Youth Development Fund (YDF)

Youth Development Fund (YDF) is sponsored by Local Government Authorities (LGAs) in Tanzania. The Youth Development Fund is a decentralized financing mechanism administered through Local Government Authorities in Tanzania, including those in the Mara Region.

Frontiers | Youth development fund program inclusion and its effects in maximizing youth agri-enterprise return on investment in Tanzania: evidence from the Mara Region ORIGINAL RESEARCH article Front. Sustain. Food Syst.

, 05 August 2026 Sec. Agricultural and Food Economics Volume 10 - 2026 | https://doi. org/10.

3389/fsufs. 2026.

1863875 Youth development fund program inclusion and its effects in maximizing youth agri-enterprise return on investment in Tanzania: evidence from the Mara Region Department of Business, Economics, and Finance, Mwalimu Nyerere University of Agriculture and Technology, Butiama, Tanzania Youth participation in agribusiness enterprises is essential for improving income, food security, and sustainable agribusiness, but limited access to finance, institutional support, and market information hinders the performance of their agri-enterprises.

Thus, this study examined the effects of the Youth Development Fund (YDF) program inclusion on return on investment (ROI) among youth agri-entrepreneurs across all districts of the Mara Region, Tanzania.

Cross-sectional data were collected from 499 (comprising 256 YDF beneficiaries and 243 non-beneficiaries) randomly selected youth using structured questionnaires and analyzed using descriptive statistics and econometric models, including covariate regression, propensity score analysis, and the Heckman two-step method.

Descriptive results indicated that YDF beneficiaries had significantly higher ROI, larger loans, better repayment terms, timely disbursement, and greater financial literacy ( p < 0. 01). They also had improved access to extension services ( p < 0.

01), market information ( p < 0. 05), and cooperatives ( p < 0. 01).

Economic analysis confirms higher revenue, profit, and ROI across all enterprise types (crops, livestock, and mixed farming). Econometric results confirmed that YDF participation significantly increased ROI by 6. 5–7.

8% ( p < 0. 05). Other significant determinants of ROI included loan size ( p < 0.

01), favorable repayment terms, timely disbursement ( p < 0. 05), financial literacy, enterprise size and experience, access to extension services and market information, cooperative membership, and access to other finance, while greater distance to market negatively affected ROI at p < 0. 05.

The study concludes that YDF participation improves youth agri-enterprise performance and recommends expanding loan size, capacity building, timely loan disbursement, credit, and market access.

Globally, youth participation in agriculture is increasingly recognized as a pivotal driver of sustainable development, employment generation, and enhanced food security, particularly in the context of achieving the Sustainable Development Goals (SDGs).

Young people are seen as a source of innovation and entrepreneurship capable of transforming agrifood systems, yet they often face multiple barriers that limit their capacity to build profitable and sustainable enterprises.

These barriers include limited access to formal finance, inadequate technical and business skills, restricted access to markets and value chains, and insufficient access to productive resources such as land and technology, which collectively constrain the profitability and long-term sustainability of youth-led agribusinesses ( FAO, 2014 ; IFPRI, 2019 ).

In response to these global challenges, international institutions and development partners have introduced targeted frameworks and action plans designed to empower youth in agriculture.

For example, the World Food Forum’s Global Youth Action Plan (GYAP) outlines region-specific priorities and actions for youth engagement in agrifood systems, emphasizing capacity development, policy advocacy, and innovation to bridge regional gaps ( FAO, 2024 ).

Similarly, global youth-focused development initiatives such as the Solutions for Youth Employment (S4YE) network engage young agripreneurs in shaping policy dialogues and promoting youth-led agricultural solutions, underscoring the importance of multisectoral partnerships to address systemic barriers ( World Bank, 2020 ).

In Sub-Saharan Africa, where agriculture remains the dominant employer and a key driver of rural livelihoods, these challenges are intensified by high youth unemployment and underdeveloped agribusiness ecosystems.

To address this, regional frameworks like the African Union’s Comprehensive Africa Agriculture Development Programme (CAADP) and the Opportunities for Youth in Africa (OYA) program have been developed to mainstream youth in agriculture by promoting inclusive agribusiness development, access to finance, and skills training ( FAO UNIDO, 2020 ).

Continental strategies such as the Africa Agribusiness Youth Strategy (AAYS) further seek to harmonize youth agribusiness interventions and align them with broader socioeconomic goals under Agenda 2063 ( African Union Commission, 2022 ). In Tanzania, agriculture remains the backbone of the economy, providing employment, food security, and rural income to millions of households ( URT, 2016 ).

Youth engagement in agriculture is particularly critical given the country’s youthful population and the urgent need for productive employment opportunities. Yet, youth-led agri-enterprises frequently record low returns on investment (ROI) due to constrained access to capital, small-scale operations with limited adoption of technology, and weak financial management skills ( FAO, 2014 ; URT, 2016 ; Isaga, 2025 ).

To address these challenges, the Government of Tanzania introduced the Youth Development Fund (YDF), a decentralized financing mechanism administered through Local Government Authorities (LGAs).

The YDF allocates 4% of LGA own-source revenue to youth groups as low-interest loans intended to improve access to productive capital, strengthen enterprise capacity, and enhance the financial performance and return on investment (ROI) of youth-led enterprises ( URT, 2007 ). At the regional level, particularly in the Mara Region, youth engagement in agribusiness is high due to the area’s fertile land and agricultural potential.

However, enterprises often struggle to maximize returns because of delayed loan disbursements, insufficient loan amounts, inadequate business training, limited financial literacy, weak technical support, and restricted market access ( URT, 2007 ; FAO, 2014 ).

Despite growing evidence that youth financing improves enterprise productivity and profitability, there is limited empirical evidence on whether participation in the YDF program translates into higher return on investment (ROI), particularly in the Mara Region, and little is known about how loan characteristics and complementary support services such as financial literacy, technical assistance, and market access influence this relationship.

This study addresses these gaps by examining whether YDF program participation improves return on investment (ROI) among youth agri-enterprises and identifying the financial and institutional factors that influence enterprise returns across all districts of the Mara Region.

Specifically, the study aims to: determine the level of youth inclusion in the YDF Program; examine the effect of YDF loan access on enterprise ROI; and analyze how loan characteristics and moderating factors (size, repayment terms, timeliness, financial literacy, technical support, and market access) influence ROI. The study is significant to multiple stakeholders.

For government bodies and LGAs, the findings provide evidence to optimize fund allocation, streamline disbursement, and strengthen monitoring mechanisms ( URT, 2007 ). For financial institutions and NGOs, the study informs the design of youth-friendly financial products and capacity-building initiatives that integrate mentorship, digital literacy, and risk-sharing strategies ( Isaga, 2025 ).

For youth and community groups, it offers practical guidance on effective loan utilization, enterprise management, and strategic adoption of agribusiness innovations to increase ROI ( FAO, 2014 ). For academia and researchers, the study contributes novel empirical evidence on the interaction between youth financing and enterprise performance, bridging a key gap in literature on agribusiness productivity and ROI measurement ( Yami et al.

, 2019 ; Yeboah et al. , 2020 ). Empirical evidence from global studies shows that youth access to credit improves agricultural productivity, profitability, and enterprise performance.

Rural credit increases investment in improved inputs and technologies, enhancing productivity and income among smallholder farmers ( Sibhatu and Qaim, 2017 ). In Asia and Latin America, financial inclusion also boosts youth entrepreneurship and reduces income volatility, particularly when combined with capacity-building programs ( Aterido et al. , 2011 ; IFAD, 2019 ).

In Sub-Saharan Africa, youth access to credit promotes adoption of mechanization, improved seeds, and market participation, as observed in Kenya, Uganda, and Ethiopia ( Ali and Deininger, 2014 ; Yami et al. , 2019 ).

Studies in Uganda and Tanzania further show that youth financing enhances profitability and investment in productivity inputs ( Isaga, 2025 ), though effectiveness is often limited by small loan sizes, low financial literacy, delayed disbursements, and lack of support services ( FAO, 2014 ; URT, 2020 ).

While previous studies largely evaluate productivity, income, or enterprise performance, few directly assess return on investment (ROI) as the principal measure of enterprise success, particularly within Tanzania’s Youth Development Fund program. This study therefore evaluates the effect of YDF participation on ROI while examining the moderating roles of loan characteristics, financial literacy, technical support, and market access.

The novelty of this study lies in evaluating whether participation in the Youth Development Fund improves return on investment (ROI) among youth agri-enterprises and identifying the program features and complementary support services that strengthen this effect.

By simultaneously analyzing loan access, loan characteristics, financial literacy, technical support, and market access, the study provides evidence on the mechanisms through which YDF participation contributes to higher enterprise returns and offers practical guidance for improving youth financing programs. 2 Theoretical framework and conceptual framework 2.

1 Theoretical framework This study is anchored in the Financial Inclusion Theory ( Demirgüç-Kunt et al. , 2018 ) and the Youth Entrepreneurship Theory ( Fatoki, 2014 ). Financial Inclusion Theory argues that access to affordable financial services enables individuals to invest in productive activities, manage risks, and improve enterprise performance.

In this study, the theory predicts that youth entrepreneurs with access to YDF loans are expected to achieve higher ROI than non-beneficiaries because loan access facilitates investment in productive inputs and business expansion. Furthermore, the positive effect of loan access is expected to be stronger when supported by timely loan disbursement, adequate loan size, financial literacy, and technical assistance.

In the Mara Region, disparities in financial access, timely loan disbursement, and technical support influence the capacity of youth to invest in agricultural inputs, adopt improved technologies, and expand market operations. Youth Entrepreneurship Theory emphasizes that entrepreneurial success depends not only on financial capital but also on entrepreneurial competencies, institutional support, and market opportunities.

Accordingly, this study expects that financial literacy, technical support, and market access positively moderate the relationship between YDF participation and ROI by improving business decision-making, technology adoption, and market performance.

Based on these theoretical propositions, the study hypothesizes that: H 1 : Participation in the Youth Development Fund has a positive effect on the return on investment (ROI) of youth agri-enterprises. H 2 : Favorable loan characteristics (loan size, repayment terms, and timely disbursement) positively influence enterprise ROI.

H 3 : Financial literacy, technical support, and market access positively moderate the relationship between YDF participation and ROI. This study conceptualizes youth agri-enterprise return on investment (ROI) as the dependent variable, influenced by YDF program inclusion and interrelated factors ( Figure 1 ).

Inclusion in the YDF program provides financial capital necessary for investment in agricultural inputs, technology adoption, and expansion of market operations ( Yami et al. , 2019 ). Conversely, limited inclusion or delays in loan disbursement reduce investment capacity and ROI.

Moderating factors play a critical role in enhancing the effect of YDF inclusion on ROI. Financial literacy, including budgeting, record-keeping, and financial planning, improves effective loan utilization ( Fatoki, 2014 ). Technical support through extension services, farmers’ cooperative membership, and training enables efficient production and better market engagement ( FAO, 2014 ).

Market access, including proximity to buyers and participation in value chains, provides avenues for higher sales and profitability ( Isaga, 2025 ). Control variables such as age, education level, enterprise size, enterprise experience, and other sources of finance influence the degree to which youth can translate YDF access into higher ROI ( URT, 2020 ).

Institutional quality, including LGA efficiency in loan administration, and gender considerations, such as equitable access for women, further mediate program effectiveness.

The key outcomes of high agri-enterprise return on investment (ROI) include improved enterprise sustainability, increased employment opportunities, enhanced household income, greater access to credit, strengthened local agribusiness ecosystems, and higher value addition and market competitiveness.

Generally, the conceptual framework illustrates the pathway through which YDF inclusion enhances youth agri-enterprise ROI: YDF inclusion → investment capacity → production and market expansion → ROI → enterprise sustainability, moderated by financial literacy, technical support, and market access, and influenced by demographic and institutional factors. 3.

1 Description of the study area This study was conducted in all districts of Mara Region: Musoma Municipal, Musoma District, Bunda, Tarime, Butiama, Rorya, and Serengeti. The region lies in northern Tanzania between latitudes 1°00′–2°31′South and longitudes 33°10′–35°15′East, covering about 30,150 km 2 , of which 10,584 km 2 is water, mainly Lake Victoria ( URT, 2020 ).

Its diverse agro-ecological conditions provide favorable opportunities for agriculture, fisheries, livestock production, and youth agribusiness development. Mara Region was purposively selected because agriculture is the dominant economic activity, youth constitute approximately 31.

2% of the regional population ( NBS, 2022 ), and the Youth Development Fund (YDF) is implemented through all Local Government Authorities in accordance with national regulations ( URT, 2007 ).

In addition, the region offers diverse agricultural production systems and active youth engagement in crop farming, livestock, fisheries, and agribusiness value chains ( URT, 2020 , 2021 ), making it an appropriate setting for examining the relationship between YDF participation and return on investment (ROI).

The region experiences a tropical climate with bimodal rainfall, receiving long rains from February to June and short rains from September to December. Annual rainfall ranges from 1,200–2,000 mm in the Tarime highlands, 900–1,300 mm in Musoma, Butiama, and central Serengeti, and 700–900 mm in the Bunda lowlands ( URT, 2021 ). Temperatures range from 14 °C to 30 °C depending on altitude and proximity to Lake Victoria ( URT, 2021 ).

According to the 2022 Population and Housing Census, Mara Region has approximately 1. 96 million people, with youth (15–35 years) accounting for about 31. 2% of the population ( NBS, 2022 ) ( Figure 2 ).

3. 2 Sampling procedures and sample size A multistage sampling procedure was employed to obtain a representative sample of youth agri-enterprises across Mara Region. In the first stage, all seven Local Government Authorities (Musoma Municipal, Musoma District, Bunda, Butiama, Tarime, Rorya, and Serengeti) were purposively included because they implement the Youth Development Fund (YDF).

In the second stage, wards were selected using simple random sampling, followed by random selection of villages within the sampled wards. Within the selected villages, lists of eligible youth agri-enterprises were compiled with the assistance of Local Government Authorities and YDF program records, from which respondents were selected using proportionate simple random sampling.

Sample allocation across districts was proportional to the number of eligible youth agri-enterprises identified in the selected areas to ensure adequate representation. To determine the appropriate sample size, the Cochran (1977) formula for cross-sectional surveys was applied.

The formula was selected because a complete and reliable sampling frame of all eligible youth agri-enterprises (both YDF beneficiaries and non-beneficiaries) across the study area was not available at the regional level at the time of sampling.

Consequently, the total target population could not be determined with certainty, making Cochran’s formula appropriate for estimating the minimum sample size required for reliable statistical inference at a 95% confidence level and a 5% margin of error. The formula is expressed in Equation 1 : 𝑛 0 is the required sample size, 𝑍 is the Z-value at 95% confidence level (1. 96), 𝑝 is the estimated population proportion (set at 0.

5 to allow for maximum variability), and 𝑒 is the margin of error (0. 05). Substituting these values into Equation 1 gives the required sample size as presented in Equation 2 : To account for potential non-response, the calculated sample size was increased by 23%.

This adjustment was based on findings from a pilot survey involving 65 eligible respondents, which achieved a 77% response rate, equivalent to a 23% non-response rate. The observed non-response was primarily due to respondent unavailability, refusal to participate, and incomplete questionnaires.

Therefore, inflating the sample size by 23% was considered appropriate to ensure that the required effective sample size would be achieved during the main survey and to minimize the risk of non-response bias ( Dillman et al. , 2014 ). Assuming a non-response rate of 23%, the adjusted sample size was computed as presented in Equation 3 : where r = 0.

23 (assumed non-response rate of 23%). Therefore, a total of 499 households were surveyed. This ensured statistical reliability while capturing the diversity of youth agribusinesses across all districts of Mara Region (Musoma Municipal, Musoma District, Bunda, Tarime, Butiama, Rorya, and Serengeti), with samples proportionally distributed according to district population and enterprise density.

3. 3 Data collection methods Data collection was conducted over 15 months (July 2024–September 2025). Primary data were collected using a pre-tested structured questionnaire administered to youth entrepreneurs in the Mara Region.

The questionnaire comprised three sections: (i) socio-economic characteristics of youth entrepreneurs; (ii) Youth Development Fund (YDF) access and utilization; and (iii) moderating factors and enterprise outcomes. To ensure content validity, the questionnaire was reviewed by academic and sector experts for clarity, relevance, and alignment with the study objectives.

A pilot test was subsequently conducted with 90 respondents from non-sampled villages in Serengeti District. Feedback from the pilot study informed revisions to the wording, sequencing, and response options of the questionnaire, thereby improving its clarity, reliability, and internal consistency before the main survey.

Notably, since many youth agri-entrepreneurs operate informal enterprises with limited financial records, investment and cost data were collected using a structured questionnaire covering key components of enterprise expenditure, including fixed assets, inputs, labor, transportation, and other operational costs.

To minimize recall bias, respondents were asked to provide information for a defined production cycle, while enumerators used probing questions and relevant local events to support accurate recall. Reported values were cross-checked where possible using available records such as loan documents, receipts, cooperative records, and other financial information.

In addition, data cleaning procedures were applied to identify and verify extreme or inconsistent observations ( Bernard, 2017 ). 3. 4 Ethical consideration Ethical clearance and research authorization were granted by the Tanzania Commission for Science and Technology (COSTECH) under research permit No. 2023/24-653-ER-2013-061.

Additional administrative approval was obtained from the Local Government Authorities (LGAs) of Musoma Municipal, Musoma District, Bunda, Tarime, Butiama, and Serengeti Councils. Before data collection, all participants were informed about the purpose of the study, confidentiality measures, and their voluntary participation rights.

Written and verbal informed consent was obtained from each respondent, and no personal identifiers were collected. This study employed both descriptive statistics and econometric models to assess the effects of the Youth Development Fund (YDF) program inclusion on youth agri-enterprise Return on Investment (ROI) in the Mara Region, Tanzania. 3.

5. 1 Descriptive statistics Descriptive statistics were used to summarize socio-economic characteristics of Youth entrepreneurs, youth development fund program access and utilization, and moderating factors and enterprise outcomes. Measures such as means, percentages, and standard deviations were applied to present the distribution and variability of key variables across the study districts.

To examine differences between recipients and non-recipients, independent sample t-tests were performed for continuous variables, while Chi-square (χ 2 ) tests were employed for categorical variables. The resulting test statistics and p -values were used to determine whether the observed differences between the two groups were statistically significant.

Additionally, Return on Investment (ROI) was calculated and compared across the three sectors under investigation. This comparison was undertaken to assess sectoral variations in investment performance and to provide a broader understanding of the distribution of returns among the sectors studied.

The study uses a combination of three econometric techniques to assess the effects of the Youth Development Fund (YDF) program inclusion on youth agri-enterprise Return on Investment (ROI). The ROI formula follows standard financial performance measurement approaches used in enterprise and agricultural investment analysis ( Gittinger, 1982 ; FAO, 2014 ).

Return on Investment (ROI) is computed as presented in Equation 4 : Net Profit = Total Revenue − Total Costs. Total Investment = All input, operating, and fixed costs associated with running the enterprise. First, a simple regression, referred to as regression on covariates, is used.

Second, regression on propensity scores accounts for selectivity bias in estimating the effects of YDF participation. Third, the Heckman treatment effect model complements the results to test robustness. The outcome variable, ROI, is modeled as a linear function of YDF participation, enterprise characteristics, and household factors.

The simplest approach is to include a YDF participation dummy variable (1 = beneficiary, 0 = non-beneficiary) in an ordinary least squares (OLS) regression, as specified in Equation 5 : Where ROI i represents the ROI of enterprise i , YDF i is the participation dummy, X i is a vector of control variables (age, education, enterprise size, enterprise experience, other sources of finance), α i and δ i are parameters to estimate, and ϵ i is the error term.

OLS may yield biased estimates if YDF participation is endogenous or non-random, as youth enterprises may self-select into the program based on unobserved characteristics such as managerial ability, motivation, or entrepreneurial skill ( Greene, 2003 ; Maddala, 1983 ).

To address this, the first approach incorporates observable covariates ( Z i ) to account for selection on observables, as specified in Equation 6 : Covariates include distance to markets, enterprise size, access to extension services, and membership in a farmers’ cooperative.

As for the determinants of YDF participation and the estimation of propensity scores: Before estimating the impact of YDF participation on enterprise Return on Investment (ROI), a Probit regression model was estimated to identify factors influencing participation in the Youth Development Fund (YDF) program and to generate propensity scores for subsequent impact estimation.

Since participation in the YDF program is a binary outcome (participant = 1, non-participant = 0), the Probit model was considered appropriate ( Greene, 2003 ).

The latent participation model is specified in Equation 7 : YDF i ^* is the latent variable representing the propensity to participate in the YDF program, X ki is a vector of explanatory variables influencing participation, β k are parameters to be estimated, and ε i is the error term.

The observed participation decision is expressed in Equation 8 : The probability of participating in the YDF program is estimated as presented in Equation 9 : where Φ(·) denotes the cumulative distribution function of the standard normal distribution.

While Xiβ represents the linear index of explanatory variables, which include the age of enterprise owner, education level; enterprise experience; size of enterprise; cooperative membership; access to extension services; access to alternative finance; and distance to market.

The estimated propensity score is given by Equation 10 : PSi = propensity score for enterprise i , representing the conditional probability of participating in YDF given observed characteristics. The Probit model served two purposes. First, it identified the determinants of participation in the YDF programme.

Second, the predicted probabilities obtained from the model were used as propensity scores in the subsequent impact estimation models to control for selection bias arising from observable characteristics. The significance of the participation model was assessed using the Likelihood Ratio (LR) Chi-square statistic, while individual coefficients were evaluated using z-statistics and associated p -values.

The second approach utilizes the propensity scores (PSi) generated from the Probit participation model described above. The propensity score represents the conditional probability of participating in the YDF programme given a set of observed characteristics. Following Alemu et al.

(2016) and Asres et al. (2013) , the estimated propensity scores were incorporated into the ROI regression model to account for selection bias arising from observable factors. The model is specified in Equation 11 : Where ϕ i is the coefficient of the propensity score.

This reduces bias from selection on observable characteristics ( Imbens, 2004 ). Notably, although propensity score matching was used to reduce observable selection bias between YDF beneficiaries and non-beneficiaries, formal post-matching balance diagnostics were not computed.

This limitation should be considered when interpreting the propensity score estimates; however, the consistent YDF treatment effects across the covariate regression, propensity score regression, and Heckman models support the robustness of the findings. The third approach employs the Heckman treatment effect model to correct for selection on unobservable factors.

The Heckman treatment effect model is one of the most widely used procedures to account for sample selection bias and offers a mean/way of correcting for biases that may arise from unobservable factors, and thus results in unbiased and consistent estimates ( Greene, 2003 ). The Heckman treatment effect model is an extension of the Heckman two-stage model.

he only difference is that the dependent variable in the selection equation becomes one of the explanatory variables in the outcome equation of the former but not in the latter model. The selection equation, usually a probit, predicts participation as a selection control factor called the inverse Mills ratio (IMR): The model can be specified in two steps.

The selection equation, which is usually a probit model, is specified in Equation 12 : Where YDF i * is a latent variable representing YDF participation, and X i is a vector of observed explanatory variables influencing programme participation. The same set of observed covariates was included in both the selection and outcome equations because no theoretically defensible exclusion restriction was available in the dataset.

Consequently, identification relies on the functional-form (nonlinearity) of the inverse Mills ratio under the assumption of jointly normally distributed error terms ( Heckman, 1979 ; Greene, 2003 ).

Although this approach provides weaker identification than models incorporating a valid exclusion restriction, the Heckman treatment effect model was estimated as a robustness check alongside Regression on Covariates and Propensity Score Matching. The consistency of the estimated treatment effects across these complementary approaches strengthens confidence in the robustness of the findings.

The substantive equation can be specified in Equation 13 : The outcome equation incorporates the inverse Mills ratio (IMR, λ ) generated from the selection equation: The formulation of the inverse Mills ratio (IMR) is presented in Equation 14 : where φ and Φ are the normal probability density function and cumulative density function, respectively, of the standard normal distribution.

Adding the IMR to Equation 9 translates into Equation 15 as: Where λ i is the inverse Mills Ratio from the selection equation, γ i is the coefficient of the IMR, and μ i is a two-sided error term with N (0, σ 2 v ). A significant γ i indicates self-selection; a non-significant value suggests no selection bias. This approach ensures that the estimated effect α i of YDF inclusion on ROI is unbiased ( Greene, 2003 ; Bushway et al.

, 2007 ). Ignoring the addition of the IMR will render the results from Eq. (8) as biased ( Heckman, 1979 ).

Thus, the inclusion of the selectivity term makes the coefficient αi (measuring the effects of the treatment variable on the outcome variables) unbiased, albeit it is inefficient as the disturbance term ( μ i ) is heteroscedastic ( Greene, 2003 ). The problem of heteroscedasticity can be corrected by the use of bootstrap standard errors or resampling.

However, the STATA software package used in generating the estimates automatically adjusts for that bias in the standard errors ( Bushway et al. , 2007 ).

Generally, in this study, the Heckman treatment effect model was employed as a robustness approach to examine whether the estimated effect of YDF participation on enterprise ROI remains consistent after controlling for potential selection bias arising from unobservable characteristics.

The coefficient of the YDF participation variable ( α i ) represents the estimated effect of program participation on ROI after accounting for the selection correction term (inverse Mills ratio). A positive and statistically significant coefficient indicates that YDF participation is associated with higher enterprise ROI, even after controlling for possible non-random selection into the program.

Table 1 below shows the hypothesized sign effects of the independent variables. Variables Measurement Hypothesized sign effect Literature basis YDF program inclusion 1 = Beneficiary, 0 = non-beneficiary + Adeyanju et al. (2023a) , Adeyanju et al.

(2023b) and Ojo et al. (2022) Loan size Amount in TZS + Akudugu et al. (2019) and Balana et al.

(2022) Favorable loan repayment terms 1 = Yes, 0 = No + Abate et al. (2016) and Balana et al. (2022) Timely loan disbursement 1 = Yes, 0 = No + Abate et al.

(2016) and Akudugu et al. (2019) Financial literacy Index score (1–5 Likert scale) + Fanta and Mutsonziwa (2021) and Fatoki (2014) Access to extension services Number of trainings and visits attended + Asres et al. (2013) and Wossen et al.

(2017) Access to market information 1 = Yes, 0 = No + Haile et al. (2019) and Aker (2010) Distance to market Kilometers – Alemu et al. (2016) and Mmbando et al.

(2015) Membership in farmers’ cooperatives 1 = Yes, 0 = No + Wossen et al. (2017) and Yeboah et al. (2020) Access to other sources of finance 1 = Yes, 0 = No + Demirgüç-Kunt et al.

(2018) and Osei-Assibey (2013) Age of youth entrepreneur Years ± Hlatshwayo et al. (2022) and Yeboah et al. (2020) Education level Years of schooling + Adeyanju et al.

(2023a) , Adeyanju et al. (2023b) , and Yami et al. (2019) Enterprise size Number of Assets + FAO (2014) and Osabohien et al.

(2021) Enterprise experience Years in operation + Akudugu et al. (2019) and Balana et al. (2022) Gender of youth 1 = Female, 0 = Male ± Alemu et al.

(2016) and Mmbando et al. (2015) Independent variables and their expected effects on Youth Agri-Enterprise return on investment (ROI). 3.

6 Multicollinearity diagnostics Variance Inflation Factors (VIFs) were computed to assess multicollinearity. All VIF values were below 10, with a mean VIF of 2. 26, indicating no multicollinearity concern ( Gujarati and Porter, 2009 ; Wooldridge, 2016 ) ( Table 2 ).

Variable VIF Tolerance (1/VIF) YDF program inclusion 2. 18 0. 46 Loan size 2.

73 0. 37 Favorable loan repayment terms 1. 95 0.

51 Timely loan disbursement 2. 04 0. 49 Financial literacy 2.

58 0. 39 Access to extension services 2. 32 0.

43 Access to market information 1. 79 0. 56 Distance to market 1.

61 0. 62 Cooperative membership 1. 83 0.

55 Access to other finance 1. 69 0. 59 Age of youth 1.

57 0. 64 Education level 2. 06 0.

49 Enterprise Size 2. 41 0. 41 Enterprise experience 1.

72 0. 58 Gender of youth 1. 29 0.

77 Mean VIF 2. 26 Variance inflation factors (VIFs). Source: Field Survey Data (2024/2025).

Pairwise correlation coefficients were computed to further assess multicollinearity ( Table 3 ). All coefficients were below the threshold of 0. 70, indicating no severe correlations ( Gujarati and Porter, 2009