Do Markets Respond to Quality Information? Evidence from Fertility Clinics
The U.S. spends far more on health care than other developed countries - 16 percent of GDP in 2007 - yet health outcomes in the U.S. are no better than those in other countries. There is increasing interest among U.S. policy makers and health care researchers in finding methods to improve the quality and value of health care services. One means of improving quality and value is to structure payments to hospitals and doctors so as to reward them financially for providing better care. A second option is to simply make quality information available to consumers. If individuals or employers who buy health insurance on their behalf "vote with their feet" and select providers with higher quality ratings, it may be possible to achieve quality improvements without changing the structure of payments to providers.
Do consumers pay attention to quality ratings when selecting health care providers? This is the question examined by researchers M. Kate Bundorf, Natalie Chun, Gopi Shah Goda, and Daniel Kessler in their new study "Do Markets Respond to Quality Information? The Case of Fertility Clinics," (NBER Working Paper 13888). Past evidence on this question is mixed, with some studies finding that more highly rated providers have higher market shares and others finding a more marginal impact of quality information on markets for health services. The authors present new evidence on this question in the context of the market for assisted reproductive therapy (ART) services.
The authors examine the effect of a federal law that requires the Centers for Disease Control (CDC) to collect and disseminate information on the success rates of all fertility clinics in the U.S. The key quality measure is the percent of treatment cycles that resulted in a live birth ("birth rate"). In the data reported publicly, this measure is lagged by three years - so for example, in 2008 the CDC reports the outcome of cycles started in 2005. The authors have data on the 3-year birth rate from 1995 to 2003, although it is available publicly starting only in 1998. Their basic empirical strategy is to compare the effect of this measure on fertility clinics' market share before and after the start of public reporting. If consumers respond to quality information, the 3-year birth rate will have a stronger positive effect on market share after the start of public reporting.
One important feature of the authors' data is that they also have data on the birth rate lagged by 1 year, a measure that is not made public by the CDC. This measure is likely to be correlated with quality information the consumer has that is unobserved by the researcher - for example, a patient's doctor may tell her that a highly regarded physician has recently joined a particular clinic and the clinic had high rates of live births as a result. With both the 1-year and 3-year birth rates included in the model, the authors can be fairly certain that any effect of the 3-year birth rate on market share is a causal effect of the reporting of quality information, since there is no other reason for the less current information to affect market share. The ability to control for unobserved quality information is an important innovation of this study relative to the previous literature.
Turning to the results, the authors find that the 3-year birth rate has a larger positive effect on market share after the adoption of the report card. The difference is statistically significant and economically important - a clinic that improves its birth rate from the 25th percentile to the 75th percentile would experience a 13 percent greater increase in its market share after public reporting vs. before. The 1-year birth rate had a similar positive effect on market share in both periods, indicating that consumers have other sources of quality information that are correlated with the 1-year birth rate that they use when choosing among clinics.
The CDC reports include data on the age mix of patients at each fertility clinic. Consumers of fertility services would almost certainly be aware that success rates fall dramatically with age. When the authors include this measure in their models, they find that clinics that treat a disproportionate number of young patients relative to their competitors have lower market shares after public reporting relative to before. This suggests that public reporting allowed consumers to adjust for patient characteristics when evaluating clinic quality.
As only some states require insurers to cover ART services, the authors next look at whether the effect of the 3-year birth rate on market share in the post-1998 period is stronger in those states. They find that it is, and suggest several possible explanations for this finding. Insurers may steer patients towards clinics with higher success rates, as it is in their interest to do so. Alternatively, patients who would not be able to afford ART services in the absence of insurance coverage may be less well informed through informal channels or more responsive to quality information than other patients. The authors find their results to be more consistent with the latter explanations and note that this may have important policy implications: "if report cards are more important to consumers pulled in to health services markets by mandates, then current health policy reforms that seek to expand insurance coverage should also be sure to provide an appropriate source of information for the newly insured."
The authors' results suggest a stronger effect of quality report cards on market share than had been found in previous studies. The authors suggest several possible reasons for the difference. First, in the fertility context quality measures may be more informative or easier to use. Second, consumers may have less access to other quality information here than in other contexts. Third, public reporting was mandatory and therefore available for all providers. Finally, consumers seeking ART services tend to be younger, more educated, and more affluent, and thus may be better able to make use of quality information.
The authors conclude "overall, our findings indicate that report cards in this market have the potential to influence provider behavior." They note that although birth rates have improved during the period of their study, it is not possible to definitively attribute this to the effect of quality reporting, and suggest that further study is necessary to establish the effect of report cards on quality.