A surreal illustration representing the interconnectedness of policy interventions across the globe.

Unlocking Policy Insights: How Synthetic Interventions Are Revolutionizing Policy Evaluation

Samir D’Costa in Politics & Policy September 2025 • 4 min read.

"Move Over, Standard Methods: A New Way to Evaluate Policies and Make Data-Driven Decisions."

In today's rapidly evolving world, effective policy evaluation is more critical than ever. Policymakers and researchers alike need robust tools to understand the true impact of their interventions. For years, the synthetic control (SC) methodology has been a go-to solution, celebrated for its ability to analyze policy effects in panel data applications. But what happens when policies become more complex, involving multiple treatments and diverse scenarios? This is where the next-generation approach, synthetic interventions (SI), steps in to transform the landscape of policy evaluation.

The traditional synthetic controls framework operates on a low-rank matrix factor model, assuming that potential outcomes can be neatly described by a limited set of factors. While powerful, this approach hits a wall when dealing with multiple treatments or when trying to answer more nuanced questions. Imagine trying to evaluate not just the impact of an anti-tobacco program, but also how that impact might change if taxes were raised simultaneously. The SC method, in its original form, simply wasn't designed to handle such complexity.

Enter synthetic interventions, a game-changing extension of the SC framework. SI leverages a low-rank tensor factor model, which enhances the matrix factor model by including a latent factorization over treatments. This unlocks the ability to explore a wider array of policy questions, offering a more comprehensive understanding of policy impacts across different scenarios. With SI, policymakers gain a sharper, more versatile tool for evidence-based decision-making.

What Are Synthetic Interventions and How Do They Work?

A surreal illustration representing the interconnectedness of policy interventions across the globe.

At its core, SI is a methodology designed to tackle the limitations of traditional synthetic control methods. It achieves this by introducing a tensor factor model, which essentially adds another layer of analysis to account for different treatments. Think of it as expanding a simple spreadsheet into a multi-dimensional database, capable of capturing more complex relationships.

The SI framework operates under a few key assumptions. First, it assumes that potential outcomes can be represented by a low-rank tensor factor model. This means that the outcomes are influenced by a relatively small number of underlying factors related to the unit (e.g., a state), time period, and treatment. Second, SI relies on the principle of latent unit factors remaining constant across both time and treatments. This allows researchers to learn relationships during a pre-treatment period and apply those insights to predict outcomes in a post-treatment setting, even when different interventions are in play.

Tensor Factor Model: SI uses a tensor factor model to analyze potential outcomes, adding a layer to account for different treatments.
Consistent Estimation: The framework provides consistent estimates, ensuring the reliability of policy evaluations.
Asymptotic Normality: Under specific conditions, SI estimators are asymptotically normal, allowing for the construction of confidence intervals.
Empirical Validation: Simulations and case studies support the theoretical results, confirming the effectiveness of SI in real-world scenarios.

By making these assumptions and employing tensor factorization, synthetic interventions provide a robust and flexible framework for evaluating policies in a wide range of contexts. The result is a more nuanced and accurate understanding of policy impacts, paving the way for better-informed decisions.

The Future of Policy Evaluation is Here

Synthetic interventions represent a significant step forward in the field of policy evaluation. By addressing the limitations of traditional methods and offering a more versatile framework, SI empowers policymakers and researchers to gain deeper insights into the true impact of their interventions. As the world becomes increasingly complex, tools like SI will be essential for making informed, evidence-based decisions that drive positive change. The future of policy evaluation is here, and it’s looking more insightful than ever.

About this Article -

This article was crafted using a human-AI hybrid and collaborative approach. AI assisted our team with initial drafting, research insights, identifying key questions, and image generation. Our human editors guided topic selection, defined the angle, structured the content, ensured factual accuracy and relevance, refined the tone, and conducted thorough editing to deliver helpful, high-quality information.See our About page for more information.

This article is based on research published under:

DOI-LINK: https://doi.org/10.48550/arXiv.2006.07691,

Title: Synthetic Interventions

Subject: econ.em cs.lg stat.ml

Authors: Anish Agarwal, Devavrat Shah, Dennis Shen

Published: 13-06-2020

Everything You Need To Know

What are synthetic interventions (SI) and why are they considered a revolutionary methodology in policy evaluation?

Synthetic interventions (SI) represent a next-generation approach to policy evaluation, building upon the foundation of synthetic control (SC) methods. SI is revolutionary because it addresses the limitations of traditional methods, especially when evaluating complex policies involving multiple treatments. The core innovation of SI lies in its use of a tensor factor model, which enhances the matrix factor model used in SC, thereby unlocking the ability to explore a wider array of policy questions and providing a more comprehensive understanding of policy impacts across different scenarios. This enhanced versatility makes it a game-changer for policymakers seeking evidence-based decision-making.

How do synthetic interventions (SI) differ from traditional synthetic control (SC) methods in evaluating policy impacts?

Traditional synthetic control (SC) methods operate on a low-rank matrix factor model, which assumes that potential outcomes can be neatly described by a limited set of factors. This approach struggles when dealing with multiple treatments or nuanced policy questions. Synthetic interventions (SI) extend this by leveraging a low-rank tensor factor model, which includes a latent factorization over treatments. This allows SI to handle more complex scenarios, offering a more comprehensive understanding of policy impacts across different interventions, something the original SC method wasn't designed to do.

What key assumptions underpin the synthetic interventions (SI) framework, and how do these assumptions contribute to its effectiveness?

The synthetic interventions (SI) framework operates under key assumptions that include representing potential outcomes using a low-rank tensor factor model, which posits that outcomes are influenced by a small number of underlying factors related to the unit, time period, and treatment. Additionally, SI relies on the principle that latent unit factors remain constant across both time and treatments. This allows researchers to learn relationships during a pre-treatment period and apply those insights to predict outcomes in a post-treatment setting, even when different interventions are in play. These assumptions enable SI to provide a robust and flexible framework for evaluating policies in diverse contexts, leading to more nuanced and accurate understanding of policy impacts.

What practical benefits do policymakers gain by using synthetic interventions (SI) for policy evaluation, and how does it improve decision-making?

By employing synthetic interventions (SI), policymakers gain a sharper and more versatile tool for evidence-based decision-making. SI enhances the matrix factor model by including a latent factorization over treatments which unlocks the ability to explore a wider array of policy questions. Also, the SI framework operates under assumptions that potential outcomes can be represented by a low-rank tensor factor model. This means that the outcomes are influenced by a relatively small number of underlying factors related to the unit (e.g., a state), time period, and treatment. It enables a deeper understanding of the true impact of their interventions. As the world becomes increasingly complex, tools like SI are essential for making informed, evidence-based decisions that drive positive change. The result is a more nuanced and accurate understanding of policy impacts, paving the way for better-informed decisions.

The synthetic interventions (SI) framework provides consistent estimates and asymptotic normality. What are the implications of these properties for policy evaluation?

The fact that synthetic interventions (SI) provides consistent estimates means that, as the amount of data increases, the estimates produced by SI will converge to the true values of the policy impacts. This ensures the reliability of policy evaluations. The property of asymptotic normality, under specific conditions, allows for the construction of confidence intervals around the estimated policy effects. These confidence intervals provide a range within which the true effect is likely to lie, giving policymakers a measure of the uncertainty associated with the estimates. Together, consistent estimation and asymptotic normality enhance the credibility and interpretability of the results obtained through SI, further supporting evidence-based decision-making.