Farmers’ Production and Marketing Response to Rice Price Increases and Fertilizer Subsidies in the Office Du Niger
December 3, 2012 - Author: David Mather and Valerie Kelly
IDWP 129. David Mather and Valerie Kelly. 2012. Farmers’ Production and Marketing Response to Rice Price Increases and Fertilizer Subsidies in the Office Du Niger.
Many African governments responded to the dramatic increases in international and domestic
grain prices of 2008 and 2009 through a mixture of trade policy changes and input/output
market subsidies. In the case of Mali, the Government put in place a rice promotion program
at the beginning of the 2008/09 production season. The program, called Initiative Riz (IR or
Rice Initiative), made subsidized fertilizer available to rice producers nationwide with a
particular focus on farmers in the Office du Niger, where roughly 50% of Mali’s rice is
produced. The goal of the program was to increase domestic rice production by 50% over the
2007/08 level, thereby increasing marketable surpluses and putting downward pressure on
Whether the program met its production goals is unclear because of conflicting numbers
reported by two different Ministry of Agriculture sources. What is not disputed, however, is
that rice prices did not fall as much as anticipated in 2008, with the government and other
observers suggesting that rice producers were ‘hoarding’ their production to take advantage
of the higher price environment– i.e., they held onto more of their annual production than
This paper aims to inform the debate about farmers’ response to the IR and rising rice prices
through empirical analysis of household survey data on crop production and marketing that
permits comparison of farmers’ behavior before (2006/07) and after the price spikes and the
introduction of the IR (2008/09 and 2009/10). The survey includes production and sales data
for both the rainy and dry seasons during the three agricultural production years mentioned
above as well as a wealth of demographic, asset, and non-farm income data. We use
descriptive and econometric analysis of this data to investigate the following research
1. When did cereal prices begin to rise in Macina markets, and to what extent did they
2. How did farmers respond to rising cereal prices and the IR with respect to their area
cultivated to rice and coarse grains?
3. How did rising cereal prices and/or the IR affect fertilizer use?
4. Did the IR and higher expected cereal prices lead to higher yields and more aggregate
rice production, as anticipated by the GOM? If not, does econometric analysis of the
observable determinants of rice yields explain why?
5. Did households reduce the percentage of their rice production that they sold over
time? If yes, does econometric analysis of the observable determinants of household
rice sales explain why?
For each of these questions, we test for differences in response by different types of farm
households. Although the focus is on how responses are shaped by farm size (access to more
or less irrigated land) and the quality of irrigation (full or partial water control), we also look
at the role of other factors such as demographic characteristics and ownership of assets (e.g.,
durable goods and agricultural equipment).
Descriptive analyses brought to the forefront some of the salient differences between farmers
located in the casiers, where most rice plots benefit from improved quality irrigation, and
bord de casiers, where access to good quality irrigation is more restricted. In terms of land
access, the average ON household of roughly ten people is thought to need at least five
hectares of irrigated land to make ends meet. The average casiers farm in the sample has
secure access to a total of only 4.2 hectares, which are predominantly irrigated parcels; this is
less than the recommended minimum size of 5 hectares, but includes substantially more good
quality irrigation land than farmers in the bord du casier villages (4 ha for casiers farms
versus only 2.2 ha for the bord de casiers). In brief, bord de casiers farmers have
significantly less high quality irrigation land with fully controlled water and more of the
lower quality parcels with partial control.
While the casiers/bord de casiers distinction does not differentiate households by food
security status (roughly 29% of casiers and bord de casiers farms fail to meet minimum
cereal needs after accounting for sales, in-kind production payments, and purchases), land
ownership patterns differentiate farms in terms of net cereal availability. Households meeting
the 214 kg/capita benchmark for food security have more total land, more land per capita, and
more irrigated land—both improved and unimproved. A similar pattern differentiates net
sellers (those who sell more cereals than they purchase) from all other farms. Net sellers have
access to 3.4 hectares of improved irrigation land and 0.7 hectares of the less productive hors
casiers land while other farmers (net buyers and autarkic households) have access to only 0.9
hectares of improved irrigation and 0.43 ha of hors casiers land, suggesting that access to
irrigated land may be an important determinant of whether one becomes a net seller.
Multivariate analyses of the Macina data base identify the factors influencing rice yields and
rice marketing decisions. The principal observable determinants of rice yield in Macina
include nitrogen, the amount of hired transplanting labor per hectare, and reported householdlevel
production problems such as poor water control. As expected, nitrogen has a strong,
significant, and positive effect on rice yield. At the mean level of nitrogen use in the sample
(79.6 kg of nitrogen/ha), an additional kg of nitrogen/ha increases rice yield by 11.3 kg/ha.
Given prevailing price relationships, the value cost ratio for this response would have been
2.3 in 2008 and 2.2 in 2009.
While the mean/median quantity of nitrogen applied to rice increased slightly from 2008 to
2009, the yield benefits from nitrogen appear to have been more than offset by various
reported household-specific production problems, which have large and significant negative
effects on rice yields. For example, problems with water control reduced yield by 477 kg/ha,
late planting reduced yield by 356 kg/ha, and ‘other’ undefined problems reduced yield by
610 kg/ha. These findings help explain the decline in rainy season rice yields from 2008 to
2009, as we found a larger percentage of households reported problems with water control
and ‘other’ problems in the latter year.
The marketing models revealed that the principal observable factors affecting the household
quantity of rice sold were household rice production and the level and source of input credit
that season. Because a principal factor explaining rice sales is rice production, it’s not
surprising that quantities of rice sold fell over time as production and yields fell. What is
perhaps surprising from this analysis is that even after controlling for the amount of rice and
coarse grains produced, the level of input credit, and demographic and wealth measures,
variation in the household’s rice sale price does not have a significant effect on the quantity
sold. This suggests that either there is considerable heterogeneity of price responsiveness
across different kinds of households or that household rice sales are simply not very
responsive to changes in the rice price. Another hypothesis might be that the price
responsiveness is linked to the production decision based on price expectations. The model
eliminates quantifying this effect by using production as an explanatory variable. Because the
ability of a farmer in the ON to change land area is limited, the main production response
would probably be through more fertilizer to increase yields. Such a response was facilitated
by the fertilizer subsidy that began in 2008/09, but there was little evidence of a substantial
increase in fertilizer use and/or yields for sample households as a result of the fertilizer
subsidy despite rising output prices. Analyses by cereal production groups also returned nonsignificant
price coefficients for both the lower 1/3 and the upper 2/3rds of cereal producers,
suggesting that the problem is not heterogeneity based on levels of cereal production.
There are a number of practical policy implications that flow from this study with respect to
the government’s goal of increasing marketed rice supply:
1) Because the study confirmed that increased fertilizer use can increase rice yields
significantly, the GOM should be able increase marketable surpluses of rice by
focusing its attention on improvements in fertilizer supply (particularly timeliness and
reducing delivery costs), input credit, and better monitoring and evaluation of the
costs and benefits of the input subsidy program for both farmers and private sector
2) Efforts to increase fertilizer use are not likely to achieve significant increases in rice
production or marketed supply unless they are accompanied by improvements in
water control and other management practices to avoid the significant yield reductions
reported in survey data. This implies a need to balance budgetary support for input
subsidies and support for services that render those inputs more effective.
3) Although this paper did not address the contribution of other technical production
issues (e.g., improved varieties, particularly for dry-season production; lower-cost
approaches to fertilizer use; improved management practices to avoid soil
acidification), continued benefits from fertilizer will be contingent on continued
research and extension on these topics to ensure that fertilizer is being used as
efficiently as possible and not having negative impacts on soil quality.
4) Roughly one third of ON farms are unable to provide for their own minimum cereal
needs of 214 kg/capita after paying for production costs; this is not a sustainable
situation and appears to be more of a problem for small farms than for large farms,
suggesting that more attention needs to be given to policies concerning access to
irrigated land for family farms and/or increasing opportunities for income
diversification through off-farm employment that does not compete with farm
demands for labor.
5) OPAMs role in rice marketing since the beginning of the IR has been unpredictable
and not very helpful to rice producers; the GOM needs to reconsider its policy of
OPAM intervention in rice markets, making it more transparent and predictable;
reliable funding must be part of the picture or marketing is better left entirely to the
6) Although more research is needed to better understand farmers’ production and
marketing responsiveness to output prices, the survey results suggest that factors such
as production costs and credit repayment scheduling (particularly fertilizer and water
payments) may be more important influences on production levels and marketing
behavior than output prices.
7) Mali is far behind many other African countries in its ability to systematically monitor
and analyze the performance of its agricultural sector through the use of longitudinal
data bases. Despite the many caveats mentioned about the panel data underlying the
analyses presented in this paper, the data set is unique in its coverage of both
production and marketing information for the same set of farms over the span of three
years. There is a need for the GOM to invest in Mali’s capacity to collect and analyze
longitudinal data on the agricultural sector at a scale that is large enough to obtain
representative results for at least the main production zones of the country; to date
these types of investments have been made by donors and have not endured.