P1: KEE103-03 MIT-EFP.cls May 4, 2006 20:6THE EFFECTS OF DEFINEDBENEFIT PENSION INCENTIVESAND WORKING CONDITIONS ONTEACHER RETIREMENTDECISIONSAbstractThe retirement behavior of Pennsylvania public schoolteachers in 1997–98 and 1998–99, a period when stateearly retirement incentives were temporarily increased,is modeled using a choice framework that emphasizesboth pecuniary and nonpecuniary factors of the retire-ment decision under a defined benefit retirement plan.We find each to have large and statistically significanteffects on the decision to retire.The present value of inflation-adjusted pension ben-efits of a public defined benefit plan is found to be an im-portant and sizable determinant of retirement. A $1,000(or .4 percent) increase in the real present value of pen-sion benefits is estimated to increase the probability ofretirement for female teachers by .02 to .08 percent-age points; this implies an elasticity of retirement forfemale teachers with respect to the present value of realpensions of between 2.0 to 3.5. These estimated definedbenefit pension elasticities for female teachers are higherthan for male teachers, whose comparable retirementelasticity was 1.9 to 2.5.A $1,000 increase in current salary is found to re-duce the mean probability of retirement by .1 percentagepoints, implying an elasticity of −1.4. Thus, substantialsalary increases systematically reduce the probability ofolder teachers retiring.Joshua FurgesonProject CoordinatorArgosy Foundation555 E. Wells St., Suite 1650Milwaukee, WI [email protected] P. Strauss(corresponding author)H. John Heinz III School ofPublic Policy and ManagementHamburg HallCarnegie-Mellon UniversityPittsburgh, PA [email protected] B. VogtH. John Heinz III School ofPublic Policy and ManagementHamburg HallCarnegie-Mellon UniversityPittsburgh, PA [email protected] 2006 American Education Finance Association1P1: KEE103-03 MIT-EFP.cls May 4, 2006 20:6THE EFFECTS OF DEFINED BENEFIT PENSION INCENTIVES AND WORKING CONDITIONSStudent achievement, but not student poverty, is also significantly relatedto teacher retirement; a one-standard-deviation increase in achievement scoresreduces the mean probability of retirement by .38 to .64 percentage points,implying an elasticity of between −.24 and −.41. Measures of school crimewere positively associated with male, but not female, retirement and weremodest in size.The estimated logistic model of the retirement decision under a public,defined benefit plan makes more accurate in-sample predictions than a sim-ple model based on age-specific retirement rates; however, the logistic model,while relatively more accurate than other approaches, is less accurate in pre-dicting the effect of previous early retirement incentive plans.INTRODUCTIONIncreasingly, public interest in school reform is focusing on improving class-room teacher quality. Indeed, the recent federal legislation No Child LeftBehind obligates school districts throughout the United States to have in placea “highly qualified” teacher force as part of a federal effort to improve studentachievement that also includes a federal system of monitoring (federally ap-proved state-implemented standardized testing), incentives via increased flexi-bility in the expenditure of federal funds, and penalties through the threatenedwithdrawal of funds if various targets surrounding student achievement andteacher quality are not met.Efforts to improve quickly the quality of the teacher profession are am-bitious, since the K–12 teaching profession is one of the largest professionaloccupations in any developed country. In the United States, the U.S. Bureau ofLabor Statistics reports that America had 4.187 million primary, secondary, andspecial education teachers in 2002. There were more such teachers in 2002than professionals in accounting and auditing (1.06 million), computer andmathematical occupations of all kinds (3.018 million), architects and engineers(2.587 million), lawyers (1.168 million), registered nurses (2.284 million), andthe entirety of protective services, including fire, police, correctional officers,and private protection services and agencies (3.116 million).1As a consequence of increased public interest in primary and secondaryschool teachers, there has been increased research on many aspects of themarket for them; however, there is virtually no literature dealing with olderteachers’ decision whether or not to retire, and the effect that unforeseen1. See Bureau of Labor Statistics 2005.2 EDUCATION FINANCE AND POLICYP1: KEE103-03 MIT-EFP.cls May 4, 2006 20:6Joshua Furgeson, Rober t P. Strauss, and William B. Vogtchanges in financial incentives to retire may have on that decision.2Hiringpractices have been explored by Ballou (1996), Ballou and Podgursky (1997),and Strauss et al. (2000), among others, while teacher attrition, mobility, andretention have been explored by Dolton and van der Klaaw (1999), Grissmerand Kirby (1992, 1997), Hanushek, Kain, and Rivkin (2001), and Manski(1987). The decision to become a teacher and issues of early retention andturnover have been examined by Brewer (1996), Dolton (1990), Hanushekand Pace (1995), and Stinebrickner (1998), while questions of initial matchingand geography have been importantly explored by Boyd et al. (2002, 2003),Lankford, Loeb, and Wyckoff (2002), Murnane et al. (1991), and Strauss et al.(1998).It is generally understood that the probability of a teacher not voluntarilyremaining in the same district next year is U-shaped; attrition or quit ratesare quite high during the first several years of teaching, and then quite lowfor a long period of time until the possibility of retirement emerges underthe terms of state personnel retirement laws. National Education Association(NEA 2003) notes that 27 percent of teachers had been in their current posi-tion for more than twenty years, compared to 8 percent in 1976. The lack ofmobility for much of a teacher’s career reflects not only likely personal circum-stances that discourage moving while raising their own children, but also thegeneral practice among school districts when hiring to not recognize all yearsof experience vis-`a-vis the salary offer to experienced teachers in other districtsseeking to change employment.In the short run, improving teacher quality may entail professional devel-opment and training of the existing teacher force, as well as