Unlocking Causality: How Angrist and Imbens Revolutionized Economics

Joshua Angrist and Guido Imbens revolutionized economics by clarifying how to interpret instrumental variable estimates. Their work provided a new lens for understanding causality, bridging the gap between potential outcomes and practical application. This involves using an 'instrument' – a variable that affects the treatment but only influences the outcome through the treatment itself. The LATE framework utilizes instrumental variables to estimate causal effects, especially when randomization isn't feasible. It helps to identify the causal effect within a specific subgroup influenced by the instrument, providing valuable insights into complex economic relationships.

Angrist and Imbens emphasized the importance of treatment effect heterogeneity, recognizing that individuals respond differently to the same intervention. This perspective is central to the LATE framework. They demonstrated that instrumental variables could recover an average causal effect for a specific subgroup influenced by the instrument. This acknowledges that causal effects vary across individuals, and the LATE approach focuses on identifying the effect within a specific subpopulation (compliers). This focus on heterogeneity allows for a more nuanced understanding of causal relationships, recognizing that interventions don't impact everyone in the same way.

The LATE framework relies on four key assumptions: 1. Exclusion Restriction: The instrument affects the outcome only through the treatment. 2. Independence: The instrument is 'as-if' randomized. 3. Relevance: The instrument has a causal effect on the treatment. 4. Monotonicity: The instrument affects the treatment in the same direction for all individuals. These assumptions are crucial because they enable researchers to isolate the causal effect of the treatment on the outcome. They ensure that any observed relationship between the instrument and the outcome is mediated through the treatment itself, providing a credible identification strategy for causal inference.

The Local Average Treatment Effect (LATE) is a core contribution of Angrist and Imbens. It is the average causal effect for a specific subgroup of individuals whose treatment status is changed by the instrument (compliers). LATE acknowledges that without restricting the heterogeneity of treatment effects, a sensible causal parameter's identification requires that the instrument affects all individuals in the same direction. It focuses on the causal effect within the subpopulation affected by the instrument, providing a localized view of the impact of the treatment. This means that the LATE estimates are not necessarily representative of the entire population but are specific to the 'compliers,' those whose treatment choice is influenced by the instrument. It is a powerful tool, particularly in settings where it is impossible to randomize the treatment.

The works of Angrist, Imbens, and Rubin have converged fundamental ideas, including potential outcomes, credible identification sources, and the acknowledgement of treatment effect heterogeneity. These have since become a dominant paradigm for discussing causal effect estimation in econometrics. The LATE framework and related methodologies, pioneered by Angrist and Imbens, have introduced a new framework, assumptions, and identification results that have become standard in econometrics. Their insights have changed the way economists approach causal inference. Their work is widely used for research and informing policy decisions, ensuring their lasting impact on the field, and advancing the understanding of causal relationships.