The following op-ed, “Banks need fewer carrots and more sticks,” was co-written by Harvard Law School Professor Hal Scott; R. Glen Hubbard, dean of Columbia Business School; and Luigi Zingales, professor at the Chicago Booth School of Business. It appeared in the May 6, 2009, edition of the Wall Street Journal.

The results of bank stress tests — expected tomorrow — will no doubt prompt calls for further government guarantees and capital injections. But continuing to prop up the banks with government cash is a mistake. There is a better approach.

A well-capitalized banking sector is a necessary ingredient for effective intermediation and economic recovery. But today’s system is not well-capitalized. How can we move in the right direction?

In a market economy, the government can create the right incentives by using a combination of carrots and sticks. Thus far, the government has only used carrots with the banks. One major carrot is the Troubled Asset Relief Program (TARP). The initial infusions were very generous — the Treasury got back securities worth $78 billion less than the $254 billion it invested — as the Congressional Oversight Panel pointed out recently. In addition, the FDIC’s guarantee of short-term debt was worth $100 billion just for the original nine TARP-participating banks. And the mortgage-related asset guarantees offered to Citibank and Bank of America were worth tens of billions of dollars more.

A new round of expensive TARP injections — by converting the government’s preferred stock into equity — may follow the release of the stress-test results. In addition, the Treasury’s Public-Private Investment Program (PPIP) plans to subsidize the purchase of banks’ “toxic assets” by hedge funds and other investors. We estimate that the government will spend $2 for every $1 the private sector will put in. Yet even with this large subsidy, PPIP’s chance for success is low because of the substantial gulf between the bid and ask prices on the toxic assets, and the reluctance of investors to partner with the government.

Not only is the carrot approach not jump-starting lending, it is also angering the American people. It’s hard to justify to taxpayers that we need to reward the same group of people who, rightly or wrongly, are perceived as responsible for the current situation.

It’s time for government to use the stick, beginning with creditors. The first step should be an announcement that the FDIC guarantees of short-term debt, set to expire at the end of October, will not be renewed. Insolvent banks — defined not by stress tests, but as those that cannot fund themselves in the private market — will be taken over by the FDIC. Of course, this takeover plan must be clear and credible. Otherwise creditors will play “chicken” with the government, knowing that at the last minute the government will flinch and fail to remove the guarantees.

Despite the clarity of such an approach, the market might be skeptical for several reasons. First of all, the FDIC lacks the staff to oversee, let alone run, several large and complex banks which may become insolvent. Second, the FDIC’s main approach so far, as with Washington Mutual and IndyMac, has been to restructure the banks for acquisition. The trouble with this plan is that it is unclear who will buy the largest banks in the near future. Finally, it is politically unappealing to have a government institution run a significant fraction of our banking sector. Waiving the specter of nationalization, the creditors may try to force the government to bail them out.

We believe these problems can largely be avoided by adopting a simple approach. Rather than taking over and running banks, the FDIC should split each bank into two parts. One part (“the bad bank”) will assume all the residential and commercial real-estate loans and securitized mortgages as assets, and all the long-term debt as liabilities. In addition, “the bad bank” will obtain a loan from the “good bank.” This loan is necessary because the long-term debt of the old bank is not likely to be sufficient to fund the assets of the bad bank. The good bank will have all the remaining assets, including derivative contracts and its loan to the bad bank. It will have all the insured deposits and the FDIC-guaranteed short-term debt as liabilities. Once the split is accomplished, the good bank can be cut loose from FDIC receivership.

On the one hand, this split separates the toxic assets, whose value is very uncertain, in an institution that has no insured or guaranteed liabilities and poses no systemic risk. The bad bank will be like a closed-end mutual fund and can be run as such. The good bank will be well-capitalized, and the value of its assets will be clear.

The losers in this reshuffling are the long-term debtholders who get stuck in the bad bank. For this reason, we propose that they be compensated by receiving all the equity of the good bank. The old shareholders will get the equity in the bad bank. (In any restructuring, bondholders should do better than equity.) And the FDIC minimizes its risk because it guarantees the deposits in the good bank.

In fact, long-term debtholders who have debt claims against the bad bank and equity claims against the good bank will be better off under this plan than if the bank were liquidated or continued to operate as one bank. If the bank were liquidated, bondholders would stand to lose almost all their investment. If the bank continues to operate with government subsidies, the benefit of the subsidies are shared by both debt and equity. Under our plan, the debtholders will get all of the equity in the “good bank” and therefore all the upside of its future performance.

One of the major objections to letting banks fail is the argument that they are not really insolvent; they are just facing a temporary dislocation in the marketplace. But if this observation were true, the bad bank would surge in value, and the old shareholders of the banks, who received the shares in the bad bank, would gain. If it is false, the bad bank would default and the old shareholders would receive nothing (as they should).

In order for this plan to work, legislation would need to take effect before the withdrawal of the FDIC guarantee in October, so that FDIC procedures for handling failed banks can be applied to bank-holding companies. FDIC Chairman Sheila Bair has called for such legislation. Most importantly, this plan won’t impose any new cost on the taxpayer.

Bold stress tests and government intervention reflect President Obama’s use of Franklin Delano Roosevelt as a model in dealing with the current crisis. But he got the wrong Roosevelt. He should instead follow the motto of Theodore Roosevelt: Speak softly and carry a big stick.

Mr. Hubbard, dean and professor of finance and economics at Columbia Business School, was chairman of the Council of Economic Advisers under President George W. Bush. Mr. Scott is a professor of international financial systems at Harvard Law School. Mr. Zingales is professor of entrepreneurship and finance at the Chicago Booth School of Business.