Are stewardship and decision usefulness
complementary of conflicting objectives of financial
accounting?
Tagung des SFB 649 “Ökonomisches Risiko” - Motzen
15 June 2007
Joachim Gassen
Institute of Accounting and Auditing
Center for Applied Statistics and Economics
School of Business and Economics
Humboldt-Universität zu Berlin
gassen@wiwi.hu-berlin.de
2
Decision Usefulness vs. Stewardship
Walkthrough
• Idea and motivation
• Background
– What is decision usefulness?
– What is stewardship?
• Research design and results
– Empirical metrics for decision usefulness and stewardship
– Data
– Test design
– Results
• Limitations and conclusions
Joachim Gassen
SFB 649 – Motzen 15 June 2007
1
3
Decision Usefulness vs. Stewardship
Idea and motivation
•
Since Beaver/Demski (1979) accounting theory tends to agree on the
information content perspective of accounting
•
Two objectives of accounting
– Providing valuation relevant and
– contracting relevant information to market participants
•
These objectives are conflicting in theory
•
But: In their current joint conceptual framework project FASB and IASB state:
“Thus, the objective of financial reporting stated in
paragraph OB2 [decision usefulness] encompasses
providing information useful in assessing management’s stewardship“
SFB 649 – Motzen 15 June 2007
Joachim Gassen
4
Decision Usefulness vs. Stewardship
What is the role of accounting information?
Two overlapping views
Valuation view (aka decision usefulness)
manager
(agent)
accounting
info
other
info
market
participants
expected return
expected risk
Contracting view (aka stewardship)
manager
(agent)
accounting
info
contract space
shareholder
(principal)
Joachim Gassen
market
participants
expected return
expected risk
other
info
SFB 649 – Motzen 15 June 2007
2
5
Decision Usefulness vs. Stewardship
So... what was the research question again?
Are stewardship and decision usefulness
complementary of conflicting objectives of financial accounting?
Joachim Gassen
SFB 649 – Motzen 15 June 2007
6
Decision Usefulness vs. Stewardship
Empirical measures for decision usefulness
•
•
•
•
•
•
•
No normative model predicting the decision usefulness of accounting items
Instead measure actual decisions which are related to arrival of accounting
information with sufficiently high probability
Being decision useful for valuation means having information content from a
valuation perspective
Focus on quarterly earnings announcements (3-day window)
Firm-specific undirectional measures
– Average abnormal turnover (ATOVER)
– Average change in closing bid-ask spread (ΔBAS)
Firm-specific directional measures
– Coefficient of a regression of 3-day CARs on change in earnings (ERC)
– Coefficient of a regression of 3-day CARs on earnings surprise (FCERC)
Combined metric based on ranks of the above (DUSCORE)
Joachim Gassen
SFB 649 – Motzen 15 June 2007
3
7
Decision Usefulness vs. Stewardship
Measure for stewardship: asymmetric earnings timeliness
• Economic news translate in accounting earnings over time, gaining verifiability
as uncertainty unravels
• Accounting earnings have to be verifiable in order to be a suitable contracting
device
• Prior: Managers are reluctant to disclose bad news
• Because of that, recognizing bad news as a loss needs a lower verifiability level
than recognizing good news as a gain
• Thus, asymmetric timeliness of accounting earnings are an overall measure for
the supply of contracting efficient accounting information
Asymmetric Timeliness
Good News
Bad News
Earnings(gain)
Earnings (loss)
t
Joachim Gassen
t
SFB 649 – Motzen 15 June 2007
8
Decision Usefulness vs. Stewardship
Demand side measures for stewardship
• Stewardship-related control risk can be expected to be in part
diversifiable
• Thus it should matter more to stakeholders with high transaction costs
(debt holders: DEBT, lenders: ACC_PAY, lessors: RENTEXP,
employees: EMPL)
• In equilibrium the contracting relevance of accounting can be expected to
increase with the relative importance of these stakeholder groups.
Joachim Gassen
SFB 649 – Motzen 15 June 2007
4
9
Decision Usefulness vs. Stewardship
Operationalized research questions
1. Do firms with more pronounced capital market reactions around quarterly
earnings announcements have more or less asymmetric timely earnings?
2. In a determinant model controlling for the information environment and other
earnings attributes:
- Do firms with more asymmetric timely earnings experience more or less
pronounced capital market reactions around quarterly earnings
announcements?
- Do firms with more high transaction cost stakeholders experience more or
less pronounced capital market reactions around quarterly earnings
announcements?
3. In a determinant model controlling for the information environment: Do the
intensity of capital market reactions around quarterly earnings announcements
and/or asymmetric earnings timeliness have an impact on the value relevance of
accounting earnings, measured by the relative return to a hedge portfolio based
on perfect accounting foresight?
Joachim Gassen
SFB 649 – Motzen 15 June 2007
10
Decision Usefulness vs. Stewardship
Data
•
•
•
•
•
•
•
Firm-specific sample based on firm time-series data
Cross-section of quarterly Compustat, IBES and daily CRSP data 1990-2005
Requiring quarterly IBES forecasts biases the sample towards big and more
visible firms
ERC and FCERC are estimated based on firm-specific time-series regressions,
requiring a minimum of 10 quarterly observations
CONS is estimated based on firm-specific time-series regressions, requiring a
minimum of 10 quarterly observations with at least 5 observations with negative
returns
Only firms with non-missing metrics are included in the analysis: Cross-sectional
sample consists of 3,425 firms, based on 119,861 firm-quarter observations
(panel sample).
Stability checks on full sample of quarterly Compustat data (1990-2006):
Up to 311,907 observations
Joachim Gassen
SFB 649 – Motzen 15 June 2007
5
11
Decision Usefulness vs. Stewardship
Some descriptives for the cross-sectional sample (n=3,425)
Variable
First
Quartile
Median
Third
Quartile
Mean
ATOVERi
0.840
2.699
5.940
4.207
4.742
ΔBASi
-15.615
-4.484
4.298
-5.664
25.533
ERCi
-0.044
0.201
0.661
0.443
1.309
FCERCi
0.239
1.969
6.840
6.522
14.025
CONSi
-1.318
0.706
4.082
2.396
8.609
Variable
ATOVERi
ATOVERi
Std. Dev.
ΔBASi
ERCi
FCERCi
DUSCOREi
CONSi
-0.021
-0.012
0.250
0.461
0.032
0.007
-0.421
0.041
0.156
0.437
-0.060
0.455
-0.076
ΔBASi
-0.063
ERCi
-0.033
-0.013
FCERCi
0.173
-0.014
DUSCOREi
0.531
-0.488
0.522
0.627
CONSi
0.062
0.017
-0.085
-0.080
0.003
0.209
-0.081
-0.057
SFB 649 – Motzen 15 June 2007
Joachim Gassen
12
Decision Usefulness vs. Stewardship
Results RQ 1
NIt ,i =
2005
10
∑ γ YEAR + ∑δ FFINDUSTRY
t =1990
t
t
j =1
j
t ,i
+ β1 NEGt ,i + β 2 DUSCOREi + β3 NEGt ,i * DUSCOREi +
β 4 RETt ,i + β5 RETt ,i * DUSCOREi + β6 RETt ,i * NEGt ,i + β 7 RETt ,i * NEGt ,i * DUSCOREi + ε t
Panel Sample
Joachim Gassen
Estimate
Prob.
NEGt,i
-0.004
0.048
DUSCOREi
0.008
0.001
NEGt,i*DUSCOREi
0.005
0.101
RETt,i
-0.078
0.000
RETt,i*DUSCOREi
0.095
0.000
RETt,i*NEGt,i
0.260
0.000
RETt,i*NEGt,i*DUSCOREi
-0.253
0.000
R2
0.079
n
119,861
SFB 649 – Motzen 15 June 2007
6
13
Decision Usefulness vs. Stewardship
Results RQ 2: CONS as supply-side stewardship metric
10
DUVARi = ∑ δ i FFINDUSTRYi + β1SIZEi + β 2 log( NUMESTi ) + β 3 MTBi +
i =1
β 4 PREDi + β 5 PERSi + β 6 ZRETURNi + β 7CONSi + ε i
DUVARi
=
ATOVERi
DUVARi
=
ΔBASi
DUVARi
=
ERCi
DUVARi
=
FCERCi
DUVARi
=
DUSCOREi
Estimate
Estimate
Estimate
Estimate
Estimate
SIZEi
-0.658
2.742
0.035
2.217
-0.007
log(NUMESTi)
0.614
0.088
-0.019
0.550
0.008
MTBi
0.236
0.571
0.033
0.705
0.002
PREDi
-0.550
4.096
0.638
-1.368
-0.012
PERSi
-0.526
-0.866
0.196
0.920
0.038
ZRETURNi
-15.335
58.288
1.402
-2.683
-0.445
CONSi
0.397
6.571
-0.408
-4.720
-0.089
R2
0.404
0.026
0.026
0.167
0.152
n
3,425
3,425
3,425
3,425
3,425
SFB 649 – Motzen 15 June 2007
Joachim Gassen
14
Decision Usefulness vs. Stewardship
Results RQ 2: Demand-side stewardship metrics, DUSCORE
Estimate
Estimate
Estimate
Estimate
Estimate
SIZEi
0.000
-0.002
-0.007
-0.004
-0.004
log(NUMESTi)
0.007
0.007
0.007
0.007
0.007
MTBi
0.001
0.002
0.004
0.003
0.003
PREDi
-0.007
-0.008
0.006
-0.002
0.008
PERSi
0.031
0.033
0.033
0.035
0.032
ZRETURNi
-0.354
-0.403
-0.472
-0.420
-0.404
DEBTi
-0.108
ACC_PAYi
-0.100
-0.056
RENTEXPi
-0.067
-0.250
EMPLi
-0.208
-1.398
CONSi
-0.425
-0.082
R2
0.133
0.125
0.146
0.138
0.166
n
2,978
2,978
2,978
2,978
2,978
Joachim Gassen
SFB 649 – Motzen 15 June 2007
7
15
Decision Usefulness vs. Stewardship
Results RQ 3: Impact on value relevance
Dep. Var = RHRET
Estimate
Prob.
Dep. Var = RHRET
Estimate
Prob.
SIZEi
-0.026
0.000
-0.030
0.000
log(NUMESTi)
-0.003
0.023
-0.001
0.294
ZRETURNi
0.155
0.002
0.132
0.005
0.084
0.000
CONSi
0.072
0.000
ATOVERi
0.005
0.000
ΔBASi
0.000
0.098
ERCi
0.010
0.000
FCERCi
-0.001
0.002
DUSCOREi
0.133
0.000
R2
0.120
0.116
n
3,425
3,425
Joachim Gassen
SFB 649 – Motzen 15 June 2007
16
Decision Usefulness vs. Stewardship
Some limitations
•
Robustness: Checked for sensitivity to
– Outliers
– Required length of the firm-specific time-series
– Alternative specifications of the earnings change variable
– Additional control variables aiming at the operational environment of the firm
included in the determinant models
•
Internal Validity
– Measures: Do they capture the underlying economic concept of decision
usefulness and stewardship?
– Potentially omitted variables in determinant models
– Functional form: determinants of value relevance, decision usefulness and
value relevance constitute a simultaneous equation problem
•
External Validity: Non IBES firms? Outside U.S.? Other time periods?
Joachim Gassen
SFB 649 – Motzen 15 June 2007
8
17
Decision Usefulness vs. Stewardship
Conclusions
•
Stressing the importance of the limitations discussed above:
– Like predicted by theory, stewardship and decision usefulness seem to be
conflicting objectives of financial accounting
– Both have a positive impact on value relevance
– The valuation relevance of accounting appears to be more pronounced for
firms from poor information environments
– For firms from richer information environments, accounting appears to play
primary a confirmatory role
•
What do we learn from this about risk?
– Information about risky assets is used by two different channels
• contracting
• valuation
– As they both at least partly conflict, information design needs to address both
– Link between information and cost of capital might be driven by both as well
SFB 649 – Motzen 15 June 2007
Joachim Gassen
18
Decision Usefulness vs. Stewardship
Do both views produce the same ranking of accounting systems?
Simple state dependent investment and incentive problem
state of nature
bad
so so
good
discomfort
m provides low effort
0.6
0.3
0.1
3
m provides high effort
0.1
0.4
0.5
5
-100
10
50
pay-off to investor
• Two possible accounting systems
– Disclose the state of nature
– Disclose the effort of the manager
• The contracting view would favor the latter, while the valuation view
would favor the former, assuming that early uncertainty unraveling is
preferred by the investor
Joachim Gassen
SFB 649 – Motzen 15 June 2007
9
19
Decision Usefulness vs. Stewardship
Decision usefulness metrics
ATOVERi
Cumulative daily turnover from the day prior to the day after the Compustat
quarterly earnings announcement date minus the average daily turnover of a
combined pre and post window, the pre window beginning 30 days and ending 5
days prior and the post window beginning 5 days and ending 30 days post the
earnings announcement date. Averaged over all available observations of firm i,
requiring a minimum of ten observations.
ΔBASi
Difference between the average BASd,i from a window beginning 2 days and
ending 5 days post the quarterly earnings announcement date and the average
BASd,i from a window beginning 5 days and ending 2 days prior the earnings
announcement date. Averaged over all available observations of firm i, requiring
a minimum of ten observations.
ERCi
Coefficient α1,i from the regression: CARt ,i = α 0,i + α1,i ΔNI t ,i + ε t ,i
over the time series of quarterly observations of firm i, requiring a minimum of
ten observations.
FCERCi
Coefficient α1,i from the regression: CARt ,i = α 0,i + α1,i FCEt ,i + ε t ,i
over the time series of quarterly observations of firm i, requiring a minimum of
ten observations.
DUSCOREi Sum of the decile ranks of ATOVERi, - ΔBASi, ERCi, and FCERCi, [0,1]
distributed.
SFB 649 – Motzen 15 June 2007
Joachim Gassen
20
Decision Usefulness vs. Stewardship
Stewardship metrics
CONSi
Kink in the regression line of the regression ,
NI t ,i = β 0,i + β1,i NEGt ,i + β 2,i RETt ,i + β 3,i NEGt ,i * RETt ,i + ε t ,i
calculated in degrees as
CONSi = arctan(β 2,i + β 3,i ) − arctan(β 2,i )
for each firm i, requiring a minimum of ten observations and a minimum of five
observations with NEG equal to one.
DEBTi
Average of total debt divided by total assets for firm i.
ACC_PAYi
Average of accounts payable divided by total assets for firm i.
RENTEXPi
Average of rental expenses divided by net sales for firm i.
EMPLi
Average of employees in thousands divided by net sales in million USD
for firm i.
Joachim Gassen
SFB 649 – Motzen 15 June 2007
10
21
Decision Usefulness vs. Stewardship
The CONS-metric
NI
Kink α in the regression line of the regression:
NI t ,i = β 0 ,i + β1,i NEGt ,i + β 2,i RETt ,i + β 3,i NEGt ,i * RETt ,i + ε t ,i
1
Calculated in degrees as
γ
β2
δ
α
RET
CONSi = arctan(β 2,i + β 3,i ) − arctan(β 2,i )
for each firm i, requiring a minimum of
ten observations and a minimum of five
observations with NEG equal to one.
β3
α = δ −γ
γ = arctan β 2
δ = arctan(β 2 + β 3 )
⇒ α = arctan(β 2 + β 3 ) − arctan β 2
SFB 649 – Motzen 15 June 2007
Joachim Gassen
22
Decision Usefulness vs. Stewardship
Results RQ2: ERC estimated on positive earnings changes only
10
POSERC i = ∑ δ i FFINDUSTRYi + β1 SIZE i + β 2 log( NUMESTi ) + β 3 MTBi +
i =1
β 4 PREDi + β 5 PERS i + β 6 ZRETURN i + β 7 CONS i + ε i
Dep. Var. = POSERC
Joachim Gassen
Estimate
Prob.
SIZEi
0.032
0.584
log(NUMESTi)
0.006
0.760
MTBi
0.017
0.468
PREDi
1.195
0.000
PERSi
0.287
0.107
ZRETURNi
1.516
0.066
CONSi
-0.678
0.010
R2
0.022
n
2,798
SFB 649 – Motzen 15 June 2007
11
23
Decision Usefulness vs. Stewardship
Some related literature
•
Analytical models comparing stewardship and decision usefulness: Gjesdal
(JAR, 1981), Liang (CAR, 2000), Arya/Glover/Mittendorf/Zhang (RAS, 2004),
Christensen/Demski (SBR, 2004)
•
Management compensation role and valuation role of earnings:
Bushman/Engel/Smith (JAR, 2006)
•
Decision usefulness and its determinants: Landsman/Maydew (JAR, 2002),
Francis/Schipper/Vincent (TAR, 2002), DeFond/Hung/Trezevant (JAE, 2007)
•
Financial accounting and complementary/competing information sources:
Francis/Schipper/Vincent (JAE, 2002), Chen/Cheng/Lo (WP, 2006)
•
Earnings attributes: Lo/Lys (WP, 2000), Schipper/Vincent (AH, 2003),
Francis/LaFond/Ohlson/Schipper (TAR, 2004)
SFB 649 – Motzen 15 June 2007
Joachim Gassen
24
Decision Usefulness vs. Stewardship
FASB/IASB definition of decision usefulness in OB2
“information that is useful to present and potential investors and creditors and
others in making investment, credit, and similar resource allocation decisions”
⇔
information that is useful in making investment, credit,
and similar resource allocation decisions
⇔
information that is useful in making resource allocation decisions
⇔
information that is useful in making decisions
⇔
information
Joachim Gassen
SFB 649 – Motzen 15 June 2007
12
25
Decision Usefulness vs. Stewardship
Narrower definition (OB3)
„financial reporting should provide information to help present and
potential investors and creditors and others to assess the amounts, timing,
and uncertainty of the entity’s future cash inflows and outflows”
•
Lending from Christensen/Demski (2003, pp. 147 ff.):
E [PVt θ t ] =
•
∑ E [CF θ ](1 + r )
T
j =t +1
j
t− j
t
Information content in a valuation setting (valuation relevance of information)
exist for any change in θ if and only if value varies systematically with the
information change
⎧⎪ δE [CF j θ t ] ⎫⎪
δE [PVt θ t ]
≠0⇒⎨j
≠ 0⎬ ≠ ∅
δθ t
δθ t
⎪⎭
⎪⎩
Joachim Gassen
SFB 649 – Motzen 15 June 2007
26
Decision Usefulness vs. Stewardship
Definition of stewardship
• FASB/IASB explain the stewardship role as follows: “Management of an
entity is accountable to owners (shareholders) for the custody and
safekeeping of the entity’s economic resources and for their efficient and
profitable use” (OB 27)
• Stewardship related information provides users with information about
how management discharged its stewardship role
• Thus, stewardship related information contains information about
management’s actions for controlling purposes (Gjesdal (1981),
Bushman et al. (2006))
• Conceptual definition of stewardship is closely related to information
content in a contracting setting (contracting relevance) as defined by
Christensen/Demski (2003, pp. 229 f.)
Joachim Gassen
SFB 649 – Motzen 15 June 2007
13