likelihood of tax increases and expenditure cuts. We also find that pension understatements are associated with higher
future labor costs. Importantly, we find that the positive relation between pension funding gap understatement and future
labor costs is associated with the inherent methodology in the GASB rules, which systematically understate the funding gap,
and is not associated with opportunistic reporting by state governments.
Our results should be of interest to governments and policymakers, as they have important implications for the current
direction of the GASB pension reporting regime. GASB recently issued Statements 67 and 68 that replace Statements 25 and 27.
The new standards are intended to improve financial reporting transparency by requiring the disclosure of a substantial
amount of information not required in the old regime. In addition, the new standards require that state and local governments
use a single discount rate that combines both a funding approach (long-term expected rate of return on plan assets) and a
liability approach (high rated municipal bond yields).59 While this is a move toward the FASB approach, it is only a partial one,
as the new pension accounting standards will continue to systematically understate the funding gap, albeit on a reduced scale.
This is noteworthy, since our results suggest that the funding approach is associated with states committing to additional
expenditures.
Appendix A. Duration estimation
Our research design requires that we re-estimate each state's pension liabilities using alternative discount rates.
To facilitate these calculations, we estimate the duration of each state's pension liability. Duration is a measure of the
weighted average time over which benefit payments are made from the plan. It provides an effective approach to adjusting
the pension liability to reflect different discount rates. In particular, a pension plan with a duration of 15 years will
experience a 15% increase in the pension liability for a 1% reduction in the discount rate. We estimate the duration by first
estimating each state's total pension liability, and then measuring the change in our estimate when we adjust the discount
rate by 1%. To estimate a state's total pension liability, we use an aggregate actuarial method based on a single hypothetical
participant whose characteristics reflect those of the plan as a whole. In this case, the pension liability for the entire plan is
simply the pension liability for this participant. The aggregate method is commonly used to estimate pension liabilities for
both public and private pension plans. When properly applied, the pension valuation under both the individual and
aggregate approaches produces virtually identical results.
We estimate the pension liability separately for the active and inactive participants. For the active participants, we need
three distinct groups of items for the calculation: information about the participants, information that relates to the benefit
formula, and information on the specific actuarial assumptions. We collect information on the total pay, average service and
average age of all the active participants to identify the attributes of the hypothetical employee for purposes of applying the
aggregate cost method. We use the benefit multiplier from the plan provisions to determine the size of the retirement
benefit. We use the actuarial assumptions for the discount rate, the salary growth assumption, the retiree cost of living
adjustment (COLA), and the average retirement date to determine the value of this retiree benefit. For the inactive
participants we follow the same approach, except that our hypothetical inactive participant is determined using the total
benefit payments (rather than total pay).
We compare our estimated pension liability with the reported amount for each state in our sample and find that the
difference is within 10 percent. This result suggests that our approach is reasonable and that the estimate of the duration we
derive from our liability estimates is reliable.