TRENDS IN PENSION CASH- OUT AT JOB SEPARATION AND THE EFFECTS ON - - PowerPoint PPT Presentation

TRENDS IN PENSION CASH- OUT AT JOB SEPARATION AND THE EFFECTS ON LONG- TERM OUTCOMES

SIEPR October 7, 2016

Philip Armour, Michael Hurd and Susann Rohwedder

Why study pension cash-out?

 US provides large tax incentives to encourage

private-pension savings

 Tax penalties on early withdrawals  Large Fraction of Pension Plans Have Cash-Out

Options  $$$ at risk to be spent before retirement

Effectiveness of policies promoting retirement-income security depends on

• importance of pension cash-outs and
• their long-term consequences

2

This paper

Using long panel from the Health and Retirement Study

 Trends in pension cash-outs among older workers  Cohort differences  Cumulative cash-out behavior over time  Individual and household-level assessments  Investigate precipitating events leading to cash-outs  Relate cash-out choices to subsequent retirement

security outcomes

3

Time Period: 1992 - 2012

4

Some important trends in

 Labor force participation  Macro economy

Increases in Labor Force Participation among older males, not among younger

5

Labor force participation, men

10 20 30 40 50 60 70 80 90 100 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 age 40-44 45-49 50-54 55-59 60-64 65-69

Females: increases throughout, except most recent younger cohorts

6

10 20 30 40 50 60 70 80 90 100 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 40-44 45-49 50-54 55-59 60-64 65-69

Labor force participation, women

Trends in the Macro Economy

7

 Rise and fall in house prices  Stock market drops  Unemployment: recession of 1991 and 2008

Macro trends and Great Recession

8

HRS Data: Cohort Comparisons in Pension Cash-out

9

3 groups: Follow for 4 waves = 8 years Year age 51-56 Age in 2008 1992 65-70 N=5,570 1998 59-64 N=3,308 2004 55-60 N=3,379 2010 (baseline comparisons) N=4,748

Worker’s Choices Upon Job Separation

10

Source: Hurd and Panis 2006

Shift from DB to DC plans

11

0% 10% 20% 30% 40% 50% DC DB None 1992 1998 2006 2010 Baseline comparisons: pension coverage and plan type, conditional on working

Majority of DC plans allow LSD,

• nly about half of DB plans

12

0% 20% 40% 60% 80% 100% DC DB 1992 1998 2004 2010

Baseline comparisons: pension plan allows LSD, conditional on work & pension on job

Number of Job Separations over 8 years among those with pension

13

Any DC Any DB Cohort % N with DC % N with DB 1992 44.8% 637 67.9% 956 1998 57.4% 614 62.5% 642 2004 70.1% 591 46.6% 400 Number and % of separations with pensions by plan type

Note: Rows sum to more than 100% because some workers have both types of plans

Pension disposition at job separation of DC PLANS with LSD (over 8 years)

14

Cohort

1992 1998 2004 Cashed Out 12.5% 12.0% 18.0% Rolled over into IRA 8.1% 11.2% 20.6% Annuitized 0.0% 0.0% 0.0% Left with Employer 29.3% 28.8% 28.7% Transferred to New Employer 57.3% 52.6% 36.4% Lost 2.7% 1.9% 2.9% Other 2.2% 3.8% 3.2%

All statistics weighted.

Pension disposition at job separation of DB PLANS with LSD (over 8 years)

15

1992 1998 2004 Cashed Out 17.1% 15.3% 20.0% Rolled over into IRA 15.5% 14.6% 27.9% Annuitized 0.0% 0.0% 0.0% Expecting Benefits 24.2% 22.0% 26.5% Drawing Benefits 57.7% 57.4% 29.8% Lost 0.7% 0.6% 3.9% Other 2.3% 5.6% 3.8%

Weighted.

What are People Doing with DB Cash-

• uts?

16

DB Cash-Out Uses by Cohort, Weighted % 1992 1998 2004 Spent 24.8% 21.9% 31.0% Saved 55.9% 55.7% 29.2% Debt 19.3% 18.6% 38.3% Durables 0.0% 3.7% 1.5%

Cohort comparison: reason for job separation among DC cash-outs

17

0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 1992 1998 2004

Cohort comparison: reason for job separation among DB cash-outs

18

0% 10% 20% 30% 40% 50% 60% 1992 1998 2004

Shocks around time of job separation: Some are associated with higher cash-out propensity

19

Shock at (or around time of) job separation Fraction cashed out

Lost Health Insurance 36.2 Got Divorced 20.7 Became Widowed 19.5 Became Work-Limited 22.1 Health Worsens 19.4 Became Poor Health 26.5 Fell Behind on Mortgage 54.6 Any Mortgage Issues 47.2 Among those separating with a pension, fraction who cashed out: Overall: 18.6%

Predictors of Pension Cash-out: DB and DC plans with lump-sum option

 Probit estimation  26 right-hand variables, including shocks  Summary of noteworthy effects  Increasing in… (and significant)

 Age  Unemployment rate  Race (African-American)

Predictors of Pension Cash-out:

Probit, DC and DB plans with lump-sum option (cont.)

 Decreasing in… (and significant)

 Wealth  Education  Planning horizon  Has health insurance  DC plan value

 Higher in later cohorts, but not significant

Long-run outcomes

 Focus on 1992 cohort, follow for 20 years  Compare cash-out with non-cash-out population

Vital status (about 20% died by 2012) Health Income Wealth

22

Long-run outcomes

23

 Distinguish 3 groups in 1992 cohort:

 Never separated from a job  Separated from a job, but never cashed-out  Separated from a job and cashed-out at least once

Note: About 20% of 1992 cohort had died by 2012 Among those still alive, 58% retired by 2012 Any differences in 2012 outcomes across these 3 groups?

Cash-out sample has worse baseline statistics and worse 2012 outcomes

24

Log HH Wealth HH Income Health 1992 Measures for those Retired by 2012 (1=poor) Never separated from Job to Non-Ret State

11.68 70311 3.82

Separated from a Job, Never Cashed-Out

11.58 69512 3.86

Separated from a Job and Cashed-Out

11.23 53779 3.49

2012 Measures for those Retired by 2012 Never separated from Job to Non-Ret State

11.75 31643 3.13

Separated from a Job, Never Cashed-Out

11.69 32110 3.07

Separated from a Job and Cashed-Out

10.48 22889 2.80

Among those retired by 2012

What is leading to worse outcomes?

25

 Shocks (economic, health etc.)  Initial conditions

• r

 Cash-outs

 compare baseline and outcomes 20 years later for those with cash-out option and those without

Option of LSD does not lead to worse

• utcomes at population level

26

LSD Option on DB Plan in ‘92 No Yes

Percent alive in 2012

80.2% 82.5%

Conditional on survival to 2012 N 1,548 893 cash-out of plan 10.5% 17.0% Means Wealth in 1992 193,614 237,083 Wealth in 2012 401,593 433,995 Household Income in 1992 60,239 65,445 Household Income in 2012 35,922 38,302

Option of LSD does not lead to worse

• utcomes at population level

27

LSD Option on DB Plan in ‘92 No Yes

N 1,548 893 cash-out of respective plan 10.5% 17.0% Medians Wealth in 1992 126,000 132,600 Wealth in 2012 220,200 249,750 Household Income in 1992 51,000 53,100 Household Income in 2012 24,461 25,050

Option of LSD does not lead to worse

• utcomes at population level

28

Ratios of 2012 values to 1992 values DB LSD option No Yes mean wealth 2.07 1.83 mean income 0.60 0.59 median wealth 1.75 1.88 median income 0.48 0.47

Cash-outs & Econ Preparation for Retirement:

29

Our Measure

Takes into account

 Differential mortality by marital status, sex and education  All resources observed shortly after retirement (pension &

Social Security income, bequeathable wealth)

 Uncertainty due to mortality and out-of-pocket spending

• n health care

 Taxes  Assessment of adequacy does not require full insurance

30

 Spending level age 66-69  Observed trajectories of spending derived from panel data  Initial wealth  Observed pension and Social Security income  Survival curves by sex, marital status and education  Stochastic out-of-pocket spending on health care  Taxes  Stochastic simulations

 Couples until widowing; then single  Singles

 Well-prepared if 95% or more of simulations end with

positive wealth at death of individual

Cash-outs & Econ Preparation for Retirement:

Percent adequately prepared in population

31

Singles: 52% Married persons: 78%

N Percent prepared Never Separated by age 67 138 75.7% Separated before age 67, but didn't cash-out 408 81.2% Separated before age 67 and cashed-out 77 72.4%

Cash-out option not associated with lower rates of econ prep for retirement

32

Percent prepared: whether pension has LSD option pension had LSD option DB pension had LSD option No

78.5% 79.6%

Yes

79.3% 81.1%

Separate effects of shocks and initial conditions from effects of LSD

33

Use availability of LSD in DB plans as instrumental variable.

Won’t use DC plans Almost all allow cash-out

Separate effects of shocks and initial conditions from effects of LSD

34

Example for wealth.

u, v correlated (say latent health)

w = wealth in 2012, C = cash-out, S = a negative shock. A is the availability of a cash-out.

First stage

35

 Effect of availability on frequency of cash-out:

0.0898 (0.0146)

Effect of cash-out on 2012 outcomes

36

Right-hand variables

 Race  Sex  Education  Wealth quartile in 1992  Earnings quartile in 1992  Marital status in 1992  Marital status in 2012

Effect of cash-out on 2012 outcomes:

No significant effects, but large standard errors

37

Dead Health (higher=better) HH Income Log HH Wealth Coeff

• 0.156

1.597 4,727 2.035 S.E. (0.233) (0.948) (37,352) (1.968)

Household analysis

38

 1992 economic measures of households in which

person lived

 2012 economic measures of households in which

person lived

 Did person or spouse of person have a cash-out

• ver 20 years?

Similar to person-level analysis

Summary and conclusions

39

For more recent cohorts

fewer job separations that were followed by

retirement

cash-out more prevalent Unemployment important for 2004 cohort

Shocks that trigger cash-out:

Lost health insurance Became poor health Fell behind on mortgage Any mortgage issues

Summary and conclusions (cont.)

40

Among those with cash-out: worse economic and

health at baseline AND even worse 20 years later

Preliminary evidence:

Worse outcomes not due to cash-out

shocks and initial conditions Cash-out option gave access to financial buffer But low power