E vide nc e in Ag ric ulture : Risk C ra ig Mc Into sh (UC Sa n - - PowerPoint PPT Presentation

E vide nc e in Ag ric ulture : Risk

C ra ig Mc Into sh (UC Sa n Die g o , AT AI) FAO ESA Se mina r T ue sd a y No ve mb e r 28, 2017

I. Q uic k Intro : AT AI a nd RC T s fo r Po lic y

• II. Risk a s c o nstra int to a g te c h a d o p tio n
• III. RC T

e vid e nc e o n risk mitig a tio n

• IV. Ethio p ia c a se stud y with FAO
• V. C o nc lusio n

Mo tiva tio n

Ag ric ultura l te c hno lo g ie s e xist tha t c a n

• b o o st p ro d uc tivity
• inc re a se p ro fits
• fo rtify the fo o d sup p ly

We ’ ve se e n a “ G re e n Re vo lutio n,” ye t a g ric ultura l p ro d uc tivity wa s no t tra nsfo rme d e ve rywhe re .

• Whe n te c hno lo g y a d o p tio n fa ils -- Why? Wha t p o lic y le ve rs c a n he lp ?
• Ho w c a n we imp ro ve sma llho ld e r fa rme rs’ p ro fits a nd we lfa re ?

2 AT AI | EVIDENC E IN AG RIC ULT

URE: RISK |

Da ta So urc e : Wo rld De ve lo p me nt Ind ic a to rs, FAO via the Wo rld Ba nk

Q : Wha t he lp s a nd wha t hind e rs sma llho ld e r fa rme rs’ a doption o f te c hno lo g ie s a nd a c c e ss to ma rke ts?

Whic h a p p ro a c he s impa c t fa rme r p ro fits a nd we lfa re ?

A: ...we ll, le t’ s ta c kle this sc ie ntific a lly

➔ Re vie w a va ila b le e vid e nc e : id e ntify ke y re se a rc h ne e d s sinc e

2009

➔ Mo b ilize re se a rc h ne two rks: “ c le a ring ho use ” ra the r tha n

c o nsulta nt mo d e l, fund c o mp e titive ly-se le c te d , hig h-q ua lity ra nd o mize d e va lua tio ns

➔ Sha re find ing s: info rm re le va nt d e c isio nma king

E va lua tio n Pro g ra m E va lua tio n I mpa c t E va lua tio n

RCT s

Ra ndo mize d Co ntro lle d T ria ls (RCT s)

4

J-PAL | WHY EVALUAT E?

RCT s: o ne type o f impa c t e va lua tio n

O the r me tho d s inc lud e :

• Pre -p o st
• Diffe re nc e in d iffe re nc e
• Ma tc hing
• Instrume nta l Va ria b le s
• Re g re ssio n Disc o ntinuity

T he se no n-e xp e rime nta l me tho d s re ly o n b e ing a b le to “mimic ” the c o unte rfa c tua l und e r c e rta in a ssump tio ns Ra nd o miza tio n c a n “ c o nstruc t” a c o unte rfa c tua l with me a sura b le o utc o me s

T ype s o f I mpa c t E va lua tio ns

5

J-PAL | WHY EVALUAT E?

Non-random assignment

HQ

Monthly inc ome , pe r c a pita

1000 500 1457 947

J-PAL | WHY EVALUAT E?

6

T C

Be fo re the pro g ra m sta rts, e lig ib le individua ls a re ra ndo m ly a ssig ne d to two o r m o re g ro ups so tha t the y a re sta tistic a lly ide ntic a l b e fo re the pro g ra m .

Ra ndo mize d e va lua tio ns pro vide a hig hly rig o ro us e stima te o f pro g ra m impa c t

7

T wo g ro ups c o ntinue to b e ide ntic a l, e xc e pt fo r tre a tme nt Any diffe re nc e s in

• utc o me s b e twe e n

the g ro ups c a n b e a ttrib ute d to the pro g ra m

Outc o me s fo r b o th g ro ups a re me a sure d

Inte rve ntio n Co mpa riso n

Po pula tio n is ra ndo mly split into 2 o r mo re g ro ups

J-PAL | WHY EVALUAT E?

Random assignment

Monthly inc ome , pe r c a pita

1000 500

T C

1257 1242

HQ

J-PAL | WHY EVALUAT E?

8

J-PAL | C EG A | AT AI

9

RC T s fo r Po lic y

Im pa c t re se a rc h imp o rta nt to id e ntify “ c a usa lity”

• Le sso ns fo r p ro g ra m a nd p o lic y d e sig n
• Sup p o rts re sults-b a se d ma na g e me nt o f inve stme nts

RC T s ha ve b e c o me a wid e ly use d me tho d o lo g y

• No t o nly a n a c a d e mic a p p ro a c h
• Stro ng d e ma nd b y d e ve lo p me nt p a rtne rs (C G IAR, NARS, O ne Ac re Fund ,

ma tc hma king e xe rc ise s)

RC T s in e c o no m ic s he lp in p a rtic ula r und e rsta nd the ro le o f b e ha vio r a nd institutio ns (a g ric ultura l syste ms) in p ro g ra m/ p o lic y o utc o me s.

Ine ffic ie nc ie s c o nstra ining a g te c h a d o p tio n

1. C re d it ma rke ts

2. Risk

3. Info rma tio n 4. Exte rna litie s 5. Inp ut a nd o utp ut ma rke ts 6. La b o r ma rke ts 7. La nd ma rke ts

11 AT AI | EVIDENC E IN AG RIC ULT

URE: RISK |

Pre vie w: risk

Risk ma tte rs

• Mo st inve stme nts in imp ro ve d inp uts inc re a se the fina nc ia l risks o f fa rming
• Fa rme rs ma ke c o nse rva tive p ro d uc tio n d e c isio ns to se lf-insure

So me p o te ntia l so lutio ns to risk:

1. Fina nc ia l instrume nts: We a the r Ind e x Insura nc e (WII)

• Lo w d e ma nd fo r mic ro -insura nc e , in p a rtic ula r we a the r ind e x insura nc e

2. T e c hno lo g y tha t struc tura lly d e c re a se s risks

• Risk-mitig a ting c ro p s, irrig a tio n: Pro mising e a rly re sults o n risk-mitig a ting c ro p s

3. C re d it p ro d uc ts with (e xp lic it o r imp lic it) limite d lia b ility in c a se o f we a the r sho c ks 4. Pub lic se c to r sa fe ty ne ts

12 AT AI | EVIDENC E IN AG RIC ULT

URE: RISK |

Ho w d o e s risk c o nstra in a d o p tio n?

• Ag ric ulture is inhe re ntly risky a c tivity

–

We a the r a nd d ise a se risks a re a g g re g a te , a ffe c ting a ll fa rme rs in g e o g ra p hic a re a

• Fa rme rs ma y lo se la rg e p o rtio n o f ha rve st to e xtre me we a the r e ve nt
• Witho ut a ny wa y to mitig a te o r insure risks, inve stme nt in c ro p s o r

te c hno lo g ie s a p p e a rs to b e a n unsa fe g a mb le

–

Hig he r-va lue c ro p s ma y a lso b e mo re se nsitive to we a the r

• Exa c e rb a te d b y risk a ve rsio n a nd a mb ig uity a ve rsio n

13 AT AI | EVIDENC E IN AG RIC ULT

URE: RISK |

Pro te c ting fa rme rs thro ug h fo rma l insura nc e

• Ag ric ultura l insura nc e to he d g e risk ub iq uito us in d e ve lo p e d c o untrie s

–

La rg e numb e r o f sma ll fa rme rs, p o o r re g ula to ry e nviro nme nts ma ke mo st tra d itio na l p ro d uc ts ill-suite d to sma llho ld e rs

• We a the r ind e x insura nc e a s inno va tio n to insure sma llho ld e rs

–

Pa yo uts ma d e o n o b se rva b le va ria b le (e .g . ra infa ll)

–

Avo id s so me d isa d va nta g e s o f c o nve ntio na l insura nc e : le ng thy c la ims p ro c e ss, a d ve rse se le c tio n, mo ra l ha za rd

–

But ha s b a sis risk: o ffic ia l o b se rva tio n d o e s no t a c c ura te ly p re d ic t fa rme rs’ lo sse s

14 AT AI | EVIDENC E IN AG RIC ULT

URE: RISK |

Stylize d ind e x insura nc e p a yo ut sc he d ule

15

Ra infa ll (mm) Pa yo ut Ma x Pa yo ut Pa yo ut inc re a se s with ra infa ll d e fic it

Ma jo r dra wb a c k to inde x insura nc e : b a sis risk

16

A d e c a d e o f e xp e rime nta tio n o n we a the r ind e x insura nc e

• 10 ra nd o mize d e va lua tio ns in

va rio us c o nte xts

–

Ind ia , Ethio p ia , G ha na , Ma la wi

–

Diffe re nc e s in c ro p s insure d , c o nd itio ns tha t trig g e re d p a yo ut, e tc .

–

Effe c ts o f d isc o unts, o the r e nc o ura g e me nts to p urc ha se insura nc e

–

Effe c ts o n p ro d uc tio n d e c isio ns

17

J-PAL 2016

De ma nd wa s lo w a t ma rke t pric e s b ut inc re a se d with la rg e disc o unts

Ka rla n e t a l 2013; Mo b a ra k & Ro se nzw e ig 2012; “ Ma ke it Ra in”

18

I nsure d fa rme rs to o k mo re risks o n the ir fa rms

• Whe n g ive n sub sid ize d insura nc e , fa rme rs to o k o n g re a te r p ro d uc tio n

risks

–

And hra Pra d e sh: Fe we r sub siste nc e c ro p s, mo re c a sh c ro p s

–

G ha na : Mo re la nd p la nte d to ma ize , g re a te r fe rtilize r use

–

T a mil Na d u: Shift fro m d ro ug ht-to le ra nt va rie tie s to hig h-yie ld va rie tie s

–

C hina : Insura nc e fo r so ws c a use d fa rme rs to mo ve into this risky b ut hig hly p ro fita b le c ro p

–

Me xic o (C ADENA): insure d fa rme rs p la nt mo re the ye a r a fte r a sho c k tha n no n-insure d fa rme rs

–

Ke nya (IBLI): insura nc e he lp s p a sto ra lists a vo id d e c a p ita lizing live sto c k in re sp o nse to d ro ug ht

C a i e t a l. 2015; C a i 2013; C o le e t a l 2014; Ka rla n e t a l. 2013; Mo b a ra k & Ro se nzwe ig 2014; d e Ja nvry e t a l 2016; Ja nze n & C a rte r 2013

19

C a n we via b ly imp ro ve d e ma nd ?

• Ma rke ting & T

ra ining ?

• Pric e Sub sid ie s?
• Inte rlinking with C re d it?

20 AT AI | EVIDENC E IN AG RIC ULT

URE: RISK |

Va ria tio ns o n tra ining , ma rke ting , a nd pro duc t de sig n ha d mo de st e ffe c ts o n ta ke -up

• Re la tive ly lo w ta ke -up with vid e o a nd flye r ma rke ting
• Fina nc ia l lite ra c y tra ining

–

Inc re a se d ta ke -up

–

No t c o st-e ffe c tive

• T

rust a nd e xp e rie ntia l le a rning

–

Mixe d re sults o n e nd o rse me nts

–

O b se rving p a yo uts o ve r time inc re a se d ta ke -up (c o nve rse a lso true )

• G ro up risk-sha ring

–

So me e vid e nc e tha t p re se nc e o f info rma l risk-sha ring ne two rks inc re a se d d e ma nd

C o le e t a l 2014; De rc o n e t a l, 2014; G a ura v e t a l 2011; Ka rla n e t a l 2014; Mo b a ra k & Ro se nzwe ig 2012

21

Dyna mic e ffe c ts o f sub sidie s

• L
• ts o f e vid e nc e fro m o the r p ro d uc ts tha t te mp o ra ry sub sid ie s

c a n ha ve d ura b le e ffe c ts o n d e ma nd :

–

Anti-ma la ria l b e d ne ts (Dup a s 2015)

–

Fe rtilize r use (C a rte r e t a l. 2014)

• Nume ro us stud ie s ha ve ra nd o mize d sub sid ie s fo r WII

–

So me d yna mic e ffe c t o f sub sid ie s, b ut p ro no unc e d o nly whe n p a yo uts

• c c ur in sub sid ize d p ro d uc ts.

–

Inte re st in d e sig ning ‘ o p tima l’ sub sid ie s to re a c h a d o p tio n ta rg e t (d e Ja nvry e t a l. 2015).

22

Dyna mic e ffe c ts o f sub sidie s

• No e vid e nc e tha t te mp o ra ry sub sid ie s will ‘ kic k-sta rt’ a p riva te

ma rke t a nd b e c o me unne c e ssa ry the re a fte r.

• So while sub sid ize d insura nc e a p p e a rs to ha ve a la rg e e ffe c t o n

fa rme r b e ha vio r, the ma rke t wo n’ t wo rk witho ut sub sid ie s.

• So : is the re a we lfa re c a se to b e ma d e fo r p e rp e tua l sub sid ie s to

WII?

23

Ca sh vs. F re e I nsura nc e

• O nc e we sta rt to think o f sub sid ie s a s a p e rma ne nt ne c e ssity in WII

ma rke ts,

–

Is it b e tte r to simp ly p ro vid e Unc o nd itio na l C a sh T ra nsfe rs tha n it is to d istrib ute the sa me re so urc e s in the fo rm o f fre e / sub sid ize d p re miums fo r WII?

• Fo rtuna te ly, this e xp e rime nt ha s b e e n p e rfo rme d b y Ka rla n e t a l.

(2013) in G ha na :

–

T wo -a rme d tria l d istrib ute s c a sh fo r inp ut p urc ha se s ve rsus fre e WII.

–

Pro vid e the o re tic a l justific a tio n fo r why WII mig ht wo rk b e tte r: T

• the e xte nt tha t risk is the o p e ra tive c o nstra int fo r inve stme nt, WII c a n

‘ unlo c k’ fa rme rs’ o wn c a p ita l b y g iving the m the c o nfid e nc e to inve st in inp uts.

24

Ca sh vs. Pre miums

• C urre nt d e b a te in so c ia l p ro te c tio n a b o ut UC T

s ve rsus va rio us typ e s o f c o nd itio na l c a sh tra nsfe rs.

• Distrib uting fre e insura nc e p re miums c a n b e tho ug ht o f a s

p ro vid ing a ve ry sp e c ific typ e o f C C T : ‘ If yo ur c ro p s fa il, we will p ro vide yo u with a c a sh tra nsfe r’

• T

he und e rlying lo g ic fo r this is tha t the re le a se o f risk c o nstra ints a llo w fa rme rs to mo ve to wa rd p ure p ro fit ma ximiza tio n a s fa rming d e c isio n-ma ke rs.

• L

inks WII to so c ia l p ro te c tio n.

25

Do wnside s o f sub sidizing risk

• Sub sta ntia l shift into risky p ro d uc tio n in se ve ra l stud ie s whe n

ind ivid ua ls a re p ro vid e d with sub sid ize d WII.

• T

his me a ns tha t the a g ric ultura l syste m a s a who le ha s g re a te r se nsitivity to ra infa ll.

• L

a nd le ss la b o re rs, who a re the mo st vulne ra b le , se e hig he r wa g e se nsitivity to ra infa ll whe n fa rme rs a re using WII.

26

Do wnside s o f sub sidizing risk

27

Mo b a ra k & Ro se nzwe ig 2014

Cro p o utput in insure d villa g e s lo se s the ‘ no rma l is b e st’ c urva ture a nd b e c o me s mo no to nic a lly re spo nsive to ra infa ll

Do wnside s o f sub sidizing risk

28

Mo b a ra k & Ro se nzwe ig 2014

Co nse q ue ntly, the a mo unt o f la b o r hire d in insure d villa g e s re spo nd s stro ng ly to ra infa ll

Do wnside s o f sub sidizing risk

29

Mo b a ra k & Ro se nzwe ig 2014

T his me a ns tha t a g ric ultura l la b o re rs a re ma d e mo re vulne ra b le if o nly a g ric ultura l pro d uc e rs a d o pt WII a nd the re fo re d e e pe n the ir struc tura l e xpo sure to we a the r.

C o nc lusio ns o n WII

• Still c le a r tha t risk is a ma jo r c o nstra int fo r sma llho ld e r fa rme rs
• Ho we ve r lo w d e ma nd me a ns we a the r ind e x insura nc e is unlike ly to

thrive a s a sta nd a lo ne ind ivid ua l c o mme rc ia l p ro d uc t

–

Pric e , d istrust, la c k o f fina nc ia l lite ra c y, b a sis risk

• Whe n fa rme rs ha ve insura nc e , the y ta ke mo re risks o n the ir fa rms

–

T his is g o o d fo r a ve ra g e yie ld s b ut e xp o se s la b o re rs to a d d itio na l inc o me risk

• So whe re d o we g o fro m he re ?

30

Linking WII to So c ia l Sa fe ty Ne ts

• Pub lic -p riva te p a rtne rship s fo r Risk La ye ring (C a rte r 2011)
• Whe n no t e xp lic itly c o mb ine d , p ub lic -se c to r p ro g ra ms suc h a s Ethio p ia ’ s PSNP

c ro wd o ut d e ma nd fo r WII (Duru 2015).

–

Ho w e ve r, if p riva te se c to r WII isn’ t via b le , this is no t a ma jo r d o w nsid e .

• Sa fe ty ne t p ro g ra ms a lso e xp o se g o ve rnme nts to p o te ntia lly hug e we a the r-

re la te d risk.

–

G o ve rnme nts sho uld use re insura nc e the mse lve s?

–

T ra nsfe r hug e a nd une xp e c te d lia b ilitie s into a p re d ic ta b le flo w o f c o sts fo r p ub lic se c to r.

–

Me xic o ’ s C ADENA p ro g ra m.

• WII a p p e a rs to b e a wa y o f p ro vid ing sa fe ty ne ts witho ut p ro b le ms o f

c lie nte listic d e ma nd s & so ft b ud g e t c o nstra ints, b ut ma y b e ha rd to a c hie ve this in p ra c tic e .

31

An a lte rna tive : risk-mitig a ting c ro p s a nd te c hno lo g ie s

• Ag ric ultura l R&D o n va rie tie s tha t

to le ra te flo o d , d ro ug ht, sa linity

–

Inc re a sing ly imp o rta nt with c lima te c ha ng e

• Swa rna -Sub 1 is a flo o d -to le ra nt

ric e va rie ty

–

No yie ld p e na lty in no rma l c o nd itio ns

–

Re se a rc he rs te ste d e ffe c t in re a l- life c o nd itio ns in O d isha , Ind ia

Da r e t a l 2015

32

Flo o d to le ra nt ric e

33

Da r e t a l 2015

34

• Mo re inve stme nt…

–

C ultiva te d mo re la nd

–

Use d mo re fe rtilize r

–

Ad o p te d imp ro ve d p la nting te c hniq ue s

–

Ad juste d the ir sa ving s a nd c re d it d e c isio ns

• … le d to hig he r yie ld s a nd hig he r re ve nue s.

–

Inc re a se d ric e yie ld s in ye a rs with a nd witho ut flo o d s

–

Hig he r yie ld s le d to inc re a se d re ve nue s a nd p ro d uc tive inve stme nts

F a rme rs g ive n Swa rna -Sub 1 inve ste d mo re

35

Sc a le -up wo uld b e ne fit ma rg ina lize d po pula tio ns the mo st

Da r e t a l 2015

36

I nte rlinking WI I with c re dit

• Why no t a d d re ss b o th c re d it

a nd risk c o nstra ints simulta ne o usly?

–

De ma nd sid e : Alle via te risk ra tio ning a nd b ring mo re ind ivid ua ls into the c re d it ma rke t

–

Sup p ly sid e : C ro wd in c re d it sup p ly if p o rtfo lio e xp o sure to we a the r risk limits le nd ing

• Pro b le ma tic in p ra c tic e :

–

‘ c ulture o f re p a yme nt’ ; ve ry ha rd to ma inta in re p a yme nt ra te s o nc e c o nd itio na lity o f re p a yme nt ha s b e e n intro d uc e d .

37

EPIIC A (Ethio p ia Pro je c t o n Inte rlinking Insura nc e with C re d it in Ag ric ulture ):

• T

wo wa ys o f thinking a b o ut fa rme r mic ro -insura nc e :

1. A susta ina b le ne w fina nc ia l p ro d uc t tha t c a n b uild a p riva te ma rke t me c ha nism to mo ve w e a the r risk a w a y fro m fa rme rs. 2. A w a y to c o ve r the lo ss to e xp e c te d o utp ut if fa rme rs a re und e rinve sting in inp uts d ue to risk (Ske e s & C o llie r, 2007).

• Ma y p ro vid e a b e tte r w a y o f p ro vid ing tra nsfe rs tha n c a sh (Ka rla n e t a l. 2014).
• T

his stud y e xa mine s b o th o f the se p re mise s in Ethio p ia , a ve ry we a the r- e xp o se d fa rming e nviro nme nt.

1. Wo rk thro ug h la rg e p riva te -se c to r c o mp a nie s, fa rme rs c o o p s in a re a s c ho se n to b e stro ng p o te ntia l ma rke ts. 2. Exp e rime nt a t ind ivid ua l le ve l va rying tra nsfe rs p ro vid e d in the fo rm o f risk sub sid ie s to fa rme rs a t p la nting time .

• Ultima te q ue stio n: c a n re mo va l o f risk g e ne ra te first-o rd e r imp ro ve me nts in

inp uts, p ro fits?

38

Ethio p ia n Pro je c t o n Inte rlinking Insura nc e & C re d it fo r Ag ric ulture (EPIIC A):

• Pro je c t is a c o lla b o ra tio n b e twe e n re se a rc he rs a nd :

–

Nya la Insura nc e C o mp a ny (la rg e st insure r in c o untry)

–

Da she n Ba nk (la rg e st p riva te -se c to r b a nk in c o untry)

–

Villa g e -le ve l a g ric ultura l c o o p e ra tive s a nd the ir C o o p e ra tive Unio ns.

–

Ethio p ia n Ec o no mic s Asso c ia tio n (fie ld wo rk/ a na lysis).

• Fie ld e d a c o mme rc ia l ind e x insura nc e p ro d uc t:

–

Ra infa ll ind e x b uilt fro m c ro p wa te r re q uire me nt mo d e l.

–

O ffe re d insura nc e to c o o p e ra tivize d fa rme rs in villa g e s within 25 km o f a re insura b le ra infa ll sta tio n.

–

Da she n lo a ns first c la ima nt o n Nya la p a yo uts fo r inte rlinke d .

–

Pro vid e d ra nd o mize d p ro mo tio n & sub sid y vo uc he rs fro m EEA.

39

Ma p o f Ra infa ll Sta tio ns a nd Stud y Wo re d a s.

40

Ethio p ia n C o nte xt.

1. Hig h risk: ra in-fe d a g ric ulture , la rg e ra infa ll va ria tio n.

–

Risk ha s b e e n d e mo nstra te d to b e a c o nstra int in Ethio p ia n sma llho ld e r inp ut use (De rc o n a nd C hristia e nse n, 2011).

2. Stro ng sta te invo lve me nt in inp ut a nd o utp ut c ha ins fo r the c o o p e ra tive fa rming se c to r.

–

T his p ro je c t a tte mp ting to b ring to g e the r p ub lic - a nd p riva te -se c to r e ntitie s in a ne w wa y.

3. La rg e g o ve rnme nt sa fe ty-ne t p ro g ra m (PSNP) ma y se rve a s a sub stitute fo r p riva te -se c to r insura nc e (Duru 2015). Ra ise s the q ue stio n: is it p o ssib le fo r the sta te to b e to o c re d ib le a t p ro vid ing d isa ste r re lie f, the re b y und e rmining p riva te -se c to r d e ma nd fo r insura nc e ?

41

Pro d uc t fie ld e d 1. Sta nd a lo ne Ra infa ll Ind e x Insura nc e :

• So ld thro ug h p rima ry (villa g e -le ve l) c o o p e ra tive s to me mb e rs a t time o f

p urc ha sing inp uts.

• Fra me d a s inp ut insura nc e , me a ning tha t it wo uld c o ve r c o st o f inp uts if

ra in fa ils.

• Pa yo ffs with trig g e r/ e xit fo r e a c h o f thre e c ro p g ro wth p ha se s,
• p timize d se p a ra te ly fo r ma ize , so rg hum, te ff, a nd whe a t fo r e a c h

insure d sta tio n.

• O nly ho use ho ld s in villa g e s who se c e nte r is le ss tha n 15km fro m a n

insure d sta tio n o ffe re d insura nc e .

42

Pro d uc t fie ld e d 2. Inte rlinke d C re d it with Ra infa ll Ind e x Insura nc e :

• C o o p e ra tive Unio ns (c o lle c tive s o f villa g e -le ve l c o o p e ra tive s) a re use d

a s c re d it inte rme d ia rie s.

• Ea c h C U sig ns sing le lo a n c o ntra c t with Da she n, who is ma d e

b e ne fic ia ry o f Nya la insura nc e p o lic y.

• C a n o nly g e t the inte rlinke d lo a n if insura nc e p urc ha se d , b ut c a n

c ho o se sta nd a lo ne p ro d uc t a lso in villa g e s whe re inte rlinke d p ro d uc t is so ld .

• Pushe s the C Us into ne w ro le , a sking the m to ta ke c o lla te ra lize d lo a ns

with c o lle c tive a sse ts.

• O nly suc c e ssful in a c hie ving re a l ta ke -o ff o f inte rlinke d insura nc e in o ne

C U; q ua lita tive stud y o f this c a se .

• Exp e rime nta l stud y is so le ly o n sta nd a lo ne insura nc e .

43

T he Ind ivid ua l-le ve l Vo uc he r Exp e rime nt:

T

• p re se rve a c le a n e xp e rime nt sub se q ue nt to a ttritio n:
• We ra nd o mize d the p ro visio n o f insura nc e p urc ha se vo uc he rs a t the

ind ivid ua l le ve l.

–

T he la rg e ma jo rity o f insura nc e c o ve ra g e issue d in the p ro je c t c o me s fro m the se vo uc he rs ra the r tha n fro m p riva te d e ma nd .

• Stud y p ro vid e s re la tive ly we ll-p o we re d e xp e rime nt o n e ffe c ts o f

ra nd o mizing tra nsfe rs to ho use ho ld s in the fo rm o f risk p ro te c tio n.

• Pro vid e s d ire c t te st o f ma rg ina l e ffe c t o f sta te -c o nting e nt c a sh tra nsfe rs.
• Is the re a multip lie r e ffe c t w he re b y re la xa tio n o f risk c o nstra int inc re a se s o ve ra ll

a p p e tite to inve st in inp uts a nd p ro d uc tivity?

• Q ua ntity o f c o ve ra g e ~ d ire c tly ra nd o mize d a t ind ivid ua l le ve l.

44

Virtua lly no ma rke t d e ma nd , vo uc he rs wo rk

45

No d yna mic b e ne fits o f sub sid ie s.

46

(1) (2) (3) (4) (5) (6) Any Voucher Year 1 0.00683 0.0364 0.0290 18.41 80.95 33.35 (0.0220) (0.0423) (0.0495) (35.52) (91.24) (88.13) Voucher Amount Year 1

• 0.00164
• 0.00154
• 3.551
• 3.218

(0.00222) (0.00221) (4.468) (4.273) Any Voucher Year 2 0.429*** 0.419*** 0.406*** 533.5*** 129.8 109.9 (0.0519) (0.0845) (0.0857) (124.2) (101.1) (96.84) Voucher Amount Year 2 0.000635 0.000867 28.66*** 28.97*** (0.00520) (0.00521) (9.146) (9.252) Insurance would have paid out Y1 0.0923 120.3 (0.0651) (117.9) Voucher Y1 * Insurance would pay Y1

• 0.0445

23.75 (0.0802) (122.8) Constant

• 0.00255
• 0.00241
• 0.0113
• 6.865
• 5.945
• 17.13

(0.00824) (0.00830) (0.00935) (13.34) (13.18) (13.64) Observations 841 841 841 841 841 841 Number of Observations 0.296 0.297 0.301 0.150 0.175 0.182 Covered by Insurance Year 2 Sum Insured Year 2 Regressions run at the household level among all cooperative members; dependent variable is the insurance purchase decision observed in the second sales season. Robust standard errors are reported in parentheses, clustered at the village level to account for the design effect. *** p<0.01, ** p<0.05, * p<0.1

Ve ry little imp a c t o f b e ing insure d o n inp ut use

–

No e vid e nc e o f a ny m e a ning ful inc re a se in inp ut use d ue to sta nd a lo ne WII.

47

Any Chemical Fertilizer KGs of Chemical Fertilizer Number of crops using Chemical Fertilizer Uses any Improved Seeds Uses any Input Credit (3) (4) (5) (6) (7) Any Voucher 0.0203

• 1.761
• 0.0319

0.0607* 0.0127 (0.0381) (4.862) (0.0826) (0.0329) (0.0339) Treated Village

• 0.0615

1.153

• 0.0691
• 0.124*

0.0374 (0.0983) (6.279) (0.182) (0.0689) (0.0393) R3 0.232** 15.99*** 0.497*** 0.0677 0.0432 (0.0920) (3.735) (0.176) (0.0532) (0.0292) R4 0.189** 17.09*** 0.359** 0.0570

• 0.0317

(0.0831) (4.038) (0.150) (0.0522) (0.0295) Constant 0.556*** 91.02*** 1.196*** 0.372*** 0.153*** (0.0272) (1.300) (0.0482) (0.0198) (0.00680) Observations 2,544 3,280 2,571 2,544 3,416 Number of Observations 0.084 0.025 0.069 0.006 0.014

Pa ne l Imp a c ts o n Ag O utp ut, Inc o me

–

No e vid e nc e o f a ny m e a ning ful inc re a se in inp ut use .

48

Total Value

• f Inputs

Used Index of Agricultural Yields HH Income per Capita (8) (9) (10) Any Voucher 1.567 0.00158 47.79 (14.31) (0.0438) (55.18) Treated Village

• 6.093
• 0.0848
• 190.8**

(19.63) (0.0932) (89.90) R3 3.940

• 0.0397

101.0** (12.02) (0.0791) (39.62) R4 11.35 0.0903 137.0*** (16.28) (0.0801) (39.08) Constant 128.3***

• 0.105***

246.5*** (4.802) (0.0166) (22.91) Observations 3,416 3,191 2,561 Number of Observations 0.000 0.014 0.004

So , why this la c k o f imp a c ts?

Sta tistic a l p o we r stro ng Diffe re nt a ttritio n b y vo uc he r sta tus d o e s no t c ha ng e re sults No stro ng imp a c ts within sub -g ro up s o f the tre a tme nt.

A se t o f o the r p o ssib le e xp la na tio ns:

1. Vo uc he r a mo unts to o sma ll?

• IV a na lysis to e stima ting the slo p e te rm o n a c tua l sum insure d .

2. Fe w a re a c tua lly risk c o nstra ine d in inve stme nt?

• Inte ra c tio n b y b a se line me a sure o f risk ra tio ning .

3. Insura nc e no t p ro p e rly p ro mo te d o r und e rsto o d ?

• Ind e p e nd e nt ra nd o mize d p ro mo tio n e xp e rime nt c o nd uc te d b y EEA a t

b a se line .

49

Preliminary results: please do not circulate.

T he Ind ivid ua l-le ve l Vo uc he r Exp e rime nt:

• O nly 21% o f fa rme rs p ut a ny o f the ir o w n mo ne y into p urc ha se ; mo st to o k the vo uc he r

a nd p urc ha se d o nly tha t muc h c o ve ra g e .

• Ind ivid ua lly ra nd o mize d re d uc tio n in risk e xp o sure .

50

20 40 60 80 Average Sum Insured 10 20 30 40 Randomized Voucher Amount Average sum insured Fitted Values

All values in 2010 US$. Size of dots proportional to number of observations at each assigned value.

Voucher Experiment & Average Sum Insured

Imp lic a tio ns o f risk c o nstra ine d fa rme rs

• Ba se d o n b a se line surve y 54.6% o f o ur sa mp le a re c re d it unc o nstra ine d , 18.8%

a re q ua ntity c o nstra ine d , 6.8% a re p ric e c o nstra ine d , a nd 19.8% a re risk c o nstra ine d . Sta nd a rd a g ric ultura l inve stme nt mo d e ls suc h a s Ba rd ha n a nd Ud ry (1999), Bo uc he r e t a l. (2008), a nd C a rte r e t a l. (2015) wo uld a ll p re d ic t tha t the first-o rd e r imp a c ts o f insura nc e

• n

e xp a nd ing the willing ne ss to b o rro w a nd inve st in inp uts will b e stro ng e st in the risk-c o nstra ine d g ro up .

• T

ho se id e ntifie d a s c re d it ra tio ne d a t b a se line ha ve sha rp ly lo we r inp ut use . T he y a re 12.5 p e rc e nta g e p o ints le ss like ly to use a ny fe rtilize r tha n the unc o nstra ine d , the y use 40 Kg s le ss fe rtilize ,r a nd use it o n ro ug hly ha lf the numb e r o f c ro p s.

• De sp ite the se la rg e c ro ss-se c tio na l d iffe re nc e s, the re a re no sig ns o f sig nific a nt

d iffe re ntia l imp a c ts o f the p ro visio n o f vo uc he rs o n the risk c o nstra ine d .

51

• 3. Imp a c t o f EEA Ba se line Pro mo tio n.

–

Nya la a tte m p te d to p ro m o te b y tra ining c o o p he a d s, e xte nsio n o ffic ia ls a s re c ruite rs to so lic it ind ivid ua l d e m a nd .

–

Re sults sug g e st tha t la c k o f p ro m o tio n wa s a b a rrie r to o ve ra ll up ta ke .

52

Preliminary results: please do not circulate.

Endline Survey Sum Insured Total Own Money Paid Sum Insured Total Own Money Paid Knowledge

• f Product

(2) (3) (6) (7) (9) Received Product Promotion at Baseline 194.8*** 2.810***

• 28.27

0.852*

• 0.00437

(69.49) (0.720) (33.42) (0.444) (0.0189) Any Voucher in Corresponding Season 428.8*** 4.745*** 551.4*** 2.108**

• 0.0325

(93.30) (0.923) (127.3) (0.891) (0.0249) Constant

• 20.97**
• 0.303***

9.135

• 0.275*

0.0884** (9.089) (0.109) (10.88) (0.154) (0.0360) Observations 847 847 835 835 588 R-squared 0.121 0.162 0.150 0.015 0.004 First Sales Season Second Sales Season

Na rra tive im p a c ts in Fe re s We g a , whe re inte rlinke d insura nc e wa s so ld :

• 53

Input: Number Increasing % Increasing Number Decreasing % Decreasing Number with No Change

Local Seeds 20 18.5% 3 2.8% 85 Improved Seeds 28 25.9% 5 4.6% 75 Organic Fertilizer 28 25.9% 5 4.6% 75 UREA 72 66.7% 9 8.3% 27 DAP 70 64.8% 9 8.3% 29 Insecticides/Herbicides 17 15.7% 2 1.9% 89 Veterinary Services 7 6.5% 0.0% 101 Other Livestock Inputs 4 3.7% 1 0.9% 103

Reported Changes in Input Use: Data come from the Round 5 survey conducted only in the village of Feres Wega where interlinked insurance was successfully sold.

EPIIC A C o nc lusio ns.

1. C o m m e rc ia l:

–

No e vid e nc e o f so lid d e ma nd fo r sta nd a lo ne WII a t ma rke t p ric e s.

–

T e mp o ra ry sub sid ie s a re no t a n e ffe c tive w a y to kic k-sta rt the ma rke t.

–

Inte rlinking insura nc e & c re d it is a c o mp le x und e rta king b ut sho w s p ro mise .

• Ultima te ly

mo re e ffe c tive to p ro vid e ind e x insura nc e to b a nks tha n to the ir b o rro w e rs?

2. T he pro duc tive po te ntia l o f tra nsfe rs via risk re duc tio n:

–

Sub sta ntia l ind ivid ua lly-ra nd o mize d va ria tio n in the e xte nt o f sta nd a lo ne WII c o ve ra g e , b ut no e vid e nc e o f me a ning ful c ha ng e s in a g ric ultura l b e ha vio r. No t e no ug h time fo r le a rning

–

No e vid e nc e fro m this stud y tha t ma king tra nsfe rs via risk re d uc tio n g e ne ra te a first-o rd e r imp ro ve me nt in inc o me .

–

Inte rlinking WII with c re d it se e ms to ha ve p o te ntia l to inc re a se p ro d uc tio n inp uts a nd shift risk, b ut it is a time c o nsuming p ro c e ss. Pro mising w a y to g o .

54

Summa ry: Risk

• Risk is a c o nstra int fo r sm a llho lde r fa rm e rs
• C o m m e rc ia l inde x insura nc e ta rg e te d dire c tly a t fa rm e rs unlike ly to

so lve the pro b le m

–

Pric e , d istrust, la c k o f fina nc ia l lite ra c y, b a sis risk

• Alte rna tive s to he lp fa rm e rs m a na g e risk

–

Re think insura nc e : p ro vid e sub sid ize d p o lic ie s a s a n a lte rna tive to c a sh tra nsfe rs

–

Se ll to institutio ns suc h a s a g le nd e rs

–

Pro mising p re limina ry re sults o n risk- m itig a ting c ro ps

55

T ha nk yo u!

a ta i@ p o ve rtya c tio nla b .o rg