Reconsidering the Reputation: Performance Relationship: A Resource-Based View
-
Upload
india-guzman -
Category
Documents
-
view
16 -
download
0
description
Transcript of Reconsidering the Reputation: Performance Relationship: A Resource-Based View
Reconsidering the Reputation: Reconsidering the Reputation: Performance Relationship: Performance Relationship:
A Resource-Based ViewA Resource-Based ViewBrian Boyd
Arizona State University
Don BerghUniversity of Denver
Dave KetchenAuburn University
Journal of Management, 2010
Evolution of PaperEvolution of Paper
Design to showroom time: One year
Genesis of project was a question regarding construct validity
Paper morphed into a hybrid methods/theory topic
University Competition: University Competition: Pre-WWIIPre-WWII
“…“…colleges struggling to find colleges struggling to find available models and competing available models and competing on terrain that was not yet stable on terrain that was not yet stable
or structured” (Washington & or structured” (Washington & Ventresca, 2004)Ventresca, 2004)
“…“…colleges struggling to find colleges struggling to find available models and competing available models and competing on terrain that was not yet stable on terrain that was not yet stable
or structured” (Washington & or structured” (Washington & Ventresca, 2004)Ventresca, 2004)
Post-War: Geographic Post-War: Geographic Integration and StratificationIntegration and Stratification
National Institutions:Compete on Differentiation
Local Institutions:Compete on Price
Financial Times Financial Times MBA Tuition 2007:MBA Tuition 2007:
Top decile: $80,000Top decile: $80,000
Bottom decile: $37,000Bottom decile: $37,000
What Is B-School Reputation?What Is B-School Reputation?
LabelsLabels Quality Prestige
MeasuresMeasures Graduate offerings Research output Faculty DGI Department size
Rindova’s 2005 Model of Rindova’s 2005 Model of ReputationReputation
Quality
Prominence
Salaries
R2 = .11
R2 = .36
0.33*
Faculty Exp
Degree Prestige
Faculty Pubs
Media Ranking
GMAT
R2 = .69
0.03
-.01
-.51*
0.17*
0.18*
0.59*
0.13*Reputation is composed of two Reputation is composed of two separate dimensions, with separate dimensions, with unique drivers and outcomesunique drivers and outcomes
SEM: The Problem of SEM: The Problem of Equivalent ModelsEquivalent Models
SEM appealSEM appeal Concurrent test of
multiple relationships Both global and local
tests of hypotheses Global fit measures
enable comparison of competing hypotheses
Equivalent modelsEquivalent models Many configurations
can yield identical fit Rarely addressed in
published studies Completely different
patterns of covariation and causation
Henley et al, ORM 2006Henley et al, ORM 2006Replication of Stimpert & Duhaime 1997Replication of Stimpert & Duhaime 1997
IndustryProfitability
R&DExpenditures
FirmDiversification
CapitalInvestment
SBUEffectiveness
Hypothesized Hypothesized ModelModel
Henley et al, ORM 2006Henley et al, ORM 2006Replication of Stimpert & Duhaime 1997Replication of Stimpert & Duhaime 1997
IndustryProfitability
R&DExpenditures
FirmDiversification
CapitalInvestment
SBUEffectiveness
Reverse Causality Reverse Causality ModelModel
Henley et al, ORM 2006Henley et al, ORM 2006Replication of Stimpert & Duhaime 1997Replication of Stimpert & Duhaime 1997
IndustryProfitability
R&DExpenditures
FirmDiversification
CapitalInvestment
SBUEffectiveness
Covariation ModelCovariation Model
Each model has a Each model has a unique logic, but unique logic, but
identical chi-square, identical chi-square, GFI, CFI, and so onGFI, CFI, and so on
Content Analysis by Henley Content Analysis by Henley and Colleagues and Colleagues
Reviewed 109 SEM articles Few alternate or equivalent models
addressed in papers 79% of articles had at least one
equivalent model Causality reduced to covariation in 71%
of studies Causality reversed in 48% of studies
EquivalentEquivalent models follow models follow a very specific set of a very specific set of
criteria.criteria.
Much broader range of Much broader range of alternatealternate models models
available.available.
Going From Equivalent to Alternate Models:Going From Equivalent to Alternate Models:Applying the RBV LensApplying the RBV Lens
RBV defines reputation as a general attribute, with ambiguous and interconnected elements
Synergies between elements combine to build a durable competitive edge
Example: Do better students attract better profs, or do better profs attract better students?
Accumulation advantage posits an ever widening disparity between ‘haves’ and ‘have-nots’
Reconfiguring Rindova’s Reconfiguring Rindova’s ComponentsComponents
Student QualityStudent Quality
GMATGMAT
Media RankMedia Rank
Faculty PubsFaculty Pubs
Degree PrestigeDegree Prestige
Reputation
.37 (3.4)
1.0
-.94 (10.5)
.93 (11.1)
.92 (11.2)
Chi-square is non-sig
GFI is 0.97
RMSR is 0.05
Direct Comparison of Direct Comparison of Competing LogicsCompeting Logics
Quality
Prominence
Salaries
R2 = .11
R2 = .36
0.33*
Faculty Exp
Degree Prestige
Faculty Pubs
Media Ranking
GMAT
R2 = .69
0.03
-.01
-.51*
0.17*
0.18*
0.59*
0.13*Quality
GMAT
MediaRank
FacPubs
DegPrestige
Reputation
Prominence
Salaries
R2 = .79
R2 = .45
0.96(12.1)
0.66(8.2)
0.37(3.4)
-.94(10.5)
1.0
0.92(11.2)
0.93(11.1)
Faculty Exp
-.20(2.9)
0.05(0.1)
Additional ConfigurationsAdditional Configurations
Quality
GMAT
MediaRank
FacPubs
DegPrestige
Reputation
Salaries
Faculty ExpProminence
(a) Prominence as part of reputation
Quality
GMAT
MediaRank
FacPubs
DegPrestige
ReputationProminence
Salaries
Faculty Exp
(b) Direct effect on salaries
Quality
GMAT
MediaRank
FacPubs
DegPrestige
Reputation
Elite
Faculty Exp
Prominence
Salaries
(c) Prominence and salaries as common dimension
Comparison of ModelsComparison of Models The usual suspects
2
- GFI
- RMSR Adjustment for model complexity
2 /df
- AIC
- Standardized 2
Benchmarks against null models- CFI
- IFI
Comparison of ModelsComparison of Models
Hypothesized model fit better than Rindova on all eight indicators
Alternative A – comparable to Rindova on all eight indicators
Alternative B – Best fit on four indicators, but less parsimonious and no significant improvement on 2
Alternative C – Better fit than Rindova, worse than hypothesized or Model b
Hypothesized ModelHypothesized ModelQuality
GMAT
MediaRank
FacPubs
DegPrestige
Reputation
Prominence
Salaries
R2 = .79
R2 = .45
0.96(12.1)
0.66(8.2)
0.37(3.4)
-.94(10.5)
1.0
0.92(11.2)
0.93(11.1)
Faculty Exp
-.20(2.9)
0.05(0.1)
ImplicationsImplications Need for devils advocacy by both authors
and reviewers RBV logic more compelling for reputation
than signaling or institutional perspectives Difficult to boost reputation simply by
pulling a lever Prominence is a critical mediator of
reputational effect