33. In STAT/LVSTK/9, Forecasting of livestock numbers, Roderick FORBES informed the Expert Consultation that there had been discontinuities between 1996 and 2002 for livestock statistics in New Zealand and that the Ministry of Agriculture and Forestry (MAF) in New Zealand undertook forecasting of prices, production and trade because of agriculture and forestry’s importance to the economy. He pointed out that production and processing of livestock and livestock products accounted for 47 percent of exports and 14 percent of GDP and that the forecast of livestock numbers underpinned agricultural production and trade.
34. He stressed that female breeding livestock numbers were crucial to forecasting production and indicated that MAF used econometric models to forecast livestock numbers. He illustrated the methodology with reference to three equations - sheep breeding flock, dairy milking herd, and beef breeding herd. He cautioned that judgment was required when making adjustments both during the running of the model and on the model outputs. Based on his experiences that survey data between 1996 and 2002, he confirmed that there was a critical need for annual surveys to validate forecasting models and their results.
35. In response to queries he pointed out that livestock forecasts were the basis for forecasting production and export quantities and that MAF made the forecasts available to the wider public on the internet, www.maf.govt.nz.
36. Mr Forbes agreed to prepare an analysis of how well MAF had forecast livestock numbers in the past, as there was interest expressed by other experts at the meeting. He informed the Expert Consultation that livestock diseases could be incorporated into the forecasting model through the use of dummy variables, but had not been included previously because losses to animal disease were very low in New Zealand.
37. The Expert Consultation agreed that forecasting should be included in the programme of a country’s livestock statistics.
38. In STAT/LVSTK/10, Treatment of data for nomadic and semi-nomadic livestock, Jean-Francois MICHEL pointed out that livestock mobility can be monitored. For cattle or small ruminants husbandry systems in the developing countries, all the intermediate degrees of mobility existed between sedentary and nomadic husbandry systems. Availability of feed and water and the onset of diseases were the main cause of movement that created problems for collection and processing of data, especially mapping.
39. He presented census methods and data processing solutions that were used in selected censuses conducted in countries in Africa with different levels of livestock mobility. He informed the Expert Consultation that in Burkina Faso a cattle census was conducted at the local level as part of an animal trypanosomiasis research project to identify the contact zones at the river system scale in a Geographic Information System (GIS) framework. For that project, it was necessary to map livestock distribution taking into account daily movements. He mentioned that a spatial model had been developed using GIS to link households to water points frequented by cattle at the end of the dry season. This simple model provided a better representation of the survey units than a simple cartography
40. Mr Michel reported that a national census of transhumant and nomadic livestock was conducted in Mali, based on concentration areas frequented by cattle at the end of the dry season. The general census method consisted of making an exhaustive ground enumeration of the animals in these areas. The annual movements of herds were defined by their location in the rainy season and the dry season. These locations were recorded in a short questionnaire. A GIS framework was built to produce static maps of transhumant and nomadic livestock distribution in the dry season, maps of the concentration areas, and dynamic maps of the annual movements of livestock.
41. He explained that national agricultural censuses with the introduction of sampling techniques were ongoing in Niger and Mauritania. In these countries, cross-border movements of livestock were very important, but it was not recommended to conduct separate nomadic and sedentary livestock censuses. Given the size of these countries and the available funds, it was decided to adopt a sampling strategy by selecting villages in Mauritania and census districts in Niger. For these spatial samplings, surveys were to be conducted when herds were the most homogeneously distributed in the territory. The end of the rainy season is the optimal survey period. The GIS of each country was currently under development.
42. He said that the census methods used for livestock depended on available resources and specific characteristics of farming systems. However, he noted that despite differences, it was possible to pinpoint common principles for livestock censuses: (i) Survey units should be well defined (e.g. the herd for livestock censuses); (ii) sampling units should be easy to identify in the field; (iii) proper information about cattle breeders was crucial, and (iv) the questionnaire should be limited to the most relevant items. He suggested that the development of the census methodology required a good knowledge of local breeding systems, a good understanding of sampling techniques, expertise in GIS, and the acquisition of basic geographic layers of the census area.
43. He stated that the integration of data on livestock in a GIS constituted a new decision tool for livestock policy and that livestock data layers were essential, but often required surveys which were complex and expensive. He believed that the identification of indicators for livestock distribution and movements would significantly reduce field effort.
44. Slamet MUKENO introduced, in STAT/LVSTK/11, the discussion about Assessment of slaughter statistics from non-official slaughterhouses. He mentioned that, for most of the population in Indonesia, religious obligations required individuals to make sacrifices to help the poor and that many animals were not slaughtered in official slaughterhouses. In many cases, these slaughterings were carried out either within the household or within the religious compound and were not reported formally.
45. He illustrated the efforts of Statistics Indonesia (BPS) to estimate the amount of non-reporting through a special study in Jakarta where households in selected census blocks were asked about animals slaughtered in a year by the household. He said that approximately 17 percent of the respondents had slaughtered animals, but that well over 50 percent of these slaughterings were not reported. He cautioned that the number of census blocks was minimal and that the results required further confirmation.
46. Joo-Hwan KIM, in STAT/LVSTK/12, Sampling considerations for livestock sector surveys in Korea, presented a paper about the new sampling design and the methodologies adopted for the “Livestock Statistics Survey”. He reported that in October 2002, the Ministry of Agriculture and Forestry in the Republic of Korea changed the sampling methods for the Livestock Statistics Survey by using the results of the “2000 Agriculture Census”.
47. He further reported that there were large changes in the livestock population between the 1995 and 2000 national agricultural censuses. He explained the survey objectives and the basis for the new sampling design, including the criteria for stratification, sample size and the estimation procedures. The sample design was a stratified sample with complete enumeration of the largest farms and without rotation of the selected sample farms. The paper included recommendations to improve the effectiveness of the survey design, including the enumeration of the large farm stratum before the other strata.
48. In STAT/LVSTK/13, Strategies for collection of data on livestock, Mr Diedhiou encouraged the use of GIS tools to geo-reference livestock data. He described sampling methods not commonly used (capture-recapture, adaptive sampling, distance sampling and double sampling) and suggested that their use for livestock surveys be explored. The use of satellite imagery was also proposed.
49. He pointed out that it was important for countries to consider appropriate means for dissemination of information about the livestock sector and listed several alternatives, such as the use of internet, television, radios, CD-ROMs and hard-copy publications.
50. The experts recognized the need to have a coordinating body in countries where more than one institution was involved in livestock statistics and the potential for the use of spatial sampling techniques.
51. The paper, Estimation of livestock products, STAT/LVSTK/14 was presented by Ashok KUMAR. He reported that in India data on livestock were collected by various agencies through censuses, surveys and reports/returns. He mentioned that the Livestock Census was conducted every five years. At present it was the responsibility of the Department of Animal Husbandry and Dairying (DAHD).
52. He described the integrated sample survey scheme, developed by the Indian Agricultural Statistics Research Institute, which was used for estimation of production of milk, eggs, wool and meat. He noted that the survey was conducted throughout the year in three seasons, each of four months called ‘rounds’. He explained that, in each season, 5 percent of the villages in the district were selected for complete enumeration. A specified number of houses were selected randomly in each selected village for the estimation of milk, eggs, wool and meat. For each household, animals were selected for calculation of the yield per animal. Complete details of the sample design and steps in the calculation to estimate all-India statistics were given in the paper. He mentioned that special studies were planned to fill data gaps.
53. Nenita YANSON presented the paper Supply-utilization accounting for meat and animals, STAT/LVSTK/15. She illustrated the value of supply-utilization accounting (SUA) for Meat and Animals. She pointed out that the livestock subsector contributes about 30 percent of the agriculture GDP and that broilers and layers agribusiness have become the catalysts of growth for the agriculture sector. She said that with the emergence of global markets, the demand for adequate, relevant and quality statistics and information was rising.
54. She indicated that SUA analysis was a tool to ensure quality and acceptability of data. She explained that in the Philippine Agriculture Statistical System, the SUA framework of analysis for animals and meat had already been set up at all geographic levels of the data review process and that this kind of analysis had been very helpful in assessing the country’s livestock numbers and their uses (in particular the meat available for consumption and the per capita level of meat). Moreover, it had enhanced usability of livestock data produced by the Bureau of Agricultural Statistics (BAS), especially by the livestock sector planners and agribusiness.
55. During discussion of the paper she confirmed that SUA output had been used in the estimation of GNP/GDP and had enhanced the image of BAS as well as establishing the trust and confidence of agribusiness.
