Sowing Supply: Compensating Responses by Rural Coca
Economies to Intervention in the Cocaine Market
Eric A. Wuestman
Rockefeller College of Public Affairs and Policy, State University of New York, Albany, N.Y. 12222
Abstract
Inrecent years, increased public awareness of the health and productivity costs associated with the use of cocaine and
its potent derivative “crack” has served to heighten concern and renew debate over the most effective strategies for
managing the drug problem. This paper presents a preliminary system dynamics model of the international cocaine
trade. The initial model incorporates the various stages of the cocaine system from source country production to
final consumption including: primary resource allocation and production; cocaine production and export; and U.S.
demand, import, pricing, and consumption. The model is used to examine an ensemble of policies proposed by the
National Strategy for Drug Control (White House 1989). Simulation results show the capacity of the system to
exhibit a wide array of behavior modes depending on the type of intervention being applied and the aspect of the
problem being targeted. Of particular interest from a policy standpoint is the implication of delays in physical and
information flows for generating divergent short and long term policy results. Findings suggest a comprehensive
approach combining demand and supply side policy leverage represents the most effective management strategy.
1, Introduction
During the past two decades, managing the drug problem has become a primary concern of U.S. citizens and
policy makers. Increased public awareness of the health, productivity and social costs associated with the use of
cocaine and its derivative “crack” has served to heighten concern and elicit increased government intervention and
expenditure for controlling the international trade. This concern has renewed debate over the most effective types of
policies for coping with the perennial drug abuse problem.
Traditionally, government intervention has emphasized drug interdiction policy to minimize the flow of
cocaine before it reaches U.S. markets. Despite increased expenditures for these programs in recent years, cocaine
consumption estimates and other indicators of domestic use exhibit dramatic growth. Evidence suggests that
intervention on the supply side of the market alone is insufficient to cope with a complex system that also
encompasses a resilient demand for cocaine. The limited success of past policies has prompted a recent National
Drug Control Strategy (White House 1989) that proposes shifting resources to well known demand side policies
including deterrence, preventive education, and rehabilitation/treatment,
This paper develops a preliminary system dynamics model to examine several of the policies proposed by the
new strategy. The simulation model describes the commodity structure of the cocaine industry from production to
final consumption. The basic structure of the model is similar to that presented in Meadow's (1969) study of
commodity production cycles. The model treats explicitly: Andean agricultural resource allocation and production;
cocaine production; import; export; and pricing; and US. demand determinants including the dynamics of user growth
and addiction. Simulation results show the capacity of the system to exhibit a wide array of behavior modes that are
sensitive to the type of intervention being applied, and the aspect of the problem being targeted. Of particular
interest from a policy standpoint is the implication of delays in physical and information flows for generating
divergent short and long term policy resultg Findings suggest a comprehensive approach combining demand and
supply side policy leverage as the most effective management strategy. Finally, the paper attempts to communicate
practical insights for selecting among the available policy options.
‘The paper proceeds in the following order. The subsequent section presents a brief overview of the dynamic
problem, Next, the simulation model is described, with the discussion focusing on the important causal
relationships and system components. The fourth section discusses simulated behavior, examining model results
under several alternative scenarios, The final section discusses findings, implications, and possible future
developments for the project.
* Thanks and acknowledgements are due to George Richardson, David Andersen, Elena Alvarez and Sheila Clifford-
Bova for helpful comments on the work presented here. Of course, all errors are mine.
13870
“
System Dynamics '90 1371
2. The Problem
Manifestations of the cocaine problem are apparent from several different aspects within the cocaine system.
There are concerns with growing supply and demand, as well as the resulting costs on society. From a domestic
standpoint, the most publicized indicators being the growth of drug use and the user population. Two surveys,
funded by the National Institute for Drug Abuse, The National Survey of Drug Abuse (1972-), and The National
Survey of High School Seniors (1975-), have begun tracking patterns across several categories of cocaine use. Two
distinct growth periods can be identified (Brower and Anglin 1987: 166-67). An initial growth phase can be traced
from the inception of each study and ends approximately in 1979. During this period, use in all prevalence and age
categories increased dramatically through the middle seventies, before leveling at the end of the decade. From 1979-
1985, nearly all categories of use exhibited a common tendency toward subsiding. While most use categories
stabilized, the overall expansion for the period remains significant. For example, the proportion of respondents in
the 12-17 year old age group who reported ever using cocaine more than tripled. Moreover, the proportion in the
same age group reporting use in the year and month prior to the survey nearly tripled. From 1975-1985, the
percentage of high school seniors who reported daily use quadrupled (Brower and Anglin 1987: 166-67).
Since 1985, evidence from the National High School survey suggests the trend may be changing. The
proportion of high school seniors reporting ever experimenting with the drug dropped thirty percent from 1985-1988
(NNICC Report 1988: 30-36). There are also signs of stabilization or reduction in several moderate use categories.
These drops may be a sign that preventive education programs are having a positive effect on the problem, however,
forecasting a permanent downturn is premature given the limited period of observation.
The dramatic growth in the popularity of cocaine use coincides with several changes in supply availability
conditions. First, both source country cultivation of coca, the base crop used to produce cocaine, and estimated
world production of cocaine have quadrupled since 1978 (NNICC Reports 1978-1988). According to straightforward
economic reasoning, a shift to the right in the world cocaine supply will decrease prices at all levels of the
production chain and increase the amount flowing into the U.S.. Increased availability and the resulting lower prices
make the drug more affordable to lower income groups and adolescents, and thereby stimulate cocaine
experimentation (Brower and Anglin 1987: 164-5). In turn, a rise in experimentation increases the potential for an
increase in cocaine abuse and addiction. The greater the number of persons entering the pool of casual users, the
larger the pool of potential addicts becomes, as a period of experimentation necessarily precedes heavier use
_ There are several reasons for the dramatic growth in supply. First, wages in the illicit coca trade are far above
those in any other industry. While the average per capita income in Peru was approximately $1,100 in 1985
(Alvarez: 1988: 8), it has been estimated that a family can earn upwards of $2,000 per year for harvesting dried coca
leaves, or $10,000 by easily converting the yield to cocaine base (Carsisle 1989: 6). Second, many of the traditional
industries in the Andean region such as mining have fallen on hard times leaving many dislocated workers in search
of alternative incomes, Finally, past government policies such as implementing food price ceilings, and dispersing
funds to favor urban development in the Andes at the preclusion of rural infrastructure development (Alvarez: 1988:
9), have kept rural farmers wages below those of the rest of the region. If producing coca allows growers a larger net
benefit than producing food, even after deducting food costs, they will be compelled to do so. Population growth in
Chapare, Bolivia’s major coca growing region, reflects the willingness of labor to revert to coca production,
increasing six fold from 1981 to 1985, faster than any other region by far (Eastwood and Pollard 1987: 166).
Another aspect of the problem emphasizes the impact of drug use on health costs and society in general. For
example, Cocaine related deaths have increased more than tenfold over the last decade. Cocaine related hospital
emergencies increased fifteen fold during the same period (NNICC Reports 1978-1988). Among adolescents, cocaine
is deemed to be a contributing factor in nearly 20 percent of drug related deaths within the 10-19 year old age group
(Brower and Anglin 1987: 165). These dramatic trends also strike home the message that society is losing its most
valuable resource --human potential-- at an exponential rate.
3. The Dynamic Hypothesis
The drug problem involves a highly interconnected set of components. Therefore, little is gained by
distinguishing between the supply and demand side of the problem from a policy standpoint, as policies aimed at
either side of the economic equation affect the whole system. Failure to account for the tight bind between
production and consumption may explain why past policies have had little success. Figure one presents a dynamic
hypothesis of the predominant forces thought to be operating in the cocaine system. In essence, the prevailing
a
1872 System Dynamics '90
forces comprise two counteracting or negative feed back loops, that mutually adjust supply and demand when market
conditions change. The left loop represents the demand side adjustment, while the right adjusts supply. Market
behavior pattems depend on several factors, including the dimension of imbalance and the relative adjustment times
of the demand and supply sides of the market. Delays and disequilibrium conditions are inherent in the adaption
process as described below
To describe how the system behaves overtime, it is easiest to imagine the reaction of the market to an
increase in our interdiction resources. In considering the overall market response to intervention policies, one begins
from the assumption that supply and demand are in an approximate state of equilibrium. A shock to the market will
upset this balance and give rise to a period in which the market adapts to the interruption in supply flow. During
the adaption process there will be both short and long term implications as the production-distribution chain adjusts
to the imbalance and effects market responses with differential time frames.
In the near term, a significant reduction in the inflow of cocaine to the U.S. would within a matter of
months, begin reducing domestic availability through attrition. In this scenario, short term price effects start in the
retail end of the distribution chain. As the ratio of supply to desired supply decreased, price will rise. This pressure
is negative or stabilizing in nature, with price acting as a governing mechanism on demand. Within this short time
frame, dealers and wholesalers will raise prices to maximize sales revenues in light of the tighter supply flow. In
turn, we expect that demand as indicated by both sales and consumption rates, will begin to fall into line with the
new supply level, as price limits accessibility to higher income groups and less elastic habits.
Figure 1, Basic dynamic
structure thought to underlie policy
resistance in the cocaine market ae asia
Policies
ieee street A
aN mam supply
Availability
Shipments
{) ©) “))- +
Price
+
User “% +. Production
population 7 capacity
Demand-Side Supply-Side
Policies Policies
A second effect occurs in the short term as higher price levels will increase the domestic market's
attractiveness and thereby stimulate increased import rates into the U.S.. The feedback relationship between
inventory, price and availability described here is also stabilizing in nature. Price thus acts simultaneously to retard
consumption and bring supply inventories up to meet the demand level being indicated by market price. In
combination, these trends tend to counteract one another and stabilize the retail market quicker then either might
alone. In the first instance, we find the desired effect one might expect from an interdiction policy, higher retail
prices, and slowing consumption and sales rates. In the latter, we find a less obvious and often overlooked role of
price in which it serves as an incentive to smugglers and wholesalers to increase importation rates.
Over a more moderate time period, sufficient interdiction policies will stimulate reactions by suppliers and
producers beyond the retail market, as price effects ripple backward, and traffickers rush to exploit a more attractive
U.S. market. In turn, suppliers inventories are depleted at an above average pace. Impacts may be delayed depending
on the policy of suppliers to stockpile excess inventory to guard against such an event. Inevitably, a successful
policy will force traffickers to increase orders to producers, bidding up price and exerting pressure to expand
System Dynamics '90 : 1873
processing capacity. Moreover, since shipping rate changes can occur quicker than changes in production, oftentimes
capacity is over expanded to make up for shortages incurred during the period it takes producers to gear up for
increased shipping rates (Mass 1980: 102). On the demand side, an interdiction policy which is sustained long
enough to erode traffickers supply inventories, should be sufficient to make inroads into the size of the user
population. However, it will likely be several months before there is any noticeable impact on consumption, and
even longer before the user population is reduced through lower experimentation rates.
In the long term, the effects of sustained losses in the pipeline ultimately filter back to the primary production
stage. This is the only stage where the losses can be replaced. In order to meet normal shipping rates plus seizures,
cocaine production must expand, thereby increasing orders for coca, bidding up price, and setting off adjustments in
Tural coca economies. Since this sector is the farthest removed from our initial policy intervention, it will be the
last affected by policies targeting the wholesale or retail sector. Adaption within the Andean agricultural sector is
slow, first there is a lag between increased attractiveness of coca production and growers perception of changing
conditions. Second, there are delays involved in planning and preparing land for cultivation, planting, crop
maturation, harvesting, processing, and transportation to the next stage in the process.
4. Model Structure
In this section a simple system dynamics model of the cocaine industry is developed. The model was written
using the STELLA software package and has been discussed in greater depth in a previous paper (Wuestman 1989),
which is available from the author upon request. The overall structure of the model and the major endogenous
variables are presented in figure 2. Each sector contains structure and equations relating variables within the sectors
bounds, as well as guidelines for interacting with other sectors. As shown, the model comprises three sectors:
Andean agricultural production, Columbian cocaine production, and U.S. consumption. The Andean agricultural
sector is further subdivided into food production, coca production and resource allocation sub sectors. The arrows
connecting the sectors show the physical and information flows linking the processes of one sector to another. With
the exception of the Andean population and agricultural resource base and the various policy levers, all variables and
dynamics are treated endogenously in the model.
Andean Agriculture Sector
“Andean Food Production Sub- Sector | [Resource Sub Sector ‘Andean Coca Production Sub. Sector
“Agricultural Production Resource ‘Coca Production Resource
‘Andean Food Inventory rer enter ty coca Loaf inventory
Food Production Rate Coca Harvesting Rate
Food Consumption Rate Coca Lea! Shipping Rate
Food Availabilty Bebecw Desired Coca Inventory
Feod Demand Wages Coca Availabilty
Food Price Coca Price
Farmer income Grower income
Price information -
Price ‘Goce
Information
USK Consumption Secwr Gatumbian Cocaine Production Secor
USA Cocaine Inventories Columbian Cocaine Inventory
Price Information
Cocaine Import Fate Lo-6-O-0-O-O-p] Cocaine Production Rate
Cocaine Saiee Rate Cocaine Shipping Rate
Cocaine Consumption Rate Desired Cocaine Inventory
Cocaine Demand Cocaine Availablity
U.S Cocaine Avaliabilty Cocaine Export Pricey
User Populations Cocaine Flow
Addiction Rate
Wholesale and Street Prices
Figure 2, Aggregate Model Structure listing major endogenous variables by sector
1374 ; System Dynamics ‘90
The focus of the model presented here is on the dynamics that arise with intervention in the cocaine market.
As stressed above, market interventions impact on both the supply and demand side of the equation, thus the model
incorporates components of each. Figure 3 presents a simplified flow diagram of the model showing the levels,
rates, and a generic view of the information flows comprising each sector. The four inventories connecting the coca
production and U.S. consumption sectors represent the levels of supply in the production-distribution chain. The
inventories are connected by a series of feedback loops which regulate the flow along the pipeline. The processes
guiding each inventory flow are represented similarly, each containing an inventory, a goal or desired inventory, an
inflow or production rate, and an outflow or consumption rate, The following sections briefly describe each sector.
4.1 Source Country Agriculture and Resource Allocation
This section describes the three sub-sectors of the Andean agricultural sector. At this stage of model
development the agricultural sector remains highly aggregated. Although the source countries produce a number of
agricultural products, the initial model restricts output to simply coca and food. The analysis focuses on two input
factors, land and labor. For simplicity, the two factors are combined into a single unit of production, Agricultural
Production Resource (APR). The agricultural resource base is finite, representing the total production possibility of
the agricultural sector. Each unit comprising this level is equivalent to the output of one producer working two
hectares of land.
Agricultural production is divided between a small coca production sub-sector and a comparatively larger food
production component. There are several important determining factors in each. Coca Production Resource (CPR)
measures the level of agricultural production units (APR) allocated to the coca sub-sector at any one time, and thus
determines coca production. The stock rises or falls according to changes in market conditions. A second important
stock is the coca inventory, which accumulates when production exceeds the shipping rate and falls when exports
surpass production. In the food production sub-sector, Food Production Resource (FPR) measures the level of
agricultural resource units and thereby determines food production. The other important level in this sub-sector is
the Andean food inventory. Excess food production is accumulated in inventory and food is depleted through
consumption. Both agricultural inventory levels provide important information in their respective sub-sectors for
determining price, consumption, and wages.
The resource allocation sub-sector receives information from the production sectors and determines resource
migration between the two components. The expansion or contraction of either sector is based on the relative
magnitude of the prevailing farmer and coca grower wages. When wages are balanced resource allocation remains
inert. When wages are unequal resources migrate toward the more attractive wage market. It is further assumed that
transfer rates are limited. The maximum transfer rate is set at 5% annually, limiting coca sector expansion to an
approximate doubling time of thirteen years. The structure of the market also serves to limit allocation. If one
sector loses a significant amount of resource, price and wages will rise, increasing sector attractiveness, and drawing
resource back into the market.
An additional factor that affects resource allocation is risk. Risk is associated with involvement in the
dangerous and illicit coca trade, and thus deflates coca wages. In the simulated scenarios the risk factor is determined
by the ratio of the government's eradication capacity to coca growers production capacity. The relationship
represents the probability of a grower’s crops being eradicated. Under initial conditions, it is assumed that farmer and
tisk adjusted coca wages are equal, and thereafter vary according to changes in the market.
4.2 The Columbian Production Sector
The Columbian cocaine production sector receives coca imports from the Andes and converts them into
cocaine for export to the U.S. consumption sector. The sector operates similarly to the other four supply
inventories along the pipeline. The major determining stock in the sector is the cocaine inventory, which
accumulates excess production and is depleted by both export and interdiction rates. Export (consumption) rate
averaged over a period determines the desired inventory and sets the goal for the sector.
Discrepancy between actual and desired inventory is captured by the availability ratio, and is a factor in
determining price. When availability is low, price rises, and conversely when high, price falls . Price movement is
then relayed in two directions simultaneously, both toward the primary production phase, setting the stage for future
production, and toward the consumption sector determining the amount that wholesalers (consumers) are willing to
purchase. The adjustment process contains two stabilizing loops centered on price, one which regulates inflow to
the sector, and another which governs consumption.
n_ Agriculture ig
( Andean Food Production Sub-Sector | Resource Transfer Sub-Sector Andean Coca Production Sub-Sector )
|
Food_Consmptn_Rt Eradictn
' Resource | Trans_Rt Policy
Shipping_Rt
Indicated
Res_Trans Y Gross_Coca
Agriculture Avail Ratio ShpRt_
Resource Risk Factor
Resourc@_Trans_Fret
ywer_Wage ~
Wage_Ratio SO Coca_Price
Coca_Shipping B J
C jan Cocaine P ion S
User Death_At Police_Etfor 8
~ Confise_Frot ~“Confiscation_Rt [interdetn fnterdct_Fret
Policy
Whs!_Pr
O) chngrt
Casual_Users
Street_Price
Whsl_Price
Net_User_Growth_Rt
Desrd_Col_Inv
Gross_Export_Rt
ExPrt_Pr_ChngRt
Export_Price
Figure 3. Simplified flow diagram: showing level and rate structure with abbreviated information flows
06, Soyureusg wi9764g
QZET
1376 System Dynamics ‘90
4.3 The U.S. Consumption Sector
Supply in the U.S. consumption sector is represented by two inventories, wholesale and street. The
wholesale inventory connects the U.S to the international pipeline. It is fed by an import rate and depleted by sales.
The street inventory is the final point of distribution and is fed by wholesales and depleted through final
consumption by the user population. Demand determination is slightly more complicated in the final retail sector
because it can be affected by price as well as changes in the size and composition of the user population. Users are
composed of several categories ranging from those who try cocaine only once during the year to those that use
several times a day. To observe how policies impact heavy users, as opposed to light and moderate users, the
population is subdivided into two categories. The heavy use category contains all those who use at least once a
month, while casual users comprise all those that use at least once during the year but less then once a month.
Growth of these two populations is treated within the model. There are two facets of user population growth,
those which affect experimentation and those which impact the addiction rate. Price and cocaine availability are
important variables in determining the growth of both groups. A shift in the supply balance causes price movement
which in turn affects the probability that potential users will experiment with cocaine. When price is high the
growth rate of casual users declines, and when low, the pool increases. The growth rate of both groups has a self-
reinforcing characteristic commonly known as the “ peer group effect”. As the population in either group grows,
growth rates are allowed to increase more rapidly. Similar to the spread of an epidemic, one person can“tum on” and
infect several others with the infliction, thus generating exponential growth. Growth occurs in the heavy use
category as a proportion of casual users are drawn into the heavy use pool each year. The fraction which become
addicted varies with market availability conditions, Periods of high availability facilitate addiction, as low price and
ease in obtaining the drug spur increased use.
The user population can be drawn down by a death rate, or through heavy users exiting the system as they
grow tired of the lifestyle. Like user growth, the death rate involves a proportionate relationship with the size of the
heavy user population. If the heavy user pool grows the death rate parallels its growth by a constant fraction and
conversely, if the number of users fall so too will will user deaths.
4.4 Price and Consumption
Consistent with classical economics price has a negative impact on demand. Given the illegal status of the
drug and problems limiting availability for conducting controlled experiments, research has made very little progress
toward providing a dependable estimate. For this analysis, a relatively inelastic -.25 is assumed. Consumption also
varies across different user groups, hence, independent habits are used for heavy and casual use. To account for
different frequency of use patterns within each population, average habits were adopted. Finally, heavy user habits
differ in that they are assumed to be inelastic to price.
5. Model Behavior
This section examines the dynamic behavior of the simulation model. The model has been tested using
various partial model tests and alternate scenarios (Wuestman 1990), The simulation results are not meant to be
accurate forecasts of future cocaine supply or use but to help clarify understanding about the cocaine problem and the
implications of implementing remedial policies. For comparison purposes, all the simulation were conducted in
similar fashion. Each simulation begins with approximately 5 million users consuming 60 metric tons of cocaine
per year. The system is initialized in a moderate growth phase with users increasing 3% annually. The market was
then disturbed with one of the following interventions: 1) stepping down supply at different points along the
Pipeline.(cradication targets coca supply, interdiction restricts cocaine import, and confiscation depletes street
availability); 2) reducing the user population through preventive education and rehabilitating heavy users; 3)
enhancing source country food production.
5.1 Base Run: Non-Intervention Scenario
Figure Sa shows the simulation results using selected demand-side variables of the base run. Figure 5b
reveals the behavior of supply-side variables of the base run. In the absence of intervention, the casual user
Population grows for the first five years of the run pulling consumption along with it and adding to the pool of
System Dynamics '90 ; 1877
heavy users, During the spurt, U.S. imports and other supply indicators remain stable, allowing the market to
tighten under growing demand, as evidenced by a 20% drop in availability. The combination of tighter availability
and high prices that accompany such a scenario lead first to a drop in consumption, and subsequently, to a decline in
the casual user pool. Note that it takes several years for price to exact a noticeable impact on casual users.
1 Heavy__Users 2 Casual__Users 3 Consumption_Rate 4 Street_Availability
? 10000000
$40009.09
4} 7800000.0
3 405000.90
4 ‘i 1.50
$000000.0
270000.00
1.00
a
—
2500000.0
190.01
135080588
3} 0.0
99 T r r T ane T
4 0.0 1985 1995 2005 2015 2020
Year
Figure 5a: showing behavior of demand indicators in a base run non-intervention scenario
1 Coca_Producers 2 Coca Shipping Rate © 3 USA_Import_Rate 4 Wage Ratio
3 cg Of
Bo 5800.9 7 T
98 T T T T T T T
0.81 1985 1995 2005 2015 2020
Year
Figure 5b: showing behavior of supply indicators in a base run non-intervention scenario
During the next fifteen years, consumption goes from a steady state into a second growth phase. Growth is
fueled by the delayed response of expansion on the supply side to the high price and low availability of the preceding
period. Years 15-20 reveal a change in the wage ratio at the commodity level; evidence that the demand signal has
reached the far end of the production chain, raising coca wages and attracting resource into coca production. Also,
note that heavy users have grown steadily over the period showing that the slight decline in the proportionally larger
casual user population had little impact on the flow into this population.
The final fifteen years of the run are characterized by a dramatic growth in both the demand and supply side of
the market. The increased supply flow from the tenth year on, as indicated by the import rate, fuels a resurgence in
casual user growth. During the final stage, the absence of any downturn in consumption indicates that the system
adapts by ordering sufficient production capacity to satisfy a much larger user population.
1378 System Dynamics '90
5.2 Multi-Supply Policy Scenario
The behavior of the model to the concurrent application of three supply side policies is shown below
(Figures 6a-6b). Policy one is a constant eradication of 10% of coca harvests needed to meet initial demand. Policy
two is an interdiction of 15% of the base run import rate. Policy three is a 10% confiscation of initial retail
inventory. All total, an interdiction rate equal to 35% of initial supply is implemented in year five and maintained
throughout the simulation.
1 Heavy__Users 2 Casual_Users B Consumption Rate 4 Street_Availability
Z 10000000 .
540009.00 7
ru 7500000.0 [
408000.90
= 405000-88
3} 5000000.0 5
3 270000.00 q
1.00
} 2500000.0 L
3 135000.00 4
4 8588
. ‘ a Bt I . igi
4} 0.0
z 23 7 <r . ‘ : :
4 00 1985 1995 2005 2015 2020
Year
Figures 6a: showing behavior of demand indicators under a multi-supply policy scenario
1 Coca_Producers 2 Coca_Shipping Rate 3 USA_Import_Rate ‘4 Wage_Ratio
H 300000.
id
T T
2005 2015 2020
ade 1985 1995
Year
Figure 6b: showing behavior of supply indicators under a multi-supply policy scenario
The first five years of the simulation are identical to those of the earlier runs. As the indicators reveal, there
is growth in both users and consumption, paralleled by falling availability. Shortly after program implementation
in 1990, all demand indicators with the exception of heavy users show dramatic decline. Reducing this group
requires substantial time, as gains must be made by first lowering the proportionately larger population of casual
users from which heavy users flow, or by treating addiction through more direct means.
Between 1995-2005, the program's secondary effects become evident as availability bottoms out and begins an
upward swing. Despite a 35% contraction of initial supply, availability comes surging back from 75% below initial
conditions to 30% above, over the ten year span. This phenomenon may seem surprising, however, contradictory
System Dynamics '90 1379
behavior is explained as the result of a decline in consumption. Although cocaine imports have been substantially
reduced, low availability has brought high price, suppressing consumption and in tum, causing the market to
actually be over supplied because less supply is desired. The upsurge in availability results as consumption slows
and expanded supply production capacity enables an increase in imports that eventually overrides interdiction
capability. As shown, there is a significant delay on the supply side of the market before coca production is able to
expand exports. In addition to a information lags throughout the market, eradication of capacity (crops) hinders
production activities. Approximately three years is required to replace the productivity of a mature coca bush.
The second half of the simulation shows the compensating tendency of the market system. As availability
rises, prices fall, and’consumption shows a resurgence. Expanded supply side production capacity allows both sides
of the market to continue growing throughout the remainder of the run.
5.3 Multi-Demand Policy Scenario
Many critics of current interdiction policies argue that the demand side of the market is the key to alleviating
the cocaine problem. Several simulations were conducted stepping down the rate of flow into the casual user pool,
the aim of preventive education, and stepping up the outflow from the heavy user population, the goal of a
successful rehabilitation policy. Figures 7a-b show the results of a combined effort of these two demand reduction
initiatives using selected indicators of supply and demand behavior.
During the five years following implementation, behavior takes the form of a slow moderate downturn in
consumption, paralleled by an opposite movement in availability conditions. The difference in short term behavior
from the previous scenario, is linked to the nature of the intervention. In this scenario, reduced consumption is
directly related to less users, and in contrast to the previous run, results are not dependent on supply shortages
moving through the pipeline and shifting price. Under cutting demand avoids the high price scenarios associated
with interdiction and supply shortages, and thus the influence of price pressure on supply production is negated.
1 Heavy__Users 2 Casual_Users % Consumption_Rate 4 Street_Avaliability
3} 10000000
2" 540009.00 7
4 2:00
3} 7500000.0
¥ 0800999
5000000.0
270000.00
4 1.00
2800000.0
135000.00
4 0.500
} A + + +
0.0 1985 1995 2005 2015 2020
z Year
Figures 7a: showing behavior of demand indicators under the multi-demand policy scenario
Over the second third of the simulation, consumption shows a rate of incline due to an increase in casual user
growth rates (Figure 7a). The rise in casual use may seem surprising in light of the implementation of a preventive
education program. The growth, however, is explainable as a side effect of reducing the number of heavy users.
Since this group consumes a large portion of supply, a glut is created by depleting their numbers. High availability,
accordingly, depresses price and stimulates experimentation and casual use.
In the final third of the simulation growth of the casual user pool begins to level off at a much higher
population level. Logic dictates that there will inevitably be a resurgence in heavy use, as a proportion of the now
larger casual use population fuels the addiction rate. In fact, the very last years of the rm show both consumption
and production indicators rising. A sufficiently large supply inflow will permit both groups to grow
simultaneously. In solving one aspect of the problem a tradeoff is created that allows the problem to return in an
amplified state.
1380 System Dynamics ’90
1 Coca_Producers 2 Coca_Shipping Rate 3 USA_Import_Rate 4 Wage_Ratio
1 300000.
3 270900000 e
54000!
4
} seenes
ZGE00880 4 r
4 1.10
001 t
5 sanneee Leaner ines
% 270000.00
1.00
1
; eee |
8588
4g
4 0.800 1985 1995 2005 2015 2020
Year
Figures 7b: showing behavior of supply indicators under a multi-demand policy scenario
5.4 Alternate Policy Scenarios
Several policy options were tested independently as well as combined into policy sets. In the majority of the
combination runs explored with the model the rehabilitation and preventive education policies were teamed with a
supply initiative. The long term results taken from year 40 of the simulated time horizon, are summarized in Table
1 below. The base run with no intervention is used as an index to which all subsequent runs are referred. A one
point deviation on the table equals a one percent change from the initial simulation result.
In policy set A, a 15% interdiction was teamed with the demand program. The two policies tended to
counteract one another. While interdiction attempts to achieve the aim of tightening the market and raising prices,
rehabilitation acts to free up supply by depleting heavy users. Comparing results with those of the multi-demand
policy it is clear that the addition of the interdiction program had little impact. The hitch in the policy lies in the
nature of supply intervention. When supply seizures target the supply in the pipeline price effects move in two
directions simultaneously. Heading toward the commodity stage price has a domino effect on the attractiveness of
each sector. In order, seizures will first drive up export price, and thereby stimulate paste and coca production.
Before slow working demand tactics can effectively suppress consumption, supply line losses are being replaced.
Moreover, as price moves to suppress consumption, excess supply supplements increased shipments and creates an
oversupply in the retail market. During the period of of high availability that follows, policy benefits are quickly
lost as price.entices new users into the market.
In policy set B, a moderate eradication policy is used as the supply side tactic. The simulation assumes the
same intensity for the policy as in the individual runs summarized in Table 1. The benefit of the policy can be
traced to the direct impact it has on coca wages. While other supply side programs tighten the market and raise
prices, new supply is always forthcoming. Eradication policy hits the pipeline where it hurts the worst by
destroying capacity which can only be replaced after lengthy delays. The results suggest that the policy is a good
supplement to a demand regime. The destruction of production capacity (crops), which take several years to mature,
insures low availability and high prices for an extended period. By halting the pipeline at the primary production
point, other sectors are also halted. A subsequent period of low supply then allows demand tactics to work on user
populations concurrently with high price. Overall, the program was the one of the most successful policy sets at
lowering both consumption and production activities simultaneously. The program was also one of only two that
were able to able to reduce heavy use without encountering offsetting gains in the casual user population.
System Dynamics '90 1881
Table 1.
Relative Magnitude of Market Indicators *
Policy Heavy Casual Toul Cocaine Coca Resource Food
Intervention Users Users Consumption Production Production InCoca _Production
Base run: no intervention 1.00 1.00 1.00 1.00 1.00 1.00 1.00
1. Coca Eradication
Moderate (10%fyear) 0.83 052 © 0.67 0.67 0.67 0.73 1.01
Strong (20%fyear) 0.61 0.07 0.23 0.22 0.21 0.29 1.04
2. Imterdiction Policy
Moderate (15%fyear) 0.91 092 0.90 0.88 0.95 0.92 1.00
Strong (25%/year) 0.86 0.86 0.84 0.79 0.92 0.88 1.00
3. Police Seizure 0.95 096 = 0.94 0.98 0.97 0.95 1.00
4, Preventive Education 0.99 097 0.99 0.99 0.98 0.98 1.02
5, Rehabilitation/Treatmant 0.25 1310.74 0.76 0.78 0.81 1.01
6._Agric. product. enhancement __0.96 0.86 0.90 0.90 0.91 0.91 1.01
7, Mult-Supply Policy 0.72 0.46 0.61 0.61 0.68 0.67 1.02
|__8. Muls-Demand Policy 0.24 126 0.71 0.73 0.74 O77 1.01
9. Policy set A 0.22 1.23 068 0.69 0.78 0.77 1.01
10, Policy set B 0.19 088 0.48 0.49 0.49 0.54 1.02
11. Policy set C 0.22 125 (0.69 0.75 0.76 0.77 1.01
12, Policy set D 0.23 118 0.66 0.68 0.69 0.72 1.12
13. Policy set E 0.79 038 0.56 0.57 0.56 0.61 1.12
Mult-Supply = eradication + interdiction + police confiscation» Indicators are ratios of the base ran simulation
Mult-Demand = rehabilitation + preventive education
. Policy set A = interdiction + rehabilitation + preventive education
Policy set B = eradication + rehabilitation + preventive education
Policy set C = confiscation + rehabilitation + preventive education
Policy set D = farmer productivity enhancement + rehabilitation + preventive education
Policy set E = farmer productivity enhancement + eradication + preventive education
Policy set C combines police confiscation with demand reduction. The program suffers the counteracting
forces of rehabilitation and preventive education policies similar to the behavior shown in the multi-demand
simulation. While the policy successfully reduces the heavy user population 80% by the end of the run, casual users
show a 25% increase over the base run scenario. On the supply side, the confiscation program proves insufficient to
control the rise in availability that accompanies reduced heavy user consumption. Lacking a strong supply country
initiative as well, the strategy also allows the shifting of resources into coca, and thus is overwhelmed by an
enlarged supply flow drawn by the growth trend in casual use. Although consumption is ultimately reduced 30%
over the base run, other combinations proved more effective in suppressing both production and consumption rates.
Policy set D teams the demand regime with a productivity enhancement of Andean food production. The
additional policy is simulated by stepping up annual farmer productivity and hence, initial farmer wages by 10%. As
intended, the policy meets with moderate success in suppressing production indicators by drawing resource away
from coca production. However, the program also encounters the offsetting influences of the demand reduction
policies and permits an 18% gain of casual use over the base run scenario. The strength of the program lies in its
beneficial effects on farmer wages and source country food production. Benefits on the demand side will only be
realized in the long term as the structure of the Andean agricultural sector is slowly altered.
Of all the policy sets simulated, combination E is strikingly superior on several accounts. The strategy
1382 System Dynamics '90
applies both the incentive of higher farmer wages and the disincentives of eradication to the Andean agricultural
market, simultaneously. On the demand side, rehabilitation is set aside while preventive education remains. The
success of the strategy relies on a combination of the long term impact on users, and a simultaneous reduction of
resource allocated to coca production. As the indicators illustrate, the program depresses both sides of the market.
Although heavy users are not reduced to the extent of some of the other combinations, the largest reduction of the
casual use population of all policy sets is achieved. The implication of the reduction is that future growth of the
heavy user pool will be a trickle as compared to other scenarios. A second noteworthy result, is the effect of the
policy on Andean food production which at the end of the simulation exhibits a 13% rise above the base run
simulation. While the accomplishment does not imply any immediate tangible gains in the U.S., it translates into a
source of substitute employment for those involved in producing coca. The alternative may hold the key to source
countries permitting the effective destruction of coca production capacity. Whether such a policy is actually
implementable is a relevant question currently being raised by experts in the area. The simulation of the policy
supports the notion that the proposal should be given serious consideration.
6. Preliminary Results and Implications for Further Study
Two general purposes are pursued in constructing and analyzing the preliminary system dynamics model of
this study. The first purpose is to draw attention to some of the complex forces present on both sides of the
economic equation that interact to create the cocaine problem's persistent nature. In particular, the model develops a
dynamic hypothesis that policy resistance is grounded in the feed back structure comprising the market. The base
run scenario supports the dynamic hypothesis that the problem is attributable to both sides of the economic
equation. When the model is initialized in a user growth pattern, endogenous market pressures passed from retail
through the distribution chain act to stimulate expansion in the coca production sector. When demand is reduced
below production levels, high supply and the accompanying price effects are sufficient to encourage growth in the
user population, and thus stimulate the demand side of the market.
The simulation model also identifies several key sensitive features in the cocaine system. The self
perpetuating characteristic associated with growth of the user population, raises important questions about the extent
of price in suppressing user growth rates and the type of policies that can help break the peer group pressures that are
typical of rapidly expanding popularity with certain drugs. A second sensitive structure identified by the model
focuses on the migration of resources within the rural Andean economy. The discrepancy between coca and other
agricultural wages provides one explanation for the dramatic increase in recent coca production trends. Intervention at
this important decision point may therefore provide a key policy lever for developing a strategy which directs
resources away from coca production rather than stimulating expansion.
The second aim of the study focuses on the implementation effects of several commonly proposed alternatives
in a dynamic framework. In general, simulation results of the preliminary model support a popular recommendation
in current literature for a shift in the emphasis on supply side programs toward a more comprehensive approach. In
searching for a strategy that maintained long term market suppression, the best results were achieved in simulations
using a balance between demand and supply track programs. Preliminary findings suggest supply intervention
tactics that raise price and reduce consumption in the short term are only useful to keep the problem from becoming
worse. Of the available options, eradication and productivity enhancement policies were superior to interdiction as
they tended to induce resources away from coca production whereas interdiction tends to foster expansion by
maintaining high market prices. However, other initiatives are necessary to address the deeper structural elements of
the problem. These types of policies would include the incentives which undermine demand and negate incentives
that drive resources into the illicit trade. In source countries, this may translate into providing rural laborers with
comparable forms of alternative legal income. Domestically, an effective strategy may require implementing several
policy levers simultaneously. As discussed above, there are several different classes of users that behave differently
and thus, it is likely that it will be necessary to identify and treat them independently.
The study also shows that long and short term policy implications can differ greatly due to compensating
responses that interventions at one point in the system tend to foster in others. This point is most clearly illustrated
in the policy scenarios involving interdiction policies. In the short term, price effects acting within the retail sector
serve to depress consumption and growth of the user population. Conversely, in the long term behavior is
dominated by the response of source country production sectors as price impacts serve to expand cocaine production
which lowers price, and ultimately fuels a resurgence in experimentation and the user population. Moreover, the
results reveal that more of a seemingly good policy is not always better. Initiatives that increased the intensity of
System Dynamics '90 . 1383
the moderate interdiction and eradication policies encountered diminishing marginal returns and therefore were less
effective than broadening the base of the attack, .
The simulated scenarios also suggest where the model can benefit from expansion and further research. First,
the success of the productivity enhancement and eradication policies imply that expansion on the supply side of the
model warrants the development of a multi-economy framework, with the addition of an industrial goods sector.
Such a framework would be more appropriate for simulating alternative income generating strategies. Second, there
would also be benefits to more accurate estimates of the price elasticity for labor migration between various labor
markets for such a study. Also, questions regarding the negative effects of the coca boom as a contributing factor to
the economic and political instability of the Andean source countries suggest expanding model boundaries to include
government accounting and foreign exchange sectors. The astronomical inflation rates in these countries may be
associated with the rapid influx of coca dollars to the region.
For simplicity, the initial model does not address the sociological or psychological aspects of drug use.
Hence, the model can be modified to treat this aspect of the problems. Further research to identify determinants and
tisk factors associated with drug abuse are necessary prerequisites for developing an effective demand reduction
strategy. Also, policy levers are exogenous in this initial version of the model. Community and government and
actions can therefore be incorporated into the feed back structure of the model in future phases. Moreover, the model
does not exhaust the list of remedial policies. Several policies were chosen for testing in the initial stages of
development. Further model refinement can add options to the list such as decriminalization, legalization, drug
testing and numerous others. To meet this purpose, the user population structure represented in the model of this
study can be disaggregated to provide a more detailed description of user dynamics.
Finally, it should be noted that the model depicts the cocaine industry as a commodity structure, comprised of
a series of interrelated markets, with availability and pricing mechanism determining market clearing rates.
Therefore, results and generalizations are limited to the extent that the industry is vertically integrated. Continued
research into the nature of the organizational structure of the industry is needed to supplement the information that is
available in this area.
7. Conclusions
This paper presented a preliminary system dynamics model for analyzing the persistent nature of the cocaine
problem. The simulation model described the commodity structure of the cocaine industry from production to final
consumption. Using this framework the paper examined the implementation effects of several remedial policies.
The model produced several interesting findings. First, alternate scenario tests of the initial model showed
why there is a need for a more comprehensive approach to the cocaine problem. Experimentation with the model
identified several key structures and behavior modes that merit future consideration in developing strategies and
researching the problem. In particular, the model illustrated how many of the currently proposed policies produce
Tesults that tend to counteract one another. Simulations also illustrated how interdiction policies can have attractive
short term effects but undesirable long term impacts by fostering expansion in drug production capacity and user
population.
The study also identifies several areas of promising future development and applications for the model.
Further refinement, may prove useful in examining problems particular to source countries. For example the model
can be used to experiment with alternative income generating or development scenarios. In addition, the model may
prove useful for examining policy options other than those selected for this study. In a different mode, the model
may prove useful for subsequent research which examines the different roles of major market players such as cartels
and interdiction forces. This line of research would focus on the decision making aspect of developing strategies in a
dynamic market in which key players are forced to adapt to one another.
1884 System Dynamics '90
REFERENCES
Alvarez, E. H. 1989. “Rural Poverty and Illegal Expansion of Coca Production in Peru.” Paper presented at the
XV Intemational Congress of the Latin American Studies Association, Miami, Fla.
Bagley, B. M. 1988, a S. Foreign Policy and the War on Drugs: Analysis of A Policy Failure.” Journal of
ris - 30:200-10.
Brower, K. J, and Anglin, M.D. 1987. “ Adolescent Cocaine Use: Epidemiology, Risk Factors, And Prevention.”
, 17;163-179.
Carlisle, C. 1989. “Confronting Narcotics: U.S. International Policy.” Policy Focus. No. 2. Wash, D.C.:
Overseas Development Council.
Craig, R. B. 1987. “Illicit Drug Traffic: Implications for South American Source Countries.” Journal of
Interamerican Studies and World Affairs, 29:1-34.
Eastwood, D., and Pollard, H. J. 1987. “The Accelerating Growth of Coca and Colonization in Bolivia.”
E.L.U. -Economist Intelligence Unit. 1989. 1986-88 Country Report. Peru, Bolivia. London: E.1.U.
Forrester, J. W. 1961. Industrial Dynamics. Cambridge, Mass.: MIT Press.
Forrester, J. W., and P. Senge. 1980. “Tests for Building Confidence in System Dynamics Models,” In Systems
Dynamics, ed. A. A. Legasto, Jr., J, W. Forrester, and T.M. Lyneis. TIMS Studies in the Management
Sciences 14: 209 -228. NewYork.
Healy, K. 1986. “The Boom Within The Crisis,” In Coca and Cocaine. Effects On People and Policy in Latin
America. ed. Pacini, D and C., Franquemont. Boston: Cultural Survival and LASP.
Johnson, L. D., P, M. O’Malley and J. G, Bachman. 1986. Illicit Drug Use, Smoking, Drinking Among
American High School Students, College Students and Other Young Adults: National Trends Through 1985.
NIDA Research Monograph, DHHS Pub. No. (ADM)86-1450, U.S.G.P.O, Wash, D.C.
Koch, J. V. and Grupp, S. E. 1980. “ The Economics of Drug Control Policies.” In The Economics of Crime.
ed. R. Andreano and J. Siegfried, Cambridge, Mass: Schenkman.
Levin, G., E. B. Roberts and G. B. Hirsch. 1975. The Persistent Poppy. Cambridge, Mass: Ballinger.
Mass, NJ. 1980. “Stock and Flow Variables and the Dynamics of Supply and Demand” in Elements of the
System Dynamics Method, J Randers, ed. Cambridge, MA: MIT Press.
Meadows, D.L. 1969. The Dynamics of Commodity Production Cycles. Cambridge, Mass.: Wright-Allen Press.
Morales, E. 1986. “ Coca and Cocaine Economy and Social Change in the Andes of Peru.” Economic
Development and Cultural Change, 35:143 -161.
National Narcotics Intelligence Consumers Committee, The NNICC Report 1981-1988, The Supply of Illicit Drugs
to the United States from Foreign and Domestic Sources, Washington, D.C. April, 1988.
Pool, D. 1988, Review of USAID/Bolivia Development Assistance Support for Coca Eradication. An A.I.D.
Report reviewing USAID Progress in Bolivia. DOC, 0841N.
Reuter, P., G. Crawford, and J. Cave. 1988. Sealing the Borders: The Effects of Increased Military Participation
in Drug Interdiction. Santa Monica, Cal.: Rand.
Reuter, P. G., and M. A. R. Kleiman, 1986. “Risks and Prices: An Economic Analysis of Drug Enforcement,” in
M. Tonry and N. Morris. Crime and Justice: An Annual Review of Research, No. 7.
Richardson, G. P. and A. Pugh, 1981. Introduction to System Dynamics Modeling With DYNAMO. Cambridge,
Mass: MIT Press.
Stenberg, L. 1980. “A Modeling Procedure for Public Policy.” Jn Elements of the System Dynamics Method, J
Randers, ed. Cambridge, MA: MIT Press.
Strug, D. 1986. “The Foreign Policy of Cocaine: Comments on a Plan to Eradicate the Coca Leaf in Peru,” In
Coca and Cocaine. Effects On People and Policy in Latin America. ed. Pacini,D and C., Franquemont.
Boston: Cultural Survival and LASP. -
Tullis, F.L. 1987. “Cocaine and Food: Likely Effects of a Burgeoning Transnational Industry on Food production
in Bolivia and Peru,” in W.Hollist and FL.Tullis, eds. Pursuing Food Security, Boulder: L.Reiner.
The White House, September, 1989. National Drug Control Strategy, G.P.O Document, Washington, D.C.
U.S. General Accounting Office, 1988. “Controlling Drug Abuse: A Status Report. A Special Report to The
Comptroller General of the United States.” March 1, 1988.
World Bank, 1988. “Bolivia Updating Economic Memorandum,” Report no.7278-B0, Wash.D.C.
‘Wuestman, E. A. 1990. "Sowing Supply: A Preliminary System Dynamics Model for Analyzing Policy
Resistance in the Coca-Cocaine Market", Master's thesis, Nelson A. Rockefeller College of Public Affairs
and Policy, S.U.N.Y. Albany.
CC BY-NC-SA 4.0