Analyzing Marketing Margins and the Direction of Price Flow in the Tomato Value Chain of Limpopo Province, South Africa

International Journal of Environmental & Agriculture Research (IJOEAR) ISSN:[2454-1850] [Vol-3, Issue-3, March- 2017]
Page | 72
Analyzing Marketing Margins and the Direction of Price Flow in
the Tomato Value Chain of Limpopo Province, South Africa
Kudzai Mandizvidza
Development Economist and International Consultant, Beijing, China
Email: kudzaimandizvidza@outlook.com
Abstract— The Limpopo Province is home to South Africa’s tomato farming giants, some of whom also occupy the position
of the largest producers of the commodity in the Southern Hemisphere. Regardless of its importance in the tomato industry of
the country, there are few studies analysing the mechanism through which prices of tomatoes are determined and transmitted
from the farm gate in Limpopo to the various provincial, local and international markets. This study attempts to fill the
knowledge gap on the performance of Limpopo Province’s tomato markets by examining the marketing margins and the
direction of price flow amongst the successive tomato marketing levels. By means of the Concurrent Marketing Margin
Analysis approach, it was established that the farmers’ portion of the consumer’s Rand is low. About 85.1% of the
consumer’s Rand goes to pay for marketing margins. Granger causality tests show that both the wholesale and retail prices
are caused by farm gate prices. The farm level is therefore key to tomato price determination in Limpopo Province of South
Africa and pertinent policies that improve the commodity’s primary sector may potentially enhance the other sectors in value
chain that depend on tomato production.
Keywords— Limpopo Province, market dominance, marketing margins, price flow, tomato markets, vertical price
linkage.
I. INTRODUCTION
Agricultural productivity has improved throughout history because of economic progression, innovation, specialization,
research and development (Fuglie et al., 2007; Fuglie and Nin-Pratt, 2012; Nin-Pratt, 2013). This has increased the
productive capacities of farmers who in time have been able to produce in excess of what is needed for home consumption
and subsistence. Such developments augmented the significance and complexity of marketing and have seen the
establishment of arenas to facilitate the exchange of marketable surplus between producers and consumers.
In the context of the tomato industry of South Africa, it is highlighted in Department of Agriculture, Forestry and Fisheries
(DAFF) (2011) that production of tomato occurs in all the nine provinces of South Africa by both commercial and emerging
farmers. However, different provinces produce varied volumes due to environmental disparities amongst other factors. The
Limpopo Province with its warmer climate plays the most vital role in the country’s tomato production. According to DAFF
(2011) the province contributes about 3 590 ha to the country’s total area planted to tomatoes. Limpopo Tourism and Parks
Board (2011), suggests that almost 60% of the country’s tomatoes are produced in Limpopo Province which is responsible
for about 45% of the Johannesburg Fresh Produce Market’s annual turnover. However, the Limpopo Province’s tomato
producing industry’s concentration is high which in economics sense means that a large share of total production is
dominated by a few large producers. Limpopo Tourism and Parks Board (2011), indicates that only one farming company is
responsible for about 40% of the province’s 60% contribution to South Africa’s total tomato production.
Even though the majority of tomato is produced in the Limpopo Province, the commodity is consumed in all parts of the
country and is also exported thus; there is some degree of spatial separation between production and consumption. Such
spatial distribution calls for the need for an efficient and effective marketing system to facilitate the movement of tomatoes
from the point of production to the end user. Bringing the agricultural product from the agricultural enterprise to the
consumer according to Saccomandi (1998), involves a series of functions and productive activities that interconnect. The
process also depends on the current state of technology, the organization of productive activities and the spatial distribution
of production and consumption.
II. PROBLEM STATEMENT
There is high industry concentration in the tomato producing industry of the Limpopo Province. Such a situation where the
industry is dominated by a few large firms paves way to major producers potentially using their market power to influence
pricing strategies from time to time. As to which of the participants in the chain will consequently suffer or benefit from

Page | 73
impending price shocks depends on the degree to which market players adjust to price signals. It also depends on the timing
in response and the extent to which their adjustment to price shocks is asymmetric as shown in Vavra and Goodwin (2005).
Particular aspects of participants’ response to price movements may also have important implications on marketing margins
of several players in the marketing chain of tomato in Limpopo Province. While farmers may view their proportion of the
consumer’s Rand as low, tomato consumers on the other hand experience continual rising prices of this commodity. It may
be questioned therefore, as to whom in the marketing chain sits at point(s) of price determination. It may also be of interest to
determine the direction of price causality along the marketing chain for tomato in Limpopo Province so as to guide any
pricing policy in the respective food market.
III. AIM AND OBJECTIVES
This study seeks to analyze marketing margins and the direction of price flow in the tomato value chain of Limpopo Province
South Africa so as to determine the performance of tomato markets in the region. It’s is imperative to estimate the point(s) of
price determination and direction of causality along the marketing chain for tomato in Limpopo Province so as to guide food
marketing and pricing policies.
IV. RESEARCH HYPOTHESES
1. Price and direction of causality along the marketing chain for tomato in Limpopo Province are determined beyond the
farm gate.
2. The proportion of the consumer’s dollar that goes to pay for marketing margins is uncertain.
V. LITERATURE REVIEW
According to FAO (2002), the percentage share of the final price, which is taken up by the marketing functions, is what is
known as the marketing margin. Guvheya et al. (1998) treated the concept of marketing margins as differences between
prices at different levels in the marketing channel which capture the proportion of the final selling price that a particular
agent in the marketing chain adds. Elitzak (1996) characterize marketing margins or the farm-to-retail price spread as the
difference between the farm value and retail price that represents payments for all assembling, processing, transporting and
retailing charges added to farm products. Cox (2009) noted that producers often wonder about the large difference between
the prices that consumers pay for food and the prices that farmers receive. However, high marketing margins can be justified
by costs involved in distributing the product from point of production to the final consumer. It can be argued though, that too
high a percentage reflects some exploitation of either farmers or consumers. Wohlgenant (2001) investigated marketing
margins and identified some of the questions that are frequently asked about the issue. These questions, which have attracted
considerable interest across literature and amongst researchers and policy makers, include; Are marketing margins too large?
What is the incidence of marketing costs on retail prices and farm prices? How quickly are farm prices transmitted to the
retail level and how quickly are retail price changes transmitted to farmers? In order to have a thorough and exact
investigation, Mojtaba et al. (2010) recommended that the marketing margin be divided into two smaller portions namely
Retailer Margin (RM) and Wholesaler Margin (WM). RM refers to the difference between the price paid by consumers and
the price that retailers pay to the wholesalers. WM refers to the difference between the price at which wholesalers sell their
product and the price that they pay to the farmers.
While price represents the equilibrium point where demand and supply meet in the market place, Schnepf (2006) emphasizes
that the general price level of an agricultural commodity, at any stage, is influenced by a variety of market forces that can
alter the current or expected balance between supply and demand. According to Holland (1998) price determination for many
consumer products is in most cases a function of the cost of production and a desired level of mark-up. Price determination
by this desired level of mark-up is what is referred to as cost-plus pricing, mark-up pricing or full-cost pricing which in
Salvatore (1993) is related to three rules of thumb. The first is termed; mark-up percent, which is the proportion of profit to
total cost. The second is the gross margin percent, which is the proportion of profit to the selling price, and lastly, profit
margin, which is the difference in selling price and total cost.
VI. STUDY AREA, DATA COLLECTION AND SAMPLING TECHNIQUE
This study is based mainly on the Limpopo Province which is the northernmost province of South Africa covering a total
surface area of 123,910 km2 making it the fifth largest province amongst the country’s nine provinces. Limpopo Province
can be described as the garden of South Africa due to its rich fruit and vegetable production. Apart from being South Africa’s

Page | 74
main tomato producer Limpopo Province is responsible for about 75% of the country’s mangoes, 65% of its papayas, 36% of
its tea, 25% of its citrus, bananas, and litchis, 60% of its avocados, 285 000 tons of potatoes, and 35% of its oranges. Most of
tomato production in the province is done by a private corporation, who is also the largest tomato farmer in South Africa.
The province is also involved in the production of coffee, nuts, guavas, sisal, cotton, tobacco, sunflower, maize, wheat,
grapes and timber. Livestock production and game ranching is also prominent particularly in most of the higher-lying areas
(vanNiekerk, 2012; Limpopo Tourism and Parks Board, 2011).
This study used both primary and secondary data of time series in nature. Daily tomato prices were collected synchronously
at farm gate, wholesale and retail levels for mixed grades of cooking tomatoes to achieve a sample size of 50 observations.
Semi structured interviews and desk studies were also conducted to explore the procurement and marketing procedures
followed by various participants in tomato marketing in the Limpopo Province.
VII. MARKETING MARGIN ANALYSIS
In analysing the marketing margins, the Concurrent Margin Method was used. As described by Singh (1998) the method is a
static analysis of the distributive margin usually adopted to calculate the price spread in one market town by considering
differences between prices prevailing at successive stages of marketing at a given point of time.
The model is defined as thus;
Mt = P t,L – Pt,L-1 where; (1)
Mt= Marketing margin between market level (L) and its preceding level (L-1) at time (t)
Pt,L = Price at market level (L) at time (t)
Pt,L-1 = Price at market level (L-1) at time (t)
Where marketing margins at different levels of the marketing chain are compared, Guvheya et al. (1998) emphasizes the use
of consumer price as the common denominator for all margins.
The two indices that were used in this study are Total Gross Marketing Margin (TGMM) and Producers’ Gross Marketing
Margin (GMMp) as given in Scott (1995) where Gross Marketing Margin is the difference between consumer’s price and
farmer’s price. TGMM and GMMp were calculated as;
TGMM =
’
x 100, (2)
and
GMMp = x 100 (3)
VIII. TESTING FOR UNIT ROOT NON-STATIONARITY
The Augmented Dickey Fuller test was performed on each of the logarithmic series of farm prices (FP), wholesale prices
(WP) and retail prices (RP) to formally ascertain whether they contained a unit root.
IX. PRICE DETERMINATION AND DIRECTION OF CAUSALITY
In assessing the points of price determination and the direction of causality along the major marketing channels for tomato in
Limpopo Province, Granger causality tests (Granger, 1969) were performed.
Three sets of Vector Autoregressive (VAR) models were formulated and estimated;
(a) Testing causality between farm gate and wholesale levels:
lnFPt = α1 + ilnWPt-i + ilnFPt-i + U1t
lnWPt = α2 + ilnWPt-i + ilnFPt-i + U2t (5)

Page | 75
(b) Testing causality between wholesale and retail levels:
lnWPt = α1 + ilnRPt-i + ilnWPt-i + U1t
lnRPt = α2 + ilnRPt-i + ilnWPt-i + U2t (6)
(c) Testing causality between farm and retail levels:
lnFPt = α1 + ilnRPt-i + ilnFPt-i + U1t
lnRPt = α2 + ilnRPt-i + ilnFPt-i + U2t (7)
*where; ln is the natural logarithm of each respective price series
FPt is the farm price at time (t)
WPt is the wholesale price at time (t)
RPt is the retail price at time (t)
FPt-i is lagged farm gate price
WPt-i is lagged wholesale price
RPt-i is lagged retail price
n is the upper limit set at the optimal lag length
ai, bi, ci, and di are coefficients to be estimated using the ordinary Least squares method and α1 and α2 are intercepts
U1t and U2t are error terms that are assumed uncorrelated and white noise
The inference in the first mathematical statement of equation 5 is that current farm prices are dependent on past farm prices
and past and present wholesale prices.
Likewise, the second mathematical statement postulates that current wholesale prices are dependent on past farm prices and
past and present wholesale prices.
The inference in the first mathematical statement of equation 6 is that current wholesale prices are dependent on past
wholesale prices and past and present retail prices.
Likewise, the second mathematical statement postulates that current retail prices are dependent on past wholesale prices and
past and present retail prices.
The inference in the first mathematical statement of equation 7 is that current farm gate prices are dependent on past farm
prices and past and present retail prices.
Likewise, the second mathematical statement postulates that current retail prices are dependent on past farm gate prices and
past and present retail prices.
In each of the cases (a), (b) and (c), four causality relationships were tested by placing the appropriate restrictions on each
model. P-values and F-tests were used to confirm statistical significance of the causality relationships.
For instance, in case (a) the following were the causality relationships tested between farm gate and wholesale prices;
(i) A unidirectional causality from wholesale to farm gate levels would be concluded if;
i ≠ 0 and i = 0
(ii) A unidirectional causality from farm gate to wholesale levels would be concluded if,
i = 0 and i ≠ 0
(iii) An absence of a causal relationship between the variables that is independence would be concluded if both
i = 0 and i = 0
This would imply that both sets of the lagged exogenous variables were not statistically different from zero.

Page | 76
(iv) a bilateral causality or feedback would exist if both
i ≠ 0 and i ≠ 0
This would imply that both sets of the lagged exogenous variables were, as a group, statistically significantly different from
zero.
In formulating the above sets of VAR models, the VAR Lag Order Selection Criteria (VLOSC) was used to determine the
optimal lag length.
X. RESULTS OF THE STUDY
10.1 Marketing Margin Analysis
Prices of the tomatoes varied across the marketing chain over time. For the ten week period the highest weekly average farm
gate price was ZAR1.85/kg, while the lowest was ZAR1/kg and the average farm gate price for the whole period was
ZAR1.37/kg. The highest wholesale price was ZAR5.45/kg; while the lowest was ZAR3.10/kg and the whole period average
was ZAR4.73/kg. The highest retail price was ZAR9.60/kg, while the lowest was ZAR8.79/kg and the average was
ZAR9.20/kg.
Table 1 shows the marketing margin structure of Limpopo Province’s tomato marketing chain on five day weekly basis for
the period of study.
TABLE 1
FIVE DAY WEEKLY MARKETING MARGINS FOR LIMPOPO PROVINCE’S TOMATO MARKET CHAIN
South African Rands (ZAR)/ kg %
Week FP WP RP
F-to-W
GMM
W-to-R
GMM
F-to-R GMM Total GMM Producer GMM
1 1.33 4.17 8.79 2.84 4.61 7.45 84.84 15.16
2 1.34 4.93 9.28 3.60 4.34 7.94 85.58 14.42
3 1.85 4.80 9.37 2.95 4.57 7.52 80.27 19.73
4 1.76 5.23 8.95 3.48 3.71 7.19 80.36 19.64
5 1.14 4.42 9.08 3.28 4.66 7.94 87.44 12.56
6 1.00 3.10 9.60 2.10 6.50 8.60 89.58 10.42
7 1.46 5.45 9.49 3.98 4.04 8.03 84.56 15.44
8 1.29 5.19 9.15 3.90 3.97 7.86 85.94 14.06
9 1.18 4.98 9.09 3.81 4.11 7.92 87.07 12.93
10 1.36 5.02 9.20 3.66 4.18 7.83 85.19 14.81
Average 1.37 4.73 9.20 3.36 4.47 7.83 85.08 14.92
The average farm-to-wholesale gross marketing margin (FWGMM) of R3.36 and the average wholesale-to-retail gross
marketing margin (WRGMM) of R4.47 in sum give us the average farm-to-retail gross marketing margin (FRGMM) of
R7.83. With the average retail price of R9.20, it follows that consumers typically had to part with R9.20 for every kilogram
of tomato they purchased from retailers. However, comparing the FRGMM of R7.83 with the average retail price we can
conclude that the total gross marketing margins constituted about 85.1% of the consumer’s Rand. With such a high
marketing margin, the common hypothesis that the farmers’ portion of the consumer’s Rand is low holds true. During the
period of analysis, tomato farmers were getting only 14.9% from every Rand spent by consumers on tomatoes at retail level.
10.2 Unit Root Tests
The information on margins, given in the previous section, prompts the need to undertake further explorations on vertical
price linkages amongst different levels in the marketing chain of tomatoes in Limpopo Province. The statistical properties of
the price series are analysed before carrying out causality tests. Table 2 presents an abstract of results of the unit root tests
performed on lnFP, lnWP, and lnRP according to the Augmented Dickey-Fuller (ADF) criteria.

Page | 77
TABLE 2
AUGMENTED DICKEY FULLER TEST RESULTS ON lnFP, lnWP, lnRP IN LEVELS
Null Hypotheses: each of lnFP, lnWP, and lnRP contain a unit root
lnFP
ADF test statistic
Test Critical Values
t-Statistic Prob.
-2.083252 0.5419
1% level -4.156734
5% level -3.504330
10% level -3.181826
Lag Length: 0 (Automatic - based
on SIC, maxlag=5)
DW stat: 1.799676
lnWP
ADF test statistic
t-Statistic Prob
-0.073823 0.6531
1% level -2.614029
5% level -1.947816
10% level -1.612492
Lag length: 1 (Automatic - based
on SIC, maxlag=5)
DW stat: 2.089489
lnRP
ADF test statistic
t-Statistic Prob
0.126524 0.7180
1% level -2.613010
5% level -1.947665
10% level -1.612573
on SIC, maxlag=5)
DW stat: 2.548604
The results in Table 2 shows that ADF test-statistic values of the three price series are greater than the MacKinnon critical
values for rejecting the hypotheses of a unit root. As a result, we cannot reject the null hypothesis of a unit root at 1%, 5%
and 10% levels of significance. This regression result can be trusted since the Durbin-Watson statistics are all significant
enough to reject the presence of serial correlation in each of the three series. With the statistical evidence generated, we can
therefore conclude that the farm gate, wholesale and retail price series of Limpopo produced tomatoes for the period of
analysis is non-stationary.
Vavra and Goodwin (2005) noted that most economic time series being non-stationary in nature need some transformation
through differencing or de-trending, otherwise the regression will be spurious. Spurious regressions occur when the mean,
variance and covariance of a time series vary with time. The classic results of a usual regression cannot be legitimate if non-
stationary series of data is used for analysis.
As such, first differences were taken on each of the three series lnFP, lnWP, lnRP to come up with a differenced set of data,
DlnFP, DlnWP, and DlnRP that is unit root free. The summary of the ADF test on the first differenced series are shown in
Table 3.
TABLE 3
AUGMENTED DICKEY FULLER TEST ON DlnFP, DlnWP, and DlnRP (FIRST DIFFERENCES)
Null Hypotheses: each of DlnFP, DlnWP and DlnRP contain a unit root
DlnFP ADF test statistic
t-Statistic Prob
-6.644587 0.0000
1% level -4.161144
5% level -3.506374
10% level -3.183002
on SIC, maxlag=3)
DW stat: 2.004682
DlnWP ADF test statistic
t-Statistic Prob
-11.33005 0.0000
1% level -4.161144
5% level -3.506374
10% level -3.183002
on SIC, maxlag=5)
DW stat: 2.097217
DlnRP ADF test statistic
t-Statistic Prob
-9.150489 0.0000
1% level -2.614029
5% level -1.947816
10% level -1.612492
Lag length: : 0 (Automatic - based
on SIC, maxlag=5)
DW stat: 2.030275

Page | 78
Table 3 shows that ADF test-statistic values of the three differenced price series are less than the MacKinnon critical values
for rejecting the hypotheses of a unit root. As a result, the null hypothesis of a unit root in first differences of lnFP, lnWP and
lnRP can be rejected at 1%, 5% and 10% levels of significance. Again, the regression result can be trusted since it passed the
Durbin-Watson test. It can therefore be concluded with 99% confidence that the DFP, DWP and DRP series of Limpopo
produced tomatoes for the period of analysis are stationary since there is enough statistical evidence to support this. The
implication of the ADF tests results is that the series generating the three price variables are all integrated of order one, that
is, I(1).
10.3 Lag Order Selection Criteria
The next step was to carry out the lag-order-selection criteria for correct specification of the VAR model to use in Granger
Causality tests. Results showing the optimal lag lengths to use in the causality tests are summarized in Table 4. For the
sample with 50 observations, lnFP and lnRP were the endogenous variables, while the constant C was the only exogenous
variable.
TABLE 4
VECTOR AUTO REGRESSIVE LAG ORDER SELECTION CRITERIA (lnFP and lnRP)
Lag LR FPE AIC SC HQ
0 NA 6.40e-05 -3.980930 -3.900634 -3.950997
1 70.25371 1.44e-05 -5.475860 -5.234972* -5.386059*
2 6.503296 1.46e-05 -5.460664 -5.059184 -5.310997
3 4.282047 1.56e-05 -5.395572 -4.833499 -5.186037
4 9.614758* 1.44e-05 -5.484871 -4.762206 -5.215469
5 7.539553 1.39e-05* -5.528845* -4.645588 -5.199575
In Table 4, the asterisk (*) indicates lag order selected by each criterion, LR is the sequential modified LR test statistic (each
tested at 5% level), FPE is Final Prediction Error, AIC is the Akaike Information Criterion, SC is the Schwarz Information
Criterion and HQ stands for Hannan-Quinn Information Criterion. The VAR lag order selection criteria results summarized
in Table 4 show that AIC and FPE chose 5 lags, SC and HQ 1 lag and only LR chose 4 lags. However, retailers indicated that
they in most cases alter their tomato prices on weekly basis. It is therefore reasonable to accept the AIC and FPE choices
given the available knowledge of the tomato retail markets in Limpopo Province. As a result, five lags were used as the
optimal lag length for testing Granger Causality and cointegration relationships between farm gate prices and retail prices.
A similar procedure was done to determine the optimal lag length between farm gate and wholesale prices and results are
shown in Table 5. For the sample with 50 observations, lnFP and lnWP were the endogenous variables, while the constant C
was the only exogenous variable.
TABLE 5
VECTOR AUTO REGRESSIVE LAG ORDER SELECTION CRITERIA (lnFP and lnWP)
0 NA 0.001598 -0.763229 -0.682933 -0.733296
1 59.44108* 0.000464* -2.000715* -1.759827* -1.910915*
2 3.811524 0.000504 -1.918226 -1.516745 -1.768558
3 2.279414 0.000569 -1.800433 -1.238360 -1.590898
4 7.039860 0.000563 -1.818206 -1.095541 -1.548804
5 4.482782 0.000594 -1.772275 -0.889018 -1.443006
Table 5 shows that the optimal lag length between the farm and wholesale is one. In the same way the VAR Lag Order
Selection Criteria results for wholesale and retail prices are shown in Table 6.

Page | 79
TABLE 6
VECTOR AUTO REGRESSIVE LAG ORDER SELECTION CRITERIA (lnWP and lnRP)
0 NA 3.93e-05 -4.467427 -4.387131 -4.437494
1 26.85320* 2.48e-05* -4.929011* -4.688123* -4.839211*
2 7.009408 2.49e-05 -4.926469 -4.524988 -4.776801
3 2.730931 2.78e-05 -4.820558 -4.258485 -4.611023
4 3.816023 3.00e-05 -4.748781 -4.026116 -4.479378
5 7.354435 2.92e-05 -4.787310 -3.904052 -4.458040
Table 6 shows that the optimal lag length between the wholesale and retail is one. lnWP and lnRP were the endogenous
variables.
10.4 Granger Causality Tests
The results for the pair wise Granger causality test on the series lnFP, lnWP and lnRP are presented in Table 7.
TABLE 7
RESULTS OF PAIR WISE GRANGER CAUSALITY TESTS FOR lnFP, lnWP, and lnRP
Null Hypothesis Lags Obs. F-stat. Prob. Decision
lnWP does not Granger Cause lnFP
lnFP does not Granger Cause lnWP
1 49
9.1E-06
8.19633
0.9976
0.0063
Do not reject
Reject
lnWP does not Granger Cause lnRP
lnRP does not Granger Cause lnWP
1 49
0.77784
0.28772
0.3824
0.5943
Do not reject
Do not reject
lnFP does not Granger Cause lnRP
lnRP does not Granger Cause lnFP
5 45
3.93156
1.12593
0.0064
0.3654
Reject
Do not reject
10.5 Price Causality between Farm and Wholesale
Table 7 shows that the p-value (0.0063) is significant at 1% level. As a result the hypothesis that lnFP does not cause lnWP
can be rejected. In support of this assertion, the p-value (0.9976) on the other hand is insignificant which means we cannot
reject the hypothesis that lnWP does not cause lnFP. With such statistical evidence, it can be concluded that there exists a
unidirectional causality from the farm level to the wholesale level. This implies that farm prices cause wholesale prices. A
high industry concentration that is present at the farm level can be the reason for the direction of price flow from the farm
gate (who has market power) to the wholesale (who use the commission system based on the prices set by farmers).
10.6 Price Causality between Wholesale and Retail
Table 7 also shows that both p- values (0.3824 and 0.5943) are insignificant at 5% level, as such we do not reject the null
hypotheses of lnWP not Granger causing lnRP and lnRP not Granger causing lnWP. However, the F-statistic probabilities
indicate a slight possibility of the hypothesis of lnWP not causing lnRP more likely to be rejected at a higher significance
level. In this sense, it is more likely that wholesale prices may perhaps cause the retail prices than the opposite even though
the chances are trivial. This weak causal relationship is represented by the dotted line in Figure 1. The empirical evidence
available at 5% level even so, still suggests that there is no extrapolative power between wholesale prices and retail prices. It
therefore follows that an independent causal relationship exists between the wholesale and the retail levels. Possible
explanations for such a causal relationship is the growing outsourcing behaviour of, and direct purchasing from the farms by
some tomato retailers in the Limpopo Province. For instance during the exploratory phase of this study, it was established
that some retailers consider the National Fresh Produce Markets as a competitor than a supplier.

Page | 80
10.7 Price Causality between Farm and Retail
The p-value (0.0064) is significant at 1% level while p-value (0.3654) is insignificant at 1%, 5% and 10% levels. The
hypothesis that lnFP does not cause lnRP can therefore be rejected while the one for lnRP not causing lnFP cannot. It can be
concluded that prices stream unidirectional from the farm level to the retail level. In other words, farm gate prices have a
predictive power on retail prices. The market structure of the tomato industry in Limpopo Province suggests a high
concentration at the farm level than at the retail. Such can be the reason for the direction of price flow from the farm (who
has more market power) to the retail level.
The causality relationships that exist amongst the three series are demonstrated in Figure 1. Arrows represent the direction of
price flow, where solid and dotted lines symbolize a strong and a weak causal relationship respectively.
FIGURE 1: POINTS OF PRICE DETERMINATION AND DIRECTION OF PRICE CAUSALITY
As shown in Figure 1, the farm gate plays a major role in the price formation process of tomato markets in Limpopo
Province. Therefore, the farm gate’s current and past price information is useful in improving the forecasts of both the
wholesale and retail prices it causes. Figure 1 also indicates that there is a weak causal relationship between the wholesale
and retail level prices. The weak dependence of Limpopo retail prices on wholesale prices can be attributed to the proximity
of retailers to tomato farms than they are to the National Fresh Produce Markets. As a result it is sensible that Johannesburg
wholesale prices may be seldom useful in predicting tomato retail prices in Polokwane considering the increasing direct
sourcing by Limpopo retailers from farmers.
XI. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS
The study attempted to fill the knowledge gap on the performance of Limpopo Province’s tomato markets by examining
prices at successive marketing levels. The gross marketing margins constitute about 85.1% of the consumer’s South African
Rand. There exists a large gap between what consumers pay for each unit of tomatoes purchased from retailers, and the
amount farmers receive for the same quantity from retailers in Limpopo Province. It therefore follows that the producers’
portion of the consumers’ purchasing power is low since according to the findings, a major part of tomato retail prices is
taken up by the marketing functions.
The VAR lag order selection criteria showed that the optimal lag lengths between the levels; farm gate and wholesale, farm
gate and retail, wholesale and retail are 1, 5 and 1 respectively. Since the reaction to a price change at the point of
determination does not reflect instantaneously at the other levels in a marketing chain, these optimal lag lengths indicate that
wholesale prices are faster in adjusting to farm gate price changes than do retail prices.
The results of the pair-wise Granger Causality tests suggested a unidirectional causality from farm to wholesale and also
from farm to retail. However, a weak causal relationship was found between wholesale and retail levels. It therefore follows
that tomato prices are determined at the farm gate in Limpopo Province. This finding may be consistent with the expectation
that whenever an industry is highly concentrated, the most dominant participant may also act as the price leader as is the
case in the tomato producing industry of Limpopo Province, South Africa.

Page | 81
An important conclusion drawn from the causal relationships between the farm gate, wholesale and retail prices is that the
farm level is key to tomato price determination in Limpopo Province. It therefore means that current and past information on
farm gate prices is useful in improving estimations of both wholesale and retail prices it causes. According to Guvheya et al.
(1998), information on causality shows the direction of price flow between levels and thus helps in the identification of
points of price determination along the marketing chain. In the light of this assertion, it can be concluded that tomato prices
in Limpopo Province are determined at the farm level. With the farm level being key to tomato price determination in
Limpopo Province of South Africa, pertinent policies that improve the commodity’s primary sector may by the ripple effect,
potentially benefit the other sectors in the value chain who are also dependent on tomato production.
This this study has shown the importance of the farm level at deciding the market price for food. Governments in the
developing economies such as South Africa are encouraged to intensify small scale farmers support programs so as to reduce
industry concentrations as well as encourage fairer competition in the food producing sectors of their respective economies.
This will ensure that agricultural produce prices will not be determined by a few major producers. A scenario where only one
producer dominates a market potentially leads to situation where all price movements are influenced by this single player
while the other farmers are simply price takers regardless of their cost structures.
XII. AREAS FOR FURTHER STUDY
The study established that the gross marketing margins of tomato in Limpopo Province are high. However the source of such
huge marketing margins has not been ascertained between marketing costs and retailers’ profits. This study only managed to
alert on the existence of high marketing margins that constitute the major portion of the consumer’s Rand. It is therefore
recommended that further research be undertaken to establish whether the retailers’ profits are unfairly high or peradventure
whether the marketing margins are as a result of high marketing costs. Such a complementary study will assist in identifying
the actual basis of the high marketing margins so that corrective policy measures can be directed appropriately and more
specifically.
ACKNOWLEDGEMENTS
Sincere appreciation goes to Mrs L. C Muchopa and Dr. P. Chaminuka for their grooming and guidance. Heartfelt gratitude
goes to Mrs Glenda (nee Nyanhongo) Mandizvidza, Mukundinashe Thaniel Kudzai and Ms. N. Shylletti Tsivai for their
support and inspiration.
REFERENCES
[1] Cox, L. (2009). Direct marketing of your agricultural products. Retrieved May 12, 2012, from
http://www.ctahr.hawaii.edu/sustainag/news/articles/V2-Cox- DM.pdf
[2] Department of Agriculture, Forestry and Fisheries (DAFF). (2011). A profile of the South African tomato market value chain.
Retrieved November 12, 2012, from http://www.nda.agric.za/docs/AMCP/TOMATOMVC2012.pdf
[3] Elitzak, H. (1996). Food cost review. Washington D.C: U.S Department of Agriculture.
[4] Food and Agriculture Organisation (FAO). (2002). Agricultural Marketing Extension. FAO TCP/SAF/0065, No 1 Pg1-24.
Retrieved February 04, 2011, from http://www.nda.agric.za/docs/GenPub/ 1AgriMarketExtension.pdf
[5] Fuglie, K. O., Macdonald, J. M., & Ball, E. (2007). Productivity Growth in U.S. Agriculture. Retrieved May 20, 2013, from United
States Department of Agriculture Economic Research Service: http://www.ers.usda.gov/media/201 254/eb9_1_.pdf
[6] Fuglie, K., & Nin-Pratt, A. (2012). 2012 Global Food Policy Report: A Changing Global Harvest. Retrieved May 20, 2013, from
International Food Policy Research Institute: http://www.ifpri.org/gfpr/2012/agricultural-productivity
[7] Granger, C. W. (1969). Investigating causal relationships by econometric models and cross-spectral methods. Econometrica: 37, 424-
38.
[8] Guvheya, G., Mabaya, E., & Christy, D. R. (1998)."Horticultural marketing in Zimbabwe:", Margins, price transmission and
spatial marketIntegration. Staff Paper. New York, Ithaca, U.S.A: Department of Agricultural, Resource, and Managerial
Economics Cornell University.
[9] Holland, R. (1998). Selling price, gross margin &mark-updetermination. ADC info# 12. Agricultural Extension Centre, The
University of Tennessee.
[10] Limpopo Tourism and Parks Board. (2011). Limpopo Province, South Africa's food basket. Retrieved November 5, 2011,from
http://www.golimpopo.com/fact-details_2_5.html
[11] Mojtaba, A., Farhad, S. B., & Hussein, E. (2010). Efficiency of date marketing system in Sistan & Bluchestan of Iran; A
marketing margin approach. Agric. Econ. – Czech, 56, 2010 (1) , 43-50.

Page | 82
[12] Nin-Pratt, A. (2013). Agricultural Productivity: Good and Bad News. Retrieved May 20, 2013, from International Food Policy
Research Institute:
http://www.ifpri.org/blog/agricultural-productivity-good-and-bad-news
[13] Saccomandi, V. (1998). Agricultural Market Economics:A Neo-Institutional Analysis of the Exchange, Circulation and Distribution
of Agricultural Products. Netherlands: Van Gorcum & Comp. B. V.
[14] Salvatore, D. (1993). Managerial Economics In A Global Economy. McGraw-Hill.
[15] Schnepf, R. (2006). Price determination in agricultural commodity markets: A Primer. U.S: Congressional Research Service.
[16] Scott, G. J. (1995). Prices, product and people: Analysing agricultural markets in developing countries. Colorado: Lynne Rienner
Publishers, Inc.
[17] Singh, B. (1998). Production and marketing of wheat in India. New Delhi: Mittal Publications.
[18] van Niekerk, L. (2012). South Africa Yearbook 2011/12- Land and its people. Pretoria: Government of South Africa.
[19] Vavra, P., & Goodwin, B. (2005). "Analysis of Price Transmission Along the Food Chain". OECD Food, Agriculture and Fisheries
Working Papers, No. 3. OECD Publishing doi:10.1787/752335872456.
[20] Wohlgenant, M. K. (2001). Marketing Margins: Empirical Analysis. Department of Agricultural & Resource Economics, North
Carolina State University, Raleigh NC.

Analyzing Marketing Margins and the Direction of Price Flow in the Tomato Value Chain of Limpopo Province, South Africa

More Related Content

What's hot

Similar to Analyzing Marketing Margins and the Direction of Price Flow in the Tomato Value Chain of Limpopo Province, South Africa

Recently uploaded

Analyzing Marketing Margins and the Direction of Price Flow in the Tomato Value Chain of Limpopo Province, South Africa