1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
A System Dynamics Model of Soybean Production in India-
B S Bisht, K C Sahu and K Vizayakumar
De of Industrial Engineering & Management
Indian Institute of Technology
Kharagpur 721 302, India
Abstract
An effort has been made to develop a system dynamics simulation model for soybean production in
India. Soybean has been a relatively new introduction to Indian agriculture as a cash crop. Soybean
industry in India is typified by high (1520\%) annual growth in area as well as production. It is also
typified by negligeble consumption on farms and far advancing rate of growth of processing
capacity as compared to the annual production. Soymeal has been a major earner of foreign exchange
for the country with advancing years. The model d here analyses the hanics of flow of
area between various crop options, productivity of soybean and competing crops, annual production,
prices and the benefitcost ratio. The values of constants were determined through survey of farmers
and expert opinion poll. Policy options have been analysed with the objective of higher profitability
for the growers in the long run.
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
A System Dynamics Model for Soybean Production in India
BACKGROUND
Soybean (Glycine max) has become one of the important oilseed crops in India. From an area of
nearly 5000 hectares (ha) during the year 1960-61, it occupied 0.6 million hectares (m ha) during
1980-81 and an estimated 3.8 m ha by the year 1993-94 (Pandey and Biswas, 1986, Bhatnagar and
Tewari, 1991, Anon, 1993). Soybean contributed 7.6 \% of the total edible oil requirements in the
country during the year 1992-93. Exports of soymeal during the same year were valued US \$ 382
million (Anon, 1993). Soybean processing industry consisted of nearly 117 plants and employed
nearly 30,000 people (Jain 1992, Ali 1992).
Soybean is basically a leguminous crop. It consists of 38-42\% proteins, 18-22\% edible oil, 22-23\%
carbohydrates, 5\% minerals and the rest as moisture. Soya protein is considered to be of high quality
(Miner 1976, Wolf 1983) as food. However, soya foods developed and launched by a number of
companies and R\&D units could not gain wide acceptance because of typical beany flavour and
unfamilier taste. Soya oil have been successful in penetrating the kitchens of the households primarily
because of its lower price as compared to conventionally used oils. Also, extruded soya flour
(chunks) have gained some degree of consumer acceptance as there were 3 manufacturers in
organised sector besides 20 small ones in unorganised sector by the year 1992-93 together producing
about 40,000 t of soya chunks annually.
Soybean is primarily raised as summer crop (called 'Kharif’) in India. It is sown in the months of
June-July. The crop gets ready for harvesting within 95-130 days of sowing (Bhatnagar and Tewari,
1991). Among the reasons cited for high growth of soybean area are; a) its potential as an edible oil
crop at a time when the country was importing edible oils to the tune of 0.5-1.0 mt annually (Anon,
1993), b) absence of any remunerative crop during Kharif
specifically in black cotton soil belts in central parts of India, and c) success of local R\&D in
adopting, and developing location specific suitable varieties, and enthusiasm and faith of farmers
towards these. Yet another reason for the unprecedented growth recorded by soybean has been the
establishment of large number of process plants which created a ready and competitive market for
soybean (Varadrajan 1992, Jain 1992).
This study was with the objective of ing various policy options using system
dynamics, to ensure long run profitability for the growers.
MODEL DESCRIPTION
Influence diagram for soybean production in India has been shown in Fig.1. It consists of basically
the following major loops;
a) Soybean area--production--price--profitability--area, and
b) Cr ‘itive crops duction--price--profitability--area.
Industry, page 11
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Fallow
se. +
Area Aten ¢
under —,
competitive
$01
ma & crepe
—— ers? >
Sey: * + in + Gapetitive
Rto Reo crops
input +f fe input production =
+ bs w
+ "
yield rate
+ Gf \o. ?
Govt. support Mor. Max. 7) nis
rice yield yield en aupatt
potentiat potential Zo
= Competing oe
Soya Seve crops A ——~~__ Competitive
demand price a denees
- he: gettting
in Competitive tout ot
per hectare reps competitive
0 income Ber Decare crops
m seene preduction /
mA \ cea a |
Area shift: ee Crops = Area shift:
fallow to 4 fab . profitability + fa oy
woyo We totic 9 competitive
+ ditterentiot a
Profitability
Area shift: — Area shitt
competitive + soyo tk
trop competitive
te soya
crops
FIG. 1: INFLUENCE DIAGRAM FOR SOYBEAN
PRODUCTION, AREA , SELLING PRICE
AND FARMERS’ INCOME .
Industry, page 12
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Land area allocated to soya (AAS), area of competitive summer crops (AAC), and area under fallow
(current fallow+converible waste land) (AUF), have been taken as the level variables (Fig. 2). A
similar work reported by Chembezi (1991) for tobacco farmers in Malavi reported that the farmers
diverted their land for tobacco from maize keeping in view the price benefits and the risk factors.
Williams et al (1974) assessed that the potential of soya area in central parts of India was 2.03 m ha.
They assumed 10\% land contributed by kharif fallow and sorghum and 5\% by minor millets and
upland paddy. Work was also reported by Wilkerson et al (1983) on development of SOYAGRO
model for short run decision making in view of pests and irrigation management. The model
presented here comprises of the following key relationships for long run policy modelling using
system dynamics. To start with, the equations for level variables representing area allocated to soya
(AAS), to competing crops (AAC) and area under fallow land (AUF) have been;
L AAS.K=AAS.J+DT*(AFFS.JIK+AFCS.JK-ADSF.JK-ADSC.JK)
L AAC.K=AAC.J+DT*(AFFC.JK+AFSC.JK-ADCS JK-ADCF.JK)
L AUF.K=AUF.J+DT*(AFSF.JIK+AFCF.JK-AAFS.JK-AAFC.JK)
The rate variables connecting the above levels have been expressed as;
R ADSF.KL=AAS.K*FSF.K
R ADSC.KL=AAS.K*FSC.K
R ADCS.KL=AAC.K*FCS.K
R ADCF,KL=AAC.K*FCF.K
R AAFS.KL=AUF.K*FFS.K
R AAFC.KL=AUF.K*FFC.K
Rate variables namely AFFS, AFCS, AFFC, AFSC, AFSF, and AFCF have been defined as the
delayed rates of AAFS, ADCS, ADSC, ADSF and ADCF respectively (Fig. 3). The auxiliary
variables FSF, FSC, FCS, FCF, FFS, and FFC have been expressed as TABHL functions of DFFS,
DFCS,. DFSC, DFFC, DFSF, and DFCF respectevely. Further these auxiliary variables have been
worked out as;
A DFFS.K=CLIP(SBCR.K,0.0,0.001,SBCR.K)
A DFCS.K=CLIP(DBCR.K,0.001,DBCR.K)
A DFSC.K=CLIP(DBCR.K,0.0,DBCR.K,0.0)
A DFFC.K=CLIP(CBCR.K,0.0,0.001,CBCR.K)
A DFSF.K=CLIP(SBCR.K,0.0,SBCR.K,0.0)
A DFCF.K=CLIP(CBCR.K,0.0,CBCR.K,0.0)
The profitability indicators SBCR, CBCR, DBCR have been defined as;
A SBCR.K=SONI.K/SCOP.K
A CBCR.K=CCNIK/CCOP.K
A DBCR.K=(SBCR.K-CBCRB.K)/SBCR.K
A CBCRB.K=CLIP(CBCR.K,0.0,CBCR.K,0.0)
Industry, page 13
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
.
; \
\ !
‘ Area
|ADSF Shift Rate:
Soya to
Fallow
IDEL 3 DEL 3}
ref au AFFS AAS AFSC
Qn. TAFFS| x TAFSC|
Fallows, ~ Area ~~-mADSC
land \. Area Shift Rate : Under Area Shift
\ Fallow to Soya Soya Rate : Soya to's,
\
Competing Crops N
Area \
“ Area Shift Rate: Competing Area Shift Rate : \
x Fallow to Crops { Competing Crops to
/ Competing Crops 4 Soya DBCR
v ‘\ Differential ,
DEL 3 ‘NS Benefit- Cost \
AFCE] Sm Ratio Soya
to Competing
AF CF
uate AOCF Crops
Area Shift Rate:
CBCR ra Competing Crops
- ca to Fallow
Competing a
\ -
Crops woo”
¥ Benefit }
~ Cost
Ratio
FIG.2: SHIFTING OF LAND AREA BETWEEN SOYA,
COMPETING CROPS AND FALLOW.
Industry, page 14
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Soya per hectare net income (SONI), and that of cmpetitive crops (CCNI) were calculated by
deducting the cost of production of these crops from their revenues (Fig. 4). Gross income of the
farmer per unit area (GRIPH) was calculated on annual basis where it was taken as a level variable
for the sake of simpli in ions and ing (Fig. 4). The selling price received for
soya (SPS) was modelled as a “function of the cost of soya production (SCOP), support price
announced by the government (SUPPS), and the TABHL function of the ratio of order backlog for
soya (OBFS) to the soyastocks (SOYSTOC) (Fig. 3). Cost of production of soya (SCOP), and that of
competitive crops (CCOP), yield rate of soya (SYPH) and that of competitive crops (CYPH) were
defined as varying with time. Gompertz growth models were fitted to the historical data to arrive at
the representative relationships as defined by Martino (1983).
The model was run by using DYSMAP2 simulation package (Vapenikova,1987). The base year for
the simulation was taken to be 1980 (starting January 01). The model was simulated for a period of
300 time units (months) ending by December 2004 A.D..
MODEL BOUNDARIES
The model is better applicable to soybean farming states in central parts of India. Efforts have been
made to keep the model simple and objective oriented (as argued by Forrester, 1987). The model does
not include the of 01 hip of land infh ing the yield rates and returns (as discussed by
Saeed, 1982) and assumes inequalities to be smoothened out on aggregation.
MODEL BEHAVIOUR
The model could reproduce cyclic behaviour of soya price (SPS), and inventory on farms
(SOYSTOC) as revealed in Fig. 5. Soya production being seasonal in nature, soya inventory builds
up on farms after harvest each year by October-November months. The farmers start releasing the
stocks gradually. The stocks get released depending upon the relative profitability (SBCR value)
prevailing during the time. Hence, soya price start falling by October and keep downward trend
touching annual lowest in the month of January (Fig. 5). The price starts rising as soon as the order
backlog to stocks ratio builds up. The price reaches its peak for the year by the month of August and
September. Soya price as well as production and stocks reveal chaotic situation when the plots are
obtained for the entire length of simulation time. This kind of behaviour have been analysed by
Anderson and Sturis (1988). They have argued that deterministic chaos may be one of the possible
modes in a number of ly occuring in ic, and
managerial systems.
The simulated values of annual average price of soybean showed an increasing trend all through the
advancing simulation time (Fig. 5). The model could not capture the erratic behaviour of price
observed historically for the years of abnormal rains and price falls on expectations of edible oil
imports.
Industry, page 15
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
RPHS
Revenues Per
Unit Area of soya
,
SOLP
Soya Oil
Price a
SOYSTOC
2 Soya Stocks
Soya Price Price Sensitivity /
‘ Leverage to Demand f
SMPR \ = 0 j . fOBFS
Soya Meal ‘ . fo n Orders
Price Sea ‘ ( PSSD } i \Backlog
. y
‘ ‘ ”
‘ t OBSRF)
\ Y
\ ‘
Maximum Price \ \ Orders Backlog
Feasible for to Stocks Ratio
Processors
“ Lowest Price . SSS
rig Acceptable to SS
s.
Farmers Bi Cost
Minimum Support Based
Price Fixed by Price
State
‘ .
‘ .
f
BSUPPS
Price
mo
x / ‘
Z \ wee SCOP ‘4
Base Year Inflation
Support
Cost of
Production
anne
FIG.3: DETERMINANTS OF SOYBEAN PRICE.
Industry, page 16
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
PMF
Production
Management
Factor
Gross Income ~
per Unit Areo
Income Rate
. Farmers? Income *S
per Unit Area m ‘
Consumption Rate \
——— 6
SPR scp
Profits Fraction Consumptien
er Contribution Fraction
‘Unit Area by Summer
Crops.
AAS
SSPS
Profits from Summer {ie Under Average Price
Agricutture 3 ‘eve “(ot Soya
% Revenue from
. a Unit Soyo
= Area
/
x SSYPH
Profits from i Soya Profits ~~} Average Soya
Competing Crops from Unit Area \ Yield Rate
/ SCOP
Soyo Average
\ Cost of Prodn.
aes SPFR Per Unit Area
Seasonality Fraction * Seco aity
of Competing * Fraction of
Crops Production practioner
sccop
Competing Crops
Cost
ef Production
Per Unit Areo
Revenue from Unit
Area of Competing
Crops
SCHP .
Average Sc SCYPH
Price of Competing Crops
Competing Average Yield
Crops
FIG. 4: FARMERS’ SAVINGS FROM SUMMER AGRICULTURE.
Industry, page 17
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Besides the base run, the following policies were tried for evaluation.
1; Step increase in support price of soybean (SUPPS) incorporated at TIME=168 i.e.
168th month from January 1980 (i.e. December 1994) by Rs. 1000/t. This was in
addition to Rs. 5800/t fixed by the government for the year.
2: Soya area growth stopped after 168th month.
3. Step increase in the demand for soybean by 2.0 mt on account of increased capacity for
processing by 168th month onwards.
4, Step increase in cost of production by Rs 500/ha at 168th month onwards in addition _to
the normally growing cost of production.
5. A joint policy (J) incorporating policy-3 and 4 together.
Effect on Soya Price:
Policy-1 and 4 resulted in price trends similar to that of the base run (Fig. 6). Policy-3 resulted in the
highest price of soya, and policies 2 and J produced lowest values of price in long run. The behaviour
could be explained by the argument that the soya price remains higher to that of the support price
announced by the government as long as the demand for soya from processors exceeds the supply.
Increased demand for soya wouldinflate the value of order backlog (OBFS) resulting in higher price
negotiated by the farmers in their favour. The farmers would therefore favour policy-3 in view of
receiving highest selling price for soya.
Effect on Soya Benefit-to-Cost Ratio:
Policy-3 resulted in highest soya benefit-to-cost ratio (SBCR) followed by policies 1, 2, J, and 4
tespectively. The base run SBCR ranked in the middle. Increasing demand would result in higher
price for soya, and hence higher profits as long as the production costs do not increase proportionally.
Policies J and 4 resulted in low SBCR as they incorp d cost of production of soya
(SCOP), hence lowered profitability. From the viewpoints of farmers therefore policy-3, and policy-1
would be desirable.
Effect on Area:
Policy-J resulted in highest area under soya cultivation followed by policies 4, 1, 3, B, and 2 (Fig. 7).
Increasing demand for soya resulting in increased price and increased profitability would motivate
farmers to put more and more area under soybean. Also the increased cost of production, turns the
crops less ‘ive as the leverage between their revenues per unit area of land may
not be able to compensate for the increased costs. This would force farmers to take away more area
from competitive crops and allocate it to soya. The policy of installing more capacity of processing
plants would therefore result in the highest soybean area as compared to other tested policies.
Industry, page 18
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
SBCR
3
a
ry 9000 F
=~
€ ¢ /
° ¢ Py
2 € 5
- fF 2 ss00+ ee /
g| 3 / / al
A ~ A
= © sooob ace
2) a Ly
Sip ©
e < 1 SOYA STOCKS
2 Q 7500 (sovstoc)
<
8
é 7000 1 pe spe sy cge openly
aM M 3 SN fF MOM OG Zz “
1992 1993 ~1996—
TIME (year and month)
Fig-5 Simulated seasonal behaviour of soya stocks on
farms, and soya price.
10) 16000} SOYA BENEFIT - COST
A RATIO (SBCR) Pa
« Bl
€ as
S
ost = 12000) 3
ad 1
e ANNUAL AVERAGE 2 1
w SOYA PRICE 8 j
0-8 G 8000 (sPs) ~\ 4 |
= —
. 7 4
Bs i B: Bose run |
o7 3 6000 a policies - Tlod |
:Policy -(34+4)
0 50 100 150 200 250 300
1980 TIME (months) 2008
Fig. 6 Soya price and benefit -cost ratio-
Industry, page 19
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
G 8: Bose run ~ ——
1: Poticy-1 i
2: Policy -2 ‘
3 Policy -3
ves Sr 4: Policy ~4 2
J+ Poliey-(344)
18-0- 4
2
2 a
£ 2
s1msh 43 AREA UNDER
£& € SOYBEAN (AAS)
re)
< <
< 70h af
AREA UNDER COMPETING 2
CROPS (AAC)
16-5} 1 8
3
+
16-0F 0
=
Lt
J
185 A 1 A 1
0 50 100 150 200 250 300
TIME (months)
Fig. 7: Area under soybean and competing crops.
& 4000
J B: Base run
= 1: Policy -1
< 2+ Policy -2
~ 3000} 3: Policy -3
wo 200 | Policy - 4 3
3 J: Policy-Q+4)
a i
gs BASE YEAR: 1980 (Jan) sean)
4 2000+ No 2
ra POLICY CHANGES EFFECTIVE 8
we FROM 1994 (168th month )
z “
Ca
“x «1000+
wd
=
4
& n 4 1
30 700 150 200 750 300
Industry, page 20
TIME (months)
Fig. 8: Farmer’s net surplus from summer crops.
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Effect on Farmers’ Net Surplus:
Farmers’ net surplus per unit area of cultivation (GRIPH), as reveled by the simulated results were
highest for policy-3 (Fig. 8). This was followed by policies J, 1, 2, B, and 4. Increased demand
resulted in increased price of soya and hence higher profitability. Under this condition the farmers
acquire additional bargaining power in view of higher competition among processors for buying
limited stocks. Policy-4 results in higher cost of production hence lowered profitability for farmers.
Thus the value of GRIPH was the lowest for this policy.
CONCLUSIONS
‘The area under soybean is expected to keep growing under all policies evaluated except for the case
when it is forced to remain static by the influence of exogenous factors. Also, with the growth in soya
area its production would grow. Increased demand of soya would enhance the growth of its area.
Soybean price showed seasonal variations within an year. The lowest and the highest being observed
to occur in the months of January and August respectively. The annual average price is expected to
increase with time. The policy of increased demand would result in the highest price increase
followed by other policies.
The benefit cost ratio for soya (SBCR) would have a declining trend in long run. The highest SBCR
was obtained for the policy of increased demand followed by the policy of increased support price for
soya and so on.
Farmers’ gross annual income from unit culti d area would be i for policy of i
demand followed by the policies of joint increase in demand and cost of production, increased
support price and so on. Hence, from the view point of farmers, the policy of increased demand could
be the most desirable among the tested policies.
REFERENCES
Ali Nawab, 1992, Soybean processing and utilization in India, unpublished background paper for the
planning p for of Asian Comp of the Global Network on Tropical and
Subtropical Soybean, Bangkok.
Anderson D.F. and Sturis J., 1988, Chaotic structures in generic management models, System
Dynamics Review, {\bf 4 (1-2)}, 218-232
Anon, 1993, Agrticulture, Economic Survey 1992-93, Govt. of India, 153-169.
Anon, 1993, Soybean paidawar main bridhi se videshi mudra ki prapti badhegi (increased soya
production would lead to increased foreign exchange), The Dainik Nai Dunia, Sept. 24.
Bhatnagar, P.S., and Tewari, S.P., 1990, Soybean varieties in India, The National Research Centre for
Soybean, Tech. bulletin no. {\bf 2}, 1-11.
Bhatnagar, S.P., and S.P. Tewari, 1991, Soybean in edible oil in India, The Agricultural
Situation in India, Ministry of Agri., Govt. of India, Aug., 295-301.
Chembezi,D.M., 1991, Modeling acreage response with risk considerations: The case of estate
tobacco in Malavi, Agricultural Systems, 427-438.
Coyle, R.G., 1977, Management system dynamics, New York, John Wiley Sons, 180-237.
Forrester J.W., 1987, Lessons from system dynamics modelling, System Dynamics Review, 3 (2),
136-139.
Jain, N.N., 1992, Soybean industry: p blems and p The Ec ic Times, Feb. 2, 1992.
Martino, J.P., 1983, Technol ing for decesion making, North Holland, 53-67.
Miner, B.D., 1976, Edible soy protei ional aspects of producing and ‘ing, Farmers
Coop. Service, U.S. Deptt. of Agri., FCS Res. Report 33 1-11.
Pandey, P.N.,and M.K. Biswas, 1986, Marketing of soybean in India, Deptt. of Rural Dev., Ministry
of Agri., Govt. of India, 34-181.
Industry, page 21
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Saeed, Khalid, 1982, Public policy and rural poverty: a system dynamics analysis of a social change
effort in Pakistan, Technology Forecasting and Social Change, 21, 325-349.
Shah, Suresh, 1992, Promishing outlook: edible oils, The Economic Times, Sept. 26.
Vapenikova, Olga, 1987, DYSMAP2 Implementation on 80386-based PC's, Salford University.
Varadrajan, V.K., 1992, Soya processing, limited crop, major hurdles, The Hindu Survey of Indian
Industry 1991, The Hindu, 389-391.
Wilkerson, G.G., J.W. Jones, K.J. Boote, K.T. Ingram, and J.W. Mishoe, 1983, Modelling soybean
growth for crop The Ti ions of the American Soc. of Agri. Engineers, 63-72.
Williams, S.W.,W.E. Hendrix, M.K. Von Oppen, 1974, Potential production of soybean in north
central India, International Soybean Program, University of Iinois, Urbana-Champaign, tech. report
no. 5.
Wolf, W.J., 1983, Edible soybean protein products, CRC dbook of p ing and utilization in
Agriculture, 1 (2), Plant Products, 23-55.
Industry, page 22
