Content uploaded by Ricardo Limongi
Author content
All content in this area was uploaded by Ricardo Limongi on May 16, 2016
Content may be subject to copyright.
1
How marketing balance the battle between branded and unbranded products in an
emerging economy? A time series analysis of brand sales
Autoria: Marcos Inácio Severo de Almeida, Rafael Barreiros Porto, Ricardo Limongi França Coelho
Agradecimentos à CAPES – Coordenação de Aperfeiçoamento de Pessoal de Nível Superior –
pelo financiamento desta pesquisa.
1. Propósito central do trabalho
For the last decade, prominent journals have been emphasizing the need for research to
address problems originated at emerging economies. This movement had started after Burgees
and Steenkamp (2006) stressed the significant environmental conditions of these contexts and
intensified after Sheth’s (2011) foundational paper about the five key characteristics of
emerging markets. Marketing decision making is based on rules of thumb decisions on
industrialized countries (Leeflang, Wieringa, Bijmolt, & Pauwels, 2015) and conditions in
emerging economies seem to intensify this characteristic, as they present very specific forms
of retail competition, marked by unbranded products. The purpose of this article is to address
these issues by developing a model to explain sales dynamics. “Dynamics” in our approach is
classified by two dimensions: brand sales behaviour in time (which can be stationary or non-
stationary) and brand sales position (negative, neutral or positive, compared to the initial
period). The combination among these dimensions determine six possible positions for brand
sales: Ascending, Promising, Steady, Unsteady, Descending and Faltering. They range from
what is considered the worst to the best scenario for managers. Our approach uses a unique
transactional data set from a multinational company and is applied to a specific situation, the
gasoline retailing. After conducting time series analysis to classify sales positions, a final
multinomial logit model was developed to predict them using marketing mix variables.
2. Marco teórico
Ever since the publication of the empirical generalizations about sales evolution and
stationarity (Dekimpe & Hanssens, 1995), researchers have been detailing marketing mix
effects to identify which elements are most critical in providing success for brand sales and in
change for a turnaround strategy on performance (Pauwels & Hanssens, 2007). Dynamic
models concentrate on the long-term of marketing strategy, usually resorting to price and
advertising, opening avenues for research dedicated to measure the impact of other variables
in different contexts.
The same empirical generalization that provided the basis for what studies recognize as basic
conditions for evolution or stationarity also highlighted the need for further research. This
request was followed by the assertions that the nature of product category and length of time
span could influence sales behaviour classification (Dekimpe & Hanssens, 1995). One of
these issues was addressed by Pauwels and Hanssens (2007) when they divided the sales
series of frozen dinners brands in rolling time windows and found unstable brand performance
in mature markets.
Our conceptual framework establishes itself on these advancements to develop a descriptive
model to identify brand sales behaviour and position in an emering economy. First, it
recognizes different patterns related to the first two moments of statistics, mean and variance.
For example: brand sales could be in a mean-reverting position (stationarity) or in a situation
where no fixed mean is observed and variance is increasing, defined as evolution (Dekimpe &
Hanssens, 1995). However, this is not the only possible outcome for performance, which can
2
be unstable (Pauwels & Hanssens, 2007). Comparatively, sales can also be, in the present, in a
negative, neutral or positive position, due to its trend significance (or insignificance).
Therefore, the descriptive model grounds on two dimensions (sales behaviour in time and
sales final position) to define six possible positions to brand sales.
3. Método de investigação, se pertinente
The setting for our study is the gasoline retailing in Brazil, a specific situation where
companies market branded and unbranded gasoline and ethanol for passenger cars and where
diesel is only authorized for heavier vehicles. Ethanol and gasoline serve as substitutables due
to the existence of flex-fuel vehicles, leading to what managers classify as switching
behaviour and what researchers specify as products functionally interchangeable and,
therefore, imperfect substitutes (Foxall, 1999). These products are sold in petrol stations,
partners that are mostly independent and under different contracts with energy companies.
Energy companies must rely on these partners and invest on strategies to increase sales, such
as service (in the form of convenience stores) and loyalty cards to induce frequency purchase
patterns. The data set comprises of sales and marketing data on 174 petrol stations during 27
months, from 2011 to 2013 (a total of 4681 observations). They are from an entire sales
region where petrol stations are located in 41 cities, distributed in five different states. Focal
variable is defined as brand sales, measured in the form of a ratio between the branded and the
unbranded gasoline. After the construction of brand sales variable, research methodology was
developed using three steps. Firstly, we classified observations according to “sales behaviour
in time” dimension. In order to identify if a brand sales was stationary or non-stationary, we
conducted formal unit root tests and tests for structural breaks. For the second dimension, we
used brand sales as a dependent variable and the natural logarithm of the time trend (for each
petrol station time series) as the only independent variable in a simple double-log regression.
The inspiration for this step was Pauwels and Hanssens (2007) least-squares estimation of a
time trend. After the categorization of the aforementioned sales positions to each petrol
station, we defined the independent variables. They refer to product, service, price, brand,
promotion, loyalty and salesforce decisions, while two control variables relate to the size of
the petrol station and to state taxation. The final methodological stage encompassed running a
multinomial logit model where we grouped the dependent variable into six mutually exclusive
and unordered groups (sales positions).
4. Resultados, conclusões e suas implicações para a área
Main results indicate that price is negatively associated to Ascending position, which means
that increases on price reduce the probability for brand sales to remain on a positive (and
stationary) position. Conversely, the existence of high levels of service (convenience stores)
increases the probability for brand sales to stay Ascending or Promising. Salesforce is also
positively associated to a Promising position. Product variables (ethanol and diesel) may
explain how to avoid negative positions in order to stay neutral. It is also important to note the
positive coefficient for a substitutable product (ethanol) and the negative coefficient for a non-
substitutable (diesel) product on Descending and Faltering. Our work provides three outputs
for marketing research and management: (1) a simple and descriptive classification that can
be useful for managers, based on sales positions in an emerging economy setting; (2) the
evidence of marketing important role to these positions; (3) and a simple and rational
framework to managers involved with decisions regarding brand sales based on time series
analysis.
3
5. Referências
Azimont, F., & Araujo, L. (2010). The making of a petrol station and the “on-the-move
consumer”: Classification devices and the shaping of markets. Industrial Marketing
Management, 39(6), 1010-1018.
Dekimpe, M. G., & Hanssens, D. M. (1995). Empirical generalizations about market
evolution and stationarity. Marketing Science, 14(3_supplement), G109-G121.
Leeflang, P. S. H., Wieringa, J. E., Bijmolt, T. H. A., & Pauwels, K. H. (2015). Modeling
Markets – Analyzing Marketing Phenomena and Improving Decision Making. New York:
Springer.
Pauwels, K., & Hanssens, D. M. (2007). Performance regimes and marketing policy shifts.
Marketing Science, 26(3), 293-311.
Sheth, J. N. (2011). Impact of emerging markets on marketing: Rethinking existing
perspectives and practices. Journal of Marketing, 75(4), 166-182.
