Bancassurance, the combination of banking and insurance in the same corporation, is an established strategic option in Europe. In the United States, the lifting of regulatory barriers-led if not inspired by Citigroup-is making bancassurance a possibility.

But a corporate merger is only a first step to merging products or operations. The extent of integration depends on three things: product features, workflow characteristics, and technology. Only where all three are favorably aligned can bancassurance really happen.

Most bankers have limited experience with insurance, and vice versa. And when the barriers come down, bankers will find themselves in a world very different from what they had expected.

Most life insurance products today are investment vehicles, with term life insurance playing an important role. All the growth is in variable, universal, or variable-universal life, with some $450 billion worth sold in 1997.

Banks already participate in the annuity business, controlling about 22% of new premiums written. Bankers also dominate the smaller credit life niche. And large percentages of big banks are now marketing life insurance and investment products.

The banker's easy familiarity with individual life insurance diminishes with group life or disability insurance, and vanishes with health insurance. The latter has transmogrified into managed care, bearing little resemblance to financial services.

Property and casualty product features are also unlike those of bank products. The insurance products are focused on accident risk (not credit risk) and on property characteristics and costs. Fully half of the business is liability, reflecting our infamous U.S. tort system.

In commercial lines, the product revolves around measuring, controlling, and pricing risk intelligently. To reduce excess risk, the entire industry uses reinsurance as a sort of lender of last resort.

Bankers avoid risk; insurers embrace it. The cultures are very different. A commercial lender could thrive and never make a bad loank, but an insurance product that never had a claim wouldn't have much of a future.

Insurers' workflows are different from banks'. Insurance is an infrequent, complex purchase, and independent agents play a large part. There only banking parallels to the expensive independent agent are specialists like mortgage brokers or financial planners.

Underwriting is the key skill in the insurance industry. Margins on mass-market products are low. In fact, most profits come from the investment income, not the underwriting per se. But such insurers as AIG or Progressive have found that higher-risk categories offer distinctive business strategies.

And no property/casualty player can avoid risk. In 1992, Hurricane Andrew cost a staggering $15 billion, making the entire property/casualty industry unprofitable that year. Recent construction and beachfront property appreciation had not been reflected in exposure analyses.

(As a result, insurers began issuing catastrophe bonds, known as Cats, in the U.S. private debt markets. Cats require the quantification of natural risk with catastrophe models, and incorporation of these risks in a financial contract. Sophisticated modelers might use data about individual properties, such as distance from a fault, shoreline, or other hazard.)

Another simple example of underwriting workflow differences is the use of pictures to document either pre-existing or damage conditions. About 50% of P&C files have a picture today. Digital cameras mean that claims adjusters or underwriters can feed the picture right into a PC and then into the permanent automated files.

Policy processing is the backbone of the insurance workflow. All insurance relationships exist as policies that describe terms, conditions, coverages, exclusions, deductibles, riders, coverage dates, and numerous other factors.

Policies are not payment accounts. Transactions are fewer but far more complex, and usually contain more data. Banks allow withdrawals or redemptions more or less at will, but "withdrawing" from an insurance policy requires a proof of need-for example, a claim.

In the case of death benefits or investment products, this is pretty straightforward. But for P&C, it can be very complicated, laden with legal ramifications, and highly sensitive to trade-offs between claims payouts and the cost of processing claims.

It is not for nothing that insurance companies employ more lawyers than any other industry. An estimated 60 million to 70 million P&C claims were filed last year. Each requires customer contact, data entry, and individual processing.

And the diversity is astonishing. Each may be a very high-dollar risk, especially with bodily injury, but these risks cannot be known when the workflow begins. Auto damage claims are the most standardized, but these are only 20% to 30% of all claims. Some claims, such as tobacco litigation, can be in court for decades.

Asset management is a familiar function. Insurers manage about $2.6 trillion of assets, primarily corporate bonds, common stocks, government bonds, and commercial real estate mortgages, in that order. There is also $100 billion of policy loans.

Insurers are the third-largest institutional source of investment capital. Bankers know the functions of reviewing market data, analyzing portfolio strategy, measuring performance, and managing market and liquidity risk, especially those from the trust department.

Management information systems are the last step in insurers' workflows. To understand risk, these systems have always been focused on tabulating and reporting losses, catastrophic events, and actuarial tables. The net result is a workflow very different from most banks' financial reporting.

Banks' and insurers' technology have more in common than either the products or the workflows. The hardware, from mainframes to PCs, is exactly the same. Systems software from the mainframe to the desktop level is the same. The issues from Y2K to Java and data mining are equivalent.

Even the costs are somewhat similar. U.S. insurance companies and agencies will spend about $28.8 billion this year on all forms of information technology. Some $11.4 billion will be in the P&C business, and the rest in life insurance. Overall growth is about 3.9%, faster for desktop and slower for mainframe. The divisions between hardware, software, services, and internal staff are roughly the same as at banks.

However, most applications are different. Agency automation systems replace branch automation-but of course don't cash checks. Policy processing systems replace deposit account processing systems. Policy production systems are needed in place of consumer loan documentation systems. Claims processing systems constitute much of the back-office, instead of payment processing systems. Underwriting and actuarial systems do that function instead of credit evaluation or scoring systems.

Investment management systems would be familiar. Insurers such as USAA that already offer banking products are on familiar ground. But even a staple like the general ledger has some differences, because insurers must produce a unique third set of books, called statutory reporting. Another unique characteristic are systems that file ratings books with every state, because of the absence of federal regulation.

Some of the industries' application architecture is also different. Insurers, like banks, once wrote many proprietary systems but have since switched to buying or modifying packages wherever possible. However, there is no direct insurance equivalent to the high-volume service bureaus, such as First Data or Alltel Mortgage.

Another key difference is the methods for electronic data transfers between entities. Payments and financial clearing have always needed extremely high-volume, totally reliable exchange between literally tens of thousands of players in a cooperative manner.

But with insurance, the essential element of data transfer is the claim, an inherently antagonistic situation. The two major insurance-specific electronic utilities-Acord and Ivans-are both focused on the needs of agents, and essentially replicate the functions of a large wide-area network. There are also some equivalents to credit bureaus, such as driver registration history data bases and auto parts cost data bases.

Most of the trends and issues in software applications are similar. The Internet is a puzzle and opportunity. Web use to get quotes, apply for coverage, conduct customer service, or file claims is inchoate. E-insurance will lag e-banking and especially e-brokerage. Web sites such as InsureMarket or InsWeb are there but have a long way to go. Even large players have not yet got functioning Internet claim capability.

So bancassurance advocates have to walk a careful road. The quick pace of technological change does not allow filing claims through an ATM or selling extended care coverage at the teller line. Bankers need to avoid the rush to put even more capacity into an already overextended industry that has historically been brutal on inexperienced underwriters. Understanding data centers and programming tools will not help with a long application backlog specific to the insurance industry.

In our rush to bancassurance, it may be well to remember that those who forget history are condemned to repeat it. There are deep historical reasons for the uniqueness of the insurance industry, and only those totally dedicated to the business have survived.

Those who forget this lesson may regret it. Mr. Teixeira is the president of Tower Group, a financial technology research firm in Needham, Mass.

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.