Applied econometrics 1

Einleitung / Introduction

The ongoing demographic ageing brings along health-related issues, for instance an increase in the number of people with cognitive impairment. Given that cognitive functioning is widely known to decrease with age, the number of individuals suffering from dementia is rising consequently. A potential source of cognitive decline in older ages may be incorporated in the duration of retirement. The reduction of cognitive engagement due to the end of employment in the occupation could be a possible mechanism. Hence, it is important to understand the mechanism of cognitive functioning and explore the effect of retirement duration as a potential factor on cognitive abilities.

Methode / Method

Our investigation focuses on the impact of retirement duration, measured in years, on the cognitive abilities of woman. We utilize data from the German National Cohort (NAKO), an ongoing examination that inspects medical evaluations, as well as sociodemographic and socioeconomic factors of individuals. Our cognitive ability measures include word recall, semantic fluency, and the stroop effect. To capture exogeneous variations in retirement, our identification strategy exploits the 1999 German pension reform that raised the early retirement age for women born after 1951 by three years from 60 to 63. This is important as the decision to retire is a personal choice and may also be influenced by the potential issue of reverse causality, specifically related to a declining mental health. We add new evidence to the literature by exploiting an inner-country reform affecting the retirement system of a single country, Germany.

Ergebnisse / Results

As a result of our analysis, we identify significant negative effects of an additional year spent in retirement on women’s cognitive abilities. This effect is primarily incorporated in the word recall test and the stroop effect. Our findings show that policy interventions that aim at postponing the transition into retirement to strengthen the workforce and the social security system may yield positive outcomes for the mental well-being of individuals.

Authors

Henrik Bergschneider, RWI - Leibniz-Institut für Wirtschaftsforschung

Hendrik Schmitz, Universität Paderborn, RWI - Leibniz-Institut für Wirtschaftsforschung

Matthias Westphal, Technische Universität Dortmund, RWI - Leibniz-Institut für Wirtschaftsforschung

Einleitung / Introduction

Health effects of retirement have been extensively studied, but the literature mainly considers own retirement and typically finds positive health effects (e.g. Frimmel and Pruckner (2020), Eibich (2015)). In this paper we study the effect of spousal retirement on health and health behavior. Spousal retirement may affect the health status through changes in lifestyle, risky health behavior, mental health or a change in health behavior, e.g. in preventative health behavior. Yet, there is another aspect of spousal retirement that did not receive a lot of attention so far: Retirement implies that newly retired spouses are available as caretakers. The impact of the availability of a caretaker on health care utilization is not well-understood so far.

Methode / Method

We study the causal effect of retirement on spouses’ health care utilization. We use administrative health register data from Upper Austria providing us with detailed individual inpatient and outpatient health care utilization information. To account for the endogenous retirement decision and to avoid reverse causality arising from spouses in need of care, we exploit exogenous variation in early retirement induced by two Austrian pension reforms in 2000 and 2003, which raised the eligibility age of early retirement stepwise for different quarter-of-births from 60 to 65 for men and 55 to 60 for women. This step-wise increase allows us to implement a fuzzy RD-design and estimate a short-run impact of spousal retirement on individual’s healthcare utilization.

Ergebnisse / Results

We find that wives’ retirement increases the probability of a dentist visit as well as health and cancer screening for their husbands. The availability of the wife as a caretaker also reduces the length of hospital stays on average by 1.1 days or 32 percent of a standard deviation. This effect is particularly driven by cardiovascular diseases associated with a higher need for care, such as strokes. Also, the probability of any hospitalisation due to mental and behavioral disorders (mostly due to alcohol intoxication and misuse) are reduced by 1.2 percentage points. In contrast, we do not find any significant effect for wives if their husband retires.

Zusammenfassung / Conclusion

Our findings show that retirement may lead to unexpected spillover effects on healthcare utilization which are distinct from (mental) health effects. The availability of a caretaker reduces uptake of health services and length of hospital stays, but may improve preventive and risky health behavior. This study contributes to a better understanding of consequences of retirement which are not necessarily beneficial for both gender. In times of labor shortages among hospital staff, we show that the availability of informal care may avert hospitals stays and reduce pressure on health care systems.

Authors

Wolfgang Frimmel, Johannes Kepler Universität Linz

Martina Zweimüller, Johannes Kepler Universität Linz

Einleitung / Introduction

Educational differences in health-related behavior are almost universal, yet there is no conclusive evidence why these disparities occur. Understanding and processing health-related information could be attributable to general cognitive ability rather than education as such. Health knowledge and health literacy are highly general in nature, and both reflect the capacity for unassisted learning and problem solving. Higher ability bolsters active and passive learning about health risks. This might explain why cognitive ability drives healthy behaviors over and above the connection via formal schooling.

Methode / Method

We use data from the Longitudinal study of Australian children (LSAC) to study whether various common indicators of general cognitive ability measured early in life (age 4, 6, 8 and 10) have an independent link with health behaviors, measured at age 16/17, such as teen smoking, drinking, healthy vs. unhealthy diet or not wearing a seatbelt. Generally, disentangling effects of schooling and cognitive ability is difficult. Our panel data allow us to deal with this difficulty in four complementary ways. First, we measure cognitive ability before or immediately after school entry. Second, all children are still in the compulsory stage of schooling. By controlling for age and grade level non-parametrically, we ensure that we estimate the link between cognitive ability and health behaviors adjusting for any effects of formal education and maturation. Third, LSAC provides information on a very large set of potentially important confounders, such as parental education and health behaviors, parenting styles, or peer quality, from which we select using double machine learning. Fourth, we compare standardized effect sizes of cognitive ability measured at different stages of childhood, i.e., with different numbers of years of formal schooling potentially acting as confounders. The scope and depth of our data set is essential, as any causal claim we make essentially rests on selection on observables.

Ergebnisse / Results

Our preliminary results show that, with the exception of alcohol consumption, adolescents‘ risky health behaviors are negatively linked with their cognitive ability measured up to ten years earlier. Effect sizes generally remain stable across measurements at different ages.

Zusammenfassung / Conclusion

Since we implicitly hold formal schooling constant, this suggests a genuine and quantitatively stable role of general cognitive ability in predicting risky health behaviors. Obviously general cognitive ability (net of education and maturation) cannot be easily targeted by policy. Rather, our findings suggest that knowledge-based interventions early in life might be particularly fruitful.

Authors

Hendrik Jürges, Universität Wuppertal

Rasheda Khanam

Einleitung / Introduction

The length of hospital stay is a main driver of hospital costs. Nonetheless, a shorter length of stay could potentially result in health decline and avoidable readmissions. Thus, the optimal length of stay is widely discussed. I study the impact of an additional day in hospital on mortality and healthcare usage.

Methode / Method

Since severe hospital cases go along with a longer length of stay in hospitals, the length of stay in hospitals in endogeneoes. I estimate causal effects of hospital stay on mortality and health care usage, by using an instrumental variable (IV) approach. Here, I use the state variation of German public holidays as an instrument for postponed discharge due to staff shortage during a public holiday. Hence, I compare individuals going to a hospital in a state with a public holiday in the upcoming week with individuals in hospitals without a public holiday in the upcoming week. I assume that the timing of admission in the pre-week is independent from the appearance of a public holiday in the upcoming week and show that the control and treatment group differ on observables only with respect to the state of the hospital and to the public holiday in the upcoming week.

Ergebnisse / Results

I document that a prolonged hospital stay leads to a reduction of the number of doctor cases and the use of rehabilitation clinics in the 30 days after discharge. In contrast, I do not find any impact on mortality or hospital readmissions in the same time period. I further analyse effect heterogeneities for different LTC modes. This subgroup is of significant interest, as for them an extra day in the hospital is associated with a higher risk of contracting a hospital-associated infection. The effects hold for individuals in nursing homes and for individuals who receive LTC at home. Additionally, I explore various primary diagnoses. The observed effects remain consistent for neoplasms and circulatory system diseases, but no significant effects are evident for primary diagnoses linked to injuries.

Zusammenfassung / Conclusion

My IV approach reveals that an additonal day in a hospital results in a reduction in the utilization of doctor services and rehabilitation clinics. However, no effects on hospital readmissions or mortality were observed.

Authors

Rebecca Leber, RWI Essen, Universität Duisburg-Essen