Subjective Well‐Being and Income: Is There Any Evidence of Satiation?*
Betsey Stevenson
The Gerald R. Ford School of Public Policy, University of Michigan & CESifo and NBER betseys@umich.edu www.nber.org/~bstevens

Justin Wolfers
Department of Economics & The Gerald R. Ford School of Public Policy, University of Michigan & Brookings, CAMA, CEPR, CESifo, IZA and NBER jwolfers@umich.edu www.nber.org/~jwolfers

ABSTRACT Many scholars have argued that once “basic needs” have been met, higher income is no longer associated with higher in subjective well-being. We assess the validity of this claim in comparisons of both rich and poor countries, and also of rich and poor people within a country. Analyzing multiple datasets, multiple definitions of “basic needs” and multiple questions about well-being, we find no support for this claim. The relationship between well-being and income is roughly linear-log and does not diminish as incomes rise. If there is a satiation point, we are yet to reach it.

This draft: Keywords: JEL codes:

April 16, 2013 Subjective well-being, happiness, satiation, basic needs, Easterlin paradox. D6, I3, N3, O1, O4

*

A shorter version of this paper will appear in the American Economic Review, Papers and Proceedings in May 2013. The authors wish to thank Angus Deaton, Daniel Kahneman, and Alan Krueger for useful discussions and The Gallup Organization, where Wolfers serves as a Senior Scientist, for providing data. The views expressed herein are those of the author(s) and do not necessarily reflect the views of the National Bureau of Economic Research.

In 1974 Richard Easterlin famously posited that increasing average income did not raise average well-being, a claim that became known as the Easterlin Paradox. However, in recent years new and more comprehensive data has allowed for greater testing of Easterlin’s claim. Studies by us and others have pointed to a robust positive relationship between well-being and income across countries and over time (Deaton, 2008; Stevenson and Wolfers, 2008; Sacks, Stevenson, and Wolfers, 2013). Yet, some researchers have argued for a modified version of Easterlin’s hypothesis, acknowledging the existence of a link between income and well-being among those whose basic needs have not been met, but claiming that beyond a certain income threshold, further income is unrelated to well-being. The existence of such a satiation point is claimed widely, although there has been no formal statistical evidence presented to support this view. For example Diener and Seligman (2004, p.5) state that “there are only small increases in well-being” above some threshold. While Clark, Frijters and Shields (2008, p.123) state more starkly that “greater economic prosperity at some point ceases to buy more happiness,” a similar claim is made by Di Tella and MacCulloch (2008, p.17): “once basic needs have been satisfied, there is full adaptation to further economic growth.” The income level beyond which further income no longer yields greater well-being is typically said to be somewhere between $8,000 and $25,000. Layard (2003, p.17) argues that “once a country has over $15,000 per head, its level of happiness appears to be independent of its income;” while in subsequent work he argued for a $20,000 threshold (Layard, 2005 p.32-33). Frey and Stutzer (2002, p.416) claim that “income provides happiness at low levels of development but once a threshold (around $10,000) is reached, the average income level in a country has little effect on average subjective well-being.”

1

Many of these claims, of a critical level of GDP beyond which happiness and GDP are no longer linked, come from cursorily examining plots of well-being against the level of per capita GDP. Such graphs show clearly that increasing income yields diminishing marginal gains in subjective well-being.2 However this relationship need not reach a point of nirvana beyond which further gains in well-being are absent. For instance Deaton (2008) and Stevenson and Wolfers (2008) find that the well-being–income relationship is roughly a linear-log relationship, such that, while each additional dollar of income yields a greater increment to measured happiness for the poor than for the rich, there is no satiation point. In this paper we provide a sustained examination of whether there is a critical income level beyond which the well-being–income relationship is qualitatively different, a claim referred to as the modified-Easterlin hypothesis.3 As a statistical claim, we shall test two versions of the hypothesis. The first, a stronger version, is that beyond some level of basic needs, income is uncorrelated with subjective well-being; the second, a weaker version, is that the well-being– income link estimated among the poor differs from that found among the rich. Claims of satiation have been made for comparisons between rich and poor people within a country, comparisons between rich and poor countries, and comparisons of average well-being in countries over time, as they grow. The time series analysis is complicated by the challenges of compiling comparable data over time and thus we focus in this short paper on the cross-sectional relationships seen within and between countries. Recent work by Sacks, Stevenson, and Wolfers
2

We should add a caveat, that this inference of “diminishing marginal well-being” requires taking a stronger stand on the appropriate cardinalization of subjective well-being (Oswald, 2008). 3 We should note that the term “modified-Easterlin hypothesis” is something of a misnomer, as Easterlin himself is not among those claiming a satiation point. Instead, Easterlin and Sawangfa (2009) make the even stronger claim rising aggregate income is not associated with rising subjective well-being at any level of income. While incorrect, it is not uncommon, however, to attribute the “modified Eaterlin hypothesis” to Easterlin, and indeed, his citation for the IZA Prize says that: “Societies with higher material wealth are on average more satisfied than poorer ones, but once the participation in the workforce ensures a certain level of material wealth, guaranteeing basic needs, individual as well as societal well-being as a whole are no longer increasing with a growth of economic wealth.”

2

(2013) provide evidence on the time series relationship that is consistent with the findings presented here. To preview, we find no evidence of a satiation point. The income–well-being link that one finds when examining only the poor, is similar to that found when examining only the rich. We show that this finding is robust across a variety of datasets, for various measures of subjective well-being, at various thresholds, and that it holds in roughly equal measure when making cross-national comparisons between rich and poor countries as when making comparisons between rich and poor people within a country.

I.

Cross-Country Comparisons We begin by evaluating whether countries at different levels of economic development have different average levels of subjective well-being. Our measure of economic development is the log of real GDP per capita, measured at purchasing power parity.4 We will follow four approaches in our analysis: following Layard (2003), we will define “rich” as those people or countries with income greater than $15,000 per capita; alternatively, following Di Tella and MacCulloch, we will contrast the income-happiness gradient in each half of the income distribution (with the median income “cutpoint” estimated separately, depending on the specific population we are studying). We will also consider lower and higher cut-points of $8,000 and $25,000. Finally—and perhaps more satisfyingly—we will, where possible, show scatter plots and non-parametric fits of the income-happiness data over the full range of variation, allowing the reader to assess visually if this relationship changes beyond any particular income level.

For most countries GDP comes from the World Bank’s World Development Indicators. Detailed information about how we fill in missing data is available in Sacks, Stevenson, and Wolfers (2013).

4

3

We want to assess well-being measured in many different data sets, thus we standardize well-being responses by subtracting the mean, and dividing by the typical cross-section of happiness within a country at a point in time.5 This approach yields “z-score” measures of wellbeing that are transparent, easy to calculate, and comparable across data sets measuring wellbeing on differing scales. It also ensures the estimated well-being–income gradient is roughly comparable to earlier research which had analyzed ordered probit regressions. However, the disadvantage of this approach is that it is clearly ad hoc, as it assumes, for instance, that the difference between being “very happy” and “pretty happy” is equivalent to the difference between “pretty happy” and “not too happy.”6 Figure 1 shows two measures of life satisfaction drawn from the Gallup World Poll: in the top panel, we analyze responses to the “ladder of life” question, while the bottom panel shows responses to a question about overall life satisfaction.7 The data are drawn from the five waves of the Gallup World Poll run between 2008 and 2012 and GDP per capita, plotted on a log scale. We have data on 155 countries, which account for over 95% of the world’s population, across the spectrum of levels of economic development. Each of these measures of subjective well-being is highly correlation with GDP per capita ( 0.79 for the 155 countries in the upper

That is, the denominator in this “z-score” is the standard deviation of well-being after controlling for country and wave fixed effects. 6 Fortunately, this issue turns out to be more troubling in theory than in practice; Stevenson and Wolfers (2008) show alternative approaches using instead ordered probits or logits yield estimates of national happiness averages that are highly correlated ( 0.99). 7 The question analyzed in the top graph is “Please imagine a ladder with steps numbered from zero at the bottom to ten at the top. Suppose we say that the top of the ladder represents the best possible life for you, and the bottom of the ladder represents the worst possible life for you. On which step of the ladder would you say you personally feel you stand at this time, assuming that the higher the step the better you feel about your life, and the lower the step the worse you feel about it? Which step comes closest to the way you feel?” The question answered in the bottom graph is “All things considered, how satisfied are you with your life as a whole these days? Use a 0 to 10 scale, where 0 is dissatisfied and 10 is satisfied.”

5

4

panel, and 0.85 for the 86 countries in the lower panel) . The solid lines show the results from a simple OLS regression, estimated for the full sample: – log (1)

The estimated well-being–income gradient ( ) is 0.33 (se=0.02) for the ladder question and 0.44 (se=0.03) for the life satisfaction question. The figure also plots a local linear regression as a dotted line, which allows for a non-parametric fit of the well-being–income relationship. If there were a “satiation point,” this non-parametric fit would flatten out once basic needs were met. Instead, the line steepens slightly among the rich nations in both graphs. Indeed, the most striking finding is simply how closely the non-parametric fit lies to the OLS regression line. That is, the well-being–income relationship among poor nations appears to extend roughly equally among rich nations.8 We repeat this exercise for using data from the World Values Survey for both life satisfaction (Appendix Figure A1) and happiness (Figure A2), as well as for the satisfaction ladder question asked in the Pew Global Attitudes Survey (Figure A3), and the 4point happiness question asked in the International Social Survey Program (Figure A4). In each case, we find qualitatively similar results.

Deaton (2008) and Stevenson and Wolfers (2008) make similar arguments using 2006 data from the Gallup World Poll.

8

5

Figure 1: Satisfaction and GDP per capita
9 8 7 6 5 4 3 2 .25 9 8
Life Satisfaction (0−10 scale)
DNK NOR CHE NLD CAN FIN SWE 1.0 AUT ISR ISL NZL AUS CRI IRL USA LUX PAN ARE BEL MEX VEN GBR OMN BRA FRA TTO DEUKWT QAT CYP SGP SAU ESP 0.5 CHL COL ARG CZE ITA GTM THA URY SVN KOR MLT TWN SVK JPN TKM SLV POL MYS MDA BOL ECU KAZ HRV GRC PRY PER MUS ALB BLRLTU UZB VNM JORJAM RUS BHR HKG DZA UNK 0.0 NIC HND LBY IDN TUR EST NGA ROM UKR MNELVA PRT PAK ZMB KGZ PHL AGO DOM LBN TUN LSO DJI IRQMAR CHNZAF GHA LAO SOM MOZ BGD SWZ BIH IND MNG IRN HUN SRB AZE MRT ETH MWI UGA TJK PSE MKD CMR MMR SDN −0.5 EGY COD ARM ZWE NPL SYR HTI CIV LBR NERMDG TCD KHM LKA BWA MLI KEN YEM COG BFA SEN GAB GEO AFG BGR RWA SLE GINTZA BDI CAF COM BEN

(Gallup World Poll, 2008−2012)
1.5
Satisfaction Ladder (normalized scale) Life Satisfaction (normalized scale) Satisfaction Ladder (0−10 scale)

Satisfaction Ladder

−1.0 −1.5

TGO

GDP$15k: Slope=0.67 (0.10)

.5

1 2 4 8 16 GDP per capita at PPP ($000s, log scale)

32

64

(Gallup World Poll, 2008−2012)
VEN TKM

Life Satisfaction

1.5
DNK IRL FIN NOR CHE AUS ISL NZL SWE AUT NLD LUX ESP 1.0 GBR BEL DEU CYP MLT SVN FRA ITA SGP QAT HKG 0.5 SVK TWN

7 6 5 4 3 2 .25
MWI UZB

KAZ SYR EGY TUN RUS PRT LTU JOR DZA DJI MDA EST BLR VNM LVA PAK IRQ TCD AZE BGD CHN TJK ARM KGZ MRT ZAF NGA GHA LKA LBN BWA SEN BGR CAF UGA CIV ZMB AGO GEO RWA AFG CMR MOZ MLI LBR NERMDG KEN COG SLE BFA ZWE BDI BEN TGO TZA GDP>$15k: Slope=0.72 (0.16)

0.0 −0.5 −1.0 −1.5

GDP