Institutions

Institutions can be regional or global and are assigned a behavioural type (see Table 1) that is responsible for their complex decision making on actions they perform (see section 2. Individual Decision Making, Institutions).

At each modelled time step, institutional agents perceive the land use system and possibly trigger decision making as defined. Then, the level of service production achieved by an agent is given a benefit value via a benefit function that relates production levels to unmet demand. Agents compete for land based on these benefit values, and this competition is affected by individual or typological behaviour. Table 3 gives an overview of the CRAFTY simulation schedule.

All of the institutional agents then have a chance to update their policies if the desired land use effects have not occurred (according to user-defined criteria related to the above updates), in particular their effects on benefit values.

Institutional decision making is inspired by the Theory of Planned Behaviour (TPB; Ajzen1991) to capture various in- fluences on their decision making and also maintain analytical tractability in terms of tracing back their decisions. Table 5 lists central assumptions regarding the representation of institutions.

Design assumptions made in CRAFTY for the modelling of institution agents.

CRAFTY models two separate kinds of decision making, namely land use behaviour of land managers on the micro level and decisions about subsidising by institutional agents on two levels (regional and global).

• Institutional agents have a fixed set of potential actions.

The set of potential actions an agent may select in decision making processes and perform afterwards needs to be defined and assigned beforehand. However, the usefulness of potential actions towards preferences changes.

The evolution of potential actions during the time span of simulation can be emu- lated by defining them beforehand and let their utility become positive not before a certain time step.

• Institutions can act by adjusting benefit values (i.e. provid- ing subsidies or disincentives).

Institutions can adapt their actions by choosing between their potential actions in response to their preferences for judging the success or failure of intervention (monitoring actions’ effects).

Numerous examples of institutions acting in these ways are available (see e.g. Huang, Wang, Zhi, Huang, & Rozelle (2011), Skinner, Kuhn, & Joseph (2001) and Swinnen & Gow (1999)), and we contend that these are sufficient for a stylised description of institutional behaviour.

Land managers pursue a satisficing approach in their behaviour. They perceive their production and become aware of the benefit it means to them. Depending on the relation of this benefit to their individual giving up and giving in thresholds (see above) they continue, may give in to another type of land use or simply abandon their land. Figure 3 illustrates the decision making model according to (Schluter2017). Spatial conditions impact on land manager’s deci- sion making as they influence their performance. Therefore, land use is adapted to service demands and changing conditions, e.g. due to climate change.

Institutional Agents pursue the objective of closing the demand supply gap within the area they are responsible for. These institutions can adjust agent benefits, which is equivalent to the provision of incentives (e.g., subsidies) for certain services (production-based). These adjustments may consider specific functional roles, so particular areas and/or land managers can be targeted for incentivisation. First, decisions are triggered in case the gap between de- mand and supply as perceived by an institution exceeds the institution’s threshold. The institutional agent then ap- plies deliberative decision making inspired by the Theory of Planned Behaviour (Ajzen1991) as shown in Figure 4. Potential behavioural options, i.e. ways to subsidise particular land managers, are assessed in terms of costs (part of institutions’ attitude), social approval (representing social norm), and their potential to close the demand supply gap (perceived behavioural control). The values are weighted by the agent’s preferences (e.g., strong preference to achieve balance between demand and supply of a certain service), and the best scoring action is performed after- wards. Modelled institutions do not adapt their behaviour via changing their behavioural options but the current context has an impact on the evaluation of these options.

The decision making submodel is realised by the integration of the Lightweight Architecture for boundedly Rational Agents (LARA)3 (Briegel et al., 2012). as depicted in Figure 5.

Several institutions may be active in a region or globally, and they have cumulative effects on benefit values. For instance, taking Iben to be the overall effect on benefits, an agent’s effective benefit value is: 𝑈􏰀,􏰁 =􏰂𝐼􏰃􏰄􏰅,􏰁(𝑓􏰀,􏰆(𝐼􏰇􏰀􏰈,􏰁(𝐶􏰁))𝑢(𝑟􏰆)) 􏰉 Institutional agents adapt their subsidising behaviour to changes in service demand, which is expressed in different demand supply gaps.

Insitutions are specific kinds of agents that influence the system and usually represent institutional actors on various scales:

• Authorities
• Governments
• Suppliers
• Farming Associations
• R&D institutions

Currently, the primary means of institutions to influence the land use system are innovations that are defined and triggered at certain time. Examples for innovations comprise:

• Regulations / Quotas
• Knowledge / new crop types
• Subsidies
• Investments in infrastructure

Possible effects are on

• Demand side
• Supply side
  • Adjusting competitiveness. This can be a general response to production levels (e.g. valuing certain outputs) or it can include the agent type in the calculation (e.g. valuing certain types of producer).
  • Forbidding certain agents in certain locations (e.g. restricting practises or land uses).
• Capital values: each Cell has a base level of capitals, and an effective level, taking into account institutional adjustment

There are a number of different kinds of institutions implemented. See [Setup of Institutions](Institutions set-up) for their details and configuration.