Can participatory systems modeling of food value chains improve food and nutrition security under climate change? 5th Global Food Security Conference Towards equitable, sustainable and resilient food systems 9th-12th April 2024 | KU Leuven, Belgium CF Nicholson, University of Wisconsin-Madison; D Mason-D’Croz, Cornell University; PK Thornton, Clim-Eat; EL Phillips, University of Wisconsin-Madison; L Cramer, International Livestock Research Institute; JL Garrett, Alliance CIAT-Bioversity; MI Gómez, Cornell University; M Herrero, Cornell University; AD Jones, University of Michigan; B Kopainsky, University of Bergen; E Monterrosa, Global Alliance for Improved Nutrition; D Parsons, Swedish Agricultural University, Umeå; R Zougmoré, Alliance CIAT-Bioversity; M Spiker, University of Washington; EC Stephens, Agriculture and Agri-Food Canada; M Wattiaux, University of Wisconsin-Madison; S Yanni, Agriculture and Agri-Food Canada Agenda • What motivates interest in PSM to address FNS under climate change? • What is PSM? • Benefits for addressing FNS under climate change • Challenges for addressing FNS under climate change • Proposed next steps A proposal, not the finished product The best way to have a good idea is to have a lot of ideas. Linus Pauling Participatory Systems Modeling to Improve Food and Nutrition Security 7 Figure 1. Potential Pathways for Impact of Climate on Undernutrition (Source: Myers et al. 2017) Fanzo et al. (2017) develop a conceptual diagram to address another potential limitation of many of the conceptual models linking climate change and FNS: most components of the food value chain are omitted or highly aggregated. Using an explicit “value chain” approach, this diagram (Figure 2) highlights the factors that could reduce “nutritional value” at multiple stages and suggest potential interventions that could maintain that value through to the stage of consumption and utilization. This diagram describes a linear flow process beginning with input supply and ending with consumption—again implying limited consideration of feedback processes or dynamics. It highlights the impacts on nutrition of multiple post-farm and processing activities, which have received less attention in the literature linking climate change to FNS. Although not clearly linked to climate change, this diagram provides a useful complement to others because it recognizes that many other activities (and actors) are involved in the production and distribution of food, and implicitly that growing populations in LMIC are not food producers or are net buyers of food rather than subsistence producers (Stephens and Barrett, 2011). Nutrition and climate change – Current state of play: Scoping review 13 PU38CH13-Myers ARI 17 March 2017 9:2 • Altered primary production, poleward shifts of species, smaller mean fish size • Coral reef degradation and shellfish declines • Altered fish nutrient content • Terrestrial temperature increase • Rainfall variability • Extreme weather events • Increased atmospheric CO2, ozone levels • Animal heat stress • Changes in forage species composition and productivity • Abiotic effects on crop yield • Greater pests, pathogenT ��� weFd�QSFTTVSF • Pollinator declines • Lower human labor capacity • Poorer crop nutrient content • Greater postharvest losses Greenhouse gas emissions • Ocean temperature increase • Ocean HCO3 – increase (acidification) • Higher frequency of conflict • Lower GDP growth • Price increases • Price volatility Altered crop yields and� reduced nutrient content Lower purchasing power of nutritionally vulnerable populations Increase in diarrheal diseases and enteric infections Altered livestock productivity Altered fish catch and� nutrient contenU Altered global nutrient supply Altered nutritional status Human activity Proximate biological consequences Climate and atmospheric shifts Impact on human socioeconomic systems Nutritional and health consequences Increased exposure to enteric pathogens [5.4] Pathways for impacts of climate change on food systems, food security, and undernutrition Figure 2 Anthropogenic greenhouse gas emissions are likely to impact human nutritional status through a cascading set of biophysical and socioeconomic changes. Details for the mechanisms and impacts of each cause may be found in the text sections provided in brackets. to as representative concentration pathway (RCP) 4.5, atmospheric CO2 concentrations would continue their rise from a 280-ppm preindustrial baseline, beyond the present 400-ppm levels, and on to values of 540 ppm by 2100 (123). Climate simulations indicate a further land warming of 1.9–4.0◦C (3.4–7.2◦F) [90% confidence interval (CI)] (37, 75, 115). Under the higher emission scenario, known as RCP8.5, CO2 concentrations would reach 940 ppm by 2100 and result in land warming of 4.0–6.8◦C (7.2–12.2◦F) (75, 115). Even a moderate emissions scenario is expected to result in average summer temperatures that exceed the most extreme temperatures currently experienced in many areas of the world (11). The availability of water resources for agriculture will be influenced by climate change in a multitude of ways, including shifting precipitation patterns, loss of glaciers and earlier seasonal snow melt, and intrusion of saltwater into coastal aquifers (78). Climate model projections gen- erally indicate less precipitation in currently arid and semiarid regions and greater precipitation in the polar latitudes (37). Rainfall events are expected to become more intense, likely increasing runoff and flooding (37). 262 Myers et al. A nn u. R ev . P ub lic H ea lth 2 01 7. 38 :2 59 -2 77 . D ow nl oa de d fro m w w w .a nn ua lre vi ew s.o rg A cc es s p ro vi de d by 2 a0 0: 23 c4 :3 d3 f:7 c0 1: c1 71 :2 29 0: 6d 9e :9 ef 8 on 0 1/ 14 /2 1. S ee c op yr ig ht fo r a pp ro ve d us e. Figure 1 Pathways for impacts of climate change on food systems, food security and undernutrition behaviours (Tirado et al., 2010) – are more likely to occur in countries where food monitoring and surveillance systems are less robust. Thus, poor detection of environmental and chemical contamination further increases the risk to public health and nutrition through the acute and chronic exposure to contaminants. Access to food may reduce through climate change impacts increasing the price of food. Modelling suggests that inflation-adjusted prices of wheat, rice and maize could increase 31%–106% by 2050 (Nelson et al., 2010). Food prices are extremely sensitive to shocks on both the supply and the demand side, as demonstrated recently by the COVID-19 pandemic (Heady et al., 2020). Fluctuations negatively affect household food availability, access and diet quality access to social and health services and the quality of infant and young child care (Fanzo, 2018). Food price rises between 2006 and 2008 were estimated to have resulted in at least 50 million more people becoming hungry in 2008 (FAO, 2008). gendered nature of natural resource management coupled with an unequal access to rights in certain countries leaves many women particularly vulnerable to the effects of biodiversity loss (Women’s Environment and Development Organization [WEDO], 2010). For more information on the links between gender, climate change and nutrition, see Box 3. Increased heat and water stress is also expected to increase the incidence of foodborne pathogens and mycotoxins during food storage, processing and transportation (Battilani et al., 2016). There is a general lack of governance and policies around food safety, while the risks of food safety and increasing food waste in rural communities are especially acute as retail infrastructure and cold storage are often basic and access to water may be restricted (Sheahan and Barrett, 2017). In addition, the various climate-related changes impacting food safety – including human, animal and vector behaviours, and changing pathogen, organism and pest survival, growth and transmission Source: Myers et al., 2017 What motivates interest in PSM? • Growing threat of climate change to FNS • Actions to both mitigate and adapt to climate change • The pathways linking climate change to FNS outcomes are complex and dynamic • Substantive gaps exist in understanding of impacts and assessment of interventions, especially at the scale of value chains Participatory Systems Modeling to Improve Food and Nutrition Security 8 Figure 2. Value Chain Conceptualization of Nutrition (Source: Fanzo et al., 2017) 2.2 Ex Post Studies of Climate Impact on FNS Many empirical studies have examined the linkages between specific climate variables (or climate variability) and FNS outcomes. WHO (2019) provided a summary of the likely effects of “environmental change” on key risk factors for undernutrition based on prior studies and hypothesized impact pathways (Figure 3). Previous studies frequently use statistical methods to determine associations between climate-related variables and FNS metrics over various time and geographic scales (e.g., Cooper et al., 2019; Belesova et al., 2019; Bahru et al. 2019). Many studies emphasize child nutritional outcomes such as stunting or wasting. Randell et al. (2021) evaluated HFIAS values for households in Nepal after the 2015 earthquake, one of the few studies to consider the potential impacts of transportation infrastructure on food security. They found that the interaction between post-earthquake precipitation and HFIAS was affected by the extent of shaking (and thus hypothesized damage to transportation infrastructure). Dasgupta and Robinson (2022) found a relationship between increased temperature and higher FIES scores across 83 countries from 2014 to 2019 controlling for precipitation and the Human Development Index (HDI). Although often there is statistically significant impact, the broader significance for human health is often not clearly described. Moreover, some effects appear to be counterintuitive: Cooper et al. found “diverging impacts of excessive of precipitation on anthropometry and household hunger in Ghana.” Belesova et al. (2019) also noted that attributing causality from these analyses can be challenging, noting that the extent to which child undernutrition is attributable to drought “has not been clearly quantified and may be context specific.” Although these studies provide useful quantitative evidence about the linkages between climate change and FNS, many such studies are limited by the degree to which causality is clear based on the data and methods, the extent to which the causal pathways are understood, and their usefulness for identification of intervention priorities. For example, Cooper et al. noted 5 Figure 1.1 Climate change and nutrition entry and exit points in the food value chain Source: Adapted with permission from Fanzo et al. (2017). Net nutrition along the food value chain Maximizing nutrition “entering” the value chain Minimizing nutrition “exiting” the value chain Source: Fanzo et al. 2017 What motivates interest in PSM? • A systems approach involving stakeholders and focused on value chains can fill critical knowledge gaps • Consider interconnected and heterogenous components of food systems across scales • Account for complex interactions over time • Integrate existing quantitative and qualitative knowledge • Identify priority actions based on a more holistic understanding and engagement of stakeholders What is PSM? • Systems methodological approach to identify and assess priority that combines: • Facilitated discussions with stakeholders – Group Model Building (GMB) – Conceptual model of causes and consequences – Focus on key outcome Participatory Systems Modeling to Improve Food and Nutrition Security 14 Figure 1. Steps in the Group Model Building (GMB) process. Source: Vennix, 1996, p. 120 A practical approach for building a CLD in a session is to return focus to the problem variable in the middle of a whiteboard or computer screen after gathering potential variables. Then, participants are asked to suggest a variable from the collected list is a cause for changes in the problem variable. This suggested variable is included in the drawing of the model to visualize what is meant (Step 2 in Figure 1), and a check is made to see if everyone agrees with this. If someone disagrees with the suggestion, it can be discussed to determine group consensus about what the relationship should be. If the discussion continues too long, this variable can be temporarily 'parked' on so that the process can continue with other variables. Typically, the participants will note not only variables that have a direct relationship with the problem variable, but logical chains of reasoning (via intervening variables, as shown in Step 2). This process should also identify the polarities (positive or negative) of the relationships. After spending some time considering causes, the GMB exercise proceeds to the consequences of changes in the problem variable (Step 3 in Figure 1). This is conducted similarly to the process for Step 2. A final step in the process is to identify where consequences affect causes (Step 4 in Figure 1). Step 4 begins the process of identifying feedback loops that comprise a series of causal linkages forming a loop. (One of the assumptions underlying both GMB and SDM is that feedback loops are a very important part of the structure of a social system.) When a feedback chain becomes closed, it is typical to check with the entire group to see if that chain as whole is correct. If so, the group proceeds to identify the polarity of the loop—is the overall effect of all of the linkages positive (reinforcing or escalating change) or negative (balancing or offsetting change). When drawing relations with a group, the rule that applies to all “convergent activities” with groups also applies here. When building a CLD, changes on the board (screen) need to be 65 Figure 7. steps in building a causal loop diagram (Vennix, 1996: 120) One of the assumptions underlying system dynamics is that feedback loops are the most important part of the structure of a social system. Adding relations as done in the previous phase will eventually close feedback loops. There are two kinds of feedback loops: positive loops that lead to escalating behaviour, and negative loops that lead to balancing behaviour. Figure 6 was developed in a project for the Dutch Ministry of Justice. The lower part shows required detention capacity, which is calculated by multiplying the number of prison sentences with the average time served. If required detention capacity is greater than available detention capacity, a shortage of capacity results. In 2000, the Dutch Prison Administration initiated a policy that made prisoners who are serving time for infractions and have completed 90% of their sentence eligible for early release. In 2003, the strictness of norms for early release was reduced and prisoners who had completed 70% of their sentence were eligible for early release. This increased the potential number of early releases and actual releases. By reducing the average time served by a reduction time, the early release policy frees up capacity for new prisoners. A judge participating in the modelling project recalled that he became aware of the policy after he recognized a suspect as someone recently convicted and imprisoned for an earlier crime. He then became concerned that the suspect’s sentence passed for the earlier crime was not served to completion. He foresaw that when judges would perceive an increase in the difference between duration of the sentence and time served (upper part of Figure 6) they would compensate by increasing the duration of sentences. Thus, there are three negative or balancing feedback loops in the causal loop diagram. In the loop at the bottom of the figure, an increase in required detention capacity leads to more prisoners released early. This reduces the average time served and thereby reduces required detention capacity. In this way, an initial increase is compensated by a decrease. Step 1 Identification problem Step 2 Adding causes Step 3 Adding consequences Step 4 Identifying feedback loops X X X Causes Problem variable Consequences X What is PSM? • Systems methodological approach to identify and assess priority that combines: • Facilitated discussions with stakeholders • Development of quantitative inter- temporal systems simulation models – System Dynamics Modeling (SDM) – Quantitative model development and evaluation Source: Nicholson et al 2021. Participatory Systems Modeling to Improve Food and Nutrition Security 21 must be considered?; c) Time horizon: How far in the future should we consider? How far back in the past lie the roots of the problem?; d) Dynamic problem definition (specification of reference modes): What is the historical behavior of the key concepts and variables? What might their behavior be in the future? This aligns with the initial stages of GMB to identify the reference mode (or problem variable); 2) Formulation of Dynamic Hypothesis. This involves a) Initial hypothesis generation: What are the current theories of the problematic behavior?; b) Endogenous focus: Formulate a dynamic hypothesis that explains the dynamics as endogenous consequences of the feedback structure, c) Mapping: Develop maps of causal structure based on initial hypotheses, key variables, reference modes, and other available data, using tools such as model boundary diagrams, causal loop diagrams, stock and flow maps, and other facilitation tools. The involvement of stakeholders in model development often focuses on this step as a component of GMB. 3) Formulation of a simulation model. This includes a) specification of structure and decision rules; b) estimation of parameters, behavioral relationships, and initial conditions, and c) tests for consistency with the purpose and boundary; Figure 3. The iterative steps of the System Dynamics modeling process 4) Model evaluation. Sometimes referred to as model testing, this involves the following at a minimum: a) Comparison to reference modes: Does the model reproduce the problem behavior adequately for its purpose? Is it robust under extreme conditions? Does the model behave realistically when stressed by extreme conditions?; b) Sensitivity: How does the model behave given uncertainty in parameters, initial conditions, model boundary, and aggregation?; 1. Problem Articulation (Boundary Selection) 3. Formulation4. Testing 5. Policy Formulation & Evaluation 2. Dynamic Hypothesis Identify a “reference mode” behavior Initial conceptualization of structure causing behavior Mathematical specification of stock-flow- feedback structure Formal model evaluation (emphasis on behavioral mode) Model analysis to inform decision making Adapted from Sterman (2000). What is PSM? Example with Value Chain Focus Analysing the potential impacts of three interventions on fruit and vegetable consumption in urban Kenya using participatory systems modelling Charles F Nicholson1,* and Eva Monterrosa2 1Department of Agricultural and Applied Economics, University of Wisconsin—Madison, 442 Animal Science Building, 1675 Observatory Drive, Madison, WI 53706, USA: 2Global Alliance for Improved Nutrition, Geneva, Switzerland Submitted 28 March 2023: Final revision received 3 August 2023: Accepted 11 September 2023 Abstract Objective: This study uses participatory modelling with stakeholders to assess the potential impacts of three interventions intended to increase fruit and vegetable (F&V) consumption in urban Kenya. Design: A participatory process using Group Model Building (GMB) developed a conceptual model of the determinants of vegetable consumption. A subsequent quantitative System Dynamics model using data from primary and secondary sources simulated vegetable consumption from 2020 to 2024 under three proposed interventions suggested by stakeholders: increasing consumer awareness, reducing post-harvest losses and increasing farm yields. Model analyses assumed mean parameter values and assessed uncertainty using 200 simulations with randomised parameter values. Setting: The research was implemented in Nairobi, Kenya with simulation analyses of mean per capita consumption in this location. Participants:Workshops convened diverse F&V value chain stakeholders (farmers, government officials, NGO staff and technical experts) to develop the conceptual model, data inputs and intervention scenarios. Results: Increasing consumer awareness was simulated to increase vegetable consumption by relatively modest amounts by 2024 (5 g/person/d from a base of 131 g/person/d) under mean assumed value of value chain response parameters. Reducing perishability was simulated to reduce consumption due to the higher costs required to reduce losses. Increasing farm yields was simulated to have the largest impact on consumption at assumed parameter values (about 40 g/person/d) but would have a negative impact on farm profits, which could undermine efforts to implement this intervention. Conclusions: The combination of GMB and simulation modelling informed intervention priorities for an important public health nutrition issue. Keywords Nutrition Group Model Building Systems modelling Fruit and vegetable consumption Kenya Fruit and vegetable (F&V) consumption is considered an important component of a healthy diet with numerous documented health benefits(1,2). In urban Kenya, a high proportion of the population consumes F&V, but the quantities are below the amount recommended by the WHO of 400 g/d for all socio-economic groups reported in the Global Dietary Database(3) (Fig. 1). Consumption of both fruits and vegetables in Kenya is also considerably below the ‘optimal’ levels proposed of 300 and 400 g/d(2), respectively. Moreover, there has been essentially no change in per capita consumption of fruits and vegetables on average in Kenya during the years 2000 to 2015 based on data from the Global Dietary Database, despite a threefold increase in GDP per capita during this period(4). Low and stagnant levels of F&V consumption in Kenya during 2000–2015 provide the motivation for an ex ante evaluation of interventions that could increase consumption to WHO recommended levels. Previous literature has identified diverse factors likely to affect F&V consumption both more generally and specifically in the Kenyan context(5,6). These factors typically fall into one of three general categories. Availability comprises Public Health Nutrition: page 1 of 12 doi:10.1017/S1368980023002033 *Corresponding author: Email cfnicholson@wisc.edu ©TheAuthor(s), 2023. Published by CambridgeUniversity Press on behalf of TheNutrition Society. This is anOpenAccess article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited. 3 9 08 8:2 / 310 87 71 / .6/: 021 7 1: :1 Potential Benefits of PSM Category Key Benefits Use of system dynamics models Facilitates application of known methods for model development (including ‘systems archetypes’) and model evaluation Provides examples of value-chain models to streamline development in new contexts Focus on value chains Fills knowledge gaps between farm production and meso-global-scale analyses Facilitates incorporation of FNS metrics other than food availability Allows assessment of intervention effectiveness and priorities Links to knowledge base from Nutrition-Sensitive Value Chains Other benefits May be lower cost, more timely and more flexible than large-scale global models Link to existing frameworks for foresight analyses (Theory of Change, Program Impact Pathways) used in nutrition intervention design and evaluation Integrates components of the “Three-Thirds” approach to facilitating action that comprises evidence, engagement, and outreach Potential Challenges of PSM Category Key Benefits How PSM May Address Challenges Data Data availability and quality often limited PSM has flexibility to use quantitative, qualitative and distributed system knowledge Working with stakeholders Developing effective working relationships with all relevant stakeholders can be challenging and time- consuming Benefits of PSM likely outweigh the costs based on previous implementations Food system complexity The complexity of food value chains can make quantitative modeling difficult PSM methods are designed to deal with systems complexity with focus on key outcomes Contrast with existing methods for analysis of FNS Observational and experimental methods are more common for the evaluation of nutrition interventions PSM complements study designs by including dynamic feedback effects Can PSM drive change and who benefits? Impact pathways from activities (modeling) to outcomes and impact are long and complex PSM enhances motivation for action and identifies tradeoffs among stakeholders Next Steps • Determine the best domain of applicability—what questions can PSM answer effectively and in what contexts? • Development of case studies using PSM LMIC settings that link FNS outcomes and climate change scenarios • Synthesize best practices for PSM, including: – Intervention priorities – Project design and implementation – Monitoring and evaluation Thank You Contact Information: C. F. Nicholson Email: cfn1@cornell.edu Available documents: White paper on PSM, Climate Change and FNS Consensus Statement from Workshop held September 2022 mailto:cfn1@cornell.edu