A discussion of “Working after Retiring from Career Jobs” Authors: Robert L Clark, Robert G. Hammond, Melinda Sandler Morrill, and Aditi Pathak Discussant: Joanna N. Lahey Texas A&M and NBER The Questions • What are retirement strategies and patterns for workers in government jobs with DB pensions and retiree health insurance preretirement? • What are bridge to retirement paths after leaving these jobs? Why important? • Government Policy – Benefit design – Return to work provisions – Predict supply of experienced workers • Public Policy more generally – Effects of benefit design on return to work – Demand for return to work after benefit payout begins – Why are people (not) returning to work? (=> change benefits) • Economic theory – Bridging to full retirement for people with DB pensions – Interactions of plan design with behavior What they did • Fielded two surveys with extensive questions on retirement plan expectations and choices, work expectations and choices, and retirement-related questions – Pre-retirement survey – Post-retirement survey (point-in-time) • Merged surveys with administrative data Findings: Summary statistics • Preliminary work– results today are summary statistics and correlations from the two surveys connected to basic demographics – Current employees plan to retire relatively young – Most plan on working (part-time) post-retirement – Most of those who do not plan on working expect to have enough money to not need to or expect to be unable to find work – The younger the age of expected retirement, the more likely expectations for post-retirement work Findings: Expected Retirement • Regression results conditioning on other variables (including tenure) – Men expect to retire later (controlling for profession) – K-12 Teachers/Admin plan to retire 1 year earlier – Police/firefighters plan to retire 3 years earlier – Professors plan to retire .7 years later Findings: Expected return to work • Regression results conditioning on other variables – Later retirement age => lower probability return to work – Male => higher probability return to work – Professional degree => higher pr(return to work) – Working spouse => higher pr(return to work) Findings: Survey of recent retirees • 69% of retirees not currently working (not at odds with expectations– different group, point in time vs. the entire future) • 14.4% working for government – 8.6% in same place – 5.8% in new place • 16.4% working in private sector • Of those not currently working, 61% unlikely to return to work Findings: Recent retirees regressions • Men: more likely to be working, less likely to be working in public sector, more likely to plan to return to work • Married less likely to work • < HS less likely to work, > MA less likely to work • High account balances less likely to work • TSERS more likely to work for public sector • People with caregiving responsibilities less likely to work Things to be careful of: Selection • Who takes the surveys compared to population (standard sampling bias)?: compare to administrative records • Who drops out of regressions because of missing data?: compare on non-missing data • Who is excluded because of a priori decisions about work? – Not a panel-- be careful when comparing one survey to another – Especially tricky because of age dimension More things to be careful about • Correlations: Conditional vs. unconditional – Ex. Men are more likely to work, controlling for police/firefighter occupation – Ex. People whose spouses have good health are less likely to work, controlling for caregiving • Small sample sizes for some populations – Ex. Men, professional degrees, # children? • Be clear on details of plan rules – Is it replacement for Social Security or in addition to? (in addition) – TSERS has waiting period, income limits for re-employed Extensions and new questions • Future work will show responses merged with more detailed administrative data – Basic demographics (gender, DOB) – Employment history – Salary information – Benefit account information • Additional survey questions not explored here – Choices made for plans, detailed $, detailed partner information, risk aversion etc. Extensions: Benefit design • TSERS has a lot of different payout options: How are these related to choices? – Joint and survivor benefits, different levels – Age 62 leveling prior to Soc Sec. – Can change mind if rehired and work for 3 additional years • TSERS has defined limits to public re-employment – 6 month wait before re-employment – Earnings limitations = max (50% of 12 mo salary, $31380) Extensions: Retirement Patterns • Follow the money – Where does retirement $ come from – Social Security claiming – Effect of homeownership – Financial fragility (2K question) – Risk aversion • Follow the work cycle – How long not working? – Paths, transitions, # jobs • Follow the spouse: Joint retirement Extensions: Retirement Patterns II • Is this a story about teachers substitute teaching? – 70-80% of participants are in TSERS • What specific jobs do re-employed do, in public and private sectors? – New careers or related jobs? • How do different professions behave differently? – Physical demands? – Ages of retirement? – Outside opportunities? Thank you!