The paper uses the ARDL bounds testing cointegration method suggested by (Banerjee et al., 1993) to explore the dynamics between industrialization, economic growth, agricultural land and food security in The Gambia. The ARDL method is found to be more appropriate for most of the classical cointegration methods in which the regression variables are mostly integrated of order I(0) and I(1). It works with small samples and at the same time estimates relationships of long-run equilibrium as well as short-run adjustment dynamics in a single reduced-form equation without the need for pre-tests for integration order as in (Banerjee et al., 1993) and (Kripfganz & Schneider, 2020).

We utilized annual time series data for the Gambia within a 54-year period spanning between 1970 and 2023. All variables were extracted from the World Bank's World Development Indicators (WDI) database that ensures international comparability and reliability of data. The variables are presented in detail with their operationalization in Table 1 below.

In empirical food security studies for developing economies, the Food Production Index (FPI) is commonly used as an indicator of food security ((Mohamed et al., 2022); (Pawlak & Kołodziejczak, 2020)). It shows the overall agricultural production compared with the 2014-2016 base period. The explanatory variables are gross domestic product per capita (in current US dollars) as a measure of economic growth. GDP per capita is a standard operational measure of economic growth in the food-security literature because it captures the income channel through which growth is theorized to affect food access and investment ((Bogmans et al., 2025); (Diao et al., 2019); (Ntiamoah et al., 2023)), and it is the measure used in the closest comparable country-specific ARDL studies ((Cao et al., 2026); (Mohamed et al., 2022)). We adopt Industry value added (as a percentage of GDP) as an indicator of industrialization, while agricultural land (in square kilometres) as a measure of land available for food production. Logarithm transformations are applied to all variables to ensure constant variance and ease of interpretation of elasticity.

The empirical model is grounded in three complementary theoretical strands, each motivating one of the explanatory variables and the expected sign of its long-run coefficient. Structural transformation theory (Lewis, 1954) predicts that industrialization reallocates labour, capital and land from low-productivity agriculture toward higher-productivity industrial activity. This realignment often enhances growth but can lead to food production tensions if the loss of land and labour is not matched by production (Wang, 2022). This leads to the assumption that industrialization has a negative long-run relationship with food production in a small or resource-constrained economy like The Gambia unless it is accompanied by robust agro-industrial linkages (Belton et al., 2021).

Moreover, economic growth theory (Ravallion, 2016) predicts that rising income raises food consumption and finances investment in agricultural technology. GDP per capita in this regard should be positively associated with food production provided growth is not driven exclusively by non-agricultural sectors (Diao et al., 2019)(Neglo et al., 2021). Third, in the classical production theory (Foss, 1997), land is considered as a primary production factor, suggesting that the land area of agriculture activity should be positively correlated with output in the economies where mechanization is limited. However, this relationship is dependent on land quality and use; where land expansion occurs on marginal or degraded land, or where competing urban and industrial pressures drive land expansion, the theoretically anticipated positive relationship could be negative in reality ((Jayne et al., 2014); (Kocur-Bera & Pszenny, 2020)).

Consistent with the theoretical framework in Section 2.3, food production is assumed to depend on economic growth, industrialization, and agricultural land. The theoretical relationship can be expressed as:

where denotes food production, represents economic growth, measures industrialization, and captures agricultural land availability.

To obtain an estimable empirical model, the functional relationship is specified in log-linear form as:

where is the intercept, are the long-run elasticities, and is the stochastic error term.

Given the possibility that the variables are integrated of mixed orders, the Autoregressive Distributed Lag (ARDL) framework developed by Pesaran et al. (2001) is employed. The unrestricted ARDL specification is expressed as:

where and denote the optimal lag lengths orders of , , and respectively selected using the Akaike Information Criterion (AIC).

Stationarity is examined using the Augmented Dickey–Fuller (ADF) and Phillips–Perron (PP) tests (Pesaran et al., 2001). The existence of a long-run relationship is assessed via the (Pesaran et al., 2001) bounds test; finite-sample critical values follow (Kripfganz & Schneider, 2020). Following Pesaran et al. (2001), we re-parameterized the ARDL model into an unrestricted Error Correction Model (ECM) to separate short-run dynamics from the speed of adjustment toward the long-run equilibrium defined above:

We have;

where is the lagged error-correction term and measures the adjustment speed toward the long-run equilibrium. A negative and significant confirms long-run stability.

Model robustness is validated through the Breusch-Godfrey LM test for serial correlation, the Breusch-Pagan test for heteroskedasticity, and the Skewness-Kurtosis test for normality assessment. Parameter stability is assessed using CUSUM and CUSUM of Squares tests (Brown et al., 1975).

Table 2 shows descriptive statistics for all variables used in our study. The food production index (lnFPI) shows moderate variation during the period of study, indicating economic and structural vulnerability. Industrialization (lnINDU) has a relatively low mean value, reflecting the relatively low speed of structural change in the industrial sector in The Gambia. The standard deviation for GDP per capita (lnGDP) is the highest, suggesting high income volatility over time. Agricultural land (lnAGL) displays minimal variation, confirming that land expansion has been largely constrained throughout the sample period.

The pairwise correlation coefficients are summarized in Table 3. Food production index is negatively correlated with industrialization and agricultural land but positively correlated with GDP per capita. The positive correlation between the industrialization and GDP per capita indicates the close relationship between industrial development and income in The Gambia. No pairwise correlation exceeds the conventional threshold for multicollinearity which indicates that the regressors are not problematically collinear.

Figure 1 illustrates the time-series trends of the logged variables. The food production index (lnFPI) alternates between periods of stagnation and mild growth and this reflects the vulnerability of food production to economic and structural shocks. Industrialization (lnINDU) exhibits a gradual upward trend while GDP per capita (lnGDP) displays long-term growth with intermittent fluctuations. Agricultural land (lnAGL) remains largely flat throughout the sample which is consistent with constrained land expansion in The Gambia. These trends collectively confirm the need for econometric modelling to uncover both short-run and long-run dynamics.

Table 4 presents the ADF and PP unit root test results. All the variables at levels fail to reject the null hypothesis of a unit root requirement. The null hypothesis is rejected at the 1% significance level for all series following first differencing, confirming that all variables are integrated of order I(1) and satisfying the preconditions for ARDL estimation.

Table 5 reports the Johansen cointegration test results required in this study. The Trace and Maximum Eigenvalue statistics fall below their respective 5% critical values at all ranks which indicate the absence of multiple cointegrating vectors under the Johansen framework. This outcome is common in small-sample studies and does not invalidate the ARDL bounds test specifically designed for such conditions.

Table 6 presents the bounds test results. The computed F-statistic of 6.788 exceeds the upper-bound critical values at all conventional significance levels, confirming a stable long-run equilibrium relationship among food production, industrialization, economic growth, and agricultural land. The null hypothesis of no cointegration is therefore rejected.

Table 7 presents the ARDL (2,0,4,0) short and long-run estimates in panel A and B respectively. It shows that all of our explanatory variables except lag GDP have no significant effect on food security in The Gambia during the study period. There’s however a statistically significant relationship between all three variables on food security in the long-run. The results indicate that industrialization and Agricultural Land negatively affect food security but Economic growth positively influences it.

Table 8 presents the post-estimation diagnostic results. All tests confirm the statistical adequacy of the estimated model: no evidence of serial correlation, homoskedastic residuals, and normally distributed errors.

Table 9 decomposes the Cameron and Trivedi IM-test. All components such as heteroskedasticity, skewness, and kurtosis are statistically insignificant, confirming that the residuals satisfy classical regression assumptions and the model is correctly specified.

The CUSUM and CUSUMSQ test statistics remain within the 5% critical bounds throughout the sample period (Figure 2), confirming structural parameter stability and reinforcing the robustness of the estimated long-run relationships.

Industrialization has a negative and statistically significant impact on the production of food in the long term. An increment in the industrial value added by 1 percent lowers the food production index by about 1.12. This indicates that an increase in industrialization in The Gambia can be at the expense of agriculture in terms of land, labor and governmental investment, thus compromising food production. Our result is contrary to the structural transformation hypothesis ((Johnston & Mellor, 1961);(Lewis, 1954)) that predicts increased productivity in agriculture due to technological spillovers and infrastructure development in the industrial sector. The findings are consistent with those of (Zhang et al., 2020) and (McMillan & Zeufack, 2022), who show that industrial growth in developing countries is likely to be at the expense of agricultural development as industrial expansion drives a loss of fertile land.

Our findings also align with those of (Ungwa & Godwin, 2024) who document asymmetric effects of industrialization on food production in Nigeria, while (Gobezie & Boka, 2023) show that industrial-related environmental degradation adversely affects food security across sub-Saharan Africa. (Segbefia et al., 2023) also suggest that industrial carbon emissions contribute to food insecurity in African countries, but that human capital investment can mitigate this effect. The magnitude of our estimated coefficient of -1.121 is significantly higher than similar estimates reported for Indonesia (Setiawan et al., 2023), reflecting the intensity of resource displacement in The Gambia's comparatively small and less diversified economy. This is pertinent to The Gambia where the industrial sector is still narrow and focused on tourism, groundnut processing and construction, suggesting even a small expansion of the industries will put additional strain on scarce land and labour resources for food production. The Gambia's industrial development policies should therefore be reconsidered to encourage agriculture sector, especially agro-based and agro-processing industries that create backward and forward linkages with food production.

The long-run effect of GDP per capita on food production has a significant positive coefficient of 0.453, indicating that an increase in income strengthens the capacity of the population to attain self-food sufficiency. This outcome is in line with the overall empirical literature. In East African economies, (Ntiamoah et al., 2023) employed a panel cointegration approach and demonstrated a positive and significant relationship between economic growth and food security. Income growth reinforces sustainable long-run economic performance with implications for food system investment (Yusuf & Setiawan, 2022). Similarly, (Ghufran et al., 2024) conclude that economic growth has a positive impact on food security across OECD and sub-Saharan African countries in the post-Paris Agreement period.

The country-specific time-series analysis results for Sudan also indicate a positive long-run relationship between economic growth and food security (Mohamed et al., 2022), and the positive relationship between GDP performance and all four pillars of food security is report for Ghana ((Debrah, 2024); (Twumasi-Ankrah et al., 2020)). The elasticity derived here are however comparatively low compared to the values of these estimates, which may be indicative of the limited absorptive capacity of The Gambia in terms of translating increased incomes into long-term investments in agriculture. This limited absorptive capacity is consistent with The Gambia's structural conditions, including low rural infrastructure, constrained access to agricultural credit, and the dominance of subsistence farming, all of which dampen the transmission of income growth into productive agricultural investment.

Agricultural land has a negative but not very strong long-run effect (p = 0.053). This contrasts with previous studies reporting positive land-production relationships ((Bambi & Pea-Assounga, 2024); (Pawlak & Kołodziejczak, 2020)), and reflects The Gambia's unique structural constraints where land development has taken place mainly on marginal or degraded land, leading to a negative productivity-adjusted relationship with food production. (Sarwar et al., 2025) highlight that sustainable land and water management is essential for increasing the long-term effectiveness of agricultural land, while (Koryo-Dabrah et al., 2021) and (Wang, 2022) show that food production is hampered even when land is available due to insufficient land management and soil degradation. Agricultural land in The Gambia is largely rain-fed, highly fragmented and increasingly subject to salinity intrusion and soil degradation. Such conditions explain why additional land area has not translated into proportional gains in food output.

This Gambian evidence extends to (Setiawan et al., 2023) and (Li et al., 2024) who show that agricultural land expansion under industrialization pressure does not necessarily increase food production. (Rahim et al., 2023) further demonstrate this evidence using Indonesian border province data and show that agriculture, infrastructure, and investment programmes jointly determine poverty and human development outcomes, underscoring the need for integrated land and agricultural policy. In line with (Jayne et al., 2014) and (Zhou et al., 2021), extensification alone is not sufficient to enhance food security in sub-Saharan Africa. Land policies should therefore focus on making land productive through effective and sustainable use practices, not on simply increasing land size. Greater institutional coordination is also needed between industrial, agricultural and land-use policy frameworks (Rahim et al., 2023).

Short-run results show that most variables in our study do not have a significant short-run impact on food security in The Gambia. In the short run, lagged changes in GDP per capita are negatively correlated with food production, implying transitional adjustment costs in response to macroeconomic change. The error correction term (λ = -0.562) is negative and statistically significant at 1 per cent, indicating a rapid adjustment speed of about 56 percent towards the long-run equilibrium following short-run shocks. This implies that food production in The Gambia reacts to disequilibrium brought about by industrialization, income, or land use changes relatively fast. The consistency of the estimated coefficients as confirmed by the CUSUM and CUSUMSQ tests enhances the plausibility of the long-run associations identified in this research In terms of relative magnitude show that industrialization exerts the largest effect on food production (-1.121), followed by agricultural land (-0.879) and GDP per capita (coefficient = 0.453). This ordering has direct policy implications in an way that industrialization management constitutes the most urgent priority for food security in The Gambia followed by land productivity improvements and income-enhancing policies.