University of Amsterdam Amsterdam Business School BSc Economics & Business
Does GDP capture welfare indicators well?
Author: Shiyao Wang
Student number: 11594756
Thesis supervisor: dr. D.F. (Dirk) Damsma Finish date: June, 2020
ABSTRACT
This paper uses Westerlund cointegration test to examine the relationship of GDP per capita, health and employment indicators from the Better Life Index for six selected countries over five years. Empirical results have shown no correlation between the GDP per capita and employment rate indicator, but a cointegrating relationship between GDP per capita and self-reported health indicator for middle-income countries.
TABLE OF CONTENTS
ABSTRACT... ii
TABLE OF CONTENTS... iiii
LIST OF TABLES...iiv
1 Introduction...1
2 Theoretical Framework...3
3 Methodology...5
3.1 Participants and measures...5
3.2 Design... 6 3.3 Data...7 4 Results...8 5 Discussion...13 6 Conclusion...14 REFERENCES... .15
LIST OF TABLES
Table 1 - Descriptive Statistics 06
Table 2 - Self-reported Health by countries 07 Table 3 - Augmented Dickey Fuller Test for Canada 08 Table 4 - Augmented Dickey Fuller Test for Japan 08 Table 5 - Augmented Dickey Fuller Test for Korea 08 Table 6 - Augmented Dickey Fuller Test for Mexico 08 Table 7 - Augmented Dickey Fuller Test for Turkey 09 Table 8 - Augmented Dickey Fuller Test for the U.S. 09 Table 9 - Breitung unit-root test for Turkey 09
Table 10 - Employment rate of Canada 09
Table 11 - Westerlund test for Canada 10
Table 12 - Westerlund test for Japan 10
Table 13 - Westerlund test for Korea 10
Table 14 - Westerlund test for Mexico 11
Table 15 - Westerlund test for Turkey 11
1 Introduction
Since the outbreak of the global pandemic coronavirus disease 2019 (COVID-19), over 9 million confirmed cases had been reported, which contains around half a million death cases (“Situation report,” 2020). Moreover, as the virus is primarily spread among people through close contact, countries and cities started to implement the “Social Distancing” policy and working from home is encouraged. However, due to the nature of jobs in service sector like restaurants and barber shops, construction sector that requires physical work, and the fact that not all families have access to internet and computers, many people are forced to become unemployed without knowing if and when they will find another job. As an example, the unemployment rate in the United States of America (the U.S) has been rising from 3.8% in February to the highest level of 14.4% in April (Kochhar, 2020).
Besides the tremendous joint efforts that have been put by countries in medical treatment to prevent and treat the disease, it is worth noticing how the life of people in different countries are influenced to a different extent by this pandemic. The performance of many countries during this crisis does not seem to match their Gross Domestic Products per capita (GDP per capita) level, which are commonly used to measure the economic welfare. For instance, Canada has a lower GDP per capita, yet has been dealing with the COVID-19 better than the U.S., which can be seen from its lower infection rate, death rate and unemployment rate (Beauchamp, 2020). Similarly, South Korea is more proactively and efficiently combating coronavirus than Japan despite a lower GDP per capita index. The political system plays an important role, but the social welfare system regarding unemployment benefit and healthcare contribute to performance during the pandemic as well.
In order to better understand whether GDP per capita captures the welfare indicators well, especially in the aspects of employment and healthcare, the concerning question is the
relationship between GDP per capita and employment and individual health status. An empirical study with a panel data cointegration tests, using data of GDP per capita and Better Life Index (BLI) from Organisation for Economic Co-operation and Development (OECD) of 6 countries over a 5-year range from 2013 to 2017 will be used to try to answer this question
This paper is organized as follows. In the Theoretical Framework section, information about the GDP per capita and OECD Better Life Index, and how the indicators are chosen will be introduced, previous literature regarding the relationships between GDP per capita,
employment and health status will be discussed, and the hypotheses will be stated. Then the details about the participants, design, data, and model of this paper will be described in the
Methodology section. After that, this paper will show and interpret the results of the Westerlund cointegration tests in the Result and Discussion section. In the last section of the Conclusion, there will be a summary of the question addressed and the conclusion of the discussion. Limitations will also be discussed.
2 Theoretical Framework
GDP measures the aggregate monetary value of final goods and service result from economic production in a given year within a country’s boundaries (Bergh, 2009; Callen, 2020). According to the definition of GDP, it is designed to compare the position of economy of
countries; however, for many years, GDP and GDP per capita are also used to assessing the economic welfare.
Economists consider GDP as a flawed measurement of welfare as it ignores factors like non-market transactions, inequalities and destruction of environment and resources (Fleurbaey, 2009; Boarini & D’ Ercole, 2013; Jones & Klenow, 2010). Efforts trying to find the alternative measurement of welfare such as “consumption equivalent” (Jones & Klenow, 2010), “corrected GDP” (Nordhaus & Tobin, 1973; Fleurbaey, 2009) and “capability approach” (Sen, 2011) are taken. This paper will focus on the indicators of the OECD Better Life Initiative, which combined 11 universally considered dimensions of welfare.
The OECD framework considers well-being at a broader range of dimension instead of sole monetary value. It has 11 dimensions which are income and wealth, jobs, housing
conditions, health status, work-life balance, education and skills, social connections, civic engagement and governance, environmental quality, personal security and life satisfaction, where the first 3 dimensions represent the material requirement of well-being and the rest indicate the non-monetary parts of welfare (Balestra, Boarini, & Tosetto, 2018; Boarini & D’ Ercole, 2013). The two dimensions of jobs and health status will be discussed in details in this paper. Not only because the health status and working life are affected most by the current global pandemic, but also the two dimensions themselves are of the topics that matter to a greater extent to people across countries (Balestra, Boarini, & Tosetto, 2018; Boarini & D’ Ercole, 2013). After analyzing responses of around 88,000 users of the OECD Better Life Index, Balestra et al. (2018) and Boarini and D’ Ercole (2013) have found that although those 11 dimensions are on average equally attractive to people, dimensions like health, education and life satisfaction are rated higher and matter the most to users. As health is an issue for older people, the weight of jobs, on the other hand, declines with ages, are more important for young and middle-aged people that consist of most of a country’s population. Therefore, focus on those two aspects would give a better understanding of the overall welfare of society.
Within each of the 11 dimensions, several indicators are chosen for each topic that is tried to be relevant and comparable (Boarini & D’ Ercole, 2013). For instance, life expectancy and self-reported health are two indicators selected by the OECD to indicate the health status
aspect of welfare. As for the health, previous paper focus on the life expectancy has shown that there is a cointegrating relationship between the life expectancy and GDP per capita (Swift, 2011), economists study health care expenditure have found a unidirectional causality from health care expenditure to GDP but not the reverse (Heshmati, 2018), and Fogel (2004) has a strong case that higher GDP per capita leads to improved health. Under the current world situation, self-reported health indicator is a more relevant health index than life expectancy as people with underlying diseases are more easily to be infected and died from the coronavirus. Moreover, based on the result of Swift (2011) it is worth to investigate if the cointegrating relationship also holds between GDP per capita and self-reported health, as the self-reported health and life expectancy are supposed to reflect a similar as they are both indicators of health dimensions under OECD.
There are 4 indicators for jobs dimensions which are labour market insecurity,
employment rate, long-term unemployment rate and personal earnings. Labour market insecurity is an essential factor of employment quality; it indicates the expected loss of earnings during the unemployed period, which contains the length of unemployment and how much unemployment subsidy can expect. Labour market insecurity would be the most appropriate indicator among those four since while many countries so are people are unsure of the time to reopen the business fully, the labour market insecurity has increased, however, due to the methodology problem that will be explained in next section, this paper will use the employment rate indicator measuring the welfare in jobs. The employment rate indicator fit the paper better than the rest two as it has a close relationship with the unemployment rate (they do not add to 1). According to economic theories like the Okun’s law, there is a negative correlation between the unemployment rate and economic growth rate; therefore, it is reasonable to assume the GDP per capita and employment rate are correlated. This makes sense from the theoretical view as well as consumer spending account for a large part of the economy, and when the employment rate is high, more people are working which stands for more money to be spending and more goods to be consumed.
Although the majority of the literature has been discussing the causality between GDP and welfare indicators, this paper focuses on the question of how well does GDP capture the indicators so correlation instead of causality will be tested. Based on the theories of previous literature mentioned above, the following hypotheses will be tested in this paper: (1) there is a correlation between GDP per capita and employment, (2) there is a correlation between GDP per capita and self-reported health indicator.
3 Methodology
3.1 Participants and measures
As mentioned above, this paper is interested in the relationship between GDP per capita and the OECD indices of health and jobs. Self-reported health will be used to measure the health aspect of welfare, and the employment rate will be used to measure the jobs aspect of welfare for a reason explained before. The data used are GDP per capita, self-reported health and
employment rate of 6 countries and over the periods of 2013-2017. The time range of the data covers only a relative period of 5 years due to the fact that the OECD Better Life Initiative was launched since 2011, the available data are limited, and since there is no Better Life Index published in 2018, data in 2019 is also disregarded to maintain the continuity of time series.
The participants are the U.S., Canada, Japan, South Korea, Mexico and Turkey. It was intended to compare groups of countries classified by country size (large, middle, small) and income level (high, middle, low) as high and low-income countries may have different correlations between GDP and welfare index, and a smaller country may be easier to manage thus contribute to better welfare index. However, due to the fact that data are not available for all countries especially for those with low income, only 6 countries are chosen that represent high-income countries with large country size (the U.S. v.s. Canada), high-high-income countries with smaller country size (Japan v.s. South Korea) and middle-income countries (Mexico v.s. Turkey).
The data of GDP per capita and welfare index are public data provided by the OECD. The data contains 6 countries that have large variations in GDP per capita. The unit of GDP per capita is the constant U.S. dollars where the base year is 2015 (“Better Life Index,” 2020). The unit of employment rate indicator and self-reported health indicator is the percentage which represents the number of employed people aged from 15 to 64 over the population of the same age as defined by the International Labour Organization (ILO). The unit of self-reported health indicator is a percentage as well since it refers to the percentage of the population aged 15 years old and over who report good or better health.
As the methodology of measuring welfare is still being fine-tuned, many of the indicators are calculated in a slightly different way for each edition of the Better Life Index. For instance, the labour market insecurity is defined as the percentage of expected loss to the previous earnings, associated with unemployment, where the loss depends on the risk of becoming unemployed, the expected duration of unemployment and the degree of mitigation against these
losses provided by government transfers to the unemployed (“Better Life Index,” 2020) since 2016. In 2014 and 2015, the labour market insecurity is calculated as the number of people who were unemployed in the previous year but was employed in two years before over the total number of employed in two years before. While in 2013, the labour market insecurity is calculated by the number of dependent employed with job tenure of fewer than 6 months over the total dependent employment. To maintain consistency and make correlation meaningful, the indicator of employment rate, which is measured with the same methodology instead of labour market insecurity is used in this paper. The GDP per capita and self-reported health both have a consistent way of measuring and calculating as well.
3.2 Design
The Westerlund cointegration tests will be used throughout this paper. There is an
underlying assumption that time series are non-stationary; therefore, an Augmented Dicke Fuller test will be performed first to check whether the panel data are non-stationary. Cointegration test suggests if a correlation between two-time series is indeed significant or due to chance. Non-stationary means the time series do not have a constant mean, variance, or autocorrelation.
For the Augmented Dickey-Fuller, the null hypothesis states there is a unit root. If the null hypothesis of a random walk with a possible drift is rejected, it can be inferred that the time series is stationary. The null hypothesis of the Westerlund test is no cointegration with the alternative hypothesis of cointegration in all panels.
3.3 Data
Table 1 - Descriptive Statistics
Variable Obs Mean Std. Dev. Min Max
Year 30 2015 1.43839 2013 2017
GDP per capita 30 37269.17 12690.49 17748.51 58271.8
Employment rate 30 64.43333 7.955407 48 74
Self-reported health 30 62.96667 23.00747 30 90
According to the descriptive statistics in Table 1, there is a considerable variation in GDP per capita and self-reported health status, as they have ranges of 40, 523.29 dollars and 60 percentage point respectively, the standard deviation of those two variables are also relatively high. The difference in GDP per capita is reasonable as the data contains middle and high-income countries. As for the variation in self-reported health, it can be seen from the Table 2 that
and Canada. This makes sense as middle-income countries are in general have a worse
healthcare system, working conditions and environment. However, the percentage of people in Japan and South Korea are surprisingly low; only around 33% of people state that they are healthy. The self-reported health percentages of Japan and South Korea are even lower than Mexico and Turkey, not to mention the U.S. and Canada, which contradicts with the health status data provided by OECD statistics where Japan, South Korea, Canada and the U.S. have a similar life length of around 83 years (“Health Status”, 2020). Possible explanations are culture
difference and difference in health indicators, the overtime work culture in East Asia (Tsai, Nitta, Kim & Wang, 2016) that may bring heavier mental burden that influences the perceived health status of people, and while the most common health status indicator is life expectancy, living longer may not necessarily imply a healthier life. The data set used penal data, which contains 6 countries over a 5 year period, which suggests less external validity due to the relatively short time range and a limited number of countries.
Table 2 - Self-reported Health by countries
Country Mean Canada 88.4 Japan 32 Korea 35.4 Mexico 66 Turkey 67.2 United States 88.8
4 Results
In the last section, the hypotheses of (1) there is a correlation between GDP per capita and employment, (2) there is a correlation between GDP per capita and self-reported health indicator were mentioned.
Table 3 - Augmented Dickey Fuller Test for Canada Test
Statistics 1% CriticalValue 5% CriticalValue 10% CriticalValue p-value GDP per capita -2.162 -4.380 -3.600 -3.240 0.5115
Employment rate · -4.380 -3.600 -3.240 1.0000
Self-reported health -0.447 -4.380 -3.600 -3.240 0.9854 Table 4 - Augmented Dickey Fuller Test for Japan
Test
Statistics 1% CriticalValue 5% CriticalValue 10% CriticalValue p-value GDP per capita -1.060 -4.380 -3.600 -3.240 0.9354
Employment rate · -4.380 -3.600 -3.240 1.0000
Self-reported health -1.789 -4.380 -3.600 -3.240 0.7101 Table 5 - Augmented Dickey Fuller Test for Korea
Test
Statistics 1% CriticalValue 5% CriticalValue 10% CriticalValue p-value GDP per capita 0.086 -4.380 -3.600 -3.240 0.9950 Employment rate -0.447 -4.380 -3.600 -3.240 0.9854 Self-reported health · -4.380 -3.600 -3.240 1.0000
Table 6 - Augmented Dickey Fuller Test for Mexico Test
Statistics 1% CriticalValue 5% CriticalValue 10% CriticalValue p-value GDP per capita -0.843 -4.380 -3.600 -3.240 0.9618 Employment rate -2.236 -4.380 -3.600 -3.240 0.4696 Self-reported health · -4.380 -3.600 -3.240 1.0000
According to Table 3, null hypotheses of the unit root test for Canada in 3 variables are not rejected as the p-values for 3 variables are 0.5115, 1.0000, 0.9854 respectively; therefore the time series is non-stationary. A similar interpretation applies for tests of the other 5 five countries that are shown from Table 4 to Table 8. In all of the 6 countries and for all variables, the null hypotheses are not rejected, so the data are random walk with possible drift except for
done in Table 9 to double-check if the time series of GDP per capita of Turkey is stationary. As a result, shown in Table 9, the p-value of the unit root test of GDP per capita in Turkey is 0.9187, which suggest that the time series is non-stationary.
It is worth noticing that there the many test statistics from Table 3 to Table 8 are missing, with a p-value of 1, which strongly suggest the non-stationarity of time series. This can be confirmed by checking the data listed in Table 10. It can be seen that the numbers of the
employment rate of Canada are very close and are even same numbers from 2013 to 2016, which are non-stationary as it has a predictable pattern in the long run.
Table 7 - Augmented Dickey Fuller Test for Turkey Test
Statistics 1% CriticalValue 5% CriticalValue 10% CriticalValue p-value GDP per capita -5.648 -4.380 -3.600 -3.240 0.0000 Employment rate -1.789 -4.380 -3.600 -3.240 0.7101 Self-reported health -0.149 -4.380 -3.600 -3.240 0.9924
Table 8 - Augmented Dickey Fuller Test for the U.S. Test
Statistics 1% CriticalValue 5% CriticalValue 10% CriticalValue p-value GDP per capita -1.482 -4.380 -3.600 -3.240 0.8353 Employment rate -0.447 -4.380 -3.600 -3.240 0.9854 Self-reported health -1.342 -4.380 -3.600 -3.240 0.8772
Table 9 - Breitung unit-root test for Turkey
Statistics p-value
lamda 1.3965 0.9187
Table 10 - Employment rate of Canada
Year Employment rate
2013 72
2014 72
2015 72
2016 72
2017 73
Having non-stationary time-series data that are checked by the Augmented Dickey-Fuller test, it is now able to perform the cointegration test. According to table 11 below, where a Westerlund cointegration test has been performed for the relationship between GDP per capita and employment rate/self-reported health in Canada. The first p-value is 0.3211, and the null hypothesis of no significant correlation between GDP per capita and employment rate in Canada
cannot be rejected. While the second p-value if 0.0203, which means there is evidence of cointegration, then the two indicators are significantly correlated. The Westerlund test is performed for the other 5 countries as well, and results are shown in Table 12 to 16 below.
Table 11 - Westerlund test for Canada
H0: No correlation # panels =1
H1: Some panels are cointegrated # periods = 5 Cointegrating vector: Panel specific
Panel means: Included Time trend: Not included AR parameter: Panel specific
Statistics p-value
GDP per capita and
Employment rate 0.4646 0.3211
GDP per capita and
Self-reported health 2.0466 0.0203
Table 12 - Westerlund test for Japan
H0: No correlation # panels =1
H1: Some panels are cointegrated # periods = 5 Cointegrating vector: Panel specific
Panel means: Included Time trend: Not included AR parameter: Panel specific
Statistics p-value
GDP per capita and
Employment rate -0.3832 0.3508
GDP per capita and
Self-reported health -0.1835 0.4272
Table 13 - Westerlund test for Korea
H0: No correlation # panels =1
H1: Some panels are cointegrated # periods = 5 Cointegrating vector: Panel specific
Panel means: Included Time trend: Not included AR parameter: Panel specific
reported health
Table 14 - Westerlund test for Mexico
H0: No correlation # panels =1
H1: Some panels are cointegrated # periods = 5 Cointegrating vector: Panel specific
Panel means: Included Time trend: Not included AR parameter: Panel specific
Statistics p-value
GDP per capita and
Employment rate -0.5724 0.2835
GDP per capita and
Self-reported health 3.0425 0.0012
Table 15 - Westerlund test for Turkey
H0: No correlation # panels =1
H1: Some panels are cointegrated # periods = 5 Cointegrating vector: Panel specific
Panel means: Included Time trend: Not included AR parameter: Panel specific
Statistics p-value
GDP per capita and
Employment rate 0.0182 0.4927
GDP per capita and
Self-reported health 3.0698 0.0011
Table 16 - Westerlund test for the U.S.
H0: No correlation # panels =1
H1: Some panels are cointegrated # periods = 5 Cointegrating vector: Panel specific
Panel means: Included Time trend: Not included AR parameter: Panel specific
Statistics p-value
GDP per capita and
Employment rate 0.5606 0.2875
GDP per capita and
According to Table 12 to Table 16, there is no correlation between GDP per capita and the employment rate for all 6 countries as all p-value are larger than 0.1 (10% level). However, for Canada, Mexico and Turkey, there is a strong correlation between GDP per capita and self-reported health, where the p-value is 0.0203, 0.0012, 0.0011, respectively.
5 Discussion
According to the results found and the analysis in the last section, the first hypothesis of this paper which states there is a correlation between GDP per capita and employment is not supported. The second hypothesis that states there is a correlation between GDP per capita and self-reported health indicator is partly supported as the significant correlation appears only for Canada, Mexico and Turkey.
It is necessary to discuss the reason for the difference between hypotheses and results. Despite the discussion before, which leads to an assumption of cointegration relationship, the empirical result of no correlation between GDP per capita and employment rate makes sense. The employment rate is defined as the number of working people between 15 to 64 over the total population at the same age; it is therefore by definition supposed to remains at a stable level in normal situations because of the stable population and working structure, significant changes in employment rate only happen in cases like the recession and global infectious diseases. On the other hand, the GDP per capita has an upward trend, and it continues to grow despite little or no increase in employment rate, the result of the cointegration test is therefore reasonable. However, due to the fact that this paper only considered data over a 5-year range and data for 6 high and middle-income countries, the external validity of the test may not be good as employment rate change may only be seen in the long run and changes may be more obvious in low-income countries.
As for the relationship between GDP per capita and self-reported health, the results confirm the hypothesis, probably for reasons that health contributes to higher productivity but less likely for the reverse. This is because according to the empirical results, both middle-income countries show a significant relationship, while only 1 high-income country has a result of strong
correlation. As many high-income countries already have a mature healthcare system and advanced medical facilities, further improvement has little effect on economic development. However, for middle-income countries that still have a large space of improvement in the healthcare system, the marginal effect of development might be large.
Comparing the data and results between the U.S., Canada with Japan and South Korea, there are little difference besides the significantly low self-reported health indicator for the latter two. Thus the culture difference instead of country size seems to play an role in social welfare, this findings agreed with other literature as well (Balestra, Boarini & Tosetto, 2018; Fleurbaey, 2009).
6 Conclusion
This paper examined the question of how well does GDP per capita capture the welfare regarding employment and health using the cointegration test. Through interpreting the results of tests, it was found that there is no evidence for a strong correlation between GDP per capita and employment rate, and for middle-income countries, there is a significant correlation between GDP per capita and the health welfare. In conclusion, GDP per capita may capture some aspects of welfare, but the correlation is different across subgroups, so in general GDP per capita lack the ability to capture welfare indicator.
The possible limitations of this paper are the data used, the measures that are chosen. The data used by this paper only covers a period of 5 years which is a relatively short period, while most of the relationships regarding macro indicators can only be seen in the long run as time lag of political or technical effects. As mentioned in previous paragraphs, this paper was intended to compare data with different income size and geographical groups of countries. However, due to the limited numbers of countries covered so far by the Better Life Index, only 6 countries are selected to make a rough comparison regarding those aspects. Moreover, as an attempt to consider the ability of GDP capturing social welfare under the current world situation, measurement of job dimensions like the labour market insecurity and measurement of health dimensions like information about the underlying diseases and healthcare insurance may be better choices than what has been chosen.
Despite the many limitations, this paper provides a different view of the question of GDP as a welfare measurement and leaves some important experience which can contribute to better designs and measures. Moreover, the finding of a strong correlation between GDP per capita and health status, especially for the possible direction from the health status to GDP per capita, may provide some insights for policymakers. Based on the limitations of this paper, future researches that study correlations in the long run, cover more countries and use measures from different aspects are encouraged.
REFERENCES
Balestra, C., Boarini, R., & Tosetto, E. (2018). What Matters Most to People? Evidence from the OECD Better Life Index Users’ Responses. Social Indicators Research, 136(3), 907–930. Beauchamp, Z. (2020, May 14). Canada succeeded on coronavirus where America failed. Why?.
Vox. Retrieved from https://www.vox.com/2020/5/4/21242750/coronavirus-covid-19-united-states-canada-trump-trudeau
Bergh, J. (2009). The GDP paradox. Journal of Economic Psychology, 30(2), 117–135. Jones, C., & Klenow, P. (2010). Beyond GDP? Welfare across Countries and Time. National
Bureau of Economic Research.
Boarini, R., & D’ Ercole, M. (2013). Going beyond GDP: An OECD Perspective*. Fiscal Studies, 34(3), 289–314.
Boarini, R., Å. Johansson and M. Mira d'Ercole (2006), "Alternative Measures of Well-Being", OECD Economics Department Working Papers, No. 476, OECD Publishing, Paris. Callen, T. (2020). Finance & Development. Finance & Development | F&D. Retrieved from
https://www.imf.org/external/pubs/ft/fandd/basics/gdp.htm
Fleurbaey, M. (2009). Beyond GDP: The Quest for a Measure of Social Welfare. Journal of Economic Literature, 47(4), 1029–1075.
Fogel, R (2004) The Escape from Premature Hunger and Death, 1700-2100: Europe, America and the Third World, Cambridge University Press.
Fuchs, V. (2013). The gross domestic product and health care spending. The New England Journal of Medicine, 369(2), 107–109.
Heshmati, A. (2018). Causality between Gross Domestic Product and Health Care Expenditure in the Augmented Solow’s Growth Model. Ukh Journal Of Social Sciences, 2(2), 19–30. Kochhar, R. (2020, June 11). Unemployment rose higher in three months of COVID-19 than it
did in two years of the Great Recession. Pew Research Center. Retrieved from https://www.pewresearch.org/fact-tank/2020/06/11/
Tsai, M., Nitta, M., Kim, S., & Wang, W. (2016). Working Overtime in East Asia: Convergence or Divergence? Journal of Contemporary Asia, 46(4), 700–722.
Nordhaus, W. & Tobin, J. (1973). “Is Growth Obsolete?” In The Measurement of Economic and Social Performance: Studies in Income and Wealth, National Bureau of Economic
Research, 38, 509–31.
OECD.Stat. (2020). Better Life Index. Retrieved from https://stats.oecd.org/Index.aspx? DataSetCode=BLI
OECD.Stat. (2020). GDP per capita levels. Retrieved from https://stats.oecd.org/Index.aspx? DataSetCode=PDB_LV
OECD.Stat. (2020). Health Status. Retrieved from https://stats.oecd.org/index.aspx? queryid=24879
Proto, E., & Rustichini, A. (2013). A Reassessment of the Relationship between GDP and Life Satisfaction.(Research Article). PLoS ONE, 8(11), e79358.
WHO. (2020). Situation Report - 160. Retrieved from https://www.who.int/emergencies/ diseases/novel-coronavirus-2019/situation-reports
Sen, A. (1992). Inequality reexamined. Russell Sage Foundation.
Swift, R. (2011). The relationship between health and GDP in OECD countries in the very long run. Health Economics, 20(3), 306–322.