Working Papers

Abstract: We study how the diffusion of telemedicine technology impacted the quality and rates of antibiotic prescriptions using Australian survey data from primary care physicians linked to administrative records on their service provision. We classify physicians based on their relative use of telemedicine consultations in response to the introduction of government-subsidized telemedicine during the COVID-19 pandemic and relate their rates of antibiotic prescriptions to indicators of pre-describing quality before and after lockdown periods in a difference-in-differences design. Our results suggest that more frequent users of telemedicine prescribe relatively fewer antibiotics while keeping prescribing quality largely unchanged. We interpret these findings as evidence that telemedicine can enhance the efficiency of service provision in primary care settings.

Abstract: Using a high-frequency panel survey, we examine the sensitivity of estimated self-reported well-being (SWB) dynamics to using monthly, quarterly, and yearly data. This is an important issue if SWB is to be used to evaluate policy. Results from autoregressive models that account for individual-level heterogeneity indicate that the estimated persistence using yearly data is near zero. However, estimated persistence from monthly and quarterly data is substantial. We estimate that persistence to shocks typically lasts around six months and has a net present value of 75–80 per cent of the contemporaneous effect. Estimates are similar for different domains of SWB.

Abstract: We study spillovers in healthcare by exploring how cardiologists’ diagnostic choices are influenced by their peers. We use rich clinical quality data from Sweden to instrument peers’ average weekly radiation output, our endogenous variable of interest, using the peer’s lagged arrival of emergency cases. Our IV estimates imply that focal cardiologists change their own radiation output by 0.5 SD for each SD change in their peers’ output. We show that our results are neither driven by endogenous peer formation nor patient selection. Effects are also stronger in academic hospitals and among younger cardiologists. These spillovers enhance patient welfare as well by increasing the share of appropriate radiation dosage and by reducing subsequent 30-day risk-adjusted mortality.

Abstract: The COVID-19 pandemic has highlighted the crucial role of internet access in pandemic prevention and response. Internet access has facilitated the rapid dissemination of vital information, provided telemedicine services, and enabled remote work and education. This study uses a wide range of data sources to investigate the geographic variation of internet access proxied by 3G mobile broadband during the COVID-19 pandemic in Indonesia. We employ several approaches to account for potential confounding factors,  including using lightning strikes as an instrumental variable, to confirm the significant role that the internet played in the spread of COVID-19 cases. Our findings suggest that increasing internet access could positively impact pandemic prevention and response efforts, particularly in regions with limited connectivity. Therefore, improving internet infrastructure in developing countries may be crucial in preventing future pandemics.

Abstract: Despite being widely used in health economics, dynamic models of health and healthcare typically assume that the persistence in these outcomes is the same for every individual. Understanding the extent and drivers of heterogeneity in persistence is essential for the design and evaluation of health interventions because persistence determines the dynamics and overall long-run effects of such interventions. This paper explores individual-level heterogeneity in the persistence of health outcomes. Using simple regression methods that do not place any restriction on the distribution of the heterogeneity in persistence, the paper documents substantial heterogeneity in health, medical expenditures, and healthcare use. We show that neglecting this heterogeneity leads to estimates that overstate the average persistence and can bias the coefficients of covariates. We find that between 75% and 87% of individuals display persistence and that this persistence is related to the individuals’ personality and socio-economic characteristics.

Abstract: We study the effect of reducing welfare assistance on migrants’ long-term integration in Australia. The policy postponed a migrant’s eligibility for benefits during their first two years in the country. It mainly affected mothers and was announced after their arrival. Using a regression discontinuity design and 21 years of administrative welfare data, we find significant reductions in welfare receipt, where the gap widened over time, and stabilized in the long run. Benefit receipt amounts reduced by 28%, and time-on-benefits by 19%, particularly in the unemployment and disability categories. We observe larger treatment effects for mothers from disadvantaged backgrounds.

Abstract: Peer-to-peer support is often a critical component of mental health programs, but evidence on the effect of peer-based support programs at scale is limited. Using quasi-experimental methods, we examine whether a prominent peer-based support campaign, “R U OK? Day”, affects short-term mental health outcomes in Australia. Using variation in daily records and differences in the campaign’s intensity over nine years, we find no evidence that “R U OK? Day” reduces suicides and suicidal behaviours in the month after the campaign. However, we find positive effects on mental wellbeing, particularly among middle-aged males, with improved social support the likely mechanism. Our results provide evidence that peer support campaigns may be a practical, low-cost approach to improve population mental wellbeing.

Abstract: We study the effects of ethnic networks on the labor market integration of asylum seekers, local firms and their employees in Switzerland. We track outcomes of individuals and firms using a novel dataset, linking 6 employer-employee matched administrative micro datasets covering the universe of residents in Switzerland and all registered firms from 2008 to 2017. Leveraging the quasi-random placement of refugees across locations upon arrival and a novel instrumental variables strategy, we show that larger migrant networks substantially and persistently increase employment and labor income of refugees. Network effects account for 38% of the variation in incomes within nationality cohorts across cantons 5 years after arrival. In line with homophily in network formation, ethnic networks that are demographically similar to the migrant, and those more economically successful have larger positive impacts. Employment effects are concentrated in firms with previous employees from the same nationality, and refugees who quasi-randomly live in the same refugee center for the first 9-12 months are 3 times more likely to be coworkers after leaving the center. This provides direct causal evidence that a network's value is at least partly due to information sharing and job referrals. Using a shift-share IV design, we then estimate how a better-matched inflow of migrants affects local firms and their employees: firms experiencing a positive shock to their network hire more migrants and natives, and there is little evidence of wage decreases for existing workers. This is consistent with networks increasing firm-worker match quality and productivity. Concerns about adverse impacts of spatially concentrated immigration are not borne out in the data, suggesting that refugee dispersal policies may not be optimal.

This paper investigates the intergenerational effect of communication barriers on child health at birth. We study refugees in Switzerland who come from French- or Italian-speaking countries and who, upon arrival, are randomly allocated to different cantons in which either German, French, or Italian is the dominant language. Children born to mothers who were exogenously allocated to a region whose dominant language matched their origin language are, on average, 72 grams (or 2.2%) heavier. Further analyses suggest that this effect is likely driven by information about health-related behavior and services. Co-ethnic networks, however, can partly compensate for communication barriers.