Model Specification

Empirical Testing of Structure-Conduct (Strategy)- Performance
• Model Specification
 In practice, researchers estimate a statistical model of

the following form where data are aggregated to the
industry level:
 Industry Profit Rates = f (Concentration, Barriers to

Entry, Demand …)
3–1
 Multiple regression analysis seeks to evaluate the degrees

to which deviations of the dependent variable (and in this
course our focus has been on profit rates as the dependent
variable) from its mean are “explained by” or associated with
variations in each of a set of independent or explanatory
variables (e.g., concentration, barriers to entry, demand, etc.)
3–2
 The nature of this association is captured by regression

coefficients relating the profit rates in the industry of each
independent variable, allowing us to determine the effect,
for example, of a 10% increase in seller concentration on
profit rates, holding all other explanatory variables constant
(i.e., “ceteris paribus”)
3–3
•
S tru c tu re -C o n d u c t-P e rfo rm a n c e
Model Specification In d u s try S tru c tu re

• N u m b e r o f b u y e rs
a n d s e lle r s
F ir m C o n d u c t
•
•
P r ic in g
A d v e r t is in g
P e rfo rm a n c e
• E c o n p r o f it s
• A c c o u n t in g
• D e g re e o f p ro d u c t • R & D p r o f it s ( r a t io s )
d if f e r e n t ia t io n • I n v e s t m e n t in
• B a r r ie r s t o e n t r y p la n t a n d • N P V /D C F
• C o s t s tru c tu re s e q u ip m e n t
• V e r t ic a l in t e g r a t io n • M V A /E V A
Variable Predicted Sign Reason

• A llia n c e s • T o b in ’s Q
CR4 + Higher concentration

enables higher prices
CAP/S + Capital-cost barrier to entry
A/S + Advertising intensity as a
product differentiation
barrier to entry
R&D/S + Technological know-how
GROW + Demand growth leads to
less likely price wars
3–4
 Note that the multiple regression results are consistent with

(but do not prove!) the structure-conduct-performance model.
 As you probably are aware from your statistics classes, there

are many potential problems that can interfere with the
reliable estimation of regression models, leading to incorrect
inference about the statistical significance and economic
importance of explanatory variables.
(1) Mis-specification Problems:
 Important Variables Omitted. In our regression, the impact

of substitute products, and the power of buyers and suppliers
have not been included in the model specification.
 Irrelevant Variables Included. If you believe in “perfect

capital markets” then you may question the idea of capital
cost entry barriers and therefore you would question the
inclusion of the independent variable [CAP/S] in the model.
(1) Mis-specification Problems:
 Model assumes a linear relationship. Since the regression

assumes a linear relationship, this may turn out to be a poor
approximation if some of the explanatory variables (e.g., ADV/S)
influence the dependent variable (i.e., ROE) in a non-linear way.
 Independent variable may not be truly independent.

For example, not only can increased concentration affect
profit rates but profit rates may affect industry
concentration.
 Multicollinearity. If independent variables such as

(ADV/S) and {R&D/S) are highly correlated, then the
validity of the t-statistics come into question.
(2) Measurement Problems:
• For example, CR4 may not be the best measure of

industry concentration, where the HHI is a better measure.
Perhaps some performance measure other than ROE
would also be better for testing the theory.
Note: If the evidence is not consistent with the theory it is not

necessarily the case that we abandon the theory. One of the
many possibilities is that we do not have good measures of the
theoretical concepts.
(3) Identification Problems:
- These problems are related to the idea that “correlation

does not imply causality.”
 For example, you might maintain that high

advertising/sales is a barrier to entry (product
differentiation) strategy that causes high profit rates.
The regression is consistent with Porter’s (1980) theory.
3-9
(3)
(3 Identification Problems
 However, one might argue instead that high profit rates

allow more discretionary spending in marketing and thus,
high profit rates cause high advertising/sales. The empirical
evidence is also consistent with this theory. Thus, we have
an “identification problem.” The data are consistent with
multiple theories and we must find more refined tests and
better econometric methods in order to advance our scientific
knowledge in strategic management.

Model Specification

Uploaded by

Copyright:

Available Formats

Model Specification

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Model Specification

Uploaded by

Copyright:

Available Formats

Empirical Testing of Structure-Conduct (Strategy)- Performance

 In practice, researchers estimate a statistical model of

 Industry Profit Rates = f (Concentration, Barriers to

 Multiple regression analysis seeks to evaluate the degrees

 The nature of this association is captured by regression

Model Specification In d u s try S tru c tu re

Variable Predicted Sign Reason

CR4 + Higher concentration

 Note that the multiple regression results are consistent with

 As you probably are aware from your statistics classes, there

(1) Mis-specification Problems:

 Important Variables Omitted. In our regression, the impact

 Irrelevant Variables Included. If you believe in “perfect

(1) Mis-specification Problems:

 Model assumes a linear relationship. Since the regression

 Independent variable may not be truly independent.

 Multicollinearity. If independent variables such as

(2) Measurement Problems:

• For example, CR4 may not be the best measure of

Note: If the evidence is not consistent with the theory it is not

(3) Identification Problems:

- These problems are related to the idea that “correlation

 For example, you might maintain that high

 However, one might argue instead that high profit rates

You might also like