How Cover Oregon fits into my discussion of complexity

I’m part way through a series of posts about the politics of implementation. In When Things Break I presented the idea of one broken piece undermining a larger whole. The Broken ATM discussed this idea in the context of doing fieldwork in a rural municipality of the northeastern state of Piauí in Brazil. Then Wiring Government in Series or in Parallel summarized some of the literature making similar arguments to my thoughts about the more places a program can break, the more likely it is to do so. I still need to summarize some key criticisms of that literature, but in this post I want to write about how that perspective might help us understand some of the problems with Cover Oregon. In the most recent post in the series I briefly mentioned Josh Schultz’s Triple Aim: Cover Oregon vs. Washington Healthplanfinder, saying that it touched on this and looked excellent, but it deserved a careful read instead of a skimming. I’ve done that and want to hi-light a few key parts that tie in with this discussion.  (He also has a more recent piece Triple Aim: Washington Health Exchange’s Single Focus……Make It Work that continues the discussion).

This paired comparison of Oregon and Washington makes sense to me for many reasons. The states are both medium sized, in the same region, and had Democratic administrations. Of course no pairing is going to be perfect, but they’re similar enough to wonder why they saw such stark differences in outcomes. Josh gives five reasons:

Five areas contributed to Washington’s success: a clearly defined and realistic project scope, a private company serving as the exchange’s systems integrator, deliverable-based contracting, coordination between different state agencies involved, and scope control with a focus on feedback from the quality assurance vendor.

I try to avoid social science jargon in this blog, but I think the term “overdetermined” is useful here. This is when there were so many things causing an outcome that we could take any one of them away and still get the same outcome. In other words, I suspect that any four out of Josh’s five explanations would be enough to leave us with a seriously flawed Cover Oregon. This was true of my study of the initial implementation of Brazil’s Zero Hunger as well. In my view, there was no one thing that went wrong that was the key and would have reversed the outcome had it gone differently.

Josh argues that Washington state limited their exchange’s scope:

Washington’s Exchange Board knew there were many things the an exchange could do, but it also recognized the short time frame for implementation and kept their exchange simple: a place where non-elderly uninsured individuals can apply for health insurance and obtain tax credits and cost sharing assistance to afford that insurance. If a person is found eligible for Medicaid or the Children’s Health Insurance Program (CHIP), the exchange sends their information to Washington’s Medicaid agency, which enrolls that person in Medicaid or CHIP, and into a Medicaid or CHIP managed health plan. That’s what the Washington Healthplanfinder website does, and it does it well.

In contrast, Oregon allowed scope creep:

Oregon tried to build an exchange website that allowed people to shop for and compare health plans, apply for tax credits, and enroll in Medicaid or CHIP. Oregon also wanted its exchange to allow consumers to sign up for food stamps, access the Temporary Assistance for Needy Families (TANF) welfare program, and apply for Employment Related Daycare Services. In its Affordable Care Act Exchange Establishment grant application, Oregon explained that it wanted to create a website functionality that would, “enable the State to create a seamless environment for clients and consumers” to help them apply for state-related benefits and entitlements that weren’t related to Obamacare’s health insurance programs.

Someone who had read that classic implementation literature from the 70’s and 80’s  would predict that, all else being equal, the Oregon approach of bundling more things together would make Cover Oregon more likely to have problems than the Washington health plan. Of course, all else was not equal, and Josh points out multiple other important differences in his articles, all of which pointed in the same direction.

He gives some more specifics about how Washington achieved this focus on doing what had to be done immediately and doing it well:

Washington’s Exchange Board emphasized that there would be an Exchange Version 1.0 and a Version 2.0. The website released for the enrollment period beginning October 1, 2013 would contain only what was necessary for the project to work. Suggestions that the project’s scope be altered had go through the robust change control process. Requests to add or modify the scope of Washington’s Exchange website were logged, tracked, and submitted to Deloitte [the company hired by Washington to be its “systems integrator”] for evaluation and “pricing.” Deloitte would then send the change request back, and the request to add to or change the project’s scope was either approved or denied by a central decision maker at Washington’s Exchange. Washington’s QA [quality assurance] vendor, Blue Crane, periodically warned about risks associated with adding to the Washington Exchange’s scope — even when Washington was not causing the project’s scope to expand.