SoNARe

Modelling social and economic influences on the decision making of farmers

in the Odra case study region

Center for Environmental Systems Research,Kassel

2CAVES Meeting, 28 March 2007

Outline

• Motivation• The CAVES Odra case study

– Spatially explicit biophysical model (developed by WUT)

• The „finer grained“ agent-based model– Explicit empirically supported farmer decision rules– Modelling economic and social aspects of decision making– First simulation run(s)

• Outlook

3CAVES Meeting, 28 March 2007

Motivation

• Asymmetric dependency relation requires collective action

• Evidence for different farmer types and their respective sets of decision rules

• Explicitly contrast social and economic influences on decision making

• Make social pressure explicit in order to model the influence of water partnership initiators (WPIs) and model general opinion dynamics

4CAVES Meeting, 28 March 2007

Biophysical model setup

• Land parcels– Located along a channel, uniform size– Upstream-downstream neighbouring relationship between

owners– LRS condition, LU type, sluice gate

• Channels– Uniform slope– No branching, no interconnections

• Number of land parcels per channel– Same for all channels

• Weather conditions– Normal, drought, flooding, set yearly– Different weather sequences

6CAVES Meeting, 28 March 2007

Agent-based model setup

• Agents– Farmer– WPI, not necessarily a farmer itself

• Networks– Dependency “network” reflects spatial neighbourhood

relationship– Farmers are embedded in an acquaintance network– WPI is acquainted with, i.e. linked to, all farmers in a star-

like fashion

7CAVES Meeting, 28 March 2007

Agent-based model setup – economic Aspects

• Farmer agents recall their past economic success– Number of years memorised – Yield threshold

• defines “good years” or “bad years”– Economic sensitivity

• determines how much “good”/”bad” yields affect the perceived economic success

– Economic success• good years memorised increase the perceived

economic success, bad years decrease it

• WPI uses social network to observe farmers’ economic success

8CAVES Meeting, 28 March 2007

Agent-based model setup – Social aspects

• Farmers exert social influence– Use outgoing network edges– Positive influence, endorsement

• acquaintances using the same LRS-strategy are supported

– Negative influence • acquaintances using the opposite LRS-strategy are

pressured into switching the strategy

• Farmers perceive their present level of social support– Use incoming network edges– (sum of) social influences received from neighbours in the

acquaintance network (including WPI)

• WPI may exert additional social influence pro LRS

9CAVES Meeting, 28 March 2007

Agent-based model setup – Network types

10CAVES Meeting, 28 March 2007

Agent Decision Making

• Farmers– IF

social support + economic success sufficiently lowTHEN switch LRS maintenance strategy (maintain/¬maintain)

– IFWP exists and maintain LRSTHENjoin / stay in WPELSEdo not join / leave WP

– (always exert social influence in favour of own strategy; possibly higher influence when member of WP)

• Water Partnership Initiator (WPI)– IF

number of farmers with big losses >= 3THENexert social influence pro LRSELSEdo not exert social influence

• Water Partnership (WP)– IF

number of farmers maintaining LRS >= 3THEN activate WPELSEdeactivate WP

11CAVES Meeting, 28 March 2007

Model Execution Cycle

• May: plant crops• October: harvest crops• December: make decisions for the coming year, i.e.

1. perceive and memorise yield 2. exert social influence3. perceive social influence and economic success4. decide (decisions are buffered => synchronised)5. commit to decisions

12CAVES Meeting, 28 March 2007

Abstract Land Parcel Map

Flow direction of channel

Agents maintaining LRS

Agents neglecting LRS

13CAVES Meeting, 28 March 2007

Scenario A

• Baseline scenario, 1 channel, 10 farmers• 2 normal years followed by 1 year of flooding• Farmers do not rate their economic success• No social influence• Thus: no opinion dynamics

14CAVES Meeting, 28 March 2007

Scenario A

12 24 36 48 60 72 84 96 108

15CAVES Meeting, 28 March 2007

Scenario B

• 1 channel, 10 farmers• 2 normal years followed by 1 year of flooding• Farmers rate their economic success

– yieldThreshold = 9.0

• No social influence

16CAVES Meeting, 28 March 2007

Scenario B

17CAVES Meeting, 28 March 2007

Scenario B

18CAVES Meeting, 28 March 2007

Scenario B

84 96 120 132 144 180 192 360... ......

19CAVES Meeting, 28 March 2007

Scenario C

• 10 channels, each 10 farmers• 2 normal years followed by 1 year of flooding• Farmers rate their economic success

– yieldThreshold = 9.0

• Scale-Free topology for acquaintance network• WPI (linked to all farmers in a star-like fashion)• Farmers and WPI exert social influence

20CAVES Meeting, 28 March 2007

Scenario C

21CAVES Meeting, 28 March 2007

Scenario C

22CAVES Meeting, 28 March 2007

Scenario C

23CAVES Meeting, 28 March 2007

Scenario C

24CAVES Meeting, 28 March 2007

Scenario C

36 48

25CAVES Meeting, 28 March 2007

Scenario C

168 180

26CAVES Meeting, 28 March 2007

Scenario C

288 300

27CAVES Meeting, 28 March 2007

Scenario C

336 480

28CAVES Meeting, 28 March 2007

Outlook

• calibrate the model• include allowances and compensation payments• include sluice gate operation / fish ponds• include additional land use types• distribute farmer types heterogenously• apply different network topologies• perform sensitivity analyses