Urban agent-based models Modelling and examining behaviour in

ARLUNZ
Fraser Morgan
Adam Daigneault
Landcare Research
Auckland
Introduction
• Fraser Morgan
• Geographer
• Modeller
• Adam Daigneault
• Economist
• Modeller
Common interest in LUCC
Merging the two approaches
Context
• Agricultural focus
–
–
–
–
12% of GDP
Sheep, Beef and Dairy
Horticulture
Forestry
• Significant rural land
use change
– “What-if” approach to
understanding policy
effects on farming
Project
• Existing LUCC models in NZ
– Economic focus / rational profit maximising
– Spatially explicit models / lack economic realism
• Development of a rural LUCC model to
examine the effect of policy
– Strong economic focus
– Spatial, temporal and behavioural focus
– Heterogeneous farmer population
• Network interactions
Enterprise Mix
Stocking Rate
Irrigation Scheme
Land Use Class
Mitigation Option
Fertilizer Regime
Environmental
payments
Nutrient Loads
Environmental
Constraints
Water
Land
Management
Land Conversion
Costs
Economics
Input Costs
GHG Emissions
Output Prices
NZFARM
Soil Type
Maximise π,
subject to input
constraints
Environmental
Outputs
Soil
Erosion
Water
Yield
GHG emissions
Economic Output
Agricultural
Production
Livestock
Products
N and P
leaching
Land-Based
Profit
Environmental
Costs
Forestry
Products
Crops and
Horticultural
Products
Farmer Decision Model
• Agent-based approach
– Behaviour, traits, communication, networks,
mobility
• Farmer agents
– Satisficing approach
– Bounded rationality
• Heterogeneous farmer population
– Various typologies
• Sector, Attribute, Production, Life stage
– Information networks (social and geographic)
• Temporal changes to the networks
ARLUNZ
Farmer
Decision
Model
Geography
& Behaviour
Agent-based Rural Land
Use New Zealand model
NZ Forestry
& Agriculture
Regional Model
Economics
& Productivity
Farmer agents in ARLUNZ
• Sector based (SnB, Dairy, Forestry)
• Satisficing approach
• Economic input from GAMS
based on farm location, size, and
local and global markets
• Acceptance tempered by agent
traits, stage of life, and
information networks
Farm Generational Model
Successor
Incumbent
Burton, Forthcoming
Life Stage
Incumbent
Successor
Type of Change
Farm life-stage 1
40-45 years
15-20 years
Succession
Farm life-stage 2
45-50 years
20-25 years
Consolidation
Farm life-stage 3
50-55 years
25-30 years
Expansion
Farm life-stage 4
55-60 years
30-35 years
Transition
Farm life-stage 5
60-65 years
35-40 years
Retirement
Networks
• Social interaction & life cycle
– Succession, expansion, transition
• Social network
– Sector based
– Endorsements (Alam et al, 2010)
• Geographic network
– Spatial based
– Imitation (Schmit & Rounsevell, 2006)
ARLUNZ Design
• NetLogo
• GAMS
• Python
ARLUNZ Setup Stages
Import GIS Data
Cadastral
Land use
Zonal
ARLUNZ Setup Stages
Import GIS Data
Create Farmer
Agents
Create Farms
Create Market
Cadastral
Decision making
Static location
Global market
Land use
Parcels > 100Ha
Organisational units
Land prices
Zonal
Sector
Soil quality
Commodity prices
Networks
Sub-catchment
Other (Revenue, etc)
ARLUNZ Run Stages
Evaluate Farmers & Age
•
•
•
•
Increment age
Find Succession (if in stage)
Update farmer attributes
Update farmer networks
Market Adjustments
• Land
• Market
• Commodity
Economic Input
• Export farmer agent
information to GAMS
• Run GAMS and evaluate
economic output and
economically ‘ideal’ sector for
each farm
• Export back to NetLogo
Evaluate sale of farms
Farmers decisions
• Farms sold
• No successor
• Financial reasons
• Priority to similar sectors in
geographic proximity
•
•
•
•
Social network
Geographical network
Farmer traits
Economic information
Experimental outputs
• GhG = $0, 12.5, 25, 37.5 & 50/tCO2e
GHG Price
Farm Profit
(Million NZD)
$0/tCO2e
$12.50/tCO2e
$25/tCO2e
$37.50/tCO2e
$50/tCO2e
19.1
-2%
-3%
-5%
-8%
$0/tCO2e
$12.50/tCO2e
$25/tCO2e
$37.50/tCO2e
$50/tCO2e
21.6
1%
-2%
-2%
-1%
$0/tCO2e
$12.50/tCO2e
$25/tCO2e
$37.50/tCO2e
$50/tCO2e
23.6
14%
23%
26%
17%
GHG
Net GHG
N Leaching
Emissions
Emissions
(tons)
(Million tCO2e)(Million tCO2e)
2015
1.0
-8%
-29%
-24%
-23%
2035
1.1
-16%
-43%
-57%
-56%
2055
1.3
-15%
-37%
-48%
-61%
P Loss
(tons)
0.8
-37%
-146%
-110%
-91%
2445
-5%
-14%
-13%
-15%
22.7
-8%
-28%
-23%
-23%
0.8
-111%
-295%
-392%
-379%
2944
-6%
-19%
-26%
-23%
26.5
-15%
-42%
-56%
-54%
0.9
-122%
-281%
-367%
-461%
3474
6%
4%
2%
-12%
31.1
-10%
-28%
-38%
-54%
18
16
14
12
10
8
6
4
2
0
20
18
16
14
12
10
8
6
4
2
0
Forest Profit
$0/tCO2e
$12.50/tCO2
e
$25/tCO2e
$37.50/tCO2
e
$50/tCO2e
Sheep and Beef Profit
18
Dairy Profit
16
14
$0/tCO2e
$12.50/tCO2e
Million NZD
Million NZD
Million NZD
Experimental outputs
12
10
$0/tCO2e
8
$12.50/tCO2e
$25/tCO2e
6
$25/tCO2e
$37.50/tCO2e
4
$37.50/tCO2e
$50/tCO2e
2
$50/tCO2e
0
Experimental outputs
GhG = $0/tCO2e
GhG = $50/tCO2e
SURVEY OF RURAL DECISION-MAKERS
IN CANTERBURY, SOUTHLAND, & WAIKATO
PIKE BROWN
BRUCE SMALL
FRASER MORGAN
OSCAR MONTES DE OCA
MARGARET BROWN
CONOR LYNCH
ROB BURTON
SURVEY OF RURAL DECISION MAKERS –
DESIGN
131 questions in total
Topics:
• respondent demographics
• farm/forest/growing operation characteristics
• succession plans
• risk tolerance
• profitability
• information sources
• objectives
• management practices
• intentions
• perceived behavioural controls
• norms
• environmental attitudes
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS
131 variables
 17,030 cross tabulations
respondent demographics
operation characteristics
succession plans
risk tolerance
profitability
information sources
objectives
management practices
intentions
perceived behavioural controls
norms
environmental attitudes
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS: AGE
younger**
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS: EDUCATION
more education*
less education*
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS: ENTERPRISE
most*
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS: ENTERPRISE
SURVEY OF RURAL DECISION MAKERS –
Trustworthiness (rank)
Trustworthiness (score)
DESCRIPTIVE RESULTS
Canterbury
Southland
Waikato
total
Newspapers, general interest magazines
12
(5.18)
12
(4.82)
12
(4.83)
12
(5.01)
Television, radio
14
(4.26)
13
(4.28)
14
(4.26)
14
(4.27)
Internet
Organizations that broadly represent primary
industries (e.g., Fed Farmers)
Industry groups (Beef+Lamb, HortNZ, DairyNZ)
11
(5.33)
11
(5.19)
11
(5.14)
11
(5.25)
5
(6.43)
7
(6.11)
6
(6.34)
5
(6.33)
7
(6.31)
5
(6.16)
5
(6.48)
6
(6.31)
Cooperatives (e.g., Zespri, Fonterra)
10
(5.77)
10
(5.72)
4
(6.55)
10
(5.93)
Central government
13
(4.35)
15
(4.19)
13
(4.41)
13
(4.32)
Regional councils
15
(4.12)
14
(4.23)
15
(4.09)
15
(4.14)
Accountants and financial advisors
3
(6.88)
2
(6.68)
2
(6.78)
2
(6.81)
Farm consultants, extension officers, contractors
4
(6.49)
8
(6.10)
6
(6.34)
4
(6.36)
Farmers' forums, agricultural shows, field days
6
(6.36)
4
(6.19)
8
(6.12)
7
(6.27)
Other farmers, farmer discussion groups
2
(6.97)
3
(6.63)
3
(6.61)
3
(6.80)
Scientists
8
(6.24)
6
(6.13)
9
(5.94)
8
(6.14)
Veterinarians
1
(7.11)
1
(7.07)
1
(7.00)
1
(7.08)
Rural retailers and their technical representatives
9
(6.11)
9
(5.96)
10
(5.75)
9
(5.99)
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS: NETWORK SIZE
most*
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS: RISK TOLERENCE
statistically
indistinguishable
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS: PRODUCTION ORIENTATION
more important**
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS: INTENTIONS
more likely***
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS: MANAGEMENT













Reduce stocking rates
Reduce N fertiliser
Winter off stock
Apply DCDs
Empliy nutrient management plan
Add/upgrade irrigation
Construct feed pad
Upgrade effluent system
Fence streams
Construct wetlands / sedimentation traps
Plant forestry blocks
Plant riparian buffers
Change primary crops/rotation
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS: MANAGEMENT BY LAND USE
reduce N
fertilizer
employ a
nutrient
mgmt plan
add or
upgrade
irrigation
system
fence
streams
7.30%
48.80%
80.50%
70.70%
68.30%
intend to do
7.30%
14.60%
9.80%
7.30%
considering
19.50%
19.50%
7.30%
4.90%
0.00%
not going to
51.20%
17.10%
2.40%
0.00%
2.40%
Canterbury
already doing
25.40%
35.10%
26.90%
23.10%
34.30%
sheep, beef
intend to do
3.70%
6.70%
16.40%
9.70%
10.40%
N=134
considering
5.20%
12.70%
13.40%
9.70%
11.20%
not going to
29.90%
14.90%
14.20%
4.50%
9.70%
variable
reduce
stocking
rates
Canterbury
already doing
dairy
N=41
✓
✓
✓
✓
0.00%
✓
SURVEY OF RURAL DECISION MAKERS –
DESCRIPTIVE RESULTS: MANAGEMENT BY REGION
reduce N
fertilizer
employ a
nutrient
mgmt plan
add or
upgrade
irrigation
system
fence
streams
7.30%
48.80%
80.50%
70.70%
68.30%
intend to do
7.30%
14.60%
9.80%
7.30%
0.00%
considering
19.50%
19.50%
7.30%
4.90%
0.00%
not going to
51.20%
17.10%
2.40%
0.00%
2.40%
Waikato
already doing
22.5%
46.5%
88.7%
21.1%
83.1%
dairy
intend to do
8.5%
11.3%
5.6%
11.3%
7.0%
N=71
considering
19.7%
14.1%
1.4%
8.5%
1.4%
not going to
36.6%
21.1%
2.8%
9.9%
1.4%
variable
reduce
stocking
rates
Canterbury
already doing
dairy
N=41
✓
✓
SURVEY OF RURAL DECISION MAKERS –
MULTIVARIATE REGRESSION ANALYSIS
Dependent variable:
“How important is
preserving the ___
of fresh water? 0-10”
SURVEY OF RURAL DECISION MAKERS
✓
✓
✓
✓
theory of planned
behaviour
demographics &
land characteristics
SURVEY OF RURAL DECISION MAKERS: BEHAVIOURS
TAKEAWAY MESSAGE
Know thy decision maker
Land use is not a sufficient predictor of behaviour
Demographics of decision makers (e.g., age and gender) matter
Land characteristics (e.g., size) matters
Location matters
Attitudes matter
Perceived control matters
Social expectations matter
Network size matters
Risk tolerance matters less
Questions & Thanks
NZ/EU Marie Curie IRSES
Contact:
[email protected]