1 s , September 7 – 8 BPEA Conference Draft , 2017 Where Have All the Workers Gone? An Inquiry into the Decline of the U.S. Labor Force Participation Rate Alan B. Krueg University er, Princeton
2 Conflict of Interest Disclosure: The author received financial support for this work from the Federal Reserve Bank of Boston and the National Institute on Aging . With the exception of the aforementioned, the author did not receive financial support from any firm or person for this paper or from any firm or person with a fi nancial or political interest in this paper. He is currently not an officer, director, or board member of any organization with an interest in this paper. No outside party had the right to review this paper prior to circulation.
3 Where Have All the Workers Gone? An Inquiry into the Decline of the U.S. Labor Force Participation Rate 1 Alan B. Krueger and NBER Princeton University August 26 , 2017 Conference Draft BPEA Abstract The labor force participation rate in the U.S. has declined since 2007 primarily because of population aging and ongoing trends that preceded the Great R ecession. The participation rate has c groups. A rise in school evolved differently and for different reasons , across demographi , enrollment has largely offset declining participation for young workers since the 1990s. Participation in the labor force has been declining for prime age men for decades, and about half of prime age men who are not in the labor force (NLF) may have a serious health condition that is a barrier to work. Nearly half of prime age NLF men take pain medication on a daily basis, and in nearly two - thirds of these cases they tak e prescription pain medication. Labor force participation has fallen more in areas where relatively more opioid pain medication is prescribed, causing the labor force participation and the opioid crisis to become intertwined. problem of depressed The labor force participation rate has stopped rising for cohorts of w omen born after 1960. Prime age men who are out of the labor force report that they experience notably low levels of emotional well - being throughout their days and that they derive relatively little meaning from their daily - being men, by contrast, report similar levels of subjective well activities. Employed and NLF wo , but NLF women who are not primarily takin g care of home responsibilities report notably low - levels of emotional well . Over the past decade retirements have increased by about the same being amount as aggregate labor force participation has declined , and the retirement rate is expected to . A meaningful rise in labor f orce participation will require continue to rise a reversal in the secular trends affecting various demographic groups, and per haps immigration reform. 1 I thank David Cho, and Amy Wickett for outstanding research assistance , and Ed Freeland for Kevin DeLuca indispensable assistance administering the survey used in Section IV B . An earlier version of this paper was th Boston Federal Reserve Bank’s 60 presented at the Economic Conference, October 14, 2016. Financial support was provided by the National Institute of Aging. Larry Katz, Matt Notowidi gdo and Jim Stock provided help ful responsible for all views and any mistakes. is comments on an earlier draft. The author 1
4 I. Introduction he labor for peaked at 67.3 percent in early T ce participation rate in the United States , and has declined at a 2000 continuous pace since then , reaching a near 40 - year low more or less of 62.4 percent in September 2015 (see Figure 1) . Italy was the only O.E.C.D. country that had a lower labor force part . Although the icipation rate of prime age men than the U.S. in 2016 participation rate has stabilized since the end of 2015, evidence on labor market flows – in particular, the continued decline in the ra te of transition of those who are out of the labor force . back into the labor force – suggests that this is likely to be a short - lived phenomenon This paper examines cyclical movements and secular trends in labor force participation particular , with a focus on the role of pain and pain medication in the lives of prime age men who are not in the labor force and prime age women who are not in the labor force not prim arily taking care of and s express the greatest degree of distress household responsibilities, because these group and dissatisfaction with their lives . The paper is organized as follows. The next section summarizes evidence on trends in overall and by various demographic groups. Care ful attention is labor force participation devoted to adjusting labor force and population data for the introduction of the 2000 and 2010 . population controls the Current Po pulation Surve y (CPS) in The main finding of this analysis is that s hifting demographic shares, mainly an increase in older workers, and trends that preceded the Great Recession (e.g., a secular decline in labor force participation of prime age men) can acco the lion share of the decline in the participation rate since th e last business cycle unt for peak . Because most of the movement in the participation rate in the last decade reflects secular examines trends in the participation rate II trends and shifting population shares, Section I 2
5 separately for yo ung workers, prime age men, women , as well as the retirement rate . The and physical and mental limitations , which could pose a barrier to role of employment for health half of prime age me n who are not in the labor force (NLF) , is highlighted and explored. around Survey evidence indicates that almost half of prime age NLF men take pain medication on a daily basis, and that prime age men who are out of the labor force spend over half of as a group their time feeling some pain. A follow - up survey finds that 40 percent of NLF prime age men report that pain prevents them from working on a full - time job for which they are qualified , and that nearly two thirds of the men who take pain medication report taking prescription medication . It is also shown that generational increases in labor force participation that have historically raised women’s labor force participation over time have come to an end, and the U.S. can no s of women to participate in the labor market at higher levels longer count on succeeding cohort ing. The section also documents that an increase in the than the cohorts they are succeed after retirement rate 2007 accounts for virtually all of the decline in participation since then , sugges . ting the persistence of labor force exits S ection I V presents evidence on the subjective well - being of employed workers, unemployed workers, and those by demographic group . Two who are out of the labor force measures of subjective well - being are used: an evaluative measure of life in general and a measure of reported emotional experience throughout the day. Young workers who are not participating in the labor force seem remarkably content with their live s, and report relatively high levels of affect during their daily routines. Prime age men who are out of the labor force, however, report less happiness and more sadness during their days than do unemployed men, although they evaluate their lives in gener al more highly than unemployed men. Prime age and e evaluations in being and lif - older women who are out of the labor force rep ort emotional well 3
6 general that are about on par with employed women the same age, suggesting a degree of contentment that may ma unlikely to see many in this group rejoin the labor force. ke it Given the high use of pain medication by NLF prime age men and women, and the opioid crisis in the U.S. since the early 2000s , Section V provides an analysis of mushrooming the connection , and labor force between the use of pain medication, opioid prescription rates participation. Evidence is first presented suggesting that local opioid prescription practices influence the use of pain medication. C onditional on individuals’ disability status, self - reported health, and demographic characteristics, pain medication is more widely used in counties where more opioid medication. Next, regression analysis finds that healthcare professionals prescribe labo r force participation fell more in counties where more opioids were prescribed, controlling for the area’s share of manufacturing employment and individual characteristics. Although it is un clear whether these correlation s represent causal effects , these findings reinforce concerns Hillbilly Elegy from anecdotal evidence. For example, in his memoir , J.D. Vance (2016, pp. 18 - 19) writes about a recent visit with h in Jackson, Kentucky: “We talked is second cousin, Rick, ve come in,’ Rick told me. ‘And nobody’s interested in about how things had changed. ‘Drugs ha holding down a job.’” findings complement Case and Deaton’s (2017) conclusion that And the “ deaths of despair ” for non - Hispanic whites “ move in tandem with other social dysfunctions, including the decline of marriage, social isolation, and detachment from the labor force.” The conclusion highlig hts the role of physical, mental and emotional health challenges as a barrier to work for many prime age men and women who are out of the labor force. Since – apart from the unemployed – this group exhibits the lowest level of emotional well - being and life evaluati on, there are potentially large gains to be had by identifying and implementing successful interven tions to help NLF prime age men and women lead more productive and fulfilling lives. 4
7 I. Trends in Participation I Figure 1 shows the seasonally adjusted labor force participation rate as published by the ) . In addition, the graph shows alternative estimates of the Bureau of Labor Statistics ( BLS participation rate using labor force and population data that were smoothed to adjust for the in 2003 Census in the C decennial population controls and introduction of the 2000 and 2010 PS 2 intercensal population adjustments introduced in January of each year . 2012, respectively , and e population adjustment s e undoubtedly occurred more grad ually over preceding months and s Th . Compared to the published series, the adjusted series indicates that the labor forc e years rose a bit less in the 1990s recovery, declined a bit more in the 2001 - 07 participation rate recovery, and fell a bit less in the current recovery, but overall the trends are similar. Henceforth, we focus on the adjusted labor force data. The aggregate labor force participation rate series masks several disparate trends for subgroups. Figure 2 shows the participation rate separately for men age 25 and older, women 24. pation rate show partici - The appendix figures age 25 and older, and young people age 16 As is well known, the participation rate for adult . trends further disaggregated by age and sex labor force data in 1948. men has been on a downward trajectory since the BLS began collecting participation of prime late 19 90s , but the decline in a bit steeper since the This trend has been in the labor force is not a new development and was not sharper after the Great age men 3 it was before it (see Figures A4 - A6 ). Recession than Workers age 55 and older are the only age 2 The population controls introduced in 2012, for example, caused an abrupt 0.3 percentage point drop in the labor force participation rate from December 2011 to J anuary 2012, largely because the population of older individuals exceeded the figure that had been assumed in intercensal years. We closely follow the procedures outlined in http://www.bls.g ov/cps/documentation.htm#pop to s mooth out changes in population controls . 3 Charles, Hurst and Notowodigdo (2016 ; 2017 ) provide evidence that the housing boom in the pre - recession period fall in the labor force participation of le masked an even greater ss educated prime age men from 2000 to 2006 due to the collapse of manufacturing. 5
8 notable rise in participation over the last two decades, albeit from a low group that has shown a age group, and the long - running rise in participation for 55 - 64 year old women base for the 65+ after the Great Recession. seems to have come to an end The aggregate participation rate rose in the half century following World War II because 4 the labor force. Beginning in the late 1990s, however, th women increasingly joined abor e l force participation rate of women age 25 and over unexpectedly reached a decade - long plateau, and sinc e 2007 women’s labor force participation has edged down almost in parallel with men’s. The plateau and then decline in women’s labor force participation is responsible for the downward trajectory of the aggregate U.S. labor force participation rate. Although age, cohort we later show that this appears more consistent and time effects cannot be separately identified, with cohort developments than time effects. Lastly, younger workers have exhibited episodic decl ines in labor force participation since the end of the 1970s. After falling sharply toward the end of the Great Recession, t he rate n . The labor force participation participation for younger individuals has stabilized since the bly responds more to the rate of young workers proba business cycle than that of older state of the workers because school is an alternative to work for many young workers in the short run. A. Decomposing the Decline in the Participation Rate At an annual frequency, after adjustments are made for the effects of changing population controls, (see Figure 3) . From 1997 to the labor force participation rate reached a peak in 1997 2017:H1 , the aggregate participation rate fell by 4 .2 percentage points, with most (2.8 poin ts) of 4 World War II labor supply. - See Goldin (1991) for an analysis of women’s post 6
9 5 the decline occurring after 2007. Several studies have shown that shifting demographics, a older population, are responsible for around half of the decline in labor force n mainly toward 6 participation. effects of shifting demographics, write the aggregate labor force participation T o see t he in year t denoted ℓ rate , , as : 푡 푝 푖푡 ∑ ∑ = (1) ℓ ( 푤 ℓ ℓ ) = 푡 푖푡 푖푡 푖 푖푡 ∑ 푝 푖푡 푖 is the participation rate for group i , 푝 ℓ is the size of the population of group i in year t, where 푖푡 푖푡 푤 and is the population share of group i. 푖푡 year t can be written as: - k and between year t The change ∑ ∑ ∑ ∑ ℓ = ℓ ∆ ℓ 푤 + (2) ∆ 푤 ℓ 푎푛푑 ∆ ℓ = ∆ 푤 + ∆ ∆ 푤 ℓ , 푖푡 푖푡 − 푘 푖 푖푡 푖 푖푡 푖푡 − 푘 푖푡 or a component due change in rates within groups (weighted by starting or ending period to the , ) and a component due to changes in population shares ( weighted by ending or population shares rates ) . staring period participation on rate and population shares for 16 ge - by - sex a Table 1 reports the labor force participati 7 groups. There are no table declines in the participation rate for young workers, both male and The population shares have also shifted over time: the share of the populat ion age 55 female. from 26.3 percent to 35. 6 percent and over rose from 1997 to 201 7 , while the share age 25 to 54 ∑ fell from percent. The bottom two rows report 57.5 percent to 49.3 ℓ where the 푤 , 0 푖푡 푖 5 Data for 2017 are only available for the first six months of the year as of this writing. Because the aggregate participation rate is not very different over the first six months and full year in the past, we do not make an here . adjustment for seasonality 6 See CEA (2014) for an excellent survey of the literature . Fernald et al. (2017) further expand the shift - share analysis by disaggregating cells by education, race and marital status. They find that from 2010 to 2016, two thirds of the decline in labor force participation occurred within groups, and one third due to the shift across groups. It is possible that membership in some of the categories, such as marital status, is endogenously determined, however. 7 We use annual data because seasonally adjusted smoothed population controls are not available for each g roup. Data for 2016 are the average of the first eight months of the year. In earlier years, the average of the first eight months of the year was close to the annual average, so no adjustment is made for seasonality. 7
10 population weights are for either 1997 or 2017. ted to ward In general, the population has shif , and this accounts for well over half of the decline in the groups with lower participation rates . Using the decompositions in equation (2), the shift in the labor force participation rate account for 65 percent (=[65.6 - 62.8]/ [67.1 - 62.8]) or 88 percent (=[67.1 - population shares can - of the decline in labor force participation from 1997 to 201 7 , d epending on 62.8]) 63.3]/[67.1 7 population shares are used to weight changes in each group’s participation whether 1997 or 201 . Clearly , the changing age distribution of the population has had a major influence on the rate . However, the decline in the participation rate of young workers , labor force participation rate popu is also quantitatively important. Regardless of whic h year’s lation especially young men, shares are d as weights , the decline in participation of young men (age 16 - 24 ) from 1997 to use or 7 accounts for almost 201 one quarter of the decline in the overall participation rate , about triple their current share of the population . A limitation of these decompositions is that there is no counterfactual comparison and no other factors considered apart from demographics . hanging population shares Furthermore, c just accounting identities could affect participation of different groups. These calculations are that highlight the potential magnitudes of various shifts in population groups. B . Continuation of Past Trends ? he decline in the participation rate was faster in the last decade than in the As mentioned, t preceding one. We next examine the extent to which the 2.8 percentage point decline in the labor force participation since the start of the Great Recession represents a continu ation of past combined with shifts in population shares, trends that were already in motion, or a new development. Specifically, for each of the 16 groups in Table 1 we estimated a linear trend from 8
11 8 1997 to 2006 This ten year period was chosen be cause it encompasses the pre - Great by OLS . 9 We then extrapolate from th e past Recession downward trend in labor force participation. trend over the next decade. To the extent that secular trends were affect ing decade’s for various groups before the Great Recession (e.g., education rising for participation trends some groups, and in turn affecting the trend in the participation rate) this approach would reflect those developments. The appendix figures show the trends for each subgroup, where the in 1997. intercept has been adjusted so the fitted line matches the actual participation rate est negative forecast residual compared with the previous The group with the bigg - 64 year old women, who were predicted to experience a 9 percentage point decade’s trend is 55 rise in their participation rate but actually experienced little change from 2007 to 2017 (see Table 1 a 15). Younger workers saw a slower downward trend in the 2007 - 17 nd Appendix Figure A - period than in the preceding decade. In general, there was a form of mean reversion, with the groups with the sharpest downward (or upward) trends from 1997 2006 experie ncing more - moderate downward (or upward) trends in the ensuing decade. he green line in Figure 3 aggregates across the group specific trends using fixed 199 7 T for each year. The red line uses the actual population shares each year to population shares 10 eight the group’s predicted participation rate to derive an aggregate rate . w T he difference s between the red and the green lines highlight the importance of shifting population shares . 8 quadratic trend fits the aggregate data better than a linear one, in 7 of the 16 Although Tables 2 and 3 suggest a subgroups the quadratic term is insignificant in the period 1997 - 2016, and a linear trend does not do much injustice the data for the other groups for describing Over such a short period, the linear extrapolation could be thought of as . a first - order approximation to a more complicated trend. 9 If a 7 - year sample period is used the results are similar and if a 15 - year period is used the trends are mostly flat. 10 each group he predicted participation rate is the weighted sum of Formally, t ’ s predicted participation rate based ̂ actual share of the population in the year : ℓ on the linear trend for that group, where the weights are the group’s = 푡 ̂ ̂ ∑ estimated linear trend. ℓ OLS 푤 , where ℓ on an extrapolation from the is based 푖푡 푖푡 푖푡 9
12 Figure 3 makes clear that the lion’s share of the decline in labor force participation since the start of the Great Recession is consistent with a continuation of past trends and shifting xtrapolating from - 2006 trends for each group and weighting by population shares. E the 1997 1997 population shares leads to a forecast that the labor force participation rate would have from 2007 to 201 fallen by as a result of pre - existing trends , or about one percentage point 7 the actual decline . Shifting population demographics can account for around 40 percent of maining gap. A similar conclusion holds if one looks at the period from 1997 almost all of the re to 2017 period: around 40 percent of the decline in labor force participation rate over the last two decades is predicted by applying the various demographic groups’ linear trends , and almost all of the rest can be attributed to shifting population shares. C of a Cyclical Recovery Should Be Expected? . How Much A key question for economic policymakers is the extent to which labor force participation two - decade long decline. As emphasized so far, most of the decline in the can recover from its (anticipated) result of an aging population and - group participation rate since 2007 is the specific 11 participation These tren ds could trends that were in motion before the Great Recession. strengthen or reverse, but an aging workforce is likely to put downward pressure on labor force participation for the next decade s . To the extent there was a cyclical negative shock to two participation, however, one might expect some recovery in the near term. The 0.6 percentage point rise in the (seasonally adjusted) participation rate from September 2015 to March 2016 gave some hope that a cyclical recovery might be taking pla ce. that there Three consi me to suspect will be only a limited and short - lived cyclical derations lead 11 the CEA (2007; Table 1 - 2 and Box 1 2), for example, predicted a 0.2 to 0.3 percentage point annual decline in - labor force participation rate from 2007 to 2012 because of the aging of the baby boom cohort. See also Aaronson, et al. (2006). 10
13 recovery , however. First, Fernald et al . (2017) find that by 2016, the cyclical in participation the lag component of the fall in labor force participation was essentially dissipated, regardless of they allow for. Second, the seasonally adjusted l abor force participation rate structure has displayed no trend since March 2016 cyclical recovery may already be over , , suggesting that the nald et al. ’s conclusion . consistent with Fer Third, throughout the recovery there has been no rise in the rate of transition of those who re out of the labor force joining the labor force. T he likelihood of transitioning into th e a labor force from out of the labor force edged down throughout the recovery , including in late 2015 and early 2016 when the participation rate retraced 0.6 percentage point (figure available on request) . Nonparticipants are increasingly a group with a lower likelihoo d of moving into the labor force. Thus, the idea that many labor force dropouts are returning to the labor market is unsupported by the data. Instead, the labor force participation rate rose in late 2015 and early 2016 because unemployed workers stayed u - term unemployed nemployed longer, especially long H istorical ly, there is no tendency for workers. transitions from out of the labor force into the labor force to be have cyclical l y (see Krueger, Cho and Cramer, 2014). Given the pre - existing downward trend in participation for most demographic groups and the aging of the U.S. for a time may population, stabilization in the labor force participation rate If a cyclical recovery in labor force represent the best one could expect for a cyclical recovery. participation is unlikely, then a reversal of secular trends toward declining labor force is the only way to achieve an increase in labor force participation. The next section focuses on secular trends toward nonparticipation for key demographic groups. I II . Secular Trends for Specific Group s Young Workers A. 11
14 Young pe ople have exhibited the largest decline in labor force partici pation in the past offset by t two decades. extent This is their increased school enrollment, o a considerable however. Figure 4 displays trends in the nonparticipation rate separately for young men and young women (age 16 - 24) from 1985 to 2016 . The share of young workers who were neither employed nor increased significantly from 1994 to 2016. In 1994, 2 9.7 percent looking for a job of young men were not participating in the labor force, a nd in 2016 this figure was 43. 0 percent. Nonparticipation in the labor force also rose for young women. However, if we remove individuals who were enrolled in school in the survey r different. eference week, the story is quite The bottom two lines in Figure 4 show the percent of men and women in this age group who were idle, defined as neither enrolled in school nor participating in the labor force. Young men still display an upward t only rose from 7.3 percent to 8.9 rend, but the share who were idle percent from 2004 to 2016, while the trend for women is downward (from 15.8 percent to 12.1 percent). A rise in school enrollment has therefore helped to offset much of the decline in rticipation. Given the significant increase in the monetary return to education that bega pa n in the early 1980s, this development could be viewed as a delayed and overdue reaction to economic incentives. Working Age Young Men (21 - 30) Aguiar, et al. (201 6) highlight the rise in non - work and non - school time by young men age 21 30, especially those with less than a college education. The share of noncollege educated - youn g men who did not work at all over the entire year rose from 10 percent in 1994 to more than 20 percent in 2015. They propose the intriguing hypothesis that the improvement in video for young men downward shift , contributing to a game technology raised the utility from leisure 12
15 12 in labor supply and response to wages . a more elasti While Aguiar and his coauthors are clear c out that demand contributed side factors may also have to point to the decline in work hours of - , and that their estimates of the shift in the labor supply curve due to changes in young men lei sure technology for video and computer games only account for 20 to 45 percent of the observed decline in market work hours of less educated young men, their hypothesis has generated keen interest. Here we briefly examine their video game hypothesis by co mparing the self reported emotional experience during video game playing, television watching, and all - activities, as well as more standard labor force, school enrollment and time use data. Preliminarily, we note that the CPS data indicate that from Octo ber 1994 to October 2014, the labor force participation rate of men age 21 - 30 fell by 7.6 points, from 89.9 percent to 82.3 percent, and this was partially offset by an increase i n school enrollment. Idleness – defined as being enrolled in school, employed, nor looking for work neither – rose by 3.5 percentage points over this period. Table 2 reports the amount of time that 21 - 30 year old men spent engaged in selected 13 activities per week in 2004 - 11, and 2012 - 15. Market work hours declined by 3.1 07, 2008 - (9 percent) from 2004 - 07 to 2012 - 15. An increase in time devoted to education hours per week over this period more than (1.3 hours), playing games (1.7 hours), and computers (0.6 hours) young rking. If we limit the sample to men who were out of offset the decline in time spent wo the labor force (not shown) , time spent on education increased by an impressive 5.3 hours, or 38 percent . Time devoted to education activities did not increase for NLF young men with a high 12 Technically, their time use measure pertains to all game playing. We follow their precedent of referring to the game playing activity in the ATUS as video game playing, as the increase in time devoted to this activity most likely is overwhelmingly the result of video game playing. 13 not add up to 168 hours because some The total amount of time per week sp ent in the listed activities does categories, such as travel, are omitted. 13
16 schoo l education or less, but conditioning on low education downwardly bias any increase would Time spent playing video games by young in school enrollment in this age group over time. NLF - 07 to 6.7 hours per week 2012 - men rose from 3.6 hours per week in 2004 5, while time spent 1 watching television fell from 23.7 to 21.7 hours over this period. As Aguiar, et al. conclude, video gaming is clearly drawing more attention from this group over time. The 2010, 2012, and 2013 ATUS’s included a supplement on subjecti ve well - being modeled on the Princeton Affect and Time Survey (see Krueger, et al., . Specifically, for 2009) to report on a scale from 0 three randomly selected episodes of each day, respondents were asked to 6, where a 0 means they did not experience the feeling at all and a 6 means the feeling was very strong , how happy, sad, tired, and stressed they felt at that time. In addition, they were asked how much pain, if any, they felt at that time, and how “ meaningful ” they considered what . they were doing Since television is a leisure activity that is probably a close substitute for video games, we explore the self - reported emotional experience during time spent playing video , and during all activities for young men . games, watching TV If video game techn improve to make engaging in the activity more ology did indeed enjoyable, one would expect to see better emotional states (e.g., higher rating of happiness) during time spent playing video games than during time spent watching TV. Moreover, with three observations per person ol for individual fixed effects and compare , it is possible to contr young men’s reported experiences as they e ngage in different activities throughout the day. Table 3 shows estimates of fixed effects regressi ons of the various affect measures on a dummy indicating time spent playing games, watching television, and using a computer. The omitted group is all other activities. To increase the sample size, the sample consists of males age 16 to associated with ame playing are 35. The results show some evidence that episodes that involve g 14
17 g reater happiness, less sadness and less fatigue than episodes of TV watching, although stress is , higher during game playing. Game playing also appears to be a more pleasant e xperience than using the computer for this group. Game playing, however, is not reported as a particularly meaningful activity by participants; indeed, it is reported as less meaningful than other activities. The ATUS also reveals that g is a social activity. A little over half the time ame playing that young men play video games they report that they were with someone while engaging in the activity, most commonly a friend. Furthermore, during 70 percent of the time that they were they repo rt they were interacting with someone (presumably playing games online when they were not present ). As a whole, these findings suggest that it is possible that, as Aguiar, et al. argue, improvements in video games have improved the enjoyment young men derive from leisure in a consequential way. B. Prime Age Men Although the participation rate of prime age men has trended down in the U.S. and other econ for many decades, by international standards the participation omically advanced countries rate of prime age men in the U.S. is notably low. Because prime age men have the highest labor rate of any demographic group, and have traditionally been the main force participation breadwinner for their families, much attention has been devoted to the decline in labor force 14 in the U.S . participation of prime age men Evidence in Juhn, Murphy and Topel (1991, 2002) and Abraham and Kearney (2017) suggest s that the secular decline in real wages of less skille d workers is a major contributor to the secular decline in their participation rate. CEA (2016) reaches a similar conclusion , as the decline in labor force participation has been steeper for less 14 calls the increase in jobless men who are not looking for work “America’s Eberstadt (2016), for example, invisible crisis.” 15
18 educa . Figure 5 age men shows that participation rate of prime age men fell for men at - ted prime all education levels, but by substantially more for those with a high school degree or less. Here we highlight a significant supply - side barrier to the employmen t prospects of prime 15 Table 4 related problems. ealth - reports the distribution of men and women age men: namely, h excellent, very good, good, fair or poor based on the 2010, 2012 and reporting their health as 16 American Time Use Survey Well - b eing Supplement (ATUS - WB) . 2013 Forty - three percent of prime age men who are out of the labor force reported their health as fair or poor, compared with Women who are out of the just 12 percent of employed men and 16 percent of unemployed men. labor force are als o more likely to report being in only fair or poor health compared with employed women , but the gap is smaller: 31 percent versus 11 percent. Thus, health appears to be a more significant issue for prime age men’s participation in the labor force than for prime age women’s , and we focus on documenting the nature , and probing the veracity, of - their health related this section. W hile it is certainly possible that extended joblessness and problems in could have exacerbated many of the physical, despair induced by weak labor demand caused or and mental health - related problems that currently afflict many prime age men who are emotional out of the labor force, the evidence in this section nonetheless suggests that these problems are a substantial barrier to work that would have to be addressed to significantly reverse their downward trend in participation. six Beginning in 2008, the BLS has regularly included a series of functional disability [in the household] questions in the monthly CPS. , the s urvey asks, “ Is anyone For example blind 15 Coglianese (2016) finds that about half of the decline in prime age male labor force participation is due to permanent exits, and that only 20 to 30 p reduced labor demand, suggesting a major ercent of the decline is due to role for supply side factors. 16 The exact question was: “Would you say your health in general is excellent, very good, good, fair or poor?” Self - have been found to correlate reasonably well with objective health outcomes in reported subjective health questions the past. 16
19 17 or does anyone have serious difficulty seeing even when wearing glasses? Pooling all of the ” 2016, results of these q data from 2008 uestions are reported in Table 5 - by labor force status for e disability was reported for 34 percent t least on of prime age men who are out A prime age men. 18 t his figure rises to 42 of the labor forc percent for the subset of men age 40 to 54 . e , and Perhaps surprisingly, white prime age men were more likely to report having at least one of the six conditions (35.8 percent) than were prime age African American (32.3 percent) or Hispanic At least one disabili ty condition was reporte d for 40 percent of (29.3 percent) men. prime age men with a high school education or less. The most commonly nonparticipating reported disabilities were “difficulty walking or climbing stairs” and “difficulty concentrating, or making decisions ” ; about half repo rted multiple disabilities. Only 2.6 percent remembering, and 5.8 percent of unemployed men in this age group reported a disability. of employed men Figure 6 a shows the probability of being out of the labor force conditional on having a disability each year from 2008 to 2017 . The probability of being out of the labor force conditional on having a disability ha s trended up over the last nine years, which suggests that the duals back to work. improvement in the job market over this period is not drawing disabled indivi 7 , Figure 6 b shows the probability of being out ther from 2008 to 201 Pooling all of the data toge of the labor force for each of the six conditions, and for those who indicate having any of the six 17 estimate of results in an underestimate or over this measure the “true” disability rate. One could question whether On the one hand, the list is restricted to just six conditions (for example, speech and language disorders are omitted). In addition, there could be a stigma attached to reporting physical, emotional and mental health conditions for - orted because it is a more socially acceptable household members. On the other hand, a disability could be self rep reason for joblessness . than the alternative 18 A natural question is whether an increase in the number of disabled military veterans returning to civilian life has contributed to the decline in the participation rate. The short answer is that this does not appear to be the case. The labor - of - the - share of out - force prime - age men who are veterans has declined, from 11.4 percent in 2008 to 9.7 percent in 2016. Moreover, the proportion of prime age men who are veterans has trended down over the last two age category. Nevertheless, about 40 decades as the large cohort of Vietnam - era veterans has aged out of the prime - percent of veterans who are out of the labor force report a significant disability, so any strategy to assist veterans to return to the labor force would need to address disability issues. 17
20 conditions and the subset with multipl e conditions. Those who have difficulty dressing, running errands, walking or concentrating have a much lower participation rate than those who are blind or have difficulty seeing or hearing. Pain and Pain Medication : ATUS and CDC Prevalence of For randomly selected episodes of the day, the ATUS - WB module asked respondents, “From 0 to 6, where a 0 means you did not feel any pain at all and a 6 means you were in severe pain, how much pain did you feel during this time if any? ” The first row of Table 6 reports the average pain rating by labor force status (weighted by episode duration ), and the second row reports the fraction of time respondents reported with a pain rating above 0, indicating the presence of some pain. The results indicate that indivi duals who are out of the labor force report experiencing a greater prevalence and intensity of pain in their daily lives. As a group, workers who are out of the labor force report feeling pain during about half of their time. And f or those – who report a disability, the prevalence and intensity of pain are higher disabled prime age men who are out of th e labor force report spending percent of their time in some pain and an 71 2.8 throughout the survey day. average pain rating of NLF men who report a disability Comparing the daily pain ratings of employed and indicates that the average p ain rating is 89 percent higher for those who are out of the labor e prevalent and force Moreover, in five of the six disability categories, reported pain is mor . more intense for those who are out of the labor force than for those who are employed. These results suggest that the disabilities reported for prime age men who are out of the labor force are more severe than those reported for employed me n , on average . The ATUS - WB also asked respondents, “ Did you take any pain medication yesterday, ” Fully 44 p ercent of prime age men such as Aspirin, Ibuprofen or prescription pain medication? 18
21 who were out of the labor force acknowledged taking pain medication on the previous day, a s . This rate was more than double that of lthough this encompasses a wide range of medication (The gap was not a s great for prime age women: 25.7 percent employed and unemployed men. of employed women reported taking pain medication on the reference day compared with 34.7 of - the percent of out labor - force women.) And if we limit the comparison to men who report a - - of the labor force were more likely to report having taken pain disability, those who were out medication (58 percent) than were those who were employed (32 percent), again suggesting the disabilities are more severe, on average, for those who are out of the labor force. he high rate T for NLF men lated to Case and Deaton’s (2015 ; of utilization of pain medication is possibly re ) finding of a rise in mortality for middle age whites due to accidental drug poisonings , 2017 from 1999 to 2013. We return to this issue below. especially from opioid overdoses, Since 1997, t he Center s for Disease Control and Prevention ’s (CDC ’s ) National Health (NHIS) Interview S urvey has asked cross sections over 300,000 individuals annually whether they experienced pain in the last three month s. Please Specifically, respondents are instructed, “ ins that are refer to pain that LASTED A WHOLE DAY OR MORE. Do not report aches and pa Figure 7 displays trends in the percent of prime age men fleeting or minor.” reporting pain in the 19 last three months by labor force status. (Beginning in 2005 the unemployed can be distinguished from other non - employed workers.) Although the data are volatile from year to year, there is a slight upward trend in the proportion of NLF and unemployed prime age men who re Despite the extraordinary rise in the use port experiencing pain in the last three months. of opioid pain medication over this period, there is certainly no evidence of a decline in the proportion of people who report feeling pain. 19 Any individual who reported lower back pain, neck pain, leg pain, or jaw pain is coded as having experienced . https://www.cdc.gov/nchs/nhis/ pain. For details of t he survey see 19
22 The NHIS data also s feeling pain have uggest that the employment consequences of increased. In 1997, prime age men who reported experiencing pain in the past three months were 6 percentage point less likely to work than were those who reported that they did not experience in; by 2015 this difference had increased to 10 percentage points. pa Prescription Pain Medication, Disability and Labor Force Dropouts: Princeton Pain Survey To better understand the role of pain and pain medication in the life of prime age men who are neither working nor looking for work, I conducted a short online panel survey of 571 NLF prime men age 25 - 54 using an internet panel provided by Survey Sampling Inc, henceforth 20 called the Princeton Pain Survey (PPS). The first wave of the survey was conducted over the period September 30 - October 2, 2016. The results of this survey underscore the role of pain in Fully the lives of non , and the widespread use of prescription pain medication . working men 47 percent of NLF prime age men responded that they took pain medicati on on the previous day, slightly higher than but not significantly different from the corresponding figure from ATUS sample . Nearly t - thirds of those who took pain medication indicated that they took wo percent of these cases, the men reported that they also took (in 36 prescription pain medication 8 - - over pain medication) ; see Figure the . Thus, on any given day, 31 percent of NLF counter prime age men take pain medication, most likely an opioid - based medication. And these figures likely understate actual proportion of men taking prescription pain medication given the the stigma and legal risk associated with reporting taking narcotics. 20 We screened for men age 25 - 54 who did not work in the previous week, were not absent from a job, and did not search for a job in the prev definition of out of the labor force requires that individuals ious week. Because the BLS did not search for a job in the past four weeks, our definition is a bit less restrictive. Weights were developed to match the 2016 CPS ASEC by age group (25 - 40, 41 - 54), race and Hispanic ethnic ity. Weighed percentages are s software. reported in the text. The survey was conducted with Qualtric 20
23 Forty percent of this sample of prime age men respo nded “Yes” when asked directly, - T wo - “ Does pain prevent you from working on a full time job for which you are qualified?” reported that they had a disability , which is about double the rate in thirds of the men in the PPS ed . Th e higher disab ility rate partly result because respondents the CPS for NLF prime age men 21 ther” in addition to the BLS’s six conditions, and 16 percent filled out other. could write “ It O also possible that men who are drawn to participate in Internet surveys are more likely to is suffer a disability, or that the CPS understates the number of prime age men with a disability. A follow - up online survey conducted July 7 - 14, 2017 attempted to interv iew the 376 who continued in the SSI panel, a little over 9 months after the initial survey. A respondents total of 156 prime age men responded to the follow - up survey, or 41 percent of those who were eligible. Six of the respondents said that they had a steady, full - time job and were dropped from the sample, so the resulting analysis sample has 150 observations. Table 7 reports a cross - tab prescription pain medication in the preceding day in wave 1 indicating the proportion who took of the survey . and 2 - tab indicate s the persistence of taking pain medication, which is The cross consistent with studies that find high rates of addiction to opioid medication (add citation) . Nearly of those who took prescription pain medication in the initial survey reported 80 percent taking it in the follow - up survey. Individuals in the follow up survey were asked, “About how often would you say that you take prescription pain medication?” Almost a quarter (24 percent) responded every day, and another 18 percent said more than once a week and 3 percent said once a week. A minority, 41 percent, responded never. All respondents except those who said they never take prescription pain medication were asked, “How do you usually pay for prescription pain medication? (Mark 21 ” included Common write - in responses for those who marked “ other : anxiety disorder; back pain; cancer; chronic pain; epilepsy; heart condition; and sleep disorder. 21
24 all that apply. ) R esults are tabulated in Table 8 . It is clear that government health insurance ” ) play a major role in providing pain medication to this programs (Medicaid, Medicare, VA - thirds of respondents used at least one of these government programs to purchase group. Two with the largest group relying on Medicaid. prescription pain medication, Respondents were asked, “Wh at is the source of pain that typically causes you to take pain ” medication? - work related injury over a work - related one: Overwhelmingly they selected non 88 percent to 12 percent. In the first wave of the PPS respondents were asked about participation in various in - come provides a tabulation of respon ses. support programs. Table 9 Half of the NLF prime age men report participating in at least one program . Thirty - five percent of NLF prime age men indicated that they were on Social Security Disability Insurance (SSDI) , compared to 25 difference is likely a result of the PPS sample percent in the May 2012 CPS supplement The . being nonrepresentative, under reporting in CPS, and an increase in SSDI participation from May 2012 to July 2017. Workers’ compensation insurance is a much less frequent source of related injuries income support than SSDI, c onsistent with work - being reported as a source of pain in only a small percent age of cases. PPS follow In the up survey, respondents who were not currently on SSDI were asked if - ever applied to SSDI in the past. Fully 30 percen t of those asked indicated that they had they 22 Many of these individuals could be in the process of applying to previously applied to SSDI. 23 or appealing a decision, which could influence their current labor supply incentives. SSDI If 22 Among the subset of individuals who were not on any income - support program, 20 percent reported that they had previously applied for SSDI. 23 : “If you’re working and your earnings average The Social Security Administration advises applicants for SSDI more than a certain amount each month, we generally won’t consider you to be disabled.” Von Wachter, Song and Manchester (2011) find that a substantial number of male applicants age 30 - 44 who are rejected fr om SSDI tend to - 64 are employed post application. work post application, while relatively few rejected applicants age 45 22
25 the fraction of NLF prime age men on SSDI is between 25 percent and 35 percent, then around half of all NLF prime age men could have applied to SSDI at some point. This suggests that the program’s reach is substantially larger than previously appreciated. reducing male labor force participation has long been debated by The role of SSDI in economists (see, e.g., Parsons, 1980 and Bound, 1989). CEA (2014) reports that the fraction of , while the labor force prime age men on DI rose from 1 to 3 percent betwee n 1967 and 2014 t DI could at participation rate of this group fell by 7.5 percentage points, which suggests tha a quarter of the decline in participation most account for over this period , and estimates of the causal effect of DI suggest that the ava less of the ilability of benefits is responsible for even on the high incidence of pain experienced . T decline in participation e he evidence reported her by the disabled, especially those who are out of the labor force, suggests that physical and mental 24 he ailments are a barrier to participating in many activities. alth C. Women in the U.S. stopped rising after As mentioned, the aggregate labor force participation 2000 because the participation rate of women stopped rising. Starting in 2007 the participation rate began to fall for women overall, although the rate had already been declining for younger women over the previous decade. America’s relative standing among economically advanced countries in terms of the participati on rate of women also slipped. A particularly interesting comparison is with Canada. The participation rate of women in Canada was equal to roughly that in the U.S. in the late 1990s , but it continued to grow for another decade in Canada while it plateau ed and then declined in the U.S. For prime age women, from 1997 to 2015 the 24 See Krueger and Stone (2008) on the relationship between pain and time use. 23
26 participation rate rose from 76 percent to 81 percent in Canada while it fell from 77 percent to 74 percent in the U.S. find that participation rate of wome n in the U.S. declined Drolet, et al. (2016) since the 1990s vels but it declined more for women with a high school at all education le , 44. In Canada, by contrast, the participa tion rate rose education or less, especially those age 25 - for all education groups. (2013) conclude that “the expansion of ‘family - friendly’ policies, Blau and Kahn - including parental leave and part ” explains 29 percent of the decrease in time work entitlements , 25 that women's labor force participation in the U.S. relative to other O.E.C.D. countries. Given the biggest gap between women’s labor force participation in Canada an d the U.S. opened up educated women of childbearing age, who are unlikely to receive paid maternity among less licies , along with the rise in the leave and other family benefits, it is plausible that family leave po - income gradient in the U.S., account for a significant share of the rising gap in education 26 participation between women in the U.S. and Canada as well. There is also evidence that generational shifts, which drew increa sing numbers of women 27 This implies that the historic gains in into the workforce, have come to an end in the U.S. that came about by the entry of new birth cohorts and women’s labor force participation of exit older ones will no longer lead to rising par ticipation. Figure 9 displays the labor force participation rate of fiv e cohorts of women based on ten year - of - birth intervals over the lifecycle from age 16 to age 79 using data from the 1962 through 2016 ASEC. The age displayed along refers to the age of the middle birth year of the cohort. So the 1941 birth the horizontal axis 25 Dahl, et al. (2016), however, find that the extension of maternity benefits from 18 to 35 weeks in Norway had little effect on labor force participation. 26 d for unmarried women without Moffitt (2012) highlights the puzzling fact that the employment rate decline children, and for higher educated women as well. 27 See Juhn and Potter (2006) for an early discussion of this issue. Goldin and Mitchell (2017) highlight that the sha lifecycle labor force participation profile of women evolved from an inverted - U - pe for cohorts born before the 1950s to a fairly flat shape with a sagging middle for those born after the mid 1950s. 24
27 cohort includes women born from 1937 to 1946, the 1951 cohort includes women born 1947 to makes clear that at all ages women in the 1951 cohort The cross cohort pattern 1956, and so on. were more likely to participate in the labor force than were women who were born a decade succeeding cohorts earlier when they were the same age. The increase in participation across 45. But the t for women age 21 - lifecycle cohort was particularly eviden profiles essentially stopped rising after the 1961 cohort, and born in the five years surrounding 1981 were women actually less likely to work at a given age than were women born a decade earlier. And while it is impossib le to separate out calendar time, age and birth year effects, these generational developments are unlikely to represent time effects because they have been occurring over several years , and because participation is not very se nsitive to the business cycle . i n Figure 9 The cohort pattern women hy it is that also helps explain another anomaly : W age 55 to ed the biggest break from trend over the last decade , as shown in Appendix 64 exhibit Figure A15. The answer appears to be that as women born in the late 1940s and early 1950s aged out of the 55 - 64 year old bracket, they were replaced by a succeeding generation of women who had about the same level of participation as the 1947 - 56 birth cohort when they were both in their late implication of this pattern is that a continuation of the sharp rise 40s and early 50s. An in participation over recent decades for women age 65 and over evident in Figure A16 is likely in jeopardy , as the 1950s birth cohort gives way to the 1960’s birth cohort that had roughly the same labor force participation rate in midlife. that the cohort participation profiles stopped rising for younger women age The finding ial for 21 to 40, who are much more likely to be engaged in raising a family, highlights the potent workplace flexibility and family friendly policies to raise participation in the future. Clearly, the 25
28 U.S. can no longer rely on the past tendency of succeeding generations of women to enter the rticipation rate in the future. lift the aggregate pa labor force at earlier ages to NLF and Not Mainly Taking Care of Home Responsibilities important distinction for women non - labor force participants involves those who say An they are not working primarily because they are taking care of home responsibilities, and those In 1991, 77 percent of NLF prime ag e women were not who are not working for other reasons. working because of home responsibilities, and in 2015 that figure had declined to 60 percent ccording to CPS and ASEC data. (Note that these questions on labor force participation relate a erence week.) Among those who cited a reason to the calendar year, as opposed to the survey ref home responsibilities as main reason for not working, the rise in nonparticipation other than the 28 (see F igure for women parallels that of men 0 ) . Excluding those who cite home 1 responsibilities, the distribution of reasons for not working for women roughly equals that of men as well , with disability/illness representing the largest category. As we shall see below, the onsibilities and other reasons also has a meaningful effect on distinction between home resp subjective wellbeing for NLF women. D. Retirees As emphasized in Section I I , a major reason for the decline in labor force participation 2007 is that the large baby boom after , as had long been cohort started to reach retirement age expected . Those born in 1946, at the beginning of the baby boom, would have qualified for Social Security retirement benefits starting in 2008. 28 Steve Hipple of BLS generously shared these tabulations with me. Also see Frank Lysy’s blog post for an analysis of these data: https://aneconomicsense.org/2016/10/14/the - structural - factors - behind - the - steady - fall - in - . labor - force - part icipation - rates - of - prime - age - workers/ 26
29 Further evidence of the profound effect of retireme nts on t he U.S. workforce is in Figure 1 , which shows the percentage of individuals age 16 and older who are classified as retired in 1 29 The share of the 16+ population that was retired hovered around 15 percent from the CPS. 1994 to 2007, and then rose fr om 15.4 percent to 1 7 . 6 percent from 2007 to 201 7 . The 2. 2 percentage point rise in the retirement rate over this period almost matches the 2.8 percentage By gender, the retirement rate increased by 2. 2 point drop in the labor force participation rate. p ercentage points for men and 2 .1 percentag e points for women since 2007. Since retirements tend to be permanent exits from the labor force, and the main reason for the decline in labor force participation over the past decade is the increasing number of retirements due to the aging of the baby boom ge neration, this is another reason to ex pect relatively little cyclical recovery in labor force participation in the near term. I - Being V. Subjective Well This section evaluates the self reported subjectiv e well - being (SWB) of various - demographic groups by labor force status. A comparison of SWB across labor force groups is of interest for two reasons. First, low levels of SWB can point to social problems for particular groups and potentially large welfare gains from successful interventions. Second, if a group that is out of the labor force exhibits a high degree of SWB it is probably unlikely that they are eager to change labor force status. Of course, SWB severely discontent with their situation, and i s difficult to measure and compare across individuals , so the usual caveats when using SWB measures apply. 29 This is based on the EMPSTAT variable in the IPUMS data. 27
30 Two types of measures of SWB are available from the ATUS WB module. The first is - a self anchoring scale which asks respondents to evaluate their life in general, the Cantril Ladder, - 30 and was included in the 2012 and 2013 waves of the survey. The exact question wording was: Please imagine a ladder with steps numbered from zero at the bottom to ten at the top. The top of the ladder represents the best possible life for you and the bottom of the ladder represents the worst possible life for you. I f the top step is 10 and the bottom step is 0, on which step of the ladder do you feel you personally stand at the present time? The second measure is th e affective rating of randomly selected episodes of the day. This includes ratings of happiness, sadness, stress , pain, meaningfulness and tiredness on a 0 to 6 scale. We compute the duration - weighted average of these affect measures as well as the U - inde x. The U - index is defined here as the percent of time in which the rating of sadness or stress the U excee Kahneman and Krueger (2006) emphasize that ds the rating of happiness. - index is robust if respondents interpret the scales differently, as long as t hey apply the same monotonic transformation to positive and negative emotions. - The mea - 10 d for men and 1 1 a a 1 1 d for women . The sures are summarized in Tables 10 second to last row of the tables reports the mean Cantril ladder rating for each group. Figu re s 1 2 a - 1 2 d further show cumulative distributions of the Cantril ladder for each group , where the horizontal axis is arrayed in reverse numerical order (from 10 to 1) so that distributions that lie above lower ones totally dominate in terms of the ladder of life. A few findings are noteworthy. First, young men and women who are out of th e labor force seem remarkably content with their lives. As a group, y oung people who are not participating in the labor force report that their lives are on a higher step of the Cantril ladder of 30 See Deaton and Kahneman (2010) for a comparison of the correlates of the Cantril ladder and daily emotional well - being. They find that the Cantril ladder is more strongly correlated with education and income, while daily being is more closely correlated with loneliness and health. - emotional well 28
31 the best possible life ho are employed. On a moment - to - than do similarly aged individuals w there are only small and statistically insignificant differences in the moment basis, typically - weighted average reported emotions across the employed, unemployed and out of the duration youth . The only statisti cally significant difference related to sadness: unemployed labor force youth reported being sad der over the course of the day than the employed or NLF youth. Second, unlike youth, prime age men who are employed are considerably more satisfied eral than are men who are out of the labor force or unemployed. Prime age with their lives in gen men who are out of the labor force report themselves between employed men and unemployed men on the Cantril The emotional experien ces ladder of life, but closer to the unemployed men. over the course of the day, however, indicate that NLF men are less happy, more sad, and more stressed than unemployed men, reversing the ranking from the Cantril ladder. Moreover , the U - index (which measures unpleasant time but omits pain) is conside rably higher for NLF men than for . This reversal suggests that there may be more adaptation in terms of unemployed men overall quality of life expectations for their moment - to - men than there is in terms of NLF moment experience . In other words, prime age men who are out of the labor force, who often have a significant disability, may have lowered their views of the best possible life they could er in relation to this compress ed ladder, while expect, and reported their step on the Cantril ladd ir reporting of emotional experience was not recalibrated with respect to expectations. If this the NLF - being of prime age men should be an even bigger is the case, then the low subjective well 31 than on the ladder of life data . social concern based on the emotional data 31 For the sample of 21 - 30 year old men who were out of the labor force we found that the Cantril ladder was closer ch lower emotional to employed men than to unemployed men, but the U - i ndex indicated that they had mu experience than employed and unemployed men. 29
32 One factor that likely contributes to the l ow level of emotional well - being of NLF prime relatively age men is the alone. Prime age NLF men spend high amount of time they spend employed prime age men, 17 nearly 30 percent of their time alone, compared with 18 percent for percent for prime age employed women, and 19 percent for prime age NLF men. Deaton and Kahneman (2010) found that alone time correlated more strongly with daily emotional well - being, while income and education correlated more strongly with evaluative well - being. Third of prime age women who are out of the labor force is closer , unlike men, the SWB to that of employed women than it is of unemployed women. In fact, the U - index is lower for prime age NLF women than for employed prime age women. NLF women report higher levels of happiness and s adness but less stress than employed women. Unlike men, women who are out of the labor force report deriving considerable meaning from their activities. These results do not paint a picture where women who are out of the labor force, as a grou p, are as a group discontent with their lives or daily routines, and therefore eager to return to work. prime age Fourth, NLF who are not working for reasons other than tak ing care of women home responsibilities report notably lower levels of subjective wellbeing th an other NLF women and employed women. The U - index for NLF women who are not employed for a reason other than taking care of home responsibilities is 0.19, as compared to 0.09 for NLF women who are not employed because they are taking care of home respons ibilities, and 0.17 for employed and unemployed women. Additionally, NLF prime age women who are not employed for a reason other than home responsibilities report a much lower average step on the Cantril ladder (6.4) and much greater incidence of pain and use of pain medication (49 percent took pain medication in the preceding day compared to 21 percent of other NLF women). Thus, NLF women are a bifurcated group, with those who cite home responsibilities as the reason they are not employed 30
33 reporting high levels of SWB and meaning in their lives, and those who are NLF for other reasons expressing high levels of distress and discomfort. women - 70 appear to be similar to prime age women in tha t the NLF group Lastly, age 55 about equal contentment with their lives as a whole and daily emotional experiences as report s 70 year old women, however, appear quite unhappy and - employed women. Unemployed 55 dissatisfied with their lives. - 70 year old gro up who are unemployed also appear to Men in the 55 be quite dissatisfied and unhappy with their lives compared with employed men the same age, NLF men appear midway while between employed and unemployed men in terms of the Cantril ladder. Men who are out of the labor f orce express relatively low levels of meaning in their daily ctivities, but their U index indicates less time spent in an unpleasant state than employed or a - unemployed men. Pain Medication, Opioid Proliferation, V. and Labor Force Participation J.D. Vance (2016) warns that, “An epidemic of prescription drug addiction has taken Many alarming root.” . According to the CDC, sales of prescription statistics bear out his fear 32 5 per capita increased by 356 percent from 1999 to 201 . opioid medication More t han one in 33 five individuals insured by Blue Cross and Blue Shield received an opioid prescription in 2015. medication is Enough opioid dispensed annually to keep every man, woman and child on (Doctor and Menchine, 2017) The numbe r of deaths from opioid painkillers for a month . quadrupled overdoses . In 2015, more than 33,000 Americans died from from 1999 to 2015 opioid overdose , more than double the number murdered in homicides . An estimated one in 32 ht tps://www.cdc.gov/vitalsigns/opioids/images/graphic - a See 1185px.png . - 33 See - - he roin - epidemic/lots - americans - prescribed - opioids - insurance http://www.nbcnews.com/storyline/americas . n777906 survey - shows - 31
34 every 550 patients who started on opioid therapy died from an opioid related cause, with the - median fatality occurring within 2.6 years of the initial prescription (Frieden and Houry, 2016). Fully 44 percent of Medicare recipients under age 65 were prescribed opioid medication in 2011 And d (Morden, et al., 2014). in the U.S., there is espite the rapid diffusion of opioid medication little evidence showing that opioid treatment is efficacious in reducing pain or improving functionality. In fact, Frieden and Houry (2016; pp. 1501 - 02) note that “several studies have showed that use of opioids for chronic pain may actually worsen pain and functioning, possibly by potentiating pain perception.” The opioid crisis preceded the Great Recession -- indeed, opioid prescriptions fell from 2010 to 2015 -- and varying p rescription rates are probably rooted in changing medical practices strategies ( and norms , and more aggressive pharmaceutical companies’ marketing Doctor and Menchine, 2017; Satel ). Doctor training also seems to affect opioid prescription rates. , 2017 Schnell and Currie (2017), for example, find that doctors from the lowest ranked medical schools write 33 times more opioid prescriptions per year than do doctors from the highest ranked hools, holding constant count y and type of medical practice fixed effects. Krause and Sawhill sc find that, “The ten counties with the highest prime - age male mortality rates due to these (2017) ‘deaths of despair’ [ alcohol, suicide, and accidental poisonings] in the CDC database had an average prime age male p articipation rate of 73 percent i n 2014, compared to 88 percent for the - prime - age male population across the country.” Although the direction of causality is unclear, Mericle (2017) notes, “ The opioid ep idemic is intertwined with the story of declining prime - age participation, especially for men, and this reinforces our doubts about a rebo und in the p articipation rate.” 32
35 There is a clear regional pattern to opioid prescription rates and drug overdoses. The average quantity of opioids prescribed per capita varies by a factor of 31 to one in the top 10 percent of counties relative to the bottom ten percent of counties, Across according to CDC data. states, per capita prescription rates vary by a factor of three to one. The CDC argues that, “ Health issues that cause people pain do not vary much from place to place, and do not explain 34 this variability in prescribing.” T his section probes the connection between the use of pain medication and local opioid p rescription rates, controlling for individual health conditions and other characteristics. Consistent with the CDC’s assertion , e vidence suggests that local opioid prescription practices influence the use of pain medication on individuals’ di sability , conditional status, self - reported health, and demographic characteristics . Leveraging local differences in prescription rates, labor force regressions indicate that the participation rate is lower and fell more in counties where more opioids were prescribed, controlling for the area’s share of manufacturing employment and individual characteristics. Use of Pain Medication and Opioid Prescription Practices To explore the relationship between local medical practices and the use of pain level medication, county - data on the volume of opioid pr escriptions per capita in 2015 from CDC were merged to the ATUS - WB supplements, which include data on whether individuals 35 any Opioid prescriptions are measured by took pain medication in the preceding day. prescribed per capita , which is a standard way of Morphine Milligram Equivalent (MME) units 34 See https://www.cdc.gov/drugoverdose/data/prescribing.html . 35 Specifically, CDC data on MME per capita were merged to the ATUS based on county FIPS codes. If the FIPS code was missing for a metro politan area in the ATUS , the average MME for the counties that comprised that metro area w and lacked a FIPS code in matched to the ATUS, and i f an individual was not residing in a metro area as A TUS he or she was linked to the average MME per capita in non - metro areas in the balance of the state. 33
36 aggregating different opioid medications. the logarithm of To ease the interpretation, we take 36 in the county . per capita MME units summarizes results of linear probability models predicting whether an Table 12 individual took pain medication in the preceding day as a function of opioid prescription rates in - reported overall health, and personal characteristics. the area, functional disab ility status, self Not surprisingly, in areas where more opioids are prescribed, individuals are more likely to report that they took pain medication on the preceding day. In column 1, a 10 percent increase in opioid s prescribed per capita is associated with a 0.6 percentage point, or the amount of 2 percent increase in the share of individuals who report taking a pain medication on any given , 37 This effect is cut roug hly in half but remains highly statistically significant when controls day. are added for functional disabilities, self - reported health and demog raphic characteristics ). Even within detailed regions, the area (column 5 wide prescription rate is a significant - p redictor of whether individuals took pain medication in the preceding day (column 6) . These findings support the CDC ’s view that differences in health conditions do not vary enough across areas to explain the large county differences in the use of p ain medication. cross Opioid Prescription Rates and Labor Force Participation - level opioid prescription rates (MME per capita) to individual - Next we link 2015 county 38 Table - level labor force data from the CPS in 1999 - 2001 and 2014 13 a reports estimates of 16. 36 Although one might expect a one to - one correspondence between opioid presc ription rates and the use of pain - medication absent other controls, there are two important reasons why such a direct relationship does not hold in these data: first, the dependent variables includes many forms of pain medication in addition to opioids; se cond, the independent variable reflects dosage as well as usage, whereas the dependent variable only reflects usage. 37 If separate regressions are estimated for men and women, the coefficient on log opioids per capita is larger for m en than for women, but the differences are not statistically significant. 38 To be more precise, i n 41 percent of observations opioid prescriptions per capita could be matched directly at the county level; in 34 percent of observations we had to aggregat e over counties to match at the metropolitan or central city level; and in the remainder of cases we used the average of counties in the balance of the state. For simplicity, we refer to these areas as counties. 34
37 linear probability models for prime age men where the dependent variable is 1 if an individual participates in the labor force and 0 if he does not. Table b has comparable estimates for 13 2014 - 16 time period. prime age women. A dummy variable indicates the Column 1 indicates that the labor force participation Consider first the results for men. rate fell by 3.2 percentage points for men from 1999 - 2001 to 2014 - 16. Column 2 adds the opioid prescription rate for 2015 and column 3 ad ds an interaction between the opioid prescription rate and the 2014 16 time period dummy variable. Both of these additional variables are negative - and significant, indicating that labor force participation is lower in areas of the U.S. with a high rate of opioid prescriptions, and labor force participation fell more over this 15 year period in areas with a high rate of opioid prescriptions. These conclusions continue to hold when additional variables are included in the model, including demographics, eigh t Census region indicators, the share of employment in the county employed in manufacturing in 1999 - 2001, and 39 the manufacturing share interacted with the 2014 - 16 time period dummy. W e continue to find a etween negative and statistically significant interaction b the 2014 and opioid - 16 time period prescriptions when unrestricted county dummies are included in column 7 to absorb /area . The fact that the coefficients on the opioid prescription variables are persistent area effects in column (6) unchanged when the manufacturing variables are included in the regression suggests that the opioid crisis is occurring in areas outside of traditional manufacturing And ngholds. w stro e find similar results (in regression not shown here) if we use t he Autor, Dorn and Hanson (2013) China import exposure variables in place of the share in manufacturing. 39 calculated with the CPS, and merged on based on The manufacturing share of employment in 1999 - 2001 was country (where available), metropolitan area (where country was not available) or state (where county and metropolitan area were not available). 35
38 These regressions are difficult to interpret , b ut if cross - county for a number of reasons opioid prescription rate s can be taken as an exogenous result of differences in differences in conditional on personal characteristics and broad region dummies , medical practices and norms the effect of the growth in opioid prescriptions on labor force can be estimated . In particular, I assume that the based opioid prescription rate coefficient reflects differences across inherent regions, and the interaction between prescriptions and time captures the effect of changes in prescriptions on labor force participation over time. This is a big leap, and ideally I would have preferred to have a baseline measure of prescriptions (country - level MME data are unavailable before 2010), so this calculation is at best considered illust rative. These caveats aside, o pioid increased by a factor of 3.5 nationwide prescriptions per capita between 1999 and 2015, which is the equivalent of 0.55 log units. Multiplying 0.55 by the coefficient on the interaction between pioids and the second p o eriod (.011), suggests that the increase in opioid prescriptions could account for perhaps a 0.6 percentage point decline in male labor force participation, which is 20 percent of the observed decline in this period. lar coefficient for the interaction term between time The results for women indicate a simi - level opioid prescription rates, but the base opioid prescription rate is posi tive. If the and county preceding calculation is conducted for women, about one quarter of the decline in labor force parti cipation can be accounted for by the growth in opioid prescriptions. An obvious concern about the labor force regressions is that omitted variables, such as and demand for pain medication , are correlated with workers’ health conditions that cause pain co unty - level opioid prescription rates. Although the basic monthly CPS does not include information on health, the CPS ASEC surveys do include information on self - reported health. If ng this ( ns pooling together men and women usi smaller) we estimate the labor force regressio 36
39 sample and control for self - level opioid prescription rate has a similar reported health, the county - effect as in the larger basic monthly CPS data. It is also worth noting that Laird and Nielsen using arguably exogenous variatio n in physicians’ practices stemming from geographic (2017), find a significant and sizable negative effect of the opioid mobility across municipalities, 40 – but not other medications – on labor force participation in Denmark . prescription rate These findings are . A useful extension of this preliminary and highly speculative analysis would be to determine whether higher prescription rates are associated with depressed flows of workers from outside the labor force back into the labor force, or with greater labor fo rce exit rates. In addition, future work could seek to identify sources of exogenous variability in prescription rates, or in treatment for opioid addiction, to estimate the causal effect of opioid medication on labor force participation. V I . Conclusion The over the past two decades is a decline in labor force participation in the U.S. her studies, this study find and a social macroeconomic . Along with several ot problem s concern that declining labor force participation since 2007 is largel y a result of an aging population and ongoing trends that preceded the Great Recession , such as increased school enrollment . Given ongoing downward pressure on labor force participation from an expected wave of members of the baby boom retirements among generation, and the fact that a substantial cyclical rebound in labor force participation is unlikely , a reversal in the slide in participation will require 40 Although it is difficult to compare the magni tudes that Laird and Nielson find with those reported here because Laird and Nielsen focus on opioid prescription rates (rather than amount prescribed per capita), their estimates imply reported here. They find that a 10 percentage large labor force effects that appear substantially larger than those point increase in a doctor’s prescription rate, which is roughly a 50 percent increase from the current U.S. average, is associated with a 1.5 percentage point decline in the labor force participation rate. 37
40 a change in secular trends affecting various demographic groups , and perhaps a major reform in . There are a few demographic groups that may be more susceptible to a rise immigration policy in labor force participation than others . First, older workers may increasingly delay retirement, bolstering the ir rise in labor force participation that has oc curred over the past two decade s . This trend may not continue for older women, however, as a cross cohort analysis shows that labor force participation stopped rising for cohorts that are about to enter their late 50s and 60s. Second, labor force parti cipation of women age 25 to 44 has been edging down for two decades, unlike their counterparts in Canada. While NLF women who report that their primary activity is taking care of home responsibilities appear satisfied with their lives, the group of women who are out of the labor force for other reasons report low levels of life satisfaction and high levels of emotional distress. More generous leave time and workplace flexibility provided by private company policies and supported by government policies cou ld possibly help reverse Corporate and government policies . the decline in labor force participation by prime age women and managerial that promote equal pay and the advancement of working women to supervisory , as well as a more robust economic recovery, may also facilitate such a reversal. positions Third, addressing the decades - long slide in labor force participation by prime age men should be a national priority. This group ex s low levels of SWB and reports finding presse relatively little meaning in their daily activities. Because nearly half of this group reported being in poor health, it may be possible for expanded health insurance coverage and preventative care under the Affordable Care Act to positively affe ct the health of prime age men going forward. The finding that nearly half of NLF prime age men take pain medication on a daily basis and that 40 percent report that pain prevents them from accepting a job suggests that pain management potentially be helpful. interventions could 38
41 Evidence presented here suggests that much of the regional variation in opioid across the U.S. differences in medical practices, rather than varying prescription rates is due to labor force participation is lower and fell health conditions that generate pain. Furthermore, in the 2000s in areas of the U.S. that have a higher volume of opioid medication prescribed more per capita than in other areas. Although some obvious suspects can be ruled out – for example, areas with high op prescription rates do not appear to be only masking historical ioid manufacturing strongholds that subsequently fell on hard times – it is unclear whether other factors underlying low labor force participation could have caused the high prescription rates of opioids in certain counties. Regardless of the direction of causality , the opioid crisis and depressed labor force participation And despite are now intertwined in many parts of the U.S. the massive rise in opioid prescriptions in the 2000s, there is no evidence that the incidence of pain has declined; in fact, the results presented here suggest a small upward trend in the incidence of pain for prime age NLF and unemployed men. Addressing the opioid crisis could forts to raise labor force participation and prevent it from falling further. help support ef Lastly, several studies have found that the rise in inequality and shift in demand against ne in labor force participati linked to the decli on less skilled workers in the U.S. are . Although labor market shifts that have lowered demand and wages for less skilled workers have not been a olicies that raise after - tax wages for low - wage workers, such as an increase focus of this study, p in the minimum wage or expansion of the Earned Income Tax Credit, would also likely help raise labor force participation. And the enormous rise in incarceration from the 1980s to the midd - 2000s and rise in males with criminal records are also likely factors that contributed to the decline in male labo r force participation and that could be addressed to reverse the trend. 39
42 References Aaronson, Stephanie, Bruce Fallick, Andrew Figura, Jonathan Pingle, and William Wascher . 2006. “The Recent Decline in the Labor Force Participation Rate and Its Implications for Brookings Papers on Economic Activity, Spring 2006: 69 154. Potential Labor Supply.” - Tomaz Cajner , Bruce Fallick , Felix Galbis - Reig, Christoph Aaronson, Stephanie, and er Smith, William Wascher. “Labor Force Participation: Recent Developments and Future Prospects.” gs Papers on Economic Activity, Fall 2014. Brookin Abraham, Katherine G. and Melissa S. Kearney. 2017. “Explaining the Decline in the U.S. Employment to Population Ratio: A Review of the Evidence,” Unpublished manuscript. Aguiar, Mark, Mark Bils, Kerwin Charles and Erik Hurst. 2016. “Leisure and Luxuries and the Labor Supply of Young Men.” Unpublished manuscript. Autor “ The China syndrome: Local labor , David, David Dorn, and Gordon Hanson. 2013. ” , The American Economic Review market effects of import competition in the United States. (6), pp.2121 103 2168. - Blau, Francine, and Lawrence Kahn. “Female Labor Supply: Why Is the United States Falling Behind?” The American Economic Review , 103 (3): May, pp. 251 - 256. Bound, John, "The Health and Earnings of Rejected Disability Insurance Applicants," American Economic Review , June 1989, 79, 482 - 503. Bureau of Labor Statistics. 2013. “Persons with a Disability: Barriers to Employment, Types of Related Issues Assistance, and Other Labor - – May 2012.” US Department of Labor. April 24, 2013. Case, Anne and Angus Deaton. 2015. “Rising Morbidity and Mortality in Midlife Among White st Non - Century.” Proceedings of the National Academy of Hispanic Americans in the 21 - 112(49): 15078 - 15083. Sciences, Case, Anne and Angus Deaton. 2017. “ Mortality a nd morbidity in the 21st century. ” Brookings Papers on Economic Activity . Charles, Kerwin Kofi, Eri k Hurst and Matthew J. Notowidigdo. 2016. "The Masking of the Decline in Manufacturing Employment by the Housing Bubble." Journal of Economic , 30(2): 179 Perspectives 200 . - Charles, Kerwin Kofi, Eri k Hurst and Matthew J. Notowidigdo. 201 7 . " Housing B ooms, Manufacturing Decline, and Labor Market Outcomes Economic Journal , forthcoming. ." Coglianese, J ohn . 2016. “ Shrinkouts versus Dropouts: Explaining Declines in Labor Force ” Working Paper, Harvard University. . Participation 40
43 (CEA) Counc il of Economic Advisers. 2007. . Economic Report of the President (CEA) Council of Economic Advisers. 2014. “The Labor Force Participation Rate Since 2007: Causes and Policy Implications.” - T (CEA) Council of Economic Advisers. 2016. “The Long - Age Male erm Decline in Prime Labor Force Participation.” Currie, Janet and Molly Schnell. 2017. “Addressing the Opioid Epidemic: Is There a Role for Physician Education?” NBER Working Paper No. 23645. Cambridge, MA: National Bureau of Economic Research. Dahl, Gordon, Katrine V. Løken, Magne Mogstad, and Kari Vea Salvanes. “What is the Case Review of Economics and Statistics (98)4, October, pp. 655 for Paid Maternity Leave?” 670. - Deaton, Angus, and Daniel Kahneman. 2010. “High income improves evaluatio n of life but not emotional well - being.” PNAS 107(38), September 21. Doctor, Jason and Michael Menchine. 2017. “Tackling the Opioid Crisis with Compassion: New Ways to Reduce Use and Treatment.” March 20. https://www.brookings.edu/blog/up - front/2017/03/20/tackling - the - opioid - crisis - with - compassion - new - ways - to - redu ce - use - and - treatment/ Drolet, Marie, Sharanjit Uppal and Sebastien LaRochelle - - U.S. Gap in Cote. 2016. “The Canada Women’s Labour Market Participation.” Statistics Canada. Unpublished manuscript. , Nicholas. 2016. “The Idle Army: America’s Un working Men.” Eberstadt , Wall Street Journal September 1, 2016. Fernald, J.G., Hall, R.E., Stock, J.H. and Watson, M.W., 2017. The Disappointing Recovery of Brookings P apers on Economic Activity . Output after 2009. Frieden, T.R. and Houry , D., 2016. Reducing the risks of relief — the CDC opioid - prescribing (374), pp.1501 guideline. New England Journal of Medicine , 2016 1504. - Goldin, Claudia World War II in the rise of women's employment. The , 1991. The role of American Economic Review 741 - 756. , pp. Goldin, C laudia and Joshua Mitchell, 2017. The New Life Cycle of Women's Employment: Disappearing Humps, Sagging Middles, Expanding Tops. The Journal of Economic Perspectives , 31 (1), pp.161 - 182. Juhn, Chinhui, Kevin M. Murphy, and Robert H. Top el. 1991. “Why Has the Natura l Rate of 126. Unemployment Increased over Time?” Brookings Paper on Economic Activity , 1:75 – 41
44 Juhn, Chinhui, Kevin M. Murphy, and Robert H. Topel. 2002. “Current Unemployment, Historically Contemplated.” , 1:79 - 116. Brookings Papers on Economic Activity and Simon Potter. 2006. “Changes in Labor Force Participation in the United Juhn, Chinhui, Journal of Economic Perspectives - 46. States.” 20(3): 27 of Subjective Kahneman, Daniel, and Alan Krueger. 2006. “Developments in the Measurement - Being.” Journal of Economic Perspectives 20(2), Winter, pp. 3 - Well 24. Eleanor Krause and Isabel Sawhill. 2017. “What We Know And Don’t Know About Declining Labor Force Participation: A Review.” Brookings Center on Children and Families. www.brookings.edu/wp - content/uploads/2017/05/ccf_20170517_declining_labor_force_participation_sawhill1.pdf Krueger, Alan B. and Arthur A. Stone. 2008. “Assessment of Pain: A Community - Based Diary Survey in the USA.” 371(9623): 1519 - 1525. Lancet, Krueger, Alan B., Daniel Kahneman, David Schkade, Norbert Schwartz, and Arthur Stone, “National Time Accounting: The Currency of Lif Measuring the Subjective Well - being of e.” In Nations , Alan Krueger, ed., University of Chicago Press. 2009 Krueger, Alan B., Judd Cramer, and David Cho. 2014. “Are the Long - Term Unemployed on the - Margins of the Labor Market?” Washington, DC: Brookings Panel on Economic Activity , 1 : 229 . 299 Laird, Jessica and Torben Nielsen, 2016. “ The Effects of Physician Prescribing Behaviors on .” Available from: Prescription Drug Use and Labor Supply: Evidence from Movers in Denmark https://scholar.harvard.edu/files/lairdja/files/Laird_JMP_1.pdf Mericle, David. 2017. “The Opioid Epidemic and the U.S. Economy.” Goldman Sachs Economics Research, July 5. Moffitt, 2012 . Robert “ The Reversal of the Employment - Population Ratio in the 2000s: Facts . and Explan ations .” Brookings Papers on Economic Activity , 43, 201 – 50 . Morden, N.E., Munson, J.C., Colla , C.H., Skinner, J.S., Bynum, J.P., Zhou, W. and Meara, E.R., 2014. Prescription Opioid Use among Disabled Medicare Beneficiaries: Intensity, Trends and Regional Variation. , 52 (9), p.852. Medical care e in male labor force participation. Journal of P olitical Parsons, Donald. 1980. The declin Economy , 88 (1), pp.117 - 134. Satel, Sally. 2017. “Taking On the Scourge of Opioids.” National Interest No. 32, Summer. Harper Collins. V ance, J. D . (2016). Hillbilly elegy: a memoir of a family and culture in crisis . 42
45 Von Wachter, T., Song, J. and Manchester, J., 2011. Trends in employment and earnings of allowed and rejected applicants to the social security disability insurance program. The American , Economic Review 3329. 101 (7), pp.3308 - 43
46 Table 1: Labor Force Participation Rates and Population Shares for Selected Demographic Groups Labor Force Participation Rate (%) Share of Population (%) 2017:H1 1997 1997 2007 2007 2017:H1 100.0 62.8 65.6 67.1 100.0 Total 100.0 Men 2.1 22.9 28.7 41.3 16-17 Years 1.8 2.0 18-19 Years 1.8 1.9 47.5 55.2 63.9 1.6 4.2 82.5 78.5 73.6 4.3 20-24 Years 4.5 8.2 8.5 9.6 88.9 92.2 92.9 25-34 Years 8.8 7.7 35-44 Years 92.5 92.2 90.8 10.7 9.1 89.4 45-54 Years 88.2 86.2 8.0 8.1 6.8 70.4 69.6 67.6 55-64 Years 7.9 5.1 65 Years & Over 6.9 6.6 23.9 20.5 17.1 8.6 Women 2.0 1.8 41.0 30.7 24.8 16-17 Years 1.9 1.8 1.7 1.5 47.5 53.7 61.2 18-19 Years 4.3 4.4 4.2 20-24 Years 72.6 70.0 68.2 9.9 8.5 75.3 74.4 76.0 25-34 Years 8.7 8.0 35-44 Years 77.7 75.5 74.8 10.9 9.2 8.4 76.0 76.0 45-54 Years 74.4 8.4 9.6 8.5 7.4 5.5 58.9 50.9 55-64 Years 58.3 9.1 10.7 65 Years & Over 8.6 12.6 15.8 9.1 Aggregate of Demographic Groups — 67.1 — 66.5 65.6 — — — — 63.3 63.4 62.8 Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control adjustments to the Current Population Survey. 2017 represents the average of data from January through June. Source: Bureau of Labor Statistics; author's calculations.
47 Table 2: Time Spent in Selected Activities by Men Ages 21-30 Average Number of Hours Spent Per Week Change From 2004-2007 to 2008-2011 2004-2007 2012-2015 Activity 2012-2015 0.78 61.40 Sleeping 60.62 60.54 34.02 33.02 30.89 -3.13 Work (Including Commuting) Watching TV 17.20 16.71 16.99 -0.21 7.39 -0.03 7.48 7.42 Eating and Drinking 4.05 0.14 3.91 Grooming 4.07 4.71 0.50 Socializing 4.66 5.16 1.64 1.42 0.51 Food/Drink Preparation 1.13 1.37 -0.05 Cleaning 1.41 1.57 0.95 0.10 0.74 0.85 Reading 1.85 1.79 -0.25 Shopping 2.04 0.45 0.40 Laundry 0.56 0.16 1.51 1.38 0.07 Relaxing/Thinking 1.44 Gardening 0.67 0.72 0.74 0.08 Child Care -0.30 2.25 2.39 1.95 3.79 4.66 1.32 3.35 Education 0.67 0.63 -0.14 Adult Care 0.78 1.56 0.60 1.86 Computer Use 1.25 Playing Games 2.05 3.28 3.72 1.67 2,638 2,308 Number of Respondents 2,705 Note: Sample is pooled from 2004 to 2015. Data are weighted using final weights. Averages include respondents who reported no time spent on an activity. Source: Bureau of Labor Statistics (American Time Use Survey).
48 Table 3: Regressions of Various Affect Measures on Activity Indicator Variables and Person Fixed Effects for Men Ages 16-35 Dependent Variable: Affect Measure Meaning Pain Tired Happy Stress Sad (5) (1) (2) (3) (4) (6) 1.540 2.208 0.582 4.209 Constant 0.523 4.168 (0.024) (0.021) (0.027) *** (0.013) *** *** *** (0.023) *** (0.017) *** -0.235 -0.860 -0.022 Gaming Indicator Variable 0.567 -0.215 0.014 (0.123) (0.231) *** (0.109) ** (0.104) * (0.209) (0.052) *** 0.085 -0.627 -0.100 -0.921 TV Indicator Variable -0.052 0.359 (0.070) *** (0.095) (0.047) *** (0.084) *** (0.064) (0.086) -0.413 0.016 -0.321 0.218 -0.252 -1.112 Computer Indicator Variable ** *** *** (0.078) (0.152) (0.154) (0.181) (0.120) ** (0.225) Yes Person Fixed Effects Yes Yes Yes Yes Yes 12,621 Number of Observations 12,594 12,621 12,603 12,618 12,618 Test of Equality of Indicator Variables: 0.000 0.255 0.075 0.005 0.297 0.809 p-value: Gaming = TV p-value: Gaming = Computer 0.030 0.421 0.000 0.067 0.651 0.365 Levels of significance: *** = 0.01, ** = 0.05, * = 0.10. Note: Sample is Well-Being Module pooled over 2010, 2012, and 2013. Regressions are weighted using Well-Being Module adjusted annual activity weights. Source: Bureau of Labor Statistics (American Time Use Survey); author's calculations.
49 Table 4: Self-Reported Health Status for Workers Ages 25-54 by Labor Force Status Not in Labor Force Unemployed Employed (%) (%) (%) Men 19.5 20.0 12.3 Excellent Very Good 36.3 29.2 20.6 24.4 Good 31.9 35.1 25.4 13.9 10.7 Fair Poor 1.2 2.3 17.3 468 683 Number of Respondents 7,277 Women 20.9 16.6 Excellent 16.3 25.6 37.0 Very Good 24.0 Good 30.9 36.3 28.0 18.1 19.3 Fair 10.0 1.1 3.7 Poor 12.1 637 2,265 7,453 Number of Respondents Sample is Well-Being Module pooled over 2010, 2012, and 2013 for individuals Note: ages 25-54. Data are weighted using Well-Being Module final weights. Source: Bureau of Labor Statistics (American Time Use Survey); author's calculations.
50 Table 5: Disability Rate for Men Ages 25-54 Conditional on Labor Force Status Not in Labor Force Employed Unemployed (%) (%) (%) Specific Disability: Difficulty Dressing or Bathing 0.4 7.4 0.2 Deaf or Difficulty Hearing 4.0 0.9 1.5 0.4 1.0 4.0 Blind or Difficulty Seeing Difficulty Doing Errands Such as Shopping 0.3 0.9 14.9 0.8 2.1 19.6 Difficulty Walking or Climbing Stairs 0.8 16.5 2.6 Difficulty Concentrating, Remembering, or Making Decisions Any Disability 2.6 6.0 33.7 0.5 1.6 18.6 Multiple Disabilities 2,130,004 143,446 280,772 Number of Respondents Note: Sample is monthly Current Population Survey data pooled from January 2009 to May 2017 for men ages 25-54. Specific disabilities are not mutually exclusive. Source: Bureau of Labor Statistics (Current Population Survey).
51 Table 6: Prevalence of Pain and Pain Medication for Men Ages 25-54 by Labor Force Status Not in Labor Force Unemployed Employed All Men Ages 25-54 1.92 0.82 0.76 Average Pain Rating (0-6) % 29.6 51.6 % 26.3 Time Spent With Pain > 0 % % Took Pain Medication Yesterday 20.2 % 18.9 43.5 % Number of Respondents 7,277 468 683 Disabled Men Ages 25-54 1.25 1.49 Average Pain Rating (0-6) 2.81 % 52.3 % Time Spent With Pain > 0 % 70.9 42.1 57.7 Took Pain Medication Yesterday 32.4 % % % 12.4 276 25 191 Number of Respondents Note: Sample is Well-Being Module pooled over 2010, 2012, and 2013 for men ages 25-54. Data are weighted using Well-Being Module adjusted annual activity weights. Source: Bureau of Labor Statistics (American Time Use Survey).
52 Table 7: Share of Men Ages 25-54 Taking Prescription Pain Medication Survey Wave 2 Survey Wave 1 No Yes 64.9 No % 8.1 % 6.1 % % Yes 20.9 Note: Sample is 150 respondents. Data are weighted using survey weights that have been adjusted to match age, race, and ethnicity figures from the March 2016 Annual Social and Economic Supplement to the Current Population Survey. Source: Princeton Pain Survey.
53 Table 8: Shares of Men Ages 25-54 Taking Prescription Pain Medication by Methods of Payment Pay by Myself, Out of Pocket 24.7 % Private Health Insurance 13.0 % 37.7 Medicaid % 29.2 % Medicare % Veterans Affairs / Tricare 9.6 Other 10.3 % Note: Sample is 94 respondents. Veterans Affairs and Tricare were not explicit categories, but were often listed if the respondent selected "Other"; so they were moved from "Other" to a separate category. Data are weighted using survey weights that have been adjusted to match age, race, and ethnicity figures from the March 2016 Annual Social and Economic Supplement to the Current Population Survey. Source: Princeton Pain Survey.
54 Table 9: Shares of Men Ages 25-54 by Participation in Income Support Programs Workers' Compensation 1.8 % % Social Security Disability Insurance 35.0 Supplemental Security Income % 10.1 Veterans Disability Compensation 6.0 % % 5.2 Disability Insurance % Other 2.4 % 49.6 None Sample is 571 respondents. The order of response categories, Note: except for "Other" and "None," were randomized across respondents. Data are weighted using survey weights that have been adjusted to match age, race, and ethnicity figures from the March 2016 Annual Social and Economic Supplement to the Current Population Survey. Source: Princeton Pain Survey (September 30, 2016-October 2, 2016).
55 Table 10(a): Subjective Well-Being for Men Ages 16-70 Not in Labor Force Employed All p-value Unemployed Happy 0.429 4.23 4.23 4.22 4.17 1.87 2.19 0.000 Tired 2.17 2.20 1.39 Stressed 0.000 1.43 1.49 1.27 0.54 0.75 0.000 0.60 Sad 0.68 0.73 0.86 1.31 0.000 Pain 0.87 4.24 4.17 0.000 Meaningful 4.04 3.96 0.13 0.13 0.799 0.13 U-Index 0.13 7.08 6.29 6.89 0.000 Cantril Ladder 6.98 29,818 8,503 41,136 2,815 Total Number of Activities Note: Sample is Well-Being Module pooled over 2010, 2012, and 2013. Emotional affects and U-Index weighted using Well-Being Module adjusted annual activity weights. Cantril Ladder question was asked in 2012 and 2013 and was weighted using Well-Being Module final weights. Each respondent was asked about three activities in Well-Being Module. p-value is from an F-test that the means for all three labor force statuses are equal. Source: Bureau of Labor Statistics (American Time Use Survey). Table 10(b): Subjective Well-Being for Men Ages 16-24 Not in Employed All p-value Unemployed Labor Force 4.25 4.16 0.570 4.23 Happy 4.30 2.23 2.23 2.27 0.935 Tired 2.24 1.19 1.24 1.12 0.492 Stressed 1.18 0.42 0.59 0.38 0.087 Sad 0.39 0.46 0.44 0.58 0.43 0.303 Pain 3.75 Meaningful 3.60 0.155 3.85 3.69 0.11 0.09 0.10 0.314 U-Index 0.12 7.06 6.94 6.81 7.36 0.028 Cantril Ladder Total Number of Activities 4,723 842 1,587 2,294 Note: Sample is Well-Being Module pooled over 2010, 2012, and 2013. Emotional affects and U-Index weighted using Well-Being Module adjusted annual activity weights. Cantril Ladder question was asked in 2012 and 2013 and was weighted using Well-Being Module final weights. Each respondent was asked about three activities in Well-Being Module. p-value is from an F-test that the means for all three labor force statuses are equal. Source: Bureau of Labor Statistics (American Time Use Survey).
56 Table 10(c): Subjective Well-Being for Men Ages 25-54 Not in Labor Force Employed All p-value Unemployed Happy 0.010 4.25 4.20 4.18 3.95 1.51 2.52 0.000 Tired 2.23 2.25 1.56 Stressed 0.038 1.59 1.57 1.81 0.55 1.15 0.000 0.62 Sad 0.74 0.76 0.82 1.92 0.000 Pain 0.87 4.27 4.24 0.002 Meaningful 4.23 3.92 0.14 0.22 0.002 0.15 U-Index 0.17 7.03 5.69 6.08 0.000 Cantril Ladder 6.87 21,661 2,025 25,079 1,393 Total Number of Activities Note: Sample is Well-Being Module pooled over 2010, 2012, and 2013. Emotional affects and U-Index weighted using Well-Being Module adjusted annual activity weights. Cantril Ladder question was asked in 2012 and 2013 and was weighted using Well-Being Module final weights. Each respondent was asked about three activities in Well-Being Module. p-value is from an F-test that the means for all three labor force statuses are equal. Source: Bureau of Labor Statistics (American Time Use Survey). Table 10(d): Subjective Well-Being for Men Ages 55-70 Not in Employed All p-value Unemployed Labor Force 4.36 4.27 0.086 4.31 Happy 4.06 1.99 1.78 1.92 0.373 Tired 1.95 1.27 1.37 1.12 0.002 Stressed 1.38 0.70 0.81 0.83 0.001 Sad 0.60 1.19 0.85 1.81 1.60 0.000 Pain 4.41 Meaningful 4.26 0.001 4.50 4.57 0.11 0.14 0.10 0.348 U-Index 0.12 6.84 6.98 5.55 6.19 0.000 Cantril Ladder 10,796 538 4,446 5,812 Total Number of Activities Note: Sample is Well-Being Module pooled over 2010, 2012, and 2013. Emotional affects and U-Index weighted using Well-Being Module adjusted annual activity weights. Cantril Ladder question was asked in 2012 and 2013 and was weighted using Well-Being Module final weights. Each respondent was asked about three activities in Well-Being Module. p-value is from an F-test that the means for all three labor force statuses are equal. Source: Bureau of Labor Statistics (American Time Use Survey).
57 Table 11(a): Subjective Well-Being for Women Ages 16-70 Not in Labor Force Employed All p-value Unemployed Happy 0.002 4.34 4.31 4.35 4.43 2.25 2.48 0.009 Tired 2.50 2.53 1.61 Stressed 0.000 1.61 1.68 1.46 0.58 0.76 0.000 0.65 Sad 0.77 0.82 0.99 1.35 0.000 Pain 1.00 4.38 4.38 0.900 Meaningful 4.38 4.39 0.16 0.14 0.002 0.15 U-Index 0.16 7.08 6.29 6.89 0.000 Cantril Ladder 6.98 31,022 15,256 49,408 3,130 Total Number of Activities Note: Sample is Well-Being Module pooled over 2010, 2012, and 2013. Emotional affects and U-Index weighted using Well-Being Module adjusted annual activity weights. Cantril Ladder question was asked in 2012 and 2013 and was weighted using Well-Being Module final weights. Each respondent was asked about three activities in Well-Being Module. p-value is from an F-test that the means for all three labor force statuses are equal. Source: Bureau of Labor Statistics (American Time Use Survey). Table 11(b): Subjective Well-Being for Women Ages 16-24 Not in Employed All p-value Unemployed Labor Force 4.29 4.40 0.211 4.37 Happy 4.52 2.80 2.28 2.57 0.017 Tired 2.63 1.48 1.50 1.45 0.897 Stressed 1.52 0.45 0.63 0.47 0.047 Sad 0.38 0.62 0.56 0.91 0.55 0.255 Pain 3.97 Meaningful 4.00 0.271 3.88 4.17 0.13 0.13 0.13 0.876 U-Index 0.14 7.06 6.97 6.92 7.29 0.116 Cantril Ladder Total Number of Activities 4,672 780 1,609 2,283 Note: Sample is Well-Being Module pooled over 2010, 2012, and 2013. Emotional affects and U-Index weighted using Well-Being Module adjusted annual activity weights. Cantril Ladder question was asked in 2012 and 2013 and was weighted using Well-Being Module final weights. Each respondent was asked about three activities in Well-Being Module. p-value is from an F-test that the means for all three labor force statuses are equal. Source: Bureau of Labor Statistics (American Time Use Survey).
58 Table 11(c): Subjective Well-Being for Women Ages 25-54 Not in Labor Force Employed All p-value Unemployed Happy 0.037 4.30 4.28 4.31 4.40 2.32 2.60 0.028 Tired 2.57 2.58 1.69 Stressed 0.001 1.72 1.77 1.57 0.60 0.78 0.000 0.66 Sad 0.85 0.83 1.05 1.43 0.000 Pain 0.98 4.40 4.43 0.007 Meaningful 4.64 4.49 0.17 0.14 0.028 0.16 U-Index 0.17 7.24 6.23 7.03 0.000 Cantril Ladder 7.13 22,192 6,736 30,825 1,897 Total Number of Activities Note: Sample is Well-Being Module pooled over 2010, 2012, and 2013. Emotional affects and U-Index weighted using Well-Being Module adjusted annual activity weights. Cantril Ladder question was asked in 2012 and 2013 and was weighted using Well-Being Module final weights. Each respondent was asked about three activities in Well-Being Module. p-value is from an F-test that the means for all three labor force statuses are equal. Source: Bureau of Labor Statistics (American Time Use Survey). Table 11(d): Subjective Well-Being for Women Ages 55-70 Not in Employed All p-value Unemployed Labor Force 4.45 4.46 0.003 4.44 Happy 3.75 2.15 1.53 2.26 0.000 Tired 2.19 1.42 1.49 1.34 0.067 Stressed 1.62 0.79 1.06 0.88 0.001 Sad 0.68 1.36 0.95 1.13 1.76 0.000 Pain 4.61 Meaningful 4.54 0.004 4.70 4.15 0.14 0.23 0.13 0.019 U-Index 0.15 7.16 7.20 6.20 7.35 0.017 Cantril Ladder Total Number of Activities 13,370 422 6,462 6,486 Note: Sample is Well-Being Module pooled over 2010, 2012, and 2013. Emotional affects and U-Index weighted using Well-Being Module adjusted annual activity weights. Cantril Ladder question was asked in 2012 and 2013 and was weighted using Well-Being Module final weights. Each respondent was asked about three activities in Well-Being Module. p-value is from an F-test that the means for all three labor force statuses are equal. Source: Bureau of Labor Statistics (American Time Use Survey).
59 Table 12: Probability Mo dels for Likeliho o d of Taking Pain Medication (1=Yes), Men and Linear Age 16-70 Women (2) (1) (5) (6) (3) Means (4) *** *** *** *** *** *** 0.047 p er 0.060 0.036 Prescrib ed 0.028 6.389 0.026 Capita Log Opioids 0.050 (0.011) (0.010) (0.010) (0.009) (0.010) [0.396] (0.010) ** * * 0.067 Diculty Dressing/Bathing 0.014 0.086 0.069 (0.038) [0.117] (0.041) (0.037) * 0.013 -0.000 0.001 0.057 Impairment Vision (0.029) (0.033) [0.111] (0.030) *** * 0.043 Diculty Hearing 0.017 0.093 0.041 (0.028) (0.026) [0.128] (0.026) *** ** ** Errands 0.025 Doing 0.066 0.065 Diculty 0.104 (0.036) (0.032) (0.031) [0.157] *** *** *** Walking 0.160 Diculty 0.160 0.333 0.057 (0.019) (0.020) [0.231] (0.020) *** 0.029 0.032 0.031 0.067 Rememb ering Diculty (0.021) (0.021) [0.167] (0.023) *** *** *** -0.294 0.192 -0.241 Health Excellent -0.240 (0.012) [0.394] (0.012) (0.012) *** *** *** -0.229 -0.187 0.341 -0.189 Health Very Go o d (0.011) (0.011) (0.011) [0.474] *** *** *** Health -0.118 -0.152 -0.119 0.297 Go o d (0.011) (0.011) [0.457] (0.011) 0.130 0.000 0.000 Health Fair 0.000 [0.336] (0.000) (0.000) (0.000) *** *** *** 0.231 Po or 0.040 0.144 Health 0.146 (0.022) (0.023) [0.195] (0.023) *** *** *** 0.080 0.067 0.073 0.784 White [0.412] (0.010) (0.010) (0.010) *** *** ** Black 0.153 0.031 0.041 0.039 (0.015) [0.360] (0.014) (0.013) *** *** *** -0.046 -0.054 Male -0.047 0.454 (0.006) (0.006) [0.498] (0.007) *** 0.002 0.002 Age 43.761 0.007 (0.001) (0.001) [14.407] (0.001) ** ** -0.003 /1000 Squared 0.032 2.123 0.032 Age (0.016) (0.016) (0.016) [1.267] *** Scho oling of -0.008 Years 0.001 0.002 13.899 (0.001) (0.001) [3.125] (0.001) *** -0.026 0.003 0.512 Married 0.002 [0.500] (0.006) (0.007) (0.007) Atlantic 0.119 -0.034 Mid (0.034) [0.323] North Central 0.159 -0.004 East (0.034) [0.366] North Central 0.081 0.000 West [0.273] (0.034) South Atlantic 0.192 -0.015 [0.394] (0.034) East South Central 0.061 0.000 [0.240] (0.036) South Central 0.002 West 0.114 (0.034) [0.317] Mountain -0.029 0.075 [0.263] (0.034) -0.039 Pacic 0.151 [0.358] (0.034) Squared 0.003 0.044 0.073 0.060 0.119 R 0.120 Note: is ATUS Well Being Supplement for years 2010, 2012, Sample 2013. Sample size is 30,073 and mean of dep endent and variable is .282. Regressions are weighted using the ATUS supplement weights. Standard errors are robust and are clustered omitted at county level, when available, and the state level otherwise. New England is the the region category. Levels of *** ** * Signicance = .01, = .05, = .1
60 Table 13A: Mo dels for Lab or Force Participation of Prime Age Males, 1999-2001 Linear Probability and 2014-2016 (3) (4) (5) (6) (7) (1) Mean (2) * *** *** * * * 0.038 (2014-2016) 0.037 Dummy 0.039 0.067 0.049 -0.032 0.511 Perio d 2 (0.002) (0.022) (0.022) (0.026) (0.026) (0.022) [0.500] *** *** *** ** *** -0.023 Log Opioids p er -0.010 Capita -0.009 by -0.009 County 6.345 -0.015 (0.005) (0.003) (0.004) (0.003) [0.429] (0.005) *** *** *** *** *** -0.016 X 3.244 -0.010 Opioids -0.011 2 -0.013 Perio d Log -0.010 (0.003) (0.003) (0.003) (0.004) [3.188] (0.004) *** *** *** *** 0.597 0.086 0.086 0.085 Married 0.086 (0.002) (0.002) (0.002) [0.491] (0.002) *** *** *** *** 0.037 0.034 0.033 0.034 0.805 White (0.003) (0.003) (0.003) [0.397] (0.003) *** *** *** *** -0.026 -0.025 -0.022 0.118 Black -0.024 (0.004) (0.005) (0.005) (0.004) [0.323] *** *** *** *** 0.038 Hispanic 0.042 0.160 0.038 0.035 (0.003) (0.003) (0.003) (0.003) [0.367] *** *** *** *** 0.012 Age 0.012 39.395 0.012 0.012 (0.001) (0.001) (0.001) (0.001) [8.558] *** *** *** *** Squared -0.186 1.625 -0.187 -0.186 -0.184 /1000 Age (0.009) (0.009) (0.009) (0.009) [0.678] *** *** *** *** 0.013 Years 0.013 of 0.013 Education 0.012 13.567 [3.079] (0.001) (0.001) (0.001) (0.001) *** Manufacturing Share (1999-2001) 0.140 0.090 [0.048] (0.033) 0.010 Share 0.070 -0.008 X Manufacturing Perio d 2 (0.037) (0.031) [0.077] 0.133 -0.009 -0.006 Mid Atlantic (0.005) (0.006) [0.340] North Central 0.153 0.007 0.003 East (0.004) [0.360] (0.004) *** *** 0.067 0.018 West North Central 0.018 [0.250] (0.004) (0.004) Atlantic 0.188 0.003 South 0.000 (0.004) (0.005) [0.391] * ** East South -0.021 Central 0.057 -0.019 (0.010) [0.233] (0.010) West Central 0.116 -0.001 0.002 South (0.006) [0.320] (0.006) Mountain 0.069 0.003 0.009 [0.254] (0.006) (0.005) * 0.168 -0.008 -0.007 Pacic (0.004) (0.004) [0.374] Eects No No County/Area No No No Yes Fixed No 1,824,890 1,824,890 Observations 1,824,890 1,810,246 1,810,246 1,788,508 1,788,508 1,824,890 R Squared 0.003 0.001 0.004 0.055 0.056 0.056 0.063 Note: Sample is Full CPS Monthly, prime age (25-54) men, p o oling years 1999-2001 and 2014-2016. Mean lab or force errors participation Regressions are weighted using the CPS nal weights. Standard 0.891. are robust and are clustered at is *** ** * when available, and the state level. Levels of Signicance: the = 0.01, 0.10. = 0.05, level, = county
61 Table Linear Mo dels for Lab or Force Participation of Prime Age Women, 1999-2001 13B: Probability and 2014-16 (3) (4) (5) (6) (7) (1) Mean (2) * *** ** * 0.087 -0.025 0.055 Dummy 0.047 0.048 0.058 (2014-2016) 0.510 Perio d 2 (0.030) (0.031) (0.031) (0.035) (0.037) (0.003) [0.500] ** * by County 6.348 Log 0.011 0.006 0.011 Opioids 0.010 p er Capita 0.002 (0.010) (0.007) (0.006) (0.005) [0.430] (0.011) *** *** *** *** *** 2 -0.018 Opioids -0.014 3.239 -0.014 Perio d -0.015 -0.016 Log X (0.006) (0.005) (0.005) (0.005) (0.005) [3.190] *** *** *** *** -0.086 Married -0.087 0.601 -0.086 -0.086 (0.005) (0.005) [0.490] (0.005) (0.005) *** *** *** *** 0.061 0.067 0.061 0.061 0.781 White (0.007) (0.007) (0.006) [0.414] (0.007) *** *** *** *** 0.070 0.069 0.077 0.137 Black 0.070 (0.008) (0.008) (0.008) [0.344] (0.007) *** *** *** *** Hispanic 0.149 -0.033 -0.015 -0.024 -0.024 (0.005) [0.356] (0.004) (0.004) (0.004) *** *** *** *** 0.012 0.012 0.012 39.478 Age 0.012 (0.001) (0.001) (0.001) [8.552] (0.001) *** *** *** *** Age -0.149 Squared -0.148 /1000 -0.148 1.632 -0.149 (0.014) (0.014) (0.014) (0.015) [0.679] *** *** *** *** 0.028 0.028 0.028 Years 0.028 of Education 13.742 [2.984] (0.001) (0.001) (0.001) (0.001) (1999-2001) Share 0.139 Manufacturing 0.059 [0.048] (0.053) Manufacturing 2 0.070 Share -0.044 X -0.043 Perio d (0.051) (0.054) [0.077] *** *** Atlantic -0.038 Mid -0.039 0.136 (0.007) [0.343] (0.007) Central 0.151 North -0.009 East -0.007 [0.358] (0.008) (0.008) *** *** 0.065 West 0.043 North 0.043 Central (0.008) [0.246] (0.008) *** *** -0.021 0.193 South Atlantic -0.020 (0.006) [0.394] (0.005) *** *** South 0.059 -0.055 -0.055 East Central (0.006) (0.006) [0.236] *** *** South West -0.033 -0.035 0.115 Central (0.007) (0.007) [0.319] *** *** Mountain -0.023 0.067 -0.026 (0.008) (0.008) [0.250] *** *** -0.032 Pacic -0.032 0.165 [0.371] (0.006) (0.006) Fixed No No No No No No Yes County/Area Eects 1,962,822 1,962,822 1,962,822 1,962,822 Observations 1,947,471 1,924,732 1,924,732 1,947,471 R Squared 0.001 0.000 0.001 0.049 0.051 0.052 0.058 Note: Sample is Full CPS Monthly, prime age (25-54) women, p o oling years 1999-2001 and 2014-2016. Mean lab or force errors participation Regressions are weighted using the CPS nal weights. Standard 0.761. are robust and are clustered at is *** ** * when available, and the state level. Levels of Signicance: the = 0.01, 0.10. = 0.05, level, = county
62 Figure 1: Labor Force Participation Rate Percent (Seasonally Adjusted) 68 67 66 Published Adjusted 65 Population Controls 64 63 Jun - 17 62 61 60 59 58 1958 1968 1978 1988 1998 2008 2018 1948 Note: Data for January 1990 to December 2016 have been adjusted to account for the effects of the annual population control adjustments to the Current Population Survey. Shading denotes recession. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author's calculations.
63 Figure 2: Labor Force Participation Rates by Age & Gender (Seasonally Adjusted) Percent 90 85 (25 Years Men & Over) 80 75 70 17 - Jun 65 Years 24 - 16 60 55 50 Women (25 Years 45 & Over) 40 35 30 25 2018 2008 1998 1988 1978 1968 1958 1948 Shading denotes recession. Note: Source: Bureau of Labor Statistics; National Bureau of Economic Research.
64 Figure 3: Labor Force Participation Rate Percent (Annual Average) 68 Trend Based on Fixed 1997 Demographic 67 Group Weights 66 Trend Based on 65 Actual Demographic Group Weights 64 Each Year 63 2017 62 61 60 59 58 1978 1988 1998 2008 2018 1948 1958 1968 Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control adjustments to the Current Population Survey. 2017 represents the average of data from January through June. Shading denotes recession. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations.
65 Figure 4: Nonparticipation & Idle Rates by Gender 24 for Ages 16 - Percent (Seasonally Adjusted) 50 45 Nonparticipation 2016 Rate: Women 40 35 30 Nonparticipation 25 Rate: Men 20 Idle Rate: Women 15 10 5 Idle Rate: Men 0 1985 2010 2000 1995 1990 2005 2015 Note: Idle refers to neither enrolled in school nor participating in labor force. Shading denotes recession. Source: Bureau of Labor Statistics; National Bureau of Economic Research.
66 Figure 5: Labor Force Participation Rate for Men Ages 25 54 by Educational Attainment - Average ) Percent (Annual 100 98 Degree Bachelor's Total & Higher 96 94 Some College or Degree Associate 92 90 88 High School or Less 86 2017 84 82 1988 1978 1998 2008 2018 1968 1948 1958 Annual averages of monthly data from the Current Population Survey. 2017 represents the average of Note: Shading denotes recession. data from January through May. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author's calculations.
67 Figure 6(a): Probability of Men Ages 25 - 54 Not Being in Labor Force Conditional on Having a Disability Percent 64 63 62 2017 61 60 59 58 57 56 2011 2015 2017 2018 2013 2012 2016 2010 2009 2014 Note: 2017 represents average of data from January through May . Source: Current Population Survey; author's calculations.
68 54 Not Being - Figure 6(b): Probability of Men Ages 25 in Labor Force Conditional by Type of Disability Percent 100 84.1 83.5 90 79.4 74.3 80 70.4 70 60.6 60 51.9 50 34.7 40 30 20 10 0 Mul tiple B lind or Dif fic ulty A ny Dif fic ulty Dif fic ulty Deaf or Dif fic ulty Concent rating, Walk ing or Dis abilit y Doing E rrands Dis abilit ies Dif fic ulty Dress ing or Dif fic ulty Hearing S eei ng R emem beri ng, B athing Clim bing S uch as or M aking S tairs S hoppi ng Deci si ons Note: Average of data from January 2009 through May 2017. Source: Current Population Survey; author's calculations.
69 - Figure 7: Share of Men Ages 25 54 Reporting Experience of Pain in Past 30 Days Percent 60 55 Not in Labor Force 50 Not Employed 45 Unemployed 40 35 30 2015 Employed 25 20 15 10 5 0 2009 2003 1997 2000 2015 2006 2012 Note: Includes back pain, neck pain, leg pain, jaw pain, or a severe headache or migraine. Bars represent Shading denotes recession. one standard error intervals for each year. (National Health Interview Survey) ; National Bureau of Source: Centers for Disease Control and Prevention Economic Research; author's calculations.
70 Figure 8: Percent of NLF Men Age 25 - 54 Taking Pain Medication
71 Figure 9: Female Labor Force Participation Rates by Birth Year and Age Percent of Population of Each Cohort 100 90 Born in 1981 Born in 1971 80 Born in 1961 70 60 Born in 1951 50 Born in 1941 40 30 20 10 0 71 76 61 56 51 46 41 36 31 26 21 66 Age of Middle Birth Year of Cohort Note: Data are from 1962 to 2016. Figure shows the labor force participation rates of five cohorts of women based on ten year - of - birth intervals over the lifecycle from age 21 to age 75. Source: Current Population Survey (Annual Social and Economic Supplement); author's calculations.
72 Figure 10: Ages 25 - 54 Not Looking for Work in Past Year for Reasons Other Than Taking Care of Home Percent 11 10 9 Men 2015 8 7 Women 6 5 4 3 2 1 0 1994 2015 2009 2006 2000 1997 2012 1991 2003 Note: Shading denotes recession. Source: Current Population Survey (Annual Social and Economic Supplement); National Bureau of Economic Research; author's calculations.
73 Figure 11: Retirement Rates by Gender for Ages 16+ Percent of Each Population 24 22 2017 20 Women 18 16 All 14 12 Men 10 8 6 4 2 0 Jan-18 Jan-02 Jan-10 Jan-94 Jan-98 Jan-14 Jan-06 . Note: 2017 represents the average of data from January through May Shading denotes recession. Source: Current Population Survey; National Bureau of Economic Research; author's calculations.
74 Figure 70 Cantril Ladder by - Gender for Ages 16 12(a): Men Women Percent Percent 100 100 90 90 Employed Employed 80 80 70 70 Unemployed 60 60 Not in Unemployed Labor Force 50 50 in Not 40 40 Labor Force 30 30 20 20 10 10 0 0 7 6 3 4 5 2 8 0 1 10 9 5 4 7 3 2 9 0 10 8 6 1 Being Module pooled over 2012 and 2013. - Note: Sample is Well Bureau of Labor Statistics (American Time Use Survey); author's calculations. Source:
75 24 Figure 12(b): Cantril Ladder by Gender for Ages 16 - Women Men Percent Percent 100 100 90 90 Employed 80 80 in Not Unemployed Labor Force 70 70 in Not Unemployed Employed Labor Force 60 60 50 50 40 40 30 30 20 20 10 10 0 0 9 10 8 7 6 5 4 3 2 1 0 3 4 5 6 0 2 10 9 1 8 7 Being Module pooled over 2012 and 2013. - Note: Sample is Well Source: Bureau of Labor Statistics (American Time Use Survey); author's calculations.
76 54 Figure 12(c): Cantril Ladder by Gender for Ages 2 5 - Men Women Percent Percent 100 100 90 90 Employed 80 80 Employed 70 70 Unemployed 60 60 Unemployed 50 50 Not in Not in Labor Force Labor Force 40 40 30 30 20 20 10 10 0 0 0 8 9 6 7 5 4 10 3 2 1 0 8 6 5 4 3 2 1 7 10 9 - Being Module pooled over 2012 and 2013. Note: Sample is Well Source: Bureau of Labor Statistics (American Time Use Survey); author's calculations.
77 Cantril Ladder by 70 - Figure 12(d): 5 Gender for Ages 5 Women Men Percent Percent 100 100 90 90 Employed 80 80 Employed 70 70 in Not Unemployed Labor Force 60 60 in Not 50 50 Labor Force Unemployed 40 40 30 30 20 20 10 10 0 0 8 7 6 5 9 3 2 1 0 10 4 0 9 10 7 8 6 5 4 3 2 1 - Being Module pooled over 2012 and 2013. Note: Sample is Well Source: Bureau of Labor Statistics (American Time Use Survey); author's calculations.
78 Appendix Figure A1: Labor Force Participation Appendix Figure A2: Labor Force Participation - 17 Rate for Men Ages 16 19 Rate for Men Ages 18 - Percent (Annual Average) Percent (Annual Average) 55 80 50 75 45 70 40 65 35 30 60 2017 25 Trend Based 55 Trend Based on Data From on Data From 20 2017 1997 2006 - 2006 1997 - 50 15 45 10 40 5 1968 1948 1988 1998 2008 2018 1958 1978 1948 2018 2008 1998 1988 1978 1968 1958 Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control adjustments to the Current Population Survey. 2017 represents the average of data from January through adjustments to the Current Population Survey. 2017 represents the average of data from January through June. Shading denotes recession. June. Shading denotes recession. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Appendix Figure A3: Labor Force Participation Appendix Figure A4: Labor Force Participation Rate for Men Ages 20 24 - Rate for Men Ages 25 - 34 Percent (Annual Average) Percent (Annual Average) 98 90 97 88 96 86 95 84 94 82 93 80 92 Trend Based Trend Based on Data From 78 91 on Data From - 2006 1997 2017 - 1997 2006 76 2017 90 74 89 88 72 1948 1968 1978 1988 1998 2008 2018 1958 1948 1958 1968 2018 2008 1998 1988 1978 Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control adjustments to the Current Population Survey. 2017 represents the average of data from January through adjustments to the Current Population Survey. 2017 represents the average of data from January through June. Shading denotes recession. June. Shading denotes recession. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations.
79 Appendix Figure A5: Labor Force Participation Appendix Figure A6: Labor Force Participation 44 Rate for Men Ages 35 - Rate for Men Ages 45 - 54 Average) Percent (Annual Percent (Annual Average) 99 97 96 98 95 97 94 93 96 92 95 91 Trend Based 90 94 on Data From - 1997 2006 89 93 88 87 2017 92 Trend Based 2017 86 on Data From 91 2006 - 1997 85 90 84 1968 1978 1988 1998 2008 2018 1958 1948 1958 2018 1948 2008 1998 1988 1978 1968 Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control adjustments to the Current Population Survey. 2017 represents the average of data from January through adjustments to the Current Population Survey. 2017 represents the average of data from January through June. Shading denotes recession. June. Shading denotes recession. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Appendix Figure A7: Labor Force Participation Appendix Figure A8: Labor Force Participation 64 - Rate for Men Ages 55 Rate for Men Ages 65 and Older Percent (Annual Average) Percent (Annual Average) 90 50 88 45 86 84 40 82 35 80 78 30 76 Trend Based 74 Trend Based 25 on Data From on Data From 72 2006 - 1997 2017 2006 1997 - 20 70 2017 68 15 66 64 10 2008 1988 1978 1968 2018 1998 1958 1948 2018 2008 1998 1978 1968 1958 1948 1988 Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control adjustments to the Current Population Survey. 2017 represents the average of data from January through adjustments to the Current Population Survey. 2017 represents the average of data from January through June. Shading denotes recession. June. Shading denotes recession. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations.
80 Appendix Figure A9: Labor Force Participation Appendix Figure A10: Labor Force Participation 17 Rate for Women Ages 16 - - Rate for Women Ages 18 19 (Annual Average) Percent Percent (Annual Average) 64 50 62 45 60 58 40 56 Trend Based 35 54 on Data From Trend Based 1997 - 2006 on Data From 52 30 2006 - 1997 50 2017 2017 25 48 46 20 44 15 42 2018 1948 1958 1968 2008 1998 1988 1978 1988 1998 1948 1968 2008 2018 1958 1978 Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control adjustments to the Current Population Survey. 2017 represents the average of data from January through adjustments to the Current Population Survey. 2017 represents the average of data from January through June. Shading denotes recession. June. Shading denotes recession. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Appendix Figure A12: Labor Force Participation Appendix Figure A11: Labor Force Participation - 34 Rate for Women Ages 25 - Rate for Women Ages 20 24 (Annual Average) Percent Percent (Annual Average) 80 75 2017 75 2017 70 Trend Based 70 Trend Based on Data From on Data From 1997 - 2006 65 65 2006 1997 - 60 60 55 55 50 45 50 40 45 35 30 40 1948 1968 1978 1988 1998 2008 2018 1958 1948 1958 1968 1978 2018 2008 1998 1988 Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control adjustments to the Current Population Survey. 2017 represents the average of data from January through adjustments to the Current Population Survey. 2017 represents the average of data from January through June. Shading denotes recession. June. Shading denotes recession. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations.
81 Appendix Figure A13: Labor Force Participation Appendix Figure A14: Labor Force Participation 44 Rate for Women Ages 35 - - 54 Rate for Women Ages 45 Percent (Annual Average) (Annual Average) Percent 80 80 2017 75 75 Trend Based Trend Based 2017 on Data From on Data From 70 70 2006 1997 - - 2006 1997 65 65 60 60 55 55 50 50 45 45 40 40 35 35 30 2018 1948 1958 1968 2008 1998 1988 1978 1978 1988 1948 1968 1998 2008 2018 1958 Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control adjustments to the Current Population Survey. 2017 represents the average of data from January through adjustments to the Current Population Survey. 2017 represents the average of data from January through June. Shading denotes recession. June. Shading denotes recession. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Appendix Figure A15: Labor Force Participation Appendix Figure A16: Labor Force Participation - Rate for Women Ages 55 64 Rate for Women Ages 65 and Older (Annual Average) Percent Percent (Annual Average) 70 17 2017 65 16 Trend Based on Data From 2017 60 15 1997 - 2006 55 14 50 13 45 12 Trend Based on Data From 40 11 2006 1997 - 35 10 30 9 25 8 20 7 1968 1978 1988 1998 2008 2018 1958 1948 1948 2008 1998 1988 1978 1968 1958 2018 Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control Note: Data for 1990 to 2016 have been adjusted to account for the effects of the annual population control adjustments to the Current Population Survey. 2017 represents the average of data from January through adjustments to the Current Population Survey. 2017 represents the average of data from January through June. Shading denotes recession. June. Shading denotes recession. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations. Source: Bureau of Labor Statistics; National Bureau of Economic Research; author’s calculations.