Intern. J.

of Research in Marketing 26 (2009) 1320

Contents lists available at ScienceDirect

Intern. J. of Research in Marketing

Creating lift versus building the base: Current trends in marketing dynamics
Peter S.H. Leeang a,, Tammo H.A. Bijmolt a, Jenny van Doorn a, Dominique M. Hanssens b, Harald J. van Heerde c, Peter C. Verhoef a, Jaap E. Wieringa a
a b c

Faculty of Economics and Business, University of Groningen, P.O. Box 800, 9700 AV Groningen, The Netherlands UCLA Anderson School of Management, 110 Westwood Plaza, Los Angeles , CA 90095-1481, USA Waikato Management School, University of Waikato, Hamilton 3240, New Zealand

a r t i c l e

i n f o

a b s t r a c t
Markets are dynamic by nature, and marketing efforts can be directed to stimulate, reduce, or to utilize these dynamics. The eld of marketing dynamics aims at modeling the effects of marketing actions and policies on short-term performance (lift) and on long-term performance (base). One of the core questions within this eld is: How do marketing efforts affect outcome metrics such as revenues, prots, or shareholder value over time? Developments in statistical modeling and new data sources allow marketing scientists to provide increasingly comprehensive answers to this question. We present an outlook on developments in modeling marketing dynamics and specify research directions. 2009 Elsevier B.V. All rights reserved.

Keywords: Marketing dynamics Time series models Dynamic linear models

This manuscript is a conference feature paper on the 2007 Marketing Dynamics Conference which the authors organized at the University of Groningen, The Netherlands. 1. Introduction The dynamic nature of markets dictates that marketing measures are often targeted at stimulating, reducing, or utilizing market responsiveness. Firms launch new products and introduce better packaging (stimulating response), retaliate against competitive moves (reducing response), monitor trends in consumer preferences and segment membership (utilizing response), and so on. The effects of marketing efforts do not necessarily end when, for example, an advertising campaign is over. The effect, or part of it, will remain noticeable for some time. In recent years, the determination of the long-term effects of marketing efforts has received much attention from practitioners and academics. Senior executives are increasingly interested in the longterm impact on sales, prots, but also on relatively new metrics such as shareholder value. They want to create sustainable competitive advantages for their brands and they want to see permanent effects of their investments in marketing efforts. For example, Gerard Kleisterlee, CEO of Royal Philips Electronics, stated that in the long-run our values and how we honor them will determine the outcome of what we strive for.1 Oswald Grbel, CEO of the Credit Suisse Group, specied his aims in a somewhat different way: Our priorities are
Corresponding author. Department of Marketing, Faculty of Economics and Business, University of Groningen, P.O. Box 800, 9700 AV, Groningen, The Netherlands. Tel.: +31 503637065. E-mail address: [email protected] (P.S.H. Leeang). 1 Gerard Kleisterlee, December 04, 2007, Innovation as driver of sustainable growth, Speech at China Central Party School. 0167-8116/$ see front matter 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.ijresmar.2008.06.006

quite clear: we want to generate long-term added value for our shareholders by offering outstanding service to our clients and by securing a leading position in the industry.2 For non-traded companies, rm value instead of shareholder value is an important metric (Gupta, Lehmann, & Stuart, 2004). Such perspectives imply that marketing resources should be allocated to maximize the long-term impact on the relevant metrics such as shareholder value. This task requires, in turn, that a valid and reliable answer is found to the paramount question: How do marketing efforts affect outcome metrics such as revenues, prots and shareholder value over time? To address this question, the discipline of marketing dynamics studies the short- and long-term effects of marketing actions and policies on relevant metrics. In the past ten years, we have witnessed important improvements in modeling marketing dynamics. These developments have led to the establishment of the annual Marketing Dynamics Conference. The rst conference was held at the Tuck School of Business at Dartmouth, USA in 2004 (Pauwels et al., 2004a), while the fourth conference was hosted by the University of Groningen, the Netherlands, in August 2007. In this feature article, we discuss the relevance and challenges of modeling marketing dynamics for marketing decision-making. A number of these challenges were summarized in the keynote speech by Dominique Hanssens at the Groningen conference, and they partly overlap with those identied by Pauwels et al. (2004a). Our review of trends is largely based on the 40 presentations at
2 Oswald J. Grubel, April 28, 2006, Speech made at Annual General Meeting of Credit Suisse Group, Zurich.

14

P.S.H. Leeang et al. / Intern. J. of Research in Marketing 26 (2009) 1320 Table 1 Creating lift or building the base (L) = 0 (L) = 0 (L) N 0 Ineffective marketing Marketing generates sales (L) N 0 Marketing builds the brand Marketing generates sales and builds the brand

the Fourth Marketing Dynamics Conference. We specify criteria that dynamic models should satisfy, indicate important developments in relevant research methodologies, and formulate research directions. 2. Challenges and methodologies To address the core question of how marketing efforts affect outcome metrics over time, we need to build suitable dynamic marketing models. Ideally, these models and methodologies: 1. use appropriate metrics, 2. disentangle temporary (short-term) from persistent (long-term) effects, 3. account for time-varying parameters, and 4. allow for cross-sectional heterogeneity. We discuss these requirements in the following subsections. 2.1. Marketing metrics The core question involves differentiating between marketing efforts that lift sales temporarily (ow) and efforts that build marketing stock, i.e., lead to a permanent shift in the base level. Many sales response models relate current sales to current and past marketing expenditures (see e.g., Leeang, Wittink, Wedel & Naert, 2000, p.8599). The demand or revenue metric is a ow metric. Ideally, marketing expenditures will also create benecial changes in stock metrics. Examples of stock metrics are cumulative sales, brand equity, customer equity, et cetera. Pauwels and Hanssens (2007) and Hanssens and Dekimpe (2008) extend the ow response models to capture the effects of marketing investments on stock metrics and specify the following relations: Sit = cit + ki LMkit + eit ;
k

Source: adapted from Hanssens & Dekimpe (2008).

sales cit evolves over time, following the repeat-purchase diffusion process as specied in the state equation:   V V cit = i ci;t1 + Z it Z it ci;t1 + wit : 4

Zit is a vector of standardized marketing strategy (marketing policy) variables. Standardization offers the opportunity that one can pool different brands across categories and controls for unobserved xed effects. The parameter i captures the brand-specic repeat- purchase rate and and capture growth and market potential due to marketing effort, respectively; wit is a random disturbance term. The marketing actions that build the brand are called marketing policies. Examples are investments in corporate and brand reputation, strategic entries in new markets (Pauwels & Hanssens, 2007), the introduction of new distribution channels (Deleersnyder, Gielens, Geyskens, & Dekimpe, 2002), new products, and quality improvements (Tellis & Johnson, 2007). Hanssens and Dekimpe (2008) use four criteria as a guide to choose appropriate metrics. Metrics should: have nancial relevance, be actionable: i.e., it must be possible, at reasonable cost, to collect data on the performance metric, and to relate it analytically to marketing investments, exhibit stable behavior, and offer reliable long-term guidance. Highly volatile metrics are less desirable because they are difcult to interpret and manage. The leading indicator aspect of a metric is reected in the criterion that the metric should have reliable longterm guidance, i.e., movements in the metric should be indicative of improving or deteriorating health for the brand or rm. We distinguish four core metrics that can be used to specify the dependent variable Sit in Eq. (1). First, sales is a commonly used metric, for instance, to understand how marketing drives prescription drug sales (Fischer & Albers, 2007) or where the demand for a new product comes from (Albuquerque & Bronnenberg, 2007; Van Heerde, Srinivasan, & Dekimpe, 2008). Second, a useful long-term metric is customer lifetime value and its rm-level aggregate, customer equity. Gupta, Lehmann and Stuart (2004) argue that customer equity can be used to value rms, and thus, to calculate the effect of marketing actions on shareholder value. Rust, Lemon, and Zeithaml (2004) and Donkers, Verhoef, and De Jong (2007) show how customer equity is affected by alternative marketing strategies. A third metric is brand equity, the incremental cash ows that can be expected from carrying branded products instead of unbranded products (Simon & Sullivan, 1993). Pauwels, Nijs, and Srinivasan (2007) look at the effects of product-line decisions on brand equity, whereas Ataman, Van Heerde, and Mela (2007) consider the impact of all relevant marketing instruments. A fourth metric is stock market value, which is frequently analyzed by VAR models. For example, Pauwels, Silva-Risso, Srinivasan and Hanssens (2004b) study the effects of new products and sales promotions, and Joshi and Hanssens (2008) assess the inuence of advertising and R&D on the stock return of rms in the PC manufacturing and sporting goods industries. Other methodologies include event-studies for a single marketing initiative and regressionbased stock return models. Event studies have looked, for example, at

1 2

cit = i ci;t1 + ki LMkit + it ;

where Sit is the outcome metric, such as the sales of brand3 or rm i, cit the baseline of unit i at time t, ki (L) represents the effectiveness of marketing efforts on baseline sales with lag L, Mkit the marketing efforts with marketing instrument k, and it and it disturbance terms. Most attention in marketing has been given to the determination of optimal marketing expenditures (M), how to improve marketing effectiveness ((L)) and how this leads to a larger ow (Sit). Relation (2) shows the development of its baseline over time. Changes in the baseline sales are interpreted as building the base. Given that baseline sales can be seen as a measure of brand equity, ki(L) N 0 indicates that marketing investments are building the brand (equity). Hence Eqs. (1) and (2) answer the question whether or not marketing efforts create demand (ki) and/or build the baseline sales (ki) of the brand (Table 1). Ataman, Mela and Van Heerde (2008) use a similar specication to explain how marketing mix activity generates growth and builds market potential for new brands. Their so-called observation equation separates short-term uctuations from long-term sales: it = cit + X V i + it ; S it 3

where Sit is the (standardized) sales of brand i at time t, Xit includes variables that may generate short-term uctuations in sales, and it is a disturbance term. Ataman et al. (2008) standardize all variables within brands and indicate this with a superscripted bar. The baseline

Instead of sales one can also work with revenues or stock prices.

P.S.H. Leeang et al. / Intern. J. of Research in Marketing 26 (2009) 1320

15

the introduction of a new channel (Geyskens, Gielens, & Dekimpe, 2002), the entrance of a major retailer such as Wal-Mart (Gielens, Van de Gucht, Steenkamp, & Dekimpe, 2008), and the effects of the evaluation of new products that are evaluated in The Wall Street Journal on abnormal returns on stock prices (Tellis & Johnson, 2007). An example of a regression-based stock return model is the brand equity analysis in Rao, Agarwal and Dahlhoff (2004). Stock-market value receives much more attention nowadays than at earlier conferences on marketing dynamics (compare Pauwels et al., 2004a). Recent developments on stock markets, however, cast doubts on the use of this metric because stock prices might be more driven by market turbulences than by rm value. 2.2. Disentangling short-term and long-term effects The VAR-X (Vector AutoRegressive with independent variables) model disentangles short-term from long-term movements (e.g., Dekimpe & Hanssens,1999, 2000; Horvth, Leeang, Wieringa, & Wittink, 2005). When the model variables are connected through a long-term equilibrium relationship, a VEC (Vector Error Correction) model is needed, which has recently gained popularity in marketing (Fok, Horvth, Paap, & Franses, 2006; Van Heerde, Helsen & Dekimpe, 2007; Van Heerde, Srinivasan & Dekimpe, 2008). Montoya, Netzer and Jedidi (2007) use Markov models to disentangle short- from long-term effects in the context of direct-to-physician marketing in the pharmaceutical industry. Short- and long-term breaks in marketing metrics are often due to discrete (marketing) events such as the entry of a new product (Albuquerque & Bronnenberg, 2007; Van Heerde et al., 2008), the use of a new channel (Verhoef, Neslin & Vroomen, 2007), the introduction of a loyalty program (Leenheer et al., 2007) or its termination (Melnyk & Bijmolt, 2007). Recently, tests have been developed to nd more than one break, where the breaks are determined endogenously (Kornelis, Dekimpe & Leeang, 2008). To disentangle cyclical or seasonal effects from short-term and long-term trends one can use a lter such as the Hodrick-Prescott lter (Hodrick & Prescott, 1997; see, for example, Deleersnyder, Steenkamp, Dekimpe & Leeang, forthcoming; Leeang et al., 2008). 2.3. Time-varying parameters Not only can the base (intercept) parameter vary over time (Eq. (2)), other response parameters (ik in Eq. (1)) can evolve as well, due to marketing activities. Time-varying response parameters imply that the lift-effects of marketing instruments Mkit in Eq. (1) vary over time. There are different methods to account for time-varying parameters. One may perform simple analyses such as moving window regression (Mahajan, Bretschneider & Bradford, 1980) or piecewise regression (Parsons, 1975). In more recent research, the structural parameters are modeled as a function of relevant independent variables through process functions (Mela, Gupta & Lehmann,1997; Foekens, Leeang & Wittink,1999). These models are also known as Time Varying Parameter Models (TVPM). In this context, Pauwels and Hanssens (2007) and Yoo and Hanssens (2008) specify performance regimes or windows of performance decline, stability and growth in sales and customer equity, respectively. The most comprehensive methods to account for varying parameters over time are the Kalman-ltering and Dynamic Linear Models (DLM). Examples of marketing applications that use the Kalman-lter approach are Xie, Song, Sirbu, and Wang (1997), Naik, Mantrala, and Sawyer (1998), Cain (2005), Van Everdingen, Aghina, and Fok (2005), Kolsarici and Vakratsas (2007), Osinga et al. (2008), Sriram and Kalwani (2007) and Sriram, Chintagunta, and Neelamegham (2006). DLMs are closely related to TVPMs and VAR models because they all have their roots in state-space modeling. State space models represent a large class of models in which the dynamic relationships between the variables of interest are expressed in two equations. The rst equation, the measurement (or observation) equation, species

how the vector of endogenous variables depends on the state of the system; see Eqs. (1) and (3). In the second equation, the transition (or state) equation, the evolution of the state vector is specied; see Eqs. (2) and (4) for examples. The generality of this type of model formulation is illustrated by the fact that it is possible to formulate state-space analogs of TVPMs and VARs (Ataman, 2007). The estimation of state-space models traditionally relies on frequentist statistical techniques, such as maximum likelihood. DLMs are Bayesian extensions of state space models. Like any other state space model, DLMs are derived from the Kalman lter. Specically, Kalman lters are equivalent to the updating equations in a DLM (Ataman, 2007; Harrison & Stevens, 1976). DLMs have the following desirable properties. First, the specication of DLMs allows for a single-stage analysis of long-term phenomena. For example, Eqs. (1) and (2) are estimated simultaneously instead of in two stages, leading to greater statistical efciency (Van Heerde, Mela & Manchanda, 2004). Second, a DLM copes naturally with missing data arising from, e.g., product introductions or deletions. Third, the Bayesian nature also allows for inclusion of subjective data, which also means that forecasts can be produced with little or no past data. Finally, DLMs accommodate longitudinal as well as cross-sectional heterogeneity. (i) Capitalizing on these advantages, recent DLM applications have provided fresh insights on: how a radical innovation affects market structure (Van Heerde et al., 2004); (ii) how the preferences for product attributes evolve over time (Neelamegham & Chintagunta 2004); (iii) how a product-harm crisis hurts marketing effectiveness (Van Heerde et al., 2007); (iv) how to use the marketing mix to manage brand equity (Ataman, Van Heerde, & Mela, 2007); (v) how the decomposition of the demand for a radical innovation varies over time (Van Heerde, Srinivasan, & Dekimpe, 2008); (vi) what strategies build new brands (Ataman, Mela, & Van Heerde, 2008). The latter study concludes that distribution breadth is the single most important marketing mix instrument in both generating growth (relative effect of 32%) and building market potential (relative effect of 54%) for a new brand; (vii) how the effects of advertising and word of mouth for new products (such as movies) evolve over sequential (such as theatre-then-video) distribution stages (Bruce & Foutz 2007). These advantages come at the cost of high computational requirements. Estimating a DLM may take several hours or days, depending on the dimensionality of the problem. Furthermore, at present, few software packages include a DLM module, so that coding in a matrix language (e.g., Gauss, Matlab, Ox, and R) is required. While there has already been much attention for time-varying parameters and the disentangling of short-term and long-term effects at earlier conferences on marketing dynamics (see Pauwels et al., 2004a), based on the number of recent (published) papers we conclude that the interest in these topics has further increased. 2.4. Cross-sectional heterogeneity Eqs. (1) and (2) allow for heterogeneous response parameters ki and ki, i.e., they are specic to each unit (e.g., brand, store, or rm) i. Andrews, Currim, Leeang and Lim (2008) investigate whether storelevel heterogeneity in marketing mix effects improve the model accuracy (estimates, t, prediction) of the widely applied SCANPRO model of store sales. Models with continuous and discrete representations of heterogeneity are empirically compared to the original, homogenous model. Contrary to expectations, accommodating storelevel heterogeneity does not improve model accuracy. Horvth and Wieringa (2008) compare several VAR modeling approaches that accommodate different levels of heterogeneity. They conclude that

16

P.S.H. Leeang et al. / Intern. J. of Research in Marketing 26 (2009) 1320

random coefcient modeling is an overall appropriate technique when the VAR model is used for forecasting only. 3. Developments, trends and research needs We now discuss trends, developments and research needs in modeling marketing dynamics. We use the papers of the conference as illustrations of these trends, and we present a summary table of these papers and related publications. In the nal subsection, we specify our future outlook. 3.1. Marketing, revenues and rm value There is increasing evidence that longer-term rm value is affected by marketing expenditures (Yoo & Pauwels, 2007). On the other hand, many corporate executives are concerned about shorterterm performance metrics. How to reconcile these seemingly contradictory behaviors is an interesting research avenue (Srinivasan & Hanssens, 2009). Osinga et al. (2008) investigated this issue in the pharmaceutical market. They developed a methodology that assesses the effect of direct-to-consumer-advertising (DTCA) on three components of shareholder value: stock return, systematic risk and idiosyncratic risk. 3.2. Normative studies In general, most studies in marketing dynamics either focus on describing how marketing works (i.e., the exact effect of sales promotions), or what drives brand performance. However, there is an acute shortage of normative studies developing navigation systems that allow managers to optimize marketing efforts, or at least investigate what-if scenarios. Notable exceptions are Naik and Raman (2003) (the impact of synergy in multimedia communications) and Naik, Raman and Winer (2005). In the latter study, the optimal advertising and promotion budgets are determined. They observe that while some brands over-promote whereas others under-promote, all brands in their study under-advertise. Montoya, Netzer and Jedidi (2007) look at how long-term protability can be managed through marketing-mix allocation. Normative studies are susceptible to the Lucas critique (see also Pauwels et al., 2004a; Van Heerde, Dekimpe & Putsis, 2005). The inclusion of varying parameters and a sharper distinction between short-term and long-term dynamics in structural models are some approaches used to deal with the Lucas critique. 3.3. Global models Within the new-product-diffusion literature there has been ample attention devoted to understanding the drivers of adoption, newproduct take-off and new product-growth across nations (e.g., Gielens & Steenkamp, 2007; Tellis, Stremersch & Yin, 2003; Stremersch & Tellis, 2004; Stremersch & Lemmens, 2009). However, in a globalizing economy we need to extend our knowledge on the short- and longterm effects of marketing efforts beyond western economies, especially with respect to the emerging economic giants China and India (Burgess & Steenkamp, 2006). 3.4. Inclusion of attitudinal (soft) data Many models within the eld of marketing dynamics are based on hard behavioral data. However, in the customer management and service marketing literature, models have been developed that link attitudinal data to both individual customer behavior (i.e., churn, customer share) and rm performance (Gupta & Zeithaml, 2006; Van Doorn & Verhoef, 2008). The inclusion of attitudinal data in dynamic models may be a fruitful new research direction given the increased

availability of longitudinal attitudinal data due to continuous surveyresearch and CRM-systems (see also Dekimpe & Hanssens, 2000). For example, Knox and Van Oest (2007) model how complaints by consumers precede churn, while Venkatesan, Reinartz and Ravishanker (2008) show that detailing is more effective among physicians with positive attitudes towards the rm. Srinivasan, Vanhuele and Pauwels (2008) combine soft customer mindset metrics (awareness, affect and purchase consideration) with hard data (sales and marketing mix) in a joint VAR model. 3.5. Model development Models within marketing dynamics have strongly evolved over time. Econometric regression-based models were most common until the mid-90s. Since the end of the 90s, time-series models (in particular VAR models) have become very inuential, and recently, new Bayesian models and DLMs have entered the eld and are gaining momentum. A particularly interesting development has been Bayesian state space models for non-metric dependent variables (e.g., Lachaab, Ansari, Jedidi, & Trabelsi, 2006). These models allow for the study of short- and long-term marketing effects on choices and other limited-dependent variables at the individual customer level, which is valuable for rms, yet has not been captured by traditional methods. Other model sophistications will emerge in the more distant future. However, we have some cautionary notes. First, the focus on model development should not be at the cost of a focus on the core issue of marketing dynamics: understanding and predicting the persistent impact of marketing efforts. Second, over-sophistication may hamper the diffusion of marketing models and knowledge into practice because the complexity may be higher than the perceived benets (Roberts, Kayande & Stremersch, 2007). Thus, new methods should balance technical and insight contributions. 3.6. New applications Pauwels et al. (2004a) review many examples of applications of dynamic models. The number of applications has increased over the last few years. The rise of the internet and the increasing availability of customer data due to CRM-systems have proven to be very fruitful research areas (e.g., Batislam, Deniziel & Filizetekin, 2007; Reimer, Rutz & Bucklin, 2007; Deleersnyder et al., 2002). The increasing use of new technologies where rms can observe product choice (i.e., MP3 music systems) provide great new data sources. For example, Chung, Rust and Wedel (2007) develop a model to optimize the music assortment for each individual customer. Rapid changes in communications technology are creating communities of customers and prospects rather than a multitude of isolated customers. Consequently, the Connected Customer is MSI's overarching research theme for 20062008 (Marketing Science Institute, 2007). Within this theme, Van der Lans, Van Bruggen, Eliashberg, & Wierenga (2007) study the hot topic of viral marketing. A particularly promising approach to studying network effects is agent-based modeling (Jager, 2007). These models represent decision rules of agents in a virtual market. Next, by means of simulations, the consequences of alternative scenarios and marketing strategies can be assessed. For example, Delre, Jager, Bijmolt and Janssen (2007) use the agent-based approach to study alternative communication strategies for new product introductions. Goldenberg, Libai, Modovan, & Muller (2007) use agent-based models to assess the net present value of bad news in conjunction with a new-product introduction. As marketing budgets gradually move to online media, there is a need for studies that assess the effects of these marketing efforts. The adoption of digital recorders (e.g., TIVO), where customers can watch television without exposure to commercials, will for instance change the allocation of advertising budgets over the media

P.S.H. Leeang et al. / Intern. J. of Research in Marketing 26 (2009) 1320 Table 2 Overview of papers, Marketing Dynamics Conference Marketing Dynamic Examples of related Modeling Methodology published papers Regression models Narayanan, Desiraju, and Chintagunta (2004) Tellis, Stremersch and Yin (2003) Foekens, Leeang and Wittink (1994) Conference papers Fischer and Albers (2007) Stremersch and Lemmens (2009) Andrews, Currim, Leeang and Lim (2008) Reimer, Rutz and Bucklin (2007) Deleersnyder, Steenkamp, Dekimpe and Leeang (forthcoming) Albuquerque and Bronnenberg (2007) Horvth and Wieringa (2008) Yoo and Pauwels (2007) Bruce and Foutz (2007) Substantive insights from Conference paper

17

Time-series

Lamey, Deleersnyder, Dekimpe and Steenkamp (2007)

Physician-oriented marketing efforts are effective in increasing primary demand. Patient-oriented marketing efforts are not: they are effective in market stealing Effects of regulation, national culture and economic wealth on international growth of new pharmaceuticals Accommodating store-level heterogeneity in the SCANPRO model does not improve the accuracy of marketing-mix elasticities relative to the homogenous model Different marketing instruments (TV, Radio, print and Internet advertising) affect customer spending on music downloads differently This study investigates the cyclical nature of advertising expenditures for different countries Combines time series with information on the distribution of consumer behavior variables to examine the effect of a new entrant on brand switching Accuracy of impulse response functions and forecasting of various pooling approaches at different levels of heterogeneity There is stronger long-term response to price increases than to decreases

VARX-models

State-space-models: DLM

Dekimpe and Hanssens (1999) Pauwels, Srinivasan and Franses (2007) Villanueva, Yoo and Hanssens (2008) Van Heerde, Mela and Manchanda (2004)

Kalman lters Choice models Gupta (1988); Van Heerde, Gupta and Wittink (2003) Bolton and Lemon (1999) Bell and Lattin (1998); Leenheer et al. (2007)

Spatial models Markov models Paas, Vermunt and Bijmolt (2007)

Agent-based models

Delre et al. (2007); Goldenberg et al. (2007)

Assessment of dynamic effects of WOM and advertising at different stages of sequentially distributed products Van Heerde, Srinivasan and Time-varying decomposition of the demand for a pioneering innovation into brand Dekimpe (2008) switching, category switching, and primary demand effects Ataman, Mela and Van Heerde Diffusion model assessing the effects of different marketing launch strategies; (2008) distribution has largest impact on new brand success Osinga, Leeang, Srinivasan and Direct-to-consumer advertising reduces systematic risk and increases idiosyncratic risk Wieringa (2008) of stocks of pharmaceutical companies Ebling and Klapper (2007) The effect of past price promotions on current price sensitivities on purchase incidence, brand choice and quantity Prins, Verhoef and Franses (2007) Early adopters tend to increase their post-adoption usage, while for late adopters the adoption usage decreases Breugelmans and Zhang (2007) Using data from an internet retailer, the authors study the effect of a category loyalty program on store visits, number of purchased items and total spending Bonfrer, Knox, Eliashberg and Using a Brownian motion model, the authors show that the change in usage over time Chiang (2007) and variability in usage are important predictors of churn Van Diepen, Donkers and Franses Direct mailing has short- and long-term impact on revenues, which also depend on (2007) competitor mailings Szymanowski and Gijsbrechts Consumption of a private label brand leads consumers to update their beliefs about the (2007) quality of other private label brands Hunneman, Bijmolt and Elhorst Spatial model predicting store performance at the zip code level (2007) Montoya, Netzer, and Jedidi Hidden Markov model for dynamics in customer behavior and the long-term impact of (2007) marketing mix Van der Lans et al. (2007) Markov process model predicting the number of participants in a viral marketing campaign Garcia (2007) Studying market dynamics through a simulation model for manufacturer actions and consumer-level decision making

landscape. In sum, these new technologies will provide more timeseries data, and may also reveal that certain traditional marketing efforts are losing their impact. At the same time, new marketing tactics will rise in application (online-advertising), which in turn calls for input from the marketing dynamics community to assess their impact. 3.7. Overview of conference papers Table 2 gives a schematic overview of most papers presented at the 2007 Marketing Dynamics Conference and their substantive insights. We classify papers using different research methodologies/models. We also provide examples of related papers that have been published recently. It is also interesting to note that several approaches and topics are absent or underrepresented in Table 2. Hazard models, purchase timing models, and structural models did not receive much explicit attention at the Groningen Marketing Dynamics Conference. This also holds for topics that drew more attention at earlier conferences and that deal with recent data richness: aggregation, level of parameterization and data pruning (Pauwels et al., 2004a).

3.8. Future outlook In this section we abstract from the specic papers presented at the 2007 Marketing Dynamics conference by providing a helicopter view on where we believe the eld of marketing dynamics is heading. A useful way to structure the discussion is to contrast four major approaches on a number of criteria. We include the three methods that have received ample attention in this article (VAR models, VEC models, State Space models), but also Dynamic Structural Models. These models are rooted in micro-economics and show how agents, on the demand side as well as the supply side, behave optimally in a context that involves dynamic relationships between variables. Chintagunta et al. (2006) and Sun (2006) provide an excellent overview of the type of marketing problems that can be studied with Dynamic Structural Models. We foresee that Dynamic Structural Models may grow in importance in the marketing literature in the future.4

4 To capitalize on this expected trend, there will be a tutorial on Dynamic Structural Models taught by Jean-Pierre Dub at the next Marketing Dynamics Conference (University of Waikato, New Zealand, 46 January 2009). ^

18

P.S.H. Leeang et al. / Intern. J. of Research in Marketing 26 (2009) 1320

Table 3 Ratings of four core dynamic approaches on six criteria Criterion Method Disentangle temporary (short-term) from persistent (long-term) effects; +: based on impulse response functions Time-varying parameters Cross-sectional heterogeneity : leads to explosion in number of parameters +: with Bayesian estimation Equilibrium modeling +/: no explicit equilibrium but it can be added +: includes an equilibrium component Limited-dependent Systems approach with many endogenous endogenous variables variables +: key purpose : endogenous variables are continuous : endogenous variables are continuous +: use a nonlinear transformation in observation equation +/: deriving optimality is easier for continuous variables

Vector Autoregressive Model

Vector Error +: has separate parameters Correction Model for short-term and long-term effects State Space Models (Kalman lters and DLMs) Dynamic Structural Model +: observation equation for short-term effects and state equation for long-term effects +: simulation for short- and long-term responses to policy changes

+/: moving window approach possible, but requires proper window choice +/: moving window approach possible , but requires proper window choice +: state equation is time varying

+/: frequentist estimation faster than Bayesian estimation : state space quickly becomes very large

: complicates deriving optimality conditions

: does not +/: possible yet time-consuming with include an equilibrium Bayesian estimation component +: the economic +: micro-economic model at the rm or optimum is often an equilibrium consumer level

: complicates deriving optimality conditions

It goes without saying that each of these four methods should capture appropriate metrics (the rst criterion discussed in Section 2.1). To contrast the approaches, we select six criteria, three of which have been discussed previously (13) and three new ones (46) that are linked to model aspects we expect to be increasingly relevant in future studies in marketing dynamics: 1. disentangle temporary (short-term) from persistent (long-term) effects; 2. account for time-varying parameters; 3. allow for cross-sectional heterogeneity; 4. equilibrium modeling (i.e., including equilibrium as a model component, or deriving the full model as an equilibrium outcome); 5. systems approach (i.e., modeling relationships between many endogenous variables), and 6. limited-dependent (i.e., non-metric or non-continuous) endogenous variables. Examples include a binomial variable for purchase incidence, a multinomial variable for brand choice, a discrete variable for purchase quantity, a duration variable for interpurchase time, and other endogenous variables that are often (but not always) the result of modeling at the individual level. Table 3 summarizes our (arguably personal) view on how well each method (in the rows) scores on the criteria (in the columns). We adopt a consumer-report style scale, with a + means it copes well, means it copes poorly, and a +/ means it copes neither well nor poorly. Of course, these ratings are somewhat generalistic. They do not reect the fact that, within some methods, there are already some developments that will eventually lead to better ratings on the criteria after all, science evolves5. Table 3 shows that none of the methods dominates all others on all criteria. Choosing a suitable method thus depends on the purpose of each research study. All methods are suited for disentangling short- from longterm effects, but their philosophies are vastly different (see the second column of Table 3). If parameter variation over time is essential, State Space models are the most natural choice. However, these models are less suited to handle many endogenous variables, in which case a systems approach (VEC or especially VAR) becomes more desirable. On the other hand, allowing for cross-sectional heterogeneity in VAR models implies a separate model for each cross-sectional unit, which leads to an explosion in the number of parameters. The other approaches, especially when

captured in a hierarchical Bayesian specication, seem more suited for handling cross-sectional heterogeneity. When the research project involves an equilibrium around which the endogenous variables are evolving, and the researcher wants to make micro-economic assumptions on how this equilibrium is obtained, Dynamic Structural Models are the best option. If a researcher has fewer prior insights on the nature/existence of an equilibrium relationship between non-stationary variables, cointegration testing and (in case cointegration is present) VEC models can be used (Dekimpe & Hanssens, 1999, 2004). However, VEC models are not only appropriate in case of cointegrated, non-stationary variables. They can also be used to make the equilibrium underlying a set of stationary variables more explicit. We refer to Hendry (1995, Section 6.5) for an in-depth discussion, and to Fok et al. (2006), Van Heerde, Helsen and Dekimpe (2007) and Van Heerde, Srinivasan and Dekimpe (2008) for recent marketing applications using Bayesian estimation. When the model involves limited-dependent endogenous variables (e.g., it is specied at the individual level), we recommend either a State Space Approach (without micro-economic assumptions) or a Dynamic Structural model (with micro-economic assumptions). We anticipate that, going forward, the relative importance of the criteria in Table 3 will determine how frequently the four different methods will be applied in studying dynamic marketing problems. 4. Conclusion The fascinating eld of marketing dynamics is developing rapidly. The issues that are tackled are typically highly relevant for senior management, the (modeling) challenges are intellectually stimulating, and the scope of new research opportunities is endless. The eld attracts studies from all paradigms. For example, the 2007 Marketing Dynamics Conference featured not only aggregate timeseries models, but also individual-level structural models (e.g., Kopalle, Neslin, Sun, Sun, & Swaminathan, 2007), consumer learning models (e.g., Szymanowski & Gijsbrechts, 2007; Loureno, Gijsbrechts & Paap, 2007), and a latent Markov model for dynamic segmentation (Paas, Vermunt & Bijmolt, 2007). We anticipate that the study of marketing dynamics (as reected in the 2007 Conference) will lead to several milestone papers in the marketing literature. We are condent that the exciting debate about modeling marketing dynamics will continue, not only in the academic journals, but in particular at future (Marketing Dynamics) conferences.

5 For example, the VAR model may be extended to a Qual-VAR model that allows for binary endogenous variables and is estimated by MCMC methods (Joshi, 2007).

P.S.H. Leeang et al. / Intern. J. of Research in Marketing 26 (2009) 1320

19

References
Albuquerque, P., & Bronnenberg, B. J. (2007). Measuring consumer switching to a new brand across local markets. Marketing Dynamics Conference 2007. Andrews, R. L., Currim, I. S., Leeang, P. S. H., & Lim, J. (2008). Estimating the SCANPRO model of store sales? HB, FM or just OLS? International Journal of Research in Marketing, 25(1), 2233. Ataman, M. B. (2007), Managing brands. Doctoral Dissertation, Tilburg University, the Netherlands. Ataman, M. B., Mela, C. F., & Van Heerde, H. J. (2007). Building brands. Marketing Science, 27(6), 10361054. Ataman, M. B., Van Heerde, H. J., & Mela, C. F. (2007). The long-term effect of marketing strategy on brand performance. Working paper, Erasmus School of Management, the Netherlands. Batislam, E. P., Deniziel, M., & Filizetekin, A. (2007). Empirical validation and comparison of models for customer base analysis. International Journal of Research in Marketing, 24(3), 201209. Bell, D. R., & Lattin, J. M. (1998). Shopping behavior and consumer preference for store price format: Why large basket shoppers prefer EDLP. Marketing Science, 17(1), 6668. Bolton, R. N., & Lemon, K. N. (1999). A dynamic model of customers' usage of services: usage as an antecedent and consequence of satisfaction. Journal of Marketing Research, 36(2), 171186. Bonfrer, A., Knox, G., Eliashberg, J., & Chiang, J. (2007). Diagnosing and predicting individual customer defection in a contractual setting. Marketing Dynamics Conference 2007. Breugelmans, E., & Zhang, J. (2007). The impact of category-based loyalty programs. Marketing Dynamics Conference 2007. Bruce, N., & Foutz, Z. (2007). Dynamic advertising and word-of-mouth effectiveness in sequential distribution of short lifecycle products. Marketing Dynamics Conference 2007. Burgess, S. M., & Steenkamp, J. B. E. M. (2006). Marketing renaissance: How research in emerging markets advances marketing science and practice. International Journal of Research in Marketing, 23(4), 337356. Cain, P. M. (2005). Modelling and forecasting brand share: A dynamic demand system approach. International Journal of Research in Marketing, 22, 203220. Chintagunta, P., Erdem, T., Rossi, P. E., & Wedel, M. (2006). Structural modeling in marketing: Review and assessment. Marketing Science, 25(6), 604616. Chung, T. S., Rust, R. T., & Wedel, M. (2007). My mobile music: an adaptive personalization system for digital audio players. Marketing Dynamics Conference 2007. Dekimpe, M. G., & Hanssens, D. M. (1999). Sustained spending and persistent response: A new look at long-term marketing protability. Journal of Marketing Research, 36(4), 397412. Dekimpe, M. G., & Hanssens, D. M. (2000). Time-series models in marketing: Past, present and future. International Journal of Research in Marketing, 17, 183193. Dekimpe, M. G., & Hanssens, D. M. (2004). Persistence modeling for assessing marketing strategy performance. In D. Lehmann & C. Moorman (Eds.), Assessing marketing strategy performance, Marketing Science Institute (pp. 6993). Deleersnyder, B., Gielens, K., Geyskens, I., & Dekimpe, M. G. (2002). How cannibalistic is the Internet channel? A study of the newspaper industry in the United Kingdom and The Netherlands. International Journal of Research in Marketing, 19(4), 337348. Deleersnyder, B., Steenkamp, J. B. E. M., Dekimpe, M. G., & Leeang, P. S. H. (forthcoming). The role of national culture in advertisings sensitivity to business cycles: An investigation across continents. Journal of Marketing Research, 46. Delre, S. A., Jager, W., Bijmolt, T. H. A., & Janssen, M. A. (2007). Targeting and timing promotional activities: An agent-based model for the take-off of new products. Journal of Business Research, 60(8), 826835. Donkers, B., Verhoef, P. C., & De Jong, M. G. (2007). Modeling CLV: A test of competing models in the insurance industry. Quantitative Marketing and Economics, 5(2), 163190. Ebling, C., & Klapper, D. (2007). Dynamic effects of promotions in choice data: Differences with respect to whether, what and how much to purchase. Marketing Dynamics Conference 2007. Fischer, M., & Albers, S. (2007). Patient- or physician-oriented marketing: What drives primary demand for prescription drugs? Marketing Dynamics Conference 2007. Foekens, E. W., Leeang, P. S. H., & Wittink, D. R. (1994). A comparison and an explanation of the forecasting accuracy of a loglinear model at different levels of aggregation. International Journal of Forecasting, 10(2), 245261. Foekens, E. W., Leeang, P. S. H., & Wittink, D. R. (1999). Varying parameter models to accommodate dynamic promotion effects. Journal of Econometrics, 89(1-2), 249268. Fok, D., Horvth, C., Paap, R., & Franses, P. H. (2006). A hierarchical Bayes error correction model to explain dynamic effects of price changes. Journal of Marketing Research, 43(3), 443462. Garcia, R. (2007). Modeling vehicle choice behavior using agent-based modeling approach. Marketing Dynamics Conference 2007. Geyskens, I., Gielens, K., & Dekimpe, M. G. (2002). The market valuation of internet channel additions. Journal of Marketing, 66(2), 102119. Gielens, K., & Steenkamp, J. B. E. M. (2007). Drivers of consumer acceptance of new packaged goods: An investigation across products and countries. International Journal of Research in Marketing, 22(2), 97111. Gielens, K., Van de Gucht, L. M., Steenkamp, J. B. E. M., & Dekimpe, M. G. (2008). Dancing with a giant: The effect of Wal-Mart's entry into the U.K. on the performance of European retailers. Journal of Marketing Research, 45(5), 519534. Goldenberg, J., Libai, B., Modovan, S., & Muller, E. (2007). The NPV of bad news. International Journal of Research in Marketing, 24(3), 186200.

Gupta, S. (1988). Impact of sales promotions on when, what, and how much to buy. Journal of Marketing Research, 25(4), 342355. Gupta, S., Lehmann, D. R., & Stuart, J. A. (2004). Valuing customers. Journal of Marketing Research, 41(1), 718. Gupta, S., & Zeithaml, V. A. (2006). Customer metrics and their impact on nancial performance. Marketing Science, 25(6), 718739. Hanssens, D. M., & Dekimpe, M. G. (2008). Modeling the nancial-performance effects of marketing. In B. Wierenga (Ed.), Handbook of Marketing Decision Models (pp. 501523). Harrison, P. J., & Stevens, C. F. (1976). Bayesian forecasting. Journal of the Royal Statistical Society. Series B (Methodological), 38(3), 205247. Hendry, D. F. (1995). Dynamic Econometrics. Oxford University Press. Hodrick, R. J., & Prescott, E. C. (1997). Postwar U.S. business cycles: An empirical investigation. Journal of Money, Credit and Banking, 29(1), 116. Horvth, C., Leeang, P. S. H., Wieringa, J. E., & Wittink, D. R. (2005). Competitive reaction- and feedback effects based on VARX models of pooled store data. International Journal of Research in Marketing, 22(4), 415426. Horvth, C., & Wieringa, J. E. (2008). Pooling data for the analysis of dynamic marketing systems. Statistica Neerlandica, 62(2), 208229. Hunneman, A., Bijmolt, T. H. A., & Elhorst, J. P. (2007). A spatial-lag random-effects hierarchical model for store location evaluation. Marketing Dynamics Conference 2007. Jager, W. (2007). Multi-agent simulation in experimenting with market dynamics: prospects for further research. Marketing Dynamics Conference 2007. Joshi, A. (2007). The Feedback Effects of Innovation: What can lead to Innovation Momentum? Working Paper : University of Central Florida. Joshi, A., & Hanssens, D. M. (2008). Advertising spending, competition and stock return. UCLA Marketing Studies Center Working paper. Knox, G., & Van Oest, R. (2007). I'm here, but maybe not for long: Using complaints to predict and value customers. Marketing Dynamics Conference 2007. Kolsarici, & Vakratsas (2007). Dynamic market-level effects of highly regulated advertising messages. Marketing Dynamics Conference 2007. Kopalle, P. K., Neslin, S. A., Sun, B., Sun, Y., & Swaminathan, V. (2007). A dynamic structural model of the impact of loyalty programs on customer behavior. Marketing Dynamics Conference 2007. Kornelis, M., Dekimpe, M. G., & Leeang, P. S. H. (2008). The structural impact of multiple entrants on long-run market growth. International Journal of Research in Marketing, 23(3), 173182. Lachaab, M., Ansari, A., Jedidi, K., & Trabelsi, A. (2006). Modeling preference evolution in discrete choice models: A Bayesian state-space approach. Quantitative Marketing and Economics, 4(3), 5781. Lamey, L., Deleersnyder, B., Dekimpe, M. G., & Steenkamp, J. -B. E. M. (2007). How business cycles contribute to private label success: Evidence from the United States and Europe. Journal of Marketing, 71(1), 115. Leeang, P. S. H., Parreno Selva, J., Van Dijk, A., & Wittink, D. R. (2008). Decomposing the sales promotion bump accounting for cross-category effects. International Journal of Research in Marketing, 25(3), 201214. Leeang, P. S. H., Wittink, D. R., Wedel, M., & Naert, P. A. (2000). Building models for marketing decisions. Boston, MA: Kluwer Academic Publishers. Leenheer, J., Van Heerde, H. J., Bijmolt, T. H. A., & Smidts, A. (2007). Do loyalty programs really enhance behavioral loyalty? An empirical analysis accounting for selfselecting members. International Journal of Research in Marketing, 24(1), 3147. Loureno, C., Gijsbrechts, E., & Paap, R. (2007). Dynamic store price image formation and category pricing. Marketing Dynamics Conference 2007. Mahajan, V., Bretschneider, S. I., & Bradford, J. W. (1980). Feedback approaches to modeling structural shifts in market response. Journal of Marketing, 44(Winter), 7180. Marketing Science Institute (2007). http://www.msi.org/research/index.cfm?id=43, accessed on December 13, 2007. Mela, C. F., Gupta, S., & Lehmann, D. R. (1997). The long-term impact of promotion and advertising on consumer brand choice. Journal of Marketing Research, 34(2), 248261. Melnyk, V., & Bijmolt, T. H. A. (2007). Effects of having and terminating a loyalty program: The role of monetary and non-monetary rewards. Working paper New Zealand: University of Waikato. Montoya, R., Netzer, O., & Jedidi, K. (2007). Managing customers through marketing mix allocation for long-term protability. Marketing Dynamics Conference 2007. Naik, P. A., & Raman, K. (2003). Understanding the impact of synergy in multimedia communications. Journal of Marketing Research, 40(4), 375388. Naik, P. A., Raman, K., & Winer, R. S. (2005). Planning marketing-mix strategies in the presence of interaction effects. Marketing Science, 24(1), 2534. Naik, P. A., Mantrala, M. K., & Sawyer, A. G. (1998). Planning media schedules in the presence of dynamic advertising quality. Marketing Science, 17(3), 214235. Narayanan, S., Desiraju, R., & Chintagunta, P. K. (2004). Return on investment implications for pharmaceutical promotional expenditures: The role of marketing-mix interactions. Journal of Marketing, 68(4), 90105. Neelamegham, R., & Chintagunta, P. K. (2004). Modeling and forecasting the sales of technology products. Quantitative Marketing and Economics, 2(3), 195232. Osinga, E. C., Leeang, P. S. H., Srinivasan, S., & Wieringa, J. E. (2008). The effects of consumer advertising: a shareholder's perspective with application to the pharmaceutical industry. Working paper. University of Groningen. Paas, L. J., Vermunt, J., & Bijmolt, T. H. A. (2007). Discrete-time discrete-state latent Markov modelling for assessing and predicting household acquisitions of nancial products. Journal of the Royal Statistical Society, Series-A, 170(4), 955974. Parsons, L. J. (1975). The product life cycle and time-varying advertising elasticities. Journal of Marketing Research, 12(4), 476480. Pauwels, K., Currim, I., Dekimpe, M. G., Hanssens, D. M., Mizik, N., Ghysels, E., & Naik, P. (2004). Modelling marketing dynamics by time series econometrics. Marketing Letters, 15(4), 167183.

20

P.S.H. Leeang et al. / Intern. J. of Research in Marketing 26 (2009) 1320 Tellis, G., & Johnson, J. (2007). The value of quality. Marketing Science, 26(6), 758773. Tellis, G., Stremersch, S., & Yin, E. (2003). The international takeoff of new products: The role of economics, culture, and country innovativeness. Marketing Science, 22(2), 188208. Van der Lans, R., Van Bruggen, G., Eliashberg, J., & Wierenga, B. (2007). Forecasting in viral marketing campaigns. Marketing Dynamics Conference 2007. Van Diepen, M., Donkers, B., & Franses, P. H. (2007). Dynamic and competitive effects of direct mailings. Marketing Dynamics Conference 2007. Van Doorn, J., & Verhoef, P. C. (2008). Negative incidents and the impact of satisfaction on customer share. Journal of Marketing, 72(4), 123142. Van Everdingen, Y. M., Aghina, W. B., & Fok, D. (2005). Forecasting cross-population innovation diffusion: A Bayesian approach. International Journal of Research in Marketing, 22(3), 293308. Van Heerde, H. J., Dekimpe, M. G., & Putsis, W. P., Jr. (2005). Marketing models and the Lucas critique. Journal of Marketing Research, 42(1), 1521. Van Heerde, H. J., Gupta, S., & Wittink, D. R. (2003). Is 75% of the sales promotion bump due to brand switching? No, only 33% is. Journal of Marketing Research, 40(4), 481491. Van Heerde, H. J., Helsen, K., & Dekimpe, M. G. (2007). The impact of a product-harm crisis on marketing effectiveness. Marketing Science, 26(2), 230245. Van Heerde, H. J., Mela, C. F., & Manchanda, P. (2004). The dynamic effect of innovation on market structure. Journal of Marketing Research, 41(2), 166183. Van Heerde, H. J., Srinivasan, S., & Dekimpe, M. G. (2008). Decomposing the demand for a pioneering innovation. Working paper, University of Waikato, Department of Marketing. Venkatesan, R., Reinartz, W., & Ravishanker, N. (2008). The role of customer attitudes in CRM activities, Working paper. Verhoef, P. C., Neslin, S. A., & Vroomen, B. (2007). Multichannel customer management: Understanding the research-shopper phenomenon. International Journal of Research in Marketing, 24(2), 129148. Villanueva, J., Yoo, S., & Hanssens, D. M. (2008). The impact of marketing-induced vs. word-of-mouth customer acquisition on customer equity growth. Journal of Marketing Research, 45(1), 4859. Xie, J., Song, M., Sirbu, M., & Wang, Q. (1997). Kalman lter estimation of new product diffusion models. Journal of Marketing Research, 34(3), 378393. Yoo, S., & Hanssens, D. M. (2008). Measuring marketing effects on customer equity for frequently purchased brands. UCLA Marketing Studies Center Working paper. Yoo, S., & Pauwels, K. (2007). The generalized long-term impact of marketing, Working Paper, Korea University.

Pauwels, K., & Hanssens, D. M. (2007). Performance regimes and marketing policy shifts. Marketing Science, 26(3), 293311. Pauwels, K., Nijs, V., & Srinivasan, S. (2007). Managing brand equity with product line extensions and contractions. Marketing Dynamics Conference 2007. Pauwels, K., Silva-Risso, J., Srinivasan, S., & Hanssens, D. M. (2004). New products, sales promotions, and rm value: The case of the automobile industry. Journal of Marketing, 68(4), 142156. Pauwels, K., Srinivasan, S., & Franses, P. H. (2007). When do price thresholds matter in retail categories? Marketing Science, 26(1), 83100. Prins, R., Verhoef, P. C., & Franses, P. H. (2007). Do early adopters use more? A longitudinal examination of the inuence of adoption timing on postadoption usage. Marketing Dynamics Conference 2007. Rao, V. R., Agarwal, M. K., & Dahlhoff, D. (2004). How is manifest branding strategy related to the intangible value of a corporation. Journal of Marketing, 68(4), 126141. Reimer, K., Rutz, O. J., & Bucklin, R. E. (2007). Got downloads? A model of the impact of advertising and promotion on customer spending in the music download industry. Marketing Dynamics Conference 2007. Roberts, J., Kayande, U., & Stremersch, S. (2007). Impact of marketing science on practice. European Marketing Association Conference Presentation. Rust, R. T., Lemon, K. N., & Zeithaml, V. A. (2004). Return on marketing: Using customer equity to focus marketing strategy. Journal of Marketing, 68(1), 109127. Simon, C. J., & Sullivan, M. W. (1993). The measurement and determinants of brand equity: A nancial approach. Marketing Science, 12(1), 2852. Srinivasan, S., & Hanssens, D. M. (2009). Marketing and rm value, Journal of Marketing Research, forthcoming. Srinivasan, S., Vanhuele, M., & Pauwels, K. (2008). Mindset metrics and marketing mix models: never the twain shall meet? Working paper, University of California, Riverside. Sriram, S., Chintagunta, P. K., & Neelamegham, R. (2006). Effects of brand preference, product attributes, and marketing mix variables in technology product markets. Marketing Science, 25(5), 440456. Sriram, S., & Kalwani, M. (2007). Optimal advertising and promotion budgets in dynamic markets with brand equity as a mediating variable. Management Science, 53(1), 4660. Stremersch, S. & Lemmens, A. (2009). Sales Growth of New Pharmaceuticals Across the Globe: The Role of Regulatory Regimes, Marketing Science, forthcoming. Stremersch, S., & Tellis, G. (2004). Understanding and managing international growth of new products. International Journal of Research in Marketing, 21(4), 421438. Sun, B. (2006). Dynamic structural consumer models and current marketing issues. Marketing Science, 25(6), 625628. Szymanowski, M., & Gijsbrechts, E. (2007). Conditional cross-brand learning: When are private labels really private? Marketing Dynamics Conference 2007.