Economic impact of obesity in the Republic of Korea

Chung, Wankyo Working Paper Economic impact of obesity in the Republic of Korea ADBI Working Paper, No. 755 Provided in Cooperation with: Asian Development Bank Institute (ADBI), Tokyo Suggested Citation: Chung, Wankyo (2017) : Economic impact of obesity in the Republic of Korea, ADBI Working Paper, No. 755, Asian Development Bank Institute (ADBI), Tokyo This Version is available at: https://hdl.handle.net/10419/179211 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/3.0/igo/ ADBI Working Paper Series ECONOMIC IMPACT OF OBESITY IN THE REPUBLIC OF KOREA Wankyo Chung No. 755 July 2017 Asian Development Bank Institute The Working Paper series is a continuation of the formerly named Discussion Paper series; the numbering of the papers continued without interruption or change. ADBI’s working papers reflect initial ideas on a topic and are posted online for discussion. ADBI encourages readers to post their comments on the main page for each working paper (given in the citation below). Some working papers may develop into other forms of publication. ADB recognizes “Korea” as the Republic of Korea. Suggested citation: Chung, W. 2017. Economic Impact of Obesity in the Republic of Korea. ADBI Working Paper 755. Tokyo: Asian Development Bank Institute. Available: https://www.adb.org/publications/economic-impact-obesity-republic-korea Please contact the authors for information about this paper. Email: [email protected] Wankyo Chung is an associate professor at the Graduate School of Public Health, Seoul National University. The views expressed in this paper are the views of the author and do not necessarily reflect the views or policies of ADBI, ADB, its Board of Directors, or the governments they represent. ADBI does not guarantee the accuracy of the data included in this paper and accepts no responsibility for any consequences of their use. Terminology used may not necessarily be consistent with ADB official terms. Working papers are subject to formal revision and correction before they are finalized and considered published. Asian Development Bank Institute Kasumigaseki Building, 8th Floor 3-2-5 Kasumigaseki, Chiyoda-ku Tokyo 100-6008, Japan Tel: +81-3-3593-5500 Fax: +81-3-3593-5571 URL: www.adbi.org E-mail: [email protected] © 2017 Asian Development Bank Institute ADBI Working Paper 755 W. Chung Abstract Obesity, defined as abnormal or excessive fat accumulation that may impair health, has a clearly measurable impact on health and health-related quality of life, and generates considerable economic burden. This study builds on previous research by examining the available evidence on obesity, estimating obesity-related medical costs using more representative and reliable data, and estimating obesity-related risk of disability in the Republic of Korea. The obesity rate is higher for males than for females and for the self-employed (including their dependents) than for employed adults (including their dependents) in the Republic of Korea. The obesity rate shows a weak U-shaped pattern in relation to income for employed adults. It has been growing over time, especially among younger adults and adolescents, indicating greater obesity in the future. Obesity (relative to having a BMI <25) is shown to be associated with about 5,000 won (in 2010 won) higher medical costs for male adults and about 77,000 won (in 2010 won) higher medical costs for female adults. Severe obesity (relative to having a BMI <30) increases medical costs far more than obesity for both males and females, indicating higher cost effects for the few who are severely obese. Moreover, obesity is shown to have increased medical costs far more for relatively unhealthy individuals at higher percentiles of medical costs. When these estimated effects of obesity in increasing medical costs were used to estimate obesityrelated aggregate medical costs in the Republic of Korea, they were 35.8 billion won (in 2010 won) for males and 306.5 billion won (in 2010 won) for females. Furthermore, obesity is positively associated with disability, indicating another cost of obesity. Therefore, obesity is associated with a significant economic burden in terms of medical costs and disability in the Republic of Korea. JEL Classification: I12, I18 ADBI Working Paper 755 W. Chung Contents 1. INTRODUCTION ......................................................................................................... 1 2. TRENDS OF OBESITY ............................................................................................... 2 3. RISK FACTORS .......................................................................................................... 8 4. OBESITY-RELATED MEDICAL COSTS ..................................................................... 9 5. OBESITY-RELATED DISABILITY ............................................................................. 15 6. CONCLUSION .......................................................................................................... 15 REFERENCES ..................................................................................................................... 17 APPENDIX ............................................................................................................................ 20 ADBI Working Paper 755 W. Chung 1. INTRODUCTION The Republic of Korea experienced a growth in real GDP from 27.3 trillion won in 1960 to 1,463.8 trillion won (at 2010 GDP price) in 2015. Although the population increased simultaneously from 25 million to 50.6 million, real per-capita GDP increased from about 1.1 million won to 28.9 million won (see Appendix Table 1). Due to this economic growth, the Republic of Korea’s per-capita GDP was 34,510 US$ (using purchasing power parity) in 2015, and was ranked 22nd out of 35 OECD countries (OECD 2016).1 Demographically, the share of the elderly aged 65 and over increased from 2.9% in 1960 to 13.1% in 2015. This change reflects the decline in fertility from 6 in 1960 to 1.2 (children per women aged 15 to 49 years old) in 2014 and the increase in life expectancy at birth from 52.4 years in 1960 to 82.2 years in 2014 (OECD 2016). Aside from economic and demographic changes, national health insurance was introduced in 1977 and expanded to universal coverage in 1989. Driven by these changes, real per-capita health expenditure increased significantly from 56,116 won (2010=100) in 1970 to 2,000,531 won in 2014 and thus the share of national health expenditure in GDP increased from 2.67% in 1970 to 7.07% in 2014 (OECD 2016). As one of the essential factors contributing to the rising medical costs and challenging financial stability of the national health insurance scheme, overweight and obesity are defined as abnormal or excessive fat accumulation that may impair health. In 2014, more than 1.9 billion adults (39% of the world’s adults) aged 18 years and older were overweight (BMI>=25) and over 600 million adults (13%) were obese (BMI>=30) (WHO 2016a). Compared with other countries, the Republic of Korea shows a lower level of overweight and obesity. In 2014, the Republic of Korea ranked 55th out of 192 countries with 33.5% of adults aged 18 years and older being overweight (BMI>=25, age-standardized estimate) (WHO 2016b). However, the Republic of Korea’s prevalence of overweight and obesity rose from 25.8% in 1998 to 31.5% in 2014 for adults aged 19 and over. Overweight and obesity are one of the leading global risks for mortality in the world, responsible for 4.8% of deaths globally (WHO 2009). The WHO estimated them to be the third-highest risk factor among high-income countries including the Republic of Korea (those with 2004 gross national income per capita in excess of US$10,066), responsible for 8.4% of deaths. Obesity also has a clearly measurable impact on physical and mental health, and health-related quality of life, and generates considerable direct and indirect costs (Dixon 2010). Many studies have estimated the economic burden of obesity (Finkelstein et al. 2009; Thorpe et al. 2004; Cawley and Meyerhoefer 2012). The most recent such paper used an instrumental variable estimation to address both the endogeneity of weight and measurement error in weight, and showed that the effect of obesity on medical care costs is much greater than that estimated previously (Cawley and Meyerhoefer 2012). For example, while Finkelstein et al. (2009) estimated that annual medical spending on the obese was $1,429 (in 2008 dollars or 41.5%) higher than that on healthy-weight individuals, Cawley and Meyerhoefer (2012) showed that obesity increased annual medical costs by $2,741 (in 2005 dollars). 1 The 35 OECD countries in the order of per-capita GDP in 2015 were Luxembourg, Norway, Switzerland, the United States, Ireland, the Netherlands, Austria, Australia, Germany, Sweden, Denmark, Iceland, Canada, Belgium, Finland, the United Kingdom, France, New Zealand, Japan, Italy, Spain, the Republic of Korea, Israel, the Czech Republic, Slovenia, Portugal, the Slovak Republic, Estonia, Greece, Poland, Hungary, Latvia, Chile, Turkey, and Mexico. 1 ADBI Working Paper 755 W. Chung Even though few studies have focused on the economic burden of obesity in the Republic of Korea, four appear to be important to mention. Jee et al. (2006) showed that the relative risk of death is higher for those with a higher BMI, especially deaths from atherosclerotic cardiovascular disease or cancer. Similarly, Hong et al. (2015) showed that a higher BMI increased the hazard ratio of death, especially vascular mortality. Lee et al. (2012) estimated that medical costs due to 16 obese-related diseases were 2,128 billion won in the Republic of Korea, 4.6% of the total cost of national health insurance in 2011. Lee et al. (2015) divided obesity into obesity (25<=BMI<30) and severe obesity (30<=BMI) and estimated their respective relative risk of incurring diseases (27 diseases for males and 31 diseases for females) in relation to normal weight (18.5<=BMI<23). The estimated obesity-related medical costs were 1,142 billion won (in 2013 won) for males and 1,952 billion won (in 2013 won) for females. This study builds on previous research by examining the available evidence on obesity, estimating obesity-related medical costs using more representative and reliable data, and estimating obesity-related risk of disability. A better and more reliable estimation of the economic burden of obesity will help us to develop and prioritize appropriate health interventions to reduce obesity. 2. TRENDS OF OBESITY Obesity is a state of excessive body fat accumulation, which is difficult to measure. Body mass index [BMI, defined by weight (kg)/height (m)2] has been used traditionally for its simplicity and data availability. Although shortcomings of BMI have been acknowledged, its correlation with the percentage of body fat and sensitivity in diagnosing obesity based on the percentage of body fat have been verified for citizens of the Republic of Korea (Chung, Park and Ryu 2016). Obesity of infants aged 4–71 months is calculated based on those who took a health examination between 2012 and 2015. There were between 1.64 and 2.04 million infants approximately for each year from 2012 to 2015. For infants aged from 4 to 24 months (three groups of 4–6, 9–12, and 18–24 months), overweight alone was defined by being greater than or equal to the 95th percentile of weight-for-height. For infants aged from 30 to 71 months (four groups of 30–36, 42–48, 54–60, and 66–71 months), overweight was defined as 1.04<=BMI z-score <1.65 and obesity was defined as 1.65<=BMI z-score (NHIS 2016). Overweight rates were 8.8, 9.0, 8.9, and 8.6% and obesity rates were 2.8, 2.7, 2.8, and 2.8% for each year from 2012 to 2015. When infants were divided by gender, overweight and obesity rates were higher for female than for male infants. For male infants, overweight rates were 8.0, 8.2, 8.1, and 8.4% and obesity rates were 2.6, 2.5, 2.6, and 2.7% for the respective years. For female infants, overweight rates were 9.8, 9.8, 9.6, 8.9% and obesity rates were 2.9, 2.9, 3.1, and 2.9 % (NHIS 2016). Interestingly, the obesity rate increased with months among infants, for example, from 3.0% (30–36 months) to 5.1% (42–48 months), 5.9% (54–60 months), and 6.9% (66–71 months) in 2015. And it declined with household income, proxied by ventiles of insurance premium (see Figure 1 and Appendix Table 2). The obesity rates were 3.7% at the first ventile but declined sharply to 2.6% at the 13th ventile and further down to 2.5% at the 20th ventile. When infants were divided by gender, the same declining pattern appeared, while the obesity rates of female infants were higher than those of male infants by about 0.1–0.3 percentage points across the ventiles of insurance premium. 2 ADBI Working Paper 755 W. Chung Figure 1: Distribution of Infant (30–71 Months) Obesity across Income in 2015 Source: National Health Insurance Service (2016). Child obesity among children aged 6–18 years was defined by BMI (>=25 or >=95 percentiles for each age) and calculated based on data from students’ health examinations at primary, junior high, and high schools (see Table 1).2 The obesity rates of students at primary schools (6 years of education) increased by 1 percentage point from 8.0% in 2009 to 9.0% in 2015. The obesity rates of students at junior high schools (3 years of education) increased by 1.2 percentage points from 12.7% to 13.9% and those of students at high schools (3 years of education) increased by 3.8 percentage points from 15.7% to 19.5% during the same period. In each year, the obesity rates increased with ageing from students at primary schools to those at junior high schools and high schools. In 2015, for example, the respective rates were 9.0%, 13.9%, and 19.5%. And male students showed higher obesity rates than female ones. In 2015, male versus female obesity rates were 9.5% and 8.5% for students at primary schools, 17.0% and 10.5% for those at junior high schools, and 23.3% and 15.3% for those at high schools. Adult obesity is defined as a BMI of 25 or more based on the suggestion by the Korean Society for the Study of Obesity, and understanding that those from the Republic of Korea develop negative health consequences at a lower BMI. Data from the Korea National Health and Nutrition Examination Survey (KNHANES) are used to describe the trends of obesity, where measured weight is available to reduce bias due to measurement errors, for the total noninstitutionalized civilian population in the Republic of Korea (Kim 2014). 2 Sample size ranges from 82,581 (756 schools) to 194,065 students (749 schools). 3 ADBI Working Paper 755 W. Chung Table 1: Child Obesity Rate (%) by Schools (6–18 years old, 2009–2015) Year Primary School Junior High School High School 2009 8.0 12.7 15.7 2010 8.3 12.6 16.3 2011 8.5 12.6 15.8 2012 9.0 13.2 16.5 2013 9.0 13.7 17.5 2014 8.9 13.5 18.2 2015 9.0 13.9 19.5 Source: Data from students’ health examination at primary, junior high, and high schools. In 2014, 31.5% of adults aged 19 and over were obese. Males showed a higher obesity rate of 37.7% than females at 25.3%. Obesity increased with age from 23.9% for those aged 19–29 to 31.8% for those aged 30–39, 31.1% for those aged 40–49, 35.4% for those aged 50–59, 36.8% for those aged 60–69, and 32.1% for those aged 70 and over. Figure 2 shows that adult obesity increased over time from 25.8% in 1998 to 31.5% in 2014 (see Appendix Table 3). This increase is driven by changes among males rather than females, because while male obesity increased from 25.7% to 37.7%, female obesity fluctuated between 25.9% and 25.3%. When the sample is limited to younger adults from 19 to 29 years old in Figure 3 (see Appendix Table 4), both males and females show an increasing trend during the period, pointing to greater obesity in the future. Obesity in younger adults increased from 15.2% in 1998 to 23.9% in 2014, with both male and female obesity increasing, from 19.3% to 32% and from 11.6% to 15%, respectively, during the period. Figure 2: Trends of Adult Obesity from 1998 to 2014 (age>=19, BMI>=25) Source: The Korea National Health and Nutrition Examination. 4 ADBI Working Paper 755 W. Chung Table 3: Descriptive Statistics of Male Adults (age>=20, 2009–2013) Mean Std. Dev. Min Max Cost (1,000 won, 2010=100) 514.2 1,834.6 0 114,213.3 Cost>0 0.996 0.065 0 1 Disability (grade 1–6) 0.053 0.224 0 1 BMI 24.39 3.115 12.76 62.3 Being obese (=1 if BMI>=25) 0.397 0.489 0 1 Being severely obese (=1 if BMI>=30) 0.045 0.207 0 1 Weight (KG) 71.22 10.667 30 162 Height (M) 1.71 0.063 1.06 1.99 Smoking (current or former) 0.708 0.455 0 1 Drinking (>=2 days per week) 0.438 0.496 0 1 Exercise (>=2 days per week, >=20 min) 0.296 0.457 0 1 Age (20~24) 0.022 0.148 0 1 Age (25~29) 0.087 0.282 0 1 Age (30~34) 0.118 0.323 0 1 Age (35~39) 0.111 0.314 0 1 Age (40~44) 0.166 0.372 0 1 Age (45~49) 0.13 0.336 0 1 Age (50~54) 0.152 0.359 0 1 Age (55~59) 0.103 0.304 0 1 Age (60~64) 0.111 0.314 0 1 No. of Obs. 209,403 Source: National Health Insurance Service-National Sample Cohort (NHIS-NSC) database. Similarly, Table 4 presents descriptive statistics for female adults (206,927 individuals). Their medical costs (including 0 medical cost) amounted to about 597,000 won (in 2010) and 99.8% of them were spent on medical care. Note that the medical costs were greater among females on average. The average BMI was 23.02, and 24.6% of adult females were obese as defined by BMI>=25, much lower than the male obesity rate of 39.7%, and 3.5% of adult females were severely obese as defined by BMI>=30, still lower than the male severe obesity rate of 4.5%.5 The average female weight was 57.29 kg and the average height was 1.58 m, and these are correlated by 0.27 (p=0.00), weaker than their correlation for males. Some 6.4% of adult females were current or former smokers, 12% drank alcohol more than twice a week, and 20.9% exercised more than 2 days per week and more than 20 minutes per day. There were more females aged 40–54 (49.6%), but fewer females aged 20–29 (13.6%) and aged 30–39 (13.3%). 5 For comparison, the prevalence of obesity (as defined by BMI>=30) was 32.2% among US adult men (aged 20 or over) and 35.5% among US adult women in 2007–2008, while the prevalence of overweight (as defined by BMI>=25) was 72.3% among US adult men and 64.1% among US adult women (Flegal et al. 2010). 11 ADBI Working Paper 755 W. Chung Table 4: Descriptive Statistics of Female Adults (age>=20, 2009–2013) Mean Std. Dev. Min Max Cost (1,000 won, 2010 년=100) 596.6 1,689 0 95,196 Cost>0 0.998 0.041 0 1 Disability (grade 1–6) 0.03 0.17 0 1 BMI 23.02 3.391292 12.19 50.84 Being obese (=1 if BMI>=25) 0.246 0.431 0 1 Being severely obese (=1 if BMI>=30) 0.035 0.183 0 1 Weight (KG) 57.29 8.617 26 141 Height (M) 1.58 0.057 1.1 1.91 Smoking (current or former) 0.064 0.244 0 1 Drinking (>=2 days per week) 0.12 0.325 0 1 Exercise (>=2 days per week, >=20 min) 0.209 0.406 0 1 Age (20~24) 0.045 0.208 0 1 Age (25~29) 0.091 0.287 0 1 Age (30~34) 0.074 0.262 0 1 Age (35~39) 0.059 0.237 0 1 Age (40~44) 0.189 0.392 0 1 Age (45~49) 0.129 0.335 0 1 Age (50~54) 0.178 0.383 0 1 Age (55~59) 0.109 0.312 0 1 Age (60~64) 0.125 0.331 0 1 No. of Obs. 206,927 Source: National Health Insurance Service-National Sample Cohort (NHIS-NSC) database. Table 5 presents marginal effects of obesity on medical costs using OLS estimation. For males, obesity (relative to having a BMI <25) does not lead to higher medical costs in a statistically significant way. However, severe obesity (relative to having a BMI <30) increases medical costs by about 39,000 and is statistically significant at 5%. For females, obesity increases medical costs by about 100,000 won and severe obesity increases medical costs by 228,000 won. Both are statistically significant at 1%. Note that the obesity effect is estimated while controlling for other related factors: smoking, drinking, exercise, nine age category dummies, 17 area dummies, five year dummies, 10 insurance premium dummies, and four categories of insurance-type dummies (the employee insured, the dependents of the employee insured, household head of the self-employed insured, other householders of the self-employed insured).6 Among other factors, smoking increases medical costs for both males and females (statistically significant at 1% both for males and females), and exercise decreases medical costs for both males and females (statistically significant at 10% for males and 1% for females). However, drinking decreases medical costs for both males and females (statistically significant at 1% both for males and females), requiring further study of self-selection bias and the types of alcoholic drink and alcohol concentration in addition to the drinking frequency. 6 The categories of the reference groups are: youngest age group of 20 to 24, area of Seoul, year of 2009, the 1st decile of health insurance premium (proxy for income or wealth), and household head of the self-employed insured. 12 ADBI Working Paper 755 W. Chung Table 5: Marginal Effects of Obesity on Medical Costs (age>=20, 2009–2013, in 2010 won) OLS TPM (Logit, GLM [Gamma, log]) Male Female Male Female Obesity –10.631 99.657*** 5.376 77.268*** (7.831) (10.001) (5.975) (7.416) Severe Obesity 39.486** 227.845*** 57.891*** 169.117*** (16.643) (26.848) (15.980) (19.399) No. of Obs. 209,403 209,403 206,927 206,927 209,202 209,202 204,663 204,663 Adjusted for smoking, drinking, exercise, nine dummies for age, 17 dummies for area, five dummies for year, 10 dummies for insurance premium, and four dummies for insurance type. Robust standard errors in parenthesis, * significant at 10%, ** significant at 5%, *** significant at 1%. Source: National Health Insurance Service-National Sample Cohort (NHIS-NSC) database. While the OLS results suggest higher medical costs for obesity, we use a two-part model to confirm the results (Jones 2000). The first part of the two-part model estimates the probability of positive medical costs and the second part estimates the level of medical costs conditional on positive costs. We use a logit mode for the first part and a Gamma Generalized Linear Model (GLM) with log link for the second part, following Cawley and Meyerhoefer (2012). The GLM can provide consistent estimation of obesity effects while the OLS-based model for logged medical costs provides inconsistent estimation unless the degree and form of heteroscedasticity are known to retransform the estimates (Manning and Mullahy 2001). The second panel of Table 5 presents marginal effects of obesity, reflecting both parts of the two-part model.7 For males, obesity (relative to having a BMI<25) increases medical costs by about 5,000 won and severe obesity (relative to having a BMI<30) increases medical costs by about 58,000 won. Again, only the latter is statistically significant at 1%. For females, obesity increases medical costs by about 77,000 won and severe obesity increases medical costs by about 169,000 won. Both are statistically significant at 1%. Severe obesity (relative to having a BMI<30) increases medical costs far more than obesity (relative to having a BMI<25) for both males and females, indicating higher marginal cost effects for the few that are severely obese. These estimates can be used next to estimate the aggregate medical cost of obesity in the Republic of Korea. For male adults aged 20 to 64, the aggregate medical cost of obesity was estimated to be 35.8 billion won [in 2010 won, =0.397 (obesity rate in Table 1)*16,716,667 (average number of males aged 20 to 64 from 2009 to 2013)*5,400 won (marginal effect of obesity in Table 5)] and the aggregate medical cost of severe obesity was 43.6 billion won (=0.045*16,716,667*57.9) on average from 2009 to 2013. For female adults aged 20 to 64, the aggregate medical cost of obesity was estimated to be 306.5 billion won (in 2010 won, =0.246*16,117,986 [average number of females aged 20 to 64 from 2009 to 2013]*77.3) and the aggregate medical cost of severe obesity was 95.4 billion won (=0.035*16,117,986*169.1) on average from 2009 to 2013. 7 Note that there are few individuals with zero medical cost both for males and females because the medical cost data in the NHIS-NSC database is obtained from the insurance claims. Note also changes in the number of observations due to the exclusion of observations with all positive values of medical costs in an area for the first part and exclusion of observations with zero cost for the second part. 13 ADBI Working Paper 755 W. Chung When compared between males and females, the aggregate medical cost of female obesity is much higher than that of male obesity—8.55 times from 35.8 to 306.5 billion won—due to a higher marginal effect of obesity on female medical costs. Even for the aggregate medical cost of severe obesity, females show higher values than males—2.19 times from 43.6 to 95.4 billion won. These estimated aggregate costs are much lower than those estimated by Lee et al. (2015), who used an epidemiological approach. 8 For the purpose of comparison, under the strict assumption that our estimated marginal effect of obesity based on adults aged 20 to 64 generalizes to the whole population, we used the average population size from 2009 to 2013 and inflated the estimated costs to 2013 won by the medical care component of the Consumer Price Index. The recalculated aggregate medical cost of obesity was 55 billion won for males and 486.3 billion won for females. These values were still lower than the 1,142 billion won for males and 1,952 billion won for females in Lee et al. (2015). We next examine the differential marginal effect of obesity across the distribution of medical costs in Table 6. For males, obesity increases medical costs by 2.7, 7.8, 14.8, 23.0, and 33.7 thousand won (in 2010 won) at the 10th, 25th, 50th, 75th, and 90th percentiles of medical cost. For females, obesity increases medical costs by 4.1, 9.7, 23.9, 64.4, and 216.8 thousand won (in 2010) at the respective percentiles of medical cost. All are statistically significant at 1%. Therefore, obesity increases medical costs far more for relatively unhealthy individuals at higher percentiles of medical cost. Similarly to the results in Table 3, females show a higher marginal effect of obesity at each percentile of medical cost.9 Table 6: Marginal Effects of Obesity at Different Percentiles of Medical Cost (age>=20, 2009–2013, in 2010 won) 10th 25th 50th 75th 90th Male (209,403) Obesity 2.728*** 7.813*** 14.770*** 23.026*** 33.700** (0.332) (0.599) (1.205) (3.361) (13.118) Severe Obesity 3.481*** 8.058*** 20.824*** 40.693*** 126.338*** (0.754) (1.417) (2.863) (7.888) (31.214) Female (206,927) Obesity 4.134*** 9.699*** 23.877*** 64.407*** 216.844*** (0.654) (0.997) (1.845) (4.757) (16.302) Severe Obesity 6.279*** 17.218*** 42.518*** 112.823*** 445.886*** (1.472) (2.306) (4.236) (10.611) (37.030) Adjusted for smoking, drinking, exercise, nine dummies for age, 17 dummies for area, five dummies for year, 10 dummies for insurance premium, and four dummies for insurance type. Standard errors in parenthesis, * significant at 10%, ** significant at 5%, *** significant at 1%. Source: National Health Insurance Service-National Sample Cohort (NHIS-NSC) database. 8 Aside from the methodology, Lee et al. (2015) used a different sample, year, and source of data. 9 Although not shown in the table, severe obesity increases medical costs by 3.5, 8.1, 20.8, 40.7, and 126.3 thousand won in 2010 at the respective percentiles of medical cost for males, and it increases medical costs by 6.3, 17.2, 42.5, 112.8, and 445.9 thousand won at the respective percentiles of medical cost for females. All are statistically significant at 1%. 14 ADBI Working Paper 755 W. Chung 5. OBESITY-RELATED DISABILITY We further examine the association between obesity and disability to provide evidence on another cost of obesity. Lakdawalla, Bhattacharya, and Goldman (2004) showed not only higher rates of disability among the obese but also higher rates of disability growth among the obese than among the nonobese. Burkhauser and Cawley (2004) found some evidence that obesity increases the probability of health-related work limitations, and Howard and Potter (2014) showed that obesity is related to higher rates of worker illness absence. The NHIS-NSC database provides data on moderate and severe disability (grade 1~6). Out of 209,403 male adults aged 20 to 64, 5.3% had a disability (0.67% having a severe disability + 4.64% having a moderate disability), while out of 206,927 female adults aged 20 to 64, 3% had a disability (0.41% having a severe disability + 2.57% having a moderate disability). The binary outcome of having a disability is used in a logistic estimation, controlling for the same risk factors used in the tables above, such as smoking, drinking, exercise, nine age categories, 17 area dummies, five year dummies, 10 insurance premium dummies, and four categories of insurance-type dummies. Table 7 presents the odds ratio results obtained from logistic regression models. For males, the adjusted odds ratio between obesity and disability was 1.032 (95% confidence interval: 0.992–1.074), indicating a statistically insignificant positive association between the two. However, the odds ratio between severe obesity and disability was 1.173 (95% confidence interval: 1.066–1.291), indicating a statistically significant positive association between the two. For females, the adjusted odds ratio was 1.434 (95% confidence interval: 1.359–1.513) between obesity and disability and it was 1.975 (95% confidence interval: 1.790–2.179) between severe obesity and disability, which were both statistically significant. Similarly to medical costs, disability shows a stronger association with obesity or severe obesity among females than males. Table 7: Effects of Obesity on Disability (Odds Ratio from logistic regression, age>=20, 2009–2013) Male Female Obesity 1.032 1.434*** (0.021) (0.039) Severe Obesity 1.173*** 1.975*** (0.057) (0.099) No. of Obs. 209,403 209,403 206,927 206,927 Adjusted for smoking, drinking, exercise, nine dummies for age, 17 dummies for area, five dummies for year, 10 dummies for insurance premium, and four dummies for insurance type. Robust standard errors in parenthesis, * significant at 10%, ** significant at 5%, *** significant at 1%. Source: National Health Insurance Service-National Sample Cohort (NHIS-NSC) database. 6. CONCLUSION The infant obesity rate was 2.8% in 2015 and it was higher for female infants. It increased with months among infants but declined with household income, proxied by ventiles of insurance premium. Child obesity increased with age and over time from 2009 to 2015, and male students presented a higher obesity rate than female ones. 15 ADBI Working Paper 755 W. Chung The obesity rate is higher for male than for female adults and for the self-employed (including their dependents) than for employed adults (including their dependents). The obesity rate shows a weak U-shaped pattern in relation to income for employed adults. It has been growing over time, especially for males and younger adults, indicating greater obesity in the future. After selecting a reliable data and estimation model, obesity (relative to having a BMI <25) is shown to be associated with about 5,000 won (in 2010 won) higher medical costs for male adults and about 77,000 won higher medical costs for female adults. Severe obesity (relative to having a BMI <30) increases medical costs far more than obesity for both males and females, indicating higher cost effects for the few that are severely obese. Moreover, obesity is shown to have increased medical costs far more for relatively unhealthy individuals at higher percentiles of medical costs. When the estimated effects of obesity in increasing medical costs were used to estimate obesity-related aggregate medical costs in the Republic of Korea, they were 35.8 billion won (in 2010 won) for males and 306.5 billion won for females. Furthermore, obesity is positively associated with disability, indicating another cost of obesity. Therefore, obesity is associated with a significant economic burden in terms of medical costs and disability in the Republic of Korea. However, to the extent that the failure to treat obesity as endogenous leads to underestimation of the links between obesity and medical costs (Cawley and Meyerhoefer 2012), and between obesity and disability (Burkhauser and Cawley 2004), our estimated economic burden of obesity can be underestimated. Without question, a better and more reliable estimation of the economic burden of obesity will help us to develop and prioritize appropriate health interventions to reduce obesity. The OECD (2010) found that interventions such as health education and promotion, regulation and fiscal measures, and counseling in primary care are all effective in tackling obesity and have favorable cost-effectiveness ratios relative to a scenario where chronic diseases are treated only as they emerge. While the concerned ministries (especially the Ministry of Health and Welfare and the Ministry of Education) and laws (about 25 laws) have introduced many interventions to improve diets and increase physical activity in the Republic of Korea, they are assessed as being uncoordinated and focused on children (Kim et al. 2009; Lee et al. 2015). For example, the Ministry of Health and Welfare provides budget support for local governments’ obesity programs, develops educational materials and publicizes them, and provides vouchers for management services related to the physical activity and diet of obese children. The National School Lunch Act, introduced in 1981, has provisions on school dietitians, nutritional requirements, and dietary consultation. However, it is noteworthy that combining interventions in a multiple-intervention strategy would provide an affordable and cost-effective solution, significantly enhancing health gains relative to isolated actions (OECD 2010). 16 ADBI Working Paper 755 W. Chung REFERENCES Burkhauser R.V. and J. Cawley (2004), Obesity, disability, and movement onto the Disability Insurance rolls, University of Michigan Retirement Research Center, WP 2004-089. Cawley, J. and C. Meyerhoefer (2012), The medical care costs of obesity: an instrumental variables approach, Journal of Health Economics, 31. Chung W., C.G. Park, and O. Ryu (2016), Association of a new measure of obesity with hypertension and health-related quality of life, PLoS One, 11(5): e0155399. Dixon, J.B. (2010), The effect of obesity on health outcomes, Molecular and Cellular Endocrinology 316(2):104–8. Finkelstein E.A., J.G. Trogdon, J.W. Cohen, and W. Dietz (2009), Annual medical spending attributable to obesity: payerand service-specific estimates, Health Affairs, 28(5): w822–31. Flegal K.M., M.D. Carroll, C.L. Ogden, L.R. Curtin (2010), Prevalence and trends in obesity among US adults, 1999–2008, Journal of the American Medical Association 303(3):235–41. Flegal K.M., O.A. Panagiotou, and B.I. Graubard (2015), Estimating population attributable fractions to quantify the health burden of obesity, Annals of Epidemiology, 25: 201–7. Hong S., S. Yi, J.W. Sull, J.S. Hong, S.H. Jee, and H. Ohrr (2015), Body mass index and mortality among Korean elderly in rural communities: Kangwha Cohort Study, PLoS One, 10(2): e0117731. Howard J.T. and L.B. Potter (2014), An assessment of the relationships between overweight, obesity, related chronic health conditions and worker absenteeism, Obesity Research & Clinical Practice, 8(1):e1–e15. Jee S.H., J.W. Sull, J. Park, S. Lee, H. Ohrr, E. Guallar, and J.M. Samet (2006), Body-mass index and mortality in Korean men and women, New England Journal of Medicine, 355:779–87. Jones A. (2000), Health Econometrics, Chapter 6 in Anthony J. Culyer and Joseph P. Newhouse, editors, Handbook of Health Economics, Volume 1A, pp. 265–336. Kim H., Y. Kang, N. Kwark, E. Kang, and E. Kim (2009), Trends in Obesity and Comprehensive Policy Strategy to Prevent Obesity in Korea, research paper 2009–07 (in Korean), Korea Institute for Health and Social Affairs. Kim Yuna (2014), The Korea National Health and Nutrition Examination Survey (KNHANES): Current Status and Challenges, Epidemiology and Health, 36:e2014002. Kong I.G., H.J. Lee, S.Y. Kim, S. Sim, and H.G. Choi (2015) Physical activity, study sitting time, leisure sitting time, and sleep time are differently associated with obesity in Korean adolescents: a population-based study. Medicine 94(44):e1965. Lakdawalla D.N., J. Bhattacharya, and D.P. Goldman (2004), Are the young becoming more disabled? Health Affairs, 23(1): 168–176. 17 ADBI Working Paper 755 W. Chung Lee H.J., J. Cho, H. Lee, C. Kim, and E. Cho (2014a) Intakes of Dairy Products and Calcium and Obesity in Korean Adults: Korean National Health and Nutrition Examination Surveys (KNHANES) 2007–2009, PLoS One 9(6): e99085. Lee H.S., S.O. Kwon, M. Yon, D. Kim, J.Y. Lee, J. Nam, S.J. Park, J.Y. Yeon, S.K. Lee, H.Y. Lee, O.S. Kwon, and C.I. Kim (2014b), Dietary total sugar intake of Koreans: based on the Korea National Health and Nutrition Examination Survey (KNHANES), 2008–2011., Journal of Nutritional Health, 47(4):268–276. Lee S., Y. Yoon, G. Hyun, and E. Lee, (2012), How to tax on health risk factors to obtain financial stability of National Health Insurance, research paper (in Korean), 2012-08, Health Insurance Policy Institute, National Health Insurance Service. Lee S., Y. Yoon, J. Baek, G. Hyun, and H. Kang, (2015), Socioeconomic impact of major health risk factors and evaluation of their regulatory policy, Research paper (in Korean), 2015-1-0009, Health Insurance Policy Institute, National Health Insurance Service. Lee, J., J.S. Lee, S.H. Park, S.A. Shin, , and K. Kim, (2016), Cohort Profile: The National Health Insurance Service-National Sample Cohort (NHIS-NSC), South Korea. International Journal of Epidemiology; doi:10.1093/ije/dyv319. Lovejoy J.C., A. Sainsbury, and the Stock Conference Working Group (2009), Sex differences in obesity and the regulation of energy homeostasis, Obesity Reviews, 10(2): 154–67. Ma S.Y., H.J. Park, and E.H. Kong (2015), The relation between the frequency of family gathered meals and the Body Mass Index-Standard Deviation Score in Elementary Students, The Korean Journal of Obesity, 24(1):51–8. Manning W.G. and J. Mullahy (2011), Estimating log models: to transform or not to transform? Journal of Health Economics, 20(4): 461–94. National Health Insurance Service (2016), 2016 A White Paper on Obesity, Wonju Korea. OECD (2010), Obesity and the Economics of Prevention: Fit not Fat, OECD Publishing, Paris. OECD (2016), OECD.Stat, (database) (accessed 11 January 2017). Oh C., B.H. Jeon, S.N.R. Storm, S. Jho, and J. No (2017), The most effective factors to offset sarcopenia and obesity in the older Korean: physical activity, vitamin D, and protein intake, Nutrition 33:169–73. Rockhill B., B. Newman, and C. Weinberg (1998), Use and misuse of population attributable fractions, American Journal of Public Health, 88:15–19. Rowe A.K., K.E. Powell, W.D. Flanders (2004), Why population attributable fractions can sum to more than one, American Journal of Preventive Medicine, 26(3):243–249. Thorpe K.E., C.S. Florence, D.H. Howard, and P. Joski (2004), The impact of obesity on rising medical spending, Health Affairs, 23(6):W4480–4486. World Health Organization (2009), Global Health Risks: Mortality and Burden of Disease Attributable to Selected Major Risks (WHO Press, Geneva). 18 ADBI Working Paper 755 W. Chung World Health Organization (2016a), Fact Sheet on Obesity and Overweight, updated June 2016. (WHO Press, Geneva). World Health Organization (2016b), Global Health Observatory data repository, updated February 2015. (WHO Press, Geneva). You J and J. Choo (2016), Adolescent overweight and obesity: links to socioeconomic status and fruit and vegetable intakes, International Journal of Environmental Research and Public Health, 13(3). 19 ADBI Working Paper 755 W. Chung APPENDIX Appendix Table 1: Economic, Demographic and Health Expenditure Trends of the Republic of Korea (1960–2015) Year Real GDP (2010=100), Million Won Real per-capita GDP (2010=100), Won Population, Thousands Share of the Elderly, % Real per-Capita Health Expenditure, Won (2010=100) Health Expenditure Share of GDP 1960 27,305,008 1,091,660 25,012 2.9 1961 29,184,235 1,132,679 25,766 2.9 1962 30,304,612 1,143,008 26,513 3 1963 33,089,704 1,213,778 27,262 3 1964 36,219,772 1,294,296 27,984 3 1965 38,821,707 1,352,453 28,705 3.1 1966 43,473,916 1,476,918 29,436 3.1 1967 47,437,372 1,574,372 30,131 3.1 1968 53,693,715 1,741,137 30,838 3 1969 61,501,442 1,949,687 31,544 3 1970 67,649,998 2,098,271 32,241 3.1 56,116 2.67 1971 74,722,595 2,272,398 32,883 3.2 52,901 2.33 1972 80,065,800 2,389,638 33,505 3.1 50,690 2.12 1973 91,937,605 2,695,869 34,103 3.2 49,053 1.82 1974 100,635,699 2,900,811 34,692 3.2 45,758 1.58 1975 108,549,204 3,076,728 35,281 3.5 72,558 2.36 1976 122,785,601 3,425,123 35,849 3.5 81,349 2.38 1977 137,860,801 3,786,158 36,412 3.6 85,325 2.25 1978 152,714,602 4,130,862 36,969 3.7 100,909 2.44 1979 165,887,197 4,419,624 37,534 3.7 140,231 3.17 1980 163,064,999 4,277,252 38,124 3.8 148,918 3.48 1981 174,773,901 4,513,410 38,723 3.9 163,250 3.62 1982 189,218,997 4,811,507 39,326 4 171,992 3.57 1983 214,275,501 5,368,914 39,910 4 186,132 3.47 1984 236,652,102 5,856,862 40,406 4.1 193,026 3.30 1985 254,991,800 6,248,919 40,806 4.3 208,286 3.33 1986 283,612,301 6,881,510 41,214 4.4 220,792 3.21 1987 318,970,998 7,663,576 41,622 4.5 239,490 3.13 1988 356,943,603 8,492,339 42,031 4.7 276,874 3.26 1989 382,035,701 8,999,867 42,449 4.8 336,308 3.74 1990 419,518,104 9,785,984 42,869 5.1 360,583 3.68 1991 462,954,800 10,692,858 43,296 5.2 376,932 3.53 1992 491,544,604 11,235,829 43,748 5.4 409,982 3.65 1993 525,199,404 11,883,784 44,195 5.5 423,309 3.56 1994 573,549,998 12,847,899 44,642 5.7 435,799 3.39 1995 628,442,203 13,936,583 45,093 5.9 480,030 3.44 1996 676,169,298 14,852,807 45,525 6.1 538,996 3.63 1997 716,213,295 15,585,582 45,954 6.4 562,974 3.61 1998 677,027,701 14,626,892 46,287 6.6 548,545 3.75 continued on next page 20