Imagine people receive an amount of money sufficient to cover basic living expenses regularly every month. Everyone receives the same amount without preconditions. They do not have to give anything in exchange and can spend the money as they like. This is the basic idea of Universal Basic Income (UBI). Proponents of UBI argue that such nonconditional benefits should have a markedly positive impact on recipients’ mental health. Meanwhile, detractors claim that one negative effect could well be the motivation to work. Now, imagine the recipients are also employed or self-employed. What happens in this case to their well-being and work involvement directly touches upon issues of work motivation, a field with high relevance to psychologists in relation to UBI (Hüffmeier & Zacher, 2021). Thus for a greater understanding of UBI and its full impact on individuals as well as a society as a whole, this paper will look at how it affects quality of life in conjunction with the motivation to work.

UBI studies have so far shown positive influences of additional income on well-being. The most well-known might be the experiment initiated by the Finnish government (Kangas et al., 2021). For a period of two years, researchers examined the health, livelihoods and experiences of about 2,000 unemployed people and their integration into the labor market. In a randomized controlled trial, people who received UBI reported significantly higher levels of general life satisfaction as well as less distress and depression than people in the control group. In a study conducted in Kenya, unconditional cash transfers given to the poorest of the population increased subjective well-being, among other outcomes (Haushofer & Shapiro, 2016). While this result seems similar to the Finnish study, the background of extreme poverty and the difference in socio-political context limits its applicability to western societies. More recent research projects on UBI, for example in the Netherlands (Roosma, 2022), Spain (García, 2022; Sekulova et al., 2023) or Canada (Ferdosi & McDowell, 2020), have also been restricted to samples from socially disadvantaged environments with participants of low socioeconomic status. Thus, a common limitation to all of these studies is that their results cannot be considered representative of working people, because the UBI trials only address questions of welfare assistance.

There are however a few studies that have examined labor supply and contradict the myth that UBI disincentives work. Evidence from pilot studies on cash transfers in developed countries have shown advantages for labor participation in Canada, Iran, and the U.S. state of Alaska. With the MINCOME experiments in Canada in the 1970s, attempts were made to evaluate social and economic consequences of UBI in a broader community. More precisely, researchers investigated in effects of a negative income tax as an alternative form of guaranteed annual income (Hum et al., 1979). With regard to work motivation in that program, Calnitsky and Latner (2017) reported an 11.3 percentage point decrease in labor-market participation – with a greater reduction for subgroups of single-households, indicating that especially the elderly, people with poor health, young people seeking education, or single-parents tented to refuse work when offered another source of income. Although this was a large pilot, it suffers generalizability issues due to outdated results from the 70 s that cannot be fully transferred to the current, modern working world (Calnitsky & Latner, 2017; Simpson et al., 2017; Widerquist, 2005). Ferdosi and McDowell (2020) evaluated basic income payments in Ontario, Canada, and reported benefits of additional income especially for low-earning employed people when compared to a continuously unemployed subgroup of the study. Mental health increased in 86% of employed people (vs. 67% of unemployed) and depression was observed less often (88% vs. 74% of unemployed). Nevertheless, the study is also limited in that it could not verify representativeness in the region, reported a relatively small sample size and merely covered descriptives. Based on data from 2010 and 2011, an Iranian national cash transfer was shown to have no effect on overall labor supply (Salehi-Isfahani & Mostafavi-Dehzooei, 2018). Differentiated for subgroups, the authors indeed found an increase in labor supply among service sectors workers and a decrease of young workers. Since 1982, citizens of Alaska have received a cash dividend from the Alaska Permanent Fund. This guaranteed income consists of up to US$ 2,000 per person per year. Jones and Marinescu (2018) analyzed the Current Population Survey of Alaskans and reported a non-significant effect of UBI on aggregate employment and an increase in part-time work by 1.7 percentage points. These findings all have in common that they are limited in their (quantitative) focus on labor participation, rather than on motivational (qualitative) aspects. The mere presence or absence of working hours cannot explain underlying mechanisms and psychological processes that lead to decisions of employed UBI recipients.

The current state of the evidence has not yet clarified UBI’s effects on the mental health of employed individuals in dependence of their work motivation. UBI studies tend to apply a broad economic lens, as opposed to a more person-centered psychological approach. Therefore, political persuasiveness for or against UBI from a psychological perspective is missing. Scarcely anything is known about the trajectories of life satisfaction, well-being, and the role of work motivation and underlying psychological needs of employed people who receive UBI. A key contribution of the present article is thus expanding this sparse area of research by focusing on mental health of working individuals.

In doing so, the present research was grounded in Self-Determination Theory (Deci & Ryan, 2000), which holds that feelings of autonomy (the feeling of choice in behavior), competence (effectiveness of one’s behavior), and relatedness (relationships with others) are prerequisites for high qualities of motivation and, in turn, enhance mental health. Thus, we took a closer look at health psychology and motivational processes in the UBI context by addressing the following questions: Do life satisfaction and well-being in employees change during UBI payment? What part of the observed changes can be explained by the prevailing quality of work motivation, specifically autonomous and controlled motivation? What is the role of basic psychological needs satisfaction at work? Do socioeconomic differences between individuals influence these associations? Answers to these questions can deepen our understanding of UBI as a special form of income security that may promote well-being and life satisfaction with direct implications for policy and practitioners. If special types of motivation at the individual level translate to high levels of mental health in combination with UBI, research in this field might contribute to best practices for designing UBI payment strategies as well as work environments.

Universal Basic Income and Work

Today, the prevailing view is that work (and motivation to work) is more than just a trade of labor for money. For many, it is a source of meaning and purpose in life (Cassar & Meier, 2018). Therefore, most would agree that satisfaction in life derives from working on something valuable and doing it well. From an empirical perspective, positive effects of meaningful work on health and well-being have been well established for some time (Csikszentmihalyi et al., 2017; Kamerāde et al., 2019). Moreover, additional support has been provided from research on unemployment (Hollederer, 2015; Jahoda et al., 2009; Kroll & Lampert, 2011; McKee-Ryan et al., 2005), which has indicated negative effects in health of long-term unemployed people that reach far beyond financial concerns and reveal further functions of work. The exchange of labor for financial security is, however, the basis for thinking about work in modern societies. Which is why UBI is such a novelty and raises such interesting questions: It decouples income from work to a certain degree. A common concern with UBI is therefore whether people would refuse work once their basic needs of financial security are met outside the labor market (Calnitsky & Latner, 2017).

Research has also supported the idea that work itself is essential to people even though financial security might exist without it. Wielers and van der Meer (2020) asked what would happen if people were not bound to the paid employment system. They argued that if needs are met and the work itself is attractive, for instance due to higher levels of autonomy in tasks, choice of working hours, and opportunities for on-the-job development, workers would continue participating in the labor market. Conversely, people who were stressed and perceived work as a burden would stop working given sufficient financial means. Similar to this is research on the “lottery question,” which showed that people who imagined winning or who did indeed win a considerable amount in the lottery made various decisions regarding their working habits (Cesarini et al., 2017; Highhouse et al., 2010; Imbens et al., 2001; Picchio et al., 2018). For instance, in a Dutch sample of lottery winners, Picchio et al. (2018) found that labor supply in terms of working hours were reduced significantly after winning, but there was no effect on the probability of being employed. In other words, sufficient financial means have not led people to give up their jobs entirely. But with regard to the time they spent at work, their priorities changed.

It is often assumed that UBI is particularly beneficial for employees with low job security, for instance, due to automation or economic decline (Reed & Lansley, 2016; Standing, 2009). By transferring a portion of national income from capital owners to others, UBI is thought to reduce questionable work incentives while still ensuring a minimum level of income for those who are truly unable to work (Hoynes & Rothstein, 2019). Proponents of UBI further suggest that unconditional cash transfers can empower people in general (e.g., van Parijs & Vanderborght, 2017). They argue UBI recipients would be able to spend more time on creative or business projects, community work, or care of family members. They may also invest additional time in their own education and improve their health by reducing stress (Gibson et al., 2020).

Nevertheless, such reactions to UBI are not readily predictable. From a psychological point of view, recipient behavior and its underlying regulatory processes can be investigated by experimentation and pilot studies. How exactly work motivation might develop or decline under UBI can be hypothesized from motivation theories. Various approaches in psychology have shed light on work motivation (for an overview, see van den Broeck et al., 2019). A particularly influential one, which explicitly focuses on predicting growth and well-being, is Self-Determination Theory (SDT, Deci & Ryan, 2000; Ryan & Deci, 2017).

Self-Determination Theory

The basic assumption of SDT is that every person tends to strive to participate in interesting and growth-oriented activities and wants to contribute to a larger social framework. According to SDT, different qualities of motivation can be distinguished depending on their level of self-determination (Deci et al., 2017). Research has often found positive effects of self-determined work motivation on mental health (Wang & Panaccio, 2022), work-family conflict, burnout, turnover intention, and job performance (Kuvaas et al., 2017; Richer et al., 2002). Further, SDT views the fulfilment of three basic psychological needs as a necessary precondition of self-determined behavior, well-being, and life satisfaction with direct impacts on autonomous work motivation (Deci & Ryan, 2000). The three psychological needs in terms of SDT are competence, autonomy, and relatedness. The need for competence is experienced when a person effectively influences the environment and achieves desirable outcomes through actions. The need for autonomy is met when a person experiences themselves as the source of their own decisions and actions. Finally, the need for relatedness refers to social connectedness and belonging to groups. The satisfaction of psychological needs is considered to be dynamic and changeable, depending on the situation and context (Ryan & Deci, 2017). Although in theory, relatedness is considered to play a more distal role for optimal functioning than autonomy and competence (Ryan & Deci, 2017), empirical findings have strengthened the importance of all three needs for maintaining autonomous work motivation (De Cooman et al., 2013; Richer et al., 2002) and well-being (Chen et al., 2015; van den Broeck et al., 2016).

In summary, we concentrated on longitudinal changes in life satisfaction and well-being (health-related outcomes, HRO) in German employees for this study. We further examined whether self-determined motivation at work and its underlying needs for autonomy, competence, and relatedness cause differences in HRO trajectories. For this purpose, we applied a framework focusing on basic psychological needs that incorporates both the quality (dimensions) and quantity (magnitude) of work motivation (Deci & Ryan, 2000; Ryan & Deci, 2017). Inspired by advocates of UBI, we will furthermore examine whether people with certain economic or social constraints benefit more than others by including a number of socioeconomic and demographic indicators.

Hypotheses

Different life events have been shown to yield changes in life satisfaction and well-being (for a meta-analysis on different life events, see Luhmann et al., 2012). Additionally, income has been identified to predict life satisfaction and well-being to a certain degree (Clark et al., 2008; Frijters et al., 2004; Mund et al., 2021; Schyns, 2001; Sirgy, 2021). In line with this, research on happiness and wealth has shown that an adequate amount of money is necessary but not in and of itself sufficient for people’s well-being (Diener & Biswas-Diener, 2011; Kesebir & Diener, 2008). This indicates that, after reaching modest affluence, increases in income no longer contribute significantly to addressing personal needs. To a certain point, gaining additional income secures the availability of commercial goods and social status, a broader choice of leisure activities, feelings of personal control, unique chances to make lasting contributions to society, all of which are seen as sources of well-being (Clark et al., 2008; Diener & Biswas-Diener, 2011; Pinquart & Sörensen, 2000).

Initial evidence for effects of additional income on psychological health have been reported in Canada’s basic income studies (Ferdosi & McDowell, 2020). Findings from a Spanish Income Experiment have shown positive effects on subjective well-being (Sekulova et al., 2023). In relevant literature, life satisfaction and well-being have often been used interchangeably. For the present purposes, we have defined life satisfaction as a cognitive evaluation of one’s life situation, whereas well-being encompasses the positive feelings associated with the evaluation of one’s life (Pinquart & Sörensen, 2000). Therefore, well-being includes an emotional component that is not covered by life satisfaction.

We consider life satisfaction and well-being to be adaptable psychological constructs that occur as a function of satisfaction in many life domains (Rojas, 2011). Based on the large body of research concerning the relationship between income and psychological health as well as initial findings from UBI, we formulated the following two hypotheses concerning the development of mental health:

  • H1 Life satisfaction increases with UBI payment.

  • H2 Well-being increases with UBI payment.

In addition to investigating whether UBI is associated with mean-level changes in adult life satisfaction and well-being, another relevant question is under what circumstances these effects are most pronounced. We evaluated work motivation on the basis of SDT and empirical findings. Instead of measuring work motivation on a bipolar scale that varies from low to high expression, SDT according to Deci and Ryan (2000) differentiates types of goal-directed behavior along a spectrum from controlled (not self-determined) to autonomous (self-determined), whereby the different qualities of motivation are not mutually exclusive (Gagné & Deci, 2005). People who identify with and believe in the value of their work exhibit a higher quality (i.e., self-determined) of motivation and their performance of activities is less dependent on outside factors.

A large body of evidence clearly shows that autonomous forms of work motivation are associated with positive work-related outcomes, including greater work satisfaction (Richer et al., 2002), work commitment (Fernet et al., 2012), self-reported work effort (De Cooman et al., 2013), and work performance (Gagné & Deci, 2005). On a more personal level, autonomous motivation positively affects well-being and life satisfaction (Deci & Ryan, 2008; Wang & Panaccio, 2022). Given this, we postulated that:

  • H3 Health-related outcomes (life satisfaction and well-being, HRO) as well as their development during UBI payment depend on self-determined qualities of work motivation in the way that autonomous motivation influences changes in HRO positively.

SDT further assumes that three basic psychological needs for autonomy, competence, and relatedness at work are necessary for development and maintenance of self-determined work motivation. Indeed, high levels of need satisfaction at work are associated with more autonomous, self-determined motivation (De Cooman et al., 2013; Richer et al., 2002). In contrast, needs thwarting will undermine motivation and produce suboptimal consequences, for example, in well-being or psychological adjustment (Baard et al., 2004; Wang & Panaccio, 2022). Wang and Panaccio (2022) investigated the direction of effects between needs satisfaction at work, autonomous motivation, and health. While the association between changes in employees’ need satisfaction corresponding changes in health was confirmed, the study failed to find a significant mediation of this effect by autonomous motivation. The present study applies this general thesis to the context of UBI by testing the directions of effects in longitudinal data of participants’ mental health, basic psychological needs satisfaction, and work motivation. In line with predictions from SDT, we have examined the importance of work motivation as mediator between need satisfaction at work and the health-related outcomes in working individuals and postulated:

  • H4 The connection between work-related need satisfaction and HRO is mediated by self-determined qualities of work motivation.

Additionally, we examined the role of age, gender, education, income, working hours per week, and number of dependent minors in predicting changes in life satisfaction and well-being to test for the robustness of effects. Given that some data was collected during the COVID-19 pandemic, we noted the influences of the pandemic on the working conditions through inclusion of pandemic indicators.

Methods

Data and Sample

The data was collected between October 2019 and October 2022. Participants of the study had voluntarily registered with the online crowdfunding platform of the German association “Mein Grundeinkommen e. V.” (in English: My Universal Basic Income).Footnote 1 The association has been running a lottery every month, wherein approximately 20 people are randomly selected to receive UBI. The number of people chosen depends on the previous funding of the association’s community. Because of this procedure, the sample consisted of different cohorts. Each of the winners received €1,000 per month for one year. There were no exclusion criteria for online registration, although individuals receiving social welfare (until the beginning of 2023 named “Hartz IV” in Germany) were advised to register with a cooperating association as additional income would lower their social benefits. Thus, those who took part were more likely to have had jobs.

The study was designed as an online panel study. Participants were invited to the study by mail from the association at three different measurement occasions. These were before the payment (T1 with 74% response rate), in month 6 (T2 with 72% response rate), and finally in month 12 when the payment ended (T3 with 54% response rate). To ensure commitment to the study, we incentivized participation by offering a personality test and accompanying personal profile, made available after completion of the last survey. Children could be registered by their parents but were excluded from the survey. The respondents received informed consent at the beginning of the survey in which they agreed to anonymized data analysis and publication of the results.

We did not exclude people with missing longitudinal data, because excluding this group would lead to sample selectivity, distort the representativeness, and limit the generalizability of the results. Missing data in latent models were accommodated using Full Information Maximum Likelihood (FIML; Enders, 2001), which takes all available information into account. Participants were asked to report their working status (employed, non-working, in training, or education). For the purpose of the present study, we selected a sample of respondents who were employed or self-employed at the first measurement occasion (n = 357). Given that FIML estimation cannot completely correct and adjust for selective dropout, we compared people who were continuously working and participated at all waves of data collection (n = 114) with the rest of the subset (n = 243) of working people at T1. Analyses of variance and Chi-square tests of independence were used to determine if the presence of missing data was related to the variables included in the study. The means and distributions of all variables in the study did not differ significantly between the groups, with one exception. Life satisfaction at T1 was slightly lower for people who were working and remained in the sample (M114 = 6.06, SD = 1.59) compared to the rest of working participants at T1 (M243 = 6.43, SD = 1.59, t(221) = 2.05, p < 0.05, d = 0.23). Sample characteristics are summarized in Table 1. Average age at the beginning of the study was approximately 40 years; more women (54%) than men participated; no participants reported a non-binary gender status.

Table 1 Descriptives
Full size table

Measures

Measures included in this study were administered in three waves. For an overview of the items, see Appendix. We used validated scales for assessment of the dependent and independent parameters. Due to language reasons, the research group had to translate the scale if necessary from English to German in collaboration with a bilingual expert.

Life Satisfaction

A single item has been shown to capture life satisfaction well. We assessed life satisfaction using L1 (Beierlein et al., 2014). The answer format is an 11‑point rating scale, ranging from 0 – not at all satisfied to 10 – absolutely satisfied. Mean scores varied between M1 = 6.31 (SD1 = 1.77), M2 = 6.99 (SD2 = 1.79), and M3 = 6.95 (SD3 = 1.84). Rank orders differed (r12 = 0.26, r13 = 29, to r23 = 0.40) indicating that change occurred.

Well-being

Well-being was assessed with the WHO‑5 Well-Being Index (Brähler et al., 2007; for an English version see Topp et al., 2015) on a 6‑point rating scale from 1 – not at any time to 6 – all the time and included descriptions of one’s mental situation in the last two weeks (Cronbach’s α = 0.83 at T1 and T2, 0.86 at T3).

Work Motivation

We assessed different facets of work motivation with the Motivation at Work Scale (Gagné et al., 2010), using 12 items on a 7-point rating scale, (1 – does not apply at all to 7 – applies fully). The subscales capture the four different qualities of motivation, namely intrinsic motivation, identified regulation, introjected regulation, and external regulation. We decided to follow Gagné and Deci (2005) in distinguishing between autonomous and controlled motivation in the working context. Therefore, we merged the subscales intrinsic motivation and identified regulation into a composite “autonomous motivation” score, as well as introjected regulation and external regulation into a second “controlled motivation” score. Each score consisted of six items. The internal consistency of the subscale scores ranged from 0.84 (T1) to 0.87 (T2) and 0.89 (T3) for autonomous motivation, and from 0.65, 0.64 to 0.65 for controlled motivation, respectively.

Basic Psychological Need Satisfaction at Work

The three subscales measuring fulfilment of psychological needs (competence, autonomy, and relatedness) consisted of nine items in total, three for each subscale (Tafvelin & Stenling, 2018). Again, a 7‑point rating scale, ranging from 1 – does not apply at all to 7 – applies fully was used. Cronbach’s α was 0.76 to 0.82 for competence, 0.86 to 0.87 for autonomy, and 0.93 to 0.95 for the need relatedness across the three waves.

Covariates

We included age, gender, working hours per week, educational status, income, and minor household members as sociodemographic covariates. Age and working hours, both continuous variables, were mean-centered, following the recommendations by Gana and Broc (2018), to facilitate interpretations of intercept and slope means.

Additionally, two single items captured how the pandemic influenced people’s relation to their work (“The pandemic situation influenced my attitude towards my job in a positive way,” and “The pandemic situation has the consequence that I am more busy at work.”). The response scale ranged from 1 – does not apply at all to 7 – applies fully. As people started participation in the study at different states during the pandemic situation, we calculated a mean of three measurements to get an average score for each item.

We assessed weekly hours worked at all three waves. The average in working hours were 37.4 at T1, 34.7 at T2 and 35.5 hours per week at T3, respectively. With the beginning of the pandemic crisis resulting in more irregular working hours, we included an extra item of “hours commonly worked” at T2 and T3, which gives the opportunity to divide between “usual” working intensity and working hours per week adapted due to the pandemic situation. The test on differences of the mean was statistically significant for T2 (t (299.42) = ‑2.24, p < 0.05), but there was no difference detected for T3 (t (161.29) = ‑0.77, p = 0.44).

Further, we assessed educational level via the highest reached qualification out of ten different categories (from “no graduation” to earning a “PhD”), income of participants excluding basic income during the 12-month period (up to €900, €901–1,300, €1,301–1,500, €1,501–2,000, €2,001–2,600, €2,601–3,200, €3,201–4,500, €4,501–6,000, above €6,000). We also controlled for the number of dependent minors in the household at T1.

Statistical Analyses

In order to test the stated hypotheses H1–H3, we applied Structural Equation Modeling (SEM) with Latent Growth Curve Models (McArdle & Epstein, 1987; Meredith & Tisak, 1990), as LGCMs estimate a fit between specific growth trajectories and the data by modeling two latent factors. These are the intercept factor, representing the initial level of mental health, and the slope factor, indicating change of mental health across measurement occasions. The latent growth models for the two measures of mental health, life satisfaction and well-being, were estimated separately. Model specifications are shown in Fig. 1.

Fig. 1
figure 1

Graphical Representation of Latent Growth Curve Analyses. Note. Fixed loadings of intercept (i) and slope (s) of the outcome variable Y. Predictors autonomous motivation (AM) and controlled motivation (CM). AM is modelled as time-invariant covariate (TIC) in Model 1 and 2, and as time-varying covariate (TVC) in Model 3 and 4

Full size image

First, we specified changes in life satisfaction and well-being as latent outcomes by creating a basic latent model with fixed loadings for both the intercept and the slope parameter. We tested this model against an intercept-only model wherein no growth was assumed. After establishing Model 1, we added the predictors and covariates. In the next step, we included both types of motivation at T1 (autonomous and controlled motivation) to estimate their predictive power for initial levels and changes in life satisfaction or well-being (Model 2). In Model 3, we switched to inclusion of autonomous motivation as a time-varying covariate (TVC), with all other variables identical to Model 2. To model the influence of TVC autonomous motivation on the change of life satisfaction or well-being, a fourth model estimated the path from autonomous motivation at T1 to the slope of each respective outcome.

To control for sociodemographic influences, models 2 to 4 included age, gender, working hours per week, education, income, and minor household members as well as the two indicators of change in relations to work due to the COVID-19 pandemic as covariates. The impact of the covariates was tested by comparison of Bayesian information criterion (BIC), which improved in all models.

For the longitudinal mediation analysis in H4, we followed the suggestion by Hamaker et al. (2015) to utilize Random Intercept Cross-Lagged Panel Models (RI-CLPMs). RI-CLPM is considered ideal for examining temporal associations. It takes various sources of error into account, such as stability of the variables and cross-sectional as well as prior associations, and compensates for shortcomings of common Cross-Lagged Panel-Models (Mund & Nestler, 2019; Orth et al., 2021). A main feature of the RI-CLPM is that it distinguishes between a within-person level and a between-person level (Mulder & Hamaker, 2021). In RI-CLPM, it is assumed that instead of individuals varying around a group mean over time, each person fluctuates around a person-specific, trait-like level over time concerning the construct under investigation. In contrast to traditional CLPM, a latent intercept factor is modelled for each construct across all time points measured. In our study, applying RI-CLPM had the advantage of differing between improvements of mental health as outcome and influences of autonomous motivation and needs satisfaction a) on a rather personal (within-effects) level and b) on a level between persons, namely testing tendencies in the entire sample. We tested the influence of basic psychological needs at work on the outcome (Model 5a) compared to a model with autonomous motivation as a mediator (Model 5b). The models are depicted in Fig. 2.

Fig. 2
figure 2

Graphical Representation of the RI-CLPMs. Note. Three wave longitudinal models. For sake of parsimony, the figure is not completely mapped. Model 5a consists of X as need for competence, autonomy, or relatedness; outcome Y stands for life satisfaction or well-being; Model 5b incorporates mediator autonomous motivation (AM) additionally

Full size image

In line with the recommendations of van den Broeck et al. (2016), we constructed separate models for each psychological need. In present mediation models, the cross-lagged coefficients express the prospective effect of a temporary deviation from an average level in needs satisfaction on the change in a temporary deviation from an average level in autonomous motivation (paths a1 and a2). Further, they express the prospective effect of a temporary deviation from an average level in autonomous motivation that leads to the change in a temporary deviation from an average level in life satisfaction or well-being (path b1 and b2). Finally, a total effect is estimated, representing the pathways from the temporary deviation from an average level in needs satisfaction on the temporary deviation from an average level in the outcomes (path c). All the covariates that reached significance in the previous LGCA were also included in the RI-CLPM.

Overall fit of SEM models was determined using the Comparative Fit Index (CFI), the Standardized Root Mean Square Residual (SRMR), and the Root Mean Square Error of Approximation (RMSEA) with 90% confidence interval. We considered models with CFI values > 0.95, SRMR < 0.08 in combination with a maximum upper bound of the RMSEA’s 90% confidence interval of < 0.08 to have good fit (Beauducel & Wittmann, 2005; Heene et al., 2011; Hu & Bentler, 1999). The likelihood ratio χ2 statistic is also reported for the sake of completeness. Model comparison was furthermore based on the BIC difference (Raftery, 1995). All analyses were conducted using the lavaan package in R version 2022.12.0 (Rosseel, 2012). Script files can be found in the online supplementary materials.

Measurement Invariance

We tested for factorial invariance to ensure that the multi-item constructs themselves remain the same across time. We established scalar invariance for well-being and autonomous motivation, but strict variance for controlled motivation as well as for the scales of the three basic psychological needs autonomy, competence, and relatedness (see Supplemental Material at https://osf.io/9b47y/).

Results

Simple correlations among the observed variables were consistent with the directions of our proposed research hypotheses (Table 2). As can be seen in Table 3, most models from LGCA and RI-CLPM adequately fit the data. Exceptions from this were one LGCA model (Model 3 without covariates for outcome well-being) and all six RI-CLPM models without mediator (Models 5a). We continued by only interpreting the models that fit the data well.

Table 2 Correlations of the Dependent and Independent Variables
Full size table
Table 3 Structural Equation Modelling Statistics for Latent Growth Curve Models and Cross-Lagged Panel Analyses
Full size table

Trajectories of Life Satisfaction and Well-Being in Latent Growth Curve Models

In H1 and H2, we investigated in the change of health-related outcomes during UBI payment and expected to see growth. The Latent Growth Curve Model (Model 1) of life satisfaction fits the data well (SRMR = 0.010, RMSEA = 0.000, RMSEA 90% CI = [ 0.000, 0.000], CFI = 1.00), significantly better than the zero-growth model according to the likelihood ratio test and BIC difference (χ2Diff (df) = 45.802 (3), p < 0.000, Δ BIC = 28.20) (H1). This was also the case with the model for well-being (SRMR = 0.013, RMSEA = 0.000, RMSEA 90% CI = [0.000, 0.060], CFI = 1.00; χ2Diff (df) = 41.437 (3), p < 0.000, Δ BIC = 23.80), indicating that growth had occurred (H2). Loadings on the slope were fixed at 0 for T1 and at 1 for T2 and T3.

Parameter estimates (see Table 4) suggested an average level of life satisfaction of 6.314 (SE = 0.085, p < 0.001) at the beginning of the study, with an increase of 0.676 (SE = 0.190, p < 0.001) from T1 to T2. The intercept variance of 0.916 (SE = 0.281, p < 0.001) suggested significant individual differences in initial levels of life satisfaction, but no variance in growth was detected. For well-being, the initial average level was 3.453, with an increase of 0.356 (p < 0.001) from T1 to T2. Again, participants varied significantly in their life satisfaction at the beginning of the study (b = 0.254, p < 0.01), but they did not differ significantly in their developmental slope parameter. Notably, for life satisfaction and well-being, the correlation between intercept and slope was negative (-0.233 for life satisfaction, -0.260 for well-being). Hence, we could confirm that there is growth in life satisfaction and well-being, and therefore H1 and H2 were supported.

Table 4 Coefficients in Latent Growth Curve Analysis for Life Satisfaction and Well-being
Full size table

Quality of Motivation and Latent Change in Life Satisfaction and Well-being

In the next step of Latent Growth Curve Analysis, we addressed our third hypothesis (H3) about autonomous motivation as predictor of individual trajectories and differences in change. For Model 2, we included autonomous and controlled motivation at T1 in the basic Model 1. First, it must be mentioned that autonomous and controlled motivation were not associated significantly, neither for the life satisfaction model (r = 0.153) nor the well-being model (r = 0.155). Indeed, the two were not associated in any of the models tested in the study. Autonomous motivation at the beginning of the study positively predicted the initial level of life satisfaction (b = 0.453, p < 0.001), but negatively predicted the change in life satisfaction (b = -0.177, p < 0.05).

Testing for robustness of effects by inserting the set of the additional covariates (age, gender, etc.), the significance of the slope path diminished. For well-being, merely the intercept path was significant (b = 0.174, p < 0.001).

When modelling autonomous motivation as time-varying covariate (TVC, Model 3), each path from autonomous motivation at time T on life satisfaction at the same occassion showed significant effects (b1 = 0.478, b2 = 0.516, b3 = 0.506, p < 0.001), even when further covariates were included. Among the set of covariates, gender (b = 0.456, p < 0.001 for life satisfaction; b = 0.371, p < 0.01 for well-being) turned out to affect the initial levels of the outcomes, in that female UBI recipients were less satisfied and reported lower well-being at the beginning of the study. Income at T1 affected the initial level of life satisfaction (b = 0.153, p < 0.01), but not well-being. Concerning prediction of the slope, working overtime due to the pandemic crisis was a significant, negative predictor, influencing the growth of well-being (b = -0.086, p < 0.01).

In Model 4, we expanded Model 3 by estimating a loading from autonomous motivation at T1 on the slope of the outcome. Autonomous motivation at T1 became a strong predictor for the slope part with a negative coefficient (b = ‑0.268 for life satisfaction, b = ‑0.147 for well-being; p < 0.01), indicating that autonomous motivation at T1 decelerates the changes in life satisfaction and well-being. The covariates predicted initial level of life satisfaction and initial level of well-being as in Model 3 but added nothing to the prediction of growth in life satisfaction or well-being. The pandemic indicator of working overtime diminished as a predictor in contrast to Model 3.

Autonomous motivation influenced psychological health positively, both when included as a time-invariant and time-varying covariate. Hence, H3 was confirmed. But at the same time, it is true that people with less autonomous motivation at the beginning of payment showed greater improvement in life satisfaction and well-being during UBI. A further result of the LGCM was the consistent predictive power of two sociodemographic covariates. The first was gender, and the second was income. Most of the LGCM models showed significant influence of both covariates on the initial level of the outcomes, therefore we continued to include gender and income as covariates in the RI-CLPMs.

Mediating Effect of Autonomous Motivation in RI-CLPM

To test the mediation effect of autonomous motivation on the relation between needs and psychological health (H4), we set up six Random-Intercept Cross-Lagged Panel Mediation Models (3 needs * autonomous motivation * 2 outcomes), referred to as Models 5b. Parallel to this, we tested models without mediator. As reported before they showed poor model fit (Model 5a). At the within-person level, few of the autoregressive and cross-lagged parameters shown in Tables 5 and 6 were significant. This indicates that when an individual is above or below his or her person-specific average level of needs satisfaction, autonomous motivation, and life satisfaction or well-being at a specific time point, this individual is not expected to score above or below his or her usual level in any of the related variables at a subsequent time point. Exceptions from this general finding occurred in two models: First in the RI-CLPM of life satisfaction and need for competence, the autoregressive × 2-path was positive (b = 0.309, p < 0.01), suggesting that a score in life satisfaction at T2 above the person-specific mean in life satisfaction was associated with a significant higher score in life satisfaction at the subsequent T3. Second, the model of well-being and the need for autonomy showed a significant, cross-lagged negative b1-path (b = -0.405, p < 0.01), indicating that an individual’s score in autonomous motivation at T1 below the person-specific level of autonomous motivation leads to a score of well-being at T2 that is expected to be above the person-specific mean of well-being. Gender and income as covariates in the model added nothing with regard to the prediction of needs, autonomous motivation and life satisfaction or well-being at T1.

Table 5 Results of the RI-CLPM for the Interplay between Needs Satisfaction, Autonomous Motivation and Life Satisfaction
Full size table
Table 6 Results of the RI-CLPM for the Interplay between Needs Satisfaction, Autonomous Motivation and Well-being
Full size table

The positive within-time correlations (Table 7) indicated that within-person increases in needs satisfaction above the person’s mean level are accompanied by person-specific increases in autonomous motivation (rx1m1, rx2m2, rx3m3) and life satisfaction (rx1y1, rx2y2, rx3y3) compared to the person’s average levels. Additionally, a within-person increase in autonomous motivation above the person’s mean level led to person-specific increases in life satisfaction (rm1y1, rm2y2, rm3y3). The same positive relations occurred for the interplay between the outcomes of well-being, autonomous motivation, and needs satisfaction.

Table 7 Correlations between random intercept factors and between within-time correlations of needs, autonomous motivation and the outcome
Full size table

At a between-person level, the random intercepts had moderate to high correlations with the strongest relations between random intercept of autonomy needs satisfaction (x) and random intercepts of both of the outcomes (rixiy = 0.816 for life satisfaction; rixiy = 0.826 for well-being, p < 0.001). This indicates that individuals with higher needs satisfaction than others tend to be more satisfied in life and report greater well-being than other individuals.

Moreover, the results indicate that there is a mediating effect between needs satisfaction, autonomous motivation and health-related outcomes (H4). One exception to this was the statistically insignificant relation between relatedness, autonomous motivation, and well-being, as both indirect paths showed no significant relations (rixim = 0.573, SE = 0.326; rimiy = 0.778, SE = 0.426). Although significant, the lowest association of the constructs under investigation occurred between needs satisfaction and autonomous motivation. In line with previous results of our study from LGCM and in line with H3, autonomous motivation showed strong positive connections to life satisfaction (rimiy = 0.556 to 0.592, p < 0.001) and well-being (rimiy = 0.651 to 0.703, p < 0.01). In sum, these results support H4.

Discussion

The present study gives unique insight into the dynamics underlying life satisfaction and well-being in working individuals who receive UBI. It investigates how, following the SDT framework (Ryan & Deci, 2017), autonomous motivation and the satisfaction of basic needs affect changes in mental health. In particular, this longitudinal study aims to extend the sparse existing findings on UBI by considering psychological explanations for motivation at work and the development of mental health.

The results imply that there is overall growth in life satisfaction and well-being during UBI payment. This is in line with a large body of research in the field of income and life satisfaction and well-being (Frijters et al., 2004; Mund et al., 2021; Schyns, 2001; Sirgy, 2021). The level of mental health cannot be increased indefinitely but reaches a plateau, an effect replicated in our data. Therefore, our findings reinforce the view that an adequate amount of money is a necessary but insufficient condition for mental health. We detected no inter-individual difference in change in LGCM, indicating that participants generally made gains in life satisfaction and well-being throughout the duration of the study, primarily in the first half of the year. Nevertheless, the finding of growth in the health-related outcomes is notable, considering that the COVID-19 pandemic was unfolding concurrently. During this period, other researchers found a decrease in well-being and life satisfaction in the general population (Jung et al., 2020; Wang et al., 2020). The results are in line with evidence from another study in Spain where subjective well-being increased during UBI payment (Sekulova et al., 2023). This sample was selected differently than ours; the monthly payment amount was dependent on neediness, and participation in the program was restricted to neighborhoods of Barcelona with high concentrations of socioeconomic vulnerability, rates of poverty, and social deprivation for the city. The same occurred for a further Finnish sample of randomly selected unemployed individuals, where people’s life satisfaction in the treatment group, receiving unconditional cash transfers for two years, was significantly higher than in the control group (Kangas et al., 2019, 2021). Our results go further to suggest a robust and generally positive effect of UBI even among a working population, especially in otherwise challenging times.

The most striking findings of our study is that autonomous motivation was a strong predictor for change in mental health, both when included as single measurement covariate (time-invariant covariate, TIC) and as time-varying covariate (TVC) in LGCM models. Additionally, this result was confirmed in RI-CLPM. A distinction between a within-person and a between-person variance showed that interrelations between needs satisfaction, autonomous motivation, and life satisfaction or well-being occurred on both levels (within- and between-person). Effects even remained stable upon inclusion of other covariates into the models. This underscores the robustness of the results. With regard to H3, autonomous motivation influenced life satisfaction and well-being in a positive way. This is in line with a large body of research on the connection between autonomous motivation and well-being (Ryan & Deci, 2017; Wang & Panaccio, 2022). Furthermore, confirming the importance of different qualities of motivation postulated by SDT, controlled motivation did not affect the trajectory of psychological health, presumably due to its lack of benefits for optimal functioning (Gagné et al., 2015).

An unexpected finding was that autonomous motivation was negatively related to growth in life satisfaction and well-being in the TVC model (Model 4). This might indicate a compensatory effect in which UBI to some extent buffers against the usual (negative) impact of low levels of autonomous motivation in working people on life satisfaction and well-being. This result must be considered preliminary until replicated in future research. At present, we can only speculate about its nature. Here, we can think of three different explanations: One might be that people with low autonomous motivation at the beginning of the study have more potential and room to develop life satisfaction and well-being in general. Second, UBI ensures satisfaction of basic financial needs for food and shelter, allowing people to focus on gaining satisfaction of their basic psychological needs – in terms of SDT – from life domains outside work, thereby increasing life satisfaction and well-being indirectly (Milyavskaya et al., 2013). As Gibson et al. (2020) argued in favor of UBI, it enables people to spend more time on creative or business projects, community work, or care of family members, invest time in their own education, and improve their health by reducing stress. A third explanation might be the direct effect of UBI on the working context. In line with Ferdosi and McDowell (2020), working people may benefit in various ways, as they receive financial security, gained better bargaining conditions and the ability to emancipate themselves from precarious working conditions. This argument is also often emphasized by advocates of UBI with the rationale behind that people get empowered by a more contemporary social security system that covers modern uncertainties (Standing, 2009; Widerquist, 2013).

To sum up, self-determined motivation is neglected in narrowly focused work for mere sustenance. Under UBI payment, work becomes less important for financial reasons such that people are not forced to invest time, energy, and effort into a life domain to which they have no inner connection – possibly even an aversion. These explanations allow for to better understanding of the growth in mental health given less autonomous motivation at work in our study.

Gender and income were also noteworthy as predictors in LGCM, at least for the initial level of mental health. Women and people with lower income reported lower life satisfaction and well-being at the beginning of the study. The finding of lower mental health in employed women is not surprising, considering the previous research on dual loads of employed women, especially during the pandemic (Jung et al., 2020; Syrek et al., 2022; Vicari et al., 2022). During payment, UBI seems to compensate somewhat for these differences in gender and income, although we did not detect significant effects on growth of life satisfaction or well-being in our study.

For basic psychological needs satisfaction, the RI-CLPMs suggested that needs predicted life satisfaction and well-being both directly and indirectly. Overall, models only fit the data when autonomous motivation was included as mediator. This replicates the results of many studies in the SDT framework, including in the context of work (Ryan & Deci, 2017; Wang & Panaccio, 2022). However, in the model predicting well-being, the indirect path from relatedness via autonomous motivation to well-being was not significant. In the literature, the findings on the need relatedness have been inconclusive to date: On the one hand, empirical work partly points on an influence of relatedness at work on self-determined motivation (Richer et al., 2002), while on the other, some see relatedness as a more distal need that is not as important as autonomy and competence (Deci & Ryan, 2000). Nonetheless, the direct path from relatedness to well-being was significant, indicating that the feeling of being connected to others at work is relevant to one’s well-being, albeit not through autonomous motivation.

At the within-person level, we detected very few effects, and all of them pointed in expected directions. For example, we found an autoregressive path of life satisfaction from T2 to T3, meaning that a high deviation of an individual’s average level of life satisfaction at the middle of UBI payment leads to deviations from the mean level in the same positive direction at the end of the year with UBI payment. This indicates a stabilization in life satisfaction during the payment. We also found a negative path from autonomous motivation at T1 to well-being at T2 meaning that deviances below the individual’s average level of autonomous motivation predicted above average well-being at the subsequent time point. Both effects are consistent with our findings in LGCA.

The results have a number of implications on a theoretical and practical basis. First, people will benefit from UBI – similarly to how additional income more generally positively influences mental health. As politicians and economists become more familiar with the costs of mental illness, they may be more open to cost-reduction plans based on the potential health benefits of UBI. Moreover, long-term projects would be ideally suited to evaluating development and changes under UBI, particularly with the inclusion of psychological evaluations. Second, our study provides further reasons to consider how autonomous motivation, and needs satisfaction in the workplace more generally, can be improved beyond offering UBI. Our results show that autonomous motivation holds the potential to influence mental health, independent of people’s age, gender, income, education, working hours per week, children living in the household, or even pandemic factors. Therefore, this should be one of the leading goals for employers, politicians, and individuals themselves to create, support and select working conditions that enhance self-determination in order to improve mental health. Finally, our results suggest that UBI should not undermine participation in the workforce in general, as long as basic needs at work are met and autonomous motivation prevails. Interestingly, we found that working individuals who showed a lack of autonomous motivation, due to lack of basic needs satisfaction at work in autonomy and competence, benefitted the most under UBI in terms of their mental health.

One limitation of our study might be the self-selective character of people in the sample. They had to register at a website for the UBI pilot program. For this reason, we may have captured people who are already interested in changes (e.g., to the security system, to working conditions, to circumstances in life in general) and, by this logic, are therefore interested in the project to a certain extent. In other words, their response behavior might be somewhat colored. Additionally, although anyone could register on the association’s website without restrictions and the annual UBI was allotted at random, our sample consisted solely of people with a graduate degree or higher. This demonstrates some initiative towards UBI, especially among the highly educated and thus limits the generalizability of the results. Naturally, a fully randomized-controlled trial would be beneficial for future investigations.

Nevertheless, our results in trajectories of life satisfaction and well-being support findings from previous UBI studies and expands the insights. To the best of our knowledge, this is the first empirically confirmed result of its kind specifically in the context of UBI. It therefore has important implications for future projects in this field, as well as for political and economic decision-making. In fact, we did not detect evidence for any negative effect of UBI on mental health or working motivation. This is directly relevant for current discussions around UBI and should encourage research exploring UBI more deeply as an alternative to prevailing social security systems in western countries.