The Impact of Wealth on Subjective Well-Being: A Comparison of Three Welfare-State Regimes

This study provides new insight on subjective well-being (hereafter SWB) and its associa­ tion with individuals’ objective economic standing. In particular, we are interested in how one’s relative position in the distribution of wealth influences his or her general satisfaction with life (hereafter GLS), representing the cognitive and most stable dimen­ sion of SWB. Most studies on the relationship between economic standing and SWB have used income as an indicator of economic standing. Yet, income seems to account for only a small part of the variation in SWB (e.g. Diener, Sandvik, Seidlitz & Diener 1993). In addition, income is restricted to a certain time interval (income per week, per month, or per year) and, to periods of labor market activity. Recent studies thus argue that other measures of economic standing such as socioeconomic status, deprivation, and wealth, might be more useful for understanding its relationship with SWB (e.g. Christoph 2010). Wealth is a stock Graph accumulated throughout a person’s life course. Additionally, in contrast to earned income, which requires time, effort, and working ability, wealth offers access to capital and goods independent of individual investment and ability, for example, through intergenerational transfers (Elmelech 2008). Wealth may also be a better indicator of an individual‘s long-term consumption potential (Spilerman 2000). Considering the unique properties of wealth, we see it fit to measure the consequences of economic standing to SWB, over and above the consequences income may have on it. A second contribution of this study is the focus it places on macro-level factors, and specifically, state-level welfare systems. The relevance of the welfare-state system to the relation between wealth and SWB is best explained through the concept of decom­ modification, referring to the extent to which citizens in a country are economically independent from the market through the provision of social benefits. These benefits can be understood as a cushion against the consequences of shortage of financial resources (Pacek & Radcliff 2008). Because the extent to which these benefits are provided by the welfare-state strongly differs between

The Impact of Wealth on Subjective Well-Being:

A Comparison of Three Welfare-State Regimes1
This study provides new insight on subjective well-being (hereafter SWB) and its associa tion with individuals' objective economic standing.In particular, we are interested in how one's relative position in the distribution o f wealth influences his or her general satisfaction w ith life (hereafter GLS), representing the cognitive and m ost stable dim en sion o f SWB.M ost studies on the relationship between economic standing and SW B have used income as an indicator o f economic standing.Yet, income seems to account fo r only a small part o f the variation in SW B (e.g.Diener, Sandvik, Seidlitz & Diener 1993).In addition, income is restricted to a certain time interval (income per week, per month, or per year) and, to periods o f labor m arket activity.Recent studies thus argue that other measures o f economic standing such as socioeconomic status, deprivation, and wealth, m ight be more useful fo r understanding its relationship w ith SW B (e.g. Christoph 2010).Wealth is a stock Graph accumulated throughout a person's life course.Additionally, in contrast to earned income, which requires time, effort, and working ability, wealth offers access to capital and goods independent o f individual investm ent and ability, fo r example, through intergenerational transfers (Elmelech 2008).Wealth m ay also be a better indicator o f an individual's long-term consumption potential (Spilerman 2000).Considering the unique properties o f wealth, we see it f i t to measure the consequences o f economic standing to SWB, over and above the consequences income m ay have on it.
A second contribution o f this study is the fo cu s it places on macro-level factors, and specifically, state-level welfare systems.The relevance o f the welfare-state system to the relation between wealth and SW B is best explained through the concept o f decom modification, referring to the extent to which citizens in a country are economically independent from the market through the provision o f social benefits.These benefits can be understood as a cushion against the consequences o f shortage o f financial resources (Pacek & R adcliff 2008).Because the extent to which these benefits are provided by the welfare-state strongly differs between the three regimes, we predict that the association between wealth and SW B will also differ.

Linking w ealth to SWB: Needs theory
Needs theory defines the relations betw een econom ic standing and SWB th ro u g h the function of econom ic standing and mostly incom e, in p roviding in d iv id u als' needs (Diener & B isw as-D iener 2002).In g en eral, needs theory assum es th a t individual income, as a principal indicator of economic standing, augm ents individual SWB prim ar ily because incom e enables people to better provide for their needs.2The m ain premise o f needs theory is that low incom e implies a disadvantage in SWB w hile high incom e leads only to a sm all advantage (Maslow 1943, 1954), if any (Veenhoven & Ouweneel 1995;V eenhoven 1993).Im portantly, in this study we move aw ay from the typi cal understanding of econom ic standing as represented by incom e, an d in v estig ate a different form o f econom ic w ell-being namely, w ealth.
In order to test for the em pirical v alidity o f the needs-based m echanism our models n ot only control for the position o f the respondents in the w ealth distribution.They additionally include a subjective m easure for econom ic hardship -th a t is, individuals' self-reported responses about having "prob lems to make ends m eet."If the relevant m echanism behind the relation of w ealth w ith SWB is the fulfillm ent o f basic needs, then we w ould expect subjective economic hardship to m ediate the supposed negative effect o f being poor on SWB.

The comparative setting: Do different contexts imply differences in the wealth-SW B relation?
One im portant aim o f our study is to inves tigate the contribution of the w elfare-state system, operationalized through the concept o f decom m odification, to the wealth-SW B relation.D ecom m odification, determ ines w hether personal w ealth is necessary for individuals in order to m aintain their SWB or not, an d w h a t is th e level of w ealth required to do so.Given our focus on the relation betw een w ealth an d SWB am ong aging individuals w ho are either close to retirem ent or already retired, w e d em on strate the w orkings o f decom m odification referring to state pension system s.State pension system s intervene in individuals' saving behavior by taking over the task of old-age provision through m andatory sav ing.Moreover, the level of m inim um p en sion benefits provided is likely to affect the intensity o f the w ealth SWB relation am ong those individuals w ho did not have the pos sibility to choose how m uch they are willing to save namely, the poor.
In order to p u t these pred ictions to an empirical test w e com pare three countries, each representing a different w elfare-state system com plying w ith a different regime type, specified in E sping-A ndersen's typol o gy (1990,1999).The so cial-dem ocratic w elfare-state represented in our study by Sweden, secures its citizens the highest level of decom m odification, w ith relatively generous public pensions (OECD 2011a).The relative m inim um pension benefits as percentage o f average earnings have been at 25% in 2008 (OECD, 2011a).In Sweden therefore, persons are less d ependent on their incom e for old-age provision.Under such conditions, the wealth-SW B relation is predicted to be com paratively weak.In the liberal model, represented in this study by Israel, the general decom m odification level is low an d accordingly, public pensions are relatively prudent.O ld-age allow ance in Israel, w hich is u niversally distributed, is insufficient even for a m inim al living standard (Dagan-Busaglo 2007), w ith m ini m um pension benefits at a level of 13% of average earnings in 2008 (OECD 2011b).Under these conditions w ealth is expected to be a necessary instrum ent to secure SWB in old age and the wealth-SW B relation is therefore expected to be rather strong.In the conservative w elfare-state, represented in our study b y Germany, the level of decom m odification is interm ediate.The German system does not have a general m inim um retirem ent pension.However, individuals w ith o u t m an d ato ry and personal savings are eligible for basic social security in retirem ent, w hich equals the standard rate o f the basic support for em ploym ent seekers (Hartz 4) th at w as 21% o f average earnings in 2008 (OECD 2011b).The conservative m odel thus suggests a relation of medium stren g th betw een w ealth and SWB.From our th eoretical considerations, w e derive the follow ing hypotheses: H1: Incom e and w ealth together account for the varian ce in GLS better th an does incom e alone.
H2: Individuals of poor w ealth have lower GLS th an do those in the middle of the w ealth distribution.The w ealthy group has only slightly higher, or the same, GLS th an does the middle group.
H3: The negative effect o f poor w ealth on GLS is strongest in Israel.It w ill be w eaker in Germ any, an d sim ilar or slightly w eaker in Sweden.
H4: The negative effect o f poor w ealth on GLS is m ediated by individuals' sub jective feelings about their respective econom ic hardship.

Data, Variables, Hypotheses & Methods
For o u r em p irical an a ly se s w e m ake use of th e Survey of H ealth, A geing and R etirem ent in Europe (SHARE) th at is an in tern atio n al, represen tativ e panel study o f the population aged 50 years and over.
Observing individuals aged 50 or more, we analyse the consequences o f differential w ealth levels, m easured as household gross w ealth on their GLS.In order to account for the socio-econom ic and socio-dem ographic factors th a t affect GLS, w e apply a linear regression m odel (OLS regression) w ith th e H uber-W hite Sandw ich estim ator for cluster sam pling (individuals clustered in households) th at provides robust standard error estim ates.The analyses are carried out separately for each of the three countries.
G eneral life satisfactio n w as m easured using an 11-point single item scale w hich w e standardized to have a m ean o f zero and a standard deviation o f one to enable international com parison.3Consequentially, a one-u n it change in one of the indepen dent variables results in a change in GLS of one standard deviation.Incom e and w ealth are m easured on the household level and presented in Euro, adjusted for purchasing pow er parity.We operationalize incom e as to tal net an n u al incom e, divided b y the root o f the num ber o f persons living in a household.W ealth was measured by positive (gross) w ealth, while controlling for house hold debt (negative wealth).4 Follow ing our theoretical considerations, the effect o f w ealth on SWB is expected to differ depending on an individual's, respec tively a household's position in a country's w ealth distribution.We account for three m ain positions: The tw o m iddle quartiles represent the middle category (hereinafter, "m iddle w ealth " or the "m iddle g ro u p "), w hich serves as the reference category.
Those in the highest quartile are hereinafter referred to as the "w ealth y " or as being of "high wealth."Those in the lowest quartile are hereinafter referred to as the "poor" or as "being of poor w ealth".
Subjective economic hardship is captured by respondents' self-evaluation o f their ability to "make ends m eet" in th eir household.
Originally, responses range from 1 ("w ith great difficulty") to 4 ("easily").We combined the responses o f 1 and 2 to create a dum m y variable representing people w ith economic hardship.Following previous research (Die ner, Suh, Lucas, & Smith 1999), we control in our models for th e respondents' labor m arket outcom es, m easured th ro u g h their educational attainm ent, their labor market status and their household income.We also control for fam ily characteristics (m arital status and children); im m igrant status (in Israel, we also control for Arab origin due to the u n ique position o f this m inority in the Israeli stratification system); and health.

Descriptive results
G  the poor reporting lower GLS th an house holds in the middle of the w ealth distribu tion and the rich reporting higher GLS.Debt is significantly associated w ith SWB only am ong the Israeli respondents, w here debt is found to slightly decrease their GLS.The R2 of Models 1 and 2 indicate th at incom e and w ealth, w hen taken together, explain a greater part of the variance in SWB th an does incom e alone.The increase in R2 is significant at the one per cent level in Ger m any and at the five per cent level in Israel (F-test).H ypothesis 1 is thus supported by our analysis fo r Germany and Israel.
In M odel 2 w e further test hypothesis 2, w hich follow ing the needs theory predicts th a t the gap in SWB betw een the middle group and the poor w ill be g reater th a n the gap betw een the middle group and the w ealthy.The regressio n m odel conveys th at in both G erm any and Israel, the poor penalty and the w ealth prem ium are similar in m agnitude.In Sweden, the GLS o f both the poor and the w ealth y are not signifi cantly different from the GLS of their middle w ealth counterparts.H ypothesis 2 is thus not supported by the SH ARE data.
H ypothesis 3, proposes th a t the negative effect of being of poor w ealth on GLS will be strongest in Israel, and w eaker in Ger m any and Sweden.The findings in Model 2 indicate th a t in Sweden individuals of poor w ealth do not differ significantly in their GLS com pared w ith individuals in the middle of the w ealth distribution.Regard ing the size of the poor effects on GLS in G erm any and Israel, interaction based coef ficients provide some evidence th a t being poor is more detrim ental for GLS in Israel th an in Germany.However, this difference is not statistically significant.H ypothesis 3 is thus not supported by the data.M odel 3 finally tests hypothesis 4, w hich suggests th a t the effect o f poor w ealth on GLS w ill be m ediated by an individual's subjective sense o f econom ic hardship.The findings presented in Model 3 indicate that respondents w ho report h av in g problem s m aking ends m eet are significantly less sat isfied w ith their life com pared to those who do not have problems.Furtherm ore, Model 3 dem onstrates th a t th e poor p en alty we found in G erm any and in Israel dim inishes after this subjective m easure o f econom ic hardship is introduced.In Germ any, the rem aining effect is no longer statistically significant im plying full m ediation, but in Israel it rem ains significant, im plying only partial m ediation.Therefore, hypothesis 4 is supported by the data fo r Germany, but only partially supported by the data fo r Israel.

Conclusions
Needs theory claims th at w ealth augm ents SWB by enabling a person to better provide for his or her basic needs, econom ic and/ or recreational.This pro p o sitio n im plies th at the poor will most likely enjoy a lower level of SWB compared to the middle w ealth group, while the w ealthy will differ from the middle w ealth group only slightly, if at all.Our findings show that net of income, gross household w ealth has a significant im pact on SWB, m easured as GLS, in G erm any and Israel.More specifically, we find a poor p en alty and a rich prem ium on GLS in these two countries.A gainst the predictions o f needs theory the two coefficients do not differ in magnitude, suggesting th at w ealth can buy happiness.In Sweden, neither the poor nor the w ealthy differ significantly in their SWB from the middle w ealth group.
Our study contributes to the contem porary literature on SWB by dem onstrating that: 1) the SWB of individuals aged 50 or older, is strongly associated w ith w ealth, repre senting an im portant source for econom ic standing, in G erm any and Israel, yet, not in Sweden; 2) needs are th e p redom inant m echanism linking econom ic stan d in g to SWB in Germany, and th ey also partially explain the w ealth -SWB relations in Israel; and 3) the degree of social support provided by the state has an im pact on the m agni tude of the association betw een w ealth and SWB.Our study m ight thus serve as a fertile ground for the on-going debate on w hether an d how w elfare policy in stru m en ts can im prove a population's SWB.

Abstieg zumeist nur um ein Einkommensquintil
Der ökonom ische A bstieg eines H aushalts k ann als Verlust einer vom M enschen w ert b a s e :S H A R E W ave 2, re le a se 2.5.0, o w n c a lc u la tio n s ; D O I: 10 .6 103/S H A R E .w2.500 .d ia n W e a lth « M e a n W e a lth D a ta b a s e :S H A R E W ave 2, re le a se 2.5.0, o w n c a lc u la tio n s ; D O I: 10 .6 103/S H A R E .w2.500 .
d ia n D e b ts ■ M e a n D e b ts D a ta b a s e :S H A R E W ave 2, re le a se 2.5.0, o w n c a lc u la tio n s : D O I: l0 .6 l0 3 /S H A R E .w 2 b a s e :S H A R E W ave 2, re le a se 2.5.0, o w n c a lc u la tio n s ; D O I: 10 .6 103/S H A R E .w2.500 .

W
e co n tro lle d fo r gender, household size, age, m ig ra n t status, A rab orig in (IL), m arried, child, e du ca tio n , em plo ye d , une m p loyed , h o m e m a ke r and health.All analyses based on 5 sets o f im p u ta tio n s (using th e 'm im '-p re fix co m m a n d in Stata 12).See H o ch m an & Skopek (2013) fo r fu ll m odels and details.*p<0.05,**p<0.01,***p<0.001.D a ta b a s e : SH AR E W ave 2, re le a se 2.5.0, o w n c a lc u la tio n s , d a ta u n w e ig h te d .56 44 Die A nalyse erfo lg t m it den D aten des Sozio-oekonom ischen Panels (SOEP), einer seit 1984 jä h rlic h durch g efü h rten rep rä sentativen Panelstudie, die sich besonders a u f die A nalyse v o n E inkom m ens-u n d B erufsverläufen k o n z e n trie rt1.Die letzte zum Zeitpunkt der A nalyse zur Verfügung stehende Welle des SOEP ist aus dem Jahr 2013, so dass Daten aus insgesam t 30 Befra gungsw ellen u n tersu ch t w erden können.U nter 20-jährige B efragte w urden aus der A nalyse ausgeschlossen, da P ersonen in dieser A ltersgruppe häu fig noch bei den Eltern w ohnen u n d keine klaren A ussagen zum ökonom ischen Status getroffen w er den können.Das erste B efragungsjahr jedes B efragten w urde ebenfalls ausgeschlossen, da die E inkom m ensangaben hier häufig u n g en au sind (Frick et al. 2006).Zudem w erden Befragte mit m axim al einem Befra gungszeitpunkt (nach Exklusion der Erstbe fragung) von der A nalyse ausgeschlossen, da in diesem Fall keine Einkom m ensm obi lität festgestellt w erden kann.Die Lebenszufriedenheit wird über die Frage gemessen: "Wie zufrieden sind Sie gegen w ärtig, alles in allem, m it Ihrem Leben?".Beim Einkom m en w ird a u f das jährliche, inflationsbereinigte u n d bedarfsgew ichtete H au sh altsn etto ein k o m m en zu rü ck g eg rif fen 2. Das E inkom m en aller B efragten zu jedem der Befragungsjahre w ird in Quintile eingeteilt u n d je d e r Befragte w ird einem der Quintile zugeordnet.Der V organg der Einkom m ensm obilität w ird somit über den Positionsw echsel der B efragten zw ischen den Quintilen in konsekutiven B efragungs jah ren gemessen3.Die Wahl dieses Einkom m en sm obilitätskonzeptes b eg rü n d et sich insbesondere darin, dass die relative Ein kom m ensposition jedes Einzelnen im Falle gleichverteilter Einkom m ensveränderungen aller invariant bleibt.Dabei beschreibt das erste Q uintil das u n terste E inkom m ensquintil u n d das fünfte Quintil das oberste Einkom m ensquintil.Ein A bstieg zw ischen den Einkom m ensquintilen w ird folglich als A bw ärtsm obilität interpretiert.