Supplementary files are available for this work. For more information about accessing
these files, follow the link from the Table of Contents to "Reading the Supplementary Files".
Table of Conten
Go Back
An Application of System Dynamics Modeling
to
The Question of a Log Export Ban
for Indonesia
with Comments on Illegal Logging
Richard Dudley °
January 2002
Introduction 3
Modeling Approach 4
Starting Point 4
Model structure 5
General Comments 5
Wood Processing and Domestic Log Demand 6
Logging and Log Supply 7
Foreign Demand and Export Price 9
Effect of Harvest on Forest 11
Model Outcomes 12
Effect of Gradually Increasing Foreign Demand 12
Application of a Log Export Ban for Five Years 13
Increasing Foreign Demand and a Log Export Ban 16
Effect of Variations in Domestic Milling Investment Response 18
Changes in Export Price 21
The Question of Ilegal Logging 22
Discussion 24
Examining Model Validity 25
Background 25
Validation Status - Log Export Ban Model 26
Literature Cited 27
Appendix 1: Modeling Conventions 28
PRELIMINARY SD MODELING OF LOG EXPORT BAN 2
Table of Figures
Figure 1. A causal loop diagram illustrating interconnections between the wood processing and logging
sectors.
Figure 2. Wood Processing and Domestic Log Demand. Model components showing the link between
domestic log processing capacity, domestic demand for logs and the current domestic purchase price
for logs.
Figure 3. Logging and Log Supply. Model components linking current domestic purchase price to log
harvest and harvest capacity (=logging teams).
Figure 4. The portion of the model used to determine export demand, export price and amount of exports.
Figure 5. Model components used to determine forest cover, forest regeneration and harvest effects on
forest. . .
Figure 6. Response of milling and harvesting capacity to increasing foreign demand. .
Figure 7. Effect of increasing foreign demand on prices and the fraction of supply exported. ...
Figure 8. Effect of a log export ban (year 20 to 25) on wood processing and harvesting capacity.
Figure 9. When a log export ban is instituted export prices rise but have little effect on domestic prices
which fall due to a sudden increase in supply. ..
Figure 10. Detail of removal of a log export ban (in year 25).
Figure 11. Domestic and export prices and the fraction of logs exported under a gradually increasing
foreign demand interrupted by an export ban. ....
Figure 12. Effect of a log export ban on harvesting capacity and forest cover during a period of increasing
foreign demand. . ...
Figure 13. Effect of a log ban on domestic use of logs. .
Figure 14. Functions used to examine the effect of different responses to changing wood product
profitability.
Figure 15. Log exports and domestic demand for logs under three different mill investment strategies. ...19
Figure 16. Effect of four mill investment strategies on forest cover during a period of increasing foreign
demand interrupted by a log export ban.
Figure 17. Domestic wood processing capacity under three investment strategies during and shortly after a
log ban in years 15 to 20.
Figure 18. Three possible relationships for the effect of the export demand ratio on export price.
Figure 19. An example of how different export demand vs. price functions can change the effects of a log
export ban. .. 22
Figure 20. Predicted timber harvests under situations where logging costs have decreased by various
percentages over an initial five year period.
Figure 21. Possible scenarios for regaining control of logging following a drop in logging c
illegal logging.
PRELIMINARY SD MODELING OF LOG ExPORT BAN 3
A Preliminary Application of System Dynamics Modeling
to
The Question of a Log Export Ban
for Indonesia
with Comments on Illegal Logging
Richard Dudley °
January 2002
Abstract
System dynamics modeling allows us to examine various scenarios within a
complex system. By using this approach we not only learn about the response
of the system to test inputs which could never be tested in the real world, but
we also learn to question our assumptions about the system itself. This leads to
a better understand of its workings.
This paper presents a preliminary system dynamics investigation into the
potential effects of a log export ban on the Indonesian forest sector. As a
preliminary model its primary purpose is to help us better visualize potential
effects of a log export ban rather than to predict, in detail, actual outcomes.
The model provides relatively simplistic overviews of the wood processing
sector, demand ~ price fee standing stock and log
availability, capacity of the harvest sector, as well as export demand, price and
log exports. In spite of its apparent simplicity the model examines important
feedbacks that must be understood if the effects of a log ban are to be properly
examined.
Introduction
Indonesia’s forestry sector has been in turmoil for the past several years following the
removal of President Soeharto from power in May 1998. Since then there has been a
major decentralization of governmental authority to the provinces and districts. This was
done without careful planning of laws and regulations, including those related to natural
resources. During this period of weakened legal control, there has been a substantial
increase in the amount of illegal logging, so that in 2001 roughly 50% of the timber
harvest was illegal. The illegal logging situation in Indonesia as of mid 2000 is well
documented by Scotland et al (2000). For a discussion of possible causal links leading to
increases in illegal logging see Dudley (2001).
A significant portion of the illegal harvest is thought to be exported. Consequently, some
concerned agencies and NGOs have suggested a log export ban as a means of limiting
illegal exports, thus reducing illegal logging. The idea of a ban on all exports to prevent
exports of illegal logs might appear illogical, since one might expect that logs would
merely be exported illegally. However, proponents of a complete log export ban believe
that terminating all exports would make an export ban workable. In their view a partial
ban (e.g. on certain species) would be ineffective since it could be easily defeated using
PRELIMINARY SD MODELING OF LOG EXPORT BAN 4
phony documentation. With a total ban, any log leaving the country would be illegal and
could thus be easily identified as such.
Several analysts have pointed out that there are many other ramifications of a log export
ban. Most often cited are the loss of export tax revenue, and possible effects on the
domestic timber related industries. The effects on domestic industry are generally
considered either: good — due to increased access to cheaper raw materials, bad —a lack of
competitiveness fostered by over protection, or both. Also, employment in the logging
and wood processing sectors: might decrease — due to a decrease in log harvest, or might
increase — due to a stimulation of the wood processing sector.
A meeting! held in September 2000, to discuss possible effects of an export ban included
participants from industry, government, academia, NGOs, and donor and donor project
representatives. Of the subjects discussed they agreed on 13 and disagreed on 28 items.
Significantly, several items where no consensus could be reached were those which
involved prediction of the outcomes of a log export ban.
Prediction of complex situations without an understandable and agreed upon framework
is difficult. Illustrated herein, using the log export ban as an example, is one established
approach for looking at this type of problem: system dynamics modeling.
Modeling Approach
Starting Point
The overall purpose of this
model is to examine, in a
preliminary way, the effect
that a log export ban might
have on the Indonesian
logging and wood processing
% Costs of
ema he Mocs Price Logging
industries. A secondary
purpose is to illustrate the
‘of Logs > /
utility of the system Aton Tinber L Logs mauels
dynamics approach for this a fom ew Haves
sort of policy analysis. o
‘ F U 5
The starting point for this t a
overview is a causal loop BSL
diagram — a type of rigorous
picture model (Figure 1)
modified from that presented
in Dudley 2001. In this Figure 1. A causal loop diagram illustrating interconnections between the
diagram four balancing wood processing and logging sectors. Our interest is to investigate how
foreign demand for logs might affect this system.
Demand for
ze Indonesian Logs
Profitability of +
Wood Products
' Roundtable On Log Export Ban. 27 September 2000. World Bank Office, Jakarta. (meeting summary).
PRELIMINARY SD MODELING OF LOG ExPORT BAN 5
feedback loops are labeled. Loop A represents activity in the logging sector where log
prices affect profit from logging which in turn affects the amount of timber cut which
then affects prices. Loop B indicates the effect of timber harvest on timber available,
which then, in operational forest management systems, effects an allowable timber
harvest which affects amount of timber cut. Similarly, the much larger loop C indicates
that allowable timber harvest will affect the number of wood processing mills which
affects the demand for logs. This demand affects the purchase price of logs, affecting
logging profits which influence the amount of timber cut. Loop D reflects how the
demand for logs affects log price and profitability of wood products manufacture which
will further affect log demand. Our real interest is in determining how this simple system
is affected by foreign demand for logs.
This simplistic starting point helps to reveal many important questions especially when
we think about the meaning of the connecting arrows. To what extent, for example, do
low log prices improve profitability of wood product manufacture? At what log price do
profits start to suffer? At what profitability level do mills start to expand? At what point
would mills shut down?
Similarly we can envision several questions in the harvesting sector. How fast can the
harvesting sector respond to changes in demand? Will log prices drop as harvesting
increases? As forests become less abundant do harvest sector profits drop or do log
prices rise enough to compensate for decreased availability?
These questions cannot be answered by the type of model presented in Figure 1. A
quantified model is necessary to get a better feel for how these model components might
behave under different circumstances.
Model structure
General Comments
The basic structure of the model consists of the following sectors:
1) links between product profitability and wood processing capacity,
2) links between domestic demand and log prices,
3) links between log supply and the profitability of logging,
4) links between profitability of logging and the size of the logging industry,
5) links between log exports, foreign demand, foreign price for Indonesian logs
and its effect on domestic prices, and
6) the effect of logging on forest cover.
For simplicity we can visualize the model as depicting a forested area of 1 million ha
with associated processing capacity. The model is initiated with a stable log harvest of
3 million m*/year (or 3 m*/ha/year), 1.5 million m?/year is exported and the other half is
processed by 1.5 million m*/year of domestic processing capacity. Log prices are $50/m’
with logging cost accounting for half this. The selling price of wood products is set at
$100/m? of raw material used, giving an initial profit margin of $50/m’.
3
PRELIMINARY SD MODELING OF LOG EXPORT BAN 6
Note that wood processing capacity, domestic demand for logs and related components
are measured in m’/year not in m®. Thus flows into and out of the relevant stocks are
measured in terms of m’/year/year, which reflects how fast the flow of logs is changing.
Wood Processing and Domestic Log Demand
domestic
demand Sane aes supply
adjusted
demand
changing
x 4 ‘demand
demand fo eee
fect of product
profitably on actual
i Ee - mill se of logs alt
seat as \ Somer
3 “roabesy on fet of demand
scrapping ri aperatons
bung mits. cn purchase pace
} ills relative profitability
py.) weumaaey \
J! Ges \
capacity needed +
desired mil
capacity
‘rofsbaty on desired
all copay
_e ilrice
changing pice
‘rom demand
changing
field
ELLRE pe eae
es rasan
Figure 2. Wood Processing and Domestic Log Demand. Model components showing the link between
domestic log processing capacity, domestic demand for logs and the current domestic purchase price for
logs. Note that some model components have been removed for clarity.
Figure 2 presents that part of the model related to domestic log processing capacity, its
effect on domestic log demand and the effect of demand on log price. In this conceptual
view the relative profitability of wood products’ determines the desired mill capacity of
the processors. This desired capacity is gradually incorporated into the actual wood
processing capacity as mills are built, expanded or improved. At the same time older or
obsolete mills and equipment are scrapped. This gradually changing capacity creates a
demand from mills.
Because wood processing capacity is not always fully used, the demand from the mills
must be modified by some measure of capacity utilization. This is also dependent on the
relative profitability of products. An effect of product profitability on the actual mill use
of logs, in connection with demand from mills determines an adjusted demand. Because
* Items in italics are names of model components.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 7
there is time needed for demand changes to be felt, actual domestic demand for logs
changes more slowly than the adjusted demand.
Domestic demand for logs will affect price because demand may exceed domestic supply
and sometimes supply may exceed demand. Thus the relative demand (the demand
compared to supply) will effect the price the mills would like to pay for their logs: the
desired mill price. This becomes one of several factors affecting the current domestic
log price which in turn will affect the profit margin due to log price and overall
profitability of wood product manufacture.
Three functions are used in this portion of the model. Functions often reflect areas where
our knowledge of relationships between model components may need to be improved.
For example what is the actual relationship between product profitability and the desire to
build new mills: the effect of relative profit profitability on desired mill capacity. At
some level of profitability mill owners will merely be happy with their current mill
capacity. At this point the effect is equal to 1.0: the desired mill capacity is the same as
the current mill capacity. But above or below this level, as profitability increases or
decreases, what is the shape and scope of the curve. For example, if profitability is 50%
higher what is the likely reaction of mill owners? Increase mill capacity by 10%, 25% or
50%? What if profitability is 50% lower? The model can be used to test the effect of
different values for this effect, but it cannot discover the correct one. If the model is not
very sensitive to this model component, then the questions are not important. If the
model is sensitive to changes in this component, then it needs to be quantified more
carefully.
Another function, the effect of relative profitability on mill operations, allows us to
determine how rapidly the mill might lay off workers or shorten work hours as
profitability falls. Or conversely how quickly mill operations will be pressed into
overtime if profitability rises.
The third function in this part of the model determines the effect of demand on purchase
price of logs. This function addresses the question ‘how much does the desired mill log
price change as demand for logs exceeds or falls short of the supply of logs?’
Logging and Log Supply
This segment of the model deals with the relation between logging and log supply and
factors which affect the profitability of logging (Figure 3). Current domestic purchase
price has a direct and substantial effect on the potential profit from log harvest which is
also affected by costs of logging. Log contractors will weigh this potential profit against
what they might view as the normal profitability of logging. This comparison, the
relative profitability of logging, will determine whether it is desirable to establish more
logging teams to cut trees. Relative profitability also determines whether it is desirable to
use existing logging teams to full capacity (capacity use) or not. The capacity use is
then combined with the actual Jogging team capacity to determine the amount of timber
PRELIMINARY SD MODELING OF LOG EXPORT BAN 8
cut each year.» Logging teams are always changing and the times for the teams to build
up or decrease are also important model components.
74
_ mill price
effect of relative difference %
supply on price
rae pe oasae
revised purchase . DeLay
ried oFloue difference a
ssa = changing price
SUPPLY from ee Ca a
changing foreign domestic
ae we roowice eld “Changing pce fort") Pee dference
adjusted ply PURCHASE PRICE price foreign demand
normal
o
wa scroornce )
‘CHANGE DELAY
“Ss potential profit
Supply of from log harvest -
Logs
effect of relative costs of
profitabiltyonhanest normal logging ete
capacity use a Profitability
relative of logging
‘amount of timber profitability of
logging effect of relative
cut per year
=e capacity us
profitability on logging
team increase
7
Logging
Teams
Desired
LOGGING TEAM
UFETIME
oF
En Mesina
‘me neevé6 To logging
INCREASE teams,
.
decreasing
£ logging teams
Figure 3. Logging and Log Supply. Model components linking current domestic purchase price to log
harvest and harvest capacity (=logging teams). Note that some model components have been removed for
clarity.
The amount of timber cut yields a supply of logs which, when compared to the adjusted
normal supply (typical supply in the recent past) provides a measure of relative supply
which affects the desired, or revised, purchase price of logs. This revised purchase price
then becomes one of the elements affecting the current domestic purchase price.
There are three functions in this segment of the model which are important model
components needing careful study. As in the first model segment, two of these are
related to the response to profitability on the part of entrepreneurs. What is the effect of
3 Ideally the amount of timber cut would also be limited by allowable cuts and forest management plans,
but this does not seem to be a relevant component at present.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 9
relative profitability on harvest capacity use? That is, to what extent are logging teams
told to slow down when profits drop. Also, what is the effect of profitability on logging
team increase when profitability rises or drops. This is particularly important in that
once /ogging teams are built up there is a lag before they can drop again. Lags of this
sort introduce instabilities into a system.
The third function in this portion of the model is the effect of relative supply on price
which determines how much of an effect excess or limited log supplies have on the price
expected by loggers. This price then becomes one factor in determining the current
domestic purchase price.
Foreign Demand and Export Price
This sector of the model deals with export demand and price and the direct effects of any
export ban. Here foreign demand for Indonesian logs is part of a feedback loop linked to
the current export price. This is similar in structure to the loop seen for domestic
demand except that here we have not included effects on foreign milling capacity but
have merely given that as a model constant, basic foreign demand for logs (Figure 4).
As in the domestic sector demand, foreign demand for Indonesian logs, when compared
to the supply, /og exports, yields a relative export demand having a value dependent on
the balance between exports and demand. The effect of export demand on export price is
dependent on the effect of export demand on price function the formulation of which is
critical. If Indonesian logs dominate the market then their removal from it will have a
significant effect on price. On the other hand, if there are many other sources of logs then
the effect of a decreased supply from Indonesia would be negligible. Whatever this effect
of demand on price might be, it acts to adjust the normal export price upward or
downward. This adjusted export price will gradually be absorbed into the current export
price.
Just as demand affects export price, the current export price can affect demand.
Increases or decreases in current export price compared to the normal export price, will
cause an effect of price on demand which will affect the basic foreign demand for logs
creating a foreign demand adjusted for price. Depending on the time needed to change
foreign demand changes this adjusted demand is gradually absorbed into foreign demand
for Indonesian logs.
Importantly, the current export price is also one factor affecting current domestic
purchase price. This is a major reason why a log export ban is being considered.
However the magnitude of this effect may be difficult to determine. If export price was
very high, but no logs could be exported then presumably the effect on domestic log price
would be minimal. The actual impact of current export price on domestic price, is
determined in part by the effect of foreign export volume on price. If export volume is
high then the effect of export price will be high.
Also, the amount of /og exports must be some function of price as well. At this point we
probably do not want to model the decision processes used by entrepreneurs to determine
if logs should be sold to a mill or be exported. Instead we can use the export price ratio
along with the function, effect of pricing on amount exported. Clearly this is an important
PRELIMINARY SD MODELING OF LOG EXPORT BAN 10
function, the design of which will significantly affect model outcome. It may be, for
example, that exports follow a direct relationship with the export price ratio: if export and
domestic prices are the same, then half of available logs will be exported. If export prices
are three times domestic prices then exports will be three times domestic log sales. On
the other hand we may believe that proportion of exports rises more rapidly than the
export price ratio. For example if export price were three times the domestic price then
exports might be 5 times domestic sales. The slope and shape of this curve is rather
important and various formulations will need to be investigated.
effect of pricing on aaa EXPORT BAN
‘amount exported Sy a“ EFFECT
log exports
export price
rat
fraction of supply
exported
domestic -
ie Pert hee
changing foreign ‘inkiiog
* Ree ey,
‘DEMAND. relative export
demand
i}
| foreign demand
} gap effect of export
| A demand on export
caren | f ie
Fer ii Hscignlomend
adjusted for price
ier
effect of price on "adjusted export
\ | we Saat sit
worn
a ™
pee |
\ xsi FOREIGN
DEMAND FOR LOGS
effect of price on
effect of foreign demand exp price
export volume on eens
price
= current
TIME NEEDED FOR
‘export price PRICE CHANGE TO
changing export BE REALIZED
~ price =
foreign domestic
price difference
oo
Figure 4. The portion of the model used to determine export demand, export price and amount
of exports. Note that other model components have been removed for clarity.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 11
Effect of Harvest on Forest
The final model sector is that dealing with the effect of log harvest on the forest. This
very simple representation of a forest is based on an average ha of land within a 1 million
ha forest (although any amount of forest could be used). The forest on this average
hectare is represented by m’ currently on each ha of forest land. This amount can be
changed by harvesting and by regeneration, which combines creation of new trees,
growth and deaths of trees. Regeneration is dependent on the stock volume already on
the land and on the ratio between it and the maximum standing stock because there is an
effect of stock ratio on regeneration. As the forest becomes more densely stocked the
regeneration approaches zero.
In this model formulation m’ currently on each ha of forest land is equivalent to the
availability of trees for harvest. Under an ideal forest management system the
availability of trees for harvest would also be limited by forest management plans.
Herein it is assumed that the only feedback from the forest to the remainder of the model
is the probable effect on logging costs of the relative availability of trees for harvest. It is
assumed that as trees become more scarce logging costs rise. This is not necessarily the
case, since some harvesting may make access to the forest easier thereby lowering
logging costs. It may also be the case, under the current situation, that average logging
costs are not yet affected by forest availability.
amount of timber
grt oy s cut per year
HA AVAILABLE continuous
amount cut per ha
#
initial stock
+ 2
regeneration haryeetiog
REGENERATION +
RATE +
e +
MAX STANDING
function of effect that sce
tock ratio hi —
poseablacr Meraral availability of
janeration rate i i
reg effect of stock ratio stock ratio trees for harvest
on regeneration
Figure 5. Model components used to determine forest cover, forest regeneration and harvest
effects on forest. An arbitrary area of 1 million ha with maximum sustainable harvest of 3m*
per ha per year has been used. Note that other model components have been removed for
clarity.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 12
Model Outcomes
Effect of Gradually Increasing Foreign Demand
Domestic Milling and Harvesting Capacity
6M BASIC DEMAND FOR FOREIGN LOGS
test input “S if)
Logging Teams
45M aging
log exports
3M
15M Z
Wood Processing Capacity
(domestic)
0
0 5 10 15 20 25 30 35 40 45 50
Time (Year)
Figure 6. Response of milling and harvesting capacity to increasing foreign demand. As log exports rise in
response to increasing foreign demand, logging teams increase substantially and domestic milling capacity
drops slightly.
Here the model is used to examine the effect of a gradual increasing foreign demand for
Indonesian logs. An increase of 200,000 m*/year for 20 years starting in year 5 is tested.
Following the 20 year period foreign demand remains constant. The model starts with
foreign and domestic demand each at 1.5 million m*/year. After the 20 years basic
foreign demand has risen to 5.5 million m3/year(Figure 6).
As foreign demand for Indonesian logs increases, log exports increase as expected.
Increasing log prices (see Figure 7) stimulate creation of more logging teams providing
the increased harvest. As domestic price increases domestic wood processing capacity
declines because higher prices limit profitability.
As export demand increases the current export price also increases causing an increase in
current domestic purchase price. Downward pressures on the domestic purchase price
are caused by dropping desired mill price and by increasing log supply due to increasing
production by the logging teams (Figure 6). Thus prices increase enough to encourage
more cutting and to discourage domestic milling, but not enough to limit exports.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 13
Log Prices and Fraction of Supply Exported
70 Sim3
0.9 Dmnl “
Current Export Price
65 Sim3
08 Dm fraction of supply exported
60 Sim3 a
0.7 Dmnl
55. SimB
0.6 Dmnl
Current Domestic Purchase Price
50 Sim3
0.5 Dmnl
45. Sim desired mill log price
0.4 Dmal
40. Sim
03° Dmnl
0 5 10 15 20 25 30 35 40 45 50
‘Time (Year)
Figure 7. Effect of increasing foreign demand on prices and the fraction of supply exported.
Application of a Log Export Ban for Five Years
A second test of the model involves the implementation of a 99% effective log export ban
for five years. Immediately after the ban is instituted /og exports drop and current export
price starts to rise but has little effect on domestic prices.* Logging team harvest capacity
decreases due to a significant decrease in demand even though domestic wood processing
capacity rises slightly (Figure 8).
When the log ban is lifted in year 25, log exports jump quickly to their pre-ban level,
drop back and then jump to a higher peak prior to stabilizing at the pre-ban export level.
Logging teams respond to the increasing demand, but because of increased current
domestic log price, and associated profitability of logging, logging team capacity
overshoots the supply and after about two years stabilizes at the pre-ban harvest capacity
(Figure 8 and Figure 9).
Figure 10 shows more detail shortly after the lifting of an export ban. Because of the
lingering high export price, most logs available in the reduced supply are exported. This
drives up domestic prices which pushes exports down, and also stimulates log production.
Both export and domestic prices drop as supplies increase. In this simulation, export
price declines more slowly than domestic price causing a stimulation of exports. Three
* Note that I have assumed a significant effect of Indonesian log supply on foreign price. With no logs
exported the current export price triples from 50 to 150 $/m*
PRELIMINARY SD MODELING OF LOG EXPORT BAN 14
years after the ban the system has returned to its pre-ban values with the following
exception: domestic wood processing capacity is about 0.5% higher than pre ban values
causing log harvesting to be slightly higher as well.
System behavior during the immediate post-ban period is dependent on the lag times used
in the stocks (e.g. time needed for price change to be realized). Since we don’t know
these lag times exactly, events during this short period should be viewed as an example of
the types of interactions that are occurring rather than an exact representation of
outcomes.
Domestic Milling and Harvesting Capacity
6M 3/Year
200. $/m3
Logging Teams
45M mB/Year
150 $im3
M 1m3/Year Current Export Price
100. $/m3
15M m8/Year
50. Sim3
log exports —¥|
Wood Processing a
0 m3/Year Capacity
0 Sim3
is 16 17 +18 19 2 21 22 23 2 2 26 «27 «428 629 «630
Time (Year)
Figure 8. Effect of a log export ban (year 20 to 25) on wood processing and harvesting capacity.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 15
Log Prices and Fraction of Supply Exported
200 $/m3
1 Dmnl
160 $/m3 Current Export Price
0.8 Dmnl
120 $/m3
0.6 Dmnl
fraction of supply
exported
80 $/m3
0.4 Dmnl
40 $/m3
0.2. Dmnl domestic prices
SimB
Dmnl
coo
15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Figure 9. When a log export ban is instituted export pric but have little effect on domestic prices which fall
due to a sudden increase in supply. As supply drops domestic prices rise to their previous level. See next figure
for detail of year 24 to 27 when the export ban is lifted. “Domestic prices” in this figure refers to current domestic
purchase price and desired mill log price.
Detail of Log Prices and Fraction of Supply Exported
200 $/m3
1 Dm Current Domestic
LZ Purchase Price
160. $/m3 '
0.8 Dmnl fraction of supply
exported
a Pp
120 $/mB
0.6 Dmal
80 S/mB
04 Dml er
40 S/m3
0.2. Dm /
S$/mB desired mill log price
Dm
24.50 25 25.50 26 26.50
Time (Year)
Current Export Price
oo
Figure 10. Detail of removal of a log export ban (in year 25). Immediate effect of dropping the ban is an
export of most available logs driving up the domestic prices. This causes a drop in exports, but as log
production is stimulated by higher current domestic purchase price exports climb again.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 16
Increasing Foreign Demand and a Log Export Ban
Indonesia has been exporting an increasing amount of logs, so if a log export ban was to
be introduced it might be considered to occur in conjunction with rising foreign demand.>
The third example will look at this situation, combining the two tests above.
Figure 11 illustrates the situation with increasing foreign demand (from year 5 to 25)
interrupted by a log export ban from years 15 to 20. For the most part details are similar
to the previous example (see Figure 10). In spite of the export ban the general trend of
increasing foreign demand and its effects on prices and exports continues after the ban is
lifted.
Detail of Log Prices and Fraction of Supply Exported
200 $/m3
2 eDin fraction of supply exported<<|
160 $/m3
0.8 Dmnl —
120 $/m3
0.6 Dmnl
‘Current Export Price
80 $/m3
0.4 Dmnl
40 $/m3
0.2. Dmnl /
$/m3 Current Domestic Purchase Price
Dmnl a
5 10 15 20 25 30
Time (Year)
oo
Figure 11. Domestic and export prices and the fraction of logs exported under a gradually increasing
foreign demand interrupted by an export ban.
Harvest capacity drops significantly during the log export ban but quickly responds in
response to foreign demand when the ban is lifted. Forest cover starts to increase during
the ban, but once the ban is removed forest cover resumes its decline Figure 12.
Domestic wood processing capacity and associated domestic demand for logs both
increase during a log export ban, but resume their decline after the ban is lifted though at
a slightly higher level (Figure 13).
* Assuming that increasing exports are due to increasing demand.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 17
Logging Teams and Forest
6M m3/Year
400. m3/ha
4.5M mB/Year
350 mB/ha Logging Team
harvest capacity
3M m3/Year
300 mB/ha
Forest
15M m3/Year
250. mB/ha
0 m3/Year
200. mB/ha
0 3 10 15 20 25 30 35 40 45 50
Time (Year)
Figure 12. Effect of a log export ban on harvesting capacity and forest cover during a period of increasing
foreign demand. Results with, and without, the ban during years 15 to 20 are shown.
Domestic Demand and Wood Processing Capacity
1.75M
Wood Processing Capacity
with export ban
Domestic Demand for Logs
15M
125M
Wood Processing Capacity
IM Domestic Demand for Logs
without export ban
750,000
500,000
0 5 10 15 20 25 30 35 40 45 50
Time (Year)
Figure 13. Effect of a log ban on domestic use of logs. An export ban causes a temporary jump in
domestic demand and processing capacity. Following removal of the ban both demand and capacity remain
somewhat higher than without the ban, although both continue to decline.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 18
Effect of Variations in Domestic Milling Investment Response
More aggressive responses to profit opportunities might alter the outcome of the above
scenarios. For example, what if mill owners aggressively expanded their operations
during periods when profitability was high and severely limited mill expansion when
profitability dropped.° This test examines that question by using four relationships
presented in Figure 14.
The Effect of Relative Product Profitability on Desired Mill Capacity
4
2p
8 a i
g iggressive
Oo
co)
Es
i y
o normal
B
®
oT
S| mild
3
i=
oO
0
0 0.50 1 1.50 2
relative product profitability
Figure 14. Functions used to examine the effect of different responses to changing
wood product profitability. For example, with the aggressive response mill owners
would be more likely to invest in new mill capacity if the profitability increases.
Specifically if profitability is twice its normal value then, with the aggressive
strategy, mill owners would desire a quadrupled mill capacity, other things being
equal (see the pertinent part of the model in Figure 2). A fourth relationship, the
aggressive and hold strategy, incorporates the right-hand portion of “aggressive” and
the left hand portion of “mild.”
The outcomes reveal that some while some “mill investment strategies” make only minor
differences to the effects of the log ban plus increasing foreign demand scenario
discussed in the previous example, other strategies can cause major differences in the
outcome. For example, the maximum difference between the aggressive and mild
° When I refer to the behavior of mill owners I am referring to the average behavior of many mill owners
not the behavior of a single owner. In other words, we are interested in the effect that the overall behavior
of mill owners has on the system.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 19
strategy, which occurs during and shortly after the log ban, is between 10 and 12 percent
of total domestic log demand (Figure 15).
A fourth strategy, called "aggressive and hold", has a much more significant effect. This
strategy is identical to the “aggressive” strategy to the right and identical to the "mild"
strategy to the left of Figure 14. It implies aggressive accumulation of processing
capacity at times when profitability is high, plus a tendency to hold on to that capacity
when profitability drops below normal.
A Comparison of Three “Mill Investment Strategies”
5M
4M I
i
1 \
\ Y =
I rt \ Via
ae log exports & Wile
aggressive and hold strategy
2M ee = BL —|
: ra aggressive strategy
— ae |
—— 4 ) -
IM ~s mild strategy
Domestic demand for logs = ane
0)
5 10 15 20 25 30
Time (Year)
Figure 15. Log exports and domestic demand for logs under three different mill investment strategies.
Different strategies can affect the overall outcome over time. For example the “aggressive and hold” strategy
produces significantly higher domestic milling capacity during and after the log ban.
With this "aggressive and hold" mill investment strategy, domestic log demand jumps
during the log ban and does not disappear when the log ban is lifted (Figure 15). This
leads to a 33% increase in domestic capacity after removal of the ban (Figure 17).
This strategy also has a significant effect on forest cover as portrayed in Figure 16. The
accumulation of domestic processing capacity causes a more rapid decline in forest cover
than under the alternate strategies examined.
While these strategies are merely rough examples of what could happen under different
scenarios, they do reveal that there is no guarantee that a temporary log ban will reduce
pressure on forest by lowering overall demand. The outcome is dependent, at least in
part, on the response of domestic mill owners to temporary increases in log supply and
decreases in domestic log price.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 20
Wood Processing Capacity Under Four Mill Investment Strategies
2.25M
aggressive and hold strategy
2M S| =
175M —
_ aggressive strategy
a 2 : normal strategy]
198M mild strategy
IM
4 15 16 17 18 19 20 21 22 23 24 25
Time (Year)
Figure 17. Domestic wood processing capacity under three investment strategies during and shortly after a
log ban in years 15 to 20.
Forest Cover Under Four Mill Investment Strategies
300
275
250
log
5 export
225 | ban |
period of rising foreign |
7 mild strategy
a
aggressive and hold strategy
m° on each ha of forest land
200
0 5 10 15 20 25 30 35 40 45 50
Time (Year)
Figure 16. Effect of four mill investment strategies on forest cover during a period of
increasing foreign demand interrupted by a log export ban.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 21
Changes in Export Price
In the previous discussions export price was permitted to rise to as much as three times
the normal export price in response to increases in foreign demand resulting from a
shortage of supply during a log export ban. However, a lower maximum price may be
more realistic. Figure 18 indicates three of many possible responses of export price to
changes in export demand.
Effect of Export Demand Ratio on Export Price
4
export price can rise to 3
3 times the normal export price ————
2
2
7
“7
°
So
eal
S
°
8
& a
al export price can rise to only 1.5
1 times the normal export price
0
0 15 3 45 6
Export Demand Ratio
Figure 18. Three possible relationships for the effect of the export demand ratio on export price. At
the point (1.0, 1.0) there is no effect. As export demand rises export price will rise but will approach
some maximum
During a log export ban export prices rise because the ban restricts the satisfaction of
export demand. During the ban increasing export prices have little effect on other parts
of the system. If logs cannot be exported the rise in price in meaningless. However,
there is a lingering effect of these high prices when the ban is lifted and this lingering
effect causes interesting dynamics in the system (see, for example Figure 10). This
lingering effect is more pronounced if export prices rise higher during the export ban.
The example in Figure 19 illustrates the different responses of amount of timber
harvested using the different relationships shown in Figure 18.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 22
Amount of Timber Harvested per Year with
Different Export Demand vs. Price Relationships
6M
Lingering effect of high export
i i i Maximum
prices leads to increased logging. Says pice ratio
3.0
2 45M 2.0
8 15
=|
2
sal
5 3M
s
aS} log export ban. ————__»
E
1.5M
Low log prices stimulate
domestic demand.
0
14 15 16 17 18 19 20 21 22 23 24
Time (Year)
Figure 19. An example of how different export demand vs. price functions can change the effects of a log
. Here the amount of timber harvested is shown for the three different demand vs. price functions
igure 18. If export price rises more during an export ban the lingering effect is more
pronounced.
The Question of Illegal Logging
A full examination of illegal logging requires a different model designed to examine a
range of social, economic, legal and political issues. Nevertheless, a few of these issues
can be examined by making minor modifications to the model presented here. One of
these modifications is based on the fact that illegal logging lowers the costs of harvesting
logs. This raises profits of logging operations, stimulating the harvest of more logs,
lowering the price of logs which makes wood processing and log exports more profitable
as well. Thus, one simple examination of the response of the system to illegal logging
can be tested by lowering logging costs.’
For example, we can assume that illegal logging lowers logging costs by between 5% and
20% over a five year period (Figure 20). As logging costs decrease below normal,
logging allows windfall profits to be made. Consequently logging continues to increase
unless something limits these profits. Although an increasing supply of logs will tend to
” Within this section, for simplicity, the model has been used with only the aggressive and hold strategy of
mill investment (see Figure 14) and only allows the export price to reach 1.5 times the normal export price
(see Figure 18).
PRELIMINARY SD MODELING OF LOG EXPORT BAN 23
lower log prices, the lower log prices will stimulate new processing capacity which will
absorb the increased log production allowing logging to increase further. Only when
supply of trees become limited does the harvest decrease.
Timber Harvested per Year with Decreased Costs of Logging
20M
15M
Change in logging
costs during years
10M 5 to 10.
Increasing illegality
lowers logging costs
over 5 years.
wn
=
Pan
million cubic meters per year
no change
0 5 10 15 20 25 30 35 40 45 50
Time (Year)
Figure 20. Predicted timber harvests under situations where logging costs have decreased by various
percentages over an initial five year period. Decreased logging costs allows windfall profits which
stimulates logging lowering log pri mulating further demand for logs. Eventually scarcity of logs
cause logging costs to rise. While it is assumed here that illegality lowers the costs, other factors could
have the same effect.
Having seen the effects of lowering logging costs in Figure 20 various scenarios to
control this effect can be examined: no action, a log export ban, re-control which raises
the logging price to normal, and a log export ban plus re-control (Figure 21).
A reduction of logging costs to 80% of their normal value causes a rapid rise in logging
until costs of logging are limited by availability of logs. A log ban does not control this
trend and merely stimulates a higher level of logging and a more rapid collapse. The
reason for this is that the profitability of logging does not change significantly until
limited log availability raises costs. If logging costs are re-established at their original
value (in year 17.5) logging levels are controlled and slowly drop. A logging ban makes
such re-control less effective. Note that the nature of the relationship between tree
availability and logging costs (which causes the crash) is only roughly estimated. It
seems likely that a crash would occur sooner than indicated here.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 24
Timber Harvested per Year Under Different Control Scenarios
20M
no action
ISM
log export ban from
year 15 to year 20
v . \
10M “ re-control of prices at A
mid-point in a log
Increasing illegality export ban
lowers logging costs by
20% over 5 years.
3M | ae
0 5 10 15 20 25 30 35 40 45 50
Time (Year)
million cubic meters per year
re-control of prices with
no log export ban
Figure 21. Possible scenarios for regaining control of logging following a drop in logging costs due to
illegal logging. Under the conditions modeled, a five year ban on log exports actually stimulates more
logging. Re-control to reestablish original logging price stabilizes the amount of logging, but does not
return it to the original value. Recall that the outcomes are highly dependent on the values used for mill
investment strategies and for foreign price rise during an export ban (see text).
Discussion
System dynamics modeling provides an excellent framework for looking at the possible
outcomes of a log export ban. This approach allows us to examine the interlinked
relationships between demand and price, harvesting capacity, milling capacity and other
factors. The critical question that arises when examining this system is how to accurately
formulate the relationships between various model components. What we believe to be
correct relationships have been presented in this paper. However, the exact nature of
these relationships is not accurately known. This modeling approach allows us to look at
several formulations of these relationships and allows us to examine the outcomes of
each. Following such an examination we could follow-up with the collection of
additional data related to relationships that are deemed most important based on the
model.
Although the model presented here is preliminary, it does give indications as to where
questions remain in examining the benefits and costs of using a log export ban. It also
provides reasonable information about the utility of an export ban plus some more
general findings related to attempts to control illegal logging with an export ban.
For example, the model indicates that different responses by mill owners to changes in
log prices can result in very different outcomes, in some cases making a temporary log
ban less beneficial for forest protection than no ban. Importantly some policymakers
PRELIMINARY SD MODELING OF LOG ExPORT BAN 25
supporting the ban believe that it will help the local log processing industry by lowering
log prices thus allowing expansion of domestic milling capacity. If this view is correct,
then the other rationale for having a log ban, prevention of over-harvest, may not be
valid. That is, if local processing capacity increases significantly due to lower prices
caused by a log ban, then the effect of the log ban on protecting forest cover will be less.
Consequently, the expected response of mill owners to any log export ban should be
examined carefully, as should any government policies which might stimulate mill
investment.
Some other areas where additional information is needed are as follows:
How much will export prices change when Indonesian logs are removed from the
export market? In other words, to what is the effect of the supply to demand ratio
on the export log price (which in the model is represented in the component,
effect of export demand on price function)? If this function is different from those
used in the model, how will model outcome be affected?
There is also the question of how entrepreneurs might respond to changing
profitability of logging (represented in the model component, effect of relative
profitability on logging teams increase). How quickly will logging teams
decrease if log prices and logging profitability drop? Will people making up the
logging teams continue to cut trees anyway in order to make a living. Will this
actually increase the amount of trees cut has price drops in order to maintain their
income level? If so, how would this affect log prices and demand?
The model, as presented here, reflects the real situation, not the ideal situation. As
presented there is no feedback whereby decreasing availability of trees for harvest
directly limits amount of timber cut per year. This feedback occurs only via a decrease
in logging profitability. If various factors conspire to maintain logging profitability in the
face of decreasing forest cover, the forest will be over-harvested. This is the sad reality
of Indonesia’s forest industry.
Examining Model Validity
Background
A system dynamics model should produce the right results for the right reasons (Barlas
1996, Coyle 1996). The structure of the model should mirror the real system to the level
of detail needed for its purpose. A system dynamics model can be viewed as a theory of
how a system works. This means that model validity is connected to our view of how
theories are justified. For the most part system dynamics practitioners do not view
models as either true or false, but view them as one of many possible representations of
reality (for further discussion see Barlas 1996). As a simplification of reality, a model
will not be identical to the system being modeled, but in general must be good enough for
the purpose intended. Its usefulness for a specific task, rather than absolute accuracy, is
what is important.
Coyle and Exelby (2000) refer to two aspects of model validation:
PRELIMINARY SD MODELING OF LOG EXPORT BAN 26
“Validation, means ensuring that the model’s structure and
which it is intended. ... verification, means ensuring that
sumptions meet the purpose for
quations are technically correct”
The first aspect of this is the most difficult, and various approaches for validating models
have been put forward (e.g. see Forrester 1961, Barlas 1996, Coyle 2000).
System dynamics models are “white box models” the workings of which should be
understandable to interested clients: people who wish to use the model and have an
interest in doing so. Validation then becomes a process of building confidence in the
usefulness of the model in cooperation with these clients. Since such models should be
built in conjunction with the clients in the first place, validation becomes, in part, an
iterative process of reviewing with the client the model and needed changes, based on
preliminary model outcomes, model diagrams and logic, using approaches such as those
presented in Forrester (1961) and Barlas (1996). Interested clients are likely to be people
who needed the modeling work in the first place, for example, business clients who have
contracted model building to help solve a specific problem.
One of the problems with public policy models, such as this log export ban model, is that
the intended clients may not be interested. These reluctant clients are people who should
be interested in the model but are either not interested, or are not aware of it, for various
reasons. Validation of models with reluctant clients is somewhat more difficult because
it also involves selling the model, and perhaps the idea of modeling, rather than working
together to build a useful model.
One reason this situation arises is that public policy makers are usually different
individuals from the system “domain experts” who were consulted during the model
building process. Policy makers typically have many other issues vying for their
attention. In building the log export ban model, expert information was obtained from
people studying various aspects of the forest industry (especially illegal logging), not
from policy makers per se. In this case validation with the domain experts does not
actually involve the ultimate client.
Public policy modeling perhaps falls somewhere in between academic modeling and
commercial modeling which have different goals and validation concerns (Coyle and
Exelby 2000). Unlike commercial modeling, public policy modeling does not
necessarily have contractual obligations and associated risks. Nevertheless there are risks
associated with making specific policy recommendations, and these risks are greater than
those associated with academic (e.g. journal article) reporting of possible outcomes under
different scenarios.
Validation Status - Log Export Ban Model
The model described in this paper is in the beginning stages of the validation process.
Information for the model structure was obtained from discussions with a number of
people working in various aspects of forestry research, industry, and from several
environmental non-governmental organizations interested in forestry issues. Initial
model ideas were reviewed with some of these domain experts, and these discussions
resulted in revisions to the model. The model is verified in that all components have
relevant units of measurement and these are consistent throughout the model calculations.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 27
This model has not been validated in detail and several validation steps remain. Most
importantly, the model includes a number of graphical relationships describing the effect
of one model component on another, but these are based on only a general knowledge of
these relationships. Each of these functions should be examined and modified as
necessary in cooperation with appropriate domain experts. Similarly, time lags also need
reexamination.
The structure of the model appears to be suitable for the task at hand: examining the
effects of a log export ban with some limited aspects of illegal logging. However, there
may be several components not currently part of the model which additional examination
and discussion with domain experts may reveal.
The model as it is currently presented is a good starting point for discussion of a log
export ban and its effects on the wood processing and logging industries. This was its
original purpose. At present it cannot be used to examine exact outcomes because, as
indicated in the previous sections, use of different values in several functions alter the
outcome of the model significantly. Thus at a minimum, better knowledge of these
functions is necessary before the model could be used for more detailed analysis. If the
model is to be used for detailed analysis it should also be validated in a more rigorous
manner.
Literature Cited
Barlas, Yaman. 1996. Formal aspects of model validity and validation in system
dynamics models. System Dynamics Review 12(3): 183-210.
Coyle, Geoff and David Exelby. 2000. The validation of commercial system dynamics
models. System Dynamics Review 16(1): 27-41.
Coyle RG. 1996. System Dynamics Modelling: A Practical Approach. Chapman and Hall.
London.
Dudley, Richard G. 2001. Dynamics of Illegal Logging Systems in Indonesia: An Initial
Investigation. Chapter 16 in Colfer, Carol J. Pierce, and Ida Aju Pradnja
Resosudarmo, Editors. 2001. Which way forward? Forests, People and
Policymaking in Indonesia. RFF. Washington, D. C.
Forrester J. W. 1961. Industrial Dynamics. MIT Press: Cambridge, MA.
Scotland, Neil (with Joyotee Smith, Hikma Lisa, Marc Hiller, Ben Jarvis, Charlotte
Kaiser, Mark Leighton, Laura Paulson, Edward Pollard, Dessy Ratnasari, Ramsey
Ravanell, Scott Stanley, Erwidodo, Dave Currey, Agus Setyarso). 2000. Indonesia
Country Paper on Illegal Logging. Report prepared for the World Bank-WWF
Workshop on Control of Illegal Logging in East Asia. Jakarta, 28 August 2000.
Editied by William Finlayson, and Neil Scotland.
PRELIMINARY SD MODELING OF LOG EXPORT BAN 28
Appendix 1: Modeling Conventions
Throughout the model the following conventions have been used. Connecting arrows are
labeled with a plus or a minus to indicate the direction of change of a variable as its
predecessor changes. A plus sign (+) indicates that the direction of change is the same —
if the first variable goes up the second will go up. A minus sign indicates that the
direction of change is opposite (other things being equal). This convention is merely a
convenience to aid in understanding the model. Each variable has an equation associated
with it which gives the actual method of calculation.
There are five types of model components: Stocks, flows, auxiliary, constants and
functions.
Stocks, also referred to as “state variables” or “levels”, are accumulations over
time. They have a lingering effect over time. These are normally presented in
boxes and are capitalized.
Flows change the value of stocks over time. These are usually identified by
special pipe-like arrows and by the fact that they flow into and out of stocks.
Auxiliary variables are model components that are calculated at an instant in time.
These are presented entirely in lower case.
Constants are modifiers that are not affected by other model components. They
are given values. These are presented in upper case.
Functions are mathematical, graphical or table-defined relationships between two
components of the model. These are lower case but have no arrow pointing to
them. I have usually started these names with “effect of ....”. In color figures
these are red.
y
Wy function
relative en effect of relative
profitability of > jodi fier profitability on logging
logging logging teams team increase
desired aie
constant
LOGGING TEAM
UFETIME
increasing .
logging i sereesng
logging teams
TIME NEEDED TO teams aging
INCREASE Flow
Back to the Top
CC BY-NC-SA 4.0