INTRODUCTION

In labour policy terms, working in a job is supposed to support everyday quality of living [1,2]. Yet since at least the start of this century, employment neither has guaranteed sufficient income to cover even basic costs of living nor has led to a good quality life and flourishing [3–6]. Whether a wage translates into household poverty or prosperity may depend further on the number of other household earners and other household dependents [7]. A study reported in the inaugural issue of this journal showed that other incomes, but not household dependents moderated the link between wage on the one hand and job attitudes, such as job satisfaction, on the other [8].

In response, the Living Wage in New Zealand (NZ) is the wage sufficient to ensure a decent quality of living, both material and social, in contemporary NZ society. It is determined by the Family Centre Social Policy Research Unit in Wellington NZ using official cost-of-living data and applies across all occupations and places in NZ. In a previous article in this journal, Carr and colleagues [8] reported that as the hourly wage rose from the legal Minimum (NZ$15.75/h, approx. US$10) toward the campaign Living Wage ($20.55/hr, approx. US$13), job attitudes tended to transform from negative (e.g., job dis-satisfaction) into positive (e.g., job satisfaction) especially for single-income households. The current study explores whether qualities of living outside of work (for example, in everyday life) also varied as a function of wage, other incomes, and/or household dependents.

By way of further background, in-work poverty has become the reality for many people in “developed” economies. Employment and hard work no longer equate to escaping poverty or offer a decent living [9]. Instead, they frequently result in perpetuated working poverty, increasing inequality, social distancing, health issues and premature death [10,11]. In Aotearoa New Zealand, in-work poverty rates have been estimated to be anywhere from 9% to 12% in 2018 [12]. The gap between rich and working poor and its detrimental consequences for individuals and their families, communities and societies [11] remain prominent and have been increasing yet are under-researched. The guarantee of a Minimum Wage was one step towards alleviating working poverty, but increasing costs of living, in particular in urban centres, also has been thwarting people's aspirations to a decent quality of life. Therefore, when COVID-19 arrived, the pre-existing wage momentum was insufficient to buffer against the consequence of the virus for most working New Zealanders [13].

In their research as part of Project GLOW (Global Living Organisational Wage), Carr and colleagues [8,14] observed a non-linear pattern in the links between wage on the one hand and quality of work life on the other. The finding of a cusp in the curve linking job attitudes to hourly pay rate indicated that, within the wage range from legal Minimum Wage to campaign Living Wage, employees tended to report a transformation in quality of work life (for example, from job dissatisfaction to satisfaction). Such transformations were broadly consistent with the concept of a poverty trap in which low wages perpetuate working poverty by consistently keeping people below the poverty line. This phenomenon means that many workers are not able to afford to pay bills on time and therefore pay extra for late payments. Many workers are unable to buy food in bulk and have to buy smaller portions, which are more costly-per-unit. Many workers may also have to borrow money from disreputable lenders, thereby fuelling a predatory debt industry and all the stresses that it can bring to everyday wellbeing [13].

Such consequences from low-waged work clearly go beyond job attitudes themselves by traversing into wider work-life balance [15]. According to Haar and colleagues [16], in theory low wages can spill over into everyday quality of living. Our first objective in this study was to explore this possibility empirically. Based on the work of Haar et al. [15,16], we expected to find that qualities of living spontaneously reported by participants would transform from negative to positive as the participants' hourly pay crossed a Living Wage threshold. The precise ways in which household composition interacts with such spill-overs remain under-researched and poorly understood [8]. In this study, we expanded on Carr's and colleagues’ [8] study with a second objective to consider if and how low wages may have interacted with added household income streams and number of household dependents not employed in paid work.

METHOD

A total of N = 1011 eligible participants were drawn from a national survey panel of lower income workers across New Zealand who were paid under NZ$60,000 per annum. By income level, the majority (39.4%) came from the NZ$40,001–60,000 band, closely followed by the NZ$20,001–40,000 band (35.9%), and then an income band up to NZ$20,000 (24.7%). Respondents were more likely to be female (69%), with ages ranging from early 20s to over 60 years. The average age category was the 30–35 age range. Weekly hours worked ranged from 10–50+, with the average in the 26–30 category. By ethnicity, 61% were NZ European, with the remainder Māori (11%), Asian (11%), Indian (7%), Pacific (6%), and 3% other.

At the organisational level, 68% of respondents were from the private sector, followed by 18.1% from the public sector, and 14% from the not-for-profit sector. Respondents worked in organisations of various sizes, with the largest group (28%) located in micro-sized organisations (1–10 employees) and the average organisation size being 50 employees. A number also worked in larger organisations, including those with more than 500 employees (9%). These proportions were reasonably representative of the lower end of the wage spectrum and economy in New Zealand at the time [16].

In addition to a range of standard demographic items, we focused on two particular sets of variables, reflecting (i) income level and (ii) job attitudes.

Income Level. In keeping with the goals of our study (above), income was assessed at both individual and household levels [17]. For individual income, we asked for annual income before tax (open-ended), annual income before tax (in increments of NZ$5000), and hourly rate of pay (if paid by the hour); for household income, we asked for total annual income before tax (“yours and that of your partner/family members”) using increments of NZ$5000.

A range of job attitudes were measured using quantitative measures (Job Satisfaction, Work Engagement, Career Satisfaction, Meaningful Work, Affective Commitment, Organisational Citizenship Behaviours [OCBs], and Work–Life Balance). With the possible exception of Work-Life balance, these measures all manifestly were not directly related to everyday life outside of work. A primary question that we added in order to address this issue, and the item on which our analysis in this paper focuses, was: “How well does your wage/salary work for you? Feel free to write any stories or comments (please do not mention anyone or organisation by name)”. A secondary, open-ended question that we added was, “Finally, please feel free to make any comments regarding your job(s), especially what you see are the main problems and how work could be made better”.

The project was funded by the Royal Society of NZ (RSNZ) through its Marsden Fund (17-MAU/137), and secured ethical approval from Massey University’s Northern Human Ethics Committee [18]. The approval included named counselling services for anyone who might have experienced any distress during the study. The survey was designed by the authors and distributed via a private research company, Qualtrics, which collected the data. Within Qualtrics panel protocols, all participants were assured of confidentiality and remained anonymous to the researchers. Qualtrics pays respondents for their time, but the nature of this arrangement is proprietary. The Qualtrics system had an estimated time for the survey (10 minutes in this example). It also ensured that each respondent only could complete the survey once. We utilized this approach specifically because Qualtrics can target income level in respondent recruitment and because the respondents are already familiar with survey formats that use multiple items and thereby increase measurement reliability and validity [8]. During piloting, lack of familiarity with multi-item measures/scales had been identified as a potential barrier to participation by lower-income groups.

RESULTS

As a basis for content analysis of responses to the central question above (“How well does your wage work for you?”), one of the authors (SC) read through all responses and suggested a coding scheme. These themes were subsequently revised by AY-H, WS, and LM, resulting in Table 1.

TABLE 1

Table 1. Content analysis themes.

Two of the authors (WS and LM) then independently coded all participant answers, allowing for more than one code per answer/respondent for each category. A confusion matrix was prepared and interrater reliability was found to be K(appa) = 0.85, classified as excellent [19]. In cases where there was disagreement that the two raters could not resolve, a third author (AY-H) resolved the final code(s).

TABLE 2

Table 2. Summary content analysis (modal categories in bold).

Next, we used the 3 × 2 classification cross-tabulation in Carr et al. [8] to create sub-analyses split by hourly wage: Up to and including $15.75, the legal Minimum Wage at the time; above Minimum Wage to the campaign Living Wage in New Zealand ($20.55 at the time); and greater than the campaign Living Wage ($20.55+). The content analysis was split further, again consistent with significant findings in Carr et al. [8], according to whether the amount of other household income, for example, over and above own hourly wage, was above or below the sample median (N = 1011). Whilst Carr et al. [8] found that number of dependents was not a significant predictor of job attitudes, it still might affect quality of life outside of work and so was included in this study, in this case through the theme G (in Table 1). The cross-tabulation with separate sub-content analyses resulted in the breakdown in Table 2. Themes G and H (dependents and flexible work) comprised just 1–4% of responses and therefore were not included (see outliers below).

In Table 2, for those with “less other household income”, as wage rose so too did the modal category move from being “sub-threshold” to “minimally sufficient” to “living a good life” (from Table 1). With “more other household income” at the lowest hourly rate, the modal category was being “buffered by others”, shifting beyond the Minimum Waged participants to being “sub-threshold” and to “a good life” once paid beyond the Living Wage. Thus, the qualitative themes in the text data broadly converged with the quantitative findings in Carr et al. [8] with the exception of Minimum Waged respondents with “more other household income”, who reported being “sub threshold”.

In order to attempt to clarify this exception, we selected illustrative quotations for each of the six cells in Table 2. These quotations have been presented in Table 3. The quotes in Table 3 suggest that there may have been a role for household dependents in the marginally modal sub-threshold mode in the Minimum-Living Wage/More Other Household income cell (in Table 2). Furthermore, what looked initially like an 'outlier' turned out on closer inspection to suggest a role for context over-and-above hourly pay per se. Research to date [8,14] has tended to find a diversity of spread above and below the curve-of-best-fit between pay on one hand and job attitudes on the other. By examining outliers in more detail, we might learn more about which life circumstances actually account for such spread.

Based on the potential informativeness of outlier cases, we decided to expand our analyses to include two types of outliers: relatively low wage but high job attitude scores and relatively high wage but relatively low job attitude scores. A series of steps then were undertaken to identify the potentially informative case outliers. First, we followed Fisher [20], who argued that attitude scores often reflect different facets of happiness at work. Carr et al. [8,14] too found that the primary cusp in pay-attitude curves tends to occur at the same point regardless of specific job attitude, suggesting that those curves also reflect underlying happiness at work.

TABLE 3

Table 3. Exemplar quotations.

With this possibility of a parsimonious single underlying factor in mind, the six original raw factor scores that used the same (6-point) scale were added together then divided by six to derive a mean score per attitude: job satisfaction, affective commitment, meaningful work, work-life balance, career satisfaction, work engagement and organisational citizenship behaviours (toward the organisation). A Harman test in the current study corroborated that the factors used in Carr et al. [8] all did load significantly on one factor. LOESS regression revealed a curve shaped like those found in previous research. Hence, we used this curve to identify thresholds for selecting cases (Figure 1).

FIGURE 1

Figure 1. Mean score per attitude to identify informative case outliers. Key: Thresholds for selecting outliers.

From the data shown in Figure 1, we identified two such case groups. First, we identified what might be termed unexpected positives, meaning the person scored relatively low on wage but still reported relatively highly on overall job attitude (happiness). Second, we identified unexpected negatives, meaning the person scored relatively high on wage but still scored relatively low on overall job attitude (happiness). The cut-off scores then used for unexpectedly positive were hourly rate of pay less than $17 per hour (on the range of NZ$15-40 per hour [8]) but mean overall attitude score greater than 5 (i.e., 5/6, which was clearly in the range of “happy”). The cut-off scores used for unexpectedly negative were $22 per hour and an attitude score less than 3/6 (i.e., below the scalar mid-point, i.e., in the unhappy range).

These cut-offs produced a small group of ten cases, with five in each of the unexpected positive and unexpected negative outlier case categories (Figure 1). We then could proceed to look more closely not only at the income and job attitude dimensions of the individual cases within these two groups but also at their other anonymised data, including narratives and numbers from their respective SPSS data files.

TABLE 4

Table 4. Demographic silhouettes.

As the data in Table 4 show, the participants in the unexpectedly positive group were younger and all female, with more children/dependents in the household (total of 20 compared to 3 in the other group). Four out of five lived outside of the main urban areas (main urban areas = Auckland, Wellington, Christchurch). In comparison, all of the participants from the unexpectedly negative group lived in the three biggest agglomerations (Auckland, Wellington and Christchurch). In these three cities, housing costs were highest (Table 4). Such points of comparison, even in such a small sub-sample, suggest that higher urban housing and living costs may have dampened any fillips from higher hourly pay whilst lower living costs may have worked in the opposite direction by buffering any deleterious effect on happiness levels from lower levels of hourly pay.

Next, we examined narrative comments from these same 10 participants in response to our two open-ended questions from the survey form (above, Measures). Responses from the unexpectedly positive outliers are provided in Table 5, first for the primary question on wage and then for our secondary question on work/job.

From Table 5, all three education sector employees said that their wage was marginal, although the work/job itself was not criticised. For the two production and service workers, happiness with the work and being trusted by a good employer were salient. As a whole, therefore, a low wage may to some extent have been offset by intrinsic aspects of the work/job itself.

TABLE 5

Table 5. Unexpectedly positive responses to “How well does your wage/salary work for you”.

Table 6 contains responses from employees who were paid relatively well but reported being relatively unhappy. Even on the question about wage, the responses revealed a range of non-pecuniary concerns, from having to work a second job, supporting wider family, financial insecurity approaching old age, workplace injustice, and the distinction between needs versus wants. These non-pecuniary, precarity-related responses are underscored by the answers to our question about work/job. These stories included under-employment, frustration with employment relations, lack of human relations, depression and work overload.

TABLE 6

Table 6. Unexpectedly negative responses to “How well does your wage/salary work for you?” and “How work could be made better”.

DISCUSSION

The purpose of this study was to explore the nuances and context for Living Wages and quality of work life. We expected to find that (1) qualities of everyday living spontaneously reported by participants would transform from negative to positive as their hourly pay crossed a Living Wage threshold. According to Haar and colleagues [16], low wages and work-related happiness in theory can spill over into everyday quality of living. However, the precise ways in which household compositions interact with such spill-overs remain under-researched and poorly understood [8]. In this study, we therefore also set out to explore (2) how low wages might interact with both added household income streams and added number of household dependents who were not employed in paid work.

As wages crossed the Minimum Wage and then the Living Wage and “other” supporting incomes rose from less to more, narrative accounts about everyday quality of living tended to transform from negative to positive. These qualitative indications that wage could make a difference to quality of life outside of work were important because previous research [8] was not able to test systematically for this possibility. At the same time, we did not have enough representation of larger numbers of dependents to probe for interactions between these qualitative indications and wage and their combined links to quality of everyday life. However, we did note that there was significant variation about the curve-of-best fit to happiness as a function of wage. Such variation is very common in psychology, where most events—including attitudes—are shaped by a plethora of contextual dimensions and their contingencies. In this study however, we were able to identify at least some of these contingencies by focusing on pre-supposed 'outliers’, both people whose hourly wage was low but who reported being relatively happy and people whose hourly wage was higher but whose attitude score (happiness-at-work level) was lower. These people's shared stories told us that Living Wage thresholds constitute more fuzzy than clear lines. Their narrative accounts as outlying cases pointed clearly at previously undetected roles for supporting dependents. Our quantitative survey data might not have picked up these roles due to under-representation of larger household units [8]. The answer to our question about the roles of dependents is that they can matter and do matter in conjunction with wage and quality of everyday life.

The outliers' narrative accounts also highlighted roles for intrinsic motivation, which commonly is found in vocational occupations such as teaching where intrinsic rewards can partly offset any deleterious effects of low pay [21]. Such forms of under-payment and all it conveys about people's value often are experienced as workplace injustices and inflexibilities. These experiences then can “spill over” negatively into job attitudes [22]. Such attitudes include job and career satisfaction, which our study showed to be reflections of work-related happiness [23].

Injustice is closely connected to workplace ethical issues, including not speaking inconvenient truths to economic power [24]. Even in perhaps the best review of happiness at work, which lists no fewer than 38 causes of happiness in organisations (Fisher, 2009, p. 394 [20]), the notion of wages, including the Living Wage, is almost completely absent. Money from Living Wages thereby runs a risk of being almost completely occluded by “psychological” constructs such as sense of autonomy, skill variety, and contact with others [20]. If money and wages do matter for enabling happiness at work, as Carr et al. [8,14] and others have shown, then these elements surely should have merited more than just one rather diminutive mention [38], namely of “availability of money” ([20], p. 395). Our study thus adds to the literature by highlighting how money does matter for transforming quality of work life alongside other less pecuniary factors. In our study these factors included work justice, mental health, workload stressors, and fatigue-related strain.

Our study has some key limitations, including a very small sample, a very limited amount of narrative data, and an arguably quite constraining methodology—an online survey form with mostly closed-option questions and answers. Such legitimate limitations notwithstanding, the observed convergence with the nomothetic LOESS curves in Carr et al. [8,14], when combined with the informative diversity of outlying cases from that curve, contains an important overriding lesson. Living wage theory and practice need to take more account of local, group, household and individual contexts. Whilst on the average a Living Wage might indeed boost quality of life, work life, and work-related happiness, it does so only in conjunction with a range of other human, organisational, biological (like sleep deprivation from shift-work) factors that surfaced in this study.

Nevertheless, perhaps the most important take-home message from this addition to previous research [8] remains that Living Wages matter for human happiness at work. The idea that only “psychological” factors matter is patently absurd. Workplace self-esteem neither pays the electricity bill nor feeds the household. Not being able to support members of the household or oneself appears to spill over negatively into workplace attitudes [15,16]. Our data indicate that without a decent wage, any intrinsic qualities of the job or work itself are ultimately mere placeholders. Sooner or later, something will crack. Even in a vocational job that was likely intrinsically rewarding (such as teaching), our qualitative data suggest that wage-related injustices can simmer and eventually can boil over. Thus, whilst a Living Wage is not likely to be sufficient in and of itself to make work decent, it might be a necessary condition to make it sustainable.

The core points from this paper are as follows. First, a Living Wage is also a quality of Living Wage, necessary but not sufficient to help more people to lift themselves out of in-work poverty and hardship in everyday life. Second, context can buffer the deleterious effects of a non-Living Wage but is not a long-term solution to inadequate income. Living Wages were especially pivotal for single-income households. Relying on other household incomes can be debilitating in its own right, just as it can be supportive when needed. Third, reliance works the other way too, with a Living Wage being all the more important as the number of dependents rises and housing costs rise. Fourth, nomothetic analysis is not in opposition to idiographic approaches that include outliers. Indeed, outliers can provide more information, not only about themselves but also about the limits of generalization. Sufficient conditions for quality of living can be considered more widely and more diversely without undermining the point that a Living Wage will tend to enable more sustainable livelihoods. Similarly, context is continually evolving, and the current pandemic is arguably a wake-up call not only for closing the gaps between Minimum and Living Wages but also for introducing Guaranteed Annual Income policies, especially for the most vulnerable in society [25].

DATA AVAILABILITY

The dataset of the study is available from the authors upon reasonable request.

AUTHOR CONTRIBUTIONS

This manuscript is the result of a joint research project of which all authors with the exceptions of WS and LM are team members. While various authors took the lead at different stages and of particular elements of the project, the workload and documents were always shared for comments and feedback. WS and LM were interns. Their work was essential in independently coding, which resulted in the matrices.

CONFLICTS OF INTEREST

The authors declare that there is no conflict of interest.

FUNDING

This research was funded by the Marsden Fund, grant number 17-MAU-137.

ACKNOWLEDGEMENT

We acknowledge the constructive and helpful input of our peer reviewers and editor(s), whose suggestions have made this paper clearer and sharper. We especially appreciated the detailed feedback kindly given in the form of annotations and helpful track changes.