Mr. Springer and other experts interviewed during the Target Markets Program Administrators Association midyear meeting in Atlanta last month all stressed the need to do a lot of heavy lifting before they even begin the process of presenting a program idea to a carrier.

“We do a huge amount of research to try and satisfy ourselves that it makes sense. We'll invest hundreds of thousands of dollars in a new program before we even bring it to market,” he said, noting that money is spent on actuarial and other types of expert analysis.

Mr. Springer, who said his firm has added five or six new programs in the past 12 months, said his staff has a list of some 25 potential program ideas, but doesn't work on all of them at one time.

He described one recent program launch called Site Pro, which targets grading and excavation contractors, as a companion program to the oldest program written by NIP–Land Pro Tree Pro, a property-liability program targeting landscapers and tree services.

“The very best ideas for new programs [come about] when we have to decline a lot of submissions,” he said, noting that many landscapers do grading and excavation–”just not as much as someone who focuses on that business.”

“We see a lot of submissions [and] we know our agents have that business to give us. So that helps us with the deployment strategy,” he said.

“A lot of the ideas for new programs just develop,” said Phil Harvey, president of Venture Programs in West Chester, Pa., who described how an existing relationship he had when he was a retail agent became the basis of a program called HomeBase when he moved to the wholesale/MGA side of the insurance business.

“Even though we're dealing on the wholesale platform through Venture, the customer–a large public real estate investment trust–didn't want to lose our expertise,” Mr. Harvey said, going on to explain how the relationship sparked a program targeting property managers of privatized military housing.

The REIT was “one of the first public companies to be on the short list with the government to qualify for home-base housing,” he said, explaining that the HomeBase program was successfully built off the REIT's data.

But what do you do if you can't get the historical data to develop a sound underwriting plan, one TMPAA conference attendee asked Ken Robinette, a 25-year veteran of the program business market who led a breakout session on putting programs into practice.

“You can find it,” said Mr. Robinette, a program administrator for Bellingham Underwriters in Bellingham, Wash.

“There's an enormous amount of data in this world. Most of us get lazy. We go on the Internet once and if we don't see it, or our agent doesn't give it to us, we don't think we can get it,” he said, describing how he typically enlists the help of research assistants at a local university to retrieve the information he needs.

Responding to a request for “every piece of information” they could find on “causation patterns for food-borne illness,” he said the Ph.D.s in research science were able to put 3,000 pages of material in front of him within four hours. “I spent the next four weeks pouring through the data,” he said, explaining that he has approached the research task for every program he's done exactly the same way.

Mr. Springer said the development of a program submission package typically takes three or four months of really focused activity, and TMPAA President Greg Thompson said a lot of wholesalers and MGAs fall down at this stage of the process.

“They don't know how to put together the submission,” said Mr. Thompson, who is also president, chairman and chief executive of THOMCO, a Kennesaw, Ga.-based program administrator specializing in industry-specific programs, including daycare centers, senior living centers, pest control services and tanning salons.

A typical THOMCO submission is a large spiral notebook with 15-to-30 tabs that has information about the industry and details about typical losses, he noted.

“You really have to get to know that industry because when you go to an insurance company and you're putting their capital at risk insuring an industry, they need to be convinced that you know what you're talking about,” he said.

“When you think about it, a program administrator is an independent contractor for underwriting. An insurance company delegates authority to us in a particular industry because they believe we understand it [and] know how to underwrite it,” he said.

Mr. Thompson said making an effort to learn everything about a particular industry can start on the Internet, while advising that program administrators should also go out and visit typical businesses in the industry they're targeting.

“Talk to the businessowners. What are their biggest concerns–even non-insurance concerns?” He added that to get your arms around the insurance exposures, program administrators need to get copies of the insurance policies and also need to find out what the competition is charging.

“If they've filed rates, you need to get copies of those. You have to understand your competition to know how your product is going to be successful,” he said.

Mr. Springer noted that once a carrier gets a program prospectus in its hands, it could take as long as a month before they say whether they're willing to work on it. After that, timing depends on a number of factors, such as whether the program administrator will need to make specific product filings or where there are rule changes that go along with the program being offered.

“Under ideal circumstances, you could conceivably have a program up 45-to-60 days after you meet with a carrier,” he said. “More often, however, because you're going to be differentiating your product somehow, it will require more work. It could be six, eight or nine months before you get a program out and you're actually soliciting business.”

A critical part of a program administrator's job is to set up a plan to select business so that its insurance company partner doesn't end up with a bad loss ratio, according to Mr. Thompson. “If they end up with a bad loss ratio, you lose your reputation–and they're going to throw the program away,” he said.

Mr. Robinette, teaching a Program Business 201 workshop on advanced concepts of successful program management, advised that “you can't guarantee a loss ratio, but you can certainly reduce the uncertainty involved in a book of business.”

He went on to give a somewhat unique perspective on how to develop an underwriting plan, basing his presentation on his own successful management of a trucking program to a 50 percent unlimited loss ratio.

“You can manage your results,” he said.

Mr. Robinette differentiates his firm by building his underwriting models around one central belief–”that to be successful, you must build plans that are in no way, shape or form involved with qualitative risk decision-making.”

“It's effectively impossible for an individual underwriting firm to pick the better [quality] risk because everyone is trying to do the same thing,” he said. “We all understand what 'better' means. It means better management, better employees, better housekeeping. Everyone understands the same rules.”

Further explaining why he creates a completely different set of rules for his programs–building rating plans around “operational” risk characteristics instead of “qualitative” ones, he described how an underwriter using the latter approach might price a risk using a bureau schedule modification plan.

Such a plan assigns 5- or 10 percent credits and debits for better and worse qualitative factors, he explained.

“The problem with that is that everyone does the exact same thing. So the market basically tells you where you're going to end up,” he said. “When the market is soft, you always end up with a credit. When it is hard, you get a debit. But you're only getting a swing of 10-to-15 percent, and it's not a way of differentiating you.”

He added that “if you're going to be an MGA, you have to differentiate yourself. We can't be like an insurance company. We have to be better than an insurance company or we don't exist because that's the only reason we're here. If we're doing what an insurance company does, they don't need us.”

Distinguishing “operational” risk characteristics from “qualitative” ones by using the example of underwriting sawmills, Mr. Robinette contrasted the idea of choosing to write a certain sawmill because management has been in place for 30 years from an approach that focuses on type of wood that's being cut–an operational distinction.

Noting that the industry rating basis for sawmills is receipts, and that a sawmill takes in $4 per foot for cutting pine but $22 per foot for red oak, he said, “I can write the absolutely most disgusting red oak mill and I get five times the premium I get with an absolutely cream-of-the-crop pine mill.”

Mr. Robinette builds his program underwriting/pricing models by collecting data, and then analyzing all the research he gathers about a particular class to determine the operational risk characteristics that are primary causes of loss. Bureau rating plans referenced by the rest of the industry are typically off the mark in this area, he said.

For example, he said that by doing an analysis of loss causation for trucking risks, he found that risk is not dependent upon where a truck is garaged or how big it is–factors included in the bureau commercial auto rating plan used by many insurers.

Instead, 87 percent of all claims are driver errors. So the driver should be in the underwriting plan somewhere, he said, also asserting that the type of road is another true causation factor.

A truck garaged in New York that drives to St. Louis gets a dramatically higher premium under the industry's plan than one that goes from St. Louis to New York, “but they're the exact same risk. They're going over the same route,” he said.

For the underwriting and pricing model for its trucking program, Bellingham creates an eight-digit class to correspond to operational factors such as type of road, age of driver and miles driven. “They have nothing to do with the quality of the risk,” Mr. Robinette said.

He said the rates he uses to populate his class plan have allowed his program to achieve a limited loss ratio (where individual losses are capped at $100,000) of 25 percent in every year since 2002, “regardless of where I write, regardless of whether its long-haul or short-haul, [and] regardless of whether it's a landscaper or a truck hauling nuclear waste.”

“It doesn't make much difference. The front of the truck is what causes the accident,” he said.

While he said that building an underwriting plan around true causation is the “only way to get away from the trends of the marketplace” from a profit perspective, his firm still has a hard time writing business when the market is soft,” he said.

Indeed, he estimated roughly 25 percent of the business in the country is priced at a level where he can write at to achieve his targeted loss ratio during a hard market, and only 3-to-5 percent is available during a soft market.

“I just have to quote more business, and I'm going to have a much lower hit ratio. But my loss ratio is going to be exactly the same today as it was during the hard market,” he said.

Mr. Robinette spent the remainder of the session demonstrating how he continually tracks the results of his model using a benchmarking method–based on limited loss ratios–that allows him to see the impact of his underwriting plan within a six-month timeframe.