The Accounting Research Frontier
Published:
What determines which topics the accounting research frontier moves toward? The usual story is competing importance: we study the issues that matter most to firms, investors, and regulators. These slides document a different force — the arrival of machine-readable data — and ask whether the frontier is shaped less by what is economically important than by what has recently become researchable.
Using the full text of roughly 7,000 articles in the top-three accounting journals (JAR, JAE, TAR, 1968–2025), a topic model recovers 37 research topics whose shares rise and fall over time. ESG and carbon disclosure, conference-call tone, proprietary disclosure, and fraud/enforcement surge in the modern era, while accruals quality, conservatism, and auditor judgment recede. The striking regularity is when the rising topics take off: each accelerates a few years after a key dataset becomes available — ESG vendor data → the ESG surge, machine-readable call transcripts → disclosure tone, the AAER database → fraud research, NBER patent data → proprietary disclosure, Dealscan → debt contracting. On the strongest emergence measure, 83% of data-vendor topics surge only after their data arrives (median lag: 8 years), and a one-year-later dataset shifts the surge nearly one-for-one, while the year the underlying issue became salient in practice adds essentially nothing.
The upshot: our collective evidence base has systematic blind spots wherever data have not yet been digitized, and the “frontier” partly reflects the supply decisions of data vendors rather than the importance of questions.