Gruber Cherry-Picks the Evidence for RomneyCare
MIT economist Jonathan Gruber touts the Massachusetts health plan, which he helped create, in a paper for the National Bureau of Economic Research (the Washington Post provides a summary). So how objective can you be when you are grading your own work? Apparently not very. In reviewing the literature, Gruber carefully cherry-picked positive data, while ignoring everything that didn’t fit the story he apparently wants to tell:
- Gruber relies on an article in Health Affairs showing a slight increase in primary care visits, but neglects to inform readers that there had been no significant change in specialty care visits. And while using the study to imply that access to care has improved as a result of the reform, he neglects to mention that there has been no apparent change in self-reported unmet needs. Remarkably, one-third of adults within 300% of the federal poverty level report that they were unable to meet a health care need within the past 12 months for “any reason.”
- He also fails to inform the reader that the same Health Affairs article shows no change in overall emergency room (ER) visits, but a statistically significant increase in ER traffic among those within 300% of the federal poverty level! This is consistent with a survey of 11 Massachusetts-area hospitals that found ER use rose four percent. Instead, Gruber cites an unpublished whitepaper claiming a modest reduction in ER use.
- Gruber cites a Massachusetts Medical Society survey finding that average wait times to see physicians were basically unchanged. But he doesn’t say how long those wait times are: New patients must wait from a month to six weeks to see a family doctor or an internist. Make that two months in Boston for a family practitioner. Gruber also finds it too uninteresting to mention that the same survey found about half of all family doctors and internists won’t see new patients or accept the insurance provided in the Commonwealth Connector. This was up sharply from 2006.
Austin Frakt (who prides himself on faithfulness to the evidence) republished Gruber’s one-sided account without asking a single skeptical question. David Henderson was more critical, pointing out that Gruber was a paid consultant to the White House to help develop ObamaCare.
You call this scholarship? I’m surprised the NBER published it.
Good job, Devon. That was a crappy report.
I find it somewhat amusing how different people can read the same reports and arrive at different conclusions depending on whether they view the glass as being half empty or half full.
I thought Gruber was suppoed to be a scholar. This sounds more like an apology.
The Mass. Med. Society survey found waiting times had changed little since implementation (waiting times have been long for years). But physicians increasingly won’t accept new patients or accept the type of insurance people get from the Commonwealth Connector. This suggests access to care by patients covered by public coverage or Commonwealth plans will likely suffer in the future.
Very sneaky. Very intellectually dishonest.
I think Austin Frakt looks to the evidence until he verifies his own world view and then he quits looking. Look at the difference in how he responded to John Goodman and how he responded to Jon Gruber.
One of the major limitations of the article by Long & Masi is that it does not count the costs to society of the very small improvements in access reported in their research.
I refer primarily to the tax hikes and government spending to achieve these outcomes. They are not “free”, although that is the unstated assumption in too many Health Affairs articles.
Plus, I’m not sure I’m comfortable with using statistically significant cross-sectional estimates (of access) to infer statistically significant differences in longitudinal estimates (although it happens all the time in the health-econ literature) over just two periods.
It seems to me that if you deliver the same exogenous shock (i.e. the 2006 Romneycare law) to the cross-section, whether it’s 50 people or 5,000 people or 50,000 people, and you observe the effect over only two time periods (2007 and 2008), you don’t have enough longitudunal obervations to catch mean reversion.
When we look at Long & Masi’s figures for the low-income (<300%) FPL sample, it looks like the output variables are beginning to revert to the mean in the second period.
Am I wrong?