1
A game heo y model o explo e he ole o coope a ion and di e si y
1
in communi y ood secu i y: he case o Sou he n Malawi.
2
S e ano Balbi1,2, *, Unai Al a ez-Rod iguez1,3, Vi o La o a3,4,5,
3
Albe o An onioni6, Fe dinando Villa1,2
4
5
1 Basque Cen e o Clima e Change (BC3), Sede Building, Campus EHU/UPV, Leioa, Bizkaia,
6
Spain. s e ano.balbi@bc3 ese ch.o g, unai.al a ez@bc3 esea ch.o g,
7
e dinando. i[email p o ec ed] g
8
2 IKERBASQUE, Basque Founda ion o Science, Bilbao, Bizkaia, Spain.
9
3 School o Ma hema ical Sciences, Queen Ma y Uni e si y o London, London E1 4NS, UK.
10
.la o [email protected]
11
4 Dipa imen o di Fisica ed As onomia, Uni e si à di Ca ania and INFN, I-95123 Ca ania, I aly.
12
5 The Alan Tu ing Ins i u e, The B i ish Lib a y, London, NW1 2DB, UK.
13
6 G upo In e disciplina de Sis emas Complejos (GISC), Depa amen o de Ma emá icas,
14
Uni e sidad Ca los III de Mad id, E-28911, Leganés, Mad id, Spain.
15
albe o.an [email protected]
16
*Co esponding au ho
17
18
Abs ac
19
The Sus ainable De elopmen Goals aim a ending ood insecu i y by 2030. The e o e,
20
ci il socie y needs o unde s and he inhe en complexi ies o bo h socio-economic and
21
ecological dynamics and hei in e dependencies. In pa icula , he beha iou al
22
dynamics ha unde pin human agen s a e c ucial in d i ing he inal ou comes in e ms
23
o communi y ood secu i y and equi e u he a en ion. Using household beha iou
24
wi hin a u al illage o Sou he n Malawi as an example, we desc ibe a game heo y
25
model ep esen ing c opping s a egies: (1) coope a ion, as d i en by o he - ega ding
26
p e e ences, and (2) con o ma ion, he endency o con e ge o simila c op plan ing
27
choices as opposed o di e en ia ion (and hus c op di e si y). We ind ha he la e
28
plays a c ucial ole in d i ing he sys em owa ds success ul s a egies: how indi iduals
29
ela e o social no ms has g ea e e ec . Coope a ion is only necessa y o communi y
30
success when he communi y con e ges on c op plan ing choices. On he con a y,
31
di e en ia ion, he a i ma ion o he indi idual unique iden i y, can succeed wi h o
32
wi hou coope a ion. We u he elabo a e on he idea ha communi y le el sus ainabili y
33
can be eached h ough di e en pa hways, which migh equi e ood exchange
34
mechanisms wi hin and beyond he sys em bounda ies.
35
36
Keywo ds
37
human beha iou ; game heo y, social ne wo ks, di e si y, social-ecological sys ems,
38
ood secu i y.
39
Leng h o Manusc ip : 6100 wo ds and 5 igu es and 1 able.
40
This documen is he Accep ed Manusc ip e sion o a Published Wo k ha appea ed in inal o m in:
Balbi S., Al a ez-Rod iguez U., La o a V., An onioni A., Villa F. 2020. A game heo y model o explo e he ole o coope a ion and di e si y in communi y ood secu i y: he
case o Sou he n Malawi. Regional En i onmen al Change. 20. (2) DOI (10.1007/s10113-020-01642-5).).
© 2020, Sp inge -Ve lag GmbH Ge many, pa o Sp inge Na u e..
2
1. In oduc ion
1
2
Food p o isioning is a key challenge o coupled human-na u al sys ems. The
3
Sus ainable De elopmen Goals (SDGs) adop ed by he Uni ed Na ions ha e se a clea
4
a ge o end ood insecu i y by 2030 (UN 2015a). In o de o achie e his a ge
5
unde s anding he inhe en complexi ies o bo h socio-economic and ecological
6
dynamics, and hei mu ual eedbacks ac oss scales is c ucial (Os om 2008). While
7
ag o-ecological dynamics a e well explo ed in he li e a u e (e.g. Chappell and LaValle
8
2011), he beha iou al dynamics ha unde pin human agen s equi e mo e esea ch o
9
unde s and how hey d i e communi y ood secu i y, especially in u al egions o he
10
Global Sou h.
11
12
Using household c op plan ing choices wi hin a u al illage as an example, we
13
in oduce a beha iou al compass based on wo dimensions: (1) coope a ion, he deg ee
14
o which indi idual success depends on neighbou s’ success due o o he - ega ding
15
p e e ences, which in u n cap u es a communi y scale objec i e unc ion, and (2)
16
con o ma ion, he endency o con e ge owa ds simila c op plan ing choices as
17
opposed o di e en ia ion, and as a consequence, c op di e si y. We a gue ha by
18
acknowledging he in e play among hese wo dimensions in shaping household
19
c opping s a egy, a u al ag icul u al sys em can be e eo ganize, spon aneously o by
20
means o policy in e en ions, o imp o e ood secu i y a he communi y le el.
21
22
In he ollowing pa ag aphs we explain he compass axes, used as a e e ence sys em
23
o cha ac e ize household beha iou and show how he compass applies o he selec ed
24
case s udy: a u al illage o Sou he n Malawi. In Me hods, we desc ibe he con ex and
25
he da a used o he applica ion and he ma hema ical o maliza ion o he model. We
26
also desc ibe i s beha iou as we change he main household pa ame e s (o he -
27
ega ding p e e ences and homophily) wi hin hei in e al o e e ence. In Resul s, we
28
assess he model pe o mance when he game pa ame e s (numbe o households,
29
numbe o c ops and ne wo k opology) a e se o ma ch he case s udy cha ac e is ics,
30
de i ing he household beha iou ha deli e s he bes ou comes. In Discussion, we
31
elabo a e on he implica ions o he model indings o he communi y ood secu i y.
32
Finally, we conclude wi h he key messages o his s udy.
33
34
Inspi ed by he poli ical compass o Maddox and Lilie (1984) and Les e (1994), we e e
35
o a beha iou al compass (Fig. 1), a ou by ou g id whe e he ho izon al axis
36
ep esen s economic beha iou and displays compe i ion a i s le side and ull
37
coope a ion a i s igh side, while he e ical axis ep esen s pe sonal beha iou and
38
displays di e en ia ion a he bo om end and con o ma ion a i s op end.
39
3
Economic beha iou desc ibes how he indi idual ela es o he o he s o he pu pose
1
o achie ing esou ces, including consump ion and u ilisa ion o na u al esou ces.
2
Coope a ion can be unde s ood as he p ocess o ac ing in coo dina ion wi h o he s o
3
maximise indi idual o mu ual bene i (Pe c e al. 2017). Compe i ion a ises when a
4
leas wo pa ies s i e o a goal ha canno be sha ed, o is desi ed indi idually bu no
5
in sha ing.
6
Pe sonal beha iou desc ibes how indi idual belie s and alues ela e o social no ms.
7
Con o ma ion is he p ocess o ma ching indi idual beha iou wi h he dominan social
8
no ms, while di e en ia ion—as in psychological di e en ia ion (Wi kin e al. 1974)—is
9
he p ocess o a i ming an indi idual unique iden i y agains social no ms. Thus, he
10
highe di e en ia ion, he highe he di e si y o beha iou s wi hin he sys em.
11
12
In an ea ly example o a simila app oach, Deu sch (1949) iden i ies h ee ypes o “goal
13
s uc u es on achie emen ”: coope a i e, compe i i e and indi idualis ic. Coope a ion
14
and compe i ion ha e been ex ensi ely esea ched in ecology and economics (game
15
heo y, in pa icula : see Feh and Fischbache 2002) as indi idual ope a ing modes
16
( om he seminal wo k o Axel od and Hamil on 1981 o mo e ecen e iews, such as
17
Pe c e al. 2017). Di e en ia ion and con o ma ion belong o he domain o beha iou al
18
and expe imen al me hods in psychology and economics and, in he con ex o his
19
s udy, a e pa icula ly close o he esea ch on how social no ms and he aming o
20
op ions in luence indi idual beha iou , wi h signi ican epe cussions o human
21
coope a ion and na u al esou ce exploi a ion (Kahneman 2011).
22
23
24
Fig. 1 The beha iou al compass
25
4
2. Me hods
1
2
In his a icle, we e e o communi y ood secu i y as an eme gen p ope y o
3
household ood secu i y, which is a consequence o how households in e ac , acqui e
4
and u ilise asse s, including na u al esou ces (Chambe s and Conway 1991). In mos
5
cases, he li elihoods o u al households emain la gely dependen on ag icul u e. This
6
a icle explo es da a o a u al egion loca ed in he sou he n pa o Malawi, a small,
7
landlocked A ican coun y home o app oxima ely 15 million people. Mo e han 90% o
8
he u al popula ion a e smallholde a me s, esponsible o cul i a ing plo s wi h an
9
a e age size o jus 0.8 hec a es (NSO 2012). Maize is he socially p e e ed c op
10
among smallholde s and he main s aple die o he popula ion (NSO 2012). Acco ding
11
o Chinsinga and Chasukwa (2012), al hough o e 97% o smallholde a me s g ow
12
maize, only 10% a e ne selle s and up o 60% a e ne buye s.
13
Typically, mos o he households wi hin a u al illage a e a leas acquain ed wi h each
14
o he and i is common p ac ice o exchange on- a m labou and ood dona ions in
15
pe iods o need (Dobbie e al. 2018). Simila dynamics ha e been wi nessed in u al
16
se ings a ound he globe (Pa el 2009) including: sha ing o unused ing edien s ha
17
would o he wise be ashed, sha ing o knowledge abou a ming p ac ices, good
18
nu i ion and ecipes, in i a ions o join meals, exchange o seeds and ood commodi ies
19
o inc ease di e si y o die , exchange o po ions o meals in di e en pe iods in ime o
20
cushion empo a y sca ci y (i.e. ood banking). In Malawi, his sys em o in o mal social
21
exchanges coexis s wi h a ma ke -o ien ed s uc u e ha hea ily elies on local and
22
egional ood ma ke s o he in eg a ion and edis ibu ion o ood commodi ies, as a
23
consequence o he libe alisa ion e o ms p omo ed by he In e na ional Mone a y Fund
24
and he Wo ld Bank (Do wa d and Kydd 2004).
25
26
2.1 The case o ood secu i y in Sou he n Malawi
27
In Malawi o e 50% o he popula ion li e on less han one US dolla a day and he
28
p opo ion o ul a-poo people (de ined as he p opo ion o popula ion below he
29
minimum le el o die a y ene gy equi emen ) is highes wi hin Sou he n Malawi a
30
app oxima ely 34.2% (Gondwe 2014). Food secu i y is pa icula ly p oblema ic in u al
31
a eas whe e ag icul u e is p ima ily ain- ed, lea ing smallholde s ulne able o clima ic
32
shocks (Sahley e al. 2005). A single ainy (g owing) season, be ween he mon hs o
33
No embe and Ma ch, is ollowed by a d y season om Ap il o Oc obe . Only a limi ed
34
numbe o a ming households wi h access o dimba ields o he alleys loca ed a he
35
sou ce o s eams, c eeks, o i e s may ake ad an age o esidual mois u e and
36
ex end cul i a ion beyond he end o he ains (O e al. 2009). High popula ion
37
densi ies, small a e age plo size and poo soil quali y u he inc ease ood insecu i y.
38
The u al popula ion is an icipa ed o g ow om app oxima ely 8.4 million in 1990 o
39
almos 20 million by 2030 (UN 2015b) and his g ow h will ha e wide anging impac s
40
5
upon land and labou a ailabili y as well as ma ke p ices and p oduc i i y. Addi ional
1
exogenous ends such as soil deg ada ion and clima e change u he compound ood
2
insecu i y (Schmidhube and Tubiello 2007).
3
4
Indeed, ood secu i y is mul idimensional issue (Connolly-Bou in and Smi 2016). A o al
5
o ou dimensions a e ecognized unde he “ ou pilla s” amewo k—c ea ed by FAO
6
and ope a ionalised o modelling pu poses in Dobbie and Balbi (2017)—including ood
7
a ailabili y, access, u ilisa ion and s abili y. The p oduc ion o ood is ela ed p ima ily o
8
ood a ailabili y (Headey and Ecke 2013). Access e e s o he amoun o ood a
9
household can p oduce, pu chase om he ma ke and/o de i e om o he means
10
(Bu chi and De Mu o 2016). Households migh d aw upon social sa e y ne s such as
11
ood o wo k p og ammes o adop coping s a egies like selling li es ock o bo owing
12
ood (De e eux 2016). A hi d dimension, u ilisa ion, e e s o he abili y o households
13
o p ocess accessible ood. This is dependen upon he household abili y o ob ain
14
su icien quan i ies o uel and clean wa e . Finally, s abili y dic a es how obus
15
a ailabili y, access and u ilisa ion dimensions a e o shocks and s esses o e ime
16
(Bu chi and De Mu o 2016), such as hose ela ed o clima ic and demog aphic change.
17
This a icle ocuses p ima ily on he i s wo pilla s: a ailabili y and access.
18
19
In he ollowing pa ag aph we p esen he al e na i e p inciples o he beha iou al
20
compass applied o he c opping s a egies o a ming households, and illus a e how
21
he combina ion o hese p inciples is ele an o unde s anding communi y ood
22
secu i y (Fig. 1). The ou s o ylines shown in Table 1, which desc ibe obse ed
23
household c opping beha iou s in u al egions o he Global Sou h, a e syn hesized
24
om he Pa icipa o y Ru al App aisals (PRA: Sch eckenbe g e al. 2016) ha ook
25
place in mul iple illages o he Zomba Region, in Sou he n Malawi.
26
27
Acco ding o he compe i ion and con o ma ion s o yline (Table 1), a household would
28
maximise he yield o maize using all he a ailable inpu s subsidized by he go e nmen
29
(e.g. hyb id maize seeds and chemical e ilize ), and sell he p oduce a ma ke . This is
30
cu en ly he mos common s a egy wi hin he illages o Sou he n Malawi, whe e
31
libe alisa ion e o ms ha e ailed o he p oduce he in ended imp o emen s and he
32
social no ms ega ding wha o g ow a e widely and olun a ily accep ed. Food
33
p e e ences a e key in his ega d. Maize has become he main ood asse in he local
34
ma ke s and is conside ed a s a egic commodi y. This social no m is no only
35
de imen al in e ms o die a ie y and heal hy nu i ion, bu also makes he egion mo e
36
a isk in he ace o clima e change by imposing a clima e- ulne able c op o he
37
majo i y. Wi h coope a ion and con o ma ion a household would plan maize and sha e
38
he p oduce wi h he neighbou s ha ha e p o ided he labou o a m he land. This
39
used o happen mo e o en in he pas when he a m clubs we e popula , be o e he
40
6
e o ms, al hough mainly o cash c ops. Wi h coope a ion and di e en ia ion a
1
household would maximise c op di e si y choosing a mixed c opping pa e n wi h
2
di e en p opo ions o maize, pigeon pea, swee po a o, cassa a, ui ees, and
3
ege ables. Wi h di e en ia ion and compe i ion a household would maximise p o i by
4
plan ing cash c ops, such as obacco, peppe s, o co on. The emphasis on p o i a he
5
han on nu ien s makes i a isky pa adigm om a ood secu i y pe spec i e, especially
6
i we assume ha sel -su iciency is ele an a he illage le el, as oppo uni y cos may
7
d i e ag icul u al en ep eneu s o in es on non- ood c ops beyond he social op imal
8
le el (e.g. Ande man e al. 2014).
9
10
Table 1. Beha iou al compass: illage ood secu i y s o ylines.
11
2nd Quad an : Compe i ion & Con o ma ion
The household p oduces he socially
p e e ed c op, maximizing possible
yield in o de o ha e mo e commodi y
o be exchanged a he ma ke . Seed
and e ilize subsidies in luence his
beha iou .
1s Quad an : Coope a ion & Con o ma ion
The household p oduces he socially
p e e ed c op, a e which he p oduce
is agg ega ed o ha o o he
p oduce s belonging o a a m club,
and sha ed equi ably.
3 d Quad an : Compe i ion & Di e en ia ion
The household chooses a c opping
pa e n ha maximizes income, wi h
he objec i e o ob aining a mo e
aluable commodi y (di e si ied
quali y) o be exchanged a he ma ke
(e.g. cash c op).
4 h Quad an : Coope a ion & Di e en ia ion
The household plan s c ops in o de o
maximize c op di e si y, di e en ia ing
c opping pa e n s. i s neighbou s.
La e ood exchanges inc ease die
di e si y.
12
13
2.2 Da a and model o maliza ion
14
Household-le el da a o a illage in Sou he n Malawi was collec ed o e a pe iod o
15
ou days in July 2015 as pa o he la ge esea ch p ojec (Sch eckenbe g e al.
16
2016). Fou ained Malawian enume a o s used a household ques ionnai e o collec
17
in o ma ion on a ming p ac ices, c ops plan ed, ha es ed and sold, o he income
18
gene a ing ac i i ies, pe cei ed ood secu i y, and socio-demog aphic cha ac e is ics o
19
he households. A e a illage mapping exe cise, in which h ee illage ep esen a i es
20
lis ed he household heads and mapped hei loca ions, all households (N = 46) we e
21
selec ed (census) o pa icipa e. This same da ase was also used o calib a e an agen -
22
based model o simula e he communi y ood secu i y in o he u u e (Dobbie e al.
23
2018). In his s udy we u ilize he in o ma ion on he c ops plan ed, ha es ed and sold,
24
which include: maize, g oundnu s, ube s, pigeon peas, peas, so ghum, beans, soy,
25
ege ables, and co on.
26
We p opose a game heo y model o desc ibing he household c opping s a egies. We
27
assume ha each household is able o change i s main c op ype, a each ime s ep,
28
7
and ha he e is only one plan ing season pe yea . The c op selec ion depends on wo
1
main ac o s, which desc ibe pe sonal and economic beha iou , as pe beha iou al
2
compass. The i s one is homophily, which de e mines he household endency o
3
imi a e i s neighbou s and is he e o e he pa ame e co esponding o he
4
di e en ia ion-con o ma ion axis (i.e. pe sonal beha iou ). The second one desc ibes
5
o he - ega ding p e e ences, he in luence o communi y le el sa is ac ion in he
6
household decision, and co esponds o he compe i ion-coope a ion axis (i.e. economic
7
beha iou ). The main ules o he game a e gi en in he ollowing bulle poin s:
8
9
● The opology o in e ac ions is gi en by an ini ial ne wo k o N nodes, each node
10
ep esen ing one household, whose connec ions, ep esen ed in a ma ix A wi h
11
elemen s 𝑎𝑖𝑗, emain cons an in ime.
12
● Each node has a s a e, σi, which ep esen s he household chosen main c op,
13
and is upda ed asynch onously a each ime s ep.
14
● The s a e, σi ∈ {1, S}, wi h S being he o al numbe o c ops, changes a each
15
ime s ep acco ding o he bes esponse mechanism. This, implies ha each
16
household es s all he possible c ops and selec s he one ha maximizes i s
17
payo . C ops a e ini ialized acco ding o a uni o m dis ibu ion.
18
● The o he - ega ding p e e ences pa ame e α ∈ [0, 1], weigh s he impo ance o
19
he communi y in he s a egy decision o each household. I s alue is ixed in
20
ime.
21
● The homophily pa ame e , h ∈ [0, 1] ep esen s he desi ed p opo ion o
22
neighbou s ma ching he same c op plan ing choices. The homophily pa ame e
23
is also ixed in ime.
24
● A each ime s ep all households upda e hei s a e σi o maximize hei payo
25
(las bulle poin ). The o al numbe o ime s eps o i e a ions is gi en by T.
26
● The sa is ac ion unc ion si ∈ [0, ki], whe e ki is he deg ee o node i, depends on
27
he dis ance ∆i, which measu es how close a e he links o a node om hei
28
op imal con igu a ion. In o he wo ds, i e ie es he dis ance wi h espec o he
29
numbe o connec ed households wi h same c op, acco ding o h.
30
31
(1) 𝛥𝑖 = |ℎ − 1
𝑘𝑖 ∑
𝑗𝛿𝜎𝑖𝜎𝑗 𝑎𝑖𝑗|
32
33
(2) 𝑠𝑖= 𝑘𝑖 𝑒 −∆𝑖
34
35
● The payo unc ion Πi ∈ [0, max ki] depends on he indi idual and he communi y
36
sa is ac ions, weigh ed by he indi idual o he - ega ding p e e ences pa ame e .
37
8
The e o e he payo is no a pu ely economic measu emen , bu i depends on
1
bo h economic and pe sonal beha iou .
2
3
(3) 𝛱𝑖= (1 − 𝛼)𝑠𝑖 + 𝛼
𝑁 𝛴 𝑠𝑖
4
5
The homophily pa ame e can cap u e bo h coo dina ion (h ≈ 1) and an i-coo dina ion
6
dynamics (h ≈ 0) and he si ua ions in be ween. The las wo poin s indica e ha
7
indi idual sa is ac ion is subjec i e o each household and d i en by own le el o
8
homophily: own sa is ac ion is assessed in compa a i e e ms. O he households a e
9
conside ed in e ms o hei s a e (i.e. he plan ed c op) and we assume ull knowledge
10
o nodes s a e wi hin he ne wo k.
11
Payo combines indi idual and communi y sa is ac ion; he e he o he - ega ding
12
p e e ences pa ame e is key: he highe he mo e impo an will be he sa is ac ion o
13
o he s. This o maliza ion allows he conside a ion o wo scales o conce n (indi idual
14
and communal) in he objec i e unc ion d i ing household beha iou .
15
We conside coope a ion as d i en by he impo ance assigned o he communi y in
16
e ms o neighbou s’ sa is ac ion. In his amewo k, high o he - ega ding p e e ences (α
17
≈ 1) means o adap c op selec ion so ha neighbou s’ sa is ac ion can maximize
18
indi idual payo in addi ion o indi idual sa is ac ion. Low o he - ega ding p e e ences
19
(α ≈ 0) means o no conside neighbou s’ sa is ac ion. Rega dless o his pa ame e all
20
households maximize hei own payo and hus a e modelled as sel -in e es ed
21
indi iduals.
22
23
In he ollowing sec ion, we desc ibe he model beha iou by means o nume ical
24
simula ions on a i icial ne wo ks o di e en kinds. Then, in Resul s, we apply he model
25
o he da a o a illage in Sou he n Malawi and summa ize he main indings.
26
27
2.3 Model beha iou
28
The game heo y model has h ee con ol pa ame e s, namely he o he - ega ding
29
p e e ences pa ame e α, he homophily pa ame e h, and he numbe o possible node
30
s a es S, which ep esen he main c op ypes. The model can be implemen ed on
31
di e en ne wo k opologies (La o a, Nicosia and Russo 2018). He e we compa e h ee
32
ne wo ks, namely Regula La ices (RL), Random Ne wo ks (RN) and Scale F ee
33
Ne wo ks (SF), wi h he same numbe N = 100 o nodes and he same a e age deg ee
34
<k> = 4. Regula La ices o <k> = 4 co espond o a uni o m dis ibu ion o he nodes in
35
he plane, whe e each node is connec ed o i s wo nea es neighbou s in each
36
dimension. In o de o ha e <k> = 4 o all he nodes, we conside ed pe iodical
37
bounda ies. Random Ne wo ks (E dos and Rényi 1960) o N nodes and <k> = K a e
38
achie ed by ixing he link p obabili y o K/(N-1). Thei deg ee dis ibu ion can be well
39
app oxima ed wi h a Poisson p obabili y dis ibu ion. Scale F ee Ne wo ks (Ba abási
40
9
and Albe 1999) a e cha ac e ized by a powe law decay in he deg ee dis ibu ion,
1
allowing he exis ence o hubs, o nodes wi h a high deg ee.
2
Simula ions ha e been pe o med o α ∈ {0, 0.5, 1}, h ∈ {0, 0.5, 1}, and S ∈ {2, 3, 4, 5},
3
and he esul s ha e been a e aged o e ensembles o 100 di e en ne wo k
4
ealiza ions. In each simula ion we ha e un he game o T = 20 i e a ions unde a bes
5
esponse upda e ule, which assumes ha each o he nodes es s all he possible
6
s a es and selec s he one wi h he highes payo . The alue o T = 20 is enough o
7
gua an ee he con e gence o he sys em o he equilib ium s a e. The combina ion o
8
he bes esponse upda e mechanism and he small size o ne wo k explains why mos
9
o he in o ma ion o he equilib ium solu ion is p o ided o small alues o T (e.g. T =
10
5).
11
12
The asymp o ic alues o payo and domina ing s a e ac ion a e shown in Fig. 2 and
13
Fig. 3. The i s is he alue o payo when he sys em app oaches equilib ium, he la e
14
ep esen s he p opo ion o households wi hin he illage ma ching he same c op
15
plan ing choices. Bo h igu es show he a e age alues o ou pu as a unc ion o o he -
16
ega ding p e e ences and homophily o di e en opologies and alues o S. The
17
columns co espond o he opology, RL, RN and SF, while he ows co espond o he
18
o al numbe o s a es. In each subplo he colou in he whi e-blue scale ep esen s he
19
alue o he payo o he speci ic selec ion o α and h.
20
21
F om he simula ions we can in e he mos ele an ea u es o he model:
22
● The ela ion be ween S (numbe o s a es/c ops) and <k> (a e age deg ee o he
23
ne wo k) de e mines he s a egy wi h he highes payo . Fo low S/<k>,
24
con o ma ion is mo e a ou able han di e en ia ion. The opposi e is ue o
25
highe alues o S/<k>.
26
● The sys em is pola ized be ween consensus (high domina ing s a e ac ion (d )),
27
o h = 1, and dissension (low d ), o h = 0. The in e media e si ua ions imply
28
lowe payo s. This indica o sugges s he exis ence o a phase ansi ion
29
be ween hese wo egimes.
30
● Coope a ion is especially ele an when he communi y con e ges on c op
31
plan ing choices: i is a necessa y ing edien o he success o households ha
32
con o m. The opposi e is no ue. Di e en ia ion can su i e wi hou coope a ion.
33
● The dynamics a e s able excep o he combina ion o h = 0.5 and α ∈ {0.5, 1}.
34
We expand on his inding when s udying he empi ical case s udy. The highe S,
35
he lowe he luc ua ions. The endency owa ds s abili y depends on he
36
opology. RL achie e i as e han SF, and hese a e as e han RN (See Online
37
Resou ce 1: Fig. 6 and Fig. 7. He e we plo he absolu e alue o he disc e e
38
de i a i e a T = 20, as a mechanism o e alua ing i he esul co esponds o
39
an equilib ium si ua ion).
40
16
I is easonable o hink ha ha ag i- ood sys ems can be mapped wi hin a con inuum
1
be ween hese wo pa adigms and dynamically mo e in one di ec ion o he o he
2
acco ding o social and ecological condi ions. Fo example, a coun y o he Global
3
Sou h like Malawi has been wi nessing a shi owa ds IFA unde he libe alisa ion
4
e o ms ha happened since he eigh ies (Do wa d and Kydd 2004). A he same ime
5
NFA is gaining momen um in he ci ies o he Global No h, whe e he masses ha e
6
been elying on he IFA pa adigm, almos since he indus ial e olu ion (Ilie a 2016).
7
8
In his s udy we adop ed a illage-cen ic pe spec i e and a sys em app oach, whe eby
9
he sys em pe o mance is s emming om he beha iou o and in e ac ions among he
10
indi idual componen s, o explo e how household beha iou can a ec ood secu i y a
11
he communi y le el. We implici ly conside ed ha sel -su iciency in ood p oduc ion is a
12
posi i e ea u e o a u al ag i- ood sys em o he Global Sou h, bu he e we discuss i s
13
consequences.
14
15
Because sa is ac ion, and hus payo , is no measu ed in e ms o nu i ional
16
equi emen s we can only sugges how human beha iou is expec ed o in luence ood
17
secu i y. Indeed, ou model is p ima ily conce ned wi h beha iou al dynamics in c op
18
adop ion and he le el o dissension/consensus wi hin he communi y. To d aw
19
conclusions on he nu i ional le el o communi y ood secu i y, we would ha e o couple
20
his beha iou al model wi h ag icul u al and en i onmen al modules (see Balbi e al.
21
2015) and a module o exchanges, including ma ke in e ac ions and ba e ing
22
dynamics, possibly conside ing ood impo s and expo s. In he wo k p esen ed he e
23
we assume ha all he households ha e he same nu i ional equi emen s, and ha he
24
p oduc ion cos s and economic p o i a e equal o all c op ypes. We also don’
25
conside any di e ence in he c ops ha a e consumed by he p oduce s and he ones
26
ha a e sold in ma ke . Accessing ood ia p oduc ion o ia ma ke doesn’ a ec he
27
bene i s and ood p ice is exac ly he p oduc ion cos , hus he communi y is assumed o
28
ope a e h ough di ec ood exchanges.
29
30
Al hough pas s udies ha e emphasized he ole o coope a ion in sel -o ganized
31
sys ems (see Pe c e al. 2017 o a e iew), ou game heo y model sugges s ha
32
homophily— he con inuum be ween con o ma ion and di e en ia ion—can play a
33
ele an ole. Bo h con o ma ion and di e en ia ion can be success ul s a egies leading
34
o high payo s a he indi idual le el, bu acco ding o which p e ails, hey d i e he
35
sys em o e y di e en communi y ou comes. In he i s case, we ha e a communi y
36
con e ging on c op plan ing choices, in he second case he communi y main ains c op
37
di e si y.
38
39
17
Bo h cases can happen ega dless o he impo ance gi en o o he s sa is ac ion, which
1
is wha d i es coope a ion in ou model. This can ha e signi ican consequences in
2
e ms o ood secu i y a he illage le el. The i s case is a specializa ion case ha
3
could succeed only i he illage is connec ed o he ou side ia ood exchange
4
mechanisms (e.g. egional ood ma ke s) h ough which households can sell hei own
5
p oduce and buy he lacking nu ien s. The al e na i e is a communi y ha can pe haps
6
sel -su icien ly mee i s own calo ic needs, bu no a heal hy die di e si y. The second
7
case is a di e si ica ion case ha can succeed unde a sel -su iciency scena io, bu
8
equi es in e nal exchanges o happen ( ia local ma ke o ba e ing) wi hin he illage
9
o mee he necessa y die di e si y a he household le el.
10
11
Howe e , he case s udy we analysed in his a icle, using empi ical da a om a u al
12
illage in Sou he n Malawi, has p o ided mo e insigh s abou he ole o coope a ion.
13
Ou analysis sugges s ha he illage is loca ed in a e y pa icula egion o he
14
pa ame e s space whe e he in luence o o he - ega ding p e e ences is ela i ely high
15
o any poin in he cu e p esen ed in Figu e 5. This cu e ac s as he bounda y
16
be ween he egions ep esen ing he wo cases men ioned abo e. Thus coope a ion
17
has a key ole o play: he highe o he - ega ding p e e ences he mo e he sys em is
18
d i en o consensus, and he analysis o expec ed payo s sugges s ha he illage is
19
be e o wi h consensus. Pa ial dissension can su i e o o he - ega ding p e e ences
20
below 0.9. I hus seems ha sac i icing some deg ee o coope a ion in a ou o mo e
21
c op di e si y would educe he indi idual expec ed payo , al hough i would make
22
sense om a nu i ional di e si y pe spec i e a he communi y le el. Indeed, he cu en
23
illage si ua ion is con i ming hese insigh s: in his pa o Malawi ood exchanges a e
24
la gely ma ke -d i en wi h egional ma ke s playing an impo an ole and maize is by a
25
he socially p e e ed c op (Sch eckenbe g e al. 2016). Mo eo e , he esul s ein o ce
26
he idea o close o pe ec knowledge among households a he illage scale, and he
27
con enience o conside ing o he s sa is ac ion in own payo , hus being mo e
28
coope a i e.
29
30
One addi ional limi a ion o ou wo k is he assump ion abou he eal opology o he
31
illage ne wo k. Al hough we expec an e o o app oxima ion o he ac ual social
32
ne wo k, ou assump ion is based on p e ious s udies and he sensi i i y analysis in
33
Sec ion 2.3 explains he simila i y o model beha iou when he opology is al e ed.
34
Acco dingly, we expec he ela ion be ween he o he - ega ding p e e ences and he
35
homophily pa ame e s o be obus agains changes in he opology. Fo wha conce ns
36
illage-le el esul s, we expec hem o hold o small communi ies (e.g. 30 o 100
37
households co esponding o 100-500 indi iduals) wi h a ela i ely high deg ee o
38
mu ual knowledge among communi y membe s and pa celled access o he esou ce—
39
in his case a ming land.
40
18
Fu he analysis will explo e he explana o y powe o he beha iou al compass a
1
mul iple spa ial scales. We a gue ha he beha iou al compass could be ex ended on a
2
hi d axis, hus ep esen ing a h ee dimensional space, o cap u e mul iple le els o
3
go e nance (Lebel e al. 2006) and hei ins i u ions ( om he mic o- o mac o-scale), in
4
line wi h he hinking o Wa ing e al. (2015).
5
6
5. Conclusions
7
8
The main inding o ou simula ions sugges s ha indi idual and communi y success is
9
mo e s ongly ela ed o pe sonal beha iou han o economic beha iou . In o he
10
wo ds, how indi iduals ela e o social no ms has g ea e e ec han how much hey
11
ca e abou he success o hei neighbou s (i.e. ou wo king de ini ion o coope a ion).
12
While we expec ed he la e o be he main d i ing o ce in de e mining communi y
13
success, we ound ha coope a ion is only necessa y o communi y success when he
14
communi y con e ges on c op plan ing choices. On he con a y, di e en ia ion, he
15
a i ma ion o he indi idual unique iden i y, can succeed wi h o wi hou coope a ion.
16
17
Unde a ood secu i y pe spec i e, when he e is a di e si y o op ions abou c op
18
adop ion, di e en ia ion is likely o deli e mo e posi i e ou comes. Only when he
19
op ions a e ew, hen con o ma ion is a winning s a egy. In a si ua ion such as a u al
20
egion o he Global Sou h—in his a icle we used he example o a illage in Sou he n
21
Malawi— his ansla es in o clea policy implica ions abou he ole o ood so e eign y,
22
and in pa icula ood knowledge sha ing. Food secu i y, among o he hings, depends
23
on a a ie y o nu ien s o a heal hy and balanced die , hus enabling biological
24
di e si y in ag icul u e can be bene icial.
25
Ins ead, i seems ha he policy e o ms o he pas ha e pushed coun ies like Malawi,
26
simila ly o o he u al A ican egions, o specialize hei ag icul u al p oduc ion owa ds
27
one o wo socially p e e ed c ops, ia c op-speci ic subsidies and o he incen i es.
28
This made sense unde a scena io o ma ke libe alisa ion ha was going o deli e
29
e icien edis ibu ion o ood commodi ies among di e en egions ( o ood secu i y
30
pu poses), hus allowing beyond-sys em ood ansac ions. E idence om he g ound
31
sugges s ha his pa adigm has only pa ially deli e ed i s p omises (see Do wa d and
32
Kydd 2004) and ha s imula ing mo e di e se c op adop ion a he household le el is a
33
easonable s a egy o imp o e communi y ood secu i y.
34
We a gue ha his p ocess is no likely o happen endogenously, because, in he
35
cu en se ing, households ega d con o ma ion as a sa e s a egy, due o he
36
dominan social no ms on c ops.
37
38
19
Acknowledgmen s
1
This esea ch is suppo ed by he Basque Go e nmen h ough he BERC 2018-2021 p og am
2
and by he Ike zaile Dok o een zako Hobekun za ako dok o e za-ondoko P og ama, and by
3
Spanish Minis y o Economy and Compe i i eness MINECO h ough BC3 Ma ía de Maez u
4
excellence acc edi a ion MDM-2017-0714 and by he Le e hulme T us Resea ch Fellowship
5
“CREATE: he ne wo k componen s o c ea i i y and success. This wo k was also inspi ed and
6
pa ially unded by he ASSETS (A aining Sus ainable Se ices om Ecosys ems) p ojec
7
(NERC/ESPA NE-J002267-1). The au ho s wish o hank he ad ice and suppo o Ma ije
8
Schaa sma (Uni e si y o Sou hamp on), who p o ided he da ase analysed, Ka e
9
Sch eckenbe g (King's College London), who led he wo k on he pa icipa o y u al app aisals
10
wi hin he ASSETS p ojec , Saman ha Dobbie o he ex ensi e wo k on sus ainable li elihoods
11
achie ed du ing he doc o al esea ch a he Uni e si y o Sou hamp on, and O zi Akizu-
12
Ga doki (Uni e si y o he Basque Coun y) o he design o Figu e 1.
13
14
Re e ences
15
Ande man TL, Remans R, Wood SA, DeRosa V, DeF ies RS (2014) Syne gies and adeo s
16
be ween cash c op p oduc ion and ood secu i y: a case s udy in u al Ghana. Food
17
Secu i y 6(4):541–554. doi: 10.1007/s12571-014-0360-6
18
Axel od R., Hamil on WD (1981) The e olu ion o coope a ion. Science 211(4489):1390–1396.
19
doi: 10.1126/science.7466396
20
Balbi S, del P ado A, Gallejones P, Gee an CP, Pa do G, Pé ez-Miñana E, Man ique R,
21
He nandez-San iago C, Villa F (2015) Modeling ade-o s among ecosys em se ices in
22
ag icul u al p oduc ion sys ems. En i on Modell So w 72:314-326. doi:
23
10.1016/j.en so .2014.12.017
24
Ba abási AL, Albe R (1999) Eme gence o scaling in andom ne wo ks. Science 286(5439):
25
509-512. doi: 10.1126/science.286.5439.509
26
Bu chi F, De Mu o P (2016) F om ood a ailabili y o nu i ional capabili ies: Ad ancing ood
27
secu i y analysis. Food Policy 60:10–19. Doi: 10.1016/j. oodpol.2015.03.008
28
Chambe s R, Conway G (1991) Sus ainable u al li elihoods: P ac ical concep s o he 21s
29
cen u y. IDS Discussion Pape 296.
30
h p://opendocs.ids.ac.uk/opendocs/handle/123456789/775 Accessed 09 Ma ch 2020.
31
Chappell MJ, LaValle LA (2011) Food secu i y and biodi e si y: can we ha e bo h? An
32
ag oecological analysis. Ag ic Hum Values 28(1):3-26. doi: 10.1007/s10460-009-9251-4
33
Chinsinga B, Chasukwa M (2012) You h, ag icul u e and land g abs in Malawi. IDS Bulle in
34
43(6): 67–77. doi: 10.1111/j.1759-5436.2012.00380.x
35
Connolly-Bou in L, Smi B (2016) Clima e change, ood secu i y, and li elihoods in Sub-Saha an
36
A ica. Reg En i on Change 16(2):385–399. doi: 10.1007/s10113-015-0761-x
37
Deu sch M (1949) An expe imen al s udy o he e ec s o coope a ion and compe i ion upon
38
g oup p ocess. Human Rela ions 2(3):199–231. doi:10.1177/001872674900200301
39
De e eux S (2016) Social p o ec ion o enhanced ood secu i y in Sub-Saha an A ica. Food
40
Policy 60: 52–62. doi: 10.1016/j. oodpol.2015.03.009
41
Dobbie S, Sch eckenbe g K, Dyke J, Schaa sma M, Balbi S (2018). Agen -based modelling o
42
assess communi y ood secu i y and sus ainable li elihoods. Jou nal o A i icial
43
Socie ies and Social Simula ion 21(1):1-23. doi: 10.18564/jasss.3639
44
20
Dobbie S, Balbi S (2017) Design o an empi ical agen -based model o explo e u al household
1
ood secu i y wi hin a de eloping coun y con ex . In: Jage W, Ve b ugge R, Flache A,
2
de Roo G, Hoogduin L, Hemel ijk (eds.), Ad ances in Social Simula ion 2015, Cham:
3
Sp inge , pp 81–94. doi: 10.1007/978-3-319-47253-9_7
4
Do wa d A, Kydd J (2004) The Malawi 2002 ood c isis: he u al de elopmen challenge. The
5
Jou nal o Mode n A ican S udies 42(03):343–361. doi:10.1017/S0022278X04000229
6
E dos P, Rényi A (1960) On he e olu ion o andom g aphs. Publ Ma h Ins Hung Acad Sci
7
5(1):17-60.
8
Feh E, Fischbache U (2002) Why social p e e ences ma e – he impac o non-sel ish mo i es
9
on compe i ion, coope a ion and incen i es. The economic jou nal 112(478):C1–C33.
10
doi: 10.1111/1468-0297.00027
11
Gliessman, S. (2009) Edi o ial: A F amewo k o he Con e sion o Food Sys em Sus ainabili y.
12
Jou nal o Sus ainable Ag icul u e 33(1):1-2. doi: 10.1080/10440040802587249
13
Gondwe GE (2014) 2014 Millennium De elopmen Goal Repo o Malawi. Minis y o Finance,
14
Economic Planning and De elopmen , Lilongwe, Malawi.
15
h p://www.undp.o g/con en /dam/malawi/docs/gene al/Malawi_MDG_Repo _2014.pd
16
Accessed 09 Ma ch 2020.
17
Headey, D, Ecke O (2013) Re hinking he measu emen o ood secu i y: F om i s p inciples
18
o bes p ac ice. Food Secu i y 5(3):327–343. doi: 10.1007/s12571-013-0253-0
19
Ilie a, RT (2016) U ban ood planning: seeds o ansi ion in he global no h. Rou ledge.
20
Kahneman, D (2011) Thinking, as and slow. Macmillan.
21
K emen C, Miles A (2012) Ecosys em se ices in biologically di e si ied e sus con en ional
22
a ming sys ems: bene i s, ex e nali ies, and ade-o s. Ecology and Socie y 17(4):40.
23
doi: 10.5751/ES-05035-170440
24
La o a V, Nicosia V, Russo G (2018) Complex Ne wo ks: P inciples, Me hods and Applica ions.
25
Camb idge Uni e si y P ess.
26
Lebel L, Ande ies JM, Campbell B, Folke C, Ha ield-Dodds V, Hughes TP, Wilson J (2006)
27
Go e nance and he capaci y o manage esilience in egional social-ecological sys ems.
28
Ecology and Socie y 11(1):19.
29
Les e JC (1994) The e olu ion o he poli ical compass (and why libe a ianism is no igh -
30
wing). Jou nal o Social and E olu iona y Sys ems 17(3):231–241.
31
Ligon E, Schech e L (2012) Mo i es o sha ing in social ne wo ks. J De Econ 99(1):13–26.
32
doi: 10.1016/j.jde eco.2011.12.002
33
Maddox WS, Lilie SA (1984). Beyond libe al and conse a i e: Reassessing he poli ical
34
spec um. Ca o Ins i u e.
35
NSO (2012) Malawi In eg a ed Household Su ey 2010-11 (IHS3). Na ional S a is ical O ice,
36
Lilongwe. h p://mic oda a.wo ldbank.o g/index.php/ca alog/1003 Accessed 09 Ma ch
37
2020.
38
O A, Mwale B, Sai i-Chi songa D (2009) Explo ing seasonal po e y aps: The ‘Six-Week
39
Window’ in Sou he n Malawi. The J De S ud 45(2):227–255. doi:
40
10.1080/00220380802265330
41
Os om E (2009) A gene al amewo k o analyzing sus ainabili y o social-ecological sys ems.
42
Science 325(5939):419–422. doi: 10.1126/science.1172133
43
21
Pa el R (2009) Food so e eign y. The Jou nal o Peasan S udies 36(3):663–706. doi:
1
10.1080/03066150903143079
2
Pe c M, Jo dan JJ, Rand DG, Wang Z, Boccale i S, Szolnoki, A. (2017). S a is ical physics o
3
human coope a ion. Physics Repo s, 687, 1-51. doi: 10.1016/j.phys ep.2017.05.004
4
Pe ini C (2003) Slow ood: The case o as e. Columbia Uni e si y P ess.
5
Rausse G, Zilbe man D, Kahn G (2015) An al e na i e pa adigm o ood p oduc ion,
6
dis ibu ion, and consump ion: A noneconomis ’s pe spec i e. Annual Re iew o
7
Resou ce Economics 7(1):309–331. doi: 10.1146/annu e - esou ce-100913-012549
8
Sahley C, G oelsema B, Ma chione T, Nelson D (2005) The go e nance dimensions o ood
9
secu i y in Malawi. USAID Bu eau o Democ acy, Con lic , and Humani a ian Assis ance,
10
Washing on, DC.
11
h ps://sa pn.o g/documen s/d0001649/P1998-USAID_Malawi_Sep 2005.pd Accessed
12
09 Ma ch 2020
13
Schmidhube J, Tubiello FN (2007) Global ood secu i y unde clima e change. PNAS
14
104(50):19703–19708. doi: 10.1073/pnas.0701976104
15
Sch eckenbe g K, To es-Vi olas CA, Willcock S, Shackle on C, Ha ey CA, Ka umba a D
16
(2016) Pa icipa o y Da a Collec ion o Ecosys em Se ices Resea ch: A P ac i ione ’s
17
Manual. ESPA Wo king Pape Se ies No 003/June 2016. ISSN 2058-9875
18
h p://www.espa.ac.uk/ iles/espa/PRA-Manual.pd Accessed 09 Ma ch 2020
19
Spielman DJ, Da is K, Negash M, Ayele G (2011) Ru al inno a ion sys ems and ne wo ks:
20
Findings om a s udy o E hiopian smallholde s. Ag ic Hum Values 28(2):195–212. doi:
21
10.1007/s10460-010-9273-y
22
Uni ed Na ions (2015a) T ans o ming ou wo ld: The 2030 agenda o sus ainable de elopmen .
23
Resolu ion adop ed by he Gene al Assembly.
24
Uni ed Na ions (2015b) Wo ld u banisa ion p ospec s: The 2014 e ision. The Popula ion
25
Di ision o he Depa men o Economic and Social a ai s o he Uni ed Na ions.
26
Wa ing TM, Kline MA, B ooks JS, Go SH, Gowdy J, Janssen MA, Smaldino PE, Jacque J
27
(2015) A mul ile el e olu iona y amewo k o sus ainabili y analysis. Ecology and
28
Socie y 20(2):34. doi: 10.5751/ES-07634-200234
29
Wa s DJ, S oga z SH 1998. Collec i e dynamics o ‘small-wo ld’ ne wo ks. Na u e 393(6684):
30
440–442
31
Wi kin HA, P ice-williams D, Be ini M, Ch is iansen B, Ol man PK, Rami ez M, Meel JV (1974)
32
Social con o mi y and psychological di e en ia ion. In e na ional Jou nal o Psychology
33
9(1):11–29.
34
