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

net/publication/312208849

DEMAND FORECASTING IN RETAIL GROCERY STORES IN THE CZECH

REPUBLIC

Conference Paper · August 2015

DOI: 10.5593/SGEMSOCIAL2015/B22/S7.089

CITATION READS

1 5,181

3 authors, including:

Michal Paták Zuzana Pecinova

University of Pardubice University of Pardubice
43 PUBLICATIONS   62 CITATIONS    32 PUBLICATIONS   20 CITATIONS   

SEE PROFILE SEE PROFILE

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

Supply Chain Collaboration in Food and Chemical Industry View project

Demand Forecast View project

All content following this page was uploaded by Michal Paták on 05 May 2017.

The user has requested enhancement of the downloaded file.

 Economics and Tourism

DEMAND FORECASTING IN RETAIL GROCERY STORES

IN THE CZECH REPUBLIC

Assist. Prof. Michal Patak, M.Sc., Ph.D.

Assoc. Prof. Lenka Branska, M.Sc., Ph.D.
Assist. Prof. Zuzana Pecinova, M.Sc., Ph.D.
University of Pardubice, Czech Republic

ABSTRACT
Demand forecasting plays a key role in the management of logistics processes in
retailing. The quality of the resulting forecasts has a direct impact on the efficiency of
processes not only in retail stores, but also throughout the supply chain they are part of.
Retail stores are in fact in daily contact with end customers in the supply chain, and
therefore they are largely involved in forecasting in the application of collaborative
strategies in supply chain management (e.g. Quick Response, Effective Consumer
Response, Collaborative Planning, Forecasting and Replenishment).
The level of demand forecasting in retail grocery stores in the Czech Republic was the
subject of the quantitative research. The main objective of the research was to find out
what methods are used in demand forecasting, what factors affecting demand for food
are considered and what sources of information retail stores use to base the forecasting
on. When analyzing the data obtained through personal inquiries in 75 retail outlets,
methods of descriptive and inferential statistics were used.
The research showed that retailers rely in demand forecasting mainly on their own
intuition and experience in retailing, and therefore qualitative methods of forecasting are
applied most frequently. Although retailers consider a variety of factors influencing the
demand, the research has identified an absence of statistical methods that are able to
effectively implement the influence of factors into the final predictions. The article also
discusses the differences among stores with sales area of different sizes and depending
on the affiliation of the store with a retail chain. A low rate of applying quantitative
forecasting methods in retail stores and the availability of a small base of input
information on the demand of end customers can have a negative impact on the
implementation of modern methods in supply chain management.
Keywords: demand forecasting, grocery stores, sales forecasting, supply chain, retail

INTRODUCTION
In the current turbulent market environment, forecasting the volume of demand and the
future development has of an irreplaceable role in the management of every retail
business. Reliable demand forecasting is a prerequisite for efficient and effective
customer satisfaction. It makes possible to efficiently use resources and reduce costs
associated with excess, or lack of goods within the supply chain, as relevant. The
quality of the demand forecasting process in the retail sector is closely linked with the
choice of appropriate forecasting methods that rely on relevant information from
SGEM 2015 International Multidisciplinary Scientific Conferences on Social Sciences and Arts

information sources available to retailers and at the same time are able to translate into
the resulting forecasts influences of all internal and external factors that significantly
affect future customer demand in the B2C markets.
In the B2C markets, demand is primarily influenced by social, cultural, psychological
and personality factors [1]. Every customer has his/her own priorities, but also reasons
to buy certain goods. The crucial point is thus the customer's personality and lifestyle.
When choosing food, customers also consider a number of factors inherent in the
particular food. Most often, they evaluate the freshness, nature and cost of the particular
food [2]. According to research [3], demand for food is influenced by many factors.
Besides the quality of goods and habitual behaviour of purchasers and price of the
commodity, it also mentions various forms of sales promotion in retail business,
recommendations from other customers, brand loyalty, etc.
Retailers can choose from a wide variety of forecasting methods. Using qualitative
forecasting methods (retailers’ opinions, expert estimations, marketing studies) is not
typical for predicting demand for a wide range of goods on the consumer markets,
because it is very time consuming and costly [1]. Retailers therefore mostly work with
time series analysis of past sales, which allows them relatively easily to identify
seasonal fluctuations in sales as well as changes in the trend during the entire product
life cycle. In order to quantify the impact of product prices and sales promotion on the
demand, regression models [4] or data mining techniques are used [5], which are based
on a large database of information about customer behaviour in the form of point of
sales data (POS data). It is a database of past sales in the context of the implementation
of various forms of sales promotion and combinations thereof, price sensitivity of
consumers, cannibalization effects, impact of weather, etc. POS data represent a
valuable source of information about demand, as data on sales in the retail trade are not
distorted by chain-related effects and are also readily available from the retail
information system [6].
The quality of the forecasting process is primarily reflected in the accuracy of forecasts
produced, which is one of the major causes of bullwhip effect in supply chains [7].
Inaccurate forecasts by retailers generate inadequate orders for links further down the
supply chain and cause many difficulties in coordinating material flows within the
supply chain. This can be demonstrated, for example, with a case study reported in the
literature [8], in which the bullwhip effect was identified in a Czech manufacturer of
food products, especially due to the frequent realization of price sales promotion, to
which partners in the supply chain failed to react effectively.
There is no study that would assess the current level of the forecasting process in retail
stores in the Czech market. Therefore, the aim of this paper is to evaluate the level of
the forecasting process, based on a quantitative research among Czech retail stores with
food goods, namely in terms of adequate forecasting techniques, considering the
relevant factors affecting the demand for food, and the use of appropriate sources of
information for making forecasts.

RESEARCH METHODOLOGY
To achieve the set objectives, a quantitative research was conducted in 75 selected retail
stores in the Czech Republic. The sample included only retail stores where groceries
 Economics and Tourism

predominate in the range of goods. Data collection took place in the period of March-
April 2015 using face-to-face interviews with predetermined questions. The respondents
first commented on the possible use of demand forecasting in inventory management of
goods in retail business. In retail stores, where demand forecasts are generated and used,
open-ended questions were used to ascertain:
- frequency of use of selected forecasting methods,
- factors considered in forecasting the demand,
- sources of information used for forecasting the demand.
In the surveyed stores, the possible affiliation with retail chains and the size of the sales
area were also identified. The sample of retail stores was comprised of 52 percent of
independent retail stores and 48 percent of stores that are part of retail chains. In terms
of the size of the sales area, four groups of retail stores can be identified in the sample,
namely stores with a sales area of up to 50 m2 (41 %), stores with an area of 51 m2 to
200 m2 (33 %), stores with an area of 201 m2 to 400 m2 (12 %) and stores with an area
greater than 400 m2 (14 %).
The data were processed by means of the statistical software IBM SPSS Statistics (v.
22). In data analysis, methods of descriptive statistics (multiple response analysis) and
inferential statistics (statistical hypothesis testing) were used. In each research area, we
first analyzed the frequency of responses for all the respondents and then analyzed the
differences between different groups of respondents. The divergence of the empirical
distribution of responses of the respondents according to the foregoing attributes was
tested using Exact Pearson χ2-test at the 0.05 level of significance (based on Monte
Carlo sampling).

RESEARCH RESULTS
Based on data analysis, it was found that the majority of retail stores (91 %) creates
their own demand forecasts and uses them to support decision making in the
management and replenishment of goods in the retail business. Other respondents who
do not make forecasts most often replenish their goods according to revenues from past
sales or work with predictions taken from the retail chain headquarters. Since the
research was focused on the method of demand forecasting in the retail trade, these
respondents were excluded from the next part of the survey. The new structure of the
respondents according to selected classification criteria is shown in Table 1.

Table 1 Structure of respondents

Classification criteria Frequency
Affiliation Independent store 50%
Retail chain 50%
Sales area <= 50 m2 41%
2
51 - 200 m 34%
201 - 400 m2 10%
2
> 400 m 15%
SGEM 2015 International Multidisciplinary Scientific Conferences on Social Sciences and Arts

In another part of the research, the respondents chose any number of the available
methods, which are used in retail demand forecasting. The ability to specify a different
forecasting method that did not occur in the menu was not used by any of the
respondents. The frequency distribution of the responses is shown in Table 2.

Table 2 Forecasting methods used

Forecasting method Responses Percent
Frequency Percent of cases
Judgmental method 63 40% 94%
(Moving) Average 33 21% 49%
Naive method 30 19% 45%
Customer expectations 14 9% 21%
Unknown methods implemented in software 8 5% 12%
Analogy method 6 4% 9%
Simple regression 4 3% 6%
Time series decomposition 1 1% 1%
Exponential smoothing 0 0% 0%
ARIMA models 0 0% 0%
Advanced forecasting models 0 0% 0%
Total 159 100% x

Retail businesses clearly prefer the judgmental method (retailers’ opinions), which is
used in forecasting by almost all respondents (94 %). Nearly half of the respondents
also rely on a naive method or the average of past sales values. With the exception of
the method of customer expectations, other prediction methods are rarely used in retail
stores, and none of the methods of advanced time series analysis was marked by the
respondents.

Table 3 Forecasting methods used depending on the type of retail store

Forecasting method % within affiliation % within sales area (m2)
Independent Retail chain <=50 51-200 200-400 >400
Judgmental method 42% 37% 45% 38% 40% 30%
Customer expectations 9% 9% 13% 5% 7% 7%
Analogy method 6% 1% 6% 4% 0% 0%
Naive method 18% 20% 16% 18% 27% 22%
(Moving) Average 19% 22% 16% 24% 27% 22%
Time series decomposition 0% 1% 0% 0% 0% 4%
Simple regression 1% 4% 0% 5% 0% 4%
Unknown methods 4% 6% 3% 5% 0% 11%
Total 100% 100% 100% 100% 100% 100%

The comparison of the frequency distribution of responses in each group of retailers

(see the results in Table 3) exposes a slightly increased tendency to use qualitative
methods in small shops and stores that are not part of retail chains. Even so the
judgmental method is the most preferred forecasting method in all of the monitored
 Economics and Tourism

groups of retailers. Due to the low number of the respondents, however, it was not
possible to prove any statistical significance of the differences between stores
depending on the affiliation with retail chains (χ2 = 5.661, df = 7, Exact Sig. = 0.600) or
depending on the size of the sales area (χ2 = 20.047, df = 21, Exact Sig. = 0.535).

In demand forecasting, we can consider a number of factors that may affect future
demand. Factors and frequency of consideration thereof in the surveyed retail stores is
part of Table 4.

Table 4 Factors considered in demand forecasting

Factor affecting demand Responses Percent
Frequency Percent of cases
Price of the product 60 13% 90%
Seasonality of sales 47 10% 70%
Freshness of the product 44 10% 66%
Origin of the product 41 9% 61%
Habitual behaviour of customers 40 9% 60%
Quality of the product 38 8% 57%
Promotion in retail trade 37 8% 55%
Ingredients of the product 36 8% 54%
Weather 33 7% 49%
Promotion by the producer 27 6% 40%
Brand of the product 23 5% 34%
Product life cycle 21 5% 31%
Other 1 1% 1%
Total 448 100% x

Virtually all factors examined were mentioned at least by one third of the respondents.
One of the respondents in the research additionally stated cannibalization effects that
were not listed in the menu of factors. On the basis of the analysis, the most often
considered factors may include the price of the product (90 % of the respondents) and
seasonality of sales (70 % of the respondents). More than a half of the respondents also
consider factors closely related to the quality of food products (freshness, origin,
ingredients and quality of the product), and habitual behaviour of purchasers. The
research results confirm good knowledge of consumer behaviour of customers at most
retailers because, in accordance with survey [3], quality, customs, and the price are the
main factors that influence the customer's choice of food, and thus have the greatest
impact on the demand. Although retailers realize that a change of the price and the
seasonal nature of the products significantly affect the demand, they do not use adequate
forecasting methods (simple regression, exponential smoothing, time series
decomposition or other techniques of time series analysis) in the quantification of the
impact of these factors on the demand for food in most cases, but usually only
qualitative estimates in combination with naive and average methods. These simple
statistical estimates, however, are applicable only in forecasting the development of
sales time-series, undistorted by the trend, changing prices of products and seasonal
fluctuations.
SGEM 2015 International Multidisciplinary Scientific Conferences on Social Sciences and Arts

Table 5 compares the differences in the consideration of factors among the surveyed
groups of retailers. Significant differences were detected between independent stores
and stores from retail chains (χ2 = 24.111, df = 12, Exact Sig. = 0.014). When making
forecasts, independent stores often consider the quality parameters of products, while
stores of retail chains put more emphasis on the habitual behaviour of purchasers, sales
seasonality, weather changes and implementation of marketing communication directly
in the retail store or through the manufacturer. The differences depending on the size of
the retail sales area cannot be considered statistically significant (χ2 = 33.497, df = 36,
Exact Sig. = 0.592).

Table 5 Factors considered depending on the type of retail store

Factor affecting demand % within affiliation % within sales area (m2)
Independent Retail chain <=50 51-200 200-400 >400
Price of the product 13% 13% 13% 13% 13% 16%
Quality of the product 10% 7% 10% 9% 7% 4%
Ingredients of the product 12% 4% 12% 7% 4% 2%
Origin of the product 10% 8% 9% 9% 11% 7%
Freshness of the product 12% 8% 12% 9% 7% 9%
Brand of the product 6% 4% 5% 6% 2% 5%
Product life cycle 4% 5% 6% 4% 4% 2%
Habitual behaviour 7% 11% 8% 9% 9% 11%
Weather 6% 9% 6% 8% 11% 7%
Seasonality of sales 9% 12% 10% 9% 13% 12%
Promotion in retail trade 7% 10% 5% 9% 11% 12%
Promotion by the producer 4% 8% 3% 6% 9% 12%
Total 100% 100% 100% 100% 100% 100%

The last area of the research focused on the selection of information sources necessary
for forecasting the demand in retail business. The respondents could mark any number
of the available resources that are used in retail demand forecasting. The ability to
specify a different source, which did not occur in the menu, was not used by any of the
respondents. The frequency distribution of the responses is part of Table 6.

Table 6 Sources of information in forecasting the demand

Source of information Responses Percent
Frequency Percent of cases
Experience of sellers 66 41% 99%
POS data 33 21% 49%
Information from suppliers 29 18% 43%
Information from the media 21 13% 31%
Primary market research 10 6% 15%
Secondary market research 1 1% 1%
Total 160 100% x
 Economics and Tourism

The retail stores consider experienced sellers to be the main source of information about
future demand (99 % of respondents reported this source of information). However,
POS data containing objective information about customer behaviour in retail stores are
only used in a half of the cases. This fact is alarming if only because it is the only
available source of quantitative data on historical sales. The use of the information
mediated by suppliers (43 %) and information from the media (31 %) was identified
relatively often with retailers. Conversely, marketing research, a representative of costly
and time-consuming data sources, is hardly used.

A subsequent analysis of the data revealed that with an increasing sales area, but also in
cases where the store is part of the retail chains, a tendency grows to abandon
qualitative information sources (own sales experience and information from suppliers)
and replace them with relevant data from the point of sale. The observed trend
corresponds with a greater willingness of these stores to use quantitative forecasting
models. As in the first part of the research, however, these differences could not be
proven depending on the affiliation with retail chains (χ2 = 9.278, df = 5, Exact
Sig. = 0.083) or depending on the size of the sales area (χ2 = 14.943, df = 15, Exact
Sig. = 0.479). The comparison of the frequency distribution of responses in each group
of retailers is part of Table 7.

Table 7 Sources of information depending on the type of retail store

Source of information % within affiliation % within sales area (m2)
Independent Retail chain <=50 51-200 200-400 >400
Experience of sellers 43% 40% 45% 42% 37% 33%
POS data 15% 26% 15% 18% 26% 38%
Information from suppliers 25% 11% 23% 18% 16% 8%
Information from the media 10% 16% 10% 16% 16% 13%
Primary market research 8% 5% 8% 5% 5% 4%
Secondary market research 0% 1% 0% 0% 0% 4%
Total 100% 100% 100% 100% 100% 100%

CONCLUSION
In the current turbulent market environment, forecasting the volume of demand and the
future development has of an irreplaceable role in the management of every retail
business. When forecasting demand in the retail food stores, the choice of appropriate
forecasting methods and sources of information play an important role, and so does the
identification of all the factors that could significantly affect future demand of
customers. Due to the nature of demand on consumer markets, as well as the availability
of accurate data on customers’ past sales (POS data), the literature recommends using
mainly quantitative methods based on the time series analysis. In order to assess the
level of demand forecasting in the Czech retail stores, a quantitative research was
conducted that revealed completely different tendencies in their approaches to
forecasting than those recommended by literature. Although a trend to replace
qualitative approaches to forecasting demand with approaches based on time series
analysis of past sales was identified in stores with a larger sales area as well as in stores
 SGEM 2015 International Multidisciplinary Scientific Conferences on Social Sciences and Arts

organized into retail networks, the differences detected cannot be considered significant.
Throughout the investigated sample of stores, there is a prevalent use of qualitative
methods, relying on intuition of retailers and their experience with the sale. Application
of such methods in retail stores with a wide assortment of food products is very
expensive and time-consuming, and therefore cannot be considered effective. An
analysis in crosstabs also revealed the use of inadequate forecasting methods depending
on factors whose impact on the demand is most often considered by the retailers.

The results of the primary study expose a low level of the forecasting process in Czech
retail stores, which may lead to low precision forecasting with all its consequences.
Besides emerging costs associated with an excess or shortage of goods in the retail
logistics chain, it may have unwanted chain-related effects that also deplete other
partners in the supply chain. Further research should therefore be focused on identifying
the causes of the current level of demand forecasting in the retail business, including
specifying the possibility of removing barriers to the implementation of more suitable
approaches to demand forecasting in the surveyed retailers.

REFERENCES
[1] Kotler P., Keller K. L., Marketing Management, 2011, pp 115-124.
[2] Solomon R. M., Consumer Behavior, 2012, pp 206-211.
[3] Hes A., Šálková D., Turčínková J., Tendence chování spotřebitelů při nákupu
potravin, Acta Universitatis Bohemiae Meridionales, The Scientific Journal for
Economics, Management and Trade, vol. 13/issue 2, pp 87-92, 2010.
[4] Theotokis A., Pramatari K., Tsiros M., Effects of Expiration Date-Based Pricing on
Brand Image Perceptions, Journal of Retailing, vol. 88/issue 1, pp 72–87, 2012.
[5] Trusov M., Bodapati A. V., Cooper L. G., Retailer Promotion Planning: Improving
Forecast Accuracy and Interpretability. Journal of Interactive Marketing, vol. 20/issue
3-4, pp 71-81, 2006.
[6] Trepte K., Approaches To Demand Planning Using Point Of Sales Data. Journal of
Business Forecasting, vol. 27/issue 4, pp 22-42, 2008.
[7] Zhang X., The Impact Of Forecasting Methods On The Bullwhip Effect.
International Journal of Production Economics, vol. 88/issue 1, pp 15-27, 2004.
[8] Patak M., Vlckova V., Demand Planning Specifics in Food Industry Enterprises, 7th
International Scientific Conference “Business and Management”, Lithuania, 2012, pp
1168-1175.

View publication stats