(Item 11 of the Agenda)
105. Two documents, APCAS/98/12 “The Introduction of a Sample Survey for the Collection of Agricultural Commodity Data in Australia” and APCAS/98/13 “Striking the Balance Between Censuses and Surveys, Their Frequency, Scope and Coverage” were presented to the Commission.
106. The Commission was informed that Document APCAS/98/12, which was presented by the Assistant Director, Economic Statistics Branch, ABS, articulated the challenges confronting the Australian Bureau of Statistics as the result of the recent shift from annual commodity censuses to quinquennial censuses and sample surveys.
107. The Commission was informed that following a decision by the Australian Federal Government in May 1997, the ABS moved from conducting annual censuses of agriculture, which it had done for more than 100 years, to quinquennial censuses, with the implementation of large sample surveys in inter-censal years. The first Agricultural Commodity Survey (ACS) was conducted in respect to the 1997-98 reference period. Prior to implementing ACS, an extensive consultation process was organized to inform users of the decision and to seek their views on the objectives, scope, coverage and contents of the new survey. Members of the User Advisory Group accepted the proposal to adopt a large stratified random sample approach for the collection of Commodity Statistics.
108. The Commission was informed that ABS endeavoured to make the scope, coverage and content of the ACS comparable with that of previous agricultural censuses. Thus, in the ACS, the reference year was retained as the year ending 31 March, similar questions were asked and the scope of the collection consisted of all agricultural establishments on the ABS’ list frame having an Estimated Value of Agricultural Operations (EVAO) of A$ 5000 or more. This strategy ensured the continued availability of the full range of commodity data required to provide estimates of the value of agricultural production, a major requirement for most users, and essential for the compilation of the national accounts.
109. The Commission noted that the 1997-98 ACS used a stratified random sample of agricultural establishments. It was informed that stratification was used to improve the efficiency of the sample and to control the quality of statistical output.
110. Some agricultural units though, were selected with certainty. Units were completely enumerated if they were: (i) located in one of Australia’s 2 territories which each contained relatively few farms; (ii) extremely significant in terms of EVAO, area of holding or their contribution to production for a given commodity; and (iii) multi-establishment units (i.e. those operating a number of properties in distinctly separate geographical areas); (iv) producers of rare commodities, defined as those commodities that are important in terms of value of production, but produced by a few farmers. ABS found this sector to be too large and was considering to have it significantly reduced in future surveys.
111. The Commission was informed that ACS used “number-raised” estimation procedure which was proven in earlier exercises to produce more reliable commodity estimates between years than ratio estimation due to the impact of significantly changed agricultural production resulting from atypical climatic conditions.
112. The Commission was informed that historically the principal source of updating information was obtained from census forms. This information related to business name and address changes and details of purchases and sales of land. In addition to Agricultural Census feedback, a range of other coverage sources were also used to update the Business Register. These included group employer registrations for farms that employ labour (from the Australian Taxation Office); details of land ownership changes from relevant state government agencies; lists of producers obtained from industry associations; and information from agriculture-related publications.
113. The move to a sample survey, had accelerated the need to investigate other coverage sources. The ABS was currently investigating the feasibility of using lists of sales tax exemption registrations from the Taxation Office as well as Australian Customs Service lists of diesel fuel rebate applicants. However the loss of coverage of the population list frame during the inter-censal period remained a major concern.
114. The Commission noted that major implications of the shift to sample surveying included a A$ 1 million reduction in the budget of the ABS’ Agriculture Program as well as significant reductions in the statistical reporting load placed on small agricultural businesses.
115. These advantages, had, however, been offset by the considerable loss of “small area” regional data. While census data were released at the Statistical Local Area (SLA) level, the ACS results would only be available at the Agro-Ecological Region (AER), and possibly, the Statistical Division level.
116. Many clients also believed that quinquennial censuses were too infrequent to accurately measure inter-year variations in agricultural production deriving from varying climatic conditions, changing farm management practices or the volatility of international markets. They were also concerned that a sample might not sufficiently identify growth in emerging industries.
117. Another implication of the move to a sample survey approach was the introduction of sampling error and the effect this might have on the quality of estimates, particularly for minor commodities at the regional level. Any reduction in the quality of estimates also had the potential to impact upon the quality of other ABS data series, such as the Value of Agricultural Commodities Produced (VACP) and the Apparent Consumption of Foodstuffs series.
118. During intercensal years there would be an additional effect on the quality of estimates due to progressive deterioration of the population list frame. A degraded list frame had implications for future ACS samples and would impact upon other surveys conducted by the ABS, such as the Agricultural Finance Survey (AFS) as well as Farm Surveys conducted by ABARE.
119. The Commission was informed that Document APCAS/98/13 provided a discussion on the issues that might be considered by member countries in planning whether to use complete enumeration or sample survey techniques for the census of agriculture.
120. The Commission was informed that in recent years, many countries had seen a growing demand for a large variety of data attributable in part to the rapid changes in the agricultural structure as the sector responded to market globalization. However, available resources for data collection had been on the decline. A tendency had therefore developed to increase the frequency of censuses or to shift to structural surveys (sample enumeration).
121. The Commission noted that the FAO Programme for the World Census of Agriculture 2000 (WCA 2000) was the eighth decennial Programme. Unlike earlier programmes on census of agriculture, the WCA 2000 did not advocate for the simultaneous conduct of the census in all countries in a given year. It recognized the wide disparity in the current status of economic and statistical development in different countries. WCA 2000 encouraged countries to develop and implement an agricultural census according to their unique situation. The Commission noted that WCA 2000 encouraged standardization of a minimum core programme and a complementary relationship between the census and the more frequent food and agricultural sample surveys.
122. The Commission also noted the need for country-specific multi-year integrated statistical programme as component of a national information system for decision-making in food, agriculture and rural development. The Commission also recognized the need for all statistical development efforts to be oriented toward the long-term goal of establishing a national statistical programme that was user-driven. This implied that the system should be able to provide a continuous flow of timely and accurate data covering all relevant aspects of food, agricultural and rural development.
123. The Commission likewise noted that at the minimum, country-specific programme on agricultural census should meet the three basic objectives of WCA 2000, thus: (i) provide aggregate totals for fundamental agricultural data to use as benchmarks for inter-censal estimates; (ii) provide a frame for other agricultural surveys; and (iii) provide data for small administrative units and detailed cross-classification of farm structural attributes.
124. The Commission recognized that the decision on whether a census or structural survey should be undertaken in any one country should consider the actual needs of the prospective users. It noted that when agricultural censuses were infrequent, and especially when they were undertaken by an agency different from that conducting agricultural surveys, loss of institutional knowledge often occurred. That, as well as the availability of a suitable frame from a population census were elements to be factored into the decision process along with the availability of resources and organizational capability of the institution.
