![Sample our Area Studies journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5922/TOC-Subject-Sample-2021-Area-Studies.jpg"})
ABSTRACT
We investigate the post-war recovery of the Rwandan coffee sector. First, we look at the recovery of export earnings at the national level and show that the role played by the rise in international coffee prices largely outweighed the one played by domestic policies to boost coffee production and quality. Second, we analyze the subnational variation in the recovery of coffee tree investment and reveal the legacy of armed conflict. In 1999 – five years after the peak of the violence – highly violence-affected regions exhibit significantly lower tree planting and maintenance. Within a decade, the gap is however closed. We discuss the role that positive externalities generated by high-profile public investments in the coffee sector might have played in the catch-up process. We frame this discussion in the wider debate on the nature of the Rwandan State.
KEYWORDS:
Acknowledgements
We gratefully acknowledge JEAS editors and two anonymous referees for many valuable comments and suggestions. We also thank Elena Briones Alonso, Jean Chrysostome Ngabitsinze, Li Fan, Ameet Morjaria and participants at the Doctoral Workshop at KU Leuven, the CSAE conference in Oxford, and the EABEW conference in Kigali for valuable comments.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. Although we focus on the period 1999–2009, we provide more recent figures when relevant, and show in Appendix that our results remain qualitatively the same when we extend the analysis to more recent years.
2. World Bank, World Development Indicators.
3. Ibid.
4. Republic of Rwanda, The Evolution of Poverty. It should be mentioned that the accuracy of Rwandan administrative data, particularly in relation to poverty figures, has been questioned (Ansoms et al., “Statistics versus Livelihoods”; ROAPE, “Rwandan Poverty Statistics”). According to the African Development Bank, Rwanda ranks 28th in terms of overall quality of its national accounts, over a list of 44 African countries (ADB, Situational Analysis).
5. Des Forges, Leave None to Tell the Story; Verpoorten, “Detecting Hidden Violence,” 44–56.
6. Verpoorten, “Leave None to Claim,” 547–63.
7. See Clay, “Fighting an Uphill,” and, World Bank, Rwanda – Country Assistance Strategy.
8. Collier, The Bottom Billion. Rwanda qualifies for at least two of Collier’s poverty traps: the experience of the civil war and the a landlocked location surrounded by conflict-prone poor economies.
9. Booth and Golooba-Mutebi, “Developmental Patrimonialism?” 379–403; Mann and Berry, “Understanding the Political,” 119–44.
10. Basinga et al., “Paying Primary.”
11. Republic of Rwanda, Vision 2020.
12. See note 9.
13. Gökgür, “Rwanda’s Ruling-Party.”
14. See note 9.
15. Lin and Monga, “Growth Identification.”
16. USAID, Assessing USAID.
17. Our analysis focuses on the first decade of the new century, immediately following the end of the conflict cycle. A new coffee census was collected in 2015 and results are confirmed whenever including that census in the analysis. We report the results in Appendix.
18. Coffee washing stations are facilities that use spring water to remove the fruity pulp from the coffee cherries, separating out the beans that are then dried and stored for marketing. The process is semi-automated, and allows for several quality control methods to be employed to sort the coffee into various grades.
19. Boudreaux, “Economic Liberalization”; Chunan-Pole and Angwafo, Yes Africa.
20. Collier, On the Economic, 168–83.
21. e.g. Brück, “War and Reconstruction,” 30–9; González and Lopez, “Political Violence,” 367–92.
22. Serneels and Verpoorten, “The Impact,” 555–92; Verpoorten, “Household Coping,” 67–86.
23. Tardif-Douglin et al., “Finding the Balance.”
24. The International Coffee Agreement signed in 1983 established a system of export quotas to secure price stability within ranges agreed upon annually by exporting and importing members.
25. Ibid.
26. Loveridge, Nyarwaya, and Shingiro, “Decaffeinated? Situation.”
27. Murezeki, Songqing, and Loveridge, “Have Coffee,” 1–12.
28. Boudreaux, “Economic Liberalization.”
29. See note 16.
30. In 2007 – the year for which we have the most detailed information – there were still 12 washing stations owned directly by the Horizon group or by para-statal entities.
31. Behuria, “Centralising Rents,” 630–47; Behuria and Goodfellow, “The Political Settlement”; Gökgür, “Rwanda’s Ruling-Party.”
32. Within the context of coffee production in Ethiopia, Minten et al. “Coffee Value” find that farm distance plays a key role in determining whether farmers will bring their beans to a wet or dry mill.
33. One of many examples is the article by L. Fraser titled “Coffee, and Hope, Grow in Rwanda,” New York Times, August 6, 2006.
34. Chunan-Pole and Angwafo, Yes Africa.
35. Easterly and Reshef, “African Export,” 4.
36. The figures are taken from the World Bank database and refer to the international price for Arabica coffee, which is the predominant variety in Rwanda. Alternative sources provide similar figures: from $2.2 to $3.1 based on the New York Coffee Price Index for fully-washed Arabica coffee, and from $2.3 to $2.9 based on the international coffee prices published by the International Coffee Organization (ICO).
37. The role of the international coffee price rise is often ignored. For instance, looking at the period 2003–2008, the World Bank publication Yes Africa Can highlights that the average export price of Rwandan green coffee increased from $1.60 to $3.10 (94%), but it fails to mention that the world market price for coffee rose from $1.41 to $3.06 (117%) over the same period.
38. Republic of Rwanda, Rwanda Coffee.
39. See also Macchiavello and Morjaria, “Competition and Relational.”
40. Barro and Sala-i-Martin, Economic Growth.
41. Organski, and Kugler, “The Costs,” 1347–66.
42. According to the World Bank, the amount of aid received by Rwanda in 1999 was close to the pre-1994 levels, at roughly 550 million (in constant 2013 US$). However, it almost doubled between 1999 and 2009, reaching 990 million. While with the data at our disposal it is impossible to identify which areas benefited the most, conflict-affected regions benefited from a number of specific programs. Within the educational sector, for instance, the largest program was the FARG (Fonds d’Assistance aux Rescapés du Génocide), which awarded scholarships for secondary schooling to genocide survivors. A nationally representative survey collected in 2000 indicates that less than 5% of the students enjoyed a scholarship in the Northern Provinces (where few Tutsi lived), compared to more than 12% in some of the Central and Southern Provinces.
43. See note 2.
44. See note 40.
45. A new coffee census round was collected in 2015. The results obtained using this latest round are in line with those obtained using the 2009 dataset and are reported in Appendix.
46. Mukashema, Veldkamp, and Vrieling, “Automated High,” 331–40.
47. The criteria for the categorization are not specified, but they most probably include best practices such as pruning, mulching, weeding, and the usage of pesticides.
48. The values are computed dividing the average number of trees per Sector (535 and 53,047, respectively for 1999 and 2009) by the average size of the old and new administrative Sectors (12.41 and 52.29 km2, respectively).
49. Verpoorten, “Detecting Hidden Violence,” 44–56.
50. Commune indicates the administrative level above the Sector. There were 154 Communes in Rwanda before the administrative reform.
51. We proxy potential yield with information on the expected yield of very well maintained trees, as recorded in the 2003 census.
52. We repeated all estimations excluding the category 3–10 years from the controls and results remained qualitatively unchanged (results available on requests).
53. Coffee trees may have been uprooted during the conflict, or in the immediate post-war years. In this case, the number of trees older than 3 years as recorded in the 1999 census would be an underestimation of the true number of trees that were planted before the conflict. If more trees were uprooted in the more affected areas, the underestimation would be related to conflict intensity and thus bias our coefficient of interest. There is no indication that the uprooting of coffee trees occurred on a large scale during the violence. In fact, in some cases the local administration explicitly established that all crops standing on the fields of killed people were to belong to the Commune and were to be protected by the people of the Sector in which they were located (Des Forges, Leave None to Tell the Story). An additional piece of evidence is provided by our robustness test reported in , which shows that genocide intensity is unrelated to the number of old trees recorded in the 1999 census.
54. The 1999 and 2003 samples include 78 Communes, while the 2009 sample includes 134 Communes.
55. Table A.3 in Appendix shows that the result is confirmed when looking at the 2015 coffee census.
56. as well as all the following tables only display the coefficients of interest. The full tables are available on request.
57. Figure A.3 shows the coverage and geographic distribution of the different conflict variables.
58. Tables A.5 to A.10 in Appendix replicate all our additional tables using the three alternative conflict measures.
59. All the controls are re-computed according to the different boundaries.
60. While there are today also a large number of privately owned washing stations in Rwanda, donors (especially USAID) and the Rwandan government initiated the wave of investments in the coffee sector (for more details, see Schilling and McConnell, “Model for Siting” and USAID, Assessing USAID). Other efforts included investments in seeds development, agricultural extension practices, the stimulation of cooperatives, and the sponsorship of international partnerships with large coffee traders.
61. The lack of an association between the placement of washing stations and conflict intensity is in line with Schilling and McConnell (Model for Siting), who describe the criteria for the locational choice as based on the suitability of the region to coffee growing, the presence of good transportation roads, the presence of plentiful clean water, the absence of other washing stations in the same service area, the availability of land for the construction of the station, and the absence of a national park in the nearby area (see also Macchiavello and Morjaria, “Competition and Relational”).
62. We consider the presence of washing stations in 2007 because the dependent variables equal coffee trees younger than 3 years by 2009, which were therefore planted from 2007 onwards. Using the presence of washing stations in 2008 delivers in any case similar results.
63. This was obtained by computing (exp(−0.639 + 0.086*10) − 1)%. The positive effect of the washing station is for Sectors with a share of Tutsi greater than 7.4% (0.639/0.086). The average change in (the log of) investments in coffee trees is equal to 6.89%.
64. Further supporting this interpretation is the fact that the interaction term between the presence of a washing stations and the intensity of the violence, if anything becomes negative when looking at the evolution in coffee trees investments between 2009 and 2015 (Table A.11, in Appendix). This confirms that the signaling value of the washing station played a role in restoring confidence only in the aftermath of the violence, when uncertainty was likely higher.
65. See for instance Tuckett, Minding the Markets.
66. Elder, Zerriffi, and Le Billon, “Effects of Fair,” 2355–67.
67. See note 9.
68. Serneels and Verpoorten, “The Impact,” 555–92.
69. See Ghura, “Macro Policies, External Forces,” 759–78.
70. Arbache and Page, “How Fragile”; Olakojo, “Export Commodity.”
71. Miguel and Roland, “The Long-Run.”
72. Boudreaux, “Economic Liberalization”; Elder, Zerriffi, and Le Billon, “Effects of Fair.”