1. 1.Read Cunha, Menichini, and Crockett (2015) “The retention effects of high years of service cliff-vesting pension plans,”
Economics Letters.
2. 2.Reproduce the 2 tables and the 3 survival curves presented in the paper using the Stata dataset “retirement_data.dta” with
R Studio. I should be able to run your script without errors by only changing the parent directory.
3. 3.Additional questions to answer:
a. Do you need to drop any observations? Why? Was there a different way to handle the missing data?
b. In a few sentences, discuss the issues surrounding causal inference in this paper. Do you believe whether the removal of the 20-year cliff vest
causes
higher attrition? What are some threats to this causal interpretation of the empirical results?
Exercise #1 Incentive effects of pension plans on retention 1. Read Cunha, Menichini, and Crockett (2015) “The retention effects of high years of service cliff-vesting pension plans,” Economics Letters. 2. Reproduce the 2 tables and the 3 survival curves presented in the paper using the Stata dataset “retirement_data.dta” with R Studio. I should be able to run your script without errors by only changing the parent directory. 3. Additional questions to answer: a. Do you need to drop any observations? Why? Was there a different way to handle the missing data? b. In a few sentences, discuss the issues surrounding causal inference in this paper. Do you believe whether the removal of the 20-year cliff vest causes higher attrition? What are some threats to this causal interpretation of the empirical results? 4. Submit (1) do-file; (2) Word or PDF file with 3 survival curve graphs copy/pasted from Stata. Include your answers to “Additional questions” in the Word file. Helpful hints: · The file “retirement_data.dta” is fully labeled with the following variables: · hiredate: This is the day/month/year in which the individual started service. · birthdate: Date of birth · sex: M==male, F==female · service: branch of service, Air Force, Army, or Navy · firstsepartiondate: Date of first separation from service (data on subsequent active duty and separations is not available) · superfund: The superannuation (retirement) fund the service member is subject to, either DFRDB or MSBS · officer: Officer==1, Enlisted==0 · If you are not familiar working with dates, reading the Stata manual https://www.stata.com/manuals/u25.pdf may be a good first step. Keep in mind that you can calculate the difference between 2 dates by subtracting them. The retention effects of high years of service cliff-vesting pension plans Economics Letters 126 (2015) 6–9 Contents lists available at ScienceDirect Economics Letters journal homepage: www.elsevier.com/locate/ecolet The retention effects of high years of service cliff-vesting pension plans✩ Jesse M. Cunha a,∗, Amilcar A. Menichini a, Adam Crockett b a Naval Postgraduate School, United States b Australian Defence Force, Australia h i g h l i g h t s • We study a major reform in the Australian military’s retirement plan. • A 20-year cliff-vesting plan is substituted by a new one-year vesting plan. • Most observed individuals choose the one-year vesting retirement plan. • Removing the 20-year cliff-vesting considerably increases attrition rates. a r t i c l e i n f o Article history: Received 26 September 2014 Received in revised form 6 November 2014 Accepted 7 November 2014 Available online 13 November 2014 JEL classification: J3 H5 Keywords: Retention Cliff vesting Retirement a b s t r a c t We study the retention effects of the Australian military’s decision to remove a 20-year cliff-vesting requirement from their retirement system in 1991.We follow to the present individualswho self-selected into and out of the 20-year cliff-vesting plan, as well as thosewhowere forced out of the plan. Eliminating the high years of service cliff-vesting provision leads to consistently higher attrition over time. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). 1. Introduction One of the main criticisms of traditional retirement systems is the delayed vesting of retirement benefits, whereby no benefits are received if the employee separates before the vesting date and full benefits are received for retirement after the vesting date (Lazear, 1990). Vesting periods of up to 20 years have been the traditional norm for many public service organizations across the world (Dis- ney and Johnson, 2001). Such ‘‘cliff-vesting’’ retirement systems are often regarded as unfair for employees who leave before the ✩ The opinions expressed are the author’s own and do not reflect the view of the US Department of Defense, or the Australian Defence Force or Australian Government. ∗ Corresponding author. E-mail addresses:
[email protected] (J.M. Cunha),
[email protected] (A.A. Menichini),
[email protected] (A. Crockett). http://dx.doi.org/10.1016/j.econlet.2014.11.005 0165-1765/Published by Elsevier B.V. This is an open access article under the CC BY-N vesting date, but are seen as an important retention tool by em- ployers (Warner, 2008). While earlier vesting periods are gaining popularity, many pub- lic service retirement plans still vest at high years-of-service (YOS). For example, the USmilitary –which employs over onemillion ser- vice members – still has a 20 YOS vesting period. In this paper, we provide what is to our knowledge the first empirical evidence on the retention effects of removing the cliff-vesting component of a public sector retirement system—useful information for organiza- tions interested in removing high YOS vesting requirements. We study a major retirement reform undertaken by the Aus- tralian Defence Force (ADF) in 1991 in which a 20 YOS cliff-vesting retirement scheme was replaced by one with a one-year vest. (The plans differed in other dimensions as well, as discussed in Sec- tion 2, but the removal of the 20 YOS vest was by far the most salient). A crucial consideration for any employer when changing a retirement plan is whether current employees will be ‘‘grandfa- thered’’ into the old retirement system. Out of a sense of fairness, C-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). http://dx.doi.org/10.1016/j.econlet.2014.11.005 http://www.elsevier.com/locate/ecolet http://www.elsevier.com/locate/ecolet http://crossmark.crossref.org/dialog/?doi=10.1016/j.econlet.2014.11.005&domain=pdf http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ mailto:
[email protected] mailto:
[email protected] mailto:
[email protected] http://dx.doi.org/10.1016/j.econlet.2014.11.005 http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ J.M. Cunha et al. / Economics Letters 126 (2015) 6–9 7 the ADF allowed members hired before October 1, 1991 to choose either the new or the old plan. All service members hired on Octo- ber 1, 1991 or after were automatically enrolled in the new plan. Comparing those who chose to stay on the old plan with those on the new plan, we show that facing a 20-year cliff-vest sig- nificantly reduced attrition before the vesting period. Comparing those who self-selected into the new plan with those who were enrolled in it automatically, we find similar attrition profiles which suggests that there was not much of a selection effect amongst the choice cohort in terms of retention probability, separate from the effect of the plan itself. This paper adds to the small empirical literature on retirement reform in the military (e.g., Simon et al., forthcoming; Cunha and Menichini, 2014), and our results are in line with the simulation exercises in Ausink and Wise (1996) and Asch et al. (2013). 2. The retirement programs We briefly describe the ADF retirement programs; for full details refer to Cole et al. (1990). DFRDB. The Australian Defence Force Retirement and Death Benefits (DFRDB), introduced in 1972, is a defined benefit pension plan with a 20 YOS cliff-vest. Members retiring with 20 or more YOS are entitled to a pension equal to a percentage of their final pay; the percentage depends on total YOS, ranging from 35 for 20 YOS to 76.5 for 40 YOS or more. Those who leave service with less than 20 YOS do not receive a pension, yet have a ‘‘soft landing’’ in the form of a lump-sum payment calculated as the sum of 5.5% of each year’s annual pay for the entirety of their tenure. As this lump sum amount does not include interest or adjustments for inflation, there is still a very large discontinuity in the present value of the retirement package at the 20 YOS mark. MSBS. In order to address perceived inequities under DFRDB, the ADF created the Military Superannuation and Benefits Scheme (MSBS) and automatically enrolled all members who entered ser- vice on or after October 1, 1991. The MSBS has both defined bene- fit and defined contribution components. Importantly, it does not have a cliff-vest; rather, all members receive retirement payments upon reaching retirement age (55 years for those born before July 1960 and increasing by one year until those born after June 1964). The defined benefit is a function of the member’s YOS and their final three-year-average salary, and is taken as a pension or a lump sum payout at retirement age. The defined contribution is a mandatory 5% of the member’s annual pay, invested in a menu of government-managed funds; it cannot be withdrawn before re- tirement age. Compared to DFRDB, MSBS is far more beneficial for those with less than 20 YOS and is slightly more beneficial for those who stay longer than 25 years. For those planning on retiring with between 20 and 25 YOS, uncertain future returns on government-managed funds would have made it difficult to determine which plan was more beneficial.1 The transition period. All members under DFRDB had to make the choice between DFRDB and MSBS by September 30, 1992. 3. Data We use individual-level administrative data from the ADF on all new enlisted service members and officers from September 1, 1990 until September 30, 1992. Thus, we observe one cohort who enlisted between September 1, 1990 and September 30, 1991 and 1 A financial comparison between DFRDB and MSBS for a representative enlisted member is available in the Online Appendix. Table 1 Descriptive statistics of the sample, by cohort. Choice cohort (FY1991) No choice cohort (FY1992) (1) (2) Chose MSBS 0.863 (0.005) Male 0.812 0.827 (0.006) (0.009) Age at enlistment 20.066 19.998 (0.049) (0.075) Officer 0.138 0.233 (0.005) (0.010) Army 0.520 0.272 (0.007) (0.011) Navy 0.223 0.436 (0.006) (0.012) Air Force 0.257 0.292 (0.006) (0.011) Observations 4586 1758 had a choice between DFRDB and MSBS – the FY1991 cohort – and another cohort who enlisted between October 1, 1991 and September 30, 1992, and were automatically enrolled in MSBS— the FY1992 cohort. Two issues with this sample are of note: first, data on the re- mainder of the force in FY1991 is not available due to limited elec- tronic record keeping at that time, which limits the scope of our analysis. Second, amongst the cohorts we do observe, retirement choice data are missing for those who left the military prior to July 1, 1994, again due to incomplete record keeping. This missing data could bias our estimates if there is differential attrition amongst those who chose different retirement schemes. While we do not have the data necessary to assess the extent of any potential bias (e.g., detailed demographic characteristics of choosers), it is rea- sonable to believe that the retirement scheme is a small compo- nent of an individual’s choice to continue employment within the first three years of their career. Table 1 contains summary statistics, by cohort, of the 6344 individuals in our sample. Amongst the choice cohort