After Sen. Kerry failed (so far) in the Finance Committee on his amendment to allow the insurance exchange administrator to negotiate with insurers, the usual health wonks have been busy explaining the importance of the proposed health insurance exchanges. It’s been helpful to get some attention on how they should be designed to avoid failure. But is it enough?

The common themes, nicely summarized by Mark Thoma, tell us that to be successful in lowering insurance premiums, the health exchange must achieve sufficient scale to be effective and that the exchange administrator have the authority/duty to demand/negotiate for consistent quality and reasonable prices.

A problem with these analyses is that they ascribe to the exchange itself results that are driven by other factors. An exchange doesn’t create competition; it creates a place where whatever competition that exists can take place. If the market is highly concentrated and dominated by market power and near-exclusive agreements between mega-insurers and oligopoly provider networks and hospitals, then that’s the market you’ll get access to in the exchange. There is nothing remotely resembling the model of efficient market competition and pricing in that picture.

What changes the picture is not the exchange itself but the elements that create market power. Individuals and small businesses seeking insurance have no market power, but if they band together, they can start to acquire it. But the exchange doesn’t do that; the mandates to purchase insurance and restrictions on escaping that mandate do that. If you suddenly become part of a group of 10 million people who are required to purchase insurance, then the pool of 10 million people could potentially have market power if it organized and a central entity bargained on behalf of the entire pool. This would be true even if there were no entity or convenient web site called "the exchange."

So despite all the rhetoric about how exchanges can create or facilitate workable competition, the key is really about creating market power through a mandated pool and using that market power to bargain for lower prices against entities that also have market power and will, unless checked, use theirs to keep provider and insurer prices higher.

That point is driven home by this excellent op-ed, A Texas-Sized Health Care Failure, by Cappy McGarr, former chairman of the Texas Insurance Purchasing Alliance, a state-created pool intended to help small businesses purchase cheaper health insurance. It failed in 1998, but why?

Nevertheless, six years after the program got off the ground, it folded. Many factors contributed to our failure. Some elements of the program, like the restriction it put on the size of eligible companies (only employers with 50 or fewer employees could join), proved unpopular. In addition, the governor who helped create the alliance, Ann Richards, was replaced in 1995 by George W. Bush, who did not consider it a priority.

I think McGarr gets the reasons right:

Most important, though, our exchange failed not because it wasn’t needed, and not because the concept wasn’t sound, but because it never attained a large enough market share to exert significant clout in the Texas insurance market. Private insurance companies, which could offer small-business policies both inside and outside the exchange, cherry-picked relentlessly, signing up all the small businesses with generally healthy employees and offloading the bad risks — companies with older or sicker employees — onto the exchange. For the insurance companies, this made business sense. But as a result, our exchange was overwhelmed with people who had high health care costs, and too few healthy people to share the risk. The premiums we offered rose significantly. Insurance on the exchange was no longer a bargain, and employers began backing away. Insurance companies, too, began leaving the alliance.

Texas wasn’t the only state to see its insurance exchange fail. Florida and North Carolina were also unsuccessful. And California, which had the first exchange (established in 1992) and the largest market, shut its doors in 2006. All these state exchanges failed for the same reason: cherry-picking by insurers outside the exchange.

What do Texas’ (and other states’) experiences tell us? It’s the mandated pool, stupid. To create and sustain market power for buyers, you have to get everyone in, require that they get their insurance through the pool, and don’t allow insurers to "cherry pick" and pull out of the pool the good risks and leave the pool with the bad risks.

It also helps to be dealing with only a few plans and require the plans be uniform; multiple products create choice, but they also dilute market power with respect to each choice.

John Kerry’s amendment to allow the exchange administrators to "bargain" for better rates from the participating insurers is fine, but that doesn’t create the negotiating market power in Massachusetts’ Connector, its version of the exchange. It’s the mandate and the restrictions on opting out of the pool that limit the cherry picking and keep the pool’s market power together.

Interestingly, these lessons are reflected in the Senate Finance bill, in the form of Senator Cantwell’s amendment to allow states to create the equivalent of a state purchasing pool, allow the state to negotiate with providers (it’s a public plan!) and/or private insurers (no it’s not!), and then use the federal subsidies eligible folks would otherwise receive in the exchange to apply towards the costs of pool insurance for the entire eligible group.

A critical piece of that approach is that Cantwell’s pool would take the entire population of the eligible group — those between 133 percent and 200 percent of federal poverty level — and keep them together: it doesn’t allow private insurers to pick off low-risk customers and remove them from the pool. They’re all in — sorta like Medicare without Medicare Advantage.

Cantwell’s amendment is, of course, optional for states and limited to that income group (though it’s an important group in capturing a good percentage of the uninsured). The exact features and implementation would be important. But what the amendment is doing is different in kind from how we’ve been thinking about the exchange (and the public option), and it draws the right lessons from the failed Texas model. Keep an eye on it.