Subscribe

Back

Who we are

With research staff from more than 70 countries, and offices across the globe, IFPRI provides research-based policy solutions to sustainably reduce poverty and end hunger and malnutrition in developing countries.

researcher spotlight

Kinya Kaibung’a

Kinya Kaibung’a is a Research Officer with the Development Strategies and Governance Unit, based in Nairobi, Kenya. She has a keen interest in leveraging machine learning, AI, and other cutting-edge technologies to boost climate resilience and food security in smart agriculture systems.

Back

What we do

Since 1975, IFPRI’s research has been informing policies and development programs to improve food security, nutrition, and livelihoods around the world.

Research topics

Projects and impact

Food Security Portal

Modeling tools

What’s new

Where we work

Back

Where we work

IFPRI currently has more than 480 employees working in over 70 countries with a wide range of local, national, and international partners.

IFPRI Regional Programs

IFPRI Country and Regional Offices

Research by country and region

Subscribe

experts

Foresight and Policy Modeling

Askar Mukashov

Research Fellow

Follow

Askar

at:

ORCiD

Bio

Askar Mukashov is a Research Fellow in IFPRI’s Foresight and Policy Modeling Unit. He is a development economist, and his research focuses on economic uncertainty and development strategies, employing computable general equilibrium (CGE) models and various statistical methods, including machine learning. His past work includes evaluating the economywide impacts of diverse policies and economic shocks in several developing countries, as well as making methodological contributions to modeling of uncertainty. Before joining IFPRI, Askar worked at the Kiel Institute for the World Economy, Germany. and the Central Bank of the Kyrgyz Republic. He received his doctoral degree in Quantitative Economics from the University of Kiel, Germany.

Publications

RePEc Author Profile

CV

Publications

Resource thumbnail

Working Paper

An agentic AI assistant for country-level economic modeling: Methods, data, and expert evaluation

2026Mukashov, Askar; Kim, Soonho; Fang, Peixun; Diao, Xinshen; Thurlow, James; Proctor, Joshua; Rennison, Alan
Details

An agentic AI assistant for country-level economic modeling: Methods, data, and expert evaluation

The demand for high-quality, rapid economic analysis to navigate complex issues faced by many low- and middle-income countries has led to the development of detailed structural simulation models, such as Computable General Equilibrium (CGE) models. Policy analysis with such models requires deep knowledge of their structure and applicability to the policy issues at hand. Policymakers in these settings often lack access to the expertise required for articulating, analyzing, and interpreting the relevant causal chains captured by the models. Attempting to circumvent these barriers by submitting complex economic questions directly to off-the-shelf large language models (LLMs) introduces severe analytical risks, including hallucinations and insufficient expert guidance. To resolve this limitation, we developed and empirically evaluated an agentic AI assistant called RIAPA-AI that integrates LLMs with a CGE model. We evaluated the performance of RIAPA-AI against expert human CGE modelers and a general-purpose plain LLM baseline across samples of complex economic scenarios, utilizing an independent panel of senior economists to grade the outputs. Our statistical analysis reveals no statistically significant difference in analytical accuracy between RIAPA-AI and human experts, while the AI accelerates reproducible policy analysis from weeks to minutes. Furthermore, by operating without manual processing limits, RIAPA-AI eliminates the 6.7% error rate observed among human modelers. Conversely, the general-purpose plain LLM exhibits profound failure rates, failing to achieve policy-ready scores in over 60% of depth evaluations. Without an underlying CGE model acting as a bounding force to reflect economic structural constraints, the general-purpose plain LLM defaults to linear economic assumptions and inserts unmodeled socio-political narratives. Crucially, by explicitly restricting the AI’s narrative interpretation solely to deterministic numerical outputs, RIAPA-AI mitigates the risk of unverified assumptions and logic hallucinations. We conclude that by deploying an agentic AI assistant that layers a generative AI over a formal CGE model, RIAPA-AI successfully delivers sensible, rigorous, and rapid policy analysis.

Year published

2026

Authors

Mukashov, Askar; Kim, Soonho; Fang, Peixun; Diao, Xinshen; Thurlow, James; Proctor, Joshua; Rennison, Alan

Citation

Mukashov, Askar; Kim, Soonho; Diao, Xinshen; Thurlow, James; Proctor, Joshua; and Rennison, Alan. 2026. An agentic AI assistant for country-level economic modeling: Methods, data, and expert evaluation. IFPRI Discussion Paper 2424. Washington, DC: International Food Policy Research Institute. https://hdl.handle.net/10568/183521

Keywords

Artificial Intelligence; Machine Learning; Modelling; Computable General Equilibrium Models; Large Language Models; Agent-based Models; Econometric Models

Language

English

Access/Licence

Open Access

Source

Source record

Record type

Working Paper

Resource thumbnail

Conference Paper

Key agricultural supply risks for global poverty and undernourishment: A systematic risk profiling approach

2026Mukashov, Askar; Thierfelder, Karen; Robinson, Sherman; Thurlow, James
Details

Key agricultural supply risks for global poverty and undernourishment: A systematic risk profiling approach

The intersection of agricultural production shocks—ranging from climate-induced yield failures to conflicts and policy responses—can trigger systemic volatility similar to the 2008 agrifood crisis. Standard Computable General Equilibrium (CGE) assessments typically rely on discrete deterministic counterfactuals that fail to capture the complex, correlated nature of real-world risk factors. This paper addresses a critical gap: How does the complex landscape of agricultural shocks shape global poverty and undernourishment? We propose a framework to move from isolated scenario analysis to a probabilistic “risk map” of global poverty using the Systematic Risk Profiling (SRP) method (Mukashov et al., 2026). Our approach relies on three pillars. Pillar 1: Monte Carlo Simulations. Utilizing historical data, we generate 10,000 large-scale Monte Carlo simulation scenarios of global agricultural production shocks across key global agricultural producers and vulnerable Global South regions. Pillar 2: Microsimulation Integration. We link the global CGE model with continuous household poverty and undernourishment microsimulation modules to dynamically update regional estimates. Pillar 3: Machine Learning for Risk Attribution. We employ Random Forest regression (Permutation Importance) to estimate the relative importance of sampled agricultural shocks on household welfare outcomes worldwide.

Year published

2026

Authors

Mukashov, Askar; Thierfelder, Karen; Robinson, Sherman; Thurlow, James

Citation

Mukashov, Askar; Thierfelder, Karen; Robinson, Sherman; and Thurlow, James. 2026. Key agricultural supply risks for global poverty and undernourishment: A systematic risk profiling approach. Presented during the 29th Annual Conference on Global Economic Analysis in Kyoto, Japan June 2026. https://www.gtap.agecon.purdue.edu/resources/res_display.asp?RecordID=7902

Keywords

Modelling; Supply; Poverty; Undernourishment; Computable General Equilibrium Models; Food Security; Risk

Language

English

Access/Licence

Open Access

Source

Source record

Project

Foresight

Record type

Conference Paper

Resource thumbnail

Blog Post

Beyond the harvest: Uncovering the hidden risks driving poverty and hunger in developing economies

2026Mukashov, Askar; Jones, Eleanor
Details

Beyond the harvest: Uncovering the hidden risks driving poverty and hunger in developing economies

Key takeaways • A new systematic risk profiling (SRP) tool analyzes the varied triggers of economic shocks in developing countries using economy-wide models, historical data, and machine learning. • SRP was used to simulate thousands of shock scenarios in 11 countries , showing that household?level welfare is often shaped more by global price and capital?flow shocks than by domestic agricultural crises. • By identifying country?specific risk profiles, SRP enables targeted policymaking , helping governments tailor responses to build resilience to shocks with local or global origins.

Year published

2026

Authors

Mukashov, Askar; Jones, Eleanor

Citation

Mukashov, Askar; Jones, Eleanor. 2026. Beyond the harvest: Uncovering the hidden risks driving poverty and hunger in developing economies. IFPRI Blog Post. https://www.ifpri.org/blog/beyond-the-harvest-uncovering-the-hidden-risks-driving-poverty-and-hunger-in-developing-economies/

Country/Region

Bangladesh; Egypt; Ethiopia; Kenya; Malawi; Nepal; Papua New Guinea; Rwanda; Uganda; Zambia

Keywords

Southern Asia; Middle East; Northern Africa; Eastern Africa; Western Africa; Southern Africa; Africa; Food Access; Fragility; Conflicts; Markets; Value Chains; Poverty

Language

English

Access/Licence

Open AccessCC-BY-4.0

Source

Source record

Record type

Blog Post

Blogs

Beyond the harvest: Uncovering the hidden risks driving poverty and hunger in developing economies
Research Post
 | 
April 24, 2026

Beyond the harvest: Uncovering the hidden risks driving poverty and hunger in developing economies

By Askar Mukashov and Eleanor Jones

Systematic risk profiling reveals wide differences in country vulnerabilities.

Revisiting development strategies under climate uncertainty: Insights from Malawi
Research Post
 | 
July 19, 2024

Revisiting development strategies under climate uncertainty: Insights from Malawi

By Askar Mukashov and Eleanor Jones

Does the threat of climate change significantly undermine strategies focusing on agriculture?

Events

AI Policy Considerations for Food, Land, and Water Systems
IFPRI Policy Seminar

 | 

September 15, 2026, 9:30 – 11:00 am

AI Policy Considerations for Food, Land, and Water Systems

Also streaming on Please type your questions into the chat box with name, affiliation, and country. The event video, presenter slides, and podcast will be available in the days following the event. Amid rapid advances in artificial intelligence (AI), there is growing interest in using AI as a tool to advance food systems transformation. AI […]

Large Language Models for Policy Makers: Exploring RIAPA-AI and PEPA-AI
IFPRI Program/Country Event

 | 

March 12, 2026, 9:00 – 10:00 am

Large Language Models for Policy Makers: Exploring RIAPA-AI and PEPA-AI

As large language models (LLMs) begin to influence how evidence is generated, synthesized, and communicated, new opportunities are emerging for policymakers and analysts working in food systems. This webinar explores how generative AI tools can support policy design, analysis, and dialogue—bridging technical modeling and real-world decision-making.  Askar Mukashov, Research Fellow at IFPRI, will introduce RIAPA-AI, an extension of the Rural Investment and […]

How vulnerable are economies to systemic risks? New approaches to assessing economic and climate-related shocks
IFPRI Program/Country Event

 | 

May 14, 2025, 9:30 – 10:30 am

How vulnerable are economies to systemic risks? New approaches to assessing economic and climate-related shocks

This webinar showcases two innovative approaches to understanding how economic and climate-related risks affect national economies and vulnerable populations across developing regions. The first presentation introduces an economywide modeling approach to assess economic vulnerabilities. It examines how national economies and key population groups are exposed to shocks—such as global market fluctuations or disruptions in domestic […]

View all Events

News

February 09, 2026 | Press Release

New IFPRI study introduces breakthrough framework to assess economic risks in developing countries

A groundbreaking approach to understanding how developing economies are affected by multiple, simultaneous shocks.