Seeds for recovery: the long-term impacts of a complex agricultural intervention on welfare, behaviour and stability in Syria

Programme of work

Evaluating complexity

Principal investigator(s)

Tilman Brück

Host institution

International Security and Development Center gGmbH (ISDC)

Other institutions

American University of Beirut (AUB)

Dates

February 2020 to January 2023

Project type

Evaluation

Country/ies

Syria

Research question

This project will evaluate the long-term impact of a complex agricultural intervention conducted by the Food and Agricultural Organization (FAO) in acutely conflict-affected Syria.

FAO currently implements an agricultural intervention, funded by the De-partment for International Development, across government and rebel-held areas in Syria by providing two packages of measures to address both emergency and early recovery needs. The composition, location and timing of the packages vary with the local context, making the intervention complex.

Research design

The study will:

  1. Use modern computational methods (machine learning and deep learning) to produce geo- and time-coded data from the satellite information;
  2. Spatially and temporally match the survey data with the processed satellite information to facilitate analyses of how contextual variation affects the intervention impacts;
  3. Use difference-in-difference techniques based on five different points in time to study the separate and additive impacts of the intervention packages and temporal evolution;
  4. Perform machine-learning estimations to assess long-term programme sustainability and heterogeneity of treatment effects;
  5. Build on these parametric and non-parametric techniques for decomposition analysis assessing impact pathways; and
  6. Conduct an out-of-sample prediction analysis that emphasises the role of contextual factors.

Data source

The study will use five-wave panel data (of which three will be post-intervention). The study sample comprises 1,447 households across Syria (in both government- and rebel-held areas), including 718 treated and 729 control households.

These individual-level data will be matched with geo- and time-coded in-formation generated from satellite data on conflict, weather, agro-ecological conditions and agricultural productivity.

The agricultural and agro-ecological information will be purchased from Worldview (32 cm resolution) and Sentinel-2 (10 m resolution).

Policy relevance

The study will generate new insights on the long-term impacts of complex agricultural interventions implemented in settings experiencing protracted crises. Evidence from this study can help fine-tune current and design future programmes to improve welfare and nutrition, and promote sustainable agriculture in Syria and beyond.