Financial Literacy, Savings Behavior, and Retirement Planning in Taiwan

Transcription

1 兩岸金融季刊第四卷第二期 (2016 年 6 月 ) Financial Literacy, Savings Behavior, and Retirement Planning in Taiwan Yu-Jen Hsiao Associate Professor of the Executive MBA Program in Biotechnology, Taipei Medical University Shih-Jie Lin* deputy director of Financial Research Institute in Taiwan Academy of Banking and Finance Chinbat Dambaravdan Master in the Department of Finance at National Dong Hwa University Abstract In this paper, we examine the relationship among financial literacy, savings behavior, and retirement planning in Taiwan. We not only find people who are more financially knowledgeable are more likely to plan for retirement but also find those who have stronger savings habits are more likely to plan for retirement. In most specifications we find a stronger role for savings habits than financial illiteracy in explaining retirement planning. Keywords:Financial Literacy, Savings Behavior, Retirement Plan, Order logit 1. Introduction The relationship between financial literacy and household behavior is important, as individual are increasingly being asked to take on responsibility for their financial well-being and their retirement preparation. In fact, many US, and Netherlands have very little savings on the verge of retirement ( Lusardi and Mitchell, 2007a; van Rooij, Lusardi and Alessie, 2011a). In Taiwan as well, 1 it has become clear that many employees hold overly pessimistic expectations about the level of their pension benefits. 2 With demographic changes toward a low birth rate and an aging society, concerning the retirement planning has become an important issue in general. 3 As to the benefits of retirement planning to the national economy, adequate pension planning not only helps to fulfill the public's economic needs, but also to achieve the goal of social stability. Pension also plays the role of long term fund supplier in the financial markets, thus proper utilization of pension funds helps capital * Shih-Jie Lin (corresponding author) is a deputy director of Financial Research Institute in Taiwan Academy of Banking and Finance ( jerrylin@tabf.org.tw). 1

2 蕭育仁 林士傑 Chinbat Dambaravdan: 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 formation and facilitates economic development. According to studies applying data of developed countries such as US, Germany and Netherlands, we know that financial literacy is a key factor affecting how the public can achieve adequate retirement planning (Bucher-Koenen and Lusardi, 2011;Lusardi and Mitchell, 2011a;van Rooij, Lusardi, and Alessie, 2011a). That means if a household does not possess proficiency in making good financial decisions, it could be short sighted and lead to insufficient retirement planning (OECD, 2005). Recently, Hsiao, Chen and Liao (2016) use Taiwan s data to show that the financial literacy has a positively significant impact on retirement planning which are consistent with those in developed countries (Bucher-Koenen and Lusardi, 2011; Dvorak and Hanley, 2010; Lusardi and Mitchell, 2007a, 2007b, 2011a;van Rooij, Lusardi and Alessie, 2011a). In other words, the Taiwanese public with more proficient financial literacy is more likely to have sufficient retirement planning. However, a low level of financial literacy does not necessarily imply that people are bound to make poor retirement planning. At least in principle, they have higher saving rate. As long as households can resort to the saving habit for their retirement planning, thus avoiding can act as a substitute for their own learning. In this paper, we not only find people who are more financially knowledgeable are more likely to plan for retirement, but also find those who have stronger savings habits are more likely to plan for retirement. In most specifications we find a stronger role for savings habits than financial illiteracy in explaining retirement planning. The outline of the paper is as follows. We review literature on financial literacy survey and retirement planning in Section 2. Section 3 illustrates pension arrangements in Taiwan. Section 4 illustrates data, financial literacy measurement and empirical method. We present the empirical results and analysis in Section 5and section 6 concludes. 2. Literature Review Previous studies show that even for workers on the edge of retirement, their retirement planning is still far beyond adequate (Lusardi and Mitchell, 2007b). With difference in factors such as restrictions of wealth and income, discount rate, joining in social retirement planning and financial literacy, households show significant divergence in the need of pension (Ameriks, Caplina and Leahy, 2003; Bernheim, Skinner and Weinberg, 2001;Lusardi and Mitchell, 2007a;Saez, 2009 and Scholz, 2

3 兩岸金融季刊第四卷第二期 (2016 年 6 月 ) Seshardi and Khitatrakun, 2006). Similarly, households reveal that concern over retirement planning vary according to the level of household wealth (Ameriks et al., 2003;Lusardi and Mitchell, 2007a, 2007b). Lusardi and Mitchell (2007a) prove that while controlling the social-economic characteristics of income and education background, people who have retirement planning also have more wealth, but those who have good fortune do not certainly have proper retirement planning. Binswanger and Carman (2012) argue that the means to prepare pension is also a factor that influences retirement planning. They discovered that people who depend on rule of thumb to conduct retirement planning excel against those who do not apply any systematic manner at all. Aggarwal and Goodell (2013) find that except for major policy, institutional and social factors, the role that females play in economy, society, religion and culture also show significant positive impact on the development of pension systems in these countries. Cocco and Gomes (2012) discuss it under life cycle theory with a tendency of extended expected life. The result shows that the rational strategies for people to counter longevity risk are to increase the holdings of savings and financial assets and also to extend the age of retirement; that is to say, investment portfolio and insurance will affect wealth accumulation, and further counter longevity risk. Mastrobuoni (2011) argue that the Social Security Statement in US since 1995 is helpful to workers to enhance their knowledge about their own benefits, but is not helpful in improving their retirement behavior. Seligman and Bose (2012) discovered that by actively managing pension accounts and attending seminars for financial education, the portfolio allocation of households and wealth accumulation would both be improved. Many previous studies suggest that workers in the United States are lacking of pension related knowledge (Bernheim and Levin, 1989;Chan and Stevens, 2008). Lusardi and Mitchell (2007a, 2007b, 2011b) and also argue that people with lower financial literacy do not have relatively sound retirement planning, while they also do not conduct savings and wealth management according to their retirement planning. In contrary, people with higher financial literacy would engage more in retirement planning, stock investing and practical borrowing behavior (Lusardi, 2008). Aside from empirical studies in the United States, a study using data from the Netherlands argues that there is a significantly positive relationship between financial literacy and retirement planning (van Rooij et al., 2011a). Dvorak and Hanley (2010) find that participants show fairly good knowledge of the basic mechanics of the plan, but are unable to differentiate among various investment options. Empirical study proves that people with 3

4 蕭育仁 林士傑 Chinbat Dambaravdan: 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 higher financial literacy are keen to engage in risky commodity investment but they also show better retirement planning, thus there is a significant positive correlation between financial literacy and household wealth (van Rooij, Lusardi and Alessie, 2012). Russian data reveals similar findings and conclusions (Klapper and Panos, 2011). Recently, Hsiao, Chen, and Liao (2016) collect data from Taiwan and find that people with a higher financial literacy tend to have more diversified fund sources for retirement planning. 3. Pension Arrangements in Taiwan 4 The aim of Labor Pension in Taiwan according to the Labor Pension Act 5 is to enhance the protection of workers' lives after retirement, strengthening the relationships between workers and employers and promoting social and economic development. Under the Act, all employers are required to deposit 6% (or more) of a worker's monthly wages into an individual labor pension account managed by the Bureau of Labor Insurance, with ownership going to the worker. Business entities employing 200 or more workers shall decide through their respective labor unions whether to adopt an annuity insurance program.if no labor unions exists,with the approval obtained through a labor-management meeting and the central competent(the Ministry of Labor), a business entity may take out an annuity insurance that complies with the Insurance Act on behalf of those workers who choose, in writing, to take out an annuity insurance. Workers adopting the new pension system may begin accumulating their pension accounts as their employers contribute to them. The accounts are portable and will be retained even if workers switch jobs or if business entities shut down or cease operations. Under the pension system, in addition to the mandatory employers' contribution, workers may contribute voluntarily additional sums, maximum 6% of their monthly wages to their pension accounts. For those who contribute beyond the required amount, the extra contributions may be tax deducted from their total annual income. A worker may begin collecting his or her pension payments upon reaching the age of 60, regardless of employment status. Workers with 15 years or more work seniority can receive monthly pension payments. Workers with less than 15 years of seniority can only receive a lump sum payment.for the purpose of this Act, the authorities are Ministry of Labor at the central government level, municipal city governments at the municipal city level, and county/city governments at the county/city level. 6 4

5 兩岸金融季刊第四卷第二期 (2016 年 6 月 ) 4. Data and Variable Measures 4.1 Data Sources In this study, we use data derived from the National Financial Literacy Survey conducted by Taiwan Financial Supervisory Commission (FSC) in 2007, 2009 and For the research purpose, the representative sample comprised people over 40 years old. After deleting observations that did not report on retirement planning or with missing values, the resulting data set had a sample size of 3, Measurement of Retirement Plan Levels In this research, we employ the results of Question which is a multiple choice question that surveys the sources of the respondent's pension funding. If the respondent's answer contains any of the three: no planning, not sure or engage in derivatives investment, 8 we define this respondent as low retirement planning; if the answer contains only one of the following: firm pension fund, labor insurance, public servant insurance or savings (bank savings or rotating savings, we define as middle retirement planning; if the answers contain one of those of middle retirement planning but also contain buying bond, stocks, mutual funds or endowment insurance, we define as high retirement planning. 4.3 Measurement of Financial Literacy So far, there is no standard measurement of financial literacy. For example, for the number of questions in their financial literacy questionnaire, Lusardi and Mitchell (2007a, 2008) designed three questions, but Stango and Zitman (2009) asked only one. Other studies employ factor analysis to further divide into basic financial literacy and advanced financial literacy (van Rooij, Lusardi and Alessie, 2011b), these questionnaires includes questions on the concepts of compounded interest rate, inflation, diversification and the features of stocks and bonds. Lusardi and Mitchell (2007b) conclude the way of measuring financial literacy in different countries. There is also a research reviewing papers in the last decade and argue that the surveys of financial literacy should cover basic knowledge (eg: time value of money), knowledge of credit loans, knowledge of savings and investment and knowledge of insurance. In this research, we designed nine different questions to measure financial literacy. Respondent scores were calculated according to the answers provided, where a correct answer was scored one and an incorrect answer was scored zero; thus, the highest attainable score was nine and the lowest was zero. 5

6 蕭育仁 林士傑 Chinbat Dambaravdan: 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 Table 1 shows the overall performance of the respondents. Panel A reports the exact wording of the nine questions. Panel B shows the number of correct answers for the entire sample. The respondents knew about some economic concepts; however, no one correctly answered all nine questions. Table 1. Financial literacy Panel A reports the proportions of interviewees providing correct and incorrect answers to each of the nine questions. Panel B reports the distribution of the number of correct answers on the nine literacy questions. The data are from the 2007, 2009, and 2011 Financial Supervisory Commission survey. Panel A: financial literacy (N=3,365) Q1: In case of interest rate on the decline, what s supposed to be chosen when handling fixed deposits? Q2: In case that the borrowing customers provide collaterals for the mortgage, then the applicable borrowing rate is normally higher than the borrowing rates without collateral? Q3: The index contained in the index mortgage of self-occupied house is referred to as capitalization weighted stock index? Q4: When adopting "installment" to buy cars, the total expenditure is usually Higher than the total expenditure through "cash pay"? Q5: A person can cope with the risks of birth, senility, illness and death by means of insurance? Q6: Purchasing insurance can reduce the risk of life, so the purchasing amount is higher, than the better? Q7: As there is universal coverage, so no other extra medical insurance is needed to be no need to be covered? Q8: What s your opinion about the relationship between risks on the stock market and remunerations; is it that the investment with higher returns would be normally accompanied with greater risks? Correct Incorrect 56.1% 43.9% 40.7% 59.3% 26.8% 73.2% 74.1% 25.9% 19.0% 81.0% 5.3% 94.7% 38.8% 61.2% 70.3% 29.7% Q9: What changes would occur to the bond prices in case of interest rates rising? 29.7% 70.3% Panel B: Summary of responses (N=3365) Number of correct (out of nine questions) None All Correct

7 兩岸金融季刊第四卷第二期 (2016 年 6 月 ) 4.4 Measurement of Savings Behavior To assess the empirical evidence for savings behavior, we tried to clarify who had satisfactory family savings behavior. The wording of the question in the survey was as follows: What is the proportion of money used for saving from the annual revenues of your family? The family savings behavior (Family savings behavior_d) was measured by a binary variable as follows: Family savings behavior_d=1 The proportion of money used for saving from the annual revenues of your family is larger than 20%. Family savings behavior_d=0 The proportion of money used for saving from the annual revenues of your family is less than 20%. Similarly, we tried to clarify who had satisfactory personal bonus savings behavior. The wording of the question in the survey was as follows: If you have a nonrecurring additional income, such as a year-end bonus or employee bonuses, what is the savings ratio? The personal bonus savings behavior (Personal savings behavior_d) was measured by a binary variable as follows: Personal savings behavior_d=1 Personal savings behavior_d=0 For a nonrecurring additional income, such as a year-end bonus or employee bonuses, the savings ratio is higher than 20%. For a nonrecurring additional income, such as a year-end bonus or employee bonuses, the savings ratio is lower than 20%. 4.5 Control Variables Our survey questions are linked to a rich set of data on demographic characteristics, such as age, gender, marital status, educational, background, living area, work status, and average annual household 7

8 蕭育仁 林士傑 Chinbat Dambaravdan: 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 income. The age are categorized: years, 50-59years and above 60 years old. Only one category can be selected: The selected category is defined as 1, and the unselected category is defined as 0. The variable MALE is a dummy variable that equals one if the respondent is female and zero if the respondent is male, and variable Married is a dummy variable that equals one if the respondent is married included those who were separated or widowed and zero if the respondent is unmarried comprised respondents who were divorced or Cohabitation. The education are categorized: below senior high school, university and above graduate school. Only one category can be selected: The selected category is defined as 1, and the unselected category is defined as 0. The variable Urban is a dummy variable that equals one if the respondent live in urban and zero if the respondent live in rural. The variable Taiwanese People is also a dummy variable. The work status are categorized: working full time, temporary worker and retired. Only one category can be selected: The selected category is defined as 1, and the unselected category is defined as 0. Finally, the variables average annual household income are categorized: below NTD 660,000, NTD660,000-1,230,000 and above NTD 1,230,000. Only one category can be selected: The selected category is defined as 1, and the unselected category is defined as 0. Table 2 reports the family and personal savings behavior of the respondents. The percentage of respondents who have family saving behavior is only 26.70% and the percentage of respondents who have personal saving behavior is 27.20%. Table 2. Responses to savings control statements This table reports the proportions of interviewees providing savings control statements to each of the two questions. The data are from the 2007, 2009, and 2011 Financial Supervisory Commission survey. Saving behavior Yes No Family saving behavior: The proportion of money used for saving against annual revenues of your family would larger than 20% Personal bonus saving behavior: If you have a non-recurring additional incomes such as year-end bonus or bonuses for employees the saving ratio would be larger than 20% 26.7% 73.3% 27.2% 72.8% 8

9 兩岸金融季刊第四卷第二期 (2016 年 6 月 ) 4.6 Regression Model We used the following ordered logit model with the sample selection: 9 P( A i 1) ( ' X i ) P( Ai 2) ( k1 ' X i ) ( ' X i )....(1) P( Ai 3) 1 ( k1 ' X 2 i ) where A presents the level of retirement planning level (i.e., A 1 for no experience, A 2 i for medium experience, A 3 for high experience). is the probability density function for the i logistic distribution. is a vector of parameters and k1 is the threshold parameter. i X i is a vector of explanatory variables that includes financial literacy, saving behavior and sociodemographic controls (age, gender, marital status, educational background, living area, work status, and average annual household income). 5. Empirical Results 5.1 Overview of Retirement Planning To examine the relationship between retirement planning and personal features, we take features such as age, gender, marriage status, education, living area, family background, profession, personal and household annual average income, and financial literacy into consideration to observe the level of retirement planning of different groups of public. In Panel A of Table 3 shows the distribution of retirement planning of different groups; we can easily find out that people with higher level of retirement planning are concentrated between ages of This illustrates that people in the prime age notice that it is not enough to depend solely on national social benefits to cover retirement living expenses. Previous empirical studies suggest that older female relative to male do not have complete retirement planning (Lusardi and Mitchell, 2008); however, in our study, the ratio of female with high retirement planning (14.2%) is slightly higher than that of male (9.6%), this finding is not consistent to the result in advanced countries. Third, as respondents with higher education, their level of retirement planning are also higher. Fourth, the differences between those who live in urban areas compared to i 9

10 蕭育仁 林士傑 Chinbat Dambaravdan: 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 those in rural areas are not significant. Fifth, the ratio of the Aboriginal group classified as high retirement planning is only 0.4%. Sixth, people who have a full time job are more likely to have high retirement planning. Finally, the second highest group of respondents in terms of personal and family average annual income shows higher ratios of high retirement planning compared to the group of highest average annual income. The result indicates that people with the highest income level might not worry about financial scarcity and thus their level of retirement planning does not correspond with income. This finding is similar to the result of empirical study made by Lusardi and Mitchell (2007a), they assumed that people who have enough wealth do not necessarily engage in retirement planning. Te last but not the least finding in this study is that respondents with higher financial literacy show higher levels of retirement planning in Panel B of Table 3. Panel C of Table 3 divides savings behavior into family and personal savings behavior. The first row of family (personal) savings behavior represents savings of less than 20% of the extra income ; the second row represents savings of higher than 20%. Table 3. Demographic, financial literacy, and savings behavior characteristics for retirement planning A marital status of single includes respondents who were divorced or living with their significant other, and married includes people who were separated or widowed. Panel A Demographic Low level Middle level High level Age % 26.4% 13.5% % 21.3% 7% 60 above 1.6% 21.6% 3.4% Gender Male 2.5% 31.6% 9.6% Female 4.4% 37.7% 14.2% Status Married 5.4% 56.9% 20.0% Unmarried 1.5% 12.4% 3.9% Education Below senior high school 6.1% 61.5% 17.4% University 0.6% 6.3% 4.7% Above graduate school 0.2% 1.6% 1.8% 10

11 兩岸金融季刊第四卷第二期 (2016 年 6 月 ) Living Area Urban 5.0% 46.1% 17.6% Rural 1.9% 23.2% 6.2% Background Taiwanese People 6.5% 67.6% 23.4% Aboriginal 0.4% 1.7% 0.4% Work status Working full time 2.6% 35.5% 16.9% Temporary worker 0.5% 3.4% 1.6% Retired 0.6% 11.6% 1.4% Others 3.1% 18.8% 3.8% Family annual average income Below NTD 660, % 30.4% 4.8% NTD660,000-1,230, % 28.6% 10.8% NTD 1,230, % 10.5% 8.3% Panel B Financial literacy None 0.5% 1.9% 0.0% 1 1.7% 7.7% 0.5% 2 1.3% 10.8% 1.0% 3 1.2% 15.0% 4.2% 4 0.9% 16.9% 5.5% 5 0.7% 10.8% 7.4% 6 0.5% 4.3% 3.6% 7 0.1% 1.9% 1.7% 8 0.0% 0.0% 0.0% All 0.0% 0.0% 0.0% Panel C Saving behavior Family saving behavior Less than 20% 5.7% 54.4% 13.5% Large than 20% 1.2% 15% 10.5% Personal saving behavior Less than 20% 5.5% 54.3% 12.9% Large than 20% 1.4% 15% 11% 5.2 The Relationship Between Financial Literacy, Saving Behavior and Retirement Planning After observing the level of retirement planning of groups with different features, we employ Order Logit regression model in this study to examine the relationship between the level of financial literacy and retirement planning 10. First of all, we analyze the influence of personal featured variables 11

12 蕭育仁 林士傑 Chinbat Dambaravdan: 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 without the variable of financial literacy to the level of retirement planning; the results are in Model (1) of Table 4. Empirical evidence shows that education background, family background, working status and average annual household income do have significant influence on respondent's level of retirement planning. People with higher education and average annual household income tend to have higher levels of retirement planning. In Model (2) of Table 4, we show the result of adding in the variable of financial literacy. The result suggest that even after controlling the personal featured variables of Model (1), the variable of financial literacy in Model (2) does have significantly positive impact on people's retirement planning. That is to say, people with higher financial literacy are more likely to have more diversified retirement planning than only depend on social insurance/ annuity. This finding is correspondent to the empirical result of developed countries (Dvorak and Hanley, 2010; Lusardi and Mitchell, 2007a, 2007b, 2011b;van Rooij, Lusardi and Alessie, 2011a; Hsiao et al., 2016) and illustrates that a person's financial literacy is a key factor affecting the level of retirement planning in Taiwan. According to Model (3) and Model (4) of Table 4, the coefficient of family savings behavior and personal savings behavior are both significantly positive, suggesting that people with a well savings behavior are more likely to have more diversified retirement planning. In other words, person's saving behavior is also a key factor affecting the level of retirement planning in Taiwan. Table 5 show that family savings behavior and financial literacy are still strong determinants of retirement planning levels when they both include into the model. However, we find a stronger role for savings habits than financial illiteracy in explaining retirement planning. The coefficient of savings habits is significantly larger than the coefficient of financial literacy. Notably, one of the variables that is missing from our empirical specification is a measure of risk tolerance. Financial risk tolerance is an import factor in the financial planning. To control for risk tolerance, we created a dummy variable, risk aversion attitude, which equals one for risk-averse respondents that only invest in deposits and zero otherwise. 11 When risk aversion was included in the model we found that both the ordered logit (Model 3) and ordered logit model (Model 4) estimates of the basic financial literacy index remained significantly positive (Table 5). Thus, the inclusion of risk aversion still supports our argument that financial literacy and saving behavior are positively related to retirement planning. 12

13 兩岸金融季刊第四卷第二期 (2016 年 6 月 ) Table 4. Baseline model Model (1) Model(2) Model(3) Model(4) Par. Z-stat. Par. Z-stat. Par. Z-stat. Par. z-stat. Financial literacy 0.378*** Family saving behavior_d 0.743*** 8.37 Personal saving behavior_d 0.794*** 8.88 Age (Base group: 60 years and older) 40~50 years 0.332** ** ** ** ~60 years Male *** *** ** Married Education (Base group: Below senior high school) University 0.640*** *** *** *** 4.75 Above graduate school 0.793*** *** *** *** 3.68 Urban Taiwanese People 0.538** ** ** ** 2.10 Profession (Base group: Other) Working full time 0.546** ** ** ** 4.28 Temporary worker 0.842*** *** *** *** 4.29 Retired Family annual average income (Base group: Below NTD 660,000) NTD660,000-1,230, *** *** *** *** 4.76 NTD 1,230, *** *** *** *** 7.79 Year Year Pseudo R-square(%) Log likelihood Note: Unmarried included: Divorce and Cohabitation; Married included: Separated and Widowed. Others in profession included: Domestic homemaker, Temporary workers and Unemployment and for some reason or patients unable to work. ***, **, and * indicate significance at the 1%, 5% and 10% levels, respectively. 13

14 蕭育仁 林士傑 Chinbat Dambaravdan: 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 Table 5. Order logit regression analysis of financial literacy, savings behavior and retirement planning Model (1) Model(2) Model(3) Model(4) Par. Z-stat. Par. Z-stat. Par. Z-stat. Par. z-stat. Financial literacy 0.367*** *** *** *** Rrisk aversion attitude *** *** Family saving behavior_d *** *** Personal saving behavior_d 0.641*** *** 6.55 Age (Base group: 60 years and older) Age dummy (40~50) Age dummy (50~60) Male *** *** *** *** Married Education (Base group: Below Vocational school) University 0.455*** *** *** *** 3.42 Above graduate school 0.629*** *** *** 2.63 Urban Taiwanese People Profession (Base group: Other) Working full time 0.426*** *** *** *** 3.07 Temporary worker 0.981*** *** *** *** 5.34 Retired Family annual average income (Base group: Below NTD 660,000) NTD660,000-1,230, *** *** *** *** 3.19 NTD 1,230, *** *** *** *** 5.15 Year ** ** ** ** Year *** *** *** *** Pseudo R-square(%) Log likelihood Note: Unmarried included: Divorce and Cohabitation; Married included: Separated and Widowed. Others in profession included: Domestic homemaker, Temporary workers and Unemployment and for some reason or patients unable to work. ***, **, and * indicate significance at the 1%, 5% and 10% levels, respectively. 5.3 Robustness test Besdies, all retirement planning, financial literacy, and savings behavior might be correlated with family income level. Higher income level might results in higher level of financial literacy, savings, and 14

15 兩岸金融季刊第四卷第二期 (2016 年 6 月 ) retirement planning. The positive and significant relationship between retirement planning and financial literacy in Table 5 might be caused by income level. Similarly, the positive and significant relationship between retirement planning and savings behavior in the same table might be also caused by income level. To resolve these common cause problems, we split the total samples into different income level groups and re-estimate the models in Table 5 for each income group to check the robustness of the results. Table 6 represents the results of this robustness test. Once again, financial literacy and saving behavior are positively related to retirement planning into different income level groups. The results of robustness test are consistent with and supportive of the empirical evidence in Table 5. Table 6. Robustness Test: Order logit regression analysis of financial literacy, savings behavior and retirement planning into different income level groups. Model (1) Model(2) Panel A: Family Income Low (Less than NTD660,000) Par. Z-stat. Par. Z-stat. Financial literacy 0.453*** *** 9.51 Family saving behavior_d 0.611*** 3.10 Personal saving behavior_d 0.478** 2.39 Control Variables Yes Yes Observations Pseudo R-square Log likelihood Panel B: Family Income Middle (NTD660,000-1,230,000) Financial literacy 0.392*** *** 8.80 Family saving behavior_d 0.494*** 3.78 Personal saving behavior_d 0.479*** 2.76 Control Variables Yes Yes Observations Pseudo R-square Log likelihood Panel C: Family Income High (larger than NTD 1,230,000) Financial literacy 0.295*** *** 4.22 Family saving behavior_d 0.716*** 4.22 Personal saving behavior_d 0.816*** 4.65 Control Variables Yes Yes Observations Pseudo R-square Log likelihood Note: ***, **, and * indicate significance at the 1%, 5% and 10% levels, respectively. 15

16 蕭育仁 林士傑 Chinbat Dambaravdan: 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 6. Conclusions As the trend grows towards an aging society with low birth rate, retirement planning is receiving worldwide attention. As a result, in this study we examine how financial literacy and saving behavior affects retirement planning and analyze the difference in retirement planning between respondents with different characteristics. According to the empirical findings from 3,365 respondents above the age of 40 in the three years National Financial Literacy Survey, we find a strong, positive relationship between financial literacy and retirement planning; people who are more financially knowledgeable are more likely to plan for retirement. We also find a strong, positive relationship between savings behavior and retirement planning; those who have stronger savings habits are more likely to plan for retirement. In most specifications we find a stronger role for savings habits than financial illiteracy in explaining retirement planning. While we consider personal features, we find that females and households classified as average annual income are more active engaging in retirement planning. Footnotes 1. In Oct 2012, news about Labor Pension Fund of Taiwan becoming insolvent in 2027, caused a wave of pension fund withdrawing thereby. According to statistics from Bureau of Labor Pension, applications for lump-sum old-age benefit amounted to more than 33 thousands on the month of October This number is four times higher than average and becomes the second highest in history. In 2012, retiring workers that choose lump-sum old-age benefit withdrawing NTD 76.8 billion in October and November. The magnitude of this drawing is only next to that four years ago when the law governing labor insurance annuity was amended. 2. In late January 2013, the Pension Fund Association published Taiwan retirement index and retirement planning opinion poll, this survey showed that over 60% of people are not confident in preparing a sufficient amount of pension; half of poll participants have not even started preparing for retirement. 3. According to a demographics study made in July 2013 by the Ministry of the Interior, the 16

17 兩岸金融季刊第四卷第二期 (2016 年 6 月 ) number of prospective retirees (age 55-59) in Taiwan are 1.65 million. The report named Republic of China population projection from 2012 to 2060 was issued by the Council for Economic Planning and Development (now National Development Council) under Executive Yuan in August This report shows that the elderly population will account for more than 14% of the total population in 2018 and Taiwan will legitimately be considered an aging society; this percentage will go beyond 20% in 2025 and will make Taiwan an extremely aging society. The ratio will steadily climb to above 39% in 2060; by that time, juvenile population will only account for 9.8% while the adult population will be 50.7% of total population. 4. Information on pensions can be found on the government labor force official website: 5. Last modified in July 20th, More details can be found on the official website: 7. The survey allocates nationwide samples according to proportionate stratified sampling as follows: First, the reference population in this survey is from 2010 demographic annual report of all counties/cities in Taiwan issued by The Department of Statistics of Ministry of the Interior and census data of age, gender and education background of Taiwanese population. Then samples gathered from every county/city are allocated by population ratio relative to whole nation, and samples of each stratum are determined by the relative ratio of gender, age and education background. Thus, the samples of this survey could reflect the characteristics of age, gender and education background in each county/city. Finally, we take the annual income distribution in Taiwan issued by Directorate-General of Budget, Accounting and Statistics for reference, in order to fit the samples into Taiwanese population characteristics distribution. 8. Financial derivatives are more complex and risky than bonds and stocks. Therefore, financial derivative is not an appropriate source of the pension funding and is defined in low level retirement 17

18 蕭育仁 林士傑 Chinbat Dambaravdan: 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 planning. However, We also conduct a robustness test by classifying financial derivatives in high level retirement planning and the findings in Appendix Table A1 proved to be qualitatively similar. 9. See Greene (2003) for a description and discussion of the ordered logit model. We also estimated our model by using ordered probit and ordinary least squares regressions and the findings proved to be qualitatively similar. We do not report the results for the sake of brevity; they are available from the authors upon request. 10. In this study, we firstly conduct correlation analysis of each variables and find that no highly correlation between the variables. Thus, we could exclude the influence of collinearity on our study. Besides, we also employ an Order Probit regression model to run a robustness test and get similar findings. In order to keep this paper concise, we do not present the results of this robustness test here. 11. The Literacy Survey also provides multiple-choice Question 9.1 for investors, which asks, Do you have the following investments. The proposed choices are: (1) Current deposits; (2) Time deposits; (3) Funds; (4) Stocks; (5) Bonds; (6) Investments in the private financial sector; (7) Insurance; (8) Real estate; (9) Options and futures; and (10) Currency. References 1. Aggarwal, R. and J. W. Goodell (2013), Political-economy of pension plans: Impact of institutions, gender, and culture, Journal of Banking & Finance, 37, Ameriks, J., Capling, A., ad Leahy, J., (2003), Wealth accumulation and the propensity to plan, Quarterly journal of economics, 118, Bernheim, D. B., and L. Levin (1989), Social security and personal saving: an analysis of expectations, American Economic Review, 79, Bernheim, D. B., J. Skinner and S. Weinberg (2001), What accounts for the variation in retirement wealth among U.S. household?, American Economic Review, 91,

19 兩岸金融季刊第四卷第二期 (2016 年 6 月 ) 5. Binswanger, J., and Carman, K. G., (2012), How real people make long-term decisions: The case of retirement preparation, Journal of Economic Behavior& Organization, 81, Bucher-Koenen, T. and A. Lusardi (2011), Financial Literacy and Retirement Planning in Germany, Journal of Pension Economics and Finance, 10, Chan, S. and A. H. Stevens (2008), What you don t know can t help you: pension knowledge and retirement decision-making, The Review of Economics and Statistics, 2, Cocco, J. F. and F. J. Gomes (2012), Longevity risk, retirement savings and financial innovation, Journal of Financial Economics, 103, Dvorak, T., and H. Hanley (2010), Financial literacy and the design of retirement plans, Journal of Socio-Economics, 39, Hsiao, Y. J., J. T. Chen and C. F. Liao (2016), The relationship between financial literacy and retirement planning, Journal of Management, 33, Lusardi, A., (2008). Financial literacy: An essential Tool for Informed Consumer Choice. NBER Working paper. 12. Lusardi, A., and O.Mitchell (2007a), Baby boomers retirement security: the role of planning, financial literacy and housing wealth, Journal of Monetary Economics, 54, Lusardi, A., and O.Mitchell (2007b), Financial literacy and retirement preparedness: evidence and implications for financial education, Business Economics, 42, Lusardi, A., and Mitchell, O., Planning and financial literacy: how do female fare? American Economics Review, 98, Lusardi, A., and O.Mitchell (2011a), Financial literacy and retirement planning in the United States, Journal of Pension Economics and Finance, 10, Lusardi, A., and O.Mitchell (2011b), Financial literacy around the world: an overview, Journal of Pension Economics and Finance, 10,

20 蕭育仁 林士傑 Chinbat Dambaravdan: 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 17. Mastrobuoni, G. (2011), The role of information for retirement behavior: Evidence based on the stepwise introduction of the Social Security Statement, Journal of Public Economics, 95, OECD, Organization for Economic Co-Operation and Development (2005), Improving financial literacy: Analysis of issues and policies. Paris, France. 19. Saez, E., (2009), Details matter: the impact of presentation and information on the take-up of financial incentives for reetirement saving, American Economic Journal: Economic Policy, 1, Scholz, J. K., Seshardi, A., and Khitatrakun, S., (2006), Are Americans saving optimally for retirement?, Journal of Political Economy, 114, Seligman, J. S. and R. Bose, (2012), Learning by doing: Active employer sponsored retirement savings plan participation and household wealth accumulation, The Quarterly Review of Economics and Finance, 52, Stango, V., and J. Zinman (2009), Exponential growth bias and household finance, Journal of Finance, 64, van Rooij, M, V., A. Lusardi and R. Alessie (2011a), Financial literacy and retirement planning in the Netherlands, Journal of Economic Psychology, 32, van Rooij, M, V., A. Lusardi and R. Alessie (2011b), Financial literacy and stock market participation, Journal of Financial Economics, 101, van Rooij, M, V., A. Lusardi and R. Alessie (2012), Financial literacy, retirement planning, and household wealth, The Economic Journal, 122, 關鍵詞 金融知識 儲蓄行為 退休規劃 Order logit 20

21 兩岸金融季刊第四卷第二期 (2016 年 6 月 ) Appendix Table A1 Robustness : Measurement of Retirement Plan Levels Model(1) Model(2) Par. Z-stat. Par. Z-stat. Financial literacy 0.376*** *** Rrisk aversion attitude *** *** Family saving behavior_d 0.642*** 7.07 Personal saving behavior_d 0.530*** 5.77 Age (Base group: 60 years and older) Age dummy (40~50) Age dummy (50~60) Male *** *** Married Education (Base group: Below Vocational school) University 0.476*** *** 3.74 Above graduate school 0.601*** *** 2.77 Urban Taiwanese People 0.537* ** 2.01 Profession (Base group: Other) Working full time 0.323*** *** 2.86 Temporary worker 0.844*** *** 4.41 Retired Family annual average income (Base group: Below NTD 660,000) NTD660,000-1,230, ****** *** 4.52 NTD 1,230, *** 5.91 Year ** *** Year *** *** Pseudo R-square(%) Log likelihood Note: We employ the results of Question which is a multiple choice question that surveys the sources of the respondent's pension funding. If the respondent's answer contains any of the three: no planning, not sure, we define this respondent as low retirement planning; if the answer contains only one of the following: firm pension fund, labor insurance, public servant insurance or savings (bank savings or rotating savings, we define as middle retirement planning; if the answers contain one of those of middle retirement planning but also contain buying bond, stocks, financial derivatives, mutual funds or endowment insurance, we define as high retirement planning. Unmarried included: Divorce and Cohabitation; Married included: Separated and Widowed. Others in profession included: Domestic homemaker, Temporary workers and Unemployment and for some reason or patients unable to work. ***, **, and * indicate significance at the 1%, 5% and 10% levels, respectively. 21

22 蕭育仁 林士傑 Chinbat Dambaravdan: 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 金融知識 儲蓄行為與退休規劃 : 台灣的實證結果 蕭育仁台北醫學大學生物科技高階管理碩士在職專班副教授 林士傑 * 台灣金融研訓院副所長 Chinbat Dambaravdan 國立東華大學財務金融學系碩士 摘 要 本文透過金融監督管理委員會的國民金融知識問卷調查資料, 採用 Order logit 探討金融知識 儲蓄行為與退休規劃之關聯性 實證結果發現, 除了金融知識愈高的台灣民眾愈容易會想要進行退休計劃外, 也發現台灣民眾具有較強的儲蓄習慣的民眾也愈可能去規劃退休生活 特別值得注意的是, 具備儲蓄習慣對於積極進行退休計劃的影響力, 更勝於金融知識的程度 * 林士傑 ( 通訊作者 :jerrylin@tabf.org.tw), 台灣金融研訓院副所長 22