Energy Efficiency refers to the ratio of useful output to energy input within production processes, reflecting how effectively energy resources are converted into industrial output (IEA, 2023). It captures both technical efficiency (the performance of machines and systems) and behavioral efficiency (how human practices influence energy use).

Policy Incentives encompass fiscal, regulatory, and informational measures introduced by governments to encourage firms to invest in energy-saving technologies or adopt efficient practices. These include tax rebates, subsidies, low-interest loans, and energy performance standards (Backlund et al., 2012).

Institutional Effectiveness denotes the capacity of public agencies to design, coordinate, and enforce energy efficiency measures. Institutional quality or governance capacity, the capacity of public agencies and regulatory institutions to design, coordinate and enforce energy efficiency measures, is recognized as a key determinant of successful energy efficiency governance. Institutions that combine legal frameworks, capable implementing agencies, stakeholder coordination, and transparent enforcement mechanisms achieve better energy efficiency outcomes. (Institutional Based Governance Framework for EE in SIDS, 2021; Chen, Pinar; Román-Collado, 2024).

Technological Innovation is defined as the introduction of new or improved methods, equipment, or digital systems that enhance energy performance. Examples include smart meters, IoT-enabled monitoring, and automated control systems (Garcés; Godoy et al., 2025)

Capacity Building includes initiatives aimed at improving the skills, knowledge, and technical competence of workers, managers, and policymakers to facilitate effective energy management and policy execution (Sun; Edziah; Sun; Kporsu, 2019).

These constructs are interdependent. Policy incentives provide direction and motivation, institutional effectiveness ensures enforcement, technological innovation delivers practical means, and capacity building sustains adoption. The interaction among them defines industrial energy efficiency outcomes.

The study uses an unbalanced panel of Nigerian industrial firms observed between 2010 and 2023. The sample comprises 150 firms drawn from three energy-intensive subsectors: manufacturing, agro-processing, and chemicals. Firms were selected on the basis of data completeness across core variables and availability of energy use records. The final panel, therefore, prioritizes reliable longitudinal coverage while preserving firm heterogeneity. Sample demographics are summarized in Table 1. Key features are as follows.

These demographic descriptors contextualize estimates and permit subgroup analysis by subsector and size.

Firm-level quantitative data were obtained from the Nigerian Bureau of Statistics (NBS). Sectoral energy use, policy event dates, and program records were drawn from the Federal Ministry of Power and NESP databases. Firm surveys and industry reports provided supplemental information on technology adoption, capacity-building participation, and training intensity. Where possible, survey responses were cross-checked against company reports and program registries.

All variables are defined and justified below. The choices reflect the theoretical premise that policy incentives, institutional capacity, technical innovation, and human capital jointly determine firm energy performance.

Descriptive statistics, variance, and pairwise correlations are reported in Table 2 and Appendix A. Missing observations are treated using multiple imputation under the assumption of missing at random for core continuous variables. Binary policy and program indicators use deterministic imputation from program registries where feasible. Sensitivity tests compare results with listwise deletion (Baum, 2023).

The baseline empirical model relates energy intensity to policy incentives, capacity-building, technological innovation, and firm controls with firm and year fixed effects:

EI_it = α + β1 Policy_it + β2 Capacity_it + β3 Innovation_it + β4 X_it + μ_i + λ_t + ε_it.

Fixed effects μ_i account for time-invariant firm heterogeneity. Year fixed effects λ_t absorb common macro shocks. The Hausman test guides choice between fixed and random effects; fixed effects are expected to be preferred given the probable correlation between firm traits and regressors (Wooldridge, 2015; Zhou; Zhao; Yan, 2021).

To test interaction effects, we estimate:

EI_it = α + β1 Policy_it + β2 Capacity_it + β3 (Policy_it × Capacity_it) + β4 Innovation_it + β5 X_it + μ_i + λ_t + ε_it.

This specification tests whether capacity-building amplifies the effect of policy incentives on energy intensity.

Endogeneity concerns include reverse causality (firms with low EI are more likely to adopt innovation) and omitted time-varying confounders. To address these, we adopt two approaches.

EI_it = γ EI_{i,t-1} + β1 Policy_it + β2 Capacity_it + β3 Innovation_it + β4 X_it + μ_i + λ_t + ε_it.

System GMM corrects for dynamic bias and uses lagged levels and differences as instruments. We report Hansen/Sargan tests and check for instrument proliferation.

Robustness exercises include: Variance Inflation Factor (VIF) to confirm the multicollinearity (O’Brien, 2021), pooled OLS, random effects, difference GMM, and alternative operationalizations of policy and innovation (binary, index, counts). We test for serial correlation (Arellano-Bond tests), heteroskedasticity (cluster-robust standard errors at the firm level), and cross-sectional dependence. Sensitivity to sample composition is tested by estimating models on subsamples by subsector and firm size.

The empirical design is grounded in two complementary theoretical lenses. First, policy instruments operate through incentive-based and regulatory channels to alter firm cost-benefit calculations and resource allocation. Second, the resource-based view and organizational learning perspectives explain how internal capabilities, training, and technology mediate policy effects. The chosen proxies thus match theoretical mechanisms: Policy_it captures external incentives, Capacity_it captures human capital and organizational readiness, and Innovation_it captures the technological capability that converts incentives into measurable energy savings.

All firm-level survey data were collected with informed consent and anonymized prior to analysis. Data use complied with the NBS and ministry data-sharing agreements.

Table 1 presents summary statistics for the variables used in the analysis. The mean energy intensity of 0.485 (SD = 0.137) shows moderate variation across firms, reflecting the heterogeneity in industrial processes and energy management. The sample includes 150 firms: 48 percent from manufacturing, 32 percent from agro-processing, and 20 percent from the chemical sector. Most are medium and large enterprises, with an average workforce of 473 employees. These sectors together account for the highest industrial energy demand in Nigeria, making them a representative base for assessing policy impact (Akinola; Ojo; Oladele, 2022).

The binary policy incentive variable (mean = 0.46) suggests that nearly half of the firms benefit from energy-efficiency schemes such as tax rebates or technology grants. Technology adoption scores (mean = 5.18 on a 0–10 scale) point to moderate uptake of efficient technologies. Firms in the manufacturing and chemical sectors recorded higher adoption levels, consistent with their greater exposure to energy audits under national programs (NESP, 2023). Mean energy cost (0.12 per energy unit) indicates significant operational expenditure pressures, particularly for firms reliant on self-generated power.

The pre-estimation diagnostic (Table 2) shows mean VIF = 1.355, confirming the absence of multicollinearity (Li; Zhao; Wang, 2020). The data thus provide a sound basis for estimation.

Results from the fixed effects panel regression (Table 3) reveal that policy incentives significantly reduce industrial energy intensity (β = -0.121, p < 0.001). Firms with access to incentives use about 12 percent less energy per unit of output. It supports prior evidence that targeted incentives correct market failures related to information gaps and financing constraints (Jaffe; Stavins, 2020).

Technology adoption also shows a strong negative relationship with energy intensity (β = -0.033, p < 0.001). Firms integrating efficient machinery, process optimization, or digital monitoring systems demonstrate better energy performance. This effect aligns with the expectation that modern technologies enhance process control and reduce wastage (Karekezi; Kithyoma, 2021; Onyeji, Okereke; Nzeadibe, 2023).

Firm size, however, exerts a small positive influence on energy intensity (β = 0.008, p = 0.047). Larger firms, especially those operating in multi-product lines or continuous process systems, may face higher absolute energy use due to production scale and equipment age (Chen; Zhu, 2022). This result underscores the need for differentiated energy management strategies across firm sizes.

The positive and significant coefficient for energy cost (β = 0.275, p < 0.001) warrants careful interpretation. While higher energy prices typically encourage conservation, many Nigerian firms, especially in manufacturing, depend heavily on diesel or gas generators due to unreliable grid supply. It raises total energy expenditure even when physical consumption remains high. Moreover, smaller firms often lack the capital to invest in efficient technologies or energy audits, causing cost increases without proportional efficiency gains (Ike; Ogundipe; Balogun, 2023). In this context, high energy costs reflect infrastructural and market inefficiencies rather than deliberate overuse.

The model explains 42.3 percent of within-firm variation in energy intensity, a reasonable level for applied industrial energy studies (Chen; Li; Zhou, 2023).

As shown in Table 4, incorporating an interaction between policy incentive and firm size slightly reduces the policy effect (β = -0.101) but raises explanatory power (R² = 0.435). It suggests that policy benefits are less pronounced for larger firms, possibly because they already operate formal energy systems or face bureaucratic barriers to accessing incentives (Olusegun; Adeniyi, 2023). Adding a lagged dependent variable (Model 3) improves model fit (R² = 0.448), confirming gradual adjustments in energy management over time (Wang; Feng, 2022).

Robustness checks in Table 5 further support the findings. The negative coefficient for policy incentives remains significant across clustered SE, random effects, and IV estimations. The IV coefficient (-0.138) strengthens the conclusion that policy participation causally improves energy efficiency, even after correcting for possible self-selection bias (Moyo; Nwokolo, 2024).

Diagnostic tests confirm the appropriateness of the fixed-effects specification. The Hausman test rejects random effects, validating the control for firm-specific heterogeneity (Baltagi, 2021). Heteroskedasticity and autocorrelation corrections through robust clustering ensure the reliability of inference.

The results demonstrate that incentive-based policies meaningfully enhance industrial energy efficiency, but their effects vary by firm size and sectoral structure. Persistent high energy costs highlight the need to address systemic supply constraints alongside fiscal incentives. Strengthening capacity-building programs and ensuring equitable access to incentive schemes would allow smaller firms to benefit more fully. Integrating policy incentives with infrastructure reliability and technology financing could generate deeper and more sustained efficiency gains across Nigeria’s industrial landscape.

This study extends existing scholarship on industrial energy efficiency by linking policy incentives, capacity-building mechanisms, and firm-level technological adoption within a unified empirical framework. Previous studies have often examined these elements in isolation or emphasized macroeconomic determinants of energy performance (Jaffe; Stavins, 2020; Karekezi; Kithyoma, 2021). By employing firm-level panel data from Nigeria’s key energy-intensive industries, this research demonstrates how the interaction between government incentives and internal capacity factors shapes energy intensity outcomes.

Theoretically, the study contributes to institutional and behavioral models of industrial energy efficiency by providing evidence that external policy incentives alone are insufficient unless accompanied by organizational learning and technological readiness. The significant interaction effects found in the extended model empirically validate the complementarity between policy structure and firm capability, an aspect often theorized but rarely tested in developing-country contexts. It advances the conceptual understanding of policy effectiveness in energy economics and offers a framework adaptable to other emerging economies facing similar industrial energy challenges.

From a managerial standpoint, the findings underscore the need for firms to integrate energy management practices into their operational strategies. Managers should not rely solely on government incentives but pair them with internal audits, staff training, and technology upgrades. Firms that adopt digital monitoring tools and build technical expertise achieve lower energy intensity, suggesting that sustained efficiency requires long-term organizational commitment rather than episodic compliance (Akinola; Adeoye; Bello, 2024).

For policymakers, the results show that incentive design must go beyond financial inducements. Effective policy should incorporate monitoring, transparency mechanisms, and technical support structures to ensure equitable access, particularly for small and medium enterprises. The positive coefficient on energy cost reflects persistent structural inefficiencies in energy supply; thus, policy reform must simultaneously address infrastructure reliability and financing barriers. Partnerships between government and industry associations can improve program visibility and enable collective investment in shared energy management systems.

Several limitations should be acknowledged. First, while the panel dataset covers a diverse range of firms, it does not fully capture the informal industrial segment where energy efficiency practices are least documented. Future studies should include informal or semi-formal enterprises to extend the representativeness of findings. Second, the study relies on proxies such as dummy variables for policy incentives and capacity-building participation, which may simplify the underlying variations in program design and implementation quality. More granular policy-level data would enable deeper causal inference.

Third, despite using instrumental variables and GMM estimators to address endogeneity, the results remain limited by the availability of suitable instruments. Future research could combine quantitative analysis with qualitative case studies to provide richer insight into behavioral and institutional dynamics driving energy efficiency adoption. Additionally, cross-country comparative analyses would help test the robustness of the theoretical model in different policy and infrastructural environments.