文章 - 全球幸福研究最新综述

（文内表格暂缺，网站改版后附PDF文件。）

David G. Blanchflower and Andrew J. Oswald

NBER Working Paper No. 16668

January 2011

JEL No. I1,I3

摘要

 This paper describes the findings from a new, and intrinsically interdisciplinary, literature on happinessand human well-being. The paper focuses on international evidence. We report the patterns in moderndata; we discuss what has been persuasively established and what has not; we suggest paths for futureresearch. Looking ahead, our instinct is that this social-science research avenue will gradually merge with a related literature -- from the medical, epidemiological, and biological sciences -- on biomarkersand health. Nevertheless, we expect that intellectual convergence to happen slowly.

本文阐述了一个新的、各学科间本质的发现，即文学上所说的幸福与人类福祉。本文侧重于国际上的证据，我们用了最新的数据来报告，我们讨论了什么是有说服力的和没有说服力的，我们提出了进一步研究的路径。展望未来，我们的本质就是社会科学的研究途径——从医学、流行病学和生物科学——论生物链健康将逐步与相关文学的融合。不过，我们认为智力的收敛将慢慢发生。

 

David G. Blanchflower

Bruce V. Rauner Professor of Economics

6106 Rockefeller Hall

Dartmouth College

Hanover, NH 03755-3514

and NBER

 

Andrew J. Oswald

Department of Economics

University of Warwick

Coventry CV4 7AL

England

 

David G. Blanchflower

Andrew J. Oswald

“Measures of both objective and subjective well-being provide key information about people’s quality of life. Statistical offices should incorporate questions to capture people’s life evaluations, hedonic experiences and priorities ...”

Stiglitz et al, Commission on the Measurement of Economic Performance and Social Progress, 2009.

“这种对客观和主观福祉的测量能提供有关人们生活质量的关键信息。统计局应纳入考虑追踪人们的生活评价、享乐经验和优先事项……”

斯蒂格利茨等人在2009年经济表现和社会进步委员会上如是说。

 

1. Introduction

引言

Standard indices of a country’s prosperity -- such as longevity, human height, the incidence of disease, the suicide rate, and the level of Gross Domestic Product -- are well-known and widely collected. Governments throughout the developed nations, and in many developing nations, regularly publish such numbers. Yet if they are to do their job effectively, politicians and policy-makers arguably have to go beyond this. They have to try to understand, and measure, the happiness and mental health of their country’s citizens. The reason is that it is overall human well-being (not simply a collection of its constituent elements) that is ultimately of interest.

一个国家繁荣的基准指标，如长寿、人类身高、疾病的发病率、自杀率，与国内生产总值的水平，是众所周知的被广泛收集。所有发达国家，以及许多发展中国家的政府，会定期公布这些数字。然而，政治家和政策制定者可以说是必须超越这一点，如果他们是有效率地去做。他们必须试着去了解、衡量本国公民的幸福和心理健康。原因是，它是人类最终利益的全部福祉（而不仅仅是其组成元素的收集）。

The last few decades have seen a body of researchers attempt to rise to the difficult challenge of how to study ‘happiness’ in a systematic, empirical way. These scholars come from a range of disciplines -- including psychology, economics, epidemiology, medicine, statistics, sociology, political science, and management science. Although their methodological approaches differ in detail (moreover, researchers in one discipline are not always good at citing related work in the journals of the other disciplines, so different sub-fields sometimes lay claim to having discovered results first), a common methodology has begun to emerge.

近几十年中所看到的一个研究机构用一个系统的、实证的方式研究“幸福” ，以试图提升对困难的挑战。这些学者来自不同的学科——包括心理学、经济学、流行病学、医学、统计学、社会学、政治学，以及管理科学。尽管他们的方式方法在细节上有所不同（此外，一门学科研究人员在引用在其他相关学科的期刊并不总是做得好，因此不同的分领域有时宣称是自己首先发现的结果），但是一个共同的方法论已经开始出现。

This article describes the core findings from the new ‘happiness’ literature. One approach for a paper of this kind would be to go through in the manner of a survey article -- listing which researcher said what, and the year they said it. Here we take a different pedagogical approach. We hope it will be helpful for a starting reader who knows little about the field. In this paper, we take the latest international data and, using the methodological insights from the literature, build up from first principles to try to show readers how conclusions are reached in this field.

本文介绍了这个新的“幸福”文献的核心结果，这类文章的一个途径是通过采用一项调查——列出每个研究员说了什么，一年中他们说了。在这里，我们采取不同的教学方法。我们希望这有助于读者对这个领域有一点点的了解。在本文中，我们采用国际上的最新数据，并利用文献中的方法的见解，建立起最初的原则，试图告诉读者如何达到这一领域的结论的。

Most researchers begin from the idea that inside a human being there is some ‘happiness’or utility function of general form:

It then draws upon quantitative methods that are formally similar to those employed in medical statistics, econometrics, and quantitative management science. Authors typically take a random sample of the population; they use multiple-regression techniques; some form of well-being measure is taken as the dependent variable in these equations; researchers calculate the size of the coefficients within so-called ‘happiness equations’ (Powdthavee 2010 is a valuable nontechnical guide to this). At a formal level, this method is like the approach of an epidemiologist who wishes to understand the myriad influences on a person’s chance of good or bad life outcomes such as getting cancer. In both literatures -- epidemiology research and happiness research -- the outcome of such research is a linear or logistic equation in which factors like a person’s age, gender, diet and smoking are shown statistically to matter.

大多数研究人员开始于这样一个构想，即一个人里有一些“幸福”或效用函数的一般形式：幸福=f*（年龄，性别，收入，教育程度，婚姻状况，饮食，其他个人特征，区域特点，国家的特性）

然后，利用类似于在医学统计、计量经济学、量化管理科学采用的定量方法。作者通常采取随机抽样的人口，他们使用多回归分析技术，有些福祉的衡量形式是作为这些方程的因变量，研究人员计算所谓的“幸福方程式”的系数的大小（ Powdthavee2010年是一个宝贵的非技术指导）。在正式的层面上，这种方法就像一个流行病学家希望了解，如患上癌症对一个人或好或坏的生活结果的影响。在这两个文献中——流行病学研究和幸福感研究——这种研究的结果是线性或逻辑方程，其因子，如一个人的年龄、性别、饮食和吸烟，统计显示其实质。

Is happiness actually measurable? It is likely that debates about the right interpretation of subjective measures will continue throughout the 21st century and beyond. But in social science they currently do so in a less strident way than decades earlier. Frey and Stutzer (2002) summarize ways to validate ‘happiness’ data. Krueger and Schkade (2008) show that people’s reported well-being numbers are reasonably stable through time. Oswald and Wu (2010) demonstrate that across the states of the USA there is a strong match between subjective and objective well-being.2

幸福实际上是可测量的吗？很可能对这种主观测量合理解释的争论将继续在整个21世纪及以后。但在社会科学方面他们目前这样做相比数十年前更少了刺耳的方式。弗雷和斯塔则（2002）总结的方法来验证“幸福”的数据。克鲁格和施卡德（2008年）表明，人们的报告福祉号码在长时间里相当稳定。奥斯瓦德和吴（2010）表明，在美国各州之间福祉存在着主观和客观的强匹配。

What is noticeable about this line of modern social-science research is not merely the discoveries that have been made but the attention that such work has garnered outside academia. People seem to be attracted to the topic. Hundreds of recent newspaper articles have appeared discussing ‘happiness research’. There are a number of popular science-of-happiness books and the Christmas edition of the latest (December 2010) issue of The Economist magazine features this research area on its cover. Politicians of left and right have shown interest, and a recent Commission led by Joseph Stiglitz and Amartya Sen produced a long report making recommendations on how, looking to the future of the industrialized nations, we might move away from simple GDP measurement.

这个现代社会科学的研究路线引人注目的不仅是已经取得的发现，而且这项工作已经获得学术界以外的重视。人们似乎被吸引到这个话题，最近报纸上出现了数百篇文章讨论“幸福研究”。有许多流行的科学的幸福书籍和最新发行在经济家杂志上的圣诞版本（2010年12月），在封面上特写该研究领域。左翼和右翼政治家都表示有兴趣，而最近由约瑟夫.斯蒂格利茨和阿马蒂亚森领导的委员会编写了一份长的关于如何看待工业化国家的未来，我们可能摆脱单纯的GDP的测量建议的报告。

Our own work in this field began in the early 1990s. We then had a connection with the London School of Economics; a third member there, Andrew E. Clark, also did early (and important) work within the same broad grouping. At that time, there was no interest from our colleagues in economics departments around the world, and at one infamous open conference in 1993 in London the three of us made up a fair proportion of the inhabitants of a large room of empty chairs. At that juncture, virtually all economists viewed such work as misguided. One long paper, on well-being over time in Britain and the USA, took a decade to get published. There was precedent for all this: the seminal work of Richard Easterlin (1974, and later) had, years before, met with even less success.

我们在这一领域的工作始于20世纪90年代初。然后，我们曾与伦敦经济学院有过联系， 有三分之一的成员，如安德鲁E.克拉克，在同样广泛的分组内也做了早期（重要的）工作。当时，我们在世界各地经济部门的同事并没有利益，在伦敦于1993年召开的一个臭名昭著的公开会议上，我们三人弥补了一个空主席位的居民的公平比例。在那个时候，几乎所有的经济学家都错误地看待这个工作。在英国和美国，随着时间的推移，关于福祉的一纸长文，花了十年终于得以出版。有这一切的先例：几年前，理查德.伊斯特林的开创性工作（1974年及以后）更是举步维艰。

Easterlin, whom we were later to meet in the late 1990s, had in the 1970s demonstrated that U.S. happiness did not seem to be rising through time with GDP growth. He had argued -- and still does -- that the likely reason was that humans are fundamentally creatures of comparison; so that when they see everyone around them becoming richer at the same time as they themselves do the net result is a kind of generalized neutrality. We go from having one Ford to having three Lexuses, and nobody is happier. Many researchers believe in some version of this idea – that utility depends on relative factors. Such an approach goes back to, for example, Duesenberry's relative-income hypothesis (1949). Recent writings on comparisons includes Hagerty (2000), Luttmer (2005), Fliessbach (2007), Clark and Senik (2010), Daly et al (2010), and Layard (2010). In its latest incarnation, some authors have begun to argue empirically that ordinal rank may be what matters to humans (Brown et al 2008 review the evidence).

我们后来在20世纪90年代末遇到的伊斯特林，70年代已表明，美国人的幸福并不像是随着GDP的增长而提高。他曾声称——也确实如此——极可能的原因是，人类是从根本上比较的动物，所以当他们看到身边的每个人都象他们自己那样同一时间变得更加富有，所做的最终结果是一种广义的中立性。我们从拥有一辆福特到有三辆雷克萨斯，没有人更快乐。许多研究者对这种想法相信某些版本，即效用依赖与相关因素，这种做法可追溯到尤森贝瑞的相对收入假说（1949年），近期的著作包括哈格蒂（2000年）、鲁迪孟尔（2005年）、弗兰斯布什（2007年）、克拉克和森尼克（2010年）、达利等人（2010年）和莱亚德（2010年）。在其最新的化身，一些作者已开始争论经验秩序的事项可能是什么人（2008年回顾布朗等人的证据）。

Despite economists’ slowness to follow in Easterlin’s footsteps, today -- at the time of writing late in 2010 -- this field within economics and social science has become one of the most quickly growing. Clark et al (2008) provides an introduction.

尽管经济学家跟随伊斯特林的脚步缓慢，但是今天——在2010年年底写作的时间，这在经济学和社会科学方面已经成为最迅速成长的领域之一。克拉克等（2008年）提供了介绍。

In this essay we touch on four themes: (i) we review the literature about what has been found in happiness research that seems true in almost all countries and is of particular interest to social scientists and management scholars; (ii) this includes issues such as which countries come out high, and which low, in happiness and well-being? (iii) what is wrong with existing approaches and why the use of national happiness is an improvement over GDP; (iv) how can the field do better, and where will research go?

在这篇文章中，我们接触四个主题：（一）我们回顾已经发现的关于幸福感研究的文献，那些在几乎所有国家看起来真实的和社会科学家和管理学者特别感兴趣的著作；（二）包括的结果，例如，对幸福和福祉，哪些国家高，哪些国家低？（三）现有的方法出了什么问题，为什么国家幸福的使用是一种对GDP的改善？（四）如何在这个领域做得更好，研究将走向何方？

The paper also discusses, and is motivated in part by, the recent Stiglitz Commission’s report on the measurement of economic and social progress in a modern economy. This report, which may not yet be known to many management scholars or social scientists, was published in 2009 and can be downloaded from www.stiglitz-sen-fitoussi.fr. The Commission was set up by Nicholas Sarkozy of France.

本文还讨论，近期斯蒂格利茨委员会关于经济和社会进步在现代经济中的测量报告的部分动机。可能还不为许多管理学者或社会科学家知道，这份报告发表于2009年，可从www.stiglitz-sen-fitoussi.fr下载。该委员会成立于法国的尼古拉.萨尔科齐。

 

2. An Introduction to the Empirical Findings

研究结论简介

 To get a feel for some of the discoveries in the happiness literature, consider Table 1, which uses the most up-to-date data available to us. It takes data on 48,000 individuals from the General Social Surveys of the United States, which since 1972 has been asking an annual random sample of Americans the question: Taken all together, how would you say things are these days: would you say that you are very happy (approximately 32% give this answer), pretty happy (56%), or not too happy (12%)? So the vast majority of respondents are quite happy or very happy, and this is a consistent finding in the literature.

为了从幸福文献的发现中得到一个线索，考虑表1，其中使用我们可利用的最新数据。它来自美国一般社会调查中的48,000样本，这些样本是从1972年以来每年美国人随机抽样的个人数据。随机抽样的问题是：合起来，你会怎么表达这些天的事情：你会说你非常快乐（约32％给出这个答案），相当快乐（56％），还是不怎么快乐（12％）？因此，绝大多数受访者都非常快乐或相当快乐，这与在文献中的发现是一致的。

Table 1 presents a regression equation that is typical of the kind estimated in the happiness literature. It has as its dependent variable a cardinal version of people’s answers, where ‘very happy’ is coded 3, ‘pretty happy’ is coded 2, and ‘not too happy’ is coded 1. The mean level of American happiness -- given this elementary cardinal numbering -- is 2.2 with standard deviation 0.6. Statistically, this approach is not ideal (an ordinal estimator, such as ordered logistic regression, is preferable); the reason is that there are not good grounds to believe that people’s happiness answers obey the rules of a cardinal scale. However, a long line of research papers has found that it makes little difference whether a cardinal or ordinal estimator is used, and the former method has the attractive feature that it is straightforward for readers to read off the size of effects in a regression equation. For ease of exposition in a review paper such as this, we use cardinal dependent variables throughout.

表1给出了回归方程，它是在幸福文献中评估的那种典型。人们的答案作为其最重要的因变量，其中“非常快乐”是编码3，“相当快乐”是编码2，“不是太快乐的”是编码1。美国幸福的平均水平作为初始基数是2.2，标准差0.6。据统计，这种方法并不理想（序估计，如有序回归为首选），原因是有充分理由不相信，人们的幸福答案服从基数规模的规则。然而，一长排的研究论文发现，是否基数或序估计是二手的并没有太多差别，前一种方法有吸引力，它是直接为读者读出了一个回归方程中的效用大小。为了便于在这样的回顾文献中论述，我们贯穿使用基数因变量。

In Table 1, we find that American happiness:

 Is U-shaped through a person’s lifespan (because Age enters, at the top of Table 1, with a quadratic shape)

 Is higher among those who are women (because Male enters with a negative coefficient of -0.0497)

 Is higher among whites, the highly educated, full-time workers, married people, and those on a high income.

 Is lower among those not in these categories or who are unemployed or temporarily not working, those who work at home, people who are widowed, divorced and separated, and those who had parents who divorced before they as children were 16 years old.

These judgments are from pooled cross-section analysis, so they describe associations in the data. We should be cautious before imputing causality into such patterns, but perhaps not so cautious that we turn a Nelson-like eye away from all such inferences. The famous and vital finding that smoking causes cancer was first observed as an elementary pattern in cross-sectional data. Moreover, those (often non-quantitative) researchers who are keen to remind us all that correlation is not causation should, in turn, occasionally be reminded that the discovery of a correlation is typically one necessary part of a proof of causation. Causation first needs correlation.

在表1中，我们发现美国幸福：

•一个人的寿命型是U型（因为年龄进入，在表1上带有二次形状）

•女性更高（因为男性为-0.0497，进入负系数）

•白人、高学历、全职工人、已婚的人和高收入的人更高。

•这些人更低：失业者或暂时没有工作的人，在家工作的人，丧偶、离婚或分居的人，以及那些在16岁之前父母就离婚的人。

这些判断是从汇集的横截面分析，因此他们所描述的数据关联。在归于这种因果关系的模式之前，我们应该谨慎，但也许不必那么谨慎，我们将纳尔逊般的眼睛转向，远离这种推论。癌吸烟会致症是著名的和重要的发现，它第一次被作为一个跨部门数据的基本模式。此外，那些（通常非量化的）研究人员热衷于提醒我们，所有的相关性不是因果关系，应该反过来，偶尔提醒的是，相关的发现是一个典型因果关系的证明，发生因果关系首先需要相关性。

Table 1 allows the size of different influences to be seen. The age and age squared coefficients reveal that Americans’ happiness reaches its low point at, on average, around 40 years old. The coefficient of -0.0497 on Male means that on average a woman reports a happiness level approximately 0.05 points higher than a man (on a scale that runs from 1.0 to 3.0). Similarly, reading down the first column of Table 1, black Americans report lower happiness than whites by approximately 0.13 points. A time trend enters negatively but, in this first column, in a way that is insignificantly different from zero. Years of schooling (i.e. the amount of education) enters strongly positively. The more educated that people are, the more they report a high score on a happiness scale.

表1允许不同大小的影响有待观察。年龄和年龄平方系数显示，美国人的平均幸福在约40岁达到其最低点。关于男人-0.0497的系数意味着，女人的平均幸福水平会比男人高大约0.05百分点（在1.0至3.0的运行规模里）。同样，看表1的第一列，美国黑人的幸福比白人低约0.13点。时间趋势进入负，某种程度在这第一列与零没什么差别。受教育年限（是教育的数额）进入强阳性。人们所受教育越多，他们的幸福水平的分数就越高。

According to the first column of Table 1, each extra year of education in the United States is associated with 0.017 extra happiness points, so that the difference, for example, between completing high school and completing a college degree is slightly greater than 0.06 happiness points. One of the (strikingly) large effects can be seen for the consequences of unemployment for happiness. Joblessness is associated here with a huge amount of unhappiness. Here, in Table 1’s first column, there is a coefficient of -0.2343 (with a t-statistic of 14.1), which is almost twice that associated with the black dummy variable or five times that associated with being male. The other particularly large coefficient is on marriage, of 0.2322. Hence married Americans are happier, ceteris paribus, by approximately one quarter of a happiness point on the 1-3 scale. ‘Separated’ is large and negative; it has a coefficient slightly exceeding -0.14. A parental-divorce variable, which measures whether the respondent at the age of sixteen lived with only one parent because they parents who were divorced, also enters negatively and with a tightly-defined standard error; its coefficient in the first column of Table 1 is -0.0436. It seems that it could be argued that American adults -- filling up their happiness forms many years after the event -- may carry a small psychological scar from parental divorce that occurred decades earlier. Again, this pattern may or may not be one of cause-and-effect.

根据表1第一列，美国教育特别的一年每次都与0.017这个额外快乐点相关联，因此这样的差异，例如，在完成高中学业和完成大学学位之间，略大于0.06幸福点。一个（惊人的）大影响可以从失业的幸福看到其后果，失业是与巨大数量的不幸相联系的。这里，在表1的第一列，有一个-0.2343系数（T-14.1统计），几乎两次与黑色的虚拟变量相关，或五次与身为男性有关。其他特别大的系数在婚姻，是0.2322。因此，其他条件不变，已婚的美国人会更幸福，在1-3规模上约高四分之一的幸福点。 “分隔”是大负的，它有一个系数略高于-0.14。父母离婚变量，由答辩人是否十六岁与单亲父母住在一起测量，因为他们的父母离婚，也进入负且紧密定义的标准误，其在表1第一列的系数是-0.0436。似乎可以说，美国的成年人——在事件发生后许多年填满了他们的幸福形式——可能随身携带一个几十年前发生父母离婚的小的心理创伤影响，从。再次，这种模式可能或可能不是原因和结果之一。

Are richer Americans happier than poorer ones? Most economists, and many ordinary

citizens, would think the answer is obvious: people strive after money so it must make them happier. Historically, there has been a huge debate on this topic in psychology journals.

美国富人比穷人更幸福？大多数经济学家和许多普通公民，会认为答案是显而易见的：人们努力追逐金钱，因此必须使他们更幸福。历史上，关于这个主题在心理学杂志出现了巨大的争论。

Some textbooks -- we deliberately omit citations on this -- have wrongly told generations of psychology undergraduates that money is not a source of happiness. In so far as regression equations can settle the question, the answer is unambiguous: yes, money buys happiness. Every extra thousand dollars of income is associated, according to column 2 of Table 1, with 0.00246 extra units of happiness. Hence $100,000 a year, for example, is the equivalent of 0.246 happiness points on the standardized GSS scale, which is slightly greater in size than the positive happiness value of marriage or the (negative) value of unemployment.

有些教科书——我们故意忽略这个引用——错误地告诉几代心理学的大学生，钱不是幸福的源泉。在迄今为止的回归方程可以解决这个问题，答案是明确的：是的，金钱可以买到幸福。根据表1的第2列，每多一千美元收入，相关联0.00246的额外幸福单位。因此，例如，10万美元是标准化安全总局的0.246幸福点，这稍大于结婚正幸福值或失业（负）幸福值。

Recent work by Kahneman and Deaton (2010) has raised doubts about whether all measures of well-being are strictly increasing in income. A great deal of future research can be expected on this important issue.

卡尼曼和迪顿（2010）在近期工作中提出质疑，是否福祉的一切测量都严格按收入提高。可以预期，对这一重要问题未来会有大量研究。

Although it looks strange at first sight, in principle these statistical methods allow researchers to work out in an exact way the deep determinants of human well-being: they allow us to put a price tag on the happiness value of a host of life’s influences. This is what the research literature has done. Much remains to be learned. But such valuations are now entering use in the ecological economics literature for the study of the environment, and may be close to being tested in the courts (where for a long time the value of emotional damages has been assigned, by judges and juries, by using intuition rather than formal techniques). We return to this later in the essay.

虽然乍看起来很奇怪，但在原则上这些统计方法使研究人员工作在一个确切的深层决定人类福祉出路：他们让我们把受生活影响主体的幸福价值贴上价格标签，这就是研究文献所做的。仍然有许多要学的，但是这样的估值已经进入了生态经济环境的研究文献使用，并可能接近被法院检验（其中很长时间的精神损害价值已分配，由法官和陪审团用直觉而不是正式的技术），我们在文章后面将回到这点。

It might be believed, on the basis of Table 1, that the explanatory power of happiness equations is low. First, the R-squared values in Table 1 are below 0.1, which implies that less than ten percent of the variance of reported happiness has been explained by the independent variables. Second, when compared to the size of one standard deviation of Americans’ reported happiness (which is 0.635), the coefficients on most of these variables look small. But such an attitude may the wrong one. A more appropriate test is on the size and statistical significance of the coefficients on the independent variables. Again an epidemiological example may be appropriate: the R-squared on equations predicting who gets any disease is very small, but since the 1950s it has been usefully realized that diet and smoking enter with substantial and statistically significant coefficients.

可能相信，在表1的基础上，这对幸福方程的解释能力低。首先，在表1中的R平方值低于0.1，这意味着只有不到百分之十的幸福感差异已由独立变量解释。其次，相比一个标准差的美国幸福感（即0.635），这对大多数变量系数看起来很小。但是，这种态度可能是错误的。一个比较合适的测试是对独立变量系数大小和统计有意义。另一个流行病例子可能恰当：对谁得到任何疾病的预测方程R平方是非常小，但自从20世纪50年代以来，它已被有效地认识到，饮食和吸烟成为重大和显著性系数。

A fundamental question, for both social science and public policy, is whether the quality of life in a country like the United States is improving through the years. What do the data say? They do not paint a particularly encouraging picture.

对社会科学和公共政策而言，一个根本的问题是一个像美国这样的国家，其生活质量是经过多年的改善。数据说明什么？他们没有画一个特别令人鼓舞的图画。

Table 2 describes the mean level of happiness in the USA in approximately every year between 1972 and 2008 (the most recent year for which data are available). As can be seen, just by inspection from Table 2 and without the need for statistical methods, Americans’ happiness is not rising. In the early to mid 1970s, one third of people said they were very happy with life; by the mid to late 2000s, that level of reported happiness was the same as, or, if anything, a little lower than, three decades earlier.

表2描述了从1972至2008年（最近一年的数据）间大约每年美国幸福的平均水平。正如可以看到的只是由表2检查而不需要统计方法，美国人的幸福是没有上升。在20世纪70年代初期至中期，三分之一的人认为他们的生活非常快乐；到2000年中后期，所报告的幸福水平与三十年前一样，或者稍微低点，如果有的话。

It might be thought that this is open to a killer objection. Perhaps humans will always renorm their answers as the years pass, which would make trends impossible. But the evidence does not greatly support such a view. For example, it is known that American women have become steadily less happy through time (Blanchflower and Oswald 2004, Stevenson and Wolfers 2009).

这可能被认为是公开对一个杀手的反对。随着岁月的过去，或许人类总是重新规范他们的答案，这种趋势是不可能的。但证据并不十分支持这种看法。例如，众所周知，随着时间推移，美国妇女变得越来越没那么开心（布兰奇福劳和奥斯瓦尔德，2004年；史蒂文森和沃尔弗斯，2009年）。

A more detailed kind of well-being equation is given in Table 3. In this case, the sample size is larger, at approximately one third of a million Americans in year 2009 (using the Behavioral Response Factor Surveillance System, organized by the CDC, the American Centers for Disease Control). Two well-being variables are available in BRFSS data: life satisfaction and reported days of bad mental health. The questions’ wordings are, respectively, “In general, how satisfied are you with your life?” Here people are able to answer one of the following: Very Satisfied, Satisfied, Dissatisfied, or Very Dissatisfied. “Thinking about your mental health, which includes stress, depression, and problems with emotions, for how many days during the past 30 days was your mental health not good?” The means (SDs) for these are 3.37 (0.63) and 3.35 (7.68).

表3给出了更详细的一种福祉方程。在这种情况下，样本较大，2009年大约是一百万美国人的三分之一（使用行为反应的因素监测系统，由疾病预防控制中心和美国疾病控制中心举办）。两个福祉变量在BRFSS数据中是可用的：生活满意度以及所申报的不良心理健康日。这些问题字眼分别是“一般情况下，对自己的生活满意吗？”这里的人们能够回答下列之一：非常满意，满意，不满意，或非常不满意。 “关于你的心理健康，其中包括与情绪压力、抑郁和精神问题，在过去的30天中，有多少天你的心理健康不好？”对这些的方法（SDS）是3.37（0.63）和3.35（ 7.68）。

Because of the large sample size, it is possible in the well-being regression equations of Table 3 to examine the statistical links between feelings of well-being and many of life’s influences, such as, among others:

Taking exercise (positive for well-being),

Being male (negative),

Native American (negative),

Children in the household (negative),

Living as married (positive, though smaller than for marriage),

Self-employment (positive, once income is held constant),

Body mass index BMI (negative),

Eating fruit and vegetables (positive),

Smoking (negative).

由于大样本的大小，表3中的福祉回归方程可能检验联系感情的福祉和许多生活中的影响之间的统计，例如，其中包括：

    体育锻炼（福祉的正面），

    身为男性（负），

    美国原住民（负），

    在家庭中的儿童（负），

    已婚的生活（正，虽然比结婚更小），

    自由职业（正，一旦收入保持不变），

    身体质量指数BMI（负），

    吃水果和蔬菜（正），

    吸烟（负）。

Here, as in the Table 1 results, which used GSS data, there is a marked positive association in Table 3 between income and psychological well-being. Because the relationship between utility and income is so fundamental to economics (and parts of management science and psychology), it is illuminating to study what is implied by the income coefficients in the lower part of the first column of Table 3. For example, the life-satisfaction coefficients rise from 0.0883 for the average level of American income (of between $20,000-$25,000) to 0.3044 for the higher income band of over $75,000 a year. This is a (major) difference of approximately 0.22 life-satisfaction points. It corresponds in size to the difference in life satisfaction between being married and been single, and is nearly as large in absolute size as the consequences for well-being of being unemployed. Thus, as a matter of correlation, it appears that money buys ‘happiness’, and not in negligible quantities.

在这里，如表1的结果，它使用安全总局的数据，与表3在收入和心理健康之间有一个显著正关联。由于公用事业和收入之间的关系对经济学（管理科学和心理学的部分）是如此的至关重要，它是研究在表3的第一列下部的收入系数暗示是什么具启发性。例如，生活满意度系数从0.0883的美国平均收入水平（20,000 - 25,000美元之间）上升为0.3044的较高的7.5万美元以上年收入频段，这是约0.22生活满意度的（主要）差异。它在结婚和已单身之间生活满意度差异大小，几乎和正被失业所带来的后果绝对大小同样大。因此，作为一个相关的问题，看来金钱购买的幸福，在数量上不是微不足道。

What is the link, in the United States, between education and life satisfaction? Here, the first two columns of Table 3 seem particularly interesting. In column 1, where there is no control for the level of income, life satisfaction is higher the higher is a person’s level of education. But in column 2 of Table 3 that disappears: there are then no statistically significant effects from the education variables.

教育与生活满意度，什么是美国的联系？在这里，表3前两列显得特别有趣。第1列中，收入水平没有控制，一个人的教育水平越高，其生活满意度也更高。但在表3 的第2列中，这种情况会消失：教育变量再无显著影响。

This is not as paradoxical as might initially be thought. What happens, in moving between column 1 and column 2 of Table 3, is that the statistical significance switches from education to income. The latter variables start to work strongly; yet the coefficient on College Graduate now drops to 0.0221 (with a t-statistic of only 0.74). Hence the extra satisfaction in life that is associated with greater education is, according to these US equations in Table 3, coming solely from the extra income that education brings. Interestingly, the coefficient on the black demographic variable is positive here, in Table 3 for the modern BRFSS data; whereas it was negative in our earlier GSS results. That is not easily explained but is consistent with the possibility that racial discrimination has declined.

这并不像最初可能是矛盾的想法。会发生什么，在表3列1和列2之间的移动，是从教育到收入统计重要性的切换。后者的变量开始工作强烈，但现在高校毕业生系数下降到0.0221（T-统计量仅为0.74），因此，在生活中额外的满意度是与更多教育相关，根据表3这些美国方程，额外的收入来自纯粹的教育。有趣的是，黑人人口变量的系数在这里为正，表3为现代BRFSS数据，而这在我们早期的GSS结果为负。这不容易解释，但与种族歧视下降的可能性是一致的。

The two right-hand columns of Table 3 give the findings for an alternative well-being variable, ‘Days of bad mental health’ in the previous month, where survey respondents can in the survey give answers from 0 (no days) to 30 (every day). This variable might be thought of as a crude measure of mental strain or depression. Most of the coefficients enter with a sign consistent with that in the life-satisfaction equation of Table 3. The only clear exception is for Male. Men are less satisfied on average, but also have fewer bouts of poor mental health. It is known from the psychiatry literature that females seem to suffer mental health problems more than males. A possible reconciliation of these -- seemingly contradictory -- facts is that women may be happier on average than men but also have a psychological-illness distribution that is more skewed to extreme values.

表3右手边两列给出了另一种福祉变量的结果，在之前的一个月，其中受访者在调查中可以给出从0（无天）到30（每天）“不良心理健康日”的回答。这个变量可以被认为是精神紧张或抑郁粗浅的衡量标准。大多数的系数进入的标志与表3的生活满意度方程一致，唯一的确切例外是男性。男性平均的满意较少，但也有少量心理健康欠佳的较量。从精神病学文献中了解到，女性似乎受到的心理健康问题比男性多。对这些可能的和解——看似矛盾的事实是，女性平均可能比男性快乐，但也有一种心理疾病的分布是比较偏向极端值。

 

3. Well-being across Different Countries in the World

世界上不同国家的幸福感

 One of the interesting facts about recent happiness research is how its empirical findings have been to found to generalize across countries. To illustrate that, Table 4 moves to international data. It sets out European well-being regression equations. The sample size is approximately 35,000 randomly selected individuals across 31 nations (from Austria to Macedonia in Table 4, where Austria is the base country against which comparisons are made) using data taken from the 2007 European Quality of Life Survey. Using these data, both happiness and life-satisfaction equations can be estimated.

关于最近的快乐研究有趣的事实之一是，实证研究结果发现如何跨国家进行概括。为了说明这个，表4换成了国际数据，它规定了欧洲福祉的回归方程。样本是在31个国家的大约35,000随机选择的个人（表4中是从奥地利到马其顿，其中奥地利为作比较的基准国家），使用从2007年欧洲生活质量调查获取的数据。利用这些数据，幸福感和生活满意度方程就能进行估计。

We learn at least three things from Table 4. One is that the statistical structure of wellbeing in the European nations looks almost exactly the same as in the United States. The same variables enter, and in almost identical ways (to see this, compare Table 4 against the earlier Table 3). It may be this robustness that has stimulated so much recent happiness work of the same kind; researchers in country Y have found, once they started to interrogate their own nation’s data, that they could replicate the conclusions from happiness research on other countries. We ourselves stumbled on this in the 1990s, though we had started such work with data sets only on Britain and the USA.

我们从表4学到至少有三件事。一是欧洲国家的福祉统计结构看起来几乎与美国完全一样。相同的变量输入，并且是几乎相同的方式（，对比表4和表3就可以看到这一点）。这可能是这个已经激发了近期同类这么多快乐工作的稳定性，在Y国的研究人员发现，一旦他们开始审问自己国家的数据，他们可以复制其他国家的研究幸福的结论。我们自己在20世纪90年代无意中发现这一点，虽然我们已经开始的数据等工作只在英国和美国套用。

The second is that it makes little difference whether the well-being measure is happiness or life satisfaction; the four columns of Table 4 illustrate that. In each variant, the qualitative structure of the two kinds of equations is the same.

第二是福祉的措施是否是幸福或生活满意度并没有太多区别，表4的第四列可以说明。每一个变量，在两种方程里的定性结构是一样的。

The third is that it seems possible to assess the happiness levels of different countries in comparable ways. This is achieved by examining regression-equation coefficients for different nations. Doing this, Table 4 paints what in the research literature is now, to researchers, a familiar cross-country pattern (seen by reading off the country-dummy coefficients).

第三是似乎可以用类似的方式来评估不同国家的快乐程度。这是通过检验不同国家的回归方程系数来实现的。这样做，现在研究人员在表4中描绘出研究文献里有什么，一个熟悉的跨国模式（通过阅读以国家为虚拟系数看到）。

In Table 4, which draws upon data from 2007, the particularly satisfied nations include

Denmark (coefficient 1.4820)

and the rather dissatisfied European countries include

Bulgaria (-1.8028)

As is clear from this list, and Table 4 more broadly, East European transition nations have citizens who are particularly unhappy with their lives.

在表4，根据来自2007年的数据，特别满意的国家包括：

丹麦（系数为1.4820）

瑞典

芬兰

挪威

卢森堡

荷兰

爱尔兰

   和相当不满意的欧洲国家，包括

保加利亚（-1.8028）

匈牙利

马其顿

拉脱维亚。

   很明显，从这个列表中和更广泛的表4，东欧转型国家的公民，对他们的生活感到特别不满。

What might be viewed as remarkable, and was unpredicted by researchers, is the relatively poor standing -- in a well-being league table -- of certain West European nations such as Italy (-0.3473) and Portugal (-0.4631). Currently it is not known why these nations come out so low.

什么可能被视为显著、研究人员不可预测的，是在一个幸福排名表相对较差地位的某些西欧国家，如意大利（-0.3473）和葡萄牙（-0.4631）。目前还不知道，这些国家为什么会这么低。

For some years now, the Human Development Index, or HDI, has been promoted by the United Nations. In a sense, it was an early attempt to go beyond the tradition of viewing GDP as the sufficient statistic for well-being. The HDI index is an amalgamation of three kinds of data: on real income; lifespan; and education. Although conceived independently, the HDI index links intellectually to the newer empirical research on happiness. Figure 1 shows that, across European nations, there is a reasonably close correlation between HDI and subjective wellbeing.

一些年来，人类发展指数，或HDI，已晋升了联合国。从某种意义上说，它是早期试图超越的观赏作为福祉充分统计国内生产总值的传统。人类发展指数的三个指数是各种数据融合：对实际收入，寿命和教育。虽然构思独立，理智的人类发展指数指数的链接到新的对幸福感的实证研究。图1显示，在欧洲国家之间有一个合理的HDI和主观幸福感密切相关。

Where does the United States lie in a world ranking of happiness levels? To answer that, it is necessary to have cross-national data on statistically representative samples of the population collected in the same way in every country. Such data sets are now available. One source (others are the World Values Survey and recent Gallup Poll data) is the International Social Survey Programme, or ISSP.

美国在世界的幸福水平排名在哪里？要回答这个问题，有必要对每个国家用同样方式收集到统计的代表性的人口样本的跨国数据，这些数据组现在是可用的。一个来源（其他是世界价值观调查与最近的盖洛普调查数据）是国际社会调查方案，或资讯系统策略计划。

In this ISSP case, the well-being question is: “If you were to consider your life in general these days, how happy or unhappy would you say you are, on the whole? [4] Very happy; [3] Fairly happy; [2] Not very happy; [1] Not at all happy.”

在这个资讯系统策略计划的情况下，福祉的问题是：“如果要你考虑这些一般的日子里，就整体而言，你的生活快乐或不快乐？ [4]非常快乐；[3]相当快乐； [2]不太快乐；[1]不快乐。”

Table 5 reports an ISSP international happiness equation, and again codes the dependent variable in a cardinal way. It covers data on 33 nations. In this case, Australia is set as the base country against which coefficient comparisons are made. The ‘happiest’ countries are

Ireland (coefficient 0.2196)

Switzerland (0.1677)

Mexico (0.1559)

USA (0.0939)

Great Britain (0.0844)

New Zealand (0.0754)

And the least happy are

Russia (-0.6096)

Bulgaria (-0.4958)

Latvia (-0.4257)

Croatia (-0.3718).

Hence the USA does very respectably. Other interesting patterns in Table 5 include the fact that Finland and France come out relatively poorly, on -0.2285 and -0.3285, respectively. South Korea also appears low down a well-being scale, with a country dummy coefficient of -0.3483.

表5报告了国际资讯系统策略计划的幸福方程式，并再次用一个基本的方式描述代码中的因变量。它涵盖33个国家的数据，在这种情况下，澳大利亚被设置为基准国家对这些系数进行了比较。“最幸福”的国家是：

爱尔兰（系数是0.2196）

瑞士（0.1677）

墨西哥（0.1559）

美国（0.0939）

英国（0.0844）

新西兰（0.0754）

    和最不幸福的

俄罗斯（-0.6096）

保加利亚（-0.4958）

拉脱维亚（-0.4257）

克罗地亚（-0.3718）。

    因此美国很体面。表5中其他有趣的方案，包括一个事实，即芬兰和法国得出相对较差，分别为-0.2285和-0.3285。韩国也出现了一个低福祉规模，-0.3483的国家虚拟系数。

As in the earlier tables, it can be seen in Table 5 that age enters in a U-shaped way (and Figure 2 shows this even in raw European data, without any control variables); disabled and unemployed variables are associated with large negative happiness values; once more, marital status variables are prominent. Being religious is positive in a happiness equation. Having a university degree comes in large and positive (there is no income variable in this ISSP equation). Interestingly, in the full sample in column 1 of Table 5, there is no gender effect on happiness. Separate equations for males and females are also given. They have the same structure, so it might be said, put loosely, that the determinants of happiness for men and for women are almost identical.

如前面的表，可以从表5看出年龄在进入U型的方式（图2显示这个，甚至是没有任何控制变量的欧洲原始数据），残疾人和失业者变量与大的负幸福价值观相关；再次，婚姻状态变量突出。宗教在幸福方程式里为正。有一个大学学位会有大正（在资讯系统策略计划方程式里没有收入变量）。有趣的是，表5第1栏的全部样本，幸福没有性别差异。给出的男性和女性的独立方程式，它们具有相同的结构，所以可以松散的说，男性和女性对幸福的决定因素是几乎完全相同。

Table 6 reports on how happiness has changed over the recent years of recession. It provides three separate regressions plus nations’ mean scores. Data are available on life satisfaction in the twenty-seven European Union countries plus Croatia, Iceland (in 2010 only), Macedonia and Turkish Cyprus, in 2007, 2009 and most recently in May 2010. It is apparent that there has been a decline in the happiness levels in Greece and Portugal, measured both by a decline in their mean scores and also in the relative rankings in the happiness regressions. These countries have arguably been especially affected by the sovereign debt crisis and rising bond yields, and their governments have implemented severe fiscal austerity measures. Interestingly, the happiness level of the Irish increased over time, as it did in the UK as well as in Germany and Sweden. It remains to be seen whether happiness levels will remain at such levels in the UK and Ireland (as the countries’ recent austerity measures start to bite). We also find evidence that the well-being of the least educated and the unemployed, relative to the employed, has fallen sharply. The happiness of men relative to women has also fallen in the recession.

表6报告了幸福在近年来经济衰退时期是如何变化的。它提供了三种不同回归方程的多个国家平均得分。数据是二十七个欧洲联盟国家加上克罗地亚、冰岛（仅2010年）、马其顿和土耳其塞浦路斯在2007年、2009年及最近的2010年5月的生活满意程度。很明显，希腊和葡萄牙的幸福水平出现了下降，测量他们的平均得分和幸福回归方程的相对排名都在下降。这些国家可以说是特别受主权债务危机和债券收益率上升，以及他们的政府已经实施了严厉的财政紧缩措施。有趣的是，爱尔兰的幸福水平随着时间的推移而提高，在英国、德国和瑞典也一样。这还有待观察幸福水平是否在英国和爱尔兰将保持等级别（如这些国家最近的紧缩措施开始实施）。我们还发现证据表明，相对于就业，最低教育程度和失业者的福祉已急剧下降。相对女性的男性幸福在经济衰退时期也下降了。

What of feelings of well-being in the workplace? Although there has been a long history to the study of data on job satisfaction, comparable estimates across nations are rarer.

在工作场所对福祉是什么感情？虽然对工作满意度已经有很长历史数据的研究，但是跨国家的估计是比较罕见的。

Table 7 provides an illustration. It shows the structure of a job satisfaction regression for the United States. Older people, holding other factors constant, are more satisfied (the coefficient in column 1 of Table 6 is 0.0084, which implies that being 60 rather than 20 is associated with approximately half an extra standard deviation on job satisfaction). There is only a slight difference between males and females: men are less satisfied with their job. An enormous difference is noticeable between self-employed individuals and others. In column 1 of Table 6, the coefficient is a remarkable 0.2, which implies that there is a major satisfaction premium from self-employment. Black people are much less satisfied with work than whites or non-whites. Over time, since the beginning of the data in 1972, there is evidence of a slight downward trend in job satisfaction in the United States (coefficient -0.0015). In Table 6, the highly educated are more satisfied; so too are part-timers.

表7提供了一个例证，它显示了在美国一个工作满意度回归的结构。保持其他因素不变，年纪大的人更满意（表6列1中的系数是0.0084，这意味着60而不是20，对工作满意度与约一半的多标准差相关联）。男性和女性之间只有细微的差别：男性很少对他们的工作感到满意。在自由职业和其他职业之间有着显著的巨大差异。在表6列1中，系数是显著的0.2，这意味着自由职业者有一个大的满意溢价。黑人比白人和非白人的工作满意少得多。随着时间的推移，自1972年开始的数据，在美国有轻微的工作满意度下降的趋势（系数为-0.0015）的证据。在表6中，受过高等教育的人更满意，兼职者也是如此。

Column 2 of Table 7 makes clear that there is a powerful link between job insecurity and low well-being. Those workers who think it ‘not at all likely’ they will lose their job are approximately 0.3 satisfaction points more content than those who believe it is fairly likely. Another negative influence (-0.1079) comes from answering: ‘it would be not at all easy to find another job’. In column 3, which adds a variable for being a union member, there is no statistically significant association between the level of job satisfaction and being in a trade union. What does matter, however, as would be anticipated, is the level of a person’s income. In column 4 of Table 7, it is clear that when the log of annual pay is entered as a control its effect is strong and positive. Moreover, the level of education then loses statistical significance. It seems that educated Americans are more satisfied with their working life principally because their education brings them a higher annual salary. There is a large literature on job satisfaction and productivity that we cannot here review (Judge et al, 2001).

表7列2中明确指出，工作不安全感和低福利之间存在着强大的联系。那些工人认为将失去他们的工作是“根本不可能”的满意点约为0.3，比那些相信这是相当可能的有更多内容。另一个负面影响（-0.1079）来自回答：“这将是并不容易找到另一份工作”。在第3列中增加了作为一个工会成员的变量，对工作的满意程度和作为工会成员之间，在统计上没有显着相关。但是无论如何，正如所预期的，是一个人的收入水平。表7列4中，很显然，当年薪的记录作为对照进入，其效果是强大而积极的。此外，教育程度就失去统计意义，受过教育的美国人似乎更满意他们的工作生活，主要是因为他们的教育给他们带来更高的年薪。我们不能在此回顾（法官等，2001）对工作满意度和生产率的大量文献。

Table 8 provides data on the level and distribution of Americans’ job satisfaction scores through the years. Little change is evident to the eye between 1972 and 2008, even though the earlier regression equation could, after adjusting for people’s characteristics, detect a small secular decline.

表8提供了多年的美国人的工作满意度得分水平的分布数据。在1972年和2008年之间，小的变化是显而易见的，在人们的特性调整后，即使较早回归方程也能检测到一个小的持续下降。

 

4. Interpreting the Patterns

解读模式

 One intriguing fact comes through strongly in these tables and has been found often in the literature. Whether using regression equations or simple tabulations, small or large samples, simple cross-sections or pooled data, a number of the small social-democratic countries of Europe are consistently estimated to be among the world’s happiest nations.

一个有趣的事实是通过强烈的这些表格，并已在文献中经常发现。无论是用回归方程或简单的统计表、小或大样本、简单的横截面或汇总数据，欧洲的一些小社会民主党国家被惯常估计是世界最幸福的国家。

While the multi-country studies’ findings are intriguing, commentators like Ostroot and Snyder (1985) and Argyle (2001) point out that it is hard to know what to make of the cross national claims. First, language differences raise the worry that words like ‘life satisfaction’ cannot be translated sufficiently consistently to ensure that the variations in reported well-being are meaningful. Second, cultural differences -- in some countries it may be less acceptable to admit to unhappiness -- further complicate inference. Third, when visited, these European nations anecdotally appear similar in wealth, and indeed in most ways of living. None of these objections is definitive; but all of them mean that there are doubts over the substantive interpretation of estimated cross-national happiness variation.

而多国研究的结论是有趣的，像奥斯楚蒂和辛迪（1985）及亚皆（2001）评论员指出，很难知道是什么原因造成的跨国主张。首先，语言的差异提高了像“生活满意度”等词不能被充分一致的翻译，以确保在报告福祉的变化是有意义的担心。第二，文化差异——在有些国家，承认不快乐可能不太接受，进一步复杂化了推论。第三，当走访这些欧洲国家，有趣地出现类似的财富，在生活确实如此。这些反对意见都不是明确的，但他们都对估计跨国的幸福差异的实质性解释表示怀疑。

Some writers have nevertheless made a case for taking cross-country subjective wellbeing data seriously in the evaluation of human welfare. This form of research may even presage for international agencies a move away from simple GDP targets of the sort that have been favoured in post-war economic policy. However, to make progress on the construction of a national well-being index, a better empirical justification for the use of subjective life satisfaction and happiness statistics may be needed.

不过，一些作家还是提出在人类福祉的评估上严肃地采取跨国主观的福祉数据，这种研究的形式，甚至可能预示着一个国际机构摆脱了在战后的经济政策中受青睐的单纯GDP目标排序。然而，要一个国家的福祉指数建设进展，为主观生活满意度和幸福的统计数字，可能需要更好的实证理由。

Blanchflower and Oswald (2008b) try to deal with that concern. Their paper takes an unconventional approach to the assessment of well-being. It explores the use of cross-national survey data on hypertension (that is, high blood-pressure). The paper estimates both psychological well-being and blood-pressure equations. Using Pearson and Spearman rank tests, it finds that the structure of the coefficients on country dummy variables is similar in both kinds of regression equations. Happy countries seem to have fewer blood-pressure problems. This has two implications. First, it suggests that there may be a case to take seriously the subjective ‘happiness’ measurements made across the world: they follow a pattern that is reassuringly like the (inverse of) high-blood-pressure estimates. Second, in constructing new kinds of economic and social policies in the future, where well-being rather than real income is likely to be a prime concern, there are grounds for economists to study people’s blood pressure.

布兰奇福劳和奥斯瓦尔德（2008b）尝试解决这一问题。他们的论文需要对福祉评估的一个非常规的方法。它探讨了对高血压病（即高血压）的跨国调查数据。该论文估算了心理健康和血压方程。利用Pearson和Spearman秩检验，发现两种回归方程里国家虚拟变量的系数的结构相似。快乐的国家似乎很少有血压问题，这有两层含义，首先，它表明，有可能出现一种情况认真对待世界各地取得的主观“幸福”的测量：他们遵循一个模式，确切地就像是（逆）高血压的估算。第二，在未来经济和社会政策的新建设中，福祉相对真正的收入，更可能是一个主要关注点，经济学家有理由研究人的血压。

The believability of subjective well-being patterns across nations can be scrutinized in other ways. Various happiness correlates can be studied, for instance, in the search for corroboration. Di Tella et al (2003) show, encouragingly for the quality of subjective data, that in a sample of Western nations there is evidence that the rises and falls in suicide rates move in the opposite direction to changes in happiness. The null hypothesis of no correlation, however, can be rejected only at the 10% level. Bray and Gunnell (2006) demonstrate that suicide is negatively correlated with happiness, and can reject the null of a zero correlation at the 5% level; but in a smaller sample Lester (2002) does not find such clear-cut results. Other national correlates studied in the literature include trust and political institutions (Hudson 2006, Helliwell 2003). None of these articles, though, is an attempt to evaluate the persuasiveness of different measures of mental well-being. There is some evidence that happiness and heart rate are negatively associated among men, and that well-being is correlated with cortisol levels and cardiovascular behaviour (Steptoe et al 2005). However, internationally comparable measures of hypertension, where the demographic and educational characteristics of randomly sampled people are known, are in short supply. Social scientists have paid little attention to the idea that heart variables could play a role as a relevant well-being variable.

跨越国界的主观幸福模式的可信度，可以在其他方面审议。幸福可以是各种相关研究，例如，迪.特利亚等人（2003年）在搜索的佐证中，显示令人鼓舞的主观数据的质量，这在西方国家的样本有证据表明，自杀率上升和下降在幸福中的变化里朝着相反的方向。没有相关的零假设，但是，可以被拒绝的只在10％的水平。 布列和甘内尔（2006年）表明，自杀与幸福为负相关，在5％的水平可以拒绝空的零相关，但在一个较小的样本，莱斯特（2002年）没有找到这样明确的结果。在文献中其他国家的相关研究，包括信托和政治相关机构（哈德森2006年，埃利韦尔2003年）。不过，这些文章没有试图评估精神福祉的不同措施的说服力。一些证据表明，幸福和心率在男性中呈负相关，而福祉与皮质醇水平和心血管行为（世强等人，2005年）相关。然而，高血压在国际上有可比性的措施，那里随机抽样的人的人口结构和教育特色是已知的，供不应求。社会科学家们很少关注心变量可以发挥有关福祉变量的作用的想法。

In the Blanchflower and Oswald (2008b) Journal of Health Economics study, the data set is a Eurobarometer survey which collected identical survey information in September and October 2001 from approximately 15,000 randomly sampled individuals in Denmark, West Germany, Greece, Italy, Spain, France, Ireland, Luxembourg, the Netherlands, Portugal, the United Kingdom, East Germany, Finland, Sweden, Austria, and Belgium.

在布兰奇福劳和奥斯瓦尔德（2008b）健康经济学研究杂志，数据集是欧洲民意调查所收集的2001年九、十月间相同的调查信息，在丹麦，西德，希腊，意大利，西班牙，法国，爱尔兰，卢森堡，荷兰，葡萄牙，英国，东德，芬兰，瑞典，奥地利和比利时抽取约15,000随机样本。

Much research has now been done that attempts to explain the cross-country pattern of subjective well-being. Credit should go especially to Ed Diener and colleagues for analysis on this (Diener et al 1995a, b). At the level of correlation, it has been argued that happy countries have:

Low inequality (Alesina et al 2004; Winkelmann and Winkelmann 2010)

High social capital and strong friendship networks (Bjornskov 2003; Vermuri and Constanza 2006; Bjornskov et al 2008)

Low unemployment and inflation (Di Tella et al 2001, 2003; Helliwell 2003; Gandelman and Hernandez-Murillo 2009; Whiteley et al 2010)

High levels of democracy and democratic participation (Frey and Stutzer 2000; Helliwell and Huang 2008)

High trust (Hudson 2006)

Strong welfare states and public spending (DiTella et al 2003, Pacek and Radcliff 2008;  Kotakorpi and Laamen 2010)

Low pollution (DiTella and MacCulloch 2008).

现已做了大量的研究，试图解释主观福祉的跨国模式。伊迪.迪纳和同事特别用信贷去分析这个（迪纳等人，1995年a，b）。在相关层面上，有人指出，幸福的国家有：

    低不等式（埃里森那等人，2004年；温克尔曼与温克尔曼，2010年）

    高社会资本和深厚的友谊网络（布卓斯科，2003年；温缪和康斯坦萨，2006年；布卓斯科等，2008年）

    低失业率和通货膨胀（迪特利亚等人，2001年，2003年；埃利韦尔，2003年；甘蒂曼和赫南迪兹-马利洛，2009年；怀特利等，2010年）

    高层次民主和民主参与（弗雷和斯图泽，2000年；埃利韦尔和黄，2008年）

    高度信任（哈德森，2006年）

    强大的福利国家和公共开支（迪特里等人，2003年，派西克和拉德克利夫，2008年；柯特科皮和拉蒙，2010年）

    低污染（迪特利亚和麦卡洛赫，2008年）。

The Bjornskov et al (2008) paper is of particular interest. It examines the statistical impact of a wide range of cross-country determinants of life satisfaction. To do so, it exploits a database of 90,000 observations in 70 countries. The authors distinguish four groups of aggregate variables as potential determinants of satisfaction: political, economic, institutional, and human development and culture. They then use ordered probits to investigate the importance of these variables on individual life satisfaction and test the robustness of the results with Extreme Bounds Analysis. Their results reveal that “only a small number of factors, such as openness, business climate, post-communism, the number of chambers in parliament, Christian majority, and infant mortality” robustly influence life satisfaction across countries while the importance of many variables suggested in the previous literature is not confirmed. This remains largely true, they argue, when the analysis splits national populations according to gender, income, and political orientation also.

布卓斯科等（2008年）的论文特别令人感兴趣，它探讨了对生活满意度的多种跨国决定因素的统计影响。为此，它利用了70个国家的90,000观测数据库。笔者对满意度潜在决定因素的总体变量区分为四个组：政治、经济、体制及人的发展和文化。然后，他们使用有序亲位，调查个人生活满意度这些变量的重要性，并测试极端界分析结果的稳健性。他们的研究结果表明，“只有少数因素，如开放性、商业环境、后共产主义、议会中的人数分庭，基督教占大多数、婴儿死亡率”强劲的影响不同国家的生活满意度，同时，许多变量的重要性建议在前面的文献中没有得到证实。他们认为，当按性别、收入和政治取向分析分裂国家人口，在很大程度上仍然如此。

Another notable study is DiTella and MacCulloch (2008). The Easterlin Paradox, as explained earlier, refers to the fact that happiness data are typically stationary in spite of considerable increases in income. This amounts, DiTella and MacCulloch argue, to a rejection of the hypothesis that current income is the only argument in the utility function. They then show that the happiness responses of around 350,000 people living in the OECD between 1975 and 1997 are positively correlated with the level of the country’s income, the welfare state and (weakly) with life expectancy; they are negatively correlated with the average number of hours worked, environmental degradation (measured by SO emissions), crime, openness to trade, inflation and unemployment. This analysis is done with controls for country and year dummies. In an original line of argument, the authors show that effects separate across groups in a pattern that appears plausible (e.g., the rich suffer environmental degradation more than the poor). Based on actual changes from 1975 to 1997, they show, only small contributions to happiness can be attributed to the increase in income in our sample. In fact, the actual changes in several of the 'omitted variables' such as life expectancy, hours worked, inflation and unemployment also contribute to happiness over this time period since life expectancy has risen and the others have, on average, fallen. Consequently, the DiTella-MacCulloch study concludes that the unexplained trend in happiness is even bigger than would be predicted if income was the only argument in the utility function. In other words, introducing omitted variables further confirm Richard Easterlin’s nation income-growth-without-happiness paradox.

另一个值得注意的研究是迪特利亚和麦卡洛赫（2008年）。如前所述，伊斯特林的悖论是指幸福的数据通常是固定的，尽管收入大幅度增加。迪特利亚和麦卡洛赫认为，这个数额是对假设的抵制，即当前收入是效用函数的唯一参数。然后，他们表明，生活在经合组织的大约35万人，在1975年和1997年的幸福反应与该国的收入水平、国家福利和（弱）的平均寿命呈正相关；与平均工作时间、环境恶化（以二氧化硫的排放量来衡量）、犯罪、开放的贸易、通货膨胀和失业问题呈负相关。这种分析是通过对国家和年份虚拟变量的控制来完成。作者表明，在原始路线的参数，在一个合理出现的模式（例如，环境退化的富人忍受比穷人更多）对各组分开。根据1975年至1997年间的实际变化，它们表明，仅仅对幸福的小贡献，就可以归因于我们样本的收入增加。事实上，在几个“遗漏的变量”，如寿命、工作时间、通货膨胀率和失业率的实际变化也有利于这个时间段的幸福感，因为平均而言，期望寿命提高，其他则下降。因此，迪特利亚-麦卡洛赫研究得出结论认为，在原因不明的幸福趋势甚至大于预测，如果收入是效用函数的唯一参数。换句话说，引进遗漏变量进一步确认理查德伊斯特林的国家收入增长与幸福无关的悖论。

 

5. The Future and Possible Links with Medical Research

医疗研究的未来与可能联系

 Other kinds of researchers study human well-being. Our instinct is that this social science literature on happiness will slowly join up with a medical and biological literature on physical well-being. Some epidemiological publishing of this kind is already visible; the connections between the mind and the body are known to be important, and well-being research by its nature has to consider both the mental and the physical.

其他类型的研究人员研究人类的福祉。我们的本能是，这对幸福的社会科学文学会慢慢加入对实体福利的生物医学文献上。已经可以看到一些流行病的公布，心灵和身体联系被认为是重要的，福祉本质的研究必须同时考虑精神和物质。

There are at least three reasons why it can be expected that in the future these different literatures and academic disciplines will draw together. One is that the distinction between happiness and mental health is growing ever more blurry (see Hu et al 2007, for example, which shows that mental health scores are measures of positive well-being and thus not just of use to psychiatrists). The second is that the distinction between physical and psychological health is arguably artificial. It is known in the happiness literature that health variables enter in statistically significant ways in well-being equations. The third is that if the health of the body and the health of the mind are connected -- and there is growing evidence of this, including in papers such as Cohen et al (2003), and Ebrecht et al (2004) -- it is scientifically unattractive to try to keep the two strands apart in research.

至少有三个原因可以预料，今后这些不同的文献和学术学科会召集。原因之一就是幸福和心理健康之间的区别日渐模糊（例如见胡适等人，2007年，这表明，心理健康得分是采取积极的福祉，因而不只是用于精神科）。二是在生理和心理健康之间的区别可以说是人为的。这是已知的幸福文献健康变量在福祉方程显着性的方式进入。三是如果身体健康和心灵健康连接——而且越来越多的这方面的证据，包括论文如科恩等人（2003年）和伊布莱慈等（2004年），试图保持在研究上分为两个方向，这在科学上是没有吸引力。

Nevertheless, it is not easy to believe that the convergence of the happiness, mental health, psychology, and medical literatures will occur quickly (even though it would be scientifically valuable if it did). A central reason is that researchers do not have strong incentives to read and cite beyond their own discipline’s journals. Also, although classical statistical theory underpins all work in these fields, different kinds of disciplines have evolved different ways of presenting data and tests. Statistical and linguistic conventions differ from one discipline to another. Compare, for instance, the style of papers in Science, the New England Journal of Medicine, the Quarterly Journal of Economics, and the Journal of Personality and Social Psychology. Few social scientists have had any training in physiology, and few physicians and hard-scientists have been trained in social science. We believe that not many of those who advocate interdisciplinary work have tried to publish in the elite journals of other disciplines than their own; the practical difficulties are intense.

不过，这是不容易相信的幸福的收敛，心理健康、心理融合和医学文献将会迅速出现（即使这将是科学价值）。一个核心原因是，研究人员还没有强烈的动机阅读和引用超越自己学科的期刊。另外，尽管传统的统计理论支撑这些领域的所有工作，不同学科发展出了当前数据和试验不同的方式。不同学科间统计和语言的惯例不同，例如，在科学、新英格兰医学杂志、经济学季刊、以及人格与社会心理学杂志文章风格比较，很少社会科学家们在生理上有任何的训练，很少有医生和硬科学家一直在社会科学的训练。我们认为，强烈的实际困难致使没有多少主张跨学科工作的人，试图在其他学科而不是他们自己的精英期刊上发表。

All this means that the language, sociological, and stylistic differences across academic disciplines act as a set of barriers that will slow the rate of intellectual convergence. Nevertheless, the study of human well-being is intrinsically as multi-disciplinary as it is fascinating. In our opinion, this is an arena in which the hard sciences and the social sciences inherently abut one another. Slowly, perhaps painfully, we will all have to learn to talk to one another.

这一切都意味着，语言、社会学、跨学科的文体差异作为一个障碍，这将减缓智力收敛的速度。然而，人类福祉的研究在本质上是多学科，是引人入胜的。我们认为，这是一个竞技场，其中硬科学和社会科学本质上相互毗邻。慢慢地，也许是痛苦的，我们都必须学会彼此对话。

 

6. Conclusions

结论

Human well-being is of intellectual and personal interest to individuals, social scientists, and policy-makers. Understanding the determinants of something as complex as happiness is difficult; attempts to do so will, inevitably, continue to provoke disagreement. However, in the last few decades, and especially through the 2000s, a new literature -- for which researchers like the economist Richard Easterlin and the psychologists Edward Diener and (the late) Michael Argyle take particularly early credit -- has sprung up. In it, scholars use data on subjective wellbeing to explore the statistical determinants of happiness, life satisfaction, and mental health.

人类福祉是智慧和个人、社会科学家及政策制定者的个人利益。理解复杂的事情，如幸福的决定因素是困难的，尝试这样做将不可避免地继续激起分歧。然而，在过去几十年，特别是2000年后，一种新文献——对此，经济学家理查德.伊斯特林和心理学家爱德华.迪纳及（已故）的研究人员迈克尔.阿盖尔采取特别早的信贷，已如雨后春笋般涌现。其中，学者用主观福祉的数据探索幸福、生活满意度、心理健康的统计因素。

Much of the considerable knowledge that has been gained is currently at the level of correlation. That does not make it wrong or misleading. But it does mean that, as is often true with observational -- rather than experimental -- social science, we have to be cautious before we can go from even very strong patterns in the data to judgments about cause-and-effect (in a form, say, that would be decisively helpful to those involved in policy-making). The very latest style of statistical research, which is seen in scholarly journals more and more frequently, draws on longitudinal well-being data, and on experimental or quasi-experimental methods. This work should gradually, in the way common in modern social-science, lead to a better causal understanding.

已取得的大量知识，大部分目前在相关的水平。这并不错误或误导。但它确实意味着，由于经常与观测事实——而不是实验——社会科学，我们甚至在可以从很强的数据模式去因果（也就是说，这将是一种政策制定涉及的决定性帮助的形式）判断之前，必须谨慎。统计研究等方面的最新风格，这是借鉴纵向福祉的数据，在学术期刊上出现越来越频繁的实验或准实验方法。这项工作应循序渐进，随着现代社会科学的普及，会带来更好的对因果关系的理解。

Currently, the main patterns -- as illustrated earlier in our tables -- that have been replicated persuasively in the data of large numbers of nations are the following. Happy people are disproportionately the young and old (not middle-aged), rich, educated, married, in work, healthy, exercise-takers, with high fruit-and-vegetable diets, and slim. Recent work on BMI and well-being includes Oswald and Powdthavee (2007) and Blanchflower et al (2009). Happy countries are disproportionately rich, educated, democratic, trusting, and low-unemployment.

目前，主要模式——正如前面我们的表中所描述的，已在各国大量的数据令人信服地复制如下。幸福的人是不成比例的年轻人和老年人（不是中年）、富有的、受过教育、结婚、工作、健康、锻炼、具有高的水果和蔬菜的饮食、以及苗条，这是最近在体重指数和福祉上的工作，包括奥斯瓦德和保德斯韦（2007年）和布兰慈弗拉尔等人（2009年）。快乐的国家是不成比例的丰富的、受过教育、民主、信任和低失业率。

Even bearing in mind the latter characteristics, some nations do noticeably well in happiness rankings. Examples include Denmark, the Netherlands and Ireland. Unfortunately, we do not yet know why these countries are so perplexingly happy (Biswas-Diener et al 2010 think that part of Denmark’s secret is that it somehow cuts down on the statistical tail, that is, has low numbers of extremely unhappy individuals). It is difficult to avoid noticing that smallness of a country seems in the data somehow to help it be a happy one. That may be a spurious correlation, or may indicate perhaps that a geographical or low-population country feels genuinely more like a single, friendly unit (one of us recently heard a Danish politician explain Denmark’s high happiness by saying ‘we feel we are all one tribe’). It is likely that many other characteristics of nations matter, at a deep level, for human happiness; our knowledge here is sparse. Nor do we yet understand how the physical health and mental health of nations are connected. That fascinating research area lies almost uncharted. It will surely provide material for hundreds of future PhDs.

即使考虑到后者的特点，一些国家在幸福排名也明显靠前，例子包括丹麦，荷兰和爱尔兰。不幸的是，我们还不知道为什么这些国家是如此令人困惑的幸福（比斯瓦斯-迪纳等人2010年认为，丹麦的秘密部分是，它在某种程度上减少了统计的尾巴，也就是拥有低数字的极为不满个人）。这是不可避免的注意到，这个狭小国家的数据，看起来在某种程度上让其成为幸福的一个。这可能是一个虚假的相关性，或可能表明或许是一个地理或低人口的国家感觉真的像一个单一的、友好的单位（我们中之一最近听说一个丹麦政治家解释说，丹麦的高幸福感是说“我们觉得我们都是一个部落”）。很可能是其他许多国家的特点问题，在深层次上讲，是人类的幸福，我们的知识在这里是稀缺的。但我们也不了解国家的身体健康和心理健康是关联的。这引人入胜的研究领域几乎未知，这必将为今后数百的博士提供材料。

By its nature, this multi-disciplinary research field is, and will remain, one of genuine significance to human society. Almost everyone is interested in happiness.

就其性质而言，这种多学科的研究领域，对人类社会具有真正意义，几乎所有人都感兴趣的幸福，将继续进行下去。