Emerging Markets after Liberalization: Evidence from the Raw Milk Market in Rural Kenya

Size: px
Start display at page:

Download "Emerging Markets after Liberalization: Evidence from the Raw Milk Market in Rural Kenya"

Transcription

1 Forthcoming in FASID Discussion Paper Series Emerging Markets after Liberalization: Evidence from the Raw Milk Market in Rural Kenya Yoko Kijima,* Takashi Yamano,* and Isabelle Baltenweck** * Foundation for Advanced Studies on International Development ** International Livestock Research Institute January 16, 2006 Abstract This article examines how the raw milk marketing system in Kenya has been evolving after the dairy sector liberalization in 1992 by using panel data of 874 rural households. From the late 1990s to 2004, the proportion of rural households who sold milk increased from 37 to 51 percent. During the same period, the proportion of households who sold milk to traders more than doubled from 19 to 41 percent, while it has drastically declined for local cooperatives from 24 to 7 percent. The price analysis indicates that the raw milk market has become more efficient. After the liberalization was initiated, it took some time for price arbitrage to be realized among milk markets because trading a perishable commodity, such as milk, over a long distance requires reliable market channels and institutions. Corresponding author: Yoko Kijima Address: GRIPS/FASID Joint Graduate Program, Roppongi, Minato-ku, Tokyo, , Japan. Tel: , FAX , kijima@grips.ac.jp Acknowledgements: Support for the data collection used in this paper is funded by the 21 st Century Center of Excellency project at National Graduate Institute for Policy Studies. We are grateful to Tetsushi Sonobe, Keijiro Otsuka, Steve Staal, and Takashi Kurosaki for useful suggestions and discussion. We also thank Francis Karin and Mburu for excellent assistance in the field, Eunice Kariuki and Pamela Ochungo for data arrangements, Yujiro Hayami, Kei Kajisa, Takeshi Sakurai, and participants of the Poverty Alleviation and Development Strategy Conference in Hakone for useful comments, and Paul Kandasamy for his editorial assistance.

2 1. Introduction Many African countries have liberalized commodity markets under structural adjustment programs since the mid-1980s. The sudden withdrawal of governmental controls and supports on market transactions has left some commodity markets in confusion, rendering evaluations of the market liberalization controversial (Badiane and Shively 1998; Jayne et al. 2002; Negassa et al. 2004; Osborne 2005). What has become clearer in recent years is that liberalizing markets by simply removing regulations does not guarantee that markets will work smoothly and that institutional supports are necessary for markets to function efficiently (Minten and Kyle 1999; Kydd and Dorward 2001; Fafchamps et al. 2003; Fafchamps 2004; Rashid 2004; Kherallah et al. 2000). The liberalization of the raw milk market in Kenya is a typical example. Before the market was liberalized in 1992, Kenya Cooperative Creamery (KCC) was the only processing company that was permitted to process milk, 1 and selling raw milk in urban areas was prohibited. 2 Thus, KCC had a monopoly on milk sales in urban areas. Although the official intent of the 1992 dairy sector liberalization was only to affect the sale of pasteurized milk products in urban areas, the interpretation by market actors was that all actors including raw milk traders could now operate in those areas (Owango et al. 1998). Since the price of processed milk tended to be set low due to political reasons, the real consumer price of the processed milk has increased because of the removal of 1 To be exact, there were two other small cooperatives that used to process milk (Ngigi 2005). 2 Ngigi (2005) notes that even before the liberalization, raw milk was directly sold to households in the neighborhood of the producing households. Such direct sales of raw milk have been a thriving activity in urban areas due to the difficulty in enforcement and the inefficiency in the single channel formal system. However, raw milk traders were harassed by the police in urban areas, and the quantity of raw milk traded was considered to be much smaller before the liberalization than after the liberalization. 1

3 the price control after the dairy sector liberalization (Karanja 2003). Due to the increased price of processed milk, urban consumers have shifted from processed to raw milk (M mboyi et al. 2005). After the collapse of KCC in the late 1990s, 3 many local dairy cooperatives, which collected raw milk from dairy households and sold it to KCC, have also collapsed. 4 Although a cooperative based marketing system was considered to decrease the transaction costs of a perishable commodity such as raw milk (Staal et al. 1997; Holloway et al. 2000), the raw milk market in Kenya was forced to restructure itself to suit the new environments. The objective of this article is to concretely describe how the structure of the raw milk market has changed since the late 1990s when KCC collapsed to 2004, and to analyze how such a structural change has affected the efficiency of the raw milk market. We use panel data of 874 rural households in western and central Kenya from the 1990s and The panel data allow us to analyze the changes of the raw milk marketing of sampled households and to estimate the milk price model to test if the raw milk market has become more efficient over time. The rest of the article is structured as follows. The next section describes the market liberalization of the dairy sector in Kenya. Then, we present the panel data used in the study and explain our hypotheses and estimation strategies. Finally we discuss the results and provide conclusions and policy implications. 3 The financial situation of KCC was not healthy even before the liberalization. Thus the liberalization itself was not the reason for KCC s collapse. However it might have accelerated the speed of the collapse. 4 Dairy cooperatives were faced by mismanagement problems that had led some to bankruptcy even before the collapse of KCC (Ngigi 2005). In 2000, KCC was revived as another company called KCC2000. However, the amount of milk collected by KCC did not increase since KCC could not regain trust from dairy farmers. This decreased the rate of operation, delayed payments to dairy farmers, and as a result, further decreased the amount sold to KCC. Because of such a vicious circle, KCC2000 ended up stopping operation just after the recovery. In 2004, KCC gained a new managerial board and restarted as New KCC. Even now, New KCC is operating under full capacity. 2

4 2. Liberalization of the Dairy Sector in Kenya Although secondary data exist on milk production in Kenya, they do not provide clear evidence to evaluate the dairy sector liberalization. Figure 1, for instance, presents the change in the total milk production and the amount of milk processed in Kenya since Figure 1 contains two sets of statistics on raw milk production: one from the Food and Agriculture Organization (FAO) and the other from the Kenya Dairy Board (KDB). As one can see in the figure, there is a large discrepancy between these two data sets especially after the liberalization in According to the FAO data, on the one hand, the total milk production increased up to the early 1990s. After the liberalization, however, milk production stagnated throughout the 1990s, though it has returned to an upward trend since The FAO data, therefore, suggest that the market liberalization has had a positive impact on milk production after a period of transition. The KDB data, on the other hand, show that the milk production has remained stagnant since the liberalization. In addition to such discrepancies of the FAO and KDB data, many specialists think that the FAO and KDB data underestimate the actual number of cattle and the amount of milk production in Kenya (SDP 2005). Thus, it is not clear from the existing data whether the liberalization of the raw milk market has helped to increase milk production or not. Since Kenyan people mainly drink milk in the form of milk tea, they boil milk even when the milk is processed. Although raw milk is traded as a final good or a raw material for processed milk, figure 1 indicates that the amount of milk which is processed by either KCC or private processing companies accounts for less than 20 3

5 percent of the total milk production. 5 The remaining amount is consumed by dairy households at home or by consumers as unprocessed milk. Thus, the informal raw milk market has the largest share of the market in Kenya. The amount of milk processed by KCC started declining after the liberalization in 1992 and became nearly zero in In contrast, the amount processed by private milk processing companies has been increasing since the liberalization. According to Karanja (2003), 42 private companies obtained licenses to process milk between 1992 and 2001, but only 22 of them were still surviving in In 2001, the top four processing companies had about 80 percent of the market share of processed milk. In figure 2, we describe the simplified milk marketing channels before and after the liberalization in Kenya. Before the liberalization (Panel A), dairy households sold milk to neighbors, dairy cooperatives, vendors, restaurants, or individual customers in nearby towns. Since selling raw milk in urban areas was prohibited before the liberalization, private traders used to sell raw milk to consumers in rural areas. Although some dairy households residing close to KCC directly sold milk to KCC, others sold milk to dairy cooperatives who subsequently sold it to KCC. KCC then processed it and sold the processed milk mainly in urban areas. When the liberalization effectively removed the ban on raw milk sales in urban areas, dairy households living close to urban areas started selling raw milk in urban areas. Some of them became vendors who collected milk from their neighbors and sold it in urban areas. Private processors who newly entered the milk market collected raw milk from dairy households or purchased raw milk from dairy cooperatives and traders. Because KCC delayed payments to dairy cooperatives and their members, 5 Though the data after 1999 cannot distinguish the milk processed in KCC from that in private processing companies, the amount processed in KCC should be very low. 4

6 some of the members have shifted their sales outlet from KCC to private processing companies. The area where traders buy milk has enlarged from peri-urban to rural areas. An important question is how such changes have affected raw milk markets in Kenya, a question which we examine next. 3. Data and Changes in Milk Marketing among Sampled Households 3.1 Data This article uses panel data of 874 households in rural Kenya (table 1). The panel comprises two periods of time. On the one hand, the first round of the panel data comprises a series of three household surveys collected in 1996, 1998, or The second round, on the other hand, was collected in 2004 for all areas. The three surveys in the first round were conducted by the Smallholder Dairy Project (SDP), a collaborative team from the Ministry of Livestock Development & Fisheries, the Kenya Agricultural Research Institute (KARI), and the International Livestock Research Institute (ILRI) (Staal et al. 2001; Waithaka et al. 2002). In 1996, SDP first conducted a survey of 334 rural households in Kiambu District (Central Province) near Nairobi. Then, in 1998, the survey was expanded to eight districts in the Central Kenya region, and covered 1,390 additional households (Staal et al. 2001). In 2000, they interviewed 1,576 households from seven districts in the Western Kenya region (Waithaka et al. 2002). In sum, a total of 3,300 rural households, including non-dairy households, were randomly selected according to similar sample selection procedures and interviewed on dairy production and other income generating activities. In 2004, as a part of the Research on Poverty and Environment and Agricultural 5

7 Technology (REPEAT) Project, sub-locations (the smallest administrative unit in Kenya) were randomly selected from sub-locations where the SDP households resided. The 100 sub-locations are scattered in central and western Kenya. In each sub-location, ten SDP households were selected for re-interviews. Out of the 1,000 targeted households, 914 households were successfully identified and 874 households were interviewed in In the 2004 REPEAT survey, households were asked about their agricultural and non-agricultural income generating activities, asset holdings, cash expenditure, and demographic characteristics. Although new questions were added in the 2004 REPEAT survey, most of the questions on livestock and dairy production were kept comparable with the SDP surveys. Thus, the panel data can be used for measuring the change in milk marketing between the late-1990s and Table 1 presents the number of sampled households, the per capita income, and the milk income share among the sample households in 2004 by province. 8 According to table 1, the income from milk production accounts for about 20 percent of total household income in the Central province and the Rift Valley province, while it is low in the Western and Nyanza provinces. Table 1 also shows the change in the proportion of households who sold milk. There is a large increase in the proportion of households selling milk from 37 to 51 percent during the period of the late 1990s to Even though the proportion rose in all provinces, the increase was especially high in Rift Valley. This suggests that the milk sales area has enlarged from the surroundings of 6 The REPEAT Project is a collaborative research project of Foundation for Advanced Studies on International Development (FASID), National Graduate Research Institute for Policy Studies (GRIPS), the World Agro-forest Center, and Tegemeo Institute in Kenya. More details on the REPEAT are available in Yamano et al. (2005). 7 The rest (40 households) were successfully found but not available for interview due to several reasons: non-contact (27 cases), moved away (6 cases), refusal (5 cases), non-available (1 case), and households dissolved (1 case). 8 We do not have compatible figures from the SDP surveys in the late 1990s. 6

8 Nairobi to Rift Valley. 3.2 Changes in Milk Marketing In the previous sub-section, we find that the proportion of milk selling households has increased during the same period that KCC and many dairy cooperatives have collapsed. Table 2 shows how such a change has taken place. In both the first and second rounds, about half of the dairy households who sold milk have listed individual customers and restaurants as the largest milk buyer. In the same period, the proportion of households who identified dairy cooperatives as the largest milk buyer has declined drastically from 24 to 7 percent, while the proportion of households who sold milk to traders has increased from 17 to 34 percent. It dropped especially in areas close to Nairobi (distance to Nairobi is less than 100km) from 46 percent to 17 percent. In contrast, the proportion of private processors has surged from 0 to 13 percent in these areas. It is also worth noticing that private traders have increased their share and become the second largest milk buyer in areas which are 100 km to 170 km away from Nairobi. It seems that private traders have gained the market share which was lost by the dairy cooperatives. Table 3 shows the average raw milk prices received by dairy households in both rounds separately for different milk buyers. 9 We can see that the overall real raw price has declined by 7 shillings (28.3%) during this period. In all areas, the price received by dairy households is higher when milk is sold to individual consumers than when milk is sold to other milk buyers. This could be because dairy households often deliver milk to individual customers and do not need to pay a margin to other agents. 9 All price levels are adjusted to the 2004 price level. We used inflation rates in the Statistical Bulletin (Central Bank of Kenya, June 2002, 2004, and 2005). 7

9 When we look at the relationship between the raw milk price and the distance to Nairobi by each milk buyer, we do not find a clear relationship between the price and distance in the SDP research period. In the REPEAT period, however, the price of raw milk has declined in areas farther away from Nairobi. The exception is the areas that are farther than 350 km from Nairobi. The average milk price in these remote markets is higher than the milk prices in the other areas. This could be because the milk markets in these remote areas are isolated from the Nairobi-based milk market, and these areas are considered as milk deficit areas. We also find in table 3 that, in given areas stratified by the distance to Nairobi, the prices paid by cooperatives and traders have no significant differences. This holds both in the late 1990s and Although this may suggest that cooperatives and traders have offered competitive prices locally, dairy cooperatives have lost their share to traders over time as we have seen earlier. We explore the reasons for such a change next. 3.3 Institutional Changes in Milk Marketing Replacing the dairy cooperatives with traders did not take place without problems. While dairy cooperatives purchase raw milk from members throughout the year, some traders tend not to buy milk from dairy households during the rainy season when the milk price is low due to the increased supply. before paying money to the dairy households, 10 In addition, some traders have vanished and some have mixed milk with water to increase the quantity of milk. Dairy cooperatives were thought to prevent such problems through repeated transactions (Staal et al. 1997; Holloway et al. 2000). 10 Some traders pay money to farmers when buying milk. Others, however, make payments after selling milk. 8

10 According to our field surveys, 11 however, the frequency of such problems has declined as private traders themselves gained experience, and trust-based and long-term relationships have been established between traders and dairy households and between traders and retailers. 12 Before the market liberalization, dairy cooperatives used to receive politically determined prices from KCC. Therefore the price paid to dairy households did not vary across areas. Since the liberalization, however, the prices received by the dairy households can differ since the dairy cooperative sell milk to various milk buyers. In addition to such a change in price setting, especially in areas where the dairy cooperatives collapsed, the dairy households who needed to find alternative sales outlets have started selling their own and their neighbors milk in nearby towns. Such small-scale traders come to a specific place (such as road sides in one block of a main market area) and sell raw milk directly to individual consumers. Since vendors tend to set the price paid to dairy households so as to subtract the margin and the transportation cost from the retail price at the town, the producer price should be closely related with the distance from the town to the dairy households. What we found in our interviews is that the activities of vendors became more active just after the dairy cooperatives and KCC collapsed in order to sell their own milk, but it took some time for large-scale traders connecting towns to urban areas to start their business in the raw milk market. This is because trading a larger quantity of 11 Authors revisited some of the study sites in March 2005 and interviewed hawkers and traders to find out how institutions have been changing. 12 As Hayami and Kawagoe (1993) discuss, transactions based on local connections and family ties have the feature of restraining business counterparts from breaking rules. In our study areas, many small vendors collect milk from their neighboring dairy farms. Because of the local connection, it should not be so difficult for vendors to supervise the farms so that the farms do not dilute their milk by adding water. Cheating in transactions between vendors and trader has been kept controlled to some extent by checking the water content randomly. 9

11 raw milk for a long distance requires capital to purchase pick-up trucks. Some large-scale traders established a chain of milk bars as well as a brand name to develop a reputation of good quality. In this way, many large-scale traders who gained capital and experience in raw milk marketing have gradually enlarged the scale of their business. In addition to such a shift from small vendors to large scale traders within the raw milk market business, large-scale traders in the other categories of business who found trading raw milk profitable have entered in the raw milk market as well. 4. Development and Efficiency of the Raw Milk Market 4.1 Hypotheses and Estimation Models As traders enter the milk markets and start transferring milk from distant rural areas to the largest milk market, Nairobi, a local milk price should reflect the distance to Nairobi. Thus, we pose the following hypotheses: Hypothesis 1: In the late 1990s when large-scale traders played a limited role and local markets were isolated, the raw milk price was not determined by the distance to Nairobi but was rather determined by the distance to nearby towns. Hypothesis 2: By 2004, however, the milk market had become more efficient so that the milk price was determined by the distance to Nairobi because of the price arbitrage sought by large-scale traders. To test Hypothesis 1 and 2, we estimate the milk price at the household level for each survey period: k N L (1) pit = α t + β t di + δ t di + γ t X it + ε it, t = 1, 2 where k p it refers to the raw milk price of household i received from milk buyer type k 10

12 in survey round t, N d i is the distance to Nairobi in kilometers from household i, L d i is the distance to the nearest local market of household i, X it contains other household and sub-location characteristics, ε it is the error term, and α,β,δ, and γ are the coefficients to be estimated. Thus, Hypothesis 1 tests for δ1 < 0 and β 1 =0; and Hypothesis 2 tests for β < 2 0. This is quite a different way to test spatial market efficiency than existing studies which use time-series price data (Fackler and Goodwin 2001). One of the reasons is the fact that it is common in developing countries that time-series data are not available for many commodities, including raw milk in Kenya. More importantly, as Barrett (1996) discussed, it is difficult to understand the structural changes of liberalized markets in developing countries from time-series price data since institutional changes need to be taken into account. Thus, we believe that the proposed method will provide a contribution to the existing literature on market analysis methods. In addition to the milk price model, we also examine in which areas traders expanded their activities most: (2) s k jt = α + β d + δ d + γ X + ε, t k t N j k t L j k t jt k jt where k s jt refers to the proportion of dairy households who sold milk to buyer type k of sub-location j in survey period t, and the other variables are as defined previously. We T expect that for trader as sales outlet, the coefficient of distance to market, β, is negative. In addition to the cross-section analysis, we investigate whether the degree of change in the market institution is different depending on the sub-location level characteristics at the initial period. This can be done by taking the change of the 11

13 shares and by using it as a dependent variable. Thus, we estimate the following model: (3) s = 1 ε, k j k N k L k α + β d j + δ d j + γ X j + k j where k s j is change in the proportion, i.e. k k s j s j1 2. Similarly, we test whether the T coefficient of distance to market for trader as sales outlet, β, is negative or not. 4.2 Variables In Equation (1), the dependent variable, p it, raw milk price, is the real producer price per liter for the major outlet. The number of sampled prices is indicated in table 4. The reason why the number of samples more than doubled from the SDP period to the REPEAT period is that for the REPEAT period, the information on milk marketing was collected twice within a year. If the major buyer changed within a REPEAT survey year, then households have price information for each buyer type. As proxies for transportation cost, we use the distance to Nairobi and the distance to the nearest town both in kilometers. In order to incorporate the effects of poor quality roads, we also use total traveling time, instead of the distance. These variables were calculated by SDP, not by the authors, using the methods developed in Staal et al. (2000) which used GIS software. As other explanatory variables, X it, we also control for demand and supply factors in the sub-location measured by the population density and average milk production. 4.3 Results The results from the milk price model in table 4 indicate that the distance to Nairobi is an important determinant of the raw milk price in 2004, while it is not in the late 1990s. 12

14 The coefficient of distance to the nearest town, however, is negative and significant in both periods. This is because vendors sold milk in nearby towns even before the liberalization at the price level depending on the transportation cost from the town. In 2004, private processors and traders, especially large-scale traders, competed to attract dairy households and arbitraged prices across areas. We obtained the same results qualitatively even when traveling time is used instead of distance (see columns B and D) 13. Therefore, the differences in price could be explained by the transportation costs required to deliver milk to Nairobi, which means that the producer price in areas far from Nairobi is lower than that in areas close to Nairobi. The results in table 5 indicate that the distance to Nairobi is a significant determinant of the share of traders in both time periods. This means that the share of traders is higher in sub-locations close to Nairobi. The change in the share of traders given in column (C) is explained by the distance to Nairobi, which suggests that the share of traders increased more in areas close to Nairobi than those far from Nairobi. Similarly, the share of dairy cooperatives can be explained by the distance to Nairobi and the share is higher where it is closer to Nairobi in both years. In the case of the share of dairy cooperatives, however, the change in the share of dairy cooperatives is not influenced by the distance to Nairobi. This may suggest that many dairy cooperatives lost their reliable sales outlet, i.e. KCC, and had to exit the raw milk market. In areas closer to Nairobi, however, dairy cooperatives could find alternative sales outlets relatively easily, thereby resulting in less decline of the share of dairy 13 We include dummy variables for different sales outlets in table 4 to recognize the fact that producer prices are different for different outlets. However, dairy households decide to whom they sell milk, which suggests that the sales outlet dummies are endogenous. Thus, we estimated the models without sales outlet dummies to check the robustness of the results. The results remain the same qualitatively and the endogeneity does not seem serious. 13

15 cooperatives. Thus, it should not be the case that dairy cooperatives newly started operating milk marketing in areas close to Nairobi. The results so far imply the following. Although the producer price of raw milk should be higher in areas closer to Nairobi, the producer price was set relatively low since it was set uniformly across the country by KCC and dairy cooperatives before the liberalization. After the liberalization when traders could sell raw milk in urban areas, however, traders purchased raw milk from dairy households at higher prices than dairy cooperatives, which increased the share of traders. As traders began playing a more active role in milk marketing, price arbitrage has occurred between Nairobi and its surrounding areas and between urban areas. As a result, the market has become more efficient. 5. Conclusions This article examines how the raw milk market in Kenya has been restructured after the market liberalization. It is known that dairy cooperatives in Kenya have contributed to decrease transaction costs and to encourage rural smallholders to adopt dairy cows since such stable milk buyers can decrease the risks associated with milk marketing. According to the panel data of 874 households, however, from the late 1990s and 2004, the proportion of dairy households who sold milk mainly to dairy cooperatives has drastically decreased. Such a restructuring of the market institution did not occur immediately after the liberalization. The milk price analyses indicate that the raw milk price was more efficiently determined in 2004 than in the late 1990s. It is likely that the improved market efficiency was achieved by large-scale traders and private processors that connected rural and urban milk markets. 14

16 As the number of traders has increased, consumers and regulating authorities have questioned the traders handling of raw milk, which may potentially cause hygiene and health problems. raw milk hygienically. Some traders do not have basic knowledge about how to handle Responding to such concerns, however, training and licensing programs for traders has already started on a small scale in Kenya. Since the license can provide a signal for consumers about the quality of milk, some traders find such a licensing system beneficial. Nonetheless, such training and licensing programs should be designed not to impede further development of the market by imposing high entry costs. 15

17 References Badiane, O., and G. Shively Spatial Integration, Transport Costs, and the Response of Local Prices to Policy Changes in Ghana. Journal of Development Economics, 56: Barrett, C Market Analysis Methods: Are Our Enriched Toolkits Well Suited to Enlivened Markets? American Journal of Agricultural Economics 78: Central Bank of Kenya. Statistical Bulletin, June 2002, 2004, Table 6.1. available at and Fackler, P., and B. Goodwin Spatial Price Analysis. In B. L. Gardner and G.. C. Rausser, ed. Handbook of Agricultural Economics. Amsterdam; London and New York: Elsevier Science, North-Holland, pp Fafchamps, M Market Institutions in Sub-Saharan Africa: Theory and Evidence. Cambridge: MIT Press. Fafchamps, M., E. Gabre-Madhin, and B. Minten Increasing Returns and Market Efficiency in Agricultural Trade. MTID Discussion Papers 66, International Food Research Institute. Hayami, Y., and T. Kawagoe The Agrarian Origins of Commerce and Industry: A Study of Peasant Marketing in Indonesia., New York: St Marin s Press. Holloway, G., C. Nicholson, C. Delgado, S. Staal, and S. Ehui Agroindustrialization through Institutional Innovation: Transaction Costs, Cooperatives and Milk-Market Development in the East-African Highlands. Agricultural Economics 23: Jayne, T.S., J. Govereh, A. Mwanaumo, J.K. Nyoro, and A. Chapoto False Promise or False Premise? The Experience of Food and Input Market Reform in Eastern and Southern Africa. World Development 28: Karanja, A The Dairy Industry in Kenya: The Post-Liberalization Agenda. Working Paper 1, Tegemeo Institute, Nairobi, Kenya. Kherallah, M., C. Delgado, E. Gabre-Madhin, N. Minot, and M. Johnson The Road Half-Traveled: Agricultural Market Reform in Sub-Saharan Africa. Food Policy Report, International Food Policy Research Institute. 16

18 Kydd, J., and A. Dorward The Washington Consensus on Poor Country Agriculture: Analysis, Prescription and Institutional Gaps, Development Policy Review 19: Minten, B., and S. Kyle The Effect of Distance and Road Quality on Food Collection, Marketing Margins, and Trader s Wages: Evidence from the Former Zaire. Journal of Development Economics 60: M mboyi, F., A.K. Gem, M. Muyanga, and P. Gamba Consumption Patterns of Dairy Products in Kenya s Urban Centres: A Tool for Leveraging Kenya s Dairy Policy. Working Paper, Tegemeo Institute of Agricultural Policy and Development, Egerton University, Kenya. Moser, C., C. Barrett, and B. Minten Spatial Integration of Rice Markets in Madagascar. Unpublished, Western Michigan University. Negassa, A., Myers, R., and E. Gabre-Madhin Grain Marketing Policy Changes and Spatial Efficiency of Maize and Wheat Markets in Ethiopia. MTID Discussion Paper 66, International Food Policy Research Institute. Ngigi, M The Case of Smallholder Dairying in Eastern Africa. EPT Discussion Paper 131, International Food Policy Research Institute. Osborne, T Imperfect Competition in Agricultural Markets: Evidence from Ethiopia. Journal of Development Economics 76: Owango, M., B. Lukuyu, S.J. Staal, M. Kenyanjui, D. Njubi, and W. Thorpe Dairy Co-operatives and Policy Reform in Kenya: Effects of Livestock Service and Milk Market Liberalization. Food Policy 23 (2): Rashid, S Spatial Integration of Maize Markets in Post-liberalised Uganda. Journal of African Economies 13: SDP The Uncertainty of Cattle Numbers in Kenya. SDP BRIEF 10, Smallholder Dairy (R&D) Project, Nairobi, Kenya. Staal, S.J., C. Delgado, and C. Nicholson Smallholder Dairying Under Transactions Costs in East Africa. World Development 25: Staal, S.J., C. Delgado, I. Baltenweck, R. Kruska Spatial Aspects of Producer Milk Price Formation in Kenya: A Joint Household-GIS Approach. Unpublished, International Livestock Research Institute. Staal, S.J., Owango, M., H. Muriuki, M. Kenyanjui, B. Lukuyu, L. Njoroge, D. Njubi, I. 17

19 Baltenweck, F. Musembi, O. Bwana, K. Nuriuki, G. Gichungu, A. Omore, and W. Thorpe Dairy Systems Cauterization of Greater Nairobi Milk Shed. SDP Collaborative Research Report, Smallholder Dairy (R&D) Project, Nairobi, Kenya. Waithaka, M.M., J.N. Nyangaga, S.J. Staal, A.W. Wokabi, D. Njubi, K.G. Muriuki, L.N. Njoroge, and P.N. Wanjohi Characterization of Dairy Systems in the Western Kenya Region. SDP Collaborative Research Report, Smallholder Dairy (R&D) Project, Nairobi, Kenya. Yamano, T., K. Otsuka, F. Place, Y. Kijima, and J. Nyoro The 2004 REPEAT Survey in Kenya (First Wave): Results. GRIPS Development Database 1, National Graduate Institute of Policy Studies, Tokyo. 18

20 Figure 1. Change in Total Milk Production and Processed Milk in Kenya Million ton total milk production (FAO) total milk production (KDB) milk processed by KCC milk processed by private processors milk processed Note: KDB and FAO mean that data come from Kenya Dairy Board and FAO, respectively. KCC intake and Private processor intake is provided by KDB. 19

21 Figure 2. Liberalization A Simplified Milk Market Structure in Kenya before and after Before Liberalization Rural consumers Dairy Households Private traders Dairy Cooperatives Kiosk(rural rural) KCC (individual, restaurant) urban consumers (individual, restaurant) After Liberalization Rural Dairy Household lds Private traders Traders Kiosk(rural rural) Kiosk(urban urban) consumers (Individual Individual, restaurant) Urban consumers (Individual Individual, Cooperatives restaurant) Processors Raw milk Processed milk 20

22 Table 1. Sampled Households in the original SDP and REPEAT Surveys Province Number of households Per capita income (2004) Milk income share (2004) Percent of households who sold milk 1 st round (1996,1998, 2000) 2 nd round (2004) (A) (B) (C) (D) (E) Number US $ % % % Eastern Central Rift Valley Western Nyanza Total

23 Table 2. Distribution of Households Sold Milk by Sales Outlet Distance to Nairobi a Number of household s Percentage of households who sold milk Local Customers b By the largest milk buyer type Cooperativ es Traders Private Processor s KCC (A) (B) (C) (D) (E) (F) (G) Number % of A % of B % of B % of B % of B % of B SDP 0-100km km km km Total REPEAT 0-100km km km km Total Note: a Distance to Nairobi is categorized so that the number of raw milk price samples in each category is similar. b Local Customers includes restaurants, hotels, kiosks, and individual customers. 22

24 Table 3. Raw Milk Producer Price by Sales Outlet and Distance to Nairobi Local Private Distance to Cooperatives Traders KCC Customers Processors Nairobi (A) (B) (C) (D) (E) Shs Shs Shs Shs Shs SDP 0-100km N.A. N.A km km km N.A. N.A. N.A Total REPEAT 0-100km km km 16.1 N.A km N.A N.A. N.A. Total Note: All milk prices are adjusted to 2004 price levels. We use the inflation rates in Statistical Bulletin, Central Bank of Kenya 2002, 2004, and Figures are averages of household-level milk prices which belong to the respective categories. 23

25 Table 4. Determinants of the Producer Price of Raw Milk (Household Level) Raw milk price (Sh) SDP REPEAT (A) (B) (C) (D) Distance to Nairobi(km) ** (-1.87) (-5.88) Remote area (in terms of distance) dummy ** (distance to Nairobi is farther than 250km) (-0.78) (-6.94) Distance to Nairobi) Remote area dummy ** (1.68) (8.26) Distance to nearby town(km) ** ** (-4.49) (-5.92) Traveling time to Nairobi(km) ** (-1.75) (-5.71) Remote area (in terms of traveling time) dummy ** (Traveling time to Nairobi is more than 3.5 hrs) (-0.78) (-6.67) Time to Nairobi Remote area dummy ** (1.73) (8.08) Traveling time to nearby town (hrs) ** ** (-4.79) (-6.16) Milk was sold to traders =1 a (-0.09) (-0.18) (0.94) (0.91) Milk was sold to neighbors, restaurants, 2.678** 2.600** 2.371** 2.399** Shops =1 a (3.41) (3.34) (3.80) (3.86) Milk was sold to KCC =1 a (0.87) (0.86) (1.50) (1.58) Milk was sold to processors =1 a (1.09) (1.14) (1.35) (1.43) Population density in sublocation (100 persons/km 2 ) ** ** (-1.53) (-1.32) (-3.06) (-2.66) Average milk production in sublocation (liter/1000) ** ** (leave-out mean) (-1.90) (-1.85) (-4.14) (-4.33) Constant 26.45** 26.42** 20.46** 20.43** (19.95) (20.40) (23.30) (23.48) F-Statistics for all distance variables Adjusted R Number of observations Note: * and ** indicate 5% and 1% significance levels, respectively. a A comparison group for sales outlet dummies is dairy cooperatives. 24

26 Table 5. Determinants of the Structure of the Raw Milk Market (Sub-location Level, Tobit Model) Share of dairy HHs sold to traders Share of dairy HHs sold to cooperative SDP REPEAT Change a SDP REPEAT Change a (A) (B) (C) (D) (E) (F) Distance to Nairobi (km) * ** * ** ** (-1.91) (-3.42) (-2.37) (-4.22) (-3.49) (-1.37) Distance to nearby town (km) (-0.15) (-0.82) (-0.99) (-1.02) (-1.01) (-1.48) Population density (persons/km 2 ) (-0.72) (-0.54) (0.24) (-0.97) (-1.48) (-0.73) Total milk production 0.037* ** (liter/1000) (1.91) (0.93) (0.89) (2.68) (1.42) (0.88) Constant * ** 0.578* (-0.59) (2.41) (1.13) (2.71) (2.39) (0.61) F-stat: all distance variables ** 4.22** 10.20** 7.59** 1.64 Log likelihood Pseudo R Number of observations Note: * and ** indicate 5% and 1% significance levels, respectively. a In regressions for change, the explanatory variables are those in the SDP period. 25

27 Notes 26