When regulators from around the world—including the U.S. and the E.U.—determined in 2008 that the banking system was terribly under-regulated, they all suggested a similar fix: Separate the departments involved in investment banking from those that deal with more run-of-the-mill insured deposit-taking. But recently there has been some back-sliding on such separation, and finance is still dominated by giant banks that mix the two.

Yet separation makes a lot of sense. Why should to two such wildly different activities as investment banking and retail banking be bundled together in the same institution? It’s obvious what JPMorgan gains: a nice piece of ballast to offset embarrassing, high-stakes trading mishaps. But what about consumers? How does an ordinary person benefit from putting her deposit into a complex global bank? I have asked that question to experts for more than a decade and have yet to hear a satisfying answer.

Breaking up the banks in this way isn’t a proposal confined to the political fringes. Neel Kashkari, the president of  the Federal Reserve Bank of Minneapolis—as well as a former banker and a Treasury official during the brunt of the crisis—argued in a recent speech that the too-big-to-fail problem has not been solved by current regulation: The failure of more than one bank in an economic downturn would still have enormous economic costs. His suggested options include breaking up large banks, turning large banks into public utilities, and taxing leverage to reduce systemic risk. He gave short shrift to those who argue that remedies like this could be disruptive, saying, “Those potential shortcomings must be weighed against the actual risks and costs that we know exist today…Better and safer are reasons enough to act.”

The strongest argument for isolating banks’ varied services is more a matter of culture. Investment banking is a volatile, exciting, high-risk, high-reward business, and should be cordoned off from the arm that deals with consumers. The ideal structure for investment banking is a partnership in which the traders’ compensation is a function of their performance, so that they share in their firm’s gains and losses.

Here’s how it could work: A too-big-to-fail, too-complex-to-manage bank such as JPMorgan Chase should be split into three parts. The investment bank could be spun off entirely. JPMorgan’s creative investment bankers would relish the chance to turn the franchise into a partnership, with the freedom to pay themselves what they please. The remaining bank would be split into a wholesale bank, for large corporate clients, and a retail bank, the only taker of insured deposits.

Regulating these three types of entities would be straightforward. The investment bank would be regulated as lightly as a hedge fund, but would not have access to financing that uses taxpayer money. The wholesale bank would provide only simple banking products, including foreign exchange and interest-rate hedges. The retail bank would be limited to consumer banking and small-business and mortgage lending, with no scope for high leverage or the big-scale securitization that sank otherwise simple banks, such as Washington Mutual, in the crisis. These two last classes of banks would be boring but safe. A housing bubble, if it came, would not be centered on the insured banking system, and so would be potentially less damaging to the taxpayer.

Just as important as a structural change is the need to eliminate the culture of bonuses at consumer-facing banks. Bonus-pool culture migrated to big commercial banks when they set up or bought investment banks in the 1980s. But, with no unlimited partners to share both gains and losses, the incentives were skewed: The bonus pool would skim off 40 to 50 percent of revenues up-front while any losses hit only shareholders and taxpayers. The latest generation of bankers, then, are not exceptionally bad people: They simply pursued, sometimes honestly, sometimes dishonestly, a set of perverse incentives that were tolerated and even encouraged by regulators.

Attempts by the European Union to cap bonuses at 100 percent of a banker’s base salary have simply encouraged banks to find new loopholes. In the U.S., for instance, a mechanism to curb top bankers’ pay at $500,000 in 2009 failed because some banks were exempted from it because they paid back some of their bailout funds quickly, while others claimed “exceptional circumstances” allowed them to exceed the cap. In my view, when banks have access to central-bank funding there should be a legal limit to what they can pay their employees ($500,000 a year wouldn’t be unreasonable). Those wanting more would be free to join a hedge fund or an investment-banking partnership, where their talents would be better appreciated and rewarded.

The failure to fix these things now is making for trouble ahead. Simply demanding higher capital ratios at banks and better behavior from bankers sadly has not put the industry on the path to serving the economy better. Moreover, cases continue to surface of cartel-like, collusive behavior by global banks: in pricing interest rates, foreign exchange, and, most recently, interest-rate swaps. A class action recently filed in New York alleges that 10 large banks have bullied clients into trading interest-rate swaps on their common trading platform. Time and again the interest of the client has been subordinated to that of the bank, or worse still, a bunch of banks.

Banks are still big, they are still outwitting their regulators, and their business model is bust. It is hard to fathom why regulators would not favor a system that improves on this. I believe that only the fear of change and the power of the bank lobby are keeping decision makers from overturning the status quo.