Financial data analytics without identification: a structured review of a fragmenting field
Year: 2026
Type: article
Abstract: We review 224 studies of financial data analytics published between 2020 and 2026 and document a mismatch between the questions the field asks and the designs it uses to answer them. Roughly three in ten titles in the corpus assert a causal relationship — analytics capability improving financial performance, digital inclusion reducing unemployment, ESG orientation raising firm value — while not one names a difference-in-differences, instrumental-variable, regression-discontinuity, or event-study design. The field is also centrifugal: 224 studies are spread across 124 outlets, of which 85 publish a single paper on the subject, and the largest research stream is shrinking rather than consolidating. Its share of annual output falls from 31.0% in 2020–2022 to 15.2% in 2025–2026, displaced by sustainability, fraud detection, and household finance applications. We interpret this pattern as premature thematic diversification: the field moved on from its foundational question about whether analytics investment pays before establishing an evidentiary basis for answering it. We propose an identification agenda built around shocks the sector already supplies — staggered data-protection regimes, open-banking mandates, supervisory technology rollouts, and cloud infrastructure failures — and specify design templates for each. The review is bounded by its reliance on title- and outlet-level coding of an author-compiled corpus, which we discuss at length.
Source: Computer and decision making.
Institution: MIT Art, Design and Technology University
Language: English
Subfield: Management Information Systems
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W7225008879