How competitive are food crop markets in sub-Saharan Africa?
March 2016
Brian Dillon
Chelsey Dambro
University of Washington1
Abstract. During the structural adjustment era of the 1980s and 1990s, governments across subSaharan Africa generally withdrew from food markets in order to encourage entry by private
traders and foster competition. Since that time, the degree to which private food crop traders
actively compete and pass on price changes to farmers has been a central concern of
policymakers, donors, and researchers. In this paper we provide a critical review of the evidence
on that topic. We begin by developing an analytical framework to guide the review, and then
discuss the empirical evidence from four categories of literature. We have two main findings.
First, there is a paucity of empirical evidence on this question, given the importance of the topic.
This hinders our ability to draw strong conclusions. Second, that point notwithstanding, the
evidence that does exist is broadly supportive of the notion that food markets are competitive.
The dominant themes in the literature are that trading profits are highly variable, which is
consistent with substantial risk, that trader entry and exit rates are high, and that price comovements between markets suggest relatively efficient levels of competitive arbitrage. It is
possible that the high costs of entry foster non-competitive conditions at the level of large-scale,
long-distance subnational trade; but we find no positive evidence to that effect, only the
satisfaction of certain necessary conditions.
1
Dillon is an Assistant Professor in the Evans School of Public Policy and Governance; email: [email protected].
Dambro is an international development consultant; email: [email protected]. We thank the Bill &
Melinda Gates Foundation for generously supporting this project. We are grateful to Jenny Aker, Sara Boettiger,
Marcel Fafchamps, Lauren Falcao, Thom Jayne, Craig McIntosh, Bart Minten, Bob Myers, Tom Reardon, and
seminar participants at the Bill and Melinda Gates Foundation for helpful discussions or comments on earlier drafts.
All errors are the sole responsibility of the authors.
1 Table of Contents
1. Introduction ...................................................................................................................................... 3 2. Background information ............................................................................................................... 8 2.1. Who is a crop trader? .............................................................................................................................. 8 2.2. Price determination in competitive markets ......................................................................................... 9 2.3. Methodology and reviewed papers ..................................................................................................... 11 3. Analytical framework .................................................................................................................. 13 3.1. Competitive markets and marketing margins .................................................................................... 13 3.1.1. Risk ..................................................................................................................................................................... 13 3.1.2. Unobserved trader costs ........................................................................................................................... 15 3.1.3. Unobserved trader services .................................................................................................................... 16 3.2. Non-competitive markets and high marketing margins .................................................................... 17 3.2.1. Natural barriers to entry .......................................................................................................................... 18 3.2.2. Barriers to entry imposed by current traders ................................................................................. 19 3.2.3. Barriers to entry due to regulation ...................................................................................................... 20 4. Empirical evidence ....................................................................................................................... 20 4.1. Group 1: Market price dispersion and co-movement of prices ....................................................... 21 4.1.1. Price co-­‐movements: empirical approach ........................................................................................ 21 4.1.2. Price co-­‐movements: evidence .............................................................................................................. 23 4.1.3. Price co-­‐movements: summary ............................................................................................................. 25 4.2. Group 2: Analysis of trader accounts ................................................................................................. 25 4.2.1. Analysis of trader accounts: empirical approach .......................................................................... 26 4.2.2. Analysis of trader accounts: evidence ................................................................................................ 27 4.2.3. Analysis of trader accounts: summary ............................................................................................... 30 4.3. Group 3: Other analyses with trader or farmer survey data ........................................................... 30 4.3.1. Measuring concentration: empirical approach ............................................................................... 31 4.3.2. Measuring concentration: evidence ..................................................................................................... 31 4.3.3. Measuring concentration: summary ................................................................................................... 32 4.3.4. Marketing margins and trader characteristics: empirical approach ..................................... 32 4.3.5. Marketing margins and trader characteristics: evidence .......................................................... 33 4.3.6. Marketing margins and trader characteristics: summary ......................................................... 33 4.3.7. Entry and exit: empirical approach ..................................................................................................... 33 4.3.8. Entry and exit: evidence ........................................................................................................................... 34 4.3.9. Entry and exit: summary .......................................................................................................................... 35 4.4. Group 4: Experiments and impact evaluations ................................................................................. 36 4.4.1. Impact evaluations: empirical approach ........................................................................................... 36 4.4.2. Impact evaluations: evidence ................................................................................................................. 37 4.4.3. Impact evaluations: summary ................................................................................................................ 38 5. Conclusion ...................................................................................................................................... 39 6. References ...................................................................................................................................... 41 7. Appendix ....................................................................................................................................... 46 7.1. Changes in equilibrium price spreads due to reductions in communication costs ....................... 46 7.2. Analyses of trader accounts ................................................................................................................. 46 7.3. Trader social capital ............................................................................................................................. 47 7.4. Inter-linkages ......................................................................................................................................... 47 7.5. Information impact evaluations ........................................................................................................... 48 2 1. Introduction
During the first two decades of the post-colonial period in sub-Saharan Africa (SSA),
governments and parastatal organizations were highly active participants in agricultural output
markets. Government involvement took a variety of forms, including the establishment of grades
and standards, restrictions on the timing and location of crop marketing, taxation of crop sales,
price controls, and direct control of the market through cooperatives, warehouses, or marketing
boards. In many settings, private traders also played a role in buying and selling crops, but they
generally operated at a small scale and with uncertain legal status. Entry by private traders was
limited by the market power of the central government and the need to remain small enough to
avoid substantial attention. The mix of public and private buyers of food crops varied between
countries, but government control was the rule, market competition the exception (Bates 1981,
Barrett 1997, Swinnen 1997).
In the 1980s and 1990s, the structure of crop output markets changed. The structural adjustment
reforms that swept through SSA in this period included as a key pillar the opening of crop
markets to private competition. The hope was that private traders would enter the agricultural
marketing sector to compete away excess rents and improve the operation of the price
mechanism. This would in turn allow farmers to respond directly to market signals emanating
from urban centers and international markets, leading to improvements in resource allocation and
overall increases in agricultural productivity. Free entry and market competition by private
traders were critical components of these reforms (Staatz et al. 1989).
It is not surprising, then, that over the last two decades the degree to which private crop traders
actively compete has been a central concern of policymakers, donors, and researchers. Many
observers assume that crop traders enjoy a privileged position in non-competitive markets, which
they use to extract rents from farmers. These claims are often based on anecdotal evidence and
on the general sense that farmers are exploited, rather than on solid empirical work to
demonstrate non-competitive pricing (Sitko and Jayne 2014). Politicians seeking to appeal to
their rural base often find traders to be a convenient scapegoat. Non-governmental organizations
and advocacy groups also frequently heap scorn on crop buyers, asserting that collusive traders
impoverish farmers by paying artificially low prices and extracting non-competitive rents. In
recent years, concerns about the efficiency and functioning of private crop markets have
motivated calls for cooperative marketing institutions to improve farmer bargaining power
(DFID 2004, Bernard et al. 2010), direct involvement in crop marketing by international
organizations such as the World Food Program through projects such as Purchase for Progress
(WFP 2015), and a return to heavy involvement by government marketing boards in some
settings (Mason and Myers 2013).
3 What does the evidence tell us about the competitiveness of food crop output markets in subSaharan Africa? The goal of this paper is to systematically review the economic research on that
question. We proceed in two steps. First, we develop a conceptual framework that lays out the
competitive and non-competitive forces that might increase the marketing margin – the
difference between the price at which a trader buys a crop and the price at which she sells it –
above the level expected by an observer. This is a theoretical exercise, intended to lay the
groundwork for the review and to highlight how difficult it can be to arrive at conclusive
evidence about market structure in this setting. Second, we divide the empirical evidence into
four main groups, making explicit connections to the conceptual framework in step one, and
review the findings in each group. Proceeding from the macro-level down to the level of
randomized, controlled trials, the empirical groups are as follows: analysis of price co-movement
between markets, detailed surveying of traders’ accounts, other studies involving data from
trader or farmer surveys, and impact evaluations based on experimental or quasi-experimental
variation.
Papers in the first empirical group rely on the intuitive notion that when there is sufficient
competitive arbitrage between two markets, the price spread between those markets should not
exceed the full cost of transferring crops from one to the other for any sustained period of time.
The tests associated with this theory can be implemented with price data alone, though data on
transfer costs is also helpful. Most of the papers in this group use data from wholesale markets,
which are not the primary point of contact for farmers, but are nevertheless informative for
understanding competitive structure along the supply chain. The large majority of findings in this
category are suggestive of substantial competition; price spreads rarely deviate from the
competitive level for long. In some settings, prices in markets that are far from each other do not
co-move, but this is usually because transport costs are so high that trade between the markets
has come to a temporary halt.
The second group of papers utilizes data from detailed surveys of traders’ accounts to estimate
the distribution of trader profits as accurately as possible. The primary takeaway from these
papers is that while some traders do appear to earn significant profits during some periods, the
average trader earnings are not out of line with expectations, given the required investments of
time and capital. Furthermore, a significant proportion of traders earn sustained losses, and exit
rates are high (over 17% per year in one study from Madagascar; Fafchamps et al. 2005). While
evidence of this nature is not definitive, as a group these papers give little reason to suspect that
traders are earning sustained non-competitive rents.
The third group of papers is diverse. We include in this group any of the less common or ad hoc
approaches to studying market competition that we found in the literature. One set of papers in
this group studies market concentration – the degree to which trading activity is channeled to a
4 small number of dominant players – and finds relatively low levels of concentration in most of
the markets studied. Another set of papers examines barriers to entry in trading or provides
empirical estimates of trader entry and exit rates. The consensus of these (albeit few) studies is
that entry at the first step in the supply chain, involving sales from farmers to traders, is low cost.
Small-scale aggregators can enter the market at will, if they can raise a little working capital.
However, entry at the large-scale or long-distance wholesale level often requires significant
investments in transport and storage capital. The inherent risks of large-scale trading and the
challenge of accessing sufficient credit can act as substantial barriers to entry. While this is not
evidence of non-competitive pricing at the level of sub-national trade, the findings do not
preclude that possibility.
The final group of empirical studies includes impact evaluations – primarily from natural
experiments or RCTs – that use the treatment effect in response to some intervention to make
inferences about the competitiveness of the market. All but one of these papers are focused on
information interventions that channel market price data to farmers under the hypothesis that
such data will either change the market at which they sell or improve their bargaining positions
vis-à-vis traders. Tests in this group can be understood as joint tests of the presence of trader
rents, the salience and relevance of the information provided, and the effect of better information
on the farmer’s bargaining position. The evidence in this group is mixed, but leans toward null
results.2 Some studies show small but significant effects of information on farm-gate prices for
some crops; others do not. The only true RCT in the group shows a small effect in the first study
year, which disappears in the second year. While we are open to the possibility that traders in
some settings use an information asymmetry to extract rents from farmers, the evidence reviewed
in this section provides little support for the idea that crop markets in SSA are rife with such
exploitation.
Looking across all of the findings reviewed here, we have two main takeaways. The first is that
given the importance of the question, the evidence is remarkably thin. No strong conclusion can
be drawn from the research that has been conducted thus far. As we detail below, only a handful
of high quality trader surveys have been conducted during the last twenty years. And while there
are ways to study the competitiveness of markets without surveying traders, as we just outlined
and will detail further below, it is difficult to draw firm conclusions about the competitiveness of
these markets without more evidence from observations of traders’ accounts. Furthermore, this
review includes findings from only 13 countries in sub-Saharan Africa, spread over 20 years and
26 papers. Some of these 13 countries are scarcely represented.3 Thus, what we do know about
the competitiveness of crop trading markets in SSA is dwarfed by what we do not know.
2
Evidence from outside sub-Saharan Africa leans strongly toward null results. See Appendix and Aker et al. (2016).
The countries are Benin, Ethiopia, Democratic Republic of the Congo, Ghana, Kenya, Madagascar, Malawi,
Mozambique, Niger, Sierra Leone, Tanzania, Uganda, and Zambia.
3
5 However, this point notwithstanding, our second takeaway is that the theme running through the
research that we reviewed is that evidence consistent with competitive markets is commonplace,
while evidence of non-competitive pricing is rare and subject to important caveats. This is not to
say that traders would not price non-competitively if given the opportunity, and it does not
preclude the possibility that there are oligopsonies in crop trading that have not been detected.
We remain sympathetic to the concern that non-competitive pricing may be a systematic feature
of some markets. But the bulk of the existing evidence suggests that food crop output markets in
sub-Saharan Africa are generally competitive.
The scope of this paper is limited in a few important ways. We focus on papers published in the
last twenty years, as this is the period after most structural adjustment reforms were complete or
well under way. We also focus our review on food crops rather than cash crops, with the
occasional exception for papers that deal squarely with the topic at hand. Finally, we concentrate
as much as possible on the first step in the market supply chain, in which farmers sell crops to
traders. We do, however, include a number of papers that use data from wholesale markets where
traders may act as both buyers and sellers (see especially Section 4.1). These studies can give
insight into the overall competitiveness of the crop-trading sector, which has implications for
market efficiency and the effectiveness of the price mechanism in allocating resources even if it
does not tell us whether farmers sell their output into competitive local markets.
There are many challenges to studying the competitiveness of crop markets. We discuss these in
greater detail when we review each type of evidence. Two bear special mention now. The first is
that it is difficult to conduct a survey with a representative sample of traders. Some traders
operate seasonally, so that a cross-sectional survey conducted at one point in time will omit
members of an important subgroup. Also, many traders are wary of opening their books to
outsiders. This tendency toward secrecy is in part a business decision; why discuss the details of
your operation with an enumerator who can provide little in return? In other cases an aversion to
survey participation may be due to a history of suspicion and discrimination against traders. This
is especially relevant in countries where the crop-trading sector is dominated by members of
ethnic minorities or non-native immigrants who have been persecuted by the government in the
past (Barrett 1994; Barrett 1997; Fafchamps and Gabre-Madhin, 2006; Sitko and Jayne 2014).
Most of the papers in this review that involve trader surveys make an effort to mitigate these
biases by including a large and varied sample of traders; nevertheless, we can never be sure
whether important subgroups are underrepresented.
The second major challenge is that many of the empirical tests that are available to researchers
cannot distinguish between non-competitive marketing margins and unobserved trader costs.
This theme emerges regularly in our review. Trading is risky, and traders face an array of costs
that are difficult to detect and measure in surveys, such as the true opportunity costs of time and
capital. Market inter-linkages, in which agents trade more than one good or service through a
6 linked set of transactions – for example, providing credit to a farmer from whom one will later
buy crops – can make it difficult to accurately measure the benefits and costs to each party in a
single transaction. Traders may provide multiple services, or just the option of multiple services,
at a cost that is difficult to measure and to value without carefully tailored research. Some papers
that we reviewed are careful to describe the assumptions about costs needed in order to make
assertions about the competitiveness of the market. Nevertheless, without a way to test or relax
those assumptions, it is difficult to find positive evidence of non-competitive pricing that is not
also consistent with the presence of unobserved costs.
What does this review imply for policymakers and researchers? For policy, perhaps the most
important takeaway is that programs designed to “fix” broken food crop markets should be
undertaken judiciously. Research provides little support for the promotion of cooperative
marketing or parastatal marketing boards in the current environment, at least as forms of
competition policy to protect farmers from exploitation. Improvements to public goods –
especially transport and communications infrastructure – seem to be more promising avenues to
reduce price spreads between markets. Such investments have potential benefits for both
producers and consumers.
A related takeaway for the development community is that continually promoting the idea that
traders in Africa are exploitative is neither helpful nor supported by evidence. As stated above,
we remain open to the possibility that some food crop markets are non-competitive; we are not
suggesting that anecdotal evidence of excessive mark-ups or trader rents is always wrong. But
we find little support for the idea that excessive rents are widespread. In the absence of broader
evidence, exploitative traders earning consistent excessive rents seem to be the exception rather
than the rule.
For researchers, the paucity of solid empirical work on this topic suggests an ongoing need for
more surveys of traders. Also, careful work and possibly new approaches are needed for
distinguishing unobserved transfer costs from non-competitive mark-ups. These components of
producer prices co-move in a manner that is exceedingly difficult to disentangle without tailored
data collection.
The remainder of the paper is organized as follows. In Section 2, we provide definitions and
background information about the scope of this review. In Section 3, we develop a theoretical
framework for understanding the competitive and non-competitive forces that determine trader
margins. In Section 4, we review the empirical evidence, making explicit connections between
the empirical literature and the analytical framework of Section 3. In Section 5, we conclude
with a brief discussion.
7 2. Background information
The definition of “crop trader” is usually context-specific and is not always made explicit in the
policy and academic literature. This section begins with an explanation of how we use the term.
We then outline the core intuition underlying the theory of price determination in a perfectly
competitive market, which serves as a benchmark for our analytical framework. The section
concludes with a brief description of our methodology and a summary of the papers covered by
this review.
2.1. Who is a crop trader?
Crop traders go by many names in the research and policy literature: aggregators, bundlers,
wholesalers, assemblers, middlemen, intermediaries, even the pejorative “briefcase
businessmen.” The meaning of these terms can vary between settings. One of the challenges in
writing about traders in SSA food markets is that supply chains often involve actors who play
multiple roles and may only occasionally engage in the buying or selling of food crops. Without
complete specialization, “traders” cannot always be clearly distinguished from farmers, retailers,
processors, transporters or others for whom trading is not the primary role in the food system.
Distinguishing among market actors is often difficult within a particular marketing chain, and
even more challenging when making comparisons across regions or countries. Some of the
papers we reviewed define traders in a way that includes a wide range of actors; others use a
narrower definition.
To fix ideas, the key, unifying characteristic of traders in this paper is that they buy agricultural
outputs for the purpose of profiting by selling those outputs at another market or at the same
market in the future. Beyond that, the notion of who is a trader varies in both the scope and type
of activities connoted. In some of the studies we review, most traders purchase crops at the farmgate, i.e., at the field or possibly at the house of the farmer. Other traders establish a buying
station at the village square during harvest periods. Yet others work through weekly markets,
traveling from one location to another to buy and sell whatever crops are available. Some traders
are seasonally active, working only when the dominant crops in their area are harvested and
marketed, while others work year round and trade a number of different crops.
We found that most of the traders discussed in the literature can be placed into one of two
categories. The first consists of local traders who buy crops from farmers in their own village or
nearby villages. These traders are often business owners or wealthy farmers, but they may also
be small-scale entrepreneurs with only a small amount of working capital. In many instances,
the trader’s objective is to aggregate the crops produced by his or her neighbors so as to reach a
threshold after which it is profitable to incur the cost of transporting the crop to a central market,
where it commands a higher price than at the farm-gate. In other cases, the trader purchases from
8 producers with the aim of re-selling directly to consumers in nearby markets. Local traders are
often seasonally active and do not travel extensively to buy crops. They also lack physical and
human capital; many do not own motor vehicles or storage facilities (Barrett, 1997; Fafchamps
and Gabre-Madhin, 2006; Gabre-Madhin, 2001).
The second group consists of professional traders who may bring a truck or cart to the farm-gate
or the village center and buy grain from farmers, or who purchase large quantities at wholesale
markets. These traders vary in size and sophistication. The least professionalized bear a close
resemblance to the local traders in the first group, operating primarily during the harvest and
marketing period and selling grain at a larger market within a day or two of purchasing. The
largest professional traders may own multiple trucks, employ a sizeable staff, manage a storage
facility to hold crops for weeks or months, and remain in operation throughout the year (Barrett,
1997; Fafchamps et al. 2005; Fafchamps and Hill, 2008). These traders are most likely to buy
from farmers and sell into wholesale markets, or to purchase in one wholesale market and re-sell
in another.
It is not uncommon in SSA for larger trading or processing companies to send agents into the
countryside to buy crops on commission. Often times these representatives are drawn from the
local community. These agents are technically traders, by our broad definition, even if they are
acting on behalf of a larger entity. However, this model of crop purchasing is less relevant for
this paper, because it is most common in the markets for cash crops and non-food export crops,
which are outside our scope.
2.2. Price determination in competitive markets
Traders earn revenue by purchasing crops at one price and selling them at a higher price. The
difference between these prices is the gross marketing margin, the trader’s revenue per unit
traded.4 The marketing margin is often expressed as the percentage mark-up on the price that the
trader pays to buy the crop. The net marketing margin is equal to the gross marketing margin
minus the trader’s costs per unit traded, i.e. the profit per unit, expressed in either level or
percentage terms.
There are surprisingly few reliable estimates of trader marketing margins in the research
literature. We found only five papers that provided estimates of margins or related metrics based
on a comprehensive analysis of trader balance sheets. In Table 1, we list the estimated marketing
margins from these papers. Estimates refer to net marketing margins, except for those from
Fafchamps and Gabre-Madhin (2006), which are estimates of annual profits. Average net
4
In some papers – though not those reviewed here – the term “marketing margin” refers to the difference between
the price that consumers pay and the price that producers receive for a good. This would be equivalent to the sum of
gross marketing margins for all traders in a particular supply chain, by our definition.
9 margins range from below 5% in Ethiopia to 37% in Malawi, with substantial variation. Direct
comparisons between studies are subject to the caveat that the net margin is not defined in a
standardized way, so that some of the variation is likely due to between-survey variation in
measurement of costs. Furthermore, these figures likely overestimate true net margins, because
every study excludes some components of trader costs. Missing from most estimates are the
value of unpaid time by family members, the opportunity cost of some or all forms of trader
capital (such as storage facilities at the trader’s home), the value of overtime work by the trader
herself, and likely other unobserved costs discussed further in Section 3. The estimate from
Minten and Kyle (1999) is explicit about this: net marketing margins in their paper account for
direct transport costs, but no other trader costs.
The vitriol directed at traders by politicians and outside observers is usually motivated by the
perception that marketing margins are too high. What, then, is a benchmark for a fair trader
margin? There is no single answer to that question. For reasons that we will detail in Section 3,
the costs of doing business in rural SSA can be substantial, and are often difficult to measure.
Marketing margins from other parts of the world are at best a lower bound on the margins that
we should expect to see in this region. Likewise, the margin from one market or crop is not
always a reliable guide for the expected margin in another setting, even within the same country,
because of variation in value-to-weight ratios, perishability, price volatility, market structure, and
other factors.
Table 1. Marketing margins from reviewed papers
Paper
Country
Timeframe
Average Margins
Dessalegn et al. (1998)
Ethiopia
8-month average
8.2%
Minten & Kyle (1999)
DR Congo
Monthly (Oct/Nov)
34.3%
Gabre-Madhin (2001)
Ethiopia
Annual
<5%
Fafchamps et al. (2005)
Benin
Madagascar
Annual
Annual
11%; Median 8%
27%; Median 11%
Malawi
Annual
37%; Median 27%
Benin
Annual
6% (profit)
Malawi
Annual
14% (profit)
Fafchamps & Gabre-Madhin
(2006)
However, theory does provide guidance in this area. A key characteristic of competitive markets
is that no agent can exert significant influence on the price. At the extreme, agents are pure pricetakers: all traders are small relative to the size of the market, revenue is constrained by the scale
at which it is profitable to operate, and the profits earned by an individual trader are equal to the
10 market return on his investments of time and capital.5 Under these circumstances, the price paid
by traders to farmers is exactly equal to the marginal revenue earned by all buyers (termed
“marginal revenue pricing”).
When competition is limited on the demand side, so that traders have some degree of market
power in their dealings with farmers, deviations from perfect competition can manifest in two
ways. The first, less extreme circumstance is one in which traders can exert some control over
the price they offer, but cannot price discriminate. That is, they must offer the same price to all
sellers. Under these circumstances traders may artificially reduce the quantity that they purchase
in order to offer lower prices, thereby attracting only the most motivated sellers from whom they
squeeze some extra profits. In the second type of non-competitive market, traders can offer
different prices to different sellers. This could be the case if traders purchase at the farm-gate and
negotiate with each farmer individually. In this situation, if the trader can infer the lowest price
that the farmer would agree to, she or he can typically enjoy an even greater average marketing
margin than under the case without price discrimination. Trading volumes are higher than under
non-discrimination, but farmers earn less per unit sold. In either of these non-competitive cases,
traders may earn “non-competitive rents,” i.e., profits greater than those consistent with perfect
competition.
For our review this distinction between different types of non-competitive pricing is not
especially relevant. It is difficult enough to determine whether markets are competitive without
also distinguishing between different types of non-competitive behavior. We mention the
distinction because it is useful for understanding why so many observers believe that traders are
exploitative. Most crop purchases are cash-and-carry transactions, and many happen in private at
the farm-gate. The impression that well-informed traders take advantage of hapless farmers
through price discrimination is difficult to resist.
2.3. Methodology and reviewed papers
The papers in this review were found by following the citation threads that originated in seminal
works on market competition following structural adjustment. We also conducted keyword
searches in Google Scholar to find works that may be missing from the citation thread. We
limited our review to papers published or forthcoming in academic journals, books from
reputable presses, and papers from high quality working paper series or research institutions.
5
This is the “zero profit” condition. A trader’s profits do not need to be mathematically zero for this condition to be
satisfied; the “zero” refers to “zero excess profits” above the market return earned by the trader. She or he must not
earn profits above those warranted by her investments of time and capital, valued at market prices.
11 Table 2. Summary of main papers in the review
Authors
Year
Country
Type of measurement
1
Barrett
1997
Madagascar
Inter-seasonal entry and exit
2
Badiane, Shively
1998
Ghana
Price co-movement between markets
3
Dessalegn, Jayne, Shaffer
1998
Ethiopia
Surveys of traders; CR4
4
Minten and Kyle
1999
DR Congo
Surveys of farmers
5
Abdulai
2000
Ghana
Price co-movement between markets
6
Gabre-Madhin
2001
Ethiopia
Surveys of traders
7
Fafchamps, Minten
2002
Madagascar
Marketing margins and social capital
8
Fafchamps, GabreMadhin, Minten
2005
Benin, Malawi,
Madagascar
Surveys of traders
9
Osborne
2005
Ethiopia
Structural model of expected prices
10
Tostão, Brorsen
2005
Mozambique
Price co-movement between markets
11
Fafchamps, Gabre-Madhin
2006
Benin, Malawi
Surveys of traders
12
Van Campenhout
2007
Tanzania
Price co-movement between markets
13
Fafchamps, Hill
2008
Uganda
Inter-seasonal entry and exit
14
Moser, Barrett, Minten
2009
Madagascar
Price co-movement between markets
15
Muto, Yamano
2009
Uganda
Impact eval: farmer information
16
Svensson, Yanagizawa
2009
Uganda
Impact eval: farmer information
17
Aker
2010
Niger
Concentration ratio (CR4)
18
Chamberlin, Jayne
2013
Kenya
Surveys of farmers
19
Myers
2013
Malawi
Price co-movement between markets
20
Casaburi, Glennerster, Suri
2013
Sierra Leone
Price impacts of road-building program
21
Casaburi, Reed
2014
Sierra Leone
Impact eval: wholesale price premia
22
Courtois, Subervie
2014
Ghana
Impact eval: farmer information
23
Sitko, Jayne
2014
Kenya, Malawi,
Mozambique, Zambia
Surveys of farmers and traders
24
Minten, Stifel, Tamru
2014
Ethiopia
Price co-movements between markets
25
Hildebrandt, Nyarko,
Romagnoli, Soldani
2015
Ghana
Impact eval: farmer information
26
Minten, Tamru, Engida,
Kuma
2015
Ethiopia
Surveys of farmers and traders
We found a surprisingly small number of empirical papers that include direct evidence on the
competitiveness of food crop markets in sub-Saharan Africa. The main papers in our review are
listed in Table 2. We reviewed roughly 25 additional papers that are not included in the main text
12 of this paper, either because they are from other regions of the world or do not deal with major
food crops. These are discussed in the Appendix.
3. Analytical framework
Suppose an outside observer believes that traders in a particular market are earning suspiciously
high marketing margins, taking into account the volume of trade and time period over which the
margins are earned. What might explain this observation? What are the categories of competitive
and non-competitive forces that could lead to “high” marketing margins? The goal of this section
is to categorize and describe the factors that influence marketing margins. We discuss
competitive markets first, followed by non-competitive markets. Our focus is on the mechanisms
that are most important in SSA, but the exposition is general enough to be applicable elsewhere.
Also, we are interested only in the factors that might drive marketing margins above the value
expected by an observer; we omit the standard set of observable supply and demand
characteristics that determine the equilibrium marketing margin in a competitive market.
The discussion in this section is theoretical. Our goal is to describe the various forces that might
be responsible for higher-than-expected marketing margins, not to assert that a particular factor
is relevant in any specific setting. In Section 4, we will match the theoretical framework of this
section to the empirical tests in the literature. However, we will see that some of the hypotheses
advanced in this section are not easily testable. In this regard, some of the ideas in this section
represent alternative ways to interpret empirical findings that might otherwise be considered
evidence of non-competitive pricing.
3.1. Competitive markets and marketing margins
Let us assume for the moment that food crop markets are competitive. There are numerous
reasons that a trader might earn what appear to be non-competitive rents even though the crop
market is fully competitive. We group these into three categories, shown in Figure 1. First,
traders might be absorbing or aggregating substantial risk in a manner that is not understood by
observers (category A). Second, traders might have costs that are unobserved by the researcher
(category B). And third, traders may provide services that are unobserved by the researcher
(category C). Category A is in essence a case of misspecification – the observer has not fully
understood the problem being solved by agents on the ground. The category B and C cases are
instances of measurement error with respect to the costs incurred by traders or the benefits
received by farmers, respectively.
3.1.1. Risk
13 Just as farmers in sub-Saharan Africa can face significant uncertainty during cultivation, so too
can traders face substantial uncertainty as they buy and sell crops. What are the risks that traders
face? Consider first those that pertain to a single crop in a single season. Short-term price
fluctuations may present a significant source of risk. Private traders overwhelmingly engage in
cash-and-carry trade, paying farmers in full at the time of exchange.6 A trader may hold crops for
days, weeks, or occasionally months, before re-selling at a larger market. During that period the
trader is exposed to any short-term price movements, which can be substantial and unpredictable,
dependent as they are on aggregate supply and demand factors that are not fully observable to the
trader. The greater the downside price risk, the lower the farm-gate price needed to induce
traders into the market (other things equal).
High marketing margins in competitive markets A Traders absorb or aggregate risk A1 Spoilage or theft A2 Quality risk A3 Price risk within-­‐
season A4 Inter-­‐seasonal variation in profits B Traders have unobserved costs B1 Fixed costs B2 Opportunity costs of time, capital B3 Hidden costs: bribes, brokerage fees, regulatory compliance C Traders provide unobserved services C1 Extension services C2 Information C3 Social capital / option of future services Figure 1. Factors that increase marketing margins in competitive markets
Uncertain crop quality represents another source of risk absorbed by traders. If traders cannot
observe crop quality at the farm-gate as easily as it can be tested at wholesale markets, they bear
the risk of any markdowns related to high moisture content, broken grains, dirt or sand in bags,
6
Contrast this with the era of government warehousing and parastatal trading, when farmers were rarely paid in full
before the parastatal intermediary sold crops onward to a wholesale market. This could take months or even years,
and farmers were not assured of the price that they would receive until final payments were disbursed (Bates, 1981).
14 or other sources of quality variation. Crop quality can also degrade during transition. Traders
bear the risk of spoilage, theft, or destruction by pests that occur between purchase and re-sale.
Traders also face higher-level risks related to the fundamental ebb and flow of business across
crops and seasons. Many traders are engaged in the buying and selling of food crops over
multiple years. Because the trader draws from a distribution of profits across space and time,
high margins in one crop-year may be, in essence, the trader’s reward for operating in many
other markets and years with thin or negative margins. Observers who are keen to detect excess
profits by traders may focus on those margins that appear to be egregiously large. Yet if there are
significant fixed costs to trading, especially at the macro-level where physical capital needs are
not trivial (trucks, storage facilities), then trading is profitable precisely because marketing
margins for some crops in some markets are higher than average. The variation in and
unpredictability of the margins earned in any given market can be a source of substantial longrun risk to the business of crop trading in SSA.
3.1.2. Unobserved trader costs
A second rationale for seemingly high marketing margins in competitive markets is that traders
may have costs that are unobserved to the researcher. First among these are the fixed costs of
operating a trading business. As mentioned in the previous subsection, traders may purchase and
maintain trucks, carts, and storage facilities, as well as other tools of the trade such as scales and
bags. Many fixed costs are both difficult to observe and difficult to value appropriately, as the
stream of services that flow from a physical asset may depreciate at an unobserved rate or be
used for multiple activities. Matters are complicated further if fixed costs are wholly or partially
recoverable. Furthermore, when traders incur fixed costs, the marketing margin consistent with
perfect competition is a function of the volume of trade undertaken. Because fixed costs must be
recovered over the course of many transactions, the competitive marketing margin must increase
as the volume of trade falls (other things equal). To the extent that the volume of trade may be
difficult to estimate precisely, a cautious trader might assume a low volume of trade at the start
of the season and therefore offer a lower initial price to farmers, in the interest of ensuring the
recovery of fixed costs.
The opportunity cost of the trader’s time and investment capital represent a potentially
significant source of additional cost that is sometimes not appreciated by observers. It is costly
for a trader to tie up his or her capital in food crops whose value may fluctuate. Failures or
imperfections in markets other than crop markets, such as those for credit or information, can be
especially important in this regard. If there are shortcomings in those markets, then a high
margin in the food crop market may be competitive conditional on the high cost of credit and
information.
15 Relatedly, there may be market inter-linkages that involve traders advancing credit to farmers
during planting or cultivation, then recovering the principal plus interest at harvest. While these
inter-linkages are less common for food crops than cash crops, they remain a potentially
important unobserved aspect of the trader-farmer relationship in some settings. A high marketing
margin may be in part due to debt repayment that is folded into the transaction price. In high
inflation countries, the equilibrium interest rate on this debt may appear large to outsiders, yet
could still be consistent with interest rates paid to microfinance institutions or savings groups.
Traders may also incur hidden costs that they are hesitant to reveal to outsiders, but which are
nonetheless relevant to the marketing margin that they actually earn. These costs can include
bribes to traffic police, bribes to customs officials, or side payments to village officials to allow
the trader to operate.
Finally, there are miscellaneous additional costs that only the most careful survey will detect,
such as brokerage fees, service on debt, and maintenance. With poorly constructed roads,
maintenance expenditures for transport capital may be high. Some of these costs may not be
wholly salient at the time of a trader interview and thus be subject to under-reporting.
If traders incur any of these costs in a way that is unobserved, marketing margins calculated on
the set of observed costs will overstate actual margins.
3.1.3. Unobserved trader services
Traders can provide other services to farmers that incur zero marginal cost to the trader but are
nonetheless valuable to the farmer. For example, information about new varieties and market
demand for certain crops is often communicated at the time of a transaction. Traders are more
aware than farmers of which crops are being bought and sold in which markets, and of the price
premiums associated with particular varietals. Such information can be of high value to farmers,
even if communicated at the time of harvest and so only relevant for the following season. If
trading involves the exchange of such valuable information, the farmer’s acceptance of a lower
output price may include a payment for information.
Similarly, small traders are often the first buyers to arrive at harvest time (Sitko and Jayne 2014).
Farming households who are liquidity constrained and desperate for cash can sell immediately to
these traders, without waiting for the buying season to pick up. Such sales often take place
directly at the farm-gate, requiring minimal transport on the part of the farmer. This flexibility in
the timing and location of sales is a valuable service that some farmers are willing to purchase in
the form of slightly lower output prices.
16 Finally, in some settings, the connection developed between farmers and traders can be viewed
as a form of social capital for which farmers are willing to pay. Traders are an important link
with larger markets and can play a critical role in channeling the services just discussed to
farmers. A farmer who develops a relationship with a trader may be willing to accept a slightly
lower output price because s/he knows that she is simultaneously purchasing an option to call
upon the trader for other services in future years. Such relationships are often not easy for
observers to measure or value in a quantitative survey.
3.2. Non-competitive markets and high marketing margins
We now turn to the non-competitive forces that could drive marketing margins above the level
that is believed to be consistent with competition. A common reaction to high marketing margins
is to assume that traders are exploiting farmers and earning monopsony or oligopsony rents. And
while many traders are embedded in the community and face the possibility of social sanction if
they try to squeeze farmers too much, traders are still businesspeople. It is not unreasonable to
expect them to increase profits when possible, in the manner taken for granted by many
policymakers and farmer advocates.
Let us assume for the moment that food crop markets are characterized by non-competitive
pricing. These are markets, then, in which marginal revenue pricing does not predominate and
traders earn non-competitive rents. Non-competitive pricing by traders can only be sustained if
there is a barrier to entry into crop trading. If entry is free, but no one chooses to enter the
market and compete away the current set of margins that traders receive, then it is difficult to
argue that the market is non-competitive or that entry would be profitable.7 In such a case, the
equilibrium marketing margins are reflective of the underlying fundamentals of the economy. Of
course, under such circumstances there may still be a case on social welfare or political grounds
for intervening to increase the prices received by farmers or to decrease the food prices paid by
end consumers. But such policies would not be motivated by a lack of competition in output
markets.
In this section, we describe three types of barriers to entry that might underlie a non-competitive
crop-trading sector. The first are “natural” barriers (category D); the second, barriers due to the
specific activities of current traders (category E); the third, barriers due to policy or regulation
(category F). Figure 2 provides the taxonomy. As in the previous section on competitive markets,
we are not asserting that these classes of explanations apply in any particular market. Rather, we
are describing the landscape of non-competitive forces that might be relevant. Later we will
match the empirical tests in the literature to some of the specific hypotheses about market
function that we develop here.
7
The exception may be a case in which the threat of retaliation by incumbent firms is sufficient to deter entrance,
even though entrance would be profitable under current conditions. See section 3.2.2.
17 3.2.1. Natural barriers to entry
Farmers in SSA are spatially dispersed and served by a patchwork of often poor, sometimes
purely seasonal roads. In years with bad weather and low output, farmers may have little
marketable surplus to sell to traders. If the fixed costs of doing business in any particular village
or remote market are high, because of bad roads, uncertain trading volume, or other conditions,
then that market may only be large enough to support profitable trading by a single trader or a
small number of traders. This is the case of a natural monopsony or natural oligopsony in
trading.8 The traders who do serve this market may enjoy some market power, and underpay
farmers as a result of the lack of competition in the market. In this case, entry is precluded not by
the actions of the traders, but by the high fixed cost of traveling to the market to compete with
currently active traders.
A related, important barrier to entry may be the scarcity of physical assets for certain types of
arbitrage, e.g., long-haul trucks or large storage facilities (Dessalegn et al. 1998; Abdulai 2000).
If incumbent traders have preferential access to transport or storage capital, and the creation or
import of new capital is hampered by port congestion, lack of maintenance infrastructure, credit
market imperfections, or the overall risk in the sector, this can limit the number of traders who
can feasibly participate in certain links of the supply chain. These barriers may be especially
relevant at the level of long-distance trading between wholesale markets (Moser et al. 2009).
When there are natural barriers to entry such as these, government intervention is typically the
only way to improve outcomes. In a rural market with limited competition, regulation could in
theory increase the prices received by farmers above the non-competitive level. Yet in such
markets, the capacity of the state to enforce such regulations is weak or nonexistent. Thus,
transaction costs and non-competitive rents are intertwined in a manner that is difficult to
disentangle.9 In the case of long-distance arbitrage, relieving the bottlenecks to the creation and
maintenance of transport and storage capital for large-scale trading takes on some of the
characteristics of a public goods problem that can only be solved through government action.
As a final point, it is important to differentiate between markets with natural limits to
competition, like those relevant to this subsection, and markets in which trader costs, due
perhaps to information shortages, credit market failures, or scale economies, drive up the
equilibrium marketing margins earned by competitive firms (as in Section 3.1.2). The critical
factor is whether the costs of doing business are so prohibitively high that only a small number
8
One can easily re-frame this as the standard case of a natural monopoly by re-defining the market as one in which
traders are on the supply side rather than the demand side, and the good in question is a service: transport and access
to a larger market.
9
See Atkin and Donaldson (2015) for a treatment of this problem for a select subgroup of imported or processed
items.
18 of traders can operate in a given market. If trading costs are high but there are many traders, each
is small relative to the market, and marginal revenue pricing dominates, then the market is
competitive given the fundamental structure of the market, even if the equilibrium marketing
margin is worryingly large from the perspective of outsiders. If trading costs are high and that
reduces the number of traders to the point that those that are present can dominate the market and
earn sustained non-competitive rents, then the market is not competitive.
High marketing margins in non-­‐competitive markets D Natural barriers to entry E Barriers caused by incumbent traders F Barriers caused by regulation D1 Poor infrastructure; transport costs D2 Uncertain production volumes E1 Collusion and exclusion E2 Preferential access E3 Threat of short-­‐
term loss pricing F1 Licensing delays F2 Space constraints at formal markets Figure 2. Factors that increase marketing margins in non-competitive markets
3.2.2. Barriers to entry imposed by current traders
Even if there are no natural barriers to entry, current traders could act in a way that restricts the
capacity for new entrants to compete in a particular market. We found no evidence of active
collusion by traders, so we can only speculate about how current traders may actively work
together to create a barrier to entry. One possibility is that established traders who are connected
to village leaders are able to lobby for privileged access to storage facilities or permission to
begin purchasing crops at an earlier date than their rivals. Alternatively, large traders with truck
fleets and sizeable catchment areas may coordinate with other large traders to divide up the
countryside, thereby reducing competition in any given village. Larger traders might also engage
in short-term loss pricing in some markets, to drive smaller competitors out of business. Even the
threat of these activities could act as a deterrent to entry. Finally, there may be situations in
which traders actively collude, by meeting to set prices before buying in particular villages, or
tacitly collude, by not competing aggressively on price. Such actions hamper the ability of
19 farmers to bargain for better prices. Farmers in such a setting must either accept the low price or
incur the cost of transportation to bring their output to a different market. Yet collusion is
difficult to maintain in equilibrium unless there is some other barrier to entry; otherwise, we
would expect entry by new traders to compete away the associated rents.
3.2.3. Barriers to entry due to regulation
Lastly, there may be barriers to entry in food crop markets because of a policy or regulation that
makes entry costly or burdensome. Examples of such restrictions include registration
requirements, licensing of vehicles, taxes collected at roadside checkpoints or weigh stations,
adherence to standards, or payments for inspections. However, these regulations should only
create barriers to entry if at least one of two other circumstances obtains. The first is one in
which regulation creates long delays in licensing or authority to trade, which have the effect of
artificially limiting the number of traders. This is unlikely to be the case in rural markets, where
regulatory capacity is weak; perhaps at ports or at the level of large-scale storage and
warehousing, such delays may be relevant. The second is one in which regulatory costs are so
excessive that they create what is effectively a natural barrier to entry by sharply reducing the
number of firms that can profitably participate in a given market, as in Section 3.2.1.10
4. Empirical evidence
The framework in the previous section suggests various avenues for testing whether output
markets are competitive. In this section, we discuss the empirical evidence that has been
generated on this question. We subdivide the literature into four groups according to the
empirical approach utilized, connect each group to the analytical framework, and review the
findings.
As mentioned above, some of the factors raised by the analytical framework of Section 3 have
not been directly tested in the literature. And there are few avenues available to researchers that
deal with all of the possible competitive and non-competitive mechanisms simultaneously. This
highlights one of the central challenges to learning about this topic: it is much easier to reject a
specific theory of non-competitive pricing than it is to simultaneously test all of the possible
factors that might increase marketing margins above the level expected by observers.
Nevertheless, progress is possible if we look at the evidence from many studies as a single body
of work that speaks to the general pattern in SSA food markets.
10
Although we have not heard of such cases, it is theoretically possible that competition in some rural markets
would be reduced due to space constraints and minor regulatory fees. Weekly markets are usually organized by a
local coordinator who charges a small fee to make use of a market stall. The number of market stalls may be limited
by the physical capacity of the market. However, in practice, it seems unlikely that these restrictions would limit the
number of active traders to the point that the market becomes non-competitive.
20 We divide the empirical literature into four groups, proceeding roughly from the macro-level
down to the level of natural or actual experiments. The first group includes studies that use
market level price data to test for co-movement of prices across space and/or time. The second
involves careful surveying of traders to measure and analyze the balance sheets of trading firms.
The third group is a catchall for empirical approaches that are not common enough to warrant
their own group, but which involve descriptive or regression-based evidence using trader- or
farmer-level data. Lastly, the fourth group encompasses impact evaluations based on randomized
controlled trials or natural experiments, almost all of which involve providing information to
farmers with the goal of increasing their bargaining power vis-à-vis traders.
4.1. Group 1: Market price dispersion and co-movement of prices
The first body of empirical evidence is based on the theory of spatial price equilibrium.11 We
describe the intuition behind this empirical approach in the following subsection, and then
review the evidence.
4.1.1. Price co-movements: empirical approach
Consider two markets, i and j, with crop prices pi and pj. Suppose that the per-unit cost of
moving the crop from one market to the other is given by τ. This is commonly referred to as the
“transfer cost” or “transport cost” (although technically the former term encompasses the
physical transport cost plus any other costs involved in moving goods between markets). If
markets are connected and competitive, then the price spread |pi – pj| cannot be maintained at a
level greater than τ in the medium- to long-term. When the spread does increase to a level greater
than τ, traders take advantage of the arbitrage opportunity by buying in one market and selling in
the other until the prices adjust. In our context, when there is sufficient competition, a trader who
fails to pass on to farmers a rising wholesale price will be out-competed by those who do. Thus,
beyond the very short run, robust competition ensures that |pi – pj| ≤ τ.
Increases in τ, due to seasonal road outages, high petrol costs, or other factors, can lead to market
segmentation. If the price spread is lower than the transfer cost then the prices are not expected
to co-move in the short run because it is not profitable for agents to arbitrage away the
difference. However, such a disruption in trade is not because of a lack of competition. The
markets are still integrated if traders are available to arbitrage away any price spreads that exceed
τ. This model of competitive arbitrage, in which there are “trading” and a “non-trading” states
11
See Ravallion (1986) for a classic treatment, and Baulch (1997), Fackler and Goodwin (2001), and Barrett (2001)
for exposition and discussions of the relevant theory and empirics. Also, see Myers and Jayne (2012) for an
application involving South Africa and Zambia that uses related methods, but is excluded from the review because
of its international focus.
21 between any two markets based on the price spread and the transfer cost, is commonly applied to
food markets in SSA.
The concept of spatial price equilibrium provides testable predictions for how prices in linked
markets should co-move when there are sufficient levels of competitive arbitrage. These tests do
not require micro-level data from traders or farmers. They are implementable with only time
series data on crop prices in potentially linked markets. Analyses that also incorporate data on
transport costs and/or trade flows, and that allow for possible breaks in trade when arbitrage is
not profitable, can produce more precise estimates.12 An important benefit of tests based on
spatial price equilibrium theory is that they typically rely on changes in prices to test the extent
of price pass-through. This mitigates, though does not eliminate, the influence of fixed costs on
equilibrium marketing margins.
How do these tests relate to the conceptual framework of the previous section? Empirical tests in
this category are usually interpreted as indirect tests of barriers to entry (categories D-F). The
null hypothesis is that markets are competitive. Barriers to entry, if they exist, reduce the number
of market participants and enable active traders to earn positive profits from arbitrage in the
medium- to long-term. If prices do not co-move in a manner consistent with competitive spatial
arbitrage, the hypothesis of competitive markets with no significant barriers to entry is rejected.
There are some important limitations to inference about competition based on studies of this
nature. The first is that these papers typically use data from larger wholesale markets, not from
farm-gates or from the smaller rural markets that directly serve farmers. Rarely do papers in this
group help us learn about competition at the very local level. Additionally, a finding of possible
non-competitive pricing using evidence of this nature does not allow the researcher to easily
distinguish between barriers to entry due to structural factors (D), collusion (E), or regulation
(F). Furthermore, variation in trader costs is not observable in market price data, so that the link
between the cost of trading in a particular market and the competitive structure of that market
cannot be studied. Rejection of competitive spatial equilibrium, then, is a necessary but not
sufficient condition for a finding of non-competitive pricing.
However, there are also statistical challenges that can bias studies of this nature toward or away
from a finding of perfect competition. Prices may co-move spuriously, because of shared
exposure to macroeconomic shocks, changes in energy prices, or other factors (Dillon and
Barrett 2016). Careful econometric work is needed to select the appropriate tests for and models
of a stationary long-run relationship between two prices – a cointegrating relationship – without
misinterpreting correlation due to shared macroeconomic shocks as competitive arbitrage.
Likewise, if trade is steady but discontinuous because of inter-temporal variation in prices and
12
See Barrett (1996) for a discussion of the typology of market level analyses of spatial price integration.
22 transfer costs, a well-functioning and competitive relationship between markets may be difficult
to detect using price data alone (Barrett 2001).
4.1.2. Price co-movements: evidence
In an early paper, Badiane and Shively (1998) use first generation empirical approaches to the
study of cointegration to test for spatial price integration between a major central maize market
and two local wholesale markets in Ghana. They employ monthly data from 1980-1993. They
find clear support for spatial integration. Adjustment is rapid, but not immediate: price shocks
typically transmit in full from the major market to the branch markets within four months.
Also in Ghana, Abdulai (2000) utilizes threshold cointegration analysis to study wholesale maize
market integration, comparing price transmission between Accra and two local markets. Abdulai
relaxes the implicit assumption of symmetry in Badiane and Shively (1998) by allowing for
possible asymmetric adjustment to price increases and decreases. Data are again monthly,
covering the period from 1980 to 1997. Findings indicate that price increases in central markets
are quickly passed on to local markets. Price decreases transmit more slowly. But all estimated
monthly transmission rates are above 34%, indicative of rapid adjustment to both increases and
decreases. Abdulai interprets the asymmetry in adjustment as evidence of menu costs or costly
inventory adjustment, rather than imperfect competition. Given the relatively short adjustment
periods in both directions, this interpretation seems warranted.
In Mozambique, Tostão and Brorsen (2005) test for integration between markets in the northern,
central, and southern regions during the period July 1994-April 2001. The question of market
efficiency is especially relevant in this setting, because southern Mozambique is a persistently
maize deficit region and there is no direct road to the maize producing areas in the north. In some
models, the authors use monthly average prices in combination with between-market transport
costs to estimate a parity bounds model (Baulch 1997). In other models, they use weekly price
data, with significant numbers of missing values, to estimate simple vector auto-regressions
without incorporating transport costs. The results of these two approaches are broadly consistent.
Within each region, there is significant trade and prices co-move in a manner consistent with
competitive arbitrage. Also, the southern and central regions are well integrated with each other.
The north, however, is effectively isolated from the other two regions, because the cost of
shipping grain from the north to the central or southern regions is usually higher than the price
spread between the markets. As the authors conclude, “The problem does not seem to be lack of
traders to ship grain from low price areas to high price areas. The problem is that transport costs
are so high that it is not profitable to ship the grain” (p.212).
Van Campenhout (2007) studies maize market integration in Tanzania. He uses a threshold
autoregressive (TAR) model to estimate the speed of adjustment of wholesale maize prices, from
23 1989 to 2000, in six markets. Major markets in northern Tanzania are not included in the paper,
but the six studied markets are spread widely across southern and central areas. The paper
includes an important methodological innovation for our context, in that it allows both
transaction costs and the estimated speed of adjustment parameters to change over time. Van
Campenhout finds that market performance has improved over time, with price differences
decaying by 50% in 1-5 weeks across market pairs. Some market pairs have slower price
adjustments possibly due to bad road conditions and police checkpoints. Overall, the findings are
consistent with rapid spatial arbitrage and a robust level of competition.
In Madagascar, Moser et al. (2009) use a modified version of the parity bounds model to study
rice market integration at the local, regional, and national level. The data differ from those used
in other papers in this section in that the time series dimension is short: prices are only available
for the four quarters of 2001. However, the data are highly disaggregated spatially, with
observations of both rice prices and transport costs for almost all of the country’s 1,394 districts.
The goal is to differentiate between three possible marketing regimes. The first is a competitive
trading equilibrium, in which the price difference between markets is equal to the transfer cost so
that there are no positive returns to arbitrage; the second is a segmented market where there are
price differentials between markets but transfer costs are too high to warrant arbitrage; and the
third is disequilibrium, or imperfect competition, in which traders are potentially earning positive
rents because the price spread exceeds the transport cost. When the authors do not allow for
unobserved variable transfer costs, they find that small sub-regional markets are usually well
integrated (regime 1), national markets are usually segmented by high transport costs (regime 2),
and regional markets are most often characterized by disequilibrium consistent with imperfect
competition (regime 3). However, when the assumption of zero unobserved transfer costs is
relaxed, findings suggest that at all three scales the most common state is competitive trading
equilibrium (regime 1). In sum, Moser et al. find what may be evidence of imperfect competition
at the regional level, but they demonstrate directly that they cannot distinguish possible imperfect
competition from unobserved trader costs. They speculate that if competition is limited at the
regional level, it is likely because of barriers to entry related to the physical capital requirements
of medium-distance trade, such as trucks and storage facilities.
Myers (2013) studies the speed of price adjustment between maize market pairs in Malawi. Data
are weekly, from July 2001 to October 2008, and cover 10 major markets. Using a model that
allows for different regimes based on the possibility of trade in either direction, Myers finds
rapid rates of adjustment. Half-lives for price spreads – the amount of time necessary for half of
a price spread to be arbitraged away – lie between 0.6 and 2.2 weeks, on par with those found in
the US (Goodwin and Piggott 2001). The conclusion is that maize market integration is
reasonably complete throughout the country, consistent with a finding of robust competition by
traders.
24 Casaburi et al. (2013) use the roll-out of a road construction program in Sierra Leone to study the
effect of lower transport costs on price spreads between markets. They bring together a variety of
data sources, including census data, farmer surveys, trader surveys, market price information,
and baseline and endline data on road construction and quality spanning the period 2003-2011.
The finding most relevant for this review is that price spreads fall with the improvement of roads
between markets, which the authors show to be consistent with a model of costly search, but not
with various models of imperfect competition.
Finally, Minten et al. (2014) use cereal market price data from the Ethiopia Grain Trade
Enterprise to study changes in price spreads between Ethiopian markets over the period 20012011. They use monthly price data from 66 markets and five crops: teff, barley, wheat, maize,
and sorghum. Using TAR models with empirical estimates of transport costs, they find that
markets in Ethiopia became significantly more integrated over the study period. The average
speed of adjustment between market pairs fell by 25% for white teff, 50% for mixed teff, 22%
for red teff, 45% for white wheat, 33% for maize, 11% for white sorghum, and 85% for mixed
barley. In 2011, the average half-life of adjustment was less than 2 months for all crops other
than white sorghum, and was less than 1 month for maize and white teff. Minten et al. use focus
groups with market actors to identify five factors that led to increased integration of cereal
markets in Ethiopia over the study period: economic growth, urbanization, road-building, access
to mobile phones, and improvements in agricultural extension and technology adoption.
4.1.3. Price co-movements: summary
Overall, recent studies using price co-movements have found that food markets in SSA are fairly
well integrated at the local and regional level. Likewise, we saw in two papers (Tostão and
Brorsen 2005; Moser et al. 2009) that market segmentation can occur at the national level when
transport costs exceed price spreads. These findings are consistent with a model of robust
competitive arbitrage. There is slightly less clarity at the level of medium-distance intra-national
trade, where relatively high transaction costs combined with extensive physical capital
requirements, e.g. for trucks and storage facilities, may dampen entry and potentially reduce
competition. Findings in Moser et al. (2009) leave open the possibility that medium-distance
trade in food crops is not fully competitive. But the evidence is far from conclusive, and the
authors acknowledge that unobserved transfer costs could be responsible for the observed price
patterns. While this literature remains subject to some important methodological caveats – e.g.,
many papers do not account for likely cointegrating relationships between market prices – the
overall message is one of substantial and widespread competitive arbitrage in wholesale food
markets, with little indication of non-competitive pricing or sustained rents to traders.
4.2. Group 2: Analysis of trader accounts
25 The second group of empirical tests involves detailed surveying of traders and analysis of their
balance sheets. This is direct, descriptive research.
4.2.1. Analysis of trader accounts: empirical approach
The goal of the studies in this section is to measure traders’ costs, revenues, and profits as
completely as possible, in order to sharpen the measure of marketing margins. This approach is
useful in two ways. First, if marketing margins that appear to be large are found to narrow
considerably upon detailed analysis of trader accounts, concerns over possible non-competitive
pricing can be mitigated. This requires that the observer have a prior about the marketing margin
against which the observed margins can be measured and classified as excessive or not.
Although there is no clear set of guidelines for determining such a prior, some progress can be
made by comparing the profitability of trading to the profitability of other activities requiring
similar levels of human and physical capital investment. More common is that the researcher and
reader must make judgment calls about whether the estimated margins appear to be greater than
merited by the need to cover the unobserved components of trader costs.
This first application of trader survey data relates most directly to categories B and C in Section
3. The null hypothesis, to the extent that one exists for descriptive evidence of this nature, is that
a high marketing margin reflects some form of non-competitive pricing. The alternative is that
marketing margins are much narrower than they appear at first, once the researcher has fully
accounted for the full set of costs to traders (category B) and benefits to farmers (category C).
A second application of the evidence from studying traders’ balance sheets relates to the
distribution of trader profits. If the sample of traders is large enough, the researcher can estimate
not only the average of trader earnings, but also its distribution. The variance and shape of this
distribution are proxies for the equilibrium level of risk in the trading sector. If traders earning
“high” marketing margins are found to operate alongside other traders who earn low or even
negative margins, this is an indicator of substantial risk in the sector. Under such circumstances,
the average margin must be great enough to compensate traders for the underlying uncertainty.
Use of trader accounts to examine the distribution of trader margins amounts to a test of the
theory associated with category A. As with the first test discussed in this subsection, there is no
fixed standard for relating the level of risk in the trading sector to the competitive marketing
margin. The researcher must either conduct extremely fine-tuned research to study that exact
issue (which has not been done), or interpret the data with respect to some prior about the riskreturn tradeoff.
Selection into survey participation is a potentially important problem for research of this kind. If
the traders who choose not to participate in surveys are those who are earning rents, then we may
underestimate the likelihood that the market is not competitive. On the other hand, the difficulty
26 of sampling intermittent or failed traders may lead us to underestimate the downside risk of
trading. Systematic measurement error is also of potential concern, as traders may be motivated
to disguise profits by overstating costs. In these regards, it is helpful that most of the surveys
discussed below were conducted by well-established research teams with the participation of
local partners. These papers represent the best efforts in the field to sample and interview a
generally (if not statistically) representative sample of traders.
4.2.2. Analysis of trader accounts: evidence
In an extensive study of food markets in Ethiopia, Dessalegn et al. (1998) provide a descriptive
characterization of grain market performance using both farmer and trader data. The data was
collected in 1996, six years after liberalization of the grain trade, 13 and includes 4,000
households and approximately 220 traders. The key piece of evidence in Dessalegn et al., for this
subsection, is their comparison of the monthly gross real return to storage for different grains
with the borrowing costs faced by traders. Based on seasonal price variation, they estimate the
monthly return to maize storage over the post-harvest period to be roughly 5.18%. Comparable
figures are 3.66% for teff and 3.24% for wheat. They compare these figures to their estimate
from the trader survey that the real opportunity cost of capital is approximately 10% per year, or
0.8% per month. The spread between the gross return to speculative storage and the cost of
borrowing may indicate that there is insufficient competitive arbitrage. But as the authors
acknowledge, these gross returns do not include the costs of storage such as rent, labor, and
fumigation. Nor does a comparison of average returns take into account the risks associated with
price movements or with depreciation. We find the estimated spreads to be surprisingly narrow,
given these factors.
Using data from 1990 – the earliest trader survey data included in this review – Minten and Kyle
(1999) study variation in trader earnings and transport costs across markets in the Democratic
Republic of the Congo (DR Congo). The sample consists of 1,405 traders selected from the port
(for river trading, not international shipping) or truck parking lot in Kinshasa. Trading in DR
Congo during this period is challenging business: transport is difficult due to poor roads, and
86% of sampled traders report dangerous conditions the further they travel from the capital. The
authors find that variation in transport costs explains a large majority of the variation in prices
paid to farmers. They use the survey data to calculate daily wages for the sample traders, then
estimate Mincerian earnings functions to examine how wages co-move with crop type, distance
traveled, and trader characteristics. They find that trader earnings are increasing in distance
traveled on poor quality roads, but not on higher quality roads. This is consistent with the
presence of greater risks and higher costs to conduct business in remote areas served by
inadequate transport infrastructure. However, the average per-transaction gross marketing
margin, exclusive of direct transport costs, is estimated to be 34.3%. This figure is high enough
13
This paper will reappear in other sections; here we discuss only the evidence relevant for this empirical category.
27 that it could be consistent with excess profits to some traders, especially those serving the more
remote areas. Unfortunately the paper cannot distinguish between an equilibrium in which some
traders are earning non-competitive rents and one in which unobserved transport costs, other
costs, and risk premia suffice to explain the high gross margins in some areas.
In another paper from Ethiopia, Gabre-Madhin (2001) uses a new institutional economics
framework to study the market structure of the Ethiopian grain trade after liberalization. The
component of the study most relevant for this paper is the analysis of trader accounts from a
survey of 152 wholesale traders conducted in the second half of 1996. The traders in this sample
had a surprisingly high level of working capital – approximately $14,000 per year, on average –
compared to traders in other settings. This primarily reflects high rates of intra-annual turnover;
the scale of operations is still small relative to that in wealthy countries. Gabre-Madhin
calculates gross and net marketing margins from traders’ most recent transaction. The variation
is substantial – in 3 of the 12 markets studied, the average trader earned negative profits on the
most recent trade. The overall mean margin is just below 5%, which is the return to the trader for
his or her time, capital, and other unaccounted costs. Using additional data on annual sales and
costs, the author estimates average annual trader profits to be $2,300, which is to compensate the
trader for all fixed costs, labor costs, interest paid on debt, and his/her own time. This figure is
inflated by the high average earnings among Addis Ababa traders ($5,364); in 10 of 12 markets
studied, average profits do not exceed $1,575. In sum, while there is some variation across
markets, the take-home earnings of the average Ethiopian trader are hardly indicative of
significant windfall profits or non-competitive rents.
Fafchamps, Gabre-Madhin, and Minten (2005) use trader surveys to estimate returns to scale in
the crop trading sector in Benin, Madagascar, and Malawi. Data is from interviews with traders
and market-level surveys conducted in 1999 and 2000. The study is motivated by the possibility
that if there are significant returns to scale in trading, due perhaps to the high fixed costs of
acquiring the physical capital needed to earn significant profits with narrow margins, then there
may be a rationale for restricting entry in order to increase the operating scale of active firms.
The authors find no evidence of returns to scale.14 In conducting the analysis, they also provide
substantial insight into trader accounts. They find that transport costs represent the largest share
of direct marketing costs, at 50-60%. A substantial fraction of these costs are to pay for onloading and off-loading trucks. The authors do not report the distributions of marketing margins,
but they do write “Margins vary dramatically across traders, however. Some respondents appear
to be incurring massive losses while others make windfall profits” (p. 418). Median annual trader
profits, which are to compensate the trader for his or her time, unpaid labor by family and
friends, working capital, transport vehicles, storage costs at self-owned facilities, and other
physical assets and equipment used in trading, are estimated to be $116 in Benin, $536 in
14
It is notable that after spending significant time in the field, the authors consider fierce competition and free entry
into trading to be so obvious that they do not allow for the possibility that markets are not competitive.
28 Madagascar, and $1147 in Malawi. The between-country variation in profits is largely due to
differences in the composition of trader types – and the associated variation in the requisite
capital investment – between the samples. These median profit levels are hardly indicative of
traders earning non-competitive rents. After subtracting costs, it is very possible that the annual
earnings of the median trader in each country are well below the poverty line.
Osborne (2005), working in Ethiopia, employs a different empirical approach than the other
papers in this section. We include this paper in this section because inference is based on careful
surveying and analysis of traders’ accounts. For the year spanning February 1994 – February
1995, Osborne surveyed 10 traders in each of two markets, collecting detailed information on
prices, quantities, and some components of cost. She develops a model of price determination
under Cournot competition and estimates the slope of the supply curve facing traders. This slope
should be zero under perfect competition. Osborne reports that for the more remote of the two
studied markets, traders appear to reduce the price offered to farmers by approximately 3%
during the post-harvest period, when supply is greatest. In the context of the Cournot model, this
is interpreted as evidence of non-competitive pricing.
Unfortunately, the result in Osborne is subject to a significant number of important caveats,
relating to each component of the analysis. The data that underlies the evidence of noncompetitive pricing is from 10 traders in one market, which makes it difficult to draw general
conclusions. The Cournot model that generates testable predictions is a risk-free model; traders
are assumed to make price offers conditional on expected net revenue, with no allowance for
risks due to uncertain supply volume, depreciation, or wholesale price movements. The empirical
results are only interpretable under the assumption that per-unit costs are declining in quantity
traded, an assumption that is by no means obvious.15 Furthermore, the core results are only
marginally statistically significant with robust standard errors. But with inference based on a
long time series of data from only ten traders, clustering at the trader or trader-day level is surely
warranted. It is doubtful that the core finding would remain statistically significant with clustered
standard errors. Based on all of these shortcomings in the empirical analysis, we find the
evidence in this paper to be of little contribution to the overall goal of determining whether food
markets in SSA are or are not competitive.
Fafchamps and Gabre-Madhin (2006) use the data from Benin and Malawi that was used in
Fafchamps et al. (2005) to further describe variation in trader characteristics and earnings. They
emphasize the significant degree of risk in the sector: some traders appear to make large losses,
while others earn substantial profits. More than 20% of surveyed traders in Malawi and Benin do
not cover all of their costs out of their annual sales. Echoing the findings of Fafchamps et al.
15
For example, the author asserts that the unit price of capital is surely declining in quantity, because of scale
economies in transport that allow for rapid repayment. But it is equally plausible that traders face an upward sloping
credit supply curve because of credit market imperfections leading to heavy reliance on family and friend networks
for loans.
29 (2005), the authors calculate that the gross trader margin in Malawi is 2.4 times than in Benin,
possibly reflecting a difference in competitiveness between the two countries. However, they
argue that this finding is driven by important structural differences between the markets in the
two countries. Traders in Malawi are more vertically integrated than those in Benin, so higher
compensation is warranted to cover the costs associated with a greater number of supply chain
activities. Traders in Malawi also have substantially higher levels of education, suggesting a
higher opportunity cost of time.
Finally, Minten et al. (2015) conduct a detailed analysis of the teff supply chain from farmers in
rural areas to retail markets in Addis Ababa, Ethiopia. Their analysis is based on surveys
conducted in 2012 with 1,200 farmers, 205 rural wholesale traders, 90 truck drivers, 75 urban
wholesalers, and 282 urban retailers. Minten et al. find that at the median, the price to farmers is
a remarkable 79.2% of the urban retail price, and that half of that margin accrues after the grain
arrives in Addis Ababa. This is suggestive of very lean and competitive marketing channels
linking farmers and the capital city. It is important not to over-interpret this result, as a high
producer share of the urban retail price is not prima facie evidence of competitive markets. We
do not know how the 21% marketing margins are distributed across individual traders, nor about
the time period over which traders earn their respective margins. Nevertheless, Minten et al. do
show that transport costs represent a significant share of trader margins, especially in remote
areas. As a whole, the evidence in this paper is strongly suggestive of robust competition among
teff traders in Ethiopia.
4.2.3. Analysis of trader accounts: summary
The discussion in the previous section covers only a small share of the detailed descriptive
information about markets in the reviewed papers. Yet there are three themes most relevant for
our purposes. The first is that agricultural trading is indeed shown to be a risky business. Margins
and profits are highly variable, with non-trivial numbers of traders earning zero or negative
profits in some settings. The second is that remote areas are generally served by fewer traders,
and this may create scope for oligopsony pricing; but the estimated margins in less accessible
areas are still too small to provide clear evidence of imperfect competition. And third, there are
potentially important components of trader costs that remain unobserved in every study, despite
the best efforts of researchers. We cannot know from studies of this kind whether the returns to
trading, after accounting for the full set of costs that traders face, are too high to be consistent
with competitive markets. But based on this evidence, we have no reason to believe so.
4.3. Group 3: Other analyses with trader or farmer survey data
The third group of empirical evidence includes any other tests using trader- or market-level data
that do not belong in the previous section. This group is the most diverse of the four empirical
30 categories. We divide the evidence in this category into three subgroups: estimation of the degree
of market concentration, examination of the correlates of trader profits, and examination of entry
rates or barriers to entry.
4.3.1. Measuring concentration: empirical approach
Two of the three papers in this subgroup study market concentration by estimating concentration
ratios, which measure the extent to which the activity in a particular market is concentrated in the
hands of a limited number of agents. The most frequently used measure, the four firm
concentration ratio or CR4, is the percentage of output sold to the four traders that purchase the
greatest volume of a crop in any village or source market. A CR4 statistic cannot be interpreted
without reference to some scale that relates ranges of estimates to levels of market competition.
Only rough guidelines exist, and they are tailored to markets in the US or Europe. Nonetheless, a
common rule of thumb is that a CR4 ratio below 20 is consistent with substantial or perfect
competition, 20-60 suggests possible non-competitive behavior (with the likelihood of imperfect
competition increasing in the CR4), and greater than 60 indicates that the large firms likely enjoy
some degree of market power.
As a proxy for market power on the part of incumbent firms, the CR4 relates to categories D, E,
and F in the taxonomy of Section 3. However, a CR4 measure alone does not suffice to
distinguish between natural oligopsonies (D), barriers to entry due to anti-competitive behavior
by incumbents (E), and barriers due to regulation (F). Furthermore, while a high CR4 is
suggestive of possible market power, it is not positive evidence that farmers receive prices below
the competitive level. And vice versa: anti-competitive behavior can coexist with a low value of
the CR4, although it would be more difficult to maintain in equilibrium.
We also include in this subgroup evidence from a 2012 paper by Chamberlin and Jayne, which
reports the number of traders active in two Kenyan towns but does not calculate the CR4.
4.3.2. Measuring concentration: evidence
Two papers provide estimates of concentration ratios from (relatively) recent trader surveys.
Dessalegn et al. (1998), using a slightly non-standard method to calculate the CR4,16 report
separate concentration ratios by market and grain type in Ethiopia. They find substantial
variation. Addis Ababa has the lowest concentration ratio for all grains (4%). Markets in Gonder
and Dire Dawa have the highest average CR4 at 44% and 34%, respectively. Concentration
ratios are typically higher for barley, which is traded in fewer locations, than for teff, wheat,
16
The measure is based primarily on the share of trade controlled by the largest firm. It is not clear whether the
method used would over or underestimate the CR4 relative to the usual approach.
31 maize, and sorghum. For 17 of 27 market-crop pairs the CR4 is below 35%. At the market level,
the all-grain CR4 is below 35% for 10 of 11 markets.
Aker (2010) estimates that at the national level, the CR4 for the grain markets of Niger ranged
between 23-26% during the period 2004 to 2006. In addition, at the regional level, CR4 estimates
were all below 25%.
Finally, Chamberlin and Jayne (2012) conducted a maize marketing survey in 33 Kenyan
villages, 19 of which were classified as “relatively remote”, with the other 14 classified as
“relatively accessible.” On average, 94 traders visited each accessible village in the post-harvest
period, and 83 visited each of the remote villages. Given that villages in Kenya typically have a
few thousand residents, these numbers of active traders are indicative of a substantial degree of
competition in the maize market.
4.3.3. Measuring concentration: summary
While these three studies are far from representative, none presents even suggestive evidence
that a small number of actors are dominating the grain trade in their respective settings. There are
sure to be food markets somewhere in sub-Saharan Africa with a small number of dominant
traders – for example, the higher CR4 reported in Dessalegn et al. (1998) is 63%, for sorghum in
Bahir Dar. Furthermore, the CR4 evidence that we have considered here is from larger markets,
not from small rural markets where there may be less competition. Yet the one study that does
include numerous small village markets, Chamberlin and Jayne (2014), finds dozens of traders
operating in every village. Overall the findings in this subsection add to the evidence of generally
robust competition in food markets.
4.3.4. Marketing margins and trader characteristics: empirical approach
The second subgroup in this section is dedicated to a single paper that includes a reduced form
study of how marketing margins vary with trader characteristics. Fafchamps and Minten (2012)
test whether traders with greater social capital enjoy higher marketing margins in the staple crop
markets of Madagascar. The underlying idea is that if marketing margins vary with trader
characteristics, this may reveal the mechanism or mechanisms by which a certain type of trader
receives preferential access in the market. Although a positive result from such a test would be
neither necessary nor sufficient to conclude anything about the competitiveness of the output
market, it would provide suggestive evidence that some traders are able to inflate their margins,
something that is not possible under intense competition. Tests of this nature are primarily
related to category E: the possibility of anti-competitive barriers to entry due to the actions of
current traders.
32 4.3.5. Marketing margins and trader characteristics: evidence
Fafchamps and Minten (2002) hypothesize that marketing margins may increase with social
capital either because connections facilitate collusion and thereby imperfect competition, or
because social capital lowers transaction costs. Data are from interviews with traders involved in
marketing staple crops in three agricultural regions of Madagascar – urban, semi-urban, and rural
– in 1997. Social capital is measured as the number of relatives who are agricultural traders, the
number of other traders known, and the number of potential informal lenders that can be
accessed. The analysis shows that both total sales and aggregate gross margins are increasing in
social capital, conditional on other variables. Traders depend on their networks to receive
accurate market price information, obtain credit, and economize on quality inspection, which has
the effect of lowering transaction costs. However, social capital has no statistically significant
effect on the gross marketing margin.
4.3.6. Marketing margins and trader characteristics: summary
In this brief section we show that in a single paper from Madagascar (Fafchamps and Minten
2002), there is no evidence that trader social capital facilitates collusion or non-competitive
pricing.
4.3.7. Entry and exit: empirical approach
The third subgroup of evidence in this category includes measures of trader entry and exit or
survey-based examination of the barriers to entry. These are descriptive measures that give a
sense of how readily new or lapsed traders can enter a particular market, and of how likely it is
that traders active in one period will choose to leave the market in the next. As mentioned above,
it is not a simple task to measure entry and exit. Because traders may be active only seasonally, a
trader who appears to have exited a market may in fact be waiting for the next relevant
marketing season to begin. Consistent and carefully timed longitudinal survey work would be the
best way to measure entry and exit. Even then, it is likely that some types of traders would be
missed during sampling.
Nevertheless, for the purposes of understanding market structure it is useful to know whether
new traders can easily begin business. This amounts to a direct test for the presence of barriers to
entry, relating to categories D-F of Section 3. As with most of the tests we have discussed, low
rates of entry are not positive evidence of non-competitive pricing; yet, they are a necessary
condition for hypotheses that posit some form of entry barrier. In addition, the firm exit rate
gives some indication of the risks inherent in the sector. If many firms are seen to trade in one
period but not in a future period, this is suggestive (but not definitive) evidence that the return to
33 trading is not sufficient to warrant ongoing activity by many traders. Thus, the exit or hazard rate
is a rough proxy for the degree of risk in the sector, which relates to category A.
4.3.8. Entry and exit: evidence
In Madagascar, Barrett (1997) uses qualitative interviews and surveys with 261 traders from
1992 and 1993 to study entry and market structure in the period immediately following
liberalization. Interviews highlight a number of the characteristics of the trading sector that we
discussed above: market inter-linkages are common, farmer-trader relationships underpin many
deals, most assembly traders are also farmers, and trading risks can be significant. Barrett divides
the agricultural marketing system into five functional areas: collection, transport, milling, interseasonal storage, and vending. He finds that following liberalization there was considerable entry
into small-scale trade. Changes in labor markets induced some farmers to enter the assembly
trade, as a means of generating additional income. Women and members of less-privileged
groups, who generally lack access to information or capital, were also found to enter the subcollection business at small scale. This highlights how low were the barriers to entry in the first
step of the food market supply chain. In contrast, survey evidence suggested that the other
functional areas – wholesale collection, interregional transport, and storage – had significantly
higher barriers to entry because of the extensive fixed cost requirements. Credit constraints were
cited as the primary barrier to the scaling up of smaller firms or the formation of new businesses
to enter these high fixed cost sectors. Social capital may also be important, as well-connected
members of the community were better able to raise funds to finance entry. The paper cannot
determine whether the higher profits earned by wholesalers and large-scale transporters were
indicative of non-competitive rents, or simply represented the market return to the requisite
investment and the associated degree of risk.
Dessalegn et al. (1998), in their extensive study of markets in Ethiopia, note that a central
concern of many formal traders is that unlicensed merchants can easily enter the trading sector.
Because unlicensed traders do not have to pay registration fees or taxes, they are able to undercut
the prices offered by the formal traders. Thus, even when regulations exist that raise the costs of
entry, weak enforcement capacity may sharply reduce the effective barrier to entry created by
such regulations. Yet, like Barrett (1997), Dessalegn et al. note possible barriers to entry at higher-­‐level marketing niches. These include access to finance, to reliable marketing information, and to storage and transport capital.
Fafchamps et al. (2005), in their study of the trading sector in Madagascar, find evidence of
substantial exit by trading firms. In 2001, only 47% of traders interviewed in 1997 were still
operating. This suggests annual exit rates in excess of 17% – hardly indicative of an entrenched
oligopsony. One of the main conclusions of the paper is that even though fixed costs and credit
constraints create barriers to entry for high-cost links in the supply chain, there is little evidence
34 that demonstrates that large enterprises have an advantage over small enterprises. With constant
returns to scale in trading, there would be no welfare gains from a policy of restricting trader
entry.
Sitko and Jayne (2014) study the competitive structure of the markets that serve farmers – the
first step in the supply chain – in four countries: Mozambique, Kenya, Malawi, and Zambia.
Data are from various sources: 205 farmer group discussions, 2,703 farmer interviews, 166 trader
interviews, and 48 interviews with processors. The authors find that on average farmers receive
over 80% of the retail/wholesale price from the next step in the supply chain. They show that
local assembly traders who purchase at the farm-gate pay 0-15% lower prices than large
wholesalers or processors, but these differences are understood to be compensation for transport
and for the convenience of making purchases at the farm-gate immediately after harvest. Farmers
in isolated villages receive between 83% and 96% of the prices that farmers receive in more
accessible locations. As has been the case with many papers in the review, there is no way to
determine whether the lower prices received by farmers in less accessible villages reflect only
higher costs, or also non-competitive mark-ups. But the survey data shows that 76% of isolated
villages are served by 10 or more traders, as compared to 82% of accessible villages; in contrast,
12% of isolated village are served by 0-5 traders, as compared to 9% of accessible villages.
These minor differences suggest that if non-competitive mark-ups are present, they are unlikely
to be the main drivers of the 4-14% lower prices received in more remote areas. Trader entry
appears to be unimpeded, even in remote areas.
Finally, we mention evidence from one study that does not involve food crops, but which is
directly relevant for this section. Fafchamps and Hill (2008) study price transmission from export
markets to the farm-gate for the coffee sector in Uganda. Data are from surveys with exporters,
traders, and farmers conducted in 2003. The key relevant finding is that when the export price
rises, wholesale prices rise proportionally, and this triggers entry by very small-scale assembly
trader who purchase at the farm-gate. The ease of entry is a central finding of the study. It is also
notable that price transmission is nearly complete at the wholesale level. This suggests that at
least in this market, any barriers to entry at the large-scale wholesale level, possibly due to risks,
credit constraints, or high fixed costs, have not led to non-competitive pricing.
4.3.9. Entry and exit: summary
The findings in this section suggest that across the various countries and time periods for which
we have evidence, barriers to trader entry at the first level of exchange – that between farmers
and traders – are minimal. There is descriptive evidence of greater barriers to entry at the level of
wholesale exchange or inter-regional trade, due primarily to the greater risks and the fixed costs
of entry at this level of trading. However, there is no positive evidence that these higher costs
create natural oligopsonies or foster collusion. In fact, the evidence of Section 4.1 shows rapid
35 price transmission between wholesale markets, which suggests that entry costs have not
dampened competition at that level.
4.4. Group 4: Experiments and impact evaluations
The final category of empirical tests includes impact evaluations based on randomized,
controlled trials or natural experiments.
4.4.1. Impact evaluations: empirical approach
All but one of the studies in this group involve an information intervention to provide farmers
with up-to-date market price data during the cultivation and/or harvest and marketing periods.
The interventions take various forms: non-uniform allocation of radio services, natural
experiments based on the rollout of mobile phone towers, or SMS-based information services
provided randomly or quasi-randomly to some households but not to others. These studies are
premised on the hypothesis that farmers and traders bargain over the sale price, with farmers at a
disadvantage because they do not know the full distribution of prices in the relevant wholesale
markets. Traders must earn non-competitive rents for these interventions to have an impact.
However, a test of the impact of an information intervention on producer prices should be
interpreted as a joint test of the presence of trader rents, the salience and relevance of the
information provided, and the effect of better information on the farmer’s bargaining position.
The studies in this group are inherently related to the possibility of barriers to entry from anticompetitive behavior by current traders (category E). The hypothesis is that traders are using an
information asymmetry to extract rents. Of course, for such an arrangement to be maintained in
equilibrium, there must be some force that prevents other traders from entering the market and
competing away the excess margins. If no such barriers exist, then there is little reason to believe
that the observed marketing margins are the result of insufficient competition. Studies of this
kind can only provide definitive evidence about the competitiveness of the pre-treatment
equilibrium if they can reject other possible interpretations of the effect, or if they are combined
with positive evidence of barriers to entry. We elaborate on this point in the next subsection.
This branch of the literature should not be confused with papers that test the impact of ICT
expansion on the price spread between markets (Jensen 2007, Aker 2010). The rollout of mobile
phones has been shown to reduce spatial price dispersion between markets, especially for
perishable goods. But the findings from those papers tell us how changes to the fundamental cost
structure of the market lead to a new set of equilibrium prices; they do not provide evidence on
the competitiveness of the market before or after the change in technology.
36 4.4.2. Impact evaluations: evidence
In an early paper for this literature, Svensson and Yanagizawa (2009) study the impact of radio
information on producer prices in Uganda. They make use of a natural experiment in which a
market information service broadcast weekly price information for a number of major district
market prices over local FM radio stations. They find that maize farmers with radios in treated
districts receive a 15% higher farm-gate price than households with radios in control districts.
The paper provides support for the identifying assumption that treated and untreated districts are
similar prior to program roll-out. However, the central challenge to inference remains the
possibility that treatment effects are driven in part by some other time-varying unobservable
difference between treatment and control areas. This concern is partly related to the fact that rollout was biased toward areas where USAID, the funder, had an active presence, and which
therefore may have been received other, confounding interventions.
Muto and Yamano (2009) study the effect of mobile phone network expansion on the change in
farm-gate prices for maize and bananas in Uganda. Inference is from panel data in 94
communities over the period 2003-2005. At the time of the study, 7 communities had not yet
received coverage from the mobile network. They find that network expansion had a positive and
significant impact on the prices that farmers receive for bananas, but no impact on prices
received for maize. They also find an increase in the sales volume of bananas in remote areas,
suggesting that treatment increased access for farmers who are farthest from the district center.
The authors interpret the differential effects for bananas and maize as evidence that crop
perishability opened up the possibility of non-competitive pricing in the pre-treatment
equilibrium, the implication being that when farmers are under pressure to sell quickly they are
at a disadvantage in their negotiations with traders.
Courtois and Subervie (2014) evaluate the Esoko program in Ghana, a price information service
for farmers. They estimate the impact of SMS price information on farmers’ prices for
groundnuts and maize. Roll-out of the program was not randomized; identification in this paper
is based on propensity-score matching. They find that the treated group received 10.4% higher
maize prices and 7.3% higher groundnut prices than the control group.
Unlike the first three papers in this group, which are based on natural experiments or ex post
approaches to identification, Hildebrandt et al (2015) study a two-year cluster-randomized
controlled trial of the Esoko service in Ghana (the same service studied by Courtois and
Subervie, 2014). Their focus is on the price that farmers receive for yams, which is typically
purchased at the farm-gate after a negotiation between farmers and traders. In Year 1, the authors
find that the average yam price received by treated farmers is a statistically significant 5%
greater than that of control farmers. However, in Year 2 the treatment effect disappears, and the
average price received by treated farmers is slightly lower than that received by control farmers
(although the difference is not significant). The authors interpret this as evidence of a
37 “bargaining spillover,” in which trader uncertainty about the information set of the farmer leads
to less aggressive trader bargaining with both treatment and control group farmers.
While this is an intriguing result that may be evidence of non-competitive pricing in the pretreatment equilibrium, there are a number of remaining puzzles or alternative interpretations.
These include (i) the possibility that trader margins on purchases from treated farmers in year 1
were unsustainably low, resulting in either trader exit or claw-backs in year 2; (ii) the possibility
that higher prices for everyone in year 2 were simply the result of higher wholesale prices for
yams; or (iii) the fact that there was substantial pre-treatment variation in both farm-gate prices
and farmer information sets, calling into question why bargaining spillovers would emerge in the
context of a relatively small intervention but not in the regular course of business (in which case,
they would cease to be spillovers and would just be the data generating process for trader
beliefs). It is also notable that Hildebrandt et al. (2015) find no price effects for maize or
groundnuts, the two crops with significant treatment effects in the Courtois and Subervie (2014)
paper that uses propensity score methods in the same setting.
One final paper bears mention in this section. Casaburi and Reed (2014) implemented a
randomized experiment in Sierra Leone to study whether and to what extent quality premia
offered to traders are passed on to cocoa farmers. The paper is somewhat outside our scope,
because it does not involve food crops. But we include it because it is to our knowledge the only
experiment in SSA implemented at the trader rather than farmer level.
The intervention in Casaburi and Reed involved a bonus to treated traders of 5.6% on top of the
wholesale price for each unit of high quality cocoa. The key hypothesis is that under
competition, such a premium would be passed on to farmers. The motivation underlying this
intervention relates clearly to the possibility of non-competitive barriers to entry (D, E, F),
though the findings are ultimately informative for understanding market inter-linkages between
traders and farmers (category C). Casaburi and Reed report no significant differences in prices
paid to farmers who sell to treated traders. However, farmers who are regular customers of the
treated traders are 14% more likely to receive credit than farmers selling to non-treated farmers.
Additionally, traders are more likely to extend credit in villages where competition is high and
there are high volumes of cocoa. The takeaway is that when markets are interlinked – as credit
and output markets frequently are for cash crops in sub-Saharan Africa – traders can pass on
higher wholesale prices via mechanisms other than the farm-gate price.
4.4.3. Impact evaluations: summary
How should we read the balance of the evidence in this section? Three of the papers that we
reviewed report significant effects from information interventions at the farmer level. Yet each is
subject to some important caveats. Svensson and Yanagizawa (2009) and Muto and Yamano
38 (2009) find positive effects for some crops, but the identification strategies in those papers do not
exclude the possibility of non-random access to the information service. The positive finding in
Courtois and Subervie (2014), using non-experimental variation, is contradicted by the null result
for the same crop and setting in Hildebrandt et al. (2015). Hildebrandt et al. (2015) find a
positive effect for one crop, but it does not last. So while there is suggestive evidence that an
information intervention may lead some farmers to receive higher prices than they would have
otherwise, particularly for perishable crops, the result is far from conclusive. The picture is even
cloudier if we take into account evidence from outside sub-Saharan Africa, where many recent
studies have shown little to no effect on prices from randomizing farmers’ access to market
prices (see the Appendix).
There are limits to what research like this can tell us about the competitiveness of food crop
markets. The motivation for these studies is sound; any honest observer would be hard-pressed to
deny that there are surely some instances in which traders take advantage of farmer ignorance in
order to bargain down farm-gate prices. Yet, without knowing the trader-level distribution of
farm-gate prices and costs, it is not possible to know whether lower trader margins on one deal –
perhaps due to the hard bargaining of a better informed farmer – are partially offset by higher
trader margins on another. Traders who need to acquire sufficient volume to reach an efficient
transport scale may offer better prices to treated farmers in order to complete a purchase. But the
researcher does not know whether those higher prices come at the expense of non-competitive
rents that traders were earning pre-treatment, or at the expense of other farmers who are
untreated or who lack the capacity to effectively bargain. And if the gains disappear quickly, as
they do for the one crop studied with an RCT that shows a positive treatment effect, that suggests
that the inter-temporal and spatial distribution of pre-treatment marketing margins did not
include substantial non-competitive trader rents. In sum, positive findings from farmer-level
price interventions would not be sufficient evidence that there exist barriers to entry in the pretreatment equilibrium large enough to sustain oligopsony rents. Furthermore, as we see in
Casaburi and Reed (2014), traders may pass on the benefits of higher wholesale prices to farmers
through a variety of channels, some of which can be difficult to detect if the research focus is
only on prices. Of course, the converse is also true: lack of a treatment effect is not evidence that
the market is competitive. If, for example, traders are oligopsonistic and are aware of farmers’
limited outside options, better information may not effectively improve the bargaining position
of farmers.
5. Conclusion
The goal of this paper has been to categorize and review the empirical evidence on the question
of whether food crop markets in sub-Saharan Africa appear to be competitive in the poststructural adjustment era. We first developed an analytical framework that summarized the
competitive and non-competitive forces that can increase marketing margins above the
39 apparently competitive level. We then reviewed the evidence in four sub-categories of the
literature. Our overall conclusion is that the small amount of evidence that does exist is generally
consistent with the notion that food markets are competitive.
There are a number of limitations to this review. For one, our focus has been primarily on food
markets, rather than export-oriented cash crops that may be characterized by different levels of
competition. Additionally, the evidence presented here is from only 13 countries – hardly a
representative picture of sub-Saharan Africa. Also, we have not addressed the possibility of
higher-level forms of non-competitive behavior, such as the lobbying of transporter
organizations to prevent the expansion of rail trade. Finally, as we have discussed, there are
substantial empirical challenges to detecting non-competition. The first stems from the
possibility of selective participation in surveys by traders. The second is due to the inherent
difficulties of distinguishing between non-competitive rents and unobserved costs borne by
traders. Because of these difficulties, evidence on this question must come in the form of general
tendencies that emerge from patterns in numerous and varied studies; “smoking guns” are
unlikely to be found.
Are there policy conclusions from this review? If anything, our review suggests that there is little
rationale for intervening in food crop markets to increase the level of competition in agricultural
trading. However, concerns about possible non-competitive trading are motivated by the
observation that farmers earn too little for their crops. Given the high rates of poverty and
significant risks to small-scale farming across the region, there is clearly scope on social welfare
grounds for some other forms of intervention. One clear takeaway from the studies reviewed
here is that efficient crop output markets will not raise the living standards of the hundreds of
millions of small-scale farmers in sub-Saharan Africa who remain in poverty if they are not
coupled with sufficient investment in public goods – roads, information services, agricultural
research stations to develop high yield cultivars – to raise productivity and the total level of
output in the region.
40 6. References
Abdulai, Awudu (2000). “Spatial Price Transmission and Asymmetry in the Ghanaian Maize
Market.” Journal of Development Economics 63(2): 327–349.
Aker, Jenny C (2010). “Information from Markets Near and Far: Mobile Phones and Agricultural
Markets in Niger.” American Economic Journal: Applied Economics 2(3): 46–59.
Aker, Jenny C., and Marcel Fafchamps (2015). “Mobile Phone Coverage and Producer Markets:
Evidence from West Africa.” World Bank Economic Review 29(2): 262-292.
Aker, Jenny C., Ishita Ghosh, and Jenna Burrell (2016). “The Promise (and Pitfalls) of ICT for
Agriculture Initiatives.” Working paper.
Asad, Saher (2015). “The Crop Connection: Impact of Cell Phone Access on Crop Choice in
Rural Pakistan.” Working paper.
Atkin, David, and Dave Donaldson (2015). “Who’s Getting Globalized? The Size and
Implication of Intra-national Trade Costs. NBER Working Paper No. 21439.
Badiane, Ousmane, and Gerald E. Shively (1998). “Spatial integration, transport costs, and the
response of local prices to policy changes in Ghana.” Journal of Development Economics
56: 411-431.
Barrett, Christopher B. (1994). “Understanding uneven agricultural liberalization in Madagascar”
Journal of Modern African Studies 32(3): 449–476.
Barrett, Christopher B. (1996). “Market Analysis Methods: Are Our Enriched Toolkits Well
Suited to Enlivened Markets?” American Journal of Agricultural Economics 78(3): 825829.
Barrett, Christopher B. (1997). “Food Marketing Liberalization and Trader Entry: Evidence from
Madagascar.” World Development 25 (5): 763–777.
Barrett, Christopher B. (2001). “Measuring Integration and Efficiency in International
Agricultural Markets” Review of Agricultural Economics 23: 19-32.
Bates, Robert (1981). Markets and States in Tropical Africa. University of California Press.
41 Baulch, Bob (1997). “Transfer costs, spatial arbitrage, and testing for food market integration.”
American Journal of Agricultural Economics 79(2): 477-487.
Bernard, Tanguy, David J. Spielman, Alemayehu Seyoum Taffesse, and Eleni Z. Gabre-Madhin
(2010). “Cooperatives for Staple Crop Marketing: Evidence from Ethiopia.” IFPRI
Research Monograph 164.
Camacho, Adriana and Emily Conover (2010). “The Impact of Receiving Price and Climate
Information in the Agricultural Sector.” Working paper.
Casaburi, Lorenzo, Rachel Glennerster, and Tavneet Suri (2013). “Rural Roads and
Intermediated Trade: Regression Discontinuity Evidence from Sierra Leone.” Working
paper.
Casaburi, Lorenzo, and Tristan Reed (2014). “Interlinked Transactions and Pass-Through:
Experimental Evidence from Sierra Leone.” Working Paper.
Chamberlin, Jordan, and T.S. Jayne (2012). “Unpacking the Meaning of ‘Market Access’:
Evidence from Rural Kenya. World Development 41: 245-264.
Cole, Shawn, Stefan Hunt, and Wentao Xiong. “Information and Expectations: Evidence from a
Field Experiment in India.” Working paper.
Courtois, Pierre, and Julie Subervie (2014). “Farmer Bargaining Power and Market Information
Services.” American Journal of Agricultural Economics 97 (3): 953–77.
Dessalegn, Gebremeskel, Thomas S. Jayne, and James D. Shaffer (1998). “Market Structure,
Conduct, and Performance: Constraints of Performance of Ethiopian Grain Markets.” Food
Security Collaborative Working Paper, Michigan State University.
DFID Department for International Development (2004). “Rethinking Tropical Agricultural
Commodities.” DFID, London.
Dillon, Brian and Christopher B. Barrett (2016). “Global Oil Prices and Local Food Prices:
Evidence from East Africa” American Journal of Agricultural Economics 98(1): 154-171.
Fackler, Paul L. and Brad K. Goodwin (2001). “Spatial Price Analysis” in B.L. Gardner and
G.C. Rausser, eds., Handbook of Agricultural Economics Volume 1. Amsterdam: Elsevier.
42 Fafchamps, Marcel, Eleni Gabre-Madhin, and Bart Minten (2005). “Increasing Returns and
Market Efficiency in Agricultural Trade.” J. of Development Economics 78 (2): 406–42.
Fafchamps, Marcel, and Eleni Gabre-Madhin (2006). “Agricultural Markets in Benin and
Malawi  : The Operation and Performance of Traders.” World Bank Policy Research
Working Paper No. 2734.
Fafchamps, Marcel, and Ruth Vargas Hill (2005). “Selling at the Farmgate or Traveling to
Market.” American Journal of Agricultural Economics 87 (3): 717–34.
Fafchamps, Marcel, and Ruth Vargas Hill (2008). “Price Transmission and Trader Entry in
Domestic Commodity Markets.” Economic Development and Cultural Change 56 (4):
729–66.
Fafchamps, Marcel, and Bart Minten (2002). “Returns to Social Network Capital among
Traders.” Oxford Economic Papers 54 (2): 173–206.
Fafchamps, Marcel, and Bart Minten (2012). “Impact of SMS-Based Agricultural Information on
Indian Farmers.” World Bank Economic Review 26(3): 383-414.
Gabre-Madhin, Eleni Zaude (2001). “Market Institutions, Transaction Costs, and Social Capital
in the Ethiopian Grain Market.” IFPRI Research Report 124.
Goyal, Aparajita (2010). “Information, Direct Access to Farmers, and Rural Market Performance
in Central India.” SSRN Scholarly Paper ID 1613083.
Hildebrandt, Nicole, Yaw Nyarko, Giorgia Romagnoli, and Emilia Soldani (2015). “Price
Information, Inter-Village Networks, and ‘Bargaining Spillovers’: Experimental Evidence
from Ghana.” Working paper.
Jensen, Robert (2007). “The Digital Provide: Information (Technology), Market Performance,
and Welfare in the South Indian Fisheries Sector.” Quarterly Journal of Economics 122(3):
879–924.
Labonne, Julien, and Robert S. Chase (2009). “The Power of Information: The Impact of Mobile
Phones on Farmers’Welfare in the Philippines.” SSRN Scholarly Paper ID 1435202.
Mason, N. M. and Robert J. Myers (2013). “The effects of the food reserve agency on maize
market prices in Zambia.” Agricultural Economics 44: 203–216.
43 Minten, Bart, and Steven Kyle (1999). “The Effect of Distance and Road Quality on Food
Collection, Marketing Margins, and Traders’ Wages: Evidence from the Former Zaire.”
Journal of Development Economics 60(2): 467–95.
Minten, Bart, Anneleen Vandeplas, and Johan F. M. Swinnen (2011). “The Broken Broker
System? Transacting on Agricultural Wholesale Markets in India (Uttarakhand).” IFPRI
Discussion Paper 1143.
Minten, Bart, David Stifel and Seneshaw Tamru (2014). “Structural Transformation of Cereal
Markets in Ethiopia.” Journal of Development Studies 50(5): 611-629.
Minten, Bart, Seneshaw Tamru, Ermias Engida, and Tadesse Kuma (2015). "Feeding Africa's
Cities: The Case of Supply Chain of Teff to Addis Ababa." Economic Development and
Cultural Change, forthcoming.
Mitra, Sandip, Dilip Mookherjee, Maximo Torero, and Sujata Visaria (2012). “Asymmetrc
Information and Middlemen Margins: An Experiment with West Bengal Potato Farmers.”
Working Paper.
Moser, Christine, Christopher Barrett, and Bart Minten (2009). “Spatial Integration at Multiple
Scales: Rice Markets in Madagascar.” Agricultural Economics 40(3): 281–94.
Muto, Megumi, and Takashi Yamano (2009). “The Impact of Mobile Phone Coverage
Expansion on Market Participation: Panel Data Evidence from Uganda.” World
Development 37 (12): 1887–96.
Myers, Robert J. and T.S. Jayne (2012). “Multiple-regime Spatial Price Transmission with an
Application to Maize Markets in Southern Africa.” American Journal of Agricultural
Economics 94(1): 174-188.
Myers, Robert J. (2013). “Evaluating the Effectiveness of Inter-Regional Trade and Storage in
Malawi’s Private Sector Maize Markets.” Food Policy 41(August): 75–84.
Nakasone, Eduardo (2013). “The Role of Price Information in Agricultural Markets:
Experimental Evidence from Rural Peru.” Working Paper.
Osborne, Theresa (2005). “Imperfect Competition in Agricultural Markets: Evidence from
Ethiopia.” Journal of Development Economics 76(2): 405–28.
44 Pokhrel, Deepak M., and Gopal B. Thapa (2007). “Are Marketing Intermediaries Exploiting
Mountain Farmers in Nepal? A Study Based on Market Price, Marketing Margin and
Income Distribution Analyses.” Agricultural Systems 94 (2): 151–64.
doi:10.1016/j.agsy.2006.08.004.
Ravallion Martin (1986). “Testing market integration.” American Journal of Agricultural
Economics. Feb(1): 102-109.
Sitko, Nicholas J. and T.S. Jayne (2014). “Exploitative Briefcase Businessmen, Parasites, and
Other Myths and Legends: Assembly Traders and the Performance of Maize Markets in
Eastern and Southern Africa. World Development 54: 56-67.
Staatz, John M., Josue Dione and N. Nango Dembele (1989). “Cereals market liberalization in
Mali.” World Development 17(5): 703–718.
Svensson, Jakob, and David Yanagizawa (2009). “Getting Prices Right: The Impact of Market
Information Service in Uganda,” Journal of the European Economic Association 7(2-3):
435-445.
Swinnen, Johan (1997). Political Economy of Agrarian Reform in Central and Eastern Europe.
New York: Ashgate.
Tostão, Emilio and B. Wade Brorsen (2005). “Spatial Price Efficiency in Mozambique’s PostReform Maize Markets” Agricultural Economics 33: 205-214.
Van Campenhout, Bjorn (2007). “Modelling Trends in Food Market Integration: Method and an
Application to Tanzanian Maize Markets.” Food Policy 32: 112–27.
World Food Programme (2015). “Purchase for Progress, The Story: Connecting Famers to
Markets.” Mimeo.
45 7. Appendix
INTENDED FOR ONLINE PUBLICATION ONLY
We briefly outline five groups of evidence that are relevant to this paper, even though they do not include direct evidence on the competitiveness of food crop markets in SSA. 7.1. Changes in equilibrium price spreads due to reductions in communication costs
Jensen (2007) uses the roll-out of mobile phone coverage from 1996-2001 as a natural
experiment to study the effect of lower communication costs on fish prices in Kerala, India. He
finds a dramatic decline in price dispersion associated with lower communication costs. Before
treatment, the difference in market pairs between markets is estimated to be around 50-60%.
Post-treatment, the price of fish differed by only the transportation costs between markets. Lower
information costs allowed for lower cost arbitrage and changed the underlying competitive
equilibrium of market price spreads.
In a similar vein, Aker (2010) studies the impact of lower communication costs on millet price
spreads between 37 randomly selected markets in 6 geographic regions of Niger. Data comes
from monthly market prices from 1999-2006 and surveys with 415 traders. Treatment is
determined by the presence of a mobile phone tower. Aker finds that 3 months after the
introduction of mobile phone coverage, there is a 10-16% reduction in price dispersion across
markets in comparison to pre-treatment levels. Mobile technology has the largest impact on
remote markets because of the associated reduction in transportation costs for traders. The effect
is also stronger when a higher percentage of markets have coverage. Aker argues that the
reduction in price dispersion is due to the uptake of mobile phones by traders, who then use the
technology to reduce search costs and allocate goods more efficiently across the countryside.
Aker and Fafchamps (2015) use the same natural experiment to study dispersion in producer
prices (rather than wholesale market prices). Using market pair fixed effects, the introduction of
mobile phone towers reduces producer price dispersion across markets for cowpeas by 6.3%.
There is no statistically significant impact on millet. Reductions in producer price dispersion are
most evident for markets that are farther than 350km apart and are connected by unpaved roads.
They also find that phone coverage leads to a 6% reduction in producer’s price risk, as proxied
by the intra-annual coefficient of variation. They do not find that farmers’ access to mobile
phone technology results in arbitrage by farmers.
7.2. Analyses of trader accounts
In Nepal, Pokhral and Thapa (2007) surveyed 23 traders operating in agricultural regions close to
Kathmandu. These traders primarily purchase mandarin, a cash crop, from farmers. The
researchers requested that traders keep a log of revenue and costs over the 2002 agricultural
46 season. In analyzing these data, the authors categorize the trader as “exploitative” if the profit
earned in crop trading is greater than the opportunity cost of capital invested in another sector.
By this measure, a marketing margin of 8-12% is considered to be fair, although it is notable that
no allowance is made for risk adjustment between crop trading and the comparison sectors. The
authors find that on average, 69% of the gross trader margin is transferred to the farmer. After
accounting for other costs, traders’ net marketing margin are estimated to be only 9% on
average, well within the competitive range.
7.3. Trader social capital
While Fafchamps and Minten (2002) directly study the effect of social capital on trader profits,
other papers from outside of SSA allude to the importance of long-term relationships built on
trust and reputation between traders and farmers. Minten et al. (2011) study broker relationships
among retailers and farmers in government-regulated mandis in Uttarakhand, India. They find
that in practice, relationships with brokers for both retailers and farmers are based more on
personal trust and less on informal credit and insurance schemes. Farmers will work with the
same broker over many transactions because they trust they can obtain fair prices without
incurring search costs for other brokers in the mandi. Mitra et al. (2013) argue that the entire
potato supply chain in West Bengal, from farmers to traders to wholesalers, is based on
trustworthiness more than credit-worthiness, scale, or size of network. Wholesalers entrust
trading agents to transact with farmers, and are unwilling to deal directly with farmers whom
they do not know personally. However, in this setting, the lack of direct competition from
wholesalers allows traders to earn higher margins.
In Nepal, mandarin farmers build strong ties with traders or farmer-collectors that extend over
generations (Pokhral and Thapa 2007). These relationships are developed from mutual assistance
in times of need. Social pressures and the need to maintain trust with farmers leads farmercollectors to rarely cheat. However, surprisingly, only half of farmers interviewed claim that they
can fully trust that trading partners are reliable or fair. The other half of farmers say that they
have been cheated by farmer-collectors, have received delayed payments, or have incurred
harvesting delays due to problems with the collector. These delays are often due to liquidity
problems faced by traders, though they can create the perception that traders are untrustworthy or
non-competitive, even in a market where the average farmer-collector earns a roughly 9%
margin (not accounting for inflation).
7.4. Inter-linkages
Inter-linkages may not only guarantee future trade, they also can transfer risk and transaction
costs from farmer to trader. Pokhral and Thapa (2007) find that farmers in Nepal sell 90% of
mandarin output to farmer-collectors before harvest, transferring risk of damage and spoilage to
47 traders and eliminating the costs of transporting the fruit to the market. These farmers receive an
advance from farmer-collectors who then get exclusive rights to the fruit. Producers also receive
higher margins because farmer-collectors will wait until after the peak season to pick and sell
mandarin in the market.
There is also evidence that intermediaries will supply credit without the expectation of trade
commitments or reduced output prices for farmers. For instance, Mitra et al. (2013) observe that
while 31% of potato farmers in West Bengal had borrowed money from a trader, there is no
expectation of selling directly to that that trader. In Minten et al. (2011), brokers will provide
loans with subsidized interest to farmers with high crop volumes but the authors do not find that
producer prices fall as a result of receiving a loan.
7.5. Information impact evaluations
There is a large literature from outside of sub-Saharan Africa evaluating the impact of
information interventions similar to those reviewed in Section 4.4.1. Aker, Ghosh, and Burrell
(2016) provide a comprehensive overview. While the estimated effects of information
experiments on farm-gate prices are mixed, the literature from outside SSA, like the literature
from SSA, leans toward null results. If anything, there is a suggestion that information can
improve farmer prices for perishable crops, but not for storable crops. It is an open question
whether the general lack of positive treatment effects is due to incorrect intervention designs,
multiple market failures that reduce the efficacy of an isolated treatment in the information
market, or ex ante efficiencies in information flows that were unobservable to researchers.
In a tangentially related study, Labonne and Chase (2009) measure impacts on farmer welfare by
combining spatially coded data on mobile coverage with household panel data of farmers in lowincome areas in the Philippines from 2003-2006. They find that farmers with a mobile phone
experienced an 11-17% increase of per capita consumption over the study period. However,
inference in this paper is limited by the lack of a reliable way to instrument for selection into
mobile phone ownership.
Goyal (2010) measures market level price differentials on soybeans after ITC, an agricultural
procurement company, introduces Internet kiosks in the state of Madhya Pradesh in India. The
kiosks were set up to give daily wholesale price information for soybean farmers both in the
local markets (mandis) and with ITC. Data comes from monthly price information and volume
sold from 2000-2005, yield and net area under production from 1998-2004, and ITC business
records to find dates for kiosk installation and soybean volume collected post-kiosk installation.
Goyal finds an immediate increase in the average monthly mandi price of soybeans by 1-3%
after kiosk introduction. Trader profits, in contrast, decreased by 4% on average. With the
increased probability of profits, farmers increased production of soy and away from other crops,
48 with the area under soy production increasing by 19%. It is also found that there is a 2% gain in
welfare that is otherwise lost in a monopsony setting, which points to the possibility of increased
competition after the introduction of the ITC kiosks.
Camacho and Conover (2011) test the effect of SMS daily price information for 3 markets and 8
products typically sold in Colombia. They randomly select 500 farmers from irrigation
associations to receive price information through SMS for the month of July, which is
immediately pre-harvest. A randomly selected control group did not receive SMS price
information. Treated farmers are 20-30% more likely to know the market prices. However, better
information about market prices did not translate to higher sales prices, as there is no effect on
producer prices.
Cole, Hunt, & Xiong (2012) study the effects of information about futures markets on the
relationship between farmers’ price expectations and actual futures market prices. They conduct
a two-stage randomized control trial. In the first stage, farmers receive weekly spot and futures
prices on a village board for cotton, castor, and guar seeds. In the second stage, farmers receive
daily price information for the same crops, via SMS. The hypothesized benefit of futures markets
is that farmers can coordinate with sellers to create a pre-specified delivery date. The futures
market is intended to reduce vulnerability to price shocks and to provide farmers with
information to help with planting, harvesting, and storage decisions. They find that the treatment
group is 3 times more likely to use futures prices to inform their price expectations, but the
results are not uniform across crops. Farmer expectations about castor prices are related more to
spot prices than futures prices; futures price have a bigger effect on cotton price expectation. The
paper does not report whether better information impacts the prices that farmers receive at
harvest.
Fafchamps and Minten (2012) conduct a randomized controlled trial of the Reuters Market Light
(RML) agricultural information system with a sample of 933 farmers in 100 villages in
Maharashtra, India. Treated farmers received SMS-based information on prices, weather, and
crops. The study includes both high-perishable and low-perishable crops (onions, wheat,
soybeans tomatoes, pomegranates). Only 59% of farmers offered the SMS service actually used
it. Some feared having to pay for it in the future, some were illiterate, and others never activated
the service. Treated farmers were more likely to change the market where they sold their output,
but changing markets did not result in better prices for farmers on average. There is no
significant difference in price received between RML users and the control group. The point
estimate of the treatment effect is negative and not statistically significant.
Mitra et al (2013) study the effect of price information on the relationship between farmers and
traders in West Bengal, India. They randomly assign 72 villages with potato farmers to three
arms. Farmers in the first arm received price information on a notice board, those in the second
49 arm received information through private phone calls, and those in the third arm served as the
control. Neither treatment led to significant average treatment effects on the prices received by
farmers selling potatoes. There is some interesting heterogeneity, however. In villages located in
market areas with low wholesale prices, traded quantities and farm-gate prices fell as a result of
added information. The opposite was found to be true in areas with high wholesale prices. A
particular feature of this market makes it difficult to generalize from these findings: wholesale
collectors that purchase at larger markets prefer to buy from traders rather than farmers. Thus,
even if a farmer knows that a trader is offering low prices, his outside options may be limited by
the fact that wholesale collectors at the market refuse to buy his product.
Nakasone (2013) is the only paper that finds unequivocally positive impacts on prices from
SMS-based information services. Nakasone provided 111 households in 26 villages in Peru with
mobile phones that provided tailored price information via SMS, and compared outcomes to
households in 32 control villages. Information was delivered immediately before the harvest, and
the phones were designed to only receive income SMS messages from the project. Farmers who
received the treatment experienced a 13-14% average increase in sales price. The impact is
especially large for farmers who are “always sellers” (19% increase in prices received) and
farmers who are typically more risk averse. On average, the proportion of output being sold to
traders was not affected by the treatment.
Asad (2015) uses growth in cell phone coverage in Punjab, Pakistan, in combination with the
existence of “dead zone” in network coverage close to the border with India, as a natural
experiment for the provision of information services. She studies effects on price for three crop
categories: extremely perishable, highly perishable, and storable. Asad finds that treatmet
increases farmer incomes by 10-15% on average. The probability of farmers and traders
coordinating in advance increases by 35-59% for extremely perishable crops and 31-48% for
highly perishable crops. Post-harvest loss decrease by 21-35% and 16-20% for extremely
perishable and highly perishable crops, respectively. These substantial differences are attributed
to a reduction in the time between harvest and sale dates, which fell by 4-7 days. Mobile phones
allow for buyer-seller coordination that is economically advantageous for both farmer and trader.
Because the treatment impacts the probability of ex ante coordination, there is now way to
determine whether there are also implications for the competitiveness of the trading sector.
50