Empirical Result international conference elgic 2014

Education and Leadership in Glocalization : What does “think globally, act locally” mean for education around the world? 21-24 2014 243 and pays for it by a tax on healthy wealthy, t 2 , in the second period also. Following Rosenman 2008, the social utility function is determined as the weighted sum of agent expected utility. Let   V V , be the relative weights placed on the wealthy and poor people utility, respectively. If such relative weights are equal for all agents then this optimal is Pareto optimal. Thus, the social planner would choose   h h t G , , , 2 to maximize the social utility function 18 subject to the fair insurance market 19 , the high income people’s resource constraint 20 and the low income people’s resource constraint 21. 1 1 1 2 1 2 2 2 2 2 2 2 2 2 2 2 , , , , [ , , , , ] 1 , , , , H L H L H L U W F F E C V n p h U W F F E C t p h U W F F E C s v                                               1 1 1 2 1 2 2 2 2 2 2 2 2 2 2 , , , , , , , , 1 , , , , H L H L H L U W F F E C V n p h U W F F E C p h U W F F E C s v G                                                18     n h p G n h p t 1 2  19 1 2 1 2 1 2 1 2 1 2 2 H L H L W W F F F F E E C C h t                20 1 2 1 2 1 2 1 2 1 2 H L H L W W F F F F E E C C h G                21 Define  as the Lagrange multiplier on fair insurance market,  as Lagrange multiplier on the high income people’s resource constraint the marginal u tility of high income people’s wealth, and  as the Lagrange multiplier on the low income people’s resource constraint the marginal utility of wealth of poor people. Thus the necessary first order condition for interior solution with respect to   h h t G , , , 2 , respectively, and yields the social optimal condition as follows:             G W t W U V h p U V U V h p U V     1 1 2 1 22                               G W t W U h p U V U h p U V 1 1 2 1 23 1 2 2 2 2 2 2 2 2 2 2 2 2 , , , , , , , , H L H L W U W F F E C t h U W F F E C s v t p h U V                         24 1 1 2 2 2 2 2 2 2 2 2 2 , , , , , , , , H H H L H L F F U W F F E C h U W F F E C s v G Gp h U W V                         25 The expression 23 states that the social planner would choose t 2 and G which satisfies the socially optimal condition. That is, the weighted marginal utility of weight of the wealthy people divided by the probability of being healthy in period 2, plus the weighted expected marginal utility of tax of the wealthy people equal to the weight marginal utility of weight of the poor people divided the probability of being sick in the second period of poor people, plus the weighted expected marginal utility of subsidy of the poor people. In other words, the marginal cost of policy is equal to the marginal benefit of the policy if the marginal utility of weight of the wealthy people and poor people are equal to zero. Equations 24 and 25 show that the marginal utility of healthy wealthy and poor people in the first period in the social optimality are larger than the marginal utility with free choice due to the externality, so the socially optimum spending on healthy lifestyle choice of wealthy and poor people   h h , exceed h .

5. Empirical Result

Table 2 presents the descriptive statistics of the body mass index, income and exercise. The data for studying come from the first, second, third and forth reports on the National Health Examination Survey of Thailand conducted by the Ministry of Public Health during 1991-1992, 1996-1997, 2003-2004, and 2008- 2009. An average BMI of 23.37 kmm 2 is overweight with standard deviation of 2.84.The maximum and minimum BMI are 31.5 and 19.3, respectively. The monthly mean individual income is equal to 4,618.61 baht with standard deviation of 1,940.71. The maximum and minimum incomes are 8,906.00 and 1,991.00 baht, respectively. Deficiency exercise has a mean of 28.49 with standard deviation of 14.87. It has maximum deficiency exercise of 59.6 and minimum deficiency exercise of 8.9. Table 2 Summary of statistics for BMI, Income and Deficiency Exercise. Education and Leadership in Glocalization : What does “think globally, act locally” mean for education around the world? 21-24 2014 244 BMI INCOME Deficiency Exercise Mean 23.36667 4618.611 28.49778 Median 23.00000 3948.500 24.52500 Maximum 31.50000 8906.000 59.60000 Minimum 19.30000 1991.000 8.900000 Std. Dev. 2.840050 1940.715 14.86976 Skewness 1.354945 0.787200 0.595271 Kurtosis 5.305849 2.731873 2.185927 Jarque-Bera 9.495336 1.912969 1.560080 Probability 0.008672 0.384241 0.458388 Sum 420.6000 83135.00 512.9600 Sum Sq. Dev. 137.1200 64028378 3758.866 Observations 18 18 18 Cross sections 6 6 6 In addition, data from a survey of 2,500 samples in over 50 administrative districts of Bangkok in 2012 showed the following characteristics: most samples were female — 52.2, and the rest are male: 47.8. 97.7 of samples were aged 13 to 75, 70.3 of those were singles, and 63.1 of those were studying. Most samples had gained diploma or bachelor degree — 76.8, currently working — 70.3, employees — 65.2, income between 10,001 to 20,000 baht — 42.2, members of families of between 1 to 11 persons — 97.2, and carried the burden of caring for a family of between 0 to 7 persons — 62.9. In particular, most samples were highly obese. In fact, an average of currently BMI is 37.10 kgm 2 with standard deviation of 7.64 kgm 2 . The average BMI last-year was 36.77 kgm 2 with standard deviation of 7.60 kgm 2 . A mean BMI three years ago was 35.76 kgm 2 with standard deviation of 7.46 kgm 2 . A mean BMI five years ago was 35.14 kgm 2 with standard deviation of 8.19 kgm 2 . Samples’ activities in their spare time was usually using computers or mobiles for surfing the internet, chatting, watching television, watching movies, etc. 72.5. In contrast, most samples chose to get some exercise 63.5 when they responded the question, “If you have only one choice to take care of your health, what choice do you choose?” The second choice was having healthy food 30.8. The mean of monthly spending on expensively healthy food was 19.60 of income with standard deviation 19.59. When we asked the question, “How much would you prefer to spend on risk alleviation?”, most respondents were willing to spend 5 monthly 32.1. Besides, the risk-mitigating spending from getting sick, which was less than or equal to 5 per month 37.8. In addition, such spending which is between 6-10 per month was 32.1. The 61.8 of respondents paid for costs of medical care via the right of social welfare such as social security, rights of government officials, and health insurance. 18.5 of respondents paid for one by themselves. Furthermore, what if the government collects income tax from high-income people in order for curative care of the lowest-income people less than 2,910 baht per month each? 42.2 of respondents agreed to income tax of  5. And if the government allocates resources to support the lowest- income people for curative care? 48.7 of respondents agreed to resource allocation of  50 of each curative payment. These results were the particular characteristics of the respondents. Further details of characteristics from such surveys are not presented here, however, because of the limitation of space. Table 3 shows the empirical result using the panel data regression model. Data come from the second, third and fourth survey of Thai Health by the Ministry of Public Health. The second survey took place during 1996-1997, the third survey took place during 2003-2004, and the fourth survey happened during 2008-2009. They are composed of a large amount of Thai people’s health data. Yet, this paper only takes income and exercise into account with the body mass index. INCOME stands for the monthly personal income of each sample. E represents deficient exercise. Table 3 Panel Data Regression Model of the Reduced Form MODE L INTERCEP T INCOM E E 2 R Pooled 19.65 11.686 0.0006 1.56 0.0347 0.703 0.2 7 Fixed Effects 19.544 9.819 0.0006 1.782 0.0385 0.573 0.7 1 Random Effects 19.59 10.047 0.0006 1.833 0.0367 0.661 0.3 1 Note: This table reports the coefficients of the panel data regressions using a reduced form. The data come from the second, third and fourth survey of Thai Health by the Ministry of Public Health. Intercept is the constant body mass index in each model. Denotes a 0.10 significance level. Denotes a 0.05 significance level. Denotes a 0.01 significance level. The empirical findings in Table 3 demonstrate a particular model that the appropriate model is the Random Effect Model. It is because the Hausman test is not rejected the null hypothesis at a 0.05 significant level. In addition, only monthly personal income has a significantly positive effect on the body mass index. The average coefficient is 0.0006. It means that once monthly personal income goes up to 1,000 baht, the body mass index will rise by 0.6 kmm 2 per person. Deficient exercise, however, does not have any effect on the body mass index in all models. It implies that Thai people should pay income to mitigate risk from Education and Leadership in Glocalization : What does “think globally, act locally” mean for education around the world? 21-24 2014 245 obesity or pay for curative care. Consequently, it results in reduction of the body mass index. Still, there are some factors that affect the body mass index because of a larger magnitude of parameters at a 0.01 significant level. This is why the next estimation will control for other explanatory variables with enough data. In addition, the next empirical findings state that there are significant effects on the body mass index BMI employed by the Logistic Regression Model. Data come from a survey of 2,500 samples in over 50 administrative districts of Bangkok in 2012. In fact, all explanatory variables can statistically explain the body mass index with a likelihood ratio statistic of 42.76, p-value of 0.000, and pseudo 2 R of 0.0769. Most importantly, the relationship between explanatory variables and the probability of being obese is consistent with the derived model in this research. That is, gender   1 X , marriage status   3 X , occupation   7 X , exercise per week   20 X , activity in spare time   22 X , curative care   23 X , and risk- mitigating spending from getting sick   25 X which are negatively related to the probability of being obese. In contrast, age   2 X , education level   5 X , personal income level per month   8 X , and monthly spending on expensively healthy food   24 X are positively correlated with the probability of being obese. Even though there are several factors which affect on the probability of being obese, the significant variables are composed of gender, age, marriage status, exercise per week, and risk- mitigating spending from getting sick. In fact, gender has a significantly negative effect on the probability of being obese. The estimated coefficient is statistically significant at 0.01significant level. In particular, a change in probability     ˆ ˆ ˆ 1 i i i p p   is equal to -0.1828. This implies that a change in gender from male to female leads to a decline of 0.1828 in probability of being obese. In addition, age is positively related to the probability of being obese. The estimated coefficient of -0.88374 is statistically significant at a 0.01 significant level. A computed change in probability is equal to 0.0139. It means that if people in Bangkok are older by one year, the probability of being obese will go up 0.0139. It also implies that the older people are, the more people get the probability of being obese. The other finding states that marriage status has a negative effect on the probability of being obese. The average slope on 3 X of -0.48438 is statistically significant at a 0.10 significant level. A computed change in probability is equal to -0.1143. It implies that if people change status from single to married, the probability of being obese will reduce by 0.01143. The day per week of exercise is negatively related to the probability of being obese. This negative relationship is statistically significant at the 0.01 level. The estimated coefficient is -0.18517. A computed change in probability is equal to - 0.0459. It means that a decline in probability of being obese for an increase in a day per week of exercise is 0.0459. Furthermore, risk-mitigating spending from getting sick has a statistically negative effect on the probability of being obese at a 0.10 significant level. The estimated coefficient is -0.16358. A computed change in probability is equal to -0.0406. It means that a decline in probability of being obese for an increase 1 in risk-mitigating spending from getting sick is 0.0406. As a result, there is only a positive relationship between age and the probability of being obese, but the others have a statistically negative effect on the probability of being obese.

6. Conclusion