How are innovation inputs and measured outputs distributed across the regions of a resource-dependent post-Soviet economy, and how closely do they correspond? We use a balanced official panel of Kazakhstan’s 17 regions for 2014–2025 (204 region–year observations) covering four indicators: R&D expenditure, innovation-active enterprises, the innovation activity rate and innovative product output. Granted patents for inventions are available by region for 2021–2025 only; this series is shorter than the panel, so it is used in the 2025 cross-sectional and descriptive analyses and does not enter the panel or machine learning models. On the panel, we estimate dispersion measures, panel regression with region-clustered standard errors and tree-based machine learning interpreted with SHapley Additive exPlanations; on the 2025 cross-section, we compute correlations, clustering and spatial association statistics. Inputs are highly concentrated: Almaty and Astana together accounted for 65.1% of national R&D expenditure in 2025, and cross-regional dispersion did not narrow over the period. In the 2025 cross-section (n = 17), R&D expenditure, innovation-active enterprises and patent grants are strongly inter-correlated (r = 0.96–0.97), but none of the three is significantly correlated with innovative product output. This input–output misalignment is a pattern whose mechanisms we do not test. The panel specification has modest explanatory power (R² = 0.188), which bounds the weight that can be placed on any single coefficient. Within it the innovation activity rate is the only significant predictor of output (β = 0.177, p < 0.001), an association that survives exclusion of the three regions affected by the 2022 administrative reform; because that indicator and the output measure are compiled within the same system of enterprise observation, it is reported as a within-dataset association requiring validation against independently measured indicators. Under region-grouped cross-validation, the machine learning models show no out-of-region predictive ability and are reported solely as in-sample interpretation. The contribution is a documented diagnosis of input–output misalignment in an understudied regional system, with an explicit account of what such data can and cannot support, and a prospective statement of what institutional readiness for algorithmic decision support in regional innovation policy would require.

Kazakhstan’s regional innovation policy, including its Smart Specialisation and Quadruple Helix approaches, relies predominantly on static indicators, leaving the separate contributions of technological change and efficiency catch-up to productivity dynamics unexamined. This study decomposes total factor productivity (TFP) change in the innovation systems of Kazakhstan’s regions into efficiency-catch-up and technological-change components using a Malmquist-DEA framework, and identifies the spatial and structural determinants of this change. The research hypothesis is that the average decline in regional innovation productivity over 2014–2024 is driven predominantly by technological regress (a contracting production frontier) rather than efficiency catch-up, and that this process is institutional rather than geographic in nature. The study draws on an official panel from the Bureau of National Statistics of the Republic of Kazakhstan for 2014–2024 across 17 regions (187 observations), combining panel regression, Malmquist-DEA decomposition, spatial autocorrelation analysis, principal component and cluster analysis, and explainable machine learning (Random Forest, gradient boosting, CatBoost with SHAP). Results show that the geometric-mean TFP change between 2014–2016 and 2022–2024 was 0.662, driven predominantly by technological regress (0.543) and partly offset by positive efficiency catch-up (1.219). No significant global spatial autocorrelation of productivity change was detected (Moran’s I = 0.148; p = 0.436), confirming an institutional rather than geographic diffusion mechanism. The findings indicate that Kazakhstan’s Smart Specialisation and GovTech/AI Governance policies should be reoriented toward technological renewal rather than efficiency improvement alone.

Resource-dependent, spatially polarised, post-Soviet regional innovation systems remain under-studied, and are rarely analysed with the spatial-econometric and explaina-ble-machine-learning toolkit now standard in smart-city research. This paper uses an eleven-year official panel (2014-2025, 17 regions of Kazakhstan) covering R&D expendi-ture, innovation-active enterprises, the innovation-activity rate, innovative product output, and patenting. We triangulate five methods: Principal Component Analysis, Entropy Weight and TOPSIS integral indices; cluster analysis; panel econometrics; Random Forest and Gradient Boosting with SHAP; and spatial econometrics (Moran's I, Geary's c, LM diagnostics for SAR/SEM/SDM choice). We find extreme, persistent concentration of inno-vation inputs in Almaty and Astana (65.1% of national R&D expenditure, 2025), signifi-cant divergence on three of four indicators over 2014-2025, and a robust structural dis-connect between innovation inputs and outputs - a 'productivity-driven diffusion para-dox' - confirmed independently by correlation, factor analysis, panel regression, machine learning, and spatial econometrics. A robust three-tier regional typology is corroborated by an independent external study, and spatial analysis identifies peripheral-anomaly regions failing to absorb proximity benefits from more developed neighbours. We interpret these findings through Regional Innovation Systems, Smart Specialisation, New Economic Ge-ography, and Mission-Oriented Innovation Policy theory, and propose an evidence-based, AI-ready regional innovation governance framework - a unified data layer, explainable analytics, spillover corridors, and cluster-differentiated policy -transferable to other re-source-dependent, spatially polarised transition economies.