Deprivation, Social Mobility Considerations, and Life Satisfaction: A Comparative Study of 33 European Countries

This study aims to provide a comparative analysis between non-transition and transition countries, with focus on exploring the life satisfaction costs of deprivation aspects, i.e. material, subjective, and relative. Relative deprivation is measured using the Gini index at the city level, since the Gini index at the country level is unable to capture the total influence of relative income inequality on life satisfaction for both sets of countries. A negative association between these measures and life satisfaction is suggestive of deprivation dimensions being quality-of-life important considerations in the EU and neighbouring candidate countries. They capture objective and subjective income inequalities among different households within the same time point. The relative importance of such indicators is also of particular interest because it is driven by social mobility considerations, which are more related with whether people think they can improve their lot in life and how easy/hard this is. This is more related to present versus future concerns about inequalities of same households across time. The study is based on a comparative analysis of data taken by a nationally representative household database from the 2016 European Quality of Life Survey. We evaluate the hypotheses using a two-level linear mixed-effects model of individual responses nested in 33 European countries (28 EU plus 5 non-EU countries—Albania, Montenegro, North Macedonia, Serbia, and Turkey). Estimates are generated for the pooled sample and separately for the non-transition (West-EU) and transition post-communist (East-EU plus non-EU) countries. The results suggest that there are significant life satisfaction costs attached to all three aspects of deprivation. However, the comparative importance of relative deprivation, as a measure of income inequality at the city/local level, is significantly larger than material and subjective deprivation, even after controlling for equivalized household income. This relationship is more pronounced for transition countries as compared to non-transition (post-communist) ones.


Introduction
This research recognizes the income-related drivers of people's happiness 1 are inherently multi-dimensional, showing that while income per se matters, there are potentially other latent aspects related to income that explain differences in happiness (Bellani and D'Ambrosio 2011;Gravelle and Sutton 2009;Jebb et al. 2018;Kahneman and Deaton 2010). Among these factors, material, subjective, and relative deprivation could potentially hold the key to the understanding of happiness differentials.
The scope of this study is to empirically investigate the relationship between deprivation aspects and subjective life satisfaction (SLfSat), after controlling for income. SLfSat is the degree to which a person positively evaluates the overall quality of their life as-a-whole, in other words, how much the person likes the life they lead (Veenhoven 1996). It is a concept central to the research field of subjective wellbeing which assumes that general satisfaction with life judgements is the sum of satisfaction with life domains and there is an assumption that this can be reached (Diener 1984;Headey et al. 1991;Leonardi et al. 2005).
In particular, we are interested in evaluating the effects of relative (comparative) deprivation (relative income, i.e. social comparison/status effects) on SLfSat and establish whether these are larger than the effects of subjective (perceived) and material deprivation. The latter have been found to have weaker effects on SLfSat (Angel et al. 2003;Christoph 2010;Fusco et al. 2011;Pepper and Nettle 2017). Relative income has often been the only measure used to capture social comparisons effects on SLfSat (Clark and Senik 2010). However, there is an ongoing debate about what the relative measures (income, wealth, health, material, etc.) and reference groups (friends, family, colleagues, employer, village, country, etc.) should be and we offer an account on this in the literature review section. Evidence suggests that groups such as family, friends, former classmates are homogenous and lack information about how other groups are ranked in the social ladder, which might in turn indicate 1 In line with much of the related research, the terms '(overall) life satisfaction', '(subjective) satisfaction with life', 'happiness' and 'subjective wellbeing' are used equivalently. The reason for this is that survey responses to questions on life satisfaction or subjective wellbeing are highly correlated with alternative indicators of happiness (Ferrer-i-Carbonell and Frijters 2004;Kahneman et al. 1999;E. Nikolova and Sanfey 2016). Measures of life satisfaction are also used extensively in empirical studies in 'economics of happiness'. Similarly, the 'post-communist happiness gap' is also mainly concerned with a gap in life satisfaction (Amini and Douarin 2020;Djankov et al. 2016).
However, it is also recognized that subjective wellbeing (SWB) represents a comprehensive concept that incorporates affective (happiness) and cognitive (life satisfaction) components of life, which, respectively, relate to the emotional and evaluative sides of SWB (Diener et al. 2003). Indeed, Frey and Stutzer (2010) argue that happiness is affect (an emotional assessment) and life satisfaction is evaluation (a cognitive assessment) which is much related to how people evaluate their current quality of life in comparison with the ideal/benchmark they would seek for themselves. Additionally, the determinants of life satisfaction and happiness, such as income (Kahneman and Deaton 2010) have been shown to differ, implying a distinction between these concepts. incidence of the so-called adaptation to income effects. On the other hand, relative income at the country level (usually measured by the Gini Index) is calculated on the assumption that everybody compares to everybody. There is a scarcity of studies that investigate the relative income effects in SLfSat at a more intermediate local level. Consequently, including relative income comparisons at the city level could, in principle, capture information on how other groups are doing and whether there are members of the group who climbed the social ladder, which could capture in fact the 'social comparisons' effects. The incidence of wealthier individuals among the population can inform individuals about their own potentials to become wealthier in the future, and this can lead to an increase in present SLfSat, before any enrichment takes place (multi-period information) (D'Ambrosio et al. 2020). Therefore, in this paper we complement the relative income at the country level with relative income at the city level. We identify it as 'relative deprivation' in terms of income as the reference groups are of local and of more heterogeneous nature.
Empirical evidence suggests that relative income concerns tend to translate considerations in relation to mobility of social status and life style (Arber et al. 2014;Cojocaru 2014;D'Ambrosio and Frick 2007;Jung 2018). Social mobility potentials stemming from inequality of income are in fact related to the subjective meaning and value that a person attributes to their own class-based identity in a multiperiodic (present vs. forthcoming) sense (Destin et al. 2017). Therefore, we also include social mobility considerations, which comprise the sociological approach, to account for concerns about the socio-economic status (class) of a given household across time. The interaction between relative income at the city level and social mobility concerns aids our attempt to pin down the incidence of the 'social comparisons' and 'adaptation to income' effects (Chen and Miller 2013;Simandan 2018). Indeed, how social mobility interacts with relative deprivation to influence SLfSat is essentially something that characterizes a society, or the individual's relationship to the society (Dardanoni 1993). Evidence suggests that these considerations are more prevalent in transition contexts which have experienced rapid changes of political regimes (Clark and Senik 2010;Guriev and Zhuravskaya 2009;Sanfey and Teksoz 2007;Zagórski 2011). Regarding relative deprivation, Senik (2009) argued that in a context of economic and social transition-such as the fall of communism-which results in a rapid proliferation of inequality combined with rapid growth, comparison to reference groups might inherently contain 'information' for one's personal future expectations (Evans and Kelley 2017;Whelan et al. 2001;Zagórski 2011).
In this regard, the contribution of this study is threefold (i) Firstly, because of the limitations of a narrower income-based measure of poverty, we go one-step further and include material, subjective, and relative deprivation that come from lack of income. We find that, altogether, deprivation dimensions have negative effects on SLfSat, but the largest size effect is captured by relative deprivation. This indicates that deprivation is a concept that goes beyond basic needs, and mostly relates to social status aspects. (ii) Secondly, social mobility considerations are found to further impair the adverse effect of relative deprivation. (iii) Thirdly, the above adverse effects are more pronounced for transition as compared to non-transition countries.
There are several reasons for these differences, partly related to the cultural background of the societies' structure which are vested in peoples perceived values and attitudes towards inequality and wellbeing prospects. Therefore, in undertaking this research, we seek to further advance theoretical views about the complex interrelationships between SLfSat, income, material deprivation, perceived financial hardship, and comparative income inequality. To provide empirical insights, we analyse data from a nationally representative household database, i.e. the 2016 wave of the European Quality of Life Survey.
The remainder of this paper is organized as follows. The next main section reviews theoretical and empirical evidence that inform the hypotheses. Section "Data and Variables" describes the data and variables employed in the empirical analysis. Section "Strategy of Empirical Analysis" presents the identification strategy. Section "Results" explains the results and main findings. Section "Discussion and Conclusion" closes with discussion and a conclusion.

Deprivation and Subjective Life Satisfaction
In this section, we review the theoretical foundations and empirical evidence on different dimensions of deprivation, i.e. material, subjective, and relative deprivation that have been used to capture some of the limitations of a narrower income-based measure of poverty in terms of SLfSat. These deprivation dimensions relate to the resource-based perspective which suggests that possession of resources and sources of support can alleviate the conditions of income poverty (Gilbert 2009). Similarly, the conservation of resources theory (Hobfoll 1989) classifies deprivation as a stressor that causes a disruption in the flow of resources. As stated by Iceland and Bauman (2004), persistent income poverty could determine multi-dimensional deprivation through three different channels: it cumulatively increases the divergence between the necessary and the available resources to fulfil basic needs (material deprivation); it gives rise to more erratic incomes and needs that make it difficult to make ends meet (subjective deprivation); and it produces long-term deficiencies in the ability to fulfil such needs thus making it impossible to improve one's social status and life style (relative deprivation). In the same fashion, within the framework of the latent deprivation model (Jahoda 1982), lack of resource not only affect material and subjective deprivation, but also produce a latent relative deprivation threat in terms of life style and socio-economic status that are needed for social and human capital accumulation necessary for upward social mobility.

Material Deprivation
Material deprivation is defined as an unacceptably low standard of living (Ringen 1988) and captures the inability to possess the goods and services, and/or engage in ordinary activities, that are socially perceived as the minimum acceptable living pattern in the society to which one belongs (Fusco et al. 2011). It is one of the key factors negatively affecting SLfSat (Bellani and D'Ambrosio 2011;Fusco et al. 2011). Indeed, income may fail to capture the mechanism that explains the unfavourable effects of material deprivation on SLfSat. Studies have shown that income does not predict one's material situation as it does not include wealth and consumption, which are central to the material situation-SLfSat nexus (Diener and Biswas-Diener 2002;Headey 2008Headey , 2019. It is possible for two individuals/households to have the same disposable income but their income alone does not measure adequately all the resources that are available to each of them (wealth) and/or if their needs (consumption) differ, and this will result in different levels of material deprivation (Whelan et al. 2001).

Subjective deprivation
Subjective deprivation refers to the individual's self-rating of their income adequacy to meet their general needs. It is thought to be associated with perceptions of financial strain and stress (Arber et al. 2014). Angel et al. (2003) argue that, due to its distinct meanings and potentially individual consequences for SLfSat, it is important to distinguish perceived financial hardship from income and material deprivation. The lack of research in this area is surprising given that Angel et al. (2003) argued the importance for both researchers and policy makers of differentiating whether it is income and material deprivation alone, or an individual's perception of their financial situation that impacts SLfSat.
Objective measures of income do not capture the meaning of income adequacy to individuals (Hazelrigg and Hardy 1997;Mirowsky and Ross 1999) with people on low incomes not always reporting financial strain, which indicates that these two measures are different and therefore may differentially affect SLfSat. Similarly, financial strain has been found to diverge for the same level of income (Kahn and Fazio 2005) as different individuals experience different inadequacies to meet their needs (Zimmerman and Katon 2005).

Relative Deprivation
While material and subjective deprivation are more related to the lack of resources necessary for the fulfilment of basic physical needs, they deal more with the survival perspective. Relative deprivation, on the other hand, is about 'not being like others'. Yitzhaki (1979) argued that relative deprivation captures the differences in economic resources in relation to reference groups that could potentially be concealing the existence of very different mechanisms to translate income into changes in living conditions. In terms of effects on SLfSat, the comparative effect of relative deprivation has been found to have larger effects than those of material and subjective deprivation (Angel et al. 2003;Arber et al. 2014;Fusco et al. 2011;Greitemeyer and Sagioglou 2019;Pepper and Nettle 2017). Indeed, social comparison and expectations may lead to differences in the perception of the adequacy of income which are likely to be related to differences with an individual's reference group (Whelan et al. 2001;Whelan and Maître 2010).
There is an ongoing debate about what the relative measure (income, wealth, health, material, etc.) and reference group (friends, family, colleagues, village, etc.) should be to calculate relative deprivation. The more frequent reference groups used in previous studies as the basis of relative comparisons were peer reference groups, arguably reflecting reference-dependent preferences (Kahneman and Tversky 1972). Reference group theory suggests that individuals' social perceptions rely on own experiences and the experiences of their families, friends, and co-workers. Individuals tend to compare themselves with those considered to be "at about the same level" on given dimensions such as age, education, gender, socio-economic status, employment status, and geographical closeness as related features for social comparisons (Alesina et al. 2004;Brock 2020;Cojocaru 2014). Empirical evidence suggests that closer groups such as former schoolmates, colleagues, family members tend to be more homogeneous (Clark and Senik 2010;Evans and Kelley 2017) and altogether lack information about society as a whole (Evans and Kelley 2017). In fact, these social circles are not representative of the local composition of society (social ladder) and create systematic differences in the perception and experience of inequality. This results in turn in the incidence of the so-called adaptation to income. Therefore, in order to be able to capture information on whether there are members of the group who succeeded/failed to climb the social ladder and how other groups are doing-in other words, the 'social comparisons' effects-more heterogeneous groups are needed to widen the scope of the relevant reference group. Copestake (2011) provides evidence that supports the link between SLfSat with locally framed aspirations (such as raising a family, find a place for better living, and professional achievements) in the poorer regions of Peru. Similarly, two-thirds of surveyed individuals in China report that their main comparison group base are the individuals in their own village (Knight et al. 2009). In Latin America, the effect of relative status on SLfSat is found to be strongest at the city level rather than the country level (Graham and Felton 2006). Genicot and Ray (2022) suggest that interpersonal inequality is shaped by the "cognitive neighbourhood" as this is where the ambitions and goals, as the basis of the reference point, are formed. Wider local comparisons within city or villages are connected to individual utility functions, and as societal distributions changes, the reference point changes as well (García-Castro et al. 2020).
Central to the measurement of relative comparison is usually a scalar entity such as income, but in practice this entity could be wealth, expenditure, wages, access to primary goods, health or some other economic quantity the distribution of which is of particular interest (Alkire and Santos 2009;Cowell 2008;Wilkinson 2001). There are studies which have also analysed comparisons with the above reference groups (e.g. family members, parents) in terms of what is considered 'fair' (e.g. a fair wage) (Clark and Senik 2010). Relative income has often been the only measure used to capture social comparisons effects on SLfSat. The most frequently used empirical measurement of inequality is the Gini Index at the country level. However, its limitation relies on the fact that while it captures relative income, it assumes that everybody compares to everybody within the nation. People consider their income and access to resources relative to that of their local reference group rather than income at the national level and adjust expectations accordingly. Including relative income comparisons at the city level, on the other hand, could in principle capture dynamics beyond the adaptation (within-group), which relate to social comparisons (betweengroups) effects. This is well aligned to Festinger's (1954) social comparison theory, which suggests that individuals might be unaware of people's circumstances outside their local place of residence and will therefore tend to compare themselves with others who have similar profile in a neighbouring locality. Consequently, in this paper we complement the relative income at the country level (Gini index at the country level) with relative income at the city level (Gini index at the city level) which we identify as 'relative deprivation' in terms of income as the reference groups are of local nature. According to Pedersen (2004), the Gini index can be interpreted and rationalized as a meaningful measure of relative deprivation if it is applied to a 'reference population', where it can be assumed that each member of the collective defines is/her social standing with reference to everybody else-in other words the city level. Guriev and Zhuravskaya (2009) suggested that the increase in income inequality in transition countries had larger negatively effects on SLfSat compared to nontransition countries. Similar findings are reported in Evans and Kelley (2017) and Bottero (2019). Therefore, while the material and subjective deprivation dimension help to capture concrete and plausible adverse SLfSat effects that go beyond those of income, it is relative deprivation of those in the lower segments of income compared to upper reference groups that accounts for experiencing continuous lower SLfSat in comparison with others in their own city or village.
In conclusions, given the importance of deprivation dimensions for SLfSat, the fact that relative deprivation is expected to have a larger relative effect on SLfSat, and whether a country did undergo the transition from planned to market economy, we hypothesize that: Hypothesis 1 All other things being equal, (a) material, subjective, and relative deprivation will have negative effects on SLfSat, (b) these adverse effects are expected to be larger for relative (vs. material and subjective) deprivation, and (c) these adverse effects are expected to be larger for transition (vs. non-transition) countries.

Relative Deprivation and SLfSat: the Moderating Role of Social Mobility Concerns
Considerations about upward mobility with respect to social status are intimately related to whether a society has a more/less rigid hierarchical class structure. Present socio-economic ranking, in other words current relative deprivation, would not be perceived as worrisome (thus affecting current SLfSat) if the prospects are brighter from a multi-period approach. This is related to the rigidity of the class structure of a society which is determined by its relative ease and frequency of social mobility, that is, moving into a different class than that into which one was born (Simandan 2018). Social class is a complex construct that seeks to measure individuals' ranking in the socio-economic ladder. Common variables determining one's social class are education, occupational prestige, and income (Loignon and Woehr 2018;Pepper and Nettle 2017).
Social mobility is assumed to be achieved by generations via a higher educational or occupational level or by finetuning work profession or position (intergenerational mobility) (Hadjar and Samuel 2015). However, this effect depends on mobility and the structure of social classes (Dhoore et al. 2019). Empirical work of previous authors has attempted to explore the effect upward social mobility in achieving a higher life satisfaction. Few authors have found contrasting results in terms of effect of upward mobility on life satisfaction (Hadjar and Samuel 2015;Dhoore et al. 2019: Hsiao et al. 2020. Two of the most influential theories that offer an account on the potential mechanisms through which social mobility considerations moderate the adverse effects of relative deprivation on SLfSat refer to the 'relative deprivation theory' proposed by Runciman (1966) and the 'tunnel/information effect' theory proposed by Hirschman and Rothschild (1973). Relative deprivation theory (Runciman 1966) suggests that the position in relative lower income segments (worse off compared to a reference group) is accompanied by feelings of anger and resentment about being worse off than others in the long run. This means that it is related to a subjective state that shapes emotions, cognitions, and beliefs (Smith and Pettigrew 2015), which altogether connect well to SLfSat. In this fashion, relative deprivation, which positions the individual in relation to the local intergroup levels of analysis, is further connected with social and psychological concerns related to upward mobility. Similarly, Hirschman and Rothschild (1973) argued that individuals in relatively low-income segments might appreciate inequality to their local reference groups if this signals upward social mobility, a phenomenon labelled by Hirschman as the "tunnel effect". Those who can observe others around them climbing on the social ladder, adjust their expectations about their own social mobility and this improves their SLfSat because it improves expectations about their own future. Consequently, mobility considerations have broader implications for SLfSat in a multi-period sense. Empirical evidence from Senik (2009) suggests a tunnel effect in transitional countries immediately after fall of communism but that this effect later disappeared; other studies indicate more persistent traces of the tunnel effect in the transition world (Evans and Kelley 2017;Whelan et al. 2001).
Perceived mobility has been identified as a central factor that affects the link between other people's income and own SLfSat, as it regulates individual beliefs about opportunities and risks (Clark and Senik 2010;Senik 2005). In this sense, current socio-economic status and prospects for upward mobility play an important role in terms of present SLfSat (Evans and Kelley 2017;Garcia 2014;Pedersen 2004). There is broad consensus that the causal mechanism supporting these statistical associations highlights the role of resources: belonging to lower social class is related to prolonged lack of resources to control the environment which in turn fuels relative deprivation concerns and triggers deteriorating physical and psychological subjective wellbeing (Ezeh et al. 2017;McEwen 2017;Ohrnberger et al. 2017;Pepper and Nettle 2017;Whitehead et al. 2016). Theoretical and empirical evidence on the 'income-health gradient' support the claim that individual socio-economic status is positively associated with health measures of physical and mental illness, even after controlling for individual socio-demographic characteristics (Deaton 2008;Mackenbach 2002). Wilkinson's (2001) argues that the consistent socio-economic class-health gradient, is explained by the fact that human beings are reflexive, and most of the time imagine themselves through others' eyes (Rose 2006). In particular, mental illness which is a subjective state, in much compliance to SLfSat, can capture information about lack of future prospects for upward social mobility (Flèche and Layard 2017;Layard 2013). Therefore, in this study, we also include social mobility considerations, which consist of the sociological approach to capture the 'social comparisons' effects. These considerations interact with relative deprivation in the sense that they represent concerns about socio-economic status (class) of same households across time.
Given the importance of social mobility considerations for SLfSat and the importance of whether a country did undergo the transition from planned to market economy, gives rise to the following testable hypothesis: Hypothesis 2 Mobility considerations interact with relative deprivation (i.e. Gini coefficient at the city level) in a way that the SLfSat of those more relatively deprived will be particularly impaired in cities where income inequality is higher (vs. lower) and these effects are expected to be larger for transition (vs. non-transition) countries.

Data
We use the 2016 wave of Eurofound's European Quality of Life Survey (EQLS) data that documents living conditions and social situations pertinent to the lives of European citizens. From September 2016 to March 2017, Eurofound carried out its fourth survey of the adult population (18+) living in private households, based on a statistical sample, and covering a cross section of society. The 2016 EQLS interviewed nearly 37,000 people in 33 countries-the 28 European Union (EU) Member States and 5 candidate countries (Albania, the Former Yugoslav Republic of Macedonia, Montenegro, Serbia, and Turkey). This unique, pan-European survey documents both the factual circumstances of Europeans' lives as well as how they feel about those circumstances and their lives in general. The survey looks at a range of issues, such as deprivation, family, and wellbeing. It also looks at subjective topics, such as people's levels of happiness, how satisfied they are with their lives and their participation in society.

Dependent Variable: Subjective Life Satisfaction (SLfSat)
Life satisfaction is a single-item measure of how participants are satisfied with their current life in general, all things considered. Responses are on 11-point scale ranging from (0) = not at all satisfied to (10) = completely satisfied. High scores are indicative of high levels of satisfaction. This measure has been shown to be a valid and reliable proxy of utility (Diener et al. 2003;Frey 2010) and a typical measure of evaluated subjective wellbeing (Kahneman and Deaton 2010). Descriptive statistics for the variable at the individual level, and at the country and region (Transition/ West EU, non-transition: East and Non-EU) levels are introduced in Tables 6 and 8 in the Appendix. The later reveals the presence of a SLfSat gap between transition and non-transition countries (unconditional).

Independent Variables
Material deprivation is an index which uses weights from principal component analysis (Maasoumi and Nickelsburg 1988;Nguefack-Tsague et al., 2011;Ram 1982) on concerns related to subjective assessments of the capacity to meet a range of needs such as (i) keep home adequately warm, (ii) pay for a week's annual holiday away from home (not staying with relatives), (iii) replace any worn-out furniture, (iv) have a meal with meat, chicken, or fish every second day if desired, (v) buy new, rather than second-hand, clothes, and (vi) have friends or family for a drink or meal at least once a month. The aggregate material deprivation index therefore measures a notion of poverty that goes beyond basic needs, as it includes some questions related to lifestyle; the use of nationally defined weights is included to reflect the relative importance of individual items in the different societies (Fusco et al. 2011).
Subjective deprivation is measured by the perceptions of financial hardship similar to the operationalization in the study of Arber et al. (2014). Respondents report their perceived level of difficulty/easiness to 'making ends meet'. The variable is recoded to binary if the respondent made ends meet with 'some difficulty', 'difficulty', and 'great difficulty'.
In choosing these items for measuring material and subjective deprivation within the EQLS, we have followed Eurostat 2 approach for measuring deprivation dimensions which suggest that non-monetary aspects of deprivation are captured by items related to basic necessities and social connections for material deprivation and monetary aspects for subjective deprivation.
While material and subjective deprivation are measured more on items that capture individual perceptions of deprivation dimensions, relative deprivation evaluated is an objective measure related to the household income variable. Relative deprivation uses reference groups at the city level, similar to the measurements included in Gravelle andSutton (2009) andCojocaru (2014) and is measured as a Gini index for household income at the city level to capture deprivation on relative terms. Cojocaru (2014) uses reference groups based on the Census Enumeration Areas (CEA). CEA represents a few thousand individuals, and the Gini indices of city level deprivation are similarly computed at the CEA level. This is an inequality measure which is more precise than the Gini index at the country level and is measured on a 0 to 1 scale 3 .

Moderating Variables
To estimate social mobility considerations, a group of four items capturing subjective evaluations of the individual's relation to the society (social class), mainly related to income and occupational status, are used. Social mobility considerations are inherent in terms of how hurtful (as captured by negative and significant effects on SLfSat) will subjective differences in social status between self and others and feelings about being left out and looked down in current (year of survey) time will result. Therefore, optimism about future economic mobility would lower negative effects of exclusion and social status differences in terms of SLfSat. Consequently, current perceptions about such items are of instrumental nature for future (inter-temporal) predictions of upward mobility. This line of thought is consistent with the 'tunnel effect' theory (Hirschman and Rothschild 1973) which includes a dynamic and inter-temporal aspect about aspirations for future mobility. Specifically, we use (i) perceptions on being left out which is captured by the question 'I feel left out of society', (ii) on employment recognition captured by the question 'I feel that the value of what I do is not recognized by others', (iii) on being looked down captured by the question 'Some people look down on me because of my job situation or income', and (iv) on feeling close to people captured by the question 'I feel close to people in the area where I live'. Responses are on 5-point Likert scale that records the level of agreeableness with the above question such 1'strongly agree', 2 'agree', 3 for 'neither agree, nor disagree', 4 'disagree', and 5 'strongly disagree'. We have further re-coded them to binary with 1 if respondent answered 'agree' or 'strongly agree' and 0 for 'neither agree, nor disagree', 'disagree', or 'strongly disagree'.

Control Variables
We controlled for participants' curvilinear effects of age (in years), gender, highest achieved level of education (measured in three categories: ISCED 1-2; ISCED 3-4; ISCED 5-6), and household characteristics such as living with a partner (as a dummy) and monthly income equivalized in PPP Euro (continuous and logarithmic). Also, labour market status, social capital, practicing of religion, and bad health valuations are included as dummy variables. These variables have been systematically found to relate to SLfSat and have been accounted for in studies as standard controls in analyses of SLfSat (Ferrer-i-Carbonell and Frijters 2004;Ferrer-i-Carbonell and Gërxhani 2004;Frey and Stutzer 2000;Litchfield et al. 2016;van Praag et al. 2003).
As the focus of this study is on the main and interaction effects of deprivation aspects and mobility considerations on SLfSat, these variables were treated only as controls. See Table 6 in the Appendix for more details.

Macroeconomic Indicators
Annual percentage growth rate of GDP per capita for 2016 is retrieved by the World Bank national accounts data and OECD National Accounts data files. GDP at purchaser's prices is the sum of gross value added by all resident producers in the economy plus any product taxes and minus any subsidies not included in the value of the products.
Standardized World Income Inequality Database (SWIID) 2016 Gini index of disposable income inequality for the 33 EU and candidate countries is used. The SWIID database offers income inequality estimates that are based on thousands of reported Gini indices from a wider range of published sources. The data collected and harmonized based on Luxembourg income study as the standard (Solt 2019).
The worldwide governance indicators (WGI) reports on six broad dimensions of governance for 2016 and includes six dimensions: (i) government effectiveness, (ii) regulatory quality, (iii) rule of law, (iv) control of corruption, (v) voice and accountability and (vi) political stability and absence of violence/terrorism (Kaufmann et al. 2010) that capture the experiences and views of citizens, entrepreneurs, and experts on the quality of various aspects of governance. The aggregated index is constrained to include all six governance indicators that measure quality of formal governing institutions directly related to the economy (Brock 2020). Kaufmann et al. (2010) argue that these six dimensions of governance should not be thought of as being somehow independent of one another. These six composite measures of governance are strongly positively correlated across countries. In this study, an average of the six index scores is generated from this database to obtain a single variable for institutional quality. Table 7 in the Appendix provides detailed definitions of all the variables at the country (macro) level included in the analysis.

Strategy of Empirical Analysis
The scope of this study requires the use of an econometric model that can handle multi-level variables and that allows us to straightforwardly test the relative gains from small changes in specification (Amini and Douarin 2020;Taht et al. 2020). If the hierarchical nature of the data is not accounted for, standard errors for higher level variables might be underestimated; hence, the significance of the coefficients estimated for these variables would be overestimated (Amini and Douarin 2020). Consequently, to test our hypotheses, we employ a two-level linear mixed-effects model of individual responses (level 1) nested in 33 European countries (level 2), as this is meant to deal with a two-level dataset. This model is advantageous because the LR-test allows us to directly compare the variance explained by different nested models. This is a log-likelihood ratio test which tests whether the newly added variables are improving the fit of our model.
Based on the approach of Taht et al. (2020), linear models for the outcome variable at the individual-level are used, assuming that both outcome variables 4 are measured cardinally. An issue that is often discussed in SLfSat literature is the need to model in the estimation the ordinal nature of the dependent variable. All surveys questions on life satisfaction ask individuals to categorically evaluate their quality of life (i.e. respondents are ask to rank their life satisfaction on a scale of 1-5 or 0-10). Pasta (2009) argues that ordinal variables can be modelled with linear estimator with no significant loss of information. Moreover, it is very rare that a significant predictor for a categorical variable would not matter if that variable was measured on a continuous scale (Deaton 2008). Similar evidence shows that this should not be a problem since results from ordinal and linear specifications (OLS) tend to not differ substantially (Ferrer-i-Carbonell and Gërxhani 2004;Frey and Stutzer 2010) and this is more the case for the 11-point response scale.
By controlling for contextual factors at the country-level, a better comprehension can be obtained on whether and how effects on SLfSat are affected by material, subjective, and relative deprivation at the individual-level. The models presented here are random intercepts models, thus allowing us to account for systematic differences in the way respondents assess their SLfSat in each of the countries investigated (i.e. the intercept is country-specific). What is most important is controlling for a set of standard time-invariant individual factors that are notably important in explaining SLfSat. This large set of controls that are unlikely to change over time can serve as proxy for unobserved individual characteristics thus diminishing the effects of any potential endogeneity which would produce a spurious correlation between SLfSat and deprivation dimensions (Amini and Douarin 2020).
We start with a shell model with two fixed terms (constant and country-fixed effects) two random terms, and no explanatory variables, nor random intercepts coefficients. This way we estimate the share of variance explained at the country level and provide an additional justification for our approach.
The country fixed effects are statistically significant. Similarly, the interclass correlations (ICCs) for the model with just these fixed country effects are above 0.05 and ICCs values exceed 0.15 (which is considered large). This provides additional justification to apply the two-level (at the individual and country) approach. The two-level model below, Eq. 2, which we refer as the baseline model, includes only individual-level control variables.
Next, country level variables are introduced in model 1, as shown in Eq. 3. The two-level model is suitable in this case as the coefficients at the country-level of aggregation are of particular interest. The standard errors may be severely underestimated otherwise. (1) In model 2, a vector of the deprivation (material, subjective, and relative) variables at the individual-level is introduced, as shown in Eq. 4. Then, in model 3, a vector of the social mobility considerations variables (subjective perceptions about being 'left out', 'lack of employment recognition', 'feeling looked down by society', and 'feeling close to people in society') at the individuallevel is introduced in model 3, as shown in Eq. 5.
Finally, we further calibrate our estimation in model 4, by including an interaction term at the individual level between the Gini Index (at the city level) and the vector of mobility considerations variables as in Eq. 6.
In the above models, SLfSat ic reflects an individual i's subjective life satisfaction in country c. The fixed part of the model includes the following terms: a grand average (intercept) β 0 ; P individual-level variables X pci , such as gender, age, age square, employment status, and logarithmic of equalised HH income; q country-level variables, Z qc , including measures of, GDP per capita, income inequality, and institutional quality. Our individual-level variables of interest r and s represent deprivation and social mobility considerations, respectively. In Eq. 6, we have extracted relative deprivation income inequality, Gini r , with reference groups at the city level. In addition, it includes cross-level interactions Gini r × Mobility sci to assess whether and how individual-level effects (i.e. slopes) of social mobility considerations are moderated by relative deprivation indicator Gini r with city level reference groups.
The random part consists of the two error terms u c (country level) and e ci (individual-level).

Results
This study consists of a comparative analysis of data collected at the individual (micro) level-the 2016 wave of European Quality of Life Survey-and at the country (macro) level-GDP growth rate from the World Bank national accounts data, the 2016 Gini index of disposable income inequality from the Standardized World (4) s Mobility sci + e ci + u c Income Inequality Database, and a composite index of institutional quality aspects comprised by the Worldwide Governance Indicators. We test our hypotheses using a two-level linear mixed-effects model with individual responses nested in 33 European countries (28 EU plus 5 non-EU countries-Albania, Montenegro, North Macedonia, Serbia, and Turkey).
We first estimate a shell-model, which includes no explanatory variables beyond the country fixed effects, nor random intercepts coefficients. This allows us to identify the share of variance explained at the country level and provide an additional justification for our approach. In this model, the constant has a coefficient of 6.03 and a standard error of 0.202. The variance of the constant is 0.88. The interclass correlation (ICC) is 0.19, which validates our use of the two-level model, instead of just using a simple regression. This modelling approach fits well with the scope of the paper that seeks to analyse the country-level coefficients before moving to the individual-level of analysis (H1 and H2). If the hierarchical nature of the data is disregarded, standard errors may be severely underestimated.
Tables 1, 2, 3, 4, and 5 show the results of our five models. Estimates are generated for the pooled sample and separately for the non-transition (West-EU) and transition post-communist (East-EU and non-EU) countries. It is important to note that for sensitivity analysis purposes, non-reported versions of the baseline and model 1-4 were estimated with the dependent variable being happiness and the WHO mental wellbeing index (measured as a continuous variable from 0 to 10), but the results were similar with the SLfSat variable.

Individual Level Variables
The baseline model includes the full list of individual controls as in Eq. 1. The main variable of interest in this model is residence in a non-transition (West-EU) or transition (East-EU and Non-EU: Albania, Montenegro, North Macedonia, and Serbia) country as compared to Turkey, the country of reference. We can notice that nontransition countries report higher levels of SLfSat with respect to their transition comparatives. This effect is statistically significant at 10% level of significance.

Country Level Variables
This baseline model shows that the effects of the control variables are consistent, both in sign and size, with previous empirical work: age has U-shaped effect, bad health and being a male has negative effects, while education and income have positive effects with respect to SLfSat (Amini and Douarin 2020;Djankov et al. 2016). These relationships between the controls and the SLfSat were robust throughout the following specifications. As the focus of this study is on the main and interaction effects of deprivation aspects and mobility considerations on SLfSat, the above individual-level variables were treated only as controls in the following models.
To test the robustness of the patterns identified in the baseline model, we move to model 1 which adds country-level variables-GDP per capita growth rate, the Gini index of disposable income, and the institutional quality index-in Table 2 below.
In column (1), we start with the comparison of 'non-transition' (West-EU) with 'transition' (East-EU) countries (thus excluding non-EU countries from the sample). Column (2) replicates this specification but including also the four Western Balkans EU candidate countries (Albania, Montenegro, North Macedonia, and Serbia) in the reference category of the dummy variable for whether the respondent comes from a non-transition economy. This way, the non-transition countries are compared with all the transition countries (East-EU and Non-EU) of the sample. Column (3) includes Turkey in the reference category, so the 'non-transition' (West-EU) cluster is compared with the rest of all the other countries in the sample. The commonalities for these three specifications of the model 1 for the pooled sample relate, on one hand, to the positive and insignificant effect of the

Individual level variables
Age ( *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively  (6)         'non-transition' dummy variable, and on the other hand, on the positive and significant effect of the 'institutional quality index' coefficient. To further explore the mechanism behind the commonalities identified in the first three specifications, in column (4) a fourth specification of model 1 is used in which Turkey is the reference country/category. This way, the other two categories, namely 'non-transition' (West-EU) and 'transition' (East-EU and Non-EU), are compared to Turkey. In this fourth specification, the size of the 'non-transition' coefficient is improved but not the significance and it is comparable in both aspects with the 'transition' coefficient, therefore suggesting a disappearance of the 'post-communist gap' which was present in the baseline model above.
Previous studies reported similar findings. Using data from wave 3 (1994-1999) and 4 (1999-2003) of the World Values Survey, Gurieve and Zhuravska (2009) showed that the gap between transition and non-transition countries 5 was 1.40 and 1.13 points with respect to SLfSat (measured on a 0-10 scale), after controlling for individual-and country-level variables. This converging pattern of SLfSat between transition and non-transition European countries was further validated by Djankov et al. (2016) for three broad time periods, 1990-2000; 2001-2007; and 2008-2014, and across several surveys with SLfSat being measured as a binary variable (0 'not satisfied' and 1 'satisfied). They report that transition countries are on average 10.4 percentage points less satisfied than their non-transition counterparts. By 2016, Guriev and Melnikov (2018), using wave 3 (2015-16) and 4 (2010-2016) of the Life in Transition Survey, report that the 'post-communist happiness gap' had closed and due to the substantial increase in SLfSat in most transition, mainly driven by middle-income young, educated individuals, regardless of gender, complement with the decrease in SLfSat in non-transition countries between 2010 and 2016.
Across all the specification, (1)-(4) in Table 2, the inclusion the 'institutional quality index' renders the non-transition (West-EU) dummy insignificant as compared to the baseline model while its own coefficient is significant and positive. Moreover, in the specification in column (4), the size effect is improved compared to the previous specification (1)-(3) and is the only country-level variable with a significant positive effect. Namely, the higher the level of institutional quality in terms of government effectiveness, and regulatory quality, rule of law, control of corruption, voice and accountability and political stability/absence of violence in the country, the higher the reported levels of SLfSat of European citizens. The significance of the effects of institutional quality on SLfSat remains even after controlling for the wealth and inequality of a nation, namely GDP per . Gini disposable income index at country level *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively capita growth and Gini index at the country level. This is a rather strong test, since institutional quality is an important determinant of economic and social success (Veenhoven 2015). The macro-level income inequality measure-the Gini index at the country level-results showed that it related negatively and insignificantly to SLfSat. The only exception is for the non-transition subsample specification in column (5) of Table 2. This might be explained by the fact that in some non-transition countries, improved levels of economic prosperity (GDP per capita economic growth) are associated with increasing income inequalities (Mikucka et al. 2017).
Institutional quality is important in explaining SLfSat (Berggren and Bjørnskov 2020;Shiroka-Pula et al. 2022;Veenhoven 2015). Nikolova (2016) argues that institutional quality (as measured by the rule of law) affects the overall SLfSat differential between the non-transition and transition societies. It played an additional role in reducing the 'post-communist happiness gap' in the 1990s and may have even reversed it in recent years. Similarly, recent empirical evidence shows that with the increase in quality of public services, transition countries converged to their nontransition comparators in terms of SLfSat (Guriev and Melnikov 2018). The adverse effect of institutional quality is significant and comparable for both transition and non-transition countries. Based on with these findings, we conclude that citizens in transition and non-transition countries are unhappy not by nature but because they suffer from poor institutional quality.
It is worth specifying that when moving from the baseline model (Table 1) in model 1, we are testing whether the baseline model is nested in the second richer one, model 1. The result of the log-likelihood ratio test that testes such hypothesis is LR Chi-Square test, which is significant for each of the versions of model 1 in Table 2, therefore confirming that these new variables are improving the fit of our model. In line with previous findings, the results in Table 2 suggest that in 2016 the post-communist happiness gap had closed, and it was driven by improvements in institutional quality perceptions. The baseline model results for the pooled sample indicate differences among transition and non-transition countries in terms of SLf-Sat. However, after controlling for per capita income and institutional quality index at the country level in model 1, we find no evidence of significant diverging effects between the non-transition and transition clusters. In other words, model 1 shows that after adjusting for a variety of country-level and individual-level variables, the citizens of transition countries no longer express significantly lower degrees of SLf-Sat. Overall, our analysis suggests that, all other things being equal, there are no differences in SLfSat between non-transition (West-EU) and transition (post-communist: East-EU and Non-EU) countries in Europe.

Deprivation Dimensions
We introduce the deprivation dimension in model 2, and the results are reported in Table 3. In line with the previous evidence (Bellani and D'Ambrosio 2011;Cheung and Chou 2019), material deprivation, subjective financial hardship and relative deprivation have significant negative effects on SLfSat across the pooled, non-transition, and transition samples, after controlling for income. This means that more materially, financially, and relatively deprived individuals reported lower levels of SLfSat. These effects are larger for the former communist countries which is also documented by Guriev and Zhuravskaya (2009), suggesting that while income per se matters, there are other income-related latent aspects that explain SLfSat differentials.
To be able to discuss about the size effects, we refer to the standardized estimates as these allow us to determine which of three deprivation items coefficients have the largest relative impact on the SLfSat (Hunter and Hamilton 2002). Namely, we use the standardized values for the 'Gini index at city level' and 'items not afforded' 6 while for the 'making ends meet' variable we have proceeded by keeping the raw values since it takes only the 0, 1 values (binary variable). In line with previous findings (Angel et al. 2003;Arber et al. 2014;Christoph 2010;D'Ambrosio and Frick 2007;Fusco et al. 2011;Pepper and Nettle 2017), the comparative effects of relative deprivation (relative income measured by the Gini Index at the city level, i.e. social comparison/status effects) on SLfSat are larger than the effects of subjective (perceived) and material deprivation. The role of relative deprivation in explaining SLfSat is also confirmed when 2016 country level Gini index from the Standardized World Income Inequality Database (SWIID) are included in the model. The results show that the estimated coefficients for Gini Index with reference groups at the city level are negative and significant and capture the largest level of variance in terms of SLfSat. Namely, the higher the level of inequality in terms of income distribution in the city/local level, the lower the reported levels of SLfSat, and this effect is larger compared to material and financial deprivation. For the pooled sample, an increase in relative deprivation by 1 standard deviation lowers SLfSat by 0.711 (approximately 1) units on the 11-point scale ranging from (0) = not at all satisfied to (10) = completely satisfied. While for the non-transition (West-EU) countries, an increase in relative deprivation by 1 standard deviation lowers SLfSat by 0.683 compared to 0.946 for the transition countries (East-EU plus Non-EU).
In Latin America, the effect of relative status is found to be strongest at the city level rather than the country level (Graham and Felton 2006). One explanation for this relates to the fact that relative deprivation is about 'not being like others' to reference groups at the city level while material and subjective deprivation are more related to the lack of resources necessary for the fulfilment of basic physical needs, they deal more with the survival perspective.
Material deprivation captures negative SLfSat effects in terms of cumulative diverging patterns between the necessary and the available resources to fulfil basic needs. Subjective deprivation on the other hand is related to the subjective perception of systematic erratic incomes that make it difficult to make ends meet. Both these aspects of deprivation are more internally oriented in the meaning that they tend to have lower explanatory power because of the 'adaptation effects' underlying the relationship. As anticipated in H1, the largest size effect is reported for relative deprivation, which is explained by the fact that persistent income poverty in comparison with city level reference groups produces long-term deficiencies in one's social status and lifestyle, which relates to the so-called social comparison effects. The mechanism that explains this relates to the 'information/tunnel' effect related to a wider and more heterogeneous comparative spectrum with whom individuals compares themselves. Further, consistent with the arguments and findings of Ayala et al. (2011), Cojocaru (2014 and Alesina et al. (2004), the results in model 2 show that the Gini index at city level, which measures relative deprivation, has the largest negative significant effect on SLfSat and this effect is stronger for the transition sample. This is related to the presence of a social ranking/ladder in the society, and it induces negative feelings in SLfSat, in particular for those on the lower income segments. Consequently, we conclude that H1 is supported.

Social Mobility Considerations and Interaction Effects
In model 3 and 4, we further calibrate our model by including main and interaction effects between social mobility considerations and relative income deprivation, which are summarized in Tables 4 and 5, respectively. In model 3, four measures of mobility considerations are introduced. These include four items related to income and occupation: (i) 'I feel left out of society', (ii) 'I feel that the value of what I do is not recognized by others', (iii) 'Some people look down on me because of my job situation or income', and (iv) 'I feel close to people in the area where I live'.
The results suggests that the sign of the first three items is negative while the sign of the last item is positive. The significance of 'being looked down due to the job situation and/or income' is significant across the pooled and subsamples. This is very important as it indicates that it is hurtful in terms of SLfSat and that individuals report subjective differences in social status between self and others and feelings about being left out and looked down result. This effect is more pronounced in nontransition countries, and this might relate to the fact that improved economic growth in more affluent economies might bring about higher level of inequality and social ranking, which in turn reduce SLfSat.
Feeling close to people has positive and significant effects across all three specifications, and this effect is larger for non-transition counties. Feeling left out of society, on the other hand, is significant only for transition countries. Employment recognition is significant for the transition countries but not for the non-transition part. This confirms the rigid class structure of a society which is indicates lower incidence of social mobility, that is, moving into a different class than that into which one was born (Simandan 2018).
We further elaborate our identification in model 4 by including interaction terms that capture the moderating effects of social mobility considerations on relative deprivation, which are then reflected in SLfSat. Again, both non-standardized and standardized coefficients indicate larger comparative negative effects of the relative deprivation dimension 7 . The results of model 4 are summarized in Table 5. These results indicate that when individuals perceive that what they do for living (employment) is not recognized and that they are looked down by society because of their job situation or income, then there are supplementary adverse effects in terms of SLfSat for those who are relatively deprived (meaning in lower income segments).
This means that both these mobility considerations have direct, as well as an indirect, significant, and negative effects SLfSat. Similarly, feeling left out of society has negative effects on SLfSat except for the transition countries subsamples. In this case, when accounting for the moderating effect, we did not observe neither a direct, nor a direct effect. However, in model 3 where interaction terms were not included, we observe a direct negative effect for this subsample indicating a direct negative effect of this measure. Finally, feeling close to people in the area where the individual lives have no indirect or direct effect on SLfSat.
The effects of these interactions hold instrumental information as they encapsulate lack of optimism about future economic mobility which lowers current SLfSat. This line of thought is consistent with the 'tunnel effect' theory (Hirschman and Rothschild 1973) which includes a dynamic and inter-temporal aspect about aspirations for future mobility. Similarly, relative deprivation theory (Runciman 1966) suggests that the position in relative lower income segments (worse off compared to reference group at the city level) is accompanied by feelings of anger and resentment about being (economically and professionally) worse off than others in the long run. This means that relative deprivation, which positions the individual in relation to the local inter-group levels of analysis, is further connected with social and psychological concerns that produce a subjective state that shapes emotions, cognitions, and beliefs related to social upward mobility, which altogether connect well to SLfSat. These patterns are more pronounced for the transition countries and altogether, these findings provide support for H2.

Discussion and Conclusion
This study sought to provide a comparative analysis between non-transition and transition European countries in terms of the subjective life satisfaction (SLfSat) implications of material, subjective, and relative deprivation while adjusting for income. Our findings suggest that different dimensions of deprivation, i.e. material deprivation, subjective financial hardship (difficulty in making ends meet) and relative deprivation (Gini index at local/city level) have significant negative effects on SLfSat, for both transition and non-transition countries. Those who did not detect potentials of social mobility from this inequality were more prone to face additional adverse effects on SLfSat from the relative deprivation with city level reference groups. Therefore, social mobility considerations interact with relative deprivation to influence SLfSat which in turn represents a crucial aspect of how fair a society is. In principle, the fairer the inequality in a city, the easier the social upward mobility should be.
However, all three dimensions of deprivation record larger negative effects on SLfSat for transition as compared to non-transition countries, which is well supported by previous evidence (Guriev and Melnikov 2018;Guriev and Zhuravskaya 2009;Zagórski 2011). This could potentially be explained by the distinctive nature of the value system which dominates the non-transition (West-EU) and transition post-communist (East-EU and non-EU) cultures. While the former is materialist and individualist (prevalence of self-achievement and merit values), the latter is more inclined towards collectivism and post-materialism (Amini et al. 2022). In individualistic countries, where values of self-achievement and meritocracy prevail, the relative inequality might yet be accepted as a natural outcome of individual differences and motivations (Fevre 2016). In these countries, people's self-direction and self-determination influence in the future the interventionist preferences (Pitlik and Rode 2017). Culture, being an important determinant of economic expectations, explains on the other hand, how in transition countries, people consider inequality as distorting factor in social order and as obstacle to social mobility. In these countries, despite the appraise for interventionist policies, the weak trust towards institutions, make people being loyal to group cohesion and redistribution and therefore, more reluctant to accept inequalities (Pitlik and Rode 2017).
These results corroborate the prepositions of the resource-based perspective on the importance of resources in relieving social status mobility and lifestyle considerations adversities that go beyond income poverty. The prominent role of relative deprivation among the other dimension of deprivation as a predictor of SLf-Sat validates the tenets of Festinger's (1954) social comparison theory and the arguments of many scholars (e.g. Yitzhaki 1979;Whelan et al. 2001;Whelan and Maître 2010) suggesting that one's perception about the adequacy of their income is often comparative in nature with reference groups being primarily local rather than national (Alesina et al. 2004;Ayala et al. 2011;Cojocaru 2014;Graham and Felton 2006). Consequently, this effect is more revealed for the less affluent transition compared to non-transition countries. Therefore, the inclusion of this more detailed measure of deprivation improves the ability to identify the drivers of SlfSat.
The results also provide empirical support for the 'tunnel/information effect' theory proposed by Hirschman and Rothschild (1973) which suggests that current deprivation becomes more stressful (in terms of SlfSat) if potentials for upward social mobility in the future are not perceptible. In other words, the negative effects of mobility consideration, such being left out, experiencing lack of employment recognition, and being looked down by society are amplified for those individuals in cities with higher rates of inequality, for the pooled, not-transition, and transition countries subsample. Therefore, it can be argued that while higher inequality rates at the local area and not seeing any social mobility potentials have adverse effects on one's life satisfaction, there are additional penalties generated by the interaction of the two: there is no light in the end of the tunnel.
These findings have strong implications for the understanding of the factors that shape the SlfSat. First, our results suggest that the inclusion of relative deprivation as an objective predictor of SlfSat can be beneficial to the use of multidimensional approach in measuring deprivation, often criticized for the reliability and face validity of some of the subjective measures. Second, the inclusion of social mobility considerations and especially, their interaction with relative deprivation as predictors of SlfSat shows how lack of multi-period (present and expected) social mobility potentials hampers people's SlfSat, and this effect is stronger in cities where inequality rates are more pronounced.
Our results have several important implications for policymakers, for both transition and non-transition countries. First, while addressing the disposable income inequality remains a significant priority to address people's dissatisfaction, particularly for transition countries, perceptions on institutional quality have contributed towards the closure of the happiness gap between transition and non-transition countries. Second, multi-dimensional measures of deprivation, including relative deprivation, offer greater detail about the range of instruments and policies that can be adopted to tackle inequality-related problems. For example, the prominent importance of relative deprivation calls for social support policies to address inequality at city or municipality-level; tackling inequality should not be considered the domain of the central government, as it is often the case for transition countries. Considering the countries institutional settings, the decentralization reforms for strengthening the role of the local government should be accompanied with regulatory interventions for raising the trust towards interventionist policies. Finally, the effects of social mobility considerations should not be neglected since they further amplify the negative effects of relative local/city deprivation on life satisfaction. Degree of urbanization Population-density measure: 1 'thinly populated' or 'rural areas' (less than 10,000 citizens); 2 'intermediate density areas', or 'towns and suburbs' (10,001-200,000 citizens); and 3 'densely populated' or 'cities' (more than 200,000) citizens and (Eurofound 2015)  GDP per capita, rate of growth Annual percentage growth rate of GDP per capita based on constant local currency. Aggregates are based on constant 2010 U.S. dollars. GDP per capita is gross domestic product divided by midyear population. GDP at purchaser's prices is the sum of gross value added by all resident producers in the economy plus any product taxes and minus any subsidies not included in the value of the products. It is calculated without making deductions for depreciation of fabricated assets or for depletion and degradation of natural resources.

1.30
Institutional quality index The Worldwide Governance Indicators (WGI) report on six broad dimensions of governance for 2016 includes six dimensions: government effectiveness, and regulatory quality, rule of law, control of corruption, voice and accountability and political stability/absence of violence. We average the index scores of only the six governance indicators (Kaufmann et al. 2010