The development of a new multi-faceted model of social wellbeing: Does income level make a difference?

Abstract

Recent research has suggested that income, while playing a part in quality of life, may have only a limited impact on a multi-faceted concept such as social wellbeing. Using data from an Australian household survey (Living in Queensland Survey), a composite Wellbeing Index was created that covered objective circumstances, with known associations to wellbeing, evaluated from the individual’s subjective viewpoint. The importance attributed to each dimension added to the robustness of the measure. The measure was then used to explore the impact of income on wellbeing using various specifications of income. The results indicate that while income is a statistically significant predictor, its effect on wellbeing is small compared with other socio-demographic variables such as health, marital status, employment status and age. The study contributes to the contemporary debate on social wellbeing and adds new evidence to a body of research that has been mainly based on European and American data.

Keywords

income quality of life social inequality social wellbeing

Policy-makers and social researchers have long been concerned with measuring and comparing the overall wellbeing of societies, communities and individuals. Key indicators of social and economic progress are designed to assist policy-makers with choices that facilitate better targeting of government actions while ensuring long-term economic and social sustainability. However, while the term ‘wellbeing’ is commonly used, it is inconsistently defined and its utility for social and economic research has been the subject of considerable reservation. Particular concerns are the subjective nature and global focus of such measures. In addition, there is growing consensus in the literature that conventional, market-based measures of income and consumption are an insufficient metric to assess social wellbeing (Diener and Seligman, 2004) and should be complemented by non-monetary indicators of people’s life circumstances. In order to address these concerns, this article considers the validity of a multi-dimensional, composite index of wellbeing designed to be a concise but comprehensive measure incorporating the objective aspects of wellbeing that people deem to be important. The relationship between wellbeing and income is then explored using data from an Australian survey, contributing to the evidence base that so far has been mostly grounded in data collected in Europe or America.

Income and wellbeing

For a number of decades, the level and change in gross domestic product (GDP) per capita have been used as the main yardsticks for measuring and comparing the wellbeing of populations across countries (Giovannini et al., 2007). However, a number of studies, reflecting what has become known as the Easterlin Paradox, have suggested that once a certain level of material needs has been met, further increases in GDP are unlikely to yield the same enhancement of wellbeing. In developing countries, wellbeing appears to rise with national income. However, in highly developed economies, per capita income growth tends to be associated with increases in wellbeing that are so small they are almost undetectable (Giovannini et al., 2007). This finding supports the premise that wellbeing is strongly influenced by social comparisons. Because relative status is a zero-sum game at a societal level, national income increases in developed countries have a lesser effect on wellbeing due to the decreasing utility of the consumption benefit (Clark et al., 2008; McBride, 2001). The other issue concerning aggregate measures such as GDP per capita is that they provide only an economy-wide measure. In contrast, the concept of wellbeing refers to the situations of individuals and households.

While the need to look beyond GDP per capita at the macro level has been recognized (Stiglitz et al., 2009), individual or household income has generally retained its place as a fundamental measure of the success of policy interventions. Yet there is increasing evidence that the relationship between life satisfaction and individual or household income is generally not strong and not proportional to differences in income (Blanchflower and Oswald, 2004; Di Tella and MacCulloch, 2008; Headey et al., 2008). In essence, the relative nature of the income effect is well established; relative both to what one already has and to what others have (Easterlin, 2003; Clark et al., 2008). Moreover, changes in individual or household income tend not to be associated with comparable changes in subjective wellbeing, largely as a result of adaptation and habituation processes (Clark et al., 2008; Kahneman and Deaton, 2010). Therefore, differences in personal income provide only a very partial explanation of differences in individual wellbeing, particularly in highly developed countries (Diener and Seligman, 2004: 2), and particularly if only linear effects of individual incomes are examined.

There is an accumulating body of evidence that, while the reported wellbeing of individuals and households is influenced by net household income, it depends to a much larger degree on a range of other non-monetary indicators of social conditions, such as employment, health, housing, personal security and family relationships (Di Tella and MacCulloch, 2008). There is important policy relevance for the incorporation of such non-monetary indicators of wellbeing into a composite indicator based on micro-data at a household level (Giovannini et al., 2007).

An approach that accounts for some of the factors influencing social wellbeing that are omitted from conventional economic accounts is provided by social indicators covering specific aspects of wellbeing (Cummins et al., 2003; OECD, 2007). Such indicators look to a broader definition of wellbeing based on composite measures of dimensions of human development, providing an overview of social trends in family characteristics, employment rates, jobless households, crime victimization, conviction rates, life expectancy, and infant mortality. However, such indicators tend to be aggregate measures better suited for national initiatives and do not easily lend themselves to analyses of changes within families or communities which are often the concern of policy-makers. Research using social indicators often focuses on the material conditions of people’s lives but these are poor predictors of the actual quality of people’s lives (Cummins et al., 2003; Veenhoven, 2005).

Some researchers have responded to these issues by focusing on more subjective global measures of life satisfaction or happiness (Hills and Argyle, 2001; Kahneman and Krueger, 2006; Layard, 2005). Other recent literature in this domain has tended to view wellbeing as a multi-dimensional set of elements covering, for example: physical, psychological, cognitive, social, and economic factors (Decancq and Lugo, 2008; Lent, 2004; Pollard and Lee, 2003); material wellbeing, health, productivity, intimacy, safety, community, and emotional wellbeing (Cummins et al., 2003); or family economic wellbeing, social relationships, health, educational attainments, community connectedness and emotional wellbeing (Land, 2010). Cummins et al. (2003) developed a national index of subjective wellbeing, the Australian Unity Wellbeing Index, comprising two subscales: Personal and National Wellbeing. The Personal subscale has seven domains: standard of living; health; achievement in life; personal relationships; how safe you feel; community connectiveness; and future security. The National subscale has three domains: economic situation; state of the environment; and social conditions. While such approaches are undoubtedly fruitful, the selection of the specific components of wellbeing is often problematic and can be somewhat arbitrary. In this article, we introduce a more transparent way of selecting wellbeing indicators, based on what people consider to be important elements of their life situation.

The contribution of this article is therefore two-fold. First, it develops and justifies an index of wellbeing based on key non-monetary indicators that incorporates a judgement of the relevance of the dimensions covered. Second, it adds to the ongoing debates on the links between wellbeing and income by providing a comprehensive treatment of the topic, using both individual and household incomes and testing both linear and non-linear forms of the relationship.

A multi-dimensional index of wellbeing

Wellbeing is a complex and multi-dimensional concept that impacts on society in both direct and indirect ways but the factors that constitute wellbeing have been, to a large extent, determined by the normative and disciplinary orientations of individual researchers. For this article, and the larger study to which it refers, we have developed a composite index of social wellbeing designed to be a concise but comprehensive measure that incorporates the objective aspects of people’s life circumstances. Crucially, the index only incorporates the components that people deem to be important to the quality of life. Wellbeing is defined, not by each person’s objective circumstances but by their subjective experience of those objective circumstances. We have eschewed more subjective global questions such as ‘I am satisfied with my life’. Instead, we focus on the elements described in the literature as critical components of quality of life: health (Frey and Stutzer, 2002), family relationships (Diener, 2000), personal security (Cummins et al., 2003), housing (Pacione, 2003), natural environment (Brereton et al., 2008; Welsch, 2006), work (Layard, 2005), financial assets (Lent, 2004), income (Stutzer, 2004), access to essential items and services (Saunders, 2011), social respect (Powdthavee, 2008), resilience to stress (Layard, 2005) and leisure opportunities (Han and Patterson, 2007).

The resulting measure of wellbeing provides us with a comprehensive measurement tool to evaluate wellbeing, which replaces what might be termed the traditional market-based measures of household income and consumption, or the more subjective cognitive approaches to happiness. The approach to wellbeing we have adopted, by aggregating the most salient measures of quality of life in a parsimonious way, provides us with a comprehensive measurement tool to evaluate wellbeing, which draws on a thorough analysis of what really matters for individuals and families and what determines the quality of their lives.

Data and methods

The data used for this article was obtained from the Study of Social Wellbeing (aka Living in Queensland Study) which involved a representative survey of Queensland private households (Boreham et al., 2009). One person per household was selected using a random sampling technique stratified by region, age and gender. The social wellbeing survey comprises two questionnaires, the Personal form and the Household form; the sampled person was responsible for completing the Personal form. An appropriate person from the sampled household was then asked to provide information about the household as a whole. The Personal form comprised questions relating to: demographics; income; employment status; workplace conditions; social participation; quality of life; and health. The Household form comprised questions relating to: household composition; housing; property and investments; household income and household expenditure and living standards. The items used to create the wellbeing measure were taken from the Personal form, while the independent and control variables used in the analyses were taken from both the Personal and Household forms.

The data used in this article was collected between May and October 2009. For the purpose of this research, only respondents of working age (18 to 65 years) were included in the analysis, which resulted in a sample of 2143 individuals.

While there are other Australian datasets such as the Household, Income and Labour Dynamics in Australia (HILDA) survey that could be used to measure wellbeing, the dataset we use has two key advantages. First, it offers a more comprehensive coverage of the facets of wellbeing compared with many other surveys, including HILDA. Second, unlike other studies, it incorporates questions about the importance of each wellbeing dimension that were pivotal for the development of our wellbeing measure.

An index of wellbeing

The Study of Social Wellbeing aimed to compile an index that was comprehensive but concise and that would cover the multiple facets of the concept discussed in the wellbeing literature. Respondents were asked to provide subjective ratings of their satisfaction with a range of aspects on a Likert scale of 1–7. Crucially, respondents were also asked to rate the importance of a number of components to their overall sense of wellbeing. Using this information, we were able to include only the items that a vast majority of people – at least three-quarters – regard as important components of wellbeing. Since the selection of such a cut-off point is inevitably somewhat arbitrary and different criteria can lead to different results (see e.g. Saunders and Naidoo, 2009), the list of selected items was thoroughly screened for conceptual consistency and generalizability to the entire population. Twelve items were selected as a result of this process, with the omitted items generally covering aspects such as involvement in the community, education or access to public services.

In order to test the construct validity of the index, specifically its convergent validity, we examined the correlations between the 12 selected items, which suggested that they measure a consistent construct. To explore the construct validity further, we ran a Principal Component Factor Analysis on the items, which elicited a two-factor solution. These factors indicated no meaningful underlying constructs and items loaded on both factors. These findings and the theoretical grounding that these items measure the construct wellbeing supported the decision to use a single-factor solution and retain all 12 items in the index. All factor loadings were greater than 0.55, and all but one were 0.6 or higher, indicating a good convergent validity of the measure. The Cronbach’s alpha coefficient was 0.89, indicating a high internal consistency of the index.

Table 1 presents the importance and satisfaction with each of the 12 items included in the model and their factor loadings. For illustrative purposes the satisfaction scores were grouped into three categories: dissatisfaction (scores 1–3), neutral (scores 4), and satisfaction (scores 5–7). For the actual index an average of the 12 scores measured on the full, 7-point scale was used.

Table 1.

Perceived importance and satisfaction percentages for each element of wellbeing – 18–65 years old (sorted by importance).

Item description	n	Importance	Dissatisfaction	Neutral	Satisfaction	Factor loading
		%	%	%	%
Your health	2069	93.80	24.26	15.22	60.51	0.60
Your family relationships	2065	92.92	7.46	8.09	84.46	0.56
Your ability to afford essential items	2065	88.62	15.59	11.86	72.54	0.77
Your housing or accommodation	2064	88.54	7.66	8.96	83.38	0.64
Your income	2068	87.66	24.61	14.56	60.83	0.73
Your savings and other financial assets	2062	86.16	32.74	12.80	54.46	0.75
Your personal security	2062	85.56	6.89	13.43	79.68	0.71
Your natural environment	2072	83.13	5.60	15.69	78.72	0.68
Your leisure opportunities	2072	81.63	16.70	16.99	66.31	0.67
The respect you are accorded by others	2066	80.81	6.24	18.73	75.02	0.72
Your job or your work	1973	80.44	17.94	19.31	62.75	0.66
The level of stress that you normally feel	2062	77.19	28.71	25.17	46.12	0.61

While deciding on the final form of the wellbeing measure, we gave considerable thought to the relationship between the satisfaction with a given item and its perceived importance. In principle, it would be possible to create an index where the satisfaction score of an item is weighted by its importance score. However, a body of research demonstrates that such procedures have little empirical benefit (Wu, 2008). Trauer and Mackinnon (2001) suggest that satisfaction ratings intrinsically incorporate the judgement of importance, thus making weighting untenable. Empirical tests (Wu et al., 2009) have also shown that unweighted satisfaction scores have a stronger predictive effect than importance-weighted satisfaction scores (Wu et al., 2009) and that weighted scores did not perform better than unweighted scores in measuring quality of life (Russell et al., 2006). However, Russell et al. (2006) found that the mean satisfaction ratings for important domains correlated more strongly with certain outcomes than did the mean satisfaction ratings for unimportant domains. Consequently, they recommend that importance should be incorporated more effectively into measures of quality of life. We have therefore used the importance attributed to these wellbeing items in our model by including the average importance score as a control variable in the regression models.

Independent variables – measuring the level of income

One of the main aims of the article is to assess the impact of the level of income on wellbeing. We consider income defined at two levels: gross household income and individual income. Including the total household income from all sources allows us to assess the effect of the financial situation of the family, which the household income approximates, on the wellbeing of individuals. However, even if we assume a full redistribution of income within the household, using the pooled family income alone implicitly ignores the possibility of the level of individual contributions affecting personal wellbeing. Taking into account both household and individual income allows us to better understand the intricate nature of the relationship between income and personal wellbeing.

Household income was collected in the Study of Social Wellbeing as gross (before-tax) dollar bands. For the purpose of analysis, each response was set equal to the midpoint of the band and then adjusted for differences in household need using the OECD-modified scale, which sets a value of 1.0 for the first adult in the household, 0.5 for each subsequent adult and 0.3 for each dependent child (under 18 years of age). Non-dependent children are treated as adults for equivalence scale purposes. The individual income was based on actual earnings from employment, transformed into an annual amount. Both the household and individual incomes were scaled down by $1000, to aid interpretation of statistical models.

Independent variables – key individual and social factors

Paid employment contributes positively to quality of life by maintaining the economic and social security of employees and their families, and also provides opportunities for intrinsic satisfaction through creative work, lifelong learning and social interaction. Conversely, unemployment is associated with reduced self-confidence, social isolation and low levels of wellbeing (Andersen, 2009; Dockery, 2005; Sen, 2000). In our analysis, employment status was defined as: employed in paid work; unemployed looking for work; and not active in the labour market (which includes housewives; volunteers; and those still studying).

Health is a critical feature shaping both the length and the quality of people’s lives and is a fundamental aspect of capabilities (Stiglitz et al., 2009). Good health has been shown to be strongly associated with individual wellbeing (Bloom and Canning, 2005; Hsiao and Heller, 2007) and enables other aspects of quality of life such as social connections (Wilkie and Young, 2009). We have used self-reported health, measured on a 5-point scale ranging from poor to excellent. Such measures have currency in national health surveys and, while there is some criticism of their ability to provide an objective measure of morbidity, they have been shown to correlate with a range of health conditions. Health was defined as: poor to fair health; good health; very good health; and excellent health.

Capabilities-based measures focus attention on endowments such as education, skill and critical reasoning that can be converted into positive outcomes for quality of life. Education has an instrumental effect on quality of life by influencing outcomes such as income and wealth. In addition, and independently of any human capital effect, higher educational qualifications are more likely to be associated with higher levels of participation in society, better health (Grossman, 2006) and greater subjective wellbeing (Helliwell, 2008). In this analysis we use three categories: schooled up to year 12; trade qualification or apprentice, certificates and diplomas; and tertiary education.

Control variables

We include a number of control variables identified in the literature as associated with wellbeing. These are: marital status; family structure; gender; age and regional differences. Marital status has been shown to be a strong predictor of wellbeing (Evans and Kelley, 2004; Hewitt and Turrell, 2011). We categorized respondents into: single, in de facto relationship, married (registered marriage), divorced, separated and widowed. Research has also shown that family structure, particularly children in the home, are strongly associated with wellbeing (Ross et al., 1990). We distinguished between households with children under 18 only, those with children 18 years and older and those without children. The relationship between marriage and wellbeing also varies by gender. Inglehart (2002) found that while significant gender-related differences in subjective wellbeing exist, they tend to be concealed by an interaction effect between age, gender and wellbeing. Age on its own has also been found to be predictive of wellbeing (Keyes and Shapiro, 2004). To account for these effects on wellbeing, we included gender and age in our models. Regional differences in living costs and access to services and jobs may affect the quality of life of households in different locations (Curran et al., 2008) and hence this variable was also included, distinguishing between remote, non-metropolitan and city locations. As with any survey data, it was not possible to capture all factors potentially associated with wellbeing. For instance, length and validity considerations precluded the inclusion of complex measures of mental health and wealth. The lack of the latter is particularly unfortunate given its close relationship with income, which plays a central role in the article.

Data analysis

Prior to the main analysis, data was screened for normality and outliers. While the Wellbeing Index was positively skewed, populations typically tend to report high wellbeing and it can therefore be considered to be naturally skewed. No transformation was applied to this index as recommended by Tabachnick and Fidell (2001) and past research on subjective wellbeing (Rashleigh, 2004). Additionally, it has been argued that skewness has little influence on samples with more than 300 participants (Keppel, 1991).

We used regression models to estimate the associations between the measure of wellbeing (our dependent variable) and the level of income (our main independent variable), while controlling for other relevant characteristics of individuals and households. Given that our dependent variable is approximately continuous (with scores ranging from 1 to 7), we used a linear regression model. The focus of the analysis is on exploring the relationship between income and wellbeing which is reflected in our modelling strategy. To better understand the complex links between income and wellbeing, the former was measured at both household and individual levels. Furthermore, if a linear effect of income was detected, we explored the relationship further by adding a quadratic term to the regression equation to test for non-linearity of the effect.

Previous research (Kahneman and Deaton, 2010), strongly suggests a non-linear association between income and wellbeing measured using a variety of indicators. This non-linearity needs to be taken into account, for example by using a logarithmic transformation or by incorporating quadratic terms into the model. While we run models with both specifications, to aid the interpretation we only report detailed results from the models with quadratic terms, which use income in dollars. Using a quadratic function, we can explicitly demonstrate the decreasing marginal utility of income measured in dollars – a property which is built into the logarithmic transformation, yet which may be easily overlooked when using log income in regression models. However, we do briefly report the results of the model with log income, and compare the two specifications.

Results and discussion

Table 2 presents descriptive statistics for the dependent variable, the independent variables, and the control variables. In order to determine the effect of income on wellbeing we estimate a four-step model, the results of which are depicted in Table 3. The baseline model only includes control variables to predict wellbeing, while subsequent models extend it by adding, in steps, household income, household income squared, and individual income. We first discuss the control variables across the models, and then move on to the relationship between wellbeing and income.

Table 2.

Wellbeing measure and income – means and standard deviations^a – 18–65 years old.

	M	SD
Dependent variable
(1) Wellbeing	5.06	0.03
Independent variable
(2) Equivalized household income (scaled by $1000)	54.23	0.93
(3) Individual income (scaled by $1000)	44.05	2.81
(4) Importance of wellbeing	5.75	0.03
(5) Employment status
Employed paid work	0.73
Unemployed looking for work	0.02
Not active in labour market	0.25
(6) Marital status
Single	0.08
De facto	0.10
Married	0.68
Divorced	0.09
Separated	0.03
Widow	0.02
(7) Children in the home
No children	0.46
Children under 18	0.41
Children 18 years up	0.13
(8) Education level
Schooled up to year 12	0.38
Trade, certificate, diploma	0.35
Degree	0.27
(9) Gender
Male	0.38
Female	0.62
(10) Region
Remote	0.08
Non-metropolitan	0.24
City	0.68
(11) Health
Poor to fair health	0.19
Good health	0.35
Very good health	0.35
Excellent health	0.11
(12) Age	47.91	0.27
N	1552

Standard deviations only reported for continuous measures.

Table 3.

Relationship between wellbeing and household income – 18–65 years old.

	Model 1No income		Model 2Model 1 and household income		Model 3Model 2 and household income²		Model 4Model 3 and individual income
	β	SE β	β	SE β	β	SE β	β	SE β
Household income (scaled by $1000)	–	–	0.004**	0.001	0.01**	0.001	0.01**	0.002
Household income² (scaled by $1000)	–	–	–	–	–0.00001**	0.00001	–0.00002**	0.00001
Individual income (scaled by $1000)	–	–	–	–	–	–	0.0002	0.0002
Importance of wellbeing	0.18**	0.02	0.18**	0.02	0.18**	0.02	0.19**	0.02
Employment status
Employed paid work	–	–	–	–	–	–	–	–
Unemployed looking for work	–0.87**	0.15	–0.72**	0.16	–0.69**	0.16	–0.67**	0.16
Not active in labour market	–0.09	0.05	–0.04	0.05	–0.01	0.05	0.01	0.06
Marital status
Single	–	–	–	–	–	–	–	–
De facto	0.13	0.10	0.03	0.10	0.01	0.10	0.01	0.10
Married	0.30**	0.08	0.22**	0.08	0.19*	0.08	0.20*	0.08
Divorced	0.05	0.10	0.01	0.10	0.003	0.10	–0.03	0.10
Separated	–0.25	0.14	–0.29*	0.14	–0.29*	0.14	–0.30*	0.14
Widow	–0.04	0.17	0.06	0.17	0.06	0.17	0.06	0.17
Children in the home
No children	–	–	–	–	–	–	–	–
Children under 18	–0.07	0.05	–0.001	0.06	0.02	0.06	0.01	0.06
Children 18 years up	0.06	0.07	0.03	0.07	0.02	0.07	0.02	0.07
Education level
Schooled up to year 12	–	–	–	–	–	–	–	–
Trade, certificate, diploma	0.04	0.05	0.04	0.05	0.03	0.05	0.01	0.05
Degree	0.11*	0.05	0.04	0.05	0.02	0.05	0.03	0.05
Gender
Male	–0.03	0.04	–0.05	0.04	–0.05	0.04	–0.06	0.04
Female	–	–	–	–	–	–	–	–
Region
Remote	–0.01	0.08	–0.04	0.08	–0.04	0.08	–0.03	0.08
Non metropolitan	–0.04	0.05	–0.07	0.05	–0.07	0.05	–0.06	0.05
City	–	–	–	–	–	–	–	–
Health
Poor to fair health	–1.27**	0.08	–1.22**	0.08	–1.22**	0.08	–1.18**	0.08
Good health	–0.73**	0.07	–0.66**	0.07	–0.66**	0.07	–0.63**	0.07
Very good health	–0.26**	0.07	–0.20**	0.07	–0.20**	0.07	–0.17*	0.07
Excellent health	–	–	–	–	–	–	–	–
Age	–0.06**	0.02	–0.07**	0.02	–0.07**	0.02	–0.07**	0.02
Age²	0.001**	0.0002	0.001**	0.0002	0.001**	0.0002	0.001**	0.0002
Constant	5.36**	0.39	5.39**	0.40	5.34**	0.40	5.21**	0.41
Adjusted R²	0.30		0.32		0.32		0.33
N	1756		1640		1640		1552

p < 0.05, ** p < 0.01, *** p < 0.001

Model 1 indicated that six control variables were significantly related to wellbeing. The importance attributed to aspects of wellbeing is positively related to the wellbeing scores, suggesting that people who attributed, on average, higher importance to the components of wellbeing, tended to have higher wellbeing levels. Being unemployed and looking for work has a negative relationship with wellbeing. It should be noted that the coefficient for unemployment is almost 10 times greater than the coefficient for not being active in the labour market. This finding points to the importance of separating these statuses when studying the effect of unemployment on wellbeing. Married individuals have significantly higher wellbeing scores than single persons. Having a tertiary degree has a positive relationship to wellbeing. Health is negatively related to wellbeing: the poorer the health, the lower the wellbeing. Age has a negative relationship to wellbeing, suggesting that wellbeing decreases with age. However, including a quadratic term for age suggests that the relationship is non-linear and that age 38 constitutes a turning point whereupon wellbeing starts to increase with age. It should be noted that once the household income was introduced in Model 2, educational level no longer has a significant relationship with wellbeing, but being separated was shown to have a significant negative relationship with wellbeing.

Contrary to some other research in the field, our models did not detect a statistically significant effect of having children on wellbeing. To cross-validate these results, we ran additional analyses, in which we substituted the marital status and the number of children with a composite ‘family type’ variable with the following categories: lone person; couple with no children; couple with children under 18; single parent with children under 18; couple with children 18 and older; and single parent with children 18 and older. There were significant differences in wellbeing between single persons and couples (whether with or without children) but not between couples with or without children, suggesting that partnership status is more important to wellbeing than if there are children in the home. A possible explanation could be that the household income used is adjusted for dependents, which may distort the link between wellbeing and having children in the home. Also, contrary to previous findings from literature, region showed no significant relationship to wellbeing in this study. In order to rule out sample bias, we created a simple post-stratification weight that matched the proportions in the sample to those observed in the population based on the Australian Bureau of Statistics (ABS) census data (ABS, 2007) with respect to regional distribution and gender composition. The final model was re-run using these weights and the findings remained stable.

We now turn to discussing the relationship between income and wellbeing. The average household income equivalent to income of a one-person household in this sample is $54,228 (Table 2), ranging from zero to $325,000. Bearing in mind that household income was adjusted for the number of persons living in the household, the household income for an average family of two adults and two children would be $113,879 and $81,342 for a couple with no children. The average individual income for the sample was $44,053.

The baseline model, or a model without any income variable (Model 1), produced an adjusted R² of .30. Adding household income (Model 2) produced an adjusted R² of .32, which explained more of the variance while the model remained stable. While household income is significantly related to wellbeing it contributes relatively little towards the variance explained.

To test for non-linearity of the effect, a quadratic term was further introduced (Model 3). This model produced an adjusted R² of .32. The coefficient for the quadratic term is statistically significant and negative. This is an important result suggesting that the effect of income on wellbeing is stronger for households with lower incomes and becomes weaker as we move along the distribution, an effect known in economics as a declining marginal utility of income. In other words, the same amount of dollars ‘buys’ more wellbeing for poorer people than it does for the rich. It is also worth noting that the model also implies that wellbeing increases with adjusted household income only until it reaches a certain level, from which point additional income does not further increase wellbeing; in fact, from this point on wellbeing actually begins to decrease with income. Based on the magnitude of the regression coefficients, this point can be estimated as $209,263 for a one-person household. It should be noted, however, that such a decrease is an intrinsic property of the quadratic function used here and as such should not be over-interpreted, particularly given that households with such a high level of income are virtually not present in the sample. To confirm the results we also run an alternative specification, where logarithmic transformation of household income was used. This model had the same fit (R² = .32) and produced substantively the same results as our original model, with the log income coefficient remaining statistically significant. This confirms a non-linear, and tapering (i.e. concave) relationship between income and wellbeing. The logarithmic specification suggests that wellbeing keeps increasing, albeit slowly, throughout the income distribution but the coefficient (0.24) suggests even weaker relationships with wellbeing than in the case of quadratic specification.

Our findings corroborate the results reported recently by Kahneman and Deaton (2010) who demonstrated a similar non-linear and tapering relationship between income and a number of wellbeing measures. However, while their ‘life evaluation’ measure was found to increase continuously with income in a non-linear manner, their measure of emotional wellbeing satiated with high income. Even though the measures of wellbeing used by Kahneman and Deaton are very different to ours and cannot be directly compared, taken together the results confirm that the association between income and wellbeing should be modelled in a non-linear fashion. The income effect is consistently weaker at the top end of the distribution regardless of the measure of wellbeing used, but whether it satiates or keeps increasing slowly depends on the wellbeing measure employed.

The final model (Model 4; R² = .33), introduces individual income. While household income remained significant, individual income did not show a significant relationship to wellbeing as one might have expected. This suggests that what matters more for the wellbeing of an individual is the overall level of income available to them through being a member of a given household, rather than the specific contribution they make towards it. In Model 4, as in preceding models, while household income is significantly related to wellbeing, its effect is weak. The results indicate that, keeping other variables constant, doubling the average income of $54,000 would only increase the wellbeing score by 0.27 while a $10,000 increase in household income for households earning $20,000 and $80,000 would only increase wellbeing by 0.08 and 0.05 respectively. Clearly, while income is important to wellbeing, the magnitude of its effect is rather small and is determined by where the household fits on the continuum of poor to wealthy. These results add weight to international findings that indicate that income has a relatively small effect on wellbeing in developed countries, such that an actual difference can only be seen between the richest and the poorest.

Conclusions

This research was based on survey data from a large sample of individuals and households who participated in the Study of Social Wellbeing in Australia. This new dataset was used to explore the relationship between various measures of individual and household income and our multi-dimensional index of social wellbeing. The findings contribute to an evidence base that has been mostly grounded in data collected in Europe or the United States.

The analysis reported in this article demonstrates the importance of viewing social wellbeing as a complex and multi-dimensional concept that incorporates those objective aspects of people’s life circumstances that they deem to be important to their quality of life. The analysis strongly supports our view that wellbeing depends on a range of social circumstances that have value for individuals and families, such as health, unemployment and marital status. All of our measures of household income are also statistically significant for wellbeing, but the magnitude of their effect is small and is determined by where the household fits on the continuum of poor to wealthy. Individual income did not show any significant relationship to wellbeing, beyond the relationship through household income.

These findings are consistent with other recent research indicating that the effect of income on wellbeing is small in developed countries. Our results therefore suggest that wellbeing cannot be easily approximated by income or consumption and should be treated as a complex and multi-faceted concept that may be used as one of a number of indicators to track social progress. Another important finding is that individual wellbeing is more affected by the income available to the household as opposed to individual earnings. This finding emphasizes the collective (household information) approach policy-makers should take when viewing the factors that impact on wellbeing.

In methodological terms, our article clearly demonstrates a non-linear association between income and wellbeing, highlighting the need for using appropriate specification of statistical models, such as logarithmic or quadratic transformations of income. The analysis reported in this article demonstrates the utility of the proposed measure of wellbeing and suggests possibilities for its incorporation in further research aimed at mapping the progress of societies. Specifically, the uniqueness and robustness of our Wellbeing Index comes from incorporating only those aspects of wellbeing that are deemed socially important.

Studies of social wellbeing have assumed an increased prominence in the lexicon of statisticians and policy-makers. In order to answer questions about the quality of life of families and communities there is a need to consider a broader range of measures of the dimensions of wellbeing that move beyond the purely economic domain and that take into account key aspects of an individual’s or family’s satisfaction with their life circumstances. The results of this analysis show that, while household and individual income may be important elements in the alleviation of poverty and deprivation, they are only a very small component of how people regard their quality of life or wellbeing. This is an important finding for policy-makers and it suggests that more attention needs to be paid to the social circumstances of individuals and families, and not just their financial means, if the quality of their lives is to be addressed.

Footnotes

Funding

The research received funding from the Australian Research Council Linkage Grant LP0775040 The Development and Application of a Conceptual and Statistical Framework for the Measurement of Non-market Factors Affecting Social Inequality and Social Wellbeing.

Author biographies

Jenny Povey is a research psychologist who has experience in test construction, survey development and evaluative research. She conducts research at the Institute for Social Science Research of the University of Queensland for academic, commercial, and public sector clients across broad, multidisciplinary activities.

Paul Boreham is an emeritus professor in the Institute for Social Science Research at the University of Queensland. He was formerly Director of the University of Queensland Social Research Centre and Head of the School of Political Science and International Studies. He has published and researched in the fields of employment and labour market analysis, policy studies, and social and economic change.

Wojtek Tomaszewski is a Research Fellow at the Institute for Social Science Research of the University of Queensland. His expertise is in quantitative research methods, statistical analysis of complex data, and in research on poverty, social exclusion, housing, homelessness and ageing.

References

ABS (Australian Bureau of Statistics) (2007) Queensland Key Statistics – 1388.3. Canberra: ABS. URL (consulted November 2010): http://www.abs.gov.au/ausstats/abs@.nsf/mf/1388.3

Andersen

S.H.

(2009) ‘Unemployment and Subjective Well-being: A Question of Class?’, Work and Occupations 36(1): 3–25.

Blanchflower

D.G.

Oswald

A.J.

(2004) ‘Well-being Over Time in Britain and the USA’, Journal of Public Economics 88: 1359–86.

Bloom

Canning

(2005) Health and Economic Growth: Reconciling the Micro and Macro Evidence. Working Paper No. 42. Stanford, CA: Stanford Institute on International Studies.

Boreham

Western

Dow

Laffan

(2009) Australian Research Council, Linkage Project LP0775040: The Development and Application of a Conceptual and Statistical Framework for the Measurement of Non-market Factors Affecting Social Inequality and Social Wellbeing (Social Wellbeing Survey 2009). Institute for Social Science Research, University of Queensland.

Brereton

Clinch

J.P.

Ferreira

(2008) ‘Happiness, Geography and the Environment’, Ecological Economics 65: 386–96.

Clark

A.E.

Frijters

Shields

M.A.

(2008) ‘Relative Income, Happiness, and Utility: An Explanation for the Easterlin Paradox and Other Puzzles’, Journal of Economic Literature 46(1): 95–144.

Cummins

R.A.

Eckersley

Pallant

Van Vugt

Misajon

(2003) ‘Developing a National Index of Subjective Wellbeing: The Australian Unity Wellbeing Index’, Social Indicators Research 64(2): 159–90.

Curran

Wolman

Hill

Furdell

(2008) Poverty, Programs and Prices: How Adjusting for Cost of Living Would Affect Federal Benefit Eligibility. Report, Metropolitan Policy Program. Washington, DC: Brookings Institution.

10.

Decancq

Lugo

M.A.

(2008) Setting Weight in Multidimensional Indices of Well-being and Deprivation. Oxford Poverty & Human Development Initiative (OPHI) Working Paper No. 18. Oxford: University of Oxford.

11.

Di Tella

MacCulloch

R.J.

(2008) ‘Gross National Happiness as an Answer to the Easterlin Paradox?’, Journal of Development Economics 86(1): 22–42.

12.

Diener

(2000) ‘Subjective Well-being: The Science of Happiness and a Proposal for a National Index’, American Psychologist 55(1): 34–43.

13.

Diener

Seligman

(2004) ‘Beyond Money: Toward an Economy of Well-being’, Psychological Science in the Public Interest 5: 1–31.

14.

Dockery

A.M.

(2005) ‘The Happiness of Young Australians: Empirical Evidence of the Role of Labour Market Experience’, The Economic Record 81(255): 322–35.

15.

Easterlin

(2003) Building a Better Theory of Well-being. Institute for the Study of Labor (IZA) Discussion paper series No. 742. Bonn: IZA.

16.

Evans

M.D.R.

Kelley

(2004) Effect of Family Structure on Life Satisfaction: Australian Evidence. Melbourne Institute Working Paper No. 24/04, Melbourne Institute of Applied Economic and Social Research, University of Melbourne.

17.

Frey

B.S.

Stutzer

(2002) Happiness and Economics. Princeton, NJ: Princeton University Press.

18.

Giovannini

Hall

Mira d’Ercole

(2007) Measuring Well-being and Societal Progress. Paris: OECD.

19.

Grossman

(2006) ‘Education and Nonmarket Outcomes’, pp. 577–634 in Hanushek

Welch

(eds) Handbook of the Economics of Education. Amsterdam: North-Holland.

20.

Han

Patterson

(2007) ‘An Analysis of the Influence that Leisure Experiences Have on a Person’s Mood State, Health and Wellbeing’, Annals of Leisure Research 10(3/4): 328–51.

21.

Headey

Muffels

Wooden

(2008) ‘Money Does Not Buy Happiness: Or Does It? A Reassessment Based on the Combined Effects of Wealth, Income and Consumption’, Social Indicators Research 87: 65–82.

22.

Helliwell

J.F.

(2008) Life Satisfaction and Quality of Development. Working Paper 14507. Cambridge: National Bureau of Economic Research.

23.

Hewitt

Turrell

(2011) ‘Short-term Functional Health and Well-being after Marital Separation: Does Initiator Status Make a Difference?’, American Journal of Epidemiology 173(11): 1308–18.

24.

Hills

Argyle

(2001) ‘Emotional Stability as a Major Dimension of Happiness’, Personality and Individual Differences 31: 1357–64.

25.

Hsiao

Heller

(2007) What Should Macroeconomists Know about Health Care Policy? IMF Working Paper, No. Wp/07/13. Washington DC: International Monetary Fund.

26.

Inglehart

(2002) ‘Gender, Aging, and Subjective Well-being’, International Journal of Comparative Sociology 43: 391–408.

27.

Kahneman

Krueger

(2006) ‘Development in the Measurement of Subjective Well-being’, Journal of Economic Perspectives 20(1): 3–24.

28.

Kahneman

Deaton

(2010) ‘High Income Improves Evaluation of Life but Not Emotional Well-being’, Proceedings of the National Academy of Sciences 107(38): 16489–93.

29.

Keppel

(1991) Design and Analysis: A Researcher’s Handbook, 3rd edn. Englewood Cliffs, NJ: Prentice Hall.

30.

Keyes

C.L.M.

Shapiro

(2004) ‘Social Well-being in the United States: A Descriptive Epidemiology’, pp. 350–72 in Brim

O.G.

Ryff

C.D.

Kessler

R.C.

(eds) How Healthy Are We? A National Study of Well-being at Midlife. Chicago: University of Chicago Press.

31.

Land

K.C.

(2010) Child and Youth Well-Being Index (CWI). Annual release. Durham, NC: Foundation for Child Development.

32.

Layard

(2005) Happiness: Lessons from a New Science. London: Allen Lane.

33.

Lent

R.W.

(2004) ‘Toward a Unifying Theoretical and Practical Perspective on Well-being and Psychosocial Adjustment’, Journal of Counseling Psychology 51(4): 482–509.

34.

McBride

(2001) ‘Relative-income Effects on Subjective Wellbeing in the Cross-section’, Journal of Economic Behaviour and Organization 45: 251–78.

35.

OECD (2007) Society at a Glance 2006: OECD Social Indicators. Paris: OECD.

36.

Pacione

(2003) ‘Urban Environmental Quality and Human Wellbeing – A Social Geographical Perspective’, Landscape and Urban Planning 65(1–2): 19–30.

37.

Pollard

Lee

(2003) ‘Child Well-being: A Systematic Review of the Literature’, Social Indicators Research 61(1): 59–78.

38.

Powdthavee

(2008) ‘Putting a Price Tag on Friends, Relatives and Neighbours: Using Surveys of Life Satisfaction to Value Social Relationships’, Journal of Socio-Economics 37(4): 1459–80.

39.

Rashleigh

(2004) ‘Relationships a Significant Factor in Depression?’, BA dissertation, Deakin University, Melbourne.

40.

Ross

Mirowsky

Goldsteen

(1990) ‘The Impact of the Family on Health: The Decade in Review’, Journal of Marriage and the Family 52(4): 1059–78.

41.

Russell

L.B.

Hubley

A.M.

Palepu

Zumbo

B.D.

(2006) ‘Does Weighting Capture What’s Important? Revisiting Subjective Importance Weighting with a Quality of Life Measure’, Social Indicators Research 75: 141–67.

42.

Saunders

(2011) Down and Out: Poverty and Exclusion in Australia. Bristol: Policy Press.

43.

Saunders

Naidoo

(2009) ‘Poverty, Deprivation and Consistent Poverty’, Economic Record 85(271): 417–32.

44.

Sen

(2000) Social Exclusion: Concept, Application, and Scrutiny. Social Development Papers No. 1. Manila: Asian Development Bank.

45.

Stiglitz

J.E.

Sen

Fitoussi

(2009) Report by the Commission on the Measurement of Economic Performance and Social Progress. Paris: Commission of the Government of France.

46.

Stutzer

(2004) ‘The Role of Income Aspirations in Individual Happiness’, Journal of Economic Behavior & Organization 54: 89–109.

47.

Tabachnick

B.G.

Fidell

L.S.

(2001) Using Multivariate Statistics, 4th edn. Sydney: Allyn and Bacon.

48.

Trauer

Mackinnon

(2001) ‘Why Are We Weighting? The Role of Importance Ratings in Quality of Life Measurement’, Quality of Life Research 10: 579–85.

49.

Veenhoven

(2005) ‘Apparent Quality of Life: How Long and Happy People Live’, Social Indicators Research 71: 61–86.

50.

Welsch

(2006) ‘Environment and Happiness: Valuation of Air Pollution Using Life Satisfaction Data’, Ecological Economics 58(4): 801–13.

51.

Wilkie

Young

(2009) ‘Why Health Matters for Economic Performance’, Economic Round-up, URL (consulted June 2011): http://www.apo.org.au/node/1239

52.

C.H.

(2008) ‘Can We Weight Satisfaction Score with Importance Ranks Across Life Domains?’, Social Indicators Research 86: 469–80.

53.

C.H.

Chen

L.H.

Tsai

Y.M.

(2009) ‘Investigating Importance Weighting of Satisfaction Scores from a Formative Model with Partial Least Squares Analysis’, Social Indicators Research 90(3): 351–63.