The Historical Determinants of Legislative Effectiveness from 1873 to 2011

• In Event: Legislative Productivity in the U.S.

Abstract

A key concept in American legislative politics is the ``effectiveness'' of representatives in proposing and advancing bills on important topics that get substantial attention in the legislature. However, existing research has limited its focus on modern periods in American political history (1970s to present). We create a new measure of effectiveness that extends back to 1873 and examine its determinants over a long period of American history. To accomplish this, we first build a new and rich data set composed of the entire corpus of all 1.1. million proposed bills in the House and Senate from 1873 to 2011 as well as numerous ``raters" capturing the attention to given them that can be measured across the entire period (e.g. mentions in the Congressional Record, number of reports, etc.).

To estimate bill significance for this huge dataset requires overcoming a few inferential challenges. Given the size of the data, running traditional methods (e.g. Markov Chain Monte Carlo) would take months on even a powerful computer. Moreover, the baseline activity of Congress has shifted dramatically over time period considered and thus the model must allow for external covariates (e.g. the size of the chamber) to ensure reliable scores that are historically comparable. Doing so makes traditional methods take even longer and thus new inferential techniques are needed. To tackle this, we develop a new variational algorithm for ideal point estimation that is tailored to the above particulars and allows for dichotomous and count outcomes. We provide simulations to show that it works well on smaller datasets, as well as comparing our results against a limited run of the traditional method where we let the model run for months.

With this bill significance measure, we aggregate the bill-level scores into MC-level scores to measure legislative effectiveness. We verify that our scores correlate well with existing measures in the period where both are defined. Substantively, we examine a long period of political history where one can exploit large variation in polarization, the nature of the political system, (unified) party control, and other factors to better understand which components of effectiveness are specific to the modern era and which ones seem to be more constant throughout time.

Authors

• Fang-Yi Chiou, Academia Sinica

• Max Goplerud, University of Pittsburgh