Summary statistics regarding our final sample are detailed in Table 3.3. Research in- dicates that physicians may interact with new technology differently than administrative staff. Hu, Chau, Sheng, and Tam (1999) consistently find that physicians differ from other types of users when accepting technology, specifically in the area of telemedicine. These differences can be attributed to their specialized training, autonomous practices, and pro- fessional work environments [Walter and Lopez, 2008]. Because of our limited physician sample size, we conduct our analysis at the clinical vs. non-clinical level. Clinical em- ployees are those actively providing medical treatment to patients, while non-clinical staff would include receptionists, billing specialists, and administrators who do not directly provide medical care. Non-clinical staff often have more experience using IT systems in the office, as they normally conduct billing and scheduling through specialized software systems; Barker et al. (2003) find prior levels of IT usage to influence adoption behavior, so this difference could influence non-clinical providers’ acceptance of a new EHR in a different manner than clinical staff members. Additionally, Yi et al. (2006) and Math- ieson (1991) posit that medical providers may behave differently in the presence of EHRs because of the availability of support staff to deal with the system on their behalf.
This essay examines topics in healtheconomics. The first study uses data obtained from the Health and Retirement Study (HRS) and the Rand HRS files, to examine the relationship between access to retiree health insurance (RHI) and the decision to leave one’s career job. This paper does not restrict attention to individual’s who choose to take a full retirement, as recent data indicates that only 51.4% of individuals leave a career job and fully retire, while nearly 25% leave their career job, and pursue a partial retirement. In this paper a Cox Proportional Hazard Model with time varying covariates is utilized to estimate the probability that an individual disengages from their career job, given they have not yet done so. Results indicate that those with access to RHI are significantly more likely to leave their career employer in all time periods than identical individuals without RHI.
for environmental risk factors that are hardly considered, at least in the economic literature. Although these factors might have a small impact on the individual level, the fact that a large share of the population is exposed to numerous risk factors may lead to substantial aggregate effects on HCE. The remainder of this thesis is structured as follows: in the next two subsections, I briefly discuss the data and methods used in the subsequent studies and provide a short summary of each of the following four chapters. Each of these chapters constitutes a self-containing empirical study in the field of healtheconomics and addresses research questions related to either efficiency aspects in health insurance markets or the understanding of the prevalence of preventable risk factors. The former is addressed by chapters 2 and 3, which consider health insurance switching behavior between SHI and PHI and within the SHI respectively in the German context. The latter is covered by chapters 4 and 5, which investigate individual and environmental risk factors respectively. Concluding remarks are given in chapter 6.
We present summary statistics of our sample in Table 4.1. Sample averages and standard deviation of non-indicator variables, as well as percentages of indicator variables are reported separately in the upper and lower panel of Table 4.1. The first two columns of Table 4.1 report information of married couples during the sample period. On average, wives have slightly higher BMI and less education than husbands. Roughly 25% of couples are obese or overweight in our sample. Two thirds of couples in our sample have at least one child and about half of them have health insurance. In the couple sample, 67% of men are smokers while only 3% of women smoke frequently; men have higher employment rate than women (82.3 % v.s. 70.7%). Around 30% of the couples come from urban areas. China uses a residence registration system called ”hukou”, which classifies people as rural or urban residents, to restrict free migration and determine eligibility to local resources such as public education, medical care and pension plan. For example, school-age children from rural areas do not have access to public schools in urban areas, even if they have been living in the urban areas. We are interested to learn whether China’s economic transformation had affected overall health condition of people with urban or rural hukou differently, given that generally urban areas have benefited more from the transformation during the sample period. For this purpose, we conduct our analysis for urban and rural areas separably.
According to the official website (http://www.cpc.unc.edu/projects/china), the China Health and Nutrition Survey (CHNS) is a collaborative project between the Carolina Population Center at the University of North Carolina at Chapel Hill and the National Institute of Nutrition and Food Safety at the Chinese Center for Disease Control and Prevention. Its main goal is to study the effects of the health, nutrition, and family planning policies and programs implemented by national and local governments and examine the dynamic impact of the economic transformation of Chinese society on the health and nutritional status of its population. The survey is collected over a 3-day period with a sample of about 4,400 households, including 26,000 individuals in nine provinces. In addition, detailed community data were collected.
Public reporting for nursing homes has existed for almost two decades since it was first introduced in 1998. Nursing Home Compare (NHC), which is a web-based nursing home report card, was launched by CMS (then known as the Health Care Financing Administration) in October 1998. Initially, only a few facility-level structural characteristics and information on health inspections were reported. The nurse staffing measure was added after 2000. After experimenting in six pilot states for six months, CMS introduced the Nursing Home Quality Initiative (NHQI) nationally in November 2002. In this version of the report card, quality indicators were introduced in addition to the existing health inspection and staffing information. Most recently in December 2008, CMS launched a newly designed five-star quality rating system that translates detailed and fragmented measures into more simplified and summarized stars ranging from one to five stars, with a higher number of stars indicating better quality.
Bünnings and Tauchmann (2015) and others and consider risk preferences as fixed over time, using the average value of an individual’s responses. Table 1.3 shows that risk aversion gives rise to advantageous selection in favour of the public sector: after controlling for observables, risk-averse individuals are less willing to opt out of public insurance and less likely to be hospitalised. Switching to private insurance implies uncertainty about future premiums, as changes in family status translate into premium changes in private insurance. For example, a privately insured couple who become parents has to pay for their child in private health insurance, whereas the child is insured free of charge in public insurance. This may explain why risk-averse individuals prefer public insurance. The finding that risk-averse individuals tend to be less risky has been observed also in other contexts (Finkelstein and McGarry 2006). Possible explanations for this result are that the risk averse use more preventive care, or that
To explore the impact of reimbursements on …rm advertising, we utilize data from Kantar Media. Kantar Ad$pender contains advertising data at the media-product-year- designated market area (DMA) level. Because DMAs are bigger than counties, we need to aggregate our reimbursement data. We create variables that represent the percentage of Medicare bene…ciaries in a DMA that live in an urban, urban ‡oor, and ‡oor county. We examine the impact of these variables on TV spot advertising spending per Medicare bene…ciary in a DMA. We de…ne this measure in two ways. In the …rst, we restrict our analysis to products with "Medicare" in their name. This includes Medicare Advantage plans, but also Part D and Medicare supplement plans as well. Furthermore, not all carriers report a speci…c Medicare line. The Kantar data does not allow us to distinguish between these products and data may be an overestimate or an underestimate of the amount of advertising for MA plans. However, we have no reason to believe that advertising for Medicare supplement or Part D plans would vary with ‡oor status. Average spending per Medicare enrollee is $5.90 per year. In the second de…nition and panel of Table 13, we take Kantar de…nition of "health insurance" as given, noting that not all Medicare products are denoted by name. While we would prefer to restrict to only Medicare Advantage products within health insurance, the products are not coded …nely enough in the data. However, Medicare products comprise the bulk of individual insurance plans sold (and, presumably, targeted advertising) within all DMAs.The dependent variable is skewed, with only about half of DMAs having advertising, but total spending in the 90th percentile of DMAs is $2.2 million per year.
To assess the influence of the new Ghanaian health insurance scheme (NHIS), we analyse cross-sectional data from the fifth Ghanaian Living Standard Survey, which is representative for the 2005 Ghanaian population. As the NHIS was implemented at the district level, we can use the variation in when individuals were interviewed at the various sub-districts. Since most districts introduced the NHIS during the survey period in 2005, we compare individuals who were interviewed before and after the introduction of the insurance scheme, conditional on fixed effects of district and interview month. Using the quasi-exogenous variation in the availability of formal health insurance, we apply ordinary least squares estimates to estimate the impact of the new insurance scheme on making and receiving transfers at the extensive and intensive margin. We also test whether the treatment effect depends on the relationship between the recipient and donor of a transfer. Our empirical findings indicate that introducing a formal health insurance scheme reduces the probability of making transfers. In addition, the number of remittances decreases significantly. We also find that a relatively close relationship between the recipient and donor of a transfer, holding everything else constant, reduces the crowding out. One potential explanation for this heterogeneity is sharing obligations, which are known to be strong in developing countries and are recognised by the contemporary literature as being a barrier to economic growth (e.g. Grimm et al., 2013). The decrease in informal transfer networks due to formal insurance is found to be lowest in kinship networks; therefore, our findings raise the question of whether formal insurance can overcome this issue, at least in the short run. Our analysis of health-related outcomes suggests that the NHIS reduces respondents’ OOP expenditures, which is in line with our expectations. Overall, our findings indicate that public health insurance schemes strongly affect how healthcare services are paid for and may also support economic development in the long run. However, the results also emphasise the effects of new policies on existing institutions, which can be an important issue in many health policy contexts.
I use Prescription Drug Plan Formulary, Pharmacy Network, and Pricing Information Files from the Centers for Medicare and Medicaid Services to test for differences in benefit design for opioids across MA-PDPs and SA-PDPs. These data contain detailed informa- tion on the universe of Part D formularies and plans, including the list of covered drugs on each formulary, utilization management rules corresponding to each formulary-drug combination, and the level of cost-sharing associated with each plan-drug combination. For formulary-drug combinations that have a quantity limit restriction, the data contain both the quantity limit amount and the days limit; for example, 90 tablets within 30 days. I use data from 2008, 2009, 2011, 2013, and 2015. These data years represent a period of significant change in the medical community’s perception of safe opioid prescribing levels; between 2009 and 2011, researchers documented a heightened risk of adverse health events linked to high daily dosages of opioids, particularly among the elderly (Dunn et al., 2010; Saunders et al., 2010; Bohnert et al., 2011).
Empirical studies that directly examine the impact of MMR-autism controversy on vaccine uptake rates generally focus on a period before 2004, when scientific evidence reached a definite consensus. Using data in the U.S., Smith et al. (2008) posit that the influence of mainstream media on MMR immunization is limited by comparing temporal correlation between MMR non- uptake rate and newspaper coverage. Employing data from the U.K., Anderberg et al. (2011) find that the uptake rate of MMR declined faster in areas where a larger fraction of parents had stayed in education past the age of 18 than in areas with less educated parents. However, both of the studies fail to explain the trend of declining MMR vaccine use after 2004. According to the newspaper, such trend is driven by well-educated parents, the mechanism of which is not examined in previous studies. In order to answer this question, we use a longer study window, which starts from the very first year of the debate till a year after the initial paper was fully retracted. We focus on differential responses in immunization decisions for their children by parental education level when information is mixed.
Economic studies have investigated the effect of MMLs on prescription medication use (Bradford and Bradford, 2016), recreational marijuana use (Anderson et al., 2015; Pacula et al., 2015; Wen et al., 2015; Chu, 2014), use of other substances, such as alcohol (Wen et al., 2015; Anderson et al., 2013), tobacco (Choi et al., 2016), and hard drugs (Wen et al., 2015; Chu, 2015), labor market outcomes, such as labor supply (Nicholas and Maclean, 2016), earnings (Sabia and Nguyen, 2016), and sickness absences (Ullman, 2016), health outcomes, such as body weight (Sabia et al., 2017), opioid addictions and opioid overdose deaths (Powell et al., 2015), suicides (Anderson et al., 2014), and traffic fatalities (Anderson et al., 2013), and even seatbelt use (Adams et al., 2017). Overall, these studies suggest that MMLs increase recreational use of marijuana among adults but not teenagers, decrease alcohol, tobacco, and heroin, but not cocaine use, decrease BMI, suicides, and traffic fatalities, and improve labor market outcomes for older adults.
The first PDMP was established in California in 1939, and as the need to collect data on prescription drugs for law enforcement and monitoring purposes grew, eight more states established this program by 1989. In this period, which is called the “Paper Era” of the PDMPs, the information was mainly used by law enforcement agencies to curtail diversion. By 1990, the “Electronic Era” of the PDMPs began, which made the sharing of data easier between providers, pharmacists, and drug agencies. In the next decade, the steady rise in the abuse and diversion of controlled substances further increased the importance of PDMPs, and eventually there was a drive to align and consolidate the programs in different states, which so far differed vastly in regulations and implementation. Thus started the “Federal Era” of the PDMPs in 2002, when the the National Alliance for Model State Drug Laws (NAMSDL) drafted a model program outlining common goals that should be shared among existing and new PDMPs (Blumenschein et al. (2010)). As a result, PDMPs that were enacted in states after 2003 were very similar, and their enactment can be viewed as a natural experiment in contrast to the early PDMPs that were started in states with high abuse rates.
Yet, this previous literature on the elderly and the near-elderly does not provide direct evidence from which to draw inferences about the non-elderly adult population, particularly lower-income adults targeted for insurance expansion under the ACA. The ACA provides additional federal financing to states for extending Medicaid coverage to non-elderly adults earning less than 138 percent of the federal poverty level (FPL). The expansion decision was later delegated by the Supreme Court to states, and as of January 1, 2017, 31 states plus Washington DC had implemented the ACA Medicaid expansion, while the remaining 19 states had not. Medicaid is a means-tested health insurance program for low-income populations that is jointly administered by the federal and state governments. Since the creation of the Medicaid program in 1965, states have had broad discretion over a range of eligibility rules, program benefits, and provider reimbursement, subject to compliance with federal minimum standards. As a result, there has long been considerable variation in Medicaid eligibility standards and program generosity across states. Even though prescription drug coverage was a state option, all states covered pharmacological treatments prior to the ACA; following the ACA’s Medicaid expansion, new enrollees must be offered so-called “benchmark” benefits, including prescription drug coverage.
The present work is related to the literature that investigates the health effects of local house prices and foreclosures (e.g., Fichera and Gathergood (2013); Ratcliffe (2015); Tekin, McClellan, and Minyard (2015)). Three recent papers have considered the effects on health. Fichera and Gathergood (2013), using the British Household Panel Survey (BHPS) from 1991 to 2008, look at the effect of the local house price movements on health. They find that the increase in home equity reduces the likelihood of homeowners having different kinds of health conditions. They further show that health effects occur through two complementary channels: an increase in purchase of private medical insurance and an increase in physical activity and leisure as a result of a decrease in work hours. Ratcliffe (2015) also uses the BHPS and documents a positive correlation between house prices and the mental wellbeing of both homeowners and nonhomeowners which is inconsistent with the pure wealth mechanism. Tekin, McClellan, and Minyard (2015) investigate the impact of foreclosures on health using the data from four US states - Arizona, California, Florida, and New Jersey that have been among the hardest hit by the foreclosure crisis. They document that zip codes with increases in foreclosures are associated with the rise in the urgent and unscheduled hospital and emergency room visits during the period 2005 to 2009.
I found that while there was no statistically significant income-gradient in expenditures on health in most years during 1996–2015, there were statistically significant differences in the overall use of health care, especially when the poor are compared to the others. Most interestingly, I discerned a distinct age-related pattern in HCU by income: for children and adolescents, HCU positively correlated with family income, but for adults, it negatively and monotonically correlated with family income. Breaking down the overall utilization to its components showed that rich people used more office-based and dental care when they were children and adolescents, but poor people went with significantly less care until curative care became a necessity, hence they ended up in emergency rooms or in hospitals. 6 If one wishes to go beyond unidirectional relationships from socioeconomic status to health (J. P. Smith 1999 ), then it can be argued that health is self-productive, in the sense that investments in health capital not only determine health status but health-related ex-
Therefore, in chapter two, it has been shown how unemployment impacts risky behaviors. Since unemployment can be seen as a strong economic change, the results of this part of the thesis reveal evidence about the general health impact of negative economic alterations. In contrast to previous literature, the results yield no evidence for a negative impact of unemployment on risky behaviors. In particular, by using data from the GSOEP four different measures for risky behaviors has been analyzed (diet, alcohol consumption, physical activity, and smoking). While the analysis shows no evidence for an adjustment of the consumption of addictive goods after becoming unemployed, individuals alter their food consumption as well as their level of physical activity in a positive way. This finding can probably be traced back on a shift in the opportunity costs of time. Since the preparation of healthy food and a high level of sports consumption is typically time consuming, but generally not expensive, additional time can be used to foster such activities.
In this part, using a great deal of information about pregnancy and delivery methods in New Jersey Vital Statistics records, I try to figure out which mother is highly likely to manipulate her baby’s birth time. Then I directly test the relationship between birth timing manipulation and parents’ characteristics and estimate the health effects of birth timing manipulation on newborns. First of all, I choose the group of all low-risk term birth, because only parents who believe their babies are healthy in the uterus and will be healthy when delivered may consider manipulating their baby’s birth timing to slightly early days by reason of holidays at the end of the year. In other words, to exclude the effects of concerns about medical problems on decision making of delivery methods, I only keep term birth whose mothers have fewer health problems during pregnancy. To get this group, from entry sample I drop all births whose mothers have any medical risk factors, such as diabetes, renal disease, cardiac disease, genital herpes, incompetent cervix, for this pregnancy. I drop some birth with some complications of labor and/or delivery, such as prolonged labor (>20 hrs.), seizures during labor, breech, cephalopelvic disproportion, anesthetic complications fetal distress and etc. Then I keep birth whose mother may have the incentive to manipulate the birth timing. Incentive means if any day of holiday break is in expectant mother’s expected term delivery period (37 weeks to 41 weeks), this expectant mother may love to avoid holiday babies.
Information on smoking behavior is from the Current Population Survey To- bacco Use Supplements (CPS-TUS) for years 1998 to 2003 and 2006 to 2007. 8 The Current Population Survey is a nationally representative, monthly house- hold survey of labor force participation. The monthly survey often includes sets of supplemental questions on particular topic such as health, schooling, fertility, and immigration. Periodically, respondents are asked a series of questions about smoking and other tobacco-related behaviors as part of the Tobacco Use Supple- ment, which is sponsored by the National Cancer Institute and the Centers for Disease Control. The CPS-TUS is a repeated cross section of individuals useful for describing smoking behavior of Americans over time. Individuals who have smoked at least 100 cigarettes in their entire life are identified as current or for- mer smokers and are asked a series of follow-up questions about smoking behavior. The smoking measures used in this paper are self and proxy reports of smoking at least some days and smoking every day.