The economics of pest and production management in small-holder cocoa: lessons from Sulawesi

Abstract

We examine pest control and production management methods used by farmers in Sulawesi to improve cocoa bean quality and increase income from cocoa. Strategies investigated include those directed at increasing the number and size of cocoa pods, those aimed at reducing hosts for pest transmission, two input-intensive approaches, and the alternative of doing nothing beyond harvesting mature cocoa pods. Using 2005 production data from 600 cocoa farms, we identify factors correlated with adoption of each treatment and, controlling for treatment, isolate factors that influence cocoa yields. To study the conditional profitability of input allocation, we compare observed factor shares with profit-maximising input levels and derive lessons for extension efforts. We conclude that the average increase in private returns arising from more intensive cocoa management appears sufficient to compensate for higher production costs, but that observed extension efforts have not been correlated with higher profits among farmers in the sample.

Acknowledgements

We thank the editor and three anonymous reviewers for substantial constructive comments on earlier versions of this paper. We have also benefited greatly from conversations with Smilja Lambert, Peter van Grinsvin, Josef Toledano, Will Masters, John Mumford, Sally Thompson and Steve Yaninek. Financial support for this research was provided by Mars Inc. and the United States Agency for International Development under the auspices of the Sustainable Agriculture and Natural Resource Management Collaborative Research Support Program (SANREM CRSP). The opinions expressed are those of the authors and do not necessarily reflect the views of the sponsoring agencies or Kraft Foods Global Inc.

Notes

¹Neilson (2007) argues that declines in quality are largely a problem for the domestic processing industry, since US and European grinders have found ways to adapt to low-quality beans.

²The PRIMA project is a joint initiative of Masterfoods Veghel BV and PSOM (Program of Cooperation with Emerging Markets), funded by the development cooperation budget of the Netherlands Ministry of Foreign Affairs.

³Sample farmers do not seem to be replacing unproductive trees at a high rate: data indicate one new tree is planted for every two trees that are deemed unproductive.

⁴Estimated coefficients from a regression of yield on age and age squared suggest that cocoa production in the sample appears to peak at a tree age of approximately 22 years. Although cocoa trees can remain productive for much longer than this under ideal conditions, old trees are not typical in the growing area.

⁵The use of outside management is a potential complicating factor in the analysis, but we observe relatively few instances in which the family relinquishes complete oversight of the farm to an outside party. Instead, the use of contracted spraying and fertilising seems to be the most regular form of outside management. We cannot reject the hypothesis that average yield and average profit are equivalent on contractor-managed and owner-managed plots, although we recognise that this may be due to the relatively small number of contractor-managed plots in the sample.

⁶In general, Indonesian cocoa farmers do not ferment cocoa beans, because the beans are valued mainly for butter content, which is unaffected by fermentation (Neilson 2007).

⁷For additional information about the PRIMA project, see Neilson et al. (2005).

⁸Viewed in terms of a two-stage least squares (2SLS) regression, the four PRIMA project variables included in the MNL regression serve in conjunction with the farmer characteristic variables (age and education) to identify the response variables for inclusion in the yield regressions. We have assumed that the influence of the project on outcomes worked solely through the indirect pathway of influencing choice of response, and had no direct impact on yields. Test results for endogeneity and over-identification restrictions are not reported here, but are available upon request.

⁹As an anonymous reviewer pointed out, the positive association of pest infestation with yields is probably due to the fact that farmers with pests are responding to the problem effectively. On the other hand, in separate regressions not reported here we find a relatively weak correlation between herbicide use and profits. This may or may not be evidence that herbicide use is unprofitable, as the weak correlation may be attributable to the fact that herbicides are used in response to severe weed infestations. In the absence of a measure of weed infestation, herbicide use can be a proxy for the intensity of weed infestation. Therefore it should not immediately be concluded that herbicides have no effect on yields or profitability.

¹⁰Given that the average tree age is 18 years, which is less than the 22-year optimum (see footnote 4), our results do confirm that farmers with trees older than 22 years obtain less than optimum yields, as do those with younger trees. Nevertheless, those who own the older trees face a problem of declining yields and are therefore justified in their concern, whereas those who own younger trees will not face this problem for some time.

¹¹Let π be profit, Y output, P the price of output, X _i the quantity used of input i and W _i the price of input i. Then . Therefore .

¹²Of course if the assumed opportunity cost of labour were higher, the apparent profit forgone would be correspondingly lower.