I examine whether working hours have a causal effect on the health behaviors of workers.
Y = alpha + beta * X vectors
health behaviors of workers = alpha + beta * working hours
In assessing the causal relationship, I estimate fixed-effects instrumental variable models by using exogenous variation in adopting a reduced workweek in South Korea as an instrument for work hours.
학력의 임금효과에 관한 실증연구는 일반적으로 횡단면 자료를 활용한 최소자승법 (OLS)나 도구변수(IV) 모형을 주로 이용. 그러나 이들 분석모형들은 일정한 한계가 있는 것으로 지적 (Card, 1999). 최소자승법은 학력의 내생성 문제로 추정계수의 편의를 야기. 이 모형에서 학력의 임금효과를 나타내는 추정계수는 학력의 내생성, 즉 학력과 개인의 미관찰 특성간에 상관관계가 있기 때문에 실질적인 임금효과보다 과대 추정될 수도 있고 과소 추정될 수도 있음. 횡단면 자료를 이용하여 학력의 내생성 문제를 해결하는 방법은 도구변수(IV) 모형을 활용. 이제까지의 실증연구들은 개인의 미관찰 이질성 중에서 능력에 의한 추정계수의 편의 문제를 해결하기 위해서 가족배경과 같은 도구변수들이 주로 사용. 그러나 도구변수 모형도 개인의 이질성에 의한 학력의 내생성 문제를 해결하는데 일정한 한계를 가짐. 개인의 미관찰 특성은 능력뿐만 아니라 지적욕구, 추진력, 인내력 등 다양하게 존재할 것이며, 이러한 특성은 학력수준과 상관관계를 가지고 있기에 추정계수의 편의를 야기시킴. 따라서 도구변수 모형으로 이러한 개인의 미관찰 이질성을 모두 통제할 수 없음.
내생성 문제를 해결하는 확실한 대안은 패널자료를 이용한 분석모형을 활용. 패널분석 모형을 이용한 학력의 임금효과를 추정하는데 유용하게 활용. 학력수준은 시간불변 변수이기 때문에 단순한 패널모형으로 학력의 임금효과를 제대로 추정할 수 없음. 임의효과 모형은 학력과 같은 설명변수와 개인의 미관찰 특성간에 상관관계가 없다고 가정한 모형이므로 내생성 문제를 해결하기 어려움. 반면, 고정효과 모형은 독립변수와 개인의 미관찰 특성이 상관관계가 있다고 가정한 모형으로 확실한 대안이 될 수 있음.
Hausman and Taylor (1981) 문제를 해결하기 위한 패널분석 모형 제시. 이들이 제시한 분석모형은 임의효과 모형에 도구변수를 포함한 모형으로 학력의 임금효과를 추정. 임의효과 모형에 포함된 설명변수를 내생적인 변수와 외생적인 변수로 구분하고 외생적인 변수를 도구변수로 활용하는 방법. Amemiya and MaCurdy (1986) 시간가변 내생적 변수들의 시기별 변수들을 도구변수로 활용하여 Hausman and Taylor 모형의 추정계수의 효율성을 더욱 향상시킨 모형을 제시.
본 연구는 학력의 임금효과에 관해 패널자료를 이용하여 엄밀하게 추정하는 것. Hausman and Taylor 의 모형과 Amemiya and MaCurdy 모형을 활용하여 개인의 미관찰 이질성과 학력과의 상관관계의 존재로 인해 발생하는 편의를 제거한 후 학력의 임금효과를 추정할 것.
The estimation results reveal that shortening work hours induces individuals to exercise regularly and decreases the likelihood of smoking, with more pronounced effects for heavy smokers. While a work-hour reduction substantially increases the probability of drinking participation, it does not significantly affect the likelihood of frequent or daily drinking habits.
In addition, the effect of a work-hour reduction on regular exercise is salient among women and older groups, and the effect on smoking behaviors is more pronounced among men and middle-aged groups.
1. INTRODUCTION
Working time has long been recognized as strongly linked with the health and well-being of working people, and it has therefore received attention from both researchers and policy makers (Bosch, 1999; Caruso, 2006). For example, long working hours have been shown to be associated with cardiovascular disease, high risk of diabetes, stress and poor mental health, and work-family conflicts (Caruso, 2006; Nakanishi et al., 2001; Sparks et al., 1997; Virtanen et al., 2012).
What are the effects of work hours on the workers' lifestyle habits that affect their health? Long working hours are associated with ill health and unhealthy behaviors such as smoking, heavy alcohol consumption, and lack of physical exercise (Maruyama and Morimoto, 1996; Siegrist and Rodel, 2006; Taris et al., 2011), which are important contributors to largely preventable chronic diseases. Studies have shown that workers compensate for overtime-related job stress by consuming more fatty and sweet food (Oliver and Wardle, 1999) and that long working hours act as a potential barrier to regular exercise by limiting the time available for nonwork activity (Schneider and Becker, 2005). Furthermore, job stress is believed to induce smokers to smoke more and to tempt those who have quit smoking to relapse because smoking is supposed to ease stress (Green and Johnson, 1990).
A related strand of research in economics emphasizes the role of the unemployment rate, another measure of economic activity, in influencing the health behaviors of individuals. Earlier studies used state-level data to examine the relationships between the local unemployment rates and the drinking and alcohol-related behaviors of workers (Ruhm, 1995; Freeman, 1999). To eliminate the unobserved local economic factors spuriously correlated with health-risk behaviors, these studies used fixed-effects models and found that alcohol consumption and drunk driving increase with economic activity. Several recent studies use individual-level data to exploit the within-person variations in health behaviors; they obtain mixed results. While some researchers find positive relationships between economic activity and alcohol consumption, physical inactivity, and smoking behaviors (Ruhm, 2000, 2005; Ruhm and Black 2002; Xu and Kaestner, 2010), others find evidence of the counter-cyclicality of drinking (Dee, 2001) or little evidence of the cyclicality of drinking, physical activity, and smoking behaviors (Charles and DeCicca, 2008).
Regarding the most recent economic downturn, using data from the American Time Use Survey, Aguiar et al (2013) document that the time spent on health care significantly increased during the Great Recession, whereas Tekin et al (2013) find little association between recent economic deterioration and health behaviors. Asgeirsdottir et al. (2012) show that the 2008 economic crisis in Iceland reduced the consumption of alcohol and cigarettes while also reducing the consumption of healthy foods.
Despite considerable evidence linking working time, economic activity, and health-risk behaviors, the causal effects of work hours are still unclear. In the absence of experimental evidence, it is difficult to assess whether workers' health behaviors are affected by working time or whether unobserved third factors such as attitudes toward smoking or drinking influence both working time and health-risk behaviors. Another possibility is reverse causality - that health habits and lifestyle choices influence employment and working time.
The goal of this paper is to investigate whether working hours have causal effects on workers' health behaviors. In assessing the causal relationship, I employ within-individual estimators (or individual fixed-effects (FE) models) to eliminate the unobserved factors that yield biased estimates. More importantly, I carry out estimation with fixed-effects models, using exogenous variation in policy adoption as an instrument for work hours. Specifically, I exploit the timing of the implementation of the legislated workweek reduction in South Korea based on establishment size. South Korea experienced a dramatic reduction in working hours during the past decade following the introduction of a 40-h workweek standard, providing a good source of variation in individuals' work time. The 40-h workweek limit was gradually adopted accordingly to establishment size from 2004 to 2011. In order to increase the robustness of the identification strategy in instrumenting work hours, I use the size data for the establishment at which each individual worked the year prior to the initial observation. This fixed-effects instrumental variable (FE-IV) estimator provides consistent results for the causal effects of work hours, addressing both individual heterogeneity and endogenous changes in working time. To the best of my knowledge, this paper is the first to provide evidence for the causal effects of working hours on health behaviors by controlling for both individual fixed factors and time-varying unobservables.
The estimation results suggest that a reduction in work hours leads to individuals' healthier lifestyles. Reducing work hours induces individuals to exercise regularly. A reduction in work hours also decreases the likelihood of smoking, with somewhat more pronounced effects for heavy smokers. While a work-hour reduction substantially increases the probability of drinking participation, it does not affect the likelihood of frequent or daily drinking habits. In addition, the results from a population subgroup analysis indicate that the effect of work-hour reduction on regular exercise is salient among women and older groups and that the effect on smoking behaviors is more pronounced among men and middle-aged groups.
2. INSTITUTIONAL BACKGROUND
South Korea's working hours have been among the highest in the industrialized world. However, South Korea has experienced the fastest decline in working hours among OECD countries over the past decade - the average annual working hours per worker decreased from 2512 in 2000 to 2090 in 2011. This dramatic reduction in working hours is mainly due to the stepwise introduction of a 40-h workweek standard implemented in 2004.
(...)
3. MODEL AND EMPIRICAL STRATEGY
3.1. Conceptual framework
(...)
Comments
Post a Comment