Int. Journal of Business Science and Applied Management, Volume 4, Issue 2, 2009
Developing global competitiveness by assessing organized
retail productivity using data envelopment analysis
Reeti Agarwal
Jaipuria Institute of Management
Vineet Khand, Gomti Nagar, Lucknow - 226010, India
Tel: +91-9335366511
Email: reeti@jiml.ac.in
Ankit Mehrotra
Jaipuria Institute of Management
Vineet Khand, Gomti Nagar, Lucknow - 226010, India
Tel: +91-9450643200
Email: ankitmeh@jiml.ac.in
Abstract
The purpose of this paper was to find out (using Regression, Data Envelopment Analysis and
Sensitivity Analysis) how efficiently some of the top organized India retail companies have been
performing relative to each other over the years and thereby to identify factors that help increase the
efficiency of a retail company. The study was conducted based on the analysis of data downloaded
from Prowess database for five Indian retail companies for the time period 2000-2007. The paper is
deemed to be helpful to enable Indian retail companies gain a competitive advantage in the face of
increased competition being faced in the emerging organized retail sector in India. The findings
brought forth Advertising and Marketing expenses as the significant performance determining factors
to be paid attention to.
Keywords: global competitive advantage, organized retail, data envelopment analysis (DEA),
performance determinants, performance indicators
Acknowledgements: Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Pellentesque et magna.
In consectetuer, libero vel vulputate feugiat, pede sem tincidunt elit, a tempus diam augue at tortor.
Donec nec quam ut diam pulvinar tristique.
Int. Journal of Business Science and Applied Management / Business-and-Management.org
2
1 INTRODUCTION
With opportunities come challenges. Retail and real estate are the two booming sectors of India in
the present times. Retail, one of India‟s upcoming industries, has presently emerged as one of the most
dynamic and fast paced industries of recent times with several players entering the market. Accounting
for over 10% of the country‟s GDP and around 8% of employment (Indian Retail Sector An Outlook
2005-2010), retailing in India is gradually inching its way towards becoming the next boom industry.
But, with this growth comes a host of challenges which existing players have to face and overcome to
remain successful in the coming onslaught of heightened competition.
1.1 The Indian retail sector
A shopping revolution is ushering in India where, a large population in the 20-34 age group in the
urban regions is boosting demand. This has resulted in huge international retail investment and a more
liberal FDI policy making India currently the most attractive destination for global retailers with a
GRDI score of 92 and a growth rate of 25 to 30% in the year 2007 (Global Retail Development Index,
2007). Since the time the Narsimha Rao Government kicked off reforms in 1991 and interest rate
deregulation became a reality, the retail sector has been like a toddler waiting to grow big. It has taken
some time but finally it seems that the evolution of organized retailing in India is picking up
momentum.
The world of retail merchandising has come a long way since the days when general stores, that
stocked everything from groceries to stationery, and small shops that sold limited varieties of products,
reigned supreme. There is a movement now from the unorganized to the organized sector. Several
companies are setting up exclusive showrooms and large format stores such as Pantaloon, Shoppers‟
Stop, Westside and several others are expanding. The whole concept of shopping has altered in terms
of format and consumer buying behavior, changing the face of shopping in India. These trends indicate
that retailing, as an industry, has come into its own.
According to the Global Edge report on Market Potential for Emerging Markets (2008), India
ranks eleventh in the list and has been able to maintain itself around this Figure for quite some years
now. Infact, according to Global Retail Development Index (2007), India is positioned as the leading
destination for retail investment topping the chart above Russia and China. Indian organized retail is
growing at a faster pace than was expected and could constitute 25% of the overall retail sector by
2011. According to a study on retail sector prepared by Deloitte Haskins and Sells, organized retail in
India had 8% share of overall retail market (total retail pie) in 2007 in comparison to 5% in the year
2006 and is expected to grow still further in the future.
1.2 Reasons for growth
Favorable demographic and psychographic changes relating to India‟s consumer class,
international exposure, increasing availability of quality retail space, wider availability of products and
brand communication are all bringing forth major opportunities in the organized retail sector in India,
which is poised for an emphatic phase of growth. For a successful retail story what is required is the
proper exploitation of these opportunities.
Over the last few years, many international retailers have entered the Indian market on the strength
of rising affluence levels of the young Indian population along with the heightened awareness of global
brands and international shopping experiences and the increased availability of retail real estate.
Development of India as a sourcing hub shall further make India as an attractive retail opportunity for
global retailers.
PricewaterhouseCoopers in its third edition of Retail & Consumer study, "From Beijing to
Budapest: New Retail & Consumer Growth Patterns in Transitional Economies," assesses growth
opportunities in fourteen countries in Asia, Central and Eastern Europe (CEE) and Russia; it has
determined six countries with "GO" recommendations in terms of investment: China, India, Turkey,
Thailand, Malaysia and Hungary. The study determines that the most immediate opportunities in the
retail and consumer sector lie in China and that India offers more long-term potential for investment in
the sector.
The biggest positive point as far as the sector is concerned is that Indian population is witnessing a
significant demographic transition. A large young working population with median age of 24 years,
nuclear families in urban areas, along with increasing working-women population and emerging
opportunities in the services sector are the key growth drivers of organized retail sector in the country.
The highly fragmented structure of the Indian retail sector is also helping the growth of the sector.
There is a great potential for the organized retail industry to prosper in.
Reeti Agarwal and Ankit Mehrotra
3
India, as a market for final consumption is very large. Many researches show that the total private
consumption market in India is about Rs.15 trillion out of which about Rs 8.5 trillion is towards retail
consumption. Though lucrative opportunities exist across product categories, food and grocery, never-
the-less, presents the most significant potential in the Indian context as consumer spending is highest
on food. While food and grocery represents about 6.5 trillion of retail consumption, clothing comes
second with consumption of about Rs 600 billion (The Indian Retail Report 2005).
The next level of opportunities in terms of product retail expansion lies in categories such as
apparel, jewellery and accessories, consumer durables, catering services and home improvement. These
sectors have already witnessed the emergence of organized formats though more players are expected
to join the bandwagon. Some of the niche categories like books, music and gifts also offer interesting
opportunities for the retail players.
Wholesale trading is another area, which has potential for rapid growth. German giant Metro AG
and South African Shoprite Holdings have already made headway in this segment by setting up stores
selling merchandise on a wholesale basis in Bangalore and Mumbai respectively.
Manufacturers in industries such as FMCG, consumer durables, paints etc are waking up to the
growing clout of retailers as a shift in bargaining power from the former to the latter becomes more
discernible. Already, a number of manufacturers in India, in line with trends in developed markets,
have set up dedicated units to service the retail channel. Also, instead of viewing retailers with
suspicion, or as a 'necessary evil' as was the case earlier, manufacturers are beginning to acknowledge
them as channel members to be partnered with for providing solutions to the end-consumer more
effectively.
Rural Retailing has also being encased into by many companies. Of late, India's large rural
population has caught the eye of retailers looking for new areas of growth. ITC launched the country's
first rural mall 'Chaupal Saga'', offering a diverse product range from FMCG to electronic appliances to
automobiles, attempting to provide farmers a one-stop destination for all their needs. There has been
yet another rural retail initiative by the DCM Sriram Group called the 'Hariyali Bazaar‟ that has
initially started off by providing farm related inputs and services but plans to introduce the complete
shopping basket in due course. Other corporate bodies include Escorts, and Tata Chemicals (with Tata
Kisan Sansar) setting up agri-stores to provide products/services targeted at the farmer in order to tap
the vast rural market.
With IT being the buzzword today how can Electronic Retailing be far behind. Videocon Group
has entered the organized retail sector through an electronic retail chain, „Next‟, under the venture
Emart India. The two other electronic retail chains in the country have a regional or city presence:
Viveks and Vijay Sales. Thus, with the growing popularity of Internet electronic retailing presents a
golden opportunity to retailers.
1.3 Challenges faced by Indian retail
During the last 10 years, many retail start-ups promised a lot. A few folded up even before they
really got started, a few others struggled and then burnt out before they could develop a sustainable
business model and others are still evolving. Pantaloon, Shoppers‟ Stop, Lifestyle, Westside and
Globus are few examples of an Indian success story in retail business.
Despite the bright picture and future prospects that Indian retail presents today, the segment is still
at a nascent stage. It faces hurdles like government regulations, logistics, low margins, vendor‟s
superior negotiating powers and fierce competition from Mom & Pop stores.
Competition from foreign players planning to enter into the country (Walmart for example has
already gained an entry in association with Bharti) represents a major threat to the Indian organized
retail sector. These foreign players have a great deal of experience in this field and their economic
power is also much stronger than that of the Indian players.
In order to achieve success, the retailing industry will also have to counter competition from the
unorganized sector. Traditional retailing is too well established in India to be wiped out. Besides,
traditional retailers have negligible real estate and labor costs and little or no taxes to pay. In contrast,
players in the organized sector have big expenses to meet, and still have to keep prices low to be able to
compete with the traditional sector.
Given the size, and the geographical, cultural and socio-economic diversity of India, there is no
role model for Indian suppliers and retailers to adapt or expand in the Indian context. Also, one must
remember that there is no right retail model. The perfect model is a question of management. The large
scale of consumer diversity, in terms of size, geography, culture and socio-economic background,
would necessitate a varied type of successful models.
Int. Journal of Business Science and Applied Management / Business-and-Management.org
4
There are other issues that are needed to make the retailing industry a force to reckon with. For
example, qualified manpower is required to look after day-to-day operations and cater to the wide
spectrum of customer expectations.
What is required at this stage is for Indian retail companies to understand the factors that have an
affect on the performance of organized retail in India so as to help them develop a strong competitive
advantage which will help them in facing and overcoming the above mentioned challenges. Thus, the
purpose of this paper was to find out the relative efficiency of some of the top retailers of India and
thereby to identify and analyze the factors which have an affect on the performance of organized retail
in India. Indian retail companies can develop global competitive advantage through a proper
understanding of these performance determining factors.
2 LITERATURE REVIEW
Retail productivity is an important issue and vast literature was found on its definitions and
measurements. A review of this literature showed that multiple methodologies have been applied to
assess productivity of individual retail stores, groups of stores, and the retail industry as a whole, but
surprisingly little attention has been given to comparing the efficiency of retail organizations in India.
Understanding and measuring the productivity and efficiency of retailers have been important
issues in retailing research (e.g., Bucklin 1978; Ingene 1982, 1984; Ratchford and Brown 1985;
Ratchford and Stoops 1988). Retail productivity has been considered important for society and for the
individual retail firm (Bucklin, 1978; Ingene, 1984). But, despite a special issue of the Journal of
Retailing in Fall, 1984 and subsequent researches, there is still no single widely accepted definition and
measurement methodology for retail productivity.
Most of the international studies of retail productivity in the 1950s were based heavily on concepts
developed in productivity assessments in the manufacturing sector. The European Productivity Agency
and the National Institute of Economic and Social Research had provided foundation studies of various
industrial sectors and economists drew on these sources (Rostas, 1948). These studies effectively set
the parameters for studies, not only related to manufacturing but also to retailing, for the next 30 years
(Deurinck, 1955). On these foundations, and comparable ones in USA, several studies of retail
productivity were undertaken. While in essence the concepts remain relevant, much has changed over
50 years in respect of both the nature of retail productivity and the factors affecting this productivity
thus requiring new and innovative methods for measuring retail productivity and efficiency.
Past researches have used and suggested the use of various measures and methods to assess retail
efficiency and productivity. Retail productivity is usually measured as ratios of outputs to inputs
(Bucklin, 1978; Ratchford and Brown, 1985; Ratchford and Stoops, 1988). Bloom (1972) defined
productivity as a ratio of output measured in specific units and any input factor also measured in
specific units. A higher ratio of measured output to measured input factors can be directly interpreted
as higher productivity. It can also be seen that the most widely used conceptualization of productivity
has been as the ratio of outputs to inputs; total input productivity is defined as the ratio of all outputs to
all inputs, and partial or single input productivity is the ratio of all outputs to a single input (Ingene,
1982, Lusch and Moon, 1984). The majority of measures of organization efficiency are input-output
ratios, such as sales per square foot or sales per employee (Kamakura, Lenartowiez, and Ratchford
1996). Good (1984) provides a list of possible measures of retail outputs and inputs. Outputs are
usually measured as the number of transactions, physical units sold, value added, and sales. Inputs are
measured as the hours of labor employed, number of employees, wages, salaries and benefits paid, area
of selling place, inventory, and advertising. Thus it can be seen that for the most part measures of
company efficiency have been developed as macro tools, such as those created by the Bureau of Labor
Statistics, and play an important role in assessing how efficiently a particular industry, or economy, is
developing, absorbing technology, or offsetting rising wages. For these purposes, the existing
techniques may be more appropriate. Apart from the industry level studies, understanding is also
required at the individual store level for which, the macro tools are not suitable. Thus, there is a need
for micro tools for use at the individual store level.
Despite its popularity in literature, the output-to-input ratio approach to retail productivity has
several problems. First, retail productivity has been used interchangeably with labor or salesperson
productivity simply because retailing is often a labor-intensive activity (Bush, Bush, Ortinau, and Hair,
1990; Ingene, 1982, 1984; Stem and El-Ansary, 1992; Thurik and Wijst, 1984), even though there is a
large non-sales portion of labor force in retail industries. As a result, retail productivity has sometimes
been treated as an issue of sales management. Focusing on an individual salesperson does not directly
meet the measurement criteria of retail productivity because labor is simply one of the input factors
(Good, 1984).
Reeti Agarwal and Ankit Mehrotra
5
Second, traditional retail productivity studies have often focused on too micro units of analysis
(e.g., salesperson evaluation; Bush, Bush, Ortinau, and Hair, 1990) or too macro units of analysis (e.g.,
retail industries or aggregation of stores; Goldman, 1992; Pilling, Henson, and Yoo, 1995). Previous
research has ignored retail productivity with respect to individual stores and has not applied macro
techniques to any extent as a managerial tool. Measuring productivity of individual stores would make
the evaluation and control of managerial activities more feasible and objective. Thus, retail managers
need such store level productivity measurement tools.
Third, most previous measures have been absolute measures of productivity. These indexes are
calculated by inserting numbers into the predetermined formulas or ratios. They do not take into
account the performance of other retail organizations or other environmental circumstances. The
productivity measurement of an individual retail organization should be "relative" and incorporate the
performances of other similar organizations.
Thus, literature related to retail productivity clearly shows that though simple to define,
assessments of retail productivity based on simple ratios of outputs to inputs have been criticized for
the following reasons: improper measurement of output (Achabal et al., 1984; Parsons, 1994; Oi,
1992); failure to account for changes in the quality of inputs or outputs over time or across stores
(Doutt, 1984; Good, 1984; Lusch and Moon, 1984; Nooteboom, 1985; Oi, 1992); failure to account for
the consumer's input to the process (Ingene, 1984; Oi, 1992); improper weighting of multiple inputs
and outputs (Parsons, 1990); inability to separate differences in productivity from scale effects
(Ratchford and Brown, 1985). In addition to these limitations, the traditional "ratio" approach to retail
productivity presents other problems when the focus is evaluation of different retailers. These retail
companies are typically located in different markets and serve a diverse population of customers,
leading to distinct operational characteristics at each organization. These differences are not taken into
account by traditional productivity indices, leading to a biased assessment of the relative efficiency of
different retail organizations.
Thus, what is required is a new approach to retail productivity measurement that focuses on one
organization relative to the best performers rather than the average performers as done in the traditional
absolute measures. There are two major advantages of relative-to-best measures. First, in contrast to
relative-to-average measures, relative-to-best measures are consistent with quality control movements
such as benchmarking. The best performing units need to be used as role models or the bases for
evaluation (Farrell, 1957). Second, in contrast to absolute measures, relative-to-best measures show
contingent productivity, which takes into account performances of other comparable units and
environmental factors. The absolute measures tend to focus only on controllable input factors such as
labor and capital (Banker and Morley, 1986).
Finally, previous techniques of retail productivity such as cost function and total factor
productivity indexes have a few drawbacks. Regression in the form of a cost function imposes a
particular functional form and total factor productivity refers to the measurement of efficiency of all
employed inputs (Bucklin, 1978), and relates net output to the associated total factor input; that is, to
the input of both labor and capital (Bloom, 1972). The weights employed in calculating indexes for
total factor productivity (weighted sums of outputs divided by weighted sums of inputs) are often
subjective.
Consequently, in order to assess the productivity of organizations of a retail firm there is a need to
develop an output-to-input ratio system which can handle multiple inputs and outputs in order to go
beyond basic labor or capital productivity measurement. Ideally such a system would measure relative-
to-best productivity or efficiency, as opposed to absolute or relative- to-average values, and resolve
problems in traditional measurement techniques (such as cost functions and total factor productivity
discussed above).
In view of the changing scenario of the Indian Retail Industry, the scarcity of studies on the
assessment of different retail organizations is not compatible with the importance of the topic. With so
many opportunities as well as challenges facing the Indian organized retail sector, the organized retail
companies of India need to develop global competitive advantage and become efficient in their
operations. Thus, given the lack of studies undertaken in this area in the Indian scenario, this study was
undertaken to gain an insight into the relative efficiency of different retail companies in India and to
identify ways to increase the efficiency of inefficient companies. In order to overcome the
shortcomings of the techniques previously used to asses productivity, Data Envelopment Analysis
technique has been used to asses the relative efficiency and productivity of some of the top retailers of
India. The study identifies and analyses the importance of performance determining factors in
improving the efficiency of a retail company.
Int. Journal of Business Science and Applied Management / Business-and-Management.org
6
3 OBJECTIVES OF THE PRESENT STUDY
The present study was undertaken to understand the factors affecting the performance of organized
retail in India so as to better understand ways to help companies develop global competitive advantage
in the retail sector. In particular, the study focused on:
1. Identifying the factors that have an affect on the performance of organized retail in India.
2. Analysis of the affect of these performance determining factors on the performance
indicating factors
3. Identifying the more significant performance determining factors
4. Analyzing the relative efficiency of some of the top organized retail companies of India.
5. Comparing the inefficient retail companies with the efficient ones in order to identify the
areas where improvement is required to help companies increase their efficiency.
4 METHODOLOGY OF RESEARCH
4.1 Data collection method & Justification of secondary source
The data used in this paper was collected from secondary sources. Data was obtained for 5 retail
companies of India for the time period 2000 to 2007. The source of data was Prowess Database.
Prowess is a database of large and medium Indian firms containing detailed information on over 20,000
firms. These comprise all companies traded on India's major stock exchanges and several others,
including, the central public sector enterprises. The database covers most of the organized industrial
activities such as banking, retailing, airlines and other service and manufacturing sectors of India.
Prowess provides detailed information on each company including a normalized database of the
financials covering 1,500 data items and ratios per company. Besides, it provides quantitative
information on production, sales, consumption of raw material and energy etc. As Prowess database
has found useful applications in places where trust and reliability matter the most, Prowess became the
preferred source of data in respect of the variables identified for the present study.
4.2 Selection of Variables
On the basis of literature studied, data was gathered in respect of 12 variables out of which 9 were
taken as performance determinants and 3 as performance indicators. The performance determinants
included Advertising Expenses, Marketing Expenses, Capital employed, Current Assets, Gross Fixed
Assets, Inventories, Power and Fuel Expenses, Salaries and Wages and Working Capital, while the
performance indicators included Sales, PBIT and Return on Capital Employed. The different variables
considered for the study have been tabled in Figure 1.
Figure 1: Conceptual Input Output framework
Power and fuel expenses
Salaries and wages
Advertising expenses
Marketing expenses
Gross fixed assets
Inventories
Current assets
Working capital
Inputs/Independent Variables
Retail Organization
Sales
PBIT
Return on capital
employed
Outputs/Dependent Variables
Reeti Agarwal and Ankit Mehrotra
7
4.3 Method of analysis
Data was analyzed using two different techniques, Regression Analysis and DEA model. For
Regression analysis, the nine performance determining factors were the independent variables while
the three performance indicating factors were taken as the dependent variables. In the DEA Model, the
performance determinants were used as the Input variables while the performance indicators were used
as the Output variables.
4.4 Justification for using DEA method of analysis
Efficiency is usually measured as ratios of outputs to inputs. A higher ratio of measured output to
measured input factors can be directly interpreted as higher efficiency. There are a number of
methodologies which can be used for evaluation of efficiency of a unit such as, output-to-input ratio
approach, regression, cost function, total factor productivity indexes and many others. DEA was chosen
as the primary technique for efficiency evaluation since it was seen that though DEA works on the
same concept as the traditional techniques of measurement, it covers lots of other aspects which the
traditional techniques lack. DEA also has certain drawbacks but its advantages overshadow its
disadvantages. The major advantages of DEA based method of efficiency evaluation includes
utilization of both output and input observations, accommodation of multiple inputs and outputs,
accommodation of both controllable and uncontrollable factors, computation of a single index of
productivity, development of a relative measure of performance for each retail outlet using best
performers as the bases, and non-imposition of any functional form on the data. Moreover, unlike total
factor productivity indexes, DEA gives each of the observations its own set of weights which make the
analysis more appropriate.
5 RESEARCH FINDINGS AND ANALYSIS
5.1 Affect of the performance determining factors of organized retail on performance
indicators using Regression Analysis
5.1.1 Affect of performance determining factors on Sales
The value of Adjusted R
2
was found to be .991 which shows that the model is a good fit. The
significance of the F-value came out to be .000 which indicates that the model is statistically significant
at 5% level of significance. In order to adjudge whether there exists multi-collinearity between the
independent variables, Durbin Watson test was administered along with regression. The value of the
Durbin-Watson test came out to be 1.629 which indicated that auto correlation was not present in the
data. Considering the correlation coefficients among predictors, it was deduced that they were not
related so data was free from multi collinearity. The Beta values and the significance levels of t-tests
for significance of individual independent variables are given in Table 1.
Table 1: Regression Analysis with Sales as dependent variable
Model
Unstandardized
Coefficients
t
Sig.
B
Std. Error
B
Std. Error
1
(Constant)
23.481
20.021
1.173
.253
Advertising Expenses
-7.243
2.844
-2.547
.018
Capital employed
.018
.185
.098
.923
Current Assets
.224
.503
.446
.660
Gross fixed assets
-.569
.594
-.957
.349
Inventories
1.110
.166
6.697
.000
Marketing expenses
2.572
1.730
1.486
.151
Power and fuel expenses
54.275
11.873
4.571
.000
Salaries and wages
-1.170
3.493
-.335
.741
Working capital
-.520
.436
-1.194
.245
Dependent Variable: Sales
Int. Journal of Business Science and Applied Management / Business-and-Management.org
8
As can be seen from Table 1, only 3 of the independent variables were found to be statistically
significant in the model at 5% significance level. These include - Advertising Expenses, Inventories
and Power & Fuel Expenses. Looking at the Beta values for all these variables, it could be seen that
Advertising Expenses was negatively related to the dependent variable i.e. Sales while the other 2
variables i.e. Inventories and Power & Fuel Expenses were both positively related to the dependent
variable. Looking at the Beta values, it can be said that in absolute terms Power & Fuel Expenses with
a Beta value of 1.069 had the maximum effect on Sales while Advertising Expenses with a Beta value
of -.203 had the least effect.
The estimated increase in sales for every unit increase or decrease in these variables is given by
the standardized Beta values of these variables. Since the Advertising Expenses were negatively related
to sales, it indicated that if advertising expenses are decreased by one unit, sales will increase by .203,
if all the other variables remain unchanged. The positive effect of Inventories and Power & Fuel
Expenses on Sales denotes that for every one unit increase in Inventories, Sales will increase by .337
other variables remaining constant and for every one unit increase in Power & Fuel Expenses, Sales
will increase by 1.069, if all other variables are unchanged.
5.1.2 Affect of performance determining factors on PBIT
The value of Adjusted R
2
was found to be .934 which shows that the model is a good fit. The
significance of the F-value came out to be .000 which indicates that the model is statistically significant
at 5% level of significance. The value of the Durbin-Watson test came out to be 1.267 showing that
auto correlation was not present in the data. Considering the correlation coefficients among predictors,
it can be said that they were not related so data was free from multi collinearity. The Beta values and
the significance levels of t-tests for significance of individual independent variables are given in Table
2.
Table 2: Regression Analysis with PBIT as dependent variable
Model
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
B
Std. Error
1
(Constant)
-10.749
8.005
-1.343
.193
Advertising Expenses
-5.500
1.137
-1.032
-4.837
.000
Capital employed
.020
.074
.055
.273
.788
Current Assets
.120
.201
.323
.598
.556
Gross fixed assets
-.317
.238
-.491
-1.333
.196
Inventories
.218
.066
.443
3.285
.003
Marketing expenses
5.479
.692
.690
7.920
.000
Power and fuel expenses
9.555
4.747
1.262
2.013
.057
Salaries and wages
-.589
1.397
-.219
-.422
.677
Working capital
.047
.174
.053
.267
.792
Dependent Variable: PBIT
As can be seen from Table 2, only 4 of the independent variables were found to be statistically
significant in the model at 5% significance level. These include - Advertising Expenses, Inventories,
Marketing Expenses and Power & Fuel Expenses. Looking at the Beta Values for all these variables, it
could be seen that Advertising Expenses was negatively related to the dependent variable i.e. PBIT
while the other 3 variables i.e. Inventories, Marketing Expenses and Power & Fuel Expenses were
positively related to the dependent variable. Looking at the Beta values it could be said that in absolute
terms Power & Fuel Expenses with a Beta value of 1.262 had the maximum effect on PBIT while
Inventories with a Beta value of .443 had the least effect on PBIT.
The negative effect of Advertising Expenses on PBIT clearly shows that an increase in
Advertising Expenses decreases PBIT and vice versa. Thus, every one unit decrease/increase in
Advertising Expenses will lead to a 1.032 increase/decrease in PBIT, other variables remaining
unchanged. The positive effect of Inventories, Marketing Expenses and Power & Fuel Expenses on
PBIT indicates, that for every one unit increase in Inventories, Marketing Expenses and Power & Fuel
Expenses, PBIT will increase by .443, .690 and 1.262 respectively, if the other variables remain
constant.
Reeti Agarwal and Ankit Mehrotra
9
5.1.3 Affect of performance determining factors on Return on Capital Employed
The value of Adjusted R
2
was found to be .748 which shows that the model is a good fit. The
significance of the F-value came out to be .000 which indicates that the model is statistically significant
at 5% level of significance. Existence of multi-collinearity between the independent variables was seen
by administering Durbin Watson test along with regression. The value of the Durbin-Watson test came
out to be 2.578 which showed that auto correlation was not present in the data. Considering the
correlation coefficients among predictors, it was deduced that they were not related so data was free
from multi collinearity. The Beta values and the significance levels of t-tests for significance of
individual independent variables are given in Table 3.
Table 3: Regression Analysis with Return on Capital Employed as dependent variable
Model
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
B
Std. Error
1
(Constant)
-22.488
11.973
-1.878
.074
Advertising Expenses
-5.206
1.701
-1.273
-3.061
.006
Capital employed
.021
.111
.074
.188
.853
Current Assets
.280
.301
.977
.930
.362
Gross fixed assets
.210
.355
.426
.592
.560
Inventories
-.043
.099
-.115
-.436
.667
Marketing expenses
5.816
1.035
.955
5.620
.000
Power and fuel expenses
4.862
7.100
.837
.685
.501
Salaries and wages
-1.733
2.089
-.839
-.829
.416
Working capital
-.171
.261
-.254
-.655
.519
Dependent Variable: Return on Capital Employed
As can be seen from Table 3, only 2 of the independent variables were statistically significant in
the model at 5% significance level. These include - Advertising Expenses and Marketing Expenses.
Looking at the Beta Values for these 2 variables, it was seen that Advertising Expenses was negatively
related to the dependent variable i.e. Return on Capital Employed while Marketing Expenses was
positively related to the dependent variable. Looking at the Beta values it could be said that in absolute
terms Advertising Expenses with a Beta Value of -1.273 had a more significant effect on the dependent
variable than Marketing Expenses.
Looking at the standardized Beta values of the 2 significant variables, it becomes clear that an
increase/decrease in Advertising Expenses leads to a decrease/increase in Return on Capital Employed,
because of the negative relation of Advertising Expenses with Return on Capital Employed, while an
increase/decrease in Marketing Expenses leads to an increase/decrease in Return on Capital Employed,
because of the positive effect of the former on the latter. Thus, for every one unit decrease/increase in
Advertising Expenses, the Return on Capital Employed will increase/decrease by 1.273 while for every
one unit increase/decrease in Marketing Expenses, Return on Capital Employed will increase/decrease
by .955.
5.2 Comparison of Retail Productivity using Data Envelopment Analysis (DEA)
In order to measure and evaluate the efficiency of some of the top retail organizations of India,
data related to five retail organizations was obtained from a well known financial software Prowess,
for a period of eight years starting from year 2000 and ending at 2007. The five retail organizations
were coded as 1, 2, 3, 4, and 5 respectively in the following analysis. Appropriateness of company and
data to DEA has been examined in this study in terms of many assumptions which were cited by Dyson
et al (2001). One of them was homogeneity assumptions relating to the homogeneity of units under
assessment. In general the units were understood to be similar in a number of ways. Retail
Organizations in this study offer similar product categories by driving similar inputs. The second
assumption according to Dyson et al. (2001) was about the input/output set. The study satisfied the
second assumption because all retail organizations were evaluated on the same input and output
parameters. The sets of factors were common to all organizations. The last assumption named as factor
measurement was on the measurement scales of inputs and outputs. According to it, they should
conform to ratio scales. The present study also supported the last assumption.
Int. Journal of Business Science and Applied Management / Business-and-Management.org
10
Since the efficiencies of various organizations were measured by DEA model, it was necessary to
solve the model three, four or five times depending on the data available for the five different
organizations under study for different years. Productivity or efficiency in the context of DEA dealt
with producing the maximum quantity of outputs for any given amount of inputs or the minimum use
of inputs for any given amount of outputs. The first task of DEA was to find the most efficient retail
organization, which produced a so-called efficient frontier, which is a series of points, a line, or a
surface connecting the most efficient retail organizations, which were determined from a comparison of
inputs and outputs of all retail organizations under consideration. Thus, DEA produced the relative
efficiency boundaries, which are called envelopes.
Retail organizations lying on the efficient frontier were given the arbitrary efficiency score of one.
In other words, any unit or organization whose efficiency score equaled one was defined as “efficient”,
otherwise “inefficient” (Bal and Örkcü, 2005). In other words, efficiency is the ratio of the weighted
sum of outputs to the weighted sum of inputs. In the present study, the different retail organizations
used 9 input variables as mentioned earlier and 3 output variables. Thus, for an organization to be
efficient:
A
1
Y
1
+ A
2
Y
2
+A
3
Y
3
E
1
= --------------------------------------------------------------------------------- <= 1
B
1
X
1
+ B
2
X
2
+ B
3
X
3
+ B
4
X
4
+ B
5
X
5
+ B
6
X
6
+ B
7
X
7
+ B
8
X
8
+ B
9
X
9
Where,
If E
i
< 1 organization is inefficient.
If E
i
= 1 organization is efficient.
and
E = efficiency of a retail organization
Y = outputs used in the DEA model
X = inputs used in the DEA model
A = weights DEA estimates for the outputs
B = weights DEA estimates for the inputs
The model was run for each organization by utilizing Solver bundled with Microsoft Excel. The
results of the analysis are discussed under headings of Efficient and Inefficient retail organizations
while areas of improvement for inefficient retail organizations were identified using Sensitivity/Gap
analysis.
5.2.1 Efficient and inefficient Retail Organizations
The results obtained from data entered in the DEA model are tabulated in Table 4. It can be seen
from this table that companies 1, 2, 3, and 5 were found to be running efficiently with company 1
showing consistency in efficiency across all the years studied. Organization 4 secured efficiency score
less than 1 in the years 2005 and 2006 showing that it was relatively inefficient in these years in
comparison to the other companies.
Table 4: Efficiency scores for companies in different years
Companies
Year
2000
2001
2002
2003
2004
2005
2006
2007
1
1
1
1
1
1
1
1
1
2
1
1
1
1
1
1
1
1
3
NA
NA
1
1
1
1
1
NA
4
1
1
1
1
1
0.463
0.616
1
5
NA
NA
NA
NA
NA
NA
1
1
Reeti Agarwal and Ankit Mehrotra
11
Figure 2: Snapshot of DEA model for an efficient retail organization
In using DEA, the weights were estimated separately for each retail organization such that its
efficiency was the maximum attainable. As can be seen in Figure 2, DEA estimated the weights 0.001,
0.043, 0.001, 0.951, 0.001, 0.001, 0.001, 0.001 and 0.001 for the input variables and 0.010, 0.000, and
0.058 for the output variables for retail organization 5 for the year 2006. DEA estimated the weights
such that the estimated efficiency of retail organization 5 (E
5
) was the maximum possible. However,
the weights estimated for retail organization 5 were such that when they were applied to the inputs (X
s
)
and outputs (Y
s
) of all other units in the analysis their ratio of weighted outputs to weighted inputs was
less than or equal to 1. Similarly, DEA estimated a separate set of weights for each retail organization
such that the estimated weights led to a maximum attainable efficiency for that organization. As seen
from Figure 2, DEA optimized on each individual retail organization‟s performance in relation to the
performance of all other retail organizations. While using DEA, the estimated weights were constrained
so that no one input or output variable dominated the efficiency estimation. Minimum limits were also
set for the estimated weights so that all inputs and outputs were forced to play a role in efficiency
computation. The efficiency computed by DEA assumed that 100% efficiency is attained for an
organization only when (1) none of the outputs can be increased without either increasing one or more
inputs or decreasing some of its other outputs and (2) none of the inputs can be decreased without
decreasing some of its outputs or increasing some of its other inputs. Hence, 100% efficiency is defined
to have been attained by a retail organization only when comparisons with other organizations do not
provide evidence of inefficiency in the use of any inputs and in creation of any outputs.
5.2.2 Sensitivity/Gap analysis for inefficient Retail Organizations
At the individual retail organization level, DEA also provided rich diagnostic information through
sensitivity analysis. For every retail organization not on the efficient frontier, DEA identified a set of
efficient reference organizations in the corresponding envelope. These efficient reference organizations
(whose efficiency is 100%) helped in identifying the inadequacies or slacks in the controllable
inputs/outputs of the inefficient organization. By comparing the controllable inputs and outputs of the
inefficient organization with the controllable inputs and outputs of a linear combination of the efficient
reference organizations that comprised the frontier (a virtual organization), the amount of slack in each
of the variables was computed. This can help the inefficient organization identify how to allocate
resources more efficiently and improve its productivity.
An inefficient organization may become efficient by increasing all outputs by an amount equal to
its corresponding slack (i.e., move towards the efficient frontier vertically in the case of a 2-
dimensional plot) or by decreasing all controllable inputs by amounts equal to its corresponding slacks
(i.e., move towards the efficient frontier horizontally in the case of a 2- dimensional plot).
Table 5: Sensitivity analysis for retail organization 4 for the year 2005
Int. Journal of Business Science and Applied Management / Business-and-Management.org
12
(Units Rs. Crore)
Inputs
Estimated
Weights
Value
Measured
Value If
Efficient
Improvement
Scope/Slack
Power and fuel expenses
0.001
6.8
4.489
-2.311
Salaries and wages
0.021
14.86
12.205
-2.655
Advertising expenses
0.001
21.17
4.032
-17.138
Marketing expenses
0.001
13.68
2.700
-10.980
Capital employed
0.001
216.97
56.832
-160.138
Gross fixed assets
0.001
80.96
65.724
-15.236
Inventories
0.001
37.63
34.199
-3.431
Current assets
0.001
183.15
62.786
-120.364
Working capital
0.001
123.62
-1.337
-124.957
Outputs
Sales
0.002
231.49
231.490
0.000
PBIT
0.001
25.72
25.720
0.000
Return on capital employed
0.000
9.49
25.982
16.492
Table 6: Sensitivity analysis for retail organization 4 for the year 2006
(Units Rs. Crore)
Inputs
Estimated
Weights
Value
Measured
Value If
Efficient
Improvement
Scope/Slack
Power and fuel expenses
0.001
9.69
7.178
-2.512
Salaries and wages
0.009
20.61
18.993
-1.617
Advertising expenses
0.001
29.85
7.532
-22.318
Marketing expenses
0.001
19.07
3.337
-15.733
Capital employed
0.001
335.32
137.685
-197.635
Gross fixed assets
0.001
97.24
82.077
-15.163
Inventories
0.001
53.36
53.360
0.000
Current assets
0.001
188.3
126.433
-61.867
Working capital
0.001
82.28
58.833
-23.447
Outputs
Sales
0.002
343.23
343.230
0.000
PBIT
0.002
38
38.000
0.000
Return on capital employed
0.000
12.74
17.174
4.434
Table 5 and 6 show the gap calculated for various inputs of the inefficient organization by
comparing them with the combined weighted inputs of all the efficient organizations for year 2005 and
year 2006 respectively. Table 5 shows the sensitivity analysis results for retail organization 4 for the
year 2005 while Table 6 has the sensitivity analysis results for retail organization 4 for the year 2006.
These tables show the amount of slack in each of the controllable input and output observations for this
retail organization. This slack was computed by comparing the input and output of retail organization 4
with the inputs and outputs of its efficient reference organizations. These efficient reference
organizations were organizations which operate under circumstances similar to that of organization 4,
but have 100% efficiency. The results show that retail organization 4 could have become efficient
(increased efficiency from 0.463 to 1.00 in year 2005 and from 0.616 to 1.00 in year 2006) by
increasing all outputs by the corresponding slack amounts or decreasing all controllable inputs by
corresponding slacks. Retail organization 4's estimated weights for the 12 variables are also shown in
Table 5 and Table 6 for the year 2005 and 2006 respectively. DEA estimated these weights such that
the estimated efficiency of 0.463 and 0.616 for retail organization 4 is the maximum attainable. No
other combination of weights would have produced a higher efficiency estimate for retail organization
4 and yet satisfied all of the constraints in the optimization.