Income Trajectories Around Men's Retirement

Table 7.1 summarizes income trajectories around men’s retirement in Germany and the United States. Men are divided into three groups according to their year of retirement, with period definitions differing slightly between Germany and the US due to the somewhat different overall observation periods. Superscripted numbers indicate if the estimate for a given period differs significantly (at the 10 per cent level, twotailed tests) from the corresponding estimates for the other periods. For example, the entry ‘—66.5 (2, 3)’ in the first cell means that, on average, the needs-adjusted pre-government income of German men who retired between 1985 and 1992 was 66.5 per cent lower after retirement than before, and that this value is statistically significantly different from the average changes experienced by German men who retired in the other periods (that is, periods 2 and 3).

Table 7.1 contains information on German—American differences as well as (within-country) changes over time. As for the country comparison,

Table 7.1 I ncome change around men's retirement by country and year of retirement

Germany

United States

Period 1 1985-1992

Period 2 1993-1998

Period 3 1999-2007

Period 1 1981-1988

Period 2 1989-1994

Period 3 1995-2002

Avg. change in pre-government income (in % of pre-retirement value)

-66.5 (2-3) [2.3]

-71.5 (1) [1.7]

-72.6 (1) [1.3]

-49.9 [2.4]

-51.9 [2.6]

-49.2 [3.2]

Avg. change in post-government income (in % of pre-retirement value)

-10.8 <2-3> [2.0]

-15.3 <'-«[1.6]

-19.6 <'-2> [1.2]

-12.8 <2> [2.0]

-19.8 <’> [2.4]

-18.5 [2.9]

% with post-government loss >1/3

23.1 (2-3) [2.3]

28.9 (1) [2.2]

31.1 (1) [2.0]

24.8 (2-3) [2.6]

37.1 (1) [3.3]

40.9 (1) [3.5]

% with post-government loss >1/2

9.5 [1.6]

10.5 [1.5]

13.0 [1.4]

7.6 (2-3) [1.5]

14.5 <'-3> [2.3]

20.7 <'-2> [2.9]

% entering income poverty

15.1 [2.0]

11.6 <3> [1.6]

15.9 (2) [1.6]

12.9 <3> [2.2]

11.8 (3) [2.3]

21.7 (1-2) [3.4]

Post-retirement obs. (person years) 643/608 all/not poor before retirement

790/749

1134/1098

519/449

362/321

343/305

Post-retirement obs. (persons) all/ not poor before retirement

347/328

416/394

702/680

280/243

196/172

212/188

Person-level cluster-robust standard errors in brackets. Superscripted numbers in parentheses indicate that the within-country period difference for a given income measure is significantly different from zero at the 10% level (two-tailed tests). For example, the superscript (1, 2) indicates that the Period 3 estimate is significantly different at the 10% level from the Period 1 and Period 2 estimates, avg. = average, obs. = observations Sources'. SOEP, PSID, CNEF, author's calculations results confirm that private income sources play a greater role for retirement income provision in the United States. For German men average changes in pre-government income lie between 66.5 and 72.6 per cent, whereas for American men they are in the neighbourhood of 50 per cent for all three periods. Changes in post-government income are more similar. Other than perhaps for the final period, there is no evidence that American men faced greater risks of falling below the poverty line upon retirement—a finding that stands in some tension with the widespread assumption that the ‘residual’ American welfare state leaves larger parts of the population economically vulnerable and disadvantaged. It should be noted, however, that the proportion of men who were poor already before retirement is higher in the United States (see Heisig 2015, chap. 6).

Turning to differences between retirement cohorts, Table 7.1 shows that income trajectories around retirement have become less favourable in both countries, but the precise pattern differs. In Germany, there is a clear trend towards larger average losses in disposable income: whereas men who retired between 1985 and 1992 saw their disposable income fall by approximately 10.8 per cent on average, men who retired between 1999 and 2007 experienced declines of almost 20 per cent—an increase that is statistically significant at the 10 per cent level. There is also relatively clear evidence that large losses of more than a third occurred more frequently among recent retirement cohorts. Very large losses of more than half appear to have increased very slightly at most. The incidence of poverty entry follows no clear trend. The overall picture is thus of a more or less broadly shared increase in financial risks that has not (yet) resulted in more men (and their partners) entering poverty or experiencing very large income losses around retirement.

The American results also indicate a trend towards greater average losses, but here growing financial risks seem to have been less equally shared: there are marked increases in the incidence of both large and very large losses, and in the risk of having a low income below the poverty threshold after retirement. Period differences for these indicators also mostly attain statistical significance. These patterns fit well with the idea that the shift towards defined-contribution raised the heterogeneity of income trajectories and hurt disadvantaged groups with higher risks of falling below the poverty line. To explore this possibility more directly, I examine below trends for different educational groups.

Table 7.2 compares the experiences of involuntary retirees who retired after a job loss and/or decline in health and voluntary retirees whose exit was not accompanied by either of these events. As argued above, involuntary retirees might be more directly affected by retrenchment of early retirement policies in Germany. Results in Table 7.2 are consistent with this idea. German involuntary retirees faced substantially higher risks of falling into poverty in the final period (retirement years 1999 to 2007), when almost 1 out of 4 men entered poverty. For voluntary retirees, there is no evidence of such a trend. In the first two periods risks of poverty entry did not differ markedly by the extent of control over retirement. A clear difference only emerged in the final period, when involuntary retirees faced substantially higher risks. Importantly, this coincides with the retrenchment of early retirement options from the late-1990s onwards. Trends for the other income measures also look less favourable for German involuntary retirees than for voluntary retirees, even though the patterns are somewhat less conclusive than in the case of poverty entries.

Overall, income trajectories are less clearly structured along the voluntary-involuntary distinction in the United States. This lack of clarity may partly reflect smaller sample sizes, but it may also reflect the fact that retirement after adverse events tends to be more selective in the United States. Retirement after a job loss or health shock is involuntary in the sense that it presumably would have occurred later in the absence of the adverse ‘trigger’—but it is not inevitable. In the US, where the costs of early retirement remain higher than in Germany even after the recent reforms and where re-employment opportunities for older workers are probably better, those who retire after a late-career job loss or health shock are possibly a more selective group than in Germany. Further analysis of work and income trajectories around late-career job loss supports this interpretation (Heisig 2015, chap. 8).

Table 7.3 differentiates men with respect to their level of education. Because of the limited number of observations, I now only differentiate between an early and a late period. In both countries, there is a steep educational gradient in the risk of entering income poverty upon retirement. In Germany, in both periods about 1 in 3 men with low education fell

Table 7.2 Income change around retirement by country, year of and control over retirement

Voluntary retirement Germany

Involuntary retirement

Period 1 1985-1992

Period 2 1993-1998

Period 3 1999-2007

Period 1 1985-1992

Period 2 1993-1998

Period 3 1999-2007

Avg. change in pre-government

-69.1 [2.3]

-70.4 [2.4]

-73.2 [1.6]

-60.1 <2-3> [5.3]

-73.3 <1> [2.4]

-71.5 <1> [2.2]

income (in % of pre-retirement value)

Avg. change in post-government

-12.3 ® [2.2]

-14.3 [2.1]

-18.2 ™ [1.5]

-7.3 <23> [4.3]

-16.9 32 [2.3]

-22.7 a 2) [1.9]

income (in % of pre-retirement value)

% with post-government loss > 1/3

23.4 и [2.7]

28.5 [2.9]

29.8 ™ [2.4]

22.4 <3> [4.2]

29.6 [3.3]

33.9 <1> [3.4]

% with post-government loss > 1/2

9.2 [2.0]

8.7 [1.8]

11.3 [1.6]

10.3 [2.7]

13.3 [2.7]

16.5 [2.8]

% entering income poverty

15.2 [2.5]

10.3 [2.1]

12.1 [1.8]

14.8 <3> [3.4]

13.9 <3> [2.6]

23.7 22 [3.1]

Post-retirement obs. (person years)

449/430

468/456

748/733

194/178

322/293

386/365

all/not poor before retirement

Post-retirement obs. (persons)

244/233

246/239

476/466

103/95

170/155

226/214

all/ not poor before retirement

United States

Period 1

Period 2

Period 3

Period 1

Period 2

Period 3

1981-1988

1989-1994

1995-2002

1981-1988

1989-1994

1995-2002

Avg. change in pre-government

-51.6 [2.7]

-49.8 [3.4]

-48.7 [4.0]

-47.1 [4.7]

-55.8 [4.0]

-50.4 [5.7]

income (in % of pre-retirement value)

Avg. change in post-government

-12.5 <2-3> [2.3]

-20.3 m [2.9]

-20.4 <1> [3.5]

-13.1 [3.9]

-19.1 [4.2]

-14.2 [5.2]

income (in % of pre-retirement value)

% with post-government loss > 1/3

21.7 <23> [3.2]

38.5 <1> [4.2]

42.4 <1> [4.3]

30.0 [4.4]

34.6 [5.6]

37.6 [6.2]

% with post-government loss > 1/2

6.1 <23> [1.5]

15.0 (13) [2.9]

24.1 <1-2> [3.6]

10.1 [3.0]

13.8 [3.8]

13.0 [4.3]

% entering income poverty

8.3 <3> [2.2]

9.9 <3> [2.6]

19.4 <'?« [4.0]

21.1 [4.5]

15.6 [4.4]

27.7 [6.4]

(continued)

Table 7.2 (continued)

Voluntary retirement Germany

Involuntary retirement

Period 1 1985-1992

Period 2 1993-1998

Period 3 1999-2007

Period 1 1985-1992

Period 2 1993-1998

Period 3 1999-2007

Post-retirement obs. (person years) all/not poor before retirement

307/279

233/217

235/216

212/170

129/104

108/89

Post-retirement obs. (persons) all/ not poor before retirement

167/152

125/116

144/132

113/91

71/56

68/56

Person-level cluster-robust standard errors in brackets. Retirement is treated as involuntary if it was preceded by job displacement or a decline in health; see Section 'Data & Methods' for further details. Years refer to the year of retirement. Superscripted numbers in parentheses indicate that the within-country period difference for a given income measure is significantly different from zero at the 10% level (two-tailed tests). For example, the superscript (1, 2) indicates that the Period 3 estimate is significantly different at the 10% level from the Period 1 and Period 2 estimates, avg. = average, obs. = observations Sources: SOEP, PSID, CNEF, author's calculations below the poverty threshold compared to approximately 1 out of 8 men in the intermediate educational group. If averaged over the two periods, educational differences are of similar magnitude in the United States.

Do the results in Table 7.3 confirm that inequalities by level of education have risen in the United States? Due to small sample sizes, results must be interpreted with some caution, but overall it seems that income trajectories became less favourable for all educational groups (^-values in the right-most column illustrate to what extent period differences are statistically meaningful). That said, there is some evidence that, at least in absolute terms, inequalities by level of education grew between the 1980s and the 2000s. The incidence of very large losses and of poverty entries rose most strongly among the least educated men in the United States. Taking the estimates for the second period at face value, every fourth low-educated man saw his disposable income drop by more than 50 per cent and almost every second fell below the poverty line in the first post-retirement years. However, the uncertainty of these estimates is considerable.

Table 7.4 shows whether the shift to defined-contribution pensions has been accompanied by growing inequalities in access to complementary pension income. For each educational group (and period), it reports the proportion of men living in households with no or low private nonlabour income (PNLI)[1] after retirement, with ‘low’ meaning that PNLI accounts for less than 10 per cent of pre-tax post-transfer household income. There is clear evidence for growing educational inequalities in access to private income sources. From the first period to the second, the share of men with no/low PNLI increased from 34 to 51 per cent among the low-educated and from 12 to 22 per cent among the intermediate group while remaining practically unchanged at 7-8 per cent for those with high levels of education. In all educational groups, men with no/ low PNLI tend to experience larger declines in disposable income, and some evidence suggests that for the low educated, this link has become stronger over time. These results underline the (growing) importance of

Low education

Intermediate education

High education

Germany

Period 1 1985-1996

Period 2 1997-2007

p-value

Period 1 1985-1996

Period 2 1997-2007

p-value

Period 1 1985-1996

Period 2 1997-2007

p-value

Avg. retirement age

58.7 [0.4]

60.1 [0.5]

0.03

60.4 [0.2]

60.8 [0.2]

0.12

61.6 [0.4]

62.5 [0.4]

0.09

Avg. change in pregovernment income (in % of pre-retirement value)

-69.5 [3.4]

-75.0 [3.0]

0.22

-68.2 [1.9]

-72.0 [1.4]

0.11

-72.6 [4.1]

-73.8 [2.7]

0.81

Avg. change in postgovernment income (in % of pre-retirement value)

-20.2 [2.6]

-23.6 [3.3]

0.41

-13.1 [1.7]

-20.0 [1.2]

0.00

-8.5 [3.3]

-14.1 [2.7]

0.19

% with post-government loss > 1/3

36.8 [4.5]

41.0 [5.3]

0.55

24.7 [2.2]

30.1 [2.1]

0.08

17.4 [3.7]

25.9 [4.1]

0.12

% with post-government loss > 1/2

14.3 [2.9]

19.1 [4.5]

0.38

10.1 [1.6]

12.1 [1.5]

0.35

4.8 [1.9]

10.0 [2.4]

0.09

% entering income poverty

29.2 [4.7]

33.4 [5.3]

0.56

11.9 [1.7]

13.0 [1.5]

0.64

5.2 [2.2]

7.8 [3.0]

0.48

Post-retirement obs. (person years) all/not poor before retirement

263/246

179/171

731/715

931/915

153/152

259/256

Post-retirement obs.

(persons) all/not poor before retirement_

153/136

111/103

394/378

555/539

80/79

170/167

United States

Period 1 1981-1992

Period 2 1993-2002

p-value

Period 1 1981-1992

Period 2 1993-2002

p-value

Period 1 1981-1992

Period 2 1993-2002

p-value

Avg. retirement age

62.9 [0.5]

64.3 [0.8]

0.10

62.6 [0.4]

62.6 [0.6]

0.95

64.1 [0.5]

63.5 [0.7]

0.47

Avg. change in pregovernment income (in % of pre-retirement value)

-58.3 [3.0]

-63.6 [5.8]

0.41

-53.3 [2.4]

-48.3 [3.7]

0.25

-40.2 [4.4]

-37.9 [5.9]

0.76

Avg. change in postgovernment income (in % of pre-retirement value)

-17.6 [2.8]

-25.9 [5.5]

0.18

-20.3 [2.1]

-17.6 [3.3]

0.49

-11.7 [3.7]

-17.3 [5.1]

0.38

% with post-government loss > 1/3

31.1 [3.9]

39.7 [7.4]

0.30

33.6 [3.6]

39.7 [4.5]

0.29

23.8 [4.3]

42.9 [6.0]

0.01

% with post-government loss > 1/2

11.0 [2.6]

25.6 [6.3]

0.03

11.6 [2.2]

19.2 [3.5]

0.07

9.4 [2.9]

17.3 [4.5]

0.14

% entering income poverty

30.4 [4.2]

46.1 [8.8]

0.11

7.2 [2.0]

17.1 [3.8]

0.02

1.1 [0.8]

7.4 [3.3]

0.07

Post-retirement obs. (person years) all/not poor before retirement

312/269

86/66

291/283

214/203

137/135

119/119

Post-retirement obs. (persons) all/not poor before retirement

188/145

58/38

162/154

133/122

75/73

71/71

Person-level cluster-robust standard errors in brackets. Education is defined as follows: Low: below upper secondary; intermediate: Germany: upper secondary (general or vocational), US: upper secondary (high school) or some college (13 or 14 years of education); High: tertiary degree. Period definitions differ slightly across countries, p-val. = two-tailed, p-value for rejecting the null hypothesis that average/proportion does not differ across the two periods, avg. = average, obs. = observations Sources'. SOEP, PSID, CNEF, author's calculations

Table 7.4 American men: income change around retirement by year of retirement, level of education and level of private non-labour income (PNLI)

Low education

Intermediate education

High education

Period 1 1981-1992

Period 2 1993-2002

p-value

Period 1 1981-1992

Period 2 1993-2002

!p-value

Period 1 1981-1992

Period 2: 1993-202

p-value

% of retirees with no/ low private non-labour income (PNLI)

34

51

0.02

12

22

0.03

8

7

0.85

Avg. % change in post-gov. income

Substantial PNLI

-14

-18

0.61

-20

-13

0.15

-12

-16

0.53

No/low PNLI % entering poverty

-18

-31

0.14

-17

-22

0.65

(-3)

(-37)

0.22

Substantial PNLI

18

23

0.58

4

10

0.09

1

7

0.04

No/low PNLI

60

(77)

0.23

31

51

0.19

(2)

(15)

0.42

% with post-gov. loss > 50%

Substantial PNLI

8

15

0.23

11

17

0.12

8

16

0.12

No/low PNLI

15

28

0.18

17

23

0.59

(20)

(28)

0.72

Post-retirement obs. (person years) all/not poor before retirement

350/265

100/64

297/279

220/199

138/131

119/119

Post-retirement obs. (persons) all/not poor before retirement

188/142

58/37

162/152

134/120

75/70

71/71

Substantial private non-labour income is defined as PNLI accounting for at least 10% of household pre-tax post-transfer income. Estimates in parentheses are based on fewer than 30 observations. Education is defined as follows: Low: below upper secondary; intermediate: Germany: upper secondary (general or vocational), US: upper secondary (high school) or some college (13 or 14 years of education); High: tertiary degree, p-val. = two-tailed p-value for rejecting the null hypothesis that average/proportion does not differ across the two periods, avg. = average, obs. = observations Sources'. SOEP, PSID, CNEF, own calculations complementary pension income for economic well-being in retirement in the US and confirm that inequalities in access to this crucial resource have risen over recent decades.

  • [1] The main constituents of PNLI are occupational/private pension and asset/capital income. Resultsare similar when only occupational/private pension income is used instead of all private non-labourincome.
 
Source
< Prev   CONTENTS   Source   Next >