Sie sind auf Seite 1von 10

This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy

Food Policy 33 (2008) 455–463

Contents lists available at ScienceDirect

Food Policy
journal homepage: www.elsevier.com/locate/foodpol

Rice marketing systems in the Philippines and Thailand: Do large numbers


of competitive traders ensure good performance?
David C. Dawe a,*, Piedad F. Moya b,1, Cheryll B. Casiwan c, Jesusa M. Cabling c
a
Food and Agriculture Organization, Rome 00153, Italy
b
International Rice Research Institute, DAPO Box 7777, Metro Manila, Philippines
c
Philippine Rice Research Institute, Maligaya, Science City of Muñoz, Nueva Ecija, Philippines

a r t i c l e i n f o a b s t r a c t

Article history: Rice marketing margins are substantially greater in the Philippines than in Thailand despite many sim-
Received 31 August 2006 ilarities between the two systems and despite the fact that Philippine rice marketing has a competitive
Received in revised form 16 January 2008 structure. We found that rice marketing costs in the Philippines are higher than in Thailand mainly
Accepted 25 February 2008
due to higher interest rates in the financial system. Other fundamental factors that also result in higher
costs include endowments of water and land, rice price and trade policy, road quality and lack of non-
farm job growth. However, the greater costs can only account for about a fourth of the difference in gross
margins, implying much higher returns to management in the Philippines despite similar levels of risk
Keywords:
Rice
and no evidence of collusion. The ‘‘excess profits” in the Philippine marketing system suggest there is
Marketing much to learn about how developing country commodity markets with competitive structures function
Philippines in actual practice.
Thailand Ó 2008 Elsevier Ltd. All rights reserved.

Introduction saw: the saw eats be it pulled up or pushed down, and we Filipinos
are the ones being eaten” (cited in Houston, 1953, p. 44). On the
Rice marketing costs in Southeast Asia have received scant other hand, a study of rice marketing conducted around this time
attention in the economics literature, probably due to the difficulty in Nueva Ecija, a province in the main rice-producing area of the Phil-
of collecting such data (Hayami et al., 1999). When such analyses ippines, concluded that ‘‘the Chinese merchants are performing
are conducted, it is generally found by economists that rice essential marketing services at a fairly low cost” (Miller, H.H.,
marketing is relatively efficient in the sense that there are large 1932. Principles of Economics applied to the Philippines, Boston,
numbers of participants, barriers to entry are low, product differ- pp. 198–199, cited in Wickizer and Bennett).
entiation is small, the ability of other agencies (e.g., government, Hayami et al. (1999) suggest that one of the reasons for this dis-
cooperatives) to compete with private traders on a cost-effective parity in viewpoints is a lack of data. This is one possible reason, but
basis is limited, and marketing margins rewarding risk and innova- it is also true that even when the data exist, they can be quite diffi-
tion are not excessive (e.g., Ruttan, 1969; Mears, 1974, 1981; Ellis cult to interpret. For example, during the past ten years, rice market-
et al., 1991; Hayami et al., 1999). However, this small literature is ing margins in Thailand have been consistently and substantially
at odds with the common belief among the general population that lower than the same margins in the Philippines. At the same time,
middlemen exploit farmers and consumers by inflating marketing rice marketing technology is broadly similar in the two countries.
margins between farmgate and retail levels. Does this, therefore, indicate collusion in rice marketing in the Phil-
These contrasting views date back many years. The first president ippines, or are there other explanations? This paper attempts to
of the Philippines, Manuel Quezon (President from 1935 to 1944), in explain this puzzle by measuring marketing costs in some detail
referring to ethnic Chinese rice traders, stated that ‘‘One of our short- in the two countries and interpreting those data in order to provide
comings is that we permit others to hold the rice industry for us, and insights into the performance of the marketing systems.
they are the ones who make a lot of profit. The profiteers are like a

General description of rice marketing and margins in the study


* Corresponding author. Fax: +39 06 5705 5522. areas
E-mail addresses: david.dawe@fao.org (D.C. Dawe), p.moya@cgiar.org (P.F.
Moya), cbcasiwa@philrice.gov.ph (C.B. Casiwan), jmcabling@philrice.gov.ph (J.M.
Cabling). Comparison of rice marketing margins between the Philippines
1
Fax: +63 2 580 5699. and Thailand at the country level is difficult. First, the Philippines is

0306-9192/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.foodpol.2008.02.001
Author's personal copy

456 D.C. Dawe et al. / Food Policy 33 (2008) 455–463

140

120

100
US$ per ton dry paddy

80

60

40

20

0
1994:WS 1995:DS 1995:WS 1996:DS 1996:WS 1997:DS 1997:WS 1998:DS 1998:WS 1999:DS 1999:WS

-20

Philippines Thailand

Fig. 1. Gross marketing margins, 1994 to 1999 Nueva Ecija to Manila (Philippines) and Suphan Buri to Bangkok (Thailand) Units are in US$ per ton dry paddy. Notes: WS
indicates wet season, DS indicates dry season. No farm price data for Thailand were collected for 1996 WS and 1997 DS. Source of farm price data: Moya et al. (2002). Sources
of secondary wholesale price data: BAS (2005) and BOT (2005). Price data were converted from local currency to US$ using exchange rates from IMF (2005).

an archipelago, with much transport being done by sea, while Thai- sale price to calculate the margin. This calculation is relatively sim-
land is located on the southeast Asian mainland. Second, rice is ple in that it ignores the value of by-products from the milling
produced and consumed in many different areas of each country, process (we have no time-series data on the value of such by-prod-
and the average distance that rice moves may be very different ucts) and assumes identical milling ratios in the two countries.
in the two countries. Because of these problems, we decided to These data show that margins are substantially larger from Cen-
compare the rice marketing margin between Suphan Buri in the tral Luzon to Manila than from the Central Plain to Bangkok (Fig. 1).
Central Plain and Bangkok (Thailand) with the margin between Although the size of the margin varies from year to year and season
Nueva Ecija in the Central Luzon region and Manila (Philippines). to season, the margins in Thailand are consistently smaller by a
Comparison of these two marketing channels is appropriate for substantial amount. In fact, for one season, margins are even neg-
several reasons. First, both production areas (the Central Plain and ative in Thailand, representing a temporary disequilibrium condi-
Central Luzon) grow similar types of modern varieties without any tion. The disadvantage of these data and this calculation is that
aroma or other salient quality characteristics, and are the major none of the marketing details are known, and there is no light shed
rice-growing areas in their respective countries.2 Second, both des- on the reasons for the differences in margins. However, the advan-
tinations (Bangkok and Manila) are the capitals and largest cities of tage of these data is that they cover several years, and show that
their countries. Third, both rice-growing areas are connected by land there is a difference in margins that is consistent and is worth
to wholesale markets in the capital city, so that no sea transportation explaining. The data in Fig. 1 are used only to establish that there
is involved. Fourth, the driving distance between the rice-growing is a consistent difference in margins over time; they are not used
area and the capital city is about 120 km in both cases. subsequently in the paper’s main analysis.
Moya et al. (2002) conducted farm surveys for nearly all rice- The large margins in the Philippines are surprising given that
growing seasons between 1994 and 1999 in Suphan Buri province Hayami et al. (1999) found the rice marketing system in the Phil-
in the Central Plain of Thailand and in Nueva Ecija province in the ippines to be competitive and that rice marketing technology is
Central Luzon region of the Philippines. Farm price data for paddy broadly similar in Central Luzon and the Central Plain. In both loca-
of comparable quality from these surveys were adjusted to dry tions, the crop is brought to mills after harvesting via trucks. Most
paddy terms, and then compared with wholesale price data from storage is of paddy, not rice. The paddy is milled predominantly by
secondary sources for average quality non-aromatic (i.e., excluding fixed mills with rubber roll hullers. Finally, the milled rice is trans-
jasmine rice) rice in the respective capital cities (Bangkok and ported to wholesale markets by large trucks. There is little hand
Manila). The wholesale price data were adjusted to dry paddy pounding, milling using mobile mills, or milling using steel hullers
terms using a milling ratio of 0.65 in both countries, and then as discussed for the case of Indonesia by Timmer (1973).
the adjusted farm price was subtracted from the adjusted whole- However, there are also some differences in the two marketing
systems that are worth noting. First, farmers in the Central Plain of-
2
ten sell their harvested paddy at wholesale markets, although recent
Thailand grows substantial quantities of jasmine rice (nearly 20% of national
production), an aromatic variety that commands a premium price on world markets.
government policies may be reducing the role played by these mar-
However, jasmine rice prefers arid conditions, and most of it is grown in the kets. To our knowledge, such markets are non-existent for paddy in
Northeast of Thailand. There is little jasmine rice grown in the Central Plain. the Philippines, although they are common for milled rice. Second,
Author's personal copy

D.C. Dawe et al. / Food Policy 33 (2008) 455–463 457

the existence of paddy wholesale markets in Thailand has strongly The second term is p0f , defined as the price at farmgate level (pf)
reduced the role played by private traders who function only as trad- adjusted to 15% moisture content, the typical moisture content
ers (not millers) and buy and sell paddy on their own account. In- when paddy is milled:
stead, Thai (Central Plain) farmers tend to sell directly to private
ð1  0:15Þ
millers or their agents. On the other hand, private, own-account, p0f ¼ pf
ð1  MCÞ
traders who are not millers are an important fixture of the Philippine
marketing system. Third, a much larger percentage of the paddy crop where MC is the moisture content of wet paddy as sold by the farm-
is dried mechanically in Thailand compared to the Philippines, con- er (as a fraction of one). The adjustment made to pf to generate p0f
sistent with the higher level of economic development and wages in does not include the costs of drying, but is just a physical adjust-
Thailand that has led to a substitution of capital for labor. Fourth, ment factor to ensure that the units of V and p0f are identical: local
trucks and mills have larger capacity in Thailand than in the Philip- currency per kilogram of dry paddy.4
pines. Subsequent analysis shows that some of these differences The main focus of the paper is to measure the costs that are
play an important role in explaining the difference in marketing responsible for the gross marketing margin (GMM). Unless other-
margins between the two sites, while others are not so important. wise specified, whenever we use the word margin in this paper,
the reference is to the GMM and does not include any costs. In
Data and methods order to make comparisons, all units of local currency (peso in
the Philippines, baht in Thailand) were converted to US dollar
We used a data gathering approach similar to that of Hayami et al. equivalents using exchange rates of 55 pesos to the dollar or 40
(1999) and traced paddy from the farm level as it moved along the baht to the dollar. These rates were typical of those prevailing in
marketing chain to the wholesale level, collecting data on both prices 2003.
and costs. Because traders are more difficult to interview than farm- Our data on marketing costs are a comprehensive measure of all
ers, the authors spent several weeks during the second half of 2003 in costs incurred in the marketing process, other than returns to man-
each of the two study areas visiting farmers, traders, millers, equip- agement. A non-exhaustive list of these costs includes the cost of
ment manufacturers, and government officials. Our questionnaires hired labor (both manual labor and skilled labor such as accoun-
were semi-structured. It was necessary to collect a large amount of tants and engineers); the annualized value of capital assets includ-
quantitative information, but it was also important to understand ing machinery, buildings, weighing scales, sacks, and cement floors
the details of the marketing systems, and most of our conversations (using the capital recovery factor described in Monke and Pearson,
departed from the questionnaires. Some of the people we inter- 1989); interest on working capital; the cost of raw materials (e.g.,
viewed were very forthcoming with lots of information, while others gasoline for transportation and electricity for mill operation);
were less so. We crosschecked all of our data both between agents of maintenance costs for machinery; and trading fees.
similar function (e.g., millers) and across different types of agents Most of our analysis compares marketing costs between the
(e.g., comparing selling prices reported by farmers with buying prices two locations for particular functions, because the marketing func-
reported by traders and millers) to make sure that answers were sim- tions of transportation, drying, storage, and milling are similar in
ilar. By following these procedures, we collected high quality data on the two areas. Although marketing agents performed different
both prices and costs in the marketing systems in the two areas. combinations of those functions in the two study areas, we do
Rice, the commodity that is cooked and eaten by consumers, is present and discuss data on net margins by agent, as this provides
produced through the milling process from paddy, the product insights into the distribution of these margins. Also, we compared
grown by farmers. In addition to the rice grain inside, freshly marketing costs in terms of currency, not percentage terms. The
harvested paddy also contains moisture, a hull, and some other main reason for this is that rice farm prices are much higher in
by-products located between the hull and the grain. the Philippines than in Thailand. Thus, if margins were compared
Our definition of gross marketing margin (GMM) for this study as a percentage of the farm price, the different base in the Philip-
was the difference between two terms, GMM ¼ ðV  p0f Þ. The first pines would complicate the analysis. Nevertheless, to supplement
term V represents the final value of product when traders or mill- this analysis, we do compare some margins and markups in
ers sell 1 kg of milled rice to wholesale markets and associated by- percentage terms because traders and millers may operate on a
products to other outlets: percentage markup basis.
X
V ¼ ðpw ÞðMRÞ þ pi qi
Results and discussion
i2B

where pw is the price of rice at wholesale markets; pi is the price of Our detailed field surveys confirmed the preliminary conclu-
by-product i belonging to the set of by-products B; qi is the yield of sions showing that gross marketing margins in Thailand were
by-product i during the milling process, as a fraction of one; and MR substantially lower than in the Philippines (Table 1). The GMM
is the milling ratio, i.e., the yield of milled rice as a fraction of one. in Thailand was calculated as just $15.52 per ton of dry paddy,
Thus, V is a weighted average price of the various products that re- compared to $66.74 per ton of dry paddy in the Philippines
sult from the processing of paddy. Wholesale rice markets were (all costs and margins in the remainder of this paper will be
chosen because they are more homogeneous than retail markets, reported per ton of dry paddy, unless otherwise specified). Mark-
facilitating data comparison across time and across countries.3 ups in the Philippines are also greater in percentage terms. The
fact that the margins calculated using our primary data are sim-
3
ilar to the margins shown in Fig. 1 (which come from indepen-
Our data collection stopped at the entrance to the wholesale market for milled
rice. This sidesteps the question of whether there is a cartel of traders at the
dent sources) suggests that our primary data on gross margins
wholesale level, a claim often heard in the Philippines and other countries (as of 2002, are credible.
there were only 97 registered wholesale grain traders in Metro Manila). Indeed, the Given the disruptions that occurred in Asian currency markets
gross margin between milled rice wholesale buying and selling prices is larger in the in the aftermath of the 1997 financial crisis, it is possible that
Philippines than in Thailand. However, the difference in that margin between the two
countries is a small fraction of the difference in gross margin between the two
countries from farmgate up to wholesale level. Thus, if the objective is to increase
4
prices for farmers and lower prices for consumers, this paper addresses the The costs of drying are not included in the calculation of the GMM, nor are any
quantitatively more important segment of the marketing system. other costs. However, the costs of drying are included in the subsequent analysis.
Author's personal copy

458 D.C. Dawe et al. / Food Policy 33 (2008) 455–463

Table 1 the Philippine marketing system are not due to some disequilibrium
Rice prices at different steps of the marketing chain, 2003, Nueva Ecija to Manila phenomenon resulting from exchange rate movements after the
(Philippines) and Suphan Buri to Bangkok (Thailand)
financial crisis.
Level, form Price (US$ per ton of % Markup over There are three main observations that emerge from an exami-
dry paddy) previous step nation of the cost data (see Table 2). First, not surprisingly, given
Philippines Thailand Philippines Thailand the smaller gross margins, marketing costs are substantially lower
Farmgatea 164 108c – – in Thailand than in the Philippines. Second, marketing costs in
Millgate 184 115 12 6 Thailand equal the gross margin, implying zero returns to manage-
Wholesaleb 231 124 25 8 ment in the marketing system.6 Third, the difference in marketing
Gross marketing margin (GMM) 67 16
costs between the two countries explains only about one-fourth of
Source of data: surveys conducted by the authors. Price data were converted from the difference in gross marketing margins. Each of these observa-
local currency to US$ at exchange rates of 55 Philippine pesos to the dollar and 40 tions requires some explanation.
Thai baht to the dollar.
a
The farmgate price was converted from a wet paddy basis of 23% moisture to a
dry paddy basis of 15% moisture. These values were obtained during our interviews. Differences in marketing costs
b
The wholesale price includes the value of by-products, and is not the price of
milled rice alone. Based on our surveys, milled rice at the wholesale level was The importance of the financial system
converted to dry paddy using conversion factors of 0.64 in the Philippines and 0.66
in Thailand. Conversion factors for by-products were also determined through our
surveys. The single most important reason why marketing costs are low-
c
Very few transactions take place at the farmgate in this part of Thailand. Thus, a er in Thailand is the level of interest rates and the services provided
farm price was imputed by subtracting farm to paddy wholesale market transport by the financial system. In Thailand, millers and traders cited nom-
costs from the price received by farmers at the paddy wholesale market. The total
inal interest rates for borrowing of about 4% per annum, and this
percentage markup in prices by the miller, from paddy wholesale market to milled
rice wholesale market, is 5%. was confirmed in interviews with several banks. Interest rates
were similar for either working or investment capital. By contrast,
interest rates from banks in the Philippines were about 15% per an-
num.7 Furthermore, millers and traders in the Philippines do not
Table 2 borrow much from banks for their working capital needs because
Marketing costs by function, 2003, Nueva Ecija to Manila (Philippines) and Suphan
Buri to Bangkok (Thailand); units are in US$ per ton of dry paddy
of excessive paperwork for such short-term loans. Instead, they bor-
row from moneylenders, who charged higher interest rates of 24% on
Item Philippines Thailand Differential an annual basis.8 A simulation of counterfactual costs assuming that
Transport costs real interest rates in the Philippines were the same as those in Thai-
Fuel 3.88 1.29 2.58 land explains 58% ($7.64 per ton) of the $13.07 per ton difference in
Trading/search 3.03 1.93 1.10
marketing costs between the two countries. This is an underestimate
Labor 3.84 1.76 2.08
Investment capital 1.26 2.10 0.84 of the true effect, because it assumes no substitution of capital for
Maintenance 0.64 0.24 0.40 labor in response to the lowering of interest rates, and thus a sub-
Subtotal 12.65 7.32 5.32 optimal capital to labor ratio. Unfortunately, it is not possible for
Drying costs us to quantitatively simulate this substitution process (as Timmer,
Fuel 0.00 1.14 1.14 1973, does for rice milling in Indonesia) because capital is used in
Labor 2.01 0.00 2.01
many different parts of the marketing process, and we do not have
Investment Capital 0.50 1.40 0.90
Maintenance 0.00 0.22 0.22
the necessary data to construct the relevant isoquants for all of these
Subtotal 2.51 2.76 0.25 processes. Thus, we must be content with an estimate of the lower
Storage costs
bound importance.
Sacks 1.85 0.96 0.89 In terms of function, the lower interest rates in Thailand make a
Investment Capital 0.60 0.25 0.35 difference in two main ways. First, lower interest rates lead to low-
Working Capital 5.11 0.10 5.02 er working capital requirements, which are shown in Table 2 as
Subtotal 7.56 1.31 6.26
reduced costs of storage. Second, lower interest rates reduce
Milling costs investment capital requirements by lowering the effective pur-
Electricity 1.29 1.27 0.02
chase price of trucks, buildings, sacks, and milling equipment.
Labor 1.88 1.82 0.07
Investment Capital 2.34 0.61 1.73
Maintenance 0.36 0.44 0.08 Seasonality of production and openness to trade
Subtotal 5.87 4.14 1.74
Total costs 28.60 15.53 13.07 While interest rates are clearly the single most important factor,
they are not the entire story. In the Central Plain of Thailand, there
Source of data: surveys conducted by the authors. Price data were converted from
local currency to US$ at exchange rates of 55 Philippine pesos to the dollar and 40
Thai baht to the dollar.
6
As a point of comparison, Fafchamps and Gabre-Madhin (2004) estimated median
marketing margins net of costs (i.e., returns to management) of $13.13/ton in Benin
the high margins in the Philippines relative to those in Thailand are and $47.39/ton in Malawi. These margins are calculated across different commodities
and types of traders. Maize, beans, and pulses were the main commodities traded in
the result of exchange rate conversions. However, since the peso
both countries.
depreciated against the baht after 1997 (compared to its relatively 7
Interest rates are determined jointly with exchange rates, and any government
constant level from 1990 to 1996),5 this factor would have made interventions in foreign exchange markets could affect the level of interest rates. In
Philippine rice marketing more competitive compared with Thailand both the Philippines and Thailand, however, foreign exchange markets are subject to
when converted to a common currency. Thus, the high margins in only very limited occasional intervention, and actual exchange rates closely approx-
imate true market-determined rates. Thus, the lower interest rates in Thailand are not
due to policies in the foreign exchange market.
8
Real interest rates were used in our analysis. They were computed from nominal
5
With the exception of a brief period in the second half of 1997 and the first half of interest rates using average annual inflation rates from 1996 to 2003 (3.2% in
1998 that is not relevant to our study. Thailand, 6.0% in the Philippines).
Author's personal copy

D.C. Dawe et al. / Food Policy 33 (2008) 455–463 459

is less seasonality of harvesting due to more abundant water sup- low Thai marketing agents to haul more tons per liter of gas. First,
plies in the Chao Phraya River delta. Some farms in the Central trucks in the Philippines typically transport only about 70% of a full
Plain harvest three crops of rice in a year, and different farms in load of grain, compared to trucks in Thailand that travel full most
nearby areas often have quite different harvesting dates. The result of the time. This underutilization appears to be due to an excessive
is a cropping system that resembles continuous factory production. number of marketing agents in the Philippines that makes it diffi-
Seasonality of rice production in Central Luzon is much less pro- cult to optimally coordinate transportation. Second, Thai trucks
nounced than it was 35 years ago (the Green Revolution led to a that shuttle between rice-growing areas and the capital typically
large expansion of cropping in the dry season), but it is still more have a capacity of 30 tons, double the capacity of the trucks used
seasonal than in the Central Plain. in the Philippines. Larger trucks are impractical in the Philippines
Furthermore, international trade plays a much larger role in because the quality of roads is considerably worse than in Thai-
Thailand’s rice economy (exports were 42% of domestic production land. Philippine roads have more potholes, tend to pass through ur-
from 1996 to 2003) than it does in the Philippines (where imports ban areas instead of passing the outskirts, and have fewer lanes. All
were 14% of domestic consumption during the same period). This of these factors make it difficult to drive large trucks. Third, higher
difference is due to policies, not endowments: despite domestic interest rates also make it more difficult for Philippine traders to
prices that are much higher than world prices, the Philippines con- finance the purchase of larger trucks, which are of course more
tinues to restrict rice imports. In Thailand, however, world prices expensive than smaller trucks.
and domestic prices (adjusted to the same point in the marketing Transportation labor costs per ton are also much lower in Thai-
chain) are equal. More openness to international trade allows for land (by $2.08 per ton), even though transportation capital costs
some of the seasonality of supply to be shunted off to (or absorbed per ton are similar in the two countries. Part of this differential
by) world markets in the face of intertemporal rice consumption can be traced to the use of smaller trucks that do not travel full
demand that is roughly constant and exhibits little seasonal in the Philippines (as noted above), which means employment of
pattern. more drivers and higher labor costs per ton of grain hauled despite
The reduced seasonality of production and greater openness to lower hourly wages. More important, there is a substantial amount
trade combine to result in reduced storage requirements in Thai- of manual loading and unloading of grain in the Philippines. This
land (two months from harvest to wholesale on average in the process has been mechanized in Thailand because of higher wages
Philippines, compared to just one month in Thailand). Confirming and lower interest rates. Of course, the lower price of labor relative
the reduced storage requirements that we found in our interviews9 to capital in the Philippines suggests that less mechanization is
is the fact that farmgate paddy prices exhibit less seasonality in Thai- optimal, but this does not appear to be the entire explanation.
land. Our simulations indicate that the reduced storage require- Mechanization would likely lower costs in the Philippines, but
ments due to endowments and policies account for $2.56 of the manual laborers are often able to resist such changes.10 Faster
difference in marketing costs between the two systems. job growth in the non-farm sector would provide alternative jobs
Lack of openness to rice trade in the Philippines raises storage for these laborers and pave the way for more mechanization, but
costs through another channel as well. Government restrictions the Philippine economy’s lack of growth has hindered the creation
on imports have raised domestic farm prices to very high levels – of such jobs. Measured in constant local currency terms, GDP per ca-
in fact, farm prices for paddy in Central Luzon from 1994 to 1999 pita in the Philippines grew just 0.2% per annum from 1980 to 2003,
were on average 89% higher than in the Central Plain during the while it grew at an annual rate of 4.5% in Thailand during the same
same period of time. High farm prices mean that millers and trad- period (raw data from World Bank, 2007).
ers in the Philippines have much larger borrowing requirements to Perhaps surprisingly, differential search costs were not large
finance purchases of the more expensive crop, even apart from the (just $1.12 per ton) despite the innovation of paddy wholesale
higher interest rates that prevail in the Philippines. Our simula- markets in Thailand. The fees paid to agents in the Philippines
tions estimate that the size of this effect amounts to $1.92 per who deliver paddy to mills or traders are just a tiny fraction of
ton in additional carrying costs. the farm price, and have not changed in years. Furthermore, paddy
wholesale markets in Thailand also charge per ton fees, in addition
Land prices, road infrastructure, lack of mechanization, and other to the commissions paid by millers to their agents who buy from
factors farmers at the wholesale market. There is no doubt that the inno-
vation of wholesale markets has reduced search costs, but the cost
Several other factors are important as well. One is land costs for reduction is not that large, perhaps because of the low opportunity
the mill and storage warehouse, which are cheaper in Suphan Buri cost of agents’ time in the Philippines (i.e., the transaction costs of
by $1.61 per ton compared to Nueva Ecija. Land is cheaper in the informal trade are small).11
rice-growing areas of Thailand, where the mills are located, for at Lower milling ratios in the Philippines (0.64) compared to Thai-
least two reasons. First, the supply of arable land per capita is three land (0.66) made no difference to differential costs. The reason for
times as large in Thailand as in the Philippines (raw data from FAO, this lies in the relative prices of milled rice to by-products and the
2006). Second, the main use of land in both Suphan Buri and Nueva greater output of husks (which were considered worthless in both
Ecija is for rice, and, as noted above, paddy prices in the Philippines countries) in Thailand (0.25 compared to 0.23 in the Philippines).
are kept substantially above world prices through import restric- We also investigated trade policy regarding intermediate inputs
tions. These very high levels of domestic rice prices inflate land (e.g., machinery) as a potential explanation for higher costs, but
values, as has occurred in Japan (Trewin et al., 2000). found that tariff rates on milling equipment and trucks are very
The Thai marketing system also has a cost advantage of $2.58 low in the Philippines at 0–3%.
per ton in terms of fuel consumption despite similar levels of fuel
prices. The difference is due to several factors that collectively al- 10
In fact, this is the situation that prevails regarding mechanization of harvesting in
the Philippines. Use of a combine harvester in the Philippines would lower rice
production costs, but we met several farmers who either had their combine sabotaged
9
Millers and traders were reluctant to discuss how long they stored paddy for fear by workers worried about loss of employment, or were hesitant to purchase one for
of being branded as hoarders: the word storage has negative connotations. However, fear of the same.
11
they were more responsive when we asked how long it took from harvest until arrival Fafchamps and Gabre-Madhin (2004) place more emphasis on search costs in
of rice at the wholesale market, without mentioning the word ‘‘storage.” In any event, their comparison of Benin and Malawi. One hypothesis is that search costs are more
the latter concept is in fact storage for the marketing system. important where population densities are lower and infrastructure is more limited.
Author's personal copy

460 D.C. Dawe et al. / Food Policy 33 (2008) 455–463

Summary of cost differences Table 3


Breakdown of marketing margins and costs by marketing agent, 2003, Nueva Ecija to
Manila, Philippines (Phil) and Suphan Buri to Bangkok, Thailand (Thai)
The sum of all these effects adds up to a difference in costs of
$19.51 per ton, which is greater than the observed difference of Gross Cost (US$ per Absolute % markup
$13.07 per ton. There are two reasons for this ‘‘over explanation”. marketing ton dry markup over over costs
margin (GMM) paddy) costs (US$ per
First, there are interaction effects between different factors that (US$ per ton ton dry paddy)
make simple addition of the effects of individual changes mislead- dry paddy)
ing. For example, the combined effects of hypothetically lower Phil Thai Phil Thai Phil Thai Phil Thai
interest rates and reduced storage time in the Philippines would Assembly trader 20.02 – 15.95 – 4.07 – 26
lead to a reduction in costs of $7.79, which is less than the sum Miller 46.72 15.52 12.64 15.53 34.07 0.01 269 0
of $7.64 and $2.56 (the effects of the two changes made individu- Total 66.74 15.52 28.60 15.53 38.14 0.01
ally). Second, the Philippines has a cost advantage in some compo- Source of data: surveys conducted by the authors. Price data were converted from
nents (e.g., drying), mainly due to lower wages. local currency to US$ at exchange rates of 55 Philippine pesos to the dollar and 40
Thai baht to the dollar.

Differences in margins
hand, since 2003 was an unusual year, our estimate of the differen-
It is interesting, however, that differential marketing costs ex- tial gross margin is biased upwards, i.e., in a normal equilibrium
plain just one-fourth of the differential gross margin. Explaining situation the differential gross margin between the two countries
this differential gross margin is important because it is so large. will be smaller than we have measured. This is a problem to some
If returns to management were equal in the two countries (i.e., extent, but Fig. 1 shows a consistent large difference in GMM be-
gross margins in the Philippines were larger only due to higher tween the two countries, with an average difference of US$56,
costs), then wholesale rice prices in the Philippines would be lower which is similar to our measurement of US$51. Thus, the extent
by 18% (15.22 pesos per kg of milled rice instead of 18.50 pesos per of the bias is small. It remains true that differential costs can ac-
kg). Most of the larger net margin in the Philippine case occurs at count for only a small percentage of the differential gross market-
the milling stage, as can be seen from an examination of the costs ing margin.
and margins disaggregated by agent in Table 3. The Philippine data are also potentially at issue. First, we
believe that the gross margin data for the Philippines are not in
Data quality dispute, because they are very similar to the five years of data re-
ported earlier in the paper in Fig. 1.15 They are also very similar to
Given the difficulties involved in measuring marketing costs, it margins calculated using only data from government secondary
is legitimate to ask whether our data are of good quality, both in sources. Second, our estimates of milling costs are similar to esti-
Thailand and in the Philippines. mates from a separate, independent study by Cabling (2002). Since
Table 2 shows that marketing costs (excluding management milling is the stage where the net margin is so large, this separate
costs) in Thailand essentially equal the gross margin, implying zero estimate of costs gives us added confidence in our results. Third,
returns to management for the marketing system. This is a striking our estimates of the returns to management are strikingly similar
finding, and may cast doubt on the quality of the data, since zero to those of Hayami et al. (1999). Converting their estimates of net in-
returns to management are not possible in the long run. However, come earned by agents, traders, and millers per ton into dollars, we
such a finding is possible in the short run.12 In fact, during our inter- calculate returns to management of $35.90 per ton (using their data
views, many millers in Thailand mentioned that competition had and the exchange rates prevalent during the period of their survey),
intensified in the past few years because of low interest rates that which is very close to our estimate of $38.14 for the Philippine mar-
facilitated new entry and expansion of existing millers. Thus, capac- keting system considered as an aggregate. Thus, we strongly believe
ity utilization had declined in spite of a record rice crop in 2003. Sec- that we have good estimates of marketing costs and margins in the
ondary data on national level farm and wholesale rice prices show Philippines (and in Thailand).16
that 2003 was indeed an unusual year, with gross margins at their
lowest point during the period from 1995 to 2004.13 Presumably, Collusion among traders?
there will need to be some consolidation of the industry during
the next few years so that either gross margins increase or costs de- One possible explanation for the differential gross margin (after
cline further, either of which would allow some positive returns to accounting for differential costs) is collusion among traders, as
management.14 found by Banerji and Meenakshi (2004) in Indian wheat wholesale
Since costs are more difficult to measure than the GMM, and markets. While there are no organized wholesale paddy markets in
our measure of costs equals the GMM, this provides confidence the Philippines at which this type of collusion could occur, we con-
that we did not miss any important cost components. On the other sidered the possibility that some market segmentation takes place
in the Philippines. For example, perhaps traders and millers parcel
up the market amongst them, and agree not to ‘‘invade” each
12
Fig. 1, which uses independent data, shows that the GMM is even negative on
other’s territory. This possibility, however, is not consistent with
occasion in Thailand, which is yet more extreme than a situation where gross repeated observations we made in the field, similar to those of
marketing margins equal costs. Hayami et al. (1999) that farmers have a wide choice of millers
13
Unfortunately, these secondary data cannot be used to quantify the level of the and traders to sell to, and that millers and traders often cross-pro-
GMM, as the farm price data are collected at national level and confound a number of
very different rice qualities at the farm level (in particular, fragrant jasmine rice and
white rice). A comparison over time, however, does indicate that 2003 was an unusual
15
year. The same holds true for Thailand.
14 16
Furthermore, because the volumes handled by Thai millers are large, only a very We tried asking millers about profit rates, but they were unwilling to answer that
small return to management on a per ton basis is necessary for viability of the mill. question (and many others, e.g., about levels of stocks) directly. Accountants formerly
For example, a $1 per ton return to management in Thailand would generate an involved in the industry told us (without prompting) that the profit rates for millers
annual income of $90,000, a high income in a developing country. Inclusion of such a were very high, without citing specific numbers. Even if they had cited specific
small margin would not affect the conclusions of the paper, and could easily be within numbers, it is not clear they would calculate profit in the same manner as an
the measurement error of our cost data. economist would. Thus, a direct quantitative comparison is not likely to be useful.
Author's personal copy

D.C. Dawe et al. / Food Policy 33 (2008) 455–463 461

vincial boundaries in search of paddy to buy. Furthermore, these explanation is probably not of major importance, as there is sub-
cross-provincial ‘‘raids” are not just confined to neighboring prov- stantial excess capacity in the Philippine milling system. Assuming
inces, but often occur over long distances.17 operation at 8 h per day, 6 days per week, and 48 weeks per year
Other possible barriers to trade include the ‘‘suki” system, (allowing four weeks for maintenance), data from PhilRice-BAS
whereby farmers preferentially sell to a specific trader because (2004) on the capacity of registered millers and total production
the farmer is familiar with that trader (most suki relationships indicate that national average capacity utilization is slightly less
are not due to credit tying). These arrangements are very informal, than 80%, which indicates a reasonable amount of unused capacity.
however, and confer no strong obligations on the farmer. In fact, Further, mills in Thailand are striving to operate two to three 8 h
farmers with a suki often sell to non-suki traders. Only a tiny pro- shifts per day, with only rare shutdowns, in marked contrast to
portion of farmer sales are made to providers of input credit. But if the vast majority of millers we interviewed in the Philippines,
neither marketing costs nor collusion are responsible for the differ- who were content to operate on 8 h days. When asked if they could
ence in gross margins, then we must examine the residual, or re- expand to two shifts per day, they expressed little interest in doing
turns to management, more closely. so but gave no reasons as to why it was difficult or impossible. If
capacity utilization in the Philippines is defined relative to 16 h
Risk and trade barriers? per day instead of eight, then capacity utilization would be less
than 40%. If Philippine mills want to increase throughput, it should
Why are returns to management so much higher in the Philip- be fairly easy to increase the number of shifts per day without re-
pines? A possible explanation is greater risk, but we find this uncon- sort to additional investment capital.
vincing. First, prices are slightly more stable in the Philippines, A more important shortcoming of capital markets in the Philip-
where the government has successfully stabilized prices vis-à-vis pines is that moneylenders (who finance working capital require-
the world market, than in Thailand, where free trade means that ments) do not typically loan the large amounts of money
domestic prices are dictated by world prices and the exchange rate. necessary to buy 90,000 tons of paddy per year to individual bor-
Second, we are unaware of any millers or traders who have had their rowers, probably because they are unwilling to take on such large
assets expropriated by the government, or were even threatened amounts of risk. And banks seem unable to perform the service of
with such. Thus, neither price risk nor risk of expropriation is able satisfying the working capital requirements of traders and millers
to explain the large returns to management. in the Philippines. We were surprised by some of our interviews
Another possibility is that the import protection of rice produc- that even very large wealthy mill owners were unable to tap the
tion in the Philippines contributes to high marketing margins banking system for adequate working capital and were forced to
through rent seeking behavior.18 But this protection is for rice pro- resort to moneylenders. These capital market constraints create a
duction, not rice marketing. Further, government marketing interven- situation where it is difficult for efficient traders to expand opera-
tions do not artificially lower farm prices or artificially raise retail tions and drive out inefficient marketing agents, thus accounting
prices, at least one of which would be necessary in order to increase for the large number of marketing agents in the Philippines, as
millers’ margins. Indeed, farm prices are high due to import restric- noted above. Ellis et al. (1991) found evidence in Indonesia that
tions, and the government further supports the high domestic prices fewer links in the rice marketing chain in some locations led to
with procurement at above market prices. In addition, the govern- lower overall margins in those locations, which is consistent with
ment tries to lower retail prices, usually unsuccessfully, by selling sub- the results here.
sidized rice. Finally, the Philippines has more than 10,000 widely While the Philippine system is characterized by large numbers
dispersed rice mills (PhilRice-BAS, 2004), and we are unaware of any of small mills and low capacity utilization, consolidation has taken
organized lobbying efforts by rice mills at the national level. It is easy place during the past 15 years. The number of mills declined by
to register a rice mill, and there are no restrictions on entry. 29% from 1987 to 2002, while total capacity grew by 7% during
To the best of our knowledge, then, it seems as though market- the same period (raw data from PhilRice-BAS, 2004). At the same
ing agents in the Philippines are earning excess profits, with the time, rice production increased 48%, so that capacity utilization in-
word ‘‘excess” being defined relative to the profits earned in Thai- creased from 57% to 79% (based on 8 h days). Thus, the system is
land.19 If true, how do these ‘‘excess” profits persist? If there is no headed in the right direction.
collusion, why are they not competed away? Undoubtedly there is
no single answer; but there are several possibilities that could be ‘‘Making a living” and business uncertainty
important.
Another way to explain the high margins in the Philippine sys-
Capital markets and capacity utilization tem is that millers (and traders) must make a living from their re-
turns to management (Harriss, 1981). If the quantities milled or
One possible reason for the persistence of ‘‘excess profits” is traded per agent are low, then margins will need to be high in or-
that high interest rates hinder the ability of efficient millers to der to properly reward people with management skills and access
build bigger mills and drive out less efficient millers. But this to capital. Thus, restricted access to working capital and the need
for traders to earn a reasonable living given their skills interact
to create high margins. As noted earlier, the magnitude of profits
17
Interestingly, some local governments at the village level try to hinder such in the Philippines as measured by Hayami et al. (1999) was similar
competition by enacting laws that prohibit traders from outside jurisdictions from to ours, and they suggested that the profits they measured pro-
entering the locality. The number of local jurisdictions with such restrictions is small, vided a reasonable standard of living for the traders and millers in-
however, and farmers are still free to move paddy outside of their villages, so these
volved (who were operating at a smaller scale than the typical
restraints on trade do not have major effects.
18
The discussion here is distinct from the other effects of protection discussed miller in our study area). They did not conclude that there were
earlier (e.g., higher working capital requirements, longer storage time and higher land ‘‘excess” profits because they did not have data from Thailand with
values). which to make a comparison and were thus not led to examine the
19
Fafchamps and Gabre-Madhin (2004) find a similar phenomenon in their possibilities for consolidation and economies of scale.
comparison of Benin and Malawi, suggesting that cross-country comparisons can be
useful for making policy recommendations to improve marketing systems (i.e., the
Lack of paddy wholesale markets may also contribute to larger
issue is more than simply the efficiency of a marketing system given its existing net margins by increasing the number of market intermediaries
constraints). who must earn a living. If wholesale paddy markets drove out
Author's personal copy

462 D.C. Dawe et al. / Food Policy 33 (2008) 455–463

assembly traders in the Philippines, and their current margins to be learned about the actual functioning of developing country
disappeared from the system (because millers are already earning commodity markets with competitive structures. It also suggests
enough to make a living), this would have an important impact on that we cannot be entirely certain in making recommendations
marketing margins. about how to reduce marketing margins in the Philippines.
Another possibility is that management skills will earn higher re- Nevertheless, despite our imperfect knowledge of the reasons
turns in an uncertain business environment. Filipino businesses are behind the differential margins, some suggestions can be made.
more likely to cite corruption (35%) and labor regulations (25%) as First, by far the greatest impact would come from capital markets
major business constraints than their counterparts in Thailand that provide lower interest rates and expanded access to working
(18% and 11%, respectively). Filipino businesses also spend a greater capital. Second, margins might be reduced if the government took
percentage of management time dealing with officials (7% versus 1% the initiative to organize wholesale markets for the purchase of
in Thailand). As a result, it is much easier to do business in Thailand paddy. This would directly lower search costs, but more impor-
than in the Philippines. Indeed, Thailand had the lowest (best) rank- tantly it might help to reduce the number of agents involved in
ing in the World Bank’s ‘‘ease of doing business” index among all the marketing system, which could lower transaction costs more
developing countries in 2005 (no data for 2003 are available), and broadly by eliminating the margins of assembly traders ($4.07
was ranked number 19 in the world, compared to the Philippines’ per ton; see Table 3). In addition, the consolidation of trading made
ranking of number 121. These data (all from World Bank, 2007) sug- possible by wholesale markets would reduce coordination prob-
gest that the business environment is more uncertain in the Philip- lems in rice transport and reduce fuel and labor costs per ton. Such
pines than in Thailand, and to the extent that Filipino businesses markets would also improve the credibility of traders’ weighing
must be compensated for these ‘‘costs of doing business”, returns scales (assuming effective government regulation) and reduce
to management will be higher than in Thailand. the opportunities for defrauding farmers. In Thailand, the first
paddy wholesale markets were organized by the government, but
Restrictions on foreign investment later private investors (often rice millers) came to dominate the
market. However, these privately owned markets are still
Finally, another possible reason is foreign investment regula- regulated by government, e.g., through unannounced inspections
tions. Before 2000, only citizens of the Philippines had the right of weighing scales.
to participate in the rice and corn industry. Republic Act 8762 re- Third, another possibility would be to further liberalize for-
pealed previous restrictions (e.g., Republic Act 3018) and allowed eign investment regulations by removing divestiture require-
foreign investors to own 100% equity in companies involved in rice ments. It is not clear that this is the obstacle that discourages
milling and trading (other than retail trade). Even with the new foreign investment in rice milling, but removing this barrier
more liberal regulations, however, non-Filipinos must divest 60% might bring some benefits. Such foreign investment could help
of the equity in any such company (not necessarily to a single to substitute for a rudimentary capital market. The entry of
investor) after 30 years. It is not clear to what extent this divesti- foreign capital would likely be facilitated by the innovation of
ture requirement hinders foreign investment. To our knowledge, paddy wholesale markets that would make it easier for outsiders
we are unaware of any foreign investors in the rice milling and without local contacts to procure large volumes of paddy for
trading industry in the Philippines, but that may be due to factors milling. A more supportive business environment in general
other than government regulations. In the near future, investors would also be helpful.
from fellow ASEAN countries (of which Thailand is one) may be ex- Fourth, a more open policy as regards rice imports would lower
empt from this regulation and be treated in essence as Filipino working capital requirements, reduce storage time, and reduce
nationals. Certainly, it would seem that Thai millers could increase land values, all of which would lower marketing costs. Fifth, im-
their profits markedly by operating in the Philippines. In order to proved road infrastructure would also reduce marketing costs.
do so successfully, they would probably need to partner with Fili- As a final note, despite looking hard, we could find no evidence
pino investors to take advantage of local contacts while using their that collusion plays an important role in the wider Philippine mar-
access to Thai capital markets. Alternatively, paddy wholesale mar- gins. Thus, enactment of anti-trust regulations seems unlikely to
kets would make it easier for foreign millers to procure local paddy create narrower margins.
supplies without having to rely on local contacts.
Acknowledgements
Conclusions
The authors would like to gratefully acknowledge the com-
The main contribution of this paper has been to document, ments and suggestions received from the editor and anonymous
through analysis of a specific case study, that marketing systems referees, which greatly helped to improve the paper.
with a competitive structure may not generate competitive out-
comes, at least when the outcomes are compared with those in References
other countries. Some analysts may find the conclusion that ‘‘ex-
Banerji, A., Meenakshi, J.V., 2004. Buyer collusion and efficiency of government
cess profits” are not competed away is difficult to believe. But we intervention in wheat markets in Northern India: an asymmetric structural
went to great lengths to collect high quality data and to cross check auctions analysis. Am. J. Agr. Econ. 86 (1), 236–253.
our data in many different ways, and our analysis clearly points to BAS (Bureau of Agricultural Statistics), 2005. Rice and corn situation and outlook,
various issues. Available from: <http://bas.gov.ph>.
very high ‘‘excess profits.” Other analysts will assert that the lack of BOT (Bank of Thailand), 2005. Real sector data. Available from: <http://
a competitive outcome is because of the market failures that per- www.bot.or.th/bothomepage/databank/EconData/EconFinance/index04e.htm>.
vade all developing countries, and is so common that it is hardly Cabling, J., 2002. Market Structure, Conduct and Performance of the Rice Milling and
Trading Industries in the Philippines, unpublished M.S. Thesis. University of the
worth noting. But this study has shown that many of the standard
Philippines at Los Banos, College, Laguna, Philippines.
market failures (e.g., risk, collusion) that are often used to augment Ellis, F., Magrath, P., Trotter, B., 1991. Indonesia Rice Marketing Study 1989–91.
models of perfect competition are not relevant, at least in this par- Badan Urusan Logistik, Jakarta, Indonesia.
ticular case.20 In turn this suggests that there is much that remains FAO, 2006. FAO Stat on-line electronic database. Available from: <http://
faostat.fao.org/faostat/collections?subset=agriculture>.
Fafchamps, M., Gabre-Madhin, E., 2004. Agricultural markets in Benin and Malawi:
operation and performance of traders. Manuscript available from: <http://
20 www.csae.ox.ac.uk/reports/pdfs/benmal-traders-rep2001-04.pdf>.
But see Fafchamps and Gabre-Madhin (2004) for another example.
Author's personal copy

D.C. Dawe et al. / Food Policy 33 (2008) 455–463 463

Harriss, B., 1981. Transitional Trade and Rural Development. Vikas Publishers, New of Intensively Irrigated Rice in Asia. SSD Discussion Paper Series No. 02-01.
Delhi. Social Sciences Division, International Rice Research Institute, Los Banos,
Hayami, Y., Kikuchi, M., Marciano, E.B., 1999. Middlemen and peasants in rice Laguna, Philippines.
marketing in the Philippines. Agr. Econ. 20 (2), 79–93. PhilRice-BAS (Philippine Rice Research Institute–Bureau of Agricultural
Houston Jr., C.O., 1953. Rice in the Philippine economy 1934–1950. J. East Asiatic Statistics), 2004. Rice Statistics Handbook, 1970–2002. Muñoz, Nueva Ecija,
Stud. 3 (1), 13–81. Philippines.
IMF (International Monetary Fund), 2005. International financial statistics. Ruttan, V.W., 1969. Agricultural product and factor markets in Southeast Asia. Econ.
Available from: <http://ifs.apdi.net/imf/>. Dev. Cult. Change 17, 501–519.
Mears, L.A., 1974. Rice Economy of the Philippines. University of the Philippine Timmer, C.P., 1973. Choice of technique on rice milling in Java. B. Indones. Econ.
Press, Quezon City, Philippines. Stud. 9, 57–76.
Mears, L.A., 1981. The New Rice Economy of Indonesia. Gadjah Mada University Trewin, R., Bosworth, M., Drysdale, P., 2000. Moving Japanese agriculture forward.
Press, Yogyakarta, Indonesia. Available from: <http://www.rirdc.gov.au/reports/GLC/00-176-Part2.pdf>.
Monke, E.A., Pearson, S.R., 1989. The Policy Analysis Matrix for Agricultural Wickizer, V.D., Bennett, M.K., 1941. Rice Economy of Monsoon Asia. Food Research
Development. Cornell University Press, Ithaca and London. Institute, Stanford University.
Moya, P.F., Dawe, D., Pabale, D., Tiongco, M., Chien, N.V., Devarajan, S., Djatiharti, A., World Bank, 2007. World Development Indicators. Available from: <http://
Lai, N.X., Niyomvit, L., Ping, H.X., Redondo, G., Wardana, P., 2002. The Economics www.worldbank.org>.

Das könnte Ihnen auch gefallen