5. Dynamic simulation model
5.4. Land competition and supply module
The land competition and supply (LCS) module simulates land supply for soybean production and its resulting land-use changes. The LCS module generates five main outputs that are successively used in the CM and LCA module to estimate soybean production and LUC GHG emissions from soybean production for biodiesel use (Table 5-4).
In the CM module, soybean production is given by the supply of land for soybean production and the average soybean yield, both simulated in the current module. Additionally, GHG emissions from land-use change and soybean cultivation also depend on the average soybean yield in each region, so, the average soybean yield is also linked to the LCA module. The LCS module then estimates the quantity of land converted from each unmanaged land-use k and pastureland to cropland. dLUC GHG emissions from cropland expansion need to be allocated to soybean production for biodiesel. Therefore the LCS module also estimates the share of soybean on cropland expansion.
Table 5-4 . Outputs and simulated variables of the LCS module.
Outputs Units Symbol Equation
Soybean land supply ha/year
sp land
q Eq. 4-41
Average soybean land yield ton/ha
sp soy
Y Eq. 4-19
Unmanaged land k reduction from cropland l expansion ha/year ql,k Eq. 4-45 Pasture land reduction from cropland expansion ha/year ql,l Eq. 4-46
Cropland supply ha/year
l land
q Eq. 4-43
Simulated variables
Soybean land unit profit US$/ha
1
n land
g Eq. 4-17
Pasture land unit profit US$/ha
2
l land
g Eq. 4-3
Competing crop unit profit US$/ha
2
n land
g Eq. 4-3
Competing crop land supply ha/year
cc land
q Eq. 4-41
Pasture land supply ha/year
2
l land
q Eq. 4-43
Simulated variables account for 1) the estimation of land-unit profits of each competing crop and managed land-use type, 2) the resulting land supply for each land-use type based on the competition among managed land-uses.
The simulation process of the LCS module is shown in Figure 5-8. The dynamics of the LCS module arise from determining the rates of change in land supply for each land-use type. The first task to be performed is the estimation of the unit land profit of each crop type (i.e. soybean, corn). Then provided that production costs and export taxes allow soybean producers to get positive profits and depending on the unit profit of competing land-uses, cropland is allocated between soybean and corn. Land stocks of competing crops increase when their own land unit profit increases and decrease when the land unit profit of the competing crop increases.
Set policy scenario § export tax of
competing land-uses
Set external factor scenario
§ Soybean and competing land-use yield trend, demand trend, price trend, production cost trend
§ Land productivity § Land availability
§ Share of unmanaged lands on cropland expansion § Share of soybean supply regions
Simulate crops land unit profit
Allocate cropland Get soybean producer price End t=T? yes t=t+1
t=t0 Set initial conditions
no
Simulate managed land unit profit
Find unmanaged land and pasture conversion from cropland expansion
Allocate managed land
Find cropland supply
Simulate cropland direct land-use change
by land-use type
Find soybean land supply Get average soybean
production cost
Find average soybean yield
Figure 5-8. Simulation procedure of the LCS module.
Managed land is then allocated between cropland and pasture based on their aggregated land unit profits. The level of cropland supply defines the effect of land productivity on the yield of soybean, corn and pasture.
Finally, direct land-use changes from cropland expansion are simulated to estimate the quantity of cropland expansion into each unmanaged-use type and pastureland. Unmanaged lands decrease with cropland and pasture expansion. No natural regeneration is assumed, consequently, unmanaged lands are only allowed to be reduced. Expanding into degraded lands may lead to lower yields but also to lower carbon stock changes. In contrast, expanding agricultural land into forest may lead to the release of significant carbon stocks which may negatively affect the biodiesel GES.
The LCS module is linked to the SPM sub-module. The SPM module is a set out auxiliary variable (Figure 9-8) which determines the production costs and the soybean yield for each soybean region (Table 5-5). Auxiliary variables are assumed constant, so that changes in
SIMULATION MODEL
soybean production costs depend only on the share of cultivation methods by region. No feedback structures are present in this sub-module.
Table 5-5. Output variables of the SPM sub- module.
Outputs Units Symbol Equation
Soybean land unit production cost by region US$/ha
sp sr soy
C , Eq. 4-21
Soybean land yield by region ton/ha
sp sr soy
Y , Eq. 4-19
The simple simulation procedure is given in Figure 5-9.
Set policy scenario § No policy insturments
Set external factor scenario § Non-land input price and quantity § Average land productivity
§ Share of soybean cultivation method by region Simulate soybean production
cost by region
Simulate soybean yield by region t=t+1
t=t0 Set initial conditions
End t=T?
yes
no
Figure 5-9. Simulation procedure of the SPM sub-module.
The complexity in land competition arises from multiple interactions among competing land- uses (Figure 5-10). Two main feedback structures are identified including the effect of competition among land-uses and the effect of land productivity on yields, namely:
§ R2, R3: Cropland and managed land competition
§ B10, B11: Effect of land productivity on managed land yield
In the first case, if cropland expansion is constrained, soybean and other crops compete for cropland based on their relative land unit profits, creating a reinforcing feedback loop (R2). The same mechanism of land competition applies for cropland and pastureland. As cropland expands, ceteris paribus, less land is available for pastureland, creating also a reinforcing loop (R3). Similarly, cropland and pastures compete for managed lands based on their aggregated land unit profit. For simplicity, Figure 5-10 avoids the representation of the price linkage with supply and demand for each land-based product. Beef, other crops and soybean prices however, adjust to changes in structural factors and the price effect, similarly to price adjustments loop in the crushing dynamics module. The stock and flow structure in Figure 9-6 shows these interactions.
expand into less productive lands, yields may decrease depending on the yields trend evolution. Consequently, the productivity effect feedbacks into the producer decision to increase land supply for managed land-uses. Alternatively, as land productivity decreases, it is more difficult to bring land into production, generating higher incentives to use non-land inputs (e.g. fertilisers). For simplicity, this effect is assumed to be captured in exogenous yield trends. The counterpart of substituting land by non-land inputs is that non-land inputs costs may increase, decreasing land profits and consequently land demand.
R3 B10 R2 cropland supply pastureland supply - - soybean land supply corn land supply - - - - cropland productivity pastureland productivity + + + + B11
Figure 5-10. Feedback loops in the LCS module.
The land competition module is also a set of interconnected stocks and flows. There are three types of stocks: competing crops, managed lands and unmanaged lands. Competing crops account for soybean and corn land stocks that add up in a cropland stock. Managed lands account for cropland and pasture land stocks. Similarly, cropland and pasture sum up in an aggregated managed lands stock (Figure 9-6). Unmanaged land is disaggregated in six stocks accounting for forest, grassland, savannas, shrubland, mixed land and degraded land (Figure 9-7). The soybean land stock changes based on land supply for cropland and the share of soybean on cropland that depends on the relative land unit profit between soybean and corn. Delays in land supply differ among land-use categories. Following land conversion possibilities, it is assumed that changes in competing crops adjust faster than changes in managed lands. Similarly, changes in managed lands adjust faster than changes in unmanaged lands. Delays are assumed to account for the time needed to take the decision of increasing or decreasing a particular land-use and the time needed to make the land conversion.