Composite performance scores measure the level of a firm’s transformation but not the efficiency with which resources are converted into it. This paper constructs the efficiency layer of the Pathway-Capability-Performance (PCP) framework by applying data envelopment analysis to a panel of six platform-years assembled from the results announcements of Alibaba Group, JD.com, and PDD Holdings. Research and development (R&D) expenditure is the single input, revenue and net income attributable to ordinary shareholders are the outputs. Under an input-oriented, constant-returns specification, two units lie on the frontier and four fall below it, with scores ranging from 0.299 to 1.000. Efficiency declined at all three platforms between their two most recent reported fiscal years, and this decline persists across every alternative specification tested: a non-Generally Accepted Accounting Principles (GAAP) profit measure, a revenue-only single-output model, and variable returns to scale. Its mechanism is transparent – research spending grew faster than revenue at all three platforms, while profit fell at all three. The ranking of the declines, by contrast, is not robust. It reverses across specifications, and the paper therefore reports it as specification-dependent rather than as a finding, correcting an interpretation advanced in an earlier version of this analysis. A scale decomposition locates almost all of Alibaba’s measured inefficiency in scale rather than in pure technical efficiency, which is consistent with – though it does not identify – a build-out interpretation. The paper documents three constraints under which frontier analysis operates in concentrated markets: units below conventional thresholds, censoring of frontier units in longitudinal use, and the inseparability of a market-wide shock from firm-specific effects. It also specifies an eight-point protocol for reporting the efficiency layer so that these constraints remain visible to users of the model.
References
[1] Verhoef, P. C., Broekhuizen, T., Bart, Y., Bhattacharya, A., Dong, J. Q., Fabian, N., Haenlein, M. (2021) Digital transformation: a multidisciplinary reflection and research agenda. Journal of Business Research, 122, 889-901.
[2] Charnes, A., Cooper, W. W., Rhodes, E. (1978) Measuring the efficiency of decision making units. European Journal of Operational Research, 2(6), 429-444.
[3] Banker, R. D., Charnes, A., Cooper, W. W. (1984) Some models for estimating technical and scale inefficiencies in data envelopment analysis. Management Science, 30(9), 1078-1092.
[4] Golany, B., Roll, Y. (1989) An application procedure for DEA. Omega, 17(3), 237-250.
[5] Dyson, R. G., Allen, R., Camanho, A. S., Podinovski, V. V., Sarrico, C. S., Shale, E. A. (2001) Pitfalls and protocols in DEA. European Journal of Operational Research, 132(2), 245-259.
[6] Cook, W. D., Tone, K., Zhu, J. (2014) Data envelopment analysis: prior to choosing a model. Omega, 44, 1-4.
[7] Simar, L., Wilson, P. W. (1998) Sensitivity analysis of efficiency scores: How to bootstrap in nonparametric frontier models. Management Science, 44(1), 49-61.
[8] Rodchenkov, M. V. (2026) The functionality of economic metrics in a multipolar world. Journal of New Economy, 27(1), 115-133.
[9] Abdallah, T., Bray, R. L., Choi, H., Glinskiy, V. (2026) The impact of market growth on delivery speed: evidence from JD. com. Service Science, 18(2), 96-119.
[10] Reust, D. (2026) Marketing strategies of Chinese online platforms in Switzerland. Innovating Business and Education for Sustainable Development, 173-187.
Share and Cite
Lin, G. (2026) The Efficiency Layer of a Performance Evaluation Model: Data Envelopment Analysis of Artificial-intelligence Investment in Chinese E-commerce Platforms. Hong Kong Financial Bulletin, 2(3), 14-22. https://doi.org/10.71052/hkfb2025/CNYJ1453