


































33-Manuscript-293-2-11-20200319


Olabisi et al.  Advancements in Agricultural Development 
  Volume 1, Issue 2, 2020 
  agdevresearch.org 

1. Laura Schmitt Olabisi, Associate Professor, Michigan State University 
480 Wilson Road, Room 138, East Lansing, MI 48824 
schmi420@msu.edu, https://orcid.org/0000-0002-6557-9469 

2. Obafemi Elegbede, Research Associate, Michigan State University 
elegbede@msu.edu, https://orcid.org/0000-0002-5935-0991 

3. Matt R. Raven, Professor, Michigan State University 
480 Wilson Road, Room 310A, East Lansing, MI 48824 
mraven@msu.edu, https://orcid.org/0000-0002-8271-2321 

1 

Insights for Farmer Training Programs from System Dynamics: 
A Case Study from Northern Michigan 

 
L. S. Olabisi1, O. Elegbede2, M. R. Raven3 

 
 

  

Abstract 
Training programs for new farmers have been proposed as a solution to rural food insecurity, rural 
development, and the recruitment and training of younger farmers simultaneously. However, 
evaluation of these programs and evidence for their individual or collective impact is sparse. In this 
paper, we use in-depth interviews combined with a system dynamics model to evaluate the current 
and potential effectiveness of a farmer training program in Michigan’s Upper Peninsula. We use the 
model to represent the theoretical progression of farmers through three subsequent stages of skill 
acquisition: training, new farmer (practicing skills on land owned by the program) and experienced 
(farming on their own). We find that recruitment, access to local markets, rapidity of skill acquisition, 
and access to start-up costs are all important factors that facilitate trainees’ transition to farming on 
their own, but of these, start-up costs for independent farming appear to be the most significant 
barrier. While this model is exploratory and not predictive, these insights can inform the design of 
effective programs for training farmers who will ultimately go on to start their own operations. In 
addition, this study also demonstrates how systems dynamics can be a valuable method to evaluate 
and maximize the effectiveness of training programs. 
 
Keywords 
Training, Systems modeling, evaluation, rural livelihoods, adoption 
 
 
 
 
 
 
 
 

Keywords 

Keyword, keyword (make sure they match what is in the system)  



Olabisi et al.  Advancements in Agricultural Development 
 

https://doi.org/10.37433/aad.v1i2.33  2 

Introduction and Problem Statement 
 
Food insecurity and a lack of affordable, nutritious foods are persistent problems in the rural 
United States (Denney et al., 2017; Ramirez et al., 2017). The farming population in the United 
States is aging, with one quarter of farm operators being 65 or older, compared with only 8% of 
the general population (United States. Congress. Senate. Special Committee on, 2008). Training 
programs for new farmers have been proposed as a solution to rural food insecurity, rural 
development, and the recruitment and training of younger farmers simultaneously. However, 
evaluation of these programs and evidence for their individual or collective impact is sparse. 
 
The Upper Peninsula of Michigan is a rural region that exemplifies the linked challenges of 
economic insecurity, an aging farmer population, and a supply for locally produced fruits and 
vegetables that does not meet consumer demand. According to a 2011 survey conducted by 
Michigan State University Extension (MSUE) on a sample of 108 Upper Peninsula (U.P.) farmers, 
only 8% of the respondents derive all of their income from their farms. For most, the farm 
generated up to 10% of the household income. Nearly three-quarters (70%) were over the age 
of 51 and three-quarters (75%) indicated they did not have someone to take over for them 
when they stop farming. Nearly half of the respondents (48%) indicated they wanted to expand 
their operations. Many producers indicated they needed additional help and technical support. 
The report pointed out that as interest in local food continues to increase so will the pressures 
on U.P. producers to keep pace with the demand. Furthermore, the report concluded that a 
vibrant local food economy was integral to the health of small rural communities (Lantz & Walk, 
2011). 
 
As a solution to these documented challenges, Michigan State University’s Upper Peninsula 
Research and Extension Center has developed formal and non-formal research-based 
educational programs targeting beginning farmers and avocational growers to increase the 
number of farmers in the U.P. growing nutrient dense food, thereby proposing to fill the gap 
between supply and demand for healthy local foods. The program provides access to 
organically managed land, equipment and hand tools, and shared infrastructure (hoop houses, 
transplant greenhouses, germination chamber, a tractor, etc.). In addition, the program 
provides subsidized on-site housing, access to markets, training in National Organic Program 
and USDA Good Agricultural Practices record keeping, guidance with farm business plans, and 
networking opportunities with other farmers and educational workshops. If successful, the 
farmer training program could be a model for other such programs in the United States seeking 
to improve healthy local food access by equipping aspiring farmers. 
 
Despite the potential benefits of the farmer training program, the enrollment and graduation 
rates have been low. In the four years since the program was established, six farmers have 
participated in the program, while the original target completion number was between 12 and 
24. We conducted a qualitative evaluation of the program, combined with an exploratory 
system dynamics model, to understand the barriers and opportunities involved in recruiting 
farmers to the program, supporting them in completing the program and in their transition to 



Olabisi et al.  Advancements in Agricultural Development 
 

https://doi.org/10.37433/aad.v1i2.33  3 

independent farming, and assessing the potential extent of the impact on agriculture in the U.P. 
In this work, we seek to capture the complex interactions and dynamic feedbacks between the 
educational program for beginning farmers, challenges to technology implementation, and 
economic barriers to entry into farming. Consequently, we are using system dynamics modeling 
to represent these dynamics (Meadows, 2008). 
 
System dynamics modeling was developed in the 1960s and has been used in a wide variety of 
fields to investigate how complex systems behave over time (Forrester, 1968). It is a 
quantitative modeling technique based on a series of differential equations, but it is also 
capable of representing qualitative information gleaned from interviews and focus groups 
through causal logic (Scott et al., 2013). System dynamics has been proposed as a useful tool for 
conducting program evaluation, because it can demonstrate potential impacts of a program 
into the future, and of outlining the effects of programmatic changes (Merrill et al., 2013). It 
has also been commonly used to represent technology adoption (Amelia et al., 2014; Fisher et 
al., 2000). We drew on the technology adoption models for this study, because we are 
representing a group of farmers acquiring and then using a set of technologies for growing 
nutrient dense produce in the U.P. As represented in the model, potential farmers learn about 
the program from educational workshops offered by the North Farm at the Upper Peninsula 
Research and Extension Center.  The farmers trained in a suite of technologies demonstrated in 
the program, including walk behind tractors, seeders, hoes, price planning and assessments, 
etc. Then the farmers decide whether to implement these technologies or not. The process of 
recruiting new trainees is driven by information exchange between potential farmers and those 
who have gone through the program, as posited by diffusion theory (Rogers, 2003).  
 
Famers that graduate from the program are assumed to be more skilled at growing nutrient 
dense produce using appropriate technology, and more skilled at developing a business plan 
that will help with their future farms. We used a system dynamics model to model these 
patterns of adoption and the ways in which the technologies could diffuse through the farming 
population of the Upper Peninsula. The research question we sought to answer with this 
model: what programmatic interventions would be most effective in increasing the 
recruitment, graduation, and establishment of beginning farmers relative to the training 
program? 
 

Methods 
 
We conducted semi-structured scoping interviews with beginner farmers to understand their 
current farming operations, goals, technology and methods learned from the program, 
profitability, productivity, challenges and constraints they face and benefits from the program. 
We carried out the study at the North Farm in Chatham, located in the Upper Peninsula of 
Michigan, and at Michigan State University, East Lansing Michigan. The study sample consisted 
of six farmers that were in the program for at least a year. The farmers in program range in 
experience from beginners to experienced farmers. We used a survey that had both structured 
and semi-structured questions, as well as an exploratory approach in the survey to get more 



Olabisi et al.  Advancements in Agricultural Development 
 

https://doi.org/10.37433/aad.v1i2.33  4 

information that might not have been revealed in the structured questions (Glesne, 2010). 
Some of the data was collected face-to-face and the rest was collected over the phone. In 
addition, we used data generated from the MSU Upper Peninsula Research and Extension 
Center: North Farm Apprentice Farmer Program Evaluation Summary report that was 
conducted in December 2018. The survey along with the report contributed to a deeper 
understanding of the issues related to the program.   
 
We used information from the scoping interviews and the 2018 evaluation report from the 
North farm to create and parameterize the system dynamics model, using Vensim® software. 
The logical framework follows Rogers’ diffusion of innovation model (Rogers, 2003) (Figure 1).  
 
Figure 1 
Structure of System Dynamics Model 
 

 
Note. Model structure depicting the progression of farmers through training, to establishment of a ‘new’ farm on 
land provided by the training program, to experienced farmer status on their own land. 
 
Potential farmers (farmers who join the training program) advance through three stages of skill 
acquisition: training, new farmer (practicing skills on land owned by the North Farm) and 
experienced (farming on their own). Their ability to advance through these stages is both 
mediated and facilitated by factors discovered through the interviews. Once the structural logic 
of the model was established, values for the parameters were derived from the interviews and 
the North Farm evaluation report. For example, we asked interviewees how many months they 
estimated for the transition from ‘new farmer’ to ‘experienced farmer’ (successfully farming on 



Olabisi et al.  Advancements in Agricultural Development 
 

https://doi.org/10.37433/aad.v1i2.33  5 

their own upon completion of the North Farm program), and the average response was 40 
months. 

Findings  
 
Interviews 
The skills acquired through the North Farm program allowed beginner farmers to learn how to 
start a business and transition to the ‘training farmers’ category. As described by the 
interviewees, these skills included networking, creating invoices for customers, and developing 
avenues to sell their products at local restaurants and through community-supported 
agriculture, in partnership with the North Farm. Trainees also learned about viable farm 
business models, including price planning, price assessment, calculations of labor, and soft 
skills, as well as soil management and safe produce handling. In addition, the farmers were 
provided with technology for sustainable production practices, such as harvesting tools, water 
wheel planters, rotor tillers, walk-behind tractors, drip irrigation systems, and hand cultivators. 
All of these factors were added into the system dynamics model as variables conditioning the 
transition rate of farmers between ‘training’, ‘new’, and ‘experienced’ categories.  
 
Model 
The baseline behavior of the system dynamics model depicting adoption of the North Farm 
technologies by the training population followed a typical S-curve growth pattern over the 30 
years of the model run (Figure 2).  
 
By the end of the model run, thirteen total farmers had successfully transitioned to farming on 
their own land, representing the cumulative impact of the North Farm program. The number of 
farmers in the ‘training farmer’ category grew as they were trained at the North Farm, then 
declined as these farmers successfully transitioned to farming on their own land. In the baseline 
model run, we assumed that all farmers entering the program had access to materials they 
needed for training: access to organically managed land, subsidized housing, access to 
equipment and hand tools, access to markets, educational workshops, business plan guidance, 
and supervised training under the guidance of a program manager. This baseline run mirrors 
the setup of the North Farm Apprentice Farmer Program, which provides this support to 
enrolled farmers. 
 
 
 
 
 
 
 
 
 
 
  



Olabisi et al.  Advancements in Agricultural Development 
 

https://doi.org/10.37433/aad.v1i2.33  6 

Figure 2 
System Dynamics Model Output  
 

 
Note. System dynamics model results depicting number of farmers in the ‘training farmers’ category (top) and the 
‘experienced farmers’ category (farmers who have successfully transitioned to farming on their own land, bottom) 
over the 15-year model run under baseline conditions. 
 
The flow of farmers into the program is ultimately limited by the amount of contact and 
positive word of mouth generated by the trainees enrolled in the program. In the context of the 
North Farm model, this was represented by the number of workshops held, driving interest 
from potential farmers in enrolling in the program. When the number of workshops was 
increased from one per quarter to one per month, this had the effect of sending a greater 



Olabisi et al.  Advancements in Agricultural Development 
 

https://doi.org/10.37433/aad.v1i2.33  7 

‘pulse’ of farmers through the system initially. Without additional support for farmers to 
transition into successfully starting their own operations, however, this did not result in 
substantially greater numbers of farmers becoming independent (Figure 3). 
 
Figure 3 
Model Output: Results of Program Interventions  
 

 
Note. System dynamics model results depicting number of farmers in the ‘training farmers’ category (top), number 
of farmers in the ‘new farmers’ category (farming on land provided by the training program, middle) and the 
‘experienced farmers’ category (farmers who have successfully transitioned to farming on their own land, bottom) 
over the 15-year model run under four scenarios: ‘high workshop’ (farmers are recruited to the training program 
through monthly rather than quarterly workshops); ‘start up’ (startup costs for farmers starting their own 
operations are subsidized); ‘experience’ (time for farmers starting their own operations to gain the necessary 
experience is reduced from 3.5 years to 1 year) and ‘more CSA’ (a 2x increase in the number of CSA customers). 



Olabisi et al.  Advancements in Agricultural Development 
 

https://doi.org/10.37433/aad.v1i2.33  8 

 
Another scenario involved decreasing start-up costs for farmers to start their own farm 
operations after training in the program. This scenario had the greatest impact on the number 
of experienced farmers by the end of the model run (increasing this number to 75), but had 
minimal impact on the number of farmers in the training program or the number of new 
farmers (Figure 3). Decreasing the amount of time needed for farmers to gain the experience to 
start their own operation from 3.5 years to 1 year also increased the number of experienced 
farmers by the end of the model run above the baseline, but did not have as great an impact on 
increasing this number as did a decrease in start-up costs. 
 
Finally, increasing direct sales of vegetables through community-supported agriculture (i.e. 
doubling the enrollment in CSAs from 25% of the population to 50%) did not have a substantial 
impact on number of farmers progressing through the system (Figure 3). This is because, in the 
absence of trainees being recruited into the program, there are not enough farmers to take 
advantage of increased CSA sales. CSA sales are also split between the new farmers and the 
experienced farmers, and new farmers are better able to take advantage of them because of 
the land and resources they can access through the North Farm program. This is a cautionary 
insight for the design of such programs, which should take care not to place their trainees and 
their graduates in competition for limited market share. This may be more important for 
programs that are large in scope relative to their local markets.  
 
When all interventions to boost the number of experienced farmers were combined into one 
model run (increasing workshops, decreasing startup costs and experience needed to start 
one’s own operation, and increasing CSAs), the number of experienced farmers after 30 years 
reached 203. This is a substantial impact for one training program at one location and could 
potentially transform the food landscape in the Upper Peninsula region, especially if duplicated 
at other sites (Figure 4).  
 
  



Olabisi et al.  Advancements in Agricultural Development 
 

https://doi.org/10.37433/aad.v1i2.33  9 

Figure 4 
Model Output: Number of Experienced Farmers  
 

 
 
Note. System dynamics model results depicting number of farmers in the experienced farmer category under 
baseline conditions, and with all interventions to increase number of farmer recruited and trained (more workshop 
recruitment, reduced startup costs, increased CSA customers, and reduced time to gain experience). 
 
This is an exploratory model, based on an in-depth qualitative study of one farmer training 
program, and is useful for gaining general systemic insights rather than crafting detailed 
predictions. For example, without a larger survey of farmer training programs across the 
country with different training models, it is difficult to place weights indicating relative 
importance on the different components of the program that allow for successful training (e.g., 
access to tools, subsidized housing, supervised training, etc.). In our model, we assumed the 
same weight for all of these elements, even though interviewees highlighted the supervisory 
guidance of a program manager as being particularly important to their learning process, which 
would imply that the selection of the person to lead the training program is particularly 
important for its success. Similarly, we were not able to generate detailed market scenarios to 
guide where new farmers might gain the best advantage in selling their products (e.g. to 
restaurants, CSAs, or farmers’ markets). A detailed economic analysis of produce markets in 
Michigan’s Upper Peninsula would be necessary to provide this type of guidance. 
 
Nevertheless, the model provided substantive insights into the design of successful farmer 
training programs, and we were able to use it to test commonly proposed interventions for 
increasing the flow of new farmers into the farming population. We found that decreasing 
start-up costs for farmers striking out on their own (e.g. through loans, subsidized land 
purchase and/or leasing arrangements) would be particularly effective in transitioning farmers 



Olabisi et al.  Advancements in Agricultural Development 
 

https://doi.org/10.37433/aad.v1i2.33  10 

from a training period to establishing their operations. At the same time, increased recruitment 
of farmers into a training program will not ultimately result in more local farmers unless their 
transition into farming on their own is also facilitated, for example through reduced or 
subsidized start-up costs, as described above.  
 

Conclusions, Discussion, and Recommendations  
 
Using a system dynamics model, we were able to depict the training of new farmers producing 
vegetables for local markets in the Upper Peninsula of Michigan as a pipeline, with barriers to 
entry and opportunities for intervention. The model represents each stage of progression from 
farmers in training, to new farmers operating on land provided by the training program, to 
experienced farmers running their own operations. We built the model using in-depth 
interviews with farmers in the North Farm Apprentice Farmer Program, combined with a 
system dynamics framework depicting adoption of new technologies. While this model is 
exploratory and not predictive, we were able to glean insights around the design of effective 
programs for training farmers who will ultimately go on to start their own operations. To 
increase the number of experienced farmers, more programs will be needed to ease the startup 
costs for farmers starting their own operations. Recruiting more farmers into a training 
program will not result in more farmers that are independent unless this startup ‘bottleneck’ is 
addressed. Programs should be careful not to engender competition for customers between 
farmers in training and farmers who have graduated the program to start their own operations. 
While modest in size, training operations like the North Farm Apprentice Farmer Program could 
have a substantive impact on recruiting and establishing new farmers over time. Finally, this 
study also demonstrates how systems dynamics can be a valuable method to evaluate and 
maximize the effectiveness of training programs. 
 

References 
 
Amelia, D. F., Kopainsky, B., & Nyanga, P. H. (2014). Exploratory model of conservation 

agriculture adoption and diffusion in Zambia: A dynamic perspective. Paper presented at 
the 32nd International Conference of the System Dynamics Society, Delft, Netherlands.  

 
Denney, J. T., Kimbro, R. T., Heck, K., & Cubbin, C. (2017). Social Cohesion and Food Insecurity: 

Insights from the Geographic Research on Wellbeing (GROW) Study. Matern Child 
Health J, 21(2), 343-350. https://doi.org/10.1007/s10995-016-2119-5  

 
Fisher, D. K., Norvell, J., Sonka, S., & Nelson, M. J. (2000). Understanding technology adoption 

through system dynamics modeling: implications for agribusiness management. The 
International Food and Agribusiness Management Review, 3(3), 281-296. 
https://doi.org/10.1016/S1096-7508(01)00048-9  

  
Forrester, J. W. (1968). Principles of Systems. Pegasus Communications. 
 



Olabisi et al.  Advancements in Agricultural Development 
 

https://doi.org/10.37433/aad.v1i2.33  11 

Lantz, N., & Walk, M. (2011). Upper Peninsula agriculture assessment. Retrieved from East 
Lansing MI:  

 
Meadows, D. (2008). Thinking in Systems: A Primer. Chelsea Green. 
 
Merrill, J. A., Deegan, M., Wilson, R. V., Kaushal, R., & Fredericks, K. (2013). A system dynamics 

evaluation model: implementation of health information exchange for public health 
reporting. Journal of the American Medical Informatics Association : JAMIA, 20(e1), 
e131-e138. https://doi.org/10.1136/amiajnl-2012-001289  

 
Ramirez, A. S., Diaz Rios, L. K., Valdez, Z., Estrada, E., & Ruiz, A. (2017). Bringing Produce to the 

People: Implementing a Social Marketing Food Access Intervention in Rural Food 
Deserts. J Nutr Educ Behav, 49(2), 166-174 e161. 
https://doi.org/10.1016/j.jneb.2016.10.017  

 
Rogers, E. (2003). The Diffusion of Innovations. The Free Press. 
 
Scott, R. J., Cavana, R. Y., & Cameron, D. (2013, July 21-25, 2013). Evaluating long-term impact 

of qualitative system dynamics workshops on participant mental models. Paper 
presented at the 31st International Conference of the System Dynamics Society, 
Cambridge, MA. 

 
United States. Congress. Senate. Special Committee on, A. (2008). Harvest over the horizon 

[electronic resource]: the challenges of aging in agriculture : hearing before the Special 
Committee on Aging, United States Senate, One Hundred Tenth Congress, first session, 
Washington, DC, June 21, 2007. 

 
 
 
 
 
© 2020 by authors. This article is an open access article distributed under the terms and conditions of 
the Creative Commons Attribution license (http://creativecommons.org/licenses/by/4.0/). 
 


