Recruiting and Retaining High-quality State and Local Workers: Do Pensions Matter?

Transcription

1 Recruiting and Retaining High-quality State and Local Workers: Do Pensions Matter? Geoffrey Sanzenbacher Research Economist Center for Retirement Research at Boston College National Tax Association Annual Conference Santa Fe, NM November 15, 2014

2 1 Overview Will state and local pension benefit cuts, typically for new hires, make it difficult for governments to recruit and retain high-quality workers? This paper exploits variance across state and local pension plans in the generosity of benefits to answer this question. Findings indicate more generous pensions help attract and keep workers who command high private sector wages.

3 2 Why pensions might affect worker quality Backloaded DB pensions are associated with high retention (e.g., Gustman and Steinmeier (1993), Munnell et. al (2010)). But DC plans are not backloaded and still increase retention. One theory is workers who value deferred compensation are low discounters (e.g., Allen, Clark, and McDermed (1993)). o Ippolito (2002) finds evidence of this and adds that these low discounters seem to be higher quality workers.

4 3 Analyzing relationship between pensions and worker quality Estimating the relationship between pension generosity and worker quality requires two pieces of data: 1. a measure of worker quality; and 2. a measure of pension generosity.

5 4 Measuring quality: data To obtain a measure of worker quality, Current Population Survey Outgoing Rotation Group (CPS-ORG) data are used. o Data follow individuals over a two-year period. o Obtain information on sector of employment, weekly wages, city and state, and demographics.

6 5 Measuring quality: constructing sample Worker quality is measured by adopting Borjas (2002) and focusing on private sector wages of two groups: 1. workers leaving the state and local sector to the private sector; and 2. workers entering the state and local sector from the private sector. If leavers private sector wages are higher than enterers, then high-quality workers may be sorting out of the public sector.

7 6 The quality gap over time Private Sector Weekly Wages of State and Local Enterers and Leavers, $800 $600 $400 $200 Leaving state/local sector to the private sector Entering state/local sector from the private sector $0 Note: Includes individuals aged Wages are normalized to year 2000 dollars. Source: U.S. Census Bureau. Current Population Survey Outgoing Rotation Groups,

8 7 Measuring pension generosity: data Pension generosity is measured using the pension s normal cost obtained from the Public Plans Database (PPD). o The normal cost is the present value of benefits accrued by active members as a share of payroll. o Both employees and employers contribute. In this study, normal cost is adjusted for differences in assumed discount rates that do not reflect benefit generosity.

9 8 Assigning sample members pensions To perform the analysis, data from the CPS-ORG are merged with the PPD based on sector, state, city, and occupation. The final data contain the normal cost of the individual s pension in the year spent in the state or local sector. The final data also contain demographic information, weekly earnings, and some state-level information.

10 9 Two analyzes are performed 1. Plan-level: dependent variable is the private sector wage gap between enterers and leavers. 2. Individual-level: dependent variable is log of private sector wage of individual in year before or after public sector.

11 10 Plan-level analysis The plan-level estimates the quality gap for plan j in time t: q j,t = w j,t+1 leave enter w j,t 1 enter w j,t 1 As a function of: o The pension s normal cost (or amount paid by employer s vs. employees); o Demographic gaps between leavers and enterers (e.g., a higher share of leavers than enterers being male); and o State-level characteristics like unemployment rates or the wage differential between private and public sector.

12 Plan-level results Quadratic regression suggests higher normal cost pensions have lower quality gap The linear-regression results are consistent with expectations, but not significant Effect of Change in Normal Cost on the Quality Gap Variable Coefficient Std. error Coefficient Std. error Total normal cost Total normal cost * Number of plan years R-squared Notes: Demographic gaps included: Male, black, marriage, less HS, HS grad, college grad, age, age. 2 State-level variables included: State unemployment rate, public-private wage gap, public vs. private wage compression, and share of all state employment in state/local sector. * Indicates significance to 10-percent level. Demographic gaps are defined as the percent of leavers with the given characteristic minus the percent of enterers with that characteristic. For example, if 50 percent of leavers are college grads and 40 percent of enterers, the college gap is 10 percent. Sources: Public Plans Database Center for Retirement Research at Boston College and Center for State and Local Government Excellence; and U.S. Census Bureau. Current Population Survey Outgoing Rotation Groups,

13 12 Individual-level analysis The individual-level analysis runs the following regression: log w i,t = γ 0 L i,t + γ 1 NC i,t + γ 2 L i,t NC i,t + X i,t β + S i,t θ + L i,t S i,t θ + ε i,t o o o L i,t indicates a person leaving the state or local sector. NC i,t indicates the normal cost in the period the individual was in the state or local sector. X i,t and S i,t are demographic and state characteristics Some specifications replace NCi,t with employer and employee contributions.

14 Individual-level results The results are consistent with expectations and in this specification, significant: o specifications without demographics were similar, but results insignificant; and o results with quadratic also insignificant. Effect of 1-percentage Point Change in Normal Cost on Log Private Sector Wage Variable Coefficient Std. error Total normal cost *** Total Normal cost for leavers ** Number of observations 17,039 R-squared Notes: Demographic variables included: Male, black, married, less HS, GS grad, college grad, age. State variables included: State unemployment rate, public-private wage gap, public vs. private wage compression, and share of all state employment in state/local sector. ** Indicates significance to -5-percent level, *** to the 1-percent level. Sources: Public Plans Database Center for Retirement Research at Boston College and Center for State and Local Government Excellence; and U.S. Census Bureau. Current Population Survey Outgoing Rotation Groups,

15 Individual-level results (cont d) Interestingly, it doesn t matter who pays the normal cost, employers or employees. Effect of 1-percentage Point Change in Contribution to Normal Cost on Log Private Sector Wage Variable Coefficient Std. error Employer contribution *** Employer contribution for leavers ** Employee contribution Employee contribution for leavers * Number of observations 17,039 R-squared Notes: Demographic variables included: Male, black, married, less HS, GS grad, college grad, age. State variables included: State unemployment rate, public-private wage gap, public vs. private wage compression, and share of all state employment in state/local sector. *Indicates significance to the 10-percent level, ** Indicates significance to 5-percent level, *** to the 1-percent level. Sources: Public Plans Database Center for Retirement Research at Boston College and Center for State and Local Government Excellence; and U.S. Census Bureau. Current Population Survey Outgoing Rotation Groups,

16 15 Conclusion States with relatively generous benefits attract workers commanding higher private wages compared to those leaving. Both employer and employee contributions seem to have a similar effect on the quality gap. o Shifting pension costs to employees may have less harmful effects. Difficult to infer causation without detailed data on other benefits

17 Appendix 16

18 17 The normal cost over time Employer s and Employee s Contribution to Normal Cost as Share of Payroll, % Employee's contribution Employer's contribution 10% 4.8% 5.2% 5.1% 5.2% 5.4% 5.4% 5.5% 5.5% 5.6% 5.6% 5.9% 6.3% 5% 6.0% 6.3% 7.5% 7.5% 7.4% 7.4% 7.4% 7.3% 7.6% 7.6% 7.0% 6.3% 0% Note: Yearly numbers weighted by active members in plan. Calculation as shown uses the standardized normal cost. Source: Public Plans Database Center for Retirement Research at Boston College and Center for State and Local Government Excellence.