Policy relevance of causal quantities in networks

TL;DR

This paper examines how different causal effect estimands in network settings relate to policy relevance and interpretability, advocating for the expected average outcome as a key estimand for policy decisions.

Contribution

It characterizes various estimands based on averaging orders, highlighting the importance of the expected average outcome for policy relevance and interpretability.

Findings

Many estimands involve two types of averaging over units and treatments.

Homogeneous exposure assumptions often misalign with heterogeneous policies.

Expected average outcome is both interpretable and policy-relevant.

Abstract

In settings where units' outcomes are affected by others' treatments, there has been a proliferation of ways to quantify effects of treatments on outcomes, including via indirect exposure to other units' treatments. Here we consider two properties we might want estimands to have: being interpretable as summaries of unit-level effects, and being relevant to choice of a policy governing treatment assignment. We characterize many estimands as involving one of two orders of averaging over units in a population and over treatment assignments under a policy. The more common representation often results in quantities that are insufficient for optimal policy choice. This occurs because these quantities summarize outcomes under homogeneous exposure to treatment, but even homogeneous policies often lead to heterogeneous exposures. The other representation often yields quantities that lack an interpretation as summaries of unit-level effects. We argue that, among various estimands, the expected average outcome, which averages over units and treatment assignments in either order, deserves further attention from researchers. This estimand, or contrasts among these estimands under different policies, is both a summary of unit-level effects and is sufficient for optimal policy choice with utilitarian welfare.