Transcript Download Presentation

Pensions and Corporate
Performance: Effects of the Shift from
Defined Benefit to Defined
Contribution Pension Plans
James Brander and Seán Finucane (Presenter)
1
Outline
•
•
•
•
•
Background
Previous literature
Theoretical model
Data
Results
2
Background
• Over $10 trillion currently in US pension
plans
• Long slow shift from Defined Benefit to
Defined Contribution
• What effect does this have on corporate
performance?
3
Previous Literature
• Incentive effects (Lazear (1983))
• Kotlikoff and Wise (1984)
• Shift from Defined Benefit to Defined
Contribution (Friedberg and Webb, 2003)
• Relationships between pensions and
productivity (Cornwell and Dorsey,
1997,2000)
4
Pension accruals – DB vs. DC
Friedberg and Webb (2003)
5
Why change from DB to DC?
•
•
•
•
•
DC plans shift market risk to employee
DC plans are perceived as less expensive
DC plans are perceived as less risky
DC plans are more portable
DC plans have smooth contributions through
the lifetime compared to DB plans
• So why doesn’t everyone go DC?
6
Why would this affect corporate
performance?
• We presume a correlation between the desire
to work longer (i.e. retire later) and worker
productivity
• Compensation may differ from the value of the
marginal product
• Those who want to work longer are more likely
to choose firms with DC plans
• Those who are unhappy can leave more easily
with a DC plan
7
Theoretical model
Vt( 0) = u(wt ,et ,α) + γE(Vt+1)
Vt(1) = u(Pt)
NVt = Vt(1) – Vt(0)
πt = st(wt,et,β) + γE(πt+1)
8
Data – P&I
• Pensions and Investments magazine survey of
top 1000 pension funds
• Survey augmented by P&I research staff
• Contains total assets, fine breakdown of asset
allocations
• Our study only uses publicly traded companies
for the sample
• Firm data from Compustat
9
Alternative Data Source – Form
5500
• Must be submitted by all firms in the US with
over 100 employees to the Department of
Labor (every 3 years for smaller firms)
• Electronic data available from 1993 to present
• Includes broad asset allocation data, actuarial
assumptions and total liabilities
10
Pension assets
6,000,000
Pension Assets in $Millions
5,000,000
4,000,000
Multi-employer
Public and NP Sector
3,000,000
Union
Corporate
2,000,000
1,000,000
1997
1998
1999
2000
2001
2002
2003
Year
11
Largest corporate sponsors
Corporate
Sponsor
Corporate
Assets ($bill.)
Pension assets
($billion)
Defined Benefit
Share (%)
Defined Contr.
Share (%)
1997
General Motors
General Electric
IBM
Lucent Tech.
Ford Motor
228.9
304.0
81.5
24.8
279.1
90.6
56.9
53.1
48.4
47.8
79.1
68.5
73.7
74.3
74.7
20.9
31.5
26.3
25.7
25.3
448.5
647.5
104.5
53.0
315.9
89.7
61.6
60.1
58.2
44.9
79.1
67.0
65.2
66.3
73.8
20.9
33.0
34.8
33.7
26.2
2003
General Motors
General Electric
IBM
Boeing
Ford Motor
12
Shift to DC plans
60.0%
50.0%
DC Share
40.0%
Top 1000
Corporate
Union
Public
Multi-Employer
30.0%
20.0%
10.0%
0.0%
1997
1998
1999
2000
Year
2001
2002
2003
13
Panel Structure
Year
Firms
# of years in data
# of firms
1997
383
1
86
1998
419
2
79
1999
423
3
50
2000
422
4
55
2001
407
5
49
2002
420
6
59
2003
405
7
238
Total
Observations
2879
Total Firms
616
14
Initial Results – Simple OLS
(ROA=ß*DCShare)
Year
1997
1998
1999
2000
2001
2002
2003
aggregate
DC share
3.24**
(2.5)
3.08**
(2.1)
8.2***
(5.8)
4.77***
(3.5)
6.27***
(3.8)
8.52***
(5.6)
6.34***
(4.9)
5.42***
(9.9)
constant
4.0***
(6.0)
3.3***
(4.2)
1.6**
(2.2)
2.4 ***
(3.3)
-.6
(.6)
-1.5*
(1.7)
.1
(0.2)
1.5***
(5.1)
adjusted R2
.01
.01
.07
.03
.03
.07
.05
.03
observations
383
419
423
422
407
420
405
2879
15
OLS with controls
Specification
DC share
1
5.42***
(9.9)
2
5.42***
(8.2)
lagged DC share
3
4
5
6.09***
(7.9)
4.73***
(5.9)
0.37***
(9.0)
6.18***
(6.4)
0.32***
(7.8)
1.14***
(3.0)
yes
.04
2231
-0.27
(0.6)
yes
0.14
2136
1997 return on
assets
lagged assets
lagged
employment
lagged PPE
lagged stock
price increase
year fixed
effects
industry fixed
effects
constant
robust errors
R2
observations
1.54***
(5.1)
No
.03
2879
1.54***
(4.6)
yes
.03
2879
6
5.76***
(5.7)
0.31***
(7.48)
-2.13 *
(1.9)
6.6***
(4.1)
20.9
(1.3)
.024***
(3.5)
F(5,11)
F(5,1911)
= 14.4***
= 12.9 ***
F(17, 2111) F(17,1911)
= 10.0***
= 9.0 ***
-2.3**
-1.39
(2.4)
(1.4)
yes
yes
.22
.25
2136
1940
16
Panel Results
Specification
DC share
year fixed
effects
industry fixed
effects
constant
R2
observations
Fixed Effects
(“within” estimator)
4.22**
(2.37)
F(6,2256) = 29.3***
Random Effects
5.09***
(6.39)
Chi2 (6) = 172***
“Between”
estimator
4.54***
(4.9)
F(6,608) = 2**
First Differences
(with robust errors)
6.28**
(2.3)
F(5,2207) = 12.4***
F(17, 2207) = .37
3.74***
(4.2)
.06 (overall)
2879
2.93***
(5.9)
.06 (overall)
2879
1.79*
(1.89)
.05 (between)
2879
-1.18
(1.2)
.04
2231
17
Balanced Panel
Dependent Variable
1998-2003
average roa
1998-2003
average roa
Change in roa:
1997-2003
1997 DC share
6.60***
(4.80)
5.50***
(3.81)
6.24**
(2.47)
1997-2002 change in DC
share
2003 roa
7.53**
(2.28)
1997 return on assets
.32***
(3.81)
-.81***
(7.0)
(1.7)
.22*
industry fixed effects
F(16,219)
= 92 ***
F(17,218)
= 1458 ***
F(16,218)
= 199***
F(17,218)
= 2393***
constant
0.58
(0.3)
-1.5
(1.0)
-0.95
(0.4)
2.1
(0.9)
robust errors
Yes
yes
yes
yes
R2
.30
.42
.41
.18
observations
238
238
238
238
18
Conclusions
• Increased use of defined contribution plans
seems to be related to increased corporate
performance (ROA, OROA, ROE)
• We interpret this to be caused by improved
worker mobility and retirement decisions
19
Comments?
Suggestions?
20