Managers in South‑East Nigeria’s public enterprises report that AI forecasting and decision automation are linked to higher efficiency and innovation; improved competitiveness mediates these gains. The evidence is associative — strong survey-based correlations but not causal proof.

Main Finding

AI-driven decision making — proxied by predictive analytics capability (PAC) and intelligent decision automation (IDA) — has a significant, positive effect on business growth in South‑East Nigerian public enterprises. These effects operate both directly (improving operational efficiency and innovation capability) and indirectly via increased organizational competitiveness (service quality and competitive advantage).

Key Points

• The study tests AI-driven decision making (PAC, IDA) → Organizational Competitiveness (OC: service quality, competitive advantage) → Business Growth (operational efficiency, innovation capability).
• All hypothesized direct paths were significant (p < 0.001):
  • PAC → Operational Efficiency: β = 0.481 (C.R. = 7.914)
  • PAC → Innovation Capability: β = 0.455 (C.R. = 7.321)
  • IDA → Operational Efficiency: β = 0.423 (C.R. = 6.884)
  • IDA → Innovation Capability: β = 0.446 (C.R. = 7.204)
  • PAC → Organizational Competitiveness: β = 0.539 (C.R. = 8.521)
  • IDA → Organizational Competitiveness: β = 0.498 (C.R. = 8.017)
• Organizational competitiveness partially mediates the relationships from PAC and IDA to both operational efficiency and innovation capability. Indirect effects (bootstrapped, 5,000 samples) were significant:
  • PAC → OC → OE: indirect = 0.214 (Boot SE = 0.041, p ≈ 0.001)
  • PAC → OC → IC: indirect = 0.208 (Boot SE = 0.039, p ≈ 0.002)
  • IDA → OC → OE: indirect = 0.201 (Boot SE = 0.038, p ≈ 0.003)
  • IDA → OC → IC: indirect = 0.192 (Boot SE = 0.036, p ≈ 0.004)
• Descriptive means for constructs were high (≈4.1–4.2 on a 5‑point scale), indicating respondents perceive strong presence/impact of AI practices.
• Reliability and convergent validity: Cronbach’s αs ≥ 0.887; Composite Reliability ≥ 0.908; AVE ≥ 0.646 for constructs.
• Confirmatory factor analysis / measurement model fit: χ²/df = 2.084, GFI = 0.931, AGFI = 0.908, CFI = 0.965, TLI = 0.958, RMSEA = 0.058, SRMR = 0.041.

Data & Methods

• Setting and sample: public enterprises in five South‑East Nigerian states (Abia, Anambra, Ebonyi, Enugu, Imo), covering State Water Corporations, State Internal Revenue Services, State Transport Corporations, and Electricity Distribution Agencies.
• Population = 1,536 management and senior technical staff. Sample size computed via Taro Yamane = 317; 301 valid responses collected (94.9% response rate).
• Design: cross‑sectional survey; multi‑stage sampling (purposive selection of enterprises, proportionate allocation, simple random selection of respondents).
• Measures (5‑point Likert): Predictive Analytics Capability (PAC), Intelligent Decision Automation (IDA), Organizational Competitiveness (OC: service quality, competitive advantage), Operational Efficiency (OE), Innovation Capability (IC). Each construct measured with 5 items.
• Validation: expert review, pilot testing. Internal reliability: Cronbach’s αs ranged ~0.887–0.914. Convergent and discriminant validity established (AVE, Fornell‑Larcker).
• Analysis: SPSS 29 and AMOS 29. Confirmatory Factor Analysis and Structural Equation Modeling used to estimate direct and mediating effects. Bootstrapping (5,000 samples) used for mediation inference.
• Model specification examples:
  • OE = f(PAC, IDA, OC)
  • IC = f(PAC, IDA, OC)
  • OC = f(PAC, IDA)

Implications for AI Economics

• Theory development:
  • Extends Resource‑Based View and Dynamic Capabilities Theory by empirically framing AI-driven decision making as both a strategic resource (valuable, rare, hard to imitate) and a reconfigurable capability that increases competitiveness and firm‑level growth in the public sector.
  • Demonstrates organizational competitiveness as a measurable mediating mechanism linking AI capabilities to performance outcomes.
• Public‑sector AI adoption economics:
  • AI investments in predictive analytics and decision automation can yield measurable returns in operational efficiency and innovation even in constrained, developing‑economy public organizations.
  • The positive mediating role of competitiveness implies that AI value accrues not only through cost/time savings but also by improving service quality and strategic positioning—important when evaluating public returns on AI expenditure.
• Policy and implementation:
  • To realize economic benefits, investments should pair AI technologies with complementary assets: digital infrastructure, workforce upskilling, data governance, and ethical/regulatory frameworks.
  • Policymakers and funders should consider funding bundles (technology + human capital + governance) rather than isolated technology procurement.
• Measurement and empirical strategy recommendations for researchers and practitioners:
  • Use both direct performance metrics and mediating constructs like competitiveness when estimating AI’s economic impact; mediation captures channels beyond short‑term cost savings.
  • Cross‑sectional self‑report evidence is suggestive but not definitive for causal inference—future research should use longitudinal designs, administrative performance data, or quasi‑experimental methods to estimate causal effects and payoffs over time.
• Broader economic considerations:
  • Scaling AI across public enterprises could improve public service delivery, with downstream economic effects (e.g., higher tax compliance, more reliable utilities, reduced transaction costs). Cost–benefit assessments should quantify these externalities.
  • Equity, governance, and accountability risks remain salient; investments should prioritize transparent algorithms and data protection to avoid welfare losses that offset efficiency gains.

Limitations noted by the study (and relevant for economists): cross‑sectional survey design, self‑reported outcomes, single geopolitical region and public‑sector scope — limit generalizability and causal claims. Future work should address these gaps and estimate macroeconomic implications of scaled public‑sector AI adoption.