Professional Documents
Culture Documents
Sinta Wardani
Ujang Sumarwan
Tb. Nur Ahmad Maulana
Institute of Agriculture Bogor
E-mail: sinta_wardani@yahoo.com; usumarwan@gmail.com; amet2001uk@yahoo.co.uk
MB-IPB Building, Pajajaran Street, Bogor 16151, West Java, Indonesia
ABSTRACT
The purposes of this study are (1) to analyze the effect of the promotion cost and distribution
cost on company’s sales partially and simultaneously (2) to formulate managerial implica-
tions related to the results of this study. Data was collected from 29 companies which sepa-
rated into 18 foreign direct investments companies and 11 domestic investment’s companies.
Secondary data was gathered through consolidated financial statements’ companies which
listed on IDX for the 2007-2012 years ended. Data panel analysis fixed effect model was used
to estimate the effects of promotion cost and distribution cost. Based on the FEM analysis, all
independent variables of foreign direct investment and domestic investment companies has
significant effect on their sales. The results of this study suggested for managerial implica-
tions to increase the efficiency of promotion cost and distribution cost.
Key words:Promotion Cost, Distribution Cost, Companies’ Sales, Fixed Effect Models.
Kata Kunci: Promotion Cost, Distribution Cost, Companies’ Sales, Fixed Effect Models.
431
ISSN 2087-3735 The Effect of Promotion … (Sinta Wardani)
432
Journal of Economics, Business, and Accountancy Ventura Volume 16, No. 3, December 2013, pages 431 – 442
Accreditation No. 80/DIKTI/Kep/2012
Table 1
Value of Print Media Ad Spending in the Consumer Goods Industry Subsectors over the
Last 6 Years (Per Million Dollars)
Category 2007 2008 2009 2010 2011 2012
Food 95.905 108.463 134.805 171.376 181.332 190.010
Beverages 277.135 350.900 397.531 497.567 478.970 504.993
Smoking & Accessories 177.247 184.984 176.033 202.985 228.681 197.699
Medicines/Pharmaceuti 293.807 325.854 552.315 744.468 970.525 840.384
cals
Toiletries & Cosmetics 252.456 289.744 354.365 322.317 360.565 361.642
Household Equipment 614.692 718.825 691.546 983.687 1.185.002 1.458.942
& Appliances
Source: Nielsen Indonesia.
There have been many studies that seek the significant influence.
effect of promotion and distribution costs on The next is the study by Simester, Hu,
total sales. This studies conducted in Brynjolfsson, and Anderson (2005) found
Indonesia are mostly just case studies that that advertising affects sales in the future,
examines and rarely an industry based on but the sign a different effect on each of the
secondary data. Has done a lot of research targeted consumers. At the company's
on various consumer goods industry in customers, increase in advertising it is
various countries but to date the authors negative in the long run it while on the
have not found a specific study divides the ordinary consumer is positive. Indrawati
industry by foreign investment companies (2000) suggested that research results either
and investment companies in the country. partially or simultaneously, the variable cost
Similarly, various studies have conflicting of sales, cost of transportation and
results although the independent and warehousing costs a positive effect on sales
dependent variables under those studies have volume. A result of the same study was also
similarities. reported by Raya (2009) that there is a very
In a study by Abdel - Khalik (1975) in strong relationship, a positive and on the
Banerjee, Siddhanta and Bandopadhyay same direction of relationship between the
(2012), found that the ads have long-term cost of distribution and sales.
impact on sales in the categories of The result of a similar study was also
industries of food, pharmaceuticals, and found by Mukodim (2007) that the costs of
cosmetics. They have short-term impact on promotion and distribution are very
the automotive categories, cigarette, soaps influential on sales. Yet, only the cost of the
and cleaners. Abiodun (2011) attempted to promotion that has a higher effect on the
find out the impact of advertising on sales cost of distribution. Similarly, in a study
and profits of a business organization conducted Hernomo (2012), it was found
Starcomms Plc. The findings revealed that a that the cost of promotion and distribution
strong advertising position in the minds of simultaneously can affect the volume of
consumers having influence in order to sales promotion but only the cost has
encourage repeat purchase products, so significant results.
competitors will not have the advantage over Based on the evidences above, this study
them and create brand loyalty and product has attempts as the following:
differentiation. However, the research by 1. To analyze the effect of the cost of
Tellis and Weiss (1995) about the influence promotion on the company's sales value in
of television advertising on sales, provides the foreign investment and domestic
evidence that the discovery has no a investment companies.
433
ISSN 2087-3735 The Effect of Promotion … (Sinta Wardani)
2. To analyze the effect of distribution costs constant across time and between
on sales of foreign investment and domestic sections.
investment companies. 2. Fixed effects model is a model that can
3. To analyze simultaneously the effect of be used by considering that the omitted
promotion and distribution costs on the variables can lead to changes in the
value of the company's sales of foreign intercepts between sections and between
investment and domestic investment. times? To enable these changes can be
added intercept dummy variables into
RESEARCH METHOD the model.
This study took public companies as the 3. Random effects model. In the fixed
objects included in the category of foreign effect model, the decision to include
investment enterprises and domestic dummy variables into the model will
investment in the consumer goods industry result in a reduced number of degrees of
and listed on the Stock Exchange prior to the freedom; to overcome this we used a
period of 2006. The number of samples random effects model.
included in the criteria for the study was 29
companies consisting of 18 foreign Selection of the Models
investment companies and 11 investment It is important in this process to obtain an
companies in the country. efficient model and alleged most excellent
These data are secondary data from the among a wide selection of models that need
company's annual report for six years i.e. to be analyzed based on statistical
2007-2012. Secondary data is data collected considerations (Gujarati 2006). There are
by others for different purposes with the several statistical tests used in the panel data
objectives of the research are formulated to determine which model is best for them
(Sumarwan, Prihartono, Sumarlin, Mamahit, are chosen:
Purnomohadi, Hasan, Ahmady, Wulandari, 1. Chow test is a statistical test which aims
Haryono 2011). The operational definitions to choose whether it is better to use the
of variables studied include as the following. PLS models or FEM.
1. Promotion costs (X1) which consists of 2. Hausmann test. In choosing the model
the cost of sales promotion and use FEM or REM models Hausmann test
advertising costs in rupiah. method can be used as a basis for testing
2. Distribution costs (X2) consists of or not statistical considerations.
limited to the elements of cost of sales 3. Langrange Multiplier test is used as a
cost of packaging and shipping and basic consideration in choosing statistic
warehousing costs in rupiah. REM and PLS models.
3. Sales of the company (Y) is the total
amount generated from the sale of goods Testing the Models
that signaled the rise and fall of sales in In testing the models, each model is
the form of units, pounds, tons or liters analyzed to test the hypothesis. The model
and is expressed in units in rupiah. used in this study is based on a review of the
influence of the theory of the cost of
Methods and Data Analysis promotion and distribution towards the value
To estimate the parameters of the model of the company's sales. The variables in this
with panel data, there are several techniques study are different scales of values between
that can be used (Gujarati 2006): the minimum value and the maximum value.
1. Pooled least square is an estimation The transformation in the form of Ln can
model of panel data regression model reduce the problem heterocedastisity. This is
with the simplest assumption that the because the transformation for the scale of
intercept and the slope coefficient is the measurement of variables, by reducing
434
Journal of Economics, Business, and Accountancy Ventura Volume 16, No. 3, December 2013, pages 431 – 442
Accreditation No. 80/DIKTI/Kep/2012
435
ISSN 2087-3735 The Effect of Promotion … (Sinta Wardani)
Table 2
Estimation Results Using PLS, FEM, and REM Method for Foreign Direct Investment
Company
Pooled Least Fixed Effect Random Effect
Square Model Model
Coefficient 9.872746 10.09198
LN_PROMOSI -0.069465 0.264370 0.251806
LN_DISTRIBUSI 1.201631 0.469167 0.472793
Prob
LN_PROMOSI 0.5012 0.0000* 0.0000
LN_DISTRIBUSI 0.0000 0.0000* 0.0000
R-squared 0.535745 0.991973 0.653437
Adjusted R-squared 0.531365 0.990239 0.646836
F-statistic 572.3405 98.98775
Prob(F-statistic) 0.000000 0.000000
Durbin-Watson stat 0.055632 1.308889 1.087780
Note: The independent variable on the dependent variable proved significant in the real level 0.05.
Table 3
Results of Chow and Hausman Tests
Chow Test Hausman Test
Prob. Cross-section F 0.0000
Cross-section Chi-square 0.0000 0.5571
d.f. Cross-section F (17.88)
Cross-section Chi-square 17 2
Chi-Sq. Statistic 1.1701
436
Journal of Economics, Business, and Accountancy Ventura Volume 16, No. 3, December 2013, pages 431 – 442
Accreditation No. 80/DIKTI/Kep/2012
Table 4
Estimation Results Using PLS Method, FEM and REM for Domestic Investment
Companies in the Consumer Goods Industry
Pooled Least Fixed Effect Random Effect
Square Model Model
Coefficient 4.573494 4.492853
LN_PROMOSI 0.223577 0.173607 0.174397
LN_DISTRIBUSI 0.907818 0.775353 0.777790
Prob
LN_PROMOSI 0.0014 0.0004* 0.00020
LN_DISTRIBUSI 0.0000 0.0000* 0.00007
R-squared 0.831212 0.996051 0.890233
Adjusted R-squared 0.828575 0.995157 0.886749
F-statistic 1114.002 255.4724
Prob(F-statistic) 0.000000 0.000000
Durbin-Watson stat 0.044551 1.422101 1.239164
Note: The independent variable on the dependent variable proved significant in the real level of 0.05.
95 percent confidence level. Yet, the results approach, the F - stat test is done to see if the
of the Hausman test with prob 0.5571 independent variables are simultaneously
greater value when compared to the significantly affecting the dependent
significance level of 5 percent. variable. As in Table 2, it is the value of
The results suggest that REM is better Prob (F - stat) FEM generated is 0.0000,
than FEM with a 95 percent confidence with a confidence level of 95 percent
level. However, with the R - squared value significant P-value (0.0000 < 0.05), which
of 99.1973 percent, FEM felt better in this means it does not reject Ha3. Thus, it can be
study because it is able to predict the FEM said that the 95 percent of significance level
model fit to almost 100 percent. Also in the for independent variable such as promotion
FEM, this can determine the value of the costs and distribution cost simultaneously
constant is different for each company that have significant effect on the value of the
describes the different characteristics of each company's sales.
company.
From the results of Table 2, the obtained Domestic Investment Company
regression equation for Foreign Investment The next is the analysis of the company's
Company is as follows. investment in the country by testing pooled
Y = 9.872746 + 0.264370 + Promotion Cost least squares, fixed effect model, and
+ 0.469167 Distribution Cost. random effect model. Table 4 shows the
From the original form of regression results of calculations based on PLS, FEM
model above, the efficiency index of the and REM on each variable promotion cost
value of sales at the company's foreign and distribution costs.
investment is equal to the intercept To get more accurate results regarding
coefficient α = 9.872746. Multiple linear the election approaches, it is first tested
regression models are seen in the impact whether PLS or FEM approach that will be
taken simultaneously cost of promotion and used based on the results of the Chow test.
distribution on the sale. Valued 9.872746 Here are presented the results of the Chow
mean that if the variable cost of promotion test and Hausmann test in Table 5. As in
and distribution costs constant (no increase Table 5, it shows the results of the Chow test
or decrease) the company's sales will well with the F - test and chi - square which
increase by 9.872746 percent. is significant (prob value of 0.0000). This is
After the selection of panel regression smaller when compared to the significance
437
ISSN 2087-3735 The Effect of Promotion … (Sinta Wardani)
Table 5
Chow Test Results and Test Hausmann Chow Test of Hausman Test
Chow Test Hausman Test
Prob. Cross-section F 0.0000
Cross-section Chi-square 0.0000 0.9699
d.f. Cross-section F (10.53)
Cross-section Chi-square 10 2
Chi-Sq. Statistic 0.061219
Table 6
Testing the Assumptions of Classical Test against Foreign Investment and Domestic
Investment Companies
Foreign Domestic
Assumption Test Results
Investment Investment
Normality Test 0.254 0.874 Data distributed normally
Multicollinearity Test VIF = 5.068 VIF = 2.099 Data has no multicolline-
Tolerance = 0.197 Tolerance = 0.476 arity symptom
Heterocedastisity Test Promotion= 0.525 Promotion= 0.924 Data free from heteroce-
Distribution= 0.858 Distribution= 0.976 dastisity
Autocorrelation test 2.099 2.075 No Autocorrelation
level α = 5 percent. Thus, FEM is better sales will increase by 4.573494 percent.
when compared with PLS with 95 percent The next step is looking at the value of
confidence level. Prob (F - stat) to see if the independent
The Hausman test result, it gained variables simultaneously significantly affect
0.9699 prob value which is greater than α the dependent variable. Table 4 looks at the
significance level of 5 percent. This value of Prob (F - stat) FEM generated is
evidence suggests that REM is better than 0.0000, with the confidence level of 95
FEM with a 95 percent confidence level. In percent significant P-value (0.0000 < 0.05)
the event of conflicting results such as this, which means that the independent variables
to decide which should be the best model together proved to significantly affect
selected, it can be seen through the R - dependent variables simultaneously (Ha3 not
squared with the largest models. From table rejected).
4 known through that the R - squared of the
FEM method is bigger thus to be preferred. Classical Assumption Test against
From Table 4, the obtained regression Foreign Investment and Domestic
equation is as follows. Investment Company
Y = 4.573494 + 0.173607 promotion cost + However, in order to meets the criteria of the
0.775353 distribution costs. BLUE (best linear unbiased estimator) clas-
The original form of regression model sic assumption test should be done to foreign
above shows that the efficiency index of the investment and domestic investment compa-
value of sales at the company's domestic ny. These tests include normality test, hete-
investment is equal to the intercept roscedascity test, multicollinearity test and
coefficient α = 4.573494. Multiple linear autocorrelation test.
regression models found the impact taken Based on normality test of kolmogorov
together variable cost of promotion and smirnov in Table 6 found that the value is
distribution costs on the sale. Valued at greater than α=5% then it can be said that the
4.573494 means that if the variable cost of data is distributed normally. Based on the
promotion, and distribution costs constant calculation of multicollinearity test using
(no increase or decrease) the company's SPSS to foreign direct company, the VIF is
438
Journal of Economics, Business, and Accountancy Ventura Volume 16, No. 3, December 2013, pages 431 – 442
Accreditation No. 80/DIKTI/Kep/2012
5.068 and tolerance is 0.197. The VIF value Interpretation of Model Effects of the
for domestic company is 2.099 and tolerance Respective Foreign Investment and
is 0.476. All the results showed that the val- Domestic Investment Companies
ues of VIF is still less than 10 and tolerance Based on the results of the analysis of FEM
greater than 0.1, therefore there is no correla- model on foreign direct investment compa-
tion between independent variables in this nies, the value prob of each independent va-
regression model. In other word there is no riable on table 2, variable promotion costs
multicollinearity in the regression model. To have a probability of 0.0000 which less than
test for multicollinearity also can be done by α=0.05 (5%). This means, the variable pro-
looking at the value of coefficient indepen- motion costs have a significant effect on the
dent variables that have values above 0.8 in- companies’ sales. Thus the first partial hypo-
dicates multicollinearity. Heterocedastisity thesis which stated that the promotion costs
test results can be seen in the level of affect the sales is not rejected (Ha1 not re-
significance of the Spearman rank correlation jected). The same results happen in domestic
coefficient. Can be said to be free from need investment companies, according to Table 4;
heterocedacity value greater than 0.05, it can it is evident that the independent variable
be said that the regression equation in this promotion cost does affect the volume sales.
study is free of heterocedastisity. It can be seen from the value prob of promo-
Through table durbin-watson noted that tion costs 0.0004 that greater than α=0.05.
the value of DW of foreign company Thus the first partial hypothesis which stated
(1.308889) is between value 0, (dl) 1.6297 that the promotion costs affect the compa-
and (dU) 1.7437 (significance value of 5 nies’ sales is not rejected (Ha1 not rejected).
percent, the number of samples is 108 and The results above are still consistent
the number of independent variables is 2) with several studies which reveal that the
there is a proven means of autocorrelation. variables affect the company's sales
To overcome these transformations can be promotion. Some research suggests that the
performed by a general distinction method results of the sales promotion costs affect the
(Generalized Differences). Once it was company including the research conducted
transformed, and then it performed the by Simester, Hu, Brynjolfsson and Anderson
Durbin - Watson test back. The value of the (2005), Abdel - Khalik (1975) in Banerjee,
Durbin -Watson test results of 2.099 Siddhanta and Bandopadhyay (2012) and
common distinctions is then compared back Abiodun (2011). In addition, the research
to the Durbin - Watson value table that is Frankenberger and Graham (2004) who
between the lower limit (dl) 2 and the upper studied the effects of advertising and
limit (dU) 2.2759 which means there is no promotional spending during a recession by
autocorrelation. Similarly, the domestic using cross- sectional time series regression
firm, the Durbin - Watson value 1.422101 is with the a sample of 2,662 companies. It
between value 0, dl = 1.5305 and = dU found that the increase in advertising and
1.6640 (5 percent significance value, the promotional spending during the recession
number of samples 66 and the number of led to a higher profit than doing it during
independent variables 2) there is a proven non - recession.
means of autocorrelation. Once transformed, Balaghar, Majidazar and Niromand
and then performed the Durbin - Watson test (2012) from Iran who conducted a study to
back, the value of the Durbin -Watson test evaluate the effectiveness of promotional
results of 2.075 common distinctions is then items for advertising, sales promotion,
compared back to the Durbin - Watson value public relations, direct sales and direct
table that is between the lower limit (dl) 2 marketing on the volume of product sales
and the upper limit (dU) 2,336 proven there companies in Iran. It was found that there is
is no autocorrelation. a significant relationship between the
439
ISSN 2087-3735 The Effect of Promotion … (Sinta Wardani)
volume of sales and promotional equipment. regression method results obtained charge
In addition, the results also showed that the distribution has positive influence on the
most effective tool is a sales promotion, sales of the business Mr. Epi fish rearing
advertising, and public relations is ranked ponds are located on the river.
third, fourth is a direct marketing and the
latter is a direct selling. Managerial Implications
Based on the results of the analysis of Based on the results of the fixed effect mod-
FEM model on foreign direct investment el’s calculation of foreign direct investment
companies, the value prob of each indepen- and domestic investment companies, namely
dent variable on table 2, variable distribu- the independent variables promotion costs
tion cost have a probability of 0.0000 which and distribution costs are affecting signifi-
less than α=0.05 (5%). This means, the va- cantly to the companies’ sales. The FEM
riable distribution costs have a significant calculations on domestic and foreign
effect on the companies’ sales. Thus the companies showed that the value of
second partial hypothesis which stated that coefficient is positive, which means the
the distribution costs affect the sales is not promotion cost and distribution cost positive
rejected (Ha2 not rejected). Either in domes- effect on the value of company's sales. Thus,
tic investment companies, according to Ta- the domestic and foreign companies, in the
ble 4, it is evident that the independent va- event of an increase in promotional costs can
riable distribution cost affect volume sales. increase the value of the company's sales.
It can be seen from the value prob of pro- However, the coefficient value of indepen-
duction costs 0.0000 that less than α=0.05. dent variables for both companies proved to
Thus the second partial hypothesis which be less than one so it can be said that ineffi-
stated that the distribution cost affect the cient. Because of that, what must be done is
companies’ sales are not rejected (Ha2 not not maximization promotion costs and dis-
rejected). In Table 4 it can be seen free of tribution costs but the optimization all of it.
charge distribution of variable capital A good promotion costs and distribution
investment company in the country (DCI) costs optimization can make the costs bur-
also affect the sales volume. This can be den to a minimum and affect the output val-
seen in the value of the cost of distribution ue produced by the companies. The compa-
0.0000 smaller than 5 percent significance ny should conduct a comprehensive evalua-
level. The results are consistent with tion program of production, promotion and
several previous studies on the relationship distribution so that the weaknesses and
of the distribution of the company's sales shortcomings in the implementation of the
are Indrawati (2000), Mukodim (2007) and program can be known early.
Raya (2009). In addition the above To overcome this problem, the company
evidence, the study by Karim (2012) related can promote a strategy in times of recession
to the effect of cost distribution channel on rather than stopping overall campaign. There
sales volume, found that the cost of are several ways to advertise, with the hope
distribution channels has a positive effect. to continued raise awareness to customers
Nasri (2010) also conducted research on the but still remain economize. The companies
effect of distribution costs to sales volume don’t have to stop advertising as a whole, it
tilapia or carp in fish ponds in the river Epi is better to reduce the ads impressions in the
Mr. Lais. The method of data collection is media-print and television to maintain what
done by noting that in the form of case has been obtained by the companies. If the
studies, and the data used in the study is companies are using several different adver-
primary data which includes the cost of tising executions that aired at the same time,
distribution and sales of 2004 to 2008. just reduce it until just one ads thing. The
Based on calculations using simple linear companies may consider 15-seconds-
440
Journal of Economics, Business, and Accountancy Ventura Volume 16, No. 3, December 2013, pages 431 – 442
Accreditation No. 80/DIKTI/Kep/2012
advertising as the occupational force to period is relatively less than the expected
maintain the mental area that has been ob- number. There must be more research that
tained. Similarly, distribution costs optimi- accommodates other independent variables to
zation can make the burden of costs to a get a more complete understanding of the sales
minimum and therefore can affect the output performance. The researchers also suggest to
produced by the company. Because the dis- add more variables to describe the research
tribution system involving many parties that that the company's sales performance more
are scattered in various areas, the less exact broadly and comprehensively.
distribution system can increase distribution
cost’s companies. Therefore, it is better if REFERENCES
the companies use the concept of decentrali- Abiodun AO 2011, ‘The Impact of Advertis-
zation for their distribution system. Howev- ing on Sales Volume of a Product: A
er, the decentralization system may threaten Case Study of Stardoms Plc’, Nigeria,
the establishment of optimization overall. Valkeakoski (NG): University of Ap-
Decentralization tends to pursue the plied Science Pr.
achievement of sub-organizations’ purposes Balaghar AA, Majidazar M, Niromand M
at the expense of the organization’s objec- 2012, ‘Evaluation of Effectiveness of
tives as a whole. To that end, organizing of Sales Promotional Tools on Sales Vo-
the distribution system should be done by lume (Case Study: Iran Tractor Manu-
combining the concept of centralization and facturing Complex (ITMC)’, Middle-
decentralization. Organizing for centraliza- East J Sci Res, 11 (4).
tion means that handling system inventories, Banerjee N, Somroop S, Gautam B 2012,
warehousing, and transportation should be ‘The Effect of Marketing Communica-
under the control of one business unit, so tion on Sales Performance – A Study
that the integration of the three systems can on the Personal Care Industry in In-
be achieved to the maximum and overall dia’, European J Soc Sci, 28(3).
companies’ objectives can be achieved bet- Frankenberger KD, Graham RC 2004,
ter. ‘Should Firms Increase Advertising
Expenditures during Recessions?’,
CONCLUSION, IMPLICATION, SUG- MSI Reports, 03(115).
GESTION, AND LIMITATIONS Ghozali, I 2001, Aplikasi Analisis Multiva-
In general, by FEM method it shows that the riate dengan program SPSS, Semarang
promotion costs and distribution costs (ID): Diponegoro Univ Pr.
partially have a positive effect on the Gujarati, D 2006, Dasar-Dasar Ekonometri-
company's sales both in domestic and foreign ka, Jakarta (ID): Erlangga.
companies. Similarly, simultaneously, it Hernomo AD 2012, ‘Pengaruh Biaya Pro-
suggests that there is a positive effect of mosi dan Biaya Distribusi terhadap
independent variables on the dependent Volume Penjualan permen tolak angin
variable that is on domestic and foreign pada PT. Muncul Mekar Semarang’, J
companies. In order that the company can Ilmu Adm Bisnis, 12(6).
increase sales, they should pay attention to Indrawati, 2000, ‘Analisis Pengaruh Biaya
promotional expenses policy and distribution Distribusi terhadap Volume Penjualan
costs in marketing its products to consumers. pada CV Mekar Sejahtera Abadi Se-
Company should retain promotion and marang’, Post Graduate Thesis, Sema-
distribution policies that have executable, but rang (ID): Diponegoro Univ.
by taking action in order to save even more Karim A 2012, ‘Pengaruh Biaya Saluran
efficient company. Distribusi Terhadap Volume Penjualan
The limitation of this study is that the Krupuk Kelempang ”Dahliah” 1 Ulu
number of observations and the analysis Darat Palembang’, ILMIAH, 4(2).
441
ISSN 2087-3735 The Effect of Promotion … (Sinta Wardani)
442