Business Strategies for Managing Non-Renewable Energy Dynamics in Saudi Arabia’s Manufacturing Sector

Abstract

Understanding the asymmetric relationship between manufacturing output and non-renewable energy consumption is critical for formulating sustainable economic policies, particularly in energy-dependent economies like Saudi Arabia (KSA). This study has two aims. First, it examines how the KSA’s manufacturing sector responds to different energy sources, emphasising non-renewable energy—unlike previous studies that primarily examined general economic growth. Second, it investigates the asymmetric impact of non-renewable energy shocks on manufacturing output. Using yearly data from 1990 to 2022, this study finds that positive shocks to non-renewable energy significantly enhance manufacturing output in both the short and long run, driven by the sector’s reliance on cheap fossil fuels. On the contrary, negative shocks disrupt supply chains, increase energy costs, and reduce output over the same periods. In addition, this study reveals that renewable energy negatively affects manufacturing output due to transition costs and operational inefficiencies. However, gross fixed capital accumulation positively affects industrial production. These findings highlight the need for strategic investments in renewable energy infrastructure to mitigate the negative impacts of non-renewable energy disruptions, enhancing Saudi Arabia’s long-term economic stability. This study also underscores the importance of integrating sustainable development goals (SDGs) into policy frameworks to ensure a balanced and sustainable energy transition.

1. Introduction

Energy is crucial to economic development and progress [1,2,3,4]. Numerous studies have demonstrated the significant impact of energy consumption on economic growth. Nonetheless, energy consumption does not uniformly influence the overall economy across all sectors and countries, resulting in various hypotheses ([5], for a detailed description of the four hypotheses regarding the relationship between energy and growth). There is an association between economic growth and energy use. The diverse effects of energy consumption can be attributed to several factors. Primarily, energy may significantly impact various components of economic growth more than the overall growth process. Certain studies have demonstrated its impact on variables, including manufacturing outputs, industrial production, employment, financial development, and carbon emissions [6,7,8,9,10,11].

Furthermore, not every economic sector responds uniformly to energy consumption, as their levels of output and energy dependence vary significantly [12,13,14]. Aggregating data across sectors may obscure sector-specific dynamics, particularly in understanding how energy use affects manufacturing output [15]. The effect of energy consumption on growth might not reflect the unique characteristics of each sector, which could produce biased results because of the aggregation [15,16]. Whether or not dormant sectors consume more/less energy, empirical findings suggest that energy use does not significantly affect growth [17]. The sectors that are not directly engaged in economic activity may excessively consume resources, thereby burdening sectors that are actively involved [12,13]. These arguments suggest that it is sufficient to investigate the impact of energy consumption on the growth components to support precise policy formulation.

This study examines how energy use affects the production of the Kingdom of Saudi Arabia (KSA)’s manufacturing sector. Most of the existing literature focuses on the relationship between energy consumption and overall economic growth [8,18,19,20,21]; however, studies that focus on manufacturing output are limited, especially for the KSA [22]. The manufacturing sector is one of the most significant contributors to the economy, both in terms of economic activity and job creation [12,14]. This sector represents a crucial element of the economy, contributing 14.79% of the Gross Domestic Product (GDP) in 2023 (World Bank). Furthermore, it generally consumes more energy than any other sector does [6,14]. Given the significance of this sector, it is critical to investigate how the manufacturing sector responds to energy use, as not all industries are directly linked. For example, while the residential sector indirectly influences aggregate demands, it does not directly increase industrial productivity. In the same vein, sectors such as public administration or financial services may show relatively dormant responses to energy fluctuations, given their lower energy dependence compared to energy-intensive sectors like construction or transport [23]. Non-renewable energy sources—including coal, oil, and gas—have historically powered Saudi Arabia’s industrial and economic expansion, particularly in manufacturing and transportation [24]. Hence, our first contribution assesses the specific impact of non-renewable energy use on the output of the KSA’s manufacturing sector.

We further augment the literature by evaluating the effects of asymmetry (shocks). The complex nature of economic systems and diverse economic crises substantially affect energy consumption trends [25]. These variations often cause nonlinear changes in energy consumption, and disregarding nonlinearity in estimating the energy consumption–output relationship may lead to inaccurate estimations [26]. We employed the nonlinear Autoregressive Distributed Lag (NARDL) model introduced by Shin Shin, Yu [27]. This model divides the independent variable into partial sums (positive and negative), making it a nonlinear form of a symmetric Autoregressive Distributed Lag (ARDL) model. The theoretical rationale lies in its ability to simultaneously estimate both short- and long-run asymmetric effects while accommodating the mixed-order integration of variables, a frequent characteristic of macroeconomics time series [10,28]. Based on the gap, the current study has two objectives: First, it looks at how the manufacturing sector in the KSA reacts to various energy sources, with a focus on non-renewable energy, in contrast to earlier studies that mostly looked at overall economic growth. Second, it investigates how non-renewable energy shocks affect manufacturing output asymmetrically. We employed Zhang, Li [29]’s nonlinear Autoregressive Distributed Lag (NARDL) model. This model, which is a nonlinear variant of a symmetric Autoregressive Distributed Lag (ARDL) model, splits the independent variable into partial sums (positive and negative). Its capacity to estimate short- and long-term asymmetric impacts concurrently while considering mixed-order variable integration, a common feature of macroeconomic time series, provides a theoretical justification [10,28].

This study provides three key theoretical contributions that enhance the understanding of energy-manufacturing dynamics in the KSA. First, it offers a sector-specific examination of energy and economic activity. Given the growing global emphasis on sustainable energy practices, it is essential to understand the complex effects of energy, especially non-renewable energy, on manufacturing output to guide energy policymakers and successfully integrate strategies needed for future development. Secondly, it provides an asymmetry analysis by quantifying how positive and negative non-renewable energy shocks differentially impact manufacturing output, a critical gap given the sector’s dependence on fossil fuels [30]. These empirical findings respond to policymakers’ imperative to anticipate industrial volatility in the context of energy price variations, especially concerning the diversification objectives of KSA Vision 2030. Third, this study creates a dual-energy integration framework that goes beyond separate analyses of renewable and non-renewable sources. It gives manufacturers a choice to find the best balance between cost, stability, and sustainability, which is missing in prior research [31]. By filling in these gaps, this study not only supports Saudi Arabia’s industrial strategy, which heavily depends on oil, but it also gives emerging economies a model they can use for economic diversification.

This paper is divided as follows: in Section 2, the literature review and the background of the topic are discussed, followed by the study methodology in Section 3. Section 4 is focused on this study’s empirical results, while the discussion and implications are discussed in Section 5. Finally, this study is concluded in Section 6.

2. Literature Review

The relationship between energy use and economic output has been studied extensively; however, analyses of manufacturing and market asymmetries remain relatively underexplored [6,13,21,32,33,34]. Most existing studies concentrate on the overall economic output, frequently overlooking the distinctive dynamics of specific sectors and their vulnerability to external shocks [20,21,35]. These deficiencies in the literature underscore the necessity for more sophisticated methodologies, particularly in understanding the asymmetric effects of energy consumption on economic performance [36].

Investigations into the asymmetric relationship between energy consumption and output have received limited attention in the literature within the energy field [37,38]. For example, Wang, Bui [39] and Dogan, Altinoz [40] investigate the asymmetric impacts of renewable energy consumption, revealing that the effects exhibit substantial variation across quantiles within OECD countries. Positive effects were identified at lower quantiles, whereas detrimental effects were predominant at higher quantiles. Similarly, Hatemi-J and Uddin [41] conducted asymmetric causality tests. They found that while positive shocks had little or no impact on output, adverse shocks to the total energy consumption had a significant effect. Later research conducted by Lin and Moubarak [42] used nonlinear causality tests, such as Hiemstra–Jones and Kyrtsou–Labys, to examine sectoral relationships. These studies highlight unidirectional and bidirectional causal links in industries, including transportation and electric power. However, such approaches frequently fail to address mixed data stationarity, capture structural changes, and produce accurate short- and long-term estimates [43].

Prior research has broadly analysed energy use and economic growth without distinguishing between sector-specific responses [18,35,44]. While recent research highlights that energy shocks affect sectors differently depending on energy intensity [14]. In the Saudi Arabia context, industrial sectors—particularly petrochemicals, metals, and fertilisers—show a higher vulnerability to energy price and supply fluctuations, while sectors such as finance and public services remain relatively insulated. This research builds on such distinctions by focusing on the manufacturing sector’s asymmetric responses to non-renewable energy shocks [32].

Some studies have examined the relationship between energy and growth in the KSA using the nonlinear ARDL model [18,21,32,45]. They primarily assess aggregate-level responses. These studies confirm that the KSA shows an asymmetrical energy–growth dynamic; however, they do not investigate whether this asymmetry extends to specific sectors—particularly the manufacturing sector, which is among the most energy-intensive in the country. Moreover, they overlook whether non-renewable energy use has a differential impact on manufacturing output, despite this sector’s reliance on fossil fuels for both energy and feedstock inputs.

The NARDL framework is especially suitable for our examination of the energy–manufacturing nexus in the KSA for three convincing reasons. First, as demonstrated in Adekoya, Ogunnusi [10], partial sum decomposition helps to obtain a unique measurement of the response of manufacturing production to increases versus declines in energy prices, a crucial differentiation considering the KSA’s dual capacity as both an energy provider and consumer. Second, the model’s resilience to potential endogeneity issues [27,46] guarantees reliable parameter estimates despite the bidirectional relationship between industrial production and energy demand. Third, empirical foundations have demonstrated NARDL’s usefulness in energy economics research, especially in scenarios where supply shocks and price volatility induce asymmetric sectoral effects [27,28,47,48].

According to Shaari, Lee [2] study, which focuses on the relationship between renewable energy, CO₂ emissions, and growth in developing countries, renewable energy positively influences economic progress, whereas CO₂ emissions exert a detrimental effect [49]. Consequently, they have promoted the integration of additional renewable energy into the infrastructure to enhance development and environmental protection. Additionally, Sharif, Raza [50] used panel data from 74 different countries from 1990 to 2015 to investigate the relationship between CO₂ emissions, non-renewable energy usage, and renewable energy. Their findings also showed that non-renewable energy use has a positive impact on environmental degradation, whereas renewable energy has a negative impact and contributes to pollution reduction. Likewise, economic expansion has a negative impact on ecological degradation. Ike, Usman [51] looked at carbon emissions, cost-effective electricity, and the utilisation of renewable energy in the G7. Their findings show that although trading significantly reduces CO₂ emissions, it has a negative impact on the price of oil and renewable energy.

The nonlinear autoregressive distributed lag (NARDL) model has emerged as a superior tool to address these limitations. This model decomposes energy consumption into positive and negative shocks, enabling the simultaneous estimation of both short- and long-run effects while accommodating mixed data integration orders [21,32,33]. It also incorporates structural breaks to address external shocks and endogeneity bias. Studies by Çıtak, Uslu [52] and others have utilised NARDL to capture sectoral dynamics in countries such as the U.S. and the G6. However, these analyses remain aggregated, with limited attention paid to sector-specific manufacturing impacts.

3. Data and Methodology

This study used annual data from the World Development Indicators, World Bank, covering 1990–2022. Moreover, manufacturing value added (MVA) in (constant USD) per capita is used for manufacturing output. For non-renewable energy (NRE), which comes from fossil fuels, the unit of measurement is a million kW, which is also per capita. The analysis uses renewable energy (RE) expressed also in per capita. Also, gross fixed capital formation (GFC) in (current USD) per capita is used. Additionally, we used the logarithm of each series in the analyses to overcome the inflated variances.

Analytical Strategy

To provide dependable and accurate results, the data-generating features of the series under investigation must be established before applying time series methods. We used two-unit root tests to see if the data were steady without structural breakdowns. In both experiments, the presence of a unit root served as the null hypothesis for stationarity. The findings have an impact on the estimate and model specification of the Autoregressive Distributed Lag (ARDL) regression used in this study [53]. We utilised Dickey [54] and Phillips and Perron [55] tests at the level and first difference, respectively; the null hypothesis was confirmed by assessing the stationarity of all variables (MO, NRE, RE, GFC) and intercepts.

Table 1. Unit root tests.

	Augmented Dickey–Fuller		Phillips–Perron
Variables	Intercept	Trend and Intercept	Intercept	Trend and Intercept
Manufacturing Output	2.577	−1.514	3.18	1.12
	(−1.000)	(0.802)	(1.000)	(0.9990)
(Manufacturing Output)	−4.816	−6.197	−3.906	−4.428
	(0.000)	(0.000)	(0.012)	(0.001)
Non-Renewable Energy_POS	0.572	−2.556	0.4593	−2.629
	(0.986)	(0.301)	(0.982)	(0.270)
Δ (Non-Renewable Energy_POS)	−4.684	−4.667	−4.688	−4.637
	(0.000)	(0.0041)	(0.000)	(0.004)
Non-Renewable Energy_NEG	−0.345	−1.723	−0.392	−1.891
	(0.906)	(0.716)	(0.898)	(0.635)
Δ (Non-Renewable Energy_NEG)	−5.074	−5.073	−5.107	−5.897
	(0.000)	(0.001)	(0.000)	(0.000)
Renewable Energy	−3.052	−3.152	−10.38	−7.941
	(0.041)	(0.031)	(0.000)	(0.000)
Δ (Renewable Energy)	−3.457	−4.119	−3.518	−4.798
	(0.050)	(0.015)	(0.014)	(0.002)
Gross Fixed Capital Formation	0.485	−2.8	0.499	−1.725
	(0.983)	(0.210)	(0.984)	(0.716)
Δ (Gross Fixed Capital Formation)	−2.698	−6.829	−6.837	−6.929
	(0.068)	(0.000)	(0.000)	(0.000)

Note: the values in parentheses show probability values. Source (s): Eviews 10. Δ refers to the difference operator.

summarises the findings of the unit root analyses with the intercept and trend for the levels and first differences. Except for RE, which is stationary at the level, the other variables are stationary at the first difference. Therefore, we can conclude that the order of the combination of all nominated variables is one I(1), except for RE, so the ARDL approach is appropriate.

After confirming stationarity, we used the Akaike Information Criterion (AIC) to determine the optimal lag structure of each variable, using a maximum lag length of 4 as a benchmark and selecting the specification that minimised the AIC value, and to ensure that there was no serial correlation in the residuals. The identified optimal lag order was then used in the NARDL model estimation.

Now, we move on to the examination of the asymmetric relationship between manufacturing output, non-renewable energy, renewable energy, and fixed capital formation in the KSA. The NARDL procedure calculates the asymmetric long-term and short-run effects of the NRE variable as recommended by Adebayo and Samour [56]. This approach is chosen for two reasons. First, the asymmetric nature of the series is not contemplated using the ARDL technique [57]. Second, these regression models only calculate linear associations between variables and do not consider the variables’ nonlinear relationships. The ARDL specification by Pesaran, Shin [58] was expanded by Shin, Yu [27] to incorporate asymmetric ARDL cointegration owing to the nonlinearity of the variables. Scholars recommend that this technique be used to identify other short- and long-run asymmetries [59,60]. Hence, we propose the following equation:

$$MO=f\left({NRE}^{+},N{RE}^{-},RE,GFC\right)$$

(1)

The nonlinear model is given as

$${MO}_t={\alpha }_0+{\alpha }_1{\left(NRE\right)}_t^{+}+{\alpha }_2{\left(NRE\right)}_t^{-}+{\alpha }_3{\left(RE\right)}_t+{{\alpha }_4\left(GFC\right)}_t+{\epsilon }_t$$

(2)

where MO refers to manufacturing output, NRE refers to non-renewable energy, RE refers to renewable energy consumption, and GFC refers to gross fixed capital construction. ${\alpha }_0$, ${\alpha }_1$, ${\alpha }_3$,${ \mathrm{and} \alpha }_4$ are parameters to be estimated, whereas N${RE}^{+}$, and ${NRE}^{-}$ are partial sums of positive and negative changes in non-renewable energy that separate the effects of positive and negative changes on manufacturing output.

The positive and negative changes in NRE are structured as follows:

$${NRE}_t^{+}=\sum _{K=1}^t{\Delta NRE}_j^{+}=\sum _{K=1}^t{\mathrm{m}\mathrm{a}\mathrm{x}(\Delta NRE}_j,0)$$

(3)

$${NRE}_t^{-}=-\sum _{K=1}^t{\Delta NRE}_j^{-}=-\sum _{K=1}^t{\mathrm{m}\mathrm{i}\mathrm{n}(\Delta NRE}_j,0)$$

(4)

To examine the asymmetric cointegration among MO, N${RE}^{+}$,${ NRE}^{-}$, RE, and GFC, we incorporate positive and negative changes in the NRE variable in the linear ARDL model established by Pesaran, Shin [58]. NARDL is obtained as follows:

Equations (3) and (4) in the nonlinear ARDL framework are as follows:

$${\Delta MO}_t={\beta }_0+\sum _{i=1}^p{\beta }_1{\Delta \left(N{RE}^{+}\right)}_{t-i}+\sum _{i=1}^p{\beta }_2{\Delta \left({NRE}^{-}\right)}_{t-i}+\sum _{i=1}^p{\beta }_3{\Delta \left(RE\right)}_{t-i}+\sum _{i=1}^p{\beta }_4{\Delta \left(GFC\right)}_{t-i}+{\lambda }_1{MO}_{t-1}+{\lambda }_2{\left(N{RE}^{+}\right)}_t+{\lambda }_3{\left({NRE}^{-}\right)}_t+{{\lambda }_4{\left(RE\right)}_t+{\lambda }_5{\left(GFC\right)}_t+\eta ECT}_{t-1}+{\epsilon }_t$$

(5)

${\beta }_0$ represents the intercept component, Δ represents the difference operator, p is the optimal number of lags for each variable, and ${\epsilon }_t$ is the normally distributed error term. The first half of the equation demonstrates the short-run dynamics, while the second component represents the long-run dynamics. Specifically, ${\beta }_1$, ${\beta }_2$, and ${\beta }_3$, and represent the dynamics of the short run, whereas ${\lambda }_2$, ${\lambda }_3$, and ${\lambda }_4$, and characterise the long-run dynamics of the model. The magnitude of η shows how much the proportion of disequilibrium adjusts each year and how quickly the equilibrium is restored if it diverges. Lags were selected using the Akaike information criterion (AIC).

The following hypothesis can be used to test the long-run relationship:

$$H_0: {\lambda }_2={\lambda }_3={\lambda }_4={\lambda }_5=0$$

$$H_1: {\lambda }_2\ne {\lambda }_3\ne {\lambda }_4\ne {\lambda }_5\ne 0$$

Equation (5) is estimated using the ordinary least-squares method.

4. Empirical Results

The long-run relationship is based on optimum lags [61]. Using too many or too few lags might result in the loss of crucial information or in an estimation that is not acceptable [62]. Considering the importance of the optimum lags, we chose four lags as the ideal value, adhering to the AIC. We eliminated every irrelevant lagged regressor following the general-to-specific strategy, since researchers [63] have proposed that doing so was essential because of the possibility of irrelevant lagged regressors creating noise in active multipliers.

Table 2. NARDL bound test results.

Signif.	F-Statistic	Upper Bound	Lower Bound	Result
10%	10.098	3.01	1.90	Co-integration
5%		3.48	2.26
2.50%		3.90	2.62
1%		4.44	3.07

Note: H0 = No Co-integration. The critical values are based on [64] a small sample size analysis. Source (s): Eviews 10.

presents the results of the bound test. The asymmetric cointegration among MO, N${RE}^{+}$,${ NRE}^{-}$, RE, and GFC are confirmed by the F-statistic value of 10.098, which is above the upper boundary critical value at the 1% significance level. Now, we can move to an asymmetric ARDL estimation.

Moreover, the long-run asymmetric in the NRE is then tested by H0, ${\lambda }_2$= ${\lambda }_3$, against the alternative hypothesis H1, ${\lambda }_2$≠ ${\lambda }_3$, by using the Wald test (see

Table 3. Asymmetric result (Wald test).

Test Statistic	Value	df	Probability
t-statistic	2.884	20	0.009
F-statistic	8.319	(1, 20)	0.009
Chi-square	8.319	1	0.003

Source (s): Eviews 10.

).

The long and short runs are presented in

Table 4. Asymmetric long-run results.

No Constant and No Trend
Variable	Coefficient	t-Statistic	Prob.
NRE_POS	0.261 *	2.922	0.008
NRE_NEG	−0.602 **	−1.98	0.061
RE	−0.174 ***	−4.316	0.000
GFC	0.621 ***	9.254	0.000

Note: ***, **, and * denote significance levels of 1%, 5%, and 10%, respectively. Source (s): Eviews 10.

and

Table 5. Asymmetric short-run results.

No Constant and No Trend
Variable	Coefficient	t-Statistic	Prob.
MO (−1)	−0.947 ***	−5.668	0.000
Δ (NRE_POS)	0.247 ***	2.352	0.029
Δ (NRE_NEG)	−0.571 *	−1.896	0.072
Δ (RE)	−3.024 ***	−2.771	0.011
Δ (RE (−1))	4.251 ***	3.871	0.000
Δ (GFC)	0.706 ***	8.039	0.000
Δ (GFC (−1))	−0.069	−0.775	0.447
Δ (GFC (−2))	−0.342 ***	−3.382	0.003
ECT (−1)	−0.947 ***	−7.784	0.000
R-squared	0.862
Adjusted R-squared	0.834
Durbin–Watson stat	2.016

Note: Δ is the first difference operator; ***, and * denote significance levels of 1%, 5%, and 10%, respectively. Source (s): Eviews 10.

. All identified variables were highly significant with the expected signs. In the long run, a positive shock to non-renewable energy significantly impacts manufacturing output, with a coefficient of 0.261, implying that a 1% increase in non-renewable use boosts manufacturing output by 0.261%. Similarly, in the short run, a 1% increase in the positive shock results in a 0.247% rise in output, and the effect is highly significant. In contrast, a negative shock has a significantly negative influence on manufacturing output, both in the short run and long run. The negative coefficient (−0.602) in the long run shows that a 1% increase in shock to non-renewable energy reduces manufacturing output by 0.602%, and in the short run, it reduces output by 0.571%. The results suggest that the positive shocks to non-renewable energy may raise manufacturing output by improving energy supply or efficiency, which can boost production capacity. Negative shocks to non-renewable energy can reduce manufacturing output in the short and long term due to energy supply disruptions or higher costs. These studies demonstrate the sensitivity of the manufacturing output to energy availability and cost.

Renewable energy reduces manufacturing output by 0.174% in the long run for each 1% increase, a result that is statistically significant at the 1% level. This suggests that renewable energy adoption lowers the output due to inefficiency or resource constraints. It has a short-term negative effect of −3.024% for a 1% increase and is statistically significant at 1%. This implies that a short-term increase in renewable energy may have serious negative effects due to adaptation costs or infrastructure scaling issues.

In contrast, gross fixed capital formation demonstrates a positive association, with a coefficient of 0.621, which is statistically significant at the 1% level. This signifies that increased investments in capital goods, such as infrastructure and physical assets, lead to enhanced manufacturing production, highlighting the advantageous economic impacts of long-term capital investments. The short-term impact is also positive, indicated by a coefficient of 0.706, which is statistically significant at the 1% level. This indicates that an increase in GFC, representing capital investment in the economy, also has an immediate positive impact on production.

We also tested for regression issues using the Jarque–Bera test for residual normality, the Breusch–Pagan–Godfrey test for heteroscedasticity, the LM test for serial correlation, the Ramsey Reset test for model specification, and the Cumulative Sum (CUSUM) and Cumulative Sum Square (CUSUMSQ) tests for model stability (

Table 6. Diagnostic analysis.

Diagnostic Tests	Serial Correlation	Heteroscedasticity	Normality	Model Specification
	F-statistic	F-statistic	Chi-square	F-statistic
Breusch–Pagan–Godfrey	1.327
	(0.285)
LM test		3.1008
		0.178
Jarque–Bera			1.418
			(0.492)
Ramsey RESET test				0.462
				(0.648)

Note: The values in parentheses show prob. values. Source (s): Eviews 10.

). Before implementing the asymmetric ARDL model, several diagnostic tests were performed to confirm that none of the concerns existed. The values for the LM and Breusch–Pagan–Godfrey tests are 0.178 and 0.285, respectively, indicating that our model has no serial association or heteroscedasticity concerns.

Additionally, the Jarque–Bera test was used to ensure residual normalcy. Our model is accurately represented by studying Ramsey RESET’s statistical insignificance, which is 0.648. To ensure that the long-run coefficients remain constant, Brown, Durbin [65] suggest looking for CUSUM and CUSUMSQ. Figure 1 depicts the plots of the CUSUM and CUSUMSQ values at the 5% level, which fall within the critical boundaries.

5. Discussion

The goal of this study was to empirically investigate how asymmetric shocks in non-renewable energy use affect manufacturing production in Saudi Arabia. The data demonstrate diverse asymmetric reactions, with positive shocks increasing manufacturing production and negative shocks having long-term negative impacts. Several contextual considerations explain these dynamics. The industrial sector is fundamentally reliant on fossil fuels, notably in energy-intensive businesses—including refining, fertiliser manufacturing, petrochemicals, metals, and cement manufacturing [10,14,66]. These sectors reflect the sector-specific relevance of energy dependence in contrast to relatively dormant sectors such as public administration or financial services.

Results demonstrate that when positive shocks occur—such as price reductions from government subsidies or supply increases—production costs decline significantly, enhancing sectoral output. The results are consistent with previous studies [67,68]; their results highlight that expenses can be decreased by a favourable shock to the fossil fuel energy sector, making Saudi manufacturing more competitive at home and abroad.

Furthermore, even if the KSA attempts to diversify its economy in line with Vision 2030 [69], fossil fuels remain essential, sustaining manufacturing growth and offering stability to expand other industries. Moreover, core manufacturing subsectors—such as chemicals, polymers, and fertiliser—which account for a substantial share of industrial output, requires fossil fuels not only for energy but as essential feedstock [70]. A positive shock in energy from fossil fuels can reduce the cost of feedstock and increase output and revenue in specific sectors. This can also trigger a broader increase in aggregate economic activity, raising the demand for manufactured goods [67,71].

Conversely, negative energy shocks—such as supply disruptions or price spikes—create compounding challenges for Saudi manufacturing. Energy-intensive sectors, such as petrochemicals, metals, and cement are especially vulnerable to such volatility. The manufacture of metals, cement, and petrochemicals are especially sensitive to fluctuations in the price of energy or disruptions in the supply of fossil fuels, which can result in increased production costs and a decline in competitiveness. These findings align with established studies demonstrating the fundamental connection between energy market stability and industrial performance [71,72], while specifically highlighting the operational expenses increase in response to rising fossil fuel prices or declining availability, which decreases output and efficiency. Manufacturing is dependent on fluctuations in energy prices and outside shocks due to the absence of diverse energy sources and the fact that renewable energy is still in its early stages of development.

Furthermore, a shortage of fossil fuels can cause disturbances in the supply chain, especially in sectors such as petrochemicals, which depend on fuel for transportation and raw materials. Production is delayed because of these interruptions, and the output is subsequently decreased. Previous studies revealed that energy shocks drastically reduce industrial productivity and production, especially in energy-dependent countries [34,66,73,74].

Regarding renewable energy, the results show a significant short- and long-term negative impact on manufacturing output in the KSA. A 1% increase in renewable energy leads to a 0.174% long-run and 3.024% short-run decrease in output. This result is consistent with previous research showing that renewable energy has a detrimental impact on growth [24]. High capital costs for infrastructure upgrades, the irregular nature of renewables like wind and solar, and technological and labour-related constraints in the manufacturing sector further exacerbate this issue [4,75].

Businesses would face financial hardship and decreased productivity owing to the significant initial capital investments needed to establish and upgrade infrastructure to transition to renewable energy. Additionally, energy-intensive sectors, such as metals and petrochemicals, are disrupted by the irregular nature of renewable energy sources, such as wind and solar, which lowers efficiency. Productivity increases are further hindered by Saudi Arabia’s manufacturing industry, which lacks the trained labour force and technology awareness needed to effectively incorporate renewables into industrial operations [4]. During the transition, inefficiencies were caused by technological difficulties in modifying fossil-fuel-based infrastructure to incorporate renewable energy. Finally, compared to fossil fuels, adopting renewable energy may not result in rapid output gains, which would delay the advantages and worsen the adverse effects of RE. This is evident from the coefficient of RE with a one-period lag of 4.251, which indicates a substantial positive effect that is highly significant.

Regarding gross fixed capital formation, a highly significant long- and short-run impact is detected, as expected [76]. According to the long-run coefficient of GFC, manufacturing output would grow by 0.621% for every 1% increase in GFC. However, this gain was 0.706% in the short run. Several essential aspects contribute to the positive and highly significant impact of gross fixed capital formation on manufacturing output. Numerous growth models, such as those of Solow [77] and Barro [78], have highlighted capital accumulation as a key factor in economic expansion. Productivity increases from capital expenditures on machinery, equipment, and technology, as highlighted in Solow’s Growth Model. Barro [78] also asserts that building infrastructure, such as transport and energy networks, is crucial because it lowers prices and encourages industry growth. Furthermore, as emphasised by Graham and Krugman [79] and de Long, Summers [80], investments in physical capital frequently result in the formation of industrial clusters, which have spillover effects that increase production and innovation across sectors. GFC helps businesses adopt best practices, save costs, and innovate, which not only improves short-run production capacity but also helps the manufacturing sector remain competitive over the long run [81,82]. Finally, the manufacturing value added deviates from its long-run equilibrium, and the significant and negative ECT coefficient (−0.947) indicates that about 94.7% of the disequilibrium is rectified in the subsequent time. Consequently, the system exhibits a strong and rapid mechanism for adjusting short-term disturbances to achieve long-term stability.

5.1. Practical Implications

This study has several implications for both practitioners and policymakers. First, in light of this study’s findings, decision-makers need to formulate environmental policies and develop strategies for achieving ecologically sustainable development goals (SDGs). For example, considering the positive association between fossil fuel energy and manufacturing value added in the KSA, policy recommendations should concentrate on maximising the advantages of this correlation while maintaining the objectives of sustainability and long-term economic diversification. Even if fossil fuels are now essential to manufacturing, the over-reliance on them can be considered risky in the long run, particularly as the world’s energy markets move towards renewable sources [4,6]. Second, the government must encourage the growth of non-energy-intensive businesses, including high-tech manufacturing, aerospace, and pharmaceuticals, to help diversify the industrial sector. This may be accomplished by developing various strategies, such as investing in infrastructure, providing targeted subsidies, and offering training courses concentrating on producing labour for these emerging industries.

Third, diversifying the KSA’s energy mix for the manufacturing sector is also advised to mitigate the negative impact of fossil fuel energy shocks on industrial output, for instance, by investing in large renewable energy infrastructure, such as solar and wind, to provide dependable and reasonably priced alternatives to fossil fuels. To encourage the industrial sector to transition to renewable energy, tax cuts or subsidies need to be offered to industries that integrate renewable energy into their operations. This strategy will increase resilience to future energy shocks by lowering dependency on fossil fuels. Manufacturing industries may become less prone to energy shocks by lowering costs and reducing their dependence on fossil fuels through increased energy efficiency. Laws and regulations encouraging the industrial sector to adopt energy-efficient practices and technologies need to be further developed.

Fourth, owing to their high energy consumption, specific industries such as petrochemicals and cement are particularly vulnerable to shocks from fossil fuels. As they move to more sustainable energy sources, the systemic risks related to negative fossil fuel shocks diminish. Developing sector-specific transition plans to gradually reduce reliance on fossil fuels, increasing energy efficiency, switching to natural gas instead of propane, which has a lower carbon footprint, and eventually switching to renewable energy sources might all be part of this.

Fifth, energy storage can mitigate the effects of an unstable energy supply, whether derived from fossil fuels or renewable sources. The government should consider funding large-scale energy storage projects like battery farms or pumped hydro storage as a buffer against energy shocks. This will enable production to continue even in the event of an interruption in the energy supply. Also, rapidly switching to renewable energy sources could lower industrial output in the KSA, as the country’s industry primarily relies on fossil fuels for production [83,84]. A phased transition strategy incorporating renewable and fossil fuels can reduce disruptions, as can enacting laws and adopting strategies that help businesses integrate renewable energy sources while allowing them to continue using fossil fuels when necessary.

Implementing these policies will position the KSA’s economy for long-term sustainable growth and mitigate the negative effects of renewable energy adoption on the industry. This will result in sustainable industrial growth and diversification, which aligns with Saudi Vision 2030 [85].

5.2. Limitations and Future Research

Although the current study provides theoretical and practical implications, it has some limitations. First, industry-specific energy consumption statistics in an industrial production model will provide more reliable insights than total energy consumption, which sometimes lacks details at the industrial level. Further, industry disaggregation is necessary to gain a deeper understanding of how disaggregated energy consumption affects industrial production. Second, by examining how non-renewable energy sources (such as coal and fossil fuels) contribute to industrial production, we may gain a better understanding of how they are used in each industry and find areas where sustainability and energy efficiency can be enhanced. Finally, this study does not conduct sensitivity analyses on crucial parameters such as renewable energy price changes and capital creation rates. Future research could use a Monte Carlo simulation to assess the robustness of the NRE_POS and NRE_NEG coefficients to oil price variations. This would improve the forecast accuracy of the model.

While this analysis provides the KSA’s asymmetric energy consumption–manufacturing production relationship, it may benefit from wider validation with other hydrocarbon-dependent economies. The lack of benchmarking against neighbours or other energy-abundant emerging countries makes it difficult to identify Saudi-specific patterns from regional trends. Asymmetric effects among economies with similar energy–industrial setups should be examined in future through cross-country comparisons, focusing on institutional characteristics and policy frameworks that may mitigate these interactions.

Accordingly, several avenues for future research can be followed to enhance our understanding of the relationship between energy use and industrial production, especially in the KSA. First, industry-specific energy consumption studies can focus on individual industries within the KSA, such as construction, petrochemicals, and manufacturing. This would entail acquiring and analysing data on energy consumption at the industry level to draw more focused insights that can inform tailored strategies for improving energy efficiency and sustainability. Second, using qualitative research methods, such as case studies or interviews, would aid in understanding how industry leaders perceive the challenges and opportunities associated with energy consumption and sustainability, as such studies would provide richer context-specific insights that could inform managerial practices and strategic planning [86,87]. Third, future research might investigate technological advancements and how upcoming technologies like smart grids, energy storage, and renewable energy integration can improve energy efficiency in specific industrial applications [88]. Finally, academics can highlight the best practices and successful case studies that could further serve as a model for other industries, fostering greater adoption of sustainable practices and ultimately facilitating the achievement of Saudi Vision 2030 [89].

By addressing the highlighted limitations and future research directions, scholars can significantly advance the understanding of energy consumption in industrial production and contribute to the formulation of practical sustainability and economic development strategies in the KSA and beyond.

6. Conclusions and Outlook

This study examines the asymmetric impact of non-renewable energy (NRE) on manufacturing output in the KSA by using yearly time series data from the World Bank for the period 1990–2022. An examination of the asymmetric relationship between them using a nonlinear autoregressive distributed lag (NARDL) model revealed how manufacturing output is influenced by positive and negative energy shocks. Positive shocks increase manufacturing output because the industry depends on fossil fuels and has low energy costs; on the other hand, negative shocks lower output by increasing energy prices and causing supply disruptions. However, because of the significant transition costs and operational difficulties associated with renewable energy, output is adversely influenced by it in both the short and long term, even though lag effects may indicate long-term advantages. Gross fixed capital formation (GFC) has a significant positive impact on manufacturing output. This demonstrates the importance of continuing to invest in infrastructure and technology to support industrial expansion.