1 American Economic Review: Papers & Proceedings 2013, 103(3): 598–604 http://dx.doi.org/10.1257/aer.103.3.598 Subjective Well-Being and Income: † Is There Any Evidence of Satiation? Betsey Stevenson and Justin Wolfers* By “once a country has over $15,000 per head, its In 1974 Richard Easterlin famously posited that increasing average income did not raise aver - level of happiness appears to be independent age well-being, a claim that became known as of its income;” while in subsequent work he argued for a $20,000 threshold the Easterlin Paradox. However, in recent years Layard 2005 ( ) 2002, p.416 ( . Frey and Stutzer new and more comprehensive data has allowed p.32–33 ) claim that “income provides happiness at low levels for greater testing of Easterlin’s claim. Studies of development but once a threshold ( around by us and others have pointed to a robust posi- is reached, the average income level in $10,000 ) tive relationship between well-being and income Deaton 2008; a country has little effect on average subjective across countries and over time ( Stevenson and Wolfers 2008; Sacks, Stevenson, well-being.” ) . Yet, some researchers have Many of these claims of a critical level of and Wolfers 2013 GDP beyond which happiness and GDP are no argued for a modified version of Easterlin’s longer linked come from cursorily examining hypothesis, acknowledging the existence of a plots of well-being against the level of per capita link between income and well-being among those whose basic needs have not been met, but GDP. Such graphs show clearly that increasing income yields diminishing marginal gains in claiming that beyond a certain income thresh- 1 old, further income is unrelated to well-being. subjective well-being. However this relation- The existence of such a satiation point is ship need not reach a point of nirvana beyond which further gains in well-being are absent. claimed widely, although there has been no for - 2008 and Stevenson and ( mal statistical evidence presented to support this ) For instance Deaton ( 2004, ( view. For example Diener and Seligman ) find that the well-being–income Wolfers 2008 p.5 state that “there are only small increases in ) relationship is roughly a linear-log relationship, such that, while each additional dollar of income well-being” above some threshold. While Clark, ( yields a greater increment to measured happi- ) state more Frijters and Shields 2008, p.123 starkly that “greater economic prosperity at ness for the poor than for the rich, there is no some point ceases to buy more happiness,” a sim- satiation point. ilar claim is made by Di Tella and MacCulloch In this paper we provide a sustained examina- ( ) tion of whether there is a critical income level 2008, p.17 : “once basic needs have been satis- fied, there is full adaptation to further economic beyond which the well-being–income relation- ship is qualitatively different, a claim referred growth.” The income level beyond which further 2 As a income no longer yields greater well-being is to as the modified-Easterlin hypothesis. typically said to be somewhere between $8,000 and $25,000. Layard argues that ) 2003, p.17 ( 1 We should add a caveat, that this inference of “dimin- ishing marginal well-being” requires taking a stronger stand on the appropriate cardinalization of subjective well-being * Stevenson: Gerald R. Ford School, University of ( . Michigan, Ann Arbor, MI, 48109 e-mail: [email protected] Oswald 2008 ( ) 2 ) edu ; Wolfers: Economics Dept and Gerald R. Ford School, We should note that the term “modified-Easterlin hypothesis” is something of a misnomer, as Easterlin him- ( e-mail: University of Michigan, Ann Arbor, MI, 48109 [email protected] . The authors wish to thank Angus ) self is not among those claiming a satiation point. Instead, Deaton, Daniel Kahneman, and Alan Krueger for useful dis- Easterlin and Sawangfa ( 2009 ) make the even stronger cussions and The Gallup Organization for providing data. claim that rising aggregate income is not associated with † To view additional materials, and author disclosure level of income. While rising subjective well-being at any ( s statement ) ,visit the article page at incorrect, it is not uncommon, however, to attribute the http://dx.doi.org/10.1257/aer.103.3.598. “modified Eaterlin hypothesis” to Easterlin, and indeed, his 598
2 VOL. 103 NO. 3 THERE ANY EVIDENCE OF SATIATION? SUBJECTIVE WELL-BEING AND INCOME:IS 599 that our findings are not sensitive to considering statistical claim, we shall test two versions of alternative thresholds. the hypothesis. The first, a stronger version, is that beyond some level of basic needs, income We want to assess well-being measured in many different datasets, thus we standardize is uncorrelated with subjective well-being; the second, a weaker version, is that the well-being– well-being responses by subtracting the mean, income link estimated among the poor differs and dividing by the typical cross-section of hap- 4 from that found among the rich. This piness within a country at a point in time. z -score” measures of well- Claims of satiation have been made for com- approach yields “ being that are transparent, easy to calculate, parisons between rich and poor people within and comparable across datasets measuring well- a country, comparisons between rich and poor countries, and comparisons of average well- being on differing scales. It also ensures the esti- mated well-being–income gradient is roughly being in countries over time, as they grow. comparable to earlier research which had ana- The time series analysis is complicated by the lyzed ordered probit regressions. However, the challenges of compiling comparable data over disadvantage of this approach is that it is clearly time and thus we focus in this short paper on ad hoc, as it assumes, for instance, that the dif- the cross-sectional relationships seen within ference between being “very happy” and “pretty and between countries. Recent work by Sacks, ) happy” is equivalent to the difference between 2013 provide evidence Stevenson, and Wolfers ( 5 “pretty happy” and “not too happy.” on the time series relationship that is consistent with the findings presented here. Figure 1 shows average levels of life satisfac- tion drawn from the five waves of the Gallup To preview, we find no evidence of a satia- tion point. The well-being–income link that one World Poll run between 2008 and 2012 and GDP per capita, plotted on a log scale. We have data finds when examining only the poor, is similar - to that found when examining only the rich. We per on 155 countries, which account for over 95 cent of the world’s population, across the spec- show that this finding is robust across a variety trum of levels of economic development. The of datasets, for various measures of subjective correlation of these variables is 0.79, remark- well-being, at various thresholds, and that it ably high. The solid line shows the results from holds in roughly equal measure when making a simple OLS regression, estimated for the full cross-national comparisons between rich and sample: poor countries as when making comparisons between rich and poor people within a country. ( bein g – Well ) 1 . ε + GD P = α + β log ) ( c c c I. Cross-Country Comparisons is (β ) The estimated well-being–income gradient ) . The figure also plots a local ( se = 0.335 We begin by evaluating whether countries at 0.018 linear regression as a dotted line, which allows different levels of economic development have different average levels of subjective well-being. for a non-parametric fit of the well-being– income relationship. If there were a “satiation Our measure of economic development is the point,” this non-parametric fit would flatten out log of real GDP per capita, measured at pur - 3 In our analysis we follow chasing power parity. once basic needs were met. Instead, the line ( 2003 , and define “rich” as those people steepens slightly among the rich nations. Indeed, Layard ) the most striking finding is simply how closely or countries with income greater than $15,000 the non-parametric fit lies to the OLS regression per capita, although the online Appendix shows line. That is, the well-being–income relationship citation for the IZA Prize says that: “Societies with higher material wealth are on average more satisfied than poorer 4 -score” is the standard z That is, the denominator in this “ ones, but once the participation in the workforce ensures a deviation of well-being after controlling for country and certain level of material wealth, guaranteeing basic needs, wave fixed effects. individual as well as societal well-being as a whole are no 5 Fortunately, this issue turns out to be more troubling in longer increasing with a growth of economic wealth.” 3 2008 For most countries GDP comes from the World Bank’s show ) ( theory than in practice; Stevenson and Wolfers alternative approaches using instead ordered probits or log- World Development Indicators. Detailed information about its yield estimates of national happiness averages that are how we fill in missing data is available in Sacks, Stevenson, . 2013 ) ( and Wolfers highly correlated (ρ > 0.99 ) .
3 MAY 2013 600 AEA PAPERS AND PROCEEDINGS ) ≥ $ among “rich” countries k ) . ( those with GDP 9 Satisfaction ladder log relative to a “cutoff,” By measuring GDP ) ( 1. 5 ( Gallup World Poll, 2008–2012 ) ) 8 this functional form allows for a change in the 1. 0 well-being–income gradient ( i.e., a “kink” in the LUX 0–10 7 ( QAT ) regression line once GDP per capita exceeds 0.5 6 the cutoff, but it rules out a discontinuous shift normalized scale ( 0 in well-being once per capita GDP exceeds 5 7 $ k This specification allows us to test both . –0.5 4 the “strong” version of the modified-Easterlin –1.0 0, and the β hypothesis, which posits that = rich 3 Satisfaction ladder “weak” v . ersion, suggesting β > β –1.5 rich poor 0.10 ( 0.67 = $15k: Slope > < ) 0.03 ( 0.25 = ) $15k: Slope GDP GDP 2 In Table 1 we report results where the cutoff 8 Satisfaction ladder 0.25 0.5 1 2 4 8 16 32 64 level of per capita GDP, $ k , is set to $15,000. GDP per capita at PPP US$ We repeat the results seen in Figure 1 in the first thousands of dollars, log scale ) ( row. Subsequent rows show the results across different questions assessing well-being and dif- Figure 1. Life Satisfaction and Income around the World ferent datasets. The well-being–income gradient in the Gallup World Poll clearly remains strong Notes: Author’s calculations, based on 2008–2012 waves of for the rich countries, and indeed, is somewhat the Gallup World Poll. Solid line shows results from a sim- stronger among countries whose per capita GDP ple OLS regression of satisfaction on log GDP per capita; exceeds $15,000. These data clearly reject both the dashed line allow the slope to shift at a per capita GDP of $15,000, respectively. The dotted line shows a lowess fit the weak and strong versions of the modified- with bandwidth set to 0.8. Easterlin hypothesis. The next ten rows repeat the analysis using five rounds of the World Values Survey for both among poor nations appears to extend roughly a life satisfaction question which mirrors that in 6 the Gallup World Poll, and a question on happi- equally among rich nations. Our more formal tests of the modified- ness. The results roughly parallel those above, 9 ypothesis come from regressions of Easterlin h albeit with less statistical power. In seven of the form: the ten rows we can reject the strong claim that 0. In two cases β and are statis- β = β rich poor rich tically significantly different from each other, bein – Well ) 2 g α = ( c however the well-being–income relationship P GD I β + is steeper among rich countries than the poor. < k ( ) poor c Indeed, in all but two cases, the estimate of β rich P × − log k actually exceeds that for GD log β rather than the ( ) ( ( ) ) ( poor c . In the two cases in which the other way around ) k β point estimate of ≥ P GD is larger, we cannot reject I + β ( ) rich c poor . β = β the null that rich poor , ε k − log + × log P GD ( ( ( ) ) ) c c 7 We obtain similar results if instead we estimate the c where the subscript denotes country, the inde- well-being–income gradient separately for rich and poor pendent variables are the interaction of log real countries. 8 GDP per capita with a dummy variable indicat- Online Appendix Table 1 shows the results using alter - native thresholds of $8,000 and $25,000, as well as the ing whether GDP per capita is above or below a median level of GDP for the sample. Stevenson and Wolfers is the well- . The coefficient k β cut-off level, $ poor show estimates of ordered probit regressions estimat- ( 2008 ) being–income gradient among “poor” countries ing the well-being–income gradient for incomes above and $ k ) , and β is the gradient ( those with GDP < below $15,000, while Deaton tested thresholds of 2008 ) ( rich $12,000 and $20,000. 9 In several countries the surveys were not nationally rep- resentative, focusing instead on urban areas or more edu- 6 2008 ( Deaton ( cated members of society. Our anaylsis drops particularly ) and Stevenson and Wolfers 2008 make ) unrepresentative observations as detailed in Stevenson and similar arguments using 2006 data from the Gallup World 2008 ) and Sacks, Stevenson, and Wolfers ( Wolfers ) . Poll. ( 2013
4 VOL. 103 NO. 3 THERE ANY EVIDENCE OF SATIATION? SUBJECTIVE WELL-BEING AND INCOME:IS 601 Table 1—Cross Country Evidence Well-being data Difference β β poor rich Panel A. Gallup World Poll 2005–2012 0.422*** Satisfaction ladder 0.252*** 0.674*** ( ( ) ) 0.123 ) 0.023 ( 0.103 0.360* 0.361*** 0.720*** Life satisfaction ) ( 0.051 ) ( 0.198 ) ( 0.160 Panel B. World Values Survey 0.185 Life satisfaction: 1981–1984 wave 0.668 − 0.484 ( 0.430 ) ) 0.772 ) ( 0.418 ( 0.515* Life satisfaction: 1989–1993 wave 0.694*** 0.179 0.241 ) ( 0.284 ) ( 0.488 ) ( 0.640*** 0.445*** Life satisfaction: 1994–1999 wave 0.195 ( ) ( 0.105 ) ( 0.259 ) 0.185 0.755*** 0.209*** Life satisfaction: 2000–2004 wave 0.546** ) 0.201 0.066 ) ( 0.152 ) ( ( 0.254*** 0.176 Life satisfaction: 2005–2009 wave − 0.078 0.137 ) ( 0.056 ) ( 0.179 ) ( 0.567 0.087 0.481 Happiness: 1981–1984 wave 0.387 ( 0.338 ) ( 0.685 ) ( ) 0.945*** 0.430 0.515 Happiness: 1989–1993 wave 0.472 ( ( 0.281 ) ( ) ) 0.231 Happiness: 1994–1999 wave 0.599*** 0.241** 0.357 ( 0.184 ) ( 0.106 ) ( 0.260 ) Happiness: 2000–2004 wave 0.796*** 0.864*** 0.068 − ) ( 0.075 ) ( 0.222 ) ( 0.164 0.055 0.332** 0.277 Happiness: 2005–2009 wave ) ( 0.061 ) ( ( ) 0.135 0.182 Panel C. Pew Global Attitudes Survey Satisfaction ladder: 2002 0.716*** 0.163** 0.552** 0.205 ) ( ( ) ( 0.270 ) 0.079 Satisfaction ladder: 2007 0.208*** 0.197 0.405** 0.175 ( 0.072 ) ( 0.233 ) ( ) 0.279** 0.248* Satisfaction ladder: 2010 0.031 ( ) ( 0.126 0.295 ( 0.411 ) ) Panel D. ISSP Happiness 2008 0.449*** 0.694** 0.245 − 0.162 ) 0.190 ) ( 0.292 ) ( ( 0.788*** 0.424*** Happiness 2007 − 0.364** ( ) ( 0.148 ) ( 0.270 ) 0.149 Happiness 2001 0.713*** 0.960*** − 0.247** ( 0.232 ) ( 0.111 ) ( 0.252 ) Happiness 1998 1.00*** 0.925*** 0.076 − 0.193 ) ( 0.223 ) ( 0.362 ) ( 0.923*** Happiness 1991 1.10*** 0.177 − 0.370 ( ) ( 0.127 ) ( 0.262 ) *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level.
5 MAY 2013 602 AEA PAPERS AND PROCEEDINGS Table 2—Income and Satisfaction in the United States when comparing rich and poor people within a country. We begin by analyzing data from the Very Not too Fairly United States, and in particular, the Gallup poll happy happy Annual household happy percent ) income ) ( ( ) percent ( percent conducted on December 6–9, 2007. These data are particularly useful because the top income <$10k 35 44 21 code is unusually high, allowing respondents 42 $10k–$20k 42 15 52 43 $20k–$30k 5 to report household income in categories up to 41 4 $30k–$40k 55 $500,000. If we are to find evidence of satia- 46 9 $40k–$50k 46 tion, these data seem like the right place to look. 5 $50k–$75k 55 40 Table 2 shows a simple cross-tab of happiness 36 60 $75k–$100k 4 and household income. The positive association $100k–$150k 60 40 0 $150k–$250k 70 30 0 between family income and reported well-being 10 0 83 $250k–$500k 17 is remarkably consistent. Online Appendix 100 0 0 > $500k Table 2 shows that a similar finding holds when asking instead about life satisfaction. Note: Author’s calculations, based on Gallup Poll conducted December 6–9, 2007. When we analyze these data more formally ) not shown ( in regressions we find no evidence of a significant break in either the happiness- income relationship, nor in the life satisfaction- There are two other useful cross-country income relationship, even at annual incomes up studies that are worth analyzing, the Pew Global Attitudes studies, which posed the satisfac- to half a million dollars. This finding contrasts with a claim made by Frey and Stutzer ( 2002, tion ladder question in 44 countries in 2002, p.409 whose informal visual assessment of data ) 47 countries in 2007, and 22 countries in 2010, and the International Social Survey Program, for 1972–1974 ( from the General Social Survey which asked a consistent happiness question in ) and 1994–1996 led them to conclude that “the same proportional increase in income yields a 1991, 1998, 2001, 2007, and 2008. Each of these datasets strongly reject the null that 0. = β lower increase in happiness at higher income rich levels.” In our re-analysis of that same dataset Moreover, to the extent that the well-being– not shown ) we used data from all years, but ( income relationship changes, it appears stron- ger for rich countries. Somewhat paradoxically, even with these larger samples could not reject the null that proportional increases in income the ISSP data appear to show a negative well- continue to yield the same increase in happiness being–income gradient among poor nations, but at higher income levels. - this is entirely due to a single influential obser Looking beyond the United States, we can use ( whose influence is even vation, the Philippines the individual country data in the Gallup World greater given that these samples contain mostly Poll to examine the within-country well-being– medium- and high-income countries ) . 2 All told, comparisons of average levels of income gradients in each nation. In Figure ( subjective well-being and GDP per capita across “lowess” we perform separate local linear ) regressions estimating the satisfaction-income countries suggest that the well-being–income relationship observed among poor countries relationship non-parametrically for each of holds in at least equal measure among rich coun- the world’s 25 most populous countries. These tries. In the few cases where we cannot reject results are shown for those respondents whose β annual household income lies between the = β . Our β 0, we also cannot reject = rich rich poor larger datasets emphatically reject the weak and tenth and ninetieth percentiles of their national strong forms of the modified-Easterlin hypoth- income distributions. While there are differ - esis, while the smaller samples are sufficiently ences in the location of these non-parametric imprecise as to provide no statistically signifi- fits, and even some differences in the slopes, ( or against the more remarkable feature is simply that for it. cant evidence in support of ) II. Within-Country Cross-Sectional Comparisons 10 While 100 percent of those reporting annual incomes over $500,000 are in the top bucket of “very happy,” it is We now turn to analyzing the relationship important to note that there are only eight individuals in this category. between well-being and income that one obtains
6 VOL. 103 NO. 3 THERE ANY EVIDENCE OF SATIATION? SUBJECTIVE WELL-BEING AND INCOME:IS 603 ) ) 8 1. 0 Above: 61 1. 0 0.8 7 0.6 0.5 $15,000 standardized ( > 6 0.4 normalized scale ( 0 0.2 5 0 –0.5 for incomes –0.2 : Well-being–income gradient 4 rich β Below: 37 –0.4 Satisfaction ladder –1.0 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 0.5 1 2 4 8 16 32 64 128 Satisfaction ladder Annual household income β : Well-being–income gradient poor ( thousands of dollars, log scale ) < for incomes $15,000 Figure 2. Well-Being and Income, within the 25 Figure 3. The Well-Being–Income among the Rich and Largest Countries Poor in Each Country every country the relationship estimated at low income distribution—it’s median, twenty-fifth incomes appears to hold in roughly equal mea- or seventy-fifth percentile. In no case do we find evidence in favor of the sure at higher incomes. In particular, there is no modified-Easterlin evidence that the slope flattens out beyond any h ypothesis. particular “satiation point” in any nation. III. In order to provide a more formal assess- Conclusions ment, we repeat the earlier exercise, estimating While the idea that there is some critical level an analog to equation ( 2 ) , but analyzing indi- vidual well-being and household income, rather of income beyond which income no longer impacts well-being is intuitively appealing, it is than national averages, and allowing the slope at odds with the data. As we have shown, there to change for household incomes above $15,000 is no major well-being dataset that supports this per annum. We repeat this exercise for 98 coun- commonly-made claim. To be clear, our analysis tries in which we have at least 200 respondents in this paper has been confined to the sorts of both above and below the threshold. We report evaluative measures of life satisfaction and hap- the results of these 98 regressions compactly in Figure 3. The vertical axis shows piness that have been the focus of proponents , the β rich Easterlin hypothesis. In an ver the modified estimated well-being–income gradient o ) ( of the interesting recent contribution, Kahneman and “rich” part of the sample, while the horizontal have shown that in the United ) 2010 ( β Deaton axis shows ver the “poor” , the gradient o poor part of the sample. The strong form of modi- States, people earning above $75,000 do not appear to enjoy either more positive affect or fied-Easterlin hypothesis suggests that the well- being–income gradient is zero for the rich part less negative affect than those earning just below that. We are intrigued by these findings, although of the sample, suggesting that the data should we conclude by noting that they are based on cluster along the horizontal axis. The weaker very different measures of well-being, and so form of this hypothesis suggests a sharp break they are not necessarily in tension with our in this gradient among the “rich,” and hence that results. Indeed, those authors also find no satia- most country-level estimates will lie beneath the 45-degree line. In fact, we find 61 nations above tion point for evaluative measures of well-being. this line, and only 37 below. We also try various alternative specifications, REFERENCES changing the cutoff level of k across countries Clark, Andrew E., Paul Frijters, and Michael A. ; ) using alternative cutoffs at $8,000 and $25,000 ( Shields. 2008. “Relative Income, Happiness, k in others, s country’ parameters of a depends on
7 MAY 2013 604 AEA PAPERS AND PROCEEDINGS and Utility: An Explanation for the Easterlin 40 Journal of Economic Literature Research?” 2 : 402–35. ) ( Paradox and Other Puzzles.” Journal of Eco- Kahneman, Daniel, and Angus Deaton. 46 1 ( ) : 95–144. 2010. nomic Literature 2008. “Income, Health, and Well- Deaton, Angus. “High Income Improves Evaluation of Life But Not Emotional Well-Being.” Being around the World: Evidence from the Proceedings of the National Academy of Sciences : 107 ( 38 ) Gallup World Poll.” Journal of Economic Per - 2 : 53–72. 16489–93. ) spectives 22 ( 2003. “Happiness: Has Social Layard, Richard. Diener, Ed, and Martin E. P. Seligman. 2004. Science a Clue?” Lionel Robbins Memorial “Beyond Money: Toward an Economy of Well- Lectures 2002/3. Lecture given at the London Being.” Psychological Science in the Public : 1–31 School of Economics, London, March 3–5. Interest 5 ( 1 ) Happiness: Lessons from Layard, Richard. 2008. 2005. Di Tella, Rafael, and Robert MacCulloch. “Happiness Adaptation to Income beyond a New Science . London: Penguin. Oswald, Andrew J. ‘Basic Needs.’” National Bureau of Economic 2008. “On the Curvature of the Reporting Function from Objective Real- Research Working Paper 14539. Easterlin, Richard A. 1974. “Does Economic ity to Subjective Feelings.” Economics Letters 100 - : 369–72. Growth Improve the Human Lot? Some Empir ) 3 ( ical Evidence.” In Sacks, Daniel W., Betsey Stevenson, and Justin Nations and Households in Wolfers. 2012. “The New Stylized Facts about Economic Growth: Essays in Honor of Moses Emotion Income and Subjective Well-Being.” Abramovitz , edited by Paul A. David and Mel- ) vin W. Reder. New York: Academic Press. : 1181–87. 6 ( 12 Sacks, Daniel W., Betsey Stevenson, and Justin Easterlin, Richard A., and Onnicha Sawangfa. Wolfers. 2009. “Happiness and Economic Growth: 2013. “Growth in Subjective Well- Does the Cross Section Predict Time Trends? being and Income over Time.” Unpublished. Evidence from Developing Countries.” Unpub- Stevenson, Betsey, and Justin Wolfers. 2008. “Eco- lished. nomic Growth and Subjective Well-Being: Brookings Frey, Bruno S., and Alois Stutzer. 2002. “What Reassessing the Easterlin Paradox.” Can Economists Learn from Happiness Papers on Economic Activity : 1–87.
Health Insurance Marketplace 6 Tier Drug List June 2019 (2018 plan year) Please consider talking to your doctor about prescribing preferred medications, which may help reduce your out - of - pocket co...More info »
Health Insurance Marketplace 6 Tier Drug List May 2019 (2018 plan year) Please consider talking to your doctor about prescribing preferred medications, which may help reduce your out - of - pocket cos...More info »
Solar F AQs piled by Edited/Com : ment of Energy Jeff Tsao (U. , Office of Ba S. Depart sic Energy Science) Nate Lewis (California Institute of Technology) y) George Crabtree (Argonne National Laborat...More info »
AEA Journals Copyright and Reprint Permissions Copyright © 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012 by the American Economic Association. Permission t...More info »
TIONAL ACADEMIES PRESS THE NA This PDF is available at http://nap.edu/24938 SHARE Thriving on Our Changing Planet: A Decadal Strategy for Earth Observation from Space DET AILS 700 pages | 8.5 ...More info »