictive analytics in the commercial underwriting space is to optimize the ordering of third party data; for example, determining which motor vehicle records will likely have information on them and are worth ordering. Carriers also use predictive analytics to determine what level of services are provided on an account, such as loss control or premium audit, will create the best results. It’s about making those decisions not just based on an account’s premium size, but on its individual characteristics,” Carnahan says.

“Analytics are being applied more and more in commercial lines to send up red flags for things that underwriters might not see—not because they aren’t skilled, but because the human brain can only look at so many different elements at one time,” Smallwood says.


Hawaii Employers’ Mutual Insurance Company (HEMIC) has deployed a combination of straight-through processing and underwriting workstation technology to increase efficiency. HEMIC first installed Valen’s predictive modeling engine in 2008, which integrates with the insurer’s policy administration system from Tropics Software. In 2013, the company upgraded to Valen’s InsureRight Platform.

“The driver of both the original implementation and the 2013 upgrade was having a cost-effective, consistent way to portfolio-manage our book of business,” says Regina Harris, HEMIC’s vice president of underwriting. “We also wanted to provide quicker responses to agents on quotes.”

The Hawaii workers’ compensation market consists of predominantly small businesses, so efficiency is important in reducing underwriting leakage. HEMIC’s average account premium is just over $8,000.

Agents can submit new business via HEMIC’s online portal or through integration with a supported agency management system. Applications can also be entered into the Tropics system by HEMIC’s underwriting staff. When application information is entered, an automated workflow processes the information through the Valen platform, with the result generating either a quote based on risk characteristics and scoring or a referral to an underwriter for further review.

For accounts that fall outside straight-thorough guidelines, the system provides HEMIC underwriters guidance about risk classification and pricing through predictive modeling and the use of industry data. That type of guidance is a critical part of today’s commercial underwriting technology, Smallwood says.

“You need a workflow that connects the scoring model to the policy administration system and can automatically trigger underwriting review, as well as a rules engine that can help automate underwriting guidelines,” she says. “Those could be guidelines around hazard scores, class codes, underwriting authority, and other rules that tend to be outside the scope of a policy administration system. Automating that process establishes consistency and targets leakage.”

HEMIC’s 2013 system upgrade provided the insurer with access to two additional, proprietary predictive models: Valen’s Misclassification Model and Premium Impact Model. Those models identify, early in the individual risk assessment, potential exposure misclassifications and policies where underwriters should consider premium adjustments.

“By running new business through those models based on risk characteristics, it identifies areas where there may be potential leakage on the front end, rather than on the back end,” Harris says. “Before implementing those models, it was incumbent on the underwriter to catch discrepancies.”

Underwriters may override the system’s classification or pricing recommendations, but that process is carefully managed.

“Accounts that go to an underwriter still go through the traditional underwriting process. However, the system records the original recommendation and the underwriter is required to include notes on why that recommendation was not followed. There are also authority levels set on who can modify pricing and how much,” says Harris.

On renewal, policies are also sent through the Valen engine. As with new business, accounts may be automatically processed for renewal or referred to an underwriter for further review.

“We are seeing a considerable number of new and renewal policies being referred for potential misclassification each week” since deploying the InsureRight models, Harris says.

“We have also provided agents the ability to quote risks not needing underwriting review online and gained underwriting efficiency by having the scoring engine perform the initial ‘triage’ on accounts that do need review,” adds Harris. “It’s helped underwriters save time in handling accounts and focus their skills on more complex underwriting cases. Our underwriting results have been favorable.”

“We’ve seen companies get anywhere from a four to ten point lift in loss ratio when they implement predictive models to drive risk assessment and pricing optimization, and it seems to be a sustainable lift,” says Carnahan. “That’s definitely why we are seeing analytics being used more and more in commercial lines to provide guidance to underwriters in assessing risk quality and provide pricing guidance to underwriters in terms of what premium modification off manual rates an account justifies based on the risk characteristics of the account.”

Despite the benefits to HEMIC’s bottom line, Harris says there was some cultural resistance to the rollout of underwriting automation. “Underwriters, particularly old-school underwriters, tend to think they need to have their eyes on every account,” she explains.

Overcoming that resistance required showing underwriters that the platform produced profitable results while allowing them to apply their experience to accounts that warranted an expert review and focus on more rewarding tasks.

“It took time to gain comfort with the system,” Harris says. “It was a cultural shift, but we’ve made that shift.”

Predictive analytics to address leakage on a book level is next on the list for HEMIC. In early 2015, the company plans to implement the Manage component of the InsureRight Platform, which incorporates anonymized industry data from Valen’s proprietary database. HEMIC plans to use this industry data to better understand how to further improve underwriting results. 

“We will be looking to that component to identify trends within our book, as well as track our KPIs against our goals,” Harris says. “We will also be able to more easily run ‘what if’ scenarios that compare our underwriting decisions across a book of business to what the results would have been if we had made different decisions.”

Plugging the Gap

Book management is important because, like a drip of water that gradually becomes a steady flow, underwriting leakage can have a compounding effect over time. Underpriced accounts are more likely to be sold than polices that are correctly priced, pushing loss ratios up. If a carrier raises rates to compensate—rather than analyzing the true cause of leakage—it can find itself less competitive on good risks, creating opportunity leakage.

Just as no single solution can plug every air leak around a home, no single technology can address every source of underwriting leakage, but Smallwood believes that insurers are committed to plugging the gap.

“As the industry completes automation of personal lines, it will further increase its investment in commercial lines underwriting technology,” she says. “I think we’re on the verge of some really exciting technology advances.” 


Featured articles

Guidewire MR



The Email Chat is a regular feature of the ITA Pro magazine and website. We send a series of questions to an insurance IT leader in search of thought-provoking responses on important issues facing the insurance industry.


The tide is up! It's time to register for ITA LIVE 2019, our annual educational and networking conference! Our theme is "The InsurTech Revolution: Cutting Through the Hype." and we'll be bringing in a torrent of industry thought leaders, amazing insight and wonderful perspectives on the world of insurtech and its impact on the insurance landscape.

ITA LIVE 2019 will present real-life examples of true startup technologies that are helping insurers gain real advantage -- and a competitive edge -- in the marketplace. We’ll highlight the more successful InsurTech partnerships, while offering case studies that demonstrate exciting innovation and cutting-edge techniques impacting all aspects of the insurance ecosystem.

Ride the wave to LIVE 2019. Sign up today! We look forward to seeing you in May, 2019!


only online

Only Online Archive

ITA Pro Buyers' Guide

Vendor Views

Partner News