How Legislative Language Reveals Bureaucratic Discretion and Executive Delegation

🔎 What the Project Does

This study introduces an automated, text-based method for measuring bureaucratic discretion and executive delegation in statutes, enabling large-scale empirical work on legal delegation across U.S. states. The approach draws on tools from computational linguistics to turn legislative language into measurable signals about who is empowered to act and how much detail lawmakers provide.

🧾 How the Texts Were Read

The procedure parses statutory text using syntactic parse trees to locate legally relevant provisions and then identifies:

• agents mentioned in the law (who may act),
• delegated actions (what those agents are authorized to do), and
• indicators of legislative detail (language that constrains or specifies agency action).

This automated pipeline is designed to replace manual coding and to scale across large corpora of state statutes for comparative analysis.

📊 Key Empirical Applications and Findings

Two applications with state-level statutes demonstrate the method's utility:

• A measure of legislative detail is produced to proxy bureaucratic discretion. Higher values of this measure indicate more specific legislative instructions (that is, less discretion for agencies). This measure is observed to increase following reforms that grant agencies greater independence. The pattern is consistent with an agency-cost model in which more independent bureaucracies receive more specific instructions to reduce bureaucratic drift.

• Text-derived measures of delegation to governors are constructed and validated. These measures show that executive delegation in state legislation rises under unified government, a result consistent with earlier estimates based on non-text indicators.

💡 Why This Matters

The methodological contribution provides a replicable, scalable way to read legal texts for delegation and discretion, bridging computational linguistics and political economy. The empirical findings both validate the text-based measures against existing results and offer new, fine-grained leverage for studying how institutional change (like agency independence or unified government) reshapes the distribution of authority in American state politics.