The following are the four most relevant studies for the present paper, al- though only the first three use mother fixed effects. Black et al. (2016) analyse birth order effects on adult health behaviours and health, with administrative data from Norway. They find that late born display higher smoking rates and lower self-assessed physical and mental health, although they score better than first born on body mass index and other biomarkers. Black et al. (2018) find that birth order negatively affects adult non-cognitive skills in a sample of Swedish men examined as part of the military enlistment. Later-born men are also less likely to be in jobs that require leadership and they score lower in all Big 5 per- sonality traits. Both cited studies find lower parental investments for higher birth order children, including behaviours such as smoking during pregnancy, breast- feeding and help with homework. A similar conclusion is reached by Lehmann et al. (2018), who use a cohort of children from the US (NLSY-C) to assess birth order effects on children cognitive and non-cognitive skills, as well as educational outcomes in adulthood. They find that lower parental investments, measured as antenatal care, behaviour in pregnancy, breastfeeding and cognitive stimula- tion, can account for the negative effect of higher birth order on cognitive skills in childhood, but find no effect on average childhood non-cognitive outcomes. Finally, Argys et al. (2006) are the only other economic study looking at adoles- cent risky behaviours. Using NLSY79, a longitudinal study of adolescents from the US, they find a positive association between higher birth order and smoking, use of marijuana, alcohol consumption and earlier initiation of sexual behaviour in adolescence. They estimate OLS regressions with birth order expressed by a dummy variable for having any older siblings. In the present paper, I use a fixed-effect strategy, netting out the effect of family size and other family-specific factors that are constant across groups of siblings. The strategy thus allows for a more robust estimation of the association between birth order, risky behaviours and non-cognitive skills for adolescents, a population that has not been examined before in studies using fixed-effect strategies. Institutionally, the ages considered correspond to the period going from the end of primary school to the end of compulsory schooling, covering a key stage of individual development.
risks in year t together and use the distribution of actual expenditures from year t+1 to generate beliefs about expenditure risk at the time the insurance plan was chosen in year t. Each person in the same risk group is assumed to have the same beliefs about his health risk. More specifically, I first group each insured individual into 60 different risk groups based on his age, sex, and severity score. I do not construct separate risk groups by year due to sample size limitations. For each of the 60 risk groups, I record the empirical proportion of individuals with zero expenditures the following year. For those with positive expenditures the following year, I fit a Weibull distribution to the observed expenditure, estimating the shape and scale parameters of this distribution. I exclude expenditures on preventive care since it was covered free of charge by all plans. I also only consider claims from in-network providers, which comprise nearly all spending. For each risk group, I then construct a “modified” Weibull distribution using the group’s estimated shape and scale parameters for positive expenditure and the empirical probability of zero expenditures. For each risk group, I take 100 draws from their corresponding modified Weibull distribution. Within each family, I sum the expenditures for each draw so that each family has 100 draws corresponding to the sum of expenditures for each of its members from that particular draw. This statistical object represents the family’s beliefs about its total expenditure risk. The family’s out-of-pocket risk is then constructed by applying the insurance plan’s de- ductible to each expenditure draw. In this setting, out-of-pocket payments simply equal spending if below the deductible, and the deductible if spending exceeds the deductible.
Despite dissipation of some rents through marketing costs, it is plausible that in- surers also capture part of the increased benchmarks. Figure 3 shows dramatic increases in stock prices for the four publicly traded health insurers with the most MA enrollment (Humana, United, Cigna, and Aetna) as a result of a surprisingly large increase in bench- marks on April 1, 2013. Interestingly, it is Humana, the most active insurer in the Medicare Advantage market from Table 3, with the biggest increase. A simple pre-post comparison of market capitalization for these four …rms, which accounted for about 44 percent of MA enrollment at the time of the policy change, indicates a market capitalization increase of approximately $2.7 billion. The announced benchmarks represented an increase of approx- imately 5.6 percent relative to what otherwise was speci…ed by legislation. Multiplying this percentage by our estimate of baseline MA revenues for each insurer (calculated by multiplying enrollment weighted benchmarks for each insurer by the average risk score of its enrollees) yields an estimated increase in annual MA revenue of about $2.9 billion.
One major advantage of the PSID is that I can identify households that contain an individual with a pre-existing health condition that would result in denial of non-group health insurance in denial states. Because the definition of a pre-existing condition is insurer specific, several definitions have been offered in the literature and are outlined in a report by the Government Accountability Office Government Accountability Office (2012a). I construct two measures based on these definitions and work by Zellers, McLaughlin, and Frick (1992), who survey insurance companies to identify conditions that result in denial. The first is a narrow definition of pre-existing conditions set to one if the head answers yes to the question “Has a doctor ever told you that you/[your wife] have or had one of the following conditions,” which include heart disease, cancer, heart attack, or stroke. I also construct a broader measure of pre- existing conditions using the same question but expanding the conditions to include high blood pressure, diabetes, lung disease, emotional problems, memory loss, and obesity. By the narrow measure, 12 percent of households contain a person with a pre-existing condition and by the broad measure, 27 percent of households contain a person with pre-existing condition. For comparison, the Government Accountability Office estimates that the fraction of the adult population with a pre-existing condition is between 20 and 66 percent (Government Accountability Office 2012a).
Does providing parents Medicaid coverage benefit children? Numerous studies exist on the relationship between children’s health insurance status and their health care utilization and health outcomes. Covering children has improved children’s ac- cess to care and furthered their health outcomes (Currie and Gruber 1996a; Dafny and Gruber 2005; Howell et al. 2011). However, even among insurance-covered chil- dren, there exists a disparity in health outcomes by socioeconomic status (Currie and Thomas 1995; Case 2000). Thus, the role of parents in determining children’s health outcomes draws the attention of researchers. Increased awareness of the importance of family-coverage has led researchers to examine the relationship between parental coverage and children’s coverage and health care access. More children are enrolled in Medicaid in states with expanded Medicaid for parents (Lambrew 2001; Dubay and Kenny 2004; Wolfe et al. 2006; Aizer and Grogger 2003; Sommers 2006). Children with insured parents have better access to health care than children with uninsured parents (Gifford et al. 2005; Guendelman 2006). However, the literature has not provided a clear answer on the relationship between children’s health outcomes and parental coverage.
The third wave of centres forms the basis of much of the empirical work that follows. Following the 2007 Darzi report, the Department of Health set up a new policy known as Equitable Access to Primary Care (EAPC). The twin aims of the policy were to improve access to primary care in the most under-doctored areas of the country, and to deliver more personalised and responsive care across England. To this end Ministers announced £250 million of new annual funding to support the establishment of 100 new GP practices in the 38 Primary Care Trusts (PCTs) with the lowest per capita GP provision, and additionally required each of the 152 PCTs to establish one new GP-led health centre. The new services were to be commissioned through competitive procurements. The policy background provides grounds to suggest these centres should form a relatively homogeneous group both in terms of the specification of services as well as the characteristics of the locations where they were sited. The centres had to offer a regular registered GP practice service as well as walk-in services for any member of the public from 8am until 8pm, 7 days a week, 365 days a year. Core criteria set out in policy documents also required them to be located in areas that maximised convenient access to services and opportunities to colocate and integrate with other local services (Department of Health, 2007).
Given modest website usage, the effect on equilibrium prices may be larger if more consumers are informed about health care prices. There are two factors that make it dif- ficult to extrapolate from reduced-form estimates. First, even though the availability of the website is exogenous, use of the website when it is available is potentially endoge- nous. If the individuals who find out about the website and choose to use it are those that receive a larger benefit, there may be decreasing savings as more individuals be- come informed about prices. Second, equilibrium prices are a function of the number of consumers that have price information. By affecting negotiated prices, price trans- parency generates spillover effects that benefit all consumers, including those that do not have price information. 6 By using the demand model from Chapter 2 and estimating the individual-specific probability of using the website and deriving a bargaining equa- tion, the empirical model presented in this paper allows me to address both issues when examining out-of-sample counterfactual scenarios.
Community health centers (CHCs) are an important type of the health care safety net provider, along with public hospitals and physicians that provide free or discounted care to the poor. CHCs specialize in the provision of outpatient primary and preventive health care services, and serve patients regardless of their insurance status and ability to pay. The majority of CHCs are non-profit organizations, but some are public. The state of California started issuing permits to this type of the provider in the 1930’s (Saviano, 2009). Today, these clinics depend on patient revenue – which mainly comes from Medicaid 1 and Medicare 2 reimbursements – as well as government grants and contracts, foundation grants, and charitable contributions. 3 In 2002, the CHC sector began a period of expansion due to increased levels of federal funding under Section 330 of the Public Health Service (PHS) Act, a federal law that governs many aspects of health care in the United States. This initiative was authorized by the George W. Bush Administration with the intent of decreasing health disparities in the U.S. 4 More than half of California’s CHCs today receive Section 330 grants, which confer them the status of federally qualified health centers (FQHCs).
The German health insurance system is characterized by the coexistence of SHI and PHI. The fact that the majority of the German population is insured under the SHI – most of them mandatory – while certain subgroups of the population, such as civil servants, the self-employed and high earners, may opt out for substitutive PHI has been the subject of intense discussions in terms of both efficiency and fairness. Potential positive selection into the PHI, mainly due to risk-adjusted health premiums under the PHI, may lead to undesirable market outcomes in the German health care market. In addition, privately insured are often considered as privileged because of different and possibly better medical treatments. Although the regulatory framework and related incentives clearly suggest who will prefer which type of insurance, the empirical evidence is still inconclusive. In chapter 2 we investigate deter- minants that led individuals to switch the type of insurance using data from the SOEP for the years of 1997-2010. Besides socioeconomic characteristics and health risks, which are suspected to be the main drivers in this decision, we also analyze the role of previously unconsidered personality traits, such as risk aversion or altruistic attitudes, that could affect the decision to opt out of the SHI. Applying a haz- ard model in discrete time and accounting for potential endogeneity of self-reported health through an IV approach, the estimation results yield robust evidence on the choice of health insurance type that is consistent with pragmatic decision making, with both incentives set by regulation and personality traits as relevant determinants. For instance, the SHI is preferred by individuals who benefit from free insurance coverage of dependents or those who would have to pay high risk-adjusted premiums un- der the PHI, i.e. bad health risks. This advantageous selection in favor of the PHI may further increase pressure on the SHI which is already severely affected by demographic change. In addition, we also find convincing empirical evidence for the notion that risk-loving individuals have a higher proba- bility of buying PHI, however, we observe no significant effect of the measure of altruism. Overall the results suggest that the choice between both systems seems to be a less emotional issue and, hence, policy debates should focus more on how to design a framework that foster more competition between both systems.
For this analysis I focus on mothers between ages 18 and 44. In an ideal dataset, family income would be reported for each birth record. However, as the Vital Statistics data do not report income, I examine effects among mothers with low education, as well as mothers whom I expect to respond more to the policy change due to other demographic characteristics such as birth parity and marital status. The standard birth certificate items in the US during the study time period were changed in 1989 and 2003. As states implemented the 2003 birth certificate revisions over a staggered timeline, 38 states and the District of Columbia complied in 2012, while in 2015 (the last year in our study period) only two out the 50 states were non-compliant. While data elements from both 1989 and 2003 versions of birth certificates were included in the files, unrevised variables that were not comparable with 2003 revisions were not recorded (National Center for Health Statistics, 2010). Among the Vital Statistics variables used in this study, only birth counts, mother’s age and parity of birth are comparable across birth certificate revisions of 1989 and 2003. Maternal educational attainment is only reported in states adopting the 2003 revisions over all or part of our sample period. As a result, for models stratified by maternal educational attainment, to obtain a balanced panel of states with data available 8 quarters before and after the policy change, I trade off geographic coverage against panel length. 1 I
Parallel to Hymer, in the 60s Vernon developed the Product life cycle theory (Vernon, 1966). The theory states that a firm’s location, and therefore its foreign expansion, is determined by the stage of the product life cycle that it is in. Vernon identifies three stages of product development. In the first stage (the introduction), the firm (in Vernon’s case it is a US entrepreneur) introduces a new product into the market. At this point, the firm’s location and targeted market coincide. Moreover, the firm faces low price elasticity of demand and it is mainly concerned with maintaining the freedom to adjust inputs and the ability for prompt and effective communication. In the second stage (growth), the product demand expands and the market becomes more standardized. In the third and final stage (standardization), the product becomes completely standardized and requires processes with high capital intensity and unskilled labour. Because of this, firms will tend to relocate abroad mainly on the basis of cost considerations. Hence, the product life cycle theory stresses that firms make direct foreign investments only after products mature and competition becomes cost-based. Like all theories, Vernon’s was a product of his time and his ideas were based on the experience of American firms relocating abroad. Vernon mainly analysed products that once in the maturity stage are characterized by a high level of standardization, such as textile products, steel, electronic items and so on. The changes in economic environment from the 1950s to the 1970s and 1980s (for example, the narrowing of macroeconomic differences between the US and Europe) made his theory less applicable. Despite the limited applicability of Vernon’s theory, Ietto-Gillies (2005) argues that it may still be useful for analysing the innovation of small firms and the spread of innovation from developed to developing countries.
Table 28 reports the marginal effects of these characteristics. OLS estimation suggests that the HMO enrollees receive 0.06 more prescriptions compared with EPO enrollees. PPO enrollees get 0.2 more prescriptions per year compare with EPO enrollees. Administrative Service Only (ASO) is another characteristic of insurance plans. In insurance plans with ASO, employers have more freedom in managing claims and benefits, which results in 0.12 fewer prescriptions for patients. Consumer Driven Health Plan (CDHP) is a code that defines different payment arrangments. It can be a Health Reimbursement Arrangement (HRA) or a health saving arrangement (HSA) or neither. In the HRA plans, the employers’ fund does not accumulate in a separate fund, and the employer only pays after the employee incurs expenses. Insurance plans with HRA reduce the number of prescriptions by about 0.15 compared with HSA or not having any payment plans. Calculation of marginal effects from NB2 model results in larger effects. The first column in the table shows that HMO and PPO significantly increase the number of visits by 0.22 and 0.61 in comparison with EPO. Indemnity (IND), Point of Service (POS) and other plans also increase the number of visits, but not significantly. Access service only plans reduce visits by 0.47. HRA plan holders visit doctors 0.44 and 0.66 less compared with HSA and people with no consumer access. This estimation suggests that we could significantly reduce the number of visits and consequently total consumption of opioids by intervening in the insurance market. A plan that leads to lower demand for opioids should provide a structure similar to EPS, ASO, and HRA in the market for opioid medications.
(iii) Control variables: In line with previous studies, I include the following control variables: age, gender, race and ethnicity, marital status, education and income levels. Group average is used for income since individual income is likely to be simultaneously determined with health status (Ruhm, 2005). 15 The BRFSS reports the income (in $) in the ranges: less than 10,000, 10,000-14,999, 15,000-19,999, 20,000-24,999, 25,000-34,999, 35,000-49,999, 50,000-74,999, and 75,000 and over. For the estimation purpose, respondent’s income is assumed to be at the midpoint of each range and 150 % of the top category and is then converted to 2012 year dollars using the all-items of CPI-U of the Bureau of Labor Statistic (BLS) (Ruhm, 2005). The weighted average incomes are calculated for 36 demographic groups stratified by sex (male versus female), age (25-29, 30-34, 35-39, 40-44, 45-49, and 50-55), and education (less than high school, high school or some college, and college) (Tekin, McClellan, and Minyard, 2015). I also control for local macroeconomic condition to alleviate the concern that it might be driving the relationship between local house prices and health (Ruhm, 2005). Monthly county-level unemployment rate from the BLS Local Area Unemployment Statistics database is used as a proxy for local macroeconomic condition . 16 In addition, I control for stock market condition using
Generally, the results of these studies find Local Average Treatment Effects estimates in excess of the OLS estimates. Crespo et al. (2014) use the SHARElife data—the third wave of the SHARE panel study which contains retrospective life history data—to shed light on the early-life characteristics of those individuals whose behaviour was altered by the implementations of the CSLs, and who may have different returns to schooling than the wider population. The authors find larger effects of the reforms on years of education for those with lower socio-economic status (measured by reports of living in a dwelling with two or fewer rooms), and those reporting better childhood health status. Overall, their results show large, positive effects of extra schooling on memory and depression. Strati- fiying the sample by early-life characteristics revealed some variation in point estimates of the causal effect of education, although these differences were not statistically significant.
The market for higher education has received significant attention in the economics literature. In particular, the effects that subsidized loan policies have on the de- mand side of the market have been widely studied, given the dramatic increase in student debt during the last two decades in the U.S. Overall, there seems to be a consensus in the literature on the fact that credit constraints explain only a small fraction of enrollment decisions in higher education in the U.S. However, this is not necessarily the case in developing countries, where student financial aid systems are weak and evidence suggests that college enrollment is highly determined by family wealth (World Bank, 2003, 2012). In this context, the implementation of subsidized student loan policies can potentially affect the demand for education, generating equilibrium effects such as increases in tuition prices and changes in the quality of services offered by colleges.
Over the past several decades there has been an increased prevalence of both adult and childhood obesity in American society. This has received significant attention from researchers and governmental agencies, as there are both public health concerns and financial consequences associated with this condition. Obesity is associated with an increased risk for heart disease, stroke, type-2 diabetes, arthritis, and certain forms of cancer—all of which are serious (and costly) conditions. It has been estimated that medical expenditures associated with obesity account for 9% of all national medical expenditures in a given year. Nearly half of this bill is paid for by the government through Medicare and Medicaid expenditures, which translates to a higher tax bill for all tax-paying citizens. Moreover, it has been estimated that obese individuals who are Medicare (Medicaid) participants cost on average $1,486 ($864) more per year than their healthy weight counterparts (Finklestein et al. 2003).
The stock of food available to children could have decreased as a result of the war-induced internal displacement of people and loss of life because these factors led to loss of crops, livestock, productive inputs and assets, especially for rural farmers (Akresh et al., 2012b). In addition to these indirect effects, the war has a direct impact on health and education infrastructure, thereby affecting relevant environmental inputs to child growth as well as health and education outcomes (Lai and Thyne, 2007). For instance, it is documented that on 5 June 1998, the Eritrean Air Force bombed Ayder School in Mekele (the capital city of the war-affected region) killing twelve children. Most, if not all, destructions of this kind took place in war-affected region, damaging physical infrastructure and significantly interrupting and decreasing the humancapital accumulation process.
Exogenous increases in parental income lead to greater investments in children and improve- ments in childhood outcomes. A number of recent studies attempt to address concerns about endogeneity in estimating the effects of exogenous changes in family income on children. Dahl and Lochner (2012) exploit expansions of the Earned Income Tax Credit (primarily over the mid-1990s) to estimate the effects of additional family income on cognitive achievement. Their instrumental variable estimates suggest that an additional $1,000 in family income raises combined math and reading scores by 6 percent of a standard deviation. Estimated effects also appear to be larger for children from more disadvantaged families. Milligan and Stabile (2011) estimate that expansions of child tax benefits in Canada led to similar improvements in child cognitive and educational outcomes as well as improvements in child and maternal health. Combining data from ten welfare and anti-poverty experiments, Duncan, Morris, and Rodrigues (2011) attempt to separately identify the effects of changes in family income from employment and other effects induced by different programs. Their analysis reaches similar conclusions regarding the impacts of income on child achievement as Dahl and Lochner (2012) and Milligan and Stabile (2011). Finally, Løken (2010) and Løken, Mogstad, and Wiswall (2012) estimate the impact of family income on Norwegian children using regional variation in the economic boom following the discovery of oil as an instrument for income. The latter study estimates that income has sizeable impacts on education and IQ for children from low-income families but much weaker effects for children from higher income families. 18
The next wave of the IFLS surveys was released early spring in 2009. With this data, we are able to follow children born during the crisis into the early stages of the education system. However, even with that data, we are not able to directly analyse the full e¤ect of the crisis on educational- and labour market outcomes as the crisis children are still too young. Inference to the potential future consequences of the crisis might be drawn using research from other settings. The medical journal Lancet recently published an overview of research on long-term consequences of child stunting (Victoria et al., 2008). The paper summarizes the results of …ve cohort studies from Brazil, Guatemala, India, the Philippines and South Africa and provides new estimates pooling the data from these separate studies. Few papers from a developing country setting follow children from birth to adulthood. Alderman et al. (2004) investigates the relationship between preschool stunting and humancapital formation in Zimbabwe. The aim is to link the loss in stature at early childhood with adult or adolescent outcomes, particularly education. This demands a lot of the data. We need early childhood height measurements and to follow children over a long period of time. There are not many papers using data that ful…l these criteria, but the two mentioned above do and will serve as the basis for what follows. The evidence used here is not going to be perfectly compatible to an Indonesian setting, but will still give us some idea of how the …nancial crisis could have long-term e¤ects working through humancapital.