Antimicrobial resistance has been growing rapidly in the United States in recent years despite government efforts to control its outbreak. Both under and overutilization of prescribed medications can lead to an increase in antimicrobial resistance. The introduction of Medicare Part D in 2006 led to an increase in prescription drug coverage, including antimicrobials, for the elderly. If cost barriers had led to underutilization of prescriptions among those without previous prescription coverage, then Medicare Part D may reduce antimicrobial resistance. On the other hand, if Medicare Part D encourages over-utilization of prescriptions, then an unintended consequence may be an increase in antimicrobial resistance. Using data from the Nationwide Inpatient Sample (NIS), Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality for years 2004 to 2011 and a difference-in-differences identification strategy, I estimate the net effect of Medicare Part D on the incidence of inpatient discharges due to antimicrobial resistance among the Medicare-eligible population. Results show that the incidence of antimicrobial resistance among the elderly as measured by inpatient discharges decreased after Medicare Part D implementation. 

Medicare Part D, implemented on January 2006, expanded subsidized outpatient prescription-drug coverage, including medication-assisted treatment (MAT) drugs, for adults aged 65 and older. Using 6.2 million discharge records from the 2001 to 2011 Healthcare Cost and Utilization Project National Inpatient Sample, I study how this policy influenced serious opioid use disorder (OUD) events. The analysis uses a difference-in-differences framework that compares OUD-related hospitalizations among 65- to 69-year-olds with those of 60- to 64-year-olds, and an event study confirms parallel trends before implementation. Part D lowered OUD-related hospital admissions by 158.3 per 100,000 discharges, a 53% decline from the pre-policy mean. The reduction is concentrated in admissions that include diagnoses of opioid abuse or dependence, whereas admissions related to opioid poisoning show little change. Larger proportional declines among Black and Hispanic adults and among men indicate that changes in drug coverage were accompanied by differences in outcomes across demographic groups. Estimates are robust across alternative specifications. By reducing out-of-pocket costs for MAT drugs covered by Part D, comprehensive prescription benefits can substantially cut avoidable hospital stays even when individual treatment uptake is not observed. These findings inform current debates on drug-benefit design for aging populations confronting opioid-related harm and illustrate how insurance coverage shapes health production later in life.

This paper estimates the effects of transport (road and rail) & energy and ICT infrastructure (telephone, mobile, and broadband) on GDP growths in neighboring countries as well as own countries. We confirm positive direct contributions of infrastructure, access to Internet, and human capital on economic growth. The spatial panel regression models indicate that there exist positive externalities of the broadband infrastructure and human capital, and these results are robust regardless of the choice of spatial weight matrices. Our findings on spillover effects of infrastructure suggest the key role of neighboring countries’ infrastructure on own country’s economic growth.

China’s “One Belt, One Road” (OBOR) initiative aims to foster connectivity and cooperation among 65 nations. Together, these countries account for about 60 percent of the world’s total population and 30 percent of the world’s gross domestic product. OBOR, also called the “21st Century Maritime Silk Road”, has two main channels that will then connect each other to Europe. These are the land-based Silk Road Economic Belt (One Belt), which connects Xi’an, China, to Rotterdam, Netherlands, and the sea-based Maritime Silk Road (One Road), which connects Venice, Italy, to Fuzhou, China, through the Suez Canal and the Indian Ocean. For countries that have officially signed to participate in OBOR and are located on these channels, the proposed priority areas for cooperation include infrastructure development and connectivity, policy dialogues, unimpeded trade, financial support, and people-to-people exchanges. Using a vector autoregression model, this paper estimates the likely effects of OBOR on Philippine trade and investments.

This paper estimates the causal effect of surgical delays on inpatient outcomes using weekend admission as an instrument for time to procedure. Using the Nationwide Inpatient Sample from 1998 to 2011, I focus my analysis on emergency admissions for acute appendicitis, acute cholecystitis, and obstructed hernia. A two-stage least squares design exploits weekend admission as a source of plausibly exogenous variation in treatment timing, addressing confounding from unobserved illness severity and hospital selection. In the first stage, weekend admission delays surgery by 0.18 days for acute cholecystitis patients (12.35 percent of the mean) and by 0.12 days for obstructed hernia patients (12.46 percent). In the second stage, each additional day of delay increases length of stay by 0.76 days for cholecystitis patients (20.94 percent of the mean) and 1.55 days for hernia patients (31.19 percent), and for hernia patients also raises the risk of postoperative wound complications by 0.027 percentage points (146 percent relative to the mean). No second-stage effect is detected for appendicitis, consistent with the weak first stage. Heterogeneity analyses show larger effects in rural, government, and low-income-serving hospitals, as well as differences across patient subgroups by payer status, race, and income. A back-of-the-envelope calculation suggests that weekend-related delays for cholecystitis and hernia admissions generate roughly 10,558 excess inpatient days annually, equivalent to about $27.5 million in additional hospitalization costs. These findings provide new causal evidence on the consequences of delayed surgical treatment and underscore the efficiency and equity gains from improving timely access to acute surgical care.

This paper examines the impact of automation on workers’ health risks, using occupational injury and hospitalization data. We first document that automation leads to a divergence in the severity of occupational health risks. Utilizing data on nonfatal and fatal workplace injuries, we show that while automation reduces nonfatal occupational injury incidence, it increases fatal occupational injury incidence. Secondly, the disparity of health risks across age groups has widened due to automation. Hospital discharge data suggests that overall hospitalizations due to injuries and despair-related conditions (e.g., substance abuse) have declined in commuting zones with higher automation exposure. Yet, the benefits are concentrated among young workers, while middle-aged workers exposed to automation experience increased hospitalizations, particularly due to despair-related conditions. Combining the empirical estimates of nonfatal and fatal injury incidence, a back-of-the-envelope calculation suggests that workplace automation overall provides significant, health-driven economic benefits. However, the rise in fatal injuries offsets between 14% and 27% of economic benefits from the reduction of nonfatal injuries. In summary, automation improves workplace safety on average, but it shifts the distribution of accidents away from minor injuries towards high-impact failures and disproportionately reduces hospitalizations for younger workers while potentially increasing risks for older workers.

We estimate the causal effect of the ACA's Medicaid expansion on physician residency training capacity using state-level National Resident Matching Program (NRMP) data from 2010 to 2019 and a staggered difference-in-differences framework. Across all specialties, expansion had no significant effect on residency slots per capita, consistent across four modern estimators. Specialty-level disaggregation reveals substantial heterogeneity: primary-care programs show no differential trend, while emergency medicine, radiology, anesthesiology, and dermatology (E-RAD), together with transitional year programs linked to them, experienced a meaningful and persistent relative shortfall in slot growth; leave-one-out analyses identify emergency medicine and dermatology as the primary within-group drivers. Fill rates in affected specialties remained stable while filled positions and U.S. graduate matches declined, pointing to a supply-side institutional contraction rather than weakening applicant demand. Medicaid GME funding grew modestly in expansion states and physician supply showed no differential shift, leaving operating margin pressure as the most plausible explanation. These findings document an unintended consequence of broad coverage expansion: a relative training slowdown concentrated in programs expensive to expand above the Medicare GME cap, lacking dedicated external funding, and facing no public health or workforce policy mandate to grow, with implications for the long-run specialist workforce.

In October 2014, the United States government implemented Section 3008 of the Affordable Care Act (ACA), which imposes a 1 percent reduction in the Medicare reimbursements of hospitals that perform poorly based on a hospital-acquired infection (HAI) measure. A limited body of literature evaluates the impact of this policy in a primarily descriptive manner. Using patient discharge data from the National Inpatient Sample (NIS), Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality and a difference-in-differences identification strategy, I contribute to the literature by estimating the causal effects this ACA provision had on the incidence of HAIs. Results suggest that the policy reduced the likelihood of acquiring an infection, with effects varying by HAI type. In addition, I find a general reduction in the likelihood of a HAI for whites, while the effects by gender or age vary on HAI type.