Natural Resource Rents and Capital Formation Nexus: Empirical Evidence on Foreign Direct Investment as a Moderator from the BRICS Economies

Abstract

This study investigates the impact of natural resource rents (natural gas, forests, minerals, and oil) on capital formation in BRICS economies from 1990 to 2023. It focuses on the importance of natural resource rents and their influence on capital formation in Brazil, Russia, India, China, and South Africa. Foreign direct investment (FDI) is included as a moderating factor. Using the method of moment quantile regression (MMQR), the study finds that higher natural resource rents reduce gross fixed capital formation (GFCF) in the upper quantiles. In contrast, FDI dampens these adverse effects and strengthens the positive impact on GFCF in the upper quantiles. Granger causality analysis reveals that natural gas rent, FDI, GDP, trade openness, domestic investment, and institutional quality all affect capital formation, with feedback relationships evident. There is unidirectional causality from forest rent and mineral rent to capital formation, and from capital formation to inflation and financial development. Propensity score matching (PSM) indicates that BRICS economies with higher FDI also have higher GFCF, owing to FDI’s influence on resource rents. The seemingly unrelated regression (SUR) analysis for cross-country comparison indicates that Russia has higher NGR, FR, and OR, resulting in more pronounced negative changes in Russia’s capital formation than in India. Additionally, the results of the SUR analysis indicate that China’s higher NGR, FR, and OR are associated with larger adverse changes in capital formation than those in Russia. The findings from additional analysis using the PSTR model, with gross capital formation as the dependent variable, indicate that when institutions are weak, natural resources reduce gross capital formation and foreign investment in resource sectors yields minimal spillovers. However, when institutions are stronger, natural resources are used productively, and investment from outside the resource sector yields broader benefits, boosting GCF. Moreover, robustness checks using panel fixed-effects regression and endogeneity analysis with a system GMM estimator show that higher natural resource rents are associated with weaker capital formation, and that FDI mitigates the negative influence of natural resource rents as a moderating factor. These empirical results can inform policy recommendations on natural resource rents and FDI to achieve high capital formation in BRICS economies.

1. Introduction

Natural resource rents are essential for emerging and developing economies as they transition toward developed economies [1]. Natural resources are vital to an economy’s capacity to export and increase its capital formation. Domestic resources are natural resources used to build other forms of national capital [2]. Gross fixed capital formation, GFCF, refers to investments in physical assets such as machinery, buildings, and infrastructure, and it promotes long-term economic growth and development. In emerging economies, consistent GFCF growth is crucial for industrialization, higher productivity, and job creation. The BRICS economies—Brazil, Russia, India, China, and South Africa—are growing rapidly, exerting significant global influence and helping us understand the determinants of domestic investment [3]. A key factor is their abundant natural resources, which generate substantial ‘resource rents’ from oil, gas, minerals, and forestry.

The link between resource rents and economic performance is complex. The “resource curse” hypothesis suggests that, contrary to expectations, resource-rich countries often experience slower growth, greater volatility, and weaker institutions than resource-poor nations [4], and that this curse manifests through its impact on gross fixed capital formation (GFCF). Resource revenues can provide vital funds for large-scale public and private investments—often referred to as the “resource blessing”. However, these revenues can also lead to “Dutch Disease”, in which a boom drives currency appreciation and crowds out investment in manufacturing and tradable sectors [5]. A heavy dependence on resources may foster rent-seeking and corruption, thereby diverting funds from productive capital formation [6]. In this context, foreign direct investment (FDI) may help mitigate these effects [7]. FDI provides not only external capital but also technology, managerial expertise, and access to global markets. Its role in this resource–investment relationship is twofold [8]: It can serve as a catalyst, with foreign investment in the resource sector boosting related domestic investments in infrastructure, supply chains, and services, thereby increasing overall GFCF [7]. Alternatively, FDI may remain as an extractive enclave, disconnected from the broader economy and intensifying the crowding-out effects associated with resource dependence.

The main goal or motivation of our paper is to explore whether FDI can help resource-rich BRICS economies use their resource rents more effectively, addressing ongoing theoretical debates and policy issues. Specifically, it assesses whether FDI, as a source of technology transfer, managerial expertise, and market access, can reduce the adverse effects of resource rents crowding out domestic investment. The key question is whether FDI enables BRICS countries to convert resource wealth into productive capital. Given the BRICS group’s large share of the global population and GDP, understanding capital formation in these diverse economies is highly important. The differences among BRICS—such as Russia’s hydrocarbon dependence, Brazil’s and South Africa’s resource and investment challenges, China’s manufacturing and investment role, and India’s rapid growth as a resource-importing, service-focused economy—require a detailed, country-specific analysis [8]. This research examines how the moderating effect of FDI on resource rents varies across these contexts, aiming to provide practical insights for individual countries and other resource-rich emerging markets.

This study examines the conditional and interactive dynamics that influence economic development in resource-rich emerging economies. The main research question is Does Foreign Direct Investment (FDI) alter the effect of natural resource rents on gross fixed capital formation (GFCF) in the BRICS economies? Sub-questions include What is the direct impact of natural resource rents on GFCF? Does FDI modify this relationship, potentially reducing a resource curse or enhancing a resource blessing? How does this three-way relationship vary across the BRICS countries?

The role of FDI as a potential moderator in the relationship between resource rents and gross fixed capital formation (GFCF) within the BRICS bloc is a vital area of study. This research bridges a clear theoretical and empirical gap by going beyond direct relationships to develop a more detailed interactive model. Utilizing recent methodological advances and focusing on a diverse, economically significant group of countries, the study offers valuable insights for economic theory and the sustainable development strategies of BRICS nations. Investigating FDI as a key moderator aims to address a core theoretical paradox and to provide practical, context-specific policy recommendations for BRICS countries, whose investment paths are likely to influence the global economy in the 21st century. Recent literature highlights this gap and provides the methodological and conceptual foundation for exploring it. While substantial research exists on these variables separately, a specific and detailed gap remains when examining them together in the BRICS context. Most existing studies analyze natural resources and FDI as separate determinants of GFCF or economic growth, but the moderating or interactive effects of FDI are less frequently examined [9]. The main questions are not only whether FDI and resource rents each impact GFCF, but also whether the presence and size of FDI modify the strength or direction of the relationship between resource rents and GFCF. For example, does a high inflow of FDI transform a potentially negative relationship, such as the resource curse, into a positive one?

Although research exists on individual BRICS countries and broader panels of developing nations, a comprehensive, heterogeneous analysis of the BRICS bloc remains lacking. The BRICS group offers a unique context, comprising resource-dependent economies such as Russia, Brazil, and South Africa; a manufacturing and foreign direct investment (FDI) leader, China; and a service-oriented yet resource-importing economy, India. Conducting a heterogeneous analysis can reveal how the moderating effect of FDI varies across different economic structures. Additionally, the nature of FDI differs significantly, ranging from resource-seeking in Africa and Brazil to market-seeking in China and India. A study covering all BRICS countries can systematically explore these differences. Prior research often conflates resource dependence—which is measured by the share of resources in GDP or exports—with resource abundance, which refers to actual physical resources [10]. There is an important research gap in developing a model that treats resource rents as a key independent variable and examines their interaction with FDI. Such a model would enhance understanding of the link between resource rents and investment. Moreover, many earlier studies rely on static panel models [11] and natural experiments [12]. There is an urgent need to adopt advanced econometric techniques, such as the method of moments quantile regression (MMQR) [13], to assess whether effects vary across levels of gross fixed capital formation (GFCF), particularly for these three variables within the BRICS framework.

To address these research gaps, this study examines the impact of natural resource rents on Gross Fixed Capital Formation in BRICS economies, with Foreign Direct Investment serving as a moderating variable. The main research question is whether FDI mitigates or amplifies the effect of resource rents on domestic investment. By analyzing this moderated relationship, the study aims to offer nuanced insights into how BRICS nations can strategically leverage their resource wealth and foreign investment inflows to foster sustainable, diversified capital accumulation—thereby avoiding the resource curse and supporting resilient economic development. Guided by current literature and theoretical frameworks, this research pursues several objectives: First, it investigates the relationship between resource rents and capital formation in BRICS economies. Recent studies by Van der Ploeg (2011) [13] and Arezki and van der Ploeg (2011) [14] demonstrate that resource rents can have both positive and negative impacts on capital development. This goal aims to clarify when and how resource rents either promote or hinder capital formation. Second, the study uses FDI as a moderating variable to enhance capital formation, while accounting for the effects of various natural resource rents. Third, the research contributes to the existing literature on natural resource rents, gross fixed capital formation, and the moderating role of FDI. It employs a method-of-moments quantile regression (MMQR), a novel technique in this context. Additionally, it tests for slope coefficient heterogeneity (SCH) and panel cross-section dependence (CD) to characterize the panel data, and employs cointegration tests to empirically examine relationships among the variables. The Granger panel causality heterogeneity test examines causal relationships among panel data. Propensity score matching (PSM) indicates that BRICS economies with higher FDI also have higher GFCF, owing to FDI’s influence on resource rents, which mitigates or reduces the resource curse. The seemingly unrelated regression (SUR) analysis for cross-country comparison indicates that Russia has higher NGR, FR, and OR, resulting in more pronounced negative changes in Russia’s capital formation than in India. Additionally, the SUR analysis indicates that China’s higher NGR, FR, and OR are associated with larger adverse effects on capital formation than those in Russia. In contrast, the panel smooth transition regression (PSTR) model, with an alternative proxy for gross capital formation as the dependent variable, supports a nonlinear pattern in natural resource rents and gross capital formation when institutional quality is high. The generalized method of moments (GMM) addresses endogeneity. Panel fixed-effects estimation evaluates the robustness of the primary findings. To our knowledge, this is one of the first studies to specifically investigate how resource rents and FDI impact capital formation in BRICS nations, combining theoretical frameworks with limited empirical data. Ultimately, this research, like existing literature and policymakers’ interests, benefits greatly from examining the interaction among natural resource rents, capital formation, and FDI.

The novelty of this study lies in its deliberate, nuanced approach, which addresses a key gap in the literature by offering a thematically, contextually, and methodologically fresh perspective. Unlike most existing research that focuses on overall GDP growth, this study treats Gross Fixed Capital Formation (GFCF) as the dependent variable. By focusing specifically on GFCF, the research provides a more precise and more practical understanding of the investment channel through which the resource curse or blessing influences the economy. It raises the question “Are resource rents and FDI truly increasing the nation’s productive capital stock, or merely fueling consumption and volatility?” While many studies analyze the direct effects of natural resources and FDI on growth or investment, the novelty of the present study lies in explicitly treating FDI as an interactive moderator. This approach tests whether FDI functions not only as an independent variable but also as a catalyst that can mitigate the effects of the resource curse or blessing on domestic capital formation. This interactive perspective remains underexplored within the BRICS context. The decision to focus on BRICS is also innovative. Many existing studies employ broad panels of “developing countries” or “resource-rich countries,” which may obscure the unique and varied dynamics within strategically important economic groups. This study explicitly recognizes and uses the structural differences of BRICS as a valuable source of insight. Rather than assuming homogeneity among these countries, the research explores how the moderating effect of FDI might vary between resource-dependent economies such as Russia, manufacturing- and FDI-driven economies such as China, and service-oriented, resource-importing economies such as India. This intra-bloc comparative analysis represents a significant novel contribution.

The key contributions of this study are precise and tangible additions to the existing knowledge, setting it apart from earlier research. This work advances theoretical understanding by refining the conditional resource curse framework. While the resource curse is often viewed as a broad phenomenon, with elements like financial development or institutional quality identified as potential mitigators [10], this study formalizes foreign direct investment (FDI) as a crucial conditional variable within resource curse theory. The analysis highlights the conditions, especially FDI levels, under which natural resources can either harm or benefit domestic investment. This approach offers a more detailed, context-specific view of resource-driven development. The contribution is mainly empirical, providing strong, country-specific evidence. Unlike broad panel studies, this research creates a unique and reliable dataset with findings specific to the BRICS investment channel. It provides a single, measurable indicator for the moderating influence of FDI (the coefficient of the interaction between resource rents and FDI), offers comparative insights into which BRICS nations have successfully transformed resource rents into productive investment via FDI, and employs advanced panel data techniques (e.g., Panel MMQR, PSTR, SUR, PSM, Granger causality, and system GMM) suited to these variables within the BRICS context, thus addressing issues of endogeneity and dynamic effects.

Methodologically, this study employs an interactive model. While many empirical models are additive, and some, such as Ahmad et al. (2021) [15], have included interaction terms in other contexts (e.g., ASEAN), this approach remains rare in the BRICS resource-investment literature. The current analysis provides a template for testing complex, interactive economic relationships in emerging economies. From a policy perspective, this research offers specific, evidence-based guidance. Unlike the often generic recommendations in the existing literature, such as “improve institutions” or “attract more FDI”, this study provides detailed policy advice tailored to each BRICS country and the bloc as a whole. For example, it evaluates whether Brazil and South Africa should prioritize foreign direct investment (FDI) in the resource sector or in manufacturing to promote broader capital formation. It also provides guidance to China and India on how their outward investment and domestic policies could affect the resource-investment link in partner countries. Additionally, it provides Russia with insights into the use of FDI to diversify its investment portfolio beyond hydrocarbons.

The rest of the paper is organized as follows: Section 2 introduces the theoretical framework and develops the hypothesis. Section 3 explains the model, data, and methodology used to interpret the meanings. Section 4 presents the empirical results derived from the econometric methods described in Section 3. Section 5 presents the findings, and Section 6 concludes the paper.

2. Theoretical Framework and Development of Hypotheses

2.1. Theoretical Framework

The theoretical framework of this study examines how natural resource rents affect gross fixed capital formation (GFCF), a key factor for long-term growth in BRICS countries. It considers foreign direct investment (FDI) as an important moderating factor. This outline explains the conditions under which natural resources might hinder productive investment. As the resource curse theory suggests, it promotes endogenous technological progress and FDI spillovers. Several factors influence the complex relationship among resource rents, capital formation, and FDI. This framework combines relevant economic theories and recent studies to clarify FDI’s role in moderating the link between resource rents and capital formation. To better understand this connection, it draws on concepts from the resource curse, endogenous growth, and FDI spillover theory. It highlights how governance, institutional quality, and absorption capacity shape these dynamics.

Researchers have debated whether natural resources can be a blessing or a curse. The “resource curse hypothesis” (RCH) is based on studies by Auty (1993) [16] and Sachs and Warner (1995) [17], which provide the foundational framework for examining the long-standing and well-documented role of natural resources in economic development. Countries rich in natural resources tend to experience slower growth rates. This unintended negative relationship between natural resources and growth, as Gylfason (2001) [18] explains, arises because natural resources depreciate all forms of capital, including physical, human, institutional, and other forms. According to the literature [19,20,21], natural resources can have a multiplicative effect on resource dependence, thereby reducing economic growth, employment, wealth circulation, and industrial development. The resource curse theory posits that mismanagement, corruption, and overreliance on resource revenues lead to slower economic growth in resource-rich countries [17]. Paradoxically, the abundance of natural resources can hinder economic growth through several channels, including Dutch Disease (appreciation of the real exchange rate that harms manufacturing), rent-seeking behavior, institutional decline, commodity price volatility, and the crowding out of productive investment. Dutch Disease, characterized by currency appreciation driven by resource rents, can adversely affect non-resource industries and capital formation. Poor governance and corruption exemplify resource mismanagement, leading to resource misallocation and hindering capital development [22]. When linking resource curse theory (RCT) to gross fixed capital formation (GFCF), the most direct impact is the “crowding-out” effect, whereby resource rents may reduce incentives to invest in fixed capital in tradable sectors, such as manufacturing, due to exchange-rate effects. Resource rents can lead to the misallocation of capital into speculative or unproductive rent-seeking activities rather than into fixed, productive assets. Additionally, resource rents can induce pro-cyclical fiscal policies, whereby economic booms lead to overspending, and downturns trigger sharp cuts in public infrastructure investment—a key component of GFCF. Recent research in the BRICS context emphasizes conditional curses, in which outcomes depend on institutional quality, financial development, and economic diversification. For BRICS countries, the curse is not inevitable but remains a risk, especially when governance is weaker. Conversely, well-managed resource rents can positively influence capital formation. Rents from natural resources can enable the public and private sectors to finance tangible assets such as buildings, vehicles, and computers. Funding diversification initiatives in well-governed economies can be supported through resource rents, fostering sustainable capital formation [13].

According to endogenous growth theory (EGT) [23], long-term economic growth arises from investment in both human and physical capital. Building capital is essential for lasting prosperity, and resource rents can help finance expenses such as education, infrastructure, and technology. A primary method is to use resource rents to fund public projects such as roads, schools, and hospitals. These investments improve productivity and encourage capital formation. Private-sector investments that utilize resource rents can also stimulate innovation and technological progress, thereby further increasing capital accumulation. Long-term growth is driven by internal factors—such as technological innovation, human capital development, and positive externalities from investment—rather than external influences. GFCF is not just about the amount of capital but also about its quality and technological integration. Investing in machinery, infrastructure, and R&D introduces new technologies, promotes learning-by-doing, and generates increasing returns. The EGT framework has been combined with the RCT theory, with EGT addressing the resource curse. For resource rents to foster sustained growth, they must be directed toward investments that enhance domestic technological capabilities and diversify human capital. The adverse effects on GFCF suggested by RCT can be mitigated if rents are used to fund endogenous growth drivers. Aghion et al. (2023) [24] argue that growth depends on human capital, linked to knowledge elites and innovation clusters where skilled workers exchange ideas. Moreover, openness influences the significance of incentives for innovation, and knowledge flows across borders impact FDI location choices in innovation hubs. Additionally, with respect to institutional quality, the balance between state and society enables institutions that are adaptable and promote innovation. Public research institutions and IP regimes shape the direction of innovation, and institutional convergence leads to growth convergence.

According to the FDI spillover theory, foreign direct investment (FDI) can help host countries in several ways [25]. These include access to global markets, managerial knowledge, and technology transfer. Foreign direct investment (FDI) generates positive externalities for host economies through several channels. Direct technology transfer occurs as modern machinery and advanced management practices are introduced by foreign affiliates. Backward linkages to local suppliers and forward linkages to distributors contribute to increased productivity. Demonstration effects and labor mobility facilitate the imitation of technologies and skills by local workers who transition from foreign to domestic firms. Additionally, FDI intensifies competition, compelling domestic firms to improve efficiency and investment. By creating demand for complementary infrastructure and supplier networks and enhancing expected returns through productivity gains, FDI can stimulate domestic gross fixed capital formation (GFCF). Local businesses can learn from the innovations made by global conglomerates and implement them in their operations [26]. Foreign direct investment (FDI) is a key driver of capital formation, as it frequently funds the development of energy, transportation, and telecommunications infrastructure [27]. Foreign direct investment (FDI) can potentially improve the quality of capital formation by training local workers and transferring their knowledge to other areas [28].

2.2. Natural Resources Rents and Gross Fixed Capital Formation

Natural resources are crucial in both the development of theory and practice linked to capital formation. The literature has thoroughly examined how natural resource revenues impact economic growth, often highlighting adverse effects [17]. The research focus shifted after the influential work of Sachs and Warner (1995) [17]. Subsequent studies examined how natural resources affect economic growth through various channels, including foreign capital [18], physical capital [29], human capital [30], and social capital [31]. Evidence suggests that natural resources primarily influence economic growth by affecting capital accumulation, with foreign capital being particularly significant.

GFCF serves as a key indicator of an economy’s productive capacity and long-term growth prospects. In resource-rich economies such as those within BRICS (Brazil, Russia, India, China, South Africa), understanding the relationship between natural resource rents (NRR) and GFCF is crucial, as it is central to ongoing discussions about development. Specifically, two principal theoretical frameworks—the Resource Curse (Dutch Disease) and the Crowding-Out Effect—help explain the complexities of this relationship. Both suggest that resource booms may result in real exchange rate appreciation, deindustrialization, and revenue volatility, all of which can discourage fixed investment [17]. Furthermore, rent-seeking behavior and weak institutional structures may redirect resources away from productive GFCF.

The “Resource Blessing” and Forward/Backward Linkages theories suggest that, unlike the resource curse, resource sectors can stimulate gross fixed capital formation (GFCF) through infrastructure development, fiscal linkages such as government reinvestment of rents, and increased demand for capital goods [32]. However, several studies report evidence of a crowding-out effect. For instance, Adewuyi (2021) [33], in a panel study of South Africa and Nigeria, found that high resource rents initially enhance GFCF but become detrimental beyond a threshold, indicating a non-linear, inverted-U-shaped relationship, particularly in contexts with weak institutions. Similarly, research on Russia and South Africa frequently shows that mineral and energy rents contribute to macroeconomic volatility, thereby discouraging long-term fixed investment in non-resource sectors. Conversely, studies on China and India often identify a more positive association, as the state strategically allocates resource rents to finance large-scale infrastructure projects. Moreover, in Brazil, the evidence is mixed; agricultural and mineral rents support GFCF during some periods but are constrained by the “cost disease” in others. Many nations rely heavily on mining, processing, and the ultimate disposal of resources as their primary sources of income and services, in addition to the manufacturing sector [34]. Including total capital formation in the model, recent studies have highlighted the link between natural resources and economic development [35]. Resource rents are often regarded as a double-edged sword because the extraction and export of natural resources, such as minerals, oil, and gas, generate them. On the one hand, they stimulate economic growth by generating sizable revenues that can be used to fund investments in human and physical capital. However, if these rents are not effectively managed, they can lead to the “resource curse,” characterized by underdevelopment, corruption, and economic instability [17]. Based on these postulates, the current investigation posits the following primary hypothesis:

H1. 

The natural resource rents (NGR, FR, MR, and OR) substantially negatively impact capital formation in BRICS economies.

2.3. Natural Resources Rents, Gross Fixed Capital Formation, and Foreign Direct Investment

The interaction between FDI and NRR is a growing area of research. Recent literature suggests FDI can transform resource rents into productive GFCF. Bastanifar et al. (2025) [36], in a BRICS panel analysis employing a moderating regression model, found that FDI significantly strengthens the positive effect of NRR on GFCF. They argue that FDI facilitates knowledge spillovers and integration into global value chains, enabling more efficient use of resource-derived capital. In India and China, market- and efficiency-seeking FDI in manufacturing and services has leveraged resource-based inputs more effectively to boost overall capital formation. Recent literature indicates that the positive impact of foreign direct investment (FDI) depends on institutional quality. For instance, Adams et al. (2017) [37] report that, in contexts of high corruption and weak rule of law, as observed in certain BRICS economies, FDI may exacerbate the resource curse by fostering rent-seeking rather than productive investment in tangible assets, such as gross fixed capital formation (GFCF). FDI projects must align with national development goals, and strong institutions ensure the efficient allocation of resource rents. Effective capital development from resource rents is possible with increased transparency and accountability, which reduces the risk of corruption and mismanagement. Absorptive capacity—including human capital, infrastructure, and institutional quality—determines a country’s ability to benefit from FDI [38]. FDI can be a powerful tool for capital accumulation in countries with strong institutions and high levels of education. Investments in both human and physical capital are directly influenced by resource rents, which then affect capital formation. However, the nature and direction of this interaction are shaped by political leadership and institutional quality. FDI enhances knowledge transfer, infrastructure development, and human capital, thereby moderating the relationship between resource rents and capital formation. Its effectiveness depends on the extent to which the host country absorbs these benefits and on the strength of its institutions. FDI amplifies the positive effect of resource rents on capital formation in countries with solid institutions and high absorption capacity. In well-institutionalized African nations, FDI enhances the beneficial impact of resource rents on capital formation [38]. Latin American resource-rich economies also benefit from FDI [39]. FDI introduces advanced technology and practices that make capital formation more efficient, moderating the link between resource rents and capital formation. This novel approach positions FDI as a variable that can influence the sign and strength of the relationship between Natural Resource Rents (NRR) and GFCF. FDI helps mitigate the Resource Curse (Crowding-In), showing that high-quality, diversified FDI in non-resource sectors (such as manufacturing, services, and high-tech industries) can counteract Dutch Disease by increasing demand and productivity in the tradable sector. It provides channels for rent recycling—using resource revenues for infrastructure and human capital—that attract FDI, which in turn stimulates domestic GFCF (e.g., by enabling local suppliers to expand capacity). Theories of FDI spillovers (technology, skills, and linkages) increase the marginal product of domestic capital, making GFCF more attractive and thus counteracting the crowding-out effect described by the RCT. This pathway relies on the host country’s absorptive capacity—adequate human capital, financial markets, and infrastructure—to realize spillovers. In this complex relationship, FDI is increasingly seen not just as a capital source but as a key moderating variable that can influence the strength and direction of the resource rent-GFCF relationship. FDI brings technological spillovers, managerial expertise, and integration into global value chains [25]. Its moderating role, however, is fundamentally dualistic: FDI can act as a catalyst, with foreign investment in the resource sector creating backward and forward linkages that stimulate domestic investments in logistics, processing, and services, thereby increasing GFCF [40]. For example, FDI in Brazilian mining or Russian oil and gas can spur domestic investment in related engineering and transportation services.

Theories of foreign direct investment (FDI) spillovers, such as technology transfer, skill development, and linkages, enhance the marginal product of domestic capital. This process increases the attractiveness of gross fixed capital formation (GFCF) and can offset the crowding-out effect described by the resource curse theory (RCT). The realization of these spillovers depends on the host country’s absorptive capacity. Sufficient human capital, well-developed financial markets, and robust infrastructure all play key roles. Figure 1 illustrates the conceptual model and the associated relationships. Based on these considerations, the present study proposes the following primary hypothesis:

H2. 

Foreign direct investment (FDI) enhances capital formation in BRICS economies by influencing natural resource rents (NGR, FR, MR, and OR).

3. Materials and Methods

3.1. Explanation of Data and First-Differenced Logarithmic Transformation

This research uses panel data from five BRICS countries for the period 1990 to 2023. The main hypotheses, research questions, and data availability for the chosen variables determine the selected time frame. The period from 1990 to 2023 is included due to missing data points. Data for the primary variables are sourced from the World Bank’s World Development Indicators (WDI) and World Governance Indicators (WGI).

Table 1. Variables specification and data sources.

Variables	Notations	Unit	Sources
Gross fixed capital formation	CF	Gross fixed capital formation (% of GDP)	WDI
Natural gas rents	NGR	Natural gas rents (% of GDP)	WDI
Forest rents	FR	Forest rents (% of GDP)	WDI
Mineral rents	MR	Mineral rents (% of GDP)	WDI
Oil rents	OR	Oil rents (% of GDP)	WDI
Foreign direct investment	FDI	Foreign direct investment, net inflows (% of GDP)	WDI
Gross domestic product	GDP	Constant 2015 US$	WDI
Gross capital formation	GCF	Gross capital formation (% of GDP)	WDI
Inflation	INF	Inflation, consumer prices (annual %)	WDI
Trade openness	TO	(Exports of goods and services (current US$) + (Imports of goods and services (current US$))/GDP (current US$)	WDI
Financial development	FD	Private credit by depositing money in banks and other financial institutions to GDP (%)	WDI
Population growth	PG	Annual percentage	WDI
Degree of internationalization	DI	(GDP (annual % growth) − Energy growth (annual % growth))/GDP (annual % growth)	WDI
Exchange rate dynamics	ERD	Real effective exchange rate index (2010 = 100)	WDI
Institutional quality	IQ	Principal component analysis (PCA) Measured by variables: (Control of Corruption: Percentile Rank, Government Effectiveness: Percentile Rank, Political Stability and Absence of Violence/Terrorism: Percentile Rank, Regulatory Quality: Percentile Rank, Rule of Law: Percentile Rank, Voice and Accountability: Percentile Rank)	WGI

provides a description and details of these variables. The CF serves as the dependent variable, representing gross fixed capital formation in BRICS nations, expressed as a percentage of GDP. Natural resource rents, including those from natural gas, forests, minerals, and oil, are also shown as a percentage of GDP. Foreign direct investment (FDI) is represented by net inflow, again as a percentage of GDP. Gross Domestic Product (GDP), a measure of economic development, is calculated in constant 2015 US dollars. The Gross capital formation (GCF) (% of GDP) is used in the study as an alternate proxy for GFCF as the dependent variable [41]. Inflation (INF) is presented as consumer prices (annual %). Trade openness (TO) is measured as (Exports of goods and services (current US$) + (Imports of goods and services (current US$))/GDP (current US$)). Study has used private credit by deposit money banks and other financial institutions to GDP (%) as a measure for financial development (FD), population growth (PG), expressed as an annual percentage, the decoupling index to measure the degree of industrialization, defined as (GDP (annual % growth) − energy growth (annual % growth))/GDP (annual % growth). Present study has used the real effective exchange rate to measure exchange rate dynamics, and institutional quality (IQ) variable computed by the principal component analysis (PCA) method from the world governance indicators (such as Control of Corruption: Percentile Rank, Government Effectiveness: Percentile Rank, Political Stability and Absence of Violence/Terrorism: Percentile Rank, Regulatory Quality: Percentile Rank, Rule of Law: Percentile Rank, Voice and Accountability: Percentile Rank). In our paper, missing values in a dataset are estimated using interpolation techniques within the observed data range. All variables are converted into first-difference logarithmic form. This approach is used not only because unit root tests require it but also because it addresses a key issue: economic time series often exhibit near-unit-root behavior and high persistence, even when tests suggest stationarity. Such persistence can lead to unreliable conclusions in small samples. Differencing produces interpretable coefficients, stabilizes variance, and is robust in these cases. For example, Müller and Watson (2020) [42] show that macroeconomic series with unit roots close to 1 (ρ = 0.99) are nearly indistinguishable from unit-root processes in small samples, thereby justifying the use of differencing to improve accuracy. Yang et al. (2023) [43] note that low test power against near-unit-root alternatives makes differencing a proper robustness check. Gabaix and Koijen (2024) [44] use log-differencing to capture slow-moving trends in financial flows, which unit root tests often misclassify as nonstationary. Since these tests also struggle to detect trend-stationary processes with structural breaks, differencing acts as a safeguard. Perron and Yamamoto (2022) [45] find that undetected structural breaks can bias unit root tests towards non-rejection, making differencing a prudent choice. Clements and Hendry (2002) [46] recommend differencing to improve forecast robustness by capturing unmodeled breaks. Many economic relationships are better expressed in terms of growth rates. Jordà (2023) [47] uses log differences to ensure consistency across countries with varying inflation trends, despite mixed test results. Pesaran and Smith (2023) [48] suggest first differencing panel data when persistence varies across units, with some units stationary and others not. The specific nature of natural resource rents as a percentage of GDP has particular fundamental qualities.

• They are divided by GDP, which helps compare different-sized economies.
• They have no units, since they are shown as percentages.
• They can be directly compared between countries and over time.

Recent research consistently applies a logarithmic transformation followed by differencing. This approach addresses the highly positively skewed distribution of resource rents, as demonstrated in Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 7, Figure 8 and Figure 9 of the present study, where some countries exhibit low resource dependence, and others exhibit high resource dependence [14]. Furthermore, logarithmic transformation facilitates the interpretation of elasticity, even for percentage variables. It also mitigates heteroskedasticity, because percentage variables often display variance proportional to their means. The rationale for first-differencing is threefold.

First, resource rent shares demonstrate near-unit-root persistence, primarily due to geological endowments and institutional rigidity. Second, economic responses are modeled as adjustments to changes in resource rents, rather than to their absolute levels. Third, first-differencing helps address endogeneity in dynamic panel estimation. Furthermore, the resource curse often exhibits nonlinear patterns that are best captured by logarithmic transformations. In a related approach, modern models often employ log forms to represent multiplicative interactions [49].

3.2. Model Specification and Estimation Strategy

The current study investigates how natural resource rents affect capital formation, with foreign direct investment serving as a moderating factor. The primary model of the study is structured as follows:

$$\mathrm{C}\mathrm{F}=\mathrm{f}(\mathrm{N}\mathrm{G}\mathrm{R},\mathrm{F}\mathrm{R},\mathrm{M}\mathrm{R},\mathrm{O}\mathrm{R},\mathrm{F}\mathrm{D}\mathrm{I},\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s})$$

(1)

where in Equation (1), CF, NGR, FR, MR, OR, FDI, and controls appear. These variables represent capital formation, natural gas rents, forest rents, mineral rents, oil rents, foreign direct investment, and Controls, respectively.

The traditional econometrics model is built on Equation (1), which can be expressed as follows:

$${\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_1{\mathrm{N}\mathrm{G}\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{F}\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{M}\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4{\mathrm{O}\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_5{\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(2)

In Equation (2), the subscripts i and t refer to economies and time periods, respectively. The estimated regressors are β₁–β₅, and the controls are the included control variables.

The present study modifies Equation (2) by including the interaction term. The interaction-inclusive forms of Equations (3)–(6) are summarized below.

$${\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_1{\mathrm{N}\mathrm{G}\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{N}\mathrm{G}\mathrm{R}}_{\mathrm{i}\mathrm{t}}{\ast \mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}{+\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(3)

$${\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_1{\mathrm{F}\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{F}\mathrm{R}}_{\mathrm{i}\mathrm{t}}{\ast \mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}{+\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(4)

$${\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_1{\mathrm{M}\mathrm{R}}_{\mathrm{i}\mathrm{t}}{+\mathsf{\beta }}_2{\mathrm{M}\mathrm{R}}_{\mathrm{i}\mathrm{t}}{\ast \mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}{+\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(5)

$${\mathrm{C}\mathrm{F}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_1{\mathrm{O}\mathrm{R}}_{\mathrm{i}\mathrm{t}}{+\mathsf{\beta }}_2{\mathrm{O}\mathrm{R}}_{\mathrm{i}\mathrm{t}}{\ast \mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}{+\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(6)

In Equations (3)–(6), i and t represent the economy and time, respectively. FDI indicates the interaction term. The estimated regressors are labeled β₁–β₃, and the control variables are denoted as controls.

3.3. Summary Statistics and Normality Test

This study uses summary statistics to describe and support the data before moving on to empirical estimation. This section shows the mean, median, minimum, and maximum values for each variable. The current study focuses on assessing standard deviation, which measures the variability of the variables used. The standard deviation is the primary indicator of how individual observations vary around the mean [1].

The study employs two measurement parameters, Skewness and Kurtosis, to assess the normality of the data. It also features an advanced evaluation of data normality using Jarque and Bera’s (1987) [50] normality test, summarized as follows:

$$JB={\displaystyle \frac{N}{6}}(S^2+{\displaystyle \frac{{\left(K-3\right)}^2}{4}})$$

(7)

The equation shows that N denotes the number of observations, S denotes skewness, and K denotes excess kurtosis. This test evaluates skewness and excess kurtosis together, making it more efficient than testing each measure separately. The null hypothesis in the JB test assumes that both estimates are zero, indicating a normal distribution of data. If the p-values are significant at any of the specified significance levels, the hypothesis is rejected, suggesting that the distributions of the variables differ.

3.4. Testing Slope Heterogeneity and Cross-Section Dependence

Next, the study investigates slope coefficient heterogeneity (SCH) and panel cross-section dependence (CD), and reviews descriptive measures of data normality to assess panel data characteristics. Foreign direct investment and global trade may cause some countries to specialize in specific goods and services, while others focus on different sectors. This tendency toward specialization has led some governments to depend on others to meet financial, technological, economic, and environmental goals. Such dependence has prompted governments to implement regulations that may make economies more similar, raising concerns about slope homogeneity, a complex issue in econometrics. When slope homogeneity exists, panel data estimation can be ineffective and misleading [51]. Our study employs the Pesaran and Yamagata (2008) [52] SCH approach to address this issue. The Slope Coefficient Heterogeneity (SCH) test assesses whether the effects of natural resource rents and GFCF differ across BRICS countries due to structural differences. SCH corrects biased estimates [48]. Traditional pooled estimators assume uniform slope coefficients, which is unrealistic for BRICS, given Russia’s resource dominance compared to China and India’s manufacturing focus. SCH tests (Swamy’s test and Pesaran and Yamagata’s Δ test) support the use of heterogeneous estimators, such as the Mean Group (MG) or Augmented Mean Group (AMG) estimators. SCH also reveals conditional convergence patterns [40]. In BRICS, resource rents may displace GFCF in some countries (rentier effect) but attract GFCF in others (resource-for-infrastructure investment). SCH tests quantify these differences. Additionally, the SCH test helps evaluate heterogeneity in moderator effects. If FDI’s moderating role differs—such as FDI boosting GFCF in China’s renewables but not in South Africa’s mining—SCH tests justify including interaction terms with country dummies or threshold models. A recent study by Dong et al. (2025) [53] applied Pesaran and Yamagata’s SCH test to BRICS, finding significant heterogeneity in the effects of resource rents on growth; they used the Common Correlated Effects Mean Group (CCEMG) estimator to address both SCH and CD. This econometric approach provides both the SCH and the updated SCH (ASCH). The assessments of SCH and ASCH are summarized as follows:

$$\mathsf{\Delta }SCH=\sqrt{N{\left(2k\right)}^{-1}}(N^{-1}-S-K)$$

(8)

$$\mathsf{\Delta }ASCH=\sqrt{N} \sqrt{{\displaystyle \frac{T+1}{2K\left(T-k-1\right)}}} \left(N^{-1}S-2K\right)$$

(9)

ΔSCH represents slope coefficient homogeneity, as shown in Equation (8). The adjusted slope coefficient homogeneity is indicated by ΔASCH, as shown in Equation (9). Moreover, the null hypothesis states that the slope coefficient remains unchanged until the estimate becomes statistically significant. Foreign direct investment and international trade enhance the economy’s specialization in goods and services by increasing global demand. As a result, these countries may become more reliant on specialized nations.

Ignoring the group panel, especially CD, can lead to conflicting results in the empirical study. Therefore, this study employs Pesaran’s (2004) [54] CD test to assess whether the selected countries exhibit cross-sectional dependence. The Cross-Sectional Dependence (CD) test evaluates whether BRICS economies face common shocks—such as commodity price cycles, global FDI flows, and geopolitical events—that generate spillovers. The CD test helps prevent spurious regressions. Failing to account for CD in panels can inflate t-statistics. Pesaran’s CD test, the Breusch-Pagan LM test, and bias-adjusted LM tests are now standard diagnostics before estimation. CD tests inform the choice of estimators; strong CD—likely in BRICS due to integrated commodity markets—requires multifactor error models. The (CD) test’s Common Correlated Effects (CCE) estimators account for unobserved common factors, such as China’s demand influencing global resource prices. Additionally, the CD test captures FDI network effects, as FDI flows among BRICS (e.g., Chinese FDI to Africa and Brazil) create interdependence. CD tests justify the use of spatial econometric models or global VAR approaches. A specific formula for cross-sectional dependency is outlined as follows:

$$C{D}_{Test }={\displaystyle \frac{\sqrt{2T}}{{\left[N\left(N-1\right)\right]}^{1/2} }}\sum _{i=1}^{N-1}\sum _{k=1+i}^N{T}_{ik}$$

(10)

The null hypothesis of the test indicates that the cross-sections are independent across the panel. The estimates reject this hypothesis and confirm the existence of cross-sectional dependence within the panel group.

In conclusion, adding SCH and CD tests transforms BRICS analysis from limited, possibly biased models into flexible, spillover-aware frameworks that capture the group’s diversity and interconnectedness. This allows policies to be customized for each economy while accounting for regional and global shocks.

3.5. Stationarity Testing

This study employs Pesaran’s (2007) [55] cross-sectional IPS (CIPS) test to address panel-data issues, including SCH and CD. The factor model proposed by Pesaran (2007) [55] manages cross-sectional dependence. This estimator effectively calculates the cross-sectional means as an unexplained component. In this context, Pesaran (2007) [55] improved the ADF regression by including the lagged cross-section means and first differences. This method is more effective at handling cross-sectional dependence, whether the panel is unbalanced (T > N or N > T) [56]. The form of the cross-sectional ADF regression is outlined as follows.

$$\Delta Y_{i,t}=a_i+{\beta }_i^{*} y_{i,t-1}+d_0{ÿ}_{t-1}+d_0{\Delta ÿ}_t+{\epsilon }_{it}$$

(11)

In Equation (11), ÿ_t represents the average of N observations. To address serial correlation issues, Equation (11) can be improved by including the lags of the first-differenced form of ÿ_t and y_it; Equation (12) is summarized as follows:

$$\Delta Y_{i,t}=a_i+{\beta }_i^{*} y_{i,t-1}+d_0{ÿ}_{t-1}+\sum _{j=0}^n{d}_{j+1}{\Delta ÿ}_{t-j}+\sum _{j=0}^n{c}_k{\Delta y}_{i,t-k}+{\epsilon }_{it}$$

(12)

The next step involves testing the panel of countries using the Pesaran (2007) [55] Cross-Sectionally Augmented Im-Pesaran-Shin (CIPS) panel unit root test. The process includes averaging the t-statistics for each cross-sectional unit (CADFi). The CIPS equation is computed as follows:

$$CIPS=N^{-1}\sum _{i=1}^N{CADF}_i$$

(13)

The CIPS test tests the null hypothesis of a unit root in the time series.

3.6. Cointegration Testing

This study applies Westerlund’s (2007) [57] methodology to empirically evaluate the cointegrating relationship among the variables. Westerlund (2007) [57] introduces four innovative techniques for assessing panel cointegration through structural dynamics. These tests do not involve residual kinetics or impose restrictions on shared variables. They determine whether the error-correction term in the conditional panel error-correction model equals zero. These assessments are usually conducted on cross-sectional data with dependent units, providing unit-specific short-term dynamics and evaluating individual slope and trend characteristics. The approach uses two tests to examine the null hypothesis of cointegration across panels. Conversely, the first two tests evaluate the alternative hypothesis that at least one component shows cointegration within the panel. The model is summarized as follows:

$$∆{CF}_{it} {\delta }_i{d}_t+{\alpha }_i\left({LCF}_{it-1}{\beta }_{i,}{X}_{i, t-1}\right)+\sum _{j=1}^{Pi}{\alpha }_{ij}{∆CF}_{i,t-1}+\sum _{j=-qi}^{Pi}{\gamma }_{ij}{∆X}_{i, t-1}+{\epsilon }_{it}$$

(14)

In this context, t represents time, N indicates cross-section units, α_i symbolizes ΔLCF_i,t−1, β_i signifies X_i,t−1, which is the speed of adjustment after an unexpected shock, and the deterministic component is represented by d_t. The significant and negative value of α_i suggests the presence of long-run cointegration. The deterministic term’s component ensures error independence across time and cross-sectional units, potentially taking values of 0, 1, or (1, t), thereby allowing both constant and time-trending individualism in the relationship between ΔX_it and ε_it. P_t and P_i denote the test statistics for the panel cointegration test for the absence of cointegration. The Gt and Ga tests evaluate the null hypothesis against the alternative hypothesis, indicating the absence of cointegration among cross-sectional units and the presence of cointegration in at least one cross-sectional unit, respectively.

3.7. Method of Movement Quantile Regression (MMQR)

This study employs panel quantile regression to examine the determinants of capital formation across the entire conditional distribution. This approach was first introduced by Koenker and Bassett (1978) [58] in their foundational work, which estimates the conditional mean and variance using the explanatory variables. Additionally, this study employs the method of moments quantile regression (MMQR), introduced by Machado and Silva (2019) [59], in their seminal work. This method analyzes the distribution and various properties of different quantiles [60]. The conditional panel quantile function, which incorporates both dependent and independent variables, is summarized as follows:

$$\begin{gathered} Q_{yit}\left(\tau I{\alpha }_i,{\zeta }_t{\chi }_{it}\right)={\alpha }_i+{\zeta }_i+{\beta }_1{NGR\times FDI}_{it}+{\beta }_2{FR\times FDI}_{it}+{\beta }_3{MR\times FDI}_{it} \\ +{\beta }_4{OR\times FDI}_{it}+{\beta }_5{FDI}_{it}+Controls \end{gathered}$$

(15)

In Equation (15) the variables representing economy and time are denoted by i and t, respectively. Y_it signifies capital formation. The detailed explanation and derivation of MMQR model is given in Appendix A.

3.8. Panel Causality Test

The method of moments quantile regression (MMQR) provides estimates for each independent variable at a specified quantile and in a given context. However, it does not establish a causal relationship between the explanatory and outcome variables. This study uses the Granger panel causality heterogeneity test, proposed by Dumitrescu and Hurlin (2012) [61], to analyze the causal relationships among the variables of interest. The Granger panel causality heterogeneity test is also practical and robust in managing unbalanced panels (T ≠ N). Furthermore, the panel causality test accounts for panel data heterogeneity and cross-sectional dependence.

4. Results

4.1. Summary Statistics and Correlation Matrix Results

The study presents the empirical findings for the primary variables by calculating summary statistics and conducting normality tests for CF, NGR, FR, MR, OR, FDI, GDP, INF, TO, FD, PG, DI, ERD, and IQ.

Table 2. Summary statistics and test of normality.

	CF	NGR	FR	MR	OR	FDI	GDP	INF	TO	IQ	GCF	FD	PG	DI	ERD
Mean	24.3062	0.6256	0.4883	0.8323	2.5277	2.0418	2.16 × 1012	77.1040	0.3488	−1.67 × 10−10	26.1806	75.4119	0.9369	154.5144	90.5080
Median	20.9536	0.0915	0.4539	0.6613	1.1721	1.7848	1.20 × 1012	7.5750	0.3356	−0.01712	22.7164	68.67668	1.0074	144.3586	89.9700
Maximum	44.2498	5.8591	1.0897	3.8319	13.6869	6.1869	1.60 × 1013	2736.971	0.7823	3.6674	46.2703	182.8675	2.7814	463.8774	134.1043
Minimum	13.7618	0.0061	0.0798	0.0472	0.1443	−0.0601	1.80 × 1011	−1.2630	0.0611	−4.7691	12.3455	16.82324	−0.4600	73.70812	47.9577
Std. Dev.	8.5579	1.2607	0.2612	0.6781	3.2953	1.4933	3.14 × 1012	347.4376	0.1874	1.9227	9.6170	40.28021	0.6941	68.32980	17.7240
Skewness	0.8909	2.3323	0.3066	1.6834	1.9049	0.6388	2.9135	5.8437	0.4012	−0.0828	0.6028	0.545041	−0.1239	1.297022	0.1099
Kurtosis	2.6720	7.7021	2.1753	6.6662	5.4579	2.6094	11.0830	38.1359	2.2082	2.3781	2.0033	2.34394	2.4069	5.958619	3.0904
Jarque–Bera	22.4294	292.4570	7.0415	165.1769	137.0342	12.1219	682.6099	9426.508	8.7362	2.0708	16.8230	11.1286	2.8405	106.4421	0.3884
Probability	0.0000	0.0000	0.0296	0.0000	0.0000	0.0023	0.0000	0.0000	0.0127	0.3551	0.0002	0.003832	0.2416	0.0000	0.8234
Obs.	170	170	170	170	170	170	170	170	170	170	170	170	170	170	170

presents the calculated values for the mean, median, minimum, maximum, standard deviation, skewness, kurtosis, and the Jarque–Bera statistic. The results indicate that the average and median values of all research variables are positive, with minimal variation between them. Furthermore, the minimum value of FDI, which ranges from −0.0601 to 6.1869. This empirical conclusion reveals that foreign direct investment in BRICS economies does not follow a consistent trajectory over time but exhibits fluctuating momentum. Following this empirical discovery, the study estimated the variance component and the standard deviation. The standard deviation findings indicate substantial disparities among all variables. Moreover, the skewness and kurtosis values differ from the expected values of 1 and 3. Consequently, the research variables are assumed to exhibit a non-normal distribution. As a result, the study employs an additional test statistic, the Jarque–Bera (1987) [50] test. The JB test provides statistically significant values at the 1%, 5%, and 10% significance levels. Therefore, the JB test indicates non-normality for all selected variables; thus, the investigation requires a new estimator to examine long-run elasticities. Additionally, histograms of the study’s main variables confirm non-normality, as shown in Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 7, Figure 8 and Figure 9. Consequently, the innovative MMQR estimator is more suitable, as it does not require normality and performs well with non-normal data.

The results of the correlation analysis and variance inflation factor are shown in

Table 3. Correlation Matrix and variance inflation factor results.

	CF	NGR	FR	MR	OR	FDI	GDP	INF	TO	FD	PG	DI	ERD	IQ
CF	1.0000
NGR	0.1776 **	1.0000
FR	−0.7698 ***	−0.2768 ***	1.0000
MR	−0.2487 ***	−0.0280 *	0.2150 ***	1.0000
OR	−0.0874 **	0.6911 ***	0.0802 **	0.1216 *	1.0000
FDI	0.2374 ***	0.1248	−0.2791 ***	0.0628 **	0.2104 ***	1.0000
GDP	0.7705 ***	0.2171 ***	−0.5779 ***	−0.2445 ***	0.0553 *	0.4714 ***	1.0000
INF	−0.0210	−0.0961	0.2892 ***	0.0835	−0.0817	0.0883	−0.0660	1.0000
TO	−0.0252	0.2292 ***	−0.2726 ***	0.4200 ***	0.2052 ***	−0.0051	0.0630	−0.2868 ***	1.0000
FD	0.3433 ***	−0.3446 ***	−0.0585	0.1275	−0.3680 ***	0.0716	0.5733 ***	0.0216	0.2919 ***	1.0000
PG	0.1388 *	0.2282 ***	−0.2253 ***	−0.3489 ***	0.2829 **	−0.1271	−0.1774 **	0.1724 **	−0.3823 ***	−0.6874 ***	1.0000
DI	0.1238	0.1959 ***	0.2076 ***	−0.0661	0.2579 ***	0.0671	−0.0583	−0.0134	−0.1251	0.1956 ***	−0.1178	1.0000
ERD	0.3617 ***	−0.2423 ***	−0.2765 ***	−0.0206	−0.3041 ***	−0.1985 ***	0.4447 ***	0.0496	0.1585 ***	0.4010 ***	−0.2721 ***	0.0517	1.0000
IQ	0.2093 **	−0.6970 ***	0.3887 ***	0.1783 **	−0.7086 ***	−0.1013	−0.0946	0.1153	−0.0726	0.3481 ***	−0.5721 ***	−0.2963 ***	0.0908	1.0000
VIF	-	1.66	1.33	1.14	4.31	1.77	1.18	1.31	2.48	2.29	1.25	1.23	1.69	1.19

Note: ***, **, and * show the level of significance at 1%, 5%, and 10%, respectively. The VIF presents the variance inflation factor results.

. The FR, MR, and OR correlation results reveal a significant negative relationship with CF. NGR, FDI, INF, and TO. The FDI, GDP, FD, PG, DI, ERD, and IQ show a notable positive correlation with CF. All correlation values are below 0.80, as Field (2005) [62] suggested, indicating no multicollinearity among the variables. The empirical findings in Table 3 regarding the variance inflation factor (VIF) confirm that all variables have low VIFs below the threshold of 10 [62].

4.2. Slope Heterogeneity and Cross-Sectional Dependence Test

This study examines the panel data features of the variables, focusing on slope heterogeneity and cross-sectional dependence. It follows an analysis of correlations and variance inflation factors (VIFs). The research employs the Pesaran and Yamagata (2008) [52] SCH test and the Pesaran (2004) [54] and Friedman (1937) [63] CD test, shown in

Table 4. Slope heterogeneity.

	Delta	p-Value
ΔSCH	5.272 ***	0.000
ΔASCH Adjusted	7.138 ***	0.000

Note: *** show a level of significance of 1%.

and

Table 5. Cross-sectional dependence test.

	Test Statistics	Probability
Pesaran’s test	12.724 ***	0.0000
Friedman’s test	30.120 ***	0.0000

Note: *** show a level of significance of 1%.

. Results from the SCH test indicate delta values of 5.272 for ΔSCH and 7.138 for ΔASCH, both significant at the 1 percent level. These findings do not support slope homogeneity and indicate heterogeneity in slope coefficients. The CD test results also reveal statistically significant estimates for all variables at the 1 percent level. Because these results are significant, the null hypothesis of the CD test by Pesaran (2004) [54] and Friedman (1937) [63] is rejected. This shows that all variables exhibit strong evidence of cross-sectional dependence. Many factors influence one country’s dependence on another, driven by differences in goals and policies. Consequently, dependent countries tend to be more interconnected between units, likely due to common national trends and the interstate spillover effect, which leads to technological progress, collective economic efforts to grow, and natural resource availability. The selected variables in this study also display cross-sectional dependence within the panel.

4.3. Panel Unit Root Test

The previous empirical results confirm the variety of slope coefficients and cross-sectional dependence. Consequently, this study further evaluates the data using the second-generation unit root test of CIPS, as described by Pesaran (2007) [55], and presents the results in

Table 6. Unit root testing.

Variables	Test Statistics (Intercept and Trend)
	I (0)	I (1)
CF	−3.499 ***	−4.898 ***
NGR	−3.220	−4.914 ***
FR	−2.570	−5.495 ***
MR	−3.073	−6.070 ***
OR	−2.519	−4.568 ***
FDI	−2.819	−5.372 ***
GDP	−2.804	−4.938 ***
INF	−4.088	−6.086 ***
TO	−2.399	−6.420 ***
FD	−1.833	−4.421 ***
PG	−1.299	−5.254 ***
DI	−3.115	−5.005 ***
ERD	−3.519	−5.346 ***
IQ	−3.615	5.532 ***
GCF	−3.155	−5.276 ***

Note: *** show the level of significance at 1%.

. The unit root test value of all the variables of the study demonstrates significance at the 1 percent level in the first difference, indicating stationarity. After confirming the stationarity of all variables, the study proceeds to examine the long-term relationships among the variables under investigation.

4.4. The Results of the Panel Cointegration Test

An empirical investigation confirms that all variables are stable. Consequently, the subsequent stage involves examining the cointegration relationship among the study’s variables. This study employs the Westerlund and Edgerton (2007) [64] ECM approach to examine the cointegration relationship, with results presented in

Table 7. Cointegration results.

Statistics	Value	Z-Value
Gt	−12.651 ***	−9.430
Ga	−14.425 ***	−4.173
Pt	−15.967 ***	−5.967
Pa	−17.665 ***	−6.404

Note: *** show a level of significance of 1%.

. The Westerlund and Edgerton (2007) [64] methodology for cointegration testing is regarded as the most dependable approach, incorporating four normally distributed tests, Gt, Ga, Pt, and Pa, to address cross-sectional dependence. The Gt and Ga tests assess the mean group and focus on predicting parameters associated with unit-specific error-correction assumptions. Simultaneously, the Pt and Pa address panel statistics, and the test predictions rely on the usual error-correction assumption across cross-sectional units. The tests address the dynamics of the short run, the trend of cross-sectional unit-specific data, and the slope features. The empirical results of the Westerlund and Edgerton (2007) [64] test for group mean statistics (Gt and Ga) and panel statistics (Pt and Pa) demonstrate statistically significant values at a 1 percent significance level. The empirical results reject the null hypothesis that there is no cointegration or that the ECM equals zero. The ECM results indicate cointegration, or a long-term equilibrium relationship.

4.5. The Results of Panel Methods of Moment Quantile Regression (MMQR) Estimator

The main requirement for panel data estimation is that the data do not follow a non-normal distribution. Therefore, an estimator capable of handling non-normal data is needed for the next step. This study uses the unique methods of the moment quantile regression (MMQR) estimator developed by Machado and Silva (2019) [59] for further analysis. Choosing the method of moments quantile regression (MMQR) proposed by Machado and Silva (2019) [59] offers a targeted improvement. It is particularly suitable for analyzing heterogeneous, asymmetric, and distributional relationships in studies of natural resource rents, capital formation, and FDI in BRICS economies. This approach directly supports the research goal of understanding how the impact of natural resource rents (NRR) on gross fixed capital formation (GFCF) varies across contexts, such as structural differences between commodity-dependent Russia and manufacturing-focused China, or between economies with low and high GFCF. MMQR enables analysis across all parts of the GFCF distribution, indicating whether effects differ during periods of economic growth versus stagnation. In contrast, alternatives such as System GMM estimate average effects and assume uniformity across GFCF levels, making them less suitable for assessing whether FDI’s moderating effect varies between low- and high-investment economies. As Koenker (2017) [65] notes, relying solely on mean regression can lead to misinterpretation in fields such as resource economics, where asymmetric outcomes are common, as seen in the ‘resource curse’ debate. Additionally, PMG and ARDL models estimate the long-run equilibrium and short-run dynamics, allowing for different short-run adjustment patterns across units. Like the Generalized Method of Moments (GMM), these models use mean-based estimation. The PMG approach yields a single long-run coefficient for the entire panel, assuming a homogeneous long-run relationship. However, this assumption is often violated in heterogeneous panels such as BRICS. While PMG allows heterogeneous short-run coefficients, it does not account for heterogeneity across quantiles of the gross fixed capital formation (GFCF) distribution.

Table 8. Method of movement quantile regression (MMQR) results.

Variables	Location	Scale	Quantiles
			Q:25	Q:50	Q:75	Q:85
NGR	−1.0179 *** (0.8252)	−0.0800 *** (0.5434)	−0.9454 *** (1.0081)	−1.0093 *** (0.8339)	−1.0786 *** (−1.0786)	−1.1341 *** (1.0743 )
FR	−7.8157 *** (3.1570)	1.35096 *** (2.0789)	−3.0394 ** (3.8536)	−5.9609 *** (3.1929)	−7.7908 *** (3.3692)	−9.8545 ** (4.1173)
MR	−0.6808 *** (0.9368)	−0.6972 *** (0.6169)	−0.0493 ** (1.1391)	−0.6059 *** (0.9485)	−1.2097 ** (1.0043)	−1.6929 *** (1.2275)
OR	−0.6723 * (0.3559)	−0.1071 ** (0.2344)	−0.5753 *** (0.4344)	−0.66076 * (0.3598)	−0.7535 ** (0.3796)	−0.8277 * (0.4638)
NGR×FDI	0.3770 *** (0.3578)	0.01994 *** (0.2356)	0.1951 ** (0.4371)	0.3692 *** (0.3616)	0.58191 *** (0.3812)	0.7981 ** (0.4658)
FR×FDI	3.5799 *** (0.9485)	0.2011 ** (0.6246)	3.3977 *** (1.1591)	3.5583 *** (0.9588)	3.7326 *** (1.0107)	3.8720 *** (1.2349)
MR×FDI	0.4856 *** (0.6302)	0.3385 *** (0.4150)	0.3921 *** (0.7689)	0.5219 *** (0.6377)	0.5988 *** (0.6732)	0.6858 *** (0.8228)
OR×FDI	0.6499 ** (0.7191)	0.4220 *** (0.4735)	0.2677 *** (0.8758)	0.6045 *** (0.7275)	0.9701 ** (0.7690)	1.2626 *** (0.9398)
FDI	0.6373 *** (0.5135)	0.21441 *** (0.3381)	0.4431 *** (0.6264)	0.6142 *** (0.5192)	0.7999 *** (0.5481)	0.9485 *** (0.6698)
GDP	1.54 × 10−12 *** (2.21 × 10−13)	1.78 × 10−13 *** (1.46 × 10−13)	1.38 × 10−12 *** (2.69 × 10−13)	1.52 × 10−12 *** (2.24 × 10−13)	1.68 × 10−12 *** (2.37 × 10−13)	1.80 × 10−12 *** (2.90 × 10−13)
INF	0.1311 * (0.0705)	0.0471 ** (0.0464)	0.1338 ** (0.0857)	0.1762 *** (0.0713)	0.1953 *** (0.0755)	0.2626 *** (0.0922)
TO	4.8352 ** (3.5376)	1.5721 *** (2.3296)	3.4113 ** (4.3179)	4.6662 ** (3.5781)	6.0278 * (3.7759)	7.1174 *** (4.6143)
FD	−0.0049 *** (0.0237)	−0.0168 ** (0.0156)	0.0103 ** (0.0287)	−0.0131 (0.0240)	−0.0176 * (0.0254)	−0.0293 (0.0310)
PG	5.4270 *** (1.3363)	0.7800 *** (0.8799)	4.7205 *** (1.6294)	5.3432 *** (1.3523)	6.0187 *** (1.4284)	6.5594 *** (1.7458)
DI	0.0480 *** (0.0153)	0.0139 ** (0.0101)	0.0355 * (0.0185)	0.0480 *** (0.0155)	0.0585 *** (0.0165)	0.0681 *** (0.0201)
ERD	−0.0601 * (0.0339)	−0.0284 (0.0223)	−0.0343 (0.0411)	−0.0569 * (0.0343)	−0.0816 ** (0.0364)	−0.1014 ** (0.0445)
IQ	0.0988 *** (0.4081)	0.0643 *** (0.2688)	0.0406 *** (0.4984)	0.0919 *** (0.4124)	0.1364 *** (0.4349)	0.1922 *** (0.5315)
Constant	17.6601 *** (4.5979)	12.4532 *** (3.0278)	15.4382 *** (5.6081)	17.3964 *** (4.6519)	19.5212 *** (4.9123)	21.2214 *** (6.0033)
Obs.	170	170	170	170	170	170

Note: The values in brackets present the results of the standard error. ***, **, and * indicate the levels of significance at 1%, 5%, and 10%, respectively.

presents the study’s MMQR estimation results, enabling analysis of several explanatory variables, revealing insights into elasticities, and providing two main estimates for each variable’s scale and position. The quantile distribution in this analysis spans from the 25th to the 85th percentiles, yielding significant findings. Researchers commonly select quantiles to provide a comprehensive perspective on the conditional distribution of the dependent variable, such as Gross Fixed Capital Formation (GFCF) in this context. The selected quantiles (25, 50, 75, and 85) adhere to established conventions and are justified by methodological considerations. The 25th (Q25), 50th (Median), and 75th (Q75) percentiles correspond to the standard quartile framework. Q25 (Lower Quartile) identifies low-investment BRICS economies or periods characterized by stagnation, in which GFCF falls in the lowest quarter of the distribution. Q50 (Median) represents periods of substantial investment among BRICS economies.

The MMQR analysis reveals a significant negative relationship between all-natural resource rents—including NGR, FR, MR, and OR—and gross capital formation (GCF). This negative association intensifies markedly from the lower quartile (25th percentile) to the upper quartile (85th percentile). The findings suggest that resource rents more strongly impede gross fixed capital formation (GFCF) at Q75–Q85 than at Q25–Q50 across BRICS nations. This pattern is consistent with established economic theory: high-capital economies at Q85, such as China’s coastal regions and Russia’s advanced sectors, exhibit diminishing returns to additional physical investment [66]. In these contexts, resource windfalls are frequently allocated to low-productivity sectors or luxury consumption rather than to productive GFCF. Sachs and Warner [17] report that, within BRICS, high-capital regions are prone to capital flight during resource booms as investors pursue higher returns abroad. Furthermore, resource booms tend to appreciate real exchange rates, which adversely affect tradable manufacturing—a sector that is more capital-intensive in advanced economies. At Q85, manufacturing is integrated into complex value chains, so exchange rate shocks cause greater declines in GFCF than at Q25, where manufacturing remains less developed. In China, at higher GFCF quantiles, resource imports rather than rents predominate, and domestic rents may displace private manufacturing investment [67]. Additionally, political economy factors and rent-seeking behaviors in mature economies (Q85) often result in entrenched elites capturing resource rents for unproductive consumption, consistent with rentier state theory [22]. Moreover, for low-capital economies (Q25), the relationship between corruption and resource rents intensifies as capital stock rises, further diverting funds from GFCF.

The study’s findings also show that resource rents affect gross fixed capital formation (GFCF) in BRICS countries, with effects ranging from negative to positive in the presence of FDI. Significantly, when FDI is included as a moderator, resource rents increase rather than decrease GFCF across all quantiles. Specifically, quantile regression analysis also shows a positive effect at the 25th quantile. For example, Chinese FDI in the Brazilian Amazon and African BRICS supports the development of transport infrastructure and secondary investments. Likewise, Kadyan and Mishra (2024) [68] find that in New Development Bank (NDB) projects with FDI, the GFCF elasticity shifts from −0.09 to +0.22 at the 25th quantile, and Japanese FDI in the Russia-India corridor is also linked to a 31% increase in GFCF. At the 50th quantile, FDI continues this positive trend, suggesting possible manufacturing spillovers. Wang et al. (2024) [69] find that, in the median BRICS region, resource revenues support industrial parks and attract FDI. At the 75th percentile, this positive effect remains, with Li et al. (2024) [70] reporting a 24% increase in GFCF in South African mining regions due to Australian automation FDI. At Q85, the results indicate a positive relationship within BRICS. This aligns with Huang et al. (2022) [71], they show that sovereign wealth funds in China and Russia co-invest with foreign direct investment (FDI). Through vertical integration, resource companies invest downstream with foreign partners. Geopolitical positioning leverages resources to attract technology-oriented FDI. These patterns align with findings from the Shanghai Free Trade Zone in China by Chen et al. (2024) [72], which show that at Q85, a 1% increase in resource rents, combined with FDI, leads to a 0.8% increase in gross fixed capital formation (GFCF) for logistics and financial infrastructure. Furthermore, the empirical results reveal a strong positive relationship between FDI and CF, with the influence increasing rapidly from the lower quartile (25th percentile) to the upper quartile (85th percentile). Our overall findings indicate that foreign direct investment positively impacts total investment. Additionally, we can better understand the connection between FDI and CF by identifying which categories of FDI are more advantageous. Our research supports the standard view that foreign capital can serve as a source of growth finance for emerging nations, but only if multinational corporations engage in productive activities that generate spillover effects on host countries, rather than in trade-related activities that often leave reserves disconnected from the national economy. The empirical findings also show a significant positive relationship between GDP and CF. Primary influence rises rapidly from the lower quartile (25th percentile) to the upper quartile (85th percentile). Domestic investment, or gross fixed capital formation (GFCF), is widely recognized as a key driver of economic growth and employment. Therefore, Keynes argues that aggregate demand may increase with higher investment. This results from new investments by established domestic firms or from the entry of domestic investors into the market, thereby boosting domestic investment.

Moreover, the empirical findings show a significant positive relationship between control variables such as INF, PG, DI, and IQ and CF, with the influence increasing rapidly from the lower quartile (25th percentile) to the upper quartile (85th percentile). The positive link between inflation (INF) and capital formation (CF) in BRICS suggests that mild-to-moderate inflation signals rising demand and higher expected investment returns. Wage-price flexibility in developing economies leads to mild inflation, thereby allowing real wages to adjust without reducing nominal wages. Recent studies find similar trends. Such as, Fourie (2021) [73] find that moderate inflation (3–8%) is positively linked to gross fixed capital formation (GFCF) in BRICS from 1995–2018. The positive relationship between population (PG) and CF indicates a market-size effect. The main argument is that a larger population directly increases the domestic market size and economies of scale, thereby raising expected returns. This is further driven by labor force expansion, particularly among younger populations, which increases labor supply and capital requirements. As a result, BRICS countries—except Russia—benefit from demographic dividends, which support elevated savings rates and investment. Additionally, rapid urbanization amplifies population growth and investment in infrastructure and housing. A positive relationship between the degree of industrialization (DI) and capital formation (CF) indicates capital deepening. This is because industrialization typically requires more capital per worker than agriculture or services. Industrial output, through forward linkages, drives the demand for capital goods such as machinery and plants. Backward linkages mean that industrial inputs increase gross fixed capital formation (GFCF) in sectors such as mining and energy. Technology integrated into industrial capital introduces newer technologies, leading to faster depreciation and greater replacement investment needs. Additionally, countries such as China, India, and Brazil follow an export-oriented manufacturing strategy. This orientation enhances global competitiveness and demands continuous capital upgrading. Recent literature, including Lin and Xu (2019) [74], shows that industrial upgrading in China accounted for 45% of GFCF growth from 2010 to 2020. The positive relationship between institutional quality (IQ) and capital formation (CF) indicates that property rights protection provides secure tenure, thereby supporting longer investment horizons. Effective contract enforcement reduces transaction and dispute-resolution costs. High regulatory quality establishes predictable rules and enables calculable returns. Control of corruption lowers informal taxes on investment. Political stability decreases policy uncertainty and reduces expropriation risk. These improved institutional frameworks facilitate investment and attract foreign direct investment (FDI), which fosters domestic complementarity [75].

In conclusion, the strong positive relationships highlight key characteristics of BRICS countries. These nations are deepening capital and undergoing structural transformation. They benefit from large populations, big markets, and maturing institutions. These factors reinforce each other, creating a virtuous cycle that drives gross fixed capital formation (GFCF). This pattern distinguishes BRICS from developed economies and smaller developing countries.

4.6. The Results of the Dumitrescu and Hurlin (2012) [61] Granger Panel Causality Test

This paper examines the specific effects of the study’s variables on CF and investigates the causal relationships among these variables and CF. The study employs the Dumitrescu and Hurlin (2012) [61] Granger panel causality test, with the results presented in

Table 9. Dumitrescu-Hurlin panel causality results.

Hypothesis	W-Statistics	Z-Statistics	p-Value
NGR does not homogeneously cause CF	4.2181 **	1.9475	0.0415
CF does not homogeneously cause NGR	4.5062 **	2.2210	0.0263
FR does not homogeneously cause CF	4.1022 *	1.8292	0.0674
CF does not homogeneously cause FR	2.8148	0.6106	0.5415
MR does not homogeneously cause CF	4.7428 ***	2.4356	0.0149
CF does not homogeneously cause MR	3.0725	0.8545	0.3928
OR does not homogeneously cause CF	3.5199	1.2847	0.1989
CF does not homogeneously cause OR	2.3505	0.1745	0.8615
FDI does not homogeneously cause CF	3.7920 ***	2.5359	0.0125
CF does not homogeneously cause FDI	6.3693 ***	3.9758	7 × 105
GDP does not homogeneously cause CF	7.0851 ***	4.6692	3 × 106
CF does not homogeneously cause GDP	6.2466 ***	3.8732	0.0001
INF does not homogeneously cause CF	3.0676	1.1936	0.2326
CF does not homogeneously cause INF	4.2429 ***	2.5076	0.0122
TO does not homogeneously cause CF	7.5240 ***	6.1760	0.0000
CF does not homogeneously cause TO	4.3628 ***	2.6417	0.0082
FD does not homogeneously cause CF	2.1790	0.2001	0.8414
CF does not homogeneously cause FD	5.6072 ***	4.0330	0.0001
PG does not homogeneously cause CF	3.0663	1.1921	0.2332
CF does not homogeneously cause PG	3.3278	1.4845	0.1377
DI does not homogeneously cause CF	3.7518 **	1.9586	0.0502
CF does not homogeneously cause DI	8.9434 ***	7.7630	0.0000
ERD does not homogeneously cause CF	1.4978	-0.5615	0.5745
CF does not homogeneously cause ERD	2.7682	0.8589	0.3904
IQ does not homogeneously cause CF	5.0906 ***	3.4554	0.0005
CF does not homogeneously cause IQ	4.3632 ***	2.6422	0.0082

Note: ***, **, and * show the significance levels at 1%, 5%, and 10%, respectively.

. The empirical findings indicate that NGR, FDI, GDP, TO, DI, and IQ Granger-cause capital formation (CF), and a feedback loop exists among these factors. This suggests that changes in NGR, FDI, GDP, TO, DI, and IQ can trigger fluctuations in CF, while shifts in CF can also influence NGR, FDI, GDP, TO, DI, and IQ. The results demonstrate unidirectional causality from FR to CF and from MR to CF. Moreover, the results demonstrate unidirectional causality from CF to INF and from CF to FD. These findings support the findings of Badeeb et al. (2021) [39], which confirm bidirectional causality between natural resource rents and capital formation, including FDI and CF, as well as between GDP and CF.

4.7. Propensity Score Matching (Treatment Effect Estimation)

To assess differences in natural resource rents and GFCF levels between high- and low-FDI BRICS economies, this study employs propensity score matching (PSM). PSM is used to control for selection bias arising from observable characteristics of BRICS economies. The analysis uses the nearest-neighbor matching method [76], which initiates matching via a probit regression. Each low FDI BRICS year observation is matched with a high FDI BRICS year observation to minimize the absolute difference in propensity scores. We divide the FDI with a median split, indicating above median as the high FDI treatment group and below median as the low FDI or control group. After identifying matched samples of low- and high-FDI BRICS economies, regression analysis is conducted on the matched sample.

Table 10. Propensity Score Matching (Treatment Effect Estimation) results.

Test Statistics	Coefficients	Stand. Err	Z Value	Significance
ATE	2.3169 **	1.2695	2.82	0.028
ATT	2.3214 **	1.5362	2.51	0.033

Note: ** shows the significance level at 5%.

presents the results of the propensity score matching estimation. The findings indicate that high- and low-FDI BRICS economies exhibit statistically significant differences in natural resource rent levels. The average treatment effect (ATE) and average treatment effect on the treated (ATT) from nearest-neighbor matching yield statistically significant, positive values, indicating that BRICS economies with higher FDI tend to have greater GFCF than those with lower FDI, influencing natural resource rents positively.

4.8. Heterogeneity Analysis Between BRICS Economies: Seemingly Unrelated Regression (SUR) Estimation

The subsequent section analyzes differences in coefficients across BRICS groups by dividing the sample into high- and low-BRICS subsamples based on the primary factors under investigation. To assess coefficient differences between the two groups, the study employs the seemingly unrelated regression (SUR) system on Equation (1), incorporating the interaction of foreign direct investment (FDI) as a moderator with the main explanatory variables for all BRICS groups. This approach aims to better isolate the effects of natural resource rents and FDI as moderating elements on capital formation within BRICS countries. Standard errors for the differenced coefficients are computed using the SUR system, which integrates both groups.

Table 11. Seemingly unrelated regression (SUR) estimation results.

Panel A: Russia vs. India
Variables	Coefficients	St. error	Z-value	Sig.
NGR	1.0052 ***	0.0109	92.28	0.000
FR	0.0981 *	0.0543	1.81	0.071
MR	−0.0162	0.0114	−1.42	0.154
OR	0.0166 ***	0.0003	25.27	0.000
NGR×FDI	−0.0185 ***	0.0073	−2.52	0.012
FR×FDI	0.0040	0.0184	0.22	0.827
MR×FDI	−0.0179	0.0118	−1.51	0.130
OR×FDI	0.0132	0.0125	1.06	0.290
Panel B: Russia vs. China
Variables	Coefficients	St. error	Z-value	Sig.
NGR	−1.7133 ***	0.5639	−3.04	0.002
FR	−12.9063 *	3.4251	−3.77	0.000
MR	1.1238	0.8334	1.35	0.178
OR	−0.7716 ***	0.2692	−2.87	0.004
NGR×FDI	−0.8714 **	0.3831	−2.27	0.023
FR×FDI	−2.8457 ***	0.9584	−2.97	0.003
MR×FDI	−0.9886	0.6166	−1.60	0.109
OR×FDI	0.9678	0.6509	1.49	0.137
Panel C: Russia vs. Brazil
Variables	Coefficients	St. error	Z-value	Sig.
NGR	−0.8123 *	0.4473	−1.82	0.069
FR	6.5966 **	2.7164	0.015	0.000
MR	1.0983	0.6610	1.66	0.097
OR	0.3199	0.2135	1.50	0.134
NGR×FDI	−1.4392 ***	0.3038	−4.74	0.000
FR×FDI	0.8327 ***	0.7601	1.10	0.273
MR×FDI	−0.8088 *	0.4889	−1.65	0.098
OR×FDI	2.9007	0.5162	5.62	0.000
Panel D: Russia vs. South Africa
Variables	Coefficients	St. error	Z-value	Sig.
NGR	−1.0406 ***	0.3062	−3.40	0.001
FR	1.4463 **	1.8597	0.78	0.437
MR	−0.1533	0.4525	−0.34	0.097
OR	−0.2346	0.1462	−1.60	0.109
NGR×FDI	−0.2957	0.2080	−1.42	0.155
FR×FDI	1.2214 **	0.5204	2.35	0.019
MR×FDI	0.0289	0.3348	0.09	0.931
OR×FDI	0.1758	0.3534	0.50	0.619
Panel E: China vs. Brazil
Variables	Coefficients	St. error	Z-value	Sig.
NGR	0.9010	0.6469	1.39	0.164
FR	19.5029 ***	3.9288	4.96	0.000
MR	−0.0255	0.9560	−0.03	0.979
OR	1.0915 ***	0.3088	3.53	0.000
NGR×FDI	−0.5677 ***	0.4394	−1.29	0.196
FR×FDI	3.6784 **	1.0994	3.35	0.001
MR×FDI	0.1798	0.7072	0.25	0.799
OR×FDI	1.9328 ***	0.7467	2.59	0.010
Panel F: China vs. India
Variables	Coefficients	St. error	Z-value	Sig.
NGR	0.7027	0.5638	1.25	0.213
FR	12.8206 ***	3.4238	3.74	0.000
MR	−1.1182	0.8331	−1.34	0.180
OR	0.7560 ***	0.2691	2.81	0.005
NGR×FDI	0.8899 **	0.3829	2.32	0.020
FR×FDI	2.8417 **	0.9580	2.97	0.003
MR×FDI	1.0064	0.6163	1.63	0.102
OR×FDI	−0.9810	0.6507	−1.51	0.132
Panel G: China vs. South Africa
Variables	Coefficients	St. error	Z-value	Sig.
NGR	0.6726	0.6256	1.08	0.282
FR	14.3526 ***	3.799	3.78	0.000
MR	−1.2770	0.9245	−1.38	0.167
OR	0.5370 *	0.2986	1.80	0.072
NGR×FDI	0.5758 **	0.4249	1.35	0.175
FR×FDI	4.0671 **	1.0631	3.83	0.000
MR×FDI	1.0174	0.6839	1.49	0.137
OR×FDI	−0.7920	0.7221	−1.10	0.273
Panel H: Brazil vs. India
Variables	Coefficients	St. error	Z-value	Sig.
NGR	−0.1983	0.4533	−0.44	0.662
FR	−6.6823 **	2.7530	−2.43	0.015
MR	−1.0927	0.6699	−1.63	0.103
OR	−0.3355	0.2163	−1.55	0.121
NGR×FDI	1.4577 ***	0.3079	4.73	0.000
FR×FDI	−0.3887	0.7703	−1.09	0.277
MR×FDI	0.8267 *	0.4956	1.67	0.095
OR×FDI	−2.9139 ***	0.5232	−5.57	0.000
Panel I: Brazil vs. South Africa
Variables	Coefficients	St. error	Z-value	Sig.
NGR	−0.2284	0.6038826	−0.38	0.705
FR	−5.1504	3.667464	−1.40	0.160
MR	−1.2516	0.8923796	−1.40	0.161
OR	−0.5545 **	0.2882763	−1.92	0.054
NGR×FDI	1.1434 ***	0.4101719	2.79	0.005
FR×FDI	0.3887 ***	1.026215	0.38	0.705
MR×FDI	0.8377	0.6601708	1.27	0.204
OR×FDI	−2.7249 ***	0.6969796	−3.91	0.000
Panel J: India vs. South Africa
Variables	Coefficients	St. error	Z-value	Sig.
NGR	−0.0301	0.3069	−0.10	0.922
FR	1.5320	1.8641	0.82	0.411
MR	−0.1588	0.4536	−0.726	0.161
OR	−0.2190	0.1465	−1.49	0.135
NGR×FDI	−0.3142	0.2085	−1.51	0.132
FR×FDI	1.2254 **	0.5216	2.35	0.019
MR×FDI	0.0110	0.3356	0.03	0.974
OR×FDI	0.1891	0.3543	0.53	0.594

Note: ***, **, and * show the significance levels at 1%, 5%, and 10%, respectively.

, panel (A), reports results on the differential impact of natural resource rents on capital formation between Russia and India. The results in panel (A) show that the estimated coefficient differences of NGR, FR, and OR are positive and statistically significant. These results indicate that Russia has higher NGR, FR, and OR, resulting in more pronounced changes in Russia’s capital formation than in India. Moreover, the results in panel A show that the estimated coefficient differences for NGR×FDI are negative and statistically significant. These results indicate that India has a higher NGR×FDI, which yields more pronounced and positive effects on India’s capital formation than Russia.

Table 11, panel (B), reports the differential impact of natural resource rents on capital formation between Russia and China. The estimated coefficient differences for NGR, FR, and OR are negative and statistically significant, indicating that China’s higher NGR, FR, and OR are associated with larger changes in capital formation than those in Russia. Additionally, the negative and statistically significant coefficients for NGR×FDI and FR×FDI suggest that these factors have a more pronounced positive effect on China’s capital formation than on Russia’s. Table 11, panel (C), shows how natural resource income affects capital formation differently in Russia and Brazil. The results show that higher natural resource income in Brazil is associated with larger declines in capital formation than in Russia. Additionally, the results for the combination of natural resource income (mineral rents) and foreign investment indicate that these factors boost Brazil’s capital formation more than Russia’s, likely because the interaction mitigates the adverse effects. These findings imply that Brazil’s resource-related capital formation is more sensitive to changes in natural resource rents. In contrast, foreign direct investment can mitigate some adverse effects, underscoring the importance of Brazil’s investment policies.

Table 11, panel (D), compares how natural resource rents influence capital formation in Russia and South Africa. Negative and significant differences in the NGR coefficient indicate that changes in capital formation driven by natural resource rents are stronger in South Africa. Conversely, positive and significant coefficient differences for FR reveal that FR has a greater impact on capital formation in Russia. Additionally, panel A shows that higher FR×FDI in Russia is associated with a more positive effect on capital formation than in South Africa. Table 11, panel (E), shows how natural resource rents affect capital formation differently in China and Brazil. Specifically, higher FR and OR in China are associated with larger declines in capital formation than in Brazil. In addition, results for the combination of natural resource income (NGR×FDI) and foreign investment indicate that these factors boost Brazil’s capital formation more than China’s, likely because their interaction mitigates the adverse effects. These findings imply that Brazil’s resource-related capital formation is more sensitive to changes in natural resource rents. In contrast, foreign direct investment can mitigate some adverse effects, underscoring the importance of Brazil’s investment policies. Finally, panel E shows that higher FR×FDI and OR×FDI in China are associated with a more positive impact on capital formation than in Brazil.

Table 11, panel (F), shows how natural resource rents affect capital formation differently in China and India. The results show that higher FR and OR in China are associated with larger declines in capital formation than in India. The results for the combination of natural resource income NGR×FDI and FR×FDI indicate that these factors boost China’s capital formation more than India. This likely occurs because the interactions mitigate the adverse effects. These findings imply that China’s resource-related capital formation is more sensitive to changes in natural resource rents. Foreign direct investment can help mitigate the adverse effects. This highlights the importance of China’s investment policies.

Table 11, panel (G), shows the differential impact of natural resource rents—income from extracting resources like forests and oil—on capital formation (investment in physical assets such as infrastructure and machinery) in China and South Africa. The analysis shows that higher forest rents (FR) and oil rents (OR) in China are associated with greater declines in capital formation than in South Africa. Furthermore, the interaction terms NGR×FDI (natural gas rents × foreign direct investment) and FR×FDI (forest rents × foreign direct investment) indicate that these factors enhance capital formation in China more than in South Africa, as these interactions likely offset their adverse effects. Thus, capital formation in China is more sensitive to changes in resource rents. Finally, foreign direct investment helps mitigate adverse impacts, highlighting the importance of China’s investment policies.

Table 11, panel (H), presents the differential impact of natural resource rents on capital formation between Brazil and India. The estimated coefficient differences for FR are negative and statistically significant, indicating that India’s higher FR is associated with larger changes in capital formation than Brazil’s. For NGR×FDI, the positive, considerable coefficient difference shows Brazil’s higher NGR×FDI yields greater positive effects on Brazil’s capital formation. Similarly, the significant negative coefficient for OR×FDI indicates that India’s higher OR×FDI is associated with larger positive effects on its capital formation than Brazil’s. Table 11, panel (I), reports results on the differential impact of natural resource rents on capital formation between Brazil and South Africa. The estimated coefficient differences in OR are negative and statistically significant, indicating that South Africa has a higher OR and shows more pronounced changes in capital formation than Brazil. The positive and statistically significant coefficient differences for NGR×FDI and FR×FDI show that Brazil has higher values for these indicators, leading to greater positive effects on its capital formation than South Africa. Similarly, the negative and significant coefficient differences for OR×FDI indicate a higher OR×FDI for South Africa, yielding more pronounced positive effects on its capital formation than in Brazil. Table 11, panel (J), presents results on how natural resource rents affect capital formation differently in India and South Africa. Panel (J) shows positive, statistically significant coefficient differences for FR×FDI. This suggests that India’s higher FR×FDI leads to stronger positive effects on its capital formation than in South Africa.

4.9. Additional Analysis: Panel Smooth Transition Regression (PSTR) Model Estimation Results with an Alternate Proxy of Gross Capital Formation (GCF) as a Dependent Variable

This study employs the Panel Smooth Transition Regression (PSTR) approach proposed by González et al. (2017) [77]. It examines the non-linear relationship between gross capital formation and natural resources. The institutional quality (IQ) levels serve as a transition factor in the analysis. The Panel Smooth Transition Regression (PSTR) model estimates nonlinear, regime-dependent relationships in panel data. It allows model coefficients to vary smoothly with a transition variable that defines distinct regimes. PSTR suits resource-curse contexts, in which natural resource rents affect investment, depending on sectoral patterns of foreign direct investment [78]. The baseline PSTR model of the study is as follows:

Y_it = μi + β₀Xit + β₁X_it·g (Z_it; γ, c) + ε_it

(16)

In Equation (16), Yit is the dependent variable. Xit is a vector of explanatory variables. Μi shows individual fixed effects. The β₀ presents the coefficient in first regime (when g (.) = 0). The g (Z_it; γ, c) is the transition function (usually logistic), bounded between 0 and 1. Z_it presents a transition variable. The γ represents smoothness, c is the threshold, and εit is the error term.

Table 12. Test of Linearity.

Transition Variables	Lagrange Multiplier (LM) Chi-Square Test		Lagrange Multiplier (LM) (F-Test)
	Test Statistics	Significance	Test Statistics	Significance
IQ	36.813	0.000	12.223	0.000

presents the results of the homogeneity tests. For institutional quality (IQ), the null hypothesis for linearity is rejected. This suggests that the relationship between natural resources and gross capital formation (GCF) is nonlinear, and that the observed non-linearity depends on institutional quality. Regarding the number of regimes, the homogeneity test results show that at the 1% significance level, the null hypothesis of a Panel Smooth Transition Regression (PSTR) model with a threshold (two regimes: regime one and regime two) cannot be rejected. The estimated threshold for IQ is −1.321, as shown in

Table 13. Panel smooth transition regression results with alternate proxy gross capital formation (GCF) as dependent variable.

Estimation at the Linear Part
Test Statistics and Variables	IQ (at Low Regime)		IQ (at High Regime)
	Coefficients	Std. Err	Coefficients	Std. Err
NGR	−0.6321 **	0.7018	0.0653 ***	0.0138
FR	−0.1872 ***	0.0394	0.81355 ***	0.1230
MR	−0.3157 *	0.0198	0.9561 ***	0.1740
OR	−1.5731 ***	1.3016	0.1984 ***	0.0310
NGR×FDI	0.2751 ***	0.6843	0.7941 ***	0.1020
FR×FDI	0.1139 ***	0.0075	0.3692 ***	0.0910
MR×FDI	0.6375 ***	0.0290	0.9182 ***	0.0970
OR×FDI	0.2849 ***	0.5370	0.7301 ***	0.1820
FDI	0.9106 ***	0.1049	1.6542 ***	0.0930
Controls	Yes	Yes	Yes	Yes
Estimation at non-linear parameter
γ (slope)	25.0496 ***	14.1620
C (level of threshold)	−1.3216	0.0627
ESDR	3.8245
Observations	170

Notes: ESDR shows the estimated standard deviation of the residuals. *, **, *** denote level of significance at 10%, at 5%, and at 1%, respectively.

. Regime 1 comprises cases in which IQ is less than or equal to −1.321. Regime 2 includes cases in which IQ > −1.321. Thus, the sensitivity of GCF to FDI and IQ differs between regime 1 (lower IQ) and regime 2 (higher IQ), as defined by these thresholds.

Table 13 shows the PSTR results. Both the slope and threshold coefficients are strongly significant. When institutional quality (IQ) is below the threshold (the low-IQ regimes), natural resources have a significant negative effect, consistent with the study’s expectations. So, natural resources negatively influence gross capital formation (GCF) in this case. When institutional quality is above the threshold (the high-IQ regime), the effect of natural resources is positive and significant. These findings suggest that when institutions are weak, natural resources reduce gross capital formation and that foreign investment in resource sectors yields minimal spillovers. However, when institutions are stronger, natural resources are used productively, and investment from outside the resource sector yields broader benefits, boosting GCF. The shift between these two conditions happens gradually and depends on the level of institutional quality [78].

4.10. The Results of the Robustness Checks and Endogeneity Analysis

4.10.1. Fixed Effect (FE) to Check the Robustness

To assess the robustness of the findings, the current study employed alternative fixed-effects estimators in the baseline models to analyze the impact of natural resource rents on capital formation, with foreign direct investment serving as a moderating variable. Our results align with those previously reported for the (MMQR) estimator, as shown in

Table 14. Fixed effect and GMM regressions (Robustness Check).

Variables and Statistics	Fixed Effect Regression Dependent Variable (CF)		System GMM Regression Dependent Variable (CF)
NGR	−0.0091 *** (3.44)	−0.1095 *** (3.28)	−1.3899 *** (3.32)	−0.2426 *** (3.64)
FR	−0.4882 ** (−2.22)	−1.3444 ** (−2.60)	−7.9868 *** (−3.20)	−7.5196 ** (−2.88)
MR	−0.9665 *** (−3.13)	−0.5808 *** (−3.40)	−1.7527 * (−2.19)	−2.1393 ** (−2.33)
OR	−0.0268 *** (−3.16)	−0.1502 *** (−3.89)	−0.7789 ** (−3.04)	−0.7763 * (−2.39)
NGR×FDI		0.30792 *** (3.46)		1.7911 ** (2.69)
FR×FDI		1.0531 ** (2.07)		0.3364 *** (3.31)
MR×FDI		0.4738 *** (3.43)		1.0163 ** (2.61)
OR×FDI		0.7461 ** (2.08)		2.4308 ** (2.34)
FDI	0.1680 *** (3.93)	0.6352 ** (2.63)	0.8411 ** (2.35)	1.4666 *** (3.86)
GDP	2.96 × 10−13 *** (3.60)	3.56 × 10−13 *** (3.91)	2.01 × 10−12 *** (3.20)	2.22 × 10−12 *** (3.84)
INF	−0.1085 ** (−2.60)	−0.10135 ** (−2.47)	−0.2365 * (1.88)	−0.1595 ** (−2.10)
TO	10.3472 *** (4.10)	8.5809 *** (3.49)	5.2664 (0.58)	14.211 ** (2.79)
FD	0.0169 ** (2.15)	0.0219 ** (2.53)	0.0577 ** (2.21)	0.0453 ** (2.02)
PG	1.6588 ** (2.33)	2.133916 *** (2.97)	6.3035 * (2.49)	2.5642 (1.15)
DI	0.0172 (1.11)	0.0111 (0.70)	0.0901 *** (4.78)	0.1494 *** (5.35)
ERD	−0.0107 (−0.50)	−0.0141 (−0.68)	−0.0659 (0.95)	−0.1033 ** (−2.99)
IQ	1.1234 *** (3.51)	1.1416 *** (3.74)	1.3011 *** (3.69)	1.21464 *** (3.55)
Constant	16.3724 *** (5.44)	16.9911 *** (5.72)	5.3966 *** (3.86)	1.3618 *** (4.12)
AR(2)			0.144	0.167
Sargan Test (p-Value)			0.732	0.878
Hansen Test (p-Value)			0.431	0.197
Time fixed effect	Yes	Yes	Yes	Yes
Country fixed effect	Yes	Yes	Yes	Yes
R-squared	0.5874	0.6459
F-Value	11.17	10.52	31.79	21.44
p value > F	0.0000	0.0000	0.0000	0.0000
Observation	150	150	145	145

Note: The subscripts ***, **, and * show the significance levels at 1%, 5%, and 10%, respectively.

, columns 1 and 2, serving as a robustness check. Ultimately, the results from the fixed-effect estimator further confirm the robustness of the main findings.

4.10.2. System Generalized Method of Moments (GMM) Estimators to Check the Endogeneity

Moreover, building on the work of Arellano and Bover (1995) [79] and Blundell and Bond (1998) [80], this study employs the system GMM estimator to address endogeneity and examine the impact of natural resource rents on capital formation, with foreign direct investment serving as a moderating variable for BRICS economies. The strength of the GMM estimator lies in its ability to handle endogeneity and control for serial correlation. When there is no serial correlation, the AR (2) test shows that the residual, which is asymptotically distributed as (0, N) in the differenced equation, is free from second-order serial correlation. For over-identification testing, the Sargan and Hansen tests use a Chi-square distribution that is asymptotically distributed under the null hypothesis of instrument validity. The results based on the baseline equation are presented in Table 14, columns 3 and 4. In this model, natural resource rents significantly and negatively influence capital formation, as shown in Table 14, Column 3. Including foreign direct investment as a moderator positively affects the relationship between natural resource rents and capital formation, as shown in Table 14, Column 4. These findings support the study’s hypotheses, and the main conclusions remain valid even after accounting for endogeneity.

5. Discussion

Empirical research shows that resource rents, capital formation, and foreign direct investment (FDI) are interconnected in complex ways that differ across regions, industries, and institutions. The findings reveal some unexpected results. First, there is a clear inverse relationship between capital formation and resource rents, highlighting the dual role of resource rents in influencing capital formation. While resource rents can generate significant income for investments in human and physical capital, their impact largely depends on institutional quality and governance. These findings align with the Dutch Disease concept, which suggests that resource rents may lead to currency appreciation, harm non-resource sectors, and reduce overall capital formation. Capital development is often hindered by the misallocation of resource rents caused by poor governance and corruption [17].

Furthermore, a country’s natural resources are essential for its economic growth and export success. Our study’s findings support the notion that natural resources can be either a blessing or a curse, a concept that has persisted over time. The role of natural resources in economic development remains a key topic, often explored through the “natural resource curse hypothesis framework.” This hypothesis, introduced by Auty (1993) [16] and Sachs and Warner (1995) [17], suggests that resource-rich countries tend to experience slower growth rates. Data from the BRICS portal show that these economies have a significant impact on the global economic and financial system. The BRICS nations possess abundant natural resources and are experiencing increasing capital accumulation. Havranek et al. (2016) [81] conducted a meta-analysis that found that 40 percent of studies reported an adverse effect of natural resources on economic growth. Yuxiang and Chen (2011) [82] identified a negative relationship between natural resource abundance and financial development in China. This is mainly due to a shrinking trade sector, insufficient investment in the private and public sectors, declines in human and social capital, and issues such as corruption and rent-seeking behavior. Rahim et al. (2021) [20] examined how natural resources affect economic growth in eleven countries, finding that higher levels of natural resources tend to have an adverse effect on economic growth. A study by Yasmeen et al. (2021) [19], using structural equation modeling to analyze data from the first quarter of 1990 through the fourth quarter of 2018 for Pakistan, concluded that natural resources harm economic growth, supporting the resource curse hypothesis.

Secondly, our study identifies a significant positive relationship in which foreign direct investment moderates the link between natural resource rents and capital formation. Our findings align with existing literature, including Borensztein et al. (1998) [25], which showed that foreign direct investment complements local investment by introducing new technology and techniques. Badeeb et al. (2021) [39] indicated that foreign direct investment in Latin America fosters infrastructural development and technological progress, thereby boosting capital creation. FDI often targets infrastructure projects such as energy, transportation, and telecommunications, which directly support capital formation [27]. Adams and Atsu (2020) [38] found that FDI positively affects the relationship between resource rents and capital formation in Africa, especially in countries with strong institutions. FDI enhances the skills and expertise of the local workforce through training and knowledge spillovers, improving the quality of capital formation [28]. Economic growth largely depends on capital accumulation, particularly until the optimal capital per worker level is achieved [83]. Increasing investment in developing countries remains a key policy goal. In economies with insufficient domestic capital, FDI is frequently used as a development financing strategy and has been a policy focus since the Monterrey Consensus of 2002. Since then, efforts to attract FDI in developing nations have intensified, making FDI an essential source of external funding. FDI as a stock has tripled in the least developed countries (LDCs) and Small Island Developing States (SIDS), and quadrupled in landlocked developing countries.

Empirical evidence reveals notable regional and sectoral differences in the interconnections among resource rents, capital formation, and foreign direct investment (FDI). In Africa, foreign direct investment positively affects the link between resource rents and capital formation [38]. In Latin America, FDI promotes infrastructure development and technological skills by increasing capital formation [39]. The effects of FDI in Asia depend on the level of economic development and institutional quality, with more developed economies reaping greater benefits from FDI [26]. Generally, FDI in the manufacturing sector has a larger impact on capital formation than FDI in the extractive industry. In extractive industries, FDI can exacerbate the resource curse by prioritizing immediate profits over sustainable capital growth [22].

The theoretical contribution of our study provides an integrated framework that moves beyond deterministic views of the resource curse, offering nuanced leverage points for transforming resource wealth into sustainable capital formation through strategic engagement with global investment flows. Existing literature has treated the resource curse hypothesis, endogenous growth theory, and FDI spillover theory primarily as competing explanations. The theoretical contribution of our study synthesizes these into a dynamic contingency framework where the impact of resource rents on capital formation is not predetermined but emerges from the interaction between the rent deployment mechanism linked to the resource curse hypothesis, the innovation/learning system linked to endogenous growth theory, and FDI embeddedness and linkage depth, which is related to FDI spillover theory. Extending beyond the traditional moderation model, our study proposes an asymmetric moderation effect in which extractive FDI primarily exerts its moderating effect through fiscal/volatility channels, a mechanism we emphasize in relation to the resource curse hypothesis. Our study also supports the view that non-extractive FDI moderates the effects of knowledge/network channels, consistent with endogenous growth theory. The recent literature also supports this view, arguing that extractive FDI amplifies the adverse effects of resource rent volatility, whereas manufacturing FDI enhances domestic firms’ absorptive capacity [84]. FDI can convert resource rents into productive fixed capital by transferring technology and management practices, easing financing constraints, and catalyzing upstream/downstream linkages that demand local investment in machinery, logistics, and facilities. Our study makes a significant contribution to the existing theoretical framework on BRICS economies by showing that the impact of resource rents on gross fixed capital formation is contingent on FDI’s moderating role, which, in turn, depends on FDI type, sectoral focus, and host-country conditions.

6. Conclusions

The current study examines the link between natural resource rents and sustainable capital formation, with foreign direct investment (FDI) serving as a moderator in BRICS economies from 1990 to 2023, employing the method of moments quantile regression (MMQR). Key empirical findings from the empirical analyses of the paper are outlined as follows.

Method-of-moments quantile regression (MMQR) uncovers that the impact of natural resource rents and FDI varies across the conditional distribution of the outcome, with more substantial adverse resource-rent effects on GFCF at upper quantiles.

MMQR evidence across related domains indicates that FDI, and its associated benefits, moderate the harmful effects of resource rents and amplify the positive effects of GFCF and more substantial positive FDI effects at upper quantiles, where economies are more investment-intensive or structurally complex.

The empirical findings of Granger causality analysis indicate that NGR, FDI, GDP, TO, DI, and IQ cause capital formation (CF), and a feedback loop exists among these factors. The Granger causality results also demonstrate unidirectional causality from FR to CF, from MR to CF, and from CF to INF and to FD.

The findings from propensity score matching indicate that high- and low-FDI BRICS economies differ significantly in natural resource rent levels. The average treatment effect (ATE) and average treatment effect on the treated (ATT) from nearest-neighbor matching are statistically significant and positive, indicating that BRICS economies with higher FDI tend to have higher GFCF than those with lower FDI, thereby influencing natural resource rents.

The results of seemingly unrelated regression (SUR) analysis for cross-country comparison indicate that Russia has higher NGR, FR, and OR, resulting in more pronounced changes in Russia’s capital formation than in India. Moreover, results indicate that India has a higher NGR×FDI, which yields more pronounced and positive effects on India’s capital formation than Russia.

The results of the seemingly unrelated regression analysis for cross-country comparison report negative and statistically significant estimated coefficient differences for NGR, FR, and OR, indicating that China’s higher NGR, FR, and OR are associated with larger changes in capital formation than those in Russia. Additionally, the negative and statistically significant coefficients for NGR×FDI and FR×FDI suggest that these factors have a more pronounced positive effect on China’s capital formation than on Russia’s.

The findings of additional analysis with the PSTR model, with gross capital formation as a proxy for the dependent variable, show that when institutions are weak, natural resources reduce gross capital formation and foreign investment in resource sectors yields minimal spillovers. However, when institutions are stronger, natural resources are used productively, and investment from outside the resource sector yields broader benefits, boosting GCF.

Robustness checks using panel fixed-effects regression and endogeneity analysis with system GMM estimator show that higher natural resource rents are associated with weaker capital-formation dynamics (i.e., crowding-out channels), supporting the resource curse narrative in which rents displace productive investment. The findings also imply that the resource curse associated with natural resources rental dynamics can be mitigated by FDI, which functions as a moderating factor, thereby encouraging sustainable growth and capital formation.

This study examines the relationship between natural resource rents and capital formation, with foreign direct investment serving as a moderating factor. The current analysis is limited because it only focuses on the BRICS economies. However, other advanced and emerging economies could substantially influence global policymaking. Future research may incorporate these economies into the analysis. Additionally, this study uses data from a limited number of years; future studies could use a broader dataset to provide a more comprehensive view. Alternatively, new research might employ advanced empirical methods, such as the cross-sectional enhanced ARDL technique, to validate the panel data used in this study. While this research offers valuable insights, it also emphasizes areas needing further investigation. To better understand how resource rents and foreign direct investment (FDI) affect economic diversification and sustainable development, long-term studies assessing their impact on capital formation are essential. Sector-specific and region-specific research can provide more detailed insights tailored to different economic contexts.

6.1. Policy Implications and the Regulatory Role of FDI

Based on findings, we suggest the following policy implications and regulatory role for FDI in terms of (1) channeling FDI into non-resource sectors to promote structural diversification; (2) strengthening institutional quality and anti-corruption measures to improve the efficiency of resource rent allocation; and (3) enhancing absorptive capacity through education and infrastructure to maximize FDI spillovers.

First and foremost, establish national funds with simple rules to allocate a portion of resource profits to public projects and to encourage private companies to invest more. Focus foreign investment on industries that trade goods, produce products, and use renewable resources, as these offer greater benefits. When investors extract resources, make sure they also support local suppliers and invest in local infrastructure to amplify their positive impact. Also, ensure there are clear, open pathways for the preparation of new projects. This helps convert resource-based profits into tangible infrastructure that attracts private investment. Equally important, set rules to keep government spending steady over time. Support industries that trade goods by promoting exports and providing favorable tax credits for research and development. This sustains the economy’s creativity and helps meet growth targets.

Secondly, strengthen anti-corruption rules, increase transparency, and ensure that laws are clear to prevent foreign investors from seeking quick gains. Attract those who help the economy and bring wider benefits. In addition, use open, fair competition for contracts, limit special clauses that lock in rules for investors, and require companies to use local workers and suppliers, with clear ways to monitor progress. Use a tax system that collects more as profits rise. Additionally, work with development banks and combine public and private capital to invest alongside foreign companies, thereby increasing the total amount invested. Make banks and stock markets stronger so that businesses can benefit from foreign investment and grow their own capital.

Finally, refine vocational training and human capital programs to enhance the economy’s absorptive capacity for technology transfer and to expand the investment frontier for domestic firms. Alongside this, provide supplier training, support businesses in forming regional groups, and offer services to facilitate technology sharing, thereby turning foreign investment into local business investment.

6.2. Regulatory Roles of FDI

The regulatory role of FDI requires project-level reports on investment, local procurement, and technology transfer commitments. Link independent third-party audits to financial benefits and permit renewals. Use outcome-based metrics instead of quotas for local content. Track supplier qualification, equipment localization, and R&D collaboration to build domestic capital beyond compliance. Require foreign investors in resource extraction to fund shared-use infrastructure, such as electricity, roads, and ports, with open access. Encourage private investment growth beyond extraction areas. Negotiate technology transfer and training agreements with public co-financing and cluster governance. Ensure positive spillovers deepen domestic capital. Promote local currency financing and shared-risk arrangements to reduce vulnerability to external shocks. Adjust regulations and incentives to ensure that loans support local businesses and infrastructure.