=,==
Journal of I=hil_plne Development
N.mber
35,
Vol.rn=
X,X,
No2ondSe.,es=,
1..2
AGGREGATE
CORN AREA RESPONSE
UNDER
___
RISK:
SOME IMPLICATIONS FOR PRICE STABILIZATION PROGRAMS
Meyra Sebello Mendoza, 13.Wade Brorsen
and Mark W. Rosegrant*
INTRODUCTION
Considerableattentionhas been focusedon theimportance of risk
in farmers' production decisions. Previous empirical studies of
farmers' response under uncertainty in developed and developing
economies have shown that agricultural production is negatively
affected by risk. In studies of American agricultural farmers, Just
(1974), Lin (1977), Traill (1978), Hurt and Garcia (1982), Brorsen,
Chavas and Grant (1987), and Aradhyula and Holt (1989) found that
risk significantly reduces output. Similarly, subsistence farmers in
developing agriculturein Thailand (Behrman 1968), Kenya (Wolgin
1975); Niger (Adesinaand Brorsen1986), and Mexico (Moscardiand
de Janvry 1977) decrease productionwithincreasedrisk. Filipinorice
farmers have likewise been shown to moderately reduce levels of
fertilizer usage with increasing risk (Rosegrant and Herdt 1981;
Rosegrantand Roumasset1985). Knowledgeaboutthe impactof risk
on Filipino farmers' supply decisions has important policy
implications.If risk has a large negative impact on area and levels of
output, stabilization policies aimed at reducing risk may generate
large social benefits. Additionally, models estimating supply
decisionsunder the risk-neutralassumptionmay be misspecified,if
risk is an important variable. Because of statistical problems in
omitting relevant variables, there could be efficiency gains in
incorporating measures of risk in modeling farmers' production
decisions.
*Mendozaisa ResearchAnalystat the International
FoodPolicyResearchInstitute
(IFPRI), Brorsenis an AssociateProfessorat the Departmentof AgriculturalEconomics,
OklahomaState University,and Rosegrantis a ResearchFellowat IFPRI.
Theauthorsacknowledgethehelpfulcommentsonearlierversionsofthispaperfrom
DeborahBrown,LeeSchrader,AkinwumiAdesinaandtwoanonymousJouma/ reviewers
forsuggestions
that improvedtheexposition.Theviewsexpressedheredonotnecessarily
representthose of IFPRI.
100
JOURNALOF PHILIPPINE DEVELOPMENT
This studyanalyzestheimportanceof riskincornarea decisions
of Filipinofarmers. About30 percentof total agriculturalarea is
plantedto corn (NationalEconomicand DevelopmentAuthority
1991). Corn alsocontributed17 percentto grossvalue added in
agriculture
in 1988(NEDA1991). It isalsoconsumedasastaplefood
carbohydratein placeof rice by some Filipinos,especiallythose
locatedin the VisayasandMindanao areas. Over the past two
decades,corn hasbeen increasinglyused in the manufactureof
feedsfor poultryandlivestockraisingandthistrendis expectedto
continuein thecomingyears.
StudiesonricearearesponseinthePhilippines
haveassumedthat
producersare riskneutral(Mangahaset al. 1965;SisonandHayami
1977).Severalfactorssuggestthatriskmaybe importantin thecorn
productiondecisionsof Filipinofarmers.First,domesticcornprices
are characterizedby wide intertemporalvariations,indicatingthe
potentialimportanceof riskin cornproduction
decisions.Second,as
a smallworldproducerandnetimporterofcorn,Filipinofarmersare
highlyexposedto priceshockscomingfrom a very volatileworld
grainsmarket.Third,theheavydependence
of farmersontradersfor
price informationsuggeststhat farmers may be susceptibleto
"exploitation"
by traders.Furthermore,becausemarketinformation
servicessupposedlyavailableto thepublicare allegedlyinadequate
toprovidefarmerswithaccurateandreliablemarketnews(Dy 1988;
Olgadoet aL !977; DeomampoandSardido1982),Filipinofarmers
may be unableto accuratelyreadand interpretmarketsignalsand
translatethemintoprofits.Riskabsorption
by farmersmaythenbe
higherastheyare unableto transfertherisktothe market.
The susceptibilityof rural incometo variabilityin corn prices
indicatesthe importanceof riskinproducers'decisions,andthishas
beenusedtojustifygovernment
intervention
inthecornmarketinthe
Philippines.
Attemptstosecureruralincomeby reducingriskthrough
price stabilizationprogramsand regulatiOnof the importmarket
dominatedgovernmentprogramsfor many years. Importationof
grainshas been regulatedby the NationalFood Authority(NFA)
throughquotarestrictions,
includingimportbans,importtariffsand
importlicensing.
To containerraticpricefluctuations
andsustainfarm
income,theNFA hasattemptedtoinfluencedomesticcornpricesby
enactingsupportpriceand ceilingpriceprograms.It has likewise
directlyprocuredcorn at the farm level,thoughwithvery limited
success.The actualvolumeof cornpurchaseddirectlyby the NFA
accountedfor only2 percentof the totalproductionper year, on
MENDOZA et al.: AGGREGATE CORN AREA RESPONSE
101
average. This low level of purchases reflects the inability of the
Philippine government to defend its domestic price policies
(Pabuayon 1985). Additionally, because government-set prices are
low, farmers were found to have sold a considerable portion of their
corn harvest in the open market (Lantican and Unnevehr 1984).
To stimulate corn production to self-sufficiency levels, price control
programs have been complemented by production-enhancing
programs. These production-augmenting programs include the
National White Corn and Feedgrains Program in 1969-1973,
Masaganang Maisan in 1974-1977, Maisan 77 Program in 19771982, Maisagana Program in 1982-1984, the Expanded Yellow Corn
Assistance Program in 1984, and the Corn Productivity Enhancement
Program in 1990. Typically, these programs contain a
comprehensive package of technological and institutional support for
small corn farmers, of which technical assistance on the proper use
of fertilizers and chemicals, and of hybrid seed varieties, and
institutional linkages to obtain access to credit, product outlets,
marketing facilities, and cheaper inputs are salient features. Like
other intervention programs, however, production programs have
fallen short of targets. Knowledge of the importance of risk in farmers'
production decisions may be valuable in understanding its likely
impact in attaining development program objectives and evaluating
program benefits, if any.
The remainder of this paper is organized as followS: the theoretical
and empirical specifications of the risk-responsive area allocation
model are presented in section II and section III, respectively. The
data are described in section IV, and the results for the riskresponsive model and comparison with a risk-neutral model are
discussed in section V. Some policy implications of the study are
summarized in section VI.
THEORETICAL MODEL
The risk-responsivearea decision model specified,here assumes
that both productionand prices are stochastic.Output is represented
by a stochasticproductionfunction:
y-f(x,z,e)
(1)
where x is a vector of variable inputs, z is a vector of quasi-fixed
inputssuch as land, and e is the stochastic error term representing
weather and other random effects on production.
102
JOURNALOF PHILIPPINEDEVELOPMENT
Let p be the probabilitydistributionof the output price, w be the
vectorof variable inputprices,and qbe the vectorof quasi-fixedinput
prices or opportunitycosts.
The farmer isassumed to maximizethe expected utilityof wealth:
Max E U (w + py. rx- qz)
(2)
where wis initial wealth, pyis gross revenues, and (rx+ zq)is cost of
production.
Since both price and yield are stochastic, the optimal riskresponsive input demand and output will be a function of the joint
probability distribution of gross revenue (py). Define _-= E (py) as
expected revenues and o as the second moment and, if necessary,
higher moments of the joint probability distribution of crop revenues,
py. The variable p7= E(py) measures crop revenue risk. The oplJmal
level of use of quasi-fixed inputs, such as land, can then be expressed
as;
z*--- f(w, r, q, py, _)
(3)
Thus, from equation (3), area planted is a function of variable input
prices, quasi-fixed input prices, expected revenue, and revenue risk.
EMPIRICAL SPECIFICATION
Based on the theoretical model in equation (3) in Section II, the
empirical model used in estimatingthecorn area response of Filipino
farmers under risk is specifiedas:
CORNAREA t = f (CORNREVt. 1,RICEREVt. 1, RICERISK t
PUREAt, PAMOSOL r WAGE,, TIME_)
(4)
I
t2.1,/2
- CORNREVt.j.,))
J (S)
i,_j
=_10j RICEREVt4
- RICEREVt_.4
CORNRISK,
= ,j_1'(30j(CORNREVt/.
RICERISK,
=
)1"
(6)
MENDOZA et al.: AGGREGATE CORN AREA RESPONSE
103
where CORNAREA isthe area planted to corn, CORNREV is gross
corn revenue (price times yield) per hectare, and RICEREV is gross
rice revenue per hectare. By definitioh, corn area is specified as a
function of revenues rather than prices as is often utilized in arearesponse analysis. As noted b_/Rosegrant and Kasryno (1992) and
Sanderson, Quilkey and Freebairn (1980), the use of revenue is
theoretically preferable to price as an exogenous variable because
price and yield are nonstationary. The relative profitability of different
crops is a function of the productivity of the crop, as well as of the
price.
Rice revenue, RICEREV, represents the opportunity costs of
growing corn, as rice is the crop that competes directly with corn in
terms of land usage. Revenue expectations assume a naive form and
are equal to rates of return observed in the previous year. More area
will be planted to corn if there is an anticipated increase in returns to
corn. Thus, a positive relationship is expected between corn area and
expected returns to corn. If returns to growing rice increase relative to
those of corn, more land will be devoted to rice and less to corn, so
that corn area is postulated to be negatively related tO increases in
returns to rice.
The variable PUREA is the price of urea fertilizer and PAMOSOL
is the price of ammonium sulfate. Urea and ammonium sulfate are
the two most popularly-used inorganic fertilizers in the Philippines.
The cost of agricultural labor is included as WAGE. To capture the
effects of other exogenous factors on corn area, the variable TIME is
included.
Following the theoretical specification in equation (3), risk is
measured as a declining weighted three-year moving average of
squared deviations of corn revenues, CORNRISK, and rice revenues,
RICERISK. The weighing factors of revenues in corn and its
competing crop, rice, are specified as 8j over the previous three
years. As in Brorsen et al. (1987) and Adesina and Brorsen (1986),
variability in the most immediate past is expected to have a larger
influence on farmers' perception of risk than those in the distant past,
so the weights used here are 0.60, 0.25, 0.15 for j = 1,2, 3.
Sandmo (1971) and Ishii (1977) and Batra and Ullah (1974)
showed in a model of perfect competition under uncertainty for a
single product farm that if producers exhibit nonincreasing absolute
risk aversion (ARA), increased risk will result in a reduction in input
use. In a multiproduct farm (as in this study), the relationship between
104
JOURNAL OF PHILIPPINE DEVELOPMENT
risk and area is ambiguous and must be tested empirically. As
demonstrated in numerous other studies, the area allocated to the
planting of a crop decreases when the farmers' perception of risk in
growing that crop increases. In terms of revenue risk for the
competing crop, the relationship is positive as planting the other crop
is perceived to be a more risky venture.
For comparison, a constrained or risk-neutral area response model
is also estimated. In this model, the revenue-risk variables are
dropped from equation (4), While all other variables are retained. The
efficiency gains, if any, of using the risk-reSponsive area allocation
model can thus be identified.
THE DATA
The period of analysis covers 15 years from 1965 to 1989. Time
series data on corn area and yield were obtainedfromthe 1989 Food
and AgricultureProductionYearbook.The pricesreceivedby farmers
for corn and rice and the prices paid by farmers for urea and
ammonium sulfate are from the 1990 FAO Producer Prices. Real
wage rate indices collected from the National Economic and
Development Authority Statistical Yearbookwere used to reconstruct
the 1965-1972 values missing from the date available on real wages
for agricultural worker. All variables were transformed to their
logarithms prior to estimation.
As with most developing countries, the data obtained here may
likely have been measured with errors, particularly corn area, the
dependent variable (David et al. 1990). Random measurement errors
in the dependent variable corn area do not, however; create serious
econometric problems (Kmenta 1986).
RESULTS AND DISCUSSION
Table 1 summarizes the results of estimates of the unconstrained
(risk-responsive)andconstrained(risk-neutral)cornarea models. As
indicated by the adjusted coefficients of determination, the overall
explanatorypower of the models in explainingthe variation in corn
area decisionsof Filipinofarmers is relatively close. The overall Fstatistics are highlysignificantat the 1 percent level, indicating the
highexplanatorypower of bothmodels. Compared to the risk-neutral
model, most coefficients of the risk-responsive model have the
correct signs consistent with a priori expectations and are highly
MENDOZA et al.: AGGREGATE CORN AREA RESPONSE
105
I
ESTIMATED
Table 1
RISK NEUTRAL AND AREA-RISK RESPONSIVE
FOR CORN: PHILIPPINES, 1965-1989,
MODELS
Constrainted Model Unconstrained Model
Exogenous Variable
(risk neutral)
Intercept
Corn Revenue, previous year
(CORN REVt._
Rice Revenue, previous year
(RICEREV,.,)
Corn Revenue Variability
(RICERISK)'
Price of Urea (PUREA)
Price of Ammonium
Sulfate
(PAMOSOL)
Cost of Labor (WAGE)
Trend (TIME)
F statistic
Adjusted Coefficient of Determination
Durbin-Watson Test for Serial Correlation
'
Defined as (_. _j rREVENUE_i
' 3
,j--1
-
(risk responsive)
-2.17
(-5.52)'**
-0.002
(-0.25) n=
0.03
(1.08) "'
n,a,
n,a,
0.007
(0.86) _'
-8,47X10 .6
-2,48
(-4.59)***
0.03
(1.79)**
-0,02
(-0.44) n=
-0.04
(-0,44)"'
-0.004
(-0.32) _'
-5.04X10 "e
(-1.24)"'
-0.03
(-0.28) I'lS
0.38
(4,73)***
32.22***
0.89
10,4
(-0.32) °'
-0,10
(-0.73) ns
0.58
(3.36)***
15,29 ....
0.86
1,82
REVENUE tk:,._
J_
where REVENUE is the rate of return at the farm level of crop k, k
**
***
ns
equals 1 and 2 where 1 is corn and 2 is rice and O/= 6, .25, .15 for j= 1,
2,3.
Values in parentheses are t-statistics.
Significant at 10% probability level, one-tailed test.
Significant at 1% probability level, one-tailed test.
Statistically not significant.
106
JOURNAL OF PHILIPPINE DEVELOPMENT
significant at conventional significance levels, suggesting that they
provide.a slightly superior fit than the constrained model. Moreover,
results of the Durbin-Watson test indicate that the residuals obtained
from the unconstrained model are white noise cofnpared to the
serially-correlated error terms from the constrained model. This result
indicates that there are efficiency gains in incorporating measures of
risk in modeling corn production because estimates of tt_e standard
errors are efficient, providing greater confidence in the power of
statistical tests employed and in the validity of inferences drawn from
the area-risk response model. Subsequent discussion is thus focused
on results of the corn area-risk :model.
Results show that expected revenues of corn, CORNREVt_1,exert
a statistically significant and positive effect on corn area, indicating
that more land is planted to corn as Ceturns to corn production
increase. As expected, anticipated increasesDin rice revenues,
RICEREVt_1,reduced the area planted to corn as farrners shifted more
land to the planting of the more profitablerice and less la_d to corn,
although the coefficient is statistically insignificant.
Although the coefficients on input prices have negative signs,.
prices of urea, PUREA, and ammonium sulfate, PAMOSOI, were not
statistically significant at conventional levels. This may be because
the level of fertilizer usage on corn farms is generally very low so that
changes in fertilizer prices do not exert a significant influence on corn
area allocation. The real wage rate, WAGE, had the expected
negative sign, but was also statistically insignificant, possibly
because of the high proportion of family labor used in corn production
which may not be as responsive to changes in wages as hired labor.
Of most interest is the result obtained for the corn revenue risk,
CORNRISK. The coefficient on corn revenue risk which was
statistically significant at the 10 percent probability level, had a
negative sign. This negative relation between risk and farmer's
production decision is consistent with the empirical findings obtained
by Behrman (1968), Lin (1977), Traill (1978), Adesina and Brorsen
(1986), and Brorsen, Chavas and Grant (1987), and with the theory
on risk of Sandmo (1971), Batra and Ullah (1974), and Ishii (1977).
This finding for Philippine corn suggests that increases in corn
revenue risk will reduce the quantity of corn locally produced as less
land will be planted by Filipino farmers to corn. Rice revenue risk,
however, does not have a significant effect on corn area decisions.
Since all variables are expressed in logarithms, the coefficients are
direct estimates of elasticities. As presented in Table 1, the negative
MENDOZAet el.: AGGREGATE CORN AREA RESPONSE
107
sign on the coefficient of CORNRISK indicates that corn area would
decrease with increases in expected variability in corn revenues, but
the magnitude of response is very small, -0.04. The inelastic
response of corn area to risk indicates that increased revenue risk
resulting from highly volatile local and world market conditions and
from the variability in corn yield does not translate into a large
decrease in land planted to corn. This result is consistent with the
findings obtained by Rosegrant and Herdt (1981) and Rosegrant and
Roumasset (!985) for Philippine rice farmers that there is a
statistically significant but small effect or risk on the level of input
usage. This inelastic area response to risk perception obtained for
Philippine corn is also comparable to that found in studies by Lin
(1977) and Brorsen, Chavas and Grant (1987) for U.S. agriculture.
SUMMARY AND POLICY IMPLICATIONS
This paper investigatedthe importanceof corn revenue risk in the
corn acreage decisions of Filipino farmers in the Philippines.As a
small world producer, Filipino farmers are exposed to the wide
variability in prices in thedomesticand worldmarkets, pointingupthe
significance of risk in corn production.The existence of imperfect
market knowledge and heavy dependence of farmers on traders for
price informationfurther suggestthat Filipinocorn farmers' exposure
to risk may be high, indicatingthe importanceof incorporatingrisk in
modelingfarmers' corn area decisions.
Results showed that perception of risk measured as variability in
corn revenue significantly reduces the area planted to corn. This
result provides evidence that revenue risk is a statisticallysignificant
factor in Filipino farmers' corn area decision making. Estimates of
cornarea response modelsthat do notaccountfor risk are likelyto be
biased, and inferences derived from their results may lead to
inappropriate policy recommendations. However, the magnitude of
area response to risk is very inelastic,as indicatedby the area-risk
elasticityof -0.04. Thus, increase in riskwould result in decreases in
land planted to corn, but the magnitudeof the reductionin corn area
would be small.
The findings on the degree of risk-responsiveness in Filipino
farmer's corn area decisions have important policy implications for
programs affecting price stability. The small negative effect of
increased revenue variability on corn area indicates that there are
likely to be only small benefits to corn price stabilizationprograms.
108
JOURNAL OF PHILIPPINE DEVELOPMENT
These results are consistent with Newberry and Stiglitz (1981), and
Spriggs and Van Kooten (1988) who showed that while there are
benefits to commodity price stabilization programs, the gains are
generally small. Net economic benefit of corn price stabilization
programs in the Philippines may be smaller considering that the total
outlays involved in implementing stabilization programs, particularly
programs on open market operations and grain stock holding, are
prohibitively high (Knudsen and Nash 1990). To the extent that the
desire is for price stabilization programs for corn to achieve other
social goals, such as stabilizing rural income, alternative policy
instruments .for achieving stability, such as variable import tariffs,
should be evaluated.
MENDOZA et al,: AGGREGATE CORN AREA RESPONSE
109
BIBLIOGRAPHY
Adesina, Adesina, and B. Waoe Brorsen. "A Risk Responsive
Acreage Response Function for Millet in Niger." Aoricu/tura/
Economics 1 (1986): 229-38.
Aradhyula, Satheesh,and MatthewHolt."Risk Behaviorand Rational
Expectations in the U.S. Broiler Market." American Journal of
Agricultural Economics 71 (1989): 892-902.
Batra, Raveendra, and Aman Ullah. "Competitive Firm and the
Theory of Input Demand Under Price Uncertainty." Journal of
Political Economy 82 (1974): 537-48.
Behrman, Jere R. Supply Response in Underdeveloped Agriculture.
Amsterdam: North Holland Publishing Company, 1968.
Brorsen, B. Wade; Jean-Paul Chavas; and Warren R. Grant. "A
Market Equilibrium Analysis of the Impact of Risk on the U.S. Rice
Industry." American Journa ! of Agricultural Economics 69 (1987):
733-39.
David, Isidoro;A. Nenette C. Mendoza; and David Besa. "Evaluation
of Philippine Corn Statistics." Journal of Ph/TippineDevelopment
17 (1990): 33-66.
Deomampo, Narciso R., and Moises Sardido. "Agricultural Marketing
in the Philippines." Papers and Proceedings .of the 28th Annual
Convention of the Philippines Agricultural Economics and
Development Association, pp. 162-83, 1982.
Dy, Rolando. "Market Information and Intelligence." Paper presented
at the International Seminar on Agricultural Marketing, Yogyakarta,
Indonesia, August 1-5, 1988.
Food and Agriculture Organization. Production Yearbook. Rome,
FAO, 1989.
• Producer Prices. Rome: FAO, 1990.
Hurt, Christopher, and Philip Garcia. 'The Impact of Risk on Sow
Farrowings, 1967-1978." American Journal of Agricultural
Economics 64 (1982): 565-68.
Ishii, Y. "On the Theory of the Competitive Firm Under Price
Uncertainty." American Economic Review 67 (1977): 565-68.
Just, Richard. "An Investigation of the Importance of Risk in Farmer's
Decisions." American Journal of Agricultural Economics 56 (1974):
14-25.
Kmenta, Jan. Elements of Econometrics. New York: Macmillan
Publishing Company, 1986.
110
JOURNAL OF PHILIPPINE DEVELOPMENT
Knudsen, O,, and J. Nash. "Domestic Price Stabilization Schemes in
Developing Countries." Economic Development and Cultural
Change 38 (1990): 539-58.
Lantican, Flor, and L. Unnevehr. "Rice Price and Marketing Policy."
Paper presented at the Workshop on Agricultural Policy, May 3-4,
1985.
Lin, William. "Measuring Aggregate Supply Response Under
Uncertainty." American Journal of Agricultural Economics 59
(1977): 903-907.
Mangahas, M.; A. Recto; and V. Ruttan. Production and Market
Relationships for Rice and Corn in the Philippines. Los BaSos:
International Rice Research Institute, 1965,
Moscardi, Edgardo, and Alain de Janvy. "Attitude toward Risk among
Peasants: An Econometric Approach." American Journal of
Agricultural Economics 59 (1977): 710-15.
National Economic and Development Authority. Philippine Statistical
Yearbook. Manila.
Newberry, D. M., and J. E. Stiglitz. The Theory of Commodity Price
Stabilization. Oxford: Clarendon Press, 19811
Olgado, S.; J. Abunyawan; and B. Domingo. Yellow Corn MarketingPart ill: Ilocos Region. Philippines: Special Studies Division, Office
of the Secretary, Department of Agriculture, 1977.
Pabuayon, Isabelita. "Corn MarReting in the Philippines: An Overview
and a Review of Trade and Pricing Policies." Journal of Agricultural
Economics and Development 15 (1985): 55-80.
Rosegrant, Mark W., and R. W. Herdt. "Simulating the Impacts of
Credit Policy and Fertilizer Subsidy on Central Luzon Rice
Farmers." American Journal of Agricultural Economics 63 (1981 ):
655-665.
and James A.Roumasett. '_l'he Effect of Fertilizer on Risk:
A Heteroskedastic ibroduction Function With Measurable
Stochastic Inputs." Australian Journal of Agricultural Economics 29
(1985): 107-121.
and Faisal Kasryno. Supply Response With Endogenous
Technology: Food Crops in Indonesia, 1992 (mimeographed).
Sanderson, B. A.; J. J. Quilkey; and J. W. Freebairn. "Supply
Response of Australian Wheat Growers." Australian Journal of
Agricultural Economics 24 (1980): 129-40.
Sandmo. A. "On the Theory of Competitive Firms Under Price
Uncertainty." American Economic Review 61 (1971): 65-73.
MENDOZA et all.:AGGREGATECORN AREA RESPONSE
111
Sison, Jerome, and Yujiro Hayami. "Structural Changes in Rice
Supply Relations in the Philippines." Journal of Agricultural
Economics and Development 7 (1977): 166-85.
Spriggs, J., and G. C. Van Kooten. "Rationale for Government
Intervention in Canadian Agriculture: A Review of Stabilization
Program." Canadian Journal of Agricultural Economics 36 (1988):
1-21.
Traill, Bruce. "Risk Variables in Econometric Supply Response
Models." Journal of Agricultural Economics 29 (1978): 53-61.
Wolgin, J. M. "Resource Allocation and Risk: A Case Study of
Smallholder Agriculture in Kenya." American Journal of
Agricultural Economics 57 (1975): 622-30.