المنتدى العالمي المعني بالأمن الغذائي والتغذية

U.S. comments on Data Collection and Analysis Tools for Food Security and Nutrition  

The United States thanks the High-Level Panel of Experts (HLPE) for their work in producing this Zero Draft Report on data collection and analysis tools for food security and nutrition. We appreciate the opportunity to provide feedback early in the process and look forward to continued engagement and consultations as the workstream develops. Our general comments are below, followed by more specific comments, and finally comments that respond to question eight posed in the consultation.

General Comments:

• The United States believes the Introduction should be re-written with a more balanced tone. Statements such as, “food systems have failed us” overlook the complexity of international food security and overlook significant food security goals that have been achieved over the last twenty-five years – despite challenges associated with COVID-19. The Introduction should set the scene, but also seek to identify achievements and opportunities rather than simply disparaging the agricultural sector writ large.

• The United States believes that it is critically important for the HLPE to use internationally recognized and agreed concepts and definitions for issues that underpin the reports it produces. The four pillars of food security are well known and internationally accepted and therefore must provide the basis for this and future reports. We note that the concept of “agency” is vaguely defined and the linkages between this concept and food security remain unclear, and the definition of “sustainability”, as utilized by Clapp et al., fails to sufficiently consider all three pillars: economic, social, and environmental and seems to already be partially captured by the dimension on stability. Further, the evidence base for the use of this new definition is severely limited. The HLPE citing itself and a journal article published in 2022 (several of the authors of which also sit on the HLPE) that simply makes the case for six dimension of food security is insufficient and inappropriate for underpinning a Report that is designed to provide the foundation for multilaterally agreed policy guidance. We strongly urge the HLPE to stick to internationally agreed definitions so as to reduce confusion and ensure the panel’s credibility as a science-policy interface.

• Overall, in the Report, there is not enough focus on available measures. Validated measures are necessary in order to collect data. If the position of the HLPE is that adequate measures are already available, that should be made clear. In reading the report, the assumption seems to be that measures are available, but that high-quality data is not being collected. We are not sure that is the intended message. We suggest adding a chapter to the report between current chapters one and two that focuses on measures or indicators that could or should form the basis of data collection initiatives. This will help to make later discussions more concrete.

• The Report recognizes that a major challenge with many data sources is the timeliness of the data.  However, we did not see explicit recognition of what can be a larger challenge: that most data collected is a lagging indicator – a snapshot of how things were at a moment in time in the past, rather than how things necessarily are now, or how they may become.  Current developments may not be reflected in data, due to sudden developments (e.g., civil strife).  Thus, the value of sets of indicators that are forward looking or predictive, in addition to those that measure at a past moment in time.

• We are glad to see discussions of data quality (4.4.3 and mentioned in 3.3) but believe the report would benefit from further consideration of data integrity. The discussion regarding interpretation in Sec 3.3 and in Sec 4.4.1 on use of the data for political purposes is on target, thought it only looks at the post-collection use of the data and not ensuring unbiased data collection. Sec 5.1 doesn’t fully elaborate on this either.

Specific Comments

• P. 10-11: The discussion on “Conceptual framework for a systemic view of FSN determinants and outcomes,” including Figure 1, does not appear to explicitly recognize the role of war and other armed conflict or civil disturbance as a proximate driver of food insecurity. Yet we know that it is a major driver in some countries / regions (as recognized in the first sentence in Box 2 on Page 20).  But it should perhaps also be reflected explicitly in the -macro and/or -meso level determinants discussion on page 10 and in Figure 1 on page 11.
• P. 10 of 63, Par 6: Why is the focus only on “local food, health, and environment systems”? This should be explained or broadened.
• P. 16 of 63, Example 1:
  • What is “ASF” intended to convey in this table?
  • Column 3, row 2: what is meant by “Relevant policy (e.g., exist, and/or enforce)”? Please explain.
  • Column 2, row 4: Inappropriate focus on “local producers”. In isolation, this undermines the importance of supply chains in addressing shared objectives.  Please broaden.
  • Column 3, row 6: what does “risk to access” mean? Please explain.
• P. 17 of 63, Example 4: Consider adding an example related to the intersection of food security and water security or an example related to ensuring food security in the context of poor sanitation or lacking clean water.
• P. 19 of 63: We do not agree that anonymizing the data will make it appropriate to be freely accessible. This is true for personally identifiable information (PII), but not for business confidentiality. Will later drafts of the report address confidential business information, including that collected by the government for regulatory, oversight, or other purposes?
• P. 20 of 63, Box 2: We agree with the premise of Box 2 that drawing crisis, fragility, and conflict data into the FSN data context would be useful. The fact that a country has had multiple humanitarian crises over time can be argued to be a salient data point in informing discussion of FSN development objectives. Without endorsing any particular data source, examples of sources of fragility and conflict data might include the Fragile States Index and the Heidelberg Institute for International Conflict Research’s Conflict Barometer.
• P. 21 of 63: Table 1 begins with a list of multi-country sources of data for FNS. Suggest starting with a list of available multi-country validated measures that can be used in data collections in Table 1.
  • One such measure that is missing from the report is the Food Insecurity Experience Scale (FIES). The FIES can be implemented in national data collections relatively inexpensively as it is a set of 8 survey questions, that are already translated into more than 170 languages. The scale has already been validated as an experiential measure of food insecurity and included successfully in other surveys. Further, FAO already has resources on their website for survey implementation and data analysis https://www.fao.org/in-action/voices-of-the-hungry/fies/en/
  • Another measure that is missing is the Water Insecurity Experience Scales. Access to safe water is closely related to food and nutrition outcomes, and the scale has already been validated. https://www.ipr.northwestern.edu/wise-scales/about-the-scales/index.html
• P. 30 of 63, para 2: “Both the European Food Safety Authority (EFSA) and Codex Alimentarius have databases containing…” – Inappropriate to call out one government’s database in this context.
• P. 39 of 63: We are glad to see reference to the potential role of artificial intelligence and machine learning, particularly as a tool for data interpretation and forecasting.

Responses to Guiding Questions

The following comments are organized around question 8 (although some of the comments may overlap with the other questions)

8. Please provide your feedback on the following:

a. Are there any major omissions or gaps in the V0-draft?

• As highlighted in the data life cycle/ data value chain, analyses of FSN data as well as dissemination of results to end users is of great importance and empirical assessment of FSN outcomes is hard. This is partly because of how FSN outcome is measured and what proxies are used to measure FSN outcomes, as it often the case in practice. For example, it is common to see various studies use varying measures of FSN outcomes, although most would agree on the conceptualization of the FSN outcomes. We find in the literature varying measures that seem to refer to the same FSN outcomes, including, but not limited to, Food Insecurity Experience Scale (FIES), the level of calorie consumed (e.g., 2100 Kcal/capita/day, and some other measures indicating the level and intensity of food and nutrition insecurity (food gap and food severity index) etc. Therefore, it will be imperative to develop a commonly agreed upon metric for measuring food security that is simple and less data intensive to help facilitate data driven decision making. The implication is that while designing data collection for FSN purposes, it will be vital to use demand driven and end user informed approach and incorporate this into the data value chain / data life cycle.
• In the conceptual framework, example in Figure 1, “Individual food security and nutrition outcomes:” it may be important to highlight the level of disaggregation (by household members as well as by gender), which may have repercussions to many aspects of the data life cycle.
• In terms of the Data Value Chain, it could prove useful to add ‘model and analyze data’ as part of the 4 components of the data life cycle (Figure 2). This is important for the conceptual framework because the modeling aspect of data life cycle informs the type of data to be collected and how it should be handled as well. 
• BOX 2: may also include some coverage in relation to countries with complete lack of data reporting systems (e.g., the Democratic People's Republic of Korea (DPRK), the State of Eritrea, etc.…)
• The introduction of new data related technologies are disruptive by their nature and there is so much unknown, especially going forward. As we go forward, newer and better digital technologies will only accelerate this disruption. Regarding the AI/ML and agricultural data in general, and FSN data, it will be good to think about putting the right data strategy along the data value chains to take advantage of future opportunities and challenges that will surely be presented because of these technologies. It is important to think beyond what is currently possible and prepare our data system to reinvent every aspect of the FSN. It is important to have agile and robust data system and well-trained workforce to handle the impact of AI’s impact in the future. 

b. Are topics under- or over-represented in relation to their importance?

• Although the text contains a note on the ‘Lack of coordination between agencies (Section 3.1.3), more could be done here including some existing examples. For example, the lessons from the U.S. Government’s global food and nutrition security initiative (https://www.feedthefuture.gov/) provides a good example of how multiple stakeholders are brought together to achieve shared objective of sustainably reducing global hunger and malnutrition, also addressing agency and sustainability aspects of FSN. The U.S. Government’s global food and nutrition security initiative as indicated in the website states that the initiative was developed by 12 U.S. Government agencies and departments, with the input of multi-sectoral partners to present an integrated, multi-disciplinary approach to combating the root causes of hunger, malnutrition, and poverty in the target countries around the world. 

• Moreover, the use of women’s empowerment, although briefly mentioned in the draft report, could be elaborated a bit more to include examples and evidence of the positive linkages between women’s empowerment (agency) and FSN indicators.
• On Page 17 of 63 of the report, “EXAMPLE (3): Emergency / conflict situation in which healthy dietary intake is compromised”, the following could be added:
  • Example (5) food consumption and dietary intake level by children (and women)
  • Example (6) women’s empowerment (agency) and intrahousehold food consumption and allocation  
• More could be said about ‘capacity and inequities’ (Section 4.4.4: Insufficient capacity and inequities), especially as one of the growing demands in the wake of such newer and better data related technologies is the manpower needed to make sense of the data. Training and upgrading the human capacity aspects could be strengthened.
• Regarding chapter 5, “INSTITUTIONS AND GOVERNANCE FOR DATA COLLECTION, ANALYSIS, AND USE”, although well outlined, it can be expanded and enriched. This aspect has always been important, but it will become even more so with the increasing introduction of new data related technologies. Specifically, the draft report may need to include a dedicated section or subsection on existing data governance conceptual frameworks (example see Abraham et al. 2019), applicable to FSN, or propose a new one that should help clarify outcomes and expectations.
• Comprehensive and up to date country level estimates on price and income elasticities, and in general supply and demand dynamics as well as comprehensive price information on food consumed by consumers is necessary. This is of utmost importance, including its use for accurate assessments of economic, social, and environmental changes related to FSN.
• Presentation of new and emerging technologies: This can be elaborated with some more relevant applications for FSN, including using machine learning in combination with publicly available data sets such as LSMS data and remote sensing data for poverty predictions and other applications in low-income countries with inherent data quality problems. One important data point that is hard to collect is prices of commodities. The use of emerging technologies such as smart phones in combination with other tools could be leveraged to collect data. The key point is to empower individual consumers to report data points, if appropriate incentives are put in place, directly to central systems where information could be aggregated and harmonized for use.
• Food accessibility: food security has been mostly measured by income, especially in the low- and middle-income countries. Food insecurity, in terms of availability and nutrition intake does occur in the developed nations. It would be a good to have a central database that captures food deserts across countries. This data could be collected from grassroot (local) level.
• Rural-Urban food price dispersion: This is a variable that can provide information about differences in rural and urban dwelling across countries. Lack of infrastructures connecting rural and urban dwellings create differences in food cost and accessibility. This is important for food aid policy towards low-income countries. Since most countries’ port of entry for food aid shipments is in major cities. Lack of infrastructure linking the rural communities can also affect accessibility to these food aid provisions. 
• Food waste and post-harvest loss: Reliable data on food waste (food loss) and post-harvest loss would be beneficial to researchers and policy makers.

c.Are there any redundant facts or statements that could be eliminated from the V0-draft?

• Although timely and relevant, the generic use of terms such as machine learning and AI related technologies seems to be overly used and may need to be refined.
• Finally, regarding the contents in Table 1 Existing initiatives on data for FSN, some additional data sources to consider are:

• USDA ERS - Food Consumption and Nutrient Intakes: ERS provides data on food consumption and nutrient intake by food source and demographic characteristics
• USDA ERS - Food Availability (Per Capita) Data System: Food Availability (Per Capita) Data System-The ERS Food Availability (Per Capita) Data System (FADS) includes three distinct but related data series on food and nutrient availability for consumption: food availability data, loss-adjusted food availability data, and nutrient availability data.
• Data.gov: The home of the U.S. Government’s open data. Here you will find data, tools, and resources to conduct research, develop web and mobile applications, design data visualizations, and more

References:

Abraham, R., Schneider, J., & Vom Brocke, J. (2019). Data governance: A conceptual framework, structured review, and research agenda. International Journal of Information Management, 49, 424-438.