As imitation has replaced imagination, the banking industry has followed fad after fad for at least the last 15 years, with disastrous results.

Let's have a look at the most pervasive fads, the lessons from each, and what bankers are falling for now.

Deposits don't matter. Wholesale funding is cheaper. Why, as John Reed said, should we worry about the little old lady in Queens when we can borrow in the market?

Lesson: The real value of a bank is in its core deposits. Deposits come with customers who build a banking business. Wholesale funding is an opiate that feels good, is unsustainable, and leaves just when you need it.

The current crisis has again revealed the value of deposits and the risk of wholesale funding. As many mortgage companies self-destructed, Countrywide Financial just managed to survive, because it had some deposit base.

The death of the branch. If deposits don't matter, then customers don't matter. Let's push the customer to electronic delivery and eliminate the high-cost branch.

Lesson: The branch is the cornerstone of any bank, not the millstone. Banks with value-added branch programs prospered by building customers, while almost every major bank that tried to cost-save its way to prosperity lost major market share. The branch has returned with a vengeance.

The 80/20 rule. As applied to banking, the wealthy 20% of your customers produce 80% of your profits, while the rest produce losses. Every customer should be put in the profitable group or thrown out.

Lesson: This is just another excuse for cutting costs. No one can realistically manage customer profitability, because the customer profile is constantly changing.

Merge your way to prosperity. Every major bank merger and acquisition has been accompanied by promises of better customer experience, dramatic cost savings, and fantastic shareholder returns.

Lesson: These major deals usually fail to produce meaningful benefits. Travelers/Citicorp, Wachovia/Golden West, and many others have destroyed service, eliminated jobs, and/or destroyed massive amounts of shareholder value.

Supermarkets. The customer wants a financial supermarket to consolidate his financial assets.

Lesson: Efforts to create these supermarkets are a failure in almost every instance, but the concept is still used as an excuse for overpriced acquisitions.

Data warehousing. How many billions of dollars have been wasted by bankers who believed the consultants who said data warehouses would serve the customer better by retaining and messaging more personal information to provide more cross-selling opportunities?

Lesson: This is a variation of the financial supermarket theory. Is there any proof that data warehousing has produced a profit?

Subprime lending. Let's lend to customers who have a proven ability not to pay. This began as a government-ordered welfare program under the pretense of lending to the underserved. Banks faced scrutiny and orders to serve the subprime market, regardless of credit and risk. If your losses were too low, you needed to take more risk.

Lesson: What a surprise. People with a demonstrated ability not to repay don't repay, 120% LTV loans perform poorly, and people have poor credit records for a reason.

Structured finance. Wall Street can engineer financial instruments that defy the laws of economic gravity. Subprime becomes triple-A, leverage has no limits, off-balance-sheet structures are really independent, and computer geeks, not bankers, control the credit flow.

Lesson: As I have often noted, lending is lending. The old rules of credit and "know your customer" will always apply.

What are the common themes? Replacing banking with financial engineering; divorcing management from basic business values; treating the customer as a number, not an individual; believing in consultants; and responding to a regulatory structure that is always fighting the last war.

The next fad is clearly risk management, the catch phrase for regulators, investors, and managers who believe that computer models can predict and control risk, and that ever-increasing controls can reduce risk.

The credit crisis is another excuse to spend more money and energy on risk management plans. The theory is with improved computer models and additional oversight, management teams will have the ability to predict and eliminate risk.

Basel II is the extreme example of computer modeling and risk management. Operational risk cannot be quantified, and economic risk cannot be predicted. Yet somehow these models are supposed to do that?

Look at how well advanced computer modeling has performed at Citigroup, Merrill Lynch, UBS, and all the rest. AIG, the self-acknowledged master of risk management, has taken $20 billion of losses with more to come. And oops, the Moody's rating model has a glitch: What was rated Aaa should really be four notches lower … sorry.

Yes, we need risk management. Yes, models can help. But like everything else in the financial services, what matters is intelligent, experienced judgment, occasionally surprised by reality.

One thing is for sure — the more the government focuses on something, the more we know it's wrong.

What do you think? Let me know!

(This article originally appeard in American Banker.)