Yves here. This post does a nice job, if unintentionally, of showing the limits of using economics as the lens for viewing well-established social norms and biological urges. Similarly, Keynes’ “animal spirits” was a polite way of acknowledging that most entrepreneurial activity was not terribly logical if you were rigorous about the risks (even now, over 90% of all new businesses fail within 3 years) but people did it anyhow. Moreover, it’s only recently that economists have decided that they are qualified to comment on family and interpersonal matters. For instance, in his 1988 book, Behind the Veil of Economics, Robert Heilbroner made an in-passing reference to family relationships falling outside the economic sphere.
By Angus Deaton, Dwight D. Eisenhower Professor of Economics and International Affairs at the Woodrow Wilson School of Public and International Affairs and the Economics Department, Princeton University, and Arthur Stone, Distinguished Professor and Director of the Applied Behavioral Medicine Research Institute, Stony Brook University. Originally published at VoxEU
Study after study has shown that those who live with children are less satisfied with their lives than those who do not. Is there something wrong with these empirical analyses? Or is it that happiness measures are unreliable? This column argues that the results are correct but that comparisons of the wellbeing of parents and non-parents are of no help at all for people trying to decide whether to have children.
It is a commonplace that new parents are overwhelmed by a “tsunami of love” when they first meet their dependent offspring. Older children, though often a source of irritation and worry, are also a source of joy, and there are few parents who can even bear to think of a world without their children. Yet, study after study has shown that those who live with children are less satisfied with their lives than those who do not; Hansen (2012) and Stanca (2012) are recent surveys. How can this be? Should governments publicise such findings, to help disabuse people of the widespread notion that children are good for them? Perhaps along with Larkin’s lines?
Man hands on misery to man/It deepens like a coastal shelf/Get out as early as you can/And don’t have any kids yourself
— Philip Larkin
Is there something wrong with these empirical analyses? Or is it that, as many economists suspect, happiness measures are unreliable? We argue here that the results are correct, as far as they go. The deeper problem is that comparisons of the wellbeing of parents and non-parents are of no help at all for people trying to decide whether or not to become parents.
One story is that people don’t have much idea of what they are doing. Daniel Gilbert, the Harvard psychologist whose Stumbling on happiness is well-characterised by its title, argues that the belief that children are good for you is transmitted from generation to generation, in spite of its being false. Parents are perpetually surprised by the unhappiness that children bring.
Before we economists reject such an explanation out of hand, we should remember that Adam Smith believed that the “pleasures of wealth and greatness” are a deception, but are necessary to keep “in continual motion the industry of mankind.” Perhaps the attractions of children are also a deception, but are necessary to keep in motion the continuation of mankind.
Our two new studies, Stone and Deaton (2013, 2014), use a large American data set from the Gallup Organization to try to get to the bottom of all this. The first paper focuses on the elderly, the second on parents and children.
If we look at everyone in the population (aged 18 and above), and consider child-at-home status and life evaluation, we get a version of the standard finding.
• People with at least one child at home evaluate their lives slightly less favourably than those with no child; the difference is equivalent to a 5% decline in income.
• Those who live with children also report more anger, stress, and worry, but also more happiness
(It is important in this work to separate life evaluation or life satisfaction on the one hand – wherein people judge their lives as a whole – from the hedonic or emotional experience of happiness on the other, and we shall use term “happiness” to refer only to the emotion, not, as is often done, to life evaluation.)
Using the whole population of adults to make these comparisons is not very useful, especially in the Gallup data, which do not tell us how (or whether) the child is related to the respondent. Among the youngest respondents, the child might be a sibling, and among old respondents, the child might be a grandchild. If we look only at adults aged from 34 to 46, for whom a child at home has a 90% or better chance of being the child of the respondent, life as a parent looks better. These adults with children have better lives than those without – equivalent to a 75% increase in income – and although they still are more likely to experience more sadness, anger, and worry, their lives also contain more happiness, smiling, and enjoyment.
But this is not the end of the story. These adults with children don’t look like adults without children. They are healthier, richer, better educated, more religious, more likely to be female or Hispanic, and less likely to smoke, all factors that improve life evaluation regardless of whether people have children. More important still, they are vastly more likely to be married, and marriage itself promotes wellbeing. If we control for all of these factors, we get back to the negative finding that, even among these selected adults, those with children have worse lives. As we change the control variables, but sticking with the 34 to 46 year olds, we can move from the uncontrolled, positive finding to the highly controlled, negative one. Such comparisons, with at least some controls, are similar to many in the literature.
Interestingly, no matter what the controls, parents experience more positive and more negative emotions. This is credible enough, that children bring both joys and sorrows. And if we look at the elderly respondents, everything looks negative for old people living with children, both life evaluation and hedonic experience. We attribute this mostly to selection. At least in the US, the elderly do not normally live with young children, and when they do, it is likely to indicate an inability to live alone, a conjecture that is supported by the much poorer health outcomes among the elderly who live with young children. We also find some evidence of a direct negative effect of children on the emotional outcomes of the elderly.
At this point, we need to stop and think harder about what we are trying to do. Should we really be controlling for marriage? Many people get married with an eye to having children, so perhaps the right comparison is between those who are neither married nor parents with those who are both. But then what about those who are happily married but do not have children? Similarly, people may work harder when children bring unanticipated expenses. But children do not cause most of the differences in people’s incomes, and it makes no sense to drop income as a control.
The results so far tell us about comparisons of wellbeing and how those comparisons differ depending on what other factors we hold constant. But it is out contention that they have no policy implications, either for individuals or for nations.
What Should We be Trying to Estimate Here?
One target might be whether a specific person or a specific couple has a better life with or without children. Because of selection into parenthood, comparisons between those with and without children are not obviously helpful unless, as psychologists (or Adam Smith) might argue, people have little idea of what they are doing. But it surely makes more sense to think that people who have children are, by and large, those who wanted to have children, and vice versa.
The standard economist’s response to a selection problem is to look for panel data, an instrument, or, under ideal conditions, to run the randomised controlled trial that such methods are supposed to mimic. Panel data are of little help here, because the change in wellbeing for parents at the time of their children’s birth is only a fraction of what we are looking for, and the data show important changes in advance of the event. As for a randomised controlled trial, whatever we are interested in knowing it is not the response of a childless couple to waking up one morning to find that a stray stork has blessed them with a “bundle of joy.” This early morning metamorphosis into parenthood is about as likely to bring a better life as was Gregor Samsa’s metamorphosis into a giant insect. And we need not even trouble ourselves with the companion study in which randomly selected children cease to exist.
If we think that people who have children are people whose lives are better because they have children, while those who are childless have better lives without them, the empirical findings are not at all surprising. Recent work by Benjamin et al (2013, 2014) confirms that people do not exactly maximise their wellbeing when making life decisions, but they often come very close. If so, those whose wellbeing will be improved by having children will have children, and those whose wellbeing would be worsened by having children will not. But the two groups have different tastes –most obviously in their preference for children – so we have no a priori basis for expecting one group to be better off than the other once they have made their choices. Would be parents who cannot have children are certainly likely to be dissatisfied relative to those who do, and people who do not want a child would no doubt be dissatisfied if they accidently acquired one. But non-parents are not failed parents, nor are parents failed non-parents.
And that is about all that we can say.
See original post for references