As discussed above, we hypothesize that firms’ decisions to freeze their DB plans
are motivated by potential changes in financial accounting standards embodied in
SFAS 158. To enhance our inferences from hypothesis testing, we use a matched-pair
research design[15]. Accordingly, we compare freeze firms with non-freeze firms that
are similar in key characteristics (industry and size) to control for their influence on the
freeze decision[16]. We examine the association of accounting motivations with
the pension freeze decision using logistic regression models and decomposed balance
sheet pension measures including the potential SFAS 158 balance sheet effect while
controlling for other important factors in the freeze decision relating to cash flows and
firm competitive pressures.
We decompose the DB plan funded status (NETPENASSET), which is the difference
between the PBO and fair value of plan assets (FVA) in two ways to test our hypothesis.
First we examine the association of separate liability (PBO) and asset (FVA) components
with the DB plan freeze decision. Second, and more importantly, we disaggregate the
funded status of the DB plan into two components – one that is recognized under the SFAS
87 regime as the net prepaid pension asset or the net accrued pension liability (BS_RECOG)
and another that measures the potential incremental balance-sheet effect of full recognition
of the funded status under SFAS 158 (SFAS158_EFFECT) – which facilitates our
investigation of the association between the incremental SFAS 158 effect and the freeze
decision. We measure our key variable, SFAS158_EFFECT, as the difference between
NETPENASSET and BS_RECOG. We measure all of these variables at the end of the
fiscal year prior to the year of the freeze decision, andwe scale by lagged total assets[17].
In our logistic regression models, we code the dichotomous dependent variable
(FREEZE) as 1 if the firm has announced a pension freeze and 0 otherwise. We also
include independent variables to control for economic and regulatory factors that are
potentially associated with the pension freeze decision.
As we discuss in Section 2.2, beginning in 2008 the PPA will require many
underfunded plans to step up their contributions. Anticipating the new PPA legislation,
firms with severely underfunded plans requiring large contributions and otherwise cashstrapped
firms may be motivated to freeze their plans. We use two variables to control
for the potential effects of the PPA on the DB plan freeze decision. First, we use
CONTRATIO, the ratio of firm cash contributions to the pension plan to cash flow from
operations, to capture the relative impact of DB plan contributions. Alternatively, we use
AVG_CFO, or cash flow from operations, to control for the overall cash flow position of
the firm. As we discuss in Section 1, a key factor often cited by management in support
of their freeze announcement is to improve the competitive position of the firm.
Accordingly, we control for this factor by employing AVG_ROA, or the income before
extraordinary items, taxes, and interest expense, and ROADIFF, which is the difference
between a particular company’s ROA and industry average ROA based on the firm’s