Funding Discipline for U.S. Public Pension Plans: An Empirical Analysis of Institutional Design

Please see PDF.

  1. [1]. Joshua Rauh, The Pension Bomb, Milken Inst. Rev., Jan. 2011, at 28, 32, available at See generally Alicia H. Munnell et al., Ctr. for Ret. Research at Bos. Coll., The Funding of State and Local Pensions: 2012–2016 (2013), available at

  2. [2]. In recent years, a number of municipalities, including Vallejo, California, in 2008, Central Falls, Rhode Island, in 2011, Stockton, California, and Mammoth Lakes, California, in 2012, and Detroit, Michigan, in 2013, have filed for bankruptcy in part due to the burden of pension obligations. See, e.g., Monica Davey & Mary Williams Walsh, Billions in Debt, Detroit Tumbles into Insolvency, N.Y. Times (July 18, 2013),; Elizabeth Trotta, Stockton Teeters on Bankruptcy Edge, but City’s Not Alone, The Exchange (June 27, 2012, 3:00 PM),

  3. [3]. In the same period, most states scaled back pensions for new employees, and a number of states reduced pension obligations to current employees and retirees. See U.S. Gov’t Accountability Office, State and Local Government Pension Plans: Economic Downturn Spurs Efforts to Address Costs and Sustainability 21 (2012), available at http://www.gao.
    gov/assets/590/589043.pdf; see also John W. Schoen, Pandemic of Pension Woes Is Plaguing the Nation, CNBC (Nov. 19, 2013, 6:00 AM),; Christine Williamson, Motor City as a Motivator: Detroit’s Woes Seen Likely to Push Others to Resolve Their Pension Underfunding, Pensions & Invs. (Aug. 5, 2013),
    PRINT/130809965/motor-city-as-a-motivator. In July of 2013, Senator Hatch proposed “a way for states and cities to exit the pension business.” Mary Williams Walsh, Pension Proposal Aims to Ease Burden on States and Cities, N.Y. Times (July 9, 2013, 12:01 AM),
    2013/07/09/pension-proposal-aims-to-ease-burden-on-states-and-cities. Under the proposal, local governments would hold annual auctions to purchase annuity contracts for their employees. In the course of the working years, an employee would collect a series of annuity contracts that, in the aggregate, would substitute for a pension. The public employer’s liability would consist only of the upfront payment to the insurer. Id.

  4. [4]. The GASB is a private, not-for-profit entity that sets standards for accounting and financial reporting by state and local governments. Facts About GASB, Gov’t Acct. Standards Board, http:// (last visited Oct. 29, 2014).

  5. [5]. State Budget Crisis Task Force, Full Report 45 (2012), available at http://www. Highlighting the challenges faced by public pension plans, the Task Force Chairmen Richard Ravitch and Paul Volcker emphasize that “[i]t is human nature to prefer the present to the future. Governments display that time preference by promising now and paying later: if they can, they will underestimate liabilities, underfund annual costs, and take on substantial investment risks to make it look like promises will be kept.” Id.

  6. [6]. Alicia H. Munnell, State and Local Pensions: What Now? 75 (2012).

  7. [7]. See, e.g., Selected Approved Changes to State Public Pensions to Restore or Preserve Plan Sustainability, Nat’l Ass’n St. Retirement Admins. (2014),

  8. [8]. For example, a worker might be entitled at retirement to 2% of final average salary for each year worked. An employee who has worked for 30 years for the particular employer would be entitled 60% of her final average salary; an employee whose final average salary was $100 per year would be entitled to $60 per year in pension benefits, not counting retiree healthcare benefits.

  9. [9]. According to the U.S. Census Bureau, there were 3418 U.S. public pension plans in 2010, of which 222 were administered at the state (rather than the municipal) level. Membership in these 222 state-administered pension plans comprised 89.6% of the total U.S. public pension plan membership. Of the 3196 locally administered plans, approximately 2200 covered local units in just three states: Pennsylvania (1425 plans), Illinois (457 plans), and Florida (303 plans). See Erica Becker-Medina, U.S. Census Bureau, Public-Employee Retirement Systems State- and Locally-Administered Pensions Summary Report: 2010, at 5 (2012), available at

  10. [10]. Jun Peng, State and Local Pension Fund Management 16–17 (2009).

  11. [11]. See, e.g., 40 Ill. Comp. Stat. 5/7-132 (2005 & Supp. 2014) (describing elective participation procedures for “any city, village, or incorporated town that attains a population over 5000 inhabitants” in the Illinois Municipal Retirement Fund); About IMRF, Ill. Municipal Retirement Fund, (noting that, with the exception of the City of Chicago and Cook County, the IMRF has provided retirement benefits to the employees of local governments and school districts in Illinois since 1941).

  12. [12]. Employee Retirement Income Security Act of 1974, Pub. L. No. 93-406, 88 Stat. 829. ERISA establishes mandatory pre-funding requirements for private employer plans and regulates the discounting of liabilities. If a plan is not fully funded, for example, the employer’s annual contribution must include the amount necessary to amortize over seven years the difference between its liabilities and its assets. Stricter rules apply to severely underfunded plans. 29 U.S.C. §§ 1001–1461 (2013).

  13. [13]. See generally Thomas J. Fitzpatrick IV & Amy B. Monahan, Who’s Afraid of Good Governance? State Fiscal Crises, Public Pension Underfunding, and the Resistance to Governance Reform, 66 Fla. L. Rev. 1317 (2014). For example, “[i]n 2006 . . . the median contribution rate was 8.5% of payroll for state and local government employers and 5% for employees [in] plans in which employees [were] also covered by Social Security;” for plans in which the employees were not covered by Social Security, the rates were 11.5% of payroll for employers and 8% for employees. Jonathan Barry Forman, Funding Public Pension Plans, 42 J. Marshall L. Rev. 837, 840 (2009) (citing Barbara D. Bovbjerg, U.S. Gov’t Accountability Office, State and Local Government Pension Plans: Current Structure and Funded Status (2008), available at http://www.

  14. [14]. See generally Olivia S. Mitchell, Public Pension Pressures in the United States, in When States Go Broke: The Origins, Context, and Solutions for the American State in Fiscal Crisis 67–71 (Peter Conti-Brown & David A. Skeel Jr., eds., 2012).

  15. [15]. See, e.g., Munnell, supra note 6, at 20 (noting that “[a] plan is fully funded when its assets equal its liabilities”).

  16. [16]. See, e.g., Jagadeesh Gokhale, Cato Inst., Funding Status, Asset Management, and a Look Ahead: State and Local Pension Plans (2012), available at
    sites/; Pew Ctr. on the States, The Widening Gap Update (2012), available at
    legacy/uploadedfiles/pcs_assets/2012/PewPensionsUpdatepdf.pdf; U.S. Gov’t Accountability Office, supra note 3; Robert Novy-Marx & Joshua D. Rauh, Policy Options for State Pension Systems and Their Impact on Plan Liabilities, 10 J. Pension Econ. & Fin. 173 (2011); Robert Novy-Marx & Joshua Rauh, Public Pension Promises: How Big Are They and What Are They Worth?, 66 J. Fin. 1211 (2011). See generally Munnell et al., supra note 1.

  17. [17]. Pew Ctr. on the States, supra note 16, at 1.

  18. [18]. Id. at 5.

  19. [19]. Rauh, supra note 1, at 32. Economists such as Brown, Wilcox, and Rauh have posited in a series of influential articles that pension plans are understating the amounts of unfunded liabilities by using a discount rate that reflects expected returns on assets instead of the riskiness of the liabilities. See Jeffrey R. Brown & David W. Wilcox, Discounting State and Local Pension Liabilities, 99 Am. Econ. Rev. 538, 538 (2009). The authors point out that linking the discount rate to the expected return allows pension plan sponsors, against all findings of conventional finance theory, to reduce their liabilities by investing in riskier securities because a higher discount rate will result in a smaller present value of liabilities. See id.; see also Munnell et al., supra note 1, at 4 (noting that using a discount rate of 5% instead of 8% increased the aggregate liability from $3.8 trillion to $5.5 trillion); Andrew G. Biggs, Understanding the True Cost of State and Local Pensions, 63 St. Tax Notes 565, 565–66 (2012). In 2012 the GASB adopted new standards that, for purposes of disclosure only, impose certain restrictions on the use of the expected rate of return as the discount rate. See infra Part VI.A.

  20. [20]. See, e.g., Josh Barro, Dodging the Pension Disaster, 7 Nat’l Aff. 3, 4–5 (2011) (finding that one “fundamental problem” with the pension plans offered by state and local governments all across America is that “in many cases, the benefits are excessively costly, insofar as they are more generous than is necessary to attract qualified talent to government work”); Biggs, supra note 19, at 565 (stating that funding shortfalls were caused by “market declines, funding holidays, and benefit increases passed during the prosperous days of the late 1990s”).

  21. [21]. See, e.g., State Budget Crisis Task Force, supra note 5, at 37.

  22. [22]. See Gov’t Accounting Standards Bd., Guide to Implementation of GASB Statements 25, 26, and 27 on Pension Reporting and Disclosure by State and Local Government Plans and Employers: Questions and Answers 13–15 (1997), available at

  23. [23]. See, e.g., Pew Ctr. on the States, supra note 16, at 6 (“Keeping up with the annual required contribution is perhaps the most effective way that states can responsibly manage their long-term liabilities for public sector retirement benefits.”). The Pew Center’s “research shows that states that consistently make their full payments have better-funded retirement systems and smaller gaps.” Id.; see also Fitch Ratings, Enhancing the Analysis of U.S. State and Local Government Pension Obligations 3 (2011), available at
    enhancing_the_analysis_of_state_local_government_pension_obligations.pdf (“The systems that pose the greatest risks are those with significant unfunded liabilities for which the government’s annual payments have been significantly less than an actuarially determined ARC over multiple years.”); Jun Peng & Ilana Boivie, Nat’l Inst. on Ret. Sec., Lessons from Well-Funded Public Pensions: An Analysis of Six Plans That Weathered the Financial Storm 6 (2011), available at (“The most fundamental principle in ensuring a plan achieves a 100% funding ratio is ensuring that the plan sponsors pay the entire amount of the annual required contribution (ARC) each year, because anything short of a full ARC payment will have a negative impact on the plan’s funding ratio in the long run.”).

  24. [24]. The variation can be seen across plans of varying benefits levels, such that the lack of funding discipline is not a problem only for relatively more generous plans. See generally Munnell, supra note 6.

  25. [25]. As used throughout this Article, “state-administered” plans include plans created by state statute that pool together, at least for administrative and investment purposes, employees of participating municipalities or other local entities, even if the pension system that administers such plans is legally separate and fiscally independent from the state.

  26. [26]. Calculated by author using data on 110 plans. For additional background on the structure of retirement plans and systems, and on the data collection techniques, see infra Part IV.

  27. [27]. Calculated by author using data from the Boston College Center for Retirement Research. For additional background on the data collection techniques, see infra Part IV.

  28. [28]. In 2012, GASB adopted a series of changes, including the elimination of the ARC disclosure requirement. For a discussion of such changes, see infra Part VI.

  29. [29]. In particular, any unfunded actuarial accrued liability must be funded over a period not to exceed 30 years (40 years prior to 2006). Additional parameters for the ARC calculation are set forth in GASB Statement No. 25. Gov’t Accounting Standards Bd., Financial Reporting for Defined Benefit Pension Plans and Note Disclosures for Defined Contribution Plans 17 (1994), available at

  30. [30]. The ARC measure is informative but imperfect. It is not completely standardized and may reflect the variation in certain actuarial assumptions and methodologies heretofore permitted by the GASB. For example, particularly rosy assumptions about expected rates of return, dim projections about salary growth, and a rolling 30-year amortization period are all means to decrease the amount of the “required contribution.” Certain tactics recently employed by a handful of states—such as permitting an employer to borrow from the plan to make the required contribution—also cloud the meaning of the ARC measure. See Danny Hakim, To Pay New York Pension Fund, Cities Borrow from It First, N.Y. TIMES (Feb. 27, 2012), http://www.

  31. [31]. For a discussion of the data collection techniques, see Part IV.

  32. [32]. See supra notes 16 & 19 (describing examples of recent scholarship).

  33. [33]. For analysis of the legal limitations to pension reform, see generally Daniel J. Kaspar, Defined Benefits, Undefined Costs: Moving Toward a More Transparent Accounting of State Public Employee Pension Plans, 3 Wm. & Mary Pol’y Rev. 129 (2011); Amy B. Monahan, Public Pension Plan Reform: The Legal Framework, 5 Educ. Fin. & Pol’y 617 (2010); Amy B. Monahan, Statutes as Contracts? The “California Rule” and Its Impact on Public Pension Reform, 97 Iowa L. Rev. 1029 (2012); Paul M. Secunda, Constitutional Contracts Clause Challenges in Public Pension Litigation, 28 Hofstra Lab. & Emp. L.J. 263 (2011); Debra Brubaker Burns, Note, Too Big to Fail and Too Big to Pay: States, Their Public-Pension Bills, and the Constitution, 39 Hastings Const. L.Q. 253 (2011); Whitney Cloud, Comment, State Pension Deficits, the Recession, and a Modern View of the Contracts Clause, 120 Yale L.J. 2199 (2011); Stuart Buck, Legal Obstacles to State Pension Reform (Ass’n for Educ. Fin. & Policy, Working Paper, 2011), available at

  34. [34]. See, e.g., Olivia S. Mitchell & Ping-Lung Hsin, Public Pension Governance and Performance, in The Economics of Pensions: Principles, Policies, and International Experience 92 (Salvador Valedés-Prieto ed., 1997); David Hess, Protecting and Politicizing Public Pension Fund Assets: Empirical Evidence on the Effects of Governance Structures and Practices, 39 U.C. Davis L. Rev. 187 (2005); John R. Nofsinger, Why Targeted Investing Does Not Make Sense!, 27 Fin. Mgmt. 87 (1998); Roberta Romano, Public Pension Fund Activism in Corporate Governance Reconsidered, 93 Colum. L. Rev. 795 (1993); Sylvester J. Schieber, Political Economy of Public Sector Retirement Plans, 10 J. Pension Econ. & Fin. 269 (2011); Michael Useem & Olivia S. Mitchell, Holders of the Purse Strings: Governance and Performance of Public Retirement Systems, 81 Soc. Sci. Q. 489 (2000).

  35. [35]. Munnell, supra note 6, at 75.

  36. [36]. Fitzpatrick & Monahan, supra note 13, at 4.

  37. [37]. See Olivia S. Mitchell & Robert S. Smith, Pension Funding in the Public Sector, 76 Rev. Econ. & Stat. 278, 288 (1994) (finding “wide variations in funding practices” and noting that “pension funding ‘habits’ do seem to persist in the public sector”); see also Tim V. Eaton & John R. Nofsinger, The Eect of Financial Constraints and Political Pressure on the Management of Public Pension Plans, 23 J. Acct. & Pub. Pol’y 161, 186–87 (2004) (finding evidence of persistent funding behaviors); Tongxuan (Stella) Yang & Olivia S. Mitchell, Public Pension Governance, Funding, and Performance: A Longitudinal Appraisal 16 (Pension Research Council, Working Paper No. 2005-2, 2005), available at (finding that “funding patterns are positively correlated over time, confirming the behavioral persistence hypothesis”).

  38. [38]. See Mitchell & Smith, supra note 37, at 282–84.

  39. [39]. Id. at 288 (finding that fiscal pressure, as measured by unemployment in a given state, “appears to cause some public employers to reduce their annual contributions below required levels” (emphasis added)). Stoycheva finds that in some periods between 2000 and 2009, “fiscal stress influenced pension funding through direct reductions in the contributions to the pension plans versus indirect reductions through manipulation of the discount rate.” Rayna. L Stoycheva, Sustainable Governance and Management of Defined Benefit Plans in the Public Sector: Lessons from the Turbulent Decade of 2000–2009, at 89 (Aug. 11, 2011) (Ph.D. dissertation, Georgia State University), available at
    context=pmap_diss; see also Barbara A. Chaney et al., The Eect of Fiscal Stress and Balanced Budget Requirements on the Funding and Measurement of State Pension Obligations, 21 J. Acct. & Pub. Pol’y 287, 296 (2002). But see Yang & Mitchell, supra note 37, at 17 (finding “no evidence...that fiscal distress due to unusually high unemployment rates prompts public employers to underfund their pension promises”).

  40. [40]. See Chaney et al., supra note 39, at 289–306 (finding that fiscally stressed states select discount rate assumptions that reduce funding requirements); see also Eaton & Nofsinger, supra note 37, at 163 (showing that “governments” facing tight financial constraints are more likely to reduce pension contributions through the manipulation of actuarial assumptions in a way that reduces required contributions); J. Fred Giertz & Leslie E. Papke, Public Pension Plans: Myths and Realities for State Budgets, 60 Nat’l Tax J. 2, 318–21 (2007) (discussing actuarial assumptions in public plans).

  41. [41]. State Budget Crisis Task Force, supra note 5, at 15–16.

  42. [42]. Alicia H. Munnell et al., Ctr. For Ret. Research at Bos. Coll., Why Don’t Some States and Localities Pay Their Required Pension Contributions? 2 (2008), available at

  43. [43]. See Alicia H. Munnell et al., Ctr. for Ret. Research at Bos. Coll., Locally-Administered Pension Plans: 2007–2011, at 1, 4 (2013), available at

  44. [44]. Peng & Boivie, supra note 23, at 7. Peng and Boivie highlight the success of the Illinois Municipal Retirement Fund (“Illinois MRF”), which, unlike the chronically underfunded Illinois pension plans for state employees and teachers, is well funded as a result of consistent payments of the full ARC amount. According to Peng and Boivie, an “important factor that contributes to Illinois MRF’s strong contribution management is that the state statute governing employer contributions gives the board of trustees broad sue in civil courts to collect delinquent payments from local employers” and also allows the comptroller to “deduct the [delinquent] amounts...from any grants of state funds to the municipality.” Id. at 21.

  45. [45]. A recent working paper by political scientists highlights the bipartisan nature of pension politics in “normal” times and shows that from 1999 to 2008, Democrats and Republicans voted together on pension-related bills. Sarah F. Anzia & Terry M. Moe, The Politics of Pensions 11–12 (Inst. for Research on Labor & Emp’t, Univ. of Cal. at Berkley, Working Paper No. 108-14, 2013), available at

  46. [46]. The former group lacks the incentive to monitor because its benefits are fixed; that is, participants do not share directly in the upside of high investment returns or consistent funding, and the most empowered participants are also unlikely to feel the downside of poor funding. Collective bargaining rights have thus not been found to be a significant predictor of funding discipline. See Munnell, supra note 6, at 8. Taxpayers, on the other hand, are ultimately affected by plan funding and performance, but face information and collective action challenges that prevent effective monitoring. The impact of “funding and investment decisions may not be felt until some point far in the future,” monitoring the board may be costly, and the benefits to any individual taxpayer from this effort are marginal. Fitzpatrick & Monahan, supra note 13, at 1330. A third group of constituents—the buyers of municipal debt—has begun, in the wake of municipal bankruptcies, to monitor pension underfunding, but the full scope and impact of its involvement remains to be seen. There is some indication, however, that state credit ratings have motivated pension funding. See, e.g., Michael B. Marois, Alaska Taps Rainy-Day Cash for Pension Gap: Muni Credit, Bloomberg (July 8, 2014, 7:00 PM), (noting that rating agencies identified Alaska’s pension liabilities as the biggest risk to the state’s credit rating and detailing the state’s subsequent plan to tap its budget-reserve account to pay unfunded pension liabilities).

  47. [47]. See, e.g., Conn. Mun. Emps. Ret. Sys., Summary Plan Description 3 (2007), available at (explaining that municipalities make contributions at rates set by the State Retirement Commission and also contribute toward the administrative costs of the plan); State of Conn., Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2010, at 73–74 (2011), available at
    2010cafr/CAFR10.pdf (noting that the state makes the contributions for the state employees plan, as well as the teachers plan, while it does not make any contributions to the municipal employees plan).

  48. [48]. Chaney et al., supra note 39, at 290 (quoting Richard M. Ennis, Is a Statewide Pension Fund a Person or a Cookie Jar? The Answer Has Implications for Investment Policy, 44 Fin. Analysts J. 21, 23 (1988)).

  49. [49]. James Clemente, N.J. Dep’t of the Treasury, Div. of Pensions & Benefits, Teachers’ Pension and Annual Report: Fiscal Year 2010, at 39 (n.d.), available at http://www. (showing state contributions to the New Jersey Teachers’ Pension and Annuity Fund for the 2005–2010 period); James Clemente, N.J. Dep’t of the Treasury, Div. of Pensions & Benefits, Teachers’ Pension and Annual Report: Fiscal Year 2005, at 37 (n.d.), available at
    reports_2005/tpaf.pdf (showing state contributions to the New Jersey Teachers’ Pension and Annuity Fund for the 2001–2005 period); Gabriel Roeder Smith & Co., Connecticut State Teachers’ Retirement System Report on the Actuarial Valuation as of June 30, 2006, at E-3, available at (showing state contributions to the Connecticut State Teachers’ plan fell below the ARC during the 2001–2006 period).

  50. [50]. Based on data assembled by the author. See infra Part IV.A; see also State of Conn., supra note 47, at 74–75 (noting that “[t]he State acts solely as the administrator and the custodian of the assets of the Connecticut Municipal Employees’ Retirement System” and describing the contributions from employees and participating municipalities).

  51. [51]. See, e.g., Conn. Mun. Emps. Ret. Sys., supra note 47, at 1–3 (noting that “[t]he State Retirement Commission is responsible for the administration of the [Connecticut Municipal Employees’ Retirement System]” and describing the contributions from the municipalities).

  52. [52]. In most cases, state governments do not administer pension plans themselves, although, as this Article shows, they often maintain control over certain funding decisions. A few state pension systems are administered by state agencies or departments within state government. In Michigan, for example,

    [t]he Office of Retirement Services [(“ORS”)] . . . administers four defined benefit pension plans, two defined contribution pension plans, and one defined benefit plus a defined contribution plan with combined net assets of nearly $50.54 billion . . . . ORS is a division of Michigan’s Department of Technology, Management and Budget.

    About ORS, Mich. Off. of Retirement Services,,4649,7-144-45490-263981--,00.html (last visited Oct. 31, 2014).

  53. [53]. Cal. Const. art. XVI, § 17(f).

  54. [54]. Cal. Gov’t Code § 20090 (West 2003) (in effect in 2010).

  55. [55]. Id.

  56. [56]. Id.

  57. [57]. Del. Code Ann. tit. 29, § 8308(e)(1) (Supp. 2008) (in effect in 2010).

  58. [58]. Id. § 8309(e)(2)–(4).

  59. [59]. 40 Ill. Comp. Stat. 5/7-174 (2008) (in effect in 2010).

  60. [60]. State government officials are typically themselves members of state pension plans.

  61. [61]. There are also more granular differences within each category, as both the actuarial approach and the statutory approach can be made more or less aligned with the GASB recommended ARC. For example, in the case of Florida, the statutory contribution rates are frequently adjusted by the legislature to preserve the required amortization period; indeed, Florida Statute Section 121.71 has been amended nearly each year between 2002 and 2010. See, e.g., 2009 Fla. Sess. Law Serv. 2009-76; 2008 Fla. Sess. Law Serv. 2008-139 (West); 2007 Fla. Sess. Law Serv. 2007-84 (West); 2006 Fla. Sess. Law Serv. 2006-35 (West); 2005 Fla. Sess. Law Serv. 2005-93 (West); 2004 Fla. Sess. Law Serv. 2004-293 (West); 2003 Fla. Sess. Law Serv. 2003-206 (West); 2002 Fla. Sess. Law Serv. 2002-402 (West). The case of Alaska Teachers Retirement System (“TRS”), on the other hand, shows the limitations of the provision for actuarial determination. A 2006 audit found that the “TRS board sought to maintain a level, long-term contribution rate of 12 percent for participating employers. Such a strategy resulted in the board often adopting rates lower than the actuarial calculated rates.” Pat Davidson, Departments of Administration and Revenue Public Employees’ Retirement System, Teachers’ Retirement System, and Alaska State Pension Investment Board 21 (2006) (citations omitted), available at

  62. [62]. Administrators of plans with statutorily determined rates often note the need for statutory adjustments to maintain actuarially sound plans. For example, employer contribution rates for the Montana Public Employers Retirement plan are fixed in statute. The 2010 CAFR for Montana PERS, prepared by the board and the actuary, points to the funding challenges of the system since the last statutory rate adjustment:

    All systems were actuarially funded within the required 30 years in 2007 and 2008. . . . Effective July 1, 2009, PERS-DBRP and SRS received the last employer contribution increase under the 2007 Legislative Session House Bill 131. Based on economic conditions of the past two years and according to the PERB’s June 30, 2010 actuarial valuations, the unfunded liability in PERS-DBRP, GWPORS and SRS will not amortize within 30 years.

    Mont. Pub. Emp. Ret. Bd., Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2010, at 30 (n.d.), available at (emphasis added).

  63. [63]. Cal. Gov’t Code § 20814(a) (West 2003) (emphasis added) (in effect in 2010).

  64. [64]. Cal. Gov’t Code § 20814(b) (West 2003) (emphasis added) (in effect in 2010).

  65. [65]. Cal. Edu. Code §22950(a) (West Supp. 2008) (emphasis added) (in effect in 2010).

  66. [66]. Cal. Edu. Code § 22951 (West 2002) (in effect 2010).

  67. [67]. Cal. Edu. Code §22954 (Supp. 2010).

  68. [68]. Cal. Edu. Code § 22955(a) (Supp. 2014) (in effect in 2010). In 2014, after “[d]ecade[s]-long efforts” to address the $73.7 billion funding gap, the California legislature approved rate increases for member, employer, and state contributions. CalSTRS 2014 Funding Plan, CalSTRS, (last visited Oct. 31, 2014).

  69. [69]. Haw. Rev. Stat. Ann. § 88-122(a), (e) (2012) (emphasis added) (in effect in 2010).

  70. [70]. Ky. Rev. Stat. Ann. §61.565(5), (6) (West Supp. 2009) (emphasis added) (in effect in 2010).

  71. [71]. N.Y. Retire. & Soc. Sec. Law §17(a) (McKinney Supp. 2004). Article VIII, section 5(D) of the New York State Constitution also provides that “[e]ach such pension or retirement system or fund thereafter shall be maintained on an actuarial reserve basis with current payments to the reserve adequate to provide for all current accruing liabilities.” N.Y. Const. art. VIII, § 5(D).

  72. [72]. See Ky. Rev. Stat. Ann. §161.550 (Supp. 2006) (in effect in 2010).

  73. [73]. Starting in 2010, however, employer contribution rates for the employee retirement system (but not the County Employees Plan) are set in statute. See id. § 161.550(6).

  74. [74]. For example, a legislative research commission note from October 19, 2004, explains that:

    2004 (1st Extra Sess.) Ky. Acts ch. 1, sec. 9, provides, ‘Notwithstanding KRS 61.565 [the statutory provision for actuarially determined rates], the employer contribution rate for an entity participating in the Kentucky Employees Retirement System or State Police Retirement System shall be as follows: (1) From July 1, 2004, through June 30, 2005, the contribution rates shall be no more than 5.89 percent for nonhazardous duty employees, 18.84 percent for hazardous duty employees . . . .’

    Ky. Rev. Stat. Ann. § 61.565.

  75. [75]. Ky. Ret. Sys., Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2010, at 2 (2010), available at

  76. [76]. See infra Table 4.

  77. [77]. Cal. Gov’t Code §20831 (West 2003) (in effect in 2010).

  78. [78]. Cal. Educ. Code §23007 (Supp. 2014) (in effect in 2010).

  79. [79]. Haw. Rev. Stat. Ann. §88-126(c) (2012) (in effect in 2010).

  80. [80]. Ky. Rev. Stat. Ann. §61.675(3)(a), (b) (Supp. 2013) (in effect in 2010).

  81. [81]. N.Y. Retire. & Soc. Sec. Law §17(d), (e) (McKinney Supp. 2004) (in effect in 2010).

  82. [82]. See supra Part III.A.2.

  83. [83]. See Alaska Const. art. 12, §7; Ariz. Const. art. 29 §1; Haw. Const. art. XVI, § 2; Ill. Const. art. XIII, §5; La. Const. art. X, §29; Mich. Const. art. IX §24; N.Y. Const. art. 5 §7.

  84. [84]. For example, in People ex rel. Illinois Federation of Teachers v. Lindberg, the court held that the constitutional provision protecting against the impairment of pension benefits did not create a contractual right to enforce a specific level of pension funding and therefore did not preclude the governor from decreasing appropriations to the pension funds. People ex rel. Ill. Fed’n of Teachers v. Lindberg, 326 N.E.2d 749, 753 (Ill. 1975). See generally Darryl B. Simko, Of Public Pensions, State Constitutional Contract Protection, and Fiscal Constraint, 69 Temp. L. Rev. 1059 (1996).

  85. [85]. See, e.g., Ariz. Const. art. XXIX, §1 (“Public retirement systems shall be funded with contributions and investment earnings using actuarial methods and assumptions that are consistent with generally accepted actuarial standards.”); Mont. Const. art. VIII, § 15 (“Public retirement systems shall be funded on an actuarially sound basis.”).

  86. [86]. La. Const. art. X, § 29 (“For public retirement systems whose benefits are guaranteed by this constitution.... The legislature shall, in each fiscal year, by law, provide an amount necessary to fund the employer portion of the normal cost... [and] provide for the amortization of the unfunded accrued liability existing as of June 30, 1988, which shall be determined in accordance with the method of valuation selected in [article X, section 29(E)(1)] above, by the year 2029, commencing with Fiscal Year 1989–1990.... [Such] amounts are... hereby guaranteed payable, each fiscal year, to each retirement system covered herein. If, for any fiscal year, the legislature fails to provide these guaranteed payments, upon warrant of the governing authority of the retirement system, following the close of said fiscal year, the state treasurer shall pay the amount guaranteed directly from the state general fund.”).

  87. [87]. In a 2014 study, Amy Monahan presents a qualitative examination of states with constitutional funding requirements and finds that “having a constitutional funding requirement in place does not guarantee sound funding methodology or funding discipline.” See Amy B. Monahan, State Fiscal Constitutions and the Law and Politics of Public Pensions, 2014 U. Ill. L. Rev. (forthcoming), available at

  88. [88]. In 2012, GASB adopted significant changes to the reporting requirements. See infra Part VI.

  89. [89]. Emps.’ Ret. Sys. of the State of Haw., Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2010, at 5 (2013), available at

  90. [90]. Id. at 66.

  91. [91]. Id.

  92. [92]. See Emps.’ Ret. Sys. of the State of Haw., Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2009, at 89 (2012), available at

  93. [93]. For example, the State of Indiana contributes to the Public Employees’ Retirement Fund, an agent multiple-employer plan. The 2010 State CAFR discloses that as of July 1, 2009, “the state employees portion of the plan was 87 percent funded. The actuarial accrued liability for benefits was $2.4 billion, and the actuarial value of assets was $2.1 billion, resulting in an unfunded actuarial accrued liability (UAAL) of $0.3 billion.” Office of Ind. Auditor of State, Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2010, at 104 (2011), available at The plan’s CAFR discloses that overall, the plan was 93.1% funded, with .9 billion in total unfunded actuarial accrued liability. Ind. Pub. Emps.’ Ret. Fund., Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2010, at 56 (2010), available at
    files/PERFCAFR2010.pdf. Similarly, the 2012 CAFR for Colorado PERA, the administrator of several cost-sharing plans, explicitly acknowledges the relatively limited disclosure obligations imposed on employers in cost-sharing plans, noting that:

    [i]f the Division Trust Funds of PERA were single employer plans or an agent multi-employer plan, [the cumulative net pension obligation], allocated to each employer, would need to be reported as a liability on the employers’ financial statements. As the employers are part of a cost-sharing multi-employer plan, they . . . only record a liability if they have not paid the statutorily required contribution rate.

    Colo. Pub. Emps.’ Ret. Ass’n, Comprehensive Annual Financial Report for the Year Ended December 31, 2012, at 35 (2013), available at

  94. [94]. Letter from the Employer Cost-Sharing Coalition to Michelle Czerkawski, GASB (Dec. 27, 2011). The Coalition is “a group of cost-sharing plans and employers including California State Teachers’ Retirement System, the Colorado Public Employees’ Retirement Association, the Kentucky State Teachers’ Retirement System, the Nevada Public Employees’ Retirement System, Ohio Public Employees’ Retirement System, and the University of Colorado.” Id.

  95. [95]. Ctr. for Ret. Research at Bos. Coll., State and Local Defined Benefit Plans, Pub. Plans Database, (last visited Oct. 31, 2014) (follow hyperlink for dataset; the CRR dataset downloaded for this Article is on file with the author).

  96. [96]. See, e.g., Gokhale, supra note 16; U.S. Gov’t Accountability Office, supra note 3, at 2; Robert K. Triest & Bo Zhao, The Role of Economic, Fiscal, and Financial Shocks in the Evolution of Public Sector Pension Funding 5 (Fed. Reserve Bank of Bos., Working Paper No. 13-26, 2013), available at

  97. [97]. Ctr. for Ret. Research at Bos. Coll., supra note 95.

  98. [98]. Id.

  99. [99]. Id.

  100. [100]. Peng observes:

    With regard to local employee coverage, there are 37 states that offer the same pension plan to both state employees and local employees and the plan is managed by a state-level retirement system. In 12 states where state pension plans do not cover local employees, there is a statewide pension plan or municipal retirement system for just local governments. Massachusetts is the only state that does not have either a state pension plan for local employees or a state-level municipal retirement system.

    Peng, supra note 10, at 16. Of the systems identified by Peng (and accordingly, the relevant plans administered by each system), all but the following are included in the final dataset used in this Article: Georgia Municipal Employee Benefit System (which provides only administrative support for local plans); Louisiana Municipal Employees Retirement System (for which the relevant data was not available), and the Vermont Municipal Employees Retirement System (for which the relevant data was not available).

  101. [101]. The five plans are: Connecticut Municipal, Delaware County and Municipal (Police), Delaware County and Municipal (Other), Mississippi Municipal, and Pennsylvania Municipal.

  102. [102]. One outlier point is excluded as it is likely the result of a coding error.

  103. [103]. Nat’l Educ. Ass’n, NEA Issue Brief on Pension Protections in State Constitutions 17–22 (2004), available at

  104. [104]. State and local governments increase borrowing as one means of coping with gaps between expenditures and revenues. See, e.g., Steven Maguire, Cong. Research Serv., 7-5700, State and Local Government Debt: An Analysis 1 (2011) (noting that state and local governments have used a “combination of rainy day fund withdrawals, tax increases, spending reductions, and in some instances, borrowing to meetthese balanced budget requirements”). In this Article, data for this debt variable is calculated using data from the U.S. Census Bureau State & Local Government Finance statistics and the Bureau of Economic Analysis for all but the 2001 and 2003 periods, for which debt data is not available from the U.S. Census Bureau on a state-by-state basis. For those years, estimates are taken from Christopher Chantrill, Government Spending in the U.S.,, (last visited Oct. 31, 2014.

  105. [105]. See, e.g., Munnell et al., supra note 42, at 4–5; see also Mitchell & Smith, supra note 37, at 278, 288.

  106. [106]. The correlation matrix reveals that there are positive correlations between certain types of institutional features, which may suggest that certain types of institutions come in packages. See infra Table 4. However, tests for multicollinearity using the variance inflation factors confirm that none of the predictors are highly collinear.

  107. [107]. See Monahan, supra note 87, at 5 (suggesting that “[a] primary weakness in these constitutional funding requirements is that there is no generally accepted standard for funding a pension on an ‘actuarially sound’ basis, and the constitutional language does not define the funding requirement any further”).

  108. [108]. For example, some of the disciplining effects of the enforcement provisions for state-administered plans with only municipal employers may be picked up by the dummy variable that isolates such plans in order to assess the impact of the separation of funding and management. Notably, in plans that include both the state and local entities as employers, the mean percentage of ARC contributed is 96.00% when there is a withholding provision and 87.75% otherwise.

  109. [109]. For example, the entire seven-member Retirement Board of the Nevada Public Employees Retirement System is appointed by the governor. Nev. Rev. Stat. § 286.120 (2013) (in effect in 2010).

  110. [110]. The five “closed” plans include: the Alaska Public Employees Plan (after 2005), see State of Alaska, Comprehensive Annual Financial Report for the Fiscal Year July 1, 2009–June 30, 2010, at 109 (n.d.), available at; Alaska Teachers Plan (after 2005), see id. at 110; Mississippi Municipal Retirement Plan, see Miss. Code Ann. § 21-29-17(d) (1999) (in effect in 2010); Washington Teachers Plan I, see State of Wash., Office of Fin. Mgmt., Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2010, at 118 (2010), available at; Washington Public Employees Plan I, see id. at 115; and the Washington Law Enforcement Officers and Fire Fighters Plan I, see id. at 122.

  111. [111]. A contribution may exceed 100% of the ARC either because statutory provisions mandate contributions in excess of the ARC amount, or because the calculations for the percentage of ARC contributed were based on actuarial data that is more recent than the data used to determine the employer contribution amounts, which, for certain plans, are set several years in advance. GASB, however, requires that financial reports be based on the results of an actuarial valuation that is performed not more than two years before the plan’s reporting date for that year. Governmental Accounting Standards Bd., Governmental Accounting Standards Series: Statement No. 25, at 15 (1994), available at

  112. [112]. It is worth reiterating that GASB cannot require plan sponsors to actually make any contributions; during the time period analyzed in this Article, its standards merely required the calculation of the “annually required contribution” using a set of parameters set forth in GASB guidelines.

  113. [113]. In considering the amortization period, it is important to distinguish between the amortization period used for purposes of the GASB ARC calculation, which must be less than 30 years after 2006, and the remaining amortization period under the funding policy of any given plan. Indeed, several plans in the sample report at various times that under the funding methodology in effect during the reporting period, the amortization period is much longer than the 30 years permitted by the GASB standards. Other plans report amortization periods that are shorter than the permitted 30 years.

  114. [114]. Pursuant to GASB Statement 25, GASB has not placed constraints on the kinds of smoothing techniques or the length of smoothing periods used in the actuarial valuation of assets. Governmental Accounting Standards Bd., supra note 111. Therefore, different systems calculate the actuarial (rather than the market) value of their plan assets differently. Some systems do not incorporate smoothing techniques and simply take the market value as the actuarial value of assets, while others use fairly straightforward smoothing approaches. For example, the Missouri Public Schools Retirement System uses an asset-smoothing method by which investment returns above or below 8% are recognized over a five-year period. See Pub. Sch. & Educ. Emp. Ret. Sys. of Mo., Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2010, at 9 (2010), available at
    2010PSRSCAFR/2010PSRSCAFR.pdf. Other systems use a fixed multi-year smoothing approach but impose a “corridor” that limits the amount by which the market and actuarial values can diverge. See State Teachers Ret. Sys. of Ohio, Comprehensive Annual Financial Report: Fiscal Years Ending June 30, 2010, at 15 (2010), available at
    _pdfs/annualreports/cafrs/2010_cafr.pdf (“Market changes in investment assets are smoothed over a four-year period for valuation purposes, except that the actuarial value of assets shall not be less than 91% nor more than 109% of market value.”). Still others base the actuarial value on the expected rate of return, with an adjustment for market performance. See Emp. Ret. Sys. of Tex., Comprehensive Annual Financial Report 99 (2012) (“The actuarial value of assets is determined as the expected value of plan assets as of the valuation date plus 20% of the difference between the market-related value and the expected value. The expected value equals the actuarial value of plan assets as of the prior valuation date, plus contributions, less benefit payments and administrative expenses, all accumulated at the assumed rate of interest to the current valuation date.”).

  115. [115]. Data is taken from the CRR PPD and from plan reports. Amortization period data is available for 992 plan-year data points; smoothing data is available for 1058 plan-year data points.

  116. [116]. The significant variation in smoothing methods complicates the classification and quantification of such techniques. The CRR database includes a numeric value for the “smoothing period” used by the plans in its sample. This approach, however, may mask certain techniques used by plans to extend the smoothing period. For example, a number of plans that are coded as having a “1 year” smoothing period employ a technique whereby the actuarial value of assets is determined as the expected value of plan assets plus 20% (or 33% or 10%) of the difference between the market-related value and the expected value. Insofar as only 20% of the difference between the market and the expected value is recognized in one year, this approach may, in effect, extend smoothing beyond one year. Furthermore, systems that choose to recognize 20% of the difference between expected and market value should not necessarily be equated with those that choose to recognize 33% of the difference.

  117. [117]. For comparison, under an alternative coding scheme, plans that are coded as having a one-year smoothing period in the CRR database but that take the smoothing approach described in supra note 114 are coded as having a smoothing period that corresponds to the percentage of the difference between expected and market returns that is recognized in each year. A plan that recognizes 20% of the difference is coded as having a five-year smoothing period.

  118. [118]. See, e.g., Jonathan Barry Forman, Funding Public Pension Plans, 42 J. Marshall L. Rev. 837, 870, 871–72 (2009) (stating that “most state and local governments will want to increase the age and service requirements for pension benefits, and many will want to increase employee contributions,” and recommending a shift to cash balance plans). In a related vein, the closely watched Detroit bankruptcy, and the proposed reorganization plan, involved scaling back of pension benefits to retirees and current employees. See Alana Semuels, Detroit Bankruptcy Plan Includes Deep Pension Cuts, L.A. Times (Feb. 22, 2014, 2:00 AM),
    nation/la-na-detroit-bankruptcy-20140222-story.html. The approved reorganization plan included smaller-than-anticipated cuts to the benefits of retirees due to state and foundation contributions to shore up the pension system. See Monica Davey & Mary Williams Walsh, Plan to Exit Bankruptcy Is Approved for Detroit, N.Y. Times, Nov. 7, 2014, at A11. Fundamentally, however, municipal bankruptcy can provide some immediate debt relief for a distressed city, but it cannot, by its very nature, impose structural changes or long-term institutional reforms to promote funding discipline. See generally Adam J. Levitin, Fiscal Federalism and the Limits of Bankruptcy, in When States Go Broke: The Origins, Context, and Solutions for the American States in Fiscal Crisis (Peter Conti-Brown & David A. Skeel, Jr. eds., 2012).

  119. [119]. Governmental Accounting Standards Bd., Governmental Accounting Standards Series: Statement No. 67 (2012), available at
    DocumentPage?cid=1176160220594&acceptedDisclaimer=true; Governmental Accounting Standards Bd., Governmental Accounting Standards Series: Statement No. 68 (2012), available at

  120. [120]. In June 2012, GASB revised its guidance for discounting pension liabilities. Starting with the 2015 fiscal year, the expected rate of return may be applied only to liabilities that will be covered by the plan’s net position and projected contributions; for the remaining liabilities, the municipal borrowing rate for a 20-year general obligation bond must be used. See Governmental Accounting Standards Bd., New GASB Pension Statements to Bring About Major Improvements in Financial Reporting (2013), available at
    GASB/Document_C/GASBDocumentPage&cid=1176160140567. For a general discussion of the impact of this change, see, for example, Caitlin Kenney, Public Pensions Are About to Look Less Healthy, Planet Money (July 20, 2012, 3:43 AM),

  121. [121]. GASB Statement 67, “Financial Reporting for Pension Plans”—an amendment of GASB Statement No. 25—is effective for financial statements for fiscal years beginning after June 15, 2013. Governmental Accounting Standards Bd., Statement No. 67, supra note 119, at v. GASB Statement No. 68, “Accounting and Financial Reporting for Pensions”—an amendment of GASB Statement No. 27—is effective for fiscal years beginning after June 15, 2014. Governmental Accounting Standards Bd., Statement No. 68, supra note 119, at ix.

  122. [122]. Governmental Accounting Standards Bd., Statement No. 68, supra note 119, at v–vi.

  123. [123]. Id. at 98.

  124. [124]. Id. at 164.

  125. [125]. Alicia H. Munnell et al., Ctr. for Ret. Research at Bos. Coll., How Would GASB Proposals Affect State and Local Pension Reporting? 6–7 (2012), available at

  126. [126]. Governmental Accounting Standards Bd., Statement No. 68, supra note 119, at vi.

  127. [127]. Id.

  128. [128]. See, e.g., Josh Barro, How Congress Can Help State Pension Reform, 2012 Nat’l Aff. 92, 102 (2012), available at; see also Roger Lowenstein, The Long, Sorry Tale of Pension Promises, Wall St. J. (Oct. 1, 2013), http:// (“Before we get more Detroits, or more Studebakers, the federal government should enact an ERISA (with teeth) for public employers.”).

  129. [129]. In Louisiana in 2011, for example, voters approved a constitutional amendment that, in addition to previously adopted constitutional requirements to fund the retirement plans in an actuarially sound manner, requires further that in fiscal years 2013–2014 and 2014–2015, 5% of money designated in the official forecast as nonrecurring be applied toward the balance of the unfunded accrued liability which existed as of June 30, 1988, for the Louisiana State Employees’ Retirement System and the Teachers Retirement System of Louisiana. See La. Const. art. VII, §10(D), amended by 2011 La. Acts 2169 (Act. No. 422), approved Oct. 22, 2011. The newly adopted constitutional provision also requires that in Fiscal Year 2015–2016 and every fiscal year thereafter, 10% of such nonrecurring revenue be applied to such purposes. La. Const. art. VII, § 10(D)(2)(b)(iii).

  130. [130]. States are increasingly concerned about the funding status of independent local plans, and some are looking at means of imposing state oversight on such plans. For example, in Tennessee, The Public Employee Defined Benefit Financial Security Act of 2014 requires local governments and authorities that are not part of the Tennessee Consolidated Retirement System to meet certain actuarial benchmarks. Tenn. Code Ann. §§ 9-3-501–07 (Supp. 2014). The new law gives the state treasurer the power to withhold governments’ share of state revenue if the contributions are not met. Id. at 9-3-507(a); Press Release, Tenn. Dep’t of the Treasury, Governor Haslam Signs Landmark Local Government Pension Reform Bill into Law (May 28, 2014), available at; see also Margaret Newkirk, Memphis Pension Blues Foretold with Tennessee Bill: Muni Credit, Bloomberg, (Apr. 13, 2014, 7:00 PM), (noting that 13 of the 31 local plans to be subject to the new law “didn’t make their full contribution in 2012,...[paying] a combined $86.4 million less than was actuarially required”).

  131. [131]. Although Rhode Island implemented major reforms to its state-administered plans in 2011, state leaders have struggled to bring into line the state’s 36 locally administered plans, a few of which have contributed to actual or threatened municipal bankruptcies in Rhode Island. Gina M. Raimondo, Op-Ed, Next Up: Reforming Local Pension Plans, Providence J. (Dec. 21, 2011), State Treasurer Gina Raimondo wrote that, “Collectively, the locally-administered plans have a reported unfunded liability of approximately $2.1 billion, and funding level of only 40 percent.” Id. She encouraged local and state leaders to consider bringing such local plans into the state system, arguing that “[t]his move would provide for uniform benefits across municipalities, pooling of assets managed by the state and a legal requirement for municipalities to make annual required contributions.” Id. Recent research on locally administered plans suggests, however, that despite negative press about such plans, they are nearly as well funded as state-administered plans. See generally Munnell et al., supra note 43. State-administered plans, however, vary in the nature of the plan sponsors. Some include only state-level employers, while others include both state and local employers. A third type includes only local employers. See supra Part III.A.1. For plans that include both state and local employers, the challenge is to allocate any underfunding, as well the funding patterns, among the different plan sponsors. More generally, the critical question is whether plans for local employers that are part of state-administered systems are better funded than similar plans that are administered at the local level, and whether the result is affected by the type of state-administered plans included in the analysis.

  132. [132]. For the number of plans in each state, see Becker-Medina, supra note 9, at 34–39 and Survey of Public Pensions: State- and Locally-Administered Defined Benefit Data, U.S. Census Bureau, (last visited Oct. 31, 2014).

  133. [133]. See generally State of Hawaii, Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2010, at 92 (2010), available at

  134. [134]. See generally Martin J. Benison, State of Mass., Off. of the Comptroller, Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2010, at 109 (2011), available at

  135. [135]. See generally State of Mo., Office of Admin. Div. of Accounting, Comprehensive Annual Financial Report for the Fiscal Year Ended June 30, 2010, at 59 (2011), available at; About Us, Mo. Loc. Gov. Emps. Retirement Sys., (last visited Oct. 31, 2014).


David and Pamela Donohue Assistant Professor, Boston College Law School. 

I would like to thank Rick Antle, Ian Ayres, Shuky Ehrenberg, Yan Epelboym, Henry Hansmann, John Langbein, Amy Monahan, John Morley, James Naughton, Joshua Rauh, Roberta Romano, Maxim Pinkovskiy, Brick Susko and the participants of the Federal Reserve Bank of Cleveland 2013 Conference on Public Pension Underfunding, the 2013 American Law & Economics Association Annual Meeting, the 2013 Netspar International Pension Workshop, the 2013 American Law & Economics Association Annual Meeting, and the Brandeis University 2013 Municipal Finance Conference for helpful comments on earlier drafts. All errors are my own.