Simulation of Production Among the Kapauku

  Background to a Simulation by Michael Fischer

This is a simulation model which attempts to describe the particular
agricultural land allocation for the production of sweet potatoes among the
Kapauku of W. Irian as described by  L. Pospisil in Kapauku Economy in
1956.  The model is very simple, treating the agricultural efforts of the
Kapauku people as  a group, and avoiding other agricultural production.
Although the model is simple, it outlines the essential elements of Kapauku
land allocation for sweet potato production.

Rainfall is the independent structuring principle for the model. Rainfall is
roughly seasonal in the valley, with a relatively dry  part of the year, and a
relatively wet part of the year. This  seasonality is not absolute with respect to
time.  The onset of the et' season is highly variable  pg.  Aside from Pospisil's
remarks, the seasonality is also suggested by comparative rainfall data from
the New Guinea/W.Irian area, all showing pronounced et'  and   periods of the
year  [Chart]

For the purposes of the model, only two rain states were considered, the 'dry'
state, and the 'wet' state.  Variability in the rainfall is expressed by varying the
onset of the 'wet' state.  Specifically,  in the model the 'wet' state is designated
to begin in up to 4 different months, specifically April, May, June, or July,
depending on the run, although the time of year is not  critical.  The
likelihood of the 'wet' season beginning in any of the above months is a linear
random function.  Once begun, the length of the wet season is fixed at months.
The onset of the rainy state is independent of any other event, including the
previous onset.

Planting represents the only decision within a given run. Planting is based on
three factors:

  Labour availability, which is taken to be 35220  hours per year,
  given by Pospisil for the year 1956.

  Time since last rain. The time interval given by Pospisil for maturity
  of a sweet potato garden is 8 months.  Pospisil states that if the
rainfall is excessive, the crop will rot. This is supported by other
literature on sweet potatoes, which report that sweet potatoes require
a hot, dry period to mature. The time since the last rain is important
because it is assumed in the model that no Kapauku would plant if the



 


  Simulation of Production Among the Kapauku     2

  destruction of the crop is certain, although the Risk Factor defines
'certainty'.

  The risk factor is a global decision.  Once made, it will hold for an
  entire run of the simulation.  This decision is the mechanism used to
test different strategies of land allocation between valley and
mountain planting.  Mountain land is considered to be impervious to
variations in rainfall, whereas valley land is assumed to be a total loss
if any part of its development falls within the 'wet' state.  The risk
factor determines how much risk will be taken with respect to
planting closer to critical times which could fall within the domain of
a 'wet' state.  A risk factor of 0 would insure planting where the
returns are certain.  Any higher risk factor involves some risk. This
risk is realized in the model by having the Kapauku plant more
months past the end of the 'wet' season.

Planting is simplified to two stages, a safe period and a risky period. The use
of the risky period is determined by the global decision, the risk factor.  The
only land which is explicitly bounded is ICC, valley garden plots.  Other land
usage will be bounded by labour availability. The ranking of the land is ICC,
ISC, and ESC, although only ESC will be used outside of the risk window, as
defined by the Risk Factor.

The model function Growth simply kills all valley crops [ISC and ICC] whose
maturation intersects the 'wet' season.  All mountain crops [ESC] are
considered to mature.

Harvest searches the crops, calculating yields based on the number of pekas
planted times the yield per peka given by Pospisil, 730 kg for ESC, 1240 kg
for ISC, and 1520 kg for ICC. The harvest yields are kept on a monthly basis
for use in the eating section.

The Eating section first subtracts human needs from the monthly yield totals.
These are estimated by Pospisil at 7400 kg per month. After humans are fed,
pigs are fed.  Pig rations are estimated by Pospisil at about 4 kg per day.  First
an attempt to feed the pig population at the rate of 4 kg is made.  If this is not
possible, an attempt to feed them 3 kg is made. If this fails, a pig dies, and a
counter is initialized which will kill more pigs in later months if they are not
fed.  At the end of each year, 5 pigs are added to the population.   The initial
pig population is set at 30, the 1956 level.

At the end of each year data is collected for the year, based on data collected



 


  Simulation of Production Among the Kapauku     3

monthly.  Data is collect for 9 variables:


  ESC. This is the total amount of ESC land [mountain land] planted during
  the year.

  ISC. This is the total amount of ISC [valley mixed plots] planted during
  year

  ICC. This is the total amount of ICC [valley garden plots] planted during a
  year.

  +ESC. This is the amount of successfully cultivated ESC land.

  +ISC. This is the amount of successfully cultivated ISC land.

  +ICC. This is the amount of successfully cultivated ICC land.

  HARVEST. This is the total harvest for the year.

  EXCESS. This is the surplus after feeding humans and pigs for the year.

  PIGS. This is the number of pigs at the end of the year.



Conclusions

This model consists of one independent variable, the onset of the rain, and one
decision, the amount of risk to be considered in the allocation of labour for
planting.  The results indicate a relatively stable structure with respect to the
various strategies [Risk factors].  A risk factor  between 1.0 and 1.1 is the best
strategy for optimizing yield, and maintaining a relative stable food supply.  A
risk level of 0 seems to suggest the optimum  level for stability of the food
supply, and higher risk levels optimize short term profits.

One conclusion of these results might be to explain the lack of planting of the
ISC land by the Kapauku for sweet potatoes. Given that such planting appears
to b solely a short term profit motivated activity, it is suggested that the
Kapauku reserve this land for luxury crops, such as sugar cane, which will
realize large profits when successful, and can be planted with less labour than
sweet potatoes.



 


  Simulation of Production Among the Kapauku     4

Another important aspect of the results is to lend support to the model
embodied in the simulation as a model of the decision structure of Kapauku
agricultural land allocation.  If it is assumed that the  Kapauku want to
maximize yield and minimize risk, then present model provides motivation for
a simple principle to accomplish this goal#: Plant valley land so long as its
'safe', and mountain land the rest of the time.  Safe is hence defined as a
period of two months after the last heavy rain [Risk Factor=1.0], a time
interval which is supported by all values of the Rain Window. The difference
between gambling and safety is thus clearly marked.   The decision for any
strategy is reduced from a labour and land management problem, to a strictly
environmental problem, plant while you can, or think you can.  It is notable
that the resulting values of ESC, ISC, and ICC land the simulation allots at the
optimum Risk Factor 1.0 is very close to the actual allotment of 1956 as
reported by Pospisil.



A run over 30 trials of the simulation: