Description
The purpose of this paper is to examine the determinants of US firms’ postretirement
benefits choices.
Accounting Research Journal
Factors affecting corporate choices of postretirement benefits in the USA
Sharad Asthana
Article information:
To cite this document:
Sharad Asthana, (2008),"Factors affecting corporate choices of postretirement benefits in the USA",
Accounting Research J ournal, Vol. 21 Iss 2 pp. 123 - 146
Permanent link to this document:
http://dx.doi.org/10.1108/10309610810905926
Downloaded on: 24 January 2016, At: 21:06 (PT)
References: this document contains references to 52 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 591 times since 2008*
Users who downloaded this article also downloaded:
Kym Boon, J ill McKinnon, Philip Ross, (2008),"Audit service quality in compulsory audit tendering:
Preparer perceptions and satisfaction", Accounting Research J ournal, Vol. 21 Iss 2 pp. 93-122 http://
dx.doi.org/10.1108/10309610810905917
Ahsan Habib, Istiaq Azim, (2008),"Corporate governance and the value-relevance of accounting
information: Evidence from Australia", Accounting Research J ournal, Vol. 21 Iss 2 pp. 167-194 http://
dx.doi.org/10.1108/10309610810905944
Beilei Zhang, David Emanuel, (2008),"The provision of non-audit services and earnings conservatism: Do
New Zealand auditors compromise their independence?", Accounting Research J ournal, Vol. 21 Iss 2 pp.
195-221 http://dx.doi.org/10.1108/10309610810905953
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Factors affecting corporate
choices of postretirement bene?ts
in the USA
Sharad Asthana
Department of Accounting, College of Business,
University of Texas at San Antonio, San Antonio, Texas, USA
Abstract
Purpose – The purpose of this paper is to examine the determinants of US ?rms’ postretirement
bene?ts choices.
Design/methodology/approach – The paper uses empirical methodology (univariate and
multivariate) to test the research hypotheses.
Findings – Industry norm, average employee age, ?nancial structure, and ?rm size are signi?cant
factors in the determination of the proportion of compensation that is deferred. Industry norm,
?nancial structure, and ?rm size are signi?cant factors that determine the percentage of deferred
compensation that is negotiated as de?ned bene?ts. Finally, industry norm, corporate tax rates, and
cash ?ow help explain the percentage of de?ned bene?ts that are paid in the form of retiree health
bene?t plans.
Research limitations/implications – Data requirements might bias the sample towards larger
sized ?rms. Data availability limits the number of observations in 2000 and 2001.
Practical implications – The trends in post-retirement bene?ts reported in this paper are important
for policy makers.
Originality/value – These ?ndings have implications for the baby boomers. The trend to offer
smaller proportion of compensation as deferred bene?ts re?ects the increasing costs of deferral to the
employers. This increases the employees’ responsibilities to save on their own. This also would shift
the retirees’ dependence on the public pension system for their retirement income. The trend to favor
de?ned-contribution plans instead of de?ned-bene?t plans re?ects the employers’ attempts to
diversify their risks of paying promised post-retirement bene?ts by transferring the risk to the
employee. On the other hand, the popularity of de?ned-contribution pension plans also re?ects the
increased Government’s incentives to encourage savings via 401-k plans and employee’s willingness
to manage their own pension portfolios.
Keywords Pensions, Retirement, Retirement insurance, Taxes, Labour market, UnitedStates of America
Paper type Research paper
1. Introduction
Research that deals with the management of post-retirement bene?ts is critical in the
light of the changing population pro?le. According to the latest Population Pro?le of the
United States (US Census Bureau), the population of 65-years old and above in the USA
is expected to more than double by 2050. The main reasons for this increase are the
Baby Boomers that retire between 2011 and 2029 and the increase in human longevity.
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
The author is thankful to Steve Balsam, Roland Lipka, Dave Ryan, and participants at the 2004
Annual American Accounting Association Conference for their helpful comments. The author is
also grateful to Shan Chen and Debbie Sinclair for their help in data collection. The author also
acknowledges ?nancial support from Baruch College and Temple University.
Factors affecting
corporate choices
123
Accounting Research Journal
Vol. 21 No. 2, 2008
pp. 123-146
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309610810905926
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Thus, planning for the welfare of the aged is one of the major priorities for the
government[1]. The government tries to ful?ll this social obligation by providing public
postretirement insurance (such as, social security, medicare, and medicaid) and
encouraging private postretirement plans (such as, pensions and retiree health care).
The current problems with the public insurance system are expected to multiply
with the aging of the Baby Boom cohort, health care cost in?ation, and medical
advances resulting in the extension of the retirement years. Projections by the US
Social Security Department show that as a result of these demographic shifts, the ratio
of Social Security bene?ciaries to workers will increase to 60 percent by 2030 (from
33 percent in the late 1990s). Expected ?nancial woes of the public insurance system in
the future would result in retirees depending more and more on their private health
care plans. This factor, combined with the expected decline in individual income as a
result of retirement and the increase in health care costs with age, make private
pension and retiree health care schemes critical determinants in individual and social
retirement planning.
The main focus of this paper is to examine how ?rms make the following three
important choices pertaining to postretirement bene?ts and their socio-economic
implications, given the changing demographics and recent amendments to the US
pension laws:
(1) What proportion of the total compensation should be deferred/cash?
(2) What proportion of the deferred compensation should be de?ned
bene?t/de?ned contribution?
(3) What proportion of the de?ned bene?t compensation should be retiree health
care/pension bene?ts?
These are important questions for several reasons. First, the total pension and retiree
health care obligations of US companies are substantial. Second, while prior research
has examined the funding, reporting, and investing policy of ?rms’ pension and post
retirement bene?t choices, little research has focused on ?rms’ bene?ts policies
themselves. Third, the US Internal Revenue Code (IRC) provides substantial tax
bene?ts for certain retirement bene?ts (e.g. de?ned bene?t pension plans) and nearly
none for other bene?ts (e.g. post-retiree health care bene?ts). It is important to
understand how tax policy affects ?rm choices.
These decisions are not trivial for US ?rms. In my sample, about 4 percent of total
compensation is deferred. Also, during the sample period, the total deferred
compensation exceeds 22 percent of total assets, 33 percent of market value, and
60 percent of book value. Given the magnitude of deferred compensation and its
alternative forms, explanations about the mix between current and deferred
compensation, the mix between de?ned bene?t and de?ned contribution plans, and
the mix between de?ned retirement health bene?ts and de?ned pension bene?ts are
important. I use publicly available data, including the Financial Accounting Standards
Board’s mandated accounting information to examine the determinants of ?rm
choices[2].
Some previous researchers have focused on ?rm choices to offer new plans and to
close old plans. Kruse (1995), Dorsey (1987) and Ippolito and Thompson (2000) provide
interesting insights into how these choices are made. The essence of their ?ndings is
that the reasons for changes in the overall mix are complex. Importantly, Kruse reports
ARJ
21,2
124
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
that companies only infrequently substitute de?ned contribution plans for de?ned
bene?t plans – only four percent of his sample switched the form of their pension plans
from bene?t to contribution. Rather, new plans are created while old plans become less
important as the participating workforce diminishes.
Researchers also report that signi?cant increases, in both absolute and relative
terms, in de?ned contribution plans occurred during the late 1980s and early 1990s
(Kruse, 1995 and Ippolito, 1995). In fact, twice as many employees currently participate
in de?ned contribution plans as do in de?ned bene?t retirement plans (Bureau of Labor
Statistics, Department of Labor, 2003, Table 3). Firms have also made changes in their
retiree health bene?ts plans primarily as a result of Financial Accounting Standard 106
(SFAS 106)[3]. In a signi?cant departure from the previous accounting practice of
accounting for non-pension post-retirement bene?ts such as health care on a
pay-as-you-go basis, SFAS 106 requires companies to recognize in the current period
the compensation expense and the corresponding liability for these bene?ts (FASB,
1990). According to Fronstin (2001, p. 9), a Mercer/Foster Higgins:
[. . .] annual survey of employers with 500 or more workers shows that those currently
expecting to continue offering health bene?ts to future early retirees declined from 46 percent
in 1993 to 31 percent in 2000.
Similarly, “another survey of larger employers . . . showed that the percentage . . .
declined from 88 percent in 1991 to 73 percent in 2000.” Diamond (2000) reports that
these costs are substantial for many companies. As a result, many companies have
taken various steps to reduce retiree health bene?ts in order to limit expenses and their
liability exposure[4].
The ?rst set of tests (univariate) corroborate Kruse’s (1995) and Fronstin’s (2001)
observations that during the period of study signi?cant changes in the de?ned bene?ts
versus de?ned contributions mix (MIX2) and in the mix of post retirement health care
bene?ts to total de?ned bene?ts (MIX3) occurred[5]. The second set of tests
(multivariate) examines the determinants of the three decisions. As postulated, I ?nd
that MIX1, the percentage of deferred compensation, is positively and reliably related
to industry norm, average age of employees, and ?rm size. The percentage deferred is
negatively and reliably related to the ?rm’s ?nancial structure (debt to assets).
MIX2, the ratio of de?ned bene?ts to total deferred compensation is, as expected,
positively and reliably related to industry norm and ?rm size and is negatively and
reliably related to ?nancial structure. How much of the de?ned bene?ts will be in the
form of retirement health care (MIX3) is found to be positively and reliably related to
industry norm and ?rm size. Corporate tax rates and operating cash ?ows have
reliably negative effects on this ratio.
The reported decline in the proportion of deferred compensation out of total
compensation is consistent with increasing costs of providing deferred compensation.
The bene?ts of deferred compensation, mainly, tax bene?ts and employee retention
through bonding are obviously becoming less attractive compared to the increasing
costs of deferral. Stricter pension accounting rules, such as SFAS 87, 106, 132, 132R,
158, combined with the funding reform provisions of The Pension Protection Act 2006
also add to the level of volatility and cost of providing deferred bene?ts (FASB, 1998,
2003)[6]. These changes in the pension laws have also lead to the decline in the
popularity of de?ned-bene?t pension plans in favor of de?ned-contribution plans, since
Factors affecting
corporate choices
125
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the former limit the liability of the employer. The improving stock market
performance, easier access to technology for managing your own investment
portfolios, and Government’s increasing incentives to contribute to 401(k) plans have
all contributed to the growing popularity of de?ned-contribution plans[7].
The rest of the paper is organized as follows. The next section discusses the theory
and develops the hypotheses. Section 3 explains the research design and Section 4
describes the sample selection procedure. The results are presented in Sections 5 and 6
concludes the paper.
2. Theory and hypotheses
2.1 Various forms of deferred compensations offered by ?rms in the USA
Total employee compensation consists of wages and salaries, current fringe bene?ts,
and the promise of future bene?ts (Stone, 1982). Future bene?ts include pension and
retiree health plans, stock options, and other post-employment bene?ts. Pension and
retiree health plans are an important device to attract, retain, motivate, and eventually
to encourage retirement (Bodie and Papke, 1992; Gustman et al., 1994)[8]. In the USA,
pension plans are classi?ed into two types: de?ned-contribution and de?ned-bene?t. In
a de?ned-contribution plan, a formula determines contributions. The employee bears
all the investment risk, and the ?rm has no formal obligation beyond making its
periodic contributions. On the other hand, in a de?ned-bene?t plan, a formula that
usually takes into account years of service and wages or salary, de?nes bene?ts.
The investment risk is borne by the sponsoring ?rm or an insurance company hired by
the ?rm.
In addition to pensions, employers offer several postretirement health care and
welfare bene?ts to retirees, their spouses, dependents, and bene?ciaries. The welfare
bene?ts include life insurance, dental care, medical care, eye care, legal and tax
services, tuition assistance, day care, and housing assistance (Kieso et al., 2008).
2.2 Prior research
A ?rm’s pension and postretirement policy has four components: funding policy,
reporting policy, investment policy and bene?ts policy. Prior research has explored
?rm funding, reporting and investment policy. Research on funding policy shows that
?nancial slack (Feldstein and Morck, 1983 and Bodie and Papke, 1992); tax rates,
capital availability, debt-equity ratios, and bonding of employees (Francis and Reiter,
1987); corporate liabilities (Friedman, 1983); and ?nancial weakening and takeover
threats (Mittelstaedt, 1989) affect the funding decisions of ?rms. Thomas (1989) and
Ippolito and James (1992) investigate the causes of plan termination. Amir and Livnat
(1996) and Amir and Ziv (1997) examine the characteristics of early adopters of SFAS
106 (FASB, 1990). Mittelstaedt et al. (1995) study the determinants of decisions to cut
retiree health bene?ts.
Prior research also examined the ?rm’s strategic reporting choices of
de?ned-bene?t pension and retiree health bene?ts. Firms can in?uence their
reported de?ned-bene?t pension obligations by adjusting their actuarial choices. Bodie
et al. (1987), Thomas (1988), Ghicas (1990), Thomas and Tung (1992), Godwin et al.
(1995) and Asthana (1999) study the determinants of actuarial choices for
de?ned-bene?t pension plans. They show that pro?tability, tax liability, working
capital, debt, rate of undertaking of new investments, management of reimbursements
ARJ
21,2
126
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
by defense contractors, funded level, contribution level, excess cash from operations,
and income management incentives motivate managers to strategically change their
reported de?ned-bene?t pension obligations.
Research has also shown that ?rms have motivations to use actuarial assumptions
to change their reported retiree health obligations. D’Souza (1998) ?nds that
competitively weak electric utilities and those that did not curtail bene?ts choose
actuarial assumptions that lower the reported retiree health obligations. Amir and
Gordon (1996) show that ?rms with larger retiree health obligations and leverage select
aggressive parameter estimates, while ?rms with extreme reported earnings-price
ratios and ongoing plan amendments use conservative parameter estimates to report
higher obligations. Membership in regulated industries and the presence of
high-quality auditors curbs the tendency of managers to manipulate actuarial
assumptions under SFAS 106 (Asthana, 2001). Bodie et al. (1987) and Friedman (1983),
among others, have examined corporate investment policies. McGill and Grubbs (1989)
and Winklevoss (1993) discuss the various components of pension fund investment
policy in detail.
Thus, the main focus of the existing research has been to explore the funding,
reporting, and investment policies of pension and retiree health plans. The
determinants of the percentage deferred and the composition of bene?ts are
relatively unexplored[9]. This study extends the existing research by examining
the factors that in?uence the decision to defer compensation, the form that the deferral
will take, and the mix between de?ned bene?t pensions and de?ned bene?t retiree
health care to its employees.
2.3 Models and hypotheses
2.3.1 Percentage of deferred compensation to total compensation (MIX1)[10]. Under
standard economic theory, wages are bargained up/down to the point where employers
and workers are, at the margin, just satis?ed. This occurs when the marginal workers
earn the value of their marginal product. Once this marginal condition is established, a
?nancing contract is elicited: in what form will this bargained wage be paid? Will it be
all cash or will it be cash and some form of savings, including deferred compensation?
Ignoring government intervention for the moment, workers should be willing to save
compensation in excess of their current out of pocket needs in a number of instruments,
including deferral accounts with their employers. Of course, the ?rm must compete
with ?nancial institutions for these funds. Consequently, the competition for workers’
deferred savings will be based on risks and returns of the employer and in the ?nancial
markets. That is, the employer is, perhaps not unwittingly, a borrower[11]. Thus,
market place interest rates will be integral in determining how much will be saved
and where.
But in the USA, the government does intervene. It does so by offering tax incentives
designed to encourage private savings. These tax incentives permit workers to enjoy a
reduction in the income tax rate on their savings with the employer vis-a` -vis
conventional savings with ?nancial intermediaries. The expected effect of this
intervention is that workers will save more because of government subsidized after-tax
returns. Another thing is clear: the “where” to save decision is now affected. At the
margin, the relative net-of-tax returns are turned in favor of saving with the employers.
Factors affecting
corporate choices
127
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Beyond this, US tax rules, which I discuss below, provide tax-based subsidies for
employers as well[12].
Companies have a vested interest in the savings of their employees to the extent that
such savings stabilizes the workforce. Employee perceptions about the value of fringe
bene?ts, such as pensions, can stabilize workers by increasing search costs for
employment changes. A stable work force bene?ts companies that face signi?cant
turnover costs. Companies also can provide savings opportunities that reduce
investment search costs for employees. Hence, an environment for ?nancial
transactions naturally exists, and when added to the income tax bene?ts attached to
deferred compensation, the pervasiveness of deferrals is hardly surprising.
However, there are some impediments for ?rms that seek to utilize their employees
as a source of ?nance. First, the cost of providing deferred plans can be prohibitive to
small ?rms. Second, the reputation of the ?rm will have an effect on the willingness of
employees to contract with the employer over deferred compensation. Also, risk
sharing is a contractual problem. Who will bear the risk depends on risk tolerance and
the willingness to compensate for risk taking. These factors will be individually
important and collectively may vary across industries. Hence, I expect that industries
that are highly pro?table and in concentrated markets are likely to offer employment
contracts that are very different from low pro?t and highly competitive industries[13].
Following Thomas (1989) and Amir and Gordon (1996), eight industries are identi?ed.
The industry classi?cation is shown in detail in Table II. These arguments suggest the
following model and hypotheses:
Model 1 : MIX1 ¼ f{IMIX1; AGE; DEBTAR; LSIZE; Set of Control Variables}½14; 15?
My arguments suggest that industry characteristics will have an important and
positive effect on the deferred compensation level that individual ?rms make.
Formally, I expect:
H1. The percentage of deferred compensation to total compensation (MIX1) for
?rms in an industry is positively associated with the average percentage of
deferred compensation for that industry (IMIX1).
I expect that ?rms with a reputation for a stable work force are more likely to provide
deferred compensation for two reasons: employees will have greater trust in these
?rms, especially since default insurance for deferred plans is not perfect[16]. Second,
such ?rms are able and willing to incur the ?xed costs associated with deferred plans
when there are bene?ts to employment longevity. Also, an older work force will
normally have higher incomes, and savings rates are positively correlated with income
levels. Finally, the age pro?le of the participants is also expected to determine the
levels of the accumulated bene?t obligation (ABO) and the accumulated postretirement
bene?t obligation (APBO). Ceteris paribus, the older the average age of the ?rm’s
employees, the closer their retirement, and the greater the present value of the cash
bene?ts expected to be paid to the employees. Hence, I posit:
H2. MIX1 is positively associated with the average age of the employees,
AGE[17].
Firms with higher debt ratios are more likely to violate their accounting-based
debt-covenants (Press and Weintrop, 1990). Since creditors may include off-balance
ARJ
21,2
128
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
sheet liabilities in their calculations of debt ratios, the ?rm’s debt ratio may affect its
mix decisions[18]. Further, I expect that a ?rm’s ?nancial structure will negatively
impact the ?rm’s ability to engage in deferred compensation plans. A risky structure
(high debt-asset ratio, DEBTAR) should also make employees less willing to rely on
the ?rm as a reliable ?nancial institution. Thus, I expect:
H3. MIX1 is negatively associated with the ?nancial structure of the ?rm,
DEBTAR.
Firm size (proxied by natural logarithm of total assets at end of the ?scal-year, LSIZE),
is an important variable in this study for several reasons. First, the number of
employees in an organization will have an effect on the amount of the bene?t
obligation. Holding everything else constant, the larger the number of employees, the
larger the values of ABO and APBO. Furthermore, Scholes et al. (2002) report that
“larger ?rms are more likely to provide (retirement health care) coverage” and Graham
(1996) shows that ?rm size is positively related to ?rm tax rates. Firm size may also
capture the effects of my arguments on age. Larger ?rms are more likely to bene?t
from economies of scale with respect to the overhead costs of deferred compensation
plans. Consequently, I expect that:
H4. MIX1 is positively associated with the size of the ?rm, LSIZE.
2.3.2 Percentage of de?ned bene?ts to total deferred compensation (MIX2). De?ned
bene?t plans typically provide unique bene?ts to employers: they cause employees to
become unsecured bondholders (Dorsey, 1987). The bene?ts from this ?nancial
arrangement are three fold. First, as unsecured bondholders, employees’ interests are
aligned with the ?rm’s, which lowers costs (Dorsey). Second, underfunding de?ned
bene?t plans is a source of low-interest loans (Kruse, 1995). Third, employee turnover
is reduced for younger employees due to the cost of losing unvested future bene?ts.
Reduced turnover lowers the riskiness of human capital investments (Dorsey).
De?ned bene?t plans permit ?rms to force early retirement without violating age
discrimination by building into the plans economic disincentives to continue working.
Firms with de?ned bene?t plans have greater control over the measurement (and
smoothing) of income (Coronado and Sharpe, 2003). Excess market gains belong to the
corporation and can be used to lower future funding (and expense) requirements. On
the cost side, de?ned bene?t plans are more expensive to administer, especially for
small plans due to the presence of ?xed set up and maintenance charges. The ?rm is
also at risk for market losses. I expect that members of the same industry will
experience similar costs and bene?ts. These observations lead to my second model and
related set of hypotheses.
Model 2 : MIX2 ¼ f{IMIX2; DEBTAR; LSIZE; Set of Control Variables} ½19?
H5 has the same rationale as H1 and is not repeated here:
H5. The ratio of de?ned bene?t plans to total deferred compensation (MIX2) for
?rms in an industry is positively associated with the average for the industry
(IMIX2).
Next, I posit that unsecured bondholder employees are risk averse and will not tolerate
large amounts of employer debt. Hence, I expect that:
Factors affecting
corporate choices
129
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
H6. MIX2 is negatively associated with the ?rm’s ?nancial structure, DEBTAR.
Conversely, large ?rms are less risky than small (Fama and French, 1992). The
likelihood of ?rm survival will be important to employee-bondholders. If true, ?rm size
should impact the choice between de?ned bene?t and de?ned contribution plans in the
following way:
H7. MIX2 is positively associated with the size of the ?rm, LSIZE.
2.3.3 Percentage of retiree health care bene?ts to total de?ned bene?ts (MIX3). The IRC
regulates the tax treatment of private pension plans. De?ned-bene?t pension plans that
meet the requirements of the IRC and federal regulations, issued from time to time, are
said to be “quali?ed” for several tax bene?ts[20]. Employer contributions to quali?ed
plans are deductible, subject to a maximum tax-deductible limit, as ordinary and
necessary business expenses for federal income tax purposes[21]. These contributions
are not taxable to participants as income until paid out as bene?ts[22]. The earnings on
the pension plan assets, including realized capital gains, are not taxable for the
sponsoring ?rm[23]. Moreover, certain distributions of bene?ts from quali?ed plans
are entitled to favorable tax treatment[24]. Finally, under the tax rules, ?rms are
allowed to overfund their de?ned bene?t pension plans by up to 150 percent of their
obligations with tax bene?ts[25]. Thus, de?ned bene?t pension plans are attractive for
?rms due to the various tax bene?ts.
Taken together, these provisions can create large opportunities for tax deferrals
(Tepper, 1981). But to the extent that a ?rm claims a current deduction when
contributing to the plan, it cannot claim a deduction in the future when bene?ts are
paid to retirees. Consequently, the ?rm’s current taxes are reduced and its future taxes
increased. The exemption from tax of the earnings of the fund means that eventually
the accumulation of untaxed earnings will reduce future plan contributions and result
in taxable income. According to Tepper (1981), “deferrals are interest free loans from
the federal government and the general rule is that the gain to the ?rm is the after-tax
(income) that accumulates on the deferral.” The value of this gain to the ?rm will be a
function of its tax rate.
Funding of retiree health plans is not tax-advantaged in the same way as funding of
pension plans since any contributions to fund future retiree health bene?ts are
generally not tax-deductible. In general, under IRC § 419, employer contributions to
any welfare bene?t fund are deductible only to the extent of the bene?ts paid in the
current year[26]. Thus, over funding of retiree health bene?ts provides no tax bene?t
and hence no gains for ?rms. Moreover, there is no minimum funding requirement for
retiree health plans. As a result, the majority of retiree health care plans in the USA are
largely under funded. Owing to the differential tax bene?ts available to de?ned-bene?t
pension and retiree health plans, ?rms with higher tax rates are expected to prefer
de?ned-bene?t pension plans to retiree health plans. The third model and the last set of
hypotheses are derived largely from the above arguments.
Model 3 : MIX3 ¼ f {IMIX3; TAXRATE; CASHAR; TAXRATE*CASHAR; LSIZE;
Set of Control Variables} ½27?
The rationale for the industry variable (IMIX3) has the same justi?cation as do H1 and
H5. The reasoning is not repeated here.
ARJ
21,2
130
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
H8. The ratio of de?ned bene?t retiree health care plans to total de?ned bene?t
plans (MIX3) for ?rms in an industry is positively associated with the average
for that industry (IMIX3).
As argued above, I expect the tax rate to in?uence the mix between de?ned pension
bene?t plans and retiree health bene?t plans:
H9. MIX3 is negatively associated with the ?rm’s tax rate, TAXRATE.
Firms that have relatively higher cash ?ow from operations (proxied by operating cash
?ow to asset ratio, CASHAR) are expected to look for opportunities to pro?tably invest
their surplus capital. Ceteris paribus, investment of the surplus cash in pension plans
provides better returns in comparison to investment in retiree health plans due to the
gains that accrue from the tax deferral available from pension plan funding. Further,
storing surplus cash in pension funds creates ?nancial slack for ?rms. Firms can utilize
this slack in periods of ?nancial need by simply revising pension plan estimates which
reduce current funding requirements. For example, if poor ?nancial performance leads
to an increased cost of capital, ?rms can avoid the higher rates by utilizing their own
?nancial reserves[28]. This provides the following hypothesis:
H10. MIX3 is negatively associated with the ?rm’s operating cash ?ow, CASHAR.
The US taxes income rather than cash ?ow. Hence, it is not uncommon to observe
high-tax-rate-low-cash-?ow companies and low-tax-rate-high-cash-?ow companies.
Consequently, I seek to determine whether the possible interaction between tax rates
and cash ?ow has an effect on MIX3. Thus, I state H11, but without direction:
H11. MIX3 is associated with the interaction of tax rate and operating cash ?ow of
the ?rm.
I expect small ?rms will be more indifferent to the mix between health care bene?t
plans and de?ned bene?t pension plans than large ?rms due to economies of scale and
larger tax bene?ts of de?ned bene?t plans relative to health care bene?ts. Hence, I
expect:
H12. MIX3 is negatively associated with the size of the ?rm, LSIZE.
3. Research design
3.1 Control variables
I include the following variables as controls in each of my regressions: a trend variable
(TREND); the natural logarithm of total other compensation (LOCOMP); the discount
rate assumed for SFAS 87 by the ?rm minus the average for the industry (DDR87); the
projected increase in future employee compensation by the ?rm minus the average for
the industry (DANNINC); the discount rate assumed for SFAS 106 by the ?rm minus
the average for the industry (DDR106); and the short- and long-term health care cost
trend rates assumed for SFAS 106 by the ?rm minus the averages for the industry
(DSTHR and DLTHR).
I include TREND to capture possible changes in retirement bene?ts over time. Other
compensation (LOCOMP) is included as a proxy for the economic well being and
independence of employees[29]. DDR87, DANNINC, DDR106, DSTHR, and DLTHR are
Factors affecting
corporate choices
131
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
included to remove the effects of actuarial assumptions under SFAS 87 and 106 from
the reported obligations[30]. All the variables used in this study are de?ned in Table I.
3.2 Simultaneous estimation approach
MIX1, MIX2, and MIX3 are jointly determined and interdependent, resulting in
endogeneity of these variables[31]. Since endogenous variables are not independent,
the error term is likely to be correlated with the explanatory variables and OLS
assumptions are violated. As a result, the regression coef?cients are inconsistent
(Maddala, 1988). To avoid this problem of misspeci?cation, Maddala’s (1988)
simultaneous estimation method is used for the estimation. Two-stage least square
technique is utilized. In the ?rst-stage, the endogenous variables, MIX1, MIX2, and
Variable De?nition (in alphabetical order)
ABO ABOs reported under SFAS No. 87
AGE Proxy for average age of employees; estimated as pension plan interest cost de?ated by
pension plan service cost (Brown, 2004)
APBO APBOs reported under SFAS No.106
ANNINC Projected annual increase in future compensation
CASHAR Cash ?ow from operations (adjusted for the effects of SFAS 87 and 106) to asset ratio
DEBTAR Total debt (adjusted for the effects of SFAS 87 and 106) to asset ratio
D When pre?xed to a variable implies the value of the variable minus its industry mean
DCPO De?ned contribution pension obligations; estimated as the present value of future
de?ned contribution pension plan payments, using annual increases of ANNINC
and discount rate of DR87 ¼ Current year’s funding to de?ned contribution pension
plan divided by (DR87-ANNINC), assuming the cash out?ow continues in perpetuity
DR87 Discount rate assumed under SFAS 87
DR106 Discount rate assumed under SFAS 106
I When pre?xed to a variable implies its industry mean
LOCOMP Natural logarithm of other compensation expenses (excluding pension and
postretirement health care bene?t expenses)
LSIZE Natural logarithm of size (total assets at the end of ?scal year)
LTHR Long-term health care cost trend rate assumed under SFAS 106
MIX1 Mix of deferred and current compensation obligations. It is equal to deferred pension and
postretirement health care obligations (ABO þ DCPO þ APBO) to total compensation
obligations (OCOMPO þ ABO þ DCPO þ APBO)
MIX2 Mix of de?ned bene?t and de?ned contribution bene?ts obligations. It is equal to (ABO
þ APBO) de?ated by (ABO þ APBO þ DCPO)
MIX3 Mix of retiree health bene?ts obligations and de?ned pension bene?t plans. It is equal to
APBO de?ated by (ABO þ APBO)
OCOMPO Other compensation obligations; estimated as the present value of future labor expenses,
other than pension and postretirement health care bene?t expenses, using annual
increases of ANNINC and discount rate of DR87 ¼ Current year’s compensation
expenses (excluding pension and postretirement health care bene?t expenses) divided by
(DR87-ANNINC), assuming the cash out?ow continues in perpetuity
STHR Short-term health care cost trend rate assumed under SFAS 106
TAXRATE Average tax rate; estimated as the federal income tax payable (net of investment tax
credit) de?ated by income before taxes. The rate is winsorized in the range (0, maximum
corporate tax rate)
TREND Trend variable proxied by the year
Table I.
Variable de?nitions
ARJ
21,2
132
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
MIX3 are estimated using only the exogenous variables in the system. This is done by
replacing MIX1, MIX2, and MIX3 with IMIX1, IMIX2, and IMIX3 as independent
variables in Regressions 1-3. Then the predicted values of MIX1, MIX2, and MIX3 from
these regressions are used as independent variables in Regressions 1-3 in the
second-stage. Maddala shows that the regression coef?cients obtained from the
two-stage least square estimation are unbiased and consistent.
4. Sample selection
A ?rm that satis?es the following criteria is included in the sample:
.
data are available on the COMPUSTAT and COMPACT-DISCLOSURE
databases for at least one year during the period 1993-2001; and
.
the ?rm has a December 31 ?scal year-end.
Some of the data in this study, namely the information on ?rms’ retiree health care
plans and de?ned contribution pension plans, were hand-collected from ?nancial
statement footnotes.
Only the ?rms with December 31 ?scal year-ends are included in the sample to align
the observations. The alignment helps in debiasing the effects of actuarial choices more
ef?ciently (Asthana, 1999, 2001). Firms were required to adopt SFAS No. 106 no later
than the ?scal years beginning after December 15, 1992. The sample period begins in
1993, since this was the ?rst ?scal year when ?rms disclosed their retiree health
obligations (there were few early adopters). The ?nal sample consists of 2,458
observations for 646 ?rms during the period 1993-2001.
5. Empirical results
5.1 Sample characteristics
Panel A of Table II presents the distribution of variables. The means for my decision
variables are: 3.9 percent (MIX1), 78.7 percent (MIX2) and 46.5 percent (MIX3). The
mean (median) TA of the sample is $10.896 ($2.072) billion. The requirement that data
for the ?rm be available on COMPUSTAT and COMPACT-DISCLOSURE may have
biased the sample in favor of large ?rms. The means (in millions) for ABO, APBO, and
DCPO are $715, $381 and $462.Thus, the average ?rm’s pension liabilities are almost
twice large as retiree health liabilities. The means for CASHAR, DEBTAR and
TAXRATE are 0.120, 0.614 and 0.299[32]. The means for the interest rates for SFAS 87
and 106 are 7.5 percent each, and the mean for the expected annual increase in
compensation is 4.8 percent. The expected short- and long-term health care cost trend
rates are 9.3 and 5.4 percent, respectively. Panel B provides the number of observations
by year. Compact Disclosure is used to hand-collect data pertaining to retiree health
plans and de?ned-contribution pension plans (not available on Compustat). This
database is only available at my institution through 2001 and the coverage of ?rms
during 2000-2001 is limited to the observations reported.
Table III reports means for MIX1, MIX2 and MIX3 across eight industries using
classi?cations similar to Thomas (1989) and Amir and Gordon (1996). The null
hypotheses of equal means for all three variables across the eight industries are
rejected at the one-percent signi?cance level. Thus, there are signi?cant variations in
the decision variables. Utilities provide the largest percentage of deferred bene?ts
(4.9 percent), and Transportation provides the lowest percentage (2.1). The construction
Factors affecting
corporate choices
133
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
industry has the highest percentage of de?ned bene?ts to total bene?ts (89.2 percent).
Consumer goods has the lowest (76.2 percent). Retiree health bene?ts as a percentage of
total de?ned bene?ts is largest for the Energy industry (56.4 percent). Construction is
lowest (42.0 percent). These patterns are quite varied across the industries and suggest
that industry membership is likely to have substantial effects on the forms of
compensation. These differences could be due to the differences in ?rms’ choices of
postretirement bene?ts in these industries, or due to the differences in the number of
employees and their age distributions.
Table IV presents the Pearson correlation coef?cients. The correlations between the
three MIX variables are signi?cant, suggesting that the use of Simultaneous Estimation
was justi?ed. Consistent with this result, the three IMIX are also signi?cantly
correlated. Next, let us look at the correlations between the independent variables
in Regressions 1-3.These coef?cients range in values from20.18 to þ0.24.Thus, none of
Panel A: distribution of variables
Variables Mean Median SD
Market value (in mill. $) 6,444.950 1,477.800 17,243.490
Book value (in mill. $) 1,999.220 681.094 4,258.220
Total assets (in mill. $) 10,895.660 2,072.020 39,729.540
No. of employees (in 000) 20.608 6.700 46.132
ABO (in mill. $) 714.580 59.303 3,598.950
APBO (in mill. $) 381.366 40.665 2,658.440
DCPO (in mill. $) 462.100 0.000 3,940.970
OCOMPO (in mill. $) 1,021.800 321.500 2,131.000
MIX1 0.039 0.021 0.107
MIX2 0.787 1.000 0.329
MIX3 0.465 0.299 0.370
AGE 2.435 2.112 2.703
CASHAR 0.120 0.098 0.286
DEBTAR 0.614 0.611 0.169
TAXRATE 0.299 0.329 0.095
LOCOMP 5.794 5.774 1.611
DR87 (percent) 7.499 7.500 0.585
ANNINC (percent) 4.832 5.000 0.624
DR106 (percent) 7.507 7.500 0.573
STHR (percent) 9.327 9.800 1.976
LTHR (percent) 5.371 5.216 0.631
Panel B: sample distribution across years
Year Observations
1993 281
1994 495
1995 635
1996 204
1997 244
1998 247
1999 200
2000 75
2001 77
Total 2,458
Notes: See Table I for variable de?nitions. Total number of sample ?rms is 646
Table II.
Sample characteristics
ARJ
21,2
134
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the explanatory variables are correlated high enough to cause multicollinearity
problems. To further rule out this possibility, variance in?ation factors (VIF) are also
calculated and reported later in the paper.
5.2 Analysis of trends in retirement bene?ts
In Table V, I report the year by year movement in the average values of the three ratios,
MIX1, MIX2, and MIX3.What I seek to know is: are the three ratios changing, and if so,
are the changes signi?cant and in what direction. The way I construct my tests is to
compute the ratios for each ?rm for all sample years and then compare the current year
to the prior year[33]. For example, if I had two companies with data for all nine years,
I would have 16 observations. If I had one company with nine years and another
company with six, the sample size would be 13.The results show a signi?cant decline
in ABO and signi?cant increases in APBO and DCPO across time. This is consistent
with de?ned-contribution pension plans becoming more popular with employers than
de?ned-bene?t pension plans. Deferred compensation as a percentage of total
compensation seems to be declining. De?ned bene?t plans (pension and retiree health
care) are becoming a smaller percentage of total deferred compensations, a trend
consistent with de?ned-contributions replacing de?ned-bene?t plans. Finally, APBO is
growing at a faster rate than ABO. Increase in the projected medical costs is the likely
reason.
Industry Observations Mean MIX1 Mean MIX2 Mean MIX3
1. Basic industries 350 0.0369 0.8069 0.4343
2. Capital goods 426 0.0370 0.7542 0.4331
3. Construction 30 0.0434 0.8917 0.4202
4. Consumer goods 695 0.0422 0.7616 0.4865
5. Energy 131 0.0438 0.7112 0.5636
6. Finance 341 0.0307 0.8128 0.4321
7. Transportation 131 0.0214 0.7783 0.4766
8. Utilities 354 0.0492 0.8557 0.4830
Total observations 2,458
F-value (H
0
: equal across industries) 44.54 5.34 3.05
Prob . F 0.0001 0.0001 0.0034
See Table I for variable de?nitions. The classi?cation into industry groups is based on Thomas (1989) as
below:
Industry group SIC codes
1. Basic industries 1000-1299, 1400-1499, 2600-2699, 2800-2829, 2870-2899, and 3300-3399
2. Capital goods 3400-3419, 3440-3599, 3670-3699, 3800-3849, 5080-5089, 5100-5129,
and 7300-7399
3. Construction 1500-1599, 2400-2499, 3220-3299, 3430-3439, and 5160-5219
4. Consumer goods 0000-0999, 2000-2399, 2500-2599, 2700-2799, 2830-2869, 3000-3219,
3420-3429, 3600-3669, 3700-3719, 3850-3879, 3880-3999, 4830-4899,
5000-5079, 5090-5099, 5130-5159, 5220-5999, 7000-7299, and 7400-9999
5. Energy 1300-1399 and 2900-2999
6. Finance 6000-6999
7. Transportation 3720-3799 and 4000-4799
8. Utilities 4800-4829 and 4900-4999
Table III.
Distribution of pension
and postretirment health
care bene?ts across
industries
Factors affecting
corporate choices
135
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
V
a
r
i
a
b
l
e
s
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)
(
7
)
(
8
)
(
9
)
(
1
0
)
(
1
1
)
(
1
2
)
(
1
3
)
(
1
4
)
(
1
5
)
(
1
6
)
(
1
7
)
M
I
X
1
(
1
)
–
m
i
x
2
(
2
)
2
0
.
1
3
m
i
x
3
(
3
)
0
.
1
1
2
0
.
3
9
i
m
i
x
1
(
4
)
0
.
0
6
0
.
0
2
0
.
0
5
i
m
i
x
2
(
5
)
0
.
0
1
0
.
1
2
2
0
.
0
3
0
.
1
7
i
m
i
x
3
(
6
)
0
.
0
3
2
0
.
0
4
0
.
0
9
0
.
4
9
2
0
.
3
3
a
g
e
(
7
)
0
.
0
0
0
.
0
3
2
0
.
0
1
0
.
0
4
2
0
.
0
3
0
.
0
1
D
E
B
T
A
R
(
8
)
2
0
.
0
2
2
0
.
0
5
0
.
0
4
2
0
.
1
8
0
.
1
5
2
0
.
0
9
2
0
.
0
3
T
R
E
N
D
(
9
)
0
.
0
2
2
0
.
4
0
0
.
6
3
0
.
0
4
2
0
.
0
6
0
.
0
8
0
.
0
5
0
.
0
6
l
s
i
z
e
(
1
0
)
2
0
.
0
9
2
0
.
0
1
2
0
.
0
2
2
0
.
1
4
0
.
1
3
2
0
.
0
1
0
.
0
0
0
.
0
6
0
.
1
1
l
O
c
o
m
p
(
1
1
)
2
0
.
1
9
2
0
.
0
2
2
0
.
0
6
2
0
.
1
7
2
0
.
1
3
2
0
.
0
6
0
.
0
4
0
.
1
7
0
.
0
7
0
.
2
4
c
a
s
h
A
R
(
1
2
)
0
.
5
0
2
0
.
1
5
0
.
0
7
0
.
0
1
2
0
.
0
8
0
.
0
1
2
0
.
0
1
2
0
.
0
3
0
.
0
9
2
0
.
0
8
2
0
.
0
3
T
A
X
R
A
T
E
(
1
3
)
2
0
.
0
4
2
0
.
0
4
2
0
.
0
6
0
.
0
9
0
.
0
5
0
.
0
7
2
0
.
0
7
2
0
.
0
3
0
.
0
6
0
.
1
0
0
.
1
3
0
.
0
6
D
D
R
8
7
(
1
4
)
0
.
0
1
0
.
0
5
2
0
.
0
3
2
0
.
0
3
2
0
.
0
4
2
0
.
0
3
0
.
0
3
0
.
0
0
2
0
.
0
1
0
.
0
4
0
.
0
7
2
0
.
0
1
0
.
0
2
D
A
N
N
I
N
C
(
1
5
)
2
0
.
0
4
2
0
.
0
5
2
0
.
0
1
0
.
0
4
0
.
0
9
2
0
.
0
1
2
0
.
0
1
2
0
.
0
1
0
.
0
4
0
.
0
0
0
.
0
1
0
.
0
5
0
.
0
6
0
.
0
7
D
D
R
1
0
6
(
1
6
)
2
0
.
0
2
0
.
0
3
2
0
.
0
1
2
0
.
0
2
0
.
0
0
2
0
.
0
4
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
4
0
.
0
4
2
0
.
0
2
0
.
0
4
0
.
2
7
0
.
0
5
D
S
T
H
R
(
1
7
)
2
0
.
0
2
2
0
.
0
1
2
0
.
0
4
0
.
0
8
0
.
0
1
2
0
.
0
1
2
0
.
0
1
2
0
.
0
4
2
0
.
0
3
2
0
.
0
9
2
0
.
0
7
2
0
.
0
3
0
.
0
2
2
0
.
0
5
0
.
0
7
2
0
.
0
2
D
L
T
H
R
(
1
8
)
0
.
0
1
2
0
.
0
7
0
.
0
1
2
0
.
0
3
0
.
0
1
0
.
0
8
2
0
.
0
1
0
.
0
5
0
.
0
4
0
.
0
2
0
.
0
0
0
.
0
6
0
.
0
5
0
.
0
7
0
.
0
3
0
.
1
0
0
.
2
3
N
o
t
e
s
:
S
e
e
T
a
b
l
e
I
f
o
r
v
a
r
i
a
b
l
e
d
e
?
n
i
t
i
o
n
s
.
I
t
a
l
i
c
n
u
m
b
e
r
s
i
m
p
l
y
s
i
g
n
i
?
c
a
n
c
e
a
t
1
0
p
e
r
c
e
n
t
l
e
v
e
l
o
r
b
e
t
t
e
r
Table IV.
Pearson correlation
matrix
ARJ
21,2
136
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
5.3 Results of tests of hypotheses
I report the regression results for MIX1 in Table VI. The adjusted R
2
for Model 1 is
23.6 percent. The F-value for the model is reliably signi?cant at the 1 percent level. The
highest VIF is below ten, which suggests that multicollinearity is not a problem[34].
Variable Current year Previous year Current – previous
Panel A: analysis of bene?ts per employee
ABO 23.6868 (15.0604) 27.1828 (20.1993)
* * *
2 3.4960 (25.1370)
APBO 14.2895 (7.7010) 13.8129 (7.3800)
*
0.4766 (0.3210)
DCPO 15.4614 (0.0000) 14.3763 (0.0000)
* *
1.0851 (0.0000)
Mean values in $ 000 per employee (median values in parentheses) (N ¼ 1,812)
Panel B: analysis of bene?t ratios
MIX1 0.0378 (0.0210) 0.0391 (0.0229)
* * *
2 0.0013 (20.0019)
MIX2 0.7541 (1.0000) 0.7866 (1.0000)
* * *
2 0.0325 (0.0000)
MIX3 0.5011 (0.3133) 0.4078 (0.2711)
* * *
0.0933 (0.0422)
Mean values (median values in parentheses) (N ¼ 1,812)
Notes: See Table I for variable de?nitions. The sample size is reduced due to differencing. As a result,
the mean values for the current year in Panel B are different from those in Panel A of Table II.
Signi?cant at the levels of
*
10,
* *
5 and
* * *
1 percent, respectively
Table V.
Trends in pension and
postretirement health
care bene?ts
Variable Hypotheses Estimate White’s t p-value
Intercept
* * *
3.7615 8.12 ,0.0001
IMIX1 þ
* * *
0.6844 7.18 ,0.0001
AGE þ
* * *
0.0010 5.5 ,0.0001
DEBTAR –
* * *
2 0.0264 27.39 ,0.0001
LSIZE þ
* * *
0.0086 15.36 ,0.0001
MIX2
* *
0.0417 2.53 0.0116
MIX3
* * *
0.1017 5.1 ,0.0001
TREND
* * *
2 0.0019 28.32 ,0.0001
LOCOMP
* * *
2 0.0075 212.8 ,0.0001
DDR87
* * *
2 0.0061 27.84 ,0.0001
DANNINC 20.0011 20.55 0.5842
DDR106
* * *
2 0.0010 23 0.0027
DSTHR 20.0005 20.59 0.5575
DLTHR
* * *
0.0072 3.71 0.0002
Observations 2,458
F-value 58.27
Prob . F 0.0001
Adj R
2
23.58 percent
White’s x
2
231.68
Prob. .x
2
0.0001
Highest VIF 4.3001
Notes: Regression 1: Mix of Deferred Post-Retirement Bene?ts and Current Compensation. MIX1 ¼
a
0
þ a
1
IMIX1 þ a
2
AGE þ a
3
DEBTARþ a
4
LSIZE þ a
5
MIX2 þ a
6
MIX3 þa
7
TREND þ a
8
LOCOMP þ a
9
DDR87 þ a
10
DANNINC þ a
11
DDR106 þ a
12
DSTHR þ a
13
DLTHR þ 1.
See Table I for variable de?nitions. Signi?cant at the levels of
*
10,
* *
5, and
* * *
1 percent, respectively
Table VI.
Regression of total
deferred compensation on
?rm and industry
characteristics
Factors affecting
corporate choices
137
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
All four null hypotheses are rejected since the estimated regression coef?cients are
statistically signi?cant and in the expected direction. I report White’s t-statistics since
there is evidence of heteroskedasticity in the data. I conclude that industry norm,
employee age, ?nancial structure of the ?rm, and ?rm size are highly in?uential in
determining how much of total compensation will be deferred by employees and
employers. I also take into account the control variables, including the joint effects of
MIX2 and MIX3 on MIX1.The effects of the control variables are signi?cant, excluding
the effects of DANNINC and DSTHR.
The results for MIX2, the percentage of de?ned bene?ts to the total package of
deferred compensation, are similar to the results for MIX1 (Table VII). All three test
variables are signi?cant at the 5 percent level or less. Again, I conclude that industry
membership, ?nancial structure, and ?rm size have a signi?cant impact on the MIX2
decision. The coef?cients of MIX1 and MIX3 are not signi?cant, which suggests that
the impact of the joint decisions is not important for MIX2.Of the control variables,
only DNNINC and DLTHR’s impact is signi?cant. The explanatory power of Model 2
(24.1 percent) is very close to model one’s power. The highest VIF is 4.4, which is
consistent with low levels of multicollinearity. White’s x
2
test indicates that
heteroskedasticity exists in the data. Consequently, I report White’s t-statistics for
signi?cance testing.
The last model estimates the signi?cance of the impact of ?ve test variables
(Table VIII) on MIX3.I ?nd that the percentage of retiree health care bene?ts to total
de?ned bene?ts (MIX3) follows most of the expectations. Again, industry norm is a
Variable Hypotheses Estimate White’s t p-value
Intercept
* * *
128.2724 24.52 ,0.0001
IMIX2 þ
* * *
0.9953 5.35 ,0.0001
DEBTAR –
* * *
2 0.2102 25.25 ,0.0001
LSIZE þ
* *
0.0133 2.11 0.0353
MIX1 21.4253 21.33 0.1833
MIX3 0.2727 1.21 0.2264
TREND
* * *
2 0.0643 224.56 ,0.0001
LOCOMP 20.0095 21.45 0.1472
DDR87 0.0134 0.59 0.5545
DANNINC
*
2 0.0154 21.78 0.0746
DDR106 0.0317 1.39 0.1656
DSTHR 0.0037 1.01 0.3125
DLTHR
* * *
2 0.0417 24.75 ,0.0001
Observations 2,458
F-value 63.27
Prob. .F 0.0001
Adj R
2
24.11 percent
White’s x
2
537.44
Prob .x
2
0.0001
Highest VIF 4.4379
Notes: See Table I for variable de?nitions. Signi?cant at the levels of
*
10,
* *
5 and
* * *
1 percent,
respectively. Regression 2: mix of de?ned bene?t and de?ned contribution bene?ts. MIX2 ¼ b
0
þb
1
IMIX2 þ b
2
DEBTAR þ b
3
LSIZE þ b
4
MIX1 þ b
5
MIX3 þ b
6
TREND þ b
7
LOCOMP þ
b
8
DDR87 þ b
9
ANNINC þ b
10
DDR106 þ b
11
DSTHR þ b
12
DLTHR þ 1
Table VII.
Regression of de?ned
bene?ts to total deferred
compensation on ?rm
and industry
characteristics
ARJ
21,2
138
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
signi?cant and positive factor. Corporate tax rates are reliably negative on the mix of
de?ned bene?ts. Cash ?ow and the interaction between tax rates and cash ?ow are
marginally signi?cant. Cash ?ow is negative as expected. The interaction between tax
rate and cash ?ow is positive, but I had no expectation. Firm size is the only test
variable that does not have a signi?cant impact on the mix. Again, the other mix
variables (MIX1 and MIX2) do not have signi?cant effects on the joint decisions.
Five of the control variables have signi?cant effects. TREND, DDR87 and DDR106
are signi?cant at the one percent level. LOCOMP and DLTHR are signi?cant at the ?ve
percent level. The effects of DANNINC and DSTHR are insigni?cant. The explanatory
power of Model 3 (64.5 percent) is much larger than the ?rst two models. The highest
VIF is 4.2 and is consistent with low levels of multicollinearity. I report White’s
t-statistics for signi?cance testing since the White’s (1980) test indicates that
heteroskedasticity exists in the data.
5.4 Sensitivity analyses
A potential model speci?cation problem in Regression 2 needs to be examined in more
detail. The potential misspeci?cation arises because more than half the sample does
not have de?ned-contribution pension plans (DCPO ¼ 0). As a result, the value of
MIX2 would be 1 for majority of ?rm-year observations. This might pose a problem in
Variable Hypotheses Estimate White’s t p-value
Intercept
* * *
2 260.3214 260.03 ,.0001
IMIX3 þ
* *
0.3791 1.99 0.0471
TAXRATE –
* * *
2 0.3287 25.71 ,.0001
CASHAR –
*
2 0.2064 21.66 0.0919
CASHAR
*
TAXRATE ?
*
0.6835 1.83 0.068
LSIZE – 0.0026 0.48 0.6334
MIX1 0.191 0.21 0.8326
MIX2 0.1651 1.07 0.2868
TREND
* * *
0.1306 60.25 ,.0001
LOCOMP
* *
2 0.0146 22.45 0.0143
DDR87
* * *
0.0670 3.59 0.0003
DANNINC 20.0037 20.5 0.6154
DDR106
* * *
0.0646 3.41 0.0007
DSTHR 20.0029 20.94 0.349
DLTHR
* *
2 0.0181 22.38 0.0174
Observations 2,458
F Value 276.12
Prob .F 0.0001
Adj R-Square 64.53 percent
White’s x
2
335.12
Prob .x
2
0.0001
Highest VIF 4.1545
Notes: Regression 3: mix of retiree health care bene?ts and de?ned-bene?t pension. MIX3 ¼ g
0
þ g
1
IMIX3 þ g
2
TAXRATE þ g
3
CASHAR þ g
4
TAXRATE
*
CASHAR þ g
5
LSIZE þg
6
MIX1 þ
g
7
MIX2 þ þg
8
TREND þ g
9
LOCOMP þ g
10
DDR87 þ g
11
DANNINC þg
12
DDR106 þ
g
13
DSTHR þg
14
DLTHR þ 1See Table I for variable de?nitions. Signi?cant at the levels of
*
10,
* *
5 and
* * *
1 percent, respectively
Table VIII.
Regression of retiree
healthcare to total de?ned
bene?ts on ?rm and
industry characteristics
Factors affecting
corporate choices
139
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Regression 2 (Table VII), since OLS estimation with a binary (or binary-like) dependent
variable could lead to estimation errors. To rule out this possibility, I ?rst estimate
regression 2 with only ?rms that offer de?ned-contribution plans (i.e. MIX2 – 1). Next,
I use Tobit estimation to account for the limited dependent variable. Table IX reports
the additional sensitivity analyses. Both the truncated and Tobit regressions yield
results very similar to those in Table VII. Thus, the reported results are robust.
6. Conclusion
This paper seeks to extend post retirement bene?t research beyond the questions of
funding and actuarial assumption choices by examining the determinants of corporate
deferred compensation policies on:
.
the proportion of total compensation deferred;
.
the proportion in the form of de?ned bene?ts; and
.
the proportion for retiree health care bene?ts.
Retirement bene?ts are essential for attracting potential employees and retaining them.
Sizable retirement bene?ts also motivate employees to retire eventually, making way
for new employees.
This research indicates that industry norm is important to all three decisions. I also
?nd that there is considerable variation in the levels of deferred compensation and the
forms of deferred bene?ts across industries. Firms appear to be constrained by their
competitors’ employment contracts. Firm size is an important determinate of total
Variable Hypotheses Truncated regression Tobit regression
Intercept
* * *
161.0790 –
IMIX2 þ
* * *
1.1355
* * *
8.3387
DEBTAR –
* * *
2 0.2758
* * *
2 1.5537
LSIZE þ
*
0.0720
* * *
0.1558
MIX1 1.4509 212.6284
MIX3 20.2588
*
3.2930
TREND
* * *
2 0.0808
* * *
2 0.4120
LOCOMP 0.0027
* * *
2 0.1566
DDR87 0.0564 20.0094
DANNINC
* * *
2 0.0541
* * *
2 0.1995
DDR106 0.0109 0.2525
DSTHR 20.0003 0.0047
DLTHR
*
2 0.0215
* * *
2 0.2677
Observations 920 2,458
F-value 66.2
Prob. .F ,0.0001
Adj R
2
0.4607
Walds x
2
459.5997
Prob .x
2
,0.0001
Notes: Regression 2: mix of De?ned Bene?t and De?ned Contribution Bene?ts. MIX2 ¼ b
0
þ b
1
IMIX2 þ b
2
DEBTAR þ b
3
LSIZE þb
4
MIX1 þ b
5
MIX3 þb
6
TREND þb
7
LOCOMP þ
b
8
DDR87 þ b
9
DANNINC þb
10
DDR106 þb
11
DSTHR þ b
12
DLTHR þ 1. See Table I for
variable de?nitions. Signi?cant at the levels of
*
10,
* *
5 and
* * *
1 percent, respectively
Table IX.
Sensitivity analyses for
Regression 2
ARJ
21,2
140
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
deferred compensation and the level of de?ned bene?ts relative to de?ned
contributions. But, it does not appear to have much in?uence in determining retiree
health care bene?ts.
Corporate ?nancial structure plays a signi?cant role in the level of total deferred
compensation and in the relative level of de?ned bene?ts. I expected this result since
deferring bene?ts effectively, especially de?ned bene?ts, creates a lender-debtor
relationship between workers and their employer. The age of the average worker is
important to the total amount of deferred compensation. Retiree healthcare bene?ts
(relative to total de?ned bene?ts) are negatively (as expected) affected by corporate tax
rates. Operating cash ?ows and the interaction between cash ?ows and tax rates are
both marginally signi?cant.
This research corroborates and extends prior literature by identifying the
determinants of corporate postretirement bene?t policies. These ?ndings have
implications for the baby boomers that have started to retire en masse. The trend to
offer smaller proportion of compensation as deferred bene?ts re?ects the increasing
costs of deferral to the employers. These costs arise due to changing US pension laws
and accounting rules. This increases the employees’ responsibilities to save on their
own and plan differently for their retirement. This also would shift the retirees’
dependence on the public pension system for their retirement income. The trend to
favor de?ned-contribution plans instead of de?ned-bene?t plans re?ects the
employers’ attempts to diversify their risks of paying promised post-retirement
bene?ts by transferring the risk to the employee. On the other hand, the popularity of
de?ned-contribution pension plans also re?ects the increased Government’s incentives
to encourage savings via 401-k plans, employee’s willingness to manage their own
pension portfolios in a favorable stock market with more advanced technology
available for self-managing their pension investments, and increased access to
diversi?ed stock and bond funds[35]. The trends in post-retirement bene?ts reported in
this paper are also important for policy makers in forging future pension and
health-care policies that secures post-retirement bene?ts for employees.
Notes
1. This concern is also re?ected in the speeches, debates, and interviews of various candidates
during the Presidential election campaigns of 2004 and 2008.
2. SFAS Nos. 87 and 106.
3. SFAS 106 was approved in 1990 and became effective for ?scal years after December 15,
1992.
4. In September 2006, FASB issued a standard (SFAS 158) that brought the pension fund crisis
back into wide-spread public awareness. SFAS 158 is an amendment to SFAS statements 87,
88, 106, and 132R and went into effect for ?scal years ending after December 2006. This
ruling requires employers to fully recognize the obligations associated with single-employer
de?ned bene?t pensions, retiree healthcare, and other postretirement plans in their ?nancial
statements (FASB, 2006). The impact of this new standard on corporate decisions is yet to be
seen.
5. Kruse ?nds that industry variability, company size and capital intensity are important
factors in the choice of plans. He also ?nds that administration costs and unionization also
play a factor. We use ?rm size as a proxy for these variables.
Factors affecting
corporate choices
141
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
6. The changes in the bene?t-mix documented in this paper have continued in the recent years
beyond the sample period (PBGC, 2007).
7. The Pension Protection Act of 2006 has signi?cantly expanded the amount of contribution to
a 401(k) plan that can be made in addition to the contribution to the de?ned-bene?t plan
(Pension Trends, 2007).
8. Gustman et al. (1994), report that de?ned bene?t plan formula often contain pension accrual
spikes that are early on designed to prevent quitting and later on are designed to encourage
early retirement. The later are negative spikes that actually reduced the pension payment.
A bene?t of this pattern is that ?rms may, without violating age discrimination, rehire
(typically as consultants) the most quali?ed of the retirees.
9. Notable exceptions are Dorsey (1987) and Kruse (1995, 1996). Dorsey’s study provides
arguments about why ?rms might choose de?ned bene?t plans over de?ned contribution
plans, including control over employee quits and alignment of employee interests with the
?rm. Kruse explains why de?ned contribution plans have gained in popularity.
10. The levels of de?ned contribution pension bene?ts offered may also affect the choices of
de?ned bene?t pension and retiree health bene?ts. Under the provisions of SFAS No. 87,
de?ned contribution plans report only their annual contributions but not the liabilities
(FASB, 1985). In the rare situation when the contribution is more (less) than the required
level, the ?rm reports an asset (liability) on its balance sheet. COMPUSTAT does not report
the contributions to these plans separately. Therefore, the annual contribution to the fund
(obtained from footnote disclosures in COMPACT-DISCLOSURE database) is used to
calculate a proxy for the level of de?ned contribution bene?ts (DCPO).
11. Dorsey (1987), among others, points out that in de?ned bene?t plans the employees actually
are unsecured bondholders. There is also growing evidence that the borrowing can be
perniciously used by employers to the harm of the employees. Corporations are currently
engaged in charges that employees received reduced returns to the bene?t of the employer.
12. US Internal Revenue Code (of 1986).
13. Brown and Medoff (2003) summarize research ?ndings about wages and ?rms
characteristics. They ?nd that compensation contracts are related to ?rm size.
14. See Table I for variable de?nitions. Table VI presents the empirical model. The test variables
in Models 1-3 are highly correlated across time with Pearson coef?cients ranging from 60 to
99.3 per cent. When I rerun the regressions with average values of these variables over the
past ?ve years instead of the current values (results not reported), the conclusions do not
change.
15. I discuss the control variables in the results later.
16. This trust is not always well placed. The pension default protections under ERISA are
substantially less than dollar for dollar.
17. My proxy for employee average age, AGE, is the pension plan interest cost de?ated by
pension plan service cost (Brown, 2004).
18. DEBTAR is adjusted for the effects of SFAS Nos. 87 and 106 by subtracting the accrued
pension cost and accrued postretirement bene?t cost from the total debt and subtracting the
prepaid pension cost and prepaid postretirement bene?t cost from the ?rm’s assets.
19. See Table I for variable de?nitions. Table VII presents the empirical model.
20. IRC § 401(a).
21. IRC § 404.
22. Treasury Regulations 1.402(a)-1(a).
23. IRC § 501(a).
ARJ
21,2
142
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
24. IRC § 402(e)(4)(A) and IRC § 402(a)(2). For a more detailed discussion of the tax bene?ts
available to quali?ed pension plans and its participants, see McGill and Grubbs (1989).
25. Under provisions of the Pension Protection Act which subsequently became part of the
Omnibus Budget Reconciliation Act of 1987.
26. Scholes et al. (2002, p. 227) discuss exceptions, such as 401(h) and Voluntary Employee
Bene?t Association (VEBA) plans, where certain tax deductions are available. However,
these plans are expensive to administer. This can make unfunded plans more desirable than
401(h) plans even when the latter are available. Moreover, the annual tax-deductible
contributions to such plans cannot exceed 25 percent of the pension contributions. As a
result, when the ?rm’s pension plan is over funded, contributions to 401(h) and VEBA plans
are not permitted at all.
27. See Table I for variable de?nitions. Table VIII presents the empirical model.
28. Direct withdrawals from pension funds are costly under the existing pension law. However,
?rms can withdraw funds indirectly, without incurring such costs, by slowing down pension
funding.
29. In situations where the ?rm does not report its labor expense, I estimate labor expense as the
mean labor expense per employee for the industry multiplied by the number of active
employees of the ?rm.
30. Asthana (1999, 2001) provide evidence that the users of ?nancial information remove the
effects of variations in actuarial assumptions by adjusting the obligations to industry
averages of actuarial assumptions.
31. The interdependence of MIX1, MIX2, and MIX3 is also evident in Table IV, where all
Pearson correlation coef?cients between these three variables are signi?cant.
32. All ratios (variables that are de?ated by another variable) are truncated in the range [21, 1]
to eliminate cases where division by very small numbers has occurred.
33. Consequently, I reduce the sample to 1,812 from 2,458.
34. Belsley et al.’s (1980) test for outliers is also conducted and in?uential outliers are deleted.
35. These trends are not limited to the USA but are also visible in Latin America, UK, Germany,
and Japan (Mitchell, 1999).
References
Amir, E. and Gordon, E.A. (1996), “Firms’ choice of estimation parameters: empirical evidence
from SFAS no. 106”, Journal of Accounting, Auditing and Finance, Vol. 11, pp. 427-49.
Amir, E. and Livnat, J. (1996), “Multiperiod analysis of adoption motives: the case of SFAS
no. 106”, The Accounting Review, Vol. 71, pp. 539-53.
Amir, E. and Ziv, A. (1997), “Recognition, disclosure, or delay: timing the adoption of SFAS 106”,
Journal of Accounting Research, Vol. 35, pp. 61-81.
Asthana, S.C. (1999), “Determinants of funding strategies and actuarial choices for
de?ned-bene?t pension plans”, Contemporary Accounting Research, Vol. 16, pp. 39-74.
Asthana, S.C. (2001), “The impact of regulatory and audit environment on managers’
discretionary accounting choices: the case of SFAS 106”, Accounting and the Public
Interest, Vol. 1/1.
Belsley, D., Kuh, E. and Welsch, R.E. (1980), Regression Diagnostics: Identifying In?uential Data
and Sources of Collinearity, Wiley, New York, NY.
Factors affecting
corporate choices
143
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Bodie, Z. and Papke, L.P. (1992), “Pension fund ?nance”, Pensions and the Economy: Sources,
Uses, and Limitations of Data, University of Pennsylvania Press, Philadelphia, PA,
pp. 149-72.
Bodie, Z., Light, J.O., Morck, R. and Taggart, R.A. Jr (1987), “Funding and asset allocation in
corporate pension plans: an empirical investigation”, in Bodie, Z. et al. (Eds), Issues in
Pension Economics, The University of Chicago Press, Chicago, IL.
Brown, C. and Medoff, J.L. (2003), “Firm age and wages”, Journal of Labor Economics, Vol. 21
No. 3, pp. 677-97.
Brown, S. (2004), “The impact of pension assumptions on ?rm value”, working paper, Emory
University, Atlanta, GA.
Bureau of Labor Statistics, Department of Labor (2003), News, USDL: 03-489. September 1-13.
Coronado, J.L. and Sharpe, S.A. (2003), “Did pension plan accounting contribute to a stock market
bubble?”, Brookings Papers on Economic Activity, Vol. 1, pp. 323-71.
Diamond, P.A. (2000), “What stock market returns expect for the future?”, Social Security
Bulletin, Vol. 63 No. 2, pp. 38-52.
Dorsey, S. (1987), “The economic functions of private pensions: an empirical analysis”, Journal of
Labor Economics, Vol. 5 No. 4, pp. S172-89.
D’Souza, J. (1998), “Rate regulated enterprises and mandated accounting changes: the case of
electric utilities and post-retirement bene?ts other than pensions (SFAS 106)”, The
Accounting Review, Vol. 73, pp. 387-410.
Fama, E.F. and French, K.R. (1992), “The cross-section of expected stock returns”, The Journal of
Finance, Vol. 47, pp. 427-65.
FASB (1985), Statement of Financial Accounting Standards No. 87: Employers’ Accounting for
Pensions, Financial Accounting Standards Board, Stamford, CT.
FASB (1990), Statement of Financial Accounting Standards No. 106: Employers’ Accounting for
Postretirement Bene?ts other than Pensions, Financial Accounting Standards Board,
Stamford, CT.
FASB (1998), Statement of Financial Accounting Standards No. 132. Employer’s Disclosures
about Pensions and Other Postretirement Bene?ts, Financial Accounting Standards Board,
Stamford, CT.
FASB (2003), Statement of Financial Accounting Standards No. 132 (Revised 2003). Employers’
Disclosures about Pensions and Other Postretirement Bene?ts, Financial Accounting
Standards Board, Stamford, CT.
FASB (2006), FASB Improves Employers’ Accounting for De?ned Bene?t Pension and Other
Postretirement Plans. SFAS No. 158, Financial Accounting Standards Board,
Stamford, CT.
Feldstein, M. and Morck, R. (1983), “Pension funding decisions, interest rate assumptions, and
share prices”, Financial Aspects of the United States Pension System, The University of
Chicago Press, Chicago, IL, pp. 177-207.
Francis, J.R. and Reiter, S.A. (1987), “Determinants of corporate pension funding strategy”,
Journal of Accounting & Economics, Vol. 9, pp. 35-59.
Friedman, B.M. (1983), “Pension funding, pension asset allocation, and corporate ?nance:
Evidence from individual company data”, Financial Aspects of the United States Pension
System, The University of Chicago Press, Chicago, IL, pp. 107-52.
Fronstin, P. (2001), “Retiree health bene?ts: trends and outlook”, Employee Bene?t Research
Institute Brief No. 236, August.
ARJ
21,2
144
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Ghicas, D.C. (1990), “Determinants of actuarial cost method changes for pension accounting and
funding”, The Accounting Review, Vol. 65, pp. 384-405.
Godwin, J.H., Goldberg, S.R. and Duchac, J.E. (1995), “An empirical analysis of factors associated
with changes in pension plan interest rate assumptions”, Journal of Accounting, Auditing
and Finance, Vol. 11, pp. 305-22.
Graham, J.R. (1996), “Debt and the marginal tax rate”, Journal of Financial Economics, Vol. 41,
pp. 41-73.
Gustman, A.L., Mitchell, O.S. and Steinmeier, T.L. (1994), “The role of pensions in the labor
market: a survey of the literature”, Industrial &Labor Relations Review, Vol. 47, pp. 417-38.
Ippolito, R.A. (1995), “Towards explaining the growth of de?ned contribution plans”, Industrial
Relations, Vol. 34 No. 1, pp. 1-20.
Ippolito, R.A. and James, W.H. (1992), “LBOs, reversions and implicit contracts”, The Journal of
Finance, Vol. 47, pp. 139-67.
Ippolito, R.A. and Thompson, J.W. (2000), “The survival rate of de?ned-bene?t plans, 1987-1995”,
Industrial Relations, Vol. 39, pp. 228-45.
Kieso, D.E., Weygandt, J.J. and War?eld, T. (2008), Intermediate Accounting, 12th ed., Wiley,
New York, NY.
Kruse, D.L. (1995), “Pension substitution in the 1980s: why the shift toward de?ned
contributions?”, Industrial Relations, Vol. 34, pp. 218-41.
Kruse, D.L. (1996), “Why do ?rms adopt pro?t-sharing and employee ownership plans?”,
British Journal of Industrial Relations, Vol. 34 No. 4, pp. 515-38.
McGill, D. and Grubbs, D.S. (1989), Fundamentals of Private Pensions, 6th ed., Irwin Publishing
Company, Homewood, IL.
Maddala, G.S. (1988), Introduction to Econometrics, Macmillan Publishing Company,
New York, NY.
Mitchell, O.S. (1999), “New trends in pension bene?t and retirement provisions”, National Bureau
of Economic Research (NBER) Working Paper Number W7381, October.
Mittelstaedt, H.F. (1989), “An empirical analysis of the factors underlying the decision to remove
excess assets from overfunded pension plans”, Journal of Accounting &Economics, Vol. 11,
pp. 399-418.
Mittelstaedt, H.F., Nichols, W. and Regier, P. (1995), “SFAS no. 106 and bene?t reductions in
employer-sponsored retiree health care plans”, The Accounting Review, Vol. 70, pp. 535-56.
PBGC (2007), Pension Insurance Data Book, Number 11, Pension Bene?t Guaranty Corporation,
Washington, DC.
Pension Trends (2007), Pension Trends, Vol. VIII No. 1.
Press, E.G. and Weintrop, J.B. (1990), “Accounting-based constraints in public and private debt
agreements: their association with leverage and impact on accounting choice”, Journal of
Accounting & Economics, Vol. 12, pp. 65-97.
Scholes, M.S., Wolfson, M.A., Erickson, M., Maydew, E. and Shevlin, T. (2002), Taxes and
Business Strategy: A Planning Approach, Prentice-Hall Inc., Englewood Cliff, NJ.
Stone, M.S. (1982), “A survey of research on the effects of corporate pension plan sponsorship:
implications for accounting”, Journal of Accounting Literature, Vol. 1, pp. 1-32.
Tepper, I. (1981), “Taxation and corporate pension policy”, The Journal of Finance, Vol. 36 No. 1,
pp. 1-13.
Thomas, J.K. (1988), “Corporate taxes and de?ned bene?t pension plans”, Journal of Accounting
& Economics, Vol. 10, pp. 199-237.
Factors affecting
corporate choices
145
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Thomas, J.K. (1989), “Why do ?rms terminate their overfunded pension plans”, Journal of
Accounting & Economics, Vol. 11 No. 4, pp. 361-98.
Thomas, J.K. and Tung, S. (1992), “Cost manipulation incentives under cost reimbursement:
pension costs for defense contracts”, The Accounting Review, Vol. 67, pp. 691-711.
White, H. (1980), “A heteroskedasticity-consistent covariance matrix estimator and a direct test
for heteroskedasticity”, Econometrica, Vol. 48, pp. 817-38.
Winklevoss, H.E. (1993), Pension Mathematics with Numerical Illustrations, University of
Pennsylvania Press, Philadelphia, PA.
Further reading
Mittelstaedt, H.F., Wolfson, M.A., Erickson, M., Maydew, E. and Shevlin, T. (2002), Taxes and
Business Strategy: A Planning Approach, Prentice-Hall Inc., Englewood Cliff, NJ.
Corresponding author
Sharad Asthana can be contacted at: [email protected]
ARJ
21,2
146
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_256956841.pdf
The purpose of this paper is to examine the determinants of US firms’ postretirement
benefits choices.
Accounting Research Journal
Factors affecting corporate choices of postretirement benefits in the USA
Sharad Asthana
Article information:
To cite this document:
Sharad Asthana, (2008),"Factors affecting corporate choices of postretirement benefits in the USA",
Accounting Research J ournal, Vol. 21 Iss 2 pp. 123 - 146
Permanent link to this document:
http://dx.doi.org/10.1108/10309610810905926
Downloaded on: 24 January 2016, At: 21:06 (PT)
References: this document contains references to 52 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 591 times since 2008*
Users who downloaded this article also downloaded:
Kym Boon, J ill McKinnon, Philip Ross, (2008),"Audit service quality in compulsory audit tendering:
Preparer perceptions and satisfaction", Accounting Research J ournal, Vol. 21 Iss 2 pp. 93-122 http://
dx.doi.org/10.1108/10309610810905917
Ahsan Habib, Istiaq Azim, (2008),"Corporate governance and the value-relevance of accounting
information: Evidence from Australia", Accounting Research J ournal, Vol. 21 Iss 2 pp. 167-194 http://
dx.doi.org/10.1108/10309610810905944
Beilei Zhang, David Emanuel, (2008),"The provision of non-audit services and earnings conservatism: Do
New Zealand auditors compromise their independence?", Accounting Research J ournal, Vol. 21 Iss 2 pp.
195-221 http://dx.doi.org/10.1108/10309610810905953
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Factors affecting corporate
choices of postretirement bene?ts
in the USA
Sharad Asthana
Department of Accounting, College of Business,
University of Texas at San Antonio, San Antonio, Texas, USA
Abstract
Purpose – The purpose of this paper is to examine the determinants of US ?rms’ postretirement
bene?ts choices.
Design/methodology/approach – The paper uses empirical methodology (univariate and
multivariate) to test the research hypotheses.
Findings – Industry norm, average employee age, ?nancial structure, and ?rm size are signi?cant
factors in the determination of the proportion of compensation that is deferred. Industry norm,
?nancial structure, and ?rm size are signi?cant factors that determine the percentage of deferred
compensation that is negotiated as de?ned bene?ts. Finally, industry norm, corporate tax rates, and
cash ?ow help explain the percentage of de?ned bene?ts that are paid in the form of retiree health
bene?t plans.
Research limitations/implications – Data requirements might bias the sample towards larger
sized ?rms. Data availability limits the number of observations in 2000 and 2001.
Practical implications – The trends in post-retirement bene?ts reported in this paper are important
for policy makers.
Originality/value – These ?ndings have implications for the baby boomers. The trend to offer
smaller proportion of compensation as deferred bene?ts re?ects the increasing costs of deferral to the
employers. This increases the employees’ responsibilities to save on their own. This also would shift
the retirees’ dependence on the public pension system for their retirement income. The trend to favor
de?ned-contribution plans instead of de?ned-bene?t plans re?ects the employers’ attempts to
diversify their risks of paying promised post-retirement bene?ts by transferring the risk to the
employee. On the other hand, the popularity of de?ned-contribution pension plans also re?ects the
increased Government’s incentives to encourage savings via 401-k plans and employee’s willingness
to manage their own pension portfolios.
Keywords Pensions, Retirement, Retirement insurance, Taxes, Labour market, UnitedStates of America
Paper type Research paper
1. Introduction
Research that deals with the management of post-retirement bene?ts is critical in the
light of the changing population pro?le. According to the latest Population Pro?le of the
United States (US Census Bureau), the population of 65-years old and above in the USA
is expected to more than double by 2050. The main reasons for this increase are the
Baby Boomers that retire between 2011 and 2029 and the increase in human longevity.
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
The author is thankful to Steve Balsam, Roland Lipka, Dave Ryan, and participants at the 2004
Annual American Accounting Association Conference for their helpful comments. The author is
also grateful to Shan Chen and Debbie Sinclair for their help in data collection. The author also
acknowledges ?nancial support from Baruch College and Temple University.
Factors affecting
corporate choices
123
Accounting Research Journal
Vol. 21 No. 2, 2008
pp. 123-146
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309610810905926
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Thus, planning for the welfare of the aged is one of the major priorities for the
government[1]. The government tries to ful?ll this social obligation by providing public
postretirement insurance (such as, social security, medicare, and medicaid) and
encouraging private postretirement plans (such as, pensions and retiree health care).
The current problems with the public insurance system are expected to multiply
with the aging of the Baby Boom cohort, health care cost in?ation, and medical
advances resulting in the extension of the retirement years. Projections by the US
Social Security Department show that as a result of these demographic shifts, the ratio
of Social Security bene?ciaries to workers will increase to 60 percent by 2030 (from
33 percent in the late 1990s). Expected ?nancial woes of the public insurance system in
the future would result in retirees depending more and more on their private health
care plans. This factor, combined with the expected decline in individual income as a
result of retirement and the increase in health care costs with age, make private
pension and retiree health care schemes critical determinants in individual and social
retirement planning.
The main focus of this paper is to examine how ?rms make the following three
important choices pertaining to postretirement bene?ts and their socio-economic
implications, given the changing demographics and recent amendments to the US
pension laws:
(1) What proportion of the total compensation should be deferred/cash?
(2) What proportion of the deferred compensation should be de?ned
bene?t/de?ned contribution?
(3) What proportion of the de?ned bene?t compensation should be retiree health
care/pension bene?ts?
These are important questions for several reasons. First, the total pension and retiree
health care obligations of US companies are substantial. Second, while prior research
has examined the funding, reporting, and investing policy of ?rms’ pension and post
retirement bene?t choices, little research has focused on ?rms’ bene?ts policies
themselves. Third, the US Internal Revenue Code (IRC) provides substantial tax
bene?ts for certain retirement bene?ts (e.g. de?ned bene?t pension plans) and nearly
none for other bene?ts (e.g. post-retiree health care bene?ts). It is important to
understand how tax policy affects ?rm choices.
These decisions are not trivial for US ?rms. In my sample, about 4 percent of total
compensation is deferred. Also, during the sample period, the total deferred
compensation exceeds 22 percent of total assets, 33 percent of market value, and
60 percent of book value. Given the magnitude of deferred compensation and its
alternative forms, explanations about the mix between current and deferred
compensation, the mix between de?ned bene?t and de?ned contribution plans, and
the mix between de?ned retirement health bene?ts and de?ned pension bene?ts are
important. I use publicly available data, including the Financial Accounting Standards
Board’s mandated accounting information to examine the determinants of ?rm
choices[2].
Some previous researchers have focused on ?rm choices to offer new plans and to
close old plans. Kruse (1995), Dorsey (1987) and Ippolito and Thompson (2000) provide
interesting insights into how these choices are made. The essence of their ?ndings is
that the reasons for changes in the overall mix are complex. Importantly, Kruse reports
ARJ
21,2
124
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
that companies only infrequently substitute de?ned contribution plans for de?ned
bene?t plans – only four percent of his sample switched the form of their pension plans
from bene?t to contribution. Rather, new plans are created while old plans become less
important as the participating workforce diminishes.
Researchers also report that signi?cant increases, in both absolute and relative
terms, in de?ned contribution plans occurred during the late 1980s and early 1990s
(Kruse, 1995 and Ippolito, 1995). In fact, twice as many employees currently participate
in de?ned contribution plans as do in de?ned bene?t retirement plans (Bureau of Labor
Statistics, Department of Labor, 2003, Table 3). Firms have also made changes in their
retiree health bene?ts plans primarily as a result of Financial Accounting Standard 106
(SFAS 106)[3]. In a signi?cant departure from the previous accounting practice of
accounting for non-pension post-retirement bene?ts such as health care on a
pay-as-you-go basis, SFAS 106 requires companies to recognize in the current period
the compensation expense and the corresponding liability for these bene?ts (FASB,
1990). According to Fronstin (2001, p. 9), a Mercer/Foster Higgins:
[. . .] annual survey of employers with 500 or more workers shows that those currently
expecting to continue offering health bene?ts to future early retirees declined from 46 percent
in 1993 to 31 percent in 2000.
Similarly, “another survey of larger employers . . . showed that the percentage . . .
declined from 88 percent in 1991 to 73 percent in 2000.” Diamond (2000) reports that
these costs are substantial for many companies. As a result, many companies have
taken various steps to reduce retiree health bene?ts in order to limit expenses and their
liability exposure[4].
The ?rst set of tests (univariate) corroborate Kruse’s (1995) and Fronstin’s (2001)
observations that during the period of study signi?cant changes in the de?ned bene?ts
versus de?ned contributions mix (MIX2) and in the mix of post retirement health care
bene?ts to total de?ned bene?ts (MIX3) occurred[5]. The second set of tests
(multivariate) examines the determinants of the three decisions. As postulated, I ?nd
that MIX1, the percentage of deferred compensation, is positively and reliably related
to industry norm, average age of employees, and ?rm size. The percentage deferred is
negatively and reliably related to the ?rm’s ?nancial structure (debt to assets).
MIX2, the ratio of de?ned bene?ts to total deferred compensation is, as expected,
positively and reliably related to industry norm and ?rm size and is negatively and
reliably related to ?nancial structure. How much of the de?ned bene?ts will be in the
form of retirement health care (MIX3) is found to be positively and reliably related to
industry norm and ?rm size. Corporate tax rates and operating cash ?ows have
reliably negative effects on this ratio.
The reported decline in the proportion of deferred compensation out of total
compensation is consistent with increasing costs of providing deferred compensation.
The bene?ts of deferred compensation, mainly, tax bene?ts and employee retention
through bonding are obviously becoming less attractive compared to the increasing
costs of deferral. Stricter pension accounting rules, such as SFAS 87, 106, 132, 132R,
158, combined with the funding reform provisions of The Pension Protection Act 2006
also add to the level of volatility and cost of providing deferred bene?ts (FASB, 1998,
2003)[6]. These changes in the pension laws have also lead to the decline in the
popularity of de?ned-bene?t pension plans in favor of de?ned-contribution plans, since
Factors affecting
corporate choices
125
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the former limit the liability of the employer. The improving stock market
performance, easier access to technology for managing your own investment
portfolios, and Government’s increasing incentives to contribute to 401(k) plans have
all contributed to the growing popularity of de?ned-contribution plans[7].
The rest of the paper is organized as follows. The next section discusses the theory
and develops the hypotheses. Section 3 explains the research design and Section 4
describes the sample selection procedure. The results are presented in Sections 5 and 6
concludes the paper.
2. Theory and hypotheses
2.1 Various forms of deferred compensations offered by ?rms in the USA
Total employee compensation consists of wages and salaries, current fringe bene?ts,
and the promise of future bene?ts (Stone, 1982). Future bene?ts include pension and
retiree health plans, stock options, and other post-employment bene?ts. Pension and
retiree health plans are an important device to attract, retain, motivate, and eventually
to encourage retirement (Bodie and Papke, 1992; Gustman et al., 1994)[8]. In the USA,
pension plans are classi?ed into two types: de?ned-contribution and de?ned-bene?t. In
a de?ned-contribution plan, a formula determines contributions. The employee bears
all the investment risk, and the ?rm has no formal obligation beyond making its
periodic contributions. On the other hand, in a de?ned-bene?t plan, a formula that
usually takes into account years of service and wages or salary, de?nes bene?ts.
The investment risk is borne by the sponsoring ?rm or an insurance company hired by
the ?rm.
In addition to pensions, employers offer several postretirement health care and
welfare bene?ts to retirees, their spouses, dependents, and bene?ciaries. The welfare
bene?ts include life insurance, dental care, medical care, eye care, legal and tax
services, tuition assistance, day care, and housing assistance (Kieso et al., 2008).
2.2 Prior research
A ?rm’s pension and postretirement policy has four components: funding policy,
reporting policy, investment policy and bene?ts policy. Prior research has explored
?rm funding, reporting and investment policy. Research on funding policy shows that
?nancial slack (Feldstein and Morck, 1983 and Bodie and Papke, 1992); tax rates,
capital availability, debt-equity ratios, and bonding of employees (Francis and Reiter,
1987); corporate liabilities (Friedman, 1983); and ?nancial weakening and takeover
threats (Mittelstaedt, 1989) affect the funding decisions of ?rms. Thomas (1989) and
Ippolito and James (1992) investigate the causes of plan termination. Amir and Livnat
(1996) and Amir and Ziv (1997) examine the characteristics of early adopters of SFAS
106 (FASB, 1990). Mittelstaedt et al. (1995) study the determinants of decisions to cut
retiree health bene?ts.
Prior research also examined the ?rm’s strategic reporting choices of
de?ned-bene?t pension and retiree health bene?ts. Firms can in?uence their
reported de?ned-bene?t pension obligations by adjusting their actuarial choices. Bodie
et al. (1987), Thomas (1988), Ghicas (1990), Thomas and Tung (1992), Godwin et al.
(1995) and Asthana (1999) study the determinants of actuarial choices for
de?ned-bene?t pension plans. They show that pro?tability, tax liability, working
capital, debt, rate of undertaking of new investments, management of reimbursements
ARJ
21,2
126
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
by defense contractors, funded level, contribution level, excess cash from operations,
and income management incentives motivate managers to strategically change their
reported de?ned-bene?t pension obligations.
Research has also shown that ?rms have motivations to use actuarial assumptions
to change their reported retiree health obligations. D’Souza (1998) ?nds that
competitively weak electric utilities and those that did not curtail bene?ts choose
actuarial assumptions that lower the reported retiree health obligations. Amir and
Gordon (1996) show that ?rms with larger retiree health obligations and leverage select
aggressive parameter estimates, while ?rms with extreme reported earnings-price
ratios and ongoing plan amendments use conservative parameter estimates to report
higher obligations. Membership in regulated industries and the presence of
high-quality auditors curbs the tendency of managers to manipulate actuarial
assumptions under SFAS 106 (Asthana, 2001). Bodie et al. (1987) and Friedman (1983),
among others, have examined corporate investment policies. McGill and Grubbs (1989)
and Winklevoss (1993) discuss the various components of pension fund investment
policy in detail.
Thus, the main focus of the existing research has been to explore the funding,
reporting, and investment policies of pension and retiree health plans. The
determinants of the percentage deferred and the composition of bene?ts are
relatively unexplored[9]. This study extends the existing research by examining
the factors that in?uence the decision to defer compensation, the form that the deferral
will take, and the mix between de?ned bene?t pensions and de?ned bene?t retiree
health care to its employees.
2.3 Models and hypotheses
2.3.1 Percentage of deferred compensation to total compensation (MIX1)[10]. Under
standard economic theory, wages are bargained up/down to the point where employers
and workers are, at the margin, just satis?ed. This occurs when the marginal workers
earn the value of their marginal product. Once this marginal condition is established, a
?nancing contract is elicited: in what form will this bargained wage be paid? Will it be
all cash or will it be cash and some form of savings, including deferred compensation?
Ignoring government intervention for the moment, workers should be willing to save
compensation in excess of their current out of pocket needs in a number of instruments,
including deferral accounts with their employers. Of course, the ?rm must compete
with ?nancial institutions for these funds. Consequently, the competition for workers’
deferred savings will be based on risks and returns of the employer and in the ?nancial
markets. That is, the employer is, perhaps not unwittingly, a borrower[11]. Thus,
market place interest rates will be integral in determining how much will be saved
and where.
But in the USA, the government does intervene. It does so by offering tax incentives
designed to encourage private savings. These tax incentives permit workers to enjoy a
reduction in the income tax rate on their savings with the employer vis-a` -vis
conventional savings with ?nancial intermediaries. The expected effect of this
intervention is that workers will save more because of government subsidized after-tax
returns. Another thing is clear: the “where” to save decision is now affected. At the
margin, the relative net-of-tax returns are turned in favor of saving with the employers.
Factors affecting
corporate choices
127
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Beyond this, US tax rules, which I discuss below, provide tax-based subsidies for
employers as well[12].
Companies have a vested interest in the savings of their employees to the extent that
such savings stabilizes the workforce. Employee perceptions about the value of fringe
bene?ts, such as pensions, can stabilize workers by increasing search costs for
employment changes. A stable work force bene?ts companies that face signi?cant
turnover costs. Companies also can provide savings opportunities that reduce
investment search costs for employees. Hence, an environment for ?nancial
transactions naturally exists, and when added to the income tax bene?ts attached to
deferred compensation, the pervasiveness of deferrals is hardly surprising.
However, there are some impediments for ?rms that seek to utilize their employees
as a source of ?nance. First, the cost of providing deferred plans can be prohibitive to
small ?rms. Second, the reputation of the ?rm will have an effect on the willingness of
employees to contract with the employer over deferred compensation. Also, risk
sharing is a contractual problem. Who will bear the risk depends on risk tolerance and
the willingness to compensate for risk taking. These factors will be individually
important and collectively may vary across industries. Hence, I expect that industries
that are highly pro?table and in concentrated markets are likely to offer employment
contracts that are very different from low pro?t and highly competitive industries[13].
Following Thomas (1989) and Amir and Gordon (1996), eight industries are identi?ed.
The industry classi?cation is shown in detail in Table II. These arguments suggest the
following model and hypotheses:
Model 1 : MIX1 ¼ f{IMIX1; AGE; DEBTAR; LSIZE; Set of Control Variables}½14; 15?
My arguments suggest that industry characteristics will have an important and
positive effect on the deferred compensation level that individual ?rms make.
Formally, I expect:
H1. The percentage of deferred compensation to total compensation (MIX1) for
?rms in an industry is positively associated with the average percentage of
deferred compensation for that industry (IMIX1).
I expect that ?rms with a reputation for a stable work force are more likely to provide
deferred compensation for two reasons: employees will have greater trust in these
?rms, especially since default insurance for deferred plans is not perfect[16]. Second,
such ?rms are able and willing to incur the ?xed costs associated with deferred plans
when there are bene?ts to employment longevity. Also, an older work force will
normally have higher incomes, and savings rates are positively correlated with income
levels. Finally, the age pro?le of the participants is also expected to determine the
levels of the accumulated bene?t obligation (ABO) and the accumulated postretirement
bene?t obligation (APBO). Ceteris paribus, the older the average age of the ?rm’s
employees, the closer their retirement, and the greater the present value of the cash
bene?ts expected to be paid to the employees. Hence, I posit:
H2. MIX1 is positively associated with the average age of the employees,
AGE[17].
Firms with higher debt ratios are more likely to violate their accounting-based
debt-covenants (Press and Weintrop, 1990). Since creditors may include off-balance
ARJ
21,2
128
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
sheet liabilities in their calculations of debt ratios, the ?rm’s debt ratio may affect its
mix decisions[18]. Further, I expect that a ?rm’s ?nancial structure will negatively
impact the ?rm’s ability to engage in deferred compensation plans. A risky structure
(high debt-asset ratio, DEBTAR) should also make employees less willing to rely on
the ?rm as a reliable ?nancial institution. Thus, I expect:
H3. MIX1 is negatively associated with the ?nancial structure of the ?rm,
DEBTAR.
Firm size (proxied by natural logarithm of total assets at end of the ?scal-year, LSIZE),
is an important variable in this study for several reasons. First, the number of
employees in an organization will have an effect on the amount of the bene?t
obligation. Holding everything else constant, the larger the number of employees, the
larger the values of ABO and APBO. Furthermore, Scholes et al. (2002) report that
“larger ?rms are more likely to provide (retirement health care) coverage” and Graham
(1996) shows that ?rm size is positively related to ?rm tax rates. Firm size may also
capture the effects of my arguments on age. Larger ?rms are more likely to bene?t
from economies of scale with respect to the overhead costs of deferred compensation
plans. Consequently, I expect that:
H4. MIX1 is positively associated with the size of the ?rm, LSIZE.
2.3.2 Percentage of de?ned bene?ts to total deferred compensation (MIX2). De?ned
bene?t plans typically provide unique bene?ts to employers: they cause employees to
become unsecured bondholders (Dorsey, 1987). The bene?ts from this ?nancial
arrangement are three fold. First, as unsecured bondholders, employees’ interests are
aligned with the ?rm’s, which lowers costs (Dorsey). Second, underfunding de?ned
bene?t plans is a source of low-interest loans (Kruse, 1995). Third, employee turnover
is reduced for younger employees due to the cost of losing unvested future bene?ts.
Reduced turnover lowers the riskiness of human capital investments (Dorsey).
De?ned bene?t plans permit ?rms to force early retirement without violating age
discrimination by building into the plans economic disincentives to continue working.
Firms with de?ned bene?t plans have greater control over the measurement (and
smoothing) of income (Coronado and Sharpe, 2003). Excess market gains belong to the
corporation and can be used to lower future funding (and expense) requirements. On
the cost side, de?ned bene?t plans are more expensive to administer, especially for
small plans due to the presence of ?xed set up and maintenance charges. The ?rm is
also at risk for market losses. I expect that members of the same industry will
experience similar costs and bene?ts. These observations lead to my second model and
related set of hypotheses.
Model 2 : MIX2 ¼ f{IMIX2; DEBTAR; LSIZE; Set of Control Variables} ½19?
H5 has the same rationale as H1 and is not repeated here:
H5. The ratio of de?ned bene?t plans to total deferred compensation (MIX2) for
?rms in an industry is positively associated with the average for the industry
(IMIX2).
Next, I posit that unsecured bondholder employees are risk averse and will not tolerate
large amounts of employer debt. Hence, I expect that:
Factors affecting
corporate choices
129
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
H6. MIX2 is negatively associated with the ?rm’s ?nancial structure, DEBTAR.
Conversely, large ?rms are less risky than small (Fama and French, 1992). The
likelihood of ?rm survival will be important to employee-bondholders. If true, ?rm size
should impact the choice between de?ned bene?t and de?ned contribution plans in the
following way:
H7. MIX2 is positively associated with the size of the ?rm, LSIZE.
2.3.3 Percentage of retiree health care bene?ts to total de?ned bene?ts (MIX3). The IRC
regulates the tax treatment of private pension plans. De?ned-bene?t pension plans that
meet the requirements of the IRC and federal regulations, issued from time to time, are
said to be “quali?ed” for several tax bene?ts[20]. Employer contributions to quali?ed
plans are deductible, subject to a maximum tax-deductible limit, as ordinary and
necessary business expenses for federal income tax purposes[21]. These contributions
are not taxable to participants as income until paid out as bene?ts[22]. The earnings on
the pension plan assets, including realized capital gains, are not taxable for the
sponsoring ?rm[23]. Moreover, certain distributions of bene?ts from quali?ed plans
are entitled to favorable tax treatment[24]. Finally, under the tax rules, ?rms are
allowed to overfund their de?ned bene?t pension plans by up to 150 percent of their
obligations with tax bene?ts[25]. Thus, de?ned bene?t pension plans are attractive for
?rms due to the various tax bene?ts.
Taken together, these provisions can create large opportunities for tax deferrals
(Tepper, 1981). But to the extent that a ?rm claims a current deduction when
contributing to the plan, it cannot claim a deduction in the future when bene?ts are
paid to retirees. Consequently, the ?rm’s current taxes are reduced and its future taxes
increased. The exemption from tax of the earnings of the fund means that eventually
the accumulation of untaxed earnings will reduce future plan contributions and result
in taxable income. According to Tepper (1981), “deferrals are interest free loans from
the federal government and the general rule is that the gain to the ?rm is the after-tax
(income) that accumulates on the deferral.” The value of this gain to the ?rm will be a
function of its tax rate.
Funding of retiree health plans is not tax-advantaged in the same way as funding of
pension plans since any contributions to fund future retiree health bene?ts are
generally not tax-deductible. In general, under IRC § 419, employer contributions to
any welfare bene?t fund are deductible only to the extent of the bene?ts paid in the
current year[26]. Thus, over funding of retiree health bene?ts provides no tax bene?t
and hence no gains for ?rms. Moreover, there is no minimum funding requirement for
retiree health plans. As a result, the majority of retiree health care plans in the USA are
largely under funded. Owing to the differential tax bene?ts available to de?ned-bene?t
pension and retiree health plans, ?rms with higher tax rates are expected to prefer
de?ned-bene?t pension plans to retiree health plans. The third model and the last set of
hypotheses are derived largely from the above arguments.
Model 3 : MIX3 ¼ f {IMIX3; TAXRATE; CASHAR; TAXRATE*CASHAR; LSIZE;
Set of Control Variables} ½27?
The rationale for the industry variable (IMIX3) has the same justi?cation as do H1 and
H5. The reasoning is not repeated here.
ARJ
21,2
130
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
H8. The ratio of de?ned bene?t retiree health care plans to total de?ned bene?t
plans (MIX3) for ?rms in an industry is positively associated with the average
for that industry (IMIX3).
As argued above, I expect the tax rate to in?uence the mix between de?ned pension
bene?t plans and retiree health bene?t plans:
H9. MIX3 is negatively associated with the ?rm’s tax rate, TAXRATE.
Firms that have relatively higher cash ?ow from operations (proxied by operating cash
?ow to asset ratio, CASHAR) are expected to look for opportunities to pro?tably invest
their surplus capital. Ceteris paribus, investment of the surplus cash in pension plans
provides better returns in comparison to investment in retiree health plans due to the
gains that accrue from the tax deferral available from pension plan funding. Further,
storing surplus cash in pension funds creates ?nancial slack for ?rms. Firms can utilize
this slack in periods of ?nancial need by simply revising pension plan estimates which
reduce current funding requirements. For example, if poor ?nancial performance leads
to an increased cost of capital, ?rms can avoid the higher rates by utilizing their own
?nancial reserves[28]. This provides the following hypothesis:
H10. MIX3 is negatively associated with the ?rm’s operating cash ?ow, CASHAR.
The US taxes income rather than cash ?ow. Hence, it is not uncommon to observe
high-tax-rate-low-cash-?ow companies and low-tax-rate-high-cash-?ow companies.
Consequently, I seek to determine whether the possible interaction between tax rates
and cash ?ow has an effect on MIX3. Thus, I state H11, but without direction:
H11. MIX3 is associated with the interaction of tax rate and operating cash ?ow of
the ?rm.
I expect small ?rms will be more indifferent to the mix between health care bene?t
plans and de?ned bene?t pension plans than large ?rms due to economies of scale and
larger tax bene?ts of de?ned bene?t plans relative to health care bene?ts. Hence, I
expect:
H12. MIX3 is negatively associated with the size of the ?rm, LSIZE.
3. Research design
3.1 Control variables
I include the following variables as controls in each of my regressions: a trend variable
(TREND); the natural logarithm of total other compensation (LOCOMP); the discount
rate assumed for SFAS 87 by the ?rm minus the average for the industry (DDR87); the
projected increase in future employee compensation by the ?rm minus the average for
the industry (DANNINC); the discount rate assumed for SFAS 106 by the ?rm minus
the average for the industry (DDR106); and the short- and long-term health care cost
trend rates assumed for SFAS 106 by the ?rm minus the averages for the industry
(DSTHR and DLTHR).
I include TREND to capture possible changes in retirement bene?ts over time. Other
compensation (LOCOMP) is included as a proxy for the economic well being and
independence of employees[29]. DDR87, DANNINC, DDR106, DSTHR, and DLTHR are
Factors affecting
corporate choices
131
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
included to remove the effects of actuarial assumptions under SFAS 87 and 106 from
the reported obligations[30]. All the variables used in this study are de?ned in Table I.
3.2 Simultaneous estimation approach
MIX1, MIX2, and MIX3 are jointly determined and interdependent, resulting in
endogeneity of these variables[31]. Since endogenous variables are not independent,
the error term is likely to be correlated with the explanatory variables and OLS
assumptions are violated. As a result, the regression coef?cients are inconsistent
(Maddala, 1988). To avoid this problem of misspeci?cation, Maddala’s (1988)
simultaneous estimation method is used for the estimation. Two-stage least square
technique is utilized. In the ?rst-stage, the endogenous variables, MIX1, MIX2, and
Variable De?nition (in alphabetical order)
ABO ABOs reported under SFAS No. 87
AGE Proxy for average age of employees; estimated as pension plan interest cost de?ated by
pension plan service cost (Brown, 2004)
APBO APBOs reported under SFAS No.106
ANNINC Projected annual increase in future compensation
CASHAR Cash ?ow from operations (adjusted for the effects of SFAS 87 and 106) to asset ratio
DEBTAR Total debt (adjusted for the effects of SFAS 87 and 106) to asset ratio
D When pre?xed to a variable implies the value of the variable minus its industry mean
DCPO De?ned contribution pension obligations; estimated as the present value of future
de?ned contribution pension plan payments, using annual increases of ANNINC
and discount rate of DR87 ¼ Current year’s funding to de?ned contribution pension
plan divided by (DR87-ANNINC), assuming the cash out?ow continues in perpetuity
DR87 Discount rate assumed under SFAS 87
DR106 Discount rate assumed under SFAS 106
I When pre?xed to a variable implies its industry mean
LOCOMP Natural logarithm of other compensation expenses (excluding pension and
postretirement health care bene?t expenses)
LSIZE Natural logarithm of size (total assets at the end of ?scal year)
LTHR Long-term health care cost trend rate assumed under SFAS 106
MIX1 Mix of deferred and current compensation obligations. It is equal to deferred pension and
postretirement health care obligations (ABO þ DCPO þ APBO) to total compensation
obligations (OCOMPO þ ABO þ DCPO þ APBO)
MIX2 Mix of de?ned bene?t and de?ned contribution bene?ts obligations. It is equal to (ABO
þ APBO) de?ated by (ABO þ APBO þ DCPO)
MIX3 Mix of retiree health bene?ts obligations and de?ned pension bene?t plans. It is equal to
APBO de?ated by (ABO þ APBO)
OCOMPO Other compensation obligations; estimated as the present value of future labor expenses,
other than pension and postretirement health care bene?t expenses, using annual
increases of ANNINC and discount rate of DR87 ¼ Current year’s compensation
expenses (excluding pension and postretirement health care bene?t expenses) divided by
(DR87-ANNINC), assuming the cash out?ow continues in perpetuity
STHR Short-term health care cost trend rate assumed under SFAS 106
TAXRATE Average tax rate; estimated as the federal income tax payable (net of investment tax
credit) de?ated by income before taxes. The rate is winsorized in the range (0, maximum
corporate tax rate)
TREND Trend variable proxied by the year
Table I.
Variable de?nitions
ARJ
21,2
132
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
MIX3 are estimated using only the exogenous variables in the system. This is done by
replacing MIX1, MIX2, and MIX3 with IMIX1, IMIX2, and IMIX3 as independent
variables in Regressions 1-3. Then the predicted values of MIX1, MIX2, and MIX3 from
these regressions are used as independent variables in Regressions 1-3 in the
second-stage. Maddala shows that the regression coef?cients obtained from the
two-stage least square estimation are unbiased and consistent.
4. Sample selection
A ?rm that satis?es the following criteria is included in the sample:
.
data are available on the COMPUSTAT and COMPACT-DISCLOSURE
databases for at least one year during the period 1993-2001; and
.
the ?rm has a December 31 ?scal year-end.
Some of the data in this study, namely the information on ?rms’ retiree health care
plans and de?ned contribution pension plans, were hand-collected from ?nancial
statement footnotes.
Only the ?rms with December 31 ?scal year-ends are included in the sample to align
the observations. The alignment helps in debiasing the effects of actuarial choices more
ef?ciently (Asthana, 1999, 2001). Firms were required to adopt SFAS No. 106 no later
than the ?scal years beginning after December 15, 1992. The sample period begins in
1993, since this was the ?rst ?scal year when ?rms disclosed their retiree health
obligations (there were few early adopters). The ?nal sample consists of 2,458
observations for 646 ?rms during the period 1993-2001.
5. Empirical results
5.1 Sample characteristics
Panel A of Table II presents the distribution of variables. The means for my decision
variables are: 3.9 percent (MIX1), 78.7 percent (MIX2) and 46.5 percent (MIX3). The
mean (median) TA of the sample is $10.896 ($2.072) billion. The requirement that data
for the ?rm be available on COMPUSTAT and COMPACT-DISCLOSURE may have
biased the sample in favor of large ?rms. The means (in millions) for ABO, APBO, and
DCPO are $715, $381 and $462.Thus, the average ?rm’s pension liabilities are almost
twice large as retiree health liabilities. The means for CASHAR, DEBTAR and
TAXRATE are 0.120, 0.614 and 0.299[32]. The means for the interest rates for SFAS 87
and 106 are 7.5 percent each, and the mean for the expected annual increase in
compensation is 4.8 percent. The expected short- and long-term health care cost trend
rates are 9.3 and 5.4 percent, respectively. Panel B provides the number of observations
by year. Compact Disclosure is used to hand-collect data pertaining to retiree health
plans and de?ned-contribution pension plans (not available on Compustat). This
database is only available at my institution through 2001 and the coverage of ?rms
during 2000-2001 is limited to the observations reported.
Table III reports means for MIX1, MIX2 and MIX3 across eight industries using
classi?cations similar to Thomas (1989) and Amir and Gordon (1996). The null
hypotheses of equal means for all three variables across the eight industries are
rejected at the one-percent signi?cance level. Thus, there are signi?cant variations in
the decision variables. Utilities provide the largest percentage of deferred bene?ts
(4.9 percent), and Transportation provides the lowest percentage (2.1). The construction
Factors affecting
corporate choices
133
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
industry has the highest percentage of de?ned bene?ts to total bene?ts (89.2 percent).
Consumer goods has the lowest (76.2 percent). Retiree health bene?ts as a percentage of
total de?ned bene?ts is largest for the Energy industry (56.4 percent). Construction is
lowest (42.0 percent). These patterns are quite varied across the industries and suggest
that industry membership is likely to have substantial effects on the forms of
compensation. These differences could be due to the differences in ?rms’ choices of
postretirement bene?ts in these industries, or due to the differences in the number of
employees and their age distributions.
Table IV presents the Pearson correlation coef?cients. The correlations between the
three MIX variables are signi?cant, suggesting that the use of Simultaneous Estimation
was justi?ed. Consistent with this result, the three IMIX are also signi?cantly
correlated. Next, let us look at the correlations between the independent variables
in Regressions 1-3.These coef?cients range in values from20.18 to þ0.24.Thus, none of
Panel A: distribution of variables
Variables Mean Median SD
Market value (in mill. $) 6,444.950 1,477.800 17,243.490
Book value (in mill. $) 1,999.220 681.094 4,258.220
Total assets (in mill. $) 10,895.660 2,072.020 39,729.540
No. of employees (in 000) 20.608 6.700 46.132
ABO (in mill. $) 714.580 59.303 3,598.950
APBO (in mill. $) 381.366 40.665 2,658.440
DCPO (in mill. $) 462.100 0.000 3,940.970
OCOMPO (in mill. $) 1,021.800 321.500 2,131.000
MIX1 0.039 0.021 0.107
MIX2 0.787 1.000 0.329
MIX3 0.465 0.299 0.370
AGE 2.435 2.112 2.703
CASHAR 0.120 0.098 0.286
DEBTAR 0.614 0.611 0.169
TAXRATE 0.299 0.329 0.095
LOCOMP 5.794 5.774 1.611
DR87 (percent) 7.499 7.500 0.585
ANNINC (percent) 4.832 5.000 0.624
DR106 (percent) 7.507 7.500 0.573
STHR (percent) 9.327 9.800 1.976
LTHR (percent) 5.371 5.216 0.631
Panel B: sample distribution across years
Year Observations
1993 281
1994 495
1995 635
1996 204
1997 244
1998 247
1999 200
2000 75
2001 77
Total 2,458
Notes: See Table I for variable de?nitions. Total number of sample ?rms is 646
Table II.
Sample characteristics
ARJ
21,2
134
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the explanatory variables are correlated high enough to cause multicollinearity
problems. To further rule out this possibility, variance in?ation factors (VIF) are also
calculated and reported later in the paper.
5.2 Analysis of trends in retirement bene?ts
In Table V, I report the year by year movement in the average values of the three ratios,
MIX1, MIX2, and MIX3.What I seek to know is: are the three ratios changing, and if so,
are the changes signi?cant and in what direction. The way I construct my tests is to
compute the ratios for each ?rm for all sample years and then compare the current year
to the prior year[33]. For example, if I had two companies with data for all nine years,
I would have 16 observations. If I had one company with nine years and another
company with six, the sample size would be 13.The results show a signi?cant decline
in ABO and signi?cant increases in APBO and DCPO across time. This is consistent
with de?ned-contribution pension plans becoming more popular with employers than
de?ned-bene?t pension plans. Deferred compensation as a percentage of total
compensation seems to be declining. De?ned bene?t plans (pension and retiree health
care) are becoming a smaller percentage of total deferred compensations, a trend
consistent with de?ned-contributions replacing de?ned-bene?t plans. Finally, APBO is
growing at a faster rate than ABO. Increase in the projected medical costs is the likely
reason.
Industry Observations Mean MIX1 Mean MIX2 Mean MIX3
1. Basic industries 350 0.0369 0.8069 0.4343
2. Capital goods 426 0.0370 0.7542 0.4331
3. Construction 30 0.0434 0.8917 0.4202
4. Consumer goods 695 0.0422 0.7616 0.4865
5. Energy 131 0.0438 0.7112 0.5636
6. Finance 341 0.0307 0.8128 0.4321
7. Transportation 131 0.0214 0.7783 0.4766
8. Utilities 354 0.0492 0.8557 0.4830
Total observations 2,458
F-value (H
0
: equal across industries) 44.54 5.34 3.05
Prob . F 0.0001 0.0001 0.0034
See Table I for variable de?nitions. The classi?cation into industry groups is based on Thomas (1989) as
below:
Industry group SIC codes
1. Basic industries 1000-1299, 1400-1499, 2600-2699, 2800-2829, 2870-2899, and 3300-3399
2. Capital goods 3400-3419, 3440-3599, 3670-3699, 3800-3849, 5080-5089, 5100-5129,
and 7300-7399
3. Construction 1500-1599, 2400-2499, 3220-3299, 3430-3439, and 5160-5219
4. Consumer goods 0000-0999, 2000-2399, 2500-2599, 2700-2799, 2830-2869, 3000-3219,
3420-3429, 3600-3669, 3700-3719, 3850-3879, 3880-3999, 4830-4899,
5000-5079, 5090-5099, 5130-5159, 5220-5999, 7000-7299, and 7400-9999
5. Energy 1300-1399 and 2900-2999
6. Finance 6000-6999
7. Transportation 3720-3799 and 4000-4799
8. Utilities 4800-4829 and 4900-4999
Table III.
Distribution of pension
and postretirment health
care bene?ts across
industries
Factors affecting
corporate choices
135
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
V
a
r
i
a
b
l
e
s
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)
(
7
)
(
8
)
(
9
)
(
1
0
)
(
1
1
)
(
1
2
)
(
1
3
)
(
1
4
)
(
1
5
)
(
1
6
)
(
1
7
)
M
I
X
1
(
1
)
–
m
i
x
2
(
2
)
2
0
.
1
3
m
i
x
3
(
3
)
0
.
1
1
2
0
.
3
9
i
m
i
x
1
(
4
)
0
.
0
6
0
.
0
2
0
.
0
5
i
m
i
x
2
(
5
)
0
.
0
1
0
.
1
2
2
0
.
0
3
0
.
1
7
i
m
i
x
3
(
6
)
0
.
0
3
2
0
.
0
4
0
.
0
9
0
.
4
9
2
0
.
3
3
a
g
e
(
7
)
0
.
0
0
0
.
0
3
2
0
.
0
1
0
.
0
4
2
0
.
0
3
0
.
0
1
D
E
B
T
A
R
(
8
)
2
0
.
0
2
2
0
.
0
5
0
.
0
4
2
0
.
1
8
0
.
1
5
2
0
.
0
9
2
0
.
0
3
T
R
E
N
D
(
9
)
0
.
0
2
2
0
.
4
0
0
.
6
3
0
.
0
4
2
0
.
0
6
0
.
0
8
0
.
0
5
0
.
0
6
l
s
i
z
e
(
1
0
)
2
0
.
0
9
2
0
.
0
1
2
0
.
0
2
2
0
.
1
4
0
.
1
3
2
0
.
0
1
0
.
0
0
0
.
0
6
0
.
1
1
l
O
c
o
m
p
(
1
1
)
2
0
.
1
9
2
0
.
0
2
2
0
.
0
6
2
0
.
1
7
2
0
.
1
3
2
0
.
0
6
0
.
0
4
0
.
1
7
0
.
0
7
0
.
2
4
c
a
s
h
A
R
(
1
2
)
0
.
5
0
2
0
.
1
5
0
.
0
7
0
.
0
1
2
0
.
0
8
0
.
0
1
2
0
.
0
1
2
0
.
0
3
0
.
0
9
2
0
.
0
8
2
0
.
0
3
T
A
X
R
A
T
E
(
1
3
)
2
0
.
0
4
2
0
.
0
4
2
0
.
0
6
0
.
0
9
0
.
0
5
0
.
0
7
2
0
.
0
7
2
0
.
0
3
0
.
0
6
0
.
1
0
0
.
1
3
0
.
0
6
D
D
R
8
7
(
1
4
)
0
.
0
1
0
.
0
5
2
0
.
0
3
2
0
.
0
3
2
0
.
0
4
2
0
.
0
3
0
.
0
3
0
.
0
0
2
0
.
0
1
0
.
0
4
0
.
0
7
2
0
.
0
1
0
.
0
2
D
A
N
N
I
N
C
(
1
5
)
2
0
.
0
4
2
0
.
0
5
2
0
.
0
1
0
.
0
4
0
.
0
9
2
0
.
0
1
2
0
.
0
1
2
0
.
0
1
0
.
0
4
0
.
0
0
0
.
0
1
0
.
0
5
0
.
0
6
0
.
0
7
D
D
R
1
0
6
(
1
6
)
2
0
.
0
2
0
.
0
3
2
0
.
0
1
2
0
.
0
2
0
.
0
0
2
0
.
0
4
0
.
0
0
0
.
0
0
0
.
0
0
0
.
0
4
0
.
0
4
2
0
.
0
2
0
.
0
4
0
.
2
7
0
.
0
5
D
S
T
H
R
(
1
7
)
2
0
.
0
2
2
0
.
0
1
2
0
.
0
4
0
.
0
8
0
.
0
1
2
0
.
0
1
2
0
.
0
1
2
0
.
0
4
2
0
.
0
3
2
0
.
0
9
2
0
.
0
7
2
0
.
0
3
0
.
0
2
2
0
.
0
5
0
.
0
7
2
0
.
0
2
D
L
T
H
R
(
1
8
)
0
.
0
1
2
0
.
0
7
0
.
0
1
2
0
.
0
3
0
.
0
1
0
.
0
8
2
0
.
0
1
0
.
0
5
0
.
0
4
0
.
0
2
0
.
0
0
0
.
0
6
0
.
0
5
0
.
0
7
0
.
0
3
0
.
1
0
0
.
2
3
N
o
t
e
s
:
S
e
e
T
a
b
l
e
I
f
o
r
v
a
r
i
a
b
l
e
d
e
?
n
i
t
i
o
n
s
.
I
t
a
l
i
c
n
u
m
b
e
r
s
i
m
p
l
y
s
i
g
n
i
?
c
a
n
c
e
a
t
1
0
p
e
r
c
e
n
t
l
e
v
e
l
o
r
b
e
t
t
e
r
Table IV.
Pearson correlation
matrix
ARJ
21,2
136
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
5.3 Results of tests of hypotheses
I report the regression results for MIX1 in Table VI. The adjusted R
2
for Model 1 is
23.6 percent. The F-value for the model is reliably signi?cant at the 1 percent level. The
highest VIF is below ten, which suggests that multicollinearity is not a problem[34].
Variable Current year Previous year Current – previous
Panel A: analysis of bene?ts per employee
ABO 23.6868 (15.0604) 27.1828 (20.1993)
* * *
2 3.4960 (25.1370)
APBO 14.2895 (7.7010) 13.8129 (7.3800)
*
0.4766 (0.3210)
DCPO 15.4614 (0.0000) 14.3763 (0.0000)
* *
1.0851 (0.0000)
Mean values in $ 000 per employee (median values in parentheses) (N ¼ 1,812)
Panel B: analysis of bene?t ratios
MIX1 0.0378 (0.0210) 0.0391 (0.0229)
* * *
2 0.0013 (20.0019)
MIX2 0.7541 (1.0000) 0.7866 (1.0000)
* * *
2 0.0325 (0.0000)
MIX3 0.5011 (0.3133) 0.4078 (0.2711)
* * *
0.0933 (0.0422)
Mean values (median values in parentheses) (N ¼ 1,812)
Notes: See Table I for variable de?nitions. The sample size is reduced due to differencing. As a result,
the mean values for the current year in Panel B are different from those in Panel A of Table II.
Signi?cant at the levels of
*
10,
* *
5 and
* * *
1 percent, respectively
Table V.
Trends in pension and
postretirement health
care bene?ts
Variable Hypotheses Estimate White’s t p-value
Intercept
* * *
3.7615 8.12 ,0.0001
IMIX1 þ
* * *
0.6844 7.18 ,0.0001
AGE þ
* * *
0.0010 5.5 ,0.0001
DEBTAR –
* * *
2 0.0264 27.39 ,0.0001
LSIZE þ
* * *
0.0086 15.36 ,0.0001
MIX2
* *
0.0417 2.53 0.0116
MIX3
* * *
0.1017 5.1 ,0.0001
TREND
* * *
2 0.0019 28.32 ,0.0001
LOCOMP
* * *
2 0.0075 212.8 ,0.0001
DDR87
* * *
2 0.0061 27.84 ,0.0001
DANNINC 20.0011 20.55 0.5842
DDR106
* * *
2 0.0010 23 0.0027
DSTHR 20.0005 20.59 0.5575
DLTHR
* * *
0.0072 3.71 0.0002
Observations 2,458
F-value 58.27
Prob . F 0.0001
Adj R
2
23.58 percent
White’s x
2
231.68
Prob. .x
2
0.0001
Highest VIF 4.3001
Notes: Regression 1: Mix of Deferred Post-Retirement Bene?ts and Current Compensation. MIX1 ¼
a
0
þ a
1
IMIX1 þ a
2
AGE þ a
3
DEBTARþ a
4
LSIZE þ a
5
MIX2 þ a
6
MIX3 þa
7
TREND þ a
8
LOCOMP þ a
9
DDR87 þ a
10
DANNINC þ a
11
DDR106 þ a
12
DSTHR þ a
13
DLTHR þ 1.
See Table I for variable de?nitions. Signi?cant at the levels of
*
10,
* *
5, and
* * *
1 percent, respectively
Table VI.
Regression of total
deferred compensation on
?rm and industry
characteristics
Factors affecting
corporate choices
137
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
All four null hypotheses are rejected since the estimated regression coef?cients are
statistically signi?cant and in the expected direction. I report White’s t-statistics since
there is evidence of heteroskedasticity in the data. I conclude that industry norm,
employee age, ?nancial structure of the ?rm, and ?rm size are highly in?uential in
determining how much of total compensation will be deferred by employees and
employers. I also take into account the control variables, including the joint effects of
MIX2 and MIX3 on MIX1.The effects of the control variables are signi?cant, excluding
the effects of DANNINC and DSTHR.
The results for MIX2, the percentage of de?ned bene?ts to the total package of
deferred compensation, are similar to the results for MIX1 (Table VII). All three test
variables are signi?cant at the 5 percent level or less. Again, I conclude that industry
membership, ?nancial structure, and ?rm size have a signi?cant impact on the MIX2
decision. The coef?cients of MIX1 and MIX3 are not signi?cant, which suggests that
the impact of the joint decisions is not important for MIX2.Of the control variables,
only DNNINC and DLTHR’s impact is signi?cant. The explanatory power of Model 2
(24.1 percent) is very close to model one’s power. The highest VIF is 4.4, which is
consistent with low levels of multicollinearity. White’s x
2
test indicates that
heteroskedasticity exists in the data. Consequently, I report White’s t-statistics for
signi?cance testing.
The last model estimates the signi?cance of the impact of ?ve test variables
(Table VIII) on MIX3.I ?nd that the percentage of retiree health care bene?ts to total
de?ned bene?ts (MIX3) follows most of the expectations. Again, industry norm is a
Variable Hypotheses Estimate White’s t p-value
Intercept
* * *
128.2724 24.52 ,0.0001
IMIX2 þ
* * *
0.9953 5.35 ,0.0001
DEBTAR –
* * *
2 0.2102 25.25 ,0.0001
LSIZE þ
* *
0.0133 2.11 0.0353
MIX1 21.4253 21.33 0.1833
MIX3 0.2727 1.21 0.2264
TREND
* * *
2 0.0643 224.56 ,0.0001
LOCOMP 20.0095 21.45 0.1472
DDR87 0.0134 0.59 0.5545
DANNINC
*
2 0.0154 21.78 0.0746
DDR106 0.0317 1.39 0.1656
DSTHR 0.0037 1.01 0.3125
DLTHR
* * *
2 0.0417 24.75 ,0.0001
Observations 2,458
F-value 63.27
Prob. .F 0.0001
Adj R
2
24.11 percent
White’s x
2
537.44
Prob .x
2
0.0001
Highest VIF 4.4379
Notes: See Table I for variable de?nitions. Signi?cant at the levels of
*
10,
* *
5 and
* * *
1 percent,
respectively. Regression 2: mix of de?ned bene?t and de?ned contribution bene?ts. MIX2 ¼ b
0
þb
1
IMIX2 þ b
2
DEBTAR þ b
3
LSIZE þ b
4
MIX1 þ b
5
MIX3 þ b
6
TREND þ b
7
LOCOMP þ
b
8
DDR87 þ b
9
ANNINC þ b
10
DDR106 þ b
11
DSTHR þ b
12
DLTHR þ 1
Table VII.
Regression of de?ned
bene?ts to total deferred
compensation on ?rm
and industry
characteristics
ARJ
21,2
138
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
signi?cant and positive factor. Corporate tax rates are reliably negative on the mix of
de?ned bene?ts. Cash ?ow and the interaction between tax rates and cash ?ow are
marginally signi?cant. Cash ?ow is negative as expected. The interaction between tax
rate and cash ?ow is positive, but I had no expectation. Firm size is the only test
variable that does not have a signi?cant impact on the mix. Again, the other mix
variables (MIX1 and MIX2) do not have signi?cant effects on the joint decisions.
Five of the control variables have signi?cant effects. TREND, DDR87 and DDR106
are signi?cant at the one percent level. LOCOMP and DLTHR are signi?cant at the ?ve
percent level. The effects of DANNINC and DSTHR are insigni?cant. The explanatory
power of Model 3 (64.5 percent) is much larger than the ?rst two models. The highest
VIF is 4.2 and is consistent with low levels of multicollinearity. I report White’s
t-statistics for signi?cance testing since the White’s (1980) test indicates that
heteroskedasticity exists in the data.
5.4 Sensitivity analyses
A potential model speci?cation problem in Regression 2 needs to be examined in more
detail. The potential misspeci?cation arises because more than half the sample does
not have de?ned-contribution pension plans (DCPO ¼ 0). As a result, the value of
MIX2 would be 1 for majority of ?rm-year observations. This might pose a problem in
Variable Hypotheses Estimate White’s t p-value
Intercept
* * *
2 260.3214 260.03 ,.0001
IMIX3 þ
* *
0.3791 1.99 0.0471
TAXRATE –
* * *
2 0.3287 25.71 ,.0001
CASHAR –
*
2 0.2064 21.66 0.0919
CASHAR
*
TAXRATE ?
*
0.6835 1.83 0.068
LSIZE – 0.0026 0.48 0.6334
MIX1 0.191 0.21 0.8326
MIX2 0.1651 1.07 0.2868
TREND
* * *
0.1306 60.25 ,.0001
LOCOMP
* *
2 0.0146 22.45 0.0143
DDR87
* * *
0.0670 3.59 0.0003
DANNINC 20.0037 20.5 0.6154
DDR106
* * *
0.0646 3.41 0.0007
DSTHR 20.0029 20.94 0.349
DLTHR
* *
2 0.0181 22.38 0.0174
Observations 2,458
F Value 276.12
Prob .F 0.0001
Adj R-Square 64.53 percent
White’s x
2
335.12
Prob .x
2
0.0001
Highest VIF 4.1545
Notes: Regression 3: mix of retiree health care bene?ts and de?ned-bene?t pension. MIX3 ¼ g
0
þ g
1
IMIX3 þ g
2
TAXRATE þ g
3
CASHAR þ g
4
TAXRATE
*
CASHAR þ g
5
LSIZE þg
6
MIX1 þ
g
7
MIX2 þ þg
8
TREND þ g
9
LOCOMP þ g
10
DDR87 þ g
11
DANNINC þg
12
DDR106 þ
g
13
DSTHR þg
14
DLTHR þ 1See Table I for variable de?nitions. Signi?cant at the levels of
*
10,
* *
5 and
* * *
1 percent, respectively
Table VIII.
Regression of retiree
healthcare to total de?ned
bene?ts on ?rm and
industry characteristics
Factors affecting
corporate choices
139
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Regression 2 (Table VII), since OLS estimation with a binary (or binary-like) dependent
variable could lead to estimation errors. To rule out this possibility, I ?rst estimate
regression 2 with only ?rms that offer de?ned-contribution plans (i.e. MIX2 – 1). Next,
I use Tobit estimation to account for the limited dependent variable. Table IX reports
the additional sensitivity analyses. Both the truncated and Tobit regressions yield
results very similar to those in Table VII. Thus, the reported results are robust.
6. Conclusion
This paper seeks to extend post retirement bene?t research beyond the questions of
funding and actuarial assumption choices by examining the determinants of corporate
deferred compensation policies on:
.
the proportion of total compensation deferred;
.
the proportion in the form of de?ned bene?ts; and
.
the proportion for retiree health care bene?ts.
Retirement bene?ts are essential for attracting potential employees and retaining them.
Sizable retirement bene?ts also motivate employees to retire eventually, making way
for new employees.
This research indicates that industry norm is important to all three decisions. I also
?nd that there is considerable variation in the levels of deferred compensation and the
forms of deferred bene?ts across industries. Firms appear to be constrained by their
competitors’ employment contracts. Firm size is an important determinate of total
Variable Hypotheses Truncated regression Tobit regression
Intercept
* * *
161.0790 –
IMIX2 þ
* * *
1.1355
* * *
8.3387
DEBTAR –
* * *
2 0.2758
* * *
2 1.5537
LSIZE þ
*
0.0720
* * *
0.1558
MIX1 1.4509 212.6284
MIX3 20.2588
*
3.2930
TREND
* * *
2 0.0808
* * *
2 0.4120
LOCOMP 0.0027
* * *
2 0.1566
DDR87 0.0564 20.0094
DANNINC
* * *
2 0.0541
* * *
2 0.1995
DDR106 0.0109 0.2525
DSTHR 20.0003 0.0047
DLTHR
*
2 0.0215
* * *
2 0.2677
Observations 920 2,458
F-value 66.2
Prob. .F ,0.0001
Adj R
2
0.4607
Walds x
2
459.5997
Prob .x
2
,0.0001
Notes: Regression 2: mix of De?ned Bene?t and De?ned Contribution Bene?ts. MIX2 ¼ b
0
þ b
1
IMIX2 þ b
2
DEBTAR þ b
3
LSIZE þb
4
MIX1 þ b
5
MIX3 þb
6
TREND þb
7
LOCOMP þ
b
8
DDR87 þ b
9
DANNINC þb
10
DDR106 þb
11
DSTHR þ b
12
DLTHR þ 1. See Table I for
variable de?nitions. Signi?cant at the levels of
*
10,
* *
5 and
* * *
1 percent, respectively
Table IX.
Sensitivity analyses for
Regression 2
ARJ
21,2
140
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
deferred compensation and the level of de?ned bene?ts relative to de?ned
contributions. But, it does not appear to have much in?uence in determining retiree
health care bene?ts.
Corporate ?nancial structure plays a signi?cant role in the level of total deferred
compensation and in the relative level of de?ned bene?ts. I expected this result since
deferring bene?ts effectively, especially de?ned bene?ts, creates a lender-debtor
relationship between workers and their employer. The age of the average worker is
important to the total amount of deferred compensation. Retiree healthcare bene?ts
(relative to total de?ned bene?ts) are negatively (as expected) affected by corporate tax
rates. Operating cash ?ows and the interaction between cash ?ows and tax rates are
both marginally signi?cant.
This research corroborates and extends prior literature by identifying the
determinants of corporate postretirement bene?t policies. These ?ndings have
implications for the baby boomers that have started to retire en masse. The trend to
offer smaller proportion of compensation as deferred bene?ts re?ects the increasing
costs of deferral to the employers. These costs arise due to changing US pension laws
and accounting rules. This increases the employees’ responsibilities to save on their
own and plan differently for their retirement. This also would shift the retirees’
dependence on the public pension system for their retirement income. The trend to
favor de?ned-contribution plans instead of de?ned-bene?t plans re?ects the
employers’ attempts to diversify their risks of paying promised post-retirement
bene?ts by transferring the risk to the employee. On the other hand, the popularity of
de?ned-contribution pension plans also re?ects the increased Government’s incentives
to encourage savings via 401-k plans, employee’s willingness to manage their own
pension portfolios in a favorable stock market with more advanced technology
available for self-managing their pension investments, and increased access to
diversi?ed stock and bond funds[35]. The trends in post-retirement bene?ts reported in
this paper are also important for policy makers in forging future pension and
health-care policies that secures post-retirement bene?ts for employees.
Notes
1. This concern is also re?ected in the speeches, debates, and interviews of various candidates
during the Presidential election campaigns of 2004 and 2008.
2. SFAS Nos. 87 and 106.
3. SFAS 106 was approved in 1990 and became effective for ?scal years after December 15,
1992.
4. In September 2006, FASB issued a standard (SFAS 158) that brought the pension fund crisis
back into wide-spread public awareness. SFAS 158 is an amendment to SFAS statements 87,
88, 106, and 132R and went into effect for ?scal years ending after December 2006. This
ruling requires employers to fully recognize the obligations associated with single-employer
de?ned bene?t pensions, retiree healthcare, and other postretirement plans in their ?nancial
statements (FASB, 2006). The impact of this new standard on corporate decisions is yet to be
seen.
5. Kruse ?nds that industry variability, company size and capital intensity are important
factors in the choice of plans. He also ?nds that administration costs and unionization also
play a factor. We use ?rm size as a proxy for these variables.
Factors affecting
corporate choices
141
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
6. The changes in the bene?t-mix documented in this paper have continued in the recent years
beyond the sample period (PBGC, 2007).
7. The Pension Protection Act of 2006 has signi?cantly expanded the amount of contribution to
a 401(k) plan that can be made in addition to the contribution to the de?ned-bene?t plan
(Pension Trends, 2007).
8. Gustman et al. (1994), report that de?ned bene?t plan formula often contain pension accrual
spikes that are early on designed to prevent quitting and later on are designed to encourage
early retirement. The later are negative spikes that actually reduced the pension payment.
A bene?t of this pattern is that ?rms may, without violating age discrimination, rehire
(typically as consultants) the most quali?ed of the retirees.
9. Notable exceptions are Dorsey (1987) and Kruse (1995, 1996). Dorsey’s study provides
arguments about why ?rms might choose de?ned bene?t plans over de?ned contribution
plans, including control over employee quits and alignment of employee interests with the
?rm. Kruse explains why de?ned contribution plans have gained in popularity.
10. The levels of de?ned contribution pension bene?ts offered may also affect the choices of
de?ned bene?t pension and retiree health bene?ts. Under the provisions of SFAS No. 87,
de?ned contribution plans report only their annual contributions but not the liabilities
(FASB, 1985). In the rare situation when the contribution is more (less) than the required
level, the ?rm reports an asset (liability) on its balance sheet. COMPUSTAT does not report
the contributions to these plans separately. Therefore, the annual contribution to the fund
(obtained from footnote disclosures in COMPACT-DISCLOSURE database) is used to
calculate a proxy for the level of de?ned contribution bene?ts (DCPO).
11. Dorsey (1987), among others, points out that in de?ned bene?t plans the employees actually
are unsecured bondholders. There is also growing evidence that the borrowing can be
perniciously used by employers to the harm of the employees. Corporations are currently
engaged in charges that employees received reduced returns to the bene?t of the employer.
12. US Internal Revenue Code (of 1986).
13. Brown and Medoff (2003) summarize research ?ndings about wages and ?rms
characteristics. They ?nd that compensation contracts are related to ?rm size.
14. See Table I for variable de?nitions. Table VI presents the empirical model. The test variables
in Models 1-3 are highly correlated across time with Pearson coef?cients ranging from 60 to
99.3 per cent. When I rerun the regressions with average values of these variables over the
past ?ve years instead of the current values (results not reported), the conclusions do not
change.
15. I discuss the control variables in the results later.
16. This trust is not always well placed. The pension default protections under ERISA are
substantially less than dollar for dollar.
17. My proxy for employee average age, AGE, is the pension plan interest cost de?ated by
pension plan service cost (Brown, 2004).
18. DEBTAR is adjusted for the effects of SFAS Nos. 87 and 106 by subtracting the accrued
pension cost and accrued postretirement bene?t cost from the total debt and subtracting the
prepaid pension cost and prepaid postretirement bene?t cost from the ?rm’s assets.
19. See Table I for variable de?nitions. Table VII presents the empirical model.
20. IRC § 401(a).
21. IRC § 404.
22. Treasury Regulations 1.402(a)-1(a).
23. IRC § 501(a).
ARJ
21,2
142
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
24. IRC § 402(e)(4)(A) and IRC § 402(a)(2). For a more detailed discussion of the tax bene?ts
available to quali?ed pension plans and its participants, see McGill and Grubbs (1989).
25. Under provisions of the Pension Protection Act which subsequently became part of the
Omnibus Budget Reconciliation Act of 1987.
26. Scholes et al. (2002, p. 227) discuss exceptions, such as 401(h) and Voluntary Employee
Bene?t Association (VEBA) plans, where certain tax deductions are available. However,
these plans are expensive to administer. This can make unfunded plans more desirable than
401(h) plans even when the latter are available. Moreover, the annual tax-deductible
contributions to such plans cannot exceed 25 percent of the pension contributions. As a
result, when the ?rm’s pension plan is over funded, contributions to 401(h) and VEBA plans
are not permitted at all.
27. See Table I for variable de?nitions. Table VIII presents the empirical model.
28. Direct withdrawals from pension funds are costly under the existing pension law. However,
?rms can withdraw funds indirectly, without incurring such costs, by slowing down pension
funding.
29. In situations where the ?rm does not report its labor expense, I estimate labor expense as the
mean labor expense per employee for the industry multiplied by the number of active
employees of the ?rm.
30. Asthana (1999, 2001) provide evidence that the users of ?nancial information remove the
effects of variations in actuarial assumptions by adjusting the obligations to industry
averages of actuarial assumptions.
31. The interdependence of MIX1, MIX2, and MIX3 is also evident in Table IV, where all
Pearson correlation coef?cients between these three variables are signi?cant.
32. All ratios (variables that are de?ated by another variable) are truncated in the range [21, 1]
to eliminate cases where division by very small numbers has occurred.
33. Consequently, I reduce the sample to 1,812 from 2,458.
34. Belsley et al.’s (1980) test for outliers is also conducted and in?uential outliers are deleted.
35. These trends are not limited to the USA but are also visible in Latin America, UK, Germany,
and Japan (Mitchell, 1999).
References
Amir, E. and Gordon, E.A. (1996), “Firms’ choice of estimation parameters: empirical evidence
from SFAS no. 106”, Journal of Accounting, Auditing and Finance, Vol. 11, pp. 427-49.
Amir, E. and Livnat, J. (1996), “Multiperiod analysis of adoption motives: the case of SFAS
no. 106”, The Accounting Review, Vol. 71, pp. 539-53.
Amir, E. and Ziv, A. (1997), “Recognition, disclosure, or delay: timing the adoption of SFAS 106”,
Journal of Accounting Research, Vol. 35, pp. 61-81.
Asthana, S.C. (1999), “Determinants of funding strategies and actuarial choices for
de?ned-bene?t pension plans”, Contemporary Accounting Research, Vol. 16, pp. 39-74.
Asthana, S.C. (2001), “The impact of regulatory and audit environment on managers’
discretionary accounting choices: the case of SFAS 106”, Accounting and the Public
Interest, Vol. 1/1.
Belsley, D., Kuh, E. and Welsch, R.E. (1980), Regression Diagnostics: Identifying In?uential Data
and Sources of Collinearity, Wiley, New York, NY.
Factors affecting
corporate choices
143
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Bodie, Z. and Papke, L.P. (1992), “Pension fund ?nance”, Pensions and the Economy: Sources,
Uses, and Limitations of Data, University of Pennsylvania Press, Philadelphia, PA,
pp. 149-72.
Bodie, Z., Light, J.O., Morck, R. and Taggart, R.A. Jr (1987), “Funding and asset allocation in
corporate pension plans: an empirical investigation”, in Bodie, Z. et al. (Eds), Issues in
Pension Economics, The University of Chicago Press, Chicago, IL.
Brown, C. and Medoff, J.L. (2003), “Firm age and wages”, Journal of Labor Economics, Vol. 21
No. 3, pp. 677-97.
Brown, S. (2004), “The impact of pension assumptions on ?rm value”, working paper, Emory
University, Atlanta, GA.
Bureau of Labor Statistics, Department of Labor (2003), News, USDL: 03-489. September 1-13.
Coronado, J.L. and Sharpe, S.A. (2003), “Did pension plan accounting contribute to a stock market
bubble?”, Brookings Papers on Economic Activity, Vol. 1, pp. 323-71.
Diamond, P.A. (2000), “What stock market returns expect for the future?”, Social Security
Bulletin, Vol. 63 No. 2, pp. 38-52.
Dorsey, S. (1987), “The economic functions of private pensions: an empirical analysis”, Journal of
Labor Economics, Vol. 5 No. 4, pp. S172-89.
D’Souza, J. (1998), “Rate regulated enterprises and mandated accounting changes: the case of
electric utilities and post-retirement bene?ts other than pensions (SFAS 106)”, The
Accounting Review, Vol. 73, pp. 387-410.
Fama, E.F. and French, K.R. (1992), “The cross-section of expected stock returns”, The Journal of
Finance, Vol. 47, pp. 427-65.
FASB (1985), Statement of Financial Accounting Standards No. 87: Employers’ Accounting for
Pensions, Financial Accounting Standards Board, Stamford, CT.
FASB (1990), Statement of Financial Accounting Standards No. 106: Employers’ Accounting for
Postretirement Bene?ts other than Pensions, Financial Accounting Standards Board,
Stamford, CT.
FASB (1998), Statement of Financial Accounting Standards No. 132. Employer’s Disclosures
about Pensions and Other Postretirement Bene?ts, Financial Accounting Standards Board,
Stamford, CT.
FASB (2003), Statement of Financial Accounting Standards No. 132 (Revised 2003). Employers’
Disclosures about Pensions and Other Postretirement Bene?ts, Financial Accounting
Standards Board, Stamford, CT.
FASB (2006), FASB Improves Employers’ Accounting for De?ned Bene?t Pension and Other
Postretirement Plans. SFAS No. 158, Financial Accounting Standards Board,
Stamford, CT.
Feldstein, M. and Morck, R. (1983), “Pension funding decisions, interest rate assumptions, and
share prices”, Financial Aspects of the United States Pension System, The University of
Chicago Press, Chicago, IL, pp. 177-207.
Francis, J.R. and Reiter, S.A. (1987), “Determinants of corporate pension funding strategy”,
Journal of Accounting & Economics, Vol. 9, pp. 35-59.
Friedman, B.M. (1983), “Pension funding, pension asset allocation, and corporate ?nance:
Evidence from individual company data”, Financial Aspects of the United States Pension
System, The University of Chicago Press, Chicago, IL, pp. 107-52.
Fronstin, P. (2001), “Retiree health bene?ts: trends and outlook”, Employee Bene?t Research
Institute Brief No. 236, August.
ARJ
21,2
144
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Ghicas, D.C. (1990), “Determinants of actuarial cost method changes for pension accounting and
funding”, The Accounting Review, Vol. 65, pp. 384-405.
Godwin, J.H., Goldberg, S.R. and Duchac, J.E. (1995), “An empirical analysis of factors associated
with changes in pension plan interest rate assumptions”, Journal of Accounting, Auditing
and Finance, Vol. 11, pp. 305-22.
Graham, J.R. (1996), “Debt and the marginal tax rate”, Journal of Financial Economics, Vol. 41,
pp. 41-73.
Gustman, A.L., Mitchell, O.S. and Steinmeier, T.L. (1994), “The role of pensions in the labor
market: a survey of the literature”, Industrial &Labor Relations Review, Vol. 47, pp. 417-38.
Ippolito, R.A. (1995), “Towards explaining the growth of de?ned contribution plans”, Industrial
Relations, Vol. 34 No. 1, pp. 1-20.
Ippolito, R.A. and James, W.H. (1992), “LBOs, reversions and implicit contracts”, The Journal of
Finance, Vol. 47, pp. 139-67.
Ippolito, R.A. and Thompson, J.W. (2000), “The survival rate of de?ned-bene?t plans, 1987-1995”,
Industrial Relations, Vol. 39, pp. 228-45.
Kieso, D.E., Weygandt, J.J. and War?eld, T. (2008), Intermediate Accounting, 12th ed., Wiley,
New York, NY.
Kruse, D.L. (1995), “Pension substitution in the 1980s: why the shift toward de?ned
contributions?”, Industrial Relations, Vol. 34, pp. 218-41.
Kruse, D.L. (1996), “Why do ?rms adopt pro?t-sharing and employee ownership plans?”,
British Journal of Industrial Relations, Vol. 34 No. 4, pp. 515-38.
McGill, D. and Grubbs, D.S. (1989), Fundamentals of Private Pensions, 6th ed., Irwin Publishing
Company, Homewood, IL.
Maddala, G.S. (1988), Introduction to Econometrics, Macmillan Publishing Company,
New York, NY.
Mitchell, O.S. (1999), “New trends in pension bene?t and retirement provisions”, National Bureau
of Economic Research (NBER) Working Paper Number W7381, October.
Mittelstaedt, H.F. (1989), “An empirical analysis of the factors underlying the decision to remove
excess assets from overfunded pension plans”, Journal of Accounting &Economics, Vol. 11,
pp. 399-418.
Mittelstaedt, H.F., Nichols, W. and Regier, P. (1995), “SFAS no. 106 and bene?t reductions in
employer-sponsored retiree health care plans”, The Accounting Review, Vol. 70, pp. 535-56.
PBGC (2007), Pension Insurance Data Book, Number 11, Pension Bene?t Guaranty Corporation,
Washington, DC.
Pension Trends (2007), Pension Trends, Vol. VIII No. 1.
Press, E.G. and Weintrop, J.B. (1990), “Accounting-based constraints in public and private debt
agreements: their association with leverage and impact on accounting choice”, Journal of
Accounting & Economics, Vol. 12, pp. 65-97.
Scholes, M.S., Wolfson, M.A., Erickson, M., Maydew, E. and Shevlin, T. (2002), Taxes and
Business Strategy: A Planning Approach, Prentice-Hall Inc., Englewood Cliff, NJ.
Stone, M.S. (1982), “A survey of research on the effects of corporate pension plan sponsorship:
implications for accounting”, Journal of Accounting Literature, Vol. 1, pp. 1-32.
Tepper, I. (1981), “Taxation and corporate pension policy”, The Journal of Finance, Vol. 36 No. 1,
pp. 1-13.
Thomas, J.K. (1988), “Corporate taxes and de?ned bene?t pension plans”, Journal of Accounting
& Economics, Vol. 10, pp. 199-237.
Factors affecting
corporate choices
145
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Thomas, J.K. (1989), “Why do ?rms terminate their overfunded pension plans”, Journal of
Accounting & Economics, Vol. 11 No. 4, pp. 361-98.
Thomas, J.K. and Tung, S. (1992), “Cost manipulation incentives under cost reimbursement:
pension costs for defense contracts”, The Accounting Review, Vol. 67, pp. 691-711.
White, H. (1980), “A heteroskedasticity-consistent covariance matrix estimator and a direct test
for heteroskedasticity”, Econometrica, Vol. 48, pp. 817-38.
Winklevoss, H.E. (1993), Pension Mathematics with Numerical Illustrations, University of
Pennsylvania Press, Philadelphia, PA.
Further reading
Mittelstaedt, H.F., Wolfson, M.A., Erickson, M., Maydew, E. and Shevlin, T. (2002), Taxes and
Business Strategy: A Planning Approach, Prentice-Hall Inc., Englewood Cliff, NJ.
Corresponding author
Sharad Asthana can be contacted at: [email protected]
ARJ
21,2
146
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_256956841.pdf