Fintech Integration and Cost Efficiency in Nigerian Deposit Money Banks: Evidence on the Moderating Role of Bank Size, 2021–2025

This paper examines whether fintech integration intensity is associated with the cost efficiency of Nigerian deposit money banks (DMBs), and whether that relationship is conditioned by bank size. Using hand-assembled panel data drawn from the audited consolidated annual financial statements of nine Nigerian-Exchange-listed DMBs over FY2021–FY2025 (45 bank-year observations), cost efficiency is proxied by the cost-to-income ratio (CIR) and fintech integration intensity by the ratio of capitalised intangible (software and core-banking) assets to total assets. A two-way fixed-effects panel model, with bank size (relative log total assets) entered as a moderator, is estimated with bank-clustered standard errors and corroborated by wild cluster bootstrap inference appropriate for the small number of cross-sectional clusters (G = 9). Descriptively, the sample-wide CIR fell from 60.8% in FY2021 to 46.1% in FY2025, but this decline is almost entirely absorbed by bank and year fixed effects: the contemporaneous coefficient on fintech integration intensity, its interaction with bank size, and the joint Wald test on both terms are all statistically insignificant across the baseline model and eight robustness checks (alternative dependent variables, an alternative fintech proxy, an alternative moderator, year-exclusion, first-differencing, and a no-controls specification). The one robust signal in the data is a one-year-lagged interaction between fintech intensity and relative size, which is negative and significant (p < 0.05, wild-cluster p < 0.01), consistent with a delayed and scale-contingent efficiency payoff. The paper concludes that, within this five-year window, industry-wide cost-efficiency gains in Nigerian banking are better explained by common time effects than by cross-bank differences in balance-sheet fintech intensity, and offers methodological and data recommendations for future work.