Petroleum Consumption and Financial Development: Evidence from Selected EMEs: Panel ARDL-PMG Approach

Abstract

This paper examines the long-term and causal relationship between petroleum consumption and financial development in selected emerging market economies (EMEs) from 2000 to 2020. Using panel cointegration and an error correction model (ECM), the study captures both the short- and long-run dynamics of the petroleum–finance nexus while accounting for cross-country heterogeneity. The results show a significant long-run elasticity of petroleum consumption with respect to financial development, while the error correction term confirms robust convergence to equilibrium. In contrast, the short-run effects are insignificant, indicating that petroleum consumption does not immediately influence financial development. These findings highlight the need for robust energy policies that strengthen financial markets and support sustainable growth. Policymakers should prioritize infrastructure investments, strengthen financial linkages in the energy sector, and promote diversification to reduce the risks associated with petroleum dependence.

1. Introduction

Petroleum is a crucial component of energy consumption in emerging market economies (EMEs) across various sectors, including transportation, manufacturing, and trade [1]. Global trading exposes petroleum to price volatility, supply disruptions, and geopolitical risks [2]. These risks impact financial development through heightened investment uncertainty, rising production costs, and higher import bills, which strain foreign exchange reserves [3].

The nexus between petroleum consumption and financial development is critical for EMEs, given their rapid economic growth, increased energy demands, and evolving financial sectors. Advanced financial markets play a crucial role in channeling resources into petroleum exploration, refining, and distribution [4,5,6]. According to [7], economies with efficient financial markets experience productivity gains of 15%.

Petroleum demand has declined in advanced economies but continues to grow in EMEs, driven by population and economic expansion. Between 2011 and 2021, fossil fuel consumption in the EMEs increased by 25%, with China and India accounting for the most significant absolute increases [8]. By 2050, Central and South America, Africa, the Middle East, and Eurasia are expected to account for one-third of global oil demand. These shifts underscore the significance of understanding how petroleum consumption impacts financial development in economies that remain heavily reliant on fossil fuels.

While prior studies have explored energy–finance linkages [9], most have focused on aggregate energy consumption or electricity, overlooking the distinct characteristics of petroleum as a globally traded commodity. The study recognizes its structural importance for various sectors of the EME economy. Isolating petroleum within the energy finance nexus reduces aggregation bias and highlights its interaction with financial development. The study adopts a dual approach that focuses on both long- and short-run dynamics, providing a stronger basis for policy design.

Although industrialization and urbanization are widely recognized as structural drivers of both financial deepening and energy consumption, this study deliberately focuses on the direct relationship between petroleum and finance. This approach is motivated by two factors: first, the empirical objective of isolating the specific channel through which petroleum consumption affects financial development in EMEs, without conflating the relationship with broader macro-structural variables; and second, the absence of consistent and comparable panel data on industrialization and urbanization across the sample countries, which could introduce measurement bias. The study, therefore, examines petroleum consumption and financial development within a controlled macroeconomic context, selecting variables that best capture the relevant short- and long-run dynamics for the research question.

This study makes several contributions to the literature. First, it examines the short- and long-run relationship between petroleum consumption and financial development in EMEs using a panel ARDL–PMG framework, while also addressing potential endogeneity and omitted variables through 2 System GMM robustness checks. Second, it situates the empirical analysis within competing theoretical perspectives, including the energy-led growth hypothesis, resource curse theory, and Dutch disease, clarifying the testable implications for EMEs. Third, the findings have direct policy relevance for energy security, fossil fuel transition strategies, and the development of financial markets.

The remainder of the paper is structured as follows: Section 2 outlines the theoretical and empirical literature; Section 3 details the estimation methodology; Section 4 presents the results and analysis; Section 5 concludes with policy recommendations.

2. Literature Review

2.1. Theoretical Framework

This section explores key theoretical frameworks that have informed this study.

2.1.1. Energy-Led Growth (ELG) Hypothesis

Aligned with endogenous growth theory, this hypothesis emphasizes the role of energy as an input in the production function. Petroleum consumption supports long-term economic growth, facilitating capital accumulation, industrial production, and technological progress [10,11,12]. In energy-intensive EMEs, petroleum consumption stimulates demand for financial services, and the financial market evolves to mobilize capital for energy investments [13].

2.1.2. Financial Development-Energy Consumption Nexus

A well-developed financial sector is essential for energy development, as it channels capital into the energy sector. This theory supports the supply leading hypothesis [14]. It enables the development of instruments such as project finance, syndicated loans, energy bonds, and commodity derivatives, which enhance petroleum consumption [15,16,17].

2.1.3. Resource Curse Hypothesis

Resource-rich economies often grow more slowly and have weaker financial systems than resource-poor ones, due to crowding-out effects, overreliance on extractive industries, and weak institutions. Volatile commodity prices discourage the development of long-term financial instruments. The authors of [18] note that resource wealth does not guarantee growth; in fact, some African and Middle Eastern economies have experienced stagnation [19]. The resource curse may operate through unproductive public spending [20] and price instability [21], thereby undermining financial stability and credit flows [22].

2.1.4. The Dutch Disease Theory

This theory explains how large resource inflows can strengthen the domestic currency and reduce export competitiveness, negatively impacting non-resource sectors [23,24,25]. These dynamics weaken financial stability indicators such as banking resilience and capital market depth [26,27,28].

2.1.5. Institutional Theory

This theory emphasizes the importance of strong institutions, governance, regulatory frameworks, and institutional quality in shaping the nexus between petroleum consumption and financial development [29]. Strong institutions ensure transparency and lower transaction costs, thereby increasing investor confidence [30,31].

2.1.6. Financialization Perspective

The concept of financialization explains the dominant role of financial motives, markets, and institutions in shaping economic outcomes [32,33]. It considers both qualitative and quantitative aspects of financial development. Qualitative issues include an increase in speculative activities and short-term capital flows as the financial market becomes more complex. EMEs have witnessed rapid financialization through the channeling of resource rents into financial markets, impacting investment allocations, exchange rates, and fiscal stability [34]. Financialization magnifies EMEs’ economic vulnerabilities, exposing them to global oil shocks and volatile capital flows.

The study empirically tests theoretical linkages using panel cointegration and error correction models to capture both long-term equilibrium and short-term dynamics in EMEs.

2.1.7. Integrated Conceptual Model

Taken together, these theories outline three competing scenarios for EMEs:

• ELG-consistent outcome: Petroleum consumption enhances financial development via energy-driven growth.
• Resource curse/Dutch disease outcome: Petroleum dependence undermines financial development through volatility, rent-seeking, and structural distortions.
• Institutionally moderated outcome: The sign and magnitude of the petroleum–finance relationship depend on the quality of governance and the maturity of the financial market.

By employing a panel ARDL–PMG approach, this study aims not to prove one theory correct, but to assess which theoretical prediction aligns most closely with the observed long-run and short-run relationships in EMEs. This integration addresses the problem of treating theories in isolation, enabling a coherent interpretation of empirical findings, including counterintuitive results, such as the possibility that economic growth can suppress petroleum consumption over the long term (which may reflect structural transformation or efficiency gains rather than a spurious correlation).

2.2. Empirical Literature Review on the Petroleum Consumption and Financial Development Nexus

Empirical studies on the petroleum–finance nexus are diverse in scope, methodology, and findings. Some support the ELG hypothesis, showing positive relationships between petroleum and finance, while others align with predictions of the resource curse or Dutch disease.

2.2.1. Positive Long-Run Effects

Ref. [1], using Driscoll–Kraay fixed effects for 12 Eastern African countries, reports that petroleum consumption has a statistically significant positive impact on economic growth, indirectly supporting ELG predictions for financial development. Similarly, [2], who applied an ARDL model to Oman’s data (1978–2017), observed a positive and statistically significant long-run relationship between petroleum and finance. However, short-run effects are insignificant due to price volatility.

2.2.2. Mixed or Negative Findings

Ref. [3] examines the relationship between financial development and mineral resource rents in China and observes a positive long-run relationship, but insignificant short-run effects, suggesting that market volatility can temporarily mute the resource–finance linkages. Ref. [35] observes a positive long- and short-run petroleum–finance nexus in Malaysia with bidirectional causality, but their results also highlight sensitivity to oil price cycles.

2.2.3. Gaps in the Literature

There are three main literature gaps identified: First, there is an overreliance on aggregate energy measures that overlook the dynamics specific to petroleum. Second, narrow country samples and finally, methodological variation can lead to inconsistent results.

This study addresses these gaps by focusing explicitly on petroleum consumption, applying a panel ARDL–PMG approach across multiple EMEs, and interpreting results through the integrated theoretical framework outlined above. This approach enables the reconciliation of conflicting findings in prior research and provides a platform to explain both expected and anomalous results in a theoretically grounded manner.

Table 1. Summary of the Petroleum Consumption—Financial Development Nexus Empirical Findings.

Authors	Time	Countries	Methodology	Results
[1]	2000–2020	12 Eastern African Countries	Driscoll–Kraay Fixed Effects Model	Petroleum consumption has a significant impact on economic growth both in the short and long term, and economies that heavily rely on it are susceptible to fluctuations in global oil prices.
[2]	1978–2017	Oman	ARDL Model	Petroleum consumption has a significant positive impact on financial development in the long run, but an insignificant effect in the short run due to fluctuations in oil prices.
[36]	1990–2020	China	ARDL Model, Toda–Yamamoto test	Financial development and mineral resource rents have a positive and statistically significant long-term relationship, but the short-term relationship is negligible.
[35]	1990–2015	Malaysia	ARDL framework, Granger Causality test	Petroleum consumption has a positive impact on financial development in both the short and long run, indicating a bidirectional causal relationship.

Source: Author, 2025.

below presents a summary of the empirical findings from previous studies.

3. Data and Methodology

3.1. Data and Variables

The study examines the relationship between petroleum consumption and financial development with foreign direct investment (FDI) and economic growth as control variables. It utilizes data from the World Bank’s Global Financial Development Database (GFDD) and the United States Energy Information Administration (EIA) for the period 2000–2020.

Table 2. Definition of Variables and Data Sources.

Variable	Definition of Variables	Data Source	Expected Sign
Dependent Variable
FD	The Financial Development Index measures the breadth and depth of financial markets.	World Bank’s Global Financial Development Database (2023)	N/A
Independent Variables
PC	Petroleum consumption (millions of barrels/d)	International Energy Agency	+/−
EG	Real GDP per capita	World Development Indicators	+/−
FDI	Foreign Direct Investment	World Development Indicators	+

presents the variables and their measurement. These sources provide consistent, high-quality indicators for financial development and petroleum consumption. The balanced panel comprises 20 EMEs selected for three key criteria: first, petroleum dependence; second, complete and consistent data coverage; and third, regional diversity to capture different institutional and economic contexts. The period was selected because it encompasses major global crises (the 2008 financial crisis and the onset of COVID-19), commodity supercycles, and structural shifts in petroleum demand. Countries span South America (Argentina, Brazil, Chile, and Colombia), Asia (China, India, Indonesia, Iran, Malaysia, the Philippines, Thailand, and the UAE), Africa (Egypt, Kenya, Nigeria, Saudi Arabia, and South Africa), Europe/Eastern Europe (Hungary, Russia, and Turkey), and North America (Mexico). The sample size ensures sufficient observations for panel econometric analysis while prioritizing data quality over exhaustive coverage.

The exclusion of industrialization and urbanization is deliberate to isolate the effects of petroleum, although this may omit specific indirect channels. Although the data sources are robust, minor measurement inconsistencies may still occur. Results should be interpreted as applicable to the sampled EMEs rather than all emerging economies globally.

3.1.1. Descriptive Statistics

Table 3. Descriptive Statistics.

	Mean	Median	Minimum	Maximum	Std Dev.	Skewness	Kurtosis	Jarque-Bera	Obser
FD	0.42	0.42	0.09	0.74	0.14	0.08	2.39	6.90	419
PC	1513.04	717.32	50.06	14,432.72	2172.24	3.50	17.27	4413.54	419
EG	8264.37	6141.83	755.48	59,986.44	9618.28	3.51	13.50	4062.22	419
FDI	3.08	219	−40.09	106.60	7.81	7.76	93.34	146,682.60	419

Source: Author’s Analysis: Descriptive statistics are calculated on all available annual data for the 2000–2020 period. Notes: Std Dev. = Standard Deviation, Obser. = Number of observations, FD = Financial Development Index, PC = Petroleum Consumption (Mb/d), EG = GDP per Capita (Constant 2015 US$), FDI = Foreign Direct Investment (Net inflows as a % of GDP).

presents the descriptive statistics of the study sample, which includes emerging markets. The summary consists of key macroeconomic variables relevant to the study.

Table 3 presents the descriptive statistics of the variables in the study. The financial development index (FD) has a mean and median of 0.42, indicating a relatively symmetrical distribution. There is a moderate variation across the countries in the sample, ranging from a minimum of 0.09 to a maximum of 0.74. The findings suggest that financial development is relatively stable across the sampled countries. Petroleum consumption (PC) has a mean of 1513.04 million barrels/day and a median of 717.32 million barrels per day, indicating a right-skewed distribution, which suggests that some countries consume significantly more petroleum than others. There is a significant disparity in the consumption of petroleum among the sampled countries, ranging from 50.06 million barrels per day (Mb/d) to a maximum of 14,432.72 million barrels per day (Mb/d). The standard deviation of 2172.24 confirms this variability. Economic growth (EG) had a mean and median of 8264.37 and 6141.83, respectively, and the distribution is positively skewed.

3.1.2. Cross-Correlation Analysis

Table 4. Correlation Analysis.

Variables	FD	PC	EG	FDI
FD	1
PC	0.2947 ***	10,000
EG	−0.2951 ***	−0.0266	1
FDI	0.0977 **	−0.0564	−0.0788	1

t statistics in parentheses, ** p < 0.01, *** p < 0.001.

reports the correlation coefficients, which measure the strength of the relationship between the variables.

From Table 4, a positive and significant relationship exists between FD and ED, with a coefficient of 0.2947 ***, indicating that higher petroleum consumption is associated with greater financial market deepening. These findings bring out the notion that energy-intensive economies require robust financial systems to facilitate investments and resource allocation. The control variables exhibit mixed relationships, reinforcing the complicated interactions that involve macroeconomic stability, investment, and financial deepening.

3.2. Methodology

Figure 1 illustrates the empirical strategy employed in this study.

3.2.1. Panel Unit Root Test

Panel unit root test results (Appendix A,

Table A1. (a–d): Unit Root Tests.

Variable	No Trend	Intercept and Trend	Individual Effects	Decision
(a): Panel Unit Root Test Using the LLC
PC	−7.72546 ***	2.95045 ***	3.08854 ***	I (1)
FD	−12.5368 ***	−5.76010 ***	−7.01951 ***	I (1)
EG	−9.00243 ***	−14.1579 ***	−14.8932 ***	I (1)
FDI	−3.73481 ***	−3.84931 ***	−3.59057 ***	I (1)
(b): Panel unit root tests using IPS
PC	-	−1.54789 ***	−3.68186 ***	I (1)
EG	-	−7.32238 ***	−8.86368 ***	I (1)
FD	-	−6.80194 ***	−8.89608 ***	I (1)
FDI	-	−3.15388 ***	−4.64037 ***	I (1)
(c): Panel unit root testing using ADF—Fisher chi-square
PC	205.413 **	121.031 ***	155.718 ***	I (1)
EG	162.126 ***	119.625 ***	165.202 ***	I (1)
PC	131.662 ***	57.1864 ***	82.3936 ***	I (1)
FDI	58.5960 ***	69.1745 ***	95.5074 ***	I (1)
(d): Panel unit root testing via PP—Fisher chi-square
FD	342.526 ***	282.882 ***	338.690 ***	I (1)
EG	288.937 ***	183.140 ***	211.156 ***	I (1)
PC	236.900 ***	142.993 ***	169.306 ***	I (1)
FDI	62.4167 ***	99.6138 ***	125.058 ***	I (1)

*** and ** indicate that the null hypothesis of unit root tests is rejected at 1%, and 5%, respectively. All tests are based on first differences (except where indicated otherwise). The probabilities for all tests assume asymptotic normality, except for Fisher tests, which are computed using the asymptotic chi-square distribution. PC is petroleum and other liquids, FD is the financial development index, EG is gross domestic product per capita, and FDI is foreign direct investment. Source: Author’s compilation using Stata.

(a–d)) confirm that all variables are non-stationary at levels but become stationary after first differencing, indicating integration of order one, I(1). Traditional unit root tests lack power in panel data, particularly in small samples [37,38].

The variables are integrated of order one, I(1), across all four testing methodologies. The test reinforces the validity of results by ensuring consistency in stationarity properties. The results demonstrate that the variables exhibit non-stationarity in levels but become stationary after first differencing, necessitating the use of first-differenced models in short-term analyses. Consequently, panel cointegration analysis is justified to determine long-run equilibrium relationships, ensuring that statistical inferences are not spurious [39]. These results justify further econometric modelling, including panel VECM or ARDL, to examine the dynamic relationships between financial development and petroleum consumption in EMEs [40].

3.2.2. Empirical Analysis

Following the methodology of [41,42], the study employed the autoregressive distributed lag (ARDL) bounds testing approach to investigate the long-run cointegration between petroleum consumption and financial development. The panel ARDL model is preferred when both the cross-sectional (N) and time-series (T) dimensions exceed 1, as opposed to a standard ARDL model, which is typically used for single time-series analysis. Refs. [42,43] argue that the method allows the estimation of long-run relationships between dependent and independent variables, even when the regressors are integrated at different levels, provided they are not I(2).

The ARDL framework estimates both short- and long-run dynamics within the same model, providing it with flexibility [44]. This is particularly important in small sample sizes, as it enables efficient parameter estimation without compromising insights into long-term equilibrium. Furthermore, the approach is more suited for diverse EMEs due to its ability to accommodate heterogeneity across cross-sections. Moreover, it can consider both immediate and equilibrium relationships over time [42]. This enhances inference by allowing for dynamic policy simulations and more accurate forecasting of how shocks to petroleum consumption or financial development propagate over time. As a result, policymakers can design targeted interventions that strike a balance between short-term adjustments and long-term stability. The AIC, BIC, and SBC were used to determine the optimal lag length, with the lowest values guiding model selection in Stata version 18.

The Hausman test [45] was conducted to determine the most suitable estimator among Pooled Mean Group (PMG), Mean Group (MG), or Dynamic Fixed Effects (DFE), as well as to assess the homogeneity of long-run coefficients across sections. The varying treatment of parameter heterogeneity necessitates comparing these estimators. The PMG allows short-run heterogeneity while assuming long-run homogeneity, the MG allows for full heterogeneity across cross-sections, and the DFE imposes homogeneity on both short- and long-run parameters. Selecting the appropriate estimator reinforces unbiased and robust inferences about the relationship between petroleum consumption and financial development in EMEs.

The equation below is estimated to examine the relationship between petroleum consumption and financial development in the selected emerging markets. The study employs the ARDL and error correction model (ECM) to capture the speed of adjustment in the presence of disequilibrium [46]. This captures both the cointegrating and the short-run effects of the variables under study [27,30,47,48].

Model Specification and Estimation Techniques

The following Error Correction Model (ECM) for financial development was tested empirically:

$$∆{\mathrm{FD}}_{\mathrm{it}}={\text{∅}}_{\mathrm{i}}\left({\mathrm{F}\mathrm{D}}_{\mathrm{i},\mathrm{t}-1}-{\mathsf{\gamma }}_{1\mathrm{i}}{\mathrm{P}\mathrm{C}}_{\mathrm{i},\mathrm{t}}-{\mathsf{\gamma }}_{2\mathrm{i}}{\mathrm{E}\mathrm{G}}_{\mathrm{i},\mathrm{t}}\right)+{\sum }_{\mathrm{j}=1}^{\mathrm{p}-1}{\mathsf{\delta }}_{\mathrm{i}\mathrm{j}}{∆\mathrm{F}\mathrm{D}}_{\mathrm{i},\mathrm{t}-\mathrm{j}}+{\sum }_{\mathrm{j}=0}^{\mathrm{q}-1}{\mathsf{\beta }}_{1\mathrm{i}}∆{\mathrm{P}\mathrm{C}}_{\mathrm{i},\mathrm{t}}+{\sum }_{\mathrm{j}=0}^{\mathrm{q}-1}{\mathsf{\beta }}_{1\mathrm{i}}∆{\mathrm{E}\mathrm{G}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\epsilon }}_{,\mathrm{i}\mathrm{t}}$$

(1)

where:

• $∆{\mathrm{FD}}_{\mathrm{it}}$ = The change in Financial Development for country i at time t.
• ${\text{∅}}_{\mathbf{i}}\left({\mathrm{F}\mathrm{D}}_{\mathrm{i},\mathrm{t}-1}-{\mathsf{\gamma }}_{1\mathrm{i}}{\mathrm{P}\mathrm{C}}_{\mathrm{i},\mathrm{t}}-{\mathsf{\gamma }}_{2\mathrm{i}}{\mathrm{E}\mathrm{G}}_{\mathrm{i},\mathrm{t}}\right)$ = The error correction term captures the long-run equilibrium relationship between financial development (FD), petroleum consumption (PC), and economic growth (EG). The term ${\text{∅}}_{\mathrm{i}}$ represents the speed of adjustment back to equilibrium.
• ${\sum }_{\mathrm{j}=1}^{\mathrm{p}-1}{\mathsf{\delta }}_{\mathrm{i}\mathrm{j}}{∆\mathrm{F}\mathrm{D}}_{\mathrm{i},\mathrm{t}-\mathrm{j}}$ = The lagged changes in financial development, accounting for short-term dynamics.
• ${\sum }_{\mathrm{j}=0}^{\mathrm{q}-1}{\mathsf{\beta }}_{1\mathrm{i}}∆{\mathrm{P}\mathrm{C}}_{\mathrm{i},\mathrm{t}}+{\sum }_{\mathrm{j}=0}^{\mathrm{q}-1}{\mathsf{\beta }}_{1\mathrm{i}}∆{\mathrm{E}\mathrm{G}}_{\mathrm{i},\mathrm{t}}$ = The short-run effects of changes in energy development and economic growth, respectively.
• ${\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\epsilon }}_{,\mathrm{it}}$ = The country-specific fixed effect.
• ${\mathsf{\epsilon }}_{,\mathrm{it}}$ = The error term or disturbance.

Robustness Check: Two-Step System GMM

Endogeneity is a major econometric problem in the analysis of the petroleum–finance nexus, arising from different sources. First, there is the likelihood of simultaneity or reverse causality between petroleum consumption and financial development. Petroleum consumption can stimulate financial deepening; it is plausible that developed financial systems provide financing channels that increase petroleum consumption. This mutual feedback creates bias in standard panel estimates. Second, unobserved country-specific factors like institutional quality, regulatory capacity, or macroeconomic shocks may give rise to omitted variable bias. These omitted variables impact both petroleum consumption and financial development but are not fully captured by the included regressors. Third, there is a dynamic panel bias emanating from the inclusion of the lagged dependent variable $\left({FD}_{i,t-1}\right)$ in dynamic models, as it is mechanically correlated with the error term [49].

To address these challenges, the study employs the two-step System Generalized Method of Moments (System GMM). This method suits the balanced panel of 20 EMEs over 21 years, where both the time dimension (T) and cross-sectional dimension (N) are moderate. The 2-System GMM offers a number of advantages: First, it corrects for simultaneity bias through the use of lagged levels and band differences of endogenous regressors as internal instruments and ameliorates [50,51] meliorate the feedback effects between petroleum consumption and financial development. Second, dynamic panel consistency, the two-system GMM addresses the Nickell bias in panels with limited time periods. Third, control for unobserved heterogeneity, ensuring that the results are not driven by unobserved cross-sectional differences. Fourth, robustness against weak instruments, as the GMM combines equations in levels and first differences, increasing efficiency relative to the difference GMM estimator.

The GMM model was specified as follows:

$${∆\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}}=(\mathsf{\alpha }-1){∆\mathrm{F}\mathrm{D}}_{\mathrm{i},\mathrm{t}-1}+{\mathsf{\beta }}_1{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{∆\mathrm{P}\mathrm{C}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{∆\mathrm{E}\mathrm{G}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_{3\mathrm{i}\mathrm{t}}{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{∆\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{D}\mathrm{u}\mathrm{m}\mathrm{m}\mathrm{y}1}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_5{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{D}\mathrm{u}\mathrm{m}\mathrm{m}\mathrm{y}2}_{\mathrm{i}\mathrm{t}}+∆{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(2)

where:

• ΔFD_it: Changes in financial development.
• $(\mathsf{\alpha }-1){∆\mathrm{F}\mathrm{D}}_{\mathrm{i},\mathrm{t}-1}$ = lagged changes in financial development.
• ${\mathsf{\beta }}_1{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{∆\mathrm{P}\mathrm{C}}_{\mathrm{i}\mathrm{t}}$ = cumulative changes in petroleum consumption.
• ${\mathsf{\beta }}_2{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{∆\mathrm{E}\mathrm{G}}_{\mathrm{i}\mathrm{t}}$ = cumulative changes in economic growth.
• ${\mathsf{\beta }}_{3\mathrm{i}\mathrm{t}}{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{∆\mathrm{X}}_{\mathrm{i}\mathrm{t}}$ = changes in other explanatory variables (FDI, infrastructure, inflation, government effectiveness, prices, natural resources, and real interest rates).
• ${\mathsf{\beta }}_4{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{D}\mathrm{u}\mathrm{m}\mathrm{m}\mathrm{y}1}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_5{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{D}\mathrm{u}\mathrm{m}\mathrm{m}\mathrm{y}2}_{\mathrm{i}\mathrm{t}}$ = dummy variables (global financial crisis and the COVID-19 pandemic).
• $∆{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$ = Error term capturing unobserved factors.

The validity of the estimator was confirmed by the diagnostic statistics. The Arrellano–Bond AR(1) test is significant, indicating first-order autocorrelation, while the AR(2) is insignificant, ruling out problematic second-order autocorrelation. The Hansen and Sargan tests of overidentifying restrictions confirm the validity of the instruments, with p-values exceeding the conventional thresholds. The diagnostic test confirms that the model is appropriately specified and free from instrument proliferation. The GMM results (Appendix A,

Table A2. Summary of Pooled Mean Group on the cointegrating and causality relationship between financial development and electricity consumption.

	PMG D.FD	PMG D.PC	PMG D.EG	PMG D. FDI	2-Step System GMM FD
Long-Run
EG	0.0252 ** (5.11)	−0.0420 *** (−3.32)		−0.216 (1.66)	0.0105 ** (0.00407)
PC	0.0664 *** (1.19)		6.644 *** (2.05)	5.586 *** (−9.63)	0.149 ** (0.0472)
FD		0.850 *** (14.68)	18.35 ** (−2.65)	0.241 (0.17)	0.532 *** (0.116)
FDI	0.0186 *** (9.91)	0.00773 *** (1.77)	−0.0815 (−1.24)		0.000201 *** (0.0000531)
ECT	−0.210 *** (−3.79)	−0.140 ** (−2.86)	−0.0224 *** (−1.34)	−0.599 *** (−7.66)
Short-Run
D.EconGR	0.0196 (0.77)	0.0584 (0.88)		−1.672 (−0.89)
D.PC	0.0264 (0.34)		0.891 (1.01)	−18.33 (−0.79)
FD		−0.0516 (−0.70)	−0.629 (−0.57)	14.16 (0.88)
D. FDI	0.000416 (−0.21)	−0.00186 (−1.15)	0.0329 (0.96)
_cons	0.00723 (1.16)	0.365 ** (2.70)	−0.138 (−1.26)	10.93 *** (7.54)
Dummy1					0.00432 (0.00497)
Dummy2					0.0205 *** (0.00333)
N	399	399	399	399	359
Groups	-	-	-	-	20
Instruments	-	-	-	-	14
Arellano Bond AR 1	-	-	-	-	−2.97
Arrellano Bond AR 2	-	-	-	-	−0.51
Sargan Test					41.38
Hansen Test	-	-	-	-	9.69
Hausman	44.75 ***	39.55 ***	64.80 ***	9.42 *

t statistics in parentheses, * p < 0.05, ** p < 0.01, *** p < 0.001.

) confirm the findings of the ARDL-PMG Framework. The convergence of the results strengthens confidence that petroleum consumption and financial development reinforce each other in some EMEs. This ensures that the reported long-run relationships are robust, not spurious, and provides a firmer empirical foundation for the policy implications drawn in Section 5.

3.2.3. PMG Estimation

Table 5. Variables Long-run Coefficient Interpretation.

PC → FD	0.0664 ***	Petroleum consumption boosts financial development
FD → PC	0.850 ***	Financial development increases petroleum consumption
EG → FD	0.0252 **	Growth stimulates the financial sector
EG → PC	−0.0420 ***	Growth reduces petroleum reliance (possible diversification)
FDI → FD	0.0186 ***	FDI supports financial development
ECT (FD)	−0.210 ***	21% adjustment speed
ECT (PC)	−0.140 **	14% adjustment speed

t statistics in parentheses, ** p < 0.01, *** p < 0.001. Notes: Appendix A, Table A1 presents a complete set of short-run coefficients, diagnostic tests, and robustness estimates.

presents a summary of the Pooled Mean Group (PMG) estimation results, analyzing the cointegrating and causal relationships between financial development (FD, as measured by the Financial Development Index), petroleum consumption (PC), and economic growth.

A complete interpretation of these results, including short-run dynamics, robustness tests, and implications for the energy–finance nexus in EMEs, is provided in Section 4 below.

4. Discussion

The study provides empirical evidence on the relationship between financial development and petroleum consumption, with economic growth and FDI as the control variables in the EMEs. These findings are validated through a robustness check using the two-step System GMM estimator, which corrects for simultaneity, omitted variable bias, and dynamic panel bias. The GMM results (Appendix A, Table A2) are consistent with the ARDL–PMG estimates.

4.1. Error Correction Model

The error correction terms (ECTs) are negative and significant for both financial development and petroleum consumption, at −0.210 and −0.140, respectively, confirming a stable long-term relationship between petroleum consumption and financial development. This shows that shocks dissipate over time, with financial development and petroleum consumption adjusting annually at rates of 21% and 14%, respectively, consistent with previous findings [39,52].

4.2. Long-Run Relationships

The results reveal a strong bidirectional relationship between petroleum consumption and financial development, with a mutual reinforcing effect. These findings confirm the energy–finance nexus.

Petroleum consumption supports financial development over the long term. A one-unit rise in petroleum consumption increases financial development by 0.0664 units. This aligns with the findings of [53], who suggest that energy consumption, including petroleum, fosters financial development by enabling economic activities, improving financial intermediation, and encouraging innovation.

Financial development significantly and positively impacts petroleum consumption. A stronger financial sector supports energy-intensive activities, such as refineries and infrastructure projects [9] supports this, noting that financial development facilitates investments in energy-demanding sectors.

Control Variables

Economic growth has a dual effect. On the one hand, it is negatively correlated with petroleum consumption. This suggests that higher economic growth may lead to increased diversification into other energy sources. This counterintuitive finding reflects the energy transition in EMEs, rather than a contradiction of the growth–energy linkage [54]. On the other hand, economic growth has a positive impact on financial development. This suggests that economic growth supports the growth of the financial sector. FDI contributes positively to financial development, and it is indicative that FDI strengthens financial systems via capital flows [55].

4.3. Short-Run Relationships

The short-term results show no significant effects of petroleum consumption on financial development but reveal a significant long-term relationship. The structural features of EMEs can explain this. Many countries are net importers of petroleum, and price fluctuations influence their trade balances rather than their financial depth [56,57]. Conversely, petroleum exporters face financial volatility associated with dependence on revenue; however, infrastructure bottlenecks restrict the transmission of this volatility to financial markets [58].

Petroleum consumption has no immediate impact on financial development. This contrasts with long-term findings. This aligns with the findings of [59], who note that the effect of energy consumption on financial development is often delayed due to the dynamics of resource allocation.

Financial development does not have an immediate influence on petroleum consumption [9] findings, which suggest that financial development has weak effects on petroleum consumption, are consistent with this. A positive and significant coefficient indicates a baseline upward trend in petroleum consumption, probably driven by structural factors such as industrialization [53].

5. Conclusions and Policy Implications

This study confirms the bidirectional relationship between petroleum consumption and financial development in EMEs. It confirms a mutually significant long-term relationship between the two variables. The findings suggest that petroleum consumption leads to financial deepening, and conversely, financial development supports higher petroleum consumption. From a theoretical perspective, by isolating petroleum consumption within the energy–finance nexus, this work reduces aggregation bias and contributes new evidence for EMEs. A critical strength of this study lies in addressing the potential endogeneity that characterizes the petroleum–finance nexus. By employing the two-step System GMM estimator as a robustness check, we corrected for reverse causality, omitted variable bias, and dynamic panel bias.

The results of the study establish a general bidirectional nexus; it is critical to acknowledge regional heterogeneity. Large EMEs such as China and India exhibit stronger energy–finance linkages due to their industrial scale and deeper capital markets. In contrast, smaller EMEs, particularly in Sub-Saharan Africa, exhibit weaker effects due to infrastructural gaps and financial constraints [23]. These differences imply that policy prescriptions must be context-specific: advanced EMEs may prioritize financial market reforms and energy diversification strategies, whereas smaller EMEs may focus on strengthening basic energy infrastructure and improving credit access.

The paper has critical policy implications: first, sustaining investment in petroleum-related infrastructure is essential for boosting financial development. Second, leverage financial development to support technological innovation and a gradual transition towards sustainable energy. Third, develop innovative financial instruments, such as green bonds and sector-specific credit facilities. Finally, petroleum-importing countries like India and Turkey can utilize green bonds to finance their energy transition. This reduces oil reliance and stabilizes both markets. Conversely, oil-exporting countries like Nigeria can benefit more from sovereign wealth fund management reforms to mitigate oil revenue volatilities while deepening financial systems.

Limitations and Future Research

This analysis has several limitations. First, it uses aggregate country-level data and does not capture sectoral heterogeneity in petroleum use, such as differences across transport, manufacturing, and services. Second, while the ARDL–PMG and GMM approaches address endogeneity and strengthen inference, they do not fully capture potential nonlinearities or structural breaks in the petroleum–finance nexus. Third, structural factors such as industrialization and urbanization, which may mediate this relationship, were not explicitly controlled for due to data constraints.

Future research should therefore pursue sector-specific analyses, adopt nonlinear ARDL models, and examine the role of renewable energy integration. In addition, greater attention should be given to financialization (the qualitative transformation of financial systems through market orientation, speculative activity, and financial innovation), which may significantly alter how petroleum consumption interacts with financial markets in EMEs.