LINC was introduced to the Rice and Diversified Crops (RDC) project in the Winter of 2017, soon after the RDC project’s launch. LINC assisted the RDC project to utilize network analysis combined with qualitative interviews to better understand systems dynamics and change in the network of grantees assisted by the RDC project. In its first full year of implementation, the activity goal of RDC was to improve food security through systemic changes that increase rural incomes by catalyzing market systems changes that promote a diversified farm management approach oriented to intensified rice production and/or diversification of higher-value nutrient rich crops.

We spent several weeks exchanging documents and becoming more familiar with the design and objectives of the RDC project. We understood that the project was interested in utilizing network analysis to both a) assess system change; and b) identification of success / promising approaches for scale up. We further defined the following parameters:

• The network analysis should include observations to inform RDC strategy;
• The network analysis should be iterative to assess change;
• The approach should be replicable, able to be integrated into RDC’s MEL system

LINC took a deliberate approach to the Bangladesh research. The work was structured in the following phases:

Assessment of systems change

We first learned that grantees of the RDC project (also referred to as “lead firms”) were expected to be the key nodes of systems change for this activity. Grantees would be funded for a range of activities, and a major focus of the project MEL activities. Successful grant activities would be scaled up over time, and understanding the relationships that they forge with other actors in the network would be an important element of assessing systems change. Approximately 60 grant projects would be funded over the course of the project, and 9 such grantees had already been developed.

Responding to an interest on the part of RDC staff, we next examined the possibility of utilizing network analysis to capture systems change attributable to the RDC project. The possibility was quickly dismissed however, given that this would require a time and resource-intensive control group approach, similar in scale to an external impact evaluation.

We then examined options for conducting a network census. This is the classic approach to network analysis, a census in which all of the actors in the network are identified and surveyed against a pre-defined network boundary. The difficulty in our case was that the RDC project at this stage sought to capture quite broad interactions among a diversity of actors, and was not able to pre-define / name all of the actors in the network. This would have required a snowball data collection approach that would have likely taken multiple months and a great deal of staff time.

Given the constraints to conducting whole network analysis, and the objectives and parameters of RDC, we ultimately decided to undertake grantee ego-net analysis. Egonets are significantly more manageable that whole networks, as they focus only on the grantee and their relationships with alters. This is a slice of the whole network, focused on an actor of high interest to the RDC project. It gives a good understanding of the ego, but has short-comings in couching the ego’s relationships within the overall system / network.

To gain a better understanding of the broader network / system of relationships beyond the ego, we married our quantitative network analysis with in-depth qualitative research. This meant fielding a qualitative researcher with expertise in Bangladesh’s market systems, to conduct follow-up interviews with surveyed egos / grantees, sharing preliminary maps with them, in some cases revising, and discussing relationships of interest. Further, the qualitative researcher interviewed several alters to gain an understanding of their relationships and perspectives, although these alters were never surveyed with the quantitative network analysis tool, only named by the egos.

Identifying successful grant activities

The RDC projects Collaborate, Learn, Adapt (CLA) approach meant that grantee success and failure should be captured on an ongoing basis, subsequently scaled-up or scaled-back, replicated and/or adapted. While we knew that network analysis could not do this on its own, we thought that the tool could have an important role to play.

Based on this, we elected to design-in an iterative / longitudinal ego network analysis approach. Conducting network analysis in iteration allows us to capture dynamic relationship qualities rather than a fixed snapshot in time. While attribution is still problematic due to the absence of a control group, appropriately targeted questions can give a strong indication of change as a result of the RDC project when assessing grantees themselves before and after grant activities. Ideally, we would conduct network analysis with each grantee before (baseline), during (midterm), and after (endline) the completion of the grant award from RDC. After learning that grant awards would normally extend 6-9 months, we determined with the field that it would most likely only be possible to undertake baseline and endline analyses.

Local transfer / uptake

SPACES is a research and development project that supplies international expertise to projects undertaking systems initiatives. Generally, this means that the project conducts its research and fieldwork, and disappears. In this case, we wanted to make sure that this didn’t happen. This consideration was particularly critical for RDC, which we agreed would utilize this as a monitoring / adaptive management tool throughout the life of project. Both RDC and SPACES thus committed to full transfer of the tool, agreeing to:

• SPACES and the local RDC MEL team should work closely together throughout the engagement, transferring skills as we go;
• SPACES should conduct a field mission to train local RDC staff in network analysis, survey administration, and integration into RDC MEL systems;
• SPACES should conduct the baseline network analysis for all grantees identified to date (n=6), providing an example and templates that staff could utilize going forward; and
• At the end of the engagement, SPACES should hand-over all finalized tools and templates and would be available for ad-hoc remote consultations.

A baseline report was authored by SPACES and finalized in November 2017, available for review here.

The report was conducted to pilot the method, present baseline quantitative data for subsequent follow-up change measurement, present qualitative analysis, and provide observations to inform RDC program strategy. The report had the further benefit of serving as a template for utilizing network analysis as a monitoring tool going forward. Below are the highlights of the insights uncovered in our baseline report.

• Utility of the egonet tool: The egonet tool was particularly useful in identifying structural dynamics and social norms and biases that appear to constrain either the egos’ operational performance and/or that of the market system. Unlike other approaches that predominate in the market systems space, the egonet approach lends the capability of visualizing and quantifying structure and relationship strength. As opposed to a whole network survey, it is also manageable from both a time and resource perspective. Marrying quantitative and qualitative approaches is essential to providing ego analysis with insights on the overall framework and structure of the system.
• Structural dynamics uncovered: The network analysis uncovered several key structural observations that may inform strategy and follow-up change measurement / adaptation, including:
  • Gaps in relations with service providers – Only eight connections revealed among all six of the egos surveyed. Notably absent are service providers for marketing, advertising and promotions, especially given the competitive pressures for promotions indicated.
  • Weak coordination between seed companies and research institutions – There are gaps in knowledge and communications, with no industry association currently positioned to streamline coordination and communications.
  • Narrow distribution and supply channels – Lead firms generally relying on large numbers of small interconnected firms, fairly established relationships, and small exclusive territories. Poses challenges for scaling and value addition.

• Social norms and biases uncovered: Overall observations suggest that all egos struggle in managing their supply or distribution channels, to shift the business strategies of their suppliers and distributors from traditional extractive ones to value-additive ones. Specifically, we see:
  • Lack of growth among suppliers and distributors (alters), as seen in the narrow distribution channels and minimal investment in upgrading of business systems, infrastructure, or staffing despite strong volumes.
  • Lead firms (egos) expressing desire for their trading partners to adopt more value-add strategies.
  • Lead firms (egos) indicating that their relationships with larger suppliers and distributors are those that are best able to satisfy their most important values and preferences.
• Systemic leverage points identified: Network analysis observations revealed that the supply and distribution channels of lead firms are predominantly narrow, and businesses largely engage in extractive strategies. In fact, it appears that these dynamics are mutually reinforcing. For example, where extractive businesses do not invest in growth or upgrades to operations, supply and distribution channels remain narrow. Furthermore, as a result, demand for support services is likely low and stagnant; businesses who are not growth-oriented have seemingly little need for expert services.

Ultimately, these patterns have negative implications for small-holder farmers who typically have difficulties accessing higher-value markets and improving productivity. Part of the RDC project’s theory of change is to rectify these dynamics such that farmers instead are connected to broad supply and distribution channels where actors compete on value-additive strategies, providing farmers with input supply channels that can respond to their needs to improve productivity and output market channels that offer opportunities and incentives to improve production. Our analysis recommends five leverage points that the RDC project may address to promote this shift, indicated in the graphic below.

As of the authoring of the case study, soon after the completion of the baseline in November 2017, we had already seen results informing application of the tool itself, program strategy and the viability of network analysis for monitoring. We anticipate this case study to be updated as we learn more from follow-up network analyses conducted by RDC, resultant change data, and the extent to which the tool informs ultimate scale-up or scaling-back of grantee activities.

Feasibility as a monitoring tool

So far the Bangladesh RDC work has demonstrated that it is possible to develop a network analysis approach that can be transferred locally and integrated into program monitoring systems in a reasonably cost-effective manner, similar to the way in which a periodic grantee survey would be incorporated into a more traditional M&E system. This is important, as there are presently few examples of international development projects that have successfully mainlined iterative quantitative network analysis data collection into project M&E systems / processes.

It is however important to note that the chief driver of this feasibility is adaptation of the network analysis method. We utilized egonet analysis, a hybrid approach that makes data collection straightforward but does not capture the entire system. Complementary qualitative approaches are required for that additional perspective.

There remains additional work to do in streamlining data collection processes for RDC so that the network analysis tool can be self-administered by grantees via internet. The online data collection system has been established and utilized in the piloting. Efforts to streamline the questionnaire instrument are in progress.

Tool transfer

While the baseline network analysis conducted by SPACES took six months to complete, we saw a high degree of enthusiasm on the part of the RDC project staff once the report was finalized. The final report showed the value of network analysis to the project on both a strategy and monitoring basis. As well, the final report along with tools and templates provided, served as a roadmap for the RDC team to undertake subsequent analyses going forward. Based on this experience and others, we strongly recommend tool transfer approaches that not only train, but co-implement, and demonstrate the utility of these tools through documentation such as this baseline report.

Knowing that transfer of the tool was a key objective, SPACES infused both theoretical and practical training modules into our engagement with RDC. We conducted an in-person two-day training early in the engagement, once the quantitative instrument was finalized prior to the data collection phase. Including thirty participants from the RDC staff, this proved to be a good tool to raise awareness of network analysis and build some basic understanding and skills. Nonetheless, significantly more practical, hands-on training was required with the MEL staff who would be responsible for taking the tool forward. This was provided on an ongoing basis, both in-country and from remote.

One thing that we lacked in local transfer of the tool was a local institution firmly rooted in Bangladesh with some level of specialization in systems and/or MEL. We worked directly with the RDC project staff, employed by two international organizations, ACDI-VOCA and Action for Enterprises (AFE). While their staff are Bangladeshi and based in Bangladesh, we would have generally preferred to transfer the skills and tools directly to a local organization that might have had stronger prospects for institutionalizing the method locally.

Strategy insights

We have known for some time that network analysis can be an excellent exploratory tool producing insights that do well to inform program strategy and design. This has again proven to be the case with this Bangladesh network analysis. Field staff have pointed out insights that were new to them, particularly those insights related to the composition of supply and distribution channels. On this basis they have initiated a number of discussions on how those might be broadened. In our own estimation, we came away from the study encouraged by the extent to which the analysis informed network structure and the social biases of network members. Importantly though, these insights were the product of combined quantitative and qualitative data collection and analytics, not one or the other in isolation. This bolsters the case that we have been making for some time, specifically that quantitative network analysis data is meaningless in the absence of qualitative insights.

Reliability of strategic insights

Strategy insights derived from our baseline report may have limited reliability, as they are derived from quantitative data that includes only six egos. While the qualitative component did much to compensate, as the population of grantees participating in RDC’s network analysis grows over LOP, we anticipate that those results will be increasingly robust and generalizable. As egos progress through the grant cycle, we anticipate that change data from follow-up analysis will inform adaptation, providing leads on what is and isn’t working, identifying promising prospects for scale-up.