As an input to the Initiative for Open Ag Funding the team at Open Data Services were commissioned to write a Landscape Analysis, looking at the current sources, and gaps, for information on funding for agricultural development.
You can access the full report to download here (3.5 MB), or in an online copy (open for comments). It includes a detailed appendix looking at current coverage of agriculture data from IATI Publishers.
We would love to get your reflections and feedback, ideally in this thread, but also as comments in the online version if there are specific points related to elements in the text.
This report, along with stakeholder engagement to identify user needs, carried out by the Open Ag Funding partners, will inform the next stage of work, exploring approaches to improve the completeness and comprehensiveness of agricultural funding data through IATI.
And in case you’ve only got a few minutes, here’s the headlines and recommendations from the Executive Summary:
Agricultural Investment Data - Landscape Analysis: Executive summary
Investments in agriculture are critical to reducing poverty and improving food security and nutrition. Although billions of dollars are spent on these investments every year, comprehensive and detailed information on these investments is still largely unavailable.
This study explores the current landscape of data on agricultural investments, defined as “public and private funding into agriculture, forestry and fisheries development, including policy development, research, land and water resources, agro-industry and extending to nutrition-sensitive agricultural interventions and developmental and emergency food aid funding.” In doing so, it identifies key opportunities and challenges for the Initiative for Open Ag Funding to address in working for a clearer and more reliable picture of agricultural investments worldwide.
Landscape overview
Information on agricultural investments can be considered at three main levels:
- Aggregate information – country or region level data on investment, including data from national statistics, national budgets, and regular or ad-hoc surveys.
- Project level – Projects funded or implemented. Drawn from donor systems, country aid management platforms, crowdsourcing, and company portfolios.
- Outputs and results – Results data, reports and research outputs, held in custom platforms, internal systems and research repositories.
Although some processes may generate aggregates by collecting project level information, and some results data is linked to project information, in general the information systems and data collection approaches for each level currently operate independently. In exploring current data sources, we identify that:
- Producers of aggregate information rely on periodic surveys - using their authority to request aggregate statistics, and, in some cases, basic project lists.
- Project level information covers only the basic fields (project names, dates, values, and broad categorisation) with only a few investors providing more detailed information.
- Tools and platforms carry out their own data enrichment through re-classifying project level information with alternative or more detailed taxonomies, adding geocoding to projects, and requesting additional information from partners. Not all the results of this are openly shared.
- A small number of tools capture implementation and results data, and link outputs, results and projects. However, these are unlikely to produce data covering a large proportion of agricultural investments over the short-term.
The most promising framework for comprehensive and timely project-level agricultural investment data is provided via the International Aid Transparency Initiative (IATI). Data is available in IATI format for the majority of DAC donors, and over 35,000 activities in the IATI corpus are tagged with an agriculture-relevant sector code. Of these, 5,000 provide meaningful locations, and 2,000 have some form of results data.
Although a range of well developed, and well interlinked, vocabularies for detailed classification exist in the agriculture field, we have not found use of these in categorising project-level information. Instead, in addition to use of the OECD DAC Sector codes, a wide range of organisation specific classification schemes (usually with a relatively low number of terms, in the tens rather than hundreds) are used, based on local policy and practice requirements.
Manual tagging ‘at source’ against shared taxonomies is unlikely to gain traction except in agriculture specialist organisations, but the increased publication of rich textual descriptions and project-associated documents has the potential to support automated tagging.
Taking IATI as the best starting point to secure comprehensive and timely information, we have identified a number of potential strategies to pursue. The relevance and value of these approaches should be judged in light of clear user-needs analysis (taking place in parallel to this study).
Recommendations
- Develop a quality framework for agricultural investment data rooted in user-need, and identifying the key fields of data, and qualities of a dataset, that are important. This should be supported by clear documentation that shows how IATI publication can be used to provide the required data.
- Pilot enhanced survey processes working to pre-fill project-level information in surveys, and request corrections and additional information, rather than requiring duplicate data collection.
- Provide data quality assurance tools that help data publishers to monitor the quality of the data they are providing, and that incentivise improved data publication.
- Provide a publisher engagement and support offer to go alongside data quality tools. This may include diagnostic tools and processes to identify the scope to improve data, followed by a managed support relationship to step publishers through the technical and policy challenges of improving their data, or publishing for the first time.
- Provide multiple entry points – recognising that not all data will be published in IATI format, and providing a minimum data model for scraped, crowd-sourced and otherwise sourced agricultural investment data.
- Improve data enrichment processes and tools – recognising that not all data will be improved at source, and human and automated enhancement will be needed. This requires research into automated classification tools, and engagement work to encourage open licensing and publication of enriched data.
- Develop and maintain a corpus of agricultural investment data – to act both as a means of measuring data quality (and ultimately, project success), and to make it easier for users to find quality assured and integrated data on agricultural investments.
The challenge ahead is not to create perfect data, but rather to ensure that data is good enough to improve planning, action and evaluation. We are starting from a good place, with increased flows of project-level information over recent years – but there is still a long road ahead. This report, its Appendices, and associated datasets provide an initial landscape map that can support the first steps on that journey.
@stevieflow and I look forward to your comments and reflections.