August 20, 2007
The Underworked American?
By Kathy G.
Lots of great posts here, but not a single really wonky one since Friday. Being that this is Ezra’s blog, that’s a state of affairs bordering on scandalous. So I thought I’d write about an important new study on trends in leisure time in the U.S.
Last month Ezra wrote an article lamenting the fact that the U.S. is the only advanced economy in the world that doesn’t guarantee its workers any vacation. That led to a discussion in the comments thread about whether Americans have more or less leisure time than they used to. I strongly implied that we have less (partly because of a well-known book from the 90s that argued this). Well, I was wrong -- it turns out that on average Americans today have significantly more leisure than they did 40 years ago. However, there is growing inequality in leisure, with less educated workers having significantly more leisure time than their more educated counterparts. I’ll explain.
The study, which is by Mark Aguiar and Erik Hurst, appears in the current
Journal of Quarterly Economics Quarterly Journal of Economics. You can find a newspaper article on the study here. The journal the study was published in is subscription-only, but I’d be happy to email a copy to anyone who requests one.
Detailed study results after the jump.
Here are the main results:
-- Using the narrowest definition of leisure, women on average gained 4.9 hours of leisure per week between 1965 and 2003. During that time, their total hours of work (which includes both market and nonmarket work) declined by 7.8 hours.
-- For men, again using the narrowest definition of leisure, leisure on average increased by 6.2 hours per week, and work declined by 8.3 hours.
-- Women’s market work increased by 2.5 hours, while men’s decreased by 12 hours
-- Women’s nonmarket work decreased by 10.3 hours, while men’s increased by 3.8 hours
-- Conditional on having a child, time spent on child care increased by 5 hours per week on average (this figure is not broken down by sex)
-- Over time, there was growing inequality in leisure that mirrors inequality in wages. The most highly educated workers had less leisure and workers with less education had more. For example, in 2003 male college graduates had 0.2 fewer hours of leisure than that group had in 1965, while male high school drop-outs had 12.2 hours more. Among women, college grads had 1.3 more hours of leisure in 2003 than in 1965, while high school drop-outs had 7.9 hours more.
-- What are we doing with all this extra leisure time? Mostly watching more television, apparently.
Critics quoted in the Globe article say one problem with the study is that it does not distinguish between voluntary and involuntary leisure, and I agree that this is a problem. Yes, there’s all that increase in leisure at the bottom of the income scale, but a lot of that comes from people, men especially, who can’t find good jobs. The authors counter that half of the increase in leisure comes from people who are employed full-time.
Others make the point that by focusing what’s going on “on average” we lose sight of the fact that certain groups are working a lot more. One expert in the article points out that there’s been a significant increase in the number of individuals who work more than 50 hours a week, and also the number of couples who work more than 100 hours a week.
Here are some more strengths and weaknesses of the study.
First, the strengths:
-- It’s based on detailed time diaries, which are more reliable than other data that has been used, such as BLS data where subjects are asked to estimate how many hours they worked in the past week (people are notoriously bad at estimating this kind of thing in retrospect).
-- The sample sizes are large – between 1600 and 15,000 for each decade’s worth of data (it’s based on five separate surveys taken in 1965, 1975-76, 1985-86, and 1992-94 and 2003).
-- The authors control for changing demographics, so we can feel more confident that these are real time trends we’re seeing, and not results that are driven by there being relatively more or less of certain groups in the population.
-- They use differing definitions and measures of leisure, market work, and nonmarket work, and their basic results (that on average Americans have more leisure) are robust to these alternatives, no matter how they’re defined.
Now the weaknesses:
-- There’s a question as to whether people who are willing to fill out time use diaries are a self-selected sample. If they have the time to do this in the first place, there’s reason to suspect they may have more leisure than the actual population does.
-- No people over 65 are included, and since many people are retiring later these days, this may have biased the results. I worked as a research assistant on a study of low-wage workers at a retail chain, and a good 6% or so of our monthly sample of about 1500 was over 65, with some workers as old as 86! In the past, I doubt that very many people that old were still working.
-- Although the authors are careful to use alternative measures, it’s often difficult to say what, exactly, is leisure, and what is work. Socializing is classified leisure, but if you go to a party mainly because you want to network with people in your field, isn’t it really more like work? Also, cooking is counted as nonmarket work, but for some people, it’s more like a recreational hobby. There are many other examples of this kind of ambiguity.
-- This isn’t a criticism of this paper per se, but it would be nice if we had more studies like this one. However, this is the only big, recent diary-based time use study in the U.S. that I know of. A study this huge and this detailed is extremely expensive and work-intensive – I mean, can you imagine inputting and coding all that data, from thousands of subjects? In general, though, it’s best to have more than one study or dataset on a particular subject, because if there’s just one there’s always the chance that it’s an outlier. Though the fact that the sample size for this one is so big by and large puts those fears to rest.
Quarterly Journal of Economics, not Journal of Quarterly Economics. The name makes more sense now, eh?
Posted by: Alex F | Aug 20, 2007 2:54:44 PM
Thanks, Alex -- I've fixed it!
Posted by: Kathy G. | Aug 20, 2007 3:12:31 PM
I'm highly suspicious of that finding of increased leisure for less well-off workers. Obviously a big chunk of this is accounted for by inability to find full-time work. My impression in general of the working poor is that they work very hard, often at more than one job. Also, how was commuting time handled? A lot of people spend a lot of time commuting now, and that's certainly not leisure.
Posted by: beckya57 | Aug 20, 2007 5:39:24 PM
Journal of Quarterly Economics? It's the Quarterly Journal of Economics! Journal of Quarterly Economics?
(Listen if you really want to read the Quarterly Journal of Economics, you really have to hate the Romans.)
The only people we hate more than the Romans are the Journal of Quarterly Economics.
Oh yeah, and the Economic Journal of Quarters.
Posted by: anon | Aug 20, 2007 5:56:19 PM
Very interesting stuff. You point out (to put it a little differently) that the increasing gap in income tracks with the decrease in market work for the low-paid. I'd like to see more about how well those match up. The portion that has seen its income increase the most, and lately the only ones to see substantial increases, are only the upper few per cent, whereas the college-educated whose market hours have increased must constitute a far larger group. I'm sure the increase in income for the upper few percent comes much more from investment and higher pay than more work. But I'd like to see how it works for the rest.
Women’s market work increased by 2.5 hours
This part surprises me. I'd think there was a far greater increase since 1965, when far more married women weren't doing market work, or so we all thought. What was all the fuss about women leaving the home for work about?
Posted by: Sanpete | Aug 20, 2007 6:48:41 PM
"No people over 65 are included, and since many people are retiring later these days, this may have biased the results."
Average retirement ages have dropped.
There's been a working version of this paper available for 18 months:
I seem to recall that when I mentioned this in comments here when Ezra first wrote about it that I was told I must be in error.
Posted by: Tim Worstall | Aug 21, 2007 6:40:03 AM
Tim, the fact that the average retirement age dropping is not inconsistent with more people working after age 65. You could have one group retiring really early, along with another group working later, and that could result in a lower average retirement age, even though higher percentages of people are working after age 65.
And since this particular survey only includes people up to 65, that could indeed bias the results. The ones who retire before 65 will be captured in the survey and their retirement will register as a gain in leisure. But the post-65s who are working at greater rates than their age group did in the past will be excluded, so their increased average work hours will not show up.
I doubt that including the older folks would change the overall results that much, but still, it's something to consider.
Posted by: Kathy G. | Aug 21, 2007 8:34:34 AM
Becky, in the study commuting was counted as work. And yes, a problem with the study is that much of the increase in "leisure" for low-income folks is forced leisure due their not being able to find full-time jobs (or any jobs).
Posted by: Kathy G. | Aug 21, 2007 8:38:30 AM
Posted by: judy | Oct 11, 2007 7:39:13 AM
The comments to this entry are closed.