Journal of Financial Risk Management
2013. Vol.2, No.3, 55-60
Published Online September 2013 in SciRes (http://www.scirp.org/journal/jfrm) http://dx.doi.org/10.4236/jfrm.2013.23009
Securitizing Area Insurance: A Risk Management Approach
Pasquale Lucio Scandizzo
University of Rome “Tor Vergata”, Rome, Italy
Received April 19th, 2013; revised May 19th, 2013; accepted May 26th, 2013
Copyright © 2013 Pasquale Lucio Scandizzo. This is an open access article distributed under the Creative
Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.
This paper examines the possibility of developing a risk management instrument by designing a financial
security whose value is linked to the average revenue of a given area. This type of program is sufficiently
general to be considered for any group of businesses that face production uncertainty. In agriculture, it has
been proposed as an alternative to multiple peril crop insurance programs, as area yield, revenue or rain-
fall insurance in order to eliminate ex ante and ex post moral hazard. While most of the literature concen-
trates on the determination of value of the indemnity and the payment of such an insurance, this paper fo-
cuses on the fact that, unlike other forms of insurance, area insurance can be cast in the form of a hedging
security and, as a consequence, rather than depending only on the demand for diversification (the beta of
the Capital Asset Price Model), it makes possible a risk shifting strategy based on the heterogeneity of
risk attitudes of the economic agents operating in a given area.
Keywords: Risk; Insurance; Moral Hazard; Security
Introduction
Objective of this paper is to present and analyze the charac-
teristics of a new type of contingent contract, aimed at securiti-
zing insurance against production risks. While the concept pro-
posed has a wider applicability, this paper considers in particu-
lar the case of area insurance. This is an insurance program that
pays an indemnity proportional to an indicator of average re-
venue or income for a given area. In the case of agricultural AI,
for example, the indicator may be directly yield, or some other
statistic, such as rainfall, that can be used as a reliable predictor
for the revenue shortfall, which is the object of insurance. The
greatest advantage of AI and GI is that, unlike other insurance
programs, and provided that the subject insured is not large
enough to affect the average revenue of the area, this type of
program is almost comple tely free of moral ha zard. In the USA,
experience with area insurance in agriculture began in 1990,
when the farm bill provided funds for the Federal Crop Insur-
ance Corporation (FCIC) to pilot-test new insurance products.
The Group Risk plan was the first of these experimental pro-
grams, introduced in 1993 as a pilot area-yield crop insurance
for soybeans. This program was followed in 1994, as a conse-
quence of budgetary provisions in the 1993 Omnibus Budget
Reconciliation Act, by a mandate to FCIC to offer area based
insurance on about 1200 counties for barley, cotton, peanuts,
grain sorghum, soybeans, and wheat. Although these contracts
have not appeared to be very popular with farmers so far, com-
mitment of FCIC has remained strong over the years, and the
coverage of the program has been considerably extended (more
than 30,000) from the limited number of farmers (less than
12,000) participating in the first two years.
A recent application of GI insurance is Group Risk Income
Protection (GRIP), an area-based revenue insurance product
that pays the insured in the event the county average per-acre
revenue falls below the insured’s “trigger revenue.” GRIP de-
rives from the Group Risk Plan of Multiple Peril Crop Insur-
ance. The addition of a price component to GRP to form a re-
venue guarantee was developed by Dr. Bruce Babcock and Dr.
Dermot Hayes, Professors of Economics, Iowa State University.
GRIP is linked to the Chicago Board of Trade (CBOT) negotia-
tions, since its expected price is defined as the simple average
of the last five final daily settlement prices in February on the
CBOT December corn futures contract and the nearby Novem-
ber soybean futures contract for the current crop year. Harvest
price, on the other hand, is defined as the simple average of the
final closing daily settlement prices in November on the CBOT
nearby December corn futures contract and in October on the
nearby CBOT November soybean futures contract for the cur-
rent crop year. A GRIP indemnity payment will occur if the
county revenue is less than the producer’s trigger revenue based
on the selected coverage level.
In this paper, I examine the possibility of designing a risk
management instrument through a financial security whose va-
lue is linked to the average revenue of a given area. This type of
program is sufficiently general to be considered for any group
of businesses that face production uncertainty. In agriculture, it
has been proposed as an alternative to multiple peril crop insu-
rance programs, as area yield, revenue or rainfall insurance
(Miranda, 1991, 1999; Skees et al., 1997) in order to eliminate
ex ante and ex post moral hazard. Within the same context, Ma-
hul (1999) and Mahul et al. (2010) have shown how to deter-
mine the optimum value of the indemnity and the payment of
such an insurance. However, I focus on the fact that, unlike
other forms of insurance, area insurance (AI) can be cast in the
form of a hedging security and, as a consequence, rather than
depending only on the demand for diversification (the beta of
the Capital Asset Price Model), it makes possible a risk shifting

P. L. SCANDIZZO
strategy based on the heterogeneity of risk attitudes of the eco-
nomic agents operating in a given area.
The Competitive Market Model
Assume that a group of farmers of a given area are engaged
in the production of several crops, i.e. corn, wheat, soybeans,
etc., and that the income resulting from production is uncertain,
because it depends on weather, pest attacks, price fluctuations
and other random events. The i-th farmer chooses the number
of acres allocated to each crop by maximizing the expected
value of utility, which is defined as function increasing in the
farmer’s net revenue (positive first derivative). The utility func-
tion is also supposed to be such that its increase declines with
income increases (negative second derivative). Formally, indi-
cating with i
U the utility index for the i-th farmer, with


iij
yy the vector of total income generated by the j-th
crop, with i
x
the vector


ij
x
of the number of acres allocat-
ed to the j-th crop, and with i

the vector


ij

of the sto-
chastic net revenue per ha of the j-th crop, we can formulate the
decision problem of the i-th farmer as follows:


max ii EUy ; subject to:
,
iii i
yxDx 
 i
b (1)
where is the expectation operator, primes denote transpos-
es, E


kj is a matrix of input-output coefficients, meas-
uring the quantity of the factor required to produce one
unit of the j-th crop, an
Dd-kth
d


iik
bb

is a vector of resource
constraints for the i-th farm

1, 2,k,

K
. Taking a second
order Taylor series approximation of the utility function around
the mean values of the random variables

ij

, or, alternative-
ly, assuming that the ij ’s are distributed according to a two
parameter distribution (e.g. the normal), we obtain the familiar
E-V utility function:
p


0.5
iiyii iyyiii
UUpxUxx

  , (2)
where ii pE

,


dd
iyi i
UUy denotes the first derivative
(i.e. the marginal utility) of each farmer with respect to total in-
come from all crops,


22
dd
iyyi i
UUy0 (3)
is the corresponding second derivative, and


iijm
is the
variance-covariance matrix of crop incomes per ha of the i-th
farm .
 
ij
Dividing both sides of (2) by , we can re-write the prob-
lem in (1) as follows:
p
i
U


max 0,5
iii iiii
EVp xxx 

  , subject to: ii
Dx b

.
where iiiy and VUU

iiyy
UU  

iy
equals Pratt coeffi-
cient of absolute risk aversion.
To solve this problem, form the Lagrangean:




0.5
iiii iiiii
L
pxxxDx b 

  (4)
where


iik

  is a vector of Lagrange multipliers.
The Kuhn-Tucker conditions for the solution of (3) can be
written as follows:
11
0 or 0, 1,2,,
JK
imiijm ijikm ikim
jk
pxdxm  


 J (5)
Expression (5) states the well known condition of optimality
requiring that for each crop to which a non zero are
ed, expected revenue per ha be equal to the risk premium plus
th
1
0 or 0, 1,2,,
J
ikj ijikik
j
dx bkK 

 
 (6)
a is allocat-
e cost per ha of all inputs evaluated at the shadow prices ik

.
Each of these shadow prices, according to condition (6), on the
other hand, are non zero only if the corresponding resource
cons traint is binding.
Let’s introduce now the possibility of acquiring (going long
or short) a security, which yields to long holders a random pay-
off

IgS


,
where
g
is a positive constant, for S

, and requires a gi-
ven payment Ic

 for S


lders. For short
holder, rsey would receive a
for long ho
d,s the
r S
the situation would be reve a
payoff from the long holders fo c

 and would pay
them


IgS


 for S

. S is thus the trigger level
for the payoff,
11
11
1nJ
ij ij nJ ij
x
ij
ij
x

 



is random revenue per ha in the area (i.e. for all farmers con-
sidered). Denoting with


G

ected vathe distribution function of area revenue per ha, the explue of the premium is
We can reformulate the i-th farmer maximization problem as
follows:
subject to:
  
0
1
R
EIgRGcG R   
.
d


,
max max0.5
ii iiiii
xq EVp xqEI

 

 (7)
 
22 2C
ov,
ii i iIiii
xxq qxI 

 
 
0
ii
Dx b


ective function for
; 0
ij
x
the i-1,2,,; 1,2inj th farm in (7) is no
, ,J
w formed The obj
of four parts: 1) the expected value of net farm revenue from
, 2) thfrom holding (long or crop productione expected gain
short) the security, 3) the risk premium for crop production, 4)
the risk premium for holding the security, 5) the conjoint risk
premium to hold a portfolio with both crops and the security.
Note that the i-th farmer decides the number of hectares ij
x
to
allocate to the j-th crop


1, 2,,jJ and the number of (ha
equivalents) i q of securities to hold. Note also that 0
i
q in
the case of “long” holding (or buying) and 0
i
q in thease
of short holding (or sellin
The formulation in (7) describes the planning problem of a
risk averse farmer (a ssuming 0
i
c
g).

). This aents offered the
opportunity to go long (i.e. to buy) or short ( an insu-
ra
g i
to sell) on
nce-like security based on the average revenue accruing from
a set of activities (e.g. agricultureithin a given area or other-
wise defined group of agents. In practice, rather than average
income, which is difficult to observe directly, the security, in
exchange for a premium c paid by the “long” agent (respective-
ly, paid to the “short” agent) in the years where an observable
variable z (for example, rainfall or any other index that can
be easily observed and is correlated with area revenue) is above
a trigger level R, pays the amount

hR z in the “bad years”.
Denoting with
) w


Fz the distribution function for z, assum-
56

P. L. SCANDIZZO
ing z is non-negative, we can compute the expected value of I
as follows:
 
0
d1
R
EIh RzFzcFR 
 (8)
The Kuhn

-Tucker conditions for the solution of problem (7)
may written by adding to the constraint in (7) the following: be




Cov 0
iiiii iii
pxqID 

 

(9)
wh


2Cov ,0
iIiii
EIq xI  
  
, (10)
ere i

that for all non
is a vector of shadow prices. Express
zero activities expected revenues per ha should
equalarginal costs, where these are given by
inclusive of insurance, and by resource opportunity costs eva-
t s
ion (9) states
m risk premiums,
luated ahadow prices. Expression (10), on its part, states that
the insurance net payoff should itself equal its risk premium.
Indicating with the sign * the vectors and the submatrices
corresponding to non zero activities, and applying the variance
and covariance definitions, we can write:




 

*
2
***Cov,**
Cov ,
iiiiii ii
iii iI
pxqzIDq
EIxz I
 
 

 

 (11)
wh is a ere *
i
p*,1
J
vels *
ij
vector of revenues per ha for non zero
farmy le activit
x
, 1,2,,*jJ
z, and i
, is
varian ee

Cov ;zI the co-
ce betwn I and

is the vector
differences b
cator z
of coeff
cien ned etween
nvn
le independent of
i-
ts obtai
farmby pro nue n
jecting linearly the
e statistical indi
the i-th reveper ha and the corresponding mean onto
the difference betwee thof total re-
venue per ha of the area iolved ithe scheme and the corre-
sponding mean according to the so called “regressability as-
sumption” (Benninga et al., 1984):

iiii
pzEzu   (13)
i
u being a (vector valued) random variab
y
.
e be-
e-
of
According to this assumption, for each crop, the differenc
farm revenue per ha and its mean can
into two parts: one proportional to the difference b
tw
nearly
tw
een be decomposed li-
een the current value and the mean value of an indicator
revenue per ha of the entire area involved in the scheme, and an
independent, idiosyncratic random component, representing
non diversifiable risk. Each component of the vector i

is a
coefficient ij

similar to the widely known coefficient of the
Capital Asset Pricing Model (CAPM) and measures the sensi-
tivity of the revenue per ha of the j-th crop to movements in
area revenues per ha.
Going bato the maximizing conditions, expression (11)
states that prices of all non zero activities should equal marginal
costs at the optimum, while expression (12) indicates that the
quantity of insurance acquir
ck
ed by the i-th firm equals the net
value of the insurance, including its utility from risk diversifi-
cation. This result can be seen more clearly, dropping the aster-
isks for simplicity, by writing it as follows:
 

02
d1
R
i
iz
hR zFz z cFR
qx
FR i
i





 (14)
where


2
Cov ,1
I
zI

 ,
But




22
Cov ,Iz zIF R   1,
so that the correlation between area renue and the insurance
is ve



Cov ,
zI
zI FR

 .
er is proportional to the ratio between her expected
gain and her risk premium (the subjectivebenefit-cost ratio”)
plus the agent’s relative gain from diversifn (her objective
benefit-cost ratio
This brings us to state the following proposition:
Proposition 1. The amount of the security purchased (sold)
by each farm “
icatio
).
Dividing (14)) by
ij i
J
x
x




where

is the *,1
J
sum vector, we obtain the expression
for the amount of coverage, defined as the ratio of the quantity
purchased or sold of
ated has of the indithe security in ha equivalents to total cul- tivvidual farm:
 


02
d1
iix
ii
z
hRFz cFR
q
xFR
Rz





 (15)
where i


denotes the relative coefficient of risk aversion of
the i-tht and agen
ii
ix
i
x
x





is the ratio between the beta of the individual crop and the land
cultivated for the i-th agent.
Corollary 1. The amount of coverage purchased (sold) will
be proportional to the sum of two terms: 1) the individual ratio
between the expected net benefit and the risk premium, and 2)
the weighted average of the agent betas.
Comment. The insurance program introduces an element of
risk due to the variability of area revenue. Those who go long
on the security, in fact, pay in every period a premium c to
those who go short on the same security, unless area revenue
(or its proxy) is below its threshold level, in which case the
“shorters” will have to pay the “longers” a compensation.
The compensation paid by the “shorters” to the “longers”, in
turn, will be a function of the difference between current area
revenue (or the current value of the proxy used) and the thresh-
old. Demand for the security will be larger, the higher the trig-
ger value for the payment of the indemnity, the smaller the risk
premium that each farmer is willing to pay to hold the security,
the lower the premium to be paid to the short holders and the
higher the correlation between the performance of the buyer
and the average performance in the area. On the other hand,
supply for the security will be larger, the larger the premium
1
  
 



 
2
2
2
Cov ,d,1
1
Var
ZzR
zIyRzF zIcF R
EyRF REycF REz
EzEzF Rz

 

 


P. L. SCANDIZZO
paid to shorters, the lower the risk premium and the larger, in
absolute value, the negative correlation between the supplier’s
revenue per ha and area revenue per ha. Note also that in order
to hold the security, it is not necessary that the agent is a pro-
ducer, i.e. the diversification component may be zero and the
quantity of the security purchased (sold) may still be positive.
In this case, the agent holding (long or short) the security will
act as a pure speculator.
Equation (15) can be interpreted as the demand for holding
long the security (for 0
i
q) and the demand for holding it
short


0
i
q.Since each unit of the security promises to yield
a net expected revenue of








1
r
E
IFREhRz cFR
to long holders and EI to short holders, the demand elastic-
ity with to the payment to be made is respect

 

2
1
iiiz
c
FR
EIxFR 



for long holding and




2
r
FREhR z

ii iz
EI xFR 


for short holding. and for insurance should equal supply so
demand
.
farms the terms in (15) and solving for c, we
In equilibrium, dem
that the sum of excess
0
n
i
i
q

Summing over all
can thus find the equilibrium value of the premium for the in-
surance policy:

 
2
1
R
z
z
FR X
cEhR
FR n
z








(16)
is the harmonic mean of the individual risk avers
cients,


where.
1
n
1
1i i
n 



 


ion coeffi-


X
n is the corresponding average relative
ion coefficient, evaluated at the average level of cul-
tivated land in the area,
risk avers
 


R
E
hR zEhR zt 
for zR, and I have used the property:,
1
,
n
iiz
i
x
X 


 IJ
ij
11ij
X
x

is total land cultivated in the area and


2
Cov ,
z
z
z

  is
artor with respect
erage
Proposition 2. The payment for the short holders of the se-
curity is equivalent to an insurium.he level of the
premium that equilibrates demand and supply is always greater
expected value of
v
curity.
he expected indemnity
an
the expected charge
sh
m
the line regression coefficient of the indica
to av area revenue per ha2. z
ance prem T
than the actuarially fair premium (i.e. thethe
payoff for the long holders) depending on the aerage (area)
risk premium from holding the se
Comment. Given the number of traders, in equilibrium the
expected value of the payoff for both long and short holders
equals the average premium for risk that the farmers involved
in the scheme are willing to pay. Therefore, the premium to be
paid (respectively, to pay) from those who buy (to those who
sell) the security will equal the sum of t
d of the average (in the harmonic mean sense) subjective risk.
The premium will be greater than its “actuarially fair” level of a
loading factor reflecting average risk aversion (in the harmonic
mean sense), where the average is taken over both long and
short security holders. An increase in the risk aversion of any
agent, in other words, increases the premium (i.e. the “take” of
the short holders) independently on whether this concerns some-
body who would buy or sell the security.
Corollary. The equilibrium expected level of the equivalent
premium (the payment to short holders) equals the expected
utility gain of the average (in the harmonic mean sense) agent.
Comment. Expression (17) shows that, in order to be feasi-
bly supported by the farmers of a given area, the contract pro-
posed must be “fair” in the sense that
ould equal the expected utility gain of a representative farmer.
Such an agent is defined as one, whose absolute (or relative
computed at the mean) risk aversion coefficient is the harmonic
ean of the risk aversion coefficients of all other farmers sup-
porting the scheme. In other words, in order for the scheme to
be sustainable, the representative agent should be unable to gain
(or lose) on average from purchasing or selling the security.
This notion of fairness does not coincide with the usual actuar-
ial notion, since a risk loading factor is added to the actuarially
fair premium.
Substituting (16) into (15), we obtain:
iii
z
i
X
qx n 








(17)
Using the definition of relative risk aversion, we can also
write:
iix
ii
qH
x

z









 (18)
where

and i

are both coefficients of relative risk aversion
respectively at the overall average and average revenue level
for the i-th agent.
Proposition 3. The equilibrium holding lev
mer of an area insurance security is ind
size of the premium and the trigger level for the indemnity. It
the e
tive risk aversion (the ratio of average to the
fa
t. Proposition 3 trans-
lat
the ratio of average to your risk aversion coefficient. The rea-
el for the i-th far-
ependent of both the
equals differnce between a measure of the demand for di-
versification (the average beta of the farmer) and the reciprocal
of a measure of rela
rmer’s own risk aversion coefficient).
Comment. In equilibrium, the expected value of the payoff
equals the average risk premium of the participants to the
scheme (Proposition 2). Thus, while the expected payoff for
long and short holders will vary with the trigger level of the in-
demnity, the equilibrium level of the amount of the security
bought and sold by each farmer will no
2Using again the regressability assumpti on, we can write:

z
zEzEv    ,
where is a randomly distributed disturbance. Substituting into (13), we
obtain:
v

iiiz i
pEv  
es into the simple rule: whatever the indemnity expected, buy
(sell) the security for a share equal to your average beta minus
i
u
58

P. L. SCANDIZZO
son for this is that the demand for the security depends in the
first instance on the degree to which it will diversify the portfo-
lio of the productive activities of the farmer. Because carrying
the security implies an additional risk, this degree, which can be
measured by the weighted beta, has to be corrected with the
ratio between average and individual risk aversion. The larger
this ratio, the larger the risk that the farmer is willing to carry
with respect to the average, thus requiring less insurance. De-
mand (and supply) for the security will be zero if all farms are
identical (all betas are equal to one) and all risk aversion coef-
ficients are also equal. In general, however, the amount of the
security demanded (offered) will be larger (smaller) the larger
(the smaller) the correlations involved (between the index and
area revenue and between own and area revenue), and the smal-
ler (the larger) the ratio between average and own risk aversion.
For example, if the i-th agent risk aversion is twice the harmo-
nic mean, and 1
i

, optimum coverage will be 12. Note, in
particular, that even if all farmers’ revenues are positively cor-
related with area revenues, it will pay for some of them to go
short on the security. A sufficient heterogeneity in risk aversion,
in other words, will ensure that a market for the hedging secu-
rity may develop even in the absence of negative correlations
across farms. For the same reason, in equilibrium, we may ex-
pect pure speculators to hold short positions.
From Equation (5), by totally differentiating with respect to
i
q, we can derive for the m-th crop:

1
2
d for all 0
dii
ii
esis is full, it will
cois the matrix given by the
product y the shadow prices.
This matrix will be singular if the number of factors is
than the number of products, but its sum with a full v
covariance matrix will also be full. Moreover, by the properties
of
iiz im
ii
xDFRx
qx










 (19)
Provided that matrix in the square parenth
also be positively definite since the first term is the variance-
variance matrix, while the second
of the input-output coefficients bsmaller
ariance
shadow prices at the optimum, an increase in crop produc-
tion may not result in a decrease in any shadow price
(i.e. 0
ik
ij
x
 
 for all ,,ik j).
Thus, we can state the following proposition:
Proposition 4. In equilibrium, for long holders of the secu-
rity, production of the crops whose revenues per ha are posi-
tively (negatively) correlated with area revenue per ha increases
(decreases). For short holderson oe crops whose
revenues per ha egativelyly) correlated with area
revenue increases (decreases). nce determines a
di
the possibility of expanding its
pr
, productif th
are n (positive
Comment. The introduction of area insura
fferentiation in farm portfolio holdings, since farms now hold,
in addition to their cultivated plots, a security whose perform-
ance is negatively correlated with average performance over all
farms and crops. If the m-th crop revenue per ha is positively
correlated with area revenue, two effects will ensue: 1) a reduc-
tion of production costs, due to
oduction and covering its risk by going long on the security,
and 2) an increase in costs due to the fact that insurance incur-
porates a risk premium depending on average risk aversion and
on the variance of area revenue. For long holders, the first ef-
fect will dominate the second one, while the opposite will occur
for short holders. By a basic property of mathematical program-
ming, since the introduction of a new dimension of optimiza-
tion is equivalent to the removal of a constraint, we have that




DqDq 

, i.e. the shadow cost of the resources de-
creases as a consequence of the introduction of the security.
This implies a higher first term on the right hand side of Equa-
tion (19). For 0
i
q, and 0
im

 (i.e. the m-th crop is posi-
tively correlated with area revenue, production of the m-th crop
will thus unequivocally increase, since shadow costs are lower
and the risk premium to hold the new security is also lower. In
ers can now hedge the crops that are posi-
tively correlated with area revenue by going long on a security
which is also negatively correlated with area revenue. If 0
i
q
other words, farm

,
and 0
im


, we a simult. In other words, farmers
who grow crops that are negatively correlated with area reve-
nue can now hedge them by going short on a security which is
also negatively correlated with area revenue. For the same rea-
sons, we will see a reduction of the production of long holders
(short holders) crops negatively (positively) correlated with
area revenue. These crops, in fact, do not offer any further
portu increase earnings because of the introduction of the
new security.
Conclusion
In this paper I have shown that area insurance may be mol-
ded in the form of a security or a contingent claim, whose pay-
off depends on the value taken by a statistic of a given popula-
tion. This statistic, which may be the value of average or me-
dian income, revenue or
e havilar res
op-
nity to
yields, or any statistic correlated with
it (e.g. rainfall), displayanges with the state of
nature. When the statisitical level, the people
w
for long holders, the security
pa
s a value that ch
tic is below a cr
ho have gone “short” on the security will pay a premium,
while when it is above the critical level, they will collect a
payment from those who have gone “long”. Actuarial group or
area insurance may be considered a special case where only one
insurer goes short on the equivalent security, by committing
herself to pay the premium in the “bad” states, in exchange for
the payment in the good states.
The rewards that the security promises to long and short hol-
ders possess some simple properties. First, the amount of the
security purchased (sold) by each agent is proportional to the
ratio between her expected indemnity and her risk premium
(the subjective “benefit-cost ratio”) plus the agent’s relative
gain from diversification (his objective benefit-cost ratio). Se-
cond, given an expected reward
y off for short holders that equilibrates demand and supply is
greater than the actuarially fair premium and depends positively
on the average (area) risk premium from buying or selling in-
surance. Third, the equilibrium expected level of the premium
charged for the security equals the expected utility gain of the
average (in the harmonic mean sense) agent. Fourth, the opti-
mal level of insurance demanded or supplied in equilibrium
will be proportional to the ratio of the beta of each agent to her
share of total activity level minus the ratio of average relative
risk aversion to the agent’s own risk aversion coefficient.
REFERENCES
Benninga, S., Eldor, R., & Zilcha, I. (1984). The optimal hedge ratio in
unbiased futures markets. Journal of Futures Markets, 4 , 155-159.
Feder, G., Just, R. E., & Schmitz, A. (1980). Futures markets and the
theory of the firm under price uncertainty. Quarterly Journal of Eco-

P. L. SCANDIZZO
60
Journal of Agricultural Economics, 81,
Mahul, O., & Stutley, Cport to agricultural in-
surance: Challenges and countries. World Bank
M
St. American Journal of Agricul-
Voption of wea-
nomics, 317-328.
Mahul, O. (1999). Optimum area yield crop insurance. The American
75-83.
. (2010). Government sup
options for developing
Publications.
iranda, M. L. (1991). Area yield crop insurance reconsidered. Ameri-
can Journal of Agricultural Economics, 73, 233-242.
kees, J. R., Black, J. R., & Barnett, B. J. (1997). Designing and rating
an area yield crop insurance contrac
tural Economics, 79, 430-438.
argas Hill, R., Kumar, N., & Hoddinott, J. (2010). Ad
ther-index insurance: Learning from willingness to pay among a pa-
nel of households in rural Ethiopia. Mim eo: IFPRI.