The Capital One acquisition of ING Direct USA is, as far as we know, the first time size alone has been made an M&A approval criterion for the Federal Reserve.

Fans of breaking up big banks think denying this deal will strike a blow against systemic risk. But, scotching this transaction solely on size would worsen systemic risk. Prudential or CRA criteria might militate against this transaction, but sheer size surely shouldn't.

Like it or not, banks are big, which means that even mid-sized ones have tens of billions invested in branch networks, product offerings and one or another subsidiary. In banking, as in cell phones, autos, and lots of other American businesses, markets have differentiated into lots of little specialty firms and a few very, very big ones.

Is this "good?" That's a subjective, if not a theological question. Asking if there should be colossal banks is a lot like asking if there should be old age. No, but what's the alternative? One might not wish to be eighty, but when you are 79, it doesn't seem all that bad anymore.

For all the rhetoric about breaking up big banks, specific how-to proposals with clear alternatives are hard to find. Those we've read often base their projected benefits on a heavenly vista of competing banks with limited reach, but they are at best sparse on the details of how to reach the blessed state.

The CapOne/ING transaction shows just how hard it is to take the abstract notion of bad big banks into the real world. Recall that ING's on-line book is on the block because its home-country regulator told it to shed $80 billion in U.S. deposits. Should the Fed say no and, thus, force a weak Dutch bank to take one on the chin? Demand that someone else pony up the $9 Billion CapOne offered?

And, who might have the pocket change — not to mention the managerial capacity and operational resilience to take on so large a deposit base with so many direct and indirect risks?

If a bank can't divest a business it no longer wants or regulators tell it to shed, then banks have no choice but to liquidate the business — cost to the bank, customers or even the health of the financial system notwithstanding.

We have a lot of large banks with big subsidiaries and investments that can’t be made to go away by fiat. If big businesses can't be acquired, they must die. If the cost of preventing one bank from getting bigger is to demand that another take so large a loss that its solvency is undermined, then that cost is too high.

Blocking transactions like CapOne/ING would send a clear distress warning to the entire global banking industry: if you do business in the U.S. and want to change the model through a sizeable divestiture whether for strategic or regulatory reasons, forget about it. Unless you can find a way to break the business into bite-size pieces, you have to eat it because you can't sell it. That, of course, exacerbates pain for troubled banks. Living wills will need to be even more of a death sentence because selling under stress will be made all the more difficult. And, for good measure, the simple-size criterion will take a chunk out of the industry's market capitalization because big banks can't run themselves like businesses any more.

Advocates of breaking up big banks wish this weren't true, but it is. We can revisit the prior FRB approvals of acquisitions like Countrywide — oops — or all the other, less-disastrous ones that gobbled up the mid-sized fry and wish it weren't so. But, it is. Drawing an arbitrary line in the sand and deciding that one bank getting bigger is a step too far is a crude approach to controlling systemic risk.

So, how to address big-bank risk? The United Kingdom is planning on one approach: segregating big banks into component parts so that retail operations are "ring-fenced" from wholesale-oriented ones. This is activity-based systemic regulation with no prior track record and little evidence of success.

To date, the biggest U.S. failures were ring-fenced – Washington Mutual was a true retail bank; indeed, it was essentially a monoline mortgage bank, as was Countrywide. Conversely, Bear Stearns and Lehman were ring-fenced into investment-banking activities. Neither of these segregated business models worked out so hot, not only since all of them were grievously mismanaged, but also because none had the risk insulation of diversification to fall back upon under stress.

Does this mean that all big banks should be allowed to buy what they want whenever another behemoth puts a business on the block? Of course not. But, size is not the only transaction criterion and, indeed, it's a poor one if systemic risk is, as it should be, a criterion for approval.

Meaningful risk criteria are resolveability — can an acquirer in its entirety absorb added risk without undermining its operational resilience or its solvency? Can the acquirer handle the managerial challenges of a new business line or an added investment in an existing one? How does the new book relate to the board-dictated risk tolerance.

All too often, supervisors haven't demanded that boards set these tolerances, meaning that a bank is basically asking the supervisors to trust it when a big deal is on the table. Supervisors of course should do nothing of the sort — they should demand that banks have a robust risk-management infrastructure well in place far in advance of a transaction so that, when one is in the offing, supervisors know that the bank’s best guess isn't a bet with the taxpayer's chips.

Finally, it's past time to look at the criteria the Federal Reserve uses to judge concentration in the banking system. The methodology for bank deals is anachronistic and, as the prior M&A history outlined above makes clear, missed a lot. The FRB should defer to antitrust law for sophisticated judgments about market impact — the Justice Department's decision to block AT&T's bid for T-Mobile makes clear that we still have antitrust laws. They, not arbitrary size criteria, should be the determining factor for bank M&A when safety and soundness is assured.

Karen Shaw Petrou is a managing partner at Federal Financial Analytics Inc.