ABSTRACT
Agricultural development is an essential engine of poverty reduction in sub-Saharan Africa, where an estimated 75 per cent of the extreme poor reside in rural areas (Livingston, Schonberger, & Delaney, 2011), and are largely engaged in agriculture-related activities. While the exact relationship between poverty reduction and agricultural growth in any country depends on the agricultural and social structure of a given location (DFID, 2004; Prowse & Chimhowu, 2007), development in the agricultural sector tends to result in greater benefits accruing to the poorest segments of the population, with a 1 per cent rise in agricultural GDP resulting in an estimated 6 per cent increase in income growth for the poorest 10 per cent of the population (Chen & Ravallion, 2007; Ligon & Sadoulet, 2008). The connection between agricultural growth and poverty reduction has been tied to various path-
ways, such as the creation of wage employment in rural areas. In particular, growth in smallholder agricultural productivity continues to be heralded as a key driver of poverty reduction: for every 10 per cent increase in farm yields, Irz, Lin, Thirtle, and Wiggins (2001) estimate that there has been a 7 per cent reduction in poverty in Africa. Given that the pool of smallholders on the continent is vast, with approximately 33 million farms of less than two hectares in size,1 policies that increase the productivity of small-scale farmers can serve as important drivers of poverty reduction and improved food security in sub-Saharan Africa. Despite the key role of smallholder agriculture in the sector and the economy as a whole, serious
weaknesses persist in the measurement of agricultural outcomes and in our understanding of the factors hampering agricultural growth among smallholders. While governments and donors alike target agriculture for large-scale investments with ambitious goals of raising agricultural productivity multifold, little is done to ensure that accurate statistics are produced to monitor agricultural development. For instance, of the 44 countries in sub-Saharan Africa rated by the Food and Agricultural
Organisation (FAO), only two are considered to have high standards in data collection, while standards in 21 countries remain low (FAO, 2008). Further compounding the problem is the fact that the poorest countries – for which agriculture is a critical source of livelihood – often have the poorest data, being least able to direct their limited resources into improving the quality of their statistics (African Development Bank, 2004). In spite of the clear need for empirical evidence, these countries lack the financial resources to
generate survey or administrative data of sufficient quality and scope to inform policy, let alone to fund these policies. In the 2003 Maputo Declaration on Agriculture and Food Security in Africa, in recognition of the importance of the sector for the ‘economic prosperity and welfare of its people’, African countries committed to allocating at least 10 per cent of national budgetary resources for the implementation of sound policies for agricultural and rural development (African Union, 2003). However, a 2011 report on financial resource flows to agriculture by the FAO found that although government spending on agriculture has increased for developing countries as a whole, it has decreased as a share of total spending. In particular, one of the key messages of the report was that ‘trends in indicators of government spending on, ODA to, and FDI in agriculture are discouraging for sub-Saharan Africa’ (FAO, 2011; pg. 37). Even with sufficient financial resources, countries often lack human resources to collect data in a
cost-effective and sustainable manner. External support from donors can provide a short-term patch, but typically has not been successful in leaving in place sufficient capacity to continue the data collection work when the support ends. The low level and inconsistency of budgetary contributions to statistics from own governments, as well as erratic and short-term donor support, directly results in inconsistencies in data collection activities in many countries. This has significant implications for data quality. As one example, if the implementation of a survey depends on irregular financing by donors, it
becomes extremely difficult to plan in advance for multiple years of survey efforts, which in turn has negative repercussions for the collection of time series and panel data. However, as much of the existing agricultural data is cross-sectional, changes across time with regard to specific indicators typically cannot be well captured. The data are unable to track the changes in indicators over time, or to follow important phenomena such as the transition out of agriculture into potentially higher-return activities. In their review of agricultural development, rural non-farm activities and rural poverty, Foster and Rosenzweig (2008; pg. 3055) note that ‘very few studies permit direct comparison over time using comparable measures’. Other studies have likewise noted that data quality issues limit analysis (Ngendaumana, 2001; Tiffen, 2003). Past investments and technical assistance efforts in the area of agricultural statistics have failed to produce sustainable systems, while existing statistics continue to suffer from poor quality, lack of relevance, and little use in national policy dialogues (Binswanger, 2008). The challenge of improving agricultural statistics worldwide is daunting. Recent efforts such as the
Global Strategy to Improve Agricultural and Rural Statistics (henceforth referred to as the Global Strategy) and the ensuing regional Action Plans are testament to the renewed commitment of the global community of researchers and practitioners to rejuvenate the sector, following decades of under-investment (World Bank, United Nations and FAO, 2010). The first pillar of the Global Strategy focuses on the identification and establishment of core data with a focus on agricultural productivity and the most important crops to global agriculture production. Due to the enormity of the task at hand, this article sets out to inform the debate in a targeted and selective fashion by addressing a number of specific issues which are the focus of a recent initiative, namely the Living Standards Measurement Study – Integrated Surveys on Agriculture (LSMS-ISA). Specifically, two claims are made. First, in the advent of new technologies becoming increasingly
available at relatively low costs, more rigorous research is needed to create and promote improved, cost-effective standards in agricultural statistics. Improvements in methods for collecting smallholder agricultural statistics have been particularly sluggish over the past three decades and present the typical market failure problem, with clear disincentives for private investments. For instance, the latest guidelines by the FAO on yield measurement date back to the early 1980s, when modern technologies
were not available. The lack of up-to-date research on survey methodologies has led to serious gaps in the existing knowledge base, limiting the identification and promotion of effective policies. Second, statistical systems for agriculture lack integration, limiting the utility of the data for examining linkages between agriculture and key issues such as poverty or nutrition, as well as linkages between socio-economic variables and environmental conditions. In order to better inform agricultural policies, approaches based on the enhanced integration of agricultural data and other types of data sources are needed. This article is not meant to be a comprehensive review of the issues plaguing agricultural statistics
but a purposive discussion of selected shortcomings of current systems. Its contribution is meant to focus on a number of well-defined issues which we believe to be both tractable and to offer a high return in terms of data quality and policy relevance. In the ensuing discussion, emphasis is placed on the African continent due to the geographic focus of the LSMS-ISA initiative, as well as the greater potential of smallholder agriculture for poverty reduction and growth, highlighting the importance of overcoming what Devarajan (2013) deems a ‘statistical tragedy’ towards creating innovative, wellinformed agricultural policies.
