1 Introduction

While knowledge is critical for organisational success (Garavelli et al. 2002; Goh 2002; Hwang et al. 2008; Karlsen and Gottschalk 2004; Othman et al. 2014), when an organisation is aware of its necessity, it will spend many resources to manage it (Iyengar et al. 2015) and facilitate its timely transfer where it can be used (Kuo and Lee 2009; Vance and Eynon 1998). Alas, barriers can persist regardless of how much organizations are fully dedicated to knowledge management (Szulanski 1996). Absorptive Capacity (ACAP), defined as the dynamic capability to absorb knowledge (Cohen and Levinthal 1990), is a particularly stubborn barrier to resolve. Research into ACAP at the individual level, which underpins organisational ACAP, has been neglected (Gao et al. 2017; Lowik et al. 2017; Minbaeva et al. 2012).

  • Agent-based modeling (ABM) is a simulation paradigm transfer using individual ACAP that is yet to be undertaken. This research uses ABM to simulate organisational knowledge transfer from individuals’ ACAP. This study takes a unique approach by applying agent-based modeling (ABM) to the individual Adaptive Capability (ACAP) process. The aim is to assess how individual learning translates to improvements in knowledge transfer throughout the organization. This innovative application provides insights into the overall impact of employee learning and broadens the potential uses of ABM simulations in the field. The study addresses the following research questions: Can organisations’ knowledge transfer be simulated using individual ACAP and ABM? How does the interaction between individuals affect organisational knowledge transfer? What approach can be used to model an ABM about knowledge transfer?

  • This paper is organized as follows: Sect. 2 describes the theoretical background for the proposed system. The methodology for implementing the ABM is discussed in Sect. 3. The experimental results and discussion have been performed in Sect. 4. Finally, the paper concludes.

2 Theoretical background and literature review

Absorptive Capacity (ACAP) refers to the dynamic capability to absorb knowledge, as defined by Cohen and Levinthal (1990). While knowledge is intangible (Lopez-Cruz and Garnica 2018), it can be structured in various abstract elements. It is also dynamic in its structure and relies on intuition and human subconscious processes to consume (Davenport and Prusak 1998; p. 5). Knowledge has been defined differently across history by many authors, from Plato (1999) famous statement that knowledge is justified true beliefs (Vance and Eynon 1998) to Davenport and Prusak (1998; pp. 2–6) recent definition that when data has meaning, it is information. When information has been experienced, it is knowledge (Garavelli et al. 2002; Karlsen and Gottschalk 2004).

While knowledge itself is important, transferring that knowledge is equally crucial. Knowledge transfer is critical for organisations (Garavelli et al. 2002; Goh 2002; Hwang et al. 2008; Karlsen and Gottschalk 2004; Othman et al. 2014). Szulanski (1996) found that among the common barriers to knowledge transfer, the most common ones were causal ambiguity, an arduous rapport between the knowledge holder and knowledge recipients (KR), and the KR’s own ACAP. Causal ambiguity is when the cause of an effect is uncertain or unknown (Szulanski 1996). An arduous rapport between the knowledge holder and the KR is believed to originate from a punitive and distrustful organisational culture (Goh 2002). Also, withholding critical information allows managers to assert power (Goh 2002). Unlike a traditional vertical structure where one manager has many employees, horizontal structures allow communication to flow across business functions (Goh 2002). They promote trust and disseminate aspirations (Goh 2002; Karlsen and Gottschalk 2004; Tang et al. 2006; Uygur 2013). They effectively solve causal ambiguity and arduous rapport between the knowledge holder and the knowledge recipients (KR) (Karlsen and Gottschalk 2004; Tang et al. 2006; Uygur 2013) but leave the KR absorptive capacity unaffected and unresolved. There have been many versions of ACAP, from Cohen and Levinthal (1990) those who first introduced it to more recent revisions. While all these different versions differ in important nuances, these portray the same narrative, which starts with “not knowing” and ends with “knowing.” The ACAP framework proposed by Zahra and George (2002) is considered a foundational version, as it introduced new elements that Cohen and Levinthal (1990) had omitted. Their framework identifies four distinct capabilities within the absorption process (see Fig. 1):

  • Acquisition is the first capability where the object of knowledge is acquired (Zahra and George 2002).

  • In the Assimilation capability, the knowledge is extracted from the object (Zahra and George 2002), which knowledge is comprehended or understood.

  • Transformation is where processes are reconfigured to exploit newly acquired knowledge (Zahra and George 2002). The amount of time and effort required to unlearn and relearn the new processes and knowledge with the old one is proportional to how much prior knowledge is ingrained (Szulanski 2000). This is where knowledge is reflected or thought.

  • Exploitation is where knowledge is used, and its value is returned (Zahra and George 2002). Exploiting knowledge is often viewed as a successful demonstration that knowledge has been absorbed (Jacobs and Buys 2010).

Fig. 1
figure 1

Absorptive capacity (Zahra and George 2002)

Full size image

Most ACAP research has been into innovation, performance and knowledge transfer from the perspective of organisations (Jansen et al. 2005; Lowik et al. 2017; Yao and Chang 2017). Cohen and Levinthal (1990; p. 131) remarked that “an organisation’s absorptive capacity will depend on the absorptive capacities of its individual members” and “a firm’s absorptive capacity is not simply the sum of the absorptive capacities of its employees”. As organisations are made of individuals (Cohen and Levinthal 1990; p. 131), individuals’ behaviour is the root of the organisation’s knowledge transfer (Minbaeva et al. 2012).

Research into ACAP as a barrier at the individual level has been neglected (Gao et al. 2017; Lowik et al. 2017; Minbaeva et al. 2012). Knowledge transfer can be simulated using the characteristics of individuals. As horizontal organisational structures allow communication to flow between departments, they resolve the barriers of causal ambiguity, and arduous rapport between knowledge holders and KR (Karlsen and Gottschalk 2004; Tang et al. 2006; Uygur 2013). Higher network diversity also enhances individuals’ innovation capability (Lowik et al. 2017). This suggests that interactions between individuals may improve KR ACAP.

Agent-based models (ABM) are computer simulations employing agents to portray social actors, facilitating interactions within a virtual environment mirroring real-world scenarios (Abdou et al. 2012; Jacoby 2002; Kennedy 2012). As individual-based models, ABMs concentrate on the actions of these agents, making them ideal for analyzing intricate systems stemming from the collective interactions of numerous individuals. Addressing the research question, we propose the following hypotheses.

Hypothesis 1

The simulation of knowledge transfer using individual characteristics will effectively replicate the knowledge transfer process.

Hypothesis 2a

Increased interactions between individuals positively impact organisational knowledge transfer, enhancing knowledge acquisition and utilization.

Hypothesis 2b

Excessive or weak individual interactions negatively impact organisational knowledge transfer, resulting in diminished knowledge acquisition and utilization.

Hypothesis 3

The construction of an Agent-Based Model (ABM) for knowledge transfer can be divided into distinct stages, allowing for a systematic and structured approach to simulate the process accurately.

3 Methodology: agent-based modelling

ABM are simulations where agents, representing social actors, react in a computer environment, representing a real-world environment (Abdou et al. 2012). ABM is catalogued as an individual-based model (Crooks and Heppenstall 2012) which aligns well with individual ACAP. While the behaviour of individuals can be inconsistent, noisy or unexpected, as individuals have different skills and knowledge, their decisions are not a random selection between different options (Kennedy 2012). Individuals decide what actions to take based on what they remember, their current state, and the information they receive (Kennedy 2012). While it is often believed that individuals behave rationally (Jacoby 2002; Kennedy 2012), rationality is judged by external norms (Jacoby 2002). ABM allows researchers to create, analyse and experiment with models made of agents interacting in an environment (Abdou et al. 2012). Simulations, where individuals make decisions, allow for a bottom-up approach to human simulation systems (Crooks and Heppenstall 2012). An agent’s behaviour is determined by their rationale, which requires knowledge to be represented, absorbed, remembered and applied (Kennedy 2012). This aligns well with ACAP.

3.1 Analysis: ROADMAP method

ROADMAP (Role Oriented Analysis and Design for Multi-Agent Programming) method separates analysis with a goal model, role model and social model from design with an agent model, interaction model and the inclusion of an ABM architecture at this stage (Kuan et al. 2005). This method allows for scalability and ease of use (Kuan et al. 2005). The goal model has been reproduced in the Overview, Design Concepts and Details (ODD) section under Objectives. The only role in this case study is that of a knowledge recipient (KR) or student. Regarding the social model, students are peers. For the agent model, each activity is an ACAP capability of Acquisition, Assimilation, Transformation, and Exploitation adopted from Zahra and George (2002). The interaction between KR would be known as knowledge spill-over.

3.2 Design: overview, design concept, and detail

ODD allows others to replicate a model by supplying the parameters, data, source code, and descriptions, which can increase its underlying assumptions and confidence (Crooks and Heppenstall 2012). ODD is standard for ABM documentation (Grimm et al. 2020). Non-trivial ABM, ODD can be very long (Grimm et al. 2020). Hence, this study’s ODD has been summarized to fit the word limit. The “Overview” provides an overview, the “Design Concepts” details the designs used, and the “Details” explains all the extra details for the model (Grimm et al. 2020).

3.2.1 Purpose

“Overview’s” first section, “Purpose and patterns”, briefly describes the model’s purpose and the patterns that are used to evaluate the model (Grimm et al. 2020). This study aims to model organisational knowledge transfer based on individuals’ ACAP using ABM.

3.2.2 Entities, variables, and parameters

The section lists the different entities in the model, their state variables and definitions, and the model’s spatial and temporal resolution (Grimm et al. 2020).

Table 1 Parameters, variables, units, or entities
Full size table

3.2.3 Process overview and scheduling

Section 3, “Process overview and scheduling”, gives an overview of the model’s processes with their order, their time, and what state variables are updated (Grimm et al. 2020). Scheduling should represent a lesson that is 180 min in total. As each lesson is assumed to be independent of the other, the scope of this case study can be of a lesson and not of a course.

3.2.4 Design concepts

Design concepts” only has one section, Sect. 4, “Design concepts,” which includes 11 concepts important for designing the ABM model (Grimm et al. 2020). For the first step of the Basic Principle, this study set is a university course, as the data collected was from the same setting. Regarding the next step of Emergence, the scenario analysis determines the emerging phenomena. Scenarios present many pictures of possible futures and aid with limited and strategic thinking (Amer et al. 2013; Guenther et al. 2017). Scenario analysis can be used as a means to represent the different paths an individual can choose based on characteristics. Gausemeier et al. (1998) propose five different phases to develop a scenario:

  1. 1.

    Scenario preparation is where the object of a scenario is decided. The object of this study is ACAP at the individual level and how it affects collectives.

  2. 2.

    Scenario-field analysis is where key factors of influence are determined. Key factors can be determined with an influence matrix or a system grid (Gausemeier et al. 1998) (see Table 1). The key factors are Acquisition, Assimilation, Transformation and Exploitation.

  3. 3.

    Scenario prognostic is where the projections are developed (Gausemeier et al. 1998) (see Table 2).

  4. 4.

    In scenario development, the first step is to perform a consistency analysis (Gausemeier et al. 1998) (see Table 3). The next step is to perform a cluster analysis (Gausemeier et al. 1998) (see Fig. 2). The last step is to analyse each cluster of scenarios and assign a prosaic description to each (Gausemeier et al. 1998).

  5. 5.

    Scenario transfer begins with the consequence analysis with the opportunities and threats (Gausemeier et al. 1998) (see Table 4). The decision-field component for this case study would be “knowledge transfer” (see Table 5).

Table 2 Scenario field analysis
Full size table
Table 3 Projection catalogue
Full size table
Table 4 Consistency matrix
Full size table
Fig. 2
figure 2

Cluster analysis

Full size image
Table 5 Consequence matrix
Full size table

A KR either absorbs knowledge or does not which is indicated by whether they have reached the exploitation capability. For Adaption, Savin and Egbetokun (2016) offer a mathematical approach to ACAP. The context of this study focuses on the knowledge transfer to a KR (i) (which Savin and Egbetokun (2016) call firm) from a knowledge source (j) (which Savin and Egbetokun (2016) call potential partner). Using this mathematical formula and making some assumptions, other values can be reverse-engineered. Savin and Egbetokun (2016) define the variable of cognitive distance (dij) between the KR i from a source j. This distance is attributed to the KR in the interval [0, 1]. For this case study, students (which are, in this case study, the KR) are assumed not to know what teachers (the source) know. As stated in Sect. 3, the cognitive distance is assumed to be of its maximum possible value of 1 for each lesson. ACAP can be represented as the function (ani, j) of knowledge absorbed by the KR i from a source j (Savin and Egbetokun 2016). The median for each record is calculated to provide a value for ACAP. A lesson is made of 1 lecture of 90 min and one tutorial of 90 min. A lesson is a period (t) of 3 h or 180 min for each lesson. As a KR would absorb the cognitive distance assumed 1 over a period of time (t), the rate (r) can be calculated as over the cognitive distance (dij = 1) a student uses their ACAP (median of each capability and, j) on a period (t):

$$r= \frac{{d}_{ij}}{{an}_{i,j}*t}$$

Equation 1: Cognitive distance over period of time rate equation.

The knowledge spill-over variable was not measured. While reverse engineering a spill-over rate may be possible, running multiple simulations and observing how knowledge spill-over might affect collective knowledge transfer would be more valuable. This would provide insight into whether collaboration enhances or diminishes collective knowledge transfer. For Objectives, as the goal model has already been defined in the ROADMAP analysis, the objectives can easily be derived from it (see Figs. 3, 4 and 5).

Fig. 3
figure 3

Goal model for KR to absorb knowledge in knowledge transfer

Full size image
Fig. 4
figure 4

State transition diagram for a KR

Full size image
Fig. 5
figure 5

Social model for KR as a student in a University course

Full size image

In this case study, the objective of a KR is to acquire, assimilate, transform and exploit knowledge. This study uses the data set from (Dolmark et al. 2021). During the process of ACAP, a KR will be in a different state. These states have already been defined as the goals of ACAP, which are acquired, assimilated, transformed, and exploited.

The data set can calculate transition durations between each state of ACAP for each agent. For interactions, these activities are defined in the agent model. The agent model has only four activities of acquisition, assimilation, transformation, and exploitation, each defined as their ACAP capability.

In this case study, a KR or student will interact with another student to transmit knowledge. This has been described before as a knowledge spill-over. A knowledge spill-over would require one agent to exploit knowledge to share it and another agent to listen. This interaction would only take place during course activities that allow students to interact with each other. The stochasticity section describes using random numbers in the model (Grimm et al. 2020). At initialization, a pseudorandom number generator can set some of the agents’ values (Grimm et al. 2020). Stochasticity or randomness used correctly would mimic the randomness of the real world, such as involuntary spill-over δn. However, in this case study, stochasticity has not been used. The collectives section describes groups or collectives affected by or affects agents (Grimm et al. 2020). For this case study, the collective is not used. The social or organisational model describes different agents’ relationships (Kuan et al. 2005). This model can help articulate relationships between agents.

KR are peers to one another. The only interaction they can have is knowledge spill-over. For Observation, running the simulation multiple times with different spill-over rates would allow for observing different collective knowledge transfers. This observation could be compared to determine the effect of knowledge spill-over. This would require at least two simulations with one spill-over rate of 0 to represent students not interacting with one another and another, which can be of a value of 1. This would provide results without having to depend on a real-life experiment.

3.2.5 Initialisation

In initialisation, the creation of different entities with their variable when the simulation begins (Grimm et al. 2020). The simulation was designed and executed for this case study using AnyLogic 8 Personal Learning Edition 8.8.5 (Build: 8.8.5.202310311100 × 64). The simulation was set to minutes. In the “Main - agent type” window, within its “Properties” tab, under the “Space and network” section, the “Layout type:” dropdown box was set to “Random”. This will display agents at one random location in the simulation.

Select the “Lesson_presentation - Agent Presentation” (by clicking the person icon), within its “properties” tab, under the “Advanced” section, the “Draw agent with offset to this position” checkbox was checked”. This will display each agent at a distinct location in the simulation. Select the “Lesson - Student” in the “Main - agent type” window; within its “properties” tab, the “Population is:” radio button list is selected to “Loaded from database”, and its dropdown box is set to the data loaded into the database (see “Sect. 11.” section for further information). This will load data from a database into the simulation. The “Mode” radio button list is selected to “One agent per database record”. This will create an agent for each record in the database. In the “Agent parameter mapping:” table, each record maps a “Parameter” from the “Student-agent type” to a “Column” from the database. Here, the parameter rates are mapped to their corresponding database rate. Statistics count the students for each different ACAP state. Thus, in the “Statistics” section, there is a statistics for each ACAP state whose “Name:” textfield is set to the name of the ACAP state, the “Type:” radio button is set to “count”, and the “Condition:” textfield is set to the instruction “item.inState(Student.” name of the ACAP state “)”. These statistics can then be used for graphs and other visualisations. In the “Student-agent type” window, the students’ ACAP flowchart is modelled. Following each state, the transition is given its corresponding rate parameter. For example, when selecting the transition between the “Acquisition” and “Assimilation”, in the “Properties” tab, in the “Transition” section, the “Triggered by:” dropdown box is set to “Rate”, and the “Rate:” textfield is set to the “acquisitionRate” parameter. The spill-over rate is modelled as an arrow from the “Exploitation” to the “Acquisition” state. The “Rate:” can be set to different fixed values, representing whether students speak to one another.

3.2.6 Input data

The risk for the data used was deemed “negligible”, and the UTS ethics committee approved using this data for this case study. The original dataset contained anonymous responses to a web survey using Likert-type scales. Participants’ information, such as name and date of birth, should have been recorded. Some items pertained to an ACAP capability: Acquisition, Assimilation, Transformation, or Exploitation. The responses were used to calculate the rate for each capability, the median, and the rate for each student. Please refer to Dolmark et al. (2021) for further details about the data used.

3.3 Implementation: execution simulation

3.3.1 Setting up the simulation

The simulation was set up in AnyLogic 8 Personal Learning Edition 8.8.5 (Build: 8.8.5.202310311100 × 64). Three simulations were run for 180 min simulation time (see Figs. 6, 7 and 8):

  • A simulation with a spill-over rate set to 0 was run to mimic a lesson taught without any student interactions. This simulation provides a baseline to compare other simulations.

  • Another simulation with a spill-over rate of 1 was also run. While this spill-over rate may not be realistic, setting it to such theoretical value allows us to observe what happens with extreme values. This is one of the advantages and purposes of using ABM.

  • A third “average of the mill” was run with a spill-over rate of 0.5. This simulation is run to observe if any average spill-over rate creates any observable phenomena of interest.

Fig. 6
figure 6

Simulation Run with Spill-over Rate set to 0

Full size image
Fig. 7
figure 7

Simulation run with spill-over rate set to 1

Full size image
Fig. 8
figure 8

Simulation run with spill-over rate set to 0.5

Full size image

3.3.2 Simulation results with spill-over rate = 0

While only 3 students have achieved the Exploitation phase, most remain in the Acquisition phase.

3.3.3 Simulation results with spill-over rate = 1

The first simulation run had a spill-over rate of 1 to mimic a lesson taught with maximum student interactions. While this spill-over rate may not be realistic, setting the spill-over rate to 1 allows us to observe the case study with extreme parameters.

With a spill-over rate set to 1, no student is in the Exploitation phase after 180 min. During the simulation, some students reached Exploitation but would return to the Acquisition phase. By the end of the simulation, there are no students in the Exploitation phase.

3.3.4 Simulation results with spill-over rate = 0.5

With a spill-over rate set to 0.5, no student is in the Exploitation phase. Again, students who reached Exploitation would return to the Acquisition phase. Again, by the end of the simulation, no students are in the Exploitation phase.

4 Discussion and implications of the results

For Hypothesis 1, “Knowledge transfer can be simulated using individual characteristics”, validation demonstrates that it is possible to simulate knowledge transfer. This also validates ABM as a suitable method to simulate knowledge transfer. While ABM was used for this case study, it may be possible to use other methods to simulate knowledge transfer.

Hypothesis

b, “Interactions between individuals diminished organisational knowledge transfer” has also been validated. Only Hypothesis 2a was invalidated. Organisation knowledge transfer slows down when individuals interact. A comparison of the results shows that when individuals interact, their ACAP returns from the Exploitation phase to the Acquisition phase. Hypothesis 2b validation has different implications. First, if humans are always learning, any extra information will cause an individual to return to the acquisition phase and begin absorbing knowledge.

While the theory is that all ACAP capability happens simultaneously, it can actually be difficult for an individual to listen and work simultaneously. This suggests that students or workers may need time to reflect to reach the exploitation phase. For a student, doing homework may fulfill this role. In academia, a suggestion would be to ensure that lectures and tutorials do not happen in succession but are adequately scheduled apart to give students time to reflect. Homework may also affect students differently. The results also suggest that students will need more than just 3 h of weekly lessons to learn. While UTS -the home university of this research-, informs students that extra homework is needed to complete a lesson, the requirement for a student is very much individual. These observations apply to the industry as well. If interactions between organisational members are detrimental to exploiting or applying knowledge, then spending too much time in group discussions such as “meetings” may not always benefit a business. The industry could use key performance indicators or other measurements to calculate how much time an organisational member, such as a manager, spends in group discussions instead of measuring the individual. However, this may not require an ABM, as this could be calculated from individual metrics.

Finally, Hypothesis 3, “The construction of an ABM for knowledge transfer can be broken down into several stages,” has been validated. A few points are noteworthy here. To begin with, while separating analysis and design simplifies the construction of an ABM model by separating distinct sections and components, the combination of the ROADMAP analysis and the ODD design may have overlapping components that cause redundancy. For example, the ROADMAP Goal Model and the ODD Objectives section are identical. More importantly, as ABM’s strength is its ability to mimic the real world, this “world” must be modelled. This system with the phenomenon, dynamics, and causal effects should be the first to be considered.

This research implies that it provides a tool for schools or businesses to simulate knowledge transfer using ABM with individual ACAP. An important consideration is how to actually model an agent. An agent comprises metrics such as parameters and variables, which should represent an individual’s reasons and an algorithm that represents an individual’s rationale. As agents interact with the system and each other, how these affect the systems and other agents’ metrics will result in the design of these considering each other. Lastly, the third element that needs to be addressed is randomness. While the system and its agents should be designed to mimic the real world, the real world contains randomness, which should be represented in an ABM as it can skew results. However, randomness remains essential as it can affect agents individually. For example, while the numbers drawn by the lottery may be a random series of numbers, the prize would affect an individual’s behaviour based on their rationale. The injection of randomness or stochasticity in an ABM should be considered at significant points. Hence, this can only happen when an essential representation has been constructed and a system and its agents have been modeled. Modellers should know this nuance and dependency to minimise “backtracking”.

4.1 Limitations

This case study has limitations. While theoretically, the design of the model should be unrestricted by the data, using different already existing data sources instead of performing data collection can free up a lot of time, effort, and uncertainty about the quantity of data to be collected. The data can be populated using stochasticity. Another area for improvement was analyzing, designing, and implementing the ABM. There needs to be a clearly defined process to develop an ABM. While the ROADMAP analysis and the ODD design provide a guide for developing an ABM, as these components are modular, they can lack cohesion which can cause “backtracking”. Changing elements from the beginning stages to the end of the development is very time and effort-consuming as these changes can affect elements that depend on them. These changes also introduce a risk of breaking the entire product. This is generally common to ABM. ABM remains vulnerable with a strong development lifecycle guide. While ABM may present itself as unreliable as the results produced are theoretical, it serves its purpose here as a simulation method. While these results may be theoretical, they provide value. How these results are interpreted and applied is in the eye of the beholder. The simulation nature of ABM is not a limitation, but how people interpret the results may be.

4.2 Further recommendation and conclusion

This research offers an approach to simulating knowledge transfer in organisations using ABM. While this case study is set in a university learning environment, the approach lays the groundwork for future case studies in other organisational settings. This research would be of benefit to:

  • The social media industry could use this approach with users’ interaction data to simulate and predict how their users behave on their systems.

  • Management could use this approach to model knowledge transfer in their organisation. They could also use this to predict members’ performances, including managers, decision-makers and anyone with high internal power.

  • The education industry could use this to model and observe how their students will behave.

  • The IT and AI discipline would benefit from another use for ABM.

  • Students, employees, and other KR would also benefit as it could give them some understanding of where and how the knowledge will come to them and to whom it may go.

  • Academia will, of course, benefit from further knowledge contribution.

The study provides some interesting results. First, the actual results provided insight into knowledge transfer dynamics. Social interactions such as group meetings can be detrimental to knowledge transfer; hence, it may be essential to consider how much time is spent interacting with others when working. More importantly, while the ROADMAP analysis and ODD design may have sections that overlap, it still provides a structured process to build an ABM. This process can be reused for other case studies. The application of ABM to observe knowledge transfer opens up many opportunities. It would be of great benefit and value to pursue these.