SlideShare a Scribd company logo
1 of 14
Download to read offline
International Journal of Forecasting 10 (1994) 339-352
Strategic marketing forecasting, market segment selection
firm performance *
Noel Capon*, Peter Palij
Graduate School of Business, Columbia University, New York, NY 10027. USA
and
Abstract
In this study we investigate several hypotheses relating strategic forecasting to market segment selection and firm
performance. In the context of a strategic marketing simulation, subjects in 14 competitive industries made
strategic forecasts for market segment size and benchmark prices. Our results show that firms differentially select
those segments with attractive characteristics; that some strategic forecasts for these targeted segments are more
accurate than for non-targeted segments; that strategic forecasts are more accurate the higher the level of
competition; and that superior forecasting performance is positively associated with superior firm performance.
Implications and limitations of the study are discussed.
Keywords: Strategic marketing forecasting; Market forecasting; Market segment selection; MARKSTRAT
1. Introduction
Perhaps the most critical decision that senior
managers in multi-product, multi-market firms
must make is the choice of which products to
produce and which markets to address. In the
197Os, several strategic planning models were
developed to assist management in addressing
this problem. Although not universally accepted,
by the early 1990s these models and/or their
successors were woven into the fabric of
* Corresponding author. Tel: (212) 854-3466 or (212) 865
5332; Fax: (212) 749-3549.
‘“’Preliminary results from the non-degree executive pro-
gram subjects were presented at the November, 1991,
ORSAiTIMS Joint National Meeting.
strategic management at many major corpora-
tions (Capon et al., 1987; Greenley and Bayus,
1993).
The basic purpose of strategic planning models
is to provide managers with a coherent frame-
work to structure long-term planning. Although
strategic planning models display idiosyncratic
differences, they are typically presented as two
dimensional matrices in which the axes represent
market attractiveness and business strengths.
The models differ from each other in the manner
in which these dimensions are operationalized.
For example, in the GE/McKinsey screen, Ar-
thur D. Little and BCG portfolio models, market
attractiveness is represented respectively as a
weighted linear combination of such variables as
market growth rate, profitability, number of
competitors, and degree of regulation; as prod-
0169-2070/94/$07.00 @ 1994 Elsevier Science B.V. All rights reserved
SSDI 0169-2070(94)09007-R
uct life cycle stage; and as the forecast long-run
market growth rate (Wind, 1981).
Across these two dimensions, each model uses
a mix of measurable current variables, and
medium- and long-term forecasts. The appro-
priate values for current variables may be dcbat-
able, but the forecast variables represent a far
more difficult problem. Indeed, Capon and Hul-
bert (1985) lament the fact that virtually all
forecasting research at the firm level has ad-
dressed short-term forecasting (e.g. sales fore-
casts for production planning and inventory
management; short-term revenue forecasts for
cash Now management).
More recently, several researchers have
attempted to address firm-level forecasting issues
by using a strategic marketing simulation meth-
odology (LarreLhC and Gatignon, 1990). For
example, Glazer et al. (1989, 1990) examined
how closely short-term forecasts based on mana-
gerial judgment conformed to the formal prop-
erties of rationality. They found that one-year
forecasts for aggregate market size, average
industry price and number of marketed products
in four ‘industries’ were efficient but biased.
However, not only were these studies concerned
with short-term forecasting, the forecast vari-
ables were not particularly strategic. By contrast,
the study reported in this paper, using a similar
methodology, focuses specifically on strategic
forecasting issues.
Capon and Huibert (1985) coin the term
‘strategic forecasting’ to embrace those types of
long-term forecasts that are strategic for the
firm. Important dimensions requiring such fore-
casts include customer needs, competitive struc-
ture, competitive action, degree of regulation
’ Rclatcd arcas are role playing (Armstrong, 1988) and
new product forecasting (Mahajan and Wind, 1988). Role
playing research focuses on developing tools to overcome
managerial biases in forecasting. Most new product forecast-
ing research has avoided firm-level strategic forecasting
issues; segment ideIlti~c~~tion and sclcction arc typically
taken as given, and forecasts focus on prospects and the
optimal marketing mix for new products.
and market and/or market segment growth rates
(Wind, 1981; Wind and Robertson, 1983; Arm-
strong et al., 1987).’
The forecasting literature has generally taken
a pessimistic view of such medium- and long-
term forecasts, based primarily on their reliance
on managerial judgment (Armstrong, 1986). For
example, although Hogarth and Makridakis
(1981) recognize the importance of judgment for
these forecasts, they list over two dozen poten-
tially dist~~rting biases; and Paich and Sterman
(1993) find that even when conditions are delib-
erately designed to increase forecasting accuracy,
managers often fail to overcome judgmental
biases. In addition, Chakravarti et al. (1981) take
a pessimistic view of the value of managerial
judgment for marketing decisions in the context
of marketing decision models.’
The expert literature reaches a similarly pes-
simistic conclusion regarding the predictive abili-
ty of experts. Beginning with Thorndike’s (1918)
study of officer candidacy selection by military
experts, a long series of studies has found
.I
. . [that] expert judgment {a) is seldom better
than that of novices and (b) is easily improved
upon when replaced with simple equations.”
(Hammond, 1987, p. 3).
Hogarth and Makridakis recommend that long-
term forecasting focus on identifying sources of
environmental uncertainty, deliberately adopt
ambiguous rather than precise goals, rely on
simple time series models whenever possible,
and generally avoid complexity. Makridakis
(1986) echoes Hogarth and ~akridakis’ recom-
mendations while emphasizing the complemen-
tary roles of judgmental and quantitative fore-
casts.
Long-term forecasting research has focused on
macroeconomic variables in areas such as busi-
ness cycles, energy, population and transporta-
tion (Ascher, 1978; Adams, 1986); medium-term
forecasting research has focused on financial
‘See .Little and Lodish (ISSI) for a response on the
relative value of statistically based and judgmental models for
decision making in marketing.
N. Capon. P. Palij I International Journal of Forecasting 10 (1994) 339-352 341
analyses and capital allocation decisions (see nexus of customers, competitors and tech-
Hogarth and Makridakis, 1981). Microeconomic nologies, the marketing strategist must segment
research has emphasized industry-level forecasts the market, select those segments in which the
as inputs to national-level macroeconomic fore- firm should compete, and discard others (Hul-
casts (Adams, 1986, pp. 161-165). Little work bert, 1985, pp. 30-31). Segment selection, and
has addressed such firm-level strategic forecast- positioning decisions within the segment, repre-
ing problems as market or market segment sent the bridge between the SBU strategy and
growth rates. Marketing research has stressed tactical marketing actions that the firm executes
product forecasts at the industry level. leaving in the market place. It is at the market segment
firm-level strategic and market segment forecasts level that critical resource allocation decisions
needing extensive exploration (Armstrong et al., are made for securing customers at the expense
1987). In a rare exception, Cortes-Rello and of competitors. The ability to predict key dimen-
Golshani (1990) use an expert system meth- sions of available market segments over the
odology to predict 10 year market demand for a medium and long term is a critical strategic
capital good. necessity.
Based on the literature, an awkward situation
arises. Strategic planning models are the princi-
pal tools available for assisting senior managers
in making strategic choices, and these models
use as inputs strategic forecasts of key variables.
However, these strategic forecasts, typically
based on managerial judgment, are generally
viewed as frequently inaccurate and biased.
Furthermore, little research has addressed the
question of improving such forecasts.
In this paper we develop several hypotheses
relating to the use of strategic planning models,
the relationship between strategic forecast ac-
curacy and market segment selection, and be-
tween strategic forecast accuracy and firm per-
formance. Next, we test these hypotheses using a
large sample of strategic market forecasts se-
cured in the context of the MARKSTRAT
simulation (Larredhe and Gatignon, 1990). Final-
ly, we discuss the results, explore some limita-
tions and make recommendations for future
research.
Strategic planning models provide a useful
framework for the marketing strategist to struc-
ture the set of decision options for selecting
those market segments in which the firm will
compete. Using the market attractiveness, busi-
ness strengths paradigm, the received wisdom of
strategic planning models states that the strateg-
ist should strive to identify, then select for effort,
those market segment options that display high
levels of market attractiveness, where the firm
also has substantial business strengths (Kotler
1984, p. 61; Day, 1990, p. 200).
Hl: The firm’s selection of market segments to
address is positively related to their market
attractiveness, and to the firm’s current business
strengths in those market segments.
2. Development of hypotheses
2.1. Hypothesis 1
Perhaps the broadest use of strategic planning
models is at the strategic marketing level. In the
context of a strategic business unit (SBU)
strategy that broadly defines the competitive
In order to test this hypothesis, we focus on
the product portfolio matrix developed by the
Boston Consulting Group (BCG). In this
strategic planning model, market attractiveness
is defined as the expected market segment
growth rate; we use two measures-the medium-
term market segment size, and the annualized
growth rate needed to reach the expected
medium-term market segment size from the
current size. The BCG approach operationalizes
business strength in the market segment as the
firm’s current relative market share (RMS), and
directly links RMS to profitability (Henderson
and Zakon, 1980; Day, 1986). The resource
342 N. Capon, P. Palij I International Journal of Forecasting IO (1994) 339-3.52
allocation actions taken by the firm in the follow-
ing periods are used to indicate management’s
decision to focus on the segment.”
The BCG approach to strategic planning has
received mixed reviews. Several writers view the
model positively (e.g. Dutton et al., 1983;
Burke, 1984) and some empirical evidence sug-
gests that it is widely used in industry (e.g.
Capon et al., 1987). However, it has fallen out of
favor among academics (e.g. Wensley 1982;
Seeger 1984) largely because of its reliance on
the experience curve and the strongly implied,
and much debated, linkage between market
share and profitability (e.g. Jacobson, 1988).
However, in the context of the MARKSTRAT
simulation, use of this framework is quite appro-
priate as strong experience curve effects are
embedded in both the production, and research
and development functions. Consequently, the
correlation between market share and profitabili-
ty is high, 0.89 for the industries in this study.
2.2. Hypotheses 2 and 3
Broadly speaking, three perspectives govern
the strategic forecasts required by strategic plan-
ning models. Espousing a normative view, tradi-
tional neoclassical economic models assume ‘hy-
perrational’ behavior (Simon, 1978): managers
are sequential, exhaustive and optimal analysts
of all available data. Managers carefully forecast
each relevant variable. in each market segment,
in each period, in order to exploit opportunities
as competitive dynamics and other conditions
change within segments. For such forecasters,
accuracy should be independent of factors such
as market segment participation and competitive
intensity.
In direct contrast, empirical forecasting evi-
’ Note that sales volume in a market segment ex post does
not imply that the firm has made the strategic decision to
target that segment for effort ex ante. for a variety of reasons
including inadequate segment specification, improper im-
plementation and variety seeking by customers, firms may
secure significant market share in non-targeted market seg-
ments
dence indicates little hope for any accuracy in
making such strategic forecasts. As noted previ-
ously, these forecasts are highly susceptible to
bias (Hogarth and Makridakis, 1981); the ac-
tion-outcome-feedback loop is weak, typically
stretching beyond a product development cycle;
relatively precise values are required for effec-
tive model use; and relevant variables are affect-
ed by complex, diverse and often unknown
forces. Since any accurate forecasts are seren-
dipitous, accuracy should be independent of
factors such as market segment participation and
competitive intensity.
Rather than adopting either polar perspective,
the traditional neoclassical economic normative
approach or the pessimism of extant empirical
studies, each of which leads to similar predic-
tions regarding forecast accuracy, we choose to
work within the bounded rationality paradigm
(Simon, 1957, 1979). Bounded rationality implies
an intermediate position on forecast accuracy.
Forecasters are unable to cope with the mag-
nitude of processing tasks required for hyper-
rationality and so use processing short-cuts such
as heuristics and schemata. However, as a prob-
lem’s salience increases, a variety of active
strategies are adopted, including those hypoth-
esized by the neoclassical perspective
(Eisenhardt and Zbaracki, 1992). Thus, as re-
sources (e.g. marketing research. product de-
velopment. production, advertising and promo-
tion) are focused on a market segment, its
increased salience leads forecasters to pay in-
creasing attention to strategic variables relevant
for that segment. Ceteris paribus, increased
attention should lead to more accurate forecasts
so that the bounded rationality perspective leads
directly to Hypothesis 2.
H2: The firm’s forecasts of strategic dimensions
for target market segments are more accurate
than for avoided segments.
The notion of salience relates both to those
segments that are targeted for effort, and to the
degree of difficulty in competing in those seg-
ments. Thus, we anticipate that, ceteris paribus,
N. Capon. P. Palij I International Journal of Forecasting 10 (1994) 339-352 343
increased effort, and therefore increased accura-
cy, should be related to increased levels of
competition in targeted segments. Increased ef-
fort under these conditions is consistent with
anecdotal evidence from managers that effort on
problems typically exceeds effort on oppor-
tunities.
H3: The accuracy of the firm’s strategic forecasts
in target market segments increases as the level
of competition increases.
2.3. Hypothesis 4
Strategic planning advocates posit a positive
relationship between the use of strategic plan-
ning models and firm performance. This positive
view of the influence of managers and strategic
planning on firm performance has been chal-
lenged by research focusing on the importance of
internal and external environmental factors (e.g.
Hannan and Freeman, 1977; Pfeffer and Salan-
cik, 1978; Fredrickson and Mitchell, 1984). Em-
pirical findings are mixed, but generally suppor-
tive of a positive link between strategic planning
and firm performance (Armstrong, 1982; Capon
et al., 1987, 1994; Boyd, 1991).
As discussed above, good strategic planning
requires the ability to make good strategic fore-
casts. Indeed, in Urban et al.‘s (1990) strategic
planning model for the automotive industry, nine
of 24 of the model’s parameters were substantial-
ly or entirely based on judgmental forecasts
made by managers. Clearly, the success of such
formal models in providing useful input to deci-
sion making depends upon the accuracy of the
underlying parameters. However, regardless of
forecasting accuracy, many strategic variables
such as level of diversification and product line
maturity are also expected to impact firm per-
formance. Thus, we expect a weak positive link
between forecast accuracy and performance.
H4: The greater the accuracy of the firm’s long-
run strategic forecasts, the better its perform-
ance.
3. Method
3.1. Direct experiment
Methodologies available for testing the hy-
potheses range from laboratory experiments to
field studies. Laboratory studies have the benefit
of control, but are seriously challenged by topics
requiring a series of complex decisions evolving
over time (Singer and Brodie, 1990). Field
studies capture complexity but typically at the
cost of lack of control. Simulations conducted as
direct experiments (Sterman, 1987) offer an
intermediate technique for testing strategic fore-
casting hypotheses related to complex decisions
(Glazer et al., 1990; Green and Ryans, 1990).
The MARKSTRAT 2 simulation (Lar-
reche and Gatignon, 1990) provides a realistic
environment in which participants seek long-run
success through a series of strategic and tactical
marketing decisions, moderated by feedback on
the results of their decisions (Kinnear and Klam-
mer, 1987). Five firms compete in an established
market (sonites) comprising five well-defined
market segments, and in a second undifferen-
tiated independent market (vodites) that is un-
tapped at the beginning of the simulation. Each
firm starts the simulation with a differing mix of
corporate strengths and weaknesses (some firms
are stronger, some firms are weaker), and with
two sonite products in different competitive
positions; each firm may offer up to five prod-
ucts. For each cycle of the simulation, a team
managing its firm assesses the results of prior
decisions, evaluates marketing research data,
and allocates its limited budget to R&D, market-
ing mix expenditures and marketing research.
The firm decisions are run, and results returned
for the next decision cycle. Firm performance
(e.g. sales, profit, market share) is a function of
firm and competitor decisions, individual firm
capabilities including budgets, and the evolving
MARKSTRAT economy. Initial firm parame-
ters, underlying economic conditions for each
industry of five firms, and the general pattern of
segment growth and preferences are set identi-
cally at the start of each simulation.
MARKSTRAT focuses on short- and medium-
term decisions; long-term decisions such as aban-
doning both sonite and vodite markets, investing
in new technologies, or adding new distribution
channels are not available. Short-term decisions
are set by the ‘annual’ decision cycle. Medium-
term decisions reflect changes in resource alloca-
tions among market segments, and investments
in product development. New product entry into
a market segment implies a 2-4 year product
development cycle: l-2 years to develop a prod-
uct, 1 year for iIltroduction and feedback on
market segment acceptance, and 1 year for
product modification.
3.2. Subjects
Subjects were participants in full-time MBA,
executive MBA and non-degree executive pro-
grams conducted by a leading business school.
Full-time MBA students, representing a range of
business experience, were participating in a
second-year elective course in marketing
strategy; executive MBA students. participating
in a required advanced course in marketing
strategy, were mid-level executives (median age
33) attending classes once per week under firm
sponsorship; non-degree students were senior
marketing executives participating in a l-week
residential advanced program in strategic mar-
keting.
3.3. Procedure
The approximately 250 subjects were placed in
70 three to five person groups. (Full-time MBA
groups were self selected; the others were ran-
domly assigned.) The 70 groups comprised 14
industries of five teams (firms): four full-time
MBA, six executive MBA, and four non-degree
students. Data were collected as part of required
classroom activities. For both MBA groups, the
eight decisions were made weekly during the
semester; the non-degree students made seven
decisions at a rate of two decisions per day.
At the start of the simulation. each firm
received a summary of results for the prior year
(zero), a complete set of aggregate market re-
search data, and a decision support package
comprising a variety of spreadsheet tools to assist
in analyzing firm and marketing research data
(James et al., 1991). After a decision was run,
each team received its firm report and the
specific marketing research data it had re-
quested; use of the decision support software
was optional. Upon completion of the simula-
tion, each team was required to make formal
oral (to the class) presentations explaining ac-
tions taken, perceptions of industry events and
lessons learned. Such presentations by teams in
previously run industries provided pretest guid-
ance.
3.4. Variables
3.4.1. Strategic marketing variables
Informal discussions with previous MARK-
STRAT participants revealed a large and com-
plex array of variables perceived as strategic:
these included internal capabilities, budget allo-
cations, size and growth of the market segments,
competitive conditions in each market segment
and changes in consumer preferences.
We selected as strategic variables both market
segment size (units) and market segment growth
rate, each of which addresses the market attrac-
tiveness dimension of strategic planning models.
In addition, we selected market segment leader’s
price. Price plays an important strategic role in
MARKSTRAT. First, it provides a measure of
profit potential, given the firm’s Iikely product-
ion cost. Second, it captures customer prefer-
ences by playing a direct role as a product
positioning attribute in the sonite market, and as
an indirect positioning attribute for vodites.
3.4.2. Forecast market segment size
This variable is a forecast of the number of
units sold into the segment by all competitors 3
years ahead.
3.4.3. Forecast market segment growth rate
This variable is the compound annual growth
rate required to reach the forecast market seg-
ment size from the current market segment size.
This variable is the forecast price, 3 years
ahead, of the product having the largest market
share in the market segment,
3.4.5. Firm performance
Direct measures of firm performance in
MARKSTRAT are aggregated across products
and segments. Several candidate variables were
examined: sales units, sales revenues, unit mar-
ket share, revenue market share, gross profit
share and net profit. The smallest of the 15
correlations between these variables was 0.88.
Revenue market share was selected because it
both permits identical treatment of the different
sonite and vodite markets, and normalizes firm
results across the sample industries.
3.4. (5;.Market attractiveness
This segment-level variable was oper-
ationalized in two ways: as the forecast market
segment size and as the implicit forecast market
segment growth rate (see above), The formal
debriefs presented by previous MARKSTRAT
teams indicated that participants generally as-
sessed market attractiveness by estimating the
future size of a market segment, then checking
to see if the growth rate seemed reasonable.
3.4.7. Business strength
This segment-level variable was oper-
ationalized as the firm’s current relative market
share. Relative market share is the firm’s market
share (of all products in a segment), divided by
the market share of the segment leader (or by
the number two share firm if the focal firm was
the segment leader).
.3.4.8. Target market segment
In the sonite market, the five market segments
are defined as a set of ideal points in a three
dimensional positioning map whose axes are
perceived price, perceived power and perceived
design (the latter two are physical attributes).
Two methods assess the target segments of
current products-product design characteristics
and advertising positioning; target segments for
future products are identified from the design
characteristics of active R&D projects. In each
case, the minimum Euclidian distance from the
product to each of the five ideal points is defined
as the market segment targeted.
The three methods vary by time horizon;
product design characteristics and ideal points
are measured contemporaneously; the advertis-
ing positioning decision in period t is compared
with the revealed actual ideal points in period
t + 1; the R&D design characteristics in period t
are compared with the actual ideal points in
period t t 3. For the undi~erentiated vodite
market, any product entry, or R&D project,
respectively, defines a targeted market.
The measure of target market segment em-
ployed for any decision period was the union of
these three measures under the rationale that for
any target market segment, good strategic fore-
casting is a critical matter.
3.4.9. Competitive activity
This variable is defined by adapting the Jac-
quemin and Berry (1979) entropy measure of
diversification. Level of competition is defined
as:
where Pi is each firm’s market share in the
market segment; the larger the index, the greater
the level of competitive activity. This measure is
sensitive to the number of competitors, the
distribution of market shares among com-
petitors, and the mix of cfose and distant com-
petitors (Palepu, 1085).
Tradition&y, competitive-type measures have
focused on market concentration (e.g. concen-
tration ratios-total market shares of the n
largest firms; Herfindahl’s index, H-sum of the
squared market shares for each firm) as in-
dicators of public policy concern, rather than on
competitiveness per se. For public policy analy-
sis, Stigler (1983) recommends the Eferfindahl
index over entropy measures, based in part on its
insensitivity to firms with small market shares.
However, since the MARKSTRAT enQiro~ment
comprises a fixed number of firms, each with the
336 N. Capon. P. Palij I International Journal of Forecasting 10 (1994) .U9-352
ability to offer multiple products to multiple
segments, the dimensions captured by the en-
tropy measure seemed more appropriate as in-
dicators of a segment’s competitiveness.
3.4.10. Forecast accuracy
Several indices of forecasting accuracy are
available. However, since market segment sizes
varied by an order of magnitude, we rejected
both the simple error and mean square error
measures (Mahmoud, 19S6, pp. 509-511). In-
stead, we selected the absolute percent error
(APE) and mean absolute percent error
(MAPE). These measures give equal weight to
over- and under-prediction, reflecting the
strategic issue that for market segment size and
growth rate, under-prediction may result in op-
portunity losses as, or even more, serious than
direct losses resulting from over-prediction.
MAPE was used to aggregate forecast accuracy
across segments.
3.5. Data
Data to test the hypotheses were drawn from
firm decisions, the results for each firm from
running the simulation, and 3-year forecasts of
unit size and leader’s price-by market segment
for sonites, by total market for vodites. Forecasts
for each sonite segment and the vodite market
were collected from each team. Forecast data
were collected at periods 0 through 5 together
with the firm decisions; the six forecast/actual
pairs comprised periods O/3; l/4; 21.5; 316; 417;
518 (five pairs for the seven period industries). In
total, 2400 forecast/actual pairs were collected
from the 14 industries.
3.6. Analysis
Ordinary least squares (OLS) and logistic
regression was used to test the hypothesized
relationships. Firm and type of student program
effects were entered as dummy covariates in
each analysis. Data were omitted in those in-
stances when students did not provide forecasts,
when actual sales in vodites were zero (even
though some groups provided forecasts) and
when the APE exceeded 200%. Less than 4% of
possible data were dropped, primarily in indus-
tries with late introductions of vodites.
For Hypothesis 1, the dependent variable was
the target market segment indicator (0, 1);
independent variables were, alternatively, fore-
cast market segment size and forecast market
segment growth rate together with relative mar-
ket segment share and the covariate dummies.
Two logistic regression models examined these
relationships.
Hypotheses 2 and 3 were tested simultaneous-
ly via two OLS regression analyses in which the
APES for forecast market segment size and
forecast market leader’s price provided alter-
native dependent variables. Independent vari-
ables were the target market segment indicator
(Hypothesis 2), level of competition and the
target market segment indicator X level of
competition interaction (Hypothesis 3). For each
regression analysis, other independent variables
were the covariate dummies, plus one-period
lagged dependent variables that were included to
account for potential autocorrelation effects.
OLS was also used to model Hypothesis 4.
The dependent variable was total firm sales in
units (across all segments) for each period.
Independent variables were the MAPE of fore-
cast market segment size, the MAPE of forecast
market segment leader’s price, a period indicator
that captured inherent trends in the MARK-
STRAT environment, and the firm and type of
student effect covariate dummies. Several addi-
tional covariates captured major strategic deci-
sions that might reasonably have been expected
to be related to firm performance: the number of
sonite market segments targeted and the pres-
ence/absence in vodites captured a diversifica-
tion dimension; the research and development
budget captured product line maturity; and the
number of salespeople, and budgets for advertis-
ing and market research captured other critical
strategic dimensions in the MARKSTRAT simu-
lation.
4. Results
Hl: The firm’s selection of market segments to
address is positively related to their market
N. Capon, P. Palij I International Journal of Forecasting 10 (1994) 339-352 347
attractiveness, and to the firm’s current business
strengths in those market segments.
This hypothesis is supported for both oper-
ationalizations of market attractiveness and for
business strength (see Table 1). The positive and
significant coefficients for forecast market seg-
ment size and forecast market segment growth
rate, and for relative market share in each of the
two analyses, indicate that firms acted according
to the received wisdom of strategic planning
models: participate in market segments provid-
ing large potential opportunities where the firm
is currently strong.
H2: The firm’s forecasts of strategic dimensions
for target market segments are more accurate
than for avoided segments.
This hypothesis was tested by two regression
analyses, one each for the APE of forecast
market segment size and forecast market seg-
ment leader’s price. Partial support was adduced
for this hypothesis inasmuch as the coefficient for
target market segment in the regression of fore-
cast market segment leader’s price was both
negative and significant (see Table 2). (The
coefficient for target market segment was not
significant for forecast market segment size.) The
negative sign indicates smaller forecast errors,
hence increased accuracy, for targeted market
segments.
Table 1
Relationship between market attractiveness, business
strengths and market segment selection
Independent variable
Forecast market segment size
Forecast market segment
growth rate
Relative market share
Firm 1
Firm 2
Firm 3
Firm 4
Full-time MBA students
Executive program participants
Constant
* pco.05; **p<o.o1.
Coefficients
0.00227**
1.272*”
1.590** 1.628**
-0.243 -0.253
-0.321 -0.294
-0.162 -0.199
-0.319 -0.336’
-0.183 -0.172
-0.159 -0.237
-0.205 0.559**
Table 2
Relationship between market segment selection and strategic
forecasting accuracy
Independent variable Coefficients
Forecast market Forecast market
segment segment
size leader’s price
Targeted market 0.286 -0.0722**
segment (TMS)
Level of competition
(LC)
-0.672** -0.0593**
TMS x LC
One year lagged size
One year lagged
leader’s price
-0.172 0.0241*
0.246”*
0.3so**
Firm 1
Firm 2
Firm 3
Firm 4
Full-time MBA
students
-0.220 0.0076
-0.0028 0.0177
-0.174 -0.0010
-0.553 0.0120
-0.204 -0.0157
Executive program
participants
0.0814 0.0107
Constant
F (10.1899)
Adjusted RZ
* P<O.O5; P<O.Ol.
1.643** 0.205**
40.08** 54.03**
0.17 0.22
H3: The accuracy of the firm’s strategic forecasts
in target market segments increases as the level
of competition increases.
This hypothesis was supported for forecast
market segment leader’s price inasmuch as the
coefficient for the target market segment x level
of competition interaction was significant (see
Table 2). However, the coefficient for the main
effect of level of competition was negative and
significant indicating that, regardless of whether
or not the segment was targeted, the accuracy of
the firm’s strategic forecasts improved as the
level of competition increased. The key factor
driving the interaction effect was the relatively
low level of forecast accuracy in low competi-
tion, non-targeted market segments. Forecast
accuracy was highest for high competition,
targeted market segments, significantly greater
than for both targeted, low competition and non-
targeted high competition segments; the latter
two segments were not significantly different
from one another.
The main effect for level of competition was
found also for forecast market segment size. The
significant coefficients for one year lagged vari-
ables in each regression, forecast market seg-
ment size and forecast market segment leader’s
price, indicate the presence of autocorrelation
that these variables removed.
H4: The greater the accuracy of the firm’s long-
run strategic forecasts. the better its perform-
ance.
This hypothesis received mixed support (see
Table 3). The MAPE for forecast market seg-
ment leader’s price was significant and negatively
related to performance, thus supporting the
hypothesized relationship, but the MAPE for
forecast market segment size was not significant.
As expected, the period was significant, as were
three of four firm dummies. The magnitude and
direction of the firm coefficients reflect the
typical rankings of firm performance; over a
large number of industries, firms 1, 3 and 4
typically outperform firms 2 and 5,
Table 3
Relationship between forecast accuracy and firm perform-
ancc
Independent variable
Forecast market segment size (MAPE)
Forecast market segment leader’s price
(MAPE)
Number of sonite segments
Vodites
Advertising hudget
R&D budget
Number of salespeople
Marketing research hudgct
Period
Firm 1
Firm 2
Firm 3
Firm 4
Full-time MBA students
Executive program participants
Constant
F (15,384)
Adjusted R’
* P < 0.05; P < 0.01.
Coefficients
0.012
--t).o665*
0.0074
0.0054
0.146**
0.0099**
0.0570**
-0.0003
-0.0240**
0.0467**
0.0093
0.0309**
0.0242*
0.0123
0.0145
-0.0306
63.6**
U.71
Three of the six strategic variables. research
and development budget reflecting product line
maturity, advertising budget and number of
salespeople were also significant. The presence
of the moderating variables in the equation did
not change the direction or significance of the
coefficient for forecast market segment leader’s
price, nor affect the significance of forecast
market segment size. The presence of these
variables was, however, important in raising the
R’ from 0.16 when they were not included, to
0.71.
The difference in results between forecast
market segment size and forecast market seg-
ment leader’s price when testing Hypotheses 2, 3
and 4 may reflect differences in forecast task
complexity. Thus, depending on the particular
segment and industry, overall segment size
changes for sonites through the simulation
ranged from - 10% to + 150%. and period-to-
period changes reached 100% on occasions.
~Period-to-period changes were even iarger for
vodites since sales were zero at the start of the
simulation.) By contrast, the market segment
leader’s price demonstrated far less variance, in
large part because the ideal point for price in
each market segment moves in a constrained
linear manner.
5. Discussion
In this paper we set out to investigate several
hypotheses related to strategic forecasting in the
context of strategic planning models. We argued
that for managers to make good strategic deci-
sions, they must be able to make good judg-
ments about the medium- and long-term en-
vironment. In a sense, our work is an extension
of Glazer et al.‘s (1989, 1990) important MARK-
STRAT studies that found a modest positive
relationship between forecast accuracy and firm
performance. However, the forecasts collected
by Glazer et al. were not particularly strategic.
They used short-term (1 year) forecasts com-
pared with our medium-term (3 year) forecasts
that reflect one or two product development
cycles. In addition, their three forecast variables
N. Capon, P. Palij I International Journal of Forecasting 10 (1994) 339-352 349
(market size, average industry price, and number
of marketed products) are not particularly re-
lated to key strategic decisions at the market
segment level, the focus of the MARKSTRAT
simulation. By contrast, our forecast variables of
market segment size and growth rate, and mar-
ket segment leader’s price, are highly strategic.
Furthermore, the majority of our subjects were
practising managers versus their exclusive re-
liance on MBA students; and, finally, our find-
ings are based on 14 industries whereas their
data is drawn from just four industries.
Despite the inherent difficulties in forecasting
beyond a development cycle, our hypotheses
were largely, though weakly, supported. In the
context of portfolio models of strategic planning,
we found that firms tended to select those
market segments that provided the greatest op-
portunities in terms of fast growth and high
potential size, where the firm also had substan-
tial business strengths as indicated by its relative
market share (Hypothesis 1). The positive impli-
cation from this finding is that portfolio dimen-
sions are relevant to market segment selection;
the negative implication is that misperceptions
about market segment size and growth rate may
lead to poor segment selection and risk the
problems associated with escalation of commit-
ment (Brockner et al., 1986).
Our results also indicate that the differentially
polar perspectives taken by neoclassical econ-
omics (positive) and traditional forecasting re-
search (negative) regarding long-term forecasting
accuracy may be premature. Consistent with a
bounded rationality perspective that managers
cannot evaluate each segment in each period
with the same degree of effort, we find that to
some extent salience of the forecasting context
impacts forecast accuracy. Thus, subjects were
better able to forecast market segment leader’s
price in those market segments to which they
had committed resources (Hypothesis 2). Such
differential forecast accuracy, as with misperceiv-
ing market segment potentials, risks a focus on
market segments to which prior commitments
have been made, thus raising the spectre of an
escalating commitment cycle with potentially
devastating results. Indeed, as managerial atten-
tion focuses on current market segments, oppor-
tunities further afield may be missed.
We also found that, consistent with our per-
spective that salience increases forecasting ac-
curacy, forecast accuracy for market segment
leader’s price was greatest in those market seg-
ments that were both targeted, and in which the
firm faced high levels of competition. An un-
expected result was the finding that managers
were better able to forecast both market segment
size and market segment leader’s price the more
competitively intense the market segment, re-
gardless of whether the segment was targeted.
One explanation for these results is that differ-
ing market segment profiles lead to differing
salience levels for managers. Targeted competi-
tive segments command the most attention
because committed resources are at high risk. By
contrast, non-targeted low competitive segments
command the least attention not only because
the firm has not committed resources but
because another firm (or firms) dominance
makes it unlikely that the firm will enter in the
future. Intermediate salience levels result from
either committed resources, but low risk because
of low competition, or in high competition seg-
ments, indicating opportunities for entry, that
are currently non-targeted. Such an explanation
accounts for both the main effect for level of
competition and for the interaction effect for
forecast market segment leader’s price.
Finally, we secured partial support for a posi-
tive relationship between strategic forecast ac-
curacy and firm performance. This result is
consistent with the generally positive, but mixed,
results on the relationship between strategic
planning and firm performance (e.g. Capon et
al., 1987, 1994). Briefly stated, this result dem-
onstrates that effort spent on the forecasting
dimension of strategic planning pays off.
An important issue in testing performance
relationships such as these is model specification
error. Thus, in his meta-analytic review of the
planning-performance literature, Boyd (1991)
highlights the absence of moderating variables as
one of six major limitations of these studies. In
an attempt to deal with this problem, Capon et
al. (1987) employed four moderating variables in
350 N. Capon, P. Palij i International Journal of Forecasting 10 (1 YW) .33Y-3.52
their study of planning-performance relation-
ships in the Fortune500(scale, degree of diversi-
fication, product line maturity and industry). In
this study, industry is an irrelevant variable as all
firms operate in the same industry. However,
firm size was accounted for by the firm and
period dummies; degree of diversification by the
number of sonite segments targeted and pres-
ence/absence of vodite variables; and product
line maturity by the research and development
budget. In addition, we captured the other major
strategic variables in the MARKSTRAT simula-
tion, namely, number of salespeople, and the
advertising and marketing research budgets.
Additional methodological issues relate to the
five other major criticisms of planning-perform-
ance studies were noted by Boyd. Notwithstand-
ing its simulation basis, compared with real
world forecasting, this study did not use small
and non-representative samples; collected data
directly from the principals competing in the
simulation rather than from surrogates; mea-
sured forecasting error as a ratio-scaled variable
compared with categorical or simple ordinal
variables; used time series versus cross-sectional
analyses; and used multiple forecasting in-
dicators.
Two important methodological issues are
raised regarding the use of MARKSTRAT simu-
lations: the need to account for firm effects and
the use of student subjects. First, the frequent,
and in the case of performance-related issues
highly significant, presence of firm effects sug-
gests that we should critically examine results
from MARKSTRAT simulations that do not test
for such effects (Ross, 1987). However, the
typical criticism of using MBA subjects in studies
such as these is somewhat mitigated by our
results. In none of the five analyses (regression
and logistic regression) was any student program
variable significant. In other words, there is no
evidence that full-time MBA students behaved
any differently from the two executive groups.
Perhaps the reason for this finding is the increas-
ing maturity of MBA students who in this study
averaged 27 years of age and had significant
industrial experience.
Notwithstanding the relatively positive results
from this study, several limitations are evident.
First, even though the MARKSTRAT simula-
tion provides a realistic simulation of the real
world, it is only a simulation. We cannot make
the leaps to real world practice; real world
results must await well-designed field studies.
Second, even though we were able to support
several hypotheses relating to strategic forecast-
ing, we did not investigate the methods by which
these strategic forecasts were made. Subjects in
the study could have used a variety of different
methods in the general categories of judgment,
counting, time-series and association (Georgoff
and Murdick, 1986). Future work should investi-
gate the effects of different ways of making
strategic forecasts. In addition, echoing Paich
and Sterman (1993), methods for measuring the
intrinsic difficulty of forecasting across market
segments are needed. Such effort would also
help address the need for a dynamic measure of
market competitiveness.
Third, the MARKSTRAT environment is rel-
atively benign. The product life cycles for sonites
and vodites are fairly predictable and, in general,
market segments evolve in an orderly manner;
major disturbances are driven by the activity of
direct competitors. These conditions correspond
to ‘convergent’ periods in Tushman and
Romanelli’s (1985) punctuated equilibrium
model of organizational development. Tushman
and Romanelli also hypothesize that firms enter
periods of ‘reorientation’ (such as IBM and
General Motors entered in 1992), driven by
sustained low performance and/or rapid environ-
mental changes, which result in discontinuous
and radical shifts in core values and beliefs, firm
mission and strategies, and in internal organiza-
tion and management processes. In their model,
organizations progress through convergent
periods, punctuated by reorientations that de-
mark and set bearings for the next convergent
period.
This study is essentially conducted during
periods of convergence where environmental
change is incremental in nature and there is
considerable opportunity for feedback learning.
We did not study periods of reorientation, driven
by major environmental shifts and shocks, where
N. Capon, P. Palij I International Journal of Forecasting IO (19%) 339-352 351
competitive challenges come from new direct
entrants, indirect competitors, and suppliers and
customers engaged in vertical integration strate-
gies. The problems of strategic forecasting in
periods of reorientation, with its limited possi-
bilities for feedback learning, are much more
difficult. However, arguably the payoff to good
strategic forecasting in these periods is at its
greatest.
Acknowledgments
The authors would like to thank Don
Lehmann, Bari Harlam and two reviewers for
comments on earlier drafts of this paper.
References
Adams, F.G., 1986, The Business Forecasting Revolution:
Nutiorz-intiustry-Firm (Oxford University Press, New
York).
Armstr~~ng, J.S., IYXZ, The value of formal planning for
strategic decisions: Review of empirical research. Sfraregic
Manugement Journal. 3, 197-211.
Armstrong. J.S., 1986, Research in forecasting: A quarter-
century review, 1960-1984, Interfaces, 16, 89-109.
Armstrong, J.S.. lY88, Research needs in forecasting, Inter-
national Journal of Forecasting. 4, 449-465.
Armstrong, J.S., R.J. Brodie and S.H. McIntyre. 1987,
Forecasting methods for marketing, International Journal
of Forecasting, 3, 355-375.
Ascher, W., 1978, Forecasting: An Appraisal for Policy
Makers und f1anner.s (Johns Hopkins University Press,
Baltimore, MD).
Boyd, R.K., 1991, Strategic planning and financial perform-
ance: A mcta-analytic review, ~oi~r~lal of management
Studies. 28. 353-374.
Brockner. J.. R. Homer. G. Birnbaum. K. Lloyd, J. Deit-
cher, S. Nathanson and J.Z. Rubin, iY86. Escalation of
commitment to an ineffective course of action: The effect
of feedback having negative implications for self-identity,
Administrative Science Quarterly, 31. 109-126.
Burke, M.C.. 1984, Strategic choice and marketing mana-
gers: An examination of business-level marketing objec-
tives, Journal of Marketing Research, 21. 345-359.
Capon, N. and J.M. Hulbert. 1985, The integration of
forecasting and strategic planning, International Journal of
Forecasting, 1. 123-133.
Capon, N., J.U. Farley and J.M. Hulbert, 1987, Corporate
Strategic Planning (Columbia University Press, New
York).
Capon. N., J.U. Farley and J.M. Hulbert, 1994, Strategic
planning and firm performance: More evidence. Journal of
Management Studies, 31, 105-110.
Chakravarti. D., A. Mitchell and R. Staelin, 1981, Judgment
based marketing decision models: Problems and possible
solutions, Journal of Marketing, 4.5, 13-23.
Cortes-Rello, E. and F. Golshani, 1990, Uncertain reasoning
using the Depster-Shafer method: An application in fore-
casting and marketing management, Expert Systems, 7,
Y-17.
Day, G.S., 1986. Analysis for Strategic Market Decisions
(West Publishing, St. Paul. MN).
Day, G.S., 1990. Market Driven Strategy: Processes for
Creating Value (Free Press. New York).
Dutton, J.E.. L. Fahey and VK. Na~~nan, 1983, Toward
understanding strategic issue diagnosis, Strategic Manuge-
ment Journal, 4, 307-323.
Eisenhardt, K.M. and M.J. Zbaracki, 1992, Strategic deci-
sion making. Strategic Management Journal, 13, 17-37.
Fredrickson. J.W. and T.R. Mitchell, 1984, Strategic decision
processes: Comprehensiveness and performance in an
industry with an unstable environment, Academy of Man-
agement Journal, 27. 399-423.
Georgoff, D.M. and R.G. Murdick, 1986, Manager’s guide
to forecasting, Narvard Business Review. 64, 110-120.
Glazer, R., J.H. Steckel and R.S. Winer. 1989, The forma-
tion of key marketing variable expectations and their
impact of firm performance: Some experimental evidence,
Marketing Science. 8. 18-34.
Glazer, R., J.H. Steckel and R.S. Winer, 1990, Judgmental
forecasts in a competitive environment: Rational vs. adap-
tive expectations, International Journal of Forecasting, 6,
149-162.
Green, D.H. and A.B. Ryans, 1990, Entry strategies and
market performance: Causal modeling of business simula-
tion, Journal of Product Innovation Management, 7, 45-
58.
Greenley, G.E. and 3.L. Bayus, 1993. Marketing planning
decision making in U.K. and U.S. companies: An empiri-
cal comparative study, Journal of Marketing Management,
forthcoming.
Hammond, K.R., 1987, Toward a unified approach to the
study of expert judgment, in: J.L. Mumpower, 0. Renn,
L.D. Phillips and V.R.R. Uppuluri, eds., Expert Jl~dgment
and Expert S.vstems. NATO ASI Series, Volume F35,
(Springer-Vertag, Berlin) l-16.
Hannan, M.T. and J.H. Freeman, 1977, The population
ecology of organizations, American Journal of Sociology,
82, 929-964.
Henderson, B.D. and A.J. Zakon. 1980, Corporate growth
strategy: How to develop and implement it, in: K.J.
Albert, ed., Handbook of Business Problem Solving
(McGraw-Hill, New York).
Hogarth, R.M. and S. Makridakis. 1981, Forecasting and
planning: An evaluation, Managemenf Science, 27, 11%
138.
352 N. Capon. P. Palij I International Journal of Forecasting 10 (1Wd) 339-352
Hulbert, J.M., IYXS. Marketing: A Strutegic Perspective
(Impact Publishing Co., Katonah, NY).
Jacobson, R., 1988, Distinguishing among competing
theories of the market share effect. Journal of Marketing.
52. 6X-80.
Jacquemin. A.P. and C.H. Berry. 1Y7Y. Entropy measure of
diversification and corporate growth. Journal of Industrial
Economics. 27. 359-36’).
James, S.. T.C. Kinnear and M. Deighan, 19Yl. ISf supple-
ment to Markstrat Z: Player’s guide (Interpretive Software
Inc., Charlottesville, VA).
Kinnear. T.C. and S.K. Klammer, lY87, Management pcr-
spectives on Markstrat: The GE expcricnce and beyond,
Journal of Business Research. IS. 4YlLSOl.
Kotler. P.. 19X4. Marketing Management: Ana1ysi.y. Planning
and Control. Fifth Edition (Prentice-Hall. Englewood
Cliffs. NJ).
Larreehe. J. and H. Gatignon. 1990. MARKSTRAT 2.
(Scientific Press. Palo Alto. CA).
Little. J.D.C. and L.M. Lodish. 1981. Commentary on
Judgment Based Marketing Decision Models, Journal of
Marketing, 45, 24-29.
Mahmoud. E.. 19X6. The evaluation of forecasts, in: S.
Makridakis and S.C. Wheelwright. eds.. The Handbook of
Forecasring: A Manager’s Guide. Second Edition (John
Wiley & Sons. New York. NY). 504-516.
Mahwan, V. and Y. Wind. 1988. New product forecasting
models: directions for research and implementation, Inter-
national Journal of Forecasting. 4, 341-358.
Makridakis, S.. 19X6, The art and science of forecasting,
International Journal of Forecasting, 2. 15-39.
Paich, M. and J.D. Sterman. lY93. Boom. bust. and failures
to learn in experimental markets. Managemenr Science. 3Y.
1439-145X.
Palepu, K.. lYX5, Diversification strategy. profit performance
and the entropy measure. Strategic Manugemeni Journal,
6. 239-25s.
Pfeffer. J. and G.R. Salancik. lY7X. The External Control of
Organizations: A Resource Dependence Perspective
(Harper & Row. New York).
Ross. W.. 1087. A re-examination of the results of Hogarth
and Makridakis’ ‘The value of decision making in a
complex environment: An experimental approach’. Man-
agement Science. 33. 2X8-296.
Secger. J.A., 19x4. Reversing the images of BCG’s
growth/share matrix. Straregic Management Journal, 5. Y3-
97.
Simon. H.A., 19.57, Models of Man: Social and Rational
(John Wiley & Sons. New York).
Simon, H.A., lY7X. “Rationality as process and as product of
thought”. American Economic Review, 68, I-16.
Simon, H.A.. 1979, Rational decision making in organiza-
tions, American Economic Review. 69, 493-513.
Singer, A.E. and R.J. Brodie. 1090. Forecasting competitors’
actions: An evaluation of alternative ways of analyzing
business competition, International Journal of Forecasting,
6, 75-88.
Sterman. J.D., lY87. Testing behavioral simulation models
by direct experiment. Manqement Science. 33. 1572-1592.
Stigler. G.J.. 1983, The Organization of Industry (University
of Chicago Press. Chicago. IL).
Thorndike. E.L.. 1918. Fundamental theorems in judging
men, Journal of Applied Psychology, 2. 67-76.
Tushman. M.L. and E. Romanclli, 1985, Organizational
evolution: A metamorphosis model of convergence and
reorientation, in: L.L. Cummings and B.M. Staw. eds..
Research in Organizational Behavior, Volume 7 (JAI Press
Inc., Greenwich, CT), 171-222.
Urban. G.L., J.R. Hauser and J.H. Roberts, 1990. Pre-
launch Forecasting of New Automobiles, Marketing Sci-
ence, 36, 401-321.
Wenslcy. R.. lYX2. PIMS and BCG: New horizons or false
dawn?. Strategic Mana,qement Journal, 3. 147-158.
Wind, Y.. IYXl, Marketing oriented strategic planning
models, in: R.L. Schultz and A.A. Zoltners, eds., Market-
ing Decision Models (North Holland, New York. NY).
207~250.
Wind, Y. and T.S. Robertson, 1YX3. Marketing strategy: New
directions for theory and research, Journal of Marketing.
47. 12-25.
Biographies: Noel CAPON is Professor of Business at
Columbia Business School. His teaching and research inter-
ests are in the areas of strategy and planning. and sales force
management. He is coauthor of Corporate Strategic Planning,
a study of planning practices in major U.S. manufacturing
corporations. A companion book on the underpinnings of
corporate financial performance is in an advanced stage of
preparation. Hc acknowledges support from the Redwoord
Foundation.
Peter PALIJ is a doctoral candidate at Columbia Business
School. His teaching and research interests are in the areas of
marketing strategy. decision making and competition. He
acknowledges support from the Institute for the Study of
Business Markets.

More Related Content

What's hot

Mtm4 white paper financial ratio and statement analysis
Mtm4 white paper   financial ratio and statement analysisMtm4 white paper   financial ratio and statement analysis
Mtm4 white paper financial ratio and statement analysisIntelCollab.com
 
Industry and Competitor Analysis | Five Competitive Forces | Five Primary Ind...
Industry and Competitor Analysis | Five Competitive Forces | Five Primary Ind...Industry and Competitor Analysis | Five Competitive Forces | Five Primary Ind...
Industry and Competitor Analysis | Five Competitive Forces | Five Primary Ind...FaHaD .H. NooR
 
location choices under strategic Interactions
location choices under strategic Interactionslocation choices under strategic Interactions
location choices under strategic InteractionsKiran Niraula
 
Strategy frameworks-and-models
Strategy frameworks-and-modelsStrategy frameworks-and-models
Strategy frameworks-and-modelsTaposh Roy
 
Media Management 2011-Strategy Module - Jan 21_2
Media Management 2011-Strategy Module - Jan 21_2Media Management 2011-Strategy Module - Jan 21_2
Media Management 2011-Strategy Module - Jan 21_2Robin Teigland
 
Explorations of strategic orientation (so) dimensions on small firm growth an...
Explorations of strategic orientation (so) dimensions on small firm growth an...Explorations of strategic orientation (so) dimensions on small firm growth an...
Explorations of strategic orientation (so) dimensions on small firm growth an...Alexander Decker
 
I M P O R T A N T Chap003 1
I M P O R T A N T  Chap003 1I M P O R T A N T  Chap003 1
I M P O R T A N T Chap003 1Ashar Azam
 
GENERAL SLP SCENARIO / TUTORIALOUTLET DOT COM
GENERAL SLP SCENARIO / TUTORIALOUTLET DOT COMGENERAL SLP SCENARIO / TUTORIALOUTLET DOT COM
GENERAL SLP SCENARIO / TUTORIALOUTLET DOT COMalbert0067
 
Entrepreneurship Chap 5
Entrepreneurship Chap 5Entrepreneurship Chap 5
Entrepreneurship Chap 5Umair Arain
 
Ge nine(9) cell matrix
Ge nine(9) cell matrixGe nine(9) cell matrix
Ge nine(9) cell matrixHpm India
 
Mtm8 white paper scenario analysis
Mtm8 white paper   scenario analysisMtm8 white paper   scenario analysis
Mtm8 white paper scenario analysisIntelCollab.com
 
Portfolio planning
Portfolio planning Portfolio planning
Portfolio planning Azeem Abbas
 

What's hot (20)

Ge
GeGe
Ge
 
Prsentation on space matrix
Prsentation on space matrixPrsentation on space matrix
Prsentation on space matrix
 
Mtm4 white paper financial ratio and statement analysis
Mtm4 white paper   financial ratio and statement analysisMtm4 white paper   financial ratio and statement analysis
Mtm4 white paper financial ratio and statement analysis
 
Bcg matrix and others
Bcg matrix and othersBcg matrix and others
Bcg matrix and others
 
Industry and Competitor Analysis | Five Competitive Forces | Five Primary Ind...
Industry and Competitor Analysis | Five Competitive Forces | Five Primary Ind...Industry and Competitor Analysis | Five Competitive Forces | Five Primary Ind...
Industry and Competitor Analysis | Five Competitive Forces | Five Primary Ind...
 
Ge Final[1]
Ge Final[1]Ge Final[1]
Ge Final[1]
 
location choices under strategic Interactions
location choices under strategic Interactionslocation choices under strategic Interactions
location choices under strategic Interactions
 
Strategy frameworks-and-models
Strategy frameworks-and-modelsStrategy frameworks-and-models
Strategy frameworks-and-models
 
Media Management 2011-Strategy Module - Jan 21_2
Media Management 2011-Strategy Module - Jan 21_2Media Management 2011-Strategy Module - Jan 21_2
Media Management 2011-Strategy Module - Jan 21_2
 
Marketing Strategy
Marketing StrategyMarketing Strategy
Marketing Strategy
 
Ppt for bcg &amp; ge nine cell matrix
Ppt for bcg &amp; ge nine cell matrixPpt for bcg &amp; ge nine cell matrix
Ppt for bcg &amp; ge nine cell matrix
 
Explorations of strategic orientation (so) dimensions on small firm growth an...
Explorations of strategic orientation (so) dimensions on small firm growth an...Explorations of strategic orientation (so) dimensions on small firm growth an...
Explorations of strategic orientation (so) dimensions on small firm growth an...
 
I M P O R T A N T Chap003 1
I M P O R T A N T  Chap003 1I M P O R T A N T  Chap003 1
I M P O R T A N T Chap003 1
 
Hofer
HoferHofer
Hofer
 
GENERAL SLP SCENARIO / TUTORIALOUTLET DOT COM
GENERAL SLP SCENARIO / TUTORIALOUTLET DOT COMGENERAL SLP SCENARIO / TUTORIALOUTLET DOT COM
GENERAL SLP SCENARIO / TUTORIALOUTLET DOT COM
 
Entrepreneurship Chap 5
Entrepreneurship Chap 5Entrepreneurship Chap 5
Entrepreneurship Chap 5
 
Ge nine(9) cell matrix
Ge nine(9) cell matrixGe nine(9) cell matrix
Ge nine(9) cell matrix
 
Mtm8 white paper scenario analysis
Mtm8 white paper   scenario analysisMtm8 white paper   scenario analysis
Mtm8 white paper scenario analysis
 
Business strategy
Business strategyBusiness strategy
Business strategy
 
Portfolio planning
Portfolio planning Portfolio planning
Portfolio planning
 

Viewers also liked

Introduction to Social Media for Executives: Job Hunting with Social Media
Introduction to Social Media for Executives: Job Hunting with Social MediaIntroduction to Social Media for Executives: Job Hunting with Social Media
Introduction to Social Media for Executives: Job Hunting with Social MediaDebra Ulrich
 
Paper hack facebook
Paper hack facebookPaper hack facebook
Paper hack facebookekagiri
 
An Introduction to Twitter in Higher Education
An Introduction to Twitter in Higher Education An Introduction to Twitter in Higher Education
An Introduction to Twitter in Higher Education Debra Ulrich
 
How Linkedin Is Transforming Business
How Linkedin Is Transforming BusinessHow Linkedin Is Transforming Business
How Linkedin Is Transforming BusinessDebra Ulrich
 
The Future of Advertising 2020
The Future of Advertising 2020The Future of Advertising 2020
The Future of Advertising 2020Debra Ulrich
 
Comprehensive Report on The Cannabis Extract Movement
Comprehensive Report on The Cannabis Extract MovementComprehensive Report on The Cannabis Extract Movement
Comprehensive Report on The Cannabis Extract MovementDebra Ulrich
 

Viewers also liked (9)

THE MOOC OF ONE
THE MOOC OF ONETHE MOOC OF ONE
THE MOOC OF ONE
 
Introduction to Social Media for Executives: Job Hunting with Social Media
Introduction to Social Media for Executives: Job Hunting with Social MediaIntroduction to Social Media for Executives: Job Hunting with Social Media
Introduction to Social Media for Executives: Job Hunting with Social Media
 
Paper hack facebook
Paper hack facebookPaper hack facebook
Paper hack facebook
 
Photography
PhotographyPhotography
Photography
 
An Introduction to Twitter in Higher Education
An Introduction to Twitter in Higher Education An Introduction to Twitter in Higher Education
An Introduction to Twitter in Higher Education
 
How Linkedin Is Transforming Business
How Linkedin Is Transforming BusinessHow Linkedin Is Transforming Business
How Linkedin Is Transforming Business
 
The Future of Advertising 2020
The Future of Advertising 2020The Future of Advertising 2020
The Future of Advertising 2020
 
Pma
PmaPma
Pma
 
Comprehensive Report on The Cannabis Extract Movement
Comprehensive Report on The Cannabis Extract MovementComprehensive Report on The Cannabis Extract Movement
Comprehensive Report on The Cannabis Extract Movement
 

Similar to Market2

A Sales Forecasting Model Based on Internal Organizational Variables.pdf
A Sales Forecasting Model Based on Internal Organizational Variables.pdfA Sales Forecasting Model Based on Internal Organizational Variables.pdf
A Sales Forecasting Model Based on Internal Organizational Variables.pdfAnna Landers
 
A Business Market Segmentation Procedure For Product Planning
A Business Market Segmentation Procedure For Product PlanningA Business Market Segmentation Procedure For Product Planning
A Business Market Segmentation Procedure For Product PlanningBrittany Brown
 
How your product solves customers problems or improves their si.docx
How your product solves customers problems or improves their si.docxHow your product solves customers problems or improves their si.docx
How your product solves customers problems or improves their si.docxpooleavelina
 
External Analysis Strategic Management Ljmu
External Analysis   Strategic Management LjmuExternal Analysis   Strategic Management Ljmu
External Analysis Strategic Management Ljmusnoozed
 
Frameworks for Global Strategic Analysis.pdf
Frameworks for Global Strategic Analysis.pdfFrameworks for Global Strategic Analysis.pdf
Frameworks for Global Strategic Analysis.pdfdrsalamdarwish
 
Uncertainty
UncertaintyUncertainty
UncertaintyJan Zika
 
Dissertation Part 2 - Academic Discussion
Dissertation Part 2 - Academic DiscussionDissertation Part 2 - Academic Discussion
Dissertation Part 2 - Academic DiscussionWill Scott
 
The Five Competitive Forces That Shape Strategyby Michael E..docx
The Five Competitive Forces That Shape Strategyby Michael E..docxThe Five Competitive Forces That Shape Strategyby Michael E..docx
The Five Competitive Forces That Shape Strategyby Michael E..docxcherry686017
 
Managing market competitive strategy successfully an empirical testing of su
Managing market competitive strategy successfully an empirical testing of suManaging market competitive strategy successfully an empirical testing of su
Managing market competitive strategy successfully an empirical testing of suIAEME Publication
 
Brief of the dimensionality of business strategy among the manufacturing orga...
Brief of the dimensionality of business strategy among the manufacturing orga...Brief of the dimensionality of business strategy among the manufacturing orga...
Brief of the dimensionality of business strategy among the manufacturing orga...Alexander Decker
 
Running head DISCUSSION .docx
Running head DISCUSSION                                    .docxRunning head DISCUSSION                                    .docx
Running head DISCUSSION .docxtodd271
 
Running head DISCUSSION .docx
Running head DISCUSSION                                        .docxRunning head DISCUSSION                                        .docx
Running head DISCUSSION .docxtodd271
 
GBS Sample 1Name_ID_GBS Task 1.pdf1 P a g e .docx
GBS Sample 1Name_ID_GBS  Task 1.pdf1  P a g e  .docxGBS Sample 1Name_ID_GBS  Task 1.pdf1  P a g e  .docx
GBS Sample 1Name_ID_GBS Task 1.pdf1 P a g e .docxshericehewat
 
chapter-4-competitive-rivalry-and-competitive-dynamics.docx
chapter-4-competitive-rivalry-and-competitive-dynamics.docxchapter-4-competitive-rivalry-and-competitive-dynamics.docx
chapter-4-competitive-rivalry-and-competitive-dynamics.docxRose Sally
 
Assignment 3 Case StudyE-Business Strategy and Models in B.docx
Assignment 3  Case StudyE-Business Strategy and Models in B.docxAssignment 3  Case StudyE-Business Strategy and Models in B.docx
Assignment 3 Case StudyE-Business Strategy and Models in B.docxbraycarissa250
 
Strategy and Vision for SMEs
Strategy and Vision for SMEsStrategy and Vision for SMEs
Strategy and Vision for SMEsindranildeb
 
IHS Consulting Services
IHS Consulting ServicesIHS Consulting Services
IHS Consulting Servicescrschena
 

Similar to Market2 (20)

A Sales Forecasting Model Based on Internal Organizational Variables.pdf
A Sales Forecasting Model Based on Internal Organizational Variables.pdfA Sales Forecasting Model Based on Internal Organizational Variables.pdf
A Sales Forecasting Model Based on Internal Organizational Variables.pdf
 
A Business Market Segmentation Procedure For Product Planning
A Business Market Segmentation Procedure For Product PlanningA Business Market Segmentation Procedure For Product Planning
A Business Market Segmentation Procedure For Product Planning
 
How your product solves customers problems or improves their si.docx
How your product solves customers problems or improves their si.docxHow your product solves customers problems or improves their si.docx
How your product solves customers problems or improves their si.docx
 
10120140507001
1012014050700110120140507001
10120140507001
 
10120140507001
1012014050700110120140507001
10120140507001
 
External Analysis Strategic Management Ljmu
External Analysis   Strategic Management LjmuExternal Analysis   Strategic Management Ljmu
External Analysis Strategic Management Ljmu
 
Frameworks for Global Strategic Analysis.pdf
Frameworks for Global Strategic Analysis.pdfFrameworks for Global Strategic Analysis.pdf
Frameworks for Global Strategic Analysis.pdf
 
Uncertainty
UncertaintyUncertainty
Uncertainty
 
Dissertation Part 2 - Academic Discussion
Dissertation Part 2 - Academic DiscussionDissertation Part 2 - Academic Discussion
Dissertation Part 2 - Academic Discussion
 
The Five Competitive Forces That Shape Strategyby Michael E..docx
The Five Competitive Forces That Shape Strategyby Michael E..docxThe Five Competitive Forces That Shape Strategyby Michael E..docx
The Five Competitive Forces That Shape Strategyby Michael E..docx
 
Managing market competitive strategy successfully an empirical testing of su
Managing market competitive strategy successfully an empirical testing of suManaging market competitive strategy successfully an empirical testing of su
Managing market competitive strategy successfully an empirical testing of su
 
Brief of the dimensionality of business strategy among the manufacturing orga...
Brief of the dimensionality of business strategy among the manufacturing orga...Brief of the dimensionality of business strategy among the manufacturing orga...
Brief of the dimensionality of business strategy among the manufacturing orga...
 
Running head DISCUSSION .docx
Running head DISCUSSION                                    .docxRunning head DISCUSSION                                    .docx
Running head DISCUSSION .docx
 
Running head DISCUSSION .docx
Running head DISCUSSION                                        .docxRunning head DISCUSSION                                        .docx
Running head DISCUSSION .docx
 
Manufacturing strategy
Manufacturing strategyManufacturing strategy
Manufacturing strategy
 
GBS Sample 1Name_ID_GBS Task 1.pdf1 P a g e .docx
GBS Sample 1Name_ID_GBS  Task 1.pdf1  P a g e  .docxGBS Sample 1Name_ID_GBS  Task 1.pdf1  P a g e  .docx
GBS Sample 1Name_ID_GBS Task 1.pdf1 P a g e .docx
 
chapter-4-competitive-rivalry-and-competitive-dynamics.docx
chapter-4-competitive-rivalry-and-competitive-dynamics.docxchapter-4-competitive-rivalry-and-competitive-dynamics.docx
chapter-4-competitive-rivalry-and-competitive-dynamics.docx
 
Assignment 3 Case StudyE-Business Strategy and Models in B.docx
Assignment 3  Case StudyE-Business Strategy and Models in B.docxAssignment 3  Case StudyE-Business Strategy and Models in B.docx
Assignment 3 Case StudyE-Business Strategy and Models in B.docx
 
Strategy and Vision for SMEs
Strategy and Vision for SMEsStrategy and Vision for SMEs
Strategy and Vision for SMEs
 
IHS Consulting Services
IHS Consulting ServicesIHS Consulting Services
IHS Consulting Services
 

Market2

  • 1. International Journal of Forecasting 10 (1994) 339-352 Strategic marketing forecasting, market segment selection firm performance * Noel Capon*, Peter Palij Graduate School of Business, Columbia University, New York, NY 10027. USA and Abstract In this study we investigate several hypotheses relating strategic forecasting to market segment selection and firm performance. In the context of a strategic marketing simulation, subjects in 14 competitive industries made strategic forecasts for market segment size and benchmark prices. Our results show that firms differentially select those segments with attractive characteristics; that some strategic forecasts for these targeted segments are more accurate than for non-targeted segments; that strategic forecasts are more accurate the higher the level of competition; and that superior forecasting performance is positively associated with superior firm performance. Implications and limitations of the study are discussed. Keywords: Strategic marketing forecasting; Market forecasting; Market segment selection; MARKSTRAT 1. Introduction Perhaps the most critical decision that senior managers in multi-product, multi-market firms must make is the choice of which products to produce and which markets to address. In the 197Os, several strategic planning models were developed to assist management in addressing this problem. Although not universally accepted, by the early 1990s these models and/or their successors were woven into the fabric of * Corresponding author. Tel: (212) 854-3466 or (212) 865 5332; Fax: (212) 749-3549. ‘“’Preliminary results from the non-degree executive pro- gram subjects were presented at the November, 1991, ORSAiTIMS Joint National Meeting. strategic management at many major corpora- tions (Capon et al., 1987; Greenley and Bayus, 1993). The basic purpose of strategic planning models is to provide managers with a coherent frame- work to structure long-term planning. Although strategic planning models display idiosyncratic differences, they are typically presented as two dimensional matrices in which the axes represent market attractiveness and business strengths. The models differ from each other in the manner in which these dimensions are operationalized. For example, in the GE/McKinsey screen, Ar- thur D. Little and BCG portfolio models, market attractiveness is represented respectively as a weighted linear combination of such variables as market growth rate, profitability, number of competitors, and degree of regulation; as prod- 0169-2070/94/$07.00 @ 1994 Elsevier Science B.V. All rights reserved SSDI 0169-2070(94)09007-R
  • 2. uct life cycle stage; and as the forecast long-run market growth rate (Wind, 1981). Across these two dimensions, each model uses a mix of measurable current variables, and medium- and long-term forecasts. The appro- priate values for current variables may be dcbat- able, but the forecast variables represent a far more difficult problem. Indeed, Capon and Hul- bert (1985) lament the fact that virtually all forecasting research at the firm level has ad- dressed short-term forecasting (e.g. sales fore- casts for production planning and inventory management; short-term revenue forecasts for cash Now management). More recently, several researchers have attempted to address firm-level forecasting issues by using a strategic marketing simulation meth- odology (LarreLhC and Gatignon, 1990). For example, Glazer et al. (1989, 1990) examined how closely short-term forecasts based on mana- gerial judgment conformed to the formal prop- erties of rationality. They found that one-year forecasts for aggregate market size, average industry price and number of marketed products in four ‘industries’ were efficient but biased. However, not only were these studies concerned with short-term forecasting, the forecast vari- ables were not particularly strategic. By contrast, the study reported in this paper, using a similar methodology, focuses specifically on strategic forecasting issues. Capon and Huibert (1985) coin the term ‘strategic forecasting’ to embrace those types of long-term forecasts that are strategic for the firm. Important dimensions requiring such fore- casts include customer needs, competitive struc- ture, competitive action, degree of regulation ’ Rclatcd arcas are role playing (Armstrong, 1988) and new product forecasting (Mahajan and Wind, 1988). Role playing research focuses on developing tools to overcome managerial biases in forecasting. Most new product forecast- ing research has avoided firm-level strategic forecasting issues; segment ideIlti~c~~tion and sclcction arc typically taken as given, and forecasts focus on prospects and the optimal marketing mix for new products. and market and/or market segment growth rates (Wind, 1981; Wind and Robertson, 1983; Arm- strong et al., 1987).’ The forecasting literature has generally taken a pessimistic view of such medium- and long- term forecasts, based primarily on their reliance on managerial judgment (Armstrong, 1986). For example, although Hogarth and Makridakis (1981) recognize the importance of judgment for these forecasts, they list over two dozen poten- tially dist~~rting biases; and Paich and Sterman (1993) find that even when conditions are delib- erately designed to increase forecasting accuracy, managers often fail to overcome judgmental biases. In addition, Chakravarti et al. (1981) take a pessimistic view of the value of managerial judgment for marketing decisions in the context of marketing decision models.’ The expert literature reaches a similarly pes- simistic conclusion regarding the predictive abili- ty of experts. Beginning with Thorndike’s (1918) study of officer candidacy selection by military experts, a long series of studies has found .I . . [that] expert judgment {a) is seldom better than that of novices and (b) is easily improved upon when replaced with simple equations.” (Hammond, 1987, p. 3). Hogarth and Makridakis recommend that long- term forecasting focus on identifying sources of environmental uncertainty, deliberately adopt ambiguous rather than precise goals, rely on simple time series models whenever possible, and generally avoid complexity. Makridakis (1986) echoes Hogarth and ~akridakis’ recom- mendations while emphasizing the complemen- tary roles of judgmental and quantitative fore- casts. Long-term forecasting research has focused on macroeconomic variables in areas such as busi- ness cycles, energy, population and transporta- tion (Ascher, 1978; Adams, 1986); medium-term forecasting research has focused on financial ‘See .Little and Lodish (ISSI) for a response on the relative value of statistically based and judgmental models for decision making in marketing.
  • 3. N. Capon. P. Palij I International Journal of Forecasting 10 (1994) 339-352 341 analyses and capital allocation decisions (see nexus of customers, competitors and tech- Hogarth and Makridakis, 1981). Microeconomic nologies, the marketing strategist must segment research has emphasized industry-level forecasts the market, select those segments in which the as inputs to national-level macroeconomic fore- firm should compete, and discard others (Hul- casts (Adams, 1986, pp. 161-165). Little work bert, 1985, pp. 30-31). Segment selection, and has addressed such firm-level strategic forecast- positioning decisions within the segment, repre- ing problems as market or market segment sent the bridge between the SBU strategy and growth rates. Marketing research has stressed tactical marketing actions that the firm executes product forecasts at the industry level. leaving in the market place. It is at the market segment firm-level strategic and market segment forecasts level that critical resource allocation decisions needing extensive exploration (Armstrong et al., are made for securing customers at the expense 1987). In a rare exception, Cortes-Rello and of competitors. The ability to predict key dimen- Golshani (1990) use an expert system meth- sions of available market segments over the odology to predict 10 year market demand for a medium and long term is a critical strategic capital good. necessity. Based on the literature, an awkward situation arises. Strategic planning models are the princi- pal tools available for assisting senior managers in making strategic choices, and these models use as inputs strategic forecasts of key variables. However, these strategic forecasts, typically based on managerial judgment, are generally viewed as frequently inaccurate and biased. Furthermore, little research has addressed the question of improving such forecasts. In this paper we develop several hypotheses relating to the use of strategic planning models, the relationship between strategic forecast ac- curacy and market segment selection, and be- tween strategic forecast accuracy and firm per- formance. Next, we test these hypotheses using a large sample of strategic market forecasts se- cured in the context of the MARKSTRAT simulation (Larredhe and Gatignon, 1990). Final- ly, we discuss the results, explore some limita- tions and make recommendations for future research. Strategic planning models provide a useful framework for the marketing strategist to struc- ture the set of decision options for selecting those market segments in which the firm will compete. Using the market attractiveness, busi- ness strengths paradigm, the received wisdom of strategic planning models states that the strateg- ist should strive to identify, then select for effort, those market segment options that display high levels of market attractiveness, where the firm also has substantial business strengths (Kotler 1984, p. 61; Day, 1990, p. 200). Hl: The firm’s selection of market segments to address is positively related to their market attractiveness, and to the firm’s current business strengths in those market segments. 2. Development of hypotheses 2.1. Hypothesis 1 Perhaps the broadest use of strategic planning models is at the strategic marketing level. In the context of a strategic business unit (SBU) strategy that broadly defines the competitive In order to test this hypothesis, we focus on the product portfolio matrix developed by the Boston Consulting Group (BCG). In this strategic planning model, market attractiveness is defined as the expected market segment growth rate; we use two measures-the medium- term market segment size, and the annualized growth rate needed to reach the expected medium-term market segment size from the current size. The BCG approach operationalizes business strength in the market segment as the firm’s current relative market share (RMS), and directly links RMS to profitability (Henderson and Zakon, 1980; Day, 1986). The resource
  • 4. 342 N. Capon, P. Palij I International Journal of Forecasting IO (1994) 339-3.52 allocation actions taken by the firm in the follow- ing periods are used to indicate management’s decision to focus on the segment.” The BCG approach to strategic planning has received mixed reviews. Several writers view the model positively (e.g. Dutton et al., 1983; Burke, 1984) and some empirical evidence sug- gests that it is widely used in industry (e.g. Capon et al., 1987). However, it has fallen out of favor among academics (e.g. Wensley 1982; Seeger 1984) largely because of its reliance on the experience curve and the strongly implied, and much debated, linkage between market share and profitability (e.g. Jacobson, 1988). However, in the context of the MARKSTRAT simulation, use of this framework is quite appro- priate as strong experience curve effects are embedded in both the production, and research and development functions. Consequently, the correlation between market share and profitabili- ty is high, 0.89 for the industries in this study. 2.2. Hypotheses 2 and 3 Broadly speaking, three perspectives govern the strategic forecasts required by strategic plan- ning models. Espousing a normative view, tradi- tional neoclassical economic models assume ‘hy- perrational’ behavior (Simon, 1978): managers are sequential, exhaustive and optimal analysts of all available data. Managers carefully forecast each relevant variable. in each market segment, in each period, in order to exploit opportunities as competitive dynamics and other conditions change within segments. For such forecasters, accuracy should be independent of factors such as market segment participation and competitive intensity. In direct contrast, empirical forecasting evi- ’ Note that sales volume in a market segment ex post does not imply that the firm has made the strategic decision to target that segment for effort ex ante. for a variety of reasons including inadequate segment specification, improper im- plementation and variety seeking by customers, firms may secure significant market share in non-targeted market seg- ments dence indicates little hope for any accuracy in making such strategic forecasts. As noted previ- ously, these forecasts are highly susceptible to bias (Hogarth and Makridakis, 1981); the ac- tion-outcome-feedback loop is weak, typically stretching beyond a product development cycle; relatively precise values are required for effec- tive model use; and relevant variables are affect- ed by complex, diverse and often unknown forces. Since any accurate forecasts are seren- dipitous, accuracy should be independent of factors such as market segment participation and competitive intensity. Rather than adopting either polar perspective, the traditional neoclassical economic normative approach or the pessimism of extant empirical studies, each of which leads to similar predic- tions regarding forecast accuracy, we choose to work within the bounded rationality paradigm (Simon, 1957, 1979). Bounded rationality implies an intermediate position on forecast accuracy. Forecasters are unable to cope with the mag- nitude of processing tasks required for hyper- rationality and so use processing short-cuts such as heuristics and schemata. However, as a prob- lem’s salience increases, a variety of active strategies are adopted, including those hypoth- esized by the neoclassical perspective (Eisenhardt and Zbaracki, 1992). Thus, as re- sources (e.g. marketing research. product de- velopment. production, advertising and promo- tion) are focused on a market segment, its increased salience leads forecasters to pay in- creasing attention to strategic variables relevant for that segment. Ceteris paribus, increased attention should lead to more accurate forecasts so that the bounded rationality perspective leads directly to Hypothesis 2. H2: The firm’s forecasts of strategic dimensions for target market segments are more accurate than for avoided segments. The notion of salience relates both to those segments that are targeted for effort, and to the degree of difficulty in competing in those seg- ments. Thus, we anticipate that, ceteris paribus,
  • 5. N. Capon. P. Palij I International Journal of Forecasting 10 (1994) 339-352 343 increased effort, and therefore increased accura- cy, should be related to increased levels of competition in targeted segments. Increased ef- fort under these conditions is consistent with anecdotal evidence from managers that effort on problems typically exceeds effort on oppor- tunities. H3: The accuracy of the firm’s strategic forecasts in target market segments increases as the level of competition increases. 2.3. Hypothesis 4 Strategic planning advocates posit a positive relationship between the use of strategic plan- ning models and firm performance. This positive view of the influence of managers and strategic planning on firm performance has been chal- lenged by research focusing on the importance of internal and external environmental factors (e.g. Hannan and Freeman, 1977; Pfeffer and Salan- cik, 1978; Fredrickson and Mitchell, 1984). Em- pirical findings are mixed, but generally suppor- tive of a positive link between strategic planning and firm performance (Armstrong, 1982; Capon et al., 1987, 1994; Boyd, 1991). As discussed above, good strategic planning requires the ability to make good strategic fore- casts. Indeed, in Urban et al.‘s (1990) strategic planning model for the automotive industry, nine of 24 of the model’s parameters were substantial- ly or entirely based on judgmental forecasts made by managers. Clearly, the success of such formal models in providing useful input to deci- sion making depends upon the accuracy of the underlying parameters. However, regardless of forecasting accuracy, many strategic variables such as level of diversification and product line maturity are also expected to impact firm per- formance. Thus, we expect a weak positive link between forecast accuracy and performance. H4: The greater the accuracy of the firm’s long- run strategic forecasts, the better its perform- ance. 3. Method 3.1. Direct experiment Methodologies available for testing the hy- potheses range from laboratory experiments to field studies. Laboratory studies have the benefit of control, but are seriously challenged by topics requiring a series of complex decisions evolving over time (Singer and Brodie, 1990). Field studies capture complexity but typically at the cost of lack of control. Simulations conducted as direct experiments (Sterman, 1987) offer an intermediate technique for testing strategic fore- casting hypotheses related to complex decisions (Glazer et al., 1990; Green and Ryans, 1990). The MARKSTRAT 2 simulation (Lar- reche and Gatignon, 1990) provides a realistic environment in which participants seek long-run success through a series of strategic and tactical marketing decisions, moderated by feedback on the results of their decisions (Kinnear and Klam- mer, 1987). Five firms compete in an established market (sonites) comprising five well-defined market segments, and in a second undifferen- tiated independent market (vodites) that is un- tapped at the beginning of the simulation. Each firm starts the simulation with a differing mix of corporate strengths and weaknesses (some firms are stronger, some firms are weaker), and with two sonite products in different competitive positions; each firm may offer up to five prod- ucts. For each cycle of the simulation, a team managing its firm assesses the results of prior decisions, evaluates marketing research data, and allocates its limited budget to R&D, market- ing mix expenditures and marketing research. The firm decisions are run, and results returned for the next decision cycle. Firm performance (e.g. sales, profit, market share) is a function of firm and competitor decisions, individual firm capabilities including budgets, and the evolving MARKSTRAT economy. Initial firm parame- ters, underlying economic conditions for each industry of five firms, and the general pattern of segment growth and preferences are set identi- cally at the start of each simulation. MARKSTRAT focuses on short- and medium-
  • 6. term decisions; long-term decisions such as aban- doning both sonite and vodite markets, investing in new technologies, or adding new distribution channels are not available. Short-term decisions are set by the ‘annual’ decision cycle. Medium- term decisions reflect changes in resource alloca- tions among market segments, and investments in product development. New product entry into a market segment implies a 2-4 year product development cycle: l-2 years to develop a prod- uct, 1 year for iIltroduction and feedback on market segment acceptance, and 1 year for product modification. 3.2. Subjects Subjects were participants in full-time MBA, executive MBA and non-degree executive pro- grams conducted by a leading business school. Full-time MBA students, representing a range of business experience, were participating in a second-year elective course in marketing strategy; executive MBA students. participating in a required advanced course in marketing strategy, were mid-level executives (median age 33) attending classes once per week under firm sponsorship; non-degree students were senior marketing executives participating in a l-week residential advanced program in strategic mar- keting. 3.3. Procedure The approximately 250 subjects were placed in 70 three to five person groups. (Full-time MBA groups were self selected; the others were ran- domly assigned.) The 70 groups comprised 14 industries of five teams (firms): four full-time MBA, six executive MBA, and four non-degree students. Data were collected as part of required classroom activities. For both MBA groups, the eight decisions were made weekly during the semester; the non-degree students made seven decisions at a rate of two decisions per day. At the start of the simulation. each firm received a summary of results for the prior year (zero), a complete set of aggregate market re- search data, and a decision support package comprising a variety of spreadsheet tools to assist in analyzing firm and marketing research data (James et al., 1991). After a decision was run, each team received its firm report and the specific marketing research data it had re- quested; use of the decision support software was optional. Upon completion of the simula- tion, each team was required to make formal oral (to the class) presentations explaining ac- tions taken, perceptions of industry events and lessons learned. Such presentations by teams in previously run industries provided pretest guid- ance. 3.4. Variables 3.4.1. Strategic marketing variables Informal discussions with previous MARK- STRAT participants revealed a large and com- plex array of variables perceived as strategic: these included internal capabilities, budget allo- cations, size and growth of the market segments, competitive conditions in each market segment and changes in consumer preferences. We selected as strategic variables both market segment size (units) and market segment growth rate, each of which addresses the market attrac- tiveness dimension of strategic planning models. In addition, we selected market segment leader’s price. Price plays an important strategic role in MARKSTRAT. First, it provides a measure of profit potential, given the firm’s Iikely product- ion cost. Second, it captures customer prefer- ences by playing a direct role as a product positioning attribute in the sonite market, and as an indirect positioning attribute for vodites. 3.4.2. Forecast market segment size This variable is a forecast of the number of units sold into the segment by all competitors 3 years ahead. 3.4.3. Forecast market segment growth rate This variable is the compound annual growth rate required to reach the forecast market seg- ment size from the current market segment size.
  • 7. This variable is the forecast price, 3 years ahead, of the product having the largest market share in the market segment, 3.4.5. Firm performance Direct measures of firm performance in MARKSTRAT are aggregated across products and segments. Several candidate variables were examined: sales units, sales revenues, unit mar- ket share, revenue market share, gross profit share and net profit. The smallest of the 15 correlations between these variables was 0.88. Revenue market share was selected because it both permits identical treatment of the different sonite and vodite markets, and normalizes firm results across the sample industries. 3.4. (5;.Market attractiveness This segment-level variable was oper- ationalized in two ways: as the forecast market segment size and as the implicit forecast market segment growth rate (see above), The formal debriefs presented by previous MARKSTRAT teams indicated that participants generally as- sessed market attractiveness by estimating the future size of a market segment, then checking to see if the growth rate seemed reasonable. 3.4.7. Business strength This segment-level variable was oper- ationalized as the firm’s current relative market share. Relative market share is the firm’s market share (of all products in a segment), divided by the market share of the segment leader (or by the number two share firm if the focal firm was the segment leader). .3.4.8. Target market segment In the sonite market, the five market segments are defined as a set of ideal points in a three dimensional positioning map whose axes are perceived price, perceived power and perceived design (the latter two are physical attributes). Two methods assess the target segments of current products-product design characteristics and advertising positioning; target segments for future products are identified from the design characteristics of active R&D projects. In each case, the minimum Euclidian distance from the product to each of the five ideal points is defined as the market segment targeted. The three methods vary by time horizon; product design characteristics and ideal points are measured contemporaneously; the advertis- ing positioning decision in period t is compared with the revealed actual ideal points in period t + 1; the R&D design characteristics in period t are compared with the actual ideal points in period t t 3. For the undi~erentiated vodite market, any product entry, or R&D project, respectively, defines a targeted market. The measure of target market segment em- ployed for any decision period was the union of these three measures under the rationale that for any target market segment, good strategic fore- casting is a critical matter. 3.4.9. Competitive activity This variable is defined by adapting the Jac- quemin and Berry (1979) entropy measure of diversification. Level of competition is defined as: where Pi is each firm’s market share in the market segment; the larger the index, the greater the level of competitive activity. This measure is sensitive to the number of competitors, the distribution of market shares among com- petitors, and the mix of cfose and distant com- petitors (Palepu, 1085). Tradition&y, competitive-type measures have focused on market concentration (e.g. concen- tration ratios-total market shares of the n largest firms; Herfindahl’s index, H-sum of the squared market shares for each firm) as in- dicators of public policy concern, rather than on competitiveness per se. For public policy analy- sis, Stigler (1983) recommends the Eferfindahl index over entropy measures, based in part on its insensitivity to firms with small market shares. However, since the MARKSTRAT enQiro~ment comprises a fixed number of firms, each with the
  • 8. 336 N. Capon. P. Palij I International Journal of Forecasting 10 (1994) .U9-352 ability to offer multiple products to multiple segments, the dimensions captured by the en- tropy measure seemed more appropriate as in- dicators of a segment’s competitiveness. 3.4.10. Forecast accuracy Several indices of forecasting accuracy are available. However, since market segment sizes varied by an order of magnitude, we rejected both the simple error and mean square error measures (Mahmoud, 19S6, pp. 509-511). In- stead, we selected the absolute percent error (APE) and mean absolute percent error (MAPE). These measures give equal weight to over- and under-prediction, reflecting the strategic issue that for market segment size and growth rate, under-prediction may result in op- portunity losses as, or even more, serious than direct losses resulting from over-prediction. MAPE was used to aggregate forecast accuracy across segments. 3.5. Data Data to test the hypotheses were drawn from firm decisions, the results for each firm from running the simulation, and 3-year forecasts of unit size and leader’s price-by market segment for sonites, by total market for vodites. Forecasts for each sonite segment and the vodite market were collected from each team. Forecast data were collected at periods 0 through 5 together with the firm decisions; the six forecast/actual pairs comprised periods O/3; l/4; 21.5; 316; 417; 518 (five pairs for the seven period industries). In total, 2400 forecast/actual pairs were collected from the 14 industries. 3.6. Analysis Ordinary least squares (OLS) and logistic regression was used to test the hypothesized relationships. Firm and type of student program effects were entered as dummy covariates in each analysis. Data were omitted in those in- stances when students did not provide forecasts, when actual sales in vodites were zero (even though some groups provided forecasts) and when the APE exceeded 200%. Less than 4% of possible data were dropped, primarily in indus- tries with late introductions of vodites. For Hypothesis 1, the dependent variable was the target market segment indicator (0, 1); independent variables were, alternatively, fore- cast market segment size and forecast market segment growth rate together with relative mar- ket segment share and the covariate dummies. Two logistic regression models examined these relationships. Hypotheses 2 and 3 were tested simultaneous- ly via two OLS regression analyses in which the APES for forecast market segment size and forecast market leader’s price provided alter- native dependent variables. Independent vari- ables were the target market segment indicator (Hypothesis 2), level of competition and the target market segment indicator X level of competition interaction (Hypothesis 3). For each regression analysis, other independent variables were the covariate dummies, plus one-period lagged dependent variables that were included to account for potential autocorrelation effects. OLS was also used to model Hypothesis 4. The dependent variable was total firm sales in units (across all segments) for each period. Independent variables were the MAPE of fore- cast market segment size, the MAPE of forecast market segment leader’s price, a period indicator that captured inherent trends in the MARK- STRAT environment, and the firm and type of student effect covariate dummies. Several addi- tional covariates captured major strategic deci- sions that might reasonably have been expected to be related to firm performance: the number of sonite market segments targeted and the pres- ence/absence in vodites captured a diversifica- tion dimension; the research and development budget captured product line maturity; and the number of salespeople, and budgets for advertis- ing and market research captured other critical strategic dimensions in the MARKSTRAT simu- lation. 4. Results Hl: The firm’s selection of market segments to address is positively related to their market
  • 9. N. Capon, P. Palij I International Journal of Forecasting 10 (1994) 339-352 347 attractiveness, and to the firm’s current business strengths in those market segments. This hypothesis is supported for both oper- ationalizations of market attractiveness and for business strength (see Table 1). The positive and significant coefficients for forecast market seg- ment size and forecast market segment growth rate, and for relative market share in each of the two analyses, indicate that firms acted according to the received wisdom of strategic planning models: participate in market segments provid- ing large potential opportunities where the firm is currently strong. H2: The firm’s forecasts of strategic dimensions for target market segments are more accurate than for avoided segments. This hypothesis was tested by two regression analyses, one each for the APE of forecast market segment size and forecast market seg- ment leader’s price. Partial support was adduced for this hypothesis inasmuch as the coefficient for target market segment in the regression of fore- cast market segment leader’s price was both negative and significant (see Table 2). (The coefficient for target market segment was not significant for forecast market segment size.) The negative sign indicates smaller forecast errors, hence increased accuracy, for targeted market segments. Table 1 Relationship between market attractiveness, business strengths and market segment selection Independent variable Forecast market segment size Forecast market segment growth rate Relative market share Firm 1 Firm 2 Firm 3 Firm 4 Full-time MBA students Executive program participants Constant * pco.05; **p<o.o1. Coefficients 0.00227** 1.272*” 1.590** 1.628** -0.243 -0.253 -0.321 -0.294 -0.162 -0.199 -0.319 -0.336’ -0.183 -0.172 -0.159 -0.237 -0.205 0.559** Table 2 Relationship between market segment selection and strategic forecasting accuracy Independent variable Coefficients Forecast market Forecast market segment segment size leader’s price Targeted market 0.286 -0.0722** segment (TMS) Level of competition (LC) -0.672** -0.0593** TMS x LC One year lagged size One year lagged leader’s price -0.172 0.0241* 0.246”* 0.3so** Firm 1 Firm 2 Firm 3 Firm 4 Full-time MBA students -0.220 0.0076 -0.0028 0.0177 -0.174 -0.0010 -0.553 0.0120 -0.204 -0.0157 Executive program participants 0.0814 0.0107 Constant F (10.1899) Adjusted RZ * P<O.O5; P<O.Ol. 1.643** 0.205** 40.08** 54.03** 0.17 0.22 H3: The accuracy of the firm’s strategic forecasts in target market segments increases as the level of competition increases. This hypothesis was supported for forecast market segment leader’s price inasmuch as the coefficient for the target market segment x level of competition interaction was significant (see Table 2). However, the coefficient for the main effect of level of competition was negative and significant indicating that, regardless of whether or not the segment was targeted, the accuracy of the firm’s strategic forecasts improved as the level of competition increased. The key factor driving the interaction effect was the relatively low level of forecast accuracy in low competi- tion, non-targeted market segments. Forecast accuracy was highest for high competition, targeted market segments, significantly greater than for both targeted, low competition and non- targeted high competition segments; the latter two segments were not significantly different from one another.
  • 10. The main effect for level of competition was found also for forecast market segment size. The significant coefficients for one year lagged vari- ables in each regression, forecast market seg- ment size and forecast market segment leader’s price, indicate the presence of autocorrelation that these variables removed. H4: The greater the accuracy of the firm’s long- run strategic forecasts. the better its perform- ance. This hypothesis received mixed support (see Table 3). The MAPE for forecast market seg- ment leader’s price was significant and negatively related to performance, thus supporting the hypothesized relationship, but the MAPE for forecast market segment size was not significant. As expected, the period was significant, as were three of four firm dummies. The magnitude and direction of the firm coefficients reflect the typical rankings of firm performance; over a large number of industries, firms 1, 3 and 4 typically outperform firms 2 and 5, Table 3 Relationship between forecast accuracy and firm perform- ancc Independent variable Forecast market segment size (MAPE) Forecast market segment leader’s price (MAPE) Number of sonite segments Vodites Advertising hudget R&D budget Number of salespeople Marketing research hudgct Period Firm 1 Firm 2 Firm 3 Firm 4 Full-time MBA students Executive program participants Constant F (15,384) Adjusted R’ * P < 0.05; P < 0.01. Coefficients 0.012 --t).o665* 0.0074 0.0054 0.146** 0.0099** 0.0570** -0.0003 -0.0240** 0.0467** 0.0093 0.0309** 0.0242* 0.0123 0.0145 -0.0306 63.6** U.71 Three of the six strategic variables. research and development budget reflecting product line maturity, advertising budget and number of salespeople were also significant. The presence of the moderating variables in the equation did not change the direction or significance of the coefficient for forecast market segment leader’s price, nor affect the significance of forecast market segment size. The presence of these variables was, however, important in raising the R’ from 0.16 when they were not included, to 0.71. The difference in results between forecast market segment size and forecast market seg- ment leader’s price when testing Hypotheses 2, 3 and 4 may reflect differences in forecast task complexity. Thus, depending on the particular segment and industry, overall segment size changes for sonites through the simulation ranged from - 10% to + 150%. and period-to- period changes reached 100% on occasions. ~Period-to-period changes were even iarger for vodites since sales were zero at the start of the simulation.) By contrast, the market segment leader’s price demonstrated far less variance, in large part because the ideal point for price in each market segment moves in a constrained linear manner. 5. Discussion In this paper we set out to investigate several hypotheses related to strategic forecasting in the context of strategic planning models. We argued that for managers to make good strategic deci- sions, they must be able to make good judg- ments about the medium- and long-term en- vironment. In a sense, our work is an extension of Glazer et al.‘s (1989, 1990) important MARK- STRAT studies that found a modest positive relationship between forecast accuracy and firm performance. However, the forecasts collected by Glazer et al. were not particularly strategic. They used short-term (1 year) forecasts com- pared with our medium-term (3 year) forecasts that reflect one or two product development cycles. In addition, their three forecast variables
  • 11. N. Capon, P. Palij I International Journal of Forecasting 10 (1994) 339-352 349 (market size, average industry price, and number of marketed products) are not particularly re- lated to key strategic decisions at the market segment level, the focus of the MARKSTRAT simulation. By contrast, our forecast variables of market segment size and growth rate, and mar- ket segment leader’s price, are highly strategic. Furthermore, the majority of our subjects were practising managers versus their exclusive re- liance on MBA students; and, finally, our find- ings are based on 14 industries whereas their data is drawn from just four industries. Despite the inherent difficulties in forecasting beyond a development cycle, our hypotheses were largely, though weakly, supported. In the context of portfolio models of strategic planning, we found that firms tended to select those market segments that provided the greatest op- portunities in terms of fast growth and high potential size, where the firm also had substan- tial business strengths as indicated by its relative market share (Hypothesis 1). The positive impli- cation from this finding is that portfolio dimen- sions are relevant to market segment selection; the negative implication is that misperceptions about market segment size and growth rate may lead to poor segment selection and risk the problems associated with escalation of commit- ment (Brockner et al., 1986). Our results also indicate that the differentially polar perspectives taken by neoclassical econ- omics (positive) and traditional forecasting re- search (negative) regarding long-term forecasting accuracy may be premature. Consistent with a bounded rationality perspective that managers cannot evaluate each segment in each period with the same degree of effort, we find that to some extent salience of the forecasting context impacts forecast accuracy. Thus, subjects were better able to forecast market segment leader’s price in those market segments to which they had committed resources (Hypothesis 2). Such differential forecast accuracy, as with misperceiv- ing market segment potentials, risks a focus on market segments to which prior commitments have been made, thus raising the spectre of an escalating commitment cycle with potentially devastating results. Indeed, as managerial atten- tion focuses on current market segments, oppor- tunities further afield may be missed. We also found that, consistent with our per- spective that salience increases forecasting ac- curacy, forecast accuracy for market segment leader’s price was greatest in those market seg- ments that were both targeted, and in which the firm faced high levels of competition. An un- expected result was the finding that managers were better able to forecast both market segment size and market segment leader’s price the more competitively intense the market segment, re- gardless of whether the segment was targeted. One explanation for these results is that differ- ing market segment profiles lead to differing salience levels for managers. Targeted competi- tive segments command the most attention because committed resources are at high risk. By contrast, non-targeted low competitive segments command the least attention not only because the firm has not committed resources but because another firm (or firms) dominance makes it unlikely that the firm will enter in the future. Intermediate salience levels result from either committed resources, but low risk because of low competition, or in high competition seg- ments, indicating opportunities for entry, that are currently non-targeted. Such an explanation accounts for both the main effect for level of competition and for the interaction effect for forecast market segment leader’s price. Finally, we secured partial support for a posi- tive relationship between strategic forecast ac- curacy and firm performance. This result is consistent with the generally positive, but mixed, results on the relationship between strategic planning and firm performance (e.g. Capon et al., 1987, 1994). Briefly stated, this result dem- onstrates that effort spent on the forecasting dimension of strategic planning pays off. An important issue in testing performance relationships such as these is model specification error. Thus, in his meta-analytic review of the planning-performance literature, Boyd (1991) highlights the absence of moderating variables as one of six major limitations of these studies. In an attempt to deal with this problem, Capon et al. (1987) employed four moderating variables in
  • 12. 350 N. Capon, P. Palij i International Journal of Forecasting 10 (1 YW) .33Y-3.52 their study of planning-performance relation- ships in the Fortune500(scale, degree of diversi- fication, product line maturity and industry). In this study, industry is an irrelevant variable as all firms operate in the same industry. However, firm size was accounted for by the firm and period dummies; degree of diversification by the number of sonite segments targeted and pres- ence/absence of vodite variables; and product line maturity by the research and development budget. In addition, we captured the other major strategic variables in the MARKSTRAT simula- tion, namely, number of salespeople, and the advertising and marketing research budgets. Additional methodological issues relate to the five other major criticisms of planning-perform- ance studies were noted by Boyd. Notwithstand- ing its simulation basis, compared with real world forecasting, this study did not use small and non-representative samples; collected data directly from the principals competing in the simulation rather than from surrogates; mea- sured forecasting error as a ratio-scaled variable compared with categorical or simple ordinal variables; used time series versus cross-sectional analyses; and used multiple forecasting in- dicators. Two important methodological issues are raised regarding the use of MARKSTRAT simu- lations: the need to account for firm effects and the use of student subjects. First, the frequent, and in the case of performance-related issues highly significant, presence of firm effects sug- gests that we should critically examine results from MARKSTRAT simulations that do not test for such effects (Ross, 1987). However, the typical criticism of using MBA subjects in studies such as these is somewhat mitigated by our results. In none of the five analyses (regression and logistic regression) was any student program variable significant. In other words, there is no evidence that full-time MBA students behaved any differently from the two executive groups. Perhaps the reason for this finding is the increas- ing maturity of MBA students who in this study averaged 27 years of age and had significant industrial experience. Notwithstanding the relatively positive results from this study, several limitations are evident. First, even though the MARKSTRAT simula- tion provides a realistic simulation of the real world, it is only a simulation. We cannot make the leaps to real world practice; real world results must await well-designed field studies. Second, even though we were able to support several hypotheses relating to strategic forecast- ing, we did not investigate the methods by which these strategic forecasts were made. Subjects in the study could have used a variety of different methods in the general categories of judgment, counting, time-series and association (Georgoff and Murdick, 1986). Future work should investi- gate the effects of different ways of making strategic forecasts. In addition, echoing Paich and Sterman (1993), methods for measuring the intrinsic difficulty of forecasting across market segments are needed. Such effort would also help address the need for a dynamic measure of market competitiveness. Third, the MARKSTRAT environment is rel- atively benign. The product life cycles for sonites and vodites are fairly predictable and, in general, market segments evolve in an orderly manner; major disturbances are driven by the activity of direct competitors. These conditions correspond to ‘convergent’ periods in Tushman and Romanelli’s (1985) punctuated equilibrium model of organizational development. Tushman and Romanelli also hypothesize that firms enter periods of ‘reorientation’ (such as IBM and General Motors entered in 1992), driven by sustained low performance and/or rapid environ- mental changes, which result in discontinuous and radical shifts in core values and beliefs, firm mission and strategies, and in internal organiza- tion and management processes. In their model, organizations progress through convergent periods, punctuated by reorientations that de- mark and set bearings for the next convergent period. This study is essentially conducted during periods of convergence where environmental change is incremental in nature and there is considerable opportunity for feedback learning. We did not study periods of reorientation, driven by major environmental shifts and shocks, where
  • 13. N. Capon, P. Palij I International Journal of Forecasting IO (19%) 339-352 351 competitive challenges come from new direct entrants, indirect competitors, and suppliers and customers engaged in vertical integration strate- gies. The problems of strategic forecasting in periods of reorientation, with its limited possi- bilities for feedback learning, are much more difficult. However, arguably the payoff to good strategic forecasting in these periods is at its greatest. Acknowledgments The authors would like to thank Don Lehmann, Bari Harlam and two reviewers for comments on earlier drafts of this paper. References Adams, F.G., 1986, The Business Forecasting Revolution: Nutiorz-intiustry-Firm (Oxford University Press, New York). Armstr~~ng, J.S., IYXZ, The value of formal planning for strategic decisions: Review of empirical research. Sfraregic Manugement Journal. 3, 197-211. Armstrong. J.S., 1986, Research in forecasting: A quarter- century review, 1960-1984, Interfaces, 16, 89-109. Armstrong, J.S.. lY88, Research needs in forecasting, Inter- national Journal of Forecasting. 4, 449-465. Armstrong, J.S., R.J. Brodie and S.H. McIntyre. 1987, Forecasting methods for marketing, International Journal of Forecasting, 3, 355-375. Ascher, W., 1978, Forecasting: An Appraisal for Policy Makers und f1anner.s (Johns Hopkins University Press, Baltimore, MD). Boyd, R.K., 1991, Strategic planning and financial perform- ance: A mcta-analytic review, ~oi~r~lal of management Studies. 28. 353-374. Brockner. J.. R. Homer. G. Birnbaum. K. Lloyd, J. Deit- cher, S. Nathanson and J.Z. Rubin, iY86. Escalation of commitment to an ineffective course of action: The effect of feedback having negative implications for self-identity, Administrative Science Quarterly, 31. 109-126. Burke, M.C.. 1984, Strategic choice and marketing mana- gers: An examination of business-level marketing objec- tives, Journal of Marketing Research, 21. 345-359. Capon, N. and J.M. Hulbert. 1985, The integration of forecasting and strategic planning, International Journal of Forecasting, 1. 123-133. Capon, N., J.U. Farley and J.M. Hulbert, 1987, Corporate Strategic Planning (Columbia University Press, New York). Capon. N., J.U. Farley and J.M. Hulbert, 1994, Strategic planning and firm performance: More evidence. Journal of Management Studies, 31, 105-110. Chakravarti. D., A. Mitchell and R. Staelin, 1981, Judgment based marketing decision models: Problems and possible solutions, Journal of Marketing, 4.5, 13-23. Cortes-Rello, E. and F. Golshani, 1990, Uncertain reasoning using the Depster-Shafer method: An application in fore- casting and marketing management, Expert Systems, 7, Y-17. Day, G.S., 1986. Analysis for Strategic Market Decisions (West Publishing, St. Paul. MN). Day, G.S., 1990. Market Driven Strategy: Processes for Creating Value (Free Press. New York). Dutton, J.E.. L. Fahey and VK. Na~~nan, 1983, Toward understanding strategic issue diagnosis, Strategic Manuge- ment Journal, 4, 307-323. Eisenhardt, K.M. and M.J. Zbaracki, 1992, Strategic deci- sion making. Strategic Management Journal, 13, 17-37. Fredrickson. J.W. and T.R. Mitchell, 1984, Strategic decision processes: Comprehensiveness and performance in an industry with an unstable environment, Academy of Man- agement Journal, 27. 399-423. Georgoff, D.M. and R.G. Murdick, 1986, Manager’s guide to forecasting, Narvard Business Review. 64, 110-120. Glazer, R., J.H. Steckel and R.S. Winer. 1989, The forma- tion of key marketing variable expectations and their impact of firm performance: Some experimental evidence, Marketing Science. 8. 18-34. Glazer, R., J.H. Steckel and R.S. Winer, 1990, Judgmental forecasts in a competitive environment: Rational vs. adap- tive expectations, International Journal of Forecasting, 6, 149-162. Green, D.H. and A.B. Ryans, 1990, Entry strategies and market performance: Causal modeling of business simula- tion, Journal of Product Innovation Management, 7, 45- 58. Greenley, G.E. and 3.L. Bayus, 1993. Marketing planning decision making in U.K. and U.S. companies: An empiri- cal comparative study, Journal of Marketing Management, forthcoming. Hammond, K.R., 1987, Toward a unified approach to the study of expert judgment, in: J.L. Mumpower, 0. Renn, L.D. Phillips and V.R.R. Uppuluri, eds., Expert Jl~dgment and Expert S.vstems. NATO ASI Series, Volume F35, (Springer-Vertag, Berlin) l-16. Hannan, M.T. and J.H. Freeman, 1977, The population ecology of organizations, American Journal of Sociology, 82, 929-964. Henderson, B.D. and A.J. Zakon. 1980, Corporate growth strategy: How to develop and implement it, in: K.J. Albert, ed., Handbook of Business Problem Solving (McGraw-Hill, New York). Hogarth, R.M. and S. Makridakis. 1981, Forecasting and planning: An evaluation, Managemenf Science, 27, 11% 138.
  • 14. 352 N. Capon. P. Palij I International Journal of Forecasting 10 (1Wd) 339-352 Hulbert, J.M., IYXS. Marketing: A Strutegic Perspective (Impact Publishing Co., Katonah, NY). Jacobson, R., 1988, Distinguishing among competing theories of the market share effect. Journal of Marketing. 52. 6X-80. Jacquemin. A.P. and C.H. Berry. 1Y7Y. Entropy measure of diversification and corporate growth. Journal of Industrial Economics. 27. 359-36’). James, S.. T.C. Kinnear and M. Deighan, 19Yl. ISf supple- ment to Markstrat Z: Player’s guide (Interpretive Software Inc., Charlottesville, VA). Kinnear. T.C. and S.K. Klammer, lY87, Management pcr- spectives on Markstrat: The GE expcricnce and beyond, Journal of Business Research. IS. 4YlLSOl. Kotler. P.. 19X4. Marketing Management: Ana1ysi.y. Planning and Control. Fifth Edition (Prentice-Hall. Englewood Cliffs. NJ). Larreehe. J. and H. Gatignon. 1990. MARKSTRAT 2. (Scientific Press. Palo Alto. CA). Little. J.D.C. and L.M. Lodish. 1981. Commentary on Judgment Based Marketing Decision Models, Journal of Marketing, 45, 24-29. Mahmoud. E.. 19X6. The evaluation of forecasts, in: S. Makridakis and S.C. Wheelwright. eds.. The Handbook of Forecasring: A Manager’s Guide. Second Edition (John Wiley & Sons. New York. NY). 504-516. Mahwan, V. and Y. Wind. 1988. New product forecasting models: directions for research and implementation, Inter- national Journal of Forecasting. 4, 341-358. Makridakis, S.. 19X6, The art and science of forecasting, International Journal of Forecasting, 2. 15-39. Paich, M. and J.D. Sterman. lY93. Boom. bust. and failures to learn in experimental markets. Managemenr Science. 3Y. 1439-145X. Palepu, K.. lYX5, Diversification strategy. profit performance and the entropy measure. Strategic Manugemeni Journal, 6. 239-25s. Pfeffer. J. and G.R. Salancik. lY7X. The External Control of Organizations: A Resource Dependence Perspective (Harper & Row. New York). Ross. W.. 1087. A re-examination of the results of Hogarth and Makridakis’ ‘The value of decision making in a complex environment: An experimental approach’. Man- agement Science. 33. 2X8-296. Secger. J.A., 19x4. Reversing the images of BCG’s growth/share matrix. Straregic Management Journal, 5. Y3- 97. Simon. H.A., 19.57, Models of Man: Social and Rational (John Wiley & Sons. New York). Simon, H.A., lY7X. “Rationality as process and as product of thought”. American Economic Review, 68, I-16. Simon, H.A.. 1979, Rational decision making in organiza- tions, American Economic Review. 69, 493-513. Singer, A.E. and R.J. Brodie. 1090. Forecasting competitors’ actions: An evaluation of alternative ways of analyzing business competition, International Journal of Forecasting, 6, 75-88. Sterman. J.D., lY87. Testing behavioral simulation models by direct experiment. Manqement Science. 33. 1572-1592. Stigler. G.J.. 1983, The Organization of Industry (University of Chicago Press. Chicago. IL). Thorndike. E.L.. 1918. Fundamental theorems in judging men, Journal of Applied Psychology, 2. 67-76. Tushman. M.L. and E. Romanclli, 1985, Organizational evolution: A metamorphosis model of convergence and reorientation, in: L.L. Cummings and B.M. Staw. eds.. Research in Organizational Behavior, Volume 7 (JAI Press Inc., Greenwich, CT), 171-222. Urban. G.L., J.R. Hauser and J.H. Roberts, 1990. Pre- launch Forecasting of New Automobiles, Marketing Sci- ence, 36, 401-321. Wenslcy. R.. lYX2. PIMS and BCG: New horizons or false dawn?. Strategic Mana,qement Journal, 3. 147-158. Wind, Y.. IYXl, Marketing oriented strategic planning models, in: R.L. Schultz and A.A. Zoltners, eds., Market- ing Decision Models (North Holland, New York. NY). 207~250. Wind, Y. and T.S. Robertson, 1YX3. Marketing strategy: New directions for theory and research, Journal of Marketing. 47. 12-25. Biographies: Noel CAPON is Professor of Business at Columbia Business School. His teaching and research inter- ests are in the areas of strategy and planning. and sales force management. He is coauthor of Corporate Strategic Planning, a study of planning practices in major U.S. manufacturing corporations. A companion book on the underpinnings of corporate financial performance is in an advanced stage of preparation. Hc acknowledges support from the Redwoord Foundation. Peter PALIJ is a doctoral candidate at Columbia Business School. His teaching and research interests are in the areas of marketing strategy. decision making and competition. He acknowledges support from the Institute for the Study of Business Markets.