Energy diversification, financial development and economic development: an examination of convergence in OECD countries

Convergence theory, first introduced by Galor (1996) and later refined by Barro and Sala-i-Martin (1992), suggests that less affluent economies, driven by diminishing marginal returns, tend to catch up with wealthier counterparts over time. This convergence is facilitated by common factors like technological advancements, investment, and saving rates. In integrated markets, where technology is widely accessible, convergence becomes more achievable.

Technological progress has been a primary driver of economic growth since the 1990s (McMillan et al., 2014). As income levels rise, societies shift focus from meeting basic needs to addressing environmental concerns. Energy, a crucial input for production, sees increased demand as economic output grows ^[1]. Financial development, which provides resources for investments and production, further drives economic activities, thereby amplifying energy demand (Shahbaz et al., 2018).

However, reliance on non-renewable sources like coal, oil, and natural gas leads to CO2 emissions, accelerating climate change. In contrast, renewable energy sources like solar, wind, biomass, hydropower, and geothermal offer cleaner alternatives, though their effectiveness is often constrained by factors like weather dependency and lower efficiency. Therefore, energy diversification – shifting from fossil fuels to renewables – is essential to achieve a balance between economic growth and environmental sustainability.

Since the 1990s, technological progress has facilitated the shift from fossil fuels to renewables in energy production (Paramati et al., 2022). Despite the faster growth rate of renewable energy consumption (3.37% annually from 1980 to 2022) compared to total energy consumption (1.91%), renewables accounted for just 6.25% of global energy consumption by 2022. In contrast, coal, natural gas, oil, and nuclear energy made up 32.30%, 24.76%, 32.43%, and 4.48%, respectively ^[2]. This highlights the need to boost renewables in the energy mix to enhance diversification.

Energy diversification not only stabilizes supply and prices but also influences economic performance and financial development. Advanced financial systems are crucial for financing clean energy projects and facilitating economic growth, thereby driving energy diversification. Meanwhile, technological progress transforms production processes, financial services, and energy sources, and globalization fosters the exchange of technology, capital, and information. However, the pace of change in energy diversification, per-capita income, and financial development varies significantly by country, largely due to resource availability and differing levels of technology diffusion. Understanding the convergence of these variables can inform policies for sustainable development goals.

Previous studies, including those by Cabral and Castellanos-Sosa (2019), Zhao and Serieux (2019), Ghatak and De (2021), and Alhassan et al. (2022), have examined the convergence of per-capita income and its association with financial development across different regions. However, research on the convergence of energy diversification and its link to economic and financial development remains limited. A few studies, such as Gozgor and Paramati (2022), Yilanci et al. (2021), have explored the relationship between energy diversification and economic growth, while others, including Shahbaz et al. (2023) and Nibedita and Irfan (2023), have examined the connections between financial development and energy diversification. These studies consistently note a narrowing income gap between countries over time and/or the formation of convergence groups, yet the relationships between energy diversification, per-capita income, and financial development and the role of technological progress remain inconclusive, possibly due to differences in contexts, dataset periods, and techniques used for investigation.

We argue that technological development and diffusion can eventually lead countries to achieve a harmonious balance of per-capita income, financial development, and energy diversification. However, the lack of international cooperation in production, financial services, and energy technology innovation hampers the sharing of R&D knowledge and information, thus slowing technology diffusion (Jørgensen et al., 2018). OECD countries, with their high-income status, advanced technology, robust finance, and sustainable development goals, are well-positioned to accelerate this technology diffusion process through coordinated policies and economic openness.

Therefore, this study addresses two key research questions:

• (1)

  Do energy diversification, per-capita income, and financial development in OECD countries converge over time?

• (2)

  How do interactions between these variables drive convergence in OECD countries?

To address these questions, we analyze data from 38 OECD countries spanning 1997–2021. This period offers comprehensive data on energy diversification, financial development, and per-capita income, while encompassing significant policy shifts and initiatives aimed at promoting sustainability. The late 1990s and early 2000s marked pivotal technological advancements in energy production and financial systems, coupled with growing globalization and economic integration among OECD nations (Kurniawati, 2020). These factors make the timeframe ideal for examining convergence, as it reflects both technological diffusion and policy impacts on the studied variables.

Our study makes several contributions to the literature. First, it is the first to test convergence across energy diversification, financial development, and per-capita income within OECD countries. While earlier research has focused on income or financial development convergence in Europe or globally, none has analyzed energy diversification – a critical aspect of sustainable development. Given the OECD’s coordinated policies promoting wealth, green initiatives, and financial stability, our findings offer insights into how advanced technologies and practices diffuse across member nations, helping policymakers enhance collaboration and technology-sharing efforts. Second, we apply the log-t convergence test by Phillips and Sul (2007, 2009), a more advanced method for analyzing heterogeneous panel data. Unlike traditional β- and σ-convergence tests, which assume uniform growth paths across countries, the log-t test accommodates individual growth trajectories while assessing overall and club convergence. Lastly, we examine the interrelationships between energy diversification, financial development, and per-capita income using updated panel data. This provides insights into how these factors interact to drive convergence. Previous studies have largely explored the links between energy consumption and either economic growth or financial development using global or European datasets. By narrowing the focus to OECD countries, our research delivers a deeper understanding of these dynamics, contributing valuable perspectives to discussions on sustainable economic development.

The paper is structured as follows: Section 2 offers a comprehensive review of the literature. Section 3 outlines the methodologies used and provides an overview of the data. Section 4 presents the empirical results. Finally, Section 5 concludes the study, highlighting its policy implications.

Despite sharing traits such as high income, advanced technology, strong financial systems, and sustainable development policies, OECD countries display significant disparities in energy diversification, financial development, and per-capita income. For instance, in 2021, Slovakia had an energy concentration index of 0.0468, while Costa Rica’s was 0.4203. Per-capita GDP ranged from US$6,423.86 in Colombia to US$110,425.89 in Luxembourg, and the financial development index varied from 0.2524 in Estonia to 0.9392 in Switzerland. Over the 1997–2021 period, these disparities have narrowed, with standard deviations decreasing. This raises the question of whether energy diversification, financial development, and per-capita income in OECD countries are converging over time.

Several methods are used to test convergence, including β-convergence, σ-convergence, stochastic convergence, and log-t convergence. β-convergence examines whether countries with lower initial levels grow faster, while σ-convergence looks at whether disparities decrease over time. Stochastic convergence assesses whether shocks to a variable are temporary. However, β and σ-convergence tests rely on restrictive assumptions that countries follow the same growth path. In contrast, log t convergence, developed by Phillips and Sul (2007, 2009), allows for individual transition paths and can identify both overall and club convergence. This method is widely used in economics and energy studies (Tomal, 2023).

Technological advancements, such as artificial intelligence, have expanded opportunities for energy diversification (Zhang et al., 2024; Zhao et al., 2024), potentially facilitating convergence among OECD countries. While no studies have directly examined the convergence in energy diversification, related research on renewable energy offers insights. Reboredo (2015) found no log t convergence in renewable energy shares, while Demir and Cergibozan (2020) observed β-convergence in alternative energy consumption within OECD countries.

Global and regional studies provide mixed findings. Bigerna et al. (2021) and Berk et al. (2020) found σ- and β-convergence in renewable energy shares globally and within the EU. However, Saba and Ngepah (2022b) and Saba and Ngepah (2022a) reported no log t convergence but identified convergence clubs. Pinar (2024) also found no global convergence in renewable energy innovation but noted the formation of two convergence clubs.

Technological advancements in finance have driven financial development (Khan et al., 2024; Misati et al., 2024), potentially promoting convergence within OECD countries. However, studies on financial development convergence among OECD countries remain limited. Narayan et al. (2011) identified β-convergence in stock market growth among less developed OECD members. Maxfield et al. (2017) found no log t convergence in financial leverage but observed convergence among large transnational institutions in OECD countries.

Globally or within regions like Europe, several studies often indicate β-convergence. For example, Fung (2009) and Stolbov and Veysov (2011) reported β-convergence in financial depth globally, Affinito (2011) identified it for institutional deposits and loans in Europe. However, studies allowing for individual growth paths suggest convergence clubs rather than uniform convergence. For instance, Niţoi and Pochea (2016) observed no log t convergence for various financial indicators in Central and Eastern Europe, while Cavallaro and Villani (2021) observed similar findings in European financial development indices, suggesting that technology alone cannot ensure uniform financial convergence across OECD countries.

Technological innovation has been linked to economic growth (Cheng et al., 2021; Khan et al., 2024), and several studies show β and stochastic convergence in per-capita income across OECD countries. Greta and Lewandowski (2015) and Margaritis et al. (2007) found evidence of β-convergence, while King and Ramlogan-Dobson (2014) identified stochastic convergence to the U.S. for half of OECD countries, and Kónya (2023) found it for all OECD nations.

Few studies have used the log-t convergence test for per-capita GDP in OECD countries. Research using broader samples, such as Cavallaro and Villani (2021), Mazzola and Pizzuto (2020), and von Lyncker and Thoennessen (2017), found no log t convergence and the formation of convergence clubs.

The literature suggests β-convergence, a lack of global log t convergence, and the formation of convergence clubs for energy diversification, financial development, and per-capita income among OECD countries. This implies that technological advancements may promote convergence but cannot guarantee uniform growth rates due to varying benefits across countries. Therefore, the hypothesis is:

Promoting energy diversification is crucial, going beyond just increasing renewable energy, though adding more renewables is key. While substantial research explores the relationships between economic growth, financial development, and renewable energy across various regions, few studies focus on how energy diversification interacts with economic or financial development, particularly within OECD countries. This section reviews the relevant literature on the interplay between energy diversification, economic development, and financial development.

Energy resources are unevenly distributed globally, leading to varying degrees of energy diversification across countries. Many nations rely on specific energy sources due to cost advantages, which can limit diversification. Overdependence on a few sources exposes economies to risks from energy price volatility, political tensions, and policy shifts, potentially harming economic growth (Çetin and Kapkaç, 2022). For example, Huntington and Liddle (2022) found that a 10% increase in energy prices reduced growth by 0.15% on average across OECD nations.

Empirical evidence on the effects of energy diversification on economic growth is mixed. Gozgor and Paramati (2022) found that greater energy diversification supports long-term economic growth in G20 economies, though some OECD nations experienced short-term growth declines. Conversely, Ahmed et al. (2022) found no significant impact of energy diversification on the economic performance of four Nordic countries. Chen et al. (2024) demonstrated that while lower levels of diversification hinder growth in BRICS nations, higher levels foster positive effects.

In OECD countries, fossil fuels still account for over 80% of energy consumption (Hu et al., 2022), prompting efforts to diversify energy sources in line with SDG 7 (Affordable and Clean Energy). This diversification encourages clean energy innovations, attracts investments, and boosts per capita income. Based on these findings, we hypothesize:

Despite high income levels in many OECD countries, fossil fuels remain dominant due to their affordability, accessibility, and established infrastructure. The energy ladder theory (Hosier and Dowd, 1987) and fuel stacking model (Masera et al., 2000) suggest that rising income typically shifts economies toward cleaner energy sources, though fossil fuels remain part of the energy mix due to renewable energy’s instability and high transition costs.

However, empirical studies show mixed results. Shahbaz et al. (2023) found that economic growth reduced energy diversification in Australia from 1995 to 2021. Nibedita and Irfan (2024) observed a U-shaped effect in E7 economies, where initial economic growth led to energy concentration before increasing diversification. In OECD countries, rising per capita income often encourages a continued reliance on fossil fuels, slowing the transition to diversified energy sources. Based on these observations, we hypothesize:

Diversifying energy sources through renewable energy is capital-intensive, requiring substantial financing for high initial costs, especially in research and development (Wang et al., 2022). Developed financial markets can ease this transition by providing easier access to debt and equity financing, lowering the cost of capital for renewable energy projects (Acheampong et al., 2020). Consequently, financial development is likely to support energy diversification.

Renewable energy projects face more significant financial challenges than fossil fuel projects. They require larger capital investments, longer payback periods, and higher information costs (Peimani, 2018; Nibedita and Irfan, 2024). Additionally, renewable assets carry greater technology-specific risks and are more difficult to liquidate in bankruptcy situations (Steffen, 2020). These factors can lead to lending hesitation or lending interest rate increases. Consequently, even in well-developed financial markets, renewable projects may struggle to secure financing or face higher capital costs, which can slow down the energy diversification process.

Empirical studies indicate mixed effects. Nibedita and Irfan (2024) found no significant impact in E7 economies, while Shahbaz et al. (2023) reported positive effects in Australia, driven by government initiatives promoting clean energy. Limi (2020) observed positive effects in sub-Saharan Africa, while Qamruzzaman (2022) noted varying effects in countries like Brazil, South Africa, Russia, India, and China.

OECD countries, with their advanced financial systems, often lead international efforts in renewable energy innovation, as seen in the 2015 Paris Agreement (Jørgensen et al., 2018). Aligning with the findings of Shahbaz et al. (2023) in Australia, we hypothesize:

Energy diversification, particularly through renewables, requires significant financing at every stage—from research and development to infrastructure and ongoing operations. This capital demand has led to the creation of specialized financial products, such as green bonds and loans, to facilitate renewable energy investment (Azhgaliyeva et al., 2020; Alharbi et al., 2023). Such diversification can attract international financial institutions and sustainable investors, enhancing the depth and resilience of the financial system (Zhe et al., 2021). Additionally, energy diversification can boost economic stability by reducing energy price volatility and improving energy security, which in turn increases demand for financial services (Hache, 2018; Maciejowska, 2020).

However, risks accompany energy diversification. Renewable energy projects face challenges like technological underperformance, inconsistent resource availability (e.g. limited sunlight or wind), and changing regulations, which could lead to financial losses. These risks may result in loan defaults and erode investor confidence, straining financial institutions.

Although no studies directly link energy diversification to financial development, research on renewable energy consumption generally shows positive effects. Zhe et al. (2021) found that expanding renewable energy in Turkey spurred financial sector growth, while Khan et al. (2020) observed similar global impacts, linking renewable energy to financial innovation and market expansion.

In OECD countries, policies such as subsidies, tax incentives, and regulations lower investment risks and attract private capital to renewable energy projects. This environment fosters financial innovations tailored to renewables, driving growth in the financial sector. Therefore, we hypothesize:

The demand-following hypothesis (Robinson, 1952) suggests that economic growth drives the development of the financial sector. As economies grow, the need for more sophisticated financial products arises to support business activities and manage risks, encouraging innovation in financial services.

Empirical evidence of this relationship is mixed. Studies like those by Dar and Nain (2023) in South Asia and Atil et al. (2020) in Pakistan support the demand-following hypothesis, showing that economic growth fosters financial sector development. However, research by Cherif and Dreger (2016) in MENA countries and Nain and Kamaiah (2014) in India suggests that economic growth does not always lead to financial development, indicating that country-specific factors can influence this dynamic.

OECD countries, with their high-income levels and advanced financial systems, offer a unique environment. Rising incomes enhance savings and investment capacities, driving demand for complex financial products. Their established financial institutions are also well-positioned to innovate and meet this demand.

Thus, we propose the following hypothesis:

The supply-leading hypothesis argues that financial development is a key driver of economic growth. A well-developed financial system fosters innovation by funding transformative ideas and technological advancements, which increase both the quality and quantity of output (Schumpeter and Opie, 1934). Additionally, an efficient financial system channels savings into productive investments, promoting economic expansion (Bencivenga and Smith, 1991; King and Levine, 1993). In this view, financial development acts as a catalyst for sustained economic progress.

Empirical evidence largely supports this hypothesis. Studies on OECD countries by Aydin and Malcioglu (2016), Madsen and Ang (2016), Purewal and Haini (2022), Çetin and Çınar (2023), and Hashemizadeh et al. (2023) all show that financial development positively influences economic growth. With well-regulated financial markets, OECD countries effectively allocate resources to productive investments, fostering long-term economic growth.

Therefore, we hypothesize:

Energy diversification is the process of shifting from heavy reliance on one energy source to a more balanced mix. Energy comes from five primary sources: coal, natural gas, petroleum, nuclear, and renewables. To gauge diversification, we use the Herfindahl-Hirschman Index (HHI), termed the Energy HHI. Traditionally employed to measure the concentration of companies’ market shares, the HHI approach has been extended to gauge the concentration of export markets by the World Bank ^[3] and to measure energy diversification by researchers such as Gozgor and Paramati (2022), Akrofi (2021) and Shahbaz et al. (2023). The Energy HHI for country i in year t is

In Eq. (1), $Ei$ represents the total primary energy consumption of country i. $eij$ denotes the amount of energy consumption from source j by country i, while $ni$ represents the number of energy sources consumed by country i. The Energy HHI falls within the range of 0–1, with a higher value indicating greater concentration and lower energy diversification. This index helps assess whether a country’s energy portfolio has diversified over time and enables cross-country comparisons.

Financial development refers to improvements in financial markets and institutions. Pei et al. (2023) and Song et al. (2021) suggest using proxies like private credit to GDP ratio and market capitalization to GDP ratio, but these may not capture the full picture. We follow Ngo et al. (2022), Acheampong et al. (2020), Lahiani et al. (2021), and Baloch et al. (2021), utilizing IMF’s financial development indices constructed from 20 indicators, with the construction method detailed by Svirydzenka (2016). Our primary proxy is the overall Financial Dev index, supplemented by the financial institution index (FI Dev) and financial market index (FM Dev), ranging from 0 to 1, with higher values indicating greater development.

Per-capita income, reflecting individual economic well-being, is measured using per-capita GDP.

Beta convergence tests whether countries with lower energy diversification, financial development, or per-capita income grow faster than those with higher levels using the following equation:

where $yit$ represents the set including energy diversification, financial development, and per-capita income. A negative and statistically significant parameter $β$ indicates beta convergence.

Sigma convergence assesses whether the gaps in energy diversification, financial development, or per-capita income levels among countries narrow over time, as shown in equation:

where $wit=ln⁡yit−1n∑i=1nyit$⁠. Sigma convergence is confirmed if $σ$ is negative and statistically significant.

We employ a fixed-effects estimator for Eqs.(1)-(2) and verify the results’ robustness using a random-effects estimator.

Phillips and Sul (2007) propose log-t convergence test, which captures the heterogeneity in panel data $Xit$ through a common factor $μt$⁠, and idiosyncratic effects $δit$ and $εit$⁠, using the following single-factor model:

Where $Xit$ is the growth rate of the variables of interest for country i in year t; $μt$ is the common factor $μt$⁠, which we interpret as the common technology available to all OECD countries; $δit$ is the idiosyncratic distance between the common factor $μt$ and the systemic part of $Xit$ for country i in year t, reflecting the individual growth factor of country i; $εit$ is the random shocks.

Convergence occurs when the individual growth factor $δit$ converges.

Assume that the common factor $μt$ is captured by the average growth rate of all countries in any year, $δit$ can be isolated by taking the ratio of a growth rate of a country and the average rate. The relative transition parameter (h_it), which measures the performance of country i (⁠$δit$⁠) relative to the common growth rate (⁠$μt$⁠) in year t, and the mean square of relative transition differences $(Ht)$ are defined as:

The cross-sectional mean of $hit$ is 1. The overall convergence of all countries occurs in the long run (t →∞) when $hit$ converge to 1, leading to $Ht$ converging to 0. This convergence coincides with $δit$ converge to $δt$⁠. As t →∞, if $Ht$ remains positive, it implies that the growth rates of all countries diverge or some countries converge despite divergence occurring among others.

To formulate an econometric test for the time-varying idiosyncratic components $δit$⁠, Phillips and Sul (2007) model $δit$ as:

where $δi$ is time-invariant growth factor for country i; $ξit∼iid(0,1)$ across i but weakly dependent on t; and L(t) is a slowly varying function (like log t) and $L(t)−1$ is a slowly decay function for which L(t) →∞ and then $L(t)−1→0$ as t →∞; $σi$ is an idiosyncratic scale parameter; and $α≥0$ is the decay rate.

The null and alternative hypotheses are:

$H0:δit=δ$ for all i and/or $α≥0$ (overall convergence)

$H1:δit≠δ$ for some i and/or $α<0$ (divergence)

The null hypothesis of convergence is tested using the following log t regression:

where $L(t)=log(t+1)$⁠; $γ=2α$⁠, measuring the convergence speed of $δ$⁠; $ut$ is the random error.

If the null hypothesis of overall convergence is rejected (i.e. $γ=2α<0$ at 5% significance level, $tγ<−1.65$⁠), the test for convergence clubs can be tested using the following club convergence algorithm:

• (1)

  Step 1: list sample countries in the ascending order of growth rate in the last year.

• (2)

  Step 2: form a core group of k* countries with the highest growth rates to form a convergence club. The optimal size of a convergence club is determined by maximizing $tk$ using k countries (⁠$2≤k<n$⁠) subject to $min{tk}>−1.65$⁠.

• (3)

  Step 3: Add one country to the core group at a time and run the log t convergence test. If $tγ>−1.65$⁠, the country is included. The core group and the included countries will form the first convergence club.

• (4)

  Step 4: Conduct log t convergence test for the non-included countries. If $tγ>−1.65$⁠, they form the second convergence club. If not, repeat Steps 1–3 to detect sub-convergence clubs. If no core group is formed in Step 2, these countries exhibit divergent behavior.

After identifying the initial convergence clubs, we apply Phillips and Sul (2009)’s club merging test. This test reformulates Phillips and Sul (2007)’s test for adjacent clubs, determining if the adjacent clubs can be merged to form a larger club, and thereby resulting in the final convergence clubs.

We also follow Phillips and Sul (2007, 2009)’s suggestion to employ Christiano and Fitzgerald (2003)’s band pass filter and discard 30% of the data to extract the trend prior to conducting log t convergence tests.

Although the convergence tests can provide evidence of possible convergence clubs, the tests do not explicitly identify the likely drivers for the formation of convergence clubs. This lead us to the next phase of our analysis.

We begin by investigating whether the growth of any of the three variables of interest is influenced by its own past growth as well as the past growth of the other two variables using Dumitrescu and Hurlin (2012)’s Granger causality procedure as follows:

where $yi,t$ represents one of the three variables of interest for country i in year t; $xi,t$ denotes one of the two remaining variables for country i in year t; $ϵi,t$ signifies the stochastic error term; the optimal lag orders $k$ are determined using the Akaike information criterion; and $αi,βi,γi$ are parameters to be estimated. This Granger model can be estimated for stationary variables, and thus, the stationarity of the variables (in both level and difference forms) needs to be tested before estimation. We conclude $x$ Granger causes $y$ if the coefficients $βi(k)$ collectively show statistical significance different from zero.

We proceed to examine the causal relationships while considering the controlling covariates:

where $yi,t$ and $xi,t$ retain their definitions from Eq. (8); $trend$ denotes the linear trend in $yi,t$ from 1997 to 2021, proxied for technological progress; $Zi,t$ is a set of controlling covariates; $μi$ represents country-specific unobserved features; $εi,t$ represents the random error component; $β$ are parameters to be estimated; all variables, except trend, are expressed in natural logarithms.

We select $Zi,t$ based on existing literature and data availability. When $yi,t$ is per-capita income, following studies by Gozgor and Paramati (2022) and Ahmed et al. (2022), $Zi,t$ includes fixed capital formation, labor participation, and international trade. When $yi,t$ is financial development, $Zi,t$ incorporates fixed capital formation, unemployment rate, and human development index (HDI), as shown in studies by Ngo et al. (2022), Ngoc et al. (2024), and Raifu et al. (2023). When $yi,t$ is energy diversification, $Zi,t$ comprises oil price and environmental perception (proxied by HDI). This choice is informed by the significant contribution of oil-based energies to CO2 emissions (about 51% of the energy mix) and the reflection of environmental awareness in HDI scores.

The estimator for Eq. (9) is chosen through various tests, including the Breusch-Pagan, Hausman, and F-tests. We determine the standard error for the estimator through multiple tests, such as the Modified Wald test for heteroskedasticity, the Pesaran test for cross-sectional dependence, and the Wooldridge test for autocorrelation.

The dataset comprises a balanced panel spanning 38 OECD countries from 1997 to 2021 (see  Appendix). Table 1 presents key statistics and data sources, revealing notable variations among countries for each variable. Notably, the Trend column, serving as a proxy for technological progress, indicates a yearly decrease of 0.2 points in energy source concentration (signaling increased diversification), a $431.9 per year increase (constant 2015) in per-capita GDP, and a 0.3-point annual enhancement in financial development. These positive trends suggest that technological progress could drive countries towards convergence in energy diversification, financial development, and per-capita income. Additionally, we observe positive growth rates in oil prices, the human development index, labor participation, employment rate, and international trade. However, there is a negative growth rate in gross fixed capital formation to GDP.

Table 2 displays the results of β-convergence, σ-convergence, and log t convergence tests for our panel of 38 OECD countries from 1997 to 2021, using the fixed-effects estimator. We find that the β and σ coefficients are all negative and statistically significant at the 1% level. This indicates that energy diversification, per-capita income, and financial development across the OECD countries tend to move towards a common level, while the dispersion in these variables decreases.

However, the t-statistic values of the log-t convergence test are all below −1.65, so the null hypothesis of panel convergence is rejected, suggesting that the 38 OECD countries are not converging to the same steady state regarding energy diversification, financial development, and per-capita income. The discrepancy in convergence results among these tests may stem from their underlying assumptions: β-convergence and σ-convergence assume the same growth path for all countries, whereas the log t convergence allows for individualized transition paths.

Phillips and Sul (2007) suggest that the lack of overall convergence might stem from divergent members within a group. To explore this, we utilize their clustering algorithm to classify countries into converging or divergent clusters. Table 3 outlines the results, with the number of countries in each club detailed in the first column. Using Phillips and Sul’s (2009) club merging algorithm, we then investigate potential merges of adjacent clubs, presented in the second column. The final convergence clubs, after both initial classification and merging, are shown in the last column.

For energy diversification, initial classification yields four convergence clubs. Merging Clubs 2 and 3 reduces the final clubs to three. Despite varying levels of energy diversification, OECD countries in these final clubs exhibit convergence, with a strong convergence speed in final Clubs 1 and 3 and a weak speed in final Club 2, as indicated by the log coefficients. Regarding per-capita income, six initial convergence clubs are identified. Merging Clubs 2, 3, and 4 results in four final clubs, showing varying convergence speeds.

Initial classification uncovers four clubs for overall financial development, which are reduced to three after merging. The final clubs exhibit varying speeds of convergence. Additionally, two convergence clubs emerge for financial institution development, while four clubs are identified for financial market development. The merging tests reject club merging for both financial development components.

The convergence tests in section 4.1 for energy diversification, per-capita income, and financial development are univariate analyses. Now, we investigate potential causal relationships between these variables. Unit root tests, including LLC, Fisher-DF, and Fisher-PP methods, reveal unit roots at the level, particularly with the Breitung test, but reject them at the first difference for all variables (see Table 4). Hence, we employ the Granger causality test for panel data, as proposed by Dumitrescu and Hurlin (2012), at the first difference, with lag selection based on the Akaike Information Criterion (AIC).

Results in Table 5 demonstrate the rejection of the null hypothesis of no association between the lag of one outcome and the current value of other outcomes. This indicates evidence of a bi-directional Granger causality between energy diversification, per-capita income, and financial development in OECD countries.

We explore the causal relationship between energy diversification, financial development, and per capita GDP while controlling for other factors. Table 6 displays the results for selecting an appropriate estimator for panel data models. The Breusch-Pagan test favors random effects over pooled OLS, while the F-test suggests potential bias in both pooled OLS and random effects estimators due to significant observed and unobserved fixed effects. The Hausman test indicates that fixed effects are more suitable, except for the energy diversification model. Consequently, we employ a fixed-effects estimator to account for unobserved country characteristics.

We then assess the residuals of the fixed effects models for heteroskedasticity, cross-sectional dependence, and autocorrelation. The tests reject the null hypotheses of no heteroskedasticity, cross-sectional dependence, and autocorrelation for the three models, except for cross-sectional dependence in the energy diversification model. Therefore, we utilize the fixed effects estimator with Driscoll and Kraay (1998) standard errors to address these issues. The results are presented in Table 7.

In Table 7, Column 1 presents the main regression results, focusing on overall financial development, while Columns 2 and 3 delve into financial institution and market development.

In the per-capita income model (Part A), a significant and positive trend coefficient implies an average annual growth rate of 19.7%, reflecting the positive contribution of technological progress to the formation of convergence clubs for per capita income among OECD countries. However, the negative and statistically insignificant coefficient for one lag of Energy HHI suggests a limited impact of energy diversification on per capita income in OECD countries. The positive and significant coefficient for one lag of Financial Dev shows the positive influence of financial development on per capita income, with financial institution development exhibiting a stronger effect than financial market development. Additionally, positive and significant coefficients for Labor Participation, Capital Formation, and International Trade indicate their contributions to wealth improvement, aligning with Gozgor and Paramati (2022) and Ahmed et al. (2022).

In the financial development model (Part B), the statistically insignificant trend coefficient suggests no improvement in OECD countries’ financial systems from 1997 to 2021. The insignificant coefficient of one lag of per capita income implies a minimal contribution of economic development to financial development. The negative and significant coefficient for one lag of Energy HHI indicates a positive influence of energy diversification on financial development. The positive and significant coefficients for HDI and Capital Formation highlight the positive contribution of human development and physical capital to the development of financial systems in OECD countries, aligning with studies by Ngo et al. (2022), Ngoc et al. (2024), and Raifu et al. (2023).

In the energy diversification model (Part C), the negative and significant trend coefficient reflects an annual decline in energy concentration, indicating enhanced diversification. The negative and significant coefficient for one lag of Financial Dev suggests that better financial system development fosters improved energy diversification. This relationship is primarily driven by financial institution development, as only the coefficient for FI Dev is negative and significant. Rising oil prices and higher human development index also correlate with increased energy diversification. Higher oil prices make fossil fuels less competitive, prompting a shift in investment and consumption towards non-fossil fuel energies. A higher HDI reflects greater awareness of energy security and environmental challenges, leading to more investment and consumption of non-fossil fuel energies, thus enhancing energy diversification.

However, the positive and significant coefficient for per-capita income suggests it acts as a barrier to energy diversification. We suspect a non-linear relationship, as suggested by the environmental Kuznets curve framework and Nibedita and Irfan (2024) for E7 economies. Part D confirms this non-linear impact, revealing a U-shaped effect, with the per-capita income turning point of $67,112.8 per year (constant 2015) ^[4]. Specifically, when per-capita income is below $67,112.8 per year, higher income reduces energy diversification. When per-capita income exceeds $67,112.8 per year, higher income increases energy diversification.

Our study first found that, assuming uniform growth, energy diversification, financial development, and per capita income across 38 OECD countries tend to converge toward a common level with decreasing variability. However, when accounting for individual country trajectories, countries do not converge to a single steady state but form distinct convergence clubs—three for energy diversification, three for financial development, and four for per capita income. This supports Hypothesis 1 and aligns with existing literature, such as Saba and Ngepah (2022b) and Saba and Ngepah (2022a) on renewable energy, Niţoi and Pochea (2016) and Cavallaro and Villani (2021) on financial indicators in Europe, and Cavallaro and Villani (2021) and Mazzola and Pizzuto (2020) on per-capita GDP in Europe.

Differences in natural resources, economic incentives, and social structures lead to varied growth paths, underscoring the importance of considering individual trajectories. The lack of overall convergence suggests that countries with lower per capita income, underdeveloped financial systems, and limited energy diversification are not closing the gap with more advanced nations. This could be due to insufficient technology sharing by advanced countries. Nevertheless, the emergence of convergence clubs indicates that countries within each club are becoming more similar, albeit at varying speeds, in terms of per-capita income, financial development, and energy diversification. While there is still considerable progress to be made, the findings suggest that lagging countries are actively leveraging their limited resources and capabilities to catch up with leading nations within their respective clubs.

Second, our study found that while energy diversification positively influences per capita income, the effect is statistically insignificant, different from Hypothesis 2a. This outcome can be attributed to the slow energy consumption growth in OECD countries, averaging just 1.24% annually from 1997 to 2022, compared to 5.93% in non-OECD countries ^[5]. Furthermore, OECD nations’ heavy reliance on fossil fuels, supported by well-established infrastructure, hampers the transition to renewable energy sources, restricting the potential economic benefits of energy diversification.

Third, our study found that financial development positively influences per capita income, supporting Hypothesis 4b and aligning with previous studies such as Aydin and Malcioglu (2016), Madsen and Ang (2016), Purewal and Haini (2022), and Hashemizadeh et al. (2023) on OECD countries. This indicates that financial development plays a critical role in driving economic growth in these nations.

Fourth, we found an insignificant effect of per capita income on financial development in OECD countries, which does not support Hypothesis 4a. This may be due to slower economic growth in OECD countries (averaging 2.04% annually) compared to the global average (3.01%) ^[6]. The robust regulatory frameworks in these countries, while ensuring stability, may also limit financial innovation and the ability to adapt to evolving market needs.

Fifth, our study found that energy diversification positively impacts financial development, supporting Hypothesis 3b. Although energy diversification is slow, it is efficient due to regulatory frameworks promoting SDG 7 (Affordable and Clean Energy). This efficiency has spurred innovation in financial products, such as green bonds and loans, and stabilized energy prices, which are beneficial for economic activities, further boosting financial sector growth.

Sixth, our study identified a positive impact of financial development on energy diversification, supporting Hypothesis 3a and aligning with the findings of Shahbaz et al. (2023) on Australia. The advanced financial systems and robust regulatory frameworks in OECD countries promote access to sustainable finance, facilitating a smoother and more efficient transition to renewable energy sources.

Seventh, our study initially found a negative impact of per capita income on energy diversification, supporting Hypothesis 2b and aligning with Shahbaz et al. (2023) on Australia. This suggests that fossil fuels remain the preferred choice for economic growth in OECD countries. However, further analysis revealed a U-shaped effect, with the turning point at $67,112.8 per year (constant 2015 USD). At higher income levels, households and corporations become more aware of environmental and energy security issues and have the financial and technological means to diversify their energy mix.

Finally, our study highlights the positive role of technological progress in fostering convergence in per capita income and energy diversification among OECD countries. Technological innovation drives economic growth (Cheng et al., 2021; Khan et al., 2024) and enables more efficient energy generation from diverse sources (Zhang et al., 2024; Zhao et al., 2024), which supports energy diversification. However, technology’s limited impact on financial development convergence may stem from the stability-focused financial systems in OECD countries post-financial crises.

This study examined whether energy diversification, financial development, and per-capita income in OECD countries converged over the period 1997–2021. Using the convergence tests of Phillips and Sul (2007, 2009), we found no evidence of overall convergence for these factors among OECD countries. However, we identified convergence clubs for each factor. We also investigated the determinants of convergence. Using Granger causality tests, we found bi-directional relationships between the three variables in the short run. Long-run panel regression analysis confirmed that technological progress significantly improved per-capita income and energy diversification. It also revealed bi-directional relationships between energy diversification and financial development, an uni-directional relationship from financial development to per-capita income, and a U-shaped effect of per-capita income on energy diversification, with a turning point at $67,112.8 per year. Additionally, fixed capital, labor, and international trade contribute to wealth improvement, while human development and fixed capital contribute to financial system development, and rising oil prices and human development contribute to energy diversification.

The findings of our study provide several important policy implications. While no clear evidence of overall convergence exists among the 38 OECD countries, the formation of convergence clubs for energy diversification, per-capita income, and financial development suggests that lagging countries can catch up with leading ones within their respective groups. By maximizing their resources and capabilities, lagging countries can enhance energy diversification, reducing dependence on fossil fuels and increasing the share of renewable energy. This transition improves energy security, productivity, and accessibility, aligning with SDG 7 (Affordable and Clean Energy). Additionally, increased renewable energy usage lowers greenhouse gas emissions, contributing to SDG 13 (Climate Action).

The significant role of technological progress in improving per-capita income and energy diversification highlights the need for policymakers in lagging countries to prioritize innovations in energy generation and production. Implementing incentives for technological advancements and sharing best practices in renewable energy adoption and economic strategies can accelerate progress.

The bi-directional relationship between energy diversification and financial development and the uni-directional impact of financial development on per-capita income underscore the importance of robust financial systems. Policymakers should develop sustainable finance mechanisms, such as green bonds, green loans, and carbon credits, by strengthening regulatory frameworks and supporting innovative financial products. Expanding access to affordable financial services can create jobs, boost productivity, and raise living standards, contributing to SDG 8 (Decent Work and Economic Growth).

The U-shaped effect of per-capita income on energy diversification indicates that lower-income OECD countries rely more on fossil fuels, while higher-income nations invest in renewables. Tailored policies are needed to accelerate energy transitions in lower-income countries. Higher-income nations can support this by sharing successful models, transferring clean energy technologies, and providing financial and technical assistance to reduce fossil fuel dependency.

The findings on human and fixed capital suggest that investing in these areas can significantly enhance per-capita income, financial development, and energy diversification, helping lagging countries close the gap with leading ones.

Data availability: Data in support of findings of this study is available from the corresponding author upon reasonable request.

Australia, Austria, Belgium, Canada, Switzerland, Chile, Colombia, Costa Rica, Czechia, Germany, Denmark, Spain, Estonia, Finland, France, United Kingdom, Greece, Hungary, Ireland, Iceland, Israel, Italy, Japan, South Korea, Lithuania, Luxembourg, Latvia, Mexico, Netherlands, Norway, New Zealand, Poland, Portugal, Slovakia, Slovenia, Sweden, Turkiye, United States