See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/46556287

Long-term Profit Impact Of Integrated Marketing Communications Program

Article  in  Review of Marketing Science · January 2004

DOI: 10.2202/1546-5616.1014 · Source: RePEc

CITATIONS READS
35 1,546

2 authors:

Kalyan Raman Prasad Naik

Northwestern University University of California, Davis
7 PUBLICATIONS   280 CITATIONS    58 PUBLICATIONS   3,410 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Sethi Advertising Model View project

All content following this page was uploaded by Prasad Naik on 04 June 2014.

The user has requested enhancement of the downloaded file.

 Review of Marketing Science
Volume 2 2004 Article 8

Long-term Profit Impact Of Integrated

Marketing Communications Program

Kalyan Raman∗ Prasad A. Naik†

∗
Loughborough University, [email protected]
†
University of California Davis, [email protected]

Copyright 2004the
c authors. All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical,
photocopying, recording, or otherwise, without the prior written permission of the publisher, be-
press, which has been given certain exclusive rights by the author. Review of Marketing Science
is produced by The Berkeley Electronic Press (bepress). http://www.bepress.com/romsjournal
 Long-term Profit Impact Of Integrated
Marketing Communications Program
Kalyan Raman and Prasad A. Naik

Abstract

The concept of Integrated Marketing Communications (IMC) emphasizes the role of synergy,
which arises when the combined effect of multiple activities exceeds the sum of their individual
effects. In this paper, we investigate the effects of synergy on the profitability of IMC programs in
uncertain markets. We develop a dynamic multimedia model that incorporates both synergy and
uncertainty, and use it to determine the optimal IMC program. Our results generalize previous
findings to uncertain markets, illuminate the profit implications of IMC programs, and explain
the catalytic effects of synergy in IMC contexts. Specifically, we find that the expected long-
term profit of the advertised brand increases as synergy increases. Furthermore, managers should
allocate a non-zero budget to a catalytic activity even if it is completely ineffective. Finally, these
findings continue to hold in an uncertain duopoly market.

KEYWORDS: Marketing Communications, Stochastic Optimal Control, Synergy

Raman and Naik: Long Term Profit Impact of IMC 1

1. INTRODUCTION

The proliferation of media into specialized magazines, cable programming, in-

transit advertisements and the Internet has fragmented the readership or viewing
audience, placing greater demands on consumers’ attention, and thereby eroding
the influence of mass advertising. In response to media proliferation, marketers
are attempting to increase the impact of their communications program through an
Integrated Marketing Communications (IMC) perspective, hoping to harness
synergies between media (see, e.g., Belch and Belch 1998, p. 11). Synergy arises
when the combined effect of two activities exceeds the sum of their individual
effects, a phenomenon in which “the whole is greater than the sum of the parts.”
Naik and Raman (2003) show how to estimate synergy and understand its effects
on media budget and allocation. Their analysis reveals that advertisers should
increase the media budget in the presence of synergy and allocate the increased
media budget disproportionately in favor of the less effective medium. More
importantly, using Naik and Raman’s (2003) IMC model, Schultz and Pilotta
(2004) further enhance our understanding of “how media advertising works in the
interactive, networked, global systems found in the 21st century media
marketplace.”
This IMC model, however, assumes that the impact of media spending on
sales is deterministic, but this assumption is sometimes untenable, for instance in
turbulent, volatile markets in which uncontrollable factors affect sales. How then
should advertisers plan IMC programs under market uncertainty? What are the
long-term profit implications of the IMC program in uncertain markets? The need
for research on the implications of uncertainty for long-term profitability has been
stressed by the MAX initiative ⎯ Managing Advertising Expenditures for
Financial Performance ⎯ sponsored by both the American Association of
Advertising Agencies and the Marketing Science Institute (Farris, Shames and
Reibstein 1998). We respond to this need by elucidating the impact of synergy on
profitability of IMC programs in uncertain markets. In addition, we augment the
extant literature with insights into the catalytic effects of synergy.
To this end, we include uncertainty in the Naik and Raman (2003) model
of IMC via the Wiener process in order to represent error in their continuous-time
dynamic model. Applying stochastic optimal control theory, we then derive the
optimal IMC program incorporating cross-media synergy under uncertainty.
Next, on the basis of comparative static analyses, we find that the expected long-
term profit of the advertised brand increases as cross-media synergy increases.
Furthermore, cross-media synergy neither increases nor decreases the variability
in long-term profit. Hence, ad agencies and brand managers can enhance
synergies between multiple media to increase the brand’s long-term profitability,
without influencing the variability of profits.

Produced by The Berkeley Electronic Press, 2008

2 Review of Marketing Science Vol. 2 [2004], Article 8

Our fundamental result pertains to the importance of ancillary activities in

the communications mix, even when such ancillary activities may appear to have
negligible impact on sales. Specifically, advertisers should use activities such as
event sponsorship, free-samples and collaterals, in-transit advertising or
merchandising because these ancillary activities enhance the effectiveness of
primary activities through synergistic interactions. For example, BMW places its
sports car in Bond movies not because product placement (an ancillary activity)
directly increases car sales, but because it interacts with other marketing
activities, thereby enhancing visibility of the brand. In other words, ancillary
activities are catalysts that positively influence sales growth although they lack
direct sales impact. We later define these so-called catalytic effects of synergy,
and explain why managers should allocate a non-zero budget to an ancillary
activity even if it is ineffective, as measured by its direct linkage to sales.
The rest of the paper is organized as follows. Section 2 reviews the extant
literature on integrated marketing communications. Section 3 incorporates
uncertainty in the IMC model and formulates the budgeting problem. Section 4
generalizes the previous results and presents new results on long-term profitability
and catalytic effects of synergy. Section 5 further extends the IMC model and
generalizes the findings to duopoly markets; it also discusses the effects of
uncertainty and advertising.

2. LITERATURE REVIEW

An abundant literature on the effects of advertising on sales exists in

marketing science. However, knowledge about the joint effects of multimedia
advertising is quite limited (see Gatignon 1993, Mantrala 2002). For example, in
their comprehensive literature review, Feichtinger, Hartl and Sethi (1994, p. 219)
observe that, “with a few exceptions, the models assume … single advertising
medium. This was already noted by Sethi (1977), and this critical remark is still
valid for the literature published subsequently.”
To gain insights into joint effects of multimedia advertising, a consortium
of radio network companies conducted a field study by sampling 500 adults, ages
20-44, and across 10 locations in Britain. The main findings indicate that 73% of
the participants remembered prime visual elements of TV ads upon hearing radio
commercials. In addition, 57% re-lived the TV ads while listening to the radio
advertisement. Thus, radio ads reinforce the imagery created by TV commercials,
resulting in synergy between television and radio advertising.1 In controlled

1
For additional information, contact the Radio Advertising Bureau or visit
www.rab.co.uk.

http://www.bepress.com/romsjournal/vol2/iss1/art8
Raman and Naik: Long Term Profit Impact of IMC 3

laboratory experiments, Edell and Keller (1999) analyze interactions between TV

and print advertising, and find significant effects of cross-media synergy. More
recently, Naik and Raman (2003) investigate the effects of synergy in an
empirically validated model of multimedia advertising, and provide several
insights based on their theoretical analyses.
Specifically, Naik and Raman (2003) estimate a dynamic sales response
model using market data on Dockers® brand advertising, and establish the
presence of synergy between TV and print ads in consumer markets. Then they
theoretically show that media budget increases as the magnitude of synergy
increases. This finding provides a new perspective on the frequently debated
issue of over-advertising. The marketing literature (see Hanssens et al. 1998, p.
260) suggests that advertisers tend to overspend on advertising. However, a
response model that ignores the effects of synergy understates the optimal budget.
Hence, what appears to be overspending would represent an appropriate spending
level when we account for synergy between multiple media.
More importantly, Naik and Raman (2003) find that budget allocation
across media differs qualitatively in the presence of synergy. As synergy between
two media increases, the proportion of budget allocated to the more (less)
effective medium decreases (increases). To understand this counter-intuitive
result, consider a market with no synergy. Here each medium independently
increases brand sales, and so the optimal spending on each medium depends on its
own effectiveness only. In the presence of synergy, however, the effectiveness of
each medium depends not only on own effectiveness, but also on the spending
level of the other medium. Consequently, as synergy increases, marginal
spending on a medium increases at a rate proportional to the spending level of the
other medium. Therefore, the optimal spending on the more effective medium
increases slowly, while the optimal spending on the less effective medium
increases rapidly. Hence, the proportion of budget allocated to the more effective
medium decreases as synergy increases.
However the above results assume that sales evolve deterministically in
response to multimedia advertising. Consequently, it is natural to question
whether these results would hold in turbulent and volatile markets, where sales
growth is uncertain. Uncertainty in sales response raises questions of significant
interest. How does synergy influence the long-term profitability of the brand’s
IMC program under uncertainty? Does synergy affect the variability in the
brand’s long-term profitability? To address these issues, we next extend the IMC
model to uncertain markets.

Produced by The Berkeley Electronic Press, 2008

4 Review of Marketing Science Vol. 2 [2004], Article 8

3. EXTENDED IMC MODEL

3.1 Incorporating Uncertainty

Consider the IMC model specified by the equation,

dS
= β1 u ( t ) + β 2 v( t ) + κu ( t ) v( t ) − (1 − λ) S( t ) (1)
dt
where S(t) denotes brand sales at time t, dS/dt is the instantaneous rate of change
in sales (i.e., sales growth), u(t)and v(t) are advertising efforts on two different
media (e.g., television and print), β1 and β2 are the effectiveness of the two
media, κ measures the magnitude of synergy between the two media, and λ
represents the usual carryover effect. See Naik and Raman (2003) for further
details. Equation (1) proposes a two-fold influence of each medium on the sales
rate: direct effects arising from media efforts (β1, β2), and an indirect effect driven
by synergy (κ). Thus, when κ > 0, the combined sales impact of (u, v) exceeds
the sum of the independent effects (β1u + β2v).
To incorporate uncertainty, we re-write equation (1) in the equivalent
differential form, suppressing the variable “t” for notational clarity,
dS = {β1 u + β 2 v + κ u v − (1 − λ) S}dt , (2)
and then introduce the continuous-time error process dW to obtain
dS = {β1 u + β 2 v + κ u v − (1 − λ) S} dt + σ dW . (3)
In equation (3), the error term dW is the differential increment of W(t), a
stochastic process known as Brownian motion (or alternatively as a Wiener
process), and defined by the twin properties: (a) for every (t1, t2), t1 < t2, the
increment W(t2) – W(t1) has zero mean; and (b) for every (t1, t2), t1 < t2, the
increment W(t2) – W(t1) is normally distributed with variance equal to (t2 – t1).
Hence, as in econometric models, the random error term dW follows the normal
distribution with E[dW] = 0 and Var[dW] = σ2dt (see Arnold 1974, p. 41, 45).
This methodology finds precedence in Rao (1986) who also used a Brownian
motion process to capture uncertainty in a continuous-time sales advertising
model.
Equation (3) thus captures the effects of both IMC and uncertainty in a
continuous-time framework. Next, we derive the optimal IMC program and
evaluate its long-term profitability using stochastic control theory (see Raman
1990, Raman and Chatterjee 1995, Mantrala, Raman and Desiraju 1997).

3.2 Optimal IMC Program and Long-term Profitability

The advertiser’s problem is to determine the dynamically optimal IMC

program under synergy and uncertainty. Let u* and v* denote the optimal effort

http://www.bepress.com/romsjournal/vol2/iss1/art8
Raman and Naik: Long Term Profit Impact of IMC 5

invested in the two media. The advertiser determines (u*, v*) by maximizing the
expected discounted profit over an infinite horizon, namely,
∞
E s [ ∫ e −ρt Π (S( t ), u ( t ), v( t ))dt ] , (4)
0

subject to sales evolution via the stochastic differential equation (3), where Es[⋅] is
the expectation conditional upon the initial sales S(0) = s, ρ denotes the discount
rate, Π(s, u, v) = ms − u2 – v2 is the profit function, and m is the profit margin
(see, e.g., Fruchter and Kalish 1998).
The optimal IMC program may be derived either by stochastic dynamic
programming or the Lagrange method. Chow (1997) observes that the Lagrange
method is often simpler to use, easier to interpret and more direct since it gives
the minimum information needed to find optimal policies. However, in our case,
the stochastic dynamic programming method is preferable for two fundamental
reasons: (i) the Lagrange method yields only the slope of the maximum expected
profit function (also known as the value function), but that would be inadequate
for our purposes because information on just the slope is insufficient to analyze
the long-term profitability of the IMC program, and (ii) the stochastic dynamic
programming approach is well-suited to find closed-loop policies under
uncertainty.
We implement stochastic dynamic programming through Hamilton-
Jacobi-Bellman theory and derive:
⎡ ∞ ⎤
u * (θ) = ArgMax ⎢E s ∫ e −ρt Π (S( t ), u (S( t )), v(S( t )))dt ⎥ , (5a)
u ,v ⎣ 0 ⎦
⎡ ∞ ⎤
v * (θ) = ArgMax ⎢E s ∫ e −ρt Π (S( t ), u (S( t )), v(S( t )))dt ⎥ , (5b)
u ,v ⎣ 0 ⎦
where the vector θ = (β1 , β 2 , κ, λ, σ, ρ, m)′ contains all the model parameters. In
equations (5a, b), the notation u(S(t)) and v(S(t)) signifies the closed-loop nature
of strategies, while u*(θ) and v*(θ) remind us that the resulting optimal strategies
depend on model parameters (e.g., magnitude of synergy, κ).
Next, we substitute the optimal strategies (u*, v*) in equation (4) to
evaluate the profitability of the IMC program. Let J(s) denote the maximum
profit attained when we use the optimal IMC strategies throughout the planning
horizon, starting from an arbitrary sales level s. Then, the expected long-term
profitability (ELP) is given by the limiting value
ELP = Lim E[J (S( t ))] , (6)
t→∞
where the expectation E[⋅] integrates over all possible sales levels S = s.
Similarly, we evaluate the variability of long-term profitability (VLP) by

Produced by The Berkeley Electronic Press, 2008

6 Review of Marketing Science Vol. 2 [2004], Article 8

VLP = Lim Var[J (S( t ))] . (7)

t →∞
Finally, we use equations (5a), (5b), (6) and (7) to derive new results.

4. NORMATIVE RESULTS

Here we present four new propositions, whose proofs we relegate to the

Appendices in order to maintain continuity of exposition.
We note that the key results of Naik and Raman (2003) were derived for
deterministic markets; that is, when σ 2 = 0 . In contrast, proposition 1 generalizes
those findings to uncertain markets where σ 2 ≠ 0 . We simply state the results
here because we already discussed their implications and intuition in the literature
review.

PROPOSITION 1. In uncertain markets, the total media budget increases as

synergy increases. Furthermore, the proportion of budget allocated to the more
(less) effective medium decreases (increases) as synergy increases.
PROOF. In Appendix A, we prove that, for every σ 2 ≠ 0 (i.e., in the presence of
uncertainty),
∂ *
(i) (u + v * ) > 0
∂κ
∂ u* ⎧− if β1 > β 2
(ii) ( *)=⎨ .
∂κ v ⎩+ if β1 < β 2

The next two propositions elucidate the bottom-line implications of IMC

programs.

PROPOSITION 2. In uncertain markets, the expected value of long-term

profitability (ELP) of the optimal IMC program increases as synergy increases.
∂ ELP
PROOF. In Appendix B, we prove that > 0 for every σ 2 ≠ 0 .
∂κ

PROPOSITION 3. In uncertain markets, the variability of long-term

profitability (VLP) of the optimal IMC program is unaffected by the magnitude of
synergy.
∂ VLP
PROOF. In Appendix C, we prove that = 0 for every σ 2 ≠ 0 .
∂κ

The pragmatic implication of Propositions 2 and 3 is that brand managers should

adopt an IMC perspective to increase the brand’s profitability. That is, they

http://www.bepress.com/romsjournal/vol2/iss1/art8
Raman and Naik: Long Term Profit Impact of IMC 7

should think of marketing communications not as a set of independent activities,

but rather as a set of interconnected activities with potential synergies. By
integrating these activities to build synergies, they not only increase the expected
profitability in the long run, but also keep profit variability unaltered. Thus
synergy imposes no tradeoff between profitability and variability. Consequently,
an IMC perspective raises profit but leaves its variability unaffected, and so it
behooves managers to build synergies.
One approach for building synergies is to capitalize on the catalytic effects
of synergy by using ancillary activities, as in the example of BMW’s usage of
product placement in Bond movies. We next define this catalysis concept, and
state its managerial significance.

DEFINITION. A marketing activity is a catalyst if it has negligible direct effect on

sales, but exerts substantial synergies with other activities. For example, in
equation (3), the activity v is a catalyst in the media mix (u, v) if the direct effect
β2 ≈ 0, but the synergy κ ≠ 0.

PROPOSITION 4. An advertiser should allocate a non-zero budget to the

catalytic activity even if it is completely ineffective.
PROOF. In Appendix D, we prove that v * ≠ 0 even if β2 = 0.

This result demonstrates that advertisers must act differently when they adopt an
IMC perspective. According to models of advertising that ignore synergy effects,
an advertiser allocates the total budget to various media in proportion to their
relative effectiveness (e.g., see proposition 1 in Naik and Raman 2003), and so the
medium that is completely ineffective receives zero budget; i.e., when β2 = 0, v* =
0. In contrast, an advertiser who implements an IMC program benefits from not
only the direct effects, but also the joint effects of various activities. Therefore,
they should not eliminate spending on an ineffective medium if it enhances the
effectiveness of other activities due to its catalytic presence. Next we present
cases from industrial and pharmaceutical marketing to exemplify catalytic
effects.2
When industrial marketers advertise in trade journals, they do not make
purchasing managers directly buy the advertised products, but rather such
advertising tends to enhance the effectiveness of sales calls by enhancing
familiarity of the brand or company. Consider the following example from the
pharmaceutical industry. The distribution of product samples or collateral

2
We thank Kash Rangan and Alvin Silk for discussions that led to these insights
and the industrial marketing example.

Produced by The Berkeley Electronic Press, 2008

8 Review of Marketing Science Vol. 2 [2004], Article 8

materials to physicians does not increase the sales of prescription medicines, but it
enhances the effectiveness of detailing efforts by sales representatives (Parsons
and Vanden Abeele 1981). Indeed, many ancillary activities such as billboards,
publicity, corporate advertising, event sponsorship, in-transit ads, merchandising,
or product placement may not increase sales directly. Yet advertisers spend
millions of dollars on them. Why? Because these activities are catalysts that
enhance the effectiveness of primary activities (e.g., advertising or salesforce
efforts) by strengthening brand knowledge in consumers’ memory (Keller 1998,
p. 257).

5. DISCUSSION

Here we further extend the above analyses to duopoly markets and elaborate both
the influence of uncertainty and the relative roles of commitment and feedback in
media buying.

5.1 Role of competition

We express the IMC model (1) in terms of market share X(t), rather than
sales, to obtain a duopoly IMC model3,
dX = {β1 u 1 + γ 1 v1 + κ1 u 1 v1 − β 2 u 2 − γ 2 v 2 − κ 2 u 2 v 2 − δ(2X −1)}dt + σ(X)dW ,
(8)
where X is market share of brand 1, ui and vi are brand i’s advertising decisions, i
= 1 and 2, δ captures the rate of “churn” in a competitive market (Prasad and
Sethi 2003), and σ(X)dW is the stochastic component. Note that, with probability
one, X(t) ∈ (0,1), provided β1 u 1 + γ 1 v1 + κ1 u 1 v1 − β 2 u 2 − γ 2 v 2 − κ 2 u 2 v 2 < δ and
σ(X) > 0 for all X, σ(0) = 0, σ(1) = 0. In words, these mathematical conditions
prevent the duopoly from collapsing into a monopoly.4 Furthermore, because
competition for brand share is a zero-sum game, we do not specify a separate

3
We thank an anonymous reviewer for suggesting this model structure.
4
More specifically, Gikhman and Skorohod (1972, p. 158) show that the process
dX = µ(X)dt + σ(X)dW will remain within (r1, r2) provided that µ(r1) > 0, µ(r2) <
0, σ(X) > 0, for X(t) ∈ (r1, r2), σ( r1) = 0, and σ( r2) = 0. In our model, µ(r1) = F1
– F2 + δ <0, µ( r2) = F1 – F2 − δ >0, where Fi = β i u i + γ i v i + κ i u i v i , i = 1, 2, r1 =
0, and r2 = 1. The conjunction of these two conditions gives |F1 – F2| <
δ. Following Sethi (1983), we can impose additional conditions to drive the
infinitesimal variance to zero as X approaches the boundary of (0, 1), for
example, σ(X) = X(1−X).

http://www.bepress.com/romsjournal/vol2/iss1/art8
Raman and Naik: Long Term Profit Impact of IMC 9

equation for the second brand. Each brand then maximizes its expected
∞
discounted net profit, E x i [ ∫ e −ρi t {m i q y i − (u i2 + v i2 )}dt ] , where y1 = x1 and y2 = 1
0
– x1, taking into account the share dynamics in (8) and the best response of the
other brand. In the above integrand, q denotes the industry sales, and ρi and mi
denote the discount rate and profit margin of brand i, respectively. To this end,
we derive the Hamilton-Jacobi-Bellman equation for each brand, follow the same
k =n
logic that led to equation (A1), consider the general polynomial J i ( x ) = ∑ J ki x ik ,
k =0

and find that n = 1 provides the optimal solution. Thus, the optimal IMC program
for the first brand is given by
* qm1{2(2δ + ρ1 )β1 + qm1 γ 1 κ1 }
u1 = 2 2
, and (9)
4(2δ + ρ1 ) 2 − q 2 m1 κ1
* qm1{2(2δ + ρ1 ) γ 1 + qm1β1 κ1 }
v1 = 2 2
. (10)
4(2δ + ρ1 ) 2 − q 2 m1 κ1
The optimal IMC mix for the other brand is obtained by replacing the subscript 1
by the subscript 2. Finally, we analyze the optimal strategies (u 1* , v1* ) and
(u *2 , v *2 ) as well as their corresponding impact on long-term profit via Ji, i = 1,2 to
discover the generalization:

PROPOSITION 5. Propositions 1 through 4 generalize to duopoly markets.

PROOF. See Appendix E.

The closed-form expressions in (9) and (10) can be differentiated with

respect to any specific parameter to gain further insights into asymmetry in a
firm’s structure and the effects of the rival’s parameters. Recent studies furnish
such results in the context of Lanchester model dynamics. Specifically, Prasad
and Sethi (2003) study a two-player dynamic competition model, solve it
explicitly, and evaluate the comparative static signs with respect to all the model
parameters for both symmetric and asymmetric firms (see their Tables 1 and 2),
but ignore multiple media and synergy between them. Complementing these
results, Naik, Raman and Winer (2005) analyze the N-brand competitive
advertising model with multiple controls, estimate the model parameters using
market data, evaluate the comparative static signs with respect to interaction
between advertising and promotion, and discuss the effects of rival’s parameters
(see their Table 4).

Produced by The Berkeley Electronic Press, 2008

10 Review of Marketing Science Vol. 2 [2004], Article 8

5.2 Influence of uncertainty

Propositions 1 through 4 and their generalization to duopoly markets via

proposition 5 seem to imply that uncertainty plays no role. This interpretation is
not accurate. Specifically, uncertainty does affect sales evolution directly via
equations (3) and (8), making the level and growth of sales less predictable in the
foreseeable future. Furthermore, uncertainty directly affects the variability in
long-term profit (see equation C2), increasing downside risks of losses and
bankruptcies. Thus, uncertainty has serious consequences on both the dimensions
of sales and profit. Hence, the role of uncertainty should not be ignored in the
analyses. When we explicitly include it in the analyses, we learn that managers
should not alter their decisions by increasing or decreasing budget in response to
the effects of uncertainty on sales and profit.
This finding is important because it sheds light on existing practice, which
is based on conflicting views. Specifically, advertising and media agencies
advocate that managers should increase spending in response to demand shocks
such as recessions (see ABP 1993). An empirical analysis of national media
spending (not presented here), however, indicates that managers reduce the media
budget during recessions on average. In contrast, our normative analysis reveals,
via Propositions 1 through 5, that managers should “stick to the course of action
in uncertain times.” Indeed, this counter-intuitive insight arises not only in the
proposed IMC models, but in many other models whose value function is linear in
the state variable (e.g., Prasad and Sethi 2003, Sorger 1989, Sethi 1983).5
In sum, the force of these propositions lies in recognizing the distinction
that “no action” on budget changes does not imply managerial “inaction”, the
former requiring knowledge of optimal decision-making under uncertainty, the
discipline not to tinker with marketing budgets in the short term, and the
commitment to building brands over the long term.

5
In a recent insightful analysis, due to Prasad and Sethi (2003), we learn another
counter-intuitive result that brands with smaller market share should spend more
aggressively on advertising than larger brands. This finding is contrary to the
conventional practice of some firms to maintain share-of-voice proportional to
market share (which implies smaller brands should spend less aggressively).
Thus, managers should re-consider the validity of their decision rules in ever-
changing dynamic markets.

http://www.bepress.com/romsjournal/vol2/iss1/art8
Raman and Naik: Long Term Profit Impact of IMC 11

5.3 Commitment versus Feedback

We note that the optimal advertising strategy for our dynamic model does
not depend on the level of sales. This fact raises two questions: (a) is the model
structure realistic, and (b) do managers buy media without weekly sales feedback?
To address issue (a), we observe that our model structure is based on
Nerlove-Arrow (NA) dynamics, which has been empirically validated in hundreds
of research studies in marketing since Palda’s (1964) dissertation. More recently,
Bucklin and Gupta (1999, p. 262) surveyed real managers and found that the
commonly used advertising model in industry is based on the Koyck model,
which is a discrete-time version of NA dynamics. In addition, using Dockers
brand data, Naik and Raman (2003) provided empirical validation for the
advertising model based on NA dynamics. Thus, the model structure we have
used in this paper enjoys both managerial usage and empirical support.
As for question (b), consider the institutional facts of buying network
time, which is sold in two markets: the upfront market and the scatter market
(Tellis 1998, p. 351, Belch and Belch 2004, p. 358). In the upfront market, an
advertiser buys network time before the season begins and commits up to a year
in advance without knowing week-by-week sales for the upcoming year. On the
other hand, the scatter market is a “spot” market where network time can be
purchased based on past sales performance. Consequently, constant or open-loop
advertising strategies are implemented in upfront markets, whereas feedback or
closed-loop strategies are implementable in the scatter markets only. It is
important to recognize that, of the total network time sold, an overwhelming
amount is bought in the upfront market relative to the scatter market. For
example, approximately 80% of this year’s network time is already sold in the
upfront market (see Brough 2004) even before the year’s Fall season began at the
time of writing this manuscript and long before the Winter, Spring and Summer
seasons that are due six to twelve months from now. Thus, substantial media
buying relies on advance commitments, without the use of feedback strategies,
and this feature of our model’s prediction corroborates with the prevailing
practice of advertisers as well as the workings of media institutions. In sum, our
model based on NA dynamics has empirical and practical validity, and the
resulting advertising strategy applies to the major proportion of an advertiser’s
spending decisions.

6. CONCLUDING REMARKS

We incorporated the roles of uncertainty and competition in an IMC

model (Naik and Raman 2003), and solved the resulting stochastic control
problem to determine the optimal IMC program. Our results elucidate the profit

Produced by The Berkeley Electronic Press, 2008

12 Review of Marketing Science Vol. 2 [2004], Article 8

implications of IMC programs, explain the catalytic effects of synergy in IMC

contexts, and generalize the findings of Naik and Raman (2003) to uncertain and
competitive markets. Much remains to be done in the area of multimedia
communications; for example, limited knowledge exists on planning the IMC
budget and allocation during recessionary times. We hope that our research
provides the impetus and methodology to investigate such managerially relevant
topics.

http://www.bepress.com/romsjournal/vol2/iss1/art8
Raman and Naik: Long Term Profit Impact of IMC 13

REFERENCES

American Business Press (1993). Making a Recession Work for You. New York:
American Business Press.

Arnold, L. (1974), Stochastic Differential Equations: Theory and Applications,

John Wiley & Sons: New York.

Belch, George E., and Michael A. Belch (2004), Advertising and Promotion: An
Integrated Marketing Communications Perspective, 6th Edition, Boston, MA:
Irwin-McGraw Hill Inc.

Brough, Charley (2004), “Networks Finish Up the Upfront,” The Red Line, Issue
61, June 17, 2004, http://www.rightplacemedia.com.

Bucklin, Randolph E., and Sunil Gupta (1999), “Commercial Use of UPC
Scanner Data: Industry and Academic Perspectives,” Marketing Science, 18
(3), 247-273.

Chow, Gregory C. (1997), Dynamic Economics: Optimization by the Lagrange

Method, New York, NY: Oxford University Press.

Edell, Julie E., and Kevin L. Keller (1999), “Analyzing Media Interactions: The
Effects of Coordinated TV-Print Advertising Campaigns,” Working Paper,
Report No. 99-120, Cambridge, MA: Marketing Science Institute.

Erickson, Gary M. (1991), Dynamic Models of Advertising Competition: Open-

and Closed-Loop Extensions, Boston, MA: Kluwer Academic Publishers.

Farris, Paul W., Ervin R. Shames, and David J. Reibstein (1998), Advertising
Budgeting: A Report from the Field, New York: American Association of
Advertising Agencies.

Feichtinger, Gustav, Richard F. Hartl, and Suresh P. Sethi (1994), “Dynamic

Optimal Control Models in Advertising: Recent Developments,” Management
Science, 40 (2), 195-226.

Fruchter, Gila E. and Shlomo Kalish (1998), “Dynamic Promotional Budgeting

and Media Allocation,” European Journal of Operational Research, 111, p. 15-
27.

Produced by The Berkeley Electronic Press, 2008

14 Review of Marketing Science Vol. 2 [2004], Article 8

Gatignon, Hubert (1993), “Marketing Mix Models,” In Handbook of OR and MS,

Eds. J. Eliashberg and G. L. Lilien, Vol. 5, p. 697-732.

Gihman, I.I. and A.V. Skorohod (1972), Stochastic Differential Equations,

Springer-Verlag: New York

Hanssens, Dominique M., Leonard J. Parsons, and Randall L. Schultz (1998),

Market Response Models: Econometric and Time Series Analysis, Boston, MA:
Kluwer Academic Publishers.

Harrison, Michael J. (1985), Brownian Motion and Stochastic Flow Systems, John
Wiley & Sons: New York.

Keller, Kevin Lane (1998), Strategic Brand Management, Upper Saddle River,
New Jersey: Prentice Hall.

Mantrala, Murali K. (2002), “Allocating Marketing Resources,” Handbook of

Marketing, Eds. Barton A. Weitz and Robin Wensley, Sage Publications,
forthcoming.

Mantrala, Murali K., Kalyan Raman, and Ramarao Desiraju (1997), “Sales Quota
Plans: Mechanisms for Adaptive Learning,” Marketing Letters, 8 (4), 393-405.

Naik, Prasad A. and Kalyan Raman (2003), “Understanding the Impact of

Synergy in Multimedia Communications,” Journal of Marketing Research, 40
(4), 375-388.

Naik, Prasad A., Kalyan Raman, and Russell S. Winer (2005), “Planning
Marketing-Mix Strategies in the Presence of Interactions,” Marketing Science,
forthcoming Special Issue on Competitive Responsiveness.

Palda, Kristian S. (1964), The Measurement of Cumulative Advertising Effects,

Englewood Clifs, N. J.: Prentice Hall.

Parsons, Leonard J., and Piet Vanden Abeele (1981), “Analysis of Sales Call
Effectiveness,” Journal of Marketing Research, 18 (1), 107-113.

Prasad, Ashutosh and Suresh P. Sethi (2003), “Competitive Advertising under

Uncertainty: A Stochastic Differential Game Approach,” Journal of
Optimization Theory and Applications, forthcoming.

http://www.bepress.com/romsjournal/vol2/iss1/art8
Raman and Naik: Long Term Profit Impact of IMC 15

Raman, Kalyan (1990), “Stochastically Optimal Advertising Policies under

Dynamic Conditions: The Ratio Rule,” Optimal Control Applications and
Methods, 11, 283-288.

Raman, Kalyan and Rabikar Chatterjee (1995), “Stochastic Optimal Control of

Monopoly Pricing Models with Dynamic Demand and Production Cost,”
Management Science, 41 (1), 144-162.

Rao, Ram C. (1986) “Estimating Continuous Time Advertising-Sales Models,”

Marketing Science, 5 (Spring) 125-142.

Schultz, Don E. and Joseph J. Pilotta (2004), “A New Approach to Understanding

How Media Advertising Works,” ESOMAR/ARF, Geneva.

Sethi, Suresh P. (1977), “Dynamic Optimal Control Models in Advertising: A

Survey,” SIAM Review, 19 (4), 685-725.

Sethi, Suresh P. (1983), “Deterministic and Stochastic Optimization of a Dynamic

Advertising Model,” Optimal Control Applications & Methods, 4 (2), April-
June, 179-184.

Sethi, Suresh P. and Gerald L. Thompson (2000), Optimal Control Theory:

Applications to Management Science and Economics, Second Edition, Kluwer
Academic Publishers, Boston.

Sorger, G. (1989), “Competitive Dynamic Advertising: A Modification of the

Case Game,” Journal of Economic Dynamics & Control, 13 (1), 55-91

Tellis, Gerard (1998), Advertising and Sales Promotion Strategy, Readings, MA:
Addison-Wesley.

Produced by The Berkeley Electronic Press, 2008

16 Review of Marketing Science Vol. 2 [2004], Article 8

APPENDIX A: PROOF OF PROPOSITION 1

Here we solve the stochastic control problem of maximizing the objective

functional in equation (4) subject to the equation (3), which explicitly recognizes
the presence of uncertainty (since variance σ 2 ≠ 0 ).

Let J(s) denote the maximum profit attained when we implement the
optimal controls u(s) and v(s), starting from the initial sales level s. Then the
Hamilton-Jacobi-Bellman (HJB) equation specifies the differential equation for J:

− ρJ + Max [Π (s, u , v) + J s f (s, u , v) + 0.5σ 2 J ss ] = 0 , (A1)

u ,v

where Js = ∂J/∂s, Jss = ∂2J/∂s2, Π(s, u, v) = ms − u2 – v2, and f(s, u, v) is the drift
term in equation (3). We differentiate terms within the square brackets in (A1)
with respect to the controls u and v, and solve the resulting first-order conditions
by using the Cramer’s rule to obtain:

J s (2β1 + β 2 κJ s ) J s (2β 2 + β1 κJ s )
u= 2 2
, v= 2
. (A2)
4 − κ Js 4 − κ2Js

We note that the controls (u, v) in equation (A2) are closed-loop because they
depend on the level of sales s via the marginal profit rate Js = ∂J/∂s, where J(s) is
an unknown function of s (to be determined).

Next, by substituting u and v in the equation (A1), we get the second-order

ordinary differential equation (ODE):

2
− 8ms − 4 σ 2 J ss + σ 2 κ 2 J s J ss + 8sJ s (1 − λ) + 8ρJ − 4σ 2 J ss
2 2 2 3 2 2
(A3)
+ J s J ss κ 2 σ 2 − 2J s β1 − 2J s κβ1β 2 − 2J s β 2 = 0

i =n
To solve this ODE analytically, we consider a polynomial solution J (s) = ∑ J i s i ,
i =0
and apply the method of undetermined coefficients to find the set of coefficients
{Ji} and the order of polynomial, n. Specifically, we discover that n = 1 and so
the resulting solution is

J(s) = J0 + J1 s, (A4)
where

http://www.bepress.com/romsjournal/vol2/iss1/art8
Raman and Naik: Long Term Profit Impact of IMC 17

2 2
m(m(1 − λ + ρ)β1 + m 2 κβ1β 2 + m(1 − λ + ρ)β 2 )
J0 = , and (A5)
ρ(1 − λ + ρ) (4(1 − λ + ρ) 2 − m 2 κ 2 )
m
J1 = . (A6)
1− λ + ρ

Since Js = ∂J/∂s = J1, we substitute (A6) in (A2) to obtain the optimal IMC
strategies:

m(β 2 κm + 2β1 (1 − λ + ρ)) m(β1κm + 2β 2 (1 − λ + ρ))

u * ( θ) = 2 2 2
, v * ( θ) = .
4(1 − λ + ρ) − κ m 4(1 − λ + ρ) 2 − κ 2 m 2
(A7)

Finally, we differentiate (A7) with respect to κ to get,

∂u * ∂v *
> 0 and > 0, (A8)
∂κ ∂κ

∂ u* ⎧− if β1 > β 2
( *)=⎨ . (A9)
∂κ v ⎩+ if β1 < β 2

This completes the proof of proposition 1.

APPENDIX B: PROOF OF PROPOSITION 2

Here we prove proposition 2 by deriving the expression for expected long-

term profitability (ELP). To this end, we first substitute s = S(t) in the equation
(A4) of Appendix A, and evaluate the expectation,

E[J(S(t))] = J0 + J1 µ(t), (B1)

where we denote µ(t) = E[S(t)]. To find Ε[S(t)], we obtain via equation (3),

E[dS] = E[(β1 u + β 2 v + κu v − (1 − λ) S) dt + σdW ]

= (β1 u + β 2 v + κu v − (1 − λ) E[S])dt + σ E[dW ] (B2)
= (β1 u + β 2 v + κu v − (1 − λ) µ)dt ,

Produced by The Berkeley Electronic Press, 2008

18 Review of Marketing Science Vol. 2 [2004], Article 8

where the last equality follows from the property E[dW] = 0 (Arnold 1974, p. 41).
In addition, it follows from Fubini’s theorem (Harrison 1985, p. 131) that

E[dS] = d[E[S]] = dµ . (B3)

Therefore, from (B2) and (B3) together, we get

dµ
= β1 u + β 2 v + κuv − (1 − λ )µ . (B4)
dt

We next implement the optimal controls u* and v* to obtain

dµ
= C * − (1 − λ )µ , (B5)
dt

where C * = β1 u * + β 2 v * + κu * v * , and thus determine the optimal steady-state

C*
Lim µ ( t ) = . (B6)
t →∞ 1− λ

We then evaluate the expected long-term profitability

ELP = Lim E[J (S( t ))]

t →∞

= Lim[J 0 + J 1µ( t )] (B7)

t →∞

C*
= J 0 + J1 ,
1− λ

where the second equality follows from (B1), the last equality follows from (B6),
and J0 and J1 are given by the equations (A5) and (A6), respectively.
Finally, by differentiating (B7) with respect to synergy, we find that
∂ELP ∂J 0 ∂C *
> 0 because both > 0 (using A5) and > 0 (using A8). This
∂κ ∂κ ∂κ
completes the proof of proposition 2.

http://www.bepress.com/romsjournal/vol2/iss1/art8
Raman and Naik: Long Term Profit Impact of IMC 19

APPENDIX C: PROOF OF PROPOSITION 3

We prove proposition 3 by evaluating the variance of long-term

profitability:

VLP = Lim Var[J (S( t ))]

t →∞

= Lim Var[J 0 + J 1S( t )] (C1)

t →∞

= Lim J 12 Var[S( t )].

t →∞

To find Var[S(t)], we require the second moment of the sales process S(t).
Hence, we apply Ito’s Lemma to the stochastic process y(t) = S(t)2, derive and
solve the differential equation satisfied by µ2(t) = E[y(t)] = E[S(t)2], and
determine Var[S(t)] through the relation Var[S(t)] = E[S(t)2] – (E[S(t)])2 = µ2(t) −
µ(t)2. Thus, we obtain

VLP = J 12 {Lim Var[S( t )]}

t →∞

m2 σ2 (C2)
=
2 (1 − λ ) (1 − λ + ρ) 2
∂VLP
It is clear from (C2) that = 0 , which completes the proof.
∂κ

APPENDIX D: PROOF OF PROPOSITION 4

Suppose that the activity v is completely ineffective so that β2 = 0. Then,

we substitute β2 = 0 in equation (A7) to find that the
β1 κm 2
optimal v * = , which is non-zero when κ ≠ 0 . This
4(1 + ρ − λ) 2 − κ 2 m 2
completes the proof.

APPENDIX E: PROOF OF PROPOSITION 5 (DUOPOLY)

We set up the Hamilton-Jacobi-Bellman equation for each of the two

brands and follow the same logic that led to equation (A1). As before, we
i=n
consider a polynomial solution J (s) = ∑ J i s i , find that n = 1, and obtain the linear
i =0

Produced by The Berkeley Electronic Press, 2008

20 Review of Marketing Science Vol. 2 [2004], Article 8

value function for each duopolist by solving for J0 and J1. Consequently, we
determine the optimal IMC strategy of the first brand, which is given by equations
(9) and (10). Using equations (9) and (10), we derive the comparative static
results:

2
( ( 2
∂u 1* q 2 m1 4β1qm1 κ1 (2δ + ρ1 ) + γ 1 q 2 m1 κ1 + 4(2δ + ρ1 ) 2
=
2
)) > 0 (E3)
∂κ1 (
2 2
q 2 m κ − 4(2δ + ρ ) 2
1 1
2
1 )
2
( ( 2
∂v1* q 2 m1 4γ 1qm1 κ1 (2δ + ρ1 ) + β1 q 2 m1 κ1 + 4(2δ + ρ1 ) 2
=
2
)) > 0 (E4)
∂κ1 (
2 2
q 2 m κ − 4(2δ + ρ ) 2
1 1
2
1 )
∂ ⎛ u 1* ⎞
⎜⎜ * ⎟⎟ = − 2 1
2
( 2
)
β − γ 1 (qm1 (2δ + ρ1 ) )
∂κ1 ⎝ v1 ⎠ (qm1β1 κ1 + 2γ 1 (2δ + ρ1 ) )2
(E5)

Thus, the expressions (E3)-(E5) establish that proposition 1 generalizes to the

duopoly market.

Denoting the value function of duopolist k by Jk, k = 1, 2, we obtain the

derivative of the first duopolist’s value function J1(θ) with respect to its synergy
parameter κ1,

3 2 2
( 2 2
∂J 1 m1 q 3 {( 2β1 κ1 m1q (2δ + ρ1 ) + 2 γ 1 κ1 m1q (2δ + ρ1 ) + β1 γ 1 q 2 m1 κ1 + 4(2δ + ρ1 ) 2
=
)
2 2
∂κ1 ρ1 (2δ + ρ1 ){− κ1 m1 q 2 − 4(2δ + ρ1 ) 2 }2
. (E6)

Because (E6) is always positive, it establishes that proposition 2 holds in the

duopoly case, following exactly the same reasoning as in Appendix B.

As in Appendix C, we apply Ito’s Lemma to the stochastic process Z(t) =

2
X(t) , derive and solve the differential equation satisfied by µ2(t) = E[Z(t)] =
E[X(t)2], and determine Var[Z(t)] through the relation Var[Z(t)] = E[Z(t)2] –
(E[Z(t)])2 = µ2(t) − µ(t)2, thus obtaining:

2
m1 q 2 σ 2
VLP 1 = . (F7)
4δ (2δ + ρ1 ) 2

http://www.bepress.com/romsjournal/vol2/iss1/art8
Raman and Naik: Long Term Profit Impact of IMC 21

∂VLP1
Because = 0 , we prove that proposition 3 holds for the duopoly case.
∂κ1

To prove that proposition 4 holds for the duopoly case, we substitute γ1 =

* (qm1 ) 2 β1 κ1
0 in equation (10) to find that the optimal v1 = 2 2
, which is
4(2δ + ρ1 ) 2 − q 2 m1 κ1
non-zero when κ1 ≠ 0 .

Finally, similar results can be obtained for the other brand by replacing the
subscript 1 with 2 in the above expressions, which completes the proofs for
proposition 5.

Produced by The Berkeley Electronic Press, 2008

View publication stats