New Banks?

The Obama administration is due to announce tomorrow what it will do with the second half of the the $700 bn financial sector rescue pot (a.k.a. the TARP). Rather than continuing to put more money into our troubled banks, Paul Romer and Willem Buiter argue for starting new ones. Romer says:

The government has $350 billion in Troubled Asset Relief Program (TARP) funds that it can use to encourage new bank lending. If this money is directed to newly created good banks with pristine balance sheets, it could support $3.5 trillion in new lending with a modest 9-to-1 leverage. Right out of the gate, the newly created banks could do what the Fed has already been doing -- buying pools of loans originated by existing banks that meet high underwriting standards.

And:

The brewing backlash against the existing players from the financial sector is almost certain to burn hotter as the recession wears on, and new election campaigns get underway. If the new administration ties its fate to the existing players, it could lose its room to maneuver on countercyclical policy and be put under political pressure to intervene in bank decisions in ever more intrusive ways.

Because they can and will borrow, new banks will be much more effective in leveraging TARP funds. They will undertake more total lending, bring more trading to financial markets, and do more to limit the depth of the recession. As a result, investing the TARP funds in new banks will do more to help the troubled but potentially viable existing banks than giving funds directly to them.

Banks that are not viable, the ones with liabilities that substantially exceed their assets, will lobby vociferously against a return to historical patterns of bank regulation. They will say anything to postpone a looming FDIC takeover. The administration should not listen to threats and pleas from these doomed banks. It does not have to rely on them to get new lending going quickly and on a large scale. New entrants could give us a few good banks. That, plus an FDIC that can do its job, is all we need.

Update: A counterargument from Felix Salmon.