1 Introduction

Income inequality and its consequences on human welfare are of common concern (Neckerman and Torche 2007; Wilkinson and Pickett 2009). The inequality-wellbeing link has been well discussed in the literature (see Ferrer-i-Carbonell and Ramos 2014; Schneider 2015 for excellent reviews of the topic). According to Nguyen et al. (2015), there are at least two distinct channels through which income inequality might affect life satisfaction: (1) via possible changes in an individual’s income compared to the mean income of a reference group (even though the individual’s absolute income remains unchanged); and (2) via the level of economic inequality in the community or society where the individual lives. In the first channel, studies have included relative income (measured, for instance, as the ratio of individual income to the median income of the whole population or the median income of a relevant comparison group) find that life satisfaction is strongly and consistently positively affected by relativities (Dolan et al. 2008; Nguyen et al. 2015). In the second channel; empirical evidence has consistently found a negative impact of income inequality (measured at the national or regional level) on life satisfaction (Schneider 2015). Living in a high unequal society tend to make people unhappy possibly because high inequality might cause additional social problems such as social conflicts and tensions, crimes and violence (Ferrer-i-Carbonell and Ramos 2014; Haller and Hadler 2006) or cause risks of substantial variations in their own income (Hochman and Rodgers 1969; Ravallion and Lokshin 2000). However, in some case, life satisfaction is not negatively affected by or even positively associated with high inequality if people perceive that inequality signals social mobility (Hirschman and Rothschild 1973). People feel happy because they expect that they would enhance their own living conditions in the future when they watch others that around themselves move upwards in the social ladder (Wang et al. 2015).

The empirical evidence for inequality-subjective welfare relationship remains inconclusive (Schneider 2015). A number of studies have found a negative association between happiness and inequality. Using the aggregated data of eight countries over 25 years, Hagerty (2000) find a reduction in income inequality (max, min, skew, the 80th/20th percentile ratio) for a country increased with its average national happiness. Alesina et al. (2004) discover a negative relationship between income inequality (Gini) and life satisfaction in both the United States and Europe. Similar results were also reported for 18 Latin American Countries (Graham and Felton 2006), Germany (Schwarze and Härpfer 2007) and Urban China (Smyth and Qian 2008). Using the data from the World Value Survey (WVS) over the period 1981–2004, Verme (2011) find that inequality (Gini) had a negative effect on happiness and this effect also hold for both the poor and non-poor as well as for all geographic regions. Schröder (2016) used hybrid effects regressions with the German Socio-Economic Panel data to examine differences in happiness between Germans during times of high and low income inequality. This study shows that short-run increases in inequality, rather than long-run levels of inequality, lower level of individual happiness. Using four waves of the European Social Survey (179,273 individuals from 29 countries) Hajdu and Hajdu (2014) examine the relationship of reduction of income inequality (redistribution) with subjective wellbeing. Their study provides evidence that income inequality has a negative effect on life satisfaction, whereas reduction of inequality has a positive effect on well-being.

In contrast, some studies have shown a positive effect of inequality on happiness. Using the data from 11 waves of British Household Panel Survey data, Clark (2003) finds that inequality, as measured by either Gini or the 90th/10th income percentile ratio, had a positive effect on happiness for the employed population. Similar evidence was reported in Japan by Ohtake and Tomioka (2004) who find that both the Gini coefficient and perceived income inequality had a weak but positive association with happiness. Utilizing a very large sample of the world citizens from WVS in 1995/1996, Haller and Hadler (2006) also indicated that inequality (Gini) had a positive effect on happiness. A further study of each subgroup (by region) revealed that southern American countries showed high inequality but were very happy, while post-communist countries had equality but were very unhappy. Nevertheless, other researchers have indicated no evidence for the inequality-happiness link. Diener et al. (1995) found no relationship between inequality and happiness for the student sample using the data from WVS. Using the data from WVS, Helliwell (2003) reported no association between inequality (Gini) and life satisfaction over the period 1980–1997. Income inequality (Gini) was also found not to be related to individual happiness in Russia (Senik 2009) and 28 European countries (Zagorski et al. 2014).

In Vietnam, the quality of life of older people has emerged as the main concern in contemporary academic research (Long and Pfau 2009; Pfau and Long 2010; Truong et al. 1997) as well as in the public policy agenda (Vietnam National Commitee on Aging [VNCA] 2012). There have been a huge number of studies analyzing the characteristics and quality of life of as well as social policies for Vietnamese older people (e.g., Evans et al. 2007; Long and Pfau 2009; Pfau and Long 2010; Truong et al. 1997; UNFPA 2011; VWU 2011). Among these studies, few have examined determinants of objective wellbeing (e.g., poverty or income) of the Vietnamese elderly (Long and Pfau 2009; Pfau and Long 2010). However, to the best of our knowledge, little econometric evidence exists for factors affecting subjective welfare (i.e. happiness or life satisfaction) of the Vietnamese older population.

In the current study, our main research objective was to examine whether inequality has any impact on happiness among older persons in rural Vietnam. Our study has several strong points: First, while most country specific studies focused their research on economically developed countries, few have analyzed the association between income inequality and happiness in developing countries (Wang et al. 2015). Second, while most studies use inequality measures at the national level, this study estimates economic inequality within relatively small geographic units (the commune level). To the extent that inequality is relevant to individuals’ wellbeing, it is likely that they are more aware of, and affected by, what happens in their proximate environment. Another advantage is that the population under study is more homogenous than in most studies that often examine the relationship in the whole population. One can thus get a better handle on the mechanisms within this context.

We provided the first evidence that expenditure inequality, as measured by any indicator, has a significantly negative effect on happiness. This suggests that inequality results in negative consequences on the quality of life of the older population in rural Vietnam.Footnote 1 In addition, we find that sensitivity to inequality varies considerably according to certain individual characteristics. Among others, a useful implication for social policy is that older rural people who are farmers or poor are more sensitive to inequality. Given that these people tend to be less happy than others, the result shows the risk that inequality further lowers their subjective welfare. This finding partly supports the view that rural Vietnam is a society that has less social mobility.

The structure of the paper is as follows: Sect. 2 describes the data sources and methods. Results are given in Sect. 3. Discussion, implications and limitations are reported in Sect. 4.

2 Data and Methods

2.1 Data

The study utilized the data from the Vietnam National Aging Survey 2011 (2011 VNAS). The main objective of VNAS was to collect data on older people (those aged 50 and over). 12 provinces were randomly selected from six ecological zones. In each selected province, 200 communes were randomly selected and then two villages were randomly selected from each selected commune. Finally, 15 people aged 50 years old and above were randomly selected. The total number of interviewed people is 4007, of those, 1218 were near-elderly (50–59) and 2789 were 60 and older. 3515 people were Kinh, and 492 people were ethnic minorities. 2887 people live in rural areas, and 1120 people live in urban areas.

The survey collected data on personal information (such as age, gender, marital status, religion, social activities, life style, education, employment, own income, assets, etc.) and household information (housing conditions, family relationship, living arrangements and household income). Especially, the survey collected information about the quality of life such as health status and life satisfaction.

The data on expenditure inequality and expenditure per capita at the commune level were calculated from the 2011 Rural, Agricultural and Fishery Census (2011 RAFC). A detailed description of how to estimate different measures of inequality are reported in Sub-Sect. 2.1.2 and “Appendix 1”.

2.1.1 Happiness Indicator

The measure of subjective satisfaction is the most commonly used in the happiness literature (Dolan et al. 2008; Ferrer-i-Carbonell and Ramos 2014; Schneider 2015). The outcome variable in our study is the life satisfaction or happiness scores of respondents, obtained from a multiple-choice question: “Taken all together, how are you satisfied with your life at present?” The five possible responses to the question are “very dissatisfied”, “dissatisfied”, “neither satisfied nor dissatisfied”, “satisfied”, and “very satisfied”. For our analysis, happiness is constructed with a value ranging from 1 to 5, corresponding to “very dissatisfied”, “dissatisfied”, “neither satisfied nor dissatisfied”, “satisfied”, and “very satisfied”, respectively.

2.1.2 Measures of Economic Inequality

In this study, we used household consumption expenditure to calculate various measures of economic inequality instead of using household income. This is because expenditure inequality indicates more about the longer-run, or lifetime, differences in living standards between people while measures of income inequality provide us with a snapshot of income disparities across the population (Goodman and Oldfield 2004). In addition, consumption expenditure is associated with less measurement error than income data and commonly regarded as a better proxy for household wellbeing (Deaton 1997). In developing countries, income data tend to be less reliable due to income fluctuations in harvest cycles in rural areas or irregular income flows from the large informal sector in urban areas and difficulty in calculating income for self-employment activities (Coudouel et al. 2002).

In our study, a challenge is to how to estimate the inequality at the commune level. To calculate the expenditure inequality of a commune, one needs data on expenditure of households within the commune. However, expenditure data are only collected in sampled surveys which have small sample size and cannot be used to estimate the expenditure inequality of communes. To address this challenge, we used a small area estimation method proposed by Elbers et al. (2002, 2003) which first estimates a model of expenditure using a household survey, and then applies this estimated model to a census to predict expenditure inequality of communes.Footnote 2

The method requires data from a census and a household survey which contain expenditure data. In this study, we used census data from the Rural Agriculture and Fishery Census (RAFC) in 2011. The RAFC was carried out by GSO in July 2011. The census covered all households in rural areas. The census contains data on individuals and households including basic demography, employment and housing, and agricultural activities. There are 16,194,218 households covered in the census. The household survey is from Vietnam Household Living Standard Survey (VHLSS) in 2010. This was also conducted by GSO. It samples 9399 households and contains detailed data on characteristics of households and individuals. More important, it contains data on household expenditure, which were collected using very detailed questionnaires on different expenditures of households.

The method can be described in the following steps. First, we define common explanatory variables in both the census and the household survey. These variables are comparable in terms of definition and mean statistics. The common variables can include household-level variables and commune-level variables. Secondly, we regress the log of per capita consumption on the selected variables using the household survey as follows:

$$\ln \left( {y_{ic} } \right) = X_{ic} \beta + Z_{c} \gamma + \varepsilon_{ic} ,$$
(1)

where ln(y iv ) is log of per capita expenditure of household i in commune c, X ic the vector of the common household-level variables, Z c is the commune-level variables, β and γ are the vector of regression coefficients, ε iv the error term. In the Elbers et al. (2002, 2003) framework, the error is allowed to be correlated within communes. Model (1) is estimated using data from the 2010 VHLSS.

In the next step, the estimated parameters of model (1) are used in the 2011 RAFC to predict expenditure of all the households:

$$y_{ic}^{Census} = \exp \left( {X_{ic}^{Census} \hat{\beta } + V_{ic}^{Census} \hat{\gamma } + \hat{\varepsilon }_{ic} } \right),$$
(2)

The predicted expenditure will be used to calculate inequality indexes of communes. The standard errors of the inequality indexes are estimated using Monte-Carlo simulations. More specifically, we first estimate the distributions of \(\hat{\beta },\;\hat{\gamma }\) and \(\hat{\varepsilon }_{iv}\), then in each simulation, values of these parameters are drawn randomly from their estimated distributions, and used to estimate of the inequality indexes of small areas. After a number of simulations, we can get the sampling distribution of the inequality indexes and calculate the standard deviations of the inequality indexes.

To measure inequality, we use most common measures of inequality: the Gini coefficient, Theil’s L index of inequality, and Theil’s T index of inequality and the ratio of the 90th percentile to the 10th percentile. See “Appendix 1” for a detailed description of different measures of inequality in the current study.

2.1.3 Other Explanatory Variables

Life satisfaction or happiness is determined by a large number of different factors. Following the literature (e.g., Brown and Tierney 2009; Cheah and Tang 2013; Dolan et al. 2008; Gray et al. 2008; Morawetz et al. 1977; Nguyen et al. 2015; Schneider 2015; Smyth and Qian 2008; Sumngern et al. 2010), a set of control variables, including individual and household characteristics, were included in the econometric models. The definition and measurements of the variables are given in Table 2. The literature indicates that both absolute and relative incomes often have a positive relationship with happiness (Ball and Chernova 2008; Celse 2016; Dolan et al. 2008; Oshio et al. 2011). Hence, we included both absolute and relative income in the models, which were expected to be positively associated with individual happiness.

The other socio-economic control variables include age, gender, ethnicity, family status, religion, social activities, education, employment, and health. Among other factors, health status is predicted to be most positively associated with happiness. Empirical evidence often shows that both religious participation and frequency of worship are positively linked with life satisfaction (Dolan et al. 2008; Krause 2003; Myers 2000). Therefore, religious individuals were expected to be more likely to be happier than those without religion. The level of happiness is also expected to have a positive relationship with frequency of worship.

2.2 Econometric Models

The following equation was used to estimate the impact of inequality on individual happiness:

$$Happiness_{ij} = \alpha + \beta_{1} Lnex_{j} + \beta_{2} Ineq_{j} + \beta_{3} X_{ij} + e_{ijt}$$
(3)

where i and j are subscripts for individual and commune, respectively. X ij is the vector of other individual, household and location variables. Happiness ij represents the respondent’s self-reported happiness. Ineq j stands for the commune-level expenditure inequality as measured by various indicators, while Lnex j denotes the log-transformed commune per capita expenditure. Controlling for the average area—level income or expenditure helps mitigate the negative relationship between inequality and subjective welfare (Oshio and Kobayashi 2011).

Happiness can be used as cardinal or ordinal, depending on researchers’ assumption (Ferrer-i-Carbonell and Ramos 2014) and the results of econometric analysis are robust to both means of a liner or an ordered categorical estimator (Ferrer-i-Carbonell and Frijters 2004). Since Ordinary Least Squares (OLS) coefficients directly denote the marginal effects (Wooldridge 2013) and thus are more intuitive and interpretable by a wide range of readers (Jiang et al. 2012), we chose OLS models to examine the effect of inequality on individual happiness. However, ordered logit models were also estimated to check for the robustness to the model specifications.

We also examined some type of heterogeneity on inequality impact. Specifically, we empirically investigated whether inequality impact is different across groups such as gender, poverty, religion, and employment. In this case, a common approach is to use an interaction term between the inequality measures with the variable describing the dimension creating heterogeneity (Ferrer-i-Carbonell and Ramos 2014). In addition, we ran different models with the inclusion of the squared terms of inequality, expenditure level, and age of respondents, to capture potential non-linear effects. However, we found no non-linear effects and therefore we do not report the results from regression models using the squared terms.

It should be noted that the data for individual happiness were collected for those aged 50 and over and thus younger adults are not included in our analysis. In addition, because the 2011 RAFC was conducted in rural areas, we did not have census data for calculating inequality indicators in urban areas. Thus, our study only examines the happiness-inequality link among old people in rural areas.

3 Results

3.1 Descriptive Statistics

Table 1 reports the sample summary statistics about happiness. About two-thirds of all respondents reported being happy or very happy while around 23% said that they were neither satisfied nor dissatisfied and about 12% being dissatisfied or very dissatisfied. A close look at the data by gender shows that men seemed to be happier than women. The happiness status of farmers and non-farmers is quite similar. The happiness status is slightly different between those with and without religion. However, the poor/non-poor disparity in happiness tended to be much larger with 69% of the non-poor self-reporting to be happy or very happy, compared with about 51% of the poor.

Table 1 Distribution of individual happiness by gender, employment, religion and poverty.
Full size table

Table 2 shows that the average age for respondents in the sample is 66.4 and men account for about 41% of the sample. 86% of the sample are the ethnic majorities (Kinh and Hoa) while 14% are ethnic minorities. Regarding employment status, 42% of individuals reported that they are farmers and 58% are not farmers. Only 19 and 21% of respondents completed the primary school and secondary school (or higher level), respectively. 30% of respondents are widowed and 33% lived with their children or grandchildren. 43% of the sample participated in at least one social activity and 19% did not engage in any religion activity. The data show that 68% of respondents self-rated their health as bad or very bad, while 57% possessed normal health and only 5% had good or very good health.

Table 2 Sample summary statistics.
Full size table

Looking at economic status, about 53% of respondents estimated that their total household income ranging from 10 million dong (VND) to less than 50 million VND. 24% earned total household income less than 10 million VND while 23% with total household income equal or higher than 50 million VND. Regarding relative income, the data reveal that half of respondents self-rated that their household income lowers than neighbours. 38% of surveyed individuals responded that their household income is about as same as neighbours while 12% reported that their household income higher than neighbours.

3.2 Impact of Inequality on Happiness

Table 3 presents the result from the model in which expenditure inequality is measured by the Gini coefficient. While age has a positive effect on happiness among older people in rural Vietnam, other variables such as gender, ethnicity and employment are not associated with happiness. The result indicates that individuals with secondary or higher schooling tend to be happier than those without primary school. However, the same relationship does not hold for those with primary school. Same results are also reported for China (Brown and Tierney 2009), where age and education are positively linked with life satisfaction among the elderly. We also find that widowed people are less happy than married people. The result is partially consistent with Chyi and Mao (2012) who find that being widowed is negatively correlated with life satisfaction among older men in China. Our research result confirms no difference in happiness between those living with and without children/grandchildren. Similar evidence is also found among the elderly in rural Thailand (Gray et al. 2008). This finding suggests that living with one’s own child might have a net zero impact on an elderly’s happiness. As discussed by Chyi and Mao (2012), on the one hand, living with their children/grandchildren helps older people with easy access to receive emotional and daily life support, which can increase their life satisfaction. On the other hand, coresiding with their children/grandchildren is likely to cause tensions and conflicts, which in turn can make the elderly less happy.Footnote 3

Table 3 Factors associated with happiness
Full size table

Involvement in social activities also has a positive association with happiness among the Vietnamese elderly. This is in accordance with previous findings that there is a positive link between having membership in groups/organizations and life satisfaction (Dolan et al. 2008; Matsushima and Matsunaga 2015). Our research result shows that there is no difference in happiness between religious people and their godless counterparts. However, happiness is found to be positively linked with a higher frequency of worship. Specifically, individuals that had weekly or daily worship were happier than those that only had worship at special events. However, this is not the case for those conducting worship monthly. Consistent to the literature (Dolan et al. 2008; Kingdon and Knight 2007; Wang et al. 2015), we find that health has a substantial and positive relationship with happiness. As expected, it is found that both absolute and relative incomes have strongly positive effects on happiness. Holding all other variables constant, individuals that belong to high income and middle income households would have life satisfaction scores that were 0.28 points and 0.13 points higher than those of individuals in low income households, respectively.

Regarding the impact of inequality on happiness, the coefficient of inequality variable in Table 3 is negative and highly statistically significant (p < 0.01). This confirms that individuals who lived in the commune with higher inequality tend to feel less happy than those living in the commune with higher equality. Holding all other variables constant, a 10 percentage point increase in the Gini coefficient associated with a 0.144 point decline in life satisfaction scores. This suggests that inequality results in negative consequences on the quality of life of the older population in rural Vietnam. The finding is in line with previous findings for Australia (Nguyen et al. 2015), Japan (Oshio and Kobayashi 2011), the United State (Ahn et al. 2015) and the Europe (Alesina et al. 2004) but contrasts with reported results for China (Wang et al. 2015), Russia and Latin America (Eggers et al. 2006; Senik 2004).

We also find that the living standards within the commune (measured by the average expenditure per capita at the commune level) are positively linked with individual happiness. The result is similar to that by Kingdon and Knight (2007) who found that the income per capita of the small local residential community has a positive effect on individual happiness in South Africa. With respect to regional variables, the result indicates that individuals with equal individual, household and other characteristics will on average have life satisfaction scores that are higher in the North than in the Central. Nevertheless, no difference in individual happiness is found between the South and the Central.

Economic inequality might affect life satisfaction via individual characteristics because inequality impact might differ across groups (Ferrer-i-Carbonell and Ramos 2014). We also examine whether heterogeneity on inequality impact exists by interacting a number of key individual characteristics with inequality in regression models. In Table 4, Column (2 and 3), the coefficient of the interaction term between inequality and farmer is negative and statistically significant at the 0.05 level. This demonstrates that inequality impact is stronger for farmers than for non-farmers. The result also indicates inequality impact is stronger among the poor than the non-poor. Looking at the interaction terms for expenditure inequality with gender, ethnicity and religion in Columns (1), (5) and (6), respectively, neither of the interactions shows statistical significance at the 0.1 level. This indicates that no difference exists in the perception of inequality between men and women, between ethnic minorities and majorities, and between those with and without religion.

Table 4 Comparing the inequality sensitivities by key individual characteristics
Full size table

3.3 Robustness Checks

In the empirical literature, inequality is always measured by the Gini coefficient in the region or country where the individual lives and few studies examined the robustness of the results to the different ways of measuring inequality (Ferrer-i-Carbonell and Ramos 2014). In this study, we checked for the robustness of the results via investigating the association between individual happiness and inequality as measured by different methods. As noted by Ferrer-i-Carbonell and Ramos (2014), a robustness test would help identify what type of inequality individuals are more sensitive to. Table 5 shows that inequality, as measured by any indicator, has a significantly and negative effect on happiness. This confirms that the result is robust to different measures of inequality.

Table 5 The association between happiness and inequality as measured by different indicators
Full size table

In addition, for the robustness test for the model specifications, we also ran several ordered logit models with the same variables. The ordered logit and OLS results are very similar: there is no difference at all in the sign and the significance levels are almost the same for each of coefficients (see “Appendix 2”). This result is similar to that in a methodological paper by Ferrer-i-Carbonell and Frijters (2004), who checked for the robustness of findings on the determinants of happiness using the German national household panel survey and concluded that the results are not sensitive to the choice between a linear and an ordered categorical estimator.

4 Discussion, Implications and Limitations

4.1 Discussion

The main purpose of this study was to explore how expenditure inequality contributes to the elderly’s happiness in rural Vietnam, based on a combination of the data from the 2011 VNAS and 2011 RAFC. Our regression analysis confirms that individuals who live in the commune of high inequality tend to report themselves as being less happy, even after controlling for several individual and household attributes. The results are robust to different measures of inequality as well as the specification of econometric models. This result is consistent to previous findings in many countries.

The result of our study indicates that farmers and the poor are more sensitive to inequality possibly because they tend to face more uncertainty about their income and economic activities than others. Psychological consequences of inequality have been argued by Neckerman and Torche (2007) that “living in a context of high inequality might intensify feelings of relative deprivation among low-income individuals”. Wilkinson and Pickett (2009) gives a perspective in which the differences in social status become more pronounced in highly unequal societies, resulting in feelings of stress and unrelatedness. Another possible explanation is that farmers and the poor might believe that social mobility is low and that they are likely to remain in their current disadvantageous status, which makes them to be more negatively affected by inequality. Our finding, in part, supports the view that rural Vietnam is not a mobile society. The finding is consistent with the reported result for Japan (Oshio and Kobayashi 2011) and Europe but inconsistent with that for America (Alesina et al. 2004). Alesina et al. (2004) find that income inequality has a stronger effect on individual happiness among low-income Europeans, but not among low-income Americans. The authors argue that these findings were in line with the perception that Europeans live in less mobile societies while Americans live in a mobile society.

The current study also answered the question: To what extend do absolute income and relative income affect individual happiness? It was found that both the income of the household and the income of other households have a significant effect on subjective well-being. Some of our estimates suggest that the latter relative to the former is even more important than the former on its own. In addition, the result of the current study confirms that the living standards within the commune (as measured by the average expenditure per capita at the commune level) have a positive effect on individual happiness. We also find that health status is an important determinant of happiness among the elderly.

4.2 Implications

The current study finds that individuals who live in higher-inequality communities tend to be less happy. This finding implies that inequality results in negative consequences on the quality of life among older people. We also find that older rural people who are farmers or poor are more sensitive to inequality. Given that these people tend to be less happy than others, the result shows the risk that inequality further lowers their welfare. A policy implication here is that social policies for the elderly should focus on farmers and the poor since they are more vulnerable to inequality. As mentioned by Oshio and Urakawa (2014), individuals with low income status tend to feel more disappointed with economic inequality, when living in a less mobile society; this is because inequality implies that they have limited chances to move up the social ladder. Thus, the finding of our study partly supports the view that rural Vietnam is not a mobile society, which is another interesting topic, should be addressed in future research. Our study suggests that while both wealth and health are much of importance to the subjective welfare of older people in rural Vietnam, improving social mobility is also vital for increasing their well-being.

4.3 Limitations

We acknowledge that our study has some limitations. First, like many other happiness studies, our study considers happiness only as a single term and is based on the survey results of a subjective assessment. Since happiness is multi-dimensional, the validity of perceived happiness as reported from the survey should be further addressed. Second, we are unable to examine the link between inequality and happiness over time due to lack of longitudinal data. As noted by Ferrer-i-Carbonell and Ramos (2014), the use of panel data in estimating a happiness equation reduces the bias because it controls for time invariant unobservable individual characteristics. This suggests that further research is needed to address this issue. Third, while our study supports the negative association between inequality and happiness, we do not give sufficient explanations about how and why it was the case. This is because the data available did not contain specific information with which we can clarify the mechanism through which inequality affects happiness. This is another interesting topic should be investigated in future research. Fourth, due to the fact that the data are available for only the elderly (those aged 50 and over) and rural areas, our findings were observed only among old people in rural Vietnam. It should be noted that different groups might have different tastes for inequality (Verme 2011). This implies that the implication should not be generalized to the whole population in rural Vietnam. Future research should examine the happiness-inequality relationship for the whole population, given that data are available for all age groups and both rural and urban areas.