Objective - In many hospital care systems, nursing staff are recognized as a major input for the delivery of hospital care. Despite being the largest clinical workforce in the English NHS, nursing staff has been under significant pressure with high vacancy numbers and increasing leavers’ rates, which have a direct impact on the organization of work within hospitals and patient outcomes. To address these concerns at the local level, NHS Improvement (NHSI) launched the nationwide Retention Direct Support Programme (RDSP) in July 2017, which aimed to reduce turnover rates of nursing staff in Acute-care Trusts and clinical staff in Mental Health Hospital Trusts in England. The programme was rolled out in a staggered fashion over 5 cohorts, and it required Trusts to develop bespoke workforce strategies. NHSI monitored the progress of the Trusts in the 12 months following the start of the programme and provided targeted support where needed. With this study, we evaluate the effectiveness of changing the non-pecuniary aspects of public-sector jobs on a large scale and offer causal evidence that changing management practices can improve employee retention.

Methods - We use detailed administrative employee-level data from 2015 to 2019, and exploit the differential timings of the programme's start dates to evaluate the RDSP’s effectiveness on nursing retention, by implementing recent methodological advances in the difference-in-difference literature with staggered treatment adoption. Results - The programme has improved nursing retention by 0.78 percentage points (pp), and helped retain, on average,  1,697 nurses and midwives who would have left their hospital organization otherwise. We also find that RDSP strategies reduced NHS-leaving rates on average by 0.41 pp. We do not find any evidence that the programme has any negative impact on mortality rates or emergency re-admissions of planned patients in treated hospitals. We also explore the strategies NHS hospitals adopted as part of the RDSP to understand the mechanisms behind the impact of the programme on retention. Our findings suggest that actions focusing on staff wellbeing, engagement, and career development are associated with higher retention for nursing staff in hospitals with above-average turnover rates. On the other hand,  managerial strategies such as supporting new starters and working on recruitment strategies are associated with improved retention for nursing staff in hospitals with below-average turnover rates.

Discussion - Non-monetary local interventions have the potential to help improve hospital workforce retention in English NHS. The RDSP’s impact might be limited in alleviating the nursing workforce challenges as a standalone policy, but such interventions can provide viable and sustainable ways to prevent the ‘heating’ of workforce pressures in publicly funded healthcare systems.

Value-based insurance design (VBID) can shape managed healthcare systems centered on value by promoting the use of high-value and disincentivizing low-value care. However, concerns arise that individuals might oppose the implementation of VBID if health plans constrain consumers’ freedom of choice. The aim of this paper is to elicit consumers’ preferences for VBID elements in a managed competition health insurance system. We fielded a discrete choice experiment to estimate the willingness-to-pay (WTP) and willingness-to-accept (WTA) for selected VBID elements in Switzerland using conditional and mixed logit models. Our results suggest a strong status quo bias when choosing insurance and resistance to higher cost-sharing or restrictions on low-value care. Posterior WTP and WTA distributions reveal substantial preference heterogeneity, which we characterize along the countries’ language regions. Based on these results, tailored communication strategies could be developed to help shape future health policy and support the integration of VBID in basic insurance.

Objectives: Sustaining access to timely health care and other care treatments is one of the major policy concerns in many countries. Especially the nursing home care sector, which mainly provides care to the older population, is expected to face increased demands for their services while supply remains limited. One important policy question is what the consequences are of limited availability of nursing home care on individuals' outcomes, and whether it generates spillovers to other sectors, such as to hospitals.

While previous attempts to empirically answer this question rely on (often regional) variation in supply or eligibility for benefits, we examine the consequences of delayed access to nursing homes in the form of wait times. With this, we aim to identify the causal impact of delayed nursing home admissions on hospital care utilization. By paying particular attention to the differences between subgroups of people on the waiting list by the degree and cause of frailty, we provide insights into an optimal allocation of nursing home care services.

Methods: We use administrative data at the individual level on nursing home and hospital care use of all individuals who are newly eligible to receive nursing home care in the Netherlands from 2015 to 2018 (n = 72,762). Using a unique combination of data on the starting date of eligibility and of the nursing admission, we identify the number of days by which an individual’s nursing home admission is being delayed and examine its impact on the probability of having an urgent hospitalization within a year after eligibility. To account for a potential selection bias in delays, either induced by the demand or supply side, we use an instrumental variable based on plausibly exogenous variation in fluctuations in queue congestion within regions over time.

Results:  Those who became eligible to receive nursing home care in a period of congestion are more likely to delay their nursing home admission and delaying a nursing home admission by one month increases the risk of an urgent hospitalizations by 1.4 percentage points. This effect is mainly driven by people with dementia care needs and can to a large extent be explained by hospitalizations after a fall, and potentially by the longer exposure to the protective environment of the nursing home.

Discussion: Our results suggest that delayed access to nursing home care generates spillovers to the hospital sector, especially among people with dementia. This implies that ageing-in-place may not save as many resources as a naive analysis may suggest. Policymakers should take this into account when allocating resources both across and within systems and should aim to minimize the spillovers by improving the alternatives for nursing home care.

We examine how coworkers influence innovation diffusion. Our matched task-worker-workplace data track physician behaviour and employment for 15 years from the introduction of an innovation to its maturity. Our estimation exploits this dynamic network of coworkers to jointly estimate peer effects from coworker take-up and effects from exposure to coworkers, including particularly influential physicians (key players), whose identities we estimate jointly with coworker effects. We find positive peer effects and positive effects of exposure to coworkers, which are considerably amplified by exposure to key players. We use our estimates to show that modest but well targeted training and secondment policies can lead to substantive increases in innovation diffusion.

Rising life expectancy has raised concerns over the sustainability of generous defined ben- efit pensions. Reforms to make defined benefit pensions less generous have been common in many countries in recent decades. But the theoretical impacts of such changes on labour supply are ambiguous. This paper provides new evidence on the impact of such reforms on the labour supply of a highly skilled group who were exposed to reform of their pension at least 10 years before retirement age and in a position to alter their labour supply. We use administrative payroll data to study the impact of a reform that moved all public sector employees from a final salary pension scheme to a less generous career average scheme on the labour supply of senior doctors in the English National Health Service. Exploiting the staggered roll-out of the new scheme across narrowly defined age groups, we show that the labour supply of senior doctors increased as a result of the reform. The effects are driven by an increase on the extensive margin, senior doctors being 3.7% more likely to remain in the NHS four years after being moved onto the new scheme. Those doctors working in hospitals with fewer outside options, those who had less performance-related pay and those working in hospitals with lower overall staff satisfaction increased their labour sup- ply most. The results show that shifts to less generous schemes do not necessarily reduce labour supply.

Objectives: Hot debates are happening in the US regarding how and whether to expand Medicaid and Medicare; developing countries are striving to increase their social insurance coverage rate following the WHO's promotion of "Universal Health Coverage". However, many proposed reforms only cover a subset of the population. Understanding the impact of social insurance expansion on the welfare and drug demand of the unintended population will help policymakers to evaluate similar programmes. Measuring the price elasticity of demand for drugs is also an academic challenge that has long been in the spotlight of economic research.

Methods: Exploiting the Medical Expenditure Panel Survey and the introduction of Medicare Part D, a US prescription drug insurance targeted at the above-65 (age) US citizens, I construct an event study to identify the programme's impact on the below-65s' drug prices. As a robustness check, the uninsured above 65s' market share of a drug's molecular class before Medicare Part D, as a proxy of a drug's exposure to Part D, is used to conduct a continuous difference-in-differences analysis. To study the impact of Part D on the below 65s' welfare, I estimate a Hanemann (1984)-style discrete-and-continuous demand model. The model enables me to distinguish between the extensive (whether to consume) and the intensive margin (how much to consume) of drug consumption, which is lacking in the existing literature. Following Huang and Rojas (2013, 2014) and Dubois (2022), I estimate the market share of the outside option by the parametric feature of the demand model. Compensating variations are then computed based on the estimated demand model.

Results: Between 2006-2010, Part D reduced the below-65 prices of drugs by 12% on average for the drugs in my sample. Price decreased less for brand-name drugs and the uninsured below 65s than for generic drugs and the insured. The average price elasticity of demand for drugs is -1.15. Elasticities are higher for generic drugs and the uninsured than for brand-name drugs and the insured. The Part-D-led price reduction resulted in a welfare increase equivalent to 61.4 USD per individual per year, nearly all of which came from the intensive margin. The uninsured and poor below 65s benefited less from the price reduction than their insured and non-poor counterparts in compensating variation. Following price reduction, per-capita consumption increased by 11.9 daily recommended doses annually on average in highly effective drugs as well as opioids and antibiotics, of which the users should be closely monitored. Part D exacerbated the potential of overdose among some uninsured insulin users.

Discussion: Large-scale social insurance expansion increases consumer welfare by reducing drug prices and increasing spending among existing users. Consumption should be monitored, especially among the uninsured.

We study a fundamental reform of the public Disability Insurance (DI) system in Germany. Effective 2001, cohorts born after 1960 are no longer eligible for “occupational DI.” Occupational DI (ODI) implies benefit eligibility when health shocks prevent employees from working in their previous occupation. For the affected “notch cohorts”, the new DI eligibility rules require work disability in any job. Using administrative data, we first show that the reform significantly reduced the inflow of new DI beneficiaries by more than 30% in the long-run. Next, we validate these findings using representative SOEP household panel data comprised of the entire underlying population. The second part studies interaction effects with the private ODI market. Using representative data, we do not find much evidence that the notch cohorts purchased individual private ODI policies at significantly higher rates to compensate for the reduced generosity of the public DI system. To explain such low take-up, we employ a general equilibrium model featuring the roles of the social safety net, administrative costs, and asymmetric information. These driving forces help explain four stylized facts in the individual experience-rated private market for ODI policies: (1) low private ODI take-up and interaction effects with the public system—despite a high lifecycle work disability risk, (2) strong and positive income and health gradients in private ODI take-up, (3) inversely related income and health gradients in the lifecycle work disability risk, and (4) adverse selection. Simulations illustrate that policy reforms to lower administrative costs have the greatest potential

This paper develops a new approach to identifying to what extent individuals strategically time their healthcare consumption under deductibles in health insurance. I set up a dynamic model of healthcare consumption where individuals exceed a high deductible after a large health shock, and have an incentive to prepone care planned for the next year. The model elicits the links between timing and classical moral hazard responses, as well as deductible choice, and highlights trade-offs for insurance policy. It also serves to show that pure timing moral hazard can be identified using random variation in the timing of the health shock within the calendar year. Empirically, I find quantitatively large timing moral hazard in the context of mandatory health insurance in Switzerland. This response can create important distortions in insurance markets by shifting out-of-pocket healthcare costs onto the risk pool. Its extent decreases with the time available until the deductible reset. The insured do re-optimise on-the-go after the shock, but face substantial frictions in retiming.

This paper investigates the potential of new technologies to reduce disparities in the provision of healthcare services. Differences in providers’ skills may cause variation in patient outcomes. The adoption of innovations, like robots, can attenuate this problem if technological gains are decreasing in users’ skills or may exacerbate existing variation in performance otherwise. I show that, in England, the diffusion of surgical robots coincided with an improvement in average surgical performance and a convergence in outcomes between high and lower-skilled surgeons for prostate cancer patients. I study whether this pattern can be attributed to the adoption of robots using the universe of inpatient admissions to the National Health Service (NHS). To identify the effects of robotic surgery on patient outcomes, I exploit quasi-random variation in the geographic allocation of robots, allowing for selection and heterogeneity in treatment effects. I find that robots shorten patients’ length of stay in hospital and decrease the incidence of adverse events from surgery, but their effects significantly depend on surgeons’ skills. The robot has little impact on the performance of highly skilled surgeons, while lower skilled surgeons gain the most from it. I also uncover a strong pattern of negative selection on both observable and unobservable characteristics. Although the attainable gains are higher for lower-skilled surgeons, they use the robot the least. My results suggest that the potential benefit of a new technology largely depends on how it combines with the skills of the individual users.

International procurement institutions have played an important role in drug supply. This paper studies price, delivery, and procurement lead time of essential drugs supplied in 106 developing countries from 2007-2017 across four procurement institution types. We focus on four major therapeutic areas for infectious diseases that disproportionately affect people living in LMIC: antiretrovirals, antimalarials, tuberculosis drugs, and antibiotics. We find that pooled procurement institutions lower prices: pooling internationally is most effective for small buyers and more concentrated markets, and pooling within-country is most effective for large buyers and less concentrated markets. Pooling can reduce delays, but at the cost of longer anticipated procurement lead times. Finally, pooled procurement is more effective for older generation drugs, compared to intellectual property licensing institutions that focus on newer, patented drugs. We corroborate the findings using multiple identification strategies, including an instrumental variable strategy as well as the Altonji-Elder-Taber-Oster method for selection on unobservables. Our results suggest that the optimal mixture of procurement institutions depends on the trade-off between costs and urgency of need, with pooled international procurement institutions particularly valuable when countries can plan well ahead of time.

Background. The European Medicines Agency (EMA) offers multiple pathways to marketing authorization, aiming to standardize and accelerate authorization procedures and patients’ access to pharmaceuticals. To date, there have been no causal analyses of the extent to which being a member of the EU and having access to centralized authorization affects the launch delay and availability of pharmaceuticals. Methods. We exploited two changes to the eligibility criteria for EMA authorization procedures. First, since 2004, several central and eastern European countries have joined the European Union (EU) and obtained access to these procedures. Second, the number of indications and drug types that fall within the compulsory scope of the so-called centralized procedure (CP) has increased. Using classic and staggered difference-in-differences (DiD) approaches, we examined the effect of (a) joining the EU and becoming subject to the regulations of the EMA, (b) voluntary participation in the CP and (c) compulsory participation in the CP on the launch delay and availability of pharmaceuticals. European countries that were not part of the EU or European Economic Area (EEA) served as controls. Results. Staggered DiDs showed that countries experienced a mean decrease in launch delay of (a) 11.6 months (p=0.003) after joining the EU, (b) 13.3 months (p=0.001) for pharmaceuticals that gained access to voluntary CP participation, and (c) 10.5 months (p=0.061) for biologics, which are subject to compulsory CP participation, and 9.9 months (p=0.081) for further indications that became subject to compulsory CP participation in 2005. However, only comparisons (a) and (b) withstood all robustness checks. Conclusion. Our results suggest that by gaining access to EMA procedures, a country can significantly reduce the time needed for patients to access new pharmaceuticals. This reduction is larger for pharmaceuticals that fall within the voluntary scope of the CP, which are often financially less attractive to and less prioritized by manufacturers than those within the compulsory scope of the CP. Expanding the compulsory scope of the CP to include these pharmaceuticals might further speed up patients’ access to innovative pharmaceuticals in the EU/EEA.

Objective. This study contributes to our understanding of the long-term health effects of early childhood education programmes. It evaluates a Spanish universal preschool reform, which implied a large-scale expansion of full-time high-quality public preschool for three-year-olds from 1991/92 school year and substituted care provided by the nuclear family, and its effects on health and healthcare use in the long run. Methods. Despite being nationally enacted, the implementation of the reform was the responsibility of the Spanish regions allowing to exploit the fact that the initial intensity of public preschool uptake varied across regions. To study the effect of the policy on long-term health, I use both survey and administrative data and employ a difference-in-differences strategy exploiting the timing and geographical variation of the implementation of the programme. I compare long-term health of cohorts aged three before to those aged three after the start of the programme, across individuals either residing or born in regions with varying initial intensity (measured as the regional increase in public enrolment rates of three-year-olds between 1990/91 and 1993/94) in public preschool implementation. Results. The findings show that the Spanish universal preschool reform has no effect on long-term health, except for two (out of ten) indicators. First, an increase of 10p.p. in the initial intensity in public preschool expansion reduces the probability of being diagnosed with asthma by 2.1p.p. for individuals aged three post-policy. Instead, the policy increases hospitalisation rates by 2.7%. The results indicate that the effect on asthma is driven by men, hospitalisation rates are higher for pregnant women, and disadvantaged children benefit the most in terms of better long-term health. Discussion. This study tests if early education policies targeting enrolment, educational attainment and maternal employment have also spillover effects on long-term health. Overall, the results show that the Spanish universal preschool programme does not affect long-term health. This finding suggests that expanding the number of places in preschool is not sufficient to affect long-term health in institutional contexts such as Spain. The policy mostly improved the health of children from low and medium socioeconomic backgrounds. This might imply that more disadvantaged children enrolled in preschool once the programme started and, thus, benefited the most due to a change from (low-quality) family care to high-quality formal out-of-home care. Universal childhood education programmes have a lower cost per child but larger overall expenditure than targeted ones, while they seem to mainly benefit disadvantaged children as targeted programmes do. These results thus suggest that universal programmes might not be as cost-effective as targeted policies are.