You are using an outdated browser.
Please upgrade your browser
and improve your visit to our site.
Skip Navigation

McKinsey Insider: Survey ‘Not a Good Tool for Prediction’

Republicans are still citing that controversial McKinsey & Company report forecasting disruptive aftershocks from health care reform. But that forecast looks less and less credible, with even some McKinsey insiders casting serious doubt upon it.

The report, which McKinsey released two weeks ago, makes an arresting and politically explosive claim: Between one-third and one-half of employers are likely to drop or significantly alter existing coverage in 2014, once the Affordable Care Act is fully operational and employees can buy insurance through the new insurance exchanges. Architects of the law, including President Obama, have long claimed the law won’t force people with insurance to give up their coverage. Among other things, the law penalizes companies who don’t offer insurance, precisely to avoid this problem. The McKinsey report, based on a survey of 1200 employers, suggests that businesses will pay the penalty or find ways to circumvent it. In other words, they’ll drop insurance anyway.

The finding is at odds with projections from more respected authorities, including the Congressional Budget Office, the Rand Corporation, and the Urban Institute. But McKinsey has declined to divulge more details about the report, despite requests from the White House and congressional Democrats, while company spokesmen have declined to answer questions from reporters, including yours truly. Why the silence? One possibility is that the conclusion can’t withstand scrutiny.

Brian Beutler of TPM Media, who like the Washington Post’s Greg Sargent has advanced this story with his own extensive reporting, has already quoted multiple sources familiar with the survey questioning its controversial finding. One of them said, “this particular study wasn’t designed in a way that would allow it to be peer review published or cited academically.” Now McKinsey employees are telling TNR similar things, sometimes in even blunter terms. “Trust me,” says one of the firm’s insiders. “The survey is not a good tool for prediction.” 

Another senior firm employee defends the report’s “quality and rigor,” describing it as an “independent, professionally conducted, extensive survey.” But this employee also says that, notwithstanding the hype, the survey was not designed to produce the sort of reliable forecasts that CBO and other authorities make. “We are not making a point prediction or forecast about employer behavior after the implementation of health reform.”

The big mystery about the study is its methodology. As the report acknowledges, McKinsey “educated” the respondents about the Affordable Care Act before asking questions about how their firms were likely to respond to the law. For such a survey to yield even remotely reliable predictions of future behavior, the “education” has to be accurate. In other words, the survey has to portray the financial trade-offs as they will really exist.

In addition, the survey questions themselves must elicit responses in ways that don’t bias the answers. Otherwise, as Time’s Kate Pickert points out, “those conducting the survey may have primed respondents to say they would keep or drop coverage.” Keep in mind that a report predicting drastic coverage changes might encourage both insurers and employers to seek help from consultants who advise health care clients, as McKinsey does.

It’s important to distinguish this report from work from the type produced by the McKinsey Global Institute, a semi-autonomous think tank that operates within the company. The Institute’s 2008 report on the high cost of U.S. health care, for example, provides a superb comparison of advanced health systems that health care experts (including me) have cited frequently. But the Institute has a formal peer review process, with a board of advisory academic luminaries that includes Nobel-winning economists Kenneth Arrow and Robert Solow. The McKinsey report on employer responses to health reform never got that kind of vetting.

It’s also important to recognize that, methodology questions aside, McKinsey’s prediction might not be entirely wrong. If the new exchanges end up working really well, across all 50 states, then it’s possible some employers would start dropping coverage, even thought it would mean paying that penalty. I’m actually among those who might welcome the change, depending on the circumstances. But under the Affordable Care Act, the transition out of employer insurance would likely be slow, if it happened at all. In Massachusetts, enrollment in employer insurance actually increased after officials there enacted a similar coverage scheme.

The Massachusetts story may not be indicative of what will happen in the nation as a whole. And even the most rigorous studies by the most respected experts could be inaccurate. But surely that experience and those findings deserve more credibility than a report by consultants who may not have designed a survey consistent with academic standards, didn’t submit their work to formal outside scrutiny, and, by the way, had a financial motive for promoting the conclusion that they did. Critics of health reform citing the report as validation have no business doing so.