Shan Huang

Shan Huang

I’m a health economist at Stockholm School of Economics and the University of Copenhagen.

Working Papers

The role of physician altruism in the physician-industry relationship
NBER Working Paper 33439 (with Anirban Basu and Jing Li)

Abstract Financial incentives can distort physicians' treatment decisions, fueling healthcare spending. Altruism, a core element of medical professionalism, may counteract these distortions. We link altruism elicited from a revealed preference experiment for 267 U.S. physicians to administrative data on industry transfers and prescribing. Non-altruistic physicians receive substantially higher payments (USD 1,775, or 111% more annually) and increase prescribing of promoted drugs after payment, whereas altruistic physicians do not. Divergence is largest in drug classes with high clinical substitutability. Our findings show that altruism moderates the influence of financial incentives in physician-industry ties, limiting the scope for agency problems in prescribing.

Publications

Provider effects in antibiotic prescribing: Evidence from physician exits
Journal of Human Resources, forthcoming (with Hannes Ullrich)

Abstract In the fight against antibiotic resistance, reducing antibiotic consumption while preserving healthcare quality presents a critical health policy challenge. We investigate the role of practice styles in patients’ antibiotic intake using exogenous variation in patient-physician assignment. Practice style heterogeneity explains 49% of the differences in overall antibiotic use and 83% of the differences in second-line antibiotic use between primary care providers. We find no evidence that high prescribing is linked to better treatment quality or fewer adverse health outcomes. Policies improving physician decision-making, particularly among high-prescribers, may be effective in reducing antibiotic consumption while sustaining healthcare quality.

Assessing the value of data for prediction policies: The case of antibiotic prescribing
Economics Letters, Vol. 213, 110360, 2022 (with Michael Allan Ribers and Hannes Ullrich)

Abstract We quantify the value of data for the prediction policy problem of reducing antibiotic prescribing to curb antibiotic resistance. Using varying combinations of administrative data, we evaluate machine learning predictions for diagnosing bacterial urinary tract infections and the outcomes of prescription rules based on these predictions. Simple patient demographics improve prediction quality substantially but larger reductions in prescribing can be achieved by making use of rich health data. Our results suggest decreasing returns to data for prediction quality and increasing returns for policy outcomes. Hence, data needs for prediction policy problems must be assessed based on the policy objective and not only on prediction quality.

The effect of a ban on gender-based pricing on risk selection in the German health insurance market
Health Economics, Vol. 29(1), pp. 3-17, 2020 (with Martin Salm)

Abstract Starting from December 2012, insurers in the European Union were prohibited from charging gender‐discriminatory prices. We examine the effect of this unisex mandate on risk segmentation in the German health insurance market. Although gender used to be a pricing factor in Germany's private health insurance (PHI) sector, it was never used as a pricing factor in the social health insurance (SHI) sector. The unisex mandate makes PHI relatively more attractive for women and less attractive for men. Based on data from the German socio‐economic panel, we analyze how the unisex mandate affects the difference between women and men in switching rates between SHI and PHI. We find that the unisex mandate increases the probability of switching from SHI to PHI for women relative to men. On the other hand, the unisex mandate has no effect on the gender difference in switching rates from PHI to SHI. Because women have on average higher health care expenditures than men, our results imply a worsening of the PHI risk pool and an improvement of the SHI risk pool. Our results demonstrate that regulatory measures such as the unisex mandate can affect risk selection between public and private health insurance sectors.

Work in Progress

Organizing Expertise with Externalities: Evidence from Primary Care
(with Amanda Dahlstrand, Guy Michaels, and Nestor Le Nestour)

Abstract Understanding the determinants and consequences of the division of labor is central to economics. Modern organizations often divide labor using knowledge hierarchies in which frontline workers handle simpler tasks and pass tasks they cannot handle to experts. We study the effects of a knowledge hierarchy (KH) in almost 500,000 patient cases from an online primary care firm, linked to outcomes and costs across Sweden's healthcare system. The firm’s algorithm quasi-randomly assigns similar patients either to a nurse–doctor hierarchy or directly to doctors, depending on short-term variation in the availability of doctors. Within the KH, nurses resolve 70% of cases and pass the remainder to doctors. Relative to doctor-only care, the KH modestly increases some follow-up care outside the firm, but has no adverse effects on patient satisfaction or higher-stakes outcomes, including hospitalizations, earnings losses, and mortality. The KH does not change mean systemwide production costs, though it raises costs for some tasks and reduces them for others. Compared to using only doctors, the firm's selective use of the KH for different symptoms yields total healthcare savings equivalent to 2.3% of online primary care costs. However, the firm does not observe the downstream outcomes of individual patients and is not incentivized to internalize the externalities it generates in the healthcare system. We find that reassigning cases between KH and direct-to-doctor care within 15-minute windows based on comparative advantage in systemwide production costs, including downstream costs, would save the equivalent of 6.7% of online primary care costs.

AI adoption by human experts: Evidence from primary care physicians
(with Renke Schmacker and Hannes Ullrich)

Abstract AI has the potential to increase productivity by extracting information from rich data and improving the quality and speed of decision-making. Yet there is limited evidence on how domain experts incorporate AI-generated signals into their decisions relative to familiar, incumbent decision aids. We study expert beliefs and decisions in primary care for urinary tract infections, a leading cause of antibiotic use, where treatment is often initiated under diagnostic uncertainty. In a nationwide survey experiment with 372 Danish primary care physicians, participants evaluate patient vignettes and make diagnostic and prescribing decisions before and after receiving diagnostic signals. We hold signal accuracy constant while varying between-subjects whether the same information is presented as an AI prediction output or as a conventional urine dipstick result. In a second stage, physicians see both signals and we randomize whether the signals are independent or correlated, allowing for a test of correlation neglect. Physicians update less in response to AI than to dipstick signals, consistent with AI skepticism and a heuristic use of the dipstick signal. When presented with multiple signals, physicians insufficiently adjust for correlation and overweight correlated signals. Physicians can be separated into one group that adopts the AI tool to varying degrees and one that does not adopt the AI tool at all. Non-adopters' mental models indicate concerns about non-informative data and suboptimal information processing by AI. Linked administrative data further show that non-adopters are more likely to practice in rural clinics with lower use of other diagnostic technologies.