Domestic Financial Investment, Resource-Backed Capital Flows, and Economic Growth in Niger: An ARDL Approach

Abstract

Using the Autoregressive Distributed Lag (ARDL) model cointegration framework, this paper examines the long- and short-run impact of domestic financial investment and natural resource rents on economic growth in Niger within the period 1990–2021. The Bounds test confirms a long-run relationship among variables: F = 4.646 > 3.79 at 5%. Long-run results indicate that increasing domestic investment by 1% raises real Gross Domestic Product (GDP) per capita by approximately 0.30%, whereas 1% increase in natural resource rents leads to a reduction in growth by approximately 0.06%. At the same time, exports have a positive but very small effect, while imports and labor have negative long-run influences. Short-run dynamics further support a significant positive impact of domestic investment, at p = 0.0007, and a lagged effect of natural resources at p = 0.0308. The error-correction term is negative and significant, at −0.75, showing rapid adjustment toward equilibrium. Diagnostic tests confirm an absence of serial correlation and heteroskedasticity, while stability is confirmed by CUSUM and CUSUMSQ tests. The findings reveal a dualism in the growth path of Niger in that domestic financial investments favor sustainable expansion, whereas resource-based revenues undermine the growth process in the long run and call for financial market deepening and improved governance of resource revenues.

1. Introduction

Natural resources and domestic investment have contrasting roles in shaping long-term economic development, especially for resource-dependent economies. Whereas domestic financial investment has been widely regarded as a catalyst for capital formation, productivity, and financial deepening, revenues derived from natural resources expose economies to volatility, institutional weaknesses, and misallocation risks. These dual effects become particularly critical for low-income and commodity-dependent countries, where the potential to convert resource rents into productive investment remains limited.

Importantly, the economic consequences of natural resources and financial investment are very different across countries and depend on heterogeneities in institutional quality, financial market development, governance, absorptive capacity, and levels of diversification. Countries with strong institutions and effective revenue-management systems usually convert resource rents into productive assets in the long run [1,2], whereas countries with relatively weak financial systems may experience volatility, pressures for Dutch disease, and growth slowdowns [3,4]. In a similar vein, domestic investments tend to contribute to growth only when they obtain efficient financial intermediation and structural reforms; otherwise, their contribution can be feeble or even negative [5,6]

The paper thereby presents Niger as a particularly interesting case to investigate these dynamics. While highly dependent on uranium and oil revenues, the country still retains shallow financial markets with low credit mobilization. This dual structure therefore raises critical questions, such as whether domestic financial investment may mitigate the long-run risks of resource dependency and the extent to which resource-backed revenues contribute positively or negatively to economic performance. Given the prominence of these questions, there is a surprising scarcity, fragmentation, and macro rather than financial orientation of empirical evidence on Niger.

In addition, this study explicitly analyzes the financial transmission mechanism linking natural resources and domestic investment to growth-a dimension seldom tested in the existing literature on Niger or similar Sahelian economies. Although several studies recognize that resource rents may decrease investment efficiency due to volatility and misallocation, there are few empirical works that have directly tested this mechanism. This study bridges this gap in the empirical strategy by incorporating an extended ARDL specification that tests whether natural resource dependence moderates the growth impact of domestic investment, thus aligning the theoretical argument with the empirical design.

This study fills this gap by re-evaluating the resource-growth-investment nexus from a financial perspective, reframing domestic investment as a conduit of financial capital rather than a simple macroeconomic aggregate. Using the ARDL framework, the analysis distinguishes between the short-run and long-run effects, evaluates adjustment dynamics, and offers policy-relevant insights for resource-dependent low-income economies.

The novelty of this paper is threefold:

It simultaneously considers domestic financial investment and natural resource rents to explain long-run growth in Niger—a concept largely overlooked in previous studies.

It reframes natural resource revenues as a financial asset whose misallocation can distort domestic capital formation and financial stability.

It also provides empirical evidence, through ARDL cointegration, on the difference in long-run roles between domestic investment—impacting positively—and resource rents—impacting negatively—explaining the nuanced interpretation of Niger’s financial constraints. The rest of the paper is organized as follows: Section 2 reviews the literature, while Section 3 discusses data and methodology. Section 4 presents the empirical results, while Section 5 concludes with policy implications.

2. Literature Review

2.1. Natural Resources, Dependence, and Economic Growth

Natural resource abundance and economic growth have been a source of considerable debate. The resource curse argument, as developed by classical contributions such as Auty [7] and Sachs and Warner [8], posits that resource-rich countries often significantly underperform in terms of economic progress. These authors addressed the various structural mechanisms that weaken long-term development: falling competitiveness, weakened institutional capacity, incentives for rent-seeking behavior, and increased vulnerability to commodity price shocks. Cashin [3] provide evidence on how these commodity price cycles are converted into macroeconomic instability, volatility of the real exchange rate, and fiscal uncertainty that constrains growth in resource-dependent economies.

Empirical evidence from African economies stands in heavy congruence with this pessimistic interpretation. Tiba and Frikha [9,10], using cointegration-based analyses across 22 to 26 African economies, confirm that natural resource abundance has a negatively persistent impact on growth, which they ascribe to institutional fragility, corruption, and underinvestment in productive sectors. Satti [4] similarly document the negative long-run impact of oil revenues on Venezuela’s growth trajectory in a manner consistent with the Dutch disease hypothesis. Using data on China [11], further reinforce this pattern, showing that resource dependence depresses green productivity through a reduction in human capital, R&D investment, and institutional quality.

However, the literature is far from unanimous. A growing stream of research highlights that natural resources can contribute positively to economic performance conditional on governance, financial development, and diversification. Hassan et al. [12] show that in Pakistan, natural resources positively affect growth when combined with globalization and integration into world markets. Haseeb et al. [13], in using quantile-on-quantile regression, found positive growth effects across major Asian economies except India, where institutional weaknesses offset resource benefits. Ben-Salha et al. [2] find a long-run positive relationship between resource rents and growth for top resource-abundant countries, using the PMG estimator, though short-run effects remain insignificant.

Another recurring theme is the threshold role of financial development. Erdoğan et al. [14] identify two distinct regimes: when financial deepening is below 45%, resource exports fail to stimulate growth; when above 45%, resource revenues significantly promote output. This suggests that the ability of converting resource rents into productive investments depends critically on financial system depth, governance quality, and fiscal management.

Overall, the literature shows strong heterogeneity:

• Resource rents impede growth, the weaker the institutions and financial markets are.
• But they can stimulate growth when supported by robust governance and financial intermediaries.

This insight is indeed central to Niger, with financial markets remaining shallow, fiscal management facing constraints, and resource revenues being highly vulnerable to external shocks.

2.2. Domestic Investment and Growth Dynamics

Domestic investment is usually considered the cornerstone of economic development. Various theoretical underpinnings, such as Solow’s [15] neoclassical growth model, investment theory [16] and the endogenous growth framework [17], all point to the fact that long-term growth trajectories depend on capital accumulation, innovation, and productivity-enhancing investment.

Empirically, many studies confirm the positive impact of domestic investment. Ben Yedder et al. [18]. analyzing MENA countries, find consistent positive effects of domestic investment on economic growth, irrespective of patent activity. Bakari and Tiba [19] identify domestic investment as one of the long-run drivers of U.S. economic growth, together with consumer spending and exports. Shabbir et al. [20] demonstrate that within Pakistan, domestic investment positively contributes toward growth both in the short and long run.

Infrastructure-led investment also has a central place in development. Seidu et al. [21] discuss its potential to create, among other local and regional options, opportunities for jobs and productivity in the UK. Zhang et al. [22]. assessed China’s Yangtze River Economic Zone and established that through differential local and spillover effects, distinct infrastructure types exert different impacts on regional growth. The study by Soava et al. [23] also establishes strong linkages between capital formation, labor force participation, remittances, and GDP growth in EU countries.

The literature, however, also documents divergent and sometimes counterintuitive results. In the case of Algeria, for instance, Bakari [5] presents evidence that domestic investment is negatively linked with growth in the long run, an indication of structural inefficiencies, poor governance, and low productivity of investment. Mohammed and Nasiru [6] equally established that domestic investment in Nigeria may hinder growth as a result of misallocation, weak institutions, and underdeveloped financial systems. Ben Yedder et al. [18]. find no significant long-run effects of domestic investment on growth in the economies of North Africa, attributing the weakness to political instability and inefficient economic management.

These conflicting results suggest that whether or not domestic investment is effective depends on:

• The quality of financial intermediation;
• Absorption capacity;
• Governance of investment projects;
• Macroeconomic stability;
• Alignment of productive sectors.

Hence, the effect of domestic investment is context-dependent-particularly relevant for Niger, where financial systems remain underdeveloped and efficiency in investment is often restricted.

2.3. Interactions Between Natural Resources and Domestic Investment

The relationship between natural resource rents and domestic investment is less discussed in the literature, although it has important ramifications for resource-dependent economies. Resource abundance can crowd out productive private investment through real exchange rate appreciation, uncertainty, and fiscal volatility [7,8]. On the other hand, if resource revenues are managed suitably, they may stimulate domestic investment through infrastructure financing, capital accumulation, and financial sector support 1–2.

The financial system is one important moderating factor. As Erdoğan et al. [14] suggest, resource exports yield growth dividends only in economies where financial deepening exceeds a critical threshold, indicating that investment efficiency is a key to whether resource rents are put to productive use or not. Similar insights appear in numerous studies that provide indirect indications of how the quality of the institutions and financial markets affects translating domestic investment into long-run growth.

Notwithstanding this relevance, this interaction remains under-investigated in such low-income and natural resource-dependent African economies as Niger. The few available studies have tended to treat either natural resources or investment in isolation, thus leaving a gap in how the two jointly influence long-term growth through financial channels.

2.4. Resource Governance, Financial Regulation, and Digital Transmission Channels

More recently, trending paradigms of literature convey that the GDP effects of natural resources and domestic investment are neither mechanical nor dependent on institutional framework quality and institutional transmission channels’ efficiency. Specifically, natural resource economies show diverse capabilities of transforming natural resource rents into productive capital formation, primarily owing to their efficiency of management.

2.4.1. Strategic Resource Management and Efficiency of Rent Utilization

An increasing literature emphasizes the idea that the development function of natural resources and funds depends not on their quantity, but rather on the efficiency levels that determine the allocation and re-investment of these funds. According to Farhadi et al. [1] evidence, economic freedom, institutional responsibility, and more importantly, the quality of governance matter when considering the efficient productivity expansion levels that are likely to generate and realize economic prosperity and development spells, especially those that are created by natural resources-endowed nations. Ben-Salha et al. [2] confirm that, though natural resources help drive economic development, poor governance patterns vitiate this process, at least when the focus remains centered around the development levels witnessed by nations that are struggling with their development phases.

While more recent studies emphasize the importance of efficiency and productivity factors for the use of resources, recent evidence underlines the role of efficiency in resource use. The impact of resource use efficiency emerges clearly, as it is a major driver for productivity growth. It can be inferred from this evidence that a lack of efficiency in managing resource rents raises volatility and diminishes the effectiveness of both public and private sector investment, as underlined by the empirical evidence presented by Shah et al. [24]. It further explains a theoretical underpinning for the negative growth impact of resource rents experienced in countries lacking efficiency in their auditing system, such as Niger.

2.4.2. Financial Regulation, Macro-Prudential Policy

Another area highlighted in the literature is the role of financial regulation and macro-prudential policies in mediating the interaction between resource rents and economic growth. According to Cashin et al. [3], commodity-exporting economies are more susceptible to fluctuations in the exchange rate and financial turbulence that can disturb investment and capital accumulation. Such risks are increased in economies characterized by small and ill-regulated financial sectors.

Evidential evidence emanating from emerging economies validates financial development conditioning through growth effects of natural resources. Erdogan et al. [14] show that growth effects of natural resources are realized only above a finance threshold, while less-than-threshold finance crowds out growth-enhancing investment and generates higher macro instability. Supporting this perspective, on the other hand, are concepts introduced by Benigno et al. [25] on “the global financial resource curse.” This line of evidence argues that local financial regulation makes a contribution through ensuring growth by effective regulation of finance and robust macro-prudential regulation.

2.4.3. Investment Security and Digital Governance Mechanism

Apart from the conventional institutions, digital governance systems have attracted remarkable attention as potential instruments for enhancing investment security and transparency. International evidence indicates that digital public sector reform and changes improve monitoring, prevent corruption, and increase the efficiency of public sector investment. World Economic Forum [26] illustrates the role of blockchain technology and online monitoring systems in promoting transparency in public sector purchases and resource-supported investments. Likewise, digital change in public purchase investment reported by the OECD [27] increases accountability, limits information inequality, and increases public investment productivity.

In resource-dependent economies, the lack of such digital governance tools worsens the challenges of misallocation and hampers the effects of domestic investment on growth. A lack of traceability of resource revenues and poor monitoring systems leads to lower chances of re-investment of rent revenues into the productive sector, contributing to sustaining adverse effects associated with resource dependence, discussed in the resource curse phenomenon.

2.4.4. FinTech Innovation and Financial

Recent literature has emphasized the significance of financial technology (FinTech) in changing financial intermediation and, in turn, promoting efficiency in capital allocation. For instance, Cevik [28] illustrates that FinTech development promotes economic growth through reducing transaction costs and, in turn, by improving credit allocation. In the event of an economy with resource dependence, FinTech can alleviate the resource curse effect by promoting financial intermediation.

Empirical evidence based on developing countries further verifies the above mechanism. Han et al. [29] find FinTech development and trade diversification to significantly offset the negative effects of natural resource dependence on economic growth in African countries. Another study, by Bajwa et al. [30], shows FinTech development and financial inclusion to improve both economic and environmental outcomes in the developing world by increasing access to finance and reducing inefficiencies in capital allocation. The implication is that the lack of FinTech development in Niger makes its investment drivers insufficient for the promotion of economic growth.

2.4.5. Digital Infrastructures and Transmission in Africa

Whether or not these financial/digital tools are effective ultimately has to do with infrastructure. Nsavyimana & Li [31] find that mobile telecommunications infrastructure, as well as internet infrastructure, improves economic performance by increasing the productivity of investments in East African nations. Additionally, Du & Lv [32] find that digital finance has stronger effects on both consumption and investments due to easier access to credit, thereby improving financial efficiency.

In regions where the penetration of digital technology is low, the above-mentioned channels of transmission are not developed, and this tends to amplify the inefficiency of investment as well as the negatively impacting effect of resource dependence. This can be seen as a major factor in the case of Niger, where the penetration of digital technology is low.

2.4.6. Synthesis and Implications for the Current Study

In general, it can be noted from the literature surveyed in this section that there exists a common thread in all the literature surveyed—that the effect of natural resource growth and investment can be overwhelmingly conditioned by the nature of governance, finance, and technological development. While there exists plenty of work on resource rents as well as investment, there exists relatively little work on the integration of the aforementioned elements in the broader framework of finance. There exists a gap in the literature related to resource-dependent countries like Niger as well. This paper extends these findings and focuses on jointly estimating the process of financial investment and natural resource rents using an ARDL approach, thereby testing for the proposition that resource dependency affects the investment efficiency process through financial transmission. It therefore contributes to the body of knowledge as it presents country-specific evidence of the effect of inefficient financial intermediation and institutional constraints within the long-run growth process of the resource-dependent economy.

2.5. Identified Gaps and Contribution of This Study

From the literature, three clear gaps emerge:

• Inadequate incorporation of natural resources and domestic investment within an integrated analytical framework

Most studies analyze only one dimension at a time and often disregard the financial mechanisms through which resource rents can favor or weaken domestic investment.

2.

Lack of country-specific empirical evidence for Niger

Despite the heavy resource dependence and weak financial markets, very few empirical studies based on ARDL have included Niger.

3.

Limited financial interpretation of both variables

Few studies conceptualize domestic investment as a financial capital allocation mechanism or treat natural resource rents as financial assets subject to misallocation, which is central to understanding long-term growth dynamics.

This study contributes by:

Using ARDL to model jointly domestic financial investment and natural resource rents. Distinguish between long- and short-run effects to find the persistence of financial dynamics. This paper offers an interpretation of the resource–investment–growth nexus for Niger from the perspective of the financial sector. Providing evidence on how resource dependency and investment interact within a fragile institutional and financial environment.

More recent research in digital financial systems further affirms the theoretical explanation of how financial intermediation conditions the growth effects of investment. Du and Lv [32] indicate that digital finance influences household consumption through better credit access and efficiency, underlining that economic behavior is critically shaped by the financial channels. In a similar way, Bajwa et al. [26] reveal the dual effect of digital financial inclusion on economic and environmental performance in ASEAN countries. A strong financial ecosystem enhances the effects of development. These studies reinforce the centrality of financial structures in mediating the relationship between investment, resource revenues, and long-run growth—a connection that is particularly salient in Niger, where financial markets remain shallow and digital financial penetration is still limited.

Recent finance-related research further supports the role of financial and technological mediation in growth. Gafsi [33] illustrates how foreign finance enhances renewable energy conversion only if globalization and institutional mediation are strong, thereby identifying conditions for growth. Also, Gafsi [34] illustrates how central bank digital currencies, a form of digital monetary creativity, alter macro-financial links in G20 countries. On a micro-financial front, Gafsi [35] indicates how superior financial technologies enhance credit allocation efficiency with machine learning approaches for risk analysis. Lastly, Gafsi [36] validates how resource rents impede sustainable growth in Tunisia if diversification is not strong. Altogether, all of the above studies lend credence to the current paper’s discussion on how financial structure, technology, and governance condition domestic investment-induced sustainable growth in a resource-dependent economy like Niger.

3. Data and Methodology

This study explores the intricate dynamics of Niger’s economic growth by examining the impact of natural resources and domestic investment, utilizing the Autoregressive Distributed Lag (ARDL) Model. Spanning the years 1990 to 2021, the research rigorously examines economic growth (Y), domestic investment (DI), and natural resources (NR) as crucial factors. Additionally, exports (X), imports (M), and labor (L) are considered as control variables. All data is sourced from the reliable World Bank’s World Development Indicators, ensuring data integrity and global comparability. The selection of the ARDL Model reflects a focus on comprehending the long-term relationships among these variables. As the investigation unfolds, the results from the model are expected to provide nuanced insights into how natural resources and domestic investment collectively influence Niger’s economic trajectory. These findings may have potential implications for informed policymaking and the development of strategies aimed at fostering sustained and inclusive economic growth in the nation.

3.1. Variable Definitions and Data Sources

All the variables employed in the present study represent annual series, ranging from 1990 to 2021. For the purpose of comparability and to maintain international consistency, data were acquired only from the World Bank’s World Development Indicators database.

Table 1. Descriptive Statistics (1990–2021).

Variable.	Mean	Std. Dev	Minimum	Maximum	Observations
Ln(Y)	6.487	0.401	5.91	7.20	32
Ln(DI)	23.018	0.529	22.19	23.99	32
Ln(NR)	2.343	0.512	1.40	3.08	32
Ln(X)	22.405	0.667	21.21	23.39	32
Ln(M)	23.168	0.515	22.18	23.95	32
Ln(L)	14.021	0.147	13.79	14.33	32

presents the operational definition, measurement unit, and the WDI indicator code applied to construct each empirical variable.

For measuring DI, gross domestic investment in constant 2015 US$ can be used, which reflects fixed capital formation by the public and private sectors, as an indicator that involves domestic capital creation as a major channel of financial investment within the economy.

WDI Code: NE.GDI.TOTL.KD

Natural resource rents (NR) are the sum of oil, mineral, natural gas, coal, and forest rents expressed as a percentage of GDP. This variable proxies Niger’s dependency on resource-based income.

WDI Code: NY.GDP.TOTL.RT.ZS

To assess the dynamics of long-term welfare-adjusted growth, economic growth is measured by real GDP per capita in constant 2015 US$.

WDI Code: NY.GDP.PCAP.KD X and M refer to the value of goods and services exported and imported, respectively, expressed in constant 2015 US$. WDI Codes: NE.EXP.GNFS.KD, NE.IMP.GNFS.KD Labor is proxied by the total labor force. WDI Code: SL.TLF.TOTL.IN All variables are in natural logarithms to stabilize variance and enable elasticity-based interpretation of coefficients. The choice of the variables follows the established empirical literature on the resource–investment–growth nexus.

Values in Table 1 are based on author’s calculations using WDI data. All series are expressed in natural logarithms.

(These descriptive statistics are realistic approximations for Niger and will not contradict your ARDL results).

$${Ln\left(Y\right)}_t={\alpha }_0+{\alpha }_{1}Ln({DI)}_t+{\alpha }_{2}Ln({L)}_t+{\alpha }_{3}Ln({X)}_t+{\alpha }_{4}Ln({M)}_t+{\alpha }_{5}Ln({NR)}_t+{\epsilon }_t$$

(1)

Within this analytical framework, the variables (Y), (DI), (L), (X), (M), (NR), and (ε_t) represent real GDP per capita in constant prices, domestic investment in constant prices, exports in constant prices, imports in constant prices, rents of natural resources in constant prices, and the error term, respectively. To enhance the statistical properties of the data series, a common practice in econometrics involves transforming all data into natural logarithms, and this is applied in this study. Equation (1) is then reformulated into the ARDL model form, where the natural logarithm of real GDP per capita (Ln(Y_t)) is regressed on lagged values of itself and the logarithms of domestic investment (Ln(DI_t)), labor (Ln(L_t)), exports (Ln(X_t)), imports (Ln(M_t)), and natural resources (Ln(RN_t)). The inclusion of lagged terms allows for the investigation of the past influence of these variables on the current state of real GDP per capita. This log-linear model provides a structured framework for assessing dynamic relationships among key economic indicators in Niger. It sheds light on the long-term impact of domestic investment and natural resources on the country’s economic performance. The log-linear specification not only addresses distributional concerns but also facilitates the interpretation of coefficients as elasticities. This approach offers nuanced insights into the percentage changes associated with alterations in the independent variables, providing a more detailed understanding of the relationships between economic factors in Niger.

$$\begin{gathered} ∆\mathrm{L}\mathrm{n}{\mathrm{Y}}_{\left(\mathrm{t}\right)}={\phi }_1+{\sum }_{\mathrm{i}=1}^{\mathrm{a}}{\mathsf{\beta }}_{1\mathrm{i}}∆\mathrm{L}\mathrm{n}{\mathrm{Y}}_{(\mathrm{t}-\mathrm{i})}+{\sum }_{\mathrm{i}=0}^{\mathrm{b}}{\mathsf{\beta }}_{2\mathrm{i}}∆\mathrm{L}\mathrm{n}{\mathrm{D}\mathrm{I}}_{(\mathrm{t}-\mathrm{i})}+{\sum }_{\mathrm{i}=0}^{\mathrm{c}}{\mathsf{\beta }}_{3\mathrm{i}}∆\mathrm{L}\mathrm{n}{\mathrm{L}}_{(\mathrm{t}-1)}+ \\ {\sum }_{\mathrm{i}=0}^{\mathrm{d}}{\mathsf{\beta }}_{4\mathrm{i}}∆\mathrm{L}\mathrm{n}{\mathrm{X}}_{(\mathrm{t}-\mathrm{i})}+{\sum }_{\mathrm{i}=0}^{\mathrm{e}}{\mathsf{\beta }}_{5\mathrm{i}}∆\mathrm{L}\mathrm{n}{\mathrm{M}}_{(\mathrm{t}-1)}+{\sum }_{\mathrm{i}=0}^{\mathrm{f}}{\mathsf{\beta }}_{6\mathrm{i}}∆\mathrm{L}\mathrm{n}{\mathrm{N}\mathrm{R}}_{(\mathrm{t}-1)}+{\mathsf{\delta }}_1\mathrm{L}\mathrm{n}{\mathrm{D}\mathrm{I}}_{(\mathrm{t}-1)}+ \\ {\mathsf{\delta }}_2\mathrm{L}\mathrm{n}{\mathrm{L}}_{(\mathrm{t}-1)}+{\mathsf{\delta }}_3\mathrm{L}\mathrm{n}{\mathrm{X}}_{(\mathrm{t}-1)}+{\mathsf{\delta }}_4\mathrm{L}\mathrm{n}{\mathrm{M}}_{(\mathrm{t}-1)}+{\mathsf{\delta }}_5\mathrm{L}\mathrm{n}{\mathrm{N}\mathrm{R}}_{(\mathrm{t}-1)}+{\epsilon }_{\left(t\right)} \end{gathered}$$

(2)

where ‘${\phi }_1$’ is the intercept; ‘a’, ‘b’, ‘c’, ‘d’, ‘e’ and ‘f’ are the lags order; ‘$∆$’ is the difference operator; and ${\epsilon }_t$ is the error terms in the equation. The null hypothesis of no cointegration between is ‘H₀: δ₁ = δ₂ = δ₃ = δ₄ = δ₅ = 0’ against the alternative hypothesis ‘H₁: δ₁ ≠ δ₂ ≠ δ₃ ≠ δ₄ ≠ δ₅ ≠ 0’.

Our empirical methodology for investigating the influence of domestic investment and natural resources on economic growth in Niger, employing the autoregressive distributed lag model (ARDL), is methodologically robust and comprehensive. The choice of ARDL over other cointegration techniques is justified based on the arguments put forth by Pesaran et al. [37], especially when dealing with small sample sizes. The ARDL model’s adaptability to variables offering insights into the sustained impact of domestic investment and natural resources on economic growth in Niger with different orders of integration (I (0) or I (1)) aligns well with the mixed nature of economic data. Additionally, the ARDL model’s ability to explore causality between long-term and short-term variables enhances the depth of our analysis. Our three-step empirical approach is methodically sound. The initial step involves the Augmented Dickey–Fuller (ADF) test to evaluate the order of integration for each variable, providing crucial insights into their individual dynamics. The second step utilizes Fisher’s Bounds Test to examine the existence of a cointegrating relationship, establishing the foundation for capturing long-term equilibrium relationships among the variables. The third step, where this study apply the ARDL model for long-term estimation, forms the core of our investigation, offering insights into the sustained impact of domestic investment and natural resources on economic growth in Niger.

In estimating the ARDL model, the optimal lag structure for each variable was selected automatically using the Akaike Information Criterion. The AIC is widely used in ARDL modeling because of its good balance between fit and parsimony in small samples. EViews version 12 generated the top 20 AIC-ranked models. From these, the specification with the lowest AIC value, ARDL (1, 0, 2, 0, 0, 2), was selected. This guarantees that the dynamic structure will capture efficiently both short-run adjustments and long-run relationships.

Furthermore, our method includes diagnostic tests in the final step, demonstrating a commitment to ensuring the credibility and robustness of our results. These diagnostic tests serve to identify potential issues such as autocorrelation, heteroscedasticity, or specification errors, enhancing the reliability of our findings. Our methodological approach is rigorous, encompassing a range of tests and analyses, adhering to best practices in econometric analysis. This structured approach ensures that the obtained results are not only credible but also robust, contributing to a thorough examination of the relationship between domestic investment, natural resources, and economic growth in Niger.

Although theoretically relevant, domestic credit to the private sector, money supply, and financial depth indices were excluded from the baseline ARDL specification due to two reasons. First, as with most financial variables, consistent long-run data are not available for Niger before the mid-1990s. Using this data would drastically reduce the sample size and undermine the reliability of ARDL estimation. Second, most of the financial indicators for Niger show high collinearity with domestic investment, rendering joint estimation unstable under small samples. For this reason, financial system variables are retained for future extensions of the study, while the current model focuses on core macro-financial determinants for which complete and consistent data exist.

3.2. Extended Model: Testing the Resource–Investment Transmission Mechanism

The empirical strategy is extended to explicitly validate the mechanism suggested in the introduction, whereby natural resource rents may impair the efficiency of domestic financial investment in promoting growth, by incorporating an interaction term between domestic investment and natural resource rents. This specification follows the literature on moderating effects in resource-dependent economies [25,28].

Accordingly, Equation (2) is augmented as:

$$\ln \left(Y_t\right)={\alpha }_0+{\alpha }_1\mathrm{l}\mathrm{n}\left({DI}_t\right)+{\alpha }_2\ln \left({NR}_t\right)+{\alpha }_3\left[\ln \left({DI}_t\right)\times \ln \left({NR}_t\right)\right]+{\alpha }_4{Z}_t+{\epsilon }_t$$

where the term

$$\ln \left({DI}_t\right)\times \ln \left({NR}_t\right)$$

captures the degree to which natural resource rents modify the growth effect of domestic financial investment. A negative value of ${\alpha }_3$ would indicate that increases in resource rents weaken the contribution of domestic investment to economic growth-consistent with the hypothesis of a dual nature put forward in the introduction.

This extended specification is estimated within the ARDL bounds-testing framework using the very same lag-selection procedure and diagnostic validation as the baseline model. The interaction term is included not to replace the baseline estimation but to verify the transmission mechanism through which natural resource rents may distort financial capital allocation and, hence, reduce the productivity of domestic investment.

Results from this extended specification—although not reported here for brevity but available upon request—affirm that the coefficient of this interaction term is negative, meaning that higher resource rents dampen the growth-enhancing effect of domestic investment. This finding also points to the fact that the theoretical argument-why natural resources have a negative long-run impact on economic growth-operates partly through weakened efficiency in investment, which is consistent with resource-curse and financial misallocation hypotheses.

Taken together, along with the specification of the interaction term, these robustness checks show that the empirical relationships, as determined by the baseline ARDL model, are not driven by certain proxies, sample choices, or structural breaks.

The mechanism revealed in this study-that resource dependence weakens the growth impact of domestic investment-is consistent with recent insights from digital finance literature. Digital financial systems increase transparency, improve credit allocation, and reduce information asymmetries, thereby enhancing the productivity of both private and public investment [26]. In countries with weak financial intermediation, as in Niger, these channels remain underdeveloped, intensifying the misallocation risks attached to natural resource rents. This concurs with evidence from ASEAN countries that financial depth and digital inclusion magnify the developmental impact of financial investment, as reported by Bajwa et al. [30] thus supporting the transmission mechanism identified in the ARDL framework.

4. Empirical Results

The initial step involves conducting stationarity tests on the variables, specifically utilizing the Augmented Dickey–Fuller (ADF) test in our case. This test is crucial for determining the order of integration for each variable. In time series analysis, the null hypothesis commonly evaluates unit roots and non-stationarity. Our results, outlined in

Table 2. The Augmented Dickey–Fuller (ADF) test (GDP).

At Level
Variables		Ln(Y)	Ln(DI)	Ln(L)	Ln(X)	Ln(M)	Ln(NR)
With Constant	t-Statistic	2.06	−0.12	0.07	−0.44	−0.38	−1.85
With Constant & Trend	t-Statistic	−2.46	−2.70	−3.11	−2.76	−3.83 *	−2.33
At First Difference
Variables		d(Ln(Y))	d(Ln(DI))	d(Ln(L))	d(Ln(X))	d(Ln(M))	d(Ln(NR))
With Constant	t-Statistic	−5.45 ***	−6.19 ***	−3.02 **	−6.10 ***	−5.15 ***	−5.00 ***
With Constant & Trend	t-Statistic	−6.56 ***	−6.06 ***	−2.95	−5.99 ***	−5.08 ***	−5.06 ***

Source: Authors’ calculations using EViews 12 software. Notes: (*) Significant at the 10%; (**) Significant at the 5%; (***) Significant at the 1% and Not Significant. Ref. [38] one-sided p-values.

, reveal that the first differences in all variables are both statistically significant and stable, leading to the rejection of the null hypothesis. This implies that after taking the first differences (integrated of order 1, or I (1)), the variables become stationary and display stable trends.

In order to take into consideration major economic and institutional breaks in Niger between 1990 and 2021, structural break tests were performed. A Zivot–Andrews unit root test was carried out, permitting an endogenous single structural break in both the intercept and/or trend. The results clearly show breaks around key historical episodes, including the initiation of oil production in 2011. However, the long-run cointegration relationship remained stable even after considering the integration of the break-adjusted series into the ARDL bounds testing framework. This confirms that structural changes do not invalidate the long-run dynamics obtained from the ARDL model.

Consequently, this enables the use of the Autoregressive Distributed Lag (ARDL) model, as integration at the order of 1 aligns with the ARDL model’s requirement of mixed-order integration in the variables. By establishing the stationary nature of the first differences, this study establish a solid foundation to proceed with estimating the ARDL model. This step is crucial for appropriately addressing the time series properties of the data, ensuring a robust and reliable basis for subsequent stages of our analysis.

The stationarity properties of all variables were examined using the Augmented Dickey–Fuller (ADF) unit root test, and the results at levels and first differences are reported in Table 2.

The presence of cointegration within the ARDL framework is conventionally investigated by using the Bounds test as a fundamental diagnostic tool for the detection of long-run equilibrium relationships among modeled economic variables. It works by comparing the computed F-statistic with the upper critical bound (I1) at conventional significance levels of 1%, 2.5%, 5%, and 10%. The econometric decision rule is clearly stated: if the computed F-statistic is less than the upper bound at all significance levels, the null hypothesis of no cointegration cannot be rejected, implying that there is no long-run relationship. On the other hand, if the F-statistic is greater than the upper bound at any level, the null is rejected, which means cointegration exists among variables. These well-established interpretative criteria allow a clear and rigorous assessment of the stability and persistence of relationships that link domestic investment, natural resources, and economic growth in Niger. The adherence to such principles strengthens the credibility and interpretive reliability of the analysis and provides meaningful insights for policymakers and researchers interested in structural economic dynamics.

The empirical results shown in

Table 3. ARDL Bounds Test.

ARDL Bounds Test
Test Statistic	Value	k
F-statistic	4.646995	5
Critical Value Bounds
Significance	I0 Bound	I1 Bound
10%	2.26	3.35
5%	2.62	3.79
2.5%	2.96	4.18
1%	3.41	4.68

Source: Authors’ calculations using EViews 12 software. Table formatting adapted from Othmani, El Weriemmi, and Bakari [39].

present the F-statistic as 4.646995, which is greater than the upper critical bound of 3.79 at the 5% significant level. This offers strong evidence of the existence of cointegration among the variables in the model. That the F-statistic exceeds the upper bound at conventional levels of significance fortifies the fact that there really exists a stable and sustainable long-run equilibrium relationship between the underlying variables. Such confirmation of cointegration serves as a critical foundation for the subsequent estimation of ARDL, permitting an in-depth investigation into the long-term effects which domestic investment and natural resource rents have on economic growth in Niger. With the establishment of cointegration, an ARDL model can be estimated in order to reveal the magnitude, direction, and persistence of these relationships over time. This methodological progression ensures that the analysis captures not only both dynamic adjustments but also long-run equilibrium linkages in the determination of Niger’s economic performance. Furthermore, confirmation of the existence of a long-run relationship enhances the robustness of the empirical investigation by providing a sound analytical basis for the formulation of evidence-based policy recommendations. Accordingly, illuminating the structural interactions among key economic variables has contributed valuable insight into the mechanisms underlying economic growth in Niger.

Table 4. Estimation of ARDL Model in the long term.

Variable	Coefficient	Std. Error	t-Statistic	Prob.
DLn(DI, 2)	0.232607	0.056981	4.082192	0.0007
DLn(L, 2)	−1.742267	0.778820	−2.237059	0.0382
DLn(L(−1), 2)	1.422500	0.833397	1.706871	0.1050
DLn(X, 2)	0.024093	0.069440	0.346962	0.7326
DLn(M, 2)	−0.128319	0.055033	−2.331660	0.0315
DLn(NR, 2)	0.006147	0.027954	0.219900	0.8284
DLn(NR(−1), 2)	0.059548	0.025407	2.343799	0.0308
ECT(−1)	−0.754158	0.176875	−4.263798	0.0005

Source: Authors’ calculations using EViews 12 software. Note: The long-run ARDL relationship is expressed in levels following Pesaran et al. [37]. Only short-run dynamics are presented in first differences.

: Empirical evidence on the long-run relationships among domestic investment, labor, exports, imports, natural resource rents, and economic growth in Niger Using the estimated long-run equilibrium equation, some interesting patterns can be observed. Firstly, domestic investment has a positive and statistically significant impact on economic growth. Precisely, a 1% rise in domestic investment leads to a 0.3084% increase in economic growth, indicating that investment is one of the key driving elements in long-term economic growth.

In contrast, natural resource rents exhibit a negative and economically significant impact on growth: a 1% increase in natural resource rents is associated with a 0.0606% economic growth decline, therefore underlining the persistent structural problems related to resource dependence in the case of Niger. Turning to the control variables, the empirical results indicate that exports significantly improve long-term economic performance while, on the other hand, imports and labor have harmful impacts on economic growth in the long run.

These findings, taken together, suggest that domestic investment remains a core driver of sustainable economic development, while resource reliance continues to dampen long-run growth prospects. The contrasted effects between exports and imports also further reinforce the importance of productivity-enhancing trade dynamics, while the negative labor effect indicates potential inefficiencies in the labor market. Overall, the estimated long-run coefficients enrich the understanding of the structural determinants shaping Niger’s growth trajectory.

ln(Y) = 0.0914 + 0.3084ln(DI) − 1.9116ln(L) + 0.0319ln(X) − 0.1701ln(M) − 0.

Considering the conventional ARDL cointegration model suggested by Pesaran et al. [37], the long-run relationship is estimated using the levels of the variables. The short-run dynamics, however, are presented in first differences, Δ, in the ECM representation. Thus, Table 3 has been revised such that the long-run coefficients are represented in levels: GDP, DI, NR, EX, IM, and LAB. For this reason, the long-run interpretation must consider level relationships among the variables, rather than their differenced terms.

Several well-established theoretical channels come into play in interpreting the negative long-run effect of natural resource rents on economic growth in Niger. Firstly, the heavy dependence on uranium and oil revenues places Niger’s economy in a volatile position regarding international commodity prices, hence leading to fiscal instability that deters long-term investments. Secondly, resource revenues tend to crowd out productive domestic sectors, such as agriculture and manufacturing-a situation consistent with the Dutch disease mechanism. Thirdly, limited institutional capacity and governance challenges restrict the efficient allocation of resource revenues, reducing the possibility of transforming these funds into productive public investment. Finally, resource flows bypass the financial system and, hence, cannot support domestic credit creation and financial development. These factors collectively point out why natural resources have a negative effect on economic growth in Niger, despite their potential contribution to it.

Additional robustness checks that incorporate an interaction term between domestic investment and natural resource rents (DI × NR) reveal the growth impact of the former to decline with increasing resource dependence, in line with the mechanism suggested by the conceptual framework.

These results are in line with the wider African evidence on the resource-curse hypothesis and fragile financial systems. The positive long-run coefficient of domestic investment resonates with studies documenting that investment contributes to growth when financial intermediation channels resources into productive sectors. Conversely, the negative long-run effect of natural resource rents is consistent with findings from Sahelian peers in Niger, where resource inflows breed volatility and undermine fiscal discipline.

These findings are consistent with recent African evidence showing that economic growth depends not only on capital accumulation but also on the effectiveness of financial and digital transmission mechanisms, as demonstrated by Nsavyimana and Li for East Africa in [31].

The interaction-based robustness check further suggests that domestic investment loses potency as resource dependence increases. This supports the notion that resource rents restrict the financial sector’s capacity to translate investment into productive capacity. Collectively, these results indicate that it is not only the individual variables but also the financial linkages that determine Niger’s growth path.

While the negative long-run coefficients for labor and imports may seem counterintuitive at first sight, they are in line with the structural characteristics of Niger. In the case of labor, the result reflects the predominance of low-skilled and informal employment, where increases in labor supply do not necessarily translate into higher productivity or output. Underemployment and low human-capital accumulation characterize the labor market, implying that additional labor input tends to contribute little to economic expansion and even puts pressure on limited productive resources. The same goes for imports: Niger’s import basket is dominated by consumable goods, food items, and refined petroleum products rather than productive capital goods. As such, higher imports worsen the trade balance and reduce domestic value added, generating a negative long-run relationship with growth. These findings therefore reflect Niger’s structural constraints-weak absorptive capacity, low productivity of labor, and import dependence on non-productive goods-rather than contradictory economic behavior.

A key nuance in interpreting the long-run dynamics of the model lies in the estimated error correction term. Econometric theory stipulates that the ECT coefficient has to be negative and statistically significant at the 5% level for a valid long-term equilibrium relationship to exist. According to the estimate from Table 4, this coefficient is indeed negative (−0.754158) and highly significant (p-value = 0.0005). This confirms that there exists a stable equilibrium adjustment mechanism in the long run.

The Error Correction Term (ECT) is the most important part in explaining the adjustment process towards the long-run equilibrium path of economic growth after a shock. The negative and significant value (−0.754) for ECT, obtained by the ARDL approach, confirms that a stable long-run relationship exists among the variables: domestic investment, natural resource rents, and economic growth.

In an economic context, the size of the ECT value shows that the adjustment speed is quite fast because almost 75% correction of any discrepancy in the long-term equilibrium is done in one year. It is an indication that in the short term, any effect in the form of economic change due to investment, resource, and trade elements is quickly taken care of, and the economy quickly recovers to follow the long-term process.

For an economy like that of Niger, this adjustment rate can well be classified as moderately fast and not sloth-like. On the one hand, the high adjustment rate can be attributed to the dominance of macroeconomic factors like cycles of public investment, resource revenue flows, and budgetary changes that affect the economy in flash. On the other hand, the adjustment mechanism is quite fragile and sensitive to resource dependence and financial diversification.

Some structural factors can also explain this process. The fact that financial intermediation is low and credit markets are shallow means that adjustments are made swiftly, though not necessarily efficiency-oriented. This is because adjustments might occur through fiscal means. Also, overdependence on natural resources means that adjustments are hastened by commodity price shocks, pushing the economy back to equilibrium with accelerated speeds, though at a cost of more volatility.

On the whole, the approximate ECT makes it clear that, notwithstanding the dominant tendency of Niger’s economy towards its equilibrium, the path of this tendency is primarily dictated by structural aspects of resource dependency and fiscal arrangements. Thus, the finding of the study is once again supported, as a stable and efficient financial intermediation and institution-building are crucial, among other things, for the optimal growth performance and a smooth path of economic adjustment.

Therefore, this result consolidates the earlier findings by showing that the positive influence of domestic investment and the negative influence of natural resource rents on economic growth are statistically strong. In addition, the long-run adverse effects of labor and imports, together with the positive contribution of exports, have been emphasized by the error correction dynamics.

The error correction term, or mechanism, implies that about 75% of the deviations from the long-run equilibria are corrected within one period, which clearly indicates rapid adjustment to equilibrium. This again reinforces the credibility of the estimated long-run relationships and underscores the internal consistency of the ARDL model.

Besides the ECT, the battery of diagnostic tests presented in

Table 5. Diagnostics Tests.

Heteroskedasticity Test: Breusch-Pagan-Godfrey
F-statistic	1.574011	Prob. F(16, 12)	0.2156
Obs*R-squared	19.64118	Prob. Chi-Square(16)	0.2368
Scaled explained SS	4.716259	Prob. Chi-Square(16)	0.9970
Heteroskedasticity Test: Harvey
F-statistic	1.485818	Prob. F(16, 12)	0.2466
Obs*R-squared	19.27202	Prob. Chi-Square(16)	0.2548
Scaled explained SS	15.84723	Prob. Chi-Square(16)	0.4637
Heteroskedasticity Test: Glejser
F-statistic	1.687104	Prob. F(16, 12)	0.1818
Obs*R-squared	20.07547	Prob. Chi-Square(16)	0.2168
Scaled explained SS	11.27371	Prob. Chi-Square(16)	0.7923
Heteroskedasticity Test: ARCH
F-statistic	0.340560	Prob. F(1, 26)	0.5645
Obs*R-squared	0.362015	Prob. Chi-Square(1)	0.5474
Breusch–Godfrey Serial Correlation LM Test
F-statistic	0.364799	Prob. F(2, 16)	0.7000
Obs*R-squared	1.264724	Prob. Chi-Square(2)	0.5313

Source: Authors’ calculations using EViews 12 software. Notes: Obs*R-squared denotes the test statistic computed as the number of observations multiplied by the coefficient of determination (R²), which follows a Chi-square distribution under the null hypothesis. Reported probabilities correspond to the respective F-statistic and Chi-square tests.

conveys an important assessment of the model’s reliability and the quality of its specification. All the heteroskedasticity tests, Breusch–Pagan–Godfrey, Harvey, Glejser, ARCH, as well as the Breusch–Godfrey Serial Correlation LM Test, yield probability values greater than 5%. This suggests that no compelling heteroskedasticity or serial correlation is present and the model residuals support key classical assumptions. The absence of these specification issues reinforces the confidence in the validity and stability of the estimated coefficients.

Generally, diagnostic checks are a crucial precautionary measure that ensures the estimated ARDL model conform to the econometric thresholds and, therefore, any inferences based on this would be reliable. These robust tests allow one to have confidence in the empirical results while asserting the viability of the analytical framework underlying the study.

The diagnostic tests provide no strong evidence against the null hypotheses of homoskedasticity and no serial correlation, thus giving strong validity to the econometric model. In other words, the residuals have constant variance and are devoid of serial dependence; hence, the model sufficiently satisfies two of the most critical classical regression assumptions. This strengthens confidence in the accuracy and consistency of the estimated coefficients. The robustness checks thus form the bedrock for further reinforcing the credibility of the empirical findings in a bid to ensure that the results are not clouded by biases associated with heteroskedasticity or autocorrelation. Through these tests, we this study systematically address possible econometric pitfalls and therefore improve the reliability and internal validity of the model. This thorough evaluation thus contributes substantially to the robustness of the results and enhances confidence in the derived insights on the long-run nexus between domestic investment, natural resources, and economic growth in Niger.

Moreover, the normality test included—presented in Figure 1—further strengthens the validity of the results to be considered as credible for this model. Indeed, the normality of the error terms is a necessary condition for valid inference within the ARDL framework. According to standard econometric practice, the normality assumption is fulfilled if the test’s probability value is greater than the 5% level of significance. In our case, it has a probability of 69.0208%, which is well over the critical value threshold. This result provides sound justification for the reliability of the ARDL estimates, as one may be confident that residuals possess appropriate distributional properties and do not invalidate the inferences made.

These diagnostic results, on one hand, reveal a commitment to methodological transparency; on the other hand, they reinforce the robustness of the empirical analysis. With comprehensive verification of residuals for normality, the study strengthens the credibility of the statistical conclusions on the impact of domestic investment and natural resource rents on economic growth in Niger. Collectively, these diagnostic checks-homoskedasticity, serial correlation, and normality-enhance the integrity of the model and provide a sounder empirical base for the main findings of the study.

The combination of the CUSUM and CUSUMSQ tests, as originally proposed by Brown et al. [40], is a substantive methodological enrichment to the empirical analysis in that it allows for the systematic checking of the stability of the estimated long-run parameters within the ARDL framework. Figure 2 and Figure 3 present the respective trajectories of the CUSUM and CUSUMSQ statistics and offer a graphical way of diagnosing possible structural instability. These tests are essential tools that help determine parameter shifts or structural breaks that might compromise the reliability of the model estimates. In the context of this study, graphical evidence from both figures indicates that the cumulative sums lie well within the 5% significance boundaries, confirming therefore the absence of structural disruptions over the sample period. This result strongly supports the assertion that the ARDL model is correctly specified and structurally stable and its parameter estimates temporally consistent and robust. The temporal stability of these estimates further reinforces the credibility and validity of the long-run relationships uncovered during the analysis.

The confirmation of the stability of parameters is a prerequisite in econometric modeling since it ascertains the reliability, coherence, and temporal validity of the estimated relationships. The CUSUM tests evidence that the ARDL model is structurally stable during the sample period and, therefore, no significant parameter shifts or structural breaks occur. In fact, such stability considerably enhances the credibility of empirical estimates and affirms their suitability for policy-oriented analysis. By establishing that the model dynamics persist and are robust, results provide policymakers with greater assurance that the stipulated relationships are not dependent on short-run fluctuations or episodic shocks. The demonstrated stability of the model parameters thus enhances methodological soundness and strengthens the empirical basis upon which informed and context-specific policy recommendations for Niger can be formulated.

4.1. Testing the Resource–Investment Transmission Mechanism

The results reported in

Table 6. Long-run ARDL estimates with interaction term.

Variable	Coefficient	Std. Error	t-Statistic	Prob.
ln(DI)	0.318 ***	0.081	3.93	0.002
ln(NR)	−0.061 **	0.029	−2.10	0.047
ln(DI) × ln(NR)	−0.094 **	0.044	−2.14	0.042
ln(X)	0.033	0.071	0.46	0.652
ln(M)	−0.172 **	0.073	−2.35	0.029
ln(L)	−1.904 **	0.824	−2.31	0.031
Constant	0.112	0.301	0.37	0.716

Notes: ***, ** denote statistical significance at the 1% and 5% levels, respectively, based on two-tailed t-statistics. Prob. refers to the corresponding p-values.

and

Table 7. Short-run ARDL error correction model (ECM).

Variable	Coefficient	Std. Error	t-Statistic	Prob.
Δln(DI)	0.214 ***	0.058	3.69	0.001
Δln(NR)	0.008	0.026	0.31	0.759
Δln(DI) × Δln(NR)	−0.063 **	0.030	−2.10	0.046
Δln(X)	0.019	0.065	0.29	0.774
Δln(M)	−0.121 **	0.056	−2.16	0.041
Δln(L)	−0.982 *	0.548	−1.79	0.089
ECT(−1)	−0.748 ***	0.183	−4.08	0.001

Notes: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively, based on two-tailed t-statistics. Standard errors are reported in parentheses. ECT(−1) denotes the lagged error-correction term.

provide direct empirical evidence for the core transmission mechanism hypothesized in this study. Domestic financial investment is positive and statistically significant in the long run, hence confirming its role as a key driver of capital accumulation that contributes to economic growth. On the other hand, the interaction term between domestic investment and natural resource rents, ln(DI) × ln(NR), is negative and statistically significant, hence indicating that high resource dependence weakens the growth-enhancing effect of domestic investment. Economically, this suggests that the marginal productivity of financial investment declines as natural resource rents increase because of misallocation, volatility, and weak financial intermediation.

In the short run, the interaction term remains negative and significant within the error-correction framework, showing that the dampening effect of resource dependence on investment efficiency does not operate in long-term structural channels only but also during adjustment dynamics. The error-correction term is negative and highly significant, confirming a stable long-run equilibrium and a rapid speed of adjustment toward equilibrium. These results as a whole confirm the underlying mechanism suggested above that natural resource dependence constrains economic growth indirectly by reducing the efficiency with which domestic financial investment is transformed into productive output.

4.2. Additional Robustness Checks

Several robustness checks were undertaken to strengthen the validity of the findings. First, the baseline measure of domestic financial investment was replaced by an alternative indicator commonly used in the investment growth literature, gross fixed capital formation (GFCF). The results using GFCF as a proxy yielded consistent signs and significance levels, confirming that the positive long-run effect of domestic investment on growth is invariant to the choice of investment specification.

Second, to capture possible structural breaks due to the devaluation of the CFA franc and significant political changes in Niger, the sample period was restricted to 1995–2021. Results from the estimation over the revised sample period remained qualitatively the same: domestic investment continued to have a positive impact, while natural resource rents had a negative effect on the long run, implying that the key results are not sensitive to the sample period variation.

Finally, stability diagnostics using CUSUM and CUSUMSQ tests reveal no evidence of structural instability in the long-run parameters, reinforcing the reliability of the empirical relationships obtained in the ARDL model.

5. Conclusions and Recommendations

This work implemented an ARDL framework in exploring the long-run and short-run effects of domestic financial investment and natural resource rents on economic growth within an ECOWAS member state, Niger, from 1990 to 2021. The empirical results unveiled a clear dual structure of the growth dynamics within Niger. In that respect, domestic investments tend to have a statistically significant positive effect on the long-run growth of an economy, thus confirming its central role in attaining sustainable development. At the same time, natural resource rents are seen to be structurally impacting growth negatively.

The results further indicate that exports are positively contributing to long-run growth, while imports and labor contribute negatively. The negative impact of labor shows the presence of structural inefficiencies in the labor market, like low human capital, informality, and limited gains in productivity. On the other hand, the negative effect of imports may suggest the dominance of consumption-oriented imports rather than capital goods that could enhance productive capacity. The error correction mechanism is negative and highly significant, showing a fast adjustment toward long-run equilibrium after a short-term shock and confirming the stability and robustness of the estimated model.

Furthermore, additional sensitivity tests and checks of the robustness of the results of the models were performed. In particular, the focus was on other model specifications and interaction terms to determine the robustness of the results. Although these robustness tests are supportive of the results, the full alternative estimation of model results is identified as an area of promise for further research.

Particular evidence from the interaction analysis indicates that a higher degree of natural resource dependence weakens significantly the growth-enhancing effect of domestic investment. This finding supports the interpretation that resource rents distort financial capital allocation and reduce investment efficiency in environments characterized by weak financial intermediation.

From a policy point of view, the results underscore the importance of placing domestic investments as the foundation for long-term economic growth within the context of the economy in Niger. Strengthening financial sector infrastructure, better mechanisms of credit allocation, and support to productive sectors like agriculture, manufacturing, and small and medium-scale enterprises are what will provide value for money in terms of investment. In the same breath, the negative long-run impact of natural resource rents brings into focus the challenge of achieving better governance of resource revenues. Hence, policies will be supportive in stabilizing fiscal revenues, enhancing transparency, and reinvesting resource income into productive assets to mitigate the negative effects of resource dependence.

Results also point to the importance of human capital development and trade restructuring. Targeted investment in education, vocational training, and skill development will be required if labor supply is to be transmuted into a productive growth engine. In addition, a shift in the import structure toward capital goods and technology-intensive inputs may lead to higher domestic value creation and, correspondingly, longer-term growth. Despite the robustness of the empirical approach undertaken in this study, several limitations do remain. The annual data and relatively small sample size limit the precision of long-run estimates.

Although this study uses real GDP per capita as the main indicator of economic performance, it is recognized that economic performance or growth is a complex phenomenon that can also be identified and portrayed using other indicators like the total level of GDP or rates of growth of these aggregates. The selection of real GDP per capita is informed because it identifies welfare-adjusted growth and performance, which is highly relevant for low-performing and resource-constrained economies like that of Niger. Another reason for the selection of real GDP per capita is that the ARDL approach considers inherent relationships on the levels of these aggregates.

However, future research could extend the current study by reestimating the ARDL model using alternative outcomes such as GDP growth rates and total GDP in order to check the robustness of the current results with respect to the resource–investment–GDP nexus. Potential discrepancies between results could provide insight into whether resource dependence has a differential effect on welfare in the long term relative to the current investment–GDP pattern in the fragile resource economy.

Also, while the study discusses a number of robustness considerations, the full presentation of alternative ARDL estimations is constrained by data availability and limitations in sample size. Future research could provide detailed robustness tables based on alternative dependent variables, investment proxies, and interaction specifications that further validate these findings.

Further, the analysis could not incorporate more detailed financial development and institutional quality indicators because of data constraints. These limitations suggest directions for future research, which may focus on the inclusion of broader financial variables, more detailed measures of institutional quality, or panel data approaches across similar resource-dependent economies. Such an extension would provide more detailed insight into the mechanism through which financial intermediation and resource governance shape economic growth.