Financial Leverage and Firm Performance in Moroccan Agricultural SMEs: Evidence of Nonlinear Dynamics

Abstract

This study investigates the nexus between leverage and financial performance in a sample of 54 Moroccan agricultural small- and medium-sized enterprises (SMEs) over the period of 2017–2022. Drawing on trade-off, pecking order, and agency theories, this analysis examines whether different levels of indebtedness influence performance, as measured by return on assets (ROA). Using panel data regression models, both linear and nonlinear specifications were tested to explore the potential curvature of the leverage–performance relationship. The empirical results reveal a significant and negative linear relationship between both short-term and long-term leverage and ROA, suggesting that increased indebtedness impairs financial performance. A quadratic specification reveals a persistently negative effect of short-term leverage and a U-shaped relationship between long-term leverage and ROA, indicating that performance may improve beyond certain debt thresholds. To address endogeneity concerns and validate the findings, dynamic panel estimation using the generalized method of moments (GMM) was employed, confirming the leverage’s adverse effects on performance. Thus, this study provides policy-relevant insights into optimal capital structure decisions for small agribusinesses and underscores the need for tailored financial strategies to support their sustainable development.

1. Introduction

Capital structure has been a cornerstone of corporate finance theory since the pioneering work of Modigliani and Miller (1958, 1963), who laid the foundation for understanding the optimal balance between equity and debt financing. Debt can enhance financial performance through leverage when its cost is lower than the economic return on assets, but excessive leverage may increase financial costs, strain liquidity, and raise default risk, particularly in volatile markets (Jensen & Meckling, 1976; Myers, 1984). This debate has given rise to several theoretical approaches—including trade-off, pecking order, and agency theories—that seek to explain how firms weigh the advantages of debt financing against its potential drawbacks.

The complexity is greater for SMEs because their structural characteristics tend to intensify both the beneficial and adverse consequences of leverage. Empirical findings on the leverage–performance nexus are mixed, with some studies reporting nonlinear relationships—often inverted U-shaped or U-shaped—depending on the leverage levels (Le & Phan, 2017; Molinari, 2013; Orabi Awad & Mohamed Ali, 2022; Zeitun & Goaied, 2022). At low debt levels, fixed interest costs and underinvestment risk may depress profitability; beyond a threshold, higher leverage may reflect stronger financial discipline or better project selection, thereby enhancing performance (Chen & Hammes, 2004). These patterns justify modeling the relationship quadratically to capture potential turning points and test the competing predictions of the trade-off and pecking order theories.

In developing countries, SMEs are often undercapitalized and face barriers to external financing due to limited collateral, weak disclosure, and institutional constraints (T. H. L. Beck et al., 2005). Such constraints hinder optimal capital structure decisions, making profitability more sensitive to changes in leverage (Abor & Biekpe, 2009). Consequently, the trade-off between the need for external funding to support growth and the financial risks associated with debt is particularly acute for these firms. Investigating the leverage–performance relationship in this segment can shed light on the specific financial behaviors of SMEs and inform public policies aimed at enhancing access to finance for small productive enterprises.

This issue remains especially salient in Morocco’s agricultural sector, which is a strategic pillar of the national economy. According to the most recent data from the High Commission for Planning, agriculture contributes approximately 12% to GDP and accounts for about 20% of total exports, while providing employment to nearly 40% of the working population, predominantly in rural areas (HCP, 2024, 2025). Within this landscape, agricultural SMEs and structured family farms continue to play vital roles in value creation, supply chain modernization, and ensuring food security. However, these enterprises face persistent financial constraints driven by income seasonality, exposure to climate risks, volatility in agricultural prices, and limited access to credit (African Development Bank Group, 2019; Louali, 2019).

Considering these challenges, the Moroccan government has made the modernization of the agricultural sector, particularly through support to SMEs and small-scale farms, a central priority of its “Green Generation” strategy. This policy draws upon the progress already achieved under the Green Morocco Plan while placing greater emphasis on sustainability, agricultural entrepreneurship, value chain development, and support for youth and high-potential small producers (Département de l’Agriculture, 2020; MAPMDREF, 2020). It also promotes the integration of modern technologies, environmentally sustainable methods, and the expansion of agri-food processing to strengthen competitiveness and resilience in the sector.

Financing is a key pillar in achieving these goals. To this end, public authorities have mobilized substantial resources through the general budget, dedicated funds such as the Agricultural Development Fund, and targeted subsidies. Simultaneously, the private banking sector has engaged in this momentum by developing credit lines tailored to agriculture, introducing financial products adapted to the sector’s inherent risks, and forming partnerships with public institutions, such as Crédit Agricole du Maroc.

Despite these efforts, the issue of optimal debt levels and their impact on the financial outcomes of agricultural SMEs remains empirically underexplored in the Moroccan context. However, a nuanced understanding of this relationship is crucial for informing agricultural entrepreneurs’ financing decisions, guiding lending strategies, and fine-tuning public support mechanisms in the agricultural sector. This study aims to address this gap by rigorously examining whether the relationship between financial leverage and performance is linear or nonlinear in a strategically important sector characterized by both structural vulnerabilities and significant opportunities for transformation.

The regression analysis reveals a consistent negative link between leverage and the performance of Moroccan agricultural SMEs, using return on assets (ROA) as the indicator. The results from both fixed effects and GMM estimators show that short-term (STLEV) and long-term (LTLEV) debt have a statistically significant detrimental impact on operating profitability. However, the analysis also reveals evidence of a non-linear relationship between long-term leverage and ROA. Specifically, while the marginal impact of LTLEV is initially negative, it becomes positive beyond a certain threshold, suggesting a U-shaped dynamic. This implies that high levels of long-term debt may eventually enhance profitability, potentially by supporting productive investments. In contrast, short-term debt consistently impairs performance at all leverage levels.

While numerous international studies have examined the leverage–performance nexus, in-depth empirical analyses remain scarce in Morocco, particularly within the agricultural sector. Moreover, recent findings in the literature are often heterogeneous and sometimes contradictory, suggesting that the leverage–performance relationship is strongly influenced by context-specific conditions (Abor, 2007; Salim & Yadav, 2012).

This study contributes to both theory and practice by providing novel insights into this relationship within the agricultural sector of a transitioning economy. It also offers a testing ground for the applicability of canonical corporate finance theories under conditions of data scarcity, institutional constraints, and economic vulnerability.

From a policy perspective, our findings can inform agricultural financing strategies at both the governmental and institutional levels. Specifically, the U-shaped nature of the leverage–performance relationship underscores the importance of not only improving credit access for SMEs but also calibrating support mechanisms to firms’ optimal debt threshold. Public agencies, such as the Agricultural Development Fund, and banks, such as Crédit Agricole du Maroc, can use these insights to better tailor credit-scoring models, subsidy schemes, and financial literacy programs. Simultaneously, SME managers and agricultural entrepreneurs can benefit from these results when structuring their capital to enhance profitability while managing risks.

A further contribution lies in the study’s exclusive focus on Moroccan agricultural SMEs—a segment that faces chronic liquidity constraints, limited access to formal finance, and pronounced information asymmetries. While much of the existing research focuses on large or listed firms, empirical studies using firm-level data in the agricultural SME sector are notably rare. To the best of our knowledge, no empirical research has examined the applicability of the predominant capital structure theories in this specific context.

Accordingly, this study provides fresh evidence on how leverage relates to performance in a context that remains both challenging and underexplored. Specifically, it examines whether financial leverage significantly affects firm performance, measured by return on assets (ROA), and whether the relationship is nonlinear, potentially indicating an optimal debt threshold. Two central research questions guide the analysis: (1) Is there a statistically significant relationship between financial leverage and the financial performance of Moroccan agricultural SMEs? and (2) Does this relationship exhibit a nonlinear pattern, indicating the presence of a turning point in the impact of leverage on performance?

The remainder of this paper is structured as follows: Section 2 presents the literature review and the development of the research hypotheses; Section 3 outlines the methodological framework; Section 4 reports the main empirical results; Section 5 analyzes these findings and includes robustness checks; and Section 6 concludes the paper by discussing the main policy implications, outlining the study’s limitations, and suggesting avenues for future research.

2. Literature Review

2.1. Theoretical Perspective

Investigating the capital structure of SMEs, especially medium-sized firms, is a complex task, as it requires assessing how debt usage influences performance through financial leverage. Most prior research has focused on large, often publicly listed corporations, leaving SMEs relatively understudied in theoretical frameworks (T. H. L. Beck et al., 2005; Colot & Croquet, 2007; Berger & Udell, 1998). As noted by Denis (2004), dominant capital structure theories only partially account for SME-specific characteristics, such as information asymmetry, agency conflicts, and limited access to external finance (Ang, 1991). Nonetheless, these theories remain relevant for understanding SMEs’ financial decisions, notably in weighing the fiscal advantages of debt against the potential risk of financial distress. Therefore, capital structure choices are strategic long-term decisions that critically influence SMEs’ growth and sustainability (Kumar & Rao, 2015). Several studies have demonstrated that financing choices play a decisive role in shaping the growth paths of SMEs (T. Beck & Demirguc-Kunt, 2006; Carpenter & Petersen, 2002), which underscores the need to adapt theoretical frameworks to the specific economic realities of these firms—especially in under-researched sectors such as agriculture.

To guide the analysis of SMEs’ financing behavior, particularly in agriculture, three dominant theoretical approaches offer insights into SMEs’ financing behavior, including those in the agricultural sector: the trade-off theory, the pecking order theory, and the agency cost theory.

The trade-off theory posits that firms establish their capital structure by weighing the tax benefits of debt against its costs, especially those related to bankruptcy and agency issues (Kraus & Litzenberger, 1973; Myers, 1984). Rooted in the foundational work of Modigliani and Miller (1958, 1963), this theory asserts that the tax deductibility of interest enhances firm value, making debt an attractive option. However, as indebtedness rises, so does the likelihood of financial distress (Titman & Wessels, 1988; Warner, 1977) and the potential for conflicts between creditors and shareholders (Jensen & Meckling, 1976), which together elevate the cost of capital. Therefore, firms are expected to identify an optimal debt ratio that maximizes returns while containing risk (Bradley et al., 1984; Miller, 1977). From this perspective, the capital structure reflects a calculated balance between tax benefits and the risks of financial strain, which is a particularly critical consideration for SMEs that must balance growth ambitions with limited financial resources.

The pecking order theory (POT), advanced by Myers (1984) and later by Myers and Majluf (1984), challenges the notion of a universally optimal capital structure. Instead, it proposes a financing hierarchy in which firms prefer to use internal resources first, debt second, and equity as the last option. This order is motivated by the desire to limit the costs associated with information asymmetries between managers and outside investors. Within this framework, more profitable firms rely less on external borrowing, leading to lower leverage ratios. Consequently, an inverse relationship is expected between profitability and indebtedness (Abel, 2018), as higher profits reduce reliance on debt and related financial costs. As noted by Baker (1973), this bidirectional dynamic between profitability and leverage continues to fuel the theoretical debate regarding the role of capital structure in shaping firm performance.

Lastly, agency theory, developed by Jensen and Meckling (1976), provides a complementary perspective by focusing on potential conflicts of interest between the firm’s stakeholders, particularly between shareholders and managers or managers and creditors. The firm is viewed as a nexus of contracts among self-interested agents, each pursuing distinct objectives. In this context, financing decisions are shaped by agency costs, which include monitoring expenses, imposed constraints, and opportunity costs stemming from suboptimal decision making (Bradley et al., 1984; Kester, 1986; Titman & Wessels, 1988). These costs tend to be higher in SMEs because of their more informal governance structures and lower transparency (Michaelas et al., 1999; Pettit & Singer, 1985). Debt financing may exacerbate some conflicts, particularly with creditors, but it can also mitigate others by limiting excess free cash flow that can be diverted to discretionary or opportunistic projects. Thus, the capital structure plays a key role in regulating agency relationships.

Taken together, these theories provide complementary perspectives for interpreting SME financing decisions under conditions of uncertainty and constraint. The trade-off theory underscores the search for an optimal leverage point at which tax advantages are balanced against bankruptcy and agency risks, a particularly sensitive issue for SMEs with constrained financing options. In contrast, the pecking order theory stresses a practical financing sequence shaped by information asymmetries, in which profitable firms tend to avoid debt, thereby explaining the commonly observed negative link between profitability and leverage. Meanwhile, agency theory underscores the governance challenges and incentive misalignments typical of SMEs, in which informal management practices heighten agency costs and influence how debt shapes managerial behavior. Together, these three theories offer a multidimensional lens to understand how debt shapes SME performance—not only as a financial resource but also as a mechanism of control, signaling, and risk management—particularly in sectors as volatile and capital-intensive as agriculture.

Building on these theoretical foundations, we develop a framework in which financial leverage can influence firm performance in both linear and nonlinear ways, depending on its level and operating context. At low levels, debt may heighten financial fragility, increase interest burdens, and amplify agency costs, thereby reducing operating efficiency and asset returns. Conversely, beyond a certain threshold, leverage can relax capital constraints, finance productivity-enhancing investments, and impose managerial discipline. In Morocco’s agricultural sector, characterized by revenue seasonality, restricted access to formal finance, and reliance on public support, such effects are likely to vary across firms. The sector’s chronic liquidity constraints and preference for internal funds make the pecking order theory (POT) particularly relevant, as SMEs often avoid debt unless it is unavoidable. Nonetheless, agency theory offers complementary insights, given informal governance structures and low transparency, which may weaken debt’s disciplinary role, while trade-off theory provides a broader lens on balancing financing costs and benefits, even if its assumptions of rational optimization are less applicable in volatile, resource-limited environments. Accordingly, this study adopts the POT as the primary interpretive framework while integrating elements of the other theories to capture the multifaceted dynamics of leverage in Moroccan agricultural SMEs.

2.2. Empirical Evidence

The link between capital structure and firm performance has been a central topic in corporate finance; however, the evidence remains inconclusive. While certain studies identify a positive association between leverage and performance, consistent with signaling theory (Myers & Majluf, 1984) and the financial discipline hypothesis (Jensen, 1986), others report either adverse or insignificant effects. For instance, Abor (2005), using a sample of listed firms in Ghana, found that debt—especially short-term debt—is positively correlated with profitability, as measured by return on equity (ROE). Similarly, Gill and Biger (2011) studied 272 listed U.S. firms and concluded that debt exerts a positive and significant effect on profitability, suggesting that the most profitable firms are those that use financial leverage effectively.

In contrast, some studies point to the detrimental effects of debt on firm performance. Salim and Yadav (2012), using data from 237 Malaysian firms, found that although leverage boosts market-based indicators such as Tobin’s Q, it reduces accounting-based measures such as return on assets and return on equity, underscoring a discrepancy between investor perception and actual profitability. Al-Taani (2013), analyzing 45 Jordanian industrial firms, also identified a significant negative impact of leverage on performance. Likewise, Hasan et al. (2014) found that in Bangladesh, financial leverage deteriorates firm performance, suggesting that in emerging economies, the costs of debt may outweigh its benefits. In Vietnam, Le and Phan (2017) reported an overall detrimental effect of leverage on all key performance measures (ROA, ROE, and Tobin’s Q), which they attributed to high financing costs and significant risk exposure. In Central Europe, Wieczorek-Kosmala et al. (2021) showed that debt is generally negatively correlated with financial performance, except for short-term debt, which may occasionally support liquidity.

Firm size also appears to moderate this relationship. Jaisinghani and Kanjilal (2017) emphasized that leverage has a positive effect on the performance of large Indian firms but a negative effect on smaller ones, highlighting the importance of considering the structural specificities of SMEs in such analyses.

In the agricultural sector, a study by Mugera and Nyambane (2015) conducted in Western Australia on an initial sample of 4000 large farms provides valuable insights into the differentiated effects of debt maturity. Their findings indicate that short-term debt positively affects technical and scale efficiency but negatively influences return on assets (ROA), whereas long-term debt shows no significant relationship with firm performance.

While extensive research has documented the leverage–performance nexus (Berger & Bonaccorsi di Patti, 2006; Dawar, 2014; Margaritis & Psillaki, 2010), relatively few studies have concentrated on SMEs. This gap is partly due to limited data availability and partly due to distinctive SME characteristics, such as restricted access to external finance, informal governance structures, and dependence on local markets (Abor & Biekpe, 2009; Psillaki & Daskalakis, 2009). Within this context, several empirical studies have attempted to examine how leverage influences SME performance while considering these contextual constraints.

Abor (2007), studying SMEs in Ghana and South Africa, concluded that both long-term and overall debt ratios reduce firm performance, implying that agency issues may drive firms to adopt excessive debt, thereby lowering their profitability. Similarly, Jha and Kumar Mittal (2024), analyzing 226 Indian SMEs, found that debt generally exerts a negative effect on financial performance, except for trade credit, which provides short-term benefits. In contrast, Kim (2022), using a sample of 300 Vietnamese SMEs, identified a positive link between leverage and performance. Finally, Wahba (2013) highlights the differentiated effect of debt according to maturity in Egyptian SMEs: long-term borrowing supports performance, whereas short-term debt undermines it.

2.3. Hypotheses’ Development

Empirical research on the relationship between debt and financial performance reveals mixed results, especially among SMEs, whose structural characteristics significantly shape this relationship. Unlike large firms, SMEs—especially those operating in the agricultural sector—face significant financing constraints, strong dependence on local environments, informal governance structures, and volatile profitability. These characteristics limit their ability to fully benefit from the theoretical advantages associated with debt.

From a theoretical standpoint, several frameworks help elucidate this relationship. Trade-off theory (Kraus & Litzenberger, 1973) posits that while debt generates tax advantages, these are counterbalanced by bankruptcy costs, which are often more significant for SMEs because of their financial vulnerability and restricted access to capital markets. As noted by Ang (1991), these firms often exhibit lower profitability and are subject to reduced tax rates, which diminishes the fiscal incentive to use debt. Pecking order theory (Myers & Majluf, 1984) suggests that SMEs prioritize internal resources because of pronounced information asymmetries and resort to external borrowing only when other options are exhausted. Consequently, firms with higher profitability usually maintain lower debt ratios. Finally, agency theory (Jensen & Meckling, 1976) highlights potential conflicts between creditors and managers, which are exacerbated in SMEs because of their informal governance and lack of transparency. This weakens the effectiveness of debt as a disciplinary mechanism.

In this context, Ray and Hutchinson (1984) argued that SMEs use debt less because they are more susceptible to bankruptcy risk. Similar arguments are advanced by McConnell and Pettit (1984) and Pettit and Singer (1985), who note that smaller firms face disproportionately high external financing costs. Financial institutions often perceive SMEs as riskier borrowers, leading to unfavorable credit conditions or partial exclusion from formal lending. Furthermore, stronger information asymmetries between lenders and SME managers exacerbate agency problems (Jensen, 1986) and limit the ability of debt to serve as an effective disciplinary tool.

These constraints may cause SMEs to rely excessively on costly financing, leading to financial rigidity and lower profitability, particularly in income-volatile sectors such as agriculture. In light of these theoretical and empirical insights, the following hypothesis is proposed:

Hypothesis 1. 

Financial leverage negatively affects the performance of Moroccan agricultural SMEs.

Recent studies increasingly point to a nonlinear association between leverage and firm performance, showing that the effect of debt depends on its magnitude. The observed U-shaped-or inverted U-shaped patterns suggest that debt can be either beneficial or harmful, depending on whether leverage remains low or becomes excessive.

Zeitun and Goaied (2022), analyzing 1670 Japanese listed firms using a fixed effects model, reported a U-shaped relationship between short-term debt (STDTD) and performance. Specifically, short-term debt ratios below 45.2% reduce performance, whereas higher ratios beyond that threshold are associated with positive effects. Similarly, in Egypt, Orabi Awad and Mohamed Ali (2022) analyzed a panel of 78 non-financial listed firms using a dynamic GMM model and a quadratic specification of leverage. Their results confirm a nonlinear association between financial leverage and performance indicators (ROA, ROE, and Tobin’s Q), showing a negative impact at lower levels of debt but a positive impact as leverage increases, thus supporting the U-shaped relationship.

Conversely, Le and Phan (2017), using data from Vietnamese firms, identified an inverted U-shaped relationship: low levels of debt are positively associated with profitability (particularly ROE), but beyond a certain point, additional leverage becomes counterproductive and deteriorates financial performance.

Such evidence highlights the need for more nuanced analyses of financial leverage, especially for SMEs. At relatively low levels, debt may impair SME performance because of bankruptcy costs (Kraus & Litzenberger, 1973; Modigliani & Miller, 1963), information asymmetry, and limited governance mechanisms that exacerbate agency problems (Jensen, 1986). However, at higher levels, leverage may also play a disciplinary role (Jensen, 1986), signal firm quality (Myers & Majluf, 1984), and provide tax advantages, as suggested by trade-off, agency, and signaling theories. This dual effect is consistent with the hypothesis that the leverage–performance relationship is non-linear.

To capture this dynamic, this study examines the quadratic relationship between leverage and the performance of Moroccan agricultural SMEs, following the approach of Berger and Bonaccorsi di Patti (2006), Margaritis and Psillaki (2010), and Le and Phan (2017). Thus, the following hypothesis is proposed:

Hypothesis 2. 

The relationship between financial leverage and the performance of Moroccan agricultural SMEs is U-shaped: leverage has a negative effect at low levels, but a positive effect at higher levels.

3. Methodology

3.1. Data

Our study is based on financial data collected from 54 Moroccan agricultural SMEs over the period 2017–2022, resulting in a balanced panel dataset of 324 firm-year observations. Data were obtained through the “Directinfo” platform of the Moroccan Office of Industrial and Commercial Property (OMPIC), which provides access to legal filings, audited financial statements, and other related reports. However, given that SME financial information in Morocco is not openly available and no public institution facilitates access for research purposes, each document had to be obtained via direct purchase.

The dataset was assembled based on strict inclusion criteria to guarantee its quality and relevance. We retained only SMEs legally registered under limited liability forms (SARL, SARL à associé unique, and GIE) and operating in the agriculture, horticulture, and livestock sectors. Firms with incomplete or inconsistent disclosures, material accounting irregularities, and negative revenue in consecutive years were excluded. The final dataset consisted of 54 SMEs with verified and complete financial records across six consecutive years.

Although challenges related to data accessibility and coverage persist, the sample is sufficiently robust for empirical analysis in light of established econometric standards. As emphasized in panel data econometrics (Greene, 2012; Wooldridge, 2010), balanced panels with adequate cross-sectional and temporal variation, such as the one used in this study, allow for consistent estimation and statistically reliable inferences, even in small- or medium-sized samples.

3.2. Variables

3.2.1. Financial Performance

Firm performance is measured using return on assets (ROA), a widely employed accounting-based indicator in empirical corporate finance studies. This measure is widely adopted in recent studies, such as those by Mugera and Nyambane (2015), Abuamsha and Shumali (2022), Orabi Awad and Mohamed Ali (2022), and Khan and Qasem (2024). ROA reflects the efficiency with which a company employs its total assets to produce operating income and is calculated as the ratio of net operating profit to total assets.

Although return on equity (ROE) is frequently used in the literature to assess financial profitability, we did not include it in our analysis due to a structural issue in the dataset: a significant number of SMEs reported negative equity across multiple years, primarily as a result of cumulative financial losses. In such cases, ROE becomes undefined or yields highly volatile and non-informative values, undermining its reliability.

Other indicators, such as earnings per share (EPS) or Tobin’s Q, are also commonly used in financial performance research but are not applicable in our case, as they require market-based data that are unavailable for unlisted agricultural SMEs in Morocco.

Table 1. Description of dependent, independent, and control variables.

Variables		Notation	Proxies	Definitions
Dependent Variable	Financial Performance	ROA	Return on assets	Net income/total assets
Independent Variables	Leverage	STLEV	Short-term debt to assets	Short-term debt/total assets
		LTLEV	Long-term debt to assets	Long-term debt/total assets
Control Variables	SME Size	SIZE	Natural logarithm (Ln) of total assets	Ln (total assets)
	SME Age	AGE	Number of years since establishment	Current year—Year of creation
	Tangibility	TANG	Fixed assets to total assets	Fixed assets/total assets
	Liquidity	LIQUID	Current ratio	Current assets/current liabilities
	Profitability Growth	PG	Change in net income over the year	[Net income (t) − Net income (t − 1)]/|Net income (t − 1)|

presents a detailed summary of the variables employed in this analysis, outlining their definitions, measurement methods, and proxies.

3.2.2. Measure of Financial Leverage

In this study, financial leverage is measured using two complementary indicators that differentiate debt by maturity. The first is the short-term debt-to-total-assets ratio (STLEV), which captures the share of assets financed by short-term liabilities. This measure reflects immediate financing pressures and liquidity risk and is widely used in SME and capital structure research (Jha & Kumar Mittal, 2024; Le & Phan, 2017; Wahba, 2013; Zeitun & Goaied, 2022). The second indicator is the long-term debt-to-total-assets ratio (LTLEV), which reflects the extent of long-term financial commitments and is an indicator of strategic investment financing. This ratio is based on studies that examine the impact of debt maturity on firm performance and financial stability (Abuamsha & Shumali, 2022; Jaisinghani & Kanjilal, 2017; Khan & Qasem, 2024; Le & Phan, 2017; Mugera & Nyambane, 2015). By differentiating between short- and long-term leverage, this study provides a more nuanced perspective on how debt maturity structure influences the financial performance of agricultural SMEs.

3.2.3. Measure of Control Variables

To capture firm-specific characteristics that may affect the financial performance of agricultural SMEs, five control variables are included in the model: firm age (AGE), size (SIZE), asset tangibility (TANG), liquidity (LIQUID), and profitability growth (PG). The selection of these variables is informed by prior studies highlighting their potential influence on financial performance, including those by Abor and Biekpe (2009), Margaritis and Psillaki (2010), Kim (2022), Zeitun and Goaied (2022), Orabi Awad and Mohamed Ali (2022), and Jha and Kumar Mittal (2024).

For instance, Jha and Kumar Mittal (2024) found a significant positive effect of firm size on return on assets (ROA), suggesting that larger SMEs may benefit from improved access to financing and investment opportunities. In the same study, asset tangibility and liquidity were shown to negatively impact financial performance, indicating that high levels of these factors may reflect inefficient resource management. Similarly, Kim (2022) found that both firm size and age positively influence SME performance in Vietnam, though the study provided little theoretical explanation for these effects.

3.3. Empirical Models

To examine how leverage affects the performance of Moroccan agricultural SMEs, we applied a panel data regression framework. This choice reflects both the structure of our dataset, which combines cross-sectional and time series elements, and the need to account for unobserved firm heterogeneity (Ngatno et al., 2021). Compared to purely cross-sectional or time series approaches, panel data provide key benefits. First, by tracking firms across several years (2017–2022), they allow for a more precise estimation of causal effects while controlling for unobservable, time-invariant firm characteristics, thereby reducing omitted variable bias (Baltagi, 2008; Wooldridge, 2010). Second, panel data expand the number of observations, which increases the degrees of freedom and enhances estimation efficiency. Third, they allow us to detect and model dynamic changes over time, such as the evolving impact of debt levels under different seasonal or policy conditions, which are particularly relevant in the agricultural sector. Panel data models are widely used in the empirical literature on capital structure and firm performance (Ahmad et al., 2022; Ayalew, 2021; Dawar, 2014; Le & Phan, 2017; Liu et al., 2019; Siddik et al., 2017).

In this study, three standard estimation techniques were considered for the linear panel data models: ordinary least squares (OLS), fixed effects (FE), and random effects (RE). The appropriate specification was determined using conventional statistical tests, namely, the Breusch–Pagan Lagrange multiplier (LM) test, to differentiate between pooled OLS and RE, and the Hausman test to decide between FE and RE.

The general panel regression model is specified as follows:

$$Y_{it}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{X}_{it}+{\mathsf{\beta }}_2{C}_{it}+{\mathsf{\epsilon }}_{it}$$

(1)

where $Y_{it}$ denotes the dependent variable, ${\mathsf{\beta }}_0$ is the intercept, ${\mathsf{\beta }}_1$ and ${\mathsf{\beta }}_2$ are vectors of coefficients for the independent variable $X_{it}$ and control variables $C_{it}$, and ${\mathsf{\epsilon }}_{it}$ is the idiosyncratic error term.

To evaluate the linear relationship between leverage and performance, we estimate the following base model:

$${FP}_{i,t}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{LEV}_{i,t}+{\mathsf{\beta }}_2{C}_{i,t}+{\mathsf{\epsilon }}_{it}$$

(2)

where ${FP}_{i,t}$ represents the financial performance of firm i in year t, measured by return on assets (ROA). The variable ${LEV}_{i,t}$ represents financial leverage, captured through two complementary indicators: the short-term debt-to-total-assets ratio (STLEV) and the long-term debt-to-total-assets ratio (LTLEV). The vector $C_{i,t}$ includes control variables.

Accordingly, the estimated empirical model is as follows:

Model—Return on assets (ROA)

$${ROA}_{it}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{STLEV}_{it}+{\mathsf{\beta }}_2{LTLEV}_{it}+{\mathsf{\beta }}_3{SIZE}_{it}+{\mathsf{\beta }}_4{AGE}_{it}+{\mathsf{\beta }}_5{LIQUID}_{it}+{\mathsf{\beta }}_6{TANG}_{it}+{\mathsf{\beta }}_7{PG}_{it}+{\mathsf{\epsilon }}_{it}$$

(3)

To control for macroeconomic shocks and unobserved temporal effects that may affect all firms in a given year, we include year-fixed effects as recommended by Wooldridge (2010). The extended model specification is as follows:

$${FP}_{i,t}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{LEV}_{i,t}+{\mathsf{\beta }}_2{C}_{i,t}+{\mathsf{\delta }}_t+{\mathsf{\epsilon }}_{it}$$

(4)

where ${\mathsf{\delta }}_t$ denotes year-specific dummy variables.

Furthermore, to capture potential nonlinearities, a quadratic term for leverage is included, following the methodology of Berger and Bonaccorsi di Patti (2006) and Margaritis and Psillaki (2010). This allows us to test whether the leverage–performance link takes a U-shaped form, whereby low debt levels reduce performance, but higher levels eventually exert a positive effect once a threshold is reached. The nonlinear specification is as follows:

$${FP}_{i,t}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{LEV}_{i,t}+{\mathsf{\beta }}_2{LEV}_{i,t}^2+{\mathsf{\beta }}_3{C}_{i,t}+{\mathsf{\epsilon }}_{it}$$

(5)

The necessary condition to validate a U-shaped relationship is ${\mathsf{\beta }}_1$ < 0 and ${\mathsf{\beta }}_2$ > 0. This functional form allows for a more nuanced understanding of how varying levels of debt impact firm performance.

3.4. Diagnostic Tests

3.4.1. Unit Root Test

Prior to panel data estimation, it is necessary to verify the stationarity of the variables to prevent spurious regression results. To this end, several panel unit root tests were employed, namely Levin–Lin–Chu (LLC) (Levin et al., 2002), Phillips–Perron (PP) (Phillips & Perron, 1988), augmented Dickey–Fuller (ADF) (Dickey & Fuller, 1981), and Kwiatkowski–Phillips–Schmidt–Shin (KPSS) (Kwiatkowski et al., 1992) tests. The corresponding results are presented in

Table 2. Stationarity check of variables.

Unit Root Tests		Time Series Tests for Variables
		ROA	STLEV	LTLEV	AGE	SIZE	TANG	LIQUID	PG
LLC	Stat.	−5.470	−8.905	−1.734	−28.400	−3.868	−5.617	−0.673	−8.589
	Prob.	0.000	0.000	0.042	0.000	0.000	0.000	0.251	0.000
PP	p-Value	0.01	0.01	0.01	0.05	0.01	0.01	0.01	0.01
ADF	p-Value	0.01	0.01	0.01	0.25	0.01	0.01	0.01	0.01
KPSS	p-Value	0.1	0.1	0.1	0.1	0.01	0.1	0.1	0.1

.

The LLC, PP, and ADF tests assume the null hypothesis of a unit root (i.e., non-stationarity), whereas the KPSS test assumes stationarity under the null hypothesis. According to the results, the variables ROA, STLEV, LTLEV, TANG, and PG are found to be stationary at level across all tests.

For a few variables, the stationarity results were mixed. AGE appears non-stationary according to the ADF test but stationary according to the LLC, PP, and KPSS tests. SIZE fails the KPSS test but is stationary under the LLC, PP, and ADF tests. LIQUID is found to be non-stationary by the LLC test but stationary under the PP, KPSS, and ADF tests.

Despite these minor inconsistencies, the overall evidence suggests that all variables are stationary in the level form, thus fulfilling a key prerequisite for panel data analysis.

3.4.2. Multicollinearity Test

To detect potential multicollinearity among explanatory variables, we computed tolerance and variance inflation factor (VIF) statistics (Ahmed et al., 2023; Jha & Kumar Mittal, 2024), as reported in

Table 3. Tolerance and variance inflation factors.

Model	Collinearity Statistics
	Tolerance	VIF
STLEV	0.598	1.673
LTLEV	0.788	1.270
SIZE	0.572	1.749
AGE	0.696	1.436
LIQUID	0.715	1.400
TANG	0.710	1.409
PG	0.970	1.031

. All VIF values fell comfortably below the conventional cutoff of 10, ranging from 1.031 to 1.749. Similarly, all tolerance values exceed the minimum benchmark of 0.1.

These results suggest that multicollinearity is not a concern, implying that the explanatory variables are not strongly collinear and that coefficient estimates remain robust.

3.4.3. Heteroscedasticity Test

The presence of heteroskedasticity, or non-constant variance of residuals, was assessed using the White test, modified Breusch–Pagan test, and the F-test (Breusch & Pagan, 1979; White, 1980; Wooldridge, 2013). As presented in

Table 4. Heteroscedasticity diagnostics.

White Test for Heteroskedasticity
Chi-Square	112.675
Sig.	0.000
Modified Breush–Pagan Test for Heteroskedasticity
Chi-Square	25.790
Sig.	0.001
F Test for Heteroskedasticity
F	4.165
Sig.	0.000

Note: These test the null hypothesis that the variance of the errors does not depend on the values of independent variables.

, the null hypothesis of homoscedasticity is rejected at the 1% level across all three tests (p-values < 0.01) in the ROA regression model.

These findings indicate the presence of heteroskedasticity, which may lead to inefficient estimates and biased standard errors if not corrected. Accordingly, robust standard errors were employed in subsequent estimations to address this issue, following the guidance of Wooldridge (2013).

3.4.4. Autocorrelation Test

Serial correlation was examined using the Durbin–Watson (DW) statistic to verify the robustness of the estimation results (Drukker, 2003). The DW value for the ROA model is 1.265, suggesting a potential positive autocorrelation.

To formally interpret these values, we refer to the critical thresholds provided by Savin and White (1977), adjusted for the number of regressors (k′) and the sample size (n). Based on these benchmarks, the DW statistic for the ROA model falls below the lower bound, indicating the presence of serial correlation. This necessitates appropriate methodological adjustments in the estimation process to ensure the robustness of standard errors and the reliability of statistical inference.

3.5. Model Specification

The selection of the appropriate econometric specification was guided by the panel data estimation protocol proposed by Dougherty (2011), which involves a three-step process: validation of the sample to determine the suitability of a fixed effects (FE) model, comparison between fixed and random effects (RE) using the Hausman test, and application of the Lagrange multiplier (LM) test to determine whether a random effects model is appropriate or if a pooled OLS specification is preferable.

As shown in

Table 5. Results of specification tests (Hausman and LM tests).

Tests
	Chi-Sq. Stat.	Prob.
Correlated Random Effects— Hausman Test	29.1068	0.0001
Lagrange Multiplier Tests for Random Effects	11.1137	(0.0009)
Breusch–Pagan

, a statistically significant Hausman test for the ROA model (p = 0.0001) rejecting the null hypothesis of no correlation between regressors and unobserved effects. Consequently, the fixed effects (FE) specification is preferred over the random effects (RE) alternative.

Following the initial estimations using pooled OLS, RE, and FE, diagnostic tests revealed violations of key OLS assumptions. Specifically, the ROA model exhibited signs of heteroskedasticity and autocorrelation. To address these econometric issues and improve the reliability of the estimates, two additional specifications were employed: (i) fixed effects with robust (white-corrected) standard errors, which account for heteroskedasticity, and (ii) fixed effects with cross-section weights. This method corrects for heteroskedasticity by assigning lower weights to units with higher error variance and mitigates serial correlation by adjusting for intra-group autocorrelation (Greene, 2012; Wooldridge, 2013).

To further validate our results and address potential endogeneity concerns—particularly reverse causality and omitted variable bias—the study also employed the Arellano and Bond (1991) dynamic panel GMM estimator as a robustness check. This approach is well suited for our dataset, which features a relatively short time dimension (six years) and a moderate cross-sectional size. By employing lagged explanatory variables as instruments, the GMM estimator helps mitigate endogeneity while enhancing the consistency of the coefficient estimates. This supplementary estimation confirms the stability of our key findings.

4. Findings

4.1. Descriptive Statistics of Data

Descriptive statistics of the key variables are provided in

Table 6. Descriptive statistics.

	N	Minimum	Maximum	Mean		Std. Deviation	Skewness	Kurtosis
	Statistic	Statistic	Statistic	Statistic	Std. Error	Statistic	Statistic	Statistic
ROA	324	−0.418	0.268	0.015	0.007	0.090	−2	7.500
STLEV	324	0.022	3.271	0.696	0.039	0.496	2	7.156
LTLEV	324	0.000	0.626	0.063	0.010	0.122	3	8.020
SIZE	324	13.377	22.980	17.800	0.176	2.244	0	−0.428
AGE	324	1.099	4.304	2.880	0.053	0.670	0	−0.273
LIQUID	324	0.288	47.439	2.008	0.331	4.210	9	86.855
TANG	324	0.000	0.845	0.268	0.018	0.229	1	−0.488
PG	324	−78.595	119.296	0.248	0.924	11.765	5	77.934

. The mean return on assets (ROA)—the main proxy for financial performance—stands at 1.5% with a standard deviation of 9%. Values range from −41.8% to 26.8%, reflecting wide disparities in operational efficiency among Moroccan agricultural SMEs.

This variability may reflect divergent business models, seasonal revenue patterns, and uneven access to capital or technical support. For comparison, Mugera and Nyambane (2015) report an average ROA of −4% for agricultural enterprises in Western Australia, while Singh et al. (2019) report an unusually high average ROA of 114% for U.S. farms. More moderate figures are reported by Pokharel et al. (2020), who indicate ROA and ROE values of 8.12% and 13.44%, respectively, for U.S. agricultural SMEs. These disparities likely stem from structural differences in productivity, institutional support, financial access, and technology adoption.

Regarding capital structure, short-term leverage (STLEV) averages 69.6% (with a std. dev. of 49.6%), indicating a strong reliance on current liabilities to finance operations. Long-term leverage (LTLEV) remains comparatively low at 6.3% on average (with a std. dev. of 12.2%), reflecting persistent challenges in accessing long-term financing.

Compared to international benchmarks, these figures reflect a distinctive financing profile. For instance, Singh et al. (2019) report an average financial leverage of 23% for U.S. farms between 2009 and 2017, while Mugera and Nyambane (2015) found average short-term and long-term debt ratios of 4% and 9%, respectively, for Western Australian farms over 1995–2005. The elevated STLEV in our sample may point to working capital constraints, underdeveloped capital markets, or limited financial planning practices among Moroccan agricultural SMEs.

Overall, the wide dispersion of leverage and performance measures in the sample indicates considerable heterogeneity in capital structure, financial health, and operational strategies, reinforcing the need for nuanced econometric modeling in the subsequent analysis.

To evaluate distributional assumptions, skewness and kurtosis metrics were examined. Following Brooks (2018), approximately normal distributions are characterized by a skewness value within ±1.9 and a kurtosis value within ±3. The observed variables do not meet these thresholds, indicating non-normality in the data distribution. However, Greene (2012) emphasizes that regression does not require normally distributed variables; what matters is that residuals approximate normality and classical assumptions hold. With a sufficiently large sample, the central limit theorem ensures that estimator distributions converge toward normality, supporting valid inferences (Greene, 2012; Wooldridge, 2013).

4.2. Correlation Analysis

Table 7. Correlation matrix.

Pearson Correlation	ROA	STLEV	LTLEV	SIZE	AGE	LIQUID	TANG	PG
ROA	––
STLEV	−0.280 **	––
	(0.000)
LTLEV	−0.103	−0.289 **	––
	(0.194)	(0.000)
SIZE	0.147	−0.392 **	0.205 **	––
	(0.061)	(0.000)	(0.009)
AGE	0.113	−0.149	−0.115	0.377 **	––
	(0.152)	(0.059)	(0.146)	(0.000)
LIQUID	0.206 **	−0.292 **	−0.076	−0.275 **	−0.060	––
	(0.009)	(0.000)	(0.339)	(0.000)	(0.451)
TANG	−0.165 *	−0.241 **	0.381 **	0.135	−0.331 **	−0.027	––
	(0.036)	(0.002)	(0.000)	(0.087)	(0.000)	(0.737)
PG	0.175 *	0.061	−0.006	0.098	0.028	−0.012	−0.057	––
	(0.026)	(0.442)	(0.935)	(0.215)	(0.721)	(0.877)	(0.472)

Note: Values in parentheses indicate significance levels. ** and * denote correlation significance at the 0.01 (two-tailed) and 0.05 (two-tailed) levels, respectively.

displays the correlation matrix for the study variables. Several significant associations emerge, offering initial insights into the linkages among firm characteristics, leverage, and financial performance.

Return on assets (ROA) is negatively correlated with short-term leverage (STLEV) (r = −0.280, p < 0.01), suggesting that an increased reliance on short-term debt tends to reduce asset-based profitability. ROA also shows positive, though moderate, correlations with liquidity (LIQUID) (r = 0.206, p < 0.01), tangibility (TANG) (r = −0.165, p < 0.05), and profit growth (PG) (r = 0.175, p < 0.05), indicating that more liquid and growing SMEs tend to exhibit better performance, while SMEs with high fixed asset intensity may experience operational inefficiencies.

A negative correlation is observed between short- and long-term leverage (r = −0.289, p < 0.01), highlighting a potential substitution effect in capital structure decisions. STLEV is also negatively correlated with firm size (SIZE) (r = −0.392, p < 0.01), liquidity (r = −0.292, p < 0.01), and asset tangibility (r = −0.241, p < 0.01), implying that smaller, less liquid SMEs with fewer tangible assets rely more heavily on short-term debt.

In contrast, LTLEV is positively associated with both SIZE (r = 0.205, p < 0.01) and TANG (r = 0.381, p < 0.01), suggesting that larger SMEs with stronger collateral assets are more likely to secure long-term financing, consistent with capital market expectations.

Moreover, firm age (AGE) shows a negative correlation with tangibility (−0.331, p < 0.01), possibly reflecting changes in asset structures as firms mature. Finally, SIZE is inversely associated with liquidity (r = −0.275, p < 0.01), indicating that smaller SMEs tend to hold higher liquidity buffers, possibly due to precautionary motives or limited reinvestment opportunities.

These correlations offer preliminary support for several expected theoretical relationships and underscore the importance of accounting for firm-specific factors in the regression analyses that follow.

4.3. Regression Results

4.3.1. Pooled OLS, Random Effects, and Fixed Effects Regressions

The estimation results from the OLS, Fixed Effects (FE), and Random Effects (RE) models are presented in

Table 8. Estimation results: pooled OLS, fixed effects, and random effects models.

Model (ROA)	OLS		FE		RE
	Coef.	t-Stat.	Coef.	t-Stat.	Coef.	t-Stat.
Intercept	0.0050	0.068	−0.0735	−0.140	0.0254	0.260
STLEV	−0.0540	−3.189	−0.2800	−5.547	−0.0767	−3.503
	***		***		***
LTLEV	−0.0964	−1.603	−0.2131	−1.789	−0.1269	−1.703
			*		*
SIZE	0.0053	1.375	0.0288	0.946	0.0064	1.221
AGE	−0.0084	−0.720	−0.0608	−1.478	−0.0153	−0.954
LIQUID	0.0030	1.624	0.0056	2.904	0.0049	2.796
			***		***
TANG	−0.0835	−2.482	−0.1993	−2.026	−0.1067	−2.348
	**		**		**
PG	0.0013	2.335	0.0011	2.234	0.0013	2.687
	**		**		***
R-square	0.2022		0.5519		0.2124
Adjusted R-square	0.1659		0.4364		0.1766
F-statistic	5.5757		4.7778		5.9331
Prob.	0.0000		0.0000		0.0000

Note: ***, **, and * represent levels of significance of 1%, 5%, and 10%, respectively.

.

Across all specifications, short-term leverage (STLEV) exhibits a negative and strongly significant effect on ROA, confirming the robustness of this relationship and underscoring the financial strain associated with elevated levels of short-term debt. Long-term leverage (LTLEV) also has a negative coefficient, reaching significance at the 10% level in the FE and RE models, pointing to a weaker yet still detrimental impact on performance.

The explanatory power of the models is moderate, with adjusted R-squared values ranging from 17% to 44% in the FE estimation. Furthermore, the F-statistics yield p-values below 1% in all regressions, confirming the joint statistical significance of the explanatory variables.

Regarding the control variables, asset tangibility (TANG) consistently displays a negative and significant effect at the 5% level across models. Profit growth (PG) positively and significantly influences ROA at the 5% level, while liquidity (LIQUID) shows a positive and highly significant association under the FE and RE specifications (1% level). By contrast, firm size (SIZE) and age (AGE) display no significant effects.

4.3.2. Fixed Effects Estimations with Robust Standard Errors and Cross-Section Weights

Based on the results from the Hausman test, Lagrange multiplier test, and Dougherty procedure (2011), the fixed effects model was adopted as the preferred specification for the ROA estimation. However, to correct for heteroskedasticity and autocorrelation in the residuals, two robust approaches were applied: FE with robust standard errors and FE with cross-section weights using estimated generalized least squares (EGLS).

Table 9. Fixed effects estimations with robust standard errors and cross-section weights.

Model (ROA)	FE with Robust Standard Errors		FE with Cross-Section Weights
	Coef.	t-Stat.	Coef.	t-Stat.
Intercept	−0.0735	−0.158	−0.0579	−0.937
STLEV	−0.2800	−4.334	−0.2135	−7.930
	***		***
LTLEV	−0.2131	−5.543	−0.1684	−6.293
	***		***
SIZE	0.02883	1.013	0.0209	2.828
			**
AGE	−0.0608	−2.385	−0.0440	−3.501
	*		***
LIQUID	0.0056	1.333	0.0060	1.255
TANG	−0.1993	−2.139	−0.0966	−3.018
	*		***
PG	0.0011	1.621	0.0010	2.875
			*
R-square	0.5519		0.8539
Adjusted R-square	0.4364		0.8162
F-statistic	4.7778		22.6692
Prob.	0.0000		0.0000

Note: ***, **, and * represent levels of significance of 1%, 5%, and 10%, respectively.

summarizes these results.

The explanatory power of the models is relatively strong, with adjusted R-squared values ranging from 44% (fixed effects with robust standard errors) to 82% (cross-section weighted estimation). In both specifications, the F-test confirms the joint statistical significance of the model, with p-values below the 1% threshold.

Short-term leverage (STLEV) and long-term leverage (LTLEV) exhibit negative and statistically significant effects on ROA at the 1% level across both estimations. Specifically, under the fixed effects model with robust standard errors, a 1% increase in STLEV is associated with a 0.28% decrease in ROA, while a 1% increase in LTLEV corresponds to a 0.21% decline in ROA. These results confirm and reinforce the earlier OLS, FE, and RE outcomes, offering consistent evidence that leverage exerts a detrimental effect on the performance of Moroccan agricultural SMEs.

Remarkably, while firm size (SIZE) and firm age (AGE) initially showed no significant impact under the OLS, FE, and RE estimations, the robust models reveal a positive and significant effect of SIZE at the 5% level and a negative effect of AGE at the 1% level, according to the model with cross-section weights.

5. Results

5.1. Financial Leverage

Within Moroccan agricultural SMEs, financial leverage shows a persistent and statistically significant negative association with firm performance, as measured by return on assets. Across all model specifications, short-term debt exerts a clearly adverse influence, indicating that reliance on short-term debt tends to undermine operational efficiency within the agricultural sector. Long-term leverage also displays a detrimental effect on performance. Although this relationship appears weaker in the standard fixed effects model, it becomes clearly significant and reinforced when more robust estimation techniques are applied. Taken together, these results suggest that both short- and long-term borrowing—despite their maturity distinctions—can weaken financial performance when used excessively. These results contrast with those reported by Kim (2022) and Abuamsha and Shumali (2022), but align with the evidence provided by Mugera and Nyambane (2015), Orabi Awad and Mohamed Ali (2022), Jha and Kumar Mittal (2024), and Khan and Qasem (2024), who find that excessive leverage impairs financial performance.

Several structural characteristics specific to agricultural SMEs may account for these results. As noted by Ang (1991), SMEs may not fully benefit from the tax shield associated with debt due to their relatively low profitability and often lower effective tax rates compared to large firms. Moreover, Ray and Hutchinson (1984) suggest that SMEs are generally less inclined to use debt because of their heightened vulnerability to bankruptcy. This view is further supported by McConnell and Pettit (1984) as well as Pettit and Singer (1985), who emphasize that SMEs typically face higher external financing costs, thereby limiting their ability to raise debt under favorable conditions. In a sector such as agriculture—characterized by climatic uncertainty and volatile yields—these constraints can amplify leverage’s detrimental effect on performance.

These empirical findings thus validate the theoretical underpinnings of both the trade-off and agency cost theories when applied to agricultural SMEs. Overall, the results support the first hypothesis ($H_1$) formulated in this study, namely that there exists a negative association between leverage and financial performance.

5.2. Control Variables

Regarding the control variables, firm size, age, asset tangibility, and profitability growth display consistent coefficient signs, although their levels of statistical significance vary across specifications. This suggests that these factors play an important—albeit context-dependent—role in shaping the financial performance of Moroccan agricultural SMEs, as measured by return on assets (ROA). Liquidity demonstrates a positive and significant contribution under the fixed effects model, implying that access to liquid resources may enhance operational stability and asset utilization, although this result is not robust across alternative estimation methods. Firm size (SIZE) and profitability growth (PG) exert a moderately positive influence on SME performance. This finding implies that larger enterprises can benefit from cost efficiencies and greater availability of formal financing (T. H. L. Beck et al., 2005), while strong past profitability reflects greater operational efficiency (Myers, 1984). Conversely, the negative effect of firm age (AGE) may reflect organizational rigidity or challenges in adapting to a changing environment (Coad & Tamvada, 2012). Similarly, asset tangibility (TANG)—typically viewed as favorable for credit access—displays a negative relationship, potentially due to the low liquidity and high maintenance costs associated with physical assets in the agricultural sector (Klinefelter, 2000). This could also indicate inefficient investment decisions or the underutilization of assets, especially given that agricultural equipment and land are often illiquid and exposed to climate-related risks (Zinyengere et al., 2013).

These findings are consistent with those of Jha and Kumar Mittal (2024) and Wahba (2013), who confirm the positive effect of firm size and the negative effect of asset tangibility on SME financial performance. However, Kim (2022) also supports the positive impact of firm size but reports a contrasting result concerning firm age, suggesting that older SMEs may experience improved performance due to accumulated experience and relational capital.

5.3. Robustness Check

To evaluate the robustness of the baseline results, two complementary strategies were implemented. First, annual time dummies were introduced to control for year-specific fixed effects, thereby capturing unobserved temporal shocks that may simultaneously affect all firms in a given year (Wooldridge, 2010). Second, to address potential endogeneity concerns—stemming from reverse causality, measurement errors, or omitted time-invariant variables—we employed a dynamic panel estimation using the generalized method of moments (GMM) proposed by Arellano and Bond (1991).

Although heteroskedasticity and autocorrelation can be mitigated through model-adjusted standard errors, fixed and random effects estimators remain vulnerable to endogeneity bias, especially in corporate finance contexts (Cameron & Trivedi, 2005; Wintoki et al., 2012). Unlike traditional instrumental-variable approaches that rely on external instruments, the Arellano–Bond generalized method of moments (GMM) procedure uses lagged values of endogenous regressors as internal instruments. This makes it well suited for short panels with many entities (N) but relatively few periods (T), as in our dataset. Moreover, GMM is advantageous for dynamic settings where past performance may influence current outcomes (Roodman, 2009).

The regression tables incorporating year dummies are omitted for brevity, as they do not yield additional insights beyond those reported in the main models. In these specifications, short-term and long-term leverage remain negatively and statistically significantly associated with firm performance at the 1% level.

The results obtained through the GMM estimation, as reported in

Table 10. Dynamic GMM estimation of the impact of financial leverage on return on assets.

Model (ROA)	GMM Estimator with Robust Standard Error
	Coef.	t-Stat.
STLEV	−0.1620	−2.625
	**
LTLEV	−0.2473	−2.311
	**
SIZE	0.0296	1.015
AGE	−0.0755	−2.107
	**
LIQUID	0.0061	25.317
	***
TANG	−0.2843	−1.900
	*
PG	0.0050	2.930
	***
L.ROA	0.0910	1.849
	*
AR(1)	0.2729
AR(2)	0.7349
J-statistic	8.0340
Prob. (J-statistic)	0.5307

Note: ***, **, and * represent levels of significance of 1%, 5%, and 10%, respectively.

, continue to confirm the primary findings of this study: both short-term and long-term leverage exhibit a negative and statistically significant relationship with return on assets, with coefficients significant at the 5% level. The lagged dependent variable is positive and marginally significant (10%), suggesting that profitable SMEs tend to sustain their financial performance over time.

Diagnostic analyses validate the consistency and reliability of the GMM specification. The Hansen test of overidentifying restrictions yields a J-statistic of 8.034 with a corresponding p-value of 0.5307, indicating that the null hypothesis of instrument validity cannot be rejected. The Arellano–Bond tests further indicate no evidence of second-order serial correlation (AR(2) p = 0.7349), while the presence of first-order autocorrelation (AR(1) p = 0.2729) remains within acceptable bounds. As noted by Arellano and Bond (1991), first-order autocorrelation is expected in first-differenced residuals, whereas the absence of second-order serial correlation is essential to ensure the validity of the moment conditions. This is consistent with the view expressed by Baltagi (2005), who argues that the AR(2) test is more informative for assessing the legitimacy of instruments in dynamic panel models.

5.4. Non-Linear Relationship Between Financial Leverage and Financial Performance

To explore the possibility of a nonlinear leverage–performance relationship in Moroccan agricultural SMEs, we adopted a quadratic specification, following the methodological approach of Berger and Bonaccorsi di Patti (2006) and Margaritis and Psillaki (2010). The results, displayed in

Table 11. Non-linear relationship between leverage and financial performance.

Model (ROA)	Coef.	t-Stat.	Coef.	t-Stat.
STLEV	−0.0889	−2.309
	**
${\mathrm{S}\mathrm{T}\mathrm{L}\mathrm{E}\mathrm{V}}^2$	−0.0442	−2.505
	**
LTLEV			−0.3085	−3.765
			***
${\mathrm{L}\mathrm{T}\mathrm{L}\mathrm{E}\mathrm{V}}^2$			0.2645	1.894
			**
SIZE	0.0091	1.409	0.0553	5.719
			***
AGE	−0.0455	−3.384	−0.0195	−1.035
	***
LIQUID	0.0102	2.040	0.0123	2.893
	**		***
TANG	−0.1363	−4.134	−0.0538	−1.175
	***
PG	0.0009	2.823	0.0012	2.908
	***		***
R-square	0.8433		0.8020
Adjusted R-square	0.8030		0.7510
F-statistic	20.8808		15.7159
Prob.	0.0000		0.0000

Note: *** and ** represent levels of significance of 1% and 5%, respectively.

, reveal mixed evidence of non-linearity, depending on the type and maturity structure of financial leverage. When leverage is measured by short-term debt (STLEV), both the linear and squared terms (STLEV²) exhibit statistically significant negative coefficients at the 5% level. This indicates that the relationship with return on assets (ROA) remains strictly adverse, with higher short-term borrowing consistently reducing efficiency and showing no sign of reversal at elevated debt levels. In other words, an increase in short-term debt consistently erodes operational efficiency, with no evidence of reversal at higher leverage levels.

In contrast, the findings for long-term leverage (LTLEV) point to a U-shaped relationship with ROA. Specifically, the linear term of LTLEV is negative and statistically significant at the 1% level, while the squared term (LTLEV²) is positive and significant at the 5% level. This pattern indicates that low levels of long-term debt negatively affect performance, but beyond a certain threshold, long-term borrowing begins to exert a positive effect on asset profitability. These findings are consistent with those of Zeitun and Goaied (2022) and Orabi Awad and Mohamed Ali (2022), who similarly identify a U-shaped pattern where moderate debt impairs performance, but high leverage levels become beneficial.

This U-shaped dynamic may be interpreted through the lens of the financial leverage effect: at low levels, debt burdens increase interest expenses and financial risk, thus depressing returns. However, once a critical threshold is exceeded, long-term leverage may enhance performance by enabling strategic investment, spreading fixed costs, or signaling financial discipline. The results for STLEV, on the other hand, indicate that short-term debt fails to provide such benefits, possibly due to its revolving nature, maturity mismatch, or the higher refinancing risks associated with it.

Alternative interpretations are also found in the literature. For instance, Le and Phan (2017) and Molinari (2013) identify inverted U-shaped relationships, where leverage initially improves performance but becomes detrimental when excessive, due to increased default risk and agency conflicts.

In the present case, our findings provide partial confirmation of the second hypothesis ${(H}_2)$, as a non-linear, U-shaped effect is identified for long-term leverage, while the relationship with short-term debt remains strictly negative. These results offer nuanced support for the trade-off theory, emphasizing that the effect of leverage depends on its maturity structure and scale.

6. Conclusions and Discussion

Overall, this study analyzes the leverage–performance nexus in Moroccan agricultural SMEs using ROA as the key performance metric. Based on panel data for 54 firms from 2017 to 2022, the results consistently show that both short- and long-term leverage reduce profitability. This finding holds across all estimation methods, including fixed effects models, robust specifications, and the dynamic GMM estimator, thereby reinforcing the conclusion that higher debt levels—regardless of maturity—tend to weaken operating profitability in this sector.

Further analysis of non-linear effects reveals that while STLEV maintains a consistently negative impact in both its linear and quadratic forms, LTLEV follows a U-shaped pattern: its linear coefficient is negative, but the squared term is positive and statistically significant. This suggests that, beyond a certain threshold, long-term debt may begin to support performance—possibly by enabling larger-scale investments or alleviating reliance on short-term liabilities. One plausible explanation is that firms reaching higher levels of long-term financing are typically more mature or better structured, allowing them to negotiate more favorable credit terms, allocate debt more efficiently, and use it to fund productivity-enhancing investments. In contrast, firms with limited access to long-term credit may be constrained by small-scale borrowing, which generates financial pressure without meaningful performance gains.

The negative impact of leverage in the context of Moroccan agricultural SMEs can be attributed to several structural and contextual factors. On the one hand, these enterprises often face unfavorable financing conditions, characterized by high borrowing costs, strict collateral requirements, and restricted access to long-term funding (T. Beck & Demirguc-Kunt, 2006; Stein et al., 2013). This results in increased pressure on cash flow and higher financial expenses, directly affecting profitability. On the other hand, weak governance structures—often marked by informal or family-based management—intensify agency costs associated with debt financing (Jensen & Meckling, 1976), thereby diminishing the potential benefits of financial leverage.

In the agricultural sector, these constraints are further exacerbated by high exposure to climate-related risks, seasonal income fluctuations, and price volatility in commodity markets (FAO, 2016; OECD, 2020). The uncertainty of cash flows heightens the risk of default, especially in developing countries where risk mitigation mechanisms remain underdeveloped. Moreover, high debt levels can lead to credit rationing, as lenders become increasingly reluctant to finance already leveraged firms (Stiglitz & Weiss, 1981), limiting their opportunities for investment, innovation, and modernization. Additionally, financial management practices within agricultural SMEs tend to be poorly formalized. These firms often operate with rudimentary budgeting processes, limited knowledge of debt management tools, and a short-term focus in financial decision making (Berger & Udell, 1998). Under such conditions, debt is frequently misused, and instead of enhancing productive capacity, it increases financial vulnerability. Finally, in emerging economies, including Morocco, agricultural credit markets remain underdeveloped and segmented, restricting financing alternatives for small farms (World Bank Group, 2019). This situation reinforces dependence on traditional bank loans, which are often ill-suited to the specific needs of the agricultural sector.

The study’s findings yield multiple theoretical and practical implications. From a theoretical perspective, they support the relevance of the pecking order theory in the context of agricultural SMEs in emerging markets, where internal financing remains the preferred option due to persistent information asymmetries and high borrowing costs. They also provide partial support for agency theory, as weak governance structures tend to amplify the cost-related disadvantages of debt. From a managerial perspective, the results suggest that modest, poorly structured long-term borrowing can harm performance, whereas larger, well-negotiated long-term credit facilities—combined with disciplined investment planning—can generate positive returns. SME owners should therefore improve financial records, strengthen governance practices, and adopt long-term strategic planning to access more favorable financing conditions. From a policy perspective, the study underscores the need for tailored long-term financing products for agricultural SMEs, the expansion of credit guarantee schemes to reduce collateral constraints, and the promotion of rural financial literacy programs to ensure debt is allocated to productive investments rather than short-term consumption smoothing. From a practical standpoint, agricultural SME managers are advised to (i) prioritize internal financing where possible to limit debt-servicing pressure; (ii) avoid maturity mismatches by refraining from financing long-term investments with short-term loans; (iii) build creditworthiness through transparent financial reporting to secure better loan terms; and (iv) diversify funding sources—such as cooperatives, microfinance, and supplier credit—to reduce dependence on traditional bank loans.

While the study makes meaningful contributions, certain limitations should be acknowledged. The dataset, consisting of 54 agricultural SMEs across six years, may not fully capture the heterogeneity of the sector, thereby limiting generalizability. Although the dataset was rigorously constructed from official financial statements obtained via the Directinfo platform, and robust panel data techniques such as fixed effects with robust standard errors and the dynamic GMM were employed to enhance estimation reliability, the sample is not intended to be statistically representative of the broader population of Moroccan agricultural SMEs. Accordingly, the findings should be interpreted with caution and considered exploratory in nature. Second, restricted access to detailed financial information from privately held agricultural SMEs constrained the range of financial indicators that could be included in the analysis, potentially introducing omitted variable bias.

Future research should aim to address these limitations by expanding the sample size, extending the observation period, and incorporating more granular data. In addition, examining the role of external factors—such as government policies, macroeconomic conditions, and access to financial markets—could provide a more comprehensive understanding of how leverage influences performance in Moroccan agricultural SMEs.