What Does Human Capital Do to Labor Force Participation of Older - - PDF document

▶

Jun 20, 2023 120 likes •331 views

What Does Human Capital Do to Labor Force Participation of Older Women? Haodong Qi a and Tommy Bengtsson b,c,d a Stockholm University Demographic Unit b Centre for Economic Demography, Lund University c IZA d Centre for Economic Policy Research

SLIDE 1

What Does Human Capital Do to Labor Force Participation of Older Women?

Haodong Qi∗a and Tommy Bengtssonb,c,d

aStockholm University Demographic Unit bCentre for Economic Demography, Lund University cIZA dCentre for Economic Policy Research (CEPR), London

Abstract

Labor force participation (LFP) of older women has been rising in many developed coun- tries since the mid-1990, its causes however have not been well understood yet. This paper determined the extent to which the trend increase in LFP of older women in Swe- den can be explained by human capital development, namely the improvement in average education attainment of the retirement-age population. Our findings suggest that educa- tion played an important role in driving the growth of older women’s labor supply, which is predominately mediated by a remarkable increase in their earning capacity. On the contrary, although average education of older men has been significantly improved, this development explained little of the trend increase in their LFP. JEL Classification: H55, I25, J26 Keywords: Education, Human Capital, Labor Supply, Older Women

∗Correspondence: Haodong.Qi@sociology.su.se

1

SLIDE 2

1 Introduction

Older worker’s labor force participation (LFP) has been rising in many developed coun- tries since the mid-1990s. Effective retirement age, on average, increased from 63.2 to 64.6 for men during 1998-2014 across 34 members of OECD; this growth is even steeper for women, from 61.1 to 63.21. While the existing literature mostly suggested that in- terventions, such as raised statutory pension age, and/or financially discouraging early retirement, played an important role, many studies did not fully explain the overall in- creases in LFP, see for example Staubli and Zweimuller (2013); Manoli and Weber (2016); Atalay and Barrett (2015); Cribb et al. (2013); Pingle (2006); Mastrobuoni (2009); Hanel and Riphahn (2012); Karlstr¨

m et al. (2008); Glans (2008). Some even showed that the

magnitude of these policy effects is very much in doubt (Qi et al., 2016; Qi, 2016). This paper seeks to understand what else (besides changing retirement policy) might explain the recent increase in retirement age. Specifically, we examine the extent to which human capital development has driven the growth in LFP of older workers. Heckman (1976) argue that there are two pathways through which initial human cap- ital endowment may affect labor supply. First, it may increase labor supply through a substitution effect of greater earning capacity. Second, it may reduce work activities due to an increase in demand for effective leisure (given that leisure is a normal good). However, the leisure and substitution effects offset each other, hence the net labor supply effect of an increase in human capital endowment is unsigned. As noted by Heckman (1976), substitution effect tends to outweigh leisure effect at younger age due to steep wage growth over early life cycle, whereas the leisure effect might dominate at older age because the endowment effect on earning capacity diminishes by age (in the presence of human capital depreciation). However, the existing empirical literature generally suggest that the net effect of greater human capital endowment still remain at older age, as many have shown that high-educated workers tend to retire later, compared to low-educated (B¨

rsch-Supan et al., 2009; Buchholz et al., 2013; Glans, 2008; Klevmarken, 2010; Komp

et al., 2010; Larsen and Pedersen, 2013; Stenberg and Westerlund, 2013). Nevertheless, the existing empirical evidence seldom distinguish between the substitution and leisure

1Source: OECD estimates based on the results of national labour force surveys, the European Union

Labour Force Survey, and, for earlier years in some countries, national censuses. Note: data is available at: http://www.oecd.org/els/emp/average-effective-age-of-retirement.htm

2

SLIDE 3

effect of an increase in human capital endowment. In this paper, we identify these two effects, respectively. Tertiary education was expanding during the second half of the 20th century world- wide (Altbach et al., 2009). Many birth cohorts who were early beneficiaries from these expansions have now reached their pensionable age. This led to a trend increase in the average education of retirement-age population over the past two decades. If the net labor supply effect of education remain positive at older age, we would expect to see a contribution of growth in older workers’ human capital endowment to the trend increase in their LFP. While this development may seem obvious to explain the rising retirement age, the existing empirical literature has rarely examined such a relation. To the best

f our knowledge, only two studies have showed that the growth in educational attain-

ment can explain a substantial part of the recent increase in older men’s LFP (Blau and Goodstein, 2010; Schirle, 2008). However, neither of these two studies provide insights into what drives the growth in labor force participation of older women. This paper fills this void. Understanding the trend in older women’s labor supply is of great importance to foresee future development of aggregate LFP. As noted earlier, in many OECD countries, the increase in effective retirement age was more profound among women during the past two decades. If this trend continues, older women may constitute a dominating source

f human resource, which is vital for many developed nations to alleviate the challenges
f population ageing. Moreover, older women’s labor supply may also have an indirect

impact on older men. As Schirle (2008) showed that a large part of the increase in older men’s LFP is attributable to the growth in their wives’ labor supply, due to a so-called “shared leisure effect”. Interestingly, although the rising older women’s LFP may be an important motive for older men to prolong their working life, what encourage older women to work longer is not well understood. While Blau and Goodstein (2010); Schirle (2008) were among the first showing that the growth in educational attainment explained a substantial part of the increase in old- age LFP. Their analyses, however, only covered three countries: the US, the UK, and

Canada. This paper adds the evidence from Sweden to this scarce literature. There

are two reasons why Sweden is of interest in this regard. First, while tertiary education 3

SLIDE 4

expansion was phenomenal worldwide during the second half of the 20th century, Sweden is a country that exemplified this process; not only did enrollment and completion rates rapidly increased, but also the number of higher education institutes (Chudnovskaya and Kolk, 2015). This radical development puts Sweden as an interesting case for studying the long-term impact of human capital development on old-age labor supply. Second, Sweden is one of the first four countries in the world introduced Notional Defined Contribution (NDC) scheme in the mid-1990s. Despite the system technically provides strong incentives for late retirement, its implementation in Sweden merely explained 5% of the recent increase in women’s retirement age (Qi et al., 2016). Hence, uncovering what drives the unexplained LFP growth holds important implication for what additional policy options might complement the current pension reform strategy.

2 Data and Method

The analysis in this paper relies on Swedish register data. we use the STAR (Sweden in Time: Activities and Relations) collections of data organized by the Demography Unit and the Swedish Institute for Social Research at Stockholm University. The data covers entire population residing in Sweden during the period 1993-2012. The large sample size ensures that we can observe a sufficient number of woman who are still engaging in work activities at older age. As shown in Qi et al. (2016), older workers born after 1937 expected to receive a lower

ld age pension, compared to older cohorts, due to the 1994 Swedish pension reform. The

magnitude of the pension reductions over these notch generations however was not large enough to explain the trend increases in retirement age. Particularly, it merely explained 5% total increase in women’s retirement age. Hence, this paper seeks to uncover what

ther factors might contribute to a large proportion of unexplained increases. For this,

we extract a subpopulation from the register who were born during the period 1937-

1945. The oldest cohort in this study population were unaffected by the Swedish pension

reform in 1994, whereas the younger ones were facing a continuous decline in their old age pension benefits. While our primary focus is on older women’s labor force participation (LFP), we include older men throughout our analysis, as a comparison group. 4

SLIDE 5

The outcome of interest in this paper is individual worker’s LFP, which is measured based on the annual information on pension and labor income in the Swedish register. We compute LFP using equation (1), in which, Pension encompasses old-age public pension and early(disability) retirement benefits, and Labor entails earned income from employ- ment, including self-employment. LFP = Labor Pension + Labor (1) The resulting value from (1) falls between 0 and 1. Had this value equal to zero, an individual is fully retired; conversely, a value of 1 indicates a status of full time

working. Any value between 0 and 1 reflects the income associated with labor supply

as a percentage share of total income from labor and pension during a given year. We interpret this value as intensity of labor supply at older age. As stated earlier, the central question we seek to address is whether the recent trend increases in old-age labor supply are attributable to the growth in average education attainment of the recent retirement-age population. Since education is typically concen- trated over early life cycle, it may be regarded as initial human capital endowment that individuals bring into labor activities, which is expected to have an positive influence on labor market outcomes. However, in his life cycle model, Heckman (1976) showed that the net effect of an increase in initial human capital on labor supply is unsigned (particu- larly at older age); there are two potential mediating factors (increased earning capacity and effective leisure) that are offsetting each other. Specifically, a higher level of initial human capital may exert a positive effect on labor supply through a substitution effect of an increase in earning capacity, it may reduce labor supply due to increasing demand for effective leisure (assuming leisure is a normal good). To identify the substitution effect, a measure of earning capacity is necessary. For this, we compute the average annual wage

ver age 56-59 for each individual in our study sample, and use it as an indicator for

earning capacity. 5

SLIDE 6

Decomposition Analysis

To examine the relation between rising initial human capital and trend increase in older worker’s labor force participation, we adopt a decomposition method developed by Oax- aca and Ransom (1994). This decomposition method quantifies the contribution of chang- ing average educational attainment to the change in LFP between cohorts. For simplicity, in the decomposition analysis, we only focus on the difference between the youngest and the oldest cohort in our study population. The oldest cohort born in 1937 was unaffected by the 1994 Swedish pension reform, and the youngest cohort born in 1945 was facing an average reduction of 10% for men and 6% for women in their pension benefits (see Qi et al. (2016) for details). We begin by quantifying the contribution of rising educational attainment to the in- crease in earning capacity over age 56-59. Let W be an individual’s earning capacity at

lder age (age 56-59), X be a set of dummy indicators for initial level of human capital

(measured by education attainment: primary, secondary, and university or higher). We first estimate cohort-specific earning regression models, which may be written as: W(i, c) = αW(c) + βW(c)X(i, c) + ǫW(i, c) (2) Differentiating (2) with respect to cohort c, we get: dW(i, c) = dαW(c) + βW(c)dX(i, c) + dβW(c)X(i, c) + dǫW(i, c) (3) For simplicity, we denote dBW to all terms that are not associated with the change in the average value of education attainment, and rewrite (3) as: dW(i, c) = dBW(c) + βW(c)dX(i, c) + dǫW(i, c) (4) Using Oaxaca decomposition, we obtain the estimate of βW(c)dX(i, c), which is the contribution of change in average level of education to change in earning capacity. 6

SLIDE 7

We then estimate a cohort-specific regression on LFP in the form of: L(i, c) = αL(c) + βL(c)X(i, c) + γL(c)W(i, c) + ǫL(i, c) (5) Differentiating (5) with respect to cohort, we get: dL(i, c) = dαL(c)+βL(c)dX(i, c)+dβL(c)X(i, c)+γL(c)dW(i, c)+dγL(c)W(i, c)+dǫL(i, c) (6) For simplicity, we denote all terms associated with coefficient change in (6) by dBL, and rewrite (6) as: dL(i, c) = dBL(c) + βL(c)dX(i, c) + γL(c)dW(i, c) + dǫL(i, c) (7) We use Oaxaca decomposition to estimate the contribution of change in education composition (βL(c)dX(i, c)) and of change in average earning capacity (γL(c)dW(i, c)), respectively, to the overall increase in LFP.

Net Effect of Rising Initial Human Capital on LFP

It is noticeable from (7) that βL(c)dX(i, c) is a direct effect of compositional change in education on LFP. This is essentially an analogy to the effective leisure effect of an increase in initial human capital on labor supply, as noted in Heckman (1976). To identify the substitution effect, we substitute (4) into (7), and denote all changes in LFP associated with cross-cohort coefficient change as dB: dL(i, c) = dB(c) + βL(c)dX(i, c) + γL(c)βW(c)dX(i, c) + γL(c)dǫW(i, c) + dǫL(i, c) (8) The third term on the right hand side of equation (8) is the effect of changing education composition on LFP that is mediated through the change in earning capacity, which is however not directly estimable by decomposition. Instead, we compute the this mediated effect using the estimate of γL(c) obtained from the regression model in (5) and the estimate of βW(c)dX(i, c) obtained from the model in (4). Note, in decomposition, we 7

SLIDE 8

set cohort 1937 as the reference, hence the coefficient γL(c) refers to the 1937 cohort, i.e. γL(1937). This coefficient essentially corresponds to the substitution effect of an increase in earning capacity. And finally, the net effect of rising initial human capital on LFP can be computed by the sum of two components: 1) a direct effect of compositional change in education (or effective leisure effect), and 2) an indirect effect mediated by change in earning capacity (or substitution effect).

3 Descriptive Results

In this section, we present an overview of how initial human capital endowment, earn- ing capacity, and labor force participation develop over the notch generations who were gradually affected by the 1994 Swedish pension reform. Figure 1 shows that, for both genders, there were visible compositional changes in initial human capital. As seen from the variation along the horizontal axis, the completion rate

f university or higher education grew substantially across cohorts born 1937-45. This

trend is likely to be the outcome of expanded tertiary education in Sweden since the 1950s. However, these improvements were more profound among women, compared to men. For example, in terms of the change between 1937 and 1945 cohort, the proportion of women completed university or higher education increased by 8.5 percentage points (43%), while the corresponding change for men was 6 percentage points (30%). In addition, earning capacity at older age also exhibits a trend increase across cohorts. As seen from the change along the vertical axis in Figure 1, individual’s average annual wage over aged 56-59 grew by 36% for women and 31% for men between 1937 and 1945 cohort. Note, wages are adjusted for inflation based on 2012 price level. These trends shown in Figure 1 indicate a positive relation between education and earning capacity in later life, which somewhat contradicts with the notion in (Heckman, 1976) that the initial endowment effect on earning capacity steadily diminishes over the life cycle, and particularly weakens at older age. Specifically, Heckman (1976) argue that higher initial human capital reduces subsequent investment in human capital accumula- tion over the life cycle, and consequently slows down the growth in human capital stock, in the presence of depreciation. On the contrary, Figure 1 shows a correlated develop- 8

SLIDE 9

ment in education and earnings, which suggests that the effect of initial human capital

n human capital stock persists over the life cycle, and remain noticeable at older age.

Figure 2 shows that not only does the growth in initial human capital increase earn- ing capacity, it also increases older workers’ labor force participation (LFP). LFP over age 60-67 increased by roughly 10 percentage points for both gender (as seen from the changes along the vertical axis). This suggests that the net effect of an increase in initial human capital on old-age labor supply is positive, or equivalently the substitution effect

f increased education attainment outweighs the growing demand for leisure (with the

assumption that leisure is a normal good). In the next section, we present the estimates for the substitution, leisure, and net effects of rising education attainment, respectively.

4 Regression Results for Earning Capacity and LFP

Table 1 presents the coefficient estimates for cohort-specific regressions of earning capacity (over age 56-59) on initial human capital endowment (measured by education levels: primary, secondary, and university+). These estimates may essentially be interpreted as the return to schooling in later life within each birth cohort. The returns, particularly to college graduates, remain substantial even over the period right before pensionable age. This suggests that higher initial human capital does increase earning capacity at all ages, and this effect is still strong at older age. Table 2 reports the estimates for cohort-specific regressions of individuals’ average an- nual LFP over age 60-67 on education attainment and earning capacity at older age (as specified in (5)). The first two rows show that higher education generally reduces LFP when the model is adjusted for earning capacity. This is in line with the notion by Heck- man (1976) that higher initial human capital increases effective leisure, and thus reduces labor supply. The third row in Table 2 indicate a positive impact of older workers’ earning capacity

n LFP. If we assume that earning capacity prior to pensionable age forms the wage

expectation during age 60-67, this positive effect is essentially a substitution effect of a wage increase on labor supply; for every thousand increase in expected wage, LFP rises by 0.005-0.006 p.p. for men, and by 0.014-0.016 p.p. for women. 9

SLIDE 10

Using the above-estimated substitution effect of wage increase, and multiplied by the return to schooling in Table 1, we can compute the substitution effect of an increase in initial human capital. Taking the 1937 cohort as an example, men and women graduated from college earn (on average) 175 and 118 more (in thousand SEK per year) during age 56-69, compared to those attained only primary education. Multiplying this wage premium by the effect of earning capacity on LFP (0.0006 for men and 0.0016 for women), we get an increase of 10.5 p.p. in LFP for older men, and 18.9 p.p. for older women. It is noticeable that the substitution effect of increased education on labor supply is 80% greater for women than for men. Given the leisure and substitution effects of higher education, we can also compute the net effect of an increase in initial human capital on labor force participation. For example, LFP of male college graduates born in 1937 is 6.4 p.p. higher, compared to workers with

nly primary education. The corresponding figure for female college graduates is 11.6

p.p..

5 Decomposition Results

As noted in the beginning, higher education expansion was phenomenal worldwide during the second half of the 20th century (Altbach et al., 2009). This development has led to a substantial increase in the average education attainment of the retirement-age population since the 1990s. In this section, we present empirical estimates which quantifies the long- term impact of human capital development on older workers’ labor force participation in Sweden. Table 3 reports the estimates for the decomposition models that were specified in (4) and (7). The second panel from the top is of our main interest, as it corresponds to the extent to which rising average education attainment contributed to the growth in LFP. As shown in Figure 1, the proportion of women completed university or higher education increased by 43% between 1937 and 1945 cohort. This increase has made women born in 1945 received an annual wage (over age 56-59) that was 10 thousand SEK (7%) higher than those born in 1937. On the contrary, the number of college graduated men increased by 30%, which increase their earning capacity by 4.8%. These figures imply that the 10

SLIDE 11

growth in initial human capital endowment (due to the expansionary tertiary education in Sweden) was steeper among older women born between 1937 and 1945, which in turn exerted a more profound long-term impact on earning capacity, compared to their male counterpart. In addition, between the cohort born in 1937 and 1945, the overall increase in older women’s earning capacity was 51 thousand SEK (36%). This earning growth created a substitution effect on labor supply, which increased LFP by 8.3 p.p., or 24%. In

ther words, the increased earning power of older women contributed to 83% of the total

increase in female labor force participation at older age. By contrast, older men’s earning capacity increased by 68 thousand SEK (31%), which has led to a 4.3 p.p. (11.4%) increase in LFP. And the rising earning capacity of older men merely contributed to 43% of the overall increase in their LFP. These figures imply that the substitution effect

f increased earning capacity is stronger on labor force participation of older women,

compared to older men. It is noticeable that the estimates for education are negative in decomposing the overall increase in LFP. This negativity essentially corresponds to the leisure effect of increase in initial human capital (as seen from the negative coefficient estimates reported in Table 2). As noted earlier, higher initial human capital may reduce LFP at older age due to increased demand for leisure, a notion of effective leisure effect by Heckman (1976). Hence, as average level of education grew across cohorts, LFP decreased proportionally due to increased aggregate demand for leisure. However, it is clear that this leisure effect is largely outweighed by a substitution effect of rising earning capacity, which is partially driven by the increase in average education attainment. To fully elaborate this point, we use (8) to identify the substitution effect of increasing average level of education. The results are reported in Table 4. As shown in (8), we identify the substitution effect of rising initial human capital on LFP by multiplying two parameters: βW(c)dX(i, c) and γL(c). The former is the effect

f changing average level of education on earning capacity, that is the estimates for the

explained part of earning capacity difference in Table 3 (13.9 and 12.7 thousand SEK for men and women, respectively). The latter is the substitution effect of an increase in earning capacity on LFP, that is the coefficient estimates for earning capacity in Table 11

SLIDE 12

2. Note, in order to identify the effect of compositional change in education attainment
n LFP, we hold the coefficient γL(c) constant across cohort, and evaluate it at γL(1937)

(i.e. 0.0006 for men and 0.0016 for women). As seen from the second row in Table 4, the substitution effect of rising initial human capital endowment is 0.02 for women, which is equivalent to 20 % of the total growth in their LFP at older age. The corresponding figure for men, however, is merely 0.008 (7,4% of the overall increase in older men’s LFP). Taking the sum of the leisure and substitution effects, we get 0.005 for men and 0.012 for women as the net effect of rising initial human capital on LFP. And these figures imply that the increase in average education attainment explained 12 % of the LFP growth for

lder women, yet only 4.6% for older men.

6 Conclusion

While many developed countries have witnessed noticeable increases in older workers’ labor force participation (LFP) since the mid-1990s, the causes of these trends have not been fully understood yet. Raised statutory pension age and/or financial disincentives for early retirement are seemingly plausible explanation for why older workers are increas- ingly engaged in labor activities, they however do not fully explain the overall growth in LFP. This paper sheds new light on what has driven the rise of old-age labor supply. Specifically, we determine the extent to which human capital development might explain the increase in old-age LFP among those who were affected by the Swedish NDC pen- sion scheme (namely those born between 1937 and 1945). In concluding this paper, we highlight a few important findings in our analysis. First, the estimates in our regression analysis are in stark contrast to the notion that the substitution effect of an increase in human capital endowment tends to be outweighed by leisure effect at older age. Specifically, we show that higher education attainment largely increases earning capacity at older age, and this wage premium further creates a significant substitution effect on older workers’ labor supply. Second, tertiary education expansion since the mid-20th century in Sweden did improve average education of current retirement-age population, particularly older workers who attained college or higher degree increased considerably between cohort born in 1937 and 12

SLIDE 13

1945. This striking development contributed to a substantial increase in older workers’

earning capacity, which in turn explained a noticeable amount of growth in LFP. Although growing education attainment did increase aggregate demand for leisure, this leisure effect however is too little to prevent LFP from rising. The last, perhaps the most striking, finding is the gender difference in what drives the growth in old-age LFP. While human capital development explained 12% of the overall growth in female LFP between the 1937 and 1945 cohort, it only explained 4,6% of total increase in male LFP. What determines this difference is the substitution effect

f increase in earning capacity on labor supply. As both male and female labor force

became increasingly educated over time, their earning capacity were rising accordingly. However, older women’s labor supply is nearly three times more responsive to wage increase, compared to men. As a result, through the substitution effect of increased earning power, the growth in older women’s education attainment explained 20% of the

verall growth in their LFP, whereas the corresponding figure for men is merely 7%.

The above-mentioned gender distinction improved our understanding on why older workers’ LFP has been rising in Sweden since the mid-1990s, and more specifically why changing incentives from public pension reform merely explained 5% growth in older women’s LFP (but one-third in older men’s), a stark contrast previously shown in Qi et al. (2016). Reconciling the evidence here with our previous findings, we conclude that while income effect of reduced pension benefits substantially increased older men’s LFP, higher education mediated by greater earning power played an important role in driving the growth in LFP of older women. This holds an important implication for what options might help increase overall old-age labor supply in addition to the existing reform strategy

f pension system; solely relying on the income effect of lowering retirement benefits might

not be sufficient (as it only incentivized older men), rather, exploiting the substitution effect of enhanced earning capacity on labor supply might be a good complementary

ption.

Regarding the future trend in old-age labor supply in Sweden, our findings present two promising news. First, older women’s labor force participation might continue to rise as younger cohorts (born after 1945) had become more educated than their older counterparts through the expansion in tertiary education. Second, the growth in older 13

SLIDE 14

women’s LFP might be an additional motive for older men to work longer, through the “shared leisure effect” suggested by Schirle (2008). This potential source of growth in labor supply may alleviate some challenges of population ageing that the Swedish welfare state confronts. However, whether the growth is realizable depends on a key factor, the earning capacity at older age. As shown in our results, the long-term impact of human capital development on older women’s LFP is entirely mediated by their increased earning

power. And more importantly, the growing earning power itself can explain 83% of the

total increase in LFP of older women. Hence, if earning power continue to rise, the trend increase in labor force participation of older women is likely to persist.

Acknowledgments

We are grateful for financial support from the Swedish Research Council (Vetenskapsrdet) via the Linnaeus Center for Social Policy and Family Dynamics in Europe (SPaDE), grant registration number 349-2007-8701.

References

Altbach, P. G., L. Reisberg, and L. E. Rumbley (2009). Trends in global higher education: Tracking an academic revolution. Atalay, K. and G. F. Barrett (2015, March). The Impact of Age Pension Eligibility Age

n Retirement and Program Dependence: Evidence from an Australian Experiment.

The Review of Economics and Statistics 97(1), 71–87. Blau, D. M. and R. M. Goodstein (2010). Can social security explain trends in labor force participation of older men in the united states? The Journal of Human Re- sources 45(2), 328–363. B¨

rsch-Supan, A., A. Brugiavini, and E. Croda (2009). The Role of Institutions and

Health in European Patterns of Work and Retirement. Journal of European Social Policy 19(4), 341–358. 14

SLIDE 15

Buchholz, S., A. Rinklake, and H. P. Blossfeld (2013). Reversing Early Retirement in Germany A Longitudinal Analysis of the Effects of Recent Pension Reforms on the Timing of the Transition to Retirement and on Pension Incomes. Comparative Popu- lation Studies 38(4). Chudnovskaya, M. and M. Kolk (2015). Educational expansion and intergenerational proximity in sweden. Population, Space and Place. Cribb, J., C. Emmerson, and G. Tetlow (2013, Mar). Incentives, shocks or signals: labour supply effects of increasing the female state pension age in the uk. Glans, E. (2008, November). Retirement Patterns during the Swedish Pension Reform. Working Paper Series 2008:9, Uppsala University, Department of Economics. Hanel, B. and R. T. Riphahn (2012). The timing of retirement new evidence from swiss female workers. Labour Economics 19(5), 718 – 728. Special Section on: Price, Wage and Employment Adjustments in 2007-2008 and Some Inferences for the Current European Crisis. Heckman, J. J. (1976). A life-cycle model of earnings, learning, and consumption. Journal

f Political Economy 84(4), S9–S44.

Karlstr¨

m, A., M. Palme, and S. Ingemar (2008). The Employment Effect of Stricter

Rules for Eligibility for DI: Evidence from A Natural Experiment in Sweden. Journal

f Public Economics 92, 2071–2082.

Klevmarken, A. (2010). Vereforms efter 65 ˚ ars ˚ alder? En statistisk analys. Report fr˚ an sociala r˚ adet, Statens Offentliga Utredningar. Komp, K., T. Tilburg, and M. B. Groenou (2010). Paid Work between Age 60 and 70 Years in Europe: A Matter of Socio-economic Status? International Journal of Ageing and Later Life. Larsen, M. and P. Pedersen (2013). To Work, to Retire, or Both? Labor Market Activity after 60. IZA Journal of European Labor Studies 2(1). 15

SLIDE 16

Manoli, D. S. and A. Weber (2016, August). The effects of the early retirement age on retirement decisions. Working Paper 22561, National Bureau of Economic Research. Mastrobuoni, G. (2009). Labor supply effects of the recent social security bene- fit cuts: Empirical estimates using cohort discontinuities. Journal of Public Eco- nomics 93(1112), 1224–1233. Oaxaca, R. L. and M. R. Ransom (1994). On Discrimination and the Decomposition of Wage Differentials. Journal of Econometrics 61(1), 5 – 21. Pingle, J. F. (2006). Social Securitys Delayed Retirement Credit and the Labor Supply of Older Men . Federal Reserve Board, Washington, D.C. Qi, H. (2016). Live Longer, Work Longer? Evidence from Sweden’s Ageing Population.

Ph. D. thesis, Lund University.

Qi, H., J. Helgertz, and T. Bengtsson (2016). Do notional defined contribution schemes prolong working life? evidence from the 1994 swedish pension reform. The Journal of the Economics of Ageing. Schirle, T. (2008). Why have the labor force participation rates of older men increased since the mid?1990s? Journal of Labor Economics 26(4), 549–594. Staubli, S. and J. Zweimuller (2013). Does raising the early retirement age increase employment of older workers? Journal of Public Economics 108, 17 – 32. Stenberg, A. and O. Westerlund (2013). Education and Retirement: Does University Education at Mid-Age Extend Working Life? IZA Journal of European Labor Stud- ies 2(1). 16

SLIDE 17

Figure 1: Education and Earning by Cohort 1937-45, Sweden

0.18 0.20 0.22 0.24 0.26 0.28 0.30 150 200 250 300 Univeristy or Higer Education Rate Individual Average Annual Wage Age 56−59 (in Thousand)

42 43 44 45 37 38 39 40 41 42 43 44 45

Women

Figure 2: Education and Labor Force Participation by Cohort

0.18 0.20 0.22 0.24 0.26 0.28 0.30 0.35 0.40 0.45 0.50 Univeristy or Higer Education Rate Labor Force Participation