Accounting research asks how information, incentives, institutions, and judgment shape economic decisions. The work may use market data, contracts, experiments, interviews, field settings, or mathematical models. It rarely resembles bookkeeping.
Two dimensions: topic and method
A topic is the part of accounting you want to understand. A method is the kind of evidence and reasoning you use. Researchers often describe themselves with both: financial–archival, audit–experimental, managerial–analytical, and so on.
| Topic | Questions might include |
|---|---|
| Financial reporting | How do disclosures affect investors, managers, analysts, or capital markets? |
| Auditing | How do incentives, judgment, regulation, and audit technology affect assurance? |
| Managerial accounting | How do measurement, targets, controls, and compensation affect decisions inside organizations? |
| Tax | How do taxpayers and firms respond to tax rules, enforcement, and disclosure? |
| Accounting information systems | How do systems, data, controls, cybersecurity, and emerging technologies alter accounting work? |
| Other settings | Nonprofit, governmental, sustainability, education, history, and interdisciplinary questions. |
The main methods
Archival
Archival researchers analyze existing records: financial statements, returns, filings, audit data, tax records, contracts, or text. The central difficulty is not running a regression; it is designing a credible comparison when the data were not generated by random assignment.
Experimental
Experimental researchers design tasks or settings that isolate a causal mechanism. Participants may be students, professionals, managers, investors, auditors, or online samples. Strong work depends on theory, valid manipulations and measures, and a design that rules out competing explanations.
Analytical
Analytical researchers build formal models to study incentives, information, contracting, and equilibrium behavior. The method requires substantial mathematical preparation and a taste for abstraction.
Qualitative, survey, and field methods
Interviews, observation, cases, surveys, field studies, and mixed methods can answer questions that large datasets or laboratory tasks cannot. “Other” does not mean less rigorous; it means the evidence and standards of inference differ.
What a project actually involves
- Find a consequential question. A gap in the literature is not automatically an important problem.
- Explain the mechanism. What should happen, why, and under what conditions?
- Design evidence that could change your mind. Decide what observation would support or weaken the explanation.
- Acquire or create the data. This may require database work, hand collection, a survey, an experiment, interviews, or a model.
- Analyze and test alternatives. The first result is the beginning of scrutiny, not the conclusion.
- Write, present, revise, and submit. Workshops and peer review expose weaknesses and usually reshape the paper several times.
Data and tools
Much North American archival research uses data reached through Wharton Research Data Services (WRDS). Common sources include Compustat for financial statements, CRSP for security returns, LSEG IBES for analyst forecasts, Ideagen Audit Analytics for audit-related data, and the SEC's free EDGAR search for filings.
Stata and SAS remain common in archival work; R and Python are increasingly useful for reproducible workflows, text, scraping, visualization, and machine learning. The right first goal is fluency in one language and comfort in one statistical environment, not shallow exposure to every tool.
WRDS subscriptions vary by institution. A paper may use a database your program does not license. Ask about data access while comparing programs, especially if your interests depend on proprietary audit, compensation, news, or transaction data.
Machine learning and generative AI
Machine learning is both a research method and a subject of study in accounting. It appears in prediction, classification, textual analysis, and measurement. Generative AI is also becoming an object of research in reporting, auditing, education, and professional judgment.
Using AI in research and writing is governed by journal and publisher policy. Requirements can differ and change quickly, so check the target journal immediately before submitting. Never upload confidential data, manuscripts, or reviews to a system unless the relevant policy and your research agreements permit it.
Policy starting points: AAA manuscript preparation guide, Contemporary Accounting Research editorial policies, and Elsevier journal policy.
How to find out whether you like it
- Choose one topic and read the abstracts in a recent issue of two leading journals.
- Select three papers whose questions interest you and read them completely.
- For each, write one paragraph answering: What is the question? Why should anyone care? What evidence changes what we believe?
- Discuss your notes with a research-active professor or doctoral student.
- Ask for a bounded research-assistant task so you experience the work behind the paper.
The field's leading general-interest journals are commonly identified as The Accounting Review, Journal of Accounting Research, Journal of Accounting and Economics, Contemporary Accounting Research, Review of Accounting Studies, and Accounting, Organizations and Society. The BYU Accounting Research Rankings rank these and other journals by publication counts across topics and methods. Specialized journals may be better entry points for audit, tax, systems, managerial, behavioral, education, or sustainability interests.
How research differs from practice
Practice asks what a client, regulator, manager, or standard requires now. Research asks what is true more generally, why it is true, and how we could know. Professional experience can generate excellent questions and institutional knowledge, but a study must contribute beyond one organization's immediate decision.