How did empirical economic research improve over the last decades: Recent trends and prospects

Abstract

We map the evolution of economics research from 1998 to 2019 to assess progress and identify remaining challenges. Using text mining of 30,675 full-text articles and an in-depth analysis of 578,132 statistical tests from 3746 articles across a broad range of journals, we document changes in research practices over time. We highlight the influence of the credibility revolution and the open science movement. Awareness of reproducibility has increased modestly, largely driven by journal adoption of data and code sharing policies. At the same time, reliance on nonpublic data has grown, especially among authors at top-five universities, posing new challenges for transparency. Sample sizes have increased substantially, while reported effect sizes have declined, consistent with reduced exaggeration and potentially more policy-relevant estimates. Econometric methods and causal inference practices have become more sophisticated, and articles now report more tables and statistical tests. However, pre-analysis plans, power analyses accompanying null results, and corrections for multiple hypothesis testing remain uncommon. Article length has remained stable or declined slightly, and authors from top-five universities continue to represent a large and stable share of publications in top journals. Our findings provide a baseline for tracking future developments and for guiding efforts to further strengthen the credibility and transparency of economics research.