the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 02 Jun 2025
An endogenous modelling framework of dietary behavioural change in the fully coupled human-climate FRIDA v2.1 model
Abstract. Integrated assessment models (IAMs) occupy a central role in understanding and assessing the intricate interlinkages within the human-climate system for informing climate mitigation and adaptation strategies. However, there has been limited work on explicitly representing the internal social system dynamics that underlie human behavioural responses to climate change within IAMs. Instead, behavioural change and demand-side strategies are assessed with external, non-probabilistic narrative-based scenario analyses. In this paper, we introduce an alternative fully endogenous behavioural change modelling framework within the FRIDA v2.1 model, operationalized with the system dynamics method. Applied to the context of dietary behaviour, the framework models behavioural change as a function of perceived accessibility, descriptive norms, and personal norms, constrained by accessibility and past behaviour. By doing so, it captures the complex social-economic-cultural-environmental feedback processes within the human-climate system that dynamically determine per capita food demand and consumption. Our results show that endogenizing human behaviour alters future projections of emission behaviours and thus climatic conditions. Importantly, using an uncertainty approach, our results account for a range of plausible behaviours within the 95 % confidence bounds, which includes scenarios where we observe reversals of sustainable behavioural change in the future. We contribute to the limited work on human behaviour in IAMs, extending the complexity of current representations. Future work will extend this framework to other domains of high-impact behaviours, enhancing the robustness of IAMs for assessing demand-side mitigation.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(3046 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3046 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-2260', Kaia Waxenberg, 09 Jun 2025
General comments
I found this to be a strong paper overall, which significantly furthers research on global integrated assessment modelling. The introduction of a more endogenous food demand model represents a significant improvement over common GDP-based models. The conclusion that current income-driven models may be overestimating future global food demand is a significant, if not expected, finding.
The majority of the text is dedicated to model description and presentation of results, with limited reflection or discussion. There could be more emphasis on the relevance of results presented, particularly in relation to the shortcomings of widely available models. Reasons why the endogenous behavioural model predicts lower future food demand when compared to the income-driven model could be discussed more fully.
Specific grammatical corrections and content suggestions
Abstract
- The abstract should state that the EBM altered the IAM results by decreasing future demand projections, specifically. That is an interesting and important finding
1 Introduction
- Line 33: Do you really mean behavioural responses to climate change here, or it is any climate-relevant behaviour, whether or not this behaviour in is direct response to climate change?
2 Existing models review
- Line 70: affects not affect
- The order of paragraphs 4 and 5 could be swapped. This way, the section would first review existing models and then conclude by describing how these models differ from the presently introduced model. This would improve flow
3 Model description
- Overall, a very clear and comprehensive model description with good justifications of modelling choices and mechanistic thinking
- This section is missing a paragraph providing an overview of FRIDA (ie. it is a global model, GDP is endogenous, what exogenous inputs are included, etc). The aggregation level is mentioned a few times, so it should be explicitly described
- For all equations, I like to see variables defined in a caption even if they are defined in the text, so that equations can be read as separate from the text (like with figure captions). Saves readers from doubling back and searching the text for variable abbreviations
- Line 167: does FRIDA allow stocks of food commodities to be carried over between years or does the market clear?
- Figure 175: The clear visualisation of feedback loops is a nice addition to this schematic diagram. However, the colour codes and labels on the loops are redundant and the schematic is a bit hard to read. Placing some more of the important values inside shapes (like DADD is now) could help with readability – particularly the values in bold now, for example
- Line 375: Are only globally relevant extreme events included? What qualifies an event as event globally relevant, if so?
4 Model Calibration
- Line 463: If the FAO data was used, shouldn’t the EAT Lancet target diet be scaled up to include calories wasted at the household and retail levels?
- Line 512: Is there a reference for all FRIDA parameters included in uncertainly analysis?
5 Simulation results
- Figures in this section are nice and clear
- It would be interesting to discuss more about the weighting factors, ie. How much did climate vs health risks influence diets in the model?
- Line 597: The discussion should include some review of how this rebound could possibly be avoided based on behavioural literature
- Line 601: typo
- Line 619: production dynamics have
6 Conclusions
- Line 639: we close one human – climate feedback. Other behavioural feedbacks remain GDP-dependent in the current model
- Line 640: addressing a current gap (there are certainly other gaps)
- Line 689: How applicable would this the EBM be to national or regional food demand? Would it be possible to calibrate it for the level of aggregation used in other common IAMS?
- Line 707: remove "lastly", as it could imply this paragraph focuses on another limitation of the model rather than an avenue for future work
- Line 709: various energy demands … are modelled
- Add one sentence to the very end reiterating the points in the first paragraph of the conclusion, to leave the reader with a strong takeaway
Citation: https://doi.org/10.5194/egusphere-2025-2260-RC1 -
RC2: 'Comment on egusphere-2025-2260', Anonymous Referee #2, 22 Jul 2025
General Comments:
The model presented in this manuscript represents a significant contribution to efforts to integrate human and climate processes by demonstrating the importance of endogenous feedback in shaping dietary choices. Simulation results demonstrate that including endogenous human behavior generates distinct projections for future food demand relative to projections that rely on GDP alone. These results make a compelling case to account for endogenous behavior through factors like perceived food accessibility and behavioral norms in addition to income when modeling food demand.
Specific Comments:
It is helpful that Figure 2 includes variable acronyms/shorthand (e.g., DADD) in parentheses along with the complete names. I recommend including more of these terms for ease of reference (e.g., STA, PSV, PCR, PCCR).
I found the logic unclear for why consumption declines after a certain level of income. Why is it that higher socioeconomic class consumers are less influenced by the symbolic status of food consumption (line 243)?
I think equation 10 would be more clearly written if it began with ωdn(t) = ωpn(t)
I count 6 climate indices, not 7, in section 3.2.1 lines 378-80. Please clarify.
In section 3.2.1 line 404, the reference to the “SLR Impacts and Adaptation module in FRIDA” could be confusing since it is not illustrated as a module in Figure 1. I suggest clarifying that it is part of the Climate module in FRIDA by using the term “sub-module” or “formulation”, such as: “The SLR Impacts and Adaptation sub-module of the Climate module in FRIDA”
In section 4 line 452-3, the phrase “uncertainty is inherent in the model” could be confusing since the model itself is deterministic, not stochastic. I suggest rephrasing to say “in the model’s parameterization” instead.
I suggest revising the section 5 header to “Simulation results and discussion” (line 519) to better reflect the contents of that section.
Were any runs analyzed individually? (see line 604-5)
In section 5 line 630, suggest clarifying by adding “between the EMB and GDP-driven simulations” to the end of this sentence.
Technical Corrections:
Include consistent use of significant figures in Appendix A. Rounding to the nearest integer fails to show the range of uncertainty for several of the time constants for which the value, min, and max are all set to 1.
Line 114: should be “lifestyle” not “lifestyles”
Line 151: suggest “Additional indirect feedback” instead of “Additionally, more indirect”
Line 156: Figure 1 legend should say “Sub-system” not “Subs-system”
The rightmost reinforcing loop in Figure 2 should say “personal norm” not “person norm”
Line 176: Figure 2 legend should include “labels” after “blue” and “purple”
Line 202: should be “such as price” not “such price”
Line 203: should be “price” not “prices” in the first instance
Line 213: suggest rephrase, changing “that” to “and is set to” so that it is clear that the sensitivity term is less than one, not relative scarcity: “βs is the sensitivity of demand to relative scarcity and is set to less than 1”
Line 218: suggest adding “the” to read “the animal products balance”
Line 224: suggest including beta in parentheses, as (βi < 1), to make it clear that the constraint is not applied directly to changes in income.
Line 232: add “demand” after “animal products”, to read “animal products demand”
Line 251: suggest “decreases at a decreasing rate” rather than “decreases decreasingly”
Line 308: should be “require” not “requires”
Line 334: should be “norms” not “norm”
Line 338: should be “emphasized” not “emphasize”
Line 351: the citation (Mathematical builtins, 2025) missing from the reference list
Line 374: suggest “(listed below)” instead of “(see below for list)” since the indices are named in text just two sentences later; the reference to a list could imply a different format.
Line 375: specify “It” - if it is the Climate module, then start the sentence “The Climate module estimates”; or if “it” is the climate indices, then start the sentence “The climate indices estimate”
Line 377: should be “reproduce” not “reproduces”
Line 388: should be “diminish” not “diminishes”
Line 408: should be “perceptions of risk or abnormality are sensed” not “is sensed”
Line 430: should be “times” not “time” in first instance, as in “smoothed three times”
Line 527: suggest “demands” not “demand”, as in “various average daily food demands”
Line 570: should be “catch” not “catches”
Line 591: should be “reinforce” not “reinforces”
Line 601: typo in “Therefore”
Line 704: should be “demands” not “demand” and “are” not “is”, as in “the various energy demands in the present version are modelled primarily as functions”
Citation: https://doi.org/10.5194/egusphere-2025-2260-RC2 - AC1: 'Response to reviewers', Jefferson K. Rajah, 06 Aug 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-2260', Kaia Waxenberg, 09 Jun 2025
General comments
I found this to be a strong paper overall, which significantly furthers research on global integrated assessment modelling. The introduction of a more endogenous food demand model represents a significant improvement over common GDP-based models. The conclusion that current income-driven models may be overestimating future global food demand is a significant, if not expected, finding.
The majority of the text is dedicated to model description and presentation of results, with limited reflection or discussion. There could be more emphasis on the relevance of results presented, particularly in relation to the shortcomings of widely available models. Reasons why the endogenous behavioural model predicts lower future food demand when compared to the income-driven model could be discussed more fully.
Specific grammatical corrections and content suggestions
Abstract
- The abstract should state that the EBM altered the IAM results by decreasing future demand projections, specifically. That is an interesting and important finding
1 Introduction
- Line 33: Do you really mean behavioural responses to climate change here, or it is any climate-relevant behaviour, whether or not this behaviour in is direct response to climate change?
2 Existing models review
- Line 70: affects not affect
- The order of paragraphs 4 and 5 could be swapped. This way, the section would first review existing models and then conclude by describing how these models differ from the presently introduced model. This would improve flow
3 Model description
- Overall, a very clear and comprehensive model description with good justifications of modelling choices and mechanistic thinking
- This section is missing a paragraph providing an overview of FRIDA (ie. it is a global model, GDP is endogenous, what exogenous inputs are included, etc). The aggregation level is mentioned a few times, so it should be explicitly described
- For all equations, I like to see variables defined in a caption even if they are defined in the text, so that equations can be read as separate from the text (like with figure captions). Saves readers from doubling back and searching the text for variable abbreviations
- Line 167: does FRIDA allow stocks of food commodities to be carried over between years or does the market clear?
- Figure 175: The clear visualisation of feedback loops is a nice addition to this schematic diagram. However, the colour codes and labels on the loops are redundant and the schematic is a bit hard to read. Placing some more of the important values inside shapes (like DADD is now) could help with readability – particularly the values in bold now, for example
- Line 375: Are only globally relevant extreme events included? What qualifies an event as event globally relevant, if so?
4 Model Calibration
- Line 463: If the FAO data was used, shouldn’t the EAT Lancet target diet be scaled up to include calories wasted at the household and retail levels?
- Line 512: Is there a reference for all FRIDA parameters included in uncertainly analysis?
5 Simulation results
- Figures in this section are nice and clear
- It would be interesting to discuss more about the weighting factors, ie. How much did climate vs health risks influence diets in the model?
- Line 597: The discussion should include some review of how this rebound could possibly be avoided based on behavioural literature
- Line 601: typo
- Line 619: production dynamics have
6 Conclusions
- Line 639: we close one human – climate feedback. Other behavioural feedbacks remain GDP-dependent in the current model
- Line 640: addressing a current gap (there are certainly other gaps)
- Line 689: How applicable would this the EBM be to national or regional food demand? Would it be possible to calibrate it for the level of aggregation used in other common IAMS?
- Line 707: remove "lastly", as it could imply this paragraph focuses on another limitation of the model rather than an avenue for future work
- Line 709: various energy demands … are modelled
- Add one sentence to the very end reiterating the points in the first paragraph of the conclusion, to leave the reader with a strong takeaway
Citation: https://doi.org/10.5194/egusphere-2025-2260-RC1 -
RC2: 'Comment on egusphere-2025-2260', Anonymous Referee #2, 22 Jul 2025
General Comments:
The model presented in this manuscript represents a significant contribution to efforts to integrate human and climate processes by demonstrating the importance of endogenous feedback in shaping dietary choices. Simulation results demonstrate that including endogenous human behavior generates distinct projections for future food demand relative to projections that rely on GDP alone. These results make a compelling case to account for endogenous behavior through factors like perceived food accessibility and behavioral norms in addition to income when modeling food demand.
Specific Comments:
It is helpful that Figure 2 includes variable acronyms/shorthand (e.g., DADD) in parentheses along with the complete names. I recommend including more of these terms for ease of reference (e.g., STA, PSV, PCR, PCCR).
I found the logic unclear for why consumption declines after a certain level of income. Why is it that higher socioeconomic class consumers are less influenced by the symbolic status of food consumption (line 243)?
I think equation 10 would be more clearly written if it began with ωdn(t) = ωpn(t)
I count 6 climate indices, not 7, in section 3.2.1 lines 378-80. Please clarify.
In section 3.2.1 line 404, the reference to the “SLR Impacts and Adaptation module in FRIDA” could be confusing since it is not illustrated as a module in Figure 1. I suggest clarifying that it is part of the Climate module in FRIDA by using the term “sub-module” or “formulation”, such as: “The SLR Impacts and Adaptation sub-module of the Climate module in FRIDA”
In section 4 line 452-3, the phrase “uncertainty is inherent in the model” could be confusing since the model itself is deterministic, not stochastic. I suggest rephrasing to say “in the model’s parameterization” instead.
I suggest revising the section 5 header to “Simulation results and discussion” (line 519) to better reflect the contents of that section.
Were any runs analyzed individually? (see line 604-5)
In section 5 line 630, suggest clarifying by adding “between the EMB and GDP-driven simulations” to the end of this sentence.
Technical Corrections:
Include consistent use of significant figures in Appendix A. Rounding to the nearest integer fails to show the range of uncertainty for several of the time constants for which the value, min, and max are all set to 1.
Line 114: should be “lifestyle” not “lifestyles”
Line 151: suggest “Additional indirect feedback” instead of “Additionally, more indirect”
Line 156: Figure 1 legend should say “Sub-system” not “Subs-system”
The rightmost reinforcing loop in Figure 2 should say “personal norm” not “person norm”
Line 176: Figure 2 legend should include “labels” after “blue” and “purple”
Line 202: should be “such as price” not “such price”
Line 203: should be “price” not “prices” in the first instance
Line 213: suggest rephrase, changing “that” to “and is set to” so that it is clear that the sensitivity term is less than one, not relative scarcity: “βs is the sensitivity of demand to relative scarcity and is set to less than 1”
Line 218: suggest adding “the” to read “the animal products balance”
Line 224: suggest including beta in parentheses, as (βi < 1), to make it clear that the constraint is not applied directly to changes in income.
Line 232: add “demand” after “animal products”, to read “animal products demand”
Line 251: suggest “decreases at a decreasing rate” rather than “decreases decreasingly”
Line 308: should be “require” not “requires”
Line 334: should be “norms” not “norm”
Line 338: should be “emphasized” not “emphasize”
Line 351: the citation (Mathematical builtins, 2025) missing from the reference list
Line 374: suggest “(listed below)” instead of “(see below for list)” since the indices are named in text just two sentences later; the reference to a list could imply a different format.
Line 375: specify “It” - if it is the Climate module, then start the sentence “The Climate module estimates”; or if “it” is the climate indices, then start the sentence “The climate indices estimate”
Line 377: should be “reproduce” not “reproduces”
Line 388: should be “diminish” not “diminishes”
Line 408: should be “perceptions of risk or abnormality are sensed” not “is sensed”
Line 430: should be “times” not “time” in first instance, as in “smoothed three times”
Line 527: suggest “demands” not “demand”, as in “various average daily food demands”
Line 570: should be “catch” not “catches”
Line 591: should be “reinforce” not “reinforces”
Line 601: typo in “Therefore”
Line 704: should be “demands” not “demand” and “are” not “is”, as in “the various energy demands in the present version are modelled primarily as functions”
Citation: https://doi.org/10.5194/egusphere-2025-2260-RC2 - AC1: 'Response to reviewers', Jefferson K. Rajah, 06 Aug 2025
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 952 | 41 | 14 | 1,007 | 18 | 31 |
- HTML: 952
- PDF: 41
- XML: 14
- Total: 1,007
- BibTeX: 18
- EndNote: 31
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Jefferson K. Rajah
Benjamin Blanz
Birgit Kopainsky
William Schoenberg
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3046 KB) - Metadata XML