R Cover Photos by: Ashok Gulati, IFPRI and Nicholas Minot, IFPRI Agricultural commercialization and diversification in Bhutan i Agricultural commercialization and diversification in Bhutan August 2010 Prepared by: Policy & Planning Division (PPD) Ministry of Agriculture and Forests (MoAF) As part of the joint SDC-IFPRI-Bhutan project “Agricultural and Food Policy Research and Capacity Strengthening” iii TABle oF ConTenTS List of Tables ........................................................................................................................................................... v List of Figures........................................................................................................................................................ vii List of Maps ........................................................................................................................................................... ix Acronyms ............................................................................................................................................................... xi Acknowledgements ............................................................................................................................................. xiii Executive summary .............................................................................................................................................. xv 1. Introduction ...................................................................................................................................................... 1 2. Data and methods ............................................................................................................................................ 1 3. Patterns and trends in diversification and commercialization ..........................................................................3 4. Diversification and commercialization within the crop sector .........................................................................8 5. Diversification and commercialization within the livestock sector .................................................................28 6. Diversification, commercialization, and welfare .............................................................................................34 7. Diversification, commercialization, and remoteness ......................................................................................39 8. Summary and conclusions .............................................................................................................................. 47 9. References ...................................................................................................................................................... 49 v lIST oF TABleS Table 1. Crop area cultivation in Bhutan, 2000 and 2008 ......................................................................................3 Table 2. Livestock production in Bhutan, 2000 and 2008 ......................................................................................6 Table 3. Evolution major agricultural commodity prices in Bhutan (Nominal; Nu/kg) ...........................................7 Table 4. Shares of crops in harvested area, total production and sales in Bhutan, 2000 and 2008 (identical product list in 2000 and 2008) ..................................................................................................8 Table 5. Measures of diversification ..................................................................................................................... 11 Table 6. Crop-wise harvested area, production, sales, and productivity for comparable crops in 2000 and 2008, in quantities ..............................................................................................................13 Table 7. Crop-wise harvested area, production, sales, and productivity for comparable crops in 2000 and 2008, in values ....................................................................................................................14 Table 8. Crop-wise harvested area, production, and sales in 2008, by farm size ................................................17 Table 9. Determinants of crop diversification and crop commercialization .........................................................19 Table 10. Herd size, value of livestock, livestock products, and sales of livestock products in 2008, by farm size ............................................................................................................................................ 30 Table 11. Determinants of commercialization of live animals and livestock products ........................................31 Table 12. Crop area, value of production, and crop sales by quality of housing indicator (2000) .......................35 Table 13. Importance of different crops in total sales by remoteness tercile (hours to motorable road), 2000 .44 Table 14. Changes in production and sales by remoteness of geogs, 2000 .........................................................46 Table 15. Decomposition of changes in production and sales by remoteness of geogs, 2000 ............................46 v vii lIST oF FIguReS Figure 1. Sales and own use of different types of crops, 2008 (Million Nu) ...........................................................9 Figure 2. Crop composition of harvested area, total production, and total sales, 2008 (100%=total) ................12 Figure 3. Crop composition in area harvested, by quintile of area harvested 2008 (100%=total) .......................15 Figure 4. Crop composition in value of production, by quintile of area harvested 2008 (100%=total) ................16 Figure 5. Crop composition in value of sales, by quintile of area harvested 2008 (100%=total) ..........................16 Figure 6. Lorenz-curve of harvested area, the value of production, and value of sales over households, 2008.....18 Figure 7. Value of output and sales (Nu/acre) by quality of housing indicator, 2000 ..........................................36 Figure 8. Prices of local white rice, potato, and tomato 1996-2006 (in Gelephu) ...............................................37 Figure 9. Seasonal index of rice prices 1996-2006 (in Gelephu) ..........................................................................38 Figure 10. Seasonal index of potato prices 1996-2006 (in Gelephu) ...................................................................38 Figure 11. Seasonal index of tomato prices 1996-2006 (in Gelephu) ..................................................................39 Figure 12. Distribution of households by distance to a motorable road, 2000 ...................................................40 Figure 13. Crop area allocation by remoteness tercile, 2000 ..............................................................................42 Figure 14. Rice yields (kg/acre) by remoteness (hours to motorable road), 2000 ..............................................43 Figure 15. Commercialization rates (value of sales/value of production), by remoteness (hours to motorable road), 2000 ........................................................................................................................ 45 Figure 16. Value of output and sales per unit of land, by remoteness (hours to motorable road), 2000 ...........45 ix lIST oF MAPS Map 1. Cropped area per geog in 2000 (RNR Census 2000)...................................................................................4 Map 2. Cropped area per geog in 2008 (RNR Census 2009)...................................................................................4 Map 3. Commercialization rate (%) in 2008 (RNR Census 2009) ............................................................................5 Map 4. Value of livestock assets in 2008 (RNR Census 2009) .................................................................................7 Map 5. Share of non-ceareals in total cropped area in 2000 (RNR Census 2000) ................................................10 Map 6. Share of non-ceareals in total cropped area in 2008 (RNR Census 2009) ................................................10 Map 7. Harvested area of maize in 2000 (RNR Census 2000) ..............................................................................21 Map 8. Harvested area of maize in 2008 (RNR Census 2009) ..............................................................................21 Map 9. Harvested area of paddy in 2000 (RNR Census 2000) ..............................................................................22 Map 10. Harvested area of paddy in 2008 (RNR Census 2009) ............................................................................22 Map 11. Harvested area of potato in 2000 (RNR Census 2000) ...........................................................................23 Map 12. Harvested area of potato in 2008 (RNR Census 2009) ...........................................................................23 Map 13. Harvested area of oranges in 2000 (RNR Census 2000) .........................................................................24 Map 14. Harvested area of oranges in 2008 (RNR Census 2009) .........................................................................24 Map 15. Harvested area of apple in 2000 (RNR Census 2000) .............................................................................25 Map 16. Harvested area of apple in 2008 (RNR Census 2009) .............................................................................25 Map 17. Total crop sales per geog in 2008 (RNR Census 2009) ............................................................................26 Map 18. Share of crop sales from paddy in 2008 (RNR Census 2009) ..................................................................26 Map 19. Share of crop sales from potato in 2008 (RNR Census 2009) .................................................................27 Map 20. Share of crop sales from oranges in 2008 (RNR Census 2009) ...............................................................27 Map 21 . Share of crop sales from apple in 2008 (RNR Census 2009) ..................................................................28 Map 22. % of households that hold livestock in 2008 (RNR Census 2009) ..........................................................32 Map 23. Number of cattle in 2008 (RNR Census 2009) ........................................................................................32 Map 24. Share of improved cattle in cattle population in 2000 (RNR Census 2009) ...........................................33 Map 25. Share of cross-bred cattle in cattle population in 2008 (RNR Census 2009) ..........................................33 Map 26. Number of yaks in 2008 (RNR Census 2009) ..........................................................................................34 Map 27. Time required to get to a motorable road in 2000 (RNR Census 2000) .................................................41 xixi ACRonyMS AMS Agricultural Marketing Services ATT Average Treatment Effect for the Treated BLSS Bhutan Living Standards Survey CoRRB Council for Renewable Natural Resources Research of Bhutan DAMC Department of Agricultural Marketing and Cooperatives DoA Department of Agriculture DoL Department of Livestock FAO Food and Agricultural Organization FCB Food Corporation of Bhutan FYP Five Year Plan GDP Gross Domestic Product IFPRI International Food Policy Research Institute kg Kilogram MoAF Ministry of Agriculture and Forests mt Metric tonnes NSB National Statistical Bureau Nu Ngultrum PPD Policy and Planning Division PSM Propensity Score Matching PSU Primary Sampling Unit SSU Secondary Sampling Unit RGoB Royal Government of Bhutan RNR Renewable Natural Resources SDC Swiss Agency for Development Cooperation xiii ACknowleDgeMenT This research was supported by the “Agricultural and Food Policy Research and Capacity Strengthening Project,” jointly funded by the Swiss Agency for Development Cooperation (SDC), the International Food Policy Research Institute (IFPRI), and the Ministry of Agriculture and Forests (MoAF) of the Royal Government of Bhutan (RGoB). The project was implemented under the overall management of Tenzin Chopel (Chief Planning Officer, Policy and Planning Division, MoAF), Ashok Gulati (Director in Asia, IFPRI), and Lisa Magnollay (Programme manager, SDC). The report was written by Bart Minten (IFPRI) and Kailash Pradhan (CoRRB). It benefitted from discussions with other members of the MoAF-IFPRI project team: Nicholas Minot (IFPRI), Chencho Dukpa (CoRRB), Phub Dem (DAMC), and Nidup Peljor (PPD). The authors would like to acknowledge the data management assistance of Reno Dewina (IFPRI). They would like to thank for the support and assistance received from the IFPRI office based at New Delhi and Washington DC (USA), the Ministry of Agriculture and Forests (MoAF), and the Council for Renewable Natural Resources Research of Bhutan (CoRRB). The authors are grateful to the National Statistical Bureau (NSB) which generously provided data from the Population Census, the Bhutan Living Standards Surveys, and other datasets. They would also like to thank Karpo Dukpa (Statistical Investigator, PPD) for providing the 2009 RNR Census data for analysis and Vaishali Dassani and Task Digital for coordinating the design, the format- ting, and the printing of this booklet. Finally, the paper benefited from constructive feedback from participants at an interim workshop on 14 January 2010 and the final project workshop on 2 July 2010, where earlier versions of the paper were presented. However, any opinions stated in this report are only those of the authors and do not necessarily reflect the policies or opinions of the SDC, the RGB, or IFPRI. xv eXeCuTIVe SuMMARy Bhutan has shown good economic growth (9% annually) in the 9th Five-Year Plan (FYP) (2002-2007) and is further continuing on this growth path in the 10th FYP. As domestic incomes rise because of this growth, Bhutanese consumers are expected to shift their consumption patterns from staple grains to fruits, vege- tables, dairy, eggs, and meat, leading to an increase in demand for these high-value agricultural products. Likewise, as Bhutan becomes integrated into the regional and global economy, farmers will diversify into high-value agricultural products that are in demand in urban areas of its South-Asian neighbors and in high- income countries. This is particularly true since, within South-Asia, Bhutan has a comparative advantage in temperate and sub-tropical commodities. In this report, we define agricultural diversification as the shift in production from low-value staple crops, such as maize and rice, into higher-value commodities such as fruits, vegetables, medicinal plants, and animal products. Commercialization refers to the trend toward increasing the proportion of agricultural production that is sold by farmers. More in particular, this theme examines the following research questions: 1/ What are the patterns and trends in agricultural diversification and commercialization in Bhutan? 2/ Are households that are more diversified into high-value commodities and more commercialized better off than other rural households? 3/ What are the constraints preventing farmers from diversifying into higher-value commodities and becoming more commercially oriented? Available data in Bhutan allow us to carry out two types of analysis on agricultural diversification and commer- cialization: a geog-level analysis of the Renewable Natural Resource (RNR) Census data of 2000 and a geog-level trend analysis comparing the 2000 and 2009 Censuses. In 2000, the Ministry of Agriculture and Forests organized a RNR Census related to rural farming. The aim was to visit all rural households in Bhutan but the actual coverage achieved was about 87%, a lower number mainly due to absenteeism of households at the time of survey. The questionnaire implemented focused on information related to area cultivated of different crops, production and yield levels, commercialization, and the use of agricultural inputs. A similar Census was conducted in the beginning of the year 2009 by the Policy and Planning Division of the Ministry of Agriculture and Forests. Data were collected for all agricultural activities in the previous year 2008. 57,606 households could be interviewed, covering about 93% of all the households that are involved in farming, livestock, or forestry activities. Crop areas in Bhutan are dominated by cereals, mainly rice and maize. Based on the RNR Census of 2000, it is estimated that cereals make up 79% of the area cultivated in Bhutan. However, they account for much less in value terms as especially fruits and vegetables are relatively much higher valued. It is thus estimated that the cereals only made up 64% of the total value of production in 2000. The importance of cereals is even less when one looks at the commercial value. In this case, especially orange, potato, and apples make up the largest share. However, only a limited number of households participate in the marketing of these crops as it is estimated that 10% of all the rural households take care of 73% of all crop sales (based on data from 2008). xvi When the results of the RNR of 2000 and 2009 are compared, we note a seemingly fast crop diversification over time. While the share of cereals accounted for 79% of the cropped area in 2000, this declined to 65% in 2008. The changes are even more pronounced when the value of crop production and of crop sales are used (given the higher outputs in value terms per unit of land from non-cereal crops). The importance of cereals declined in the value of production from 65% to 53% between 2000 and 2008. We see an increasing share of high-value crops, leading to higher outputs in value terms per unit of land, commonly referred to as land intensification. On the other hand, the livestock population seems to be stabilizing but the share of cross-bred and improved cattle is on the rise, leading to higher rates of commercialization. When we look at the association between diversification, commercialization and welfare, it is found that richer geogs are better willing and able to diversify their sales in a larger number of crops than the poorer ones. As commercial cropping is riskier, only the relatively wealthier geogs are able to take on these risks but these geogs that diversify into more market-oriented crops might earn higher profits and be better off because of that. It is also found that richer geogs utilize their land much more intensively than the poorer ones. Remoteness is an important factor associated with diversification and commercialization in Bhutan. More remote geogs grow different crops and have lower yields than the less remote geogs, leading to much higher values of output per unit of land. The more remote geogs also have much lower commercialization rates of their outputs: keeping prices of the same product constant, the value of sales per unit of land is almost four times as high for the close by geogs compared to the most remote ones. Diversification in other crops than cereals is an economic process driven by increasing incomes in Bhutan and its neighbors as well as by better access to markets. Diversification away from staples is also associated with increasing trade between Bhutan and other countries. Those geogs that are able to participate in this commer- cial agriculture are associated with higher welfare levels. As most growth in the agricultural sector in the next decade is expected from the high-value sector, it is important that a shift in policies (e.g. emphasizing market information, grades and standards, contract farming, extension and cold storages) follows the shift in consump- tion and production patterns as to allow farmers to better profit from these newly emerging opportunities in the agricultural sector. 1 1. InTRoDuCTIon Bhutan has been characterized by good economic growth (9% annually) in the 9th Five-Year Plan (2002-2007) and is further continuing on this growth path in the 10th FYP. As domestic incomes supposedly rise because of this growth, Bhutanese consumers are expected to shift their consumption patterns from staple grains to fruits, vegetables, dairy, eggs, and meat, leading to an increase in demand for these high-value agricultural products. Likewise, as Bhutan becomes integrated into the regional and global economy, farmers will diversify into high- value agricultural products that are in demand in urban areas of its South-Asian neighbors and in high-income countries. This is particularly true since, within South-Asia, Bhutan has a comparative advantage in temperate and sub-tropical commodities. In this report, we define agricultural diversification as the shift in production from low-value staple crops, such as maize and rice, into higher-value commodities such as fruits, vegetables, medicinal plants, and animal products. While diversification sometimes refers to an increase in the number of crops or income sources, we will devote relatively less attention to this concept, partly because it is not related to income growth and partly because existing data sources do not allow us to generate good indicators. Commercialization refers to the trend toward increasing the proportion of agricultural production that is sold by farmers. More in particular, this theme will examine the following research questions: What are the patterns and trends in agricultural diversification and commercialization in Bhutan? � Are households that are more diversified into high-value commodities and more commercialized better � off than other rural households? What are the main constraints preventing farmers from diversifying into higher-value commodities and � becoming more commercially oriented? More specifically, what is the role that access to roads plays in agricultural diversification and commercialization? The structure of the document is as follows. In Section 2, we discuss the data and methods that are used in the analysis. Section 3 looks at patterns and trends in diversification and commercialization. Section 4 and 5 discuss diversification and commercialization in the crop and livestock sector respectively. Section 6 describes the link between welfare, diversification and commercialization. Section 7 analyzes the effect of remoteness on diversification and commercialization. We finish with summary and conclusions in Section 8. 2. DATA AnD MeThoDS Available data in Bhutan allow us to carry out three types of analysis with respect to agricultural diversification and commercialization: a geog-level analysis of the Renewable Natural Resource (RNR) Census data of 2009, a geog-level trend analysis comparing the 2000 and 2009 RNR Census data, and an analysis looking at the link of remoteness and welfare with diversification and commercialization at the geog level. In 2000, the Ministry of Agriculture and Forests organized a Renewable Natural Resource (RNR) Census related to rural farming (RGoB, 2000). The aim was to visit all rural households in Bhutan but the actual coverage achieved was about 87%, a lower number mainly due to absenteeism of households at the time of survey (RGB, 2000). The questionnaire implemented focused on information related to area cultivated of different crops, production and yield levels, commercialization, and the use of agricultural inputs. The census also collected 2 information on livestock ownership and sales of livestock products. Dzongkhag and geog level estimates were generated from this survey for different indicators related to agriculture. A similar Census was fielded in the beginning of 2009. This RNR Census was coordinated by the Policy and Planning Division (PPD) of the Ministry of Agriculture and Forest (MoAF). About 700 enumerators and 60 supervisors were employed to implement the survey. 57,606 households were interviewed, covering 93% of all households in Bhutan that are involved in farming, livestock, or forestry activities. The data that were collected concerned land holdings and tenure, crop and livestock productions, forest production and utili- zation, agricultural inputs, accessibility, quantities marketed and prices received, and farming constraints. These two datasets will be used in our study on agricultural diversification and commercialization. In the analysis of these data, extrapolation coefficients are used that allow for the estimation of agricultural indica- tors at the national level. In the study, we use several measures of diversification and commercialization for each geog as follows: Diversification – The share of crop land in fruits and vegetables. � Diversification – The share of crop land in non-grain crops. � Diversification – The herd size and the value of livestock owned. � Commercialization – The share of the value of crop production that is marketed. � Commercialization – The sale of livestock products. � Most of these variables are mapped using geog-level digitized maps. In addition, the analysis examines the correlations among indicators of diversification and commercialization on the one hand, and well-being and remoteness on the other hand. The analysis of trends in commercialization is based on a comparison of the data from the 2000 and 2009 RNR Census. Given that no price data were collected in the RNR Census 2000, we used the price data for the different crops obtained at the geog level in 2008 to value agricultural production as well as livestock products in 2000 as well. While not perfect, this allows however for easy comparison over time. None of the surveys collected any information on the area under fruit trees. To allow for comparison of areas under fruit trees with other crops, we assumed a fixed number of fruit trees per unit area as per the technical recommendations. This assumption thus allows us to aggregate all cultivated crop areas. We will present most of the analysis in the tables by region. Four regions are commonly used in Bhutan, i.e. West, West-Central, East-Central, and East. Different dzongkhags were thus assigned to these different regions as follows: the West includes the dzongkhags of Chhukha, Haa, Paro, Samtse, and Thimphu; The West- Central includes the dzongkhags of Dagana, Gasa, Punakha, Tsirang, and Wangdue; The East-Central includes the dzongkhags of Bumthang, Sarpang, Trongsa, and Zhemgang; and the Eastern region includes the dzongkhags of Lhuentse, Mongar, Pemagatshel, Samdrupjongkhar, Trashigang, and Tashiyangtse. For the analysis with respect to remoteness and welfare, we constructed two variables based on data collected at the geog level. First, we constructed a housing quality index as a proxy for welfare at the geog level. The housing quality index is a simple sum of the percentage of households whose houses show five “quality” characteristics related to type of light, water use, roof material, wall material, and toilets. Second, informa- tion was collected in the census related to the number of households that had to walk less than 30 minutes, between 30 and 60 minutes, between 1 and 2 hours, between 2 and 3 hours, between 3 and 4 hours, between 4 and 5 hours, between 5 and 6 hours, and more than 6 hours to a motorable road. A remoteness index was then constructed by using the time in hours that average households in that geog had to walk to get to the motorable road. 3 3. PATTeRnS AnD TRenDS In DIVeRSIFICATIon AnD CoMMeRCIAlIzATIon To start off the analysis on patterns and trends in diversification and commercialization of the agricultural sector in Bhutan, we first present a comparison of the crop sector with the livestock sector overall. Table 1 shows the cropped area and crop production statistics for 2000 and 2008 at the regional level. Given that more crops were covered in the Census of 2008, we present statistics based on the crops that are comparable to 2000 and on all crops reported in the Census of 2008. The comparable numbers illustrate that crop production has increased by 31% (higher than population growth that was estimated around 15% over that period), seemingly driven by two processes. First, there is a trend of land extensification given the slightly larger areas under crop cultivation in 2008 compared to 2000. A comparison between 2008 and 2000 indicates an increase of almost 7% if we look at comparable crops in 2008 and 2000. Given that overall reported cultivated land is going down (MoAF, 2010), this indicates that there is relatively more double cropping in 2008 than in 2000 (where vegetables are for example grown in off-season on the wetlands). Second, there is a process of land intensification, as the value of output generated on an acre of land has significantly increased over time. The average land intensifica- tion increase in value term at the national level over that period is estimated at about 22% (an increase of 65,000 Nu/ha in 2000 to 80,000 Nu/ha in 2008). This increase has happened in all four regions, but apparently less so in the East-central region. Note that prices were kept constant over time and this increase is thus purely driven by changes in crop composition as well as by higher physical yields. This is discussed further in Section 4. Table 1. Crop area cultivation in Bhutan, 2000 and 2008 West West-central East-central East Bhutan Total Share (%) Total Share (%) Total Share (%) Total Share (%) Total Share (%) Crop area 2000 (ha) 23,856 28 17,527 21 16,067 19 26,426 32 83,876 100 Crop area 2008 (ha), comparable crops as in 2000 23,578 26 21,196 24 13,744 15 31,222 35 89,739 100 Crop area 2008 (ha), all crops reported in 2008 25,651 26 23,032 23 14,977 15 34,529 35 98,190 100 Value production 2000 (mill Nu) 1,561 29 1,414 26 1,007 18 1,488 27 5,471 100 Value production 2008 (mill Nu), comparable crops in 2000 1,890 26 2,115 30 1,077 15 2,064 29 7,146 100 Value production 2008 (mill Nu), all crops reported in 2008 2,005 26 2,239 29 1,151 15 2,247 29 7,642 100 Land productivity 2000 (1000 Nu/ha) 65 81 63 56 65 Land productivity 2008 (1000 Nu/ha), comparable crops in 2000 80 100 78 66 80 Values sales 2000 (mill Nu) 533 40 245 18 275 21 275 21 1,328 100 Values sales 2008 (mill Nu), comparable crops in 2000 533 32 460 28 263 16 387 24 1,643 100 Values sales 2008 (mill Nu), all crops reported in 2008 549 32 482 28 270 16 402 24 1,702 100 Commercialization rate 2000 (%) 34 17 27 19 24 Commercialization rate 2008 (%), comparable crops in 2000 28 22 24 19 23 Commercialization rate 2008 (%), all crops reported in 2008 27 22 24 18 22 Source: Authors' calculations based on RNR Census 2000 and RNR Census 2009 4 Maps 1 and 2 show how total cropped area is distributed and how it has changed over time for the different geogs, based on the RNR Census of 2000 and 2008. The maps show that, given agro-ecological constraints, most of the cropped area is situated in the southern part of the country as this area is characterized by most shaded geogs on the maps. The maps also show where total cropped area is expanding over time. We see especially expansion in the North-east, Central-west and the extreme West of the country. Map 1 Cropped area per geog in 2000 (RNR Census 2000) Hectares 700 − 2000 500 − 700 300 − 500 100 − 300 0 − 100 No data Cropped area per geog in 2000 (RNR Census 2000) Map 2 Cropped area per geog in 2008 (RNR Census 2009) Hectares 700 − 2000 500 − 700 300 − 500 100 − 300 0 − 100 Cropped area per geog in 2008 (RNR Census 2009) 5 Sales data for crops were also available in both 2000 and 2008. Using these data, the value of crop sales amounted to Nu 1.3 billion in 2000 and increased to Nu 1.6 billion in 2008, an increase of 24% (Table 1). When the sales value is divided by the production value, it is estimated that slightly less than one quarter of the value of production was marketed in 2000. This commercialization rate has changed little over time indicating absolute increases in own use of crop production as well as in total sales. The ‘commercialization rate’ is highest in the West of country (32%) and lowest in the East-Central (16%) (Table 1).The value of the crop sales in the West makes up 40% of the total sales of the country. Map 3 shows that commercialization rates differ significantly over geogs. Especially the South-western and North-central region of the country is most integrated in the commercial agricultural economy as seen by the high prevalence of darkly shaded geogs (the two darkest geogs are geogs where between 50 and 100% of crop production is commercialized). Little of the crop production is commercial- ized in the North-western part and the extreme East of the country. Map 3 Commercialization rate (%) in 2008 (RNR Census 2009) % 75 − 100 50 − 75 25 − 50 15 − 25 5 − 15 0 − 5 Commercialization rate (%) in 2008 (RNR Census 2009) While crop production seems to expand faster than population growth over time, this is not the case in the livestock sector as herd sizes are rather stable or are declining over time. The herd size of large cattle1 in 2000 was estimated at 357,246 heads and it decreased to 326,525 heads in 2008, a decrease of 8% (Table 2). The value of the livestock decreased slightly over that same period. The value of livestock is highest in the East and the West. Each of these regions represents about one third of the value of livestock in the country. Given the differences in the sales of livestock products recorded in the two surveys, it is difficult to evaluate their changes over time. However, for the comparable products, we see a significant increase over time of commercialized quantities (more than a doubling). 1 Including cattle, buffaloes and yaks 6 Table 2. livestock production in Bhutan, 2000 and 2008 West West-central East-central East Bhutan Total Share (%) Total Share (%) Total Share (%) Total Share (%) Total Share (%) Herd size 2000 of major livestock* 101,297 28 70,493 20 63,202 18 122,254 34 357,246 100 Herd size 2008 of major livestock* 104,332 32 78,684 24 36,513 11 106,996 33 326,525 100 Value of livestock 2000 (mill Nu) 913 32 794 27 416 14 770 27 2,892 100 Value of livestock 2008 (mill Nu) 1,119 36 718 23 431 14 817 26 3,086 100 Value of livestock products 2000 (mill Nu) 260 25 200 19 188 18 393 38 1,041 100 Value of livestock products 2008 (mill Nu), comparable products as in 2000 292 27 214 20 174 16 390 36 1,071 100 Value of livestock products 2008 (mill Nu), all reported products as in 2008 299 28 218 20 175 16 392 36 1,084 100 Values sales livestock product 2000 (mill Nu) 27 25 17 16 18 17 47 43 110 100 Values sales livestock product 2008 (mill Nu), comparable products as in 2000 107 28 73 19 63 16 145 37 388 100 Values sales livestock product 2008 (mill Nu), all reported products as in 2008 111 28 74 19 64 16 146 37 394 100 Source: Authors' calculations based on RNR Census 2000 and RNR Census 2009 * including cattle, yaks, and buffaloes; based on the official published numbers of RNR Census 2000 and 2009 Table 2 shows that the sale of livestock products was valued at 19% of the value of crop sales in 2008 and the livestock sector thus seems less important for monetary income than the crop sector. However, this number does not include the sales of live animals as these data were not available for the year 2000.2 As we used same prices for the crop and livestock sector, we are unable to effectively value changes over time. Table 3 shows the nominal price evolution of some major agricultural commodities between 2003 and 2008. Taking a simple average of the changes between 2008 and 2003 over these commodities, it is seen that livestock prices have risen faster, at 54%, than prices in the crop sector, at 31%. It is thus possible that the relative value of livestock sector compared to the crop sector has increased over time. In contrast with crop agriculture that is very concentrated in the southern geogs, livestock production is more evenly spread over the country. Map 4 shows the spatial distribution of the value of livestock in 2008. The heaviest concentration of the livestock is in the North and in the West of the country. The areas in the south are in general ‘lighter’ and seem thus more specialized into crop agriculture than livestock. 2 There might also been some underreporting of the production and sales of livestock products as other sources indicate that livestock products are more important: a/ In their National Accounts’ Statistics, the RGoB (2009) estimate that the GDP of the livestock sector makes up 63% of the GDP of the crop sector in 2008 (1886 Million Nu versus 2987 Million Nu respectively); b/ Data from the national household survey (BLSS 2007) show that the value of the consumption of livestock products (meat, dairy products, and fish) make up 64% of the value of crop consumption (cereals, pulses, vegetables, oil, spices); c/ There are some discrepancies with the numbers published by the Ministry of Livestock (2009) and the numbers of the Census. For example, cheese production and marketing is estimated to be more than twice as high in 2008 by the Ministry of Livestock (2009) than data from the Census. More detailed analysis is required to understand the differences between the different datasets. They are partly explained by differences in methodology as National Accounting Statistics are generated based on fixed conversion ratios. 7 Table 3. evolution major agricultural commodity prices in Bhutan (nominal; nu/kg) year change 2008/2000 2003 2004 2005 2006 2007 2008 % Crop sector Wheat 9.0 8.1 8.1 9.6 9.2 12.8 42.2 Rice 20.8 25.6 25.1 28.2 30.3 35.4 70.2 Maize 9.3 9.5 9.5 9.5 9.5 10.3 10.8 Potato 8.8 9.5 8.5 11.0 10.9 11.6 31.8 Spices 34.6 36.7 38.1 37.2 39.0 40.3 16.5 Vegetables 23.9 23.5 21.2 24.7 23.4 31.0 29.7 Apples 45.4 48.8 48.0 45.6 46.9 49.7 9.4 Oranges 35.8 30.2 29.1 27.9 24.6 26.3 -26.5 Other fruits 29.4 30.1 28.3 26.9 26.4 24.4 -17.2 Simple average crop sector 30.8 Livestock sector Butter 146.0 150.0 155.0 160.0 165.0 180.0 23.3 Cheese 60.0 63.0 65.0 67.0 70.0 80.0 33.3 Beef 52.0 55.0 60.0 66.0 70.0 90.0 73.1 Pork 71.0 73.0 75.0 80.0 84.0 100.0 40.8 Mutton 110.0 120.0 122.0 130.0 140.0 140.0 27.3 Chicken 62.0 65.0 70.0 74.0 80.0 100.0 61.3 Egg (dozen) 27.0 30.0 33.0 35.0 40.0 60.0 122.2 Simple average livestock sector 54.5 Source: RNR Countrystat Bhutan Map 4 Value of livestock assets in 2008 (RNR Census 2009) Million Nu 30 − 400 20 − 30 10 − 20 5 − 10 0 − 5 Value of livestock assets in 2008 (RNR Census 2009) 8 4. DIVeRSIFICATIon AnD CoMMeRCIAlIzATIon wIThIn The CRoP SeCToR While we have noted the changes in total cropped area over time, we see also changes within the crop sector itself. Table 4 shows the importance of the different crops in harvested area (and thus potentially including two harvest per year for the same plot), in total production, and in total sales in 2000 and 2008. In 2008, almost two-thirds of the cropped area was allocated to cereals (including paddy, wheat, barley, buckwheat, millet, and maize). However, in value terms, their share was only 53%. Vegetables came second in 2008, estimated at 25% of the total value while the share of fruits is almost 20%. Cereals accounted for 79% of the cropped area in 2000 and for 64% of the value of crop production. Their share had declined in 2008 in the cropped area to 65% and in the total value of production to 53%. The impor- tance of cereals in value terms decreased thus by almost 11% over a 8-year span. Especially vegetable produc- tion has increased significantly: its share in total area increased from 9% to 22% between 2000 and 2008 and its share in total value increased from 16% in 2000 to 25% in 2008. When we look at the reported sales of crops, the importance of fruits is overwhelming (they are however less important in the cropped share, at 7% of the total area). Fruits count for 56% of the value of sales, vegetables for 36%, and cereals only count for 6%. While the latter might be an underestimation, it however shows to what extent fruits are major cash earners in crop agriculture in Bhutan. Table 4. Shares of crops in harvested area, total production and sales in Bhutan, 2000 and 2008 (identical product list in 2000 and 2008) West West-central East-central East Bhutan 2000 2008 2000 2008 2000 2008 2000 2008 2000 2008 Cropped area Cereals 76.9 60.2 78.8 67.9 83.7 69.9 78.4 63.2 79.1 64.5 Vegetables 10.6 25.0 9.2 18.8 5.2 16.2 9.3 23.6 8.9 21.7 Fruits 7.8 8.7 3.9 6.2 6.0 8.5 3.0 5.0 5.1 6.8 Other 4.7 6.1 8.1 7.1 5.0 5.4 9.3 8.3 6.9 7.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Value of production Cereals 55.8 52.2 73.2 60.8 64.5 53.1 63.1 44.6 63.9 52.7 Vegetables 19.8 26.1 14.0 18.4 8.0 17.7 17.9 33.3 15.6 24.6 Fruits 22.8 19.2 11.5 17.9 26.5 26.4 16.3 18.5 18.8 19.7 Other 1.5 2.5 1.3 2.8 0.9 2.8 2.8 3.6 1.7 2.9 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Value of sales Cereals 1.8 5.1 6.4 10.8 1.6 5.6 2.6 4.1 2.8 6.5 Vegetables 34.5 46.2 36.7 32.6 8.4 22.5 31.1 36.2 28.8 36.3 Fruits 62.2 47.1 56.4 55.0 89.8 71.5 64.0 58.4 67.2 56.0 Other 1.5 1.7 0.4 1.6 0.3 0.4 2.2 1.3 1.2 1.2 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: Authors' calculations based on RNR Census 2000 and RNR Census 2009 9 The different commercialization rates by crop category are shown in Figure 1. The value of production of cereals is by far the most important of all crop categories (almost 3,769 million Nu in 2008 compared to only 1,760 million Nu for vegetables) and it is larger than the three other crop categories combined. Cereals are however mostly produced for home consumption and little is sold in the market place. It seems that most of the cereals that are commercialized in Bhutan are thus the imported ones, mostly rice. Sales are relatively much more important for vegetables and fruits. Commercialization rates for these two categories of crops are 34% and 65% respectively. Figure 1. Sales and own use of different types of crops, 2008 (Million nu) 0 500 1000 1500 2000 2500 3000 3500 4000 Cereals Vegetables Fruits Others M ill io n N u Sales Own use Maps 5 and 6 illustrate the importance and the spread of non-cereals over space and time. The maps show per geog the share of non-cereals in total harvested area. There is relatively little difference between the different regions in terms of area allocation. The share of the area allocated to cereal production varied in 2008 between 60% in the West and 70% in East-central (Table 3). The share of fruits in total value of production was highest in East-central where it accounted for 26%. The share of vegetables was highest in the East, at 33% of the value of total production. For the analysis on crop diversification, we use two measures. First, we look at the numbers of crops grown and sold in a geog. Second, we rely on the Simpson Index of Diversity (SID) which is defined as: SID Pi i = -å1 2 where Pi is the proportion of area/production/sales that is coming from crop i. The value of SID varies between 0 and 1. If there is only one crop grown, SID will be 0 (the least diversified situation). The more diverse the number of the crops grown, the more the SID will approach 1. 10 Map 5 Share of non-ceareals in total cropped area in 2000 (RNR Census 2000) % 75 − 100 50 − 75 25 − 50 0 − 25 No data Share of non−cereals in total cropped area in 2000 (RNR Census 2000) Map 6 Share of non-ceareals in total cropped area in 2008 (RNR Census 2009) % 75 − 100 50 − 75 25 − 50 0 − 25 Share of non−cereals in total cropped area in 2008 (RNR Census 2009) 11 The measures of crop diversification for Bhutan are shown in Table 5. To ensure having comparable measures over time, we present the SID for comparable crops in 2008 and for all reported crops (information on more crops was collected in 2008 but we mostly discard that information for this analysis). Table 5 shows that on average 24 crops were grown per geog in 2000. There is especially a small number of crops grown in the East-central region of the country. The number of crops grown seemed to have dropped slightly in 2008. This might be partly due to the “1 geog, 3 products” policy where the government encourages geogs to specialize in specific activities, be it agriculture or off-farm. The number of crops that were reported to have been sold in the geog was as high as 17 in 2000 and 15 in 2008. The SID decreases when we compare cropped area (0.86) with the value of production (0.79) and the value of sales (0.73), indicating that the value of production and sales is more concentrated on a fewer number of crops than the area allocation. The Eastern region seems the least diversified in area terms and in sales composition as it has the lowest SID of the four regions. Table 5. Measures of diversification West West-central East-central East Bhutan 2000 2008 2000 2008 2000 2008 2000 2008 2000 2008 Crop area Number of crops grown 23.3 20.3 22.0 20.8 14.4 13.1 34.7 34.2 23.6 22.1 Simpson index of diversity, comparable crops 2000/2008 0.93 0.88 0.93 0.83 0.88 0.83 0.93 0.81 0.92 0.84 Simpson index of diversity, all reported crops in 2008 - 0.90 - 0.85 - 0.86 - 0.84 - 0.86 Value of production Simpson index of diversity, comparable crops 2000/2008 0.93 0.79 0.88 0.67 0.83 0.78 0.95 0.83 0.90 0.77 Simpson index of diversity, all reported crops in 2008 - 0.81 - 0.70 - 0.80 - 0.85 - 0.79 Value of sales Number of crops sold 16.7 14.1 15.5 14.3 10.3 7.6 25.8 22.2 17.1 14.6 Simpson index of diversity, comparable crops 2000/2008 0.90 0.87 0.80 0.69 0.61 0.66 0.71 0.65 0.75 0.72 Simpson index of diversity, all reported crops in 2008 - 0.87 - 0.71 - 0.67 - 0.68 - 0.73 Source: Authors' calculations based on RNR Census 2000 and RNR Census 2009 Table 6 and Figure 2 show the relative importance of different crops in harvested area, production, and sales in the country.3 We see a dramatic shift between crops over these three categories. While maize makes up over 30% of the harvested area in Bhutan, its share in the total value of production and sales is 7% and 0% respectively (the latter might however been an underestimate). Paddy is the second most important crop in area allocation representing 22% of total harvested area. It was also highly valued (especially in 2008; see Table 3) and represented 42% of the value of production. However, it is not very much commercialized and it is estimated that it only makes up 7% of total sales income in the country. The most important crop in sales income is oranges, accounting for 40% of total sales income. It however makes up only 5% of the harvested area. Table 6 further shows how absolute areas allocated to different crops have changed over time. Maize areas and production have declined by 13% and 14% respectively. Paddy production on the other hand increased by 13%, seemingly driven by a slight increase in cropped area but most importantly by an increase in yield: rice yields were 11% higher in 2008 than in 2000. The area harvested of other cereals such as barley, buckwheat, fingermillet, foxmillet, and wheat are all on the decline. This is seemingly driven by the low income incomes 3 As the purpose of the analysis is to see dynamics over time, we limit ourselves to statistics on the comparable crops. This does not influence the overall trends significantly given that these additional crops are minor in area and value. 12 elasticities for these products as there is a relatively decreasing demand for these products when people become richer (see PPD/IFPRI, 2010). On the other hand, most vegetables are on the increase. Especially potatoes, chili, and radish have become more important in area harvested over time. As can be deducted from the results of the previous Tables, non-cereals are usually associated with higher outputs, in value terms, per unit of land. They are thus commonly referred to as high-value crops (Minot et al., 2006; Joshi et al., 2006). Using average numbers on the value of production and area cultivated, Table 6 shows that the output per unit of land, based on the 2008 RNR Census, varies from 65,000 Nu/ha for cereals, to 90,000 Nu/ha for vegetables, and 232,000 Nu/ha for fruits4. Fruits thus have a land productivity that is almost four times as high as cereals. The relatively higher value in the case of fruits is partly explained by the larger up-front investments required for its cultivation given that it might take significant time before trees start bearing fruits. While these results indicate more intensive use of land in value terms if there is a move to high-value crops, they however do not reflect rewards to labor as labor use in vegetable production is generally significantly higher than on cereals, leading to smaller differences in labor productivity between the two. There is also significant variability within these three crop sectors (cereals, vegetables, fruits). For example, while paddy has a land productivity of 156,000 Nu/ha, this is only as high as 18,000 in the case of maize. This is driven by lower physical yields (2.45 tons/ha for maize compared to 3.99 tons/ha for paddy (Table 6)) as well as by significantly higher prices for paddy in the year 2008 (Table 3 shows that the price of rice is 3.5 times as high as this of maize). Similar large variations exist between different crops in the vegetables’ category. All fruits on the other hand show high land productivity rates. The legumes category has the lowest land productivity. Table 7 also shows that when overall land productivity is compared over time, we see an average increase of 23%, i.e. from 65,000 Nu/ha to 80,000 Nu/ha. This is explained by increasing physical land productivity of specific 4 We calculated the value of output using the area under fruit bearing trees. Figure 2. Crop composition of harvested area, total production, and total sales, 2008 (100%=total) 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 Harvested area Total produc�on Total sales Others Other fruits Oranges Apples Other vegetables Potato Other cereals Maize Paddy 13 Table 6. Crop-wise harvested area, production, sales, and productivity for comparable crops in 2000 and 2008, in quantities 2000 2008 Harvested Total Total Land Harvested Total Total Land area production sales productivity area production sales productivity (ha) (ton) (ton) (ton/ha) (ha) (ton) (ton) (ton/ha) Cereals: Paddy 19,158 68,500 794 3.58 19,329 77,214 2,743 3.99 Maize 31,161 77,300 513 2.48 27,141 66,592 - 2.45 Barley 1,498 1,735 8 1.16 1,313 2,048 6 1.56 Buckwheat 3,674 2,819 6 0.77 3,423 5,117 4 1.49 Fingermillet & Foxmillet 6,166 3,793 40 0.62 3,510 5,009 62 1.43 Wheat 4,688 4,352 36 0.93 3,185 5,642 18 1.77 All cereals 66,345 158,499 1,397 2.39 57,903 161,623 2,832 2.79 Vegetables: Potato 3,192 35,000 19,400 10.96 5,554 52,967 26,495 9.54 Chili 935 2,846 926 3.04 3,811 7,283 1,807 1.91 Sag 440 519 85 1.18 - - - - Eggplant 76 192 42 2.54 341 497 58 1.46 Tomato 67 331 177 4.90 421 752 46 1.79 Carrot 31 149 96 4.87 180 363 117 2.01 Radish 779 3,381 443 4.34 3,243 5,957 340 1.84 Turnip 346 2,638 67 7.62 937 5,075 21 5.41 Cassava 249 818 19 3.28 186 352 27 1.90 Ginger 625 1,268 448 2.03 1,381 3,129 1,615 2.27 Garlic 185 423 61 2.29 1,486 2,072 140 1.39 Cardamom 504 509 482 1.01 1,936 1,017 693 0.53 All vegetables 7,430 48,073 22,245 6.47 19,477 79,464 31,359 4.08 Fruits: Apple 710 5,113 4,763 7.20 921 5,531 3,927 6.01 Oranges 3,276 29,616 27,200 9.04 4,382 37,758 25,987 8.62 Pear 29 705 161 24.66 72 1,108 96 15.30 Peach 74 1,085 107 14.62 121 948 64 7.82 Plum 19 276 78 14.63 33 402 30 12.27 Walnut 30 233 23 7.68 77 333 85 4.33 Areca nut 89 1,327 1,199 14.95 326 3,847 2,546 11.79 Guava 64 655 158 10.26 144 770 113 5.35 All fruits 4,291 39,010 33,689 9.09 6,076 50,696 32,847 8.34 Others: Mustard 3,434 1,689 150 0.49 1,937 3,573 36 1.84 Soyabean 931 572 97 0.61 566 784 6 1.39 Rajmabeans 309 340 103 1.10 683 743 126 1.09 Beans 861 1,144 - 1.33 2,585 3,697 573 1.43 Peas 274 622 405 2.27 513 798 267 1.56 All others 5,810 4,367 755 0.75 6,284 9,595 1,008 1.53 Total 83,876 249,949 58,086 2.98 89,739 301,378 68,046 3.36 Source: Authors' calculations based on RNR Census 2000 and RNR Census 2009 14 Table 7. Crop-wise harvested area, production, sales, and productivity for comparable crops in 2000 and 2008, in values 2000 2008 Harvested Total Total Land Harvested Total Total Land area production sales productivity area production sales productivity (ha) (million Nu) (million Nu) (1000 Nu/ha) (ha) (million Nu) (million Nu) (1000 Nu/ha) Cereals: Paddy 19,158 2,733 32 143 19,329 3,011 106 156 Maize 31,161 579 4 19 27,141 499 - 18 Barley 1,498 23 0 15 1,313 26 0 20 Buckwheat 3,674 52 0 14 3,423 91 0 27 Fingermillet & Foxmillet 6,166 43 0 7 3,510 57 1 16 Wheat 4,688 65 1 14 3,185 85 0 27 All cereals 66,345 3,495 37 53 57,903 3,769 107 65 Vegetables: Potato 3,192 367 203 115 5,554 535 268 96 Chili 935 285 93 304 3,811 724 179 190 Sag 440 0 - - - - - - Eggplant 76 3 1 34 341 7 1 20 Tomato 67 8 4 123 421 19 1 45 Carrot 31 4 3 146 180 11 3 60 Radish 779 34 4 43 3,243 62 4 19 Turnip 346 17 0 48 937 30 0 32 Cassava 249 8 0 33 186 4 0 22 Ginger 625 38 13 61 1,381 95 49 69 Garlic 185 32 5 173 1,486 157 11 105 Cardamom 504 59 55 116 1,936 117 80 60 All vegetables 7,430 854 382 115 19,477 1,760 595 90 Fruits: Apple 710 152 142 215 921 168 120 183 Oranges 3,276 738 678 225 4,382 960 661 219 Pear 29 19 4 654 72 28 2 393 Peach 74 22 2 298 121 20 1 161 Plum 19 2 1 130 33 3 0 104 Walnut 30 5 1 176 77 8 2 100 Arecanut 89 66 60 748 326 194 129 595 Guava 64 23 5 355 144 27 4 188 All fruits 4,291 1,028 893 240 6,076 1,409 919 232 Others: Mustard 3,434 33 3 10 1,937 73 1 38 Soyabean 931 17 3 18 566 23 0 41 Rajmabeans 309 7 2 23 683 17 3 24 Beans 861 24 - 27 2,585 78 12 30 Peas 274 12 8 45 513 16 5 31 All others 5,810 93 16 16 6,284 207 21 33 Total 83,876 5,470 1,328 65 89,739 7,146 1,643 80 Source: Authors' calculations based on RNR Census 2000 and RNR Census 2009 15 crops, such as rice and maize where yields have improved over time (the land productivity was 3.36 tons/ha in 2008; this compares to 2.98 tons/ha in 2000), as well as by a switch in crop composition to higher-valued crops such as fruits and vegetables. Calculations are further done on the importance of different crops in area harvested (Figure 3), value of production (Figure 4), and value of sales (Figure 5) by quintile of area harvested. This helps us understand if smaller farmers cultivate and sell different crops than bigger ones. We note relatively few differences in area allocation between small and larger farmers (Figure 3). Figure 4 shows that larger farmers rely relatively more on paddy than smaller ones. Paddy makes up 43% of the value of production of the large farmers. This compares to only 34% for the small ones. Figure 5 illustrates, interestingly, that fruits are relatively even more important a source of income for the small farmers (70% of their sales’ income) than for the larger ones (60% of their income). However, the larger farmers rely mostly on income from oranges while the smaller ones earn more from other types of fruits. Vegetables are an important source of income for quintiles 2 to 4, but less so for the smaller and the biggest quintile. However, when we look at absolute numbers, incomes from any crop are significantly higher for the larger farmers (Table 8). Area harvested, value of production, and value of sales are respectively 15 times, 12 times, and almost 14 times larger for the largest quintile compared to the smallest one. Figure 3. Crop composition in area harvested, by quintile of area harvested 2008 (100%=total) 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 Quin�le 1 Quin�le 2 Quin�le 3 Quin�le 4 Quin�le 5 All others Other fruits Oranges Apples Other vegetables Potato Other cereals Paddy Maize Figure 6 presents a Lorenz-curve of harvested area and the value of sales and production over households, based on the RNR data of 2008. It shows the pattern of distribution of households in the total outcome indicators. To make such a graph, households are first ranked from the lowest to the highest and their importance in the total land allocation, and value of production and of sales is then calculated and shown on the y-axis. The closer (farther away) 16 Figure 4. Crop composition in value of production, by quintile of area harvested 2008 (100%=total) 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 Quin�le 1 Quin�le 2 Quin�le 3 Quin�le 4 Quin�le 5 All others Other fruits Oranges Apples Other vegetables Potato Other cereals Paddy Maize Figure 5. Crop composition in value of sales, by quintile of area harvested 2008 (100%=total) 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 Quin�le 1 Quin�le 2 Quin�le 3 Quin�le 4 Quin�le 5 All others Other fruits Oranges Apples Other vegetables Potato Other cereals Paddy Maize 17 Table 8. Crop-wise harvested area, production, and sales in 2008, by farm size Farm size of area harvested Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Total Area harvested (hectares/household) Maize 0.059 0.274 0.394 0.546 0.920 0.439 Paddy 0.046 0.157 0.241 0.375 0.735 0.311 Other cereals 0.018 0.086 0.161 0.253 0.429 0.190 All cereals 0.123 0.517 0.796 1.173 2.085 0.940 Potato 0.022 0.077 0.102 0.113 0.137 0.090 Other vegetables 0.056 0.204 0.315 0.410 0.679 0.333 All vegetables 0.079 0.282 0.417 0.523 0.816 0.424 Apples 0.004 0.010 0.011 0.015 0.035 0.015 Oranges 0.010 0.025 0.037 0.066 0.217 0.071 Other fruits 0.005 0.010 0.014 0.020 0.049 0.020 All fruits 0.020 0.045 0.062 0.100 0.301 0.106 All others 0.013 0.057 0.109 0.165 0.279 0.125 Total 0.234 0.901 1.384 1.961 3.480 1.595 Quantity produced (Nu/household) Maize 929 4,378 6,807 10,210 18,031 8,084 Paddy 7,699 25,242 37,722 58,094 115,201 48,865 Other cereals 514 2,112 3,616 5,453 9,599 4,265 All cereals 9,142 31,733 48,145 73,757 142,831 61,214 Potato 1,972 7,013 9,643 10,862 13,931 8,695 Other vegetables 5,056 18,622 26,509 33,266 42,942 25,312 All vegetables 7,028 25,635 36,152 44,128 56,872 34,007 Apples 774 1,645 1,929 2,709 6,602 2,735 Oranges 2,454 6,370 9,187 16,265 43,577 15,595 Other fruits 2,803 3,933 4,525 6,771 11,952 6,003 All fruits 6,031 11,948 15,640 25,745 62,131 24,333 All others 476 2,143 3,795 5,666 8,481 4,119 Total 22,677 71,458 103,733 149,297 270,315 123,673 Quantity sold (Nu/household) Maize 0 0 0 0 0 0 Paddy 8 277 808 1,767 5,724 1,720 Other cereals 1 5 9 19 64 20 All cereals 9 282 817 1,787 5,788 1,740 Potato 642 2,931 4,542 5,380 8,237 4,352 Other vegetables 863 3,255 4,984 7,701 12,638 5,897 All vegetables 1,505 6,186 9,526 13,081 20,875 10,249 Apples 424 1,047 1,293 1,918 5,015 1,942 Oranges 1,602 4,091 6,054 10,640 31,234 10,741 Other fruits 1,508 1,615 1,673 2,829 5,209 2,569 All fruits 3,534 6,753 9,020 15,387 41,457 15,252 All others 34 228 341 565 816 397 Total 5,083 13,448 19,705 30,820 68,936 27,639 Source: Authors' calculations based on RNR Census 2009 18 the line would be to the diagonal, the more (un)equal the distribution. The results in Figure 6 shows that the crop area allocation is most equally distributed of the three. 50% of the households cultivate 27% of the crop area. The 10% largest farmers cultivate 25% of the crop area. The value of production shows a distribution that is slightly more unequal. As could be expected, the value of sales is the most unequal of the three, indicating that some households are much better integrated in the commercial system than others. 10% of the households are responsible for 73% of all crop sales in Bhutan while half of the interviewed rural households do not report any crop sales at all. Figure 6. lorenz-curve of harvested area, the value of production, and value of sales over households, 2008 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 diagonal area produc�on sales Table 9 presents the results of a multi-variate regression of measures of crop diversification and commercial- ization. To measure these variables, we use the share of non-cereals in total area harvested and the share of the value sold over the value of production. Given that a significant number of observations are bounded at 0 or at 1, we use a tobit regression as to improve the estimates. Both measures are significantly affected by distance to a motorable road. The share of non-cereals in the most remote category is 11% lower than for households that are located close to a motorable road. As could be expected, distance to a motorable road has an even higher effect on commercialization rates. The commercialization rate is 20% lower in the most remote category compared to the least remote households. The size of the farm is also significantly associated with crop diversification and commercialization. Large farmers diversify relatively less in non-cereals but do have a higher a commercialization rate, i.e. one hectare extra increases the commercialization rate by 5%. The quality of the land further also significantly affects diversification as well as commercialization. Wetlands are associated with relatively less diversification to non-cereals and less commercialization while orchards are associated with more. 19 Leased-in land is not associated with crop commercialization as well as diversification. Livestock on the other hand is positively associated with both measures. Livestock might lead to higher productivity (through access to manure) and consequently more quantities for sale. Households that own more livestock (and are thus wealthier) might also be willing and able to take on more risk to diversify in the riskier non-cereal crops. We further see significant variation over dzongkhags. Thimphu and Paro have the highest level of diversifica- tion as well as commercialization, ceteris paribus. Zhemgang and Samdrupjongkhar are the dzongkhags with the lowest level of diversification while Lhuentse and Thrashiyangtse are the dzongkhags that have the lowest commercialization rates. Table 9. Determinants of crop diversification and crop commercialization Share non-cereals in area (Tobit) Commercialization rate (Tobit) Coefficient t-value P>t Coefficient t-value P>t Default=distance to motorable road<1 hour Distance road 1-3 hours yes=1 -0.032 -10.680 0.000 -0.052 -12.650 0.000 Distance road 3-6 hours yes=1 -0.028 -7.340 0.000 -0.070 -13.200 0.000 Distance road 6 hours - 1day yes=1 -0.031 -7.030 0.000 -0.122 -19.580 0.000 Distance road >1 day yes=1 -0.108 -26.300 0.000 -0.197 -32.720 0.000 Land cultivated hectares -0.073 -53.900 0.000 0.056 29.570 0.000 Wetlands over cultivated land share -0.227 -58.070 0.000 -0.157 -28.620 0.000 Orchards over cultivated land share 0.368 53.200 0.000 0.359 40.800 0.000 Leased-in land share -0.008 -1.920 0.055 -0.008 -1.500 0.135 Number of livestock owned log(x+1) 0.020 16.160 0.000 0.028 16.850 0.000 Risk of crop damage by wildlife index 0.101 13.170 0.000 0.021 2.010 0.045 Default=Thimphu Paro yes=1 -0.128 -12.450 0.000 0.031 2.330 0.020 Ha yes=1 -0.228 -18.710 0.000 -0.164 -10.300 0.000 Chhukha yes=1 -0.281 -27.480 0.000 -0.108 -8.230 0.000 Samtse yes=1 -0.268 -27.650 0.000 -0.186 -14.870 0.000 Punakha yes=1 -0.243 -23.150 0.000 -0.100 -7.330 0.000 Gasa yes=1 -0.343 -19.410 0.000 -0.328 -12.160 0.000 Wangdue yes=1 -0.139 -13.650 0.000 -0.070 -5.370 0.000 Tsirang yes=1 -0.299 -29.200 0.000 -0.185 -14.050 0.000 Dagana yes=1 -0.362 -35.560 0.000 -0.209 -15.860 0.000 Bumthang yes=1 -0.148 -12.480 0.000 -0.042 -2.780 0.005 Trongsa yes=1 -0.351 -30.350 0.000 -0.275 -17.570 0.000 Zhemgang yes=1 -0.378 -34.010 0.000 -0.257 -17.140 0.000 Sarpang yes=1 -0.362 -36.330 0.000 -0.217 -16.850 0.000 Lhuentse yes=1 -0.327 -30.080 0.000 -0.352 -23.440 0.000 Mongar yes=1 -0.371 -37.470 0.000 -0.270 -20.910 0.000 Trashigang yes=1 -0.325 -33.370 0.000 -0.262 -20.650 0.000 Trashiyangtse yes=1 -0.286 -27.170 0.000 -0.304 -21.520 0.000 Pemagatshel yes=1 -0.345 -33.860 0.000 -0.070 -5.350 0.000 SamdrupJongkhar yes=1 -0.376 -36.790 0.000 -0.093 -7.050 0.000 Intercept 0.669 67.090 0.000 0.240 18.620 0.000 Number of obs 49084 49078 LR chi2(29) 14517 10592 Prob > chi2 0.00 0.00 Pseudo R2 0.68 0.21 Source: Authors' calculations based on RNR Census 2009 20 The maps 7 to 16 show how the different major crops are faring between the years 2000 and 2008. On these maps, the total harvested area of five major crops in Bhutan (maize, rice, apples, oranges, and potatoes) is shown for the years 2000 and 2008. By presenting the maps this way, we are able to show the importance of different crops spatially and the evolution over time. The maps show overall that the importance of cereals (maize and paddy) in total production is relatively on the decline while fruits and vegetables (apple, orange and potato) have expanded over time. This pattern is similar to trends that are seen in a number of other Asian countries (Joshi et al., 2006; Minot et al., 2006; Gulati et al., 2008). Maize is the most important crop with respect to crop area allocation. It was estimated to have occupied 37% of the crop area in 2000. The total maize area has declined over the years and its relative share has come down to 30% in 2008. Maize is a crop that is especially important in the Eastern part of the country where it is an important part of the local diet (Maps 7 and 8). Maize counts for 7% of the total value of production in 2008, down from 10.5% in 2000. The relative crop area under paddy cultivation stayed rather constant over the years as paddy occupied 23% of the crop area in 2000 and 22% in 2008 (Maps 9 and 10). Despite this importance in crop area, Bhutan is still a large importer of rice and it is estimated that Bhutan obtains about half of its consumed rice from imports. The maps show that paddy is especially an important crop in the West-central part of the country. Potato has shown significant growth rates in the last decade and it has now become an important cash crop for Bhutan. While its area share is significantly lower than for rice and maize, i.e. 4% in 2000 and 6% in 2008, its share in the value of production is more important, i.e. about 7% in 2000 and 2008. Maps 11 and 12 show that the importance of potato seems especially to have increased in the Eastern and Northeastern part of the country. Using reasonable assumption on the number of trees per unit of land (112 trees per acre), it is estimated that oranges are by far the most important fruit in Bhutan and it occupies 4% and 5% of the harvested area in 2000 and 2008 respectively. However, given its high value, its importance in the total value of crop production is relatively much higher. It is evaluated at 13% in 2000 as well as in 2008. Map 13 and 14 show that the majority of orange production is situated in the Southern belt of the country. Apple is the second most important fruit. Apple accounts for 1% of the crop area and its share in the total value of crop production is as high as 2-3%. The apple production is rather localized in the country and is mostly situated in the western part of the country (Maps 15 and 16) at those altitudes that allow the crop to flourish. Next, we look at total crop sales and the composition of these crop sales. Map 17 shows that it is especially the Southern and Western belt of the country that are most integrated in commercial crop agriculture as they show the highest level of crop sales in 2008. Only a few crops count for the majority of crop sales income in Bhutan. They are orange, representing 40% of total sales income; apple, counting for 7% of total sales income; and potato, counting for 16%. Maps 18 to 21 show that orange is by far the most important crop in sales value for most of the country, except for the Western region. In the Western region, apple is slightly more important at 38% of the sales value compared to 35% for oranges. The importance of potato as a cash crop is important in the Eastern geogs of the country. Potatoes count there for about almost 20% of the sales income. The striking minor importance of the staples paddy and maize in total crop sales, despite their importance in area allocation, might be partly due to underestimation of the products that enter in the commercial circuit. However, they do also reflect the fact that a larger part of the production of these crops is grown towards auto-consumption. Maps 18 show paddy for sale is mostly produced in the West-central region. 21 Map 7 Harvested area of maize in 2000 (RNR Census 2000) ha 200 − 600 100 − 200 25 − 100 5 − 25 0 − 5 Harvested area of maize in 2000 (RNR Census 2000) Map 8 Harvested area of maize in 2008 (RNR Census 2009) ha 200 − 600 100 − 200 25 − 100 5 − 25 0 − 5 Harvested area of maize in 2008 (RNR Census 2009) 22 Map 9 Harvested area of paddy in 2000 (RNR Census 2000) ha 200 − 400 100 − 200 25 − 100 5 − 25 0 − 5 Harvested area of paddy in 2000 (RNR Census 2000) Map 10 Harvested area of paddy in 2008 (RNR Census 2009) ha 200 − 400 100 − 200 25 − 100 5 − 25 0 − 5 Harvested area of paddy in 2008 (RNR Census 2009) 23 Map 11 Harvested area of potatoes in 2000 (RNR Census 2000) ha 50 − 300 25 − 50 10 − 25 5 − 10 0 − 5 Harvested area of potato in 2000 (RNR Census 2000) Map 12 Harvested area of potatoes in 2008 (RNR Census 2009) ha 50 − 300 25 − 50 10 − 25 5 − 10 0 − 5 Harvested area of potato in 2008 (RNR Census 2009) 24 Map 13 Harvested area of oranges in 2000 (RNR Census 2000) ha 50 − 400 25 − 50 5 − 25 1 − 5 0 − 1 Harvested area of oranges in 2000 (RNR Census 2000) Map 14 Harvested area of oranges in 2008 (RNR Census 2009) ha 50 − 400 25 − 50 5 − 25 1 − 5 0 − 1 Harvested area of oranges in 2008 (RNR Census 2009) 25 Map 15 Harvested area of apples in 2000 (RNR Census 2000) ha 75 − 200 25 − 75 10 − 25 1 − 10 0 − 1 No data Harvested area of apple in 2000 (RNR Census 2000) Map 16 Harvested area of apples in 2008 (RNR Census 2009) ha 75 − 200 25 − 75 10 − 25 1 − 10 0 − 1 Harvested area of apple in 2008 (RNR Census 2009) 26 Map 17 Total crop sales per geog in 2008 (RNR Census 2009) Million Nu 20 − 51 5 − 20 3 − 5 1 − 3 0 − 1 Total crop sales per geog in 2008 (RNR Census 2009) Map 18 Share of crop sales from paddy in 2008 (RNR Census 2009) % 50 − 100 20 − 50 5 − 20 1 − 5 0 − 1 No data Share of crop sales from paddy in 2008 (RNR Census 2009) 27 Map 19 Share of crop sales from potatoes in 2008 (RNR Census 2009) % 50 − 100 20 − 50 5 − 20 1 − 5 0 − 1 No data Share of crop sales from potato in 2008 (RNR Census 2009) Map 20 share of crop sales from oranges in 2008 (RNR Census 2009) % 50 − 100 20 − 50 5 − 20 1 − 5 0 − 1 No data Share of crop sales from oranges in 2008 (RNR Census 2009) 28 Map 21 share of crop sales from apples in 2008 (RNR Census 2009) % 50 − 100 20 − 50 5 − 20 1 − 5 0 − 1 No data Share of crop sales from apple in 2008 (RNR Census 2009) 5. DIVeRSIFICATIon AnD CoMMeRCIAlIzATIon wIThIn The lIVeSToCk SeCToR Within the livestock sector, we look at the importance of different types of livestock and livestock products in herd size, the value of production, and the value of sales, by farm size. Table 10 shows, based on the RNR data of 2008, that 83% of rural households own livestock. There is a strong link between land cultivation and livestock ownership as larger farms are more likely to also have livestock. 96% of the largest farms hold livestock. This compares to only 49% of the small farmers. However, when we look at the value of live animals, we see that the value is relatively high for the smallest and the largest quintile. It seems that there are in the smallest quintile a number of farmers that do focus almost exclusively on livestock activities (mostly yak cultivation) with little attention towards crop cultivation. Looking at the average number of livestock per farm, cattle are by far the most important category in livestock (Table 10). They account for about 69% of the herd size of livestock in Bhutan in 2008.5 However, the number of cattle seems to slightly decline over the years. The RNR Census of 2000 estimates that there were about 336,000 cattle in 2000. This compares to 310,071 in 2008 (MoAF, 2010), or a decline of 8% over a period of 8 years. Still, an average farm holds almost 5 cattle, illustrating the importance of livestock in rural Bhutan. Yak counts for 0.6 heads per farm, mostly held by the smallest quintile. 5 Excluding poultry. 29 Dairy products make up the bulk of the value of livestock products in Bhutan. Based on the data of 20086, it is estimated that milk is the most important livestock product, counting for 47% of the total value of livestock products. Second comes cheese (22%) and then butter (19%). Beef accounts for 3% of the total value of livestock products. The importance of beef seems to be on decline over time, possibly because of increasing religious sentiments on the killing of animals. Butter and cheese seem especially an important livestock product for the smaller farmers. The share of milk over quintiles is on the other hand stable. When we look at the importance of different products in total sales, milk is much less important indicating that most of its production is for own consumption. Cheese is the most important marketed livestock product, 39% of the total sales value, followed by butter, at 31%. Regression results on the importance of the sales of live animals and the sales of animal products are reported in Table 11. We see that the sale of live animals is especially prevalent for the households that are located far from a motorable road. The most remote households sell almost 2.5 times more livestock than the least remote. The least remote households specialize on the other hand significantly more in the commercialization of live- stock products such as cheese, butter, and milk as these products are more perishable and are thus harder to commercialize by remote households. We also note the higher earnings of cross-bred cattle compared to local cattle (Nublang breed) as well as other cattle (Mithun Pure, Jersey Pure, and Brown Swiss Pure). Pigs and poultry have the lowest returns of all kind of livestock. There are again strong spatial differences, with the dzongkhags of Tsirang and Dagana prominent in the sales of live animals and the dzongkhag of Mongar best for the sales of livestock products. Maps 22 through 26 show some of regional specialization in livestock cultivation. Map 22 shows the share of the rural households that held livestock in 2008. The share is relatively low in the areas around Paro and Thimpu but high in the rest of the country. Map 23 shows where cattle in particular are held in Bhutan. Especially the Center and the Southwest has most of the cattle population. The government has done significant investments to improve the quality of cattle in Bhutan. This is showing up in an improved ratio of improved cattle over local cattle between the two years of the Census. While there might be definitional issues to allow good comparison over time, the ratio of improved cattle was as high as 19% in 2000 (Map 24) while the share of cross-breds was as high as 40% in 2008 (Map 25 and Table 10). Yaks are important livestock that can withstand colder temperature and can thus live in the higher altitude geogs, i.e. the alpine regions. In the winter, yaks often move to lower altitude geogs in search of feed. It is esti- mated that they make up 10% of the herd size of large animals in Bhutan. They are especially important in the Northern and Western geogs of the country (Map 26). 6 They include beef, chicken, egg, pork, milk, meat yak, mutton, wool, fish, butter and cheese. 30 Table 10. herd size, value of livestock, livestock products, and sales of livestock products in 2008, by farm size Farm size of area harvested Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Total Share of households that own livestock 0.49 0.84 0.92 0.94 0.96 0.83 Value of live animals (Nu/household) 62,955 36,553 40,747 48,046 62,327 50,112 Sales of live animals (Nu/household) 832 843 1,121 1,778 2,781 1,472 Livestock products (Nu/household) 14,407 13,042 16,093 19,553 24,919 17,609 Livestock products sold (Nu/household) 5,583 4,465 5,874 6,996 9,093 6,404 Herdsize (Number per household) Local cattle (Nublang breed) 1.3 2.3 2.7 3.3 4.2 2.8 Cross-bred cattle* 0.8 1.6 2.1 2.4 2.7 1.9 Other cattle** 0.0 0.1 0.1 0.1 0.2 0.1 Total cattle 2.1 4.0 4.9 5.9 7.1 4.8 Yak 2.3 0.3 0.2 0.1 0.1 0.6 Other large livestock*** 0.5 0.5 0.6 0.8 1.0 0.7 Small ruminants**** 0.2 0.3 0.4 0.6 1.2 0.6 Pigs 0.2 0.1 0.2 0.2 0.3 0.2 Poultry 1.4 2.6 2.8 3.8 5.4 3.2 Livestock products produced (% of value) Milk 45.8 48.9 46.6 46.2 47.1 46.9 Butter 20.4 18.6 19.2 18.4 18.0 18.8 Cheese 26.3 22.3 20.9 22.6 20.3 22.2 Eggs 1.8 4.4 5.7 5.6 6.1 5.0 Wool 0.2 0.1 0.1 0.1 0.1 0.1 Fish 0.0 0.1 0.1 0.1 0.3 0.2 Pork 0.4 1.2 1.5 1.7 2.8 1.7 Beef 1.3 2.9 3.9 3.6 2.6 2.9 Mutton 0.1 0.2 0.3 0.4 0.7 0.4 Chicken 0.8 0.8 1.0 0.6 1.5 1.0 Yak meat 2.7 0.4 0.7 0.5 0.4 0.9 Total 100.0 100.0 100.0 100.0 100.0 100.0 Livestock products sold (% of value) Milk 10.4 13.4 11.7 11.3 11.3 11.5 Butter 33.6 31.1 30.4 30.4 30.2 31.0 Cheese 44.2 39.6 36.9 39.7 36.6 39.1 Eggs 3.0 7.3 9.6 9.3 10.6 8.4 Wool 0.1 0.0 0.0 0.0 0.0 0.0 Fish 0.0 0.1 0.1 0.2 0.4 0.2 Pork 0.7 2.3 2.7 2.8 4.9 2.9 Beef 1.7 3.7 5.2 4.7 3.0 3.6 Mutton 0.1 0.1 0.1 0.2 0.4 0.2 Chicken 1.7 1.4 1.7 0.3 1.9 1.4 Yak meat 4.5 0.7 1.5 1.0 0.7 1.6 Total 100.0 100.0 100.0 100.0 100.0 100.0 Source: Authors' calculations based on RNR Census 2009 *: Mithun Cross, Jersey Cross, Brown Swiss Cross **: Mithun Pure, Jersey Pure, Brown Swiss Pure ***: Zo/Zom, Buffaloes, Horses, Mules, Donkeys ****: Goats, Sheep 31 Table 11. Determinants of commercialization of live animals and livestock products Sales of live animals in log (Nu per household) - Tobit Sales of livestock products in log (Nu per household) - Tobit Coefficient t-value P>t Coefficient t-value P>t Default=distance to motorable road<1 hour Distance road 1-3 hours yes=1 1.566 5.680 0.000 0.115 1.080 0.280 Distance road 3-6 hours yes=1 1.282 3.570 0.000 -0.231 -1.700 0.090 Distance road 6 hours - 1day yes=1 1.481 3.770 0.000 -1.350 -8.680 0.000 Distance road >1 day yes=1 2.561 7.150 0.000 -0.535 -3.740 0.000 Local cattle (Nublang breed) Number 0.431 122.010 0.000 Cross-bred cattle* Number 0.550 59.240 0.000 Other cattle** Number 0.504 13.570 0.000 Yak Number 0.151 25.630 0.000 Other large livestock*** Number 0.477 19.920 0.000 Small ruminants**** Number 0.292 12.460 0.000 Pigs Number 0.068 4.350 0.000 Poultry Number 0.035 17.800 0.000 Default=Thimphu Paro yes=1 -4.875 -5.300 0.000 -0.605 -1.850 0.064 Ha yes=1 -2.256 -2.140 0.032 1.382 3.680 0.000 Chhukha yes=1 4.553 5.710 0.000 1.578 4.960 0.000 Samtse yes=1 5.413 7.160 0.000 -0.650 -2.140 0.033 Punakha yes=1 -0.135 -0.150 0.878 1.629 4.910 0.000 Gasa yes=1 -6.576 -3.930 0.000 -1.253 -2.350 0.019 Wangdue yes=1 -2.602 -3.000 0.003 3.506 11.110 0.000 Tsirang yes=1 9.123 11.540 0.000 0.475 1.470 0.141 Dagana yes=1 8.888 11.230 0.000 1.292 4.020 0.000 Bumthang yes=1 -4.119 -3.700 0.000 -0.765 -2.010 0.044 Trongsa yes=1 -4.841 -4.380 0.000 2.146 5.810 0.000 Zhemgang yes=1 -0.183 -0.200 0.842 0.844 2.370 0.018 Sarpang yes=1 7.383 9.490 0.000 1.913 6.150 0.000 Lhuentse yes=1 -6.382 -6.130 0.000 1.594 4.570 0.000 Mongar yes=1 -5.668 -6.730 0.000 3.136 10.240 0.000 Trashigang yes=1 -3.112 -3.910 0.000 1.448 4.860 0.000 Trashiyangtse yes=1 -1.892 -2.130 0.033 -0.895 -2.640 0.008 Pemagatshel yes=1 -2.009 -2.370 0.018 0.469 1.450 0.148 SamdrupJongkhar yes=1 -0.408 -0.500 0.620 2.326 7.330 0.000 Intercept -19.614 -26.230 0.000 -4.272 -15.380 0.000 Number of obs 57606 57606 LR chi2(30) 2695 10373 Prob > chi2 0.00 0.00 Pseudo R2 0.03 0.05 Source: Authors' calculations based on RNR Census 2009 *: Mithun Cross, Jersey Cross, Brown Swiss Cross **: Mithun Pure, Jersey Pure, Brown Swiss Pure ***: Zo/Zom, Buffaloes, Horses, mules, Donkeys ****: Goats, Sheep 32 Map 22 % of households that hold livestock in 2008 (RNR Census 2009) % 90 − 100 80 − 90 70 − 80 60 − 70 0 − 60 % of households that hold livestock in 2008 (RNR Census 2009) Map 23 Number of cattle in 2008 (RNR Census 2009) Numbers 2000 − 4200 1500 − 2000 1000 − 1500 400 − 1000 0 − 400 Number of cattle in 2008 (RNR Census 2009) 33 Map 24 Share of improved cattle in cattle population in 2000 (RNR Census 2009) % 80 − 100 60 − 80 40 − 60 20 − 40 0 − 20 No data Share of improved cattle in cattle population in 2000 (RNR Census 2009) Map 25 Share of cross-bred cattle in cattle population in 2008 (RNR Census 2009) % 80 − 100 60 − 80 40 − 60 20 − 40 0 − 20 Share of cross−bred cattle in cattle population in 2008 (RNR Census 2009) 34 Map 26 Number of yaks in 2008 (RNR Census 2009) Numbers 2000 − 3500 1000 − 2000 250 − 1000 100 − 250 0 − 100 Number of yaks in 2008 (RNR Census 2009) 6. DIVeRSIFICATIon, CoMMeRCIAlIzATIon, AnD welFARe In this section, we look at the link between welfare, diversification, and commercialization, based on the data of the RNR Census of 2000. This is an important topic given the ongoing discussion in the country on food security and food self-sufficiency and what role agricultural diversification can play for improved access to food as well as for welfare in general. For the welfare indicator, a quality of housing indicator was created for each geog, i.e. an index of equally weighted quality housing characteristics of roofing, wall, lighting, water access, and type of toilet. In Table 12, we rank the quality of housing index by terciles and then cross-tabulate these terciles with some measures of diversification and commercialization. The Table shows that geogs that have a better housing indi- cator are associated with significantly different crop production and sales activities. The poorer geogs allocate more land towards maize. The poorest geogs have 42% of their cropped area used for maize cultivation. This compares to 24 % for the richest tercile. Richer geogs grow more vegetables. Almost one quarter of the value of crop production for the richest geogs comes from vegetables, compared to only 8% for the poorest geogs. Interestingly, fruits are as important for the richest and the poorest geogs, possibly indicating that fewer of the rewards of fruit cultivation lead to large-scale poverty reduction because fruit cultivation might be more associ- ated with larger land owners than other high-value crops. While fruits are an important source of sales income for richer and poorer geogs alike, richer geogs rely much more heavily on vegetables for sales income. Almost one quarter of the sales income from crops in the richest geogs comes from vegetables. This compares to only 6% for the poorest tercile. It is interesting to note that there is a strong association between the wealth of the geog and the Simpson Index of Diversity for sales as its value is significantly higher for the richest two terciles compared to the poorest ones. 35 Looking at outputs per unit land expressed in Nu. per acre, Figure 7 shows that richer geogs have signifi- cantly higher outputs per unit land, almost twice as high. Wealth is thus associated with land intensifica- tion. The sales per unit of land are also significantly higher for the richer geogs than for the poorer ones. Interestingly, we find a U-curve association between sales per unit of land and wealth, where poorer geogs sell a much larger share of what they earn on their land than the richest geogs. The difference between sales and output grows from poor to rich indicating that geogs consume in absolute terms more from their own production the richer they are. These results taken together seem to suggest that increasing dependence on the market, be it selling of agricultural products and purchasing of other products, is not associated with lower levels of welfare in Bhutan. Such results have also been found in numerous other settings. The commercialization of agriculture (in input as well as output markets) has been the cornerstone of economic development for many developing countries and it is generally found that a shift away from subsistence agriculture leads generally to better welfare and nutrition indicators (von Braun and Kennedy, 1994). Table 12. Crop area, value of production, and crop sales by quality of housing indicator (2000) Housing indicator Poorest Tercile Middle Tercile Richest Tercile Cropped area Distribution over crops (%) Paddy 15.2 24.3 27.1 Maize 42.3 35.9 24.5 Other cereals 17.3 18.8 18.3 Vegetables 5.5 8.2 14.3 Fruits 14.3 7.5 9.6 Other 5.4 5.3 6.1 Total 100.0 100.0 100.0 Simpson indicator of diversity 0.690 0.682 0.690 Value of production Distribution over crops (%) Paddy 19.2 33.8 29.4 Maize 23.4 21.1 14.1 Other cereals 3.7 3.3 2.9 Vegetables 7.2 14.5 23.6 Fruits 44.0 25.0 27.8 Other 2.5 2.3 2.1 Total 100.0 100.0 100.0 Simpson indicator of diversity 0.648 0.628 0.628 Value of sales Distribution over crops (%) Cereals 0.5 1.9 1.8 Vegetables 5.7 19.6 30.8 Fruits 93.0 77.7 66.5 Other 0.9 0.8 0.9 Total 100.0 100.0 100.0 Simpson indicator of diversity 0.297 0.470 0.497 Source: Authors’ calculations based on RNR Census 2000 36 While the data show that households and geogs seem overall to benefit from higher commercialization rates and from the switch to high-value crops, not all households or geogs are however able or willing to make the switch. This might be due to several reasons (Minot et al., 2006): 1/ Lack of credit might limit some households to invest in more remunerative activities such as fruit cultivation as they need upfront investments that some poor households cannot afford; 2/ Households might lack information or skills in rightly applying production technologies to allow them to switch crops; 3/ The lack of transport infrastructure might limit access to markets, might change the incentives for the cultivation of high-value crops, or might make the cultivation of perishable crops prohibitive; 4/ Lack of access to input markets, such as for seeds and for appropriate pesticides, that might be more important for these high-value crops than for staples might not be functional for poorer households. Minot et al. (2006) argue that some of these constraints could possibly be removed through the establishment of cooperatives and farmer organizations. Figure 7. Value of output and sales (nu/acre) by quality of housing indicator, 2000 0 50 00 10 00 0 15 00 0 20 00 0 N u pe r ac re 20 40 60 80 100 Housing indicator index output sales A shift away from staples towards high-value products such as fruits and vegetables often exposes the farmer also to more market risks, in particular because of the greater perishability of fruits and vegetables but also because of higher price risks. We illustrate this price variability in the case of Bhutan. Two types of volatility characterize fruits and vegetables compared to staples: 1/ In the longer-run, the former show bigger changes, 37 for better or worse for the farmers; 2/ In the medium-run, i.e. over the length of the season, prices of fruits or vegetables are significantly more volatile. Figure 8 shows the nominal price evolution in the case of local rice (in Gelephu only as time series were only available for this market). Over a ten-year period, prices changed only a little bit: they hovered between 10 Nu. per kg and 14 Nu. per kg. When calculating a seasonal price index (Figure 9), no significant seasonal pattern can be detected illustrating the relative stability of prices over the long- and medium term. When we do a similar exercise in the case of two important vegetables, i.e. potato and tomato, over the same period, we see that there have been rather significant changes in the long-run where the prices of potato have changed between 2 and 20 Nu./kg and those of tomato between 4 and 42 Nu./kg (Figure 8). We also find some important seasonal indices in both cases, reflecting the respective production periods for both products in the year (Figures 10 and 11).7 Figure 8. Prices of local white rice, potato, and tomato 1996-2006 (in gelephu) 0 5 10 15 20 25 30 35 40 45 96 97 98 99 00 01 02 03 04 05 06 N u/ kg Price rice Price potato Price tomato 7 The formula for a Central Moving Average (CMA) used to smoothen the prices in the graph can be expressed as follows: CMA P Pt i i i t i t i t i t 12 6 5 6 5 24= + é ë ê ê ù û ú ú = - = + = - = + åå / The seasonal index (SI) can be calculated as a division of the original prices by the CMA: SI P CMAi i= æ è ççç ö ø ÷÷÷÷*100 38 Figure 9. Seasonal index of rice prices 1996-2006 (in gelephu) 0 0.2 0.4 0.6 0.8 1 1.2 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec month in de x mean-std mean mean+std Figure 10. Seasonal index of potato prices 1996-2006 (in gelephu) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec month In de x mean-std mean mean+std 39 Figure 11. Seasonal index of tomato prices 1996-2006 (in gelephu) 0 0.5 1 1.5 2 2.5 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec month In de x mean-std mean mean+std 7. DIVeRSIFICATIon, CoMMeRCIAlIzATIon, AnD ReMoTeneSS Remoteness is an important determinant of access to input and output markets and might thus often lead to significantly lower productivity and commercialization levels and consequently to higher poverty (Jacoby and Minten, 2009). Remoteness is felt to be an important constraint for poverty alleviation in Bhutan and the new government has pledged to connect each geog center to redress this situation. It is especially expected that agricultural production and commercialization systems, as well as off-farm earning opportunities, will change because of these investments. In this analysis, we will focus on the influence of remoteness on the agricultural sector solely. Transportation- induced transaction costs can generally influence agricultural activities in two ways (Stifel and Minten, 2008). First, they can change the price of inputs versus outputs so that the incentives to use inputs are changed. Given that costs of inputs that have to be imported in the geogs (such as fertilizer) are likely to go up with remoteness, input use per unit of land is likely to fall, and consequently so are yields. As output prices often also fall with remoteness, this leads even to further drops in productivity (Jacoby and Minten, 2007). Second, these input-output price differences will also affect the crop choice. Higher transaction costs might lead to sometimes seemingly inefficient cropping choices of farmers where greater resources are devoted to low- yielding food crops instead of to cash crops that have higher market returns. For example, it has been found in multiple settings that farming households generally shift out of perishable cash crops (e.g. vegetables and fruits) into more storable crops such as staples and pulses as they get farther away from the market centers (von Thunen, 1966). 40 To understand the importance of this phenomenon on agricultural diversification and commercialization in Bhutan, we created a remoteness variable for each geog, based on the RNR Census data of 2000. In the Census, a question was asked on the number of the households that live less than 30 minutes, between 30 and 60 minutes, between 1 and 2 hours, between 2 and 3 hours, between 3 and 4 hours, between 4 and 5 hours, between 5 and 6 hours, and more than 6 hours from a main road. Figure 2 shows the distribution of the households by category. 47% of agricultural households have rather easy access to transport as they live within 30 minutes of a motor- able road (Figure 12). On the other hand, 14% of agricultural households have to walk for more than 6 hours to get to a motorable road. There are strong regional differences in remoteness. While 13% of the households in the West-central region have to walk for more than 4 hours to get to a main road, this is as high as 30% in the Eastern region. Based on this information from the Census, we create a remoteness measure for an average household in each geog. This indicator is used in our further analysis. Map 27 shows the variation of the remoteness indicator by geog in the country. There is no strong regional pattern that emerges. Dark shades indicating more remote regions are present in the North and the South, and in the East and far West. However, the map indicates espe- cially that the rural population in the Southeast and the Northwest of the country requires most time to get to a motorable road. Figure 12. Distribution of households by distance to a motorable road, 2000 0 5 10 15 20 25 30 35 40 45 50 <30 min 30-60min 1-2 h 2-3 h 3-4 h 4-5 h 5-6 h >6 h Distance to main road % o f h ou se ho ld s 41 Map 27 Time required to get to a motorable road in 2000 (RNR Census 2000) Hours 5 − 8 2 − 5 .5 − 2 .25 − .5 0 − .25 Time required to get to a motorable road in 2000 (RNR Census 2000) We rely on a decomposition exercise to evaluate to what extent changes in the crop sales between more and less remote areas are achieved by higher yields, crop diversification, and higher commercialization rates. We use a variant of the method developed by Minot et al (2006). If Y is the production per unit area, a is the share of crop i in total cropped area, C is the commercialization rate and P is the price per unit of production, we can write S, the total sales per unit of land and measure of commercialization of a geog, as follows: S a Y PCi i i i i n = = å 1 To measure changes in sales, we take total derivatives of both sides of the equation, yielding: dS a Y PdC a Y C dP a PC dY Y PC dai i i i i i i i i n i n i i i I i i i i i n i @ + + + == = åå å 11 1== å 1 n Changes in sales per unit land are thus explained by changes in commercialization rates (first term), changes in price (second term), changes in crop yields (third term), and changes in crop composition or the diversification term (fourth term). In the case of our remoteness analysis, where we impose the same price for all geogs, the second term disappears from our analysis and we thus link changes in sales to three factors, yields, diversification, and commer- cialization rates. When we divide both sides by dS, we obtain the proportional contribution of each component. A similar decomposition can be done to explain the variation in the value of output (O) between more and less remote areas. Using the same definitions of above, changes in the value of output per land can be explained as follows: O a Y Pi i i i n = = å 1 42 To measure changes in output, we take total derivatives of both sides of the equation, yielding: dO a Y dP a PdY Y Pdai i i i i i i i i i n i n i n @ + + === ååå 111 Changes in output per unit land are thus explained by changes in price (first term), changes in crop yields (second term), and changes in crop composition (third term). To understand these separate effects, we first link the remoteness measure to several agricultural indi- cators (crop area allocation, commercial rates, and crop diversification) through descriptive and graphical analysis. First, we compare the area allocation to different crops between different terciles of remoteness (Figure 13). The importance of cereals over the three terciles is overwhelming, accounting in all three cases for over 70% of the area planted. Within the cereals category, we see however an important shift in composi- tion. While almost 30% of the cropped area in the least remote area is allocated to paddy, this drops to almost 10% in the most remote tercile. The maize area on the other hand accounts for the 26% for the least remote and increases to almost half of the cropped area in the most remote tercile. There is surprising little difference in the area allocation of fruits and vegetables: it changes from 21% in the least remote tercile to 16% for the most remote tercile. However, the relative importance of vegetables changes quite dramatically by remote- ness tercile. While vegetables account for 11% of the area in the least remote tercile, this drops to 4% in the most remote one. Figure 13. Crop area allocation by remoteness tercile, 2000 0 10 20 30 40 50 60 70 80 90 100 Least remote Medium remote Most remote % o f c ro p al lo ca �o n others vegetables fruits other cereals maize paddy 43 Second, a change in the value of output and sales per acre is partly driven by a change in the productivity of land for the same crop. This is illustrated in the case of paddy in Figure 14. Using the data from the census, rice yields are shown to drop from over 1.5 tons per acre for the least remote geogs to 0.75 tons per acre for the most remote geogs. The difference in rice productivity might be driven by higher input use, better technology adoption, more intensive labor cultivation, and/or better irrigation infrastructure. Given that we lack data on each of these inputs, we are unable to unravel the exact contribution of each towards the higher productivity. Figure 14. Rice yields (kg/acre) by remoteness (hours to motorable road), 2000 0 20 0 40 0 60 0 80 0 10 00 12 00 14 00 16 00 kg s pe r ac re 0 2 4 6 8 remoteness Third, commercialization changes by distances to roads and commercial centers (Fafchamps and Shilpi, 2003). The census data show overall that a low percentage of the cereal production is being commercialized. While some of these values might seemingly be too low (as compared to the recent RGB-FAO value chain studies), we will however use these data as they are the only source available at the national level. Table 13 shows that there is a strong concentration in the importance of crops in the total value of sales. In the least remote tercile, apples make up almost 50% of the total sales. This is in strong contrast with the most remote tercile where the sales of oranges constitute over 80% of the total sales. The sales of potato is not important in the most remote tercile while it makes up between 14 % and 18 % of the total sales of the two closest terciles. 44 Table 13. Importance of different crops in total sales by remoteness tercile (hours to motorable road), 2000 Least Remote Medium remote Most remote Cereals 1.68 1.62 0.44 Potato 13.96 18.44 3.29 Chili 4.08 3.23 0.34 Other vegetables 4.21 3.26 2.32 Apple 39.64 1.66 0.27 Orange 31.93 66.80 8