The Impact of Agricultural Value Chains on Sustainable Industrial Development in Sub-Saharan Africa

The panel unit root test employed in this study is a well-known method introduced by Levin et al. (2002). Their approach involves an ADF test within a panel framework, where the parameters βi are constrained to be the same across different cross sections.

The study adopted the Johansen Fisher Panel cointegration test for its relative advantage in comparing two different approaches. Johansen (1988) introduces two distinct methods for detecting cointegration vectors in non-stationary time series and panel data: the likelihood ratio trace statistics and the maximum eigenvalue statistics.

The trace statistics and maximum eigenvalue statistics are presented in equations (iii) and (iv) respectively. (iii) λtracer=−TΣi=r+1nln1−λi {\lambda _{{\rm{trace}}}}\left( r \right) = - T\Sigma _{i = r + 1}^n{l_n}\left( {1 - {\lambda _{\rm{i}}}} \right) (iv) λmaxr,r+1=−Tln1−λr+1 {\lambda _{\max }}\left( {r,r + 1} \right) = - T{l_n}\left( {1 - {\lambda _{r + 1}}} \right) Where:

• T is the sample size,

• n = 10 variables (IND, AGV, FRT, MCH, LAND, EXP, LNDR, ARME, TRF and ALCT)

• λ_i is the i^th largest canonical correlation between residuals from the ten-dimensional processes and residuals from the derivatives of the ten-dimensional processes.

The trace tests the null hypothesis of at most “r” cointegration vectors against the alternative hypothesis of the full rank r = n cointegration vector. The null and alternative hypotheses of the maximum eigenvalue statistics are intended to check the r cointegrating vectors against the alternative hypothesis of r + 1 cointegrating vectors.

The study explores panel least square (fixed effect and random effect model) methods to analyze the impact of agricultural value chains on industrial development in Sub-Saharan Africa, following the methodology employed by Gujarati (2003), Atif et al. (2011), and Osinowo and Sanusi (2018). These models consider the impact of endogenous factors such as agricultural value added (AGV), fertilizer (FRT), farm machinery (MCH), arable land (LAND), government agricultural expenditure (GEXP), lending rate (LNDR), agricultural raw materials exports (ARME), human capital (HCAP) and access to electricity (ELCT) on industry value added (IND).

By introducing other endogenous factors into the Cobb-Douglass production function in line with Brownson et al. (2012), as recently adopted by Osinowo and Sanusi (2018), the industry growth equation was specified as follows: (v) InINDt=α0+α1InAGVt+α2InFRTt+α3InMCHt++ α4InLANDt+α5InGEXPt+α6InLNDt++ α7InAGEXt+α8InHCAPt+α9InELCTt+Ut \matrix{ {{I_n}{IND_t} = {\alpha _0} + {\alpha _1}{I_n}{AGV_t} + {\alpha _2}{I_n}{FRT_t} + {\alpha _3}{I_n}{MCH_t}\, + } \cr { + \;{\alpha _4}{I_n}{LAND_t} + {\alpha _5}{I_n}{GEXP_t} + {\alpha _6}{I_n}{LND_t} \,+ } \cr { + \;{\alpha _7}{I_n}{AGEX_t} + {\alpha _8}{I_n}{HCAP_t} + {\alpha _9}{I_n}{ELCT_t} + {U_t}} \cr } Where:

• IND_t = industry growth

• AGV_t = agricultural value added

• FRT_t = fertilizer usage

• MCH_t = farm machinery

• LAND_t = arable land

• GEXP_t = government agricultural expenditure

• LND_t = lending rate

• AGRE_t = agricultural raw materials exports

• HCAP_t = human capital

• ELCT_t = access to electricity

• U_t = error term; all in time t (between-country error)

• t = 1990 to 2022.

Equation (v) above is the fixed-effects panel data estimation of the model for this study. Data for each country on the ten variables mentioned above was taken for the period 1990–2022. In total, there were thirty three (33) time periods with thirty eight (38) cross-sectional units. In all, there were 1,254 observations for this study.

A fixed effects (FE) analysis investigates how predictor and outcome variables relate within an entity. Each entity possesses unique characteristics that could influence the predictor variables (Bartel, 2008). FE assumes that internal aspects of an individual entity might affect or skew the predictor and outcome variables, necessitating control. This underpins the assumption of correlation between an entity’s error term and the predictor variables. FE eliminates the impact of time-invariant characteristics, enabling the evaluation of predictors’ net effect on the outcome variable (Osinowo and Sanusi, 2008).

The FE model assumes unique, uncorrelated time-invariant traits for each entity, essential for accurate results. If the variables are correlated, FE might not be suitable, prompting a Hausman test and potential use of random-effects modeling. The fixed-effects model rectifies biases caused by omitted time-invariant traits (Gujarati, 2003). In essence, fixed-effects models are structured to investigate the factors driving shifts within an entity. As features that are constant for each individual, time-invariant characteristics lack the capacity to bring about such changes (Osinowo and Sanusi, 2018).

The fundamental idea behind the random effects model is that, unlike the fixed effects model, the variability among entities is regarded as random and unrelated to the included predictor or independent variables. The key difference between fixed and random effects hinges on whether the unobserved individual effect contains components that correlate with the model’s regressors, not on the stochastic nature of these effects (Greene, 2005). If you suspect that variations among entities impact your dependent variable, opting for random effects is advisable. An advantage of this approach is the ability to incorporate time-invariant variables. In the fixed effects model, these variables are absorbed by the intercept. The random-effects model for equation (v) previously mentioned above was specified as follows: (vi) InINDt=α0+α1InAGVt+α2InFRTt+α3InMCHt++ α4InLANDt+α5InGEXPt+α6InLNDt++ α7InAGEXt+α8InHCAPt+α9InELCTt+et+Ut \matrix{ {{I_n}{IND_t} = {\alpha _0} + {\alpha _1}{I_n}{AGV_t} + {\alpha _2}{I_n}{FRT_t} + {\alpha _3}{I_n}{MCH_t} + } \cr { + \;{\alpha _4}{I_n}{LAND_t} + {\alpha _5}{I_n}{GEXP_t} + {\alpha _6}{I_n}{LND_t}\, + } \cr { + \;{\alpha _7}{I_n}{AGEX_t} + {\alpha _8}{I_n}{HCAP_t} + {\alpha _9}{I_n}{ELCT_t} + {e_t} + {U_t}} \cr }

Equation (vi) captures both within-country and between-country errors, differing from the fixed-effects model, which solely includes the between-country error. In this equation, U_it accounts for within-country error, while e_it represents between-country error.

The Hausman specification test examines whether the efficient random effects estimator’s coefficients match those from the consistent fixed effects estimator. The null hypothesis posits that the preferred model is random effects, while the alternative hypothesis suggests fixed effects (Greene, 2005).

Levin-Lin-Chu tests were applied to all panel variables to assess stationarity. As indicated in Table 2, the outcomes revealed that some variables are stationary at their levels, while others exhibit stationarity in their first differences.

Table 3 presents the outcome of the Johansen-Fisher Panel Cointegration test. Comparing Fisher trace and Fisher max-eigen tests, at most 8 variables exhibit a long-run relationship in both scenarios. In both scenarios, the Johansen-Fisher Panel Cointegration test indicated the rejection of the null hypothesis of no cointegration at a significance level of 5%. The highly significant p-values at the 1% significance level strongly support the existence of a long-run relationship among these variables.

Table 4 displays the results of both the fixed-effects and random-effects models. The interpretation of empirical findings primarily relies on the fixed-effects model due to the Hausman specification test result. This test rejected the null hypothesis, suggesting that the fixed-effects model is suitable, while the random-effects model is inconsistent.

Examining the panel least squares result, as presented in Table 4, the estimated coefficient of determination (R-squared) for fixed effects is 76.7%. This suggests that the model accounts for approximately 77% of the total variance in industrial sector output (IND) within the SSA. The strong fit of the model is confirmed by our findings. Additionally, the F-statistic outcome for fixed effects, along with its associated probability value, indicates the collective significance of these explanatory variables in explaining the fluctuations in industrial sector output (IND) within the SSA.

From the information presented in Table 4, it is evident that the coefficient of AGV exhibits a positive and statistically significant relationship with industrial sector output at the 1% significance level. The study establishes that a 1% increase in agricultural value added results in an approximately 0.0729 rise in industrial sector output (IND) within the SSA. This finding aligns with our a priori expectations and is consistent with the conclusions reached by Nyamekye et al. (2021), as well as Bakari and El Weriemmi (2022). These scholars posit that investments in the agricultural sector significantly influence economic growth in Ghana and France respectively. Therefore, this study emphasizes the importance of fostering agricultural value addition to drive substantial growth in the industrial sector.

The farm machinery coefficient (MCH) demonstrates a significant positive impact on industrial sector at the 5% level. This outcome is in agreement with our a priori expectation, and its implications are significant. It emphasizes that increased use of machinery has the capacity to significantly boost agricultural productivity within a given economic framework (Osinowo and Sanusi, 2018). The deployment of farm machinery helps reduce the need for extensive manual labor and encourages the uptake of large-scale farming methods, thereby augmenting the incremental yield for farmers. This, in turn, triggers a multiplier effect on the expansion of raw materials for the industrial sector.

The coefficient for agricultural raw materials exports (ARME) in the estimated fixed-effects model also exhibited statistical significance at the 1% level. This finding indicates that the export of agricultural raw materials has an adverse effect on the growth of the industrial sector. This aligns with the findings of a previous investigation by Waziri et al. (2017), which highlighted that the focus of the transformation agenda should extend beyond solely increasing the export of raw agricultural materials; it should also encompass the promotion of exports of processed agricultural products to attain more comprehensive economic growth and development. The research conducted by Gilbert et al. (2013) arrived at a similar conclusion, advocating for the addition of extra value to cocoa and coffee beans before they were exported. The implementation of this approach has the potential to achieve a higher rate of economic growth, as demonstrated in the case of Cameroon.

The coefficient associated with human capital (HCAP) displayed a positive relationship with industrial sector output, which was significant at the 1 percent level. A one percent increase in the HCAP level was associated with a 0.1082 percent rise in industrial sector output. This finding aligns with and strengthens the conclusions drawn by previous researchers such as Sabir and Ahmed (2008), and Khalil and Anthony (2012). The result supports the concepts advanced within the realm of endogenous growth theory, a theoretical framework that posits that sustained economic growth is driven by internal factors such as technological progress, innovation, and investment in human capital. The concept of endogenous growth emphasizes that growth is not solely reliant on external factors, but is rather influenced by deliberate efforts to improve productivity through internal mechanisms. In particular, the increased investment in human capital, encompassing education, skills, and knowledge, plays a pivotal role in enhancing the workforce’s ability to adopt and adapt to new technologies, thereby increasing productivity (Osinowo et al., 2021).

Finally, this study revealed that the coefficient associated with electricity (ELCT) demonstrated a positive relationship with industrial sector output, which was significant at the 1% level. This finding is in line with our initial expectations, as it aligns with the anticipated outcome that improved access to electricity is likely to augment the productivity of the industrial sector, thereby fostering an increase in overall output. Access to a stable and consistent supply of electricity is a crucial factor that can significantly impact industrial operations. When factories, manufacturing units, and industrial facilities have access to reliable electricity, it enables them to operate efficiently, utilize machinery and equipment optimally, and maintain consistent production schedules. This, in turn, contributes to higher productivity levels and an increased capacity for the production of goods and services. The result is also consistent with the insights derived from Signe (2018). Improved access to electricity directly contributes to the total value generated within the industrial sector. This underscores the pivotal role of electricity as a facilitator of industrial growth and reinforces the notion that a reliable energy supply has the potential to act as a catalyst for economic development (Signe, 2018).

In conclusion, our study reveals key factors influencing industrial sector growth. Agricultural value addition (AGV) and farm machinery (MCH) are recognized as essential for industrial growth. The study emphasizes the significance of investing in human capital (HCAP), underlining the importance of education and skills for industrial development. Furthermore, it highlights the critical role of a reliable electricity (ELCT) supply in improving industrial operations and overall productivity. However, the negative impact of exporting raw materials underscores the importance of not only emphasizing raw material exports but also promoting processed agricultural products. This approach, as recommended by previous studies, holds the potential to contribute to a more inclusive and comprehensive economic growth strategy.

Building upon these findings, we propose the following policy recommendations.

• Enhancing agricultural value addition: Policymakers should focus on initiatives that encourage value addition in the agricultural sector. This can involve supporting agro-processing industries, improving post-harvest handling, and promoting research and innovation to increase the value of agricultural products.

• Promoting technological advancements: Encouraging the adoption of advanced farm machinery and technology in agriculture can lead to higher productivity. Government support in the form of subsidies, training, and access to technology can facilitate this transition.

• Diversification of agricultural exports: To mitigate the adverse impact of raw agricultural material exports, policies should prioritize the promotion of processed agricultural products in international markets. This can add value and contribute to more balanced industrial sector growth.

• Investment in human capital: Governments should invest in education and skills development programs to enhance human capital. A skilled workforce is crucial for the effective utilization of new technologies and to drive industrial growth.

• Ensuring a reliable electricity supply: Policies should prioritize investments in energy infrastructure to ensure a consistent and reliable supply of electricity. This will empower industries to operate efficiently and foster increased productivity.

By addressing these recommendations, policymakers can create an environment conducive to sustainable industrial growth, technological progress, and economic prosperity.