demo7.gms : Nonlinear Simple Agricultural Sector Model

Description

This is the last in a series of agricultural farm level and sector
models, this model simulates the market behavior of the sector
using a partial equilibrium framework. The technique is
the maximization of consumers and producers surplus.


Small Model of Types : NLP lp


Category : GAMS Model library


Main file : demo7.gms

$title Nonlinear simple Agricultural Sector Model (DEMO7,SEQ=92a)

$onText
This is the last in a series of agricultural farm level and sector
models, this model simulates the market behavior of the sector
using a partial equilibrium framework. The technique is
the maximization of consumers and producers surplus.


Kutcher, G P, Meeraus, A, and O'Mara, G T, Agriculture Sector and
Policy Models. The World Bank, 1988.

Keywords: linear programming, nonlinear programming, farming, agricultural
          economics, market behavior
$offText

Set
   c       'crop'  / wheat, clover, beans, onions, cotton, maize, tomato /
   cl      'livestock feed' / clover, straw /
   t       'month' / jan, feb, mar, apr, may, jun, jul, aug, sep, oct, nov, dec /
   r       'feeding recipes' / rec-1 , rec-2  /
   s       'seasons'         / summer, winter /
   sc(s,c) 'season crop mapping' / summer.(cotton,maize,tomato)
                                   winter.(wheat,clover,beans,onions) /
   cn(c)   'crops sold in national market'
   ce(c)   'export commodities'
   cm(c)   'import commodities';

Table a(t,c) 'months of land occupation by crop (hectares)'
         wheat  clover  beans  onions  cotton  maize  tomato
   jan     1.     1.     1.      1.
   feb     1.     1.     1.      1.
   mar     1.      .5    1.      1.       .5
   apr     1.            1.      1.      1.
   may     1.                     .25    1.      .25
   jun                                   1.     1.
   jul                                   1.     1.       .75
   aug                                   1.     1.      1.
   sep                                   1.     1.      1.
   oct                                   1.      .5     1.
   nov      .5     .25    .25     .5      .75            .75
   dec     1.     1.     1.      1.

Table lc(t,c) 'crop labor requirements (man-days per hectare)'
         wheat  clover  beans  onions  cotton  maize  tomato
   jan    1.72     4.5    .75    5.16
   feb     .5      1.     .75    5.
   mar    1.       8.     .75    5.       5.
   apr    1.              16.   19.58     5.
   may   17.16                   2.42     9.    4.3
   jun    2.34                            2.    5.04
   jul                                    1.5   7.16     17.
   aug                                    2.    7.97     15.
   sep                                    1.    4.41     12.
   oct                                   26.    1.12      7.
   nov    2.43     2.5    7.5   11.16    12.              6.
   dec    1.35     7.5    .75    4.68                       ;

Table lio(cl,r) 'livestock input output matrix'
            rec-1  rec-2
   clover     1.3    2.0
   straw      1.6     .8;

Table demdat(c,*) 'demand data'
            ref-p   ref-q  elas  exp-p  imp-p
*              ($) (1000t)          ($)    ($)
   wheat      100    2700   -.8           140
   beans      200     900   -.4           270
   onions     125     700  -1.      40    inf
   cotton     350    2100  -1.     300    inf
   maize       70    3800   -.5            85
   tomato     120     500  -1.2     60    inf;

Scalar
   fnum   'number  of  farms in sector'                   / 1000    /
   land   'farm size                          (hectares)' /    4.   /
   famlab 'family labor available       (days per month)' /   25    /
   dpm    'work days per month'                           /   25    /
   rwage  'reservation wage rate       (dollars per day)' /    3    /
   twage  'temporary labor wage        (dollars per day)' /    4    /
   llab   'livestock labor requirements (days per month)' /    2    /
   trent  'tractor rental cost      (dollar per hectare)' /   40    /
   hpa    'land plowed by animals  (hectares per animal)' /    2    /
   straw  'straw yield from wheat'                        /    1.75 /;

Parameters
   yield(c)   'crop yield         (tons per hectare)'
              / wheat  1.5, clover 6, beans  1, onions 3
                cotton 1.5, maize  2, tomato 3           /
   miscost(c) 'misc cash costs (dollars per hectare)'
              / wheat  10, beans 5, onions 50
                cotton 80, maize 5, tomato 50 /
   price(c)   'reference (observed) price  (dollars)'
   pe(c)      'commodity export prices     (dollars)'
   pm(c)      'commodity import prices     (dollars)'
   alpha(c)   'demand curve intercept'
   beta(c)    'demand curve gradient';

cn(c) = yes$demdat(c,"ref-p");
ce(c) = yes$demdat(c,"exp-p");
cm(c) = yes$(demdat(c,"imp-p") < inf);
cm("clover") = no;
price(c) = demdat(c,"ref-p");
pe(ce)   = demdat(ce,"exp-p");
pm(cm)   = demdat(cm,"imp-p");

beta(cn)$demdat(cn,"ref-q") = demdat(cn,"ref-p")/demdat(cn,"ref-q")/demdat(cn,"elas");

alpha(cn)          = demdat(cn,"ref-p") - beta(cn)*demdat(cn,"ref-q");
demdat(cn,"dem-a") = alpha(cn);
demdat(cn,"dem-b") = beta(cn);

display cn, cm, ce, price, pe, beta, alpha, demdat;

Variable
   xcrop(c)     'cropping activity                 (hectares)'
   yfarm        'farm income                        (dollars)'
   revenue      'value of production                (dollars)'
   mcost        'misc cash cost                     (dollars)'
   pcost        'tractor plowing cost'
   labcost      'labor cost                         (dollars)'
   rescost      'family labor reservation wage cost (dollars)'
   tcost        'total farm cost including rescost'
   flab(t)      'family labor use                      (days)'
   tlab(t)      'temporary labor                       (days)'
   xlive(r)     'livestock activity                   (units)'
   natprod(c)   'net production                        (tons)'
   thire(s)     'tractor rental             (hectares plowed)'
   natcon(c)    'domestic consumption             (1000 tons)'
   natprice(c)  'domestic price             (dollars per ton)'
   exports(c)   'national exports                 (1000 tons)'
   imports(c)   'national imports                 (1000 tons)'
   cps          'consumers and producers surplus'
   valpro       'value of net production at ref prices'
   employ       'employment generated             (man-years)'
   tradebal     'net exports                         (1000 $)';

Positive Variable xcrop, xlive, thire, flab, tlab, natcon, natprod, exports, imports;

Equation
   landbal(t)   'land balance                        (hectares)'
   laborbal(t)  'labor balance                           (days)'
   flabor(t)    'family labor balance                    (days)'
   plow(s)      'land plowed              (hectares per season)'
   arev         'revenue accounting                   (dollars)'
   ares         'reservation labor cost               (dollars)'
   acost        'total cost accounting                (dollars)'
   amisc        'misc cost accounting'
   aplow
   alab         'labor cost accounting                (dollars)'
   lclover      'clover balance'
   lstraw       'straw balance'
   income       'income definition                    (dollars)'
   dem(c)       'national demand balance            (1000 tons)'
   pdef(c)      'price definition'
   arevn        'revenue accounting                    (1000 $)'
   arevf        'revenue accounting: fixed price model (1000 $)'
   valproc      'value of net production definition    (1000 $)'
   proc(c)      'net production definition               (tons)'
   employc      'employment definition              (man-years)'
   tradebalc    'trade balance definition              (1000 $)'
   objn         'objective function';

landbal(t)..  sum(c, xcrop(c)*a(t,c)) =l= land*fnum;

laborbal(t).. sum(c, xcrop(c)*lc(t,c)) + sum(r, xlive(r))*llab =l= flab(t) + tlab(t);

amisc..       mcost   =e= sum(c, xcrop(c)*miscost(c));

alab..        labcost =e= sum(t, tlab(t)*twage);

ares..        rescost =e= sum(t, flab(t)*rwage);

aplow..       pcost   =e= sum(s, thire(s)*trent);

acost..       tcost   =e= mcost + labcost + rescost + pcost;

lclover..     xcrop("clover")*yield("clover") =g= sum(r,xlive(r)*lio("clover",r));

lstraw..      xcrop("wheat")*straw =g= sum(r,xlive(r)*lio("straw",r));

plow(s)..     sum(c$sc(s,c), xcrop(c)) =l= sum(r, xlive(r))*hpa + thire(s);

proc(c)..     natprod(c) =e= xcrop(c)*yield(c);

dem(cn)..     natcon(cn) =e= natprod(cn) + imports(cn)$cm(cn) - exports(cn)$ce(cn);

objn..        cps =e= sum(cn, alpha(cn)*natcon(cn) + .5*beta(cn)*sqr(natcon(cn)))
                   +  sum(ce, exports(ce)*pe(ce))
                   -  sum(cm, imports(cm)*pm(cm))
                   -  tcost;

* the next 5 equations are only reporting or accounting definitions.
* they do not effect the model.

arevn.. revenue =e= sum(cn, alpha(cn)*natcon(cn) + beta(cn)*sqr(natcon(cn))
                 +  exports(cn)*pe(cn)$ce(cn));

arevf.. revenue =e= sum(cn, natprod(cn)*price(cn)
                 + (exports(cn)*(pe(cn)-price(cn)))$ce(cn));

valproc.. valpro =e= sum(c, natprod(c)*price(c));

income..  yfarm  =e= revenue - tcost + rescost;

employc.. employ =e= ( sum(t, sum(c, xcrop(c)*lc(t,c)))
                     + sum(r, xlive(r)*llab))/dpm/12;

tradebalc.. tradebal =e= sum(cn, exports(cn)*pe(cn)$ce(cn)
                               - imports(cn)*pm(cn)$cm(cn));

pdef(cn)..  natprice(cn) =e= alpha(cn) + beta(cn)*natcon(cn);

flab.up(t) = famlab*fnum;

Model
   demo7f 'fixed price' / landbal, laborbal , plow , ares   , arevf
                          alab   , acost    , dem  , employc, proc
                          valproc, amisc    , aplow, lclover, lstraw
                          income , tradebalc                         /
   demo7n 'nonlinear'   / landbal, laborbal , plow , ares   , arevn
                          alab   , acost    , dem  , employc, proc
                          valproc, amisc    , aplow, lclover, lstraw
                          income , tradebalc, pdef , objn            /;

* find a good starting point for the NLP by solving a fixed price LP model.
* the LP has many solutions: find one with moderate domestic consumption
natcon.up(c) = 1e5;
solve demo7f maximizing yfarm using lp;
natcon.up(c) = INF;

natprice.l(cn) = demdat(cn,"ref-p");

* the first lp found a good primal starting point. we can help
* the nlp step to provide some additional dual information.
* the equation arevf is replaced by arevn and pdef is new and
* should be binding.

pdef.m(cn) = 1;
arevn.m    = arevf.m;

solve demo7n maximizing cps using nlp;