ECONOMIC VIABILITY OF CERRADO VEGETATION MANAGEMENT UNDER CONDITIONS OF RISK

Cerrado vegetation is Brazil’s second largest biome, comprising about 388 municipalities in Minas Gerais state alone and serving as an important source of natural resources. A large share of the wood charcoal produced in Minas Gerais is sourced from Cerrado vegetation. The objective of this work is to assess the economic viability of Cerrado vegetation management for wood charcoal production, under conditions of risk. The study site is a fragment of Cerrado subjected to fi ve levels of intervention as to basal area removal. For risk analysis, the Monte Carlo method was applied, using charcoal price, interest rate and land value as input variables, and using Net Present Value as output variable over an infi nite planning horizon. It was concluded that introducing risk in the economic analysis of the various Cerrado management regimes helped provide additional information to that obtained by deterministic analysis, improving understanding and ensuring safety in decision-making about the economic viability of such regimes. For all treatments, the probability of VPL being negative increases with increasing cutting cycle lengths. For all treatments, the optimal cutting cycle is ten years. Treatments where a larger volume of wood was removed proved less prone to risks of economic inviability since they secure more revenue than treatments where less wood was removed.


INTRODUCTION
The Brazilian Cerrado covers an area of roughly two million km 2 , equivalent to 23% of the Brazilian territory.It is the country's second largest biome, second only to the Amazon Forest, extending from marginal areas of the Amazon Forest down to Paraná state.In other parts of South America, the Cerrado is also present in Bolivia, Paraguay and Venezuela (RATTER et al., 1997).
In Minas Gerais alone, the Cerrado covers an area of 384,366 km 2 and embraces 388 municipalities (PEREIRA et al., 1997).Minas Gerais state is the largest producer and consumer of wood charcoal for the steel industry, having consumed 21.17 million meters of charcoal in 2008 (ASSOCIAÇÃO MINEIRA DE SILVICULTURA -AMS, 2009).
The Cerrado was the fi rst supplier of wood for charcoal production to assist the steel industry of Minas Silva, S. C. da et al.
Gerais state.In 1980, 85.8% of the total charcoal input consumed in Brazil was sourced from native forests.This percentage decreased in the following years, reaching 24.6% in 1997 (REZENDE et al., 2002).However, the supply of charcoal sourced from nonnative forests failed to meet the increase in demand by steel plants.Thus, in 2008, 36.2% of the charcoal consumed in Minas Gerais was sourced from native forests (AMS, 2009).
Several authors have analyzed the economic viability of using wood from Cerrado vegetation to manufacture charcoal for the steel industry, including Oliveira et al. (1998Oliveira et al. ( , 2002) ) and Rezende et al. (1986).In these studies, economic analysis followed a deterministic approach based on classic methods of investment analysis that include Net Present Value (VPL) and Internal Rate of Return (TIR).
Actual reality may nevertheless be poorly captured by these economic indicators (BRUNI et al., 1998).There is no certainty that expected project estimates will match reality, since there is not enough ability to predict every factor affecting the future, whether favorably or unfavorably.Thus, acceptance of a project with a positive VPL also entails uncertainty because it is based on a cash fl ow which in turn relies on uncertain estimates (LAPPONI, 2007).
Globalization and the complexity of reality cause uncertainties and market risks to hinder assessment of project effi ciency.According to Securato (2007), risk is the probability of events occurring that will generate loss or uncertainty.
Within this context, risk measurement and analysis techniques provide not only additional information to VPL or TIR results but also a perception of the intrinsic characteristic of a project and the impact of likely future events on the decision of accepting the project (LAPPONI, 2007).Moore and Weatherford (2005) argue that the Monte Carlo method can be used as an alternative method for project assessment whereby risks involved are plainly expressed and easy to understand so as to help decisionmaking.Indicators thus go from being deterministic to becoming stochastic or probabilistic.
This work aims to assess the economic viability of Cerrado vegetation management for charcoal production, under conditions of risk.

Database
The study site is a fragment of Fazenda Vitória, an estate owned by V&M Florestal and located in the municipality of Coração de Jesus, Minas Gerais state.
The above fragment falls into the 'Cerrado sensu stricto' vegetation category, having been subjected to exploration in past decades, though the intensity of intervention is unknown.The area lies in the northern part of Minas Gerais state, 490 km away from Belo Horizonte, at an a altitude of 800 m, having fl at relief.
According to Köppen classification, the study region falls into the 'Aw' group: dry tropical climate, average annual precipitation of 820 mm and average annual temperature of 25°C.
In 1986, the State Forest Institute (IEF) set up an experiment in the above study site consisting of six treatments, with 50%, 70%, 80%, 90% and 100% removal of the total basal area, besides a control treatment (no intervention).The experiment covered an area of 30 hectares and treatments were distributed randomly in fi ve blocks, to a total of 30 plots 1 ha each.

Investment analysis under condition of risk
Project investments presuppose the existence of economic, fi nancial, technological, administrative, legal and natural risks.Risks presuppose the possibility of something going wrong within an estimated probability distribution (COELHO JUNIOR et al., 2008).
According to the above authors, due to its ease of use, the Monte Carlo method provides several estimation alternatives for the probability distribution prior to decision-making, offering advantages over other simulation theories.
The risk analysis using the Monte Carlo simulation technique is presented in four stages, as follows.

Model development
In order to assess investment, taking into account the relevant risks and uncertainties, the following cash fl ow was used, as provided in Figure 1.
Table 1 provides the volume of removed timber in m 3 /ha at the moment of intervention in 1986, in inventories (1996, 1998 and 2004)

Identifi cation of risks and uncertainties
To conduct the analysis it is necessary to identify opportunities and threats infl uencing the relevant project variables.Independent variables considered as input variables (inputs) included value of land, charcoal price and interest rate.To measure input uncertainties, a triangular distribution was used attributing maximum, minimum and most likely values to these variables (Table 3).

Identifi cation of analysis variables or output variables
The net present value method over an infi nite planning horizon (VPL ∞ ) was used in order to defi ne the optimal economic cutting cycle and to assess the different Cerrado management regimes, under deterministic conditions.The formula used for VPL ∞ calculation was developed by Oliveira ( 2006

Silva, S. C. da et al.
The VPL is given by the difference between the value of revenues and costs occurring in the year when treatments were implemented (1986 = year 0) and the VFL is given by the difference between the value of revenues and costs occurring in the base year defi ning the cutting cycle (1996 = 10-year cutting cycle = year 10; 1998 = 12-year cutting cycle = year 12; ...; 2029 = 43-year cutting cycle = year 43).
The VPL ∞ of different management regimes was considered an output variable (outputs).

Simulation and model analysis
For the risk analysis, @Risk software was used (PALISADE CORPORATION, 1995).According to Bentes-Gama (2003), this program enables applying the Monte Carlo method to simulate values of random and independent variables (revenue and cost) and, as a result, to obtain values of the profi t variable.
After assembling the cash fl ow, 10,000 simulations were run for the input variable (VPL ∞ ), using pseudorandom numbers, in other words, a series of values was generated for this variable so as to obtain its simple and cumulative frequency distribution.obtained when shorter cutting cycles were used, regardless of treatment.For a given cutting cycle, the higher the level of basal area intervention, the higher profi tability is, as expressed by VPL ∞ .

RESULTS AND DISCUSSION
Table 5 provides results of simulations run in order to obtain VPL ∞ occurrence probabilities for each treatment and cutting cycle being studied, while Figure 2 provides simple and cumulative frequency distributions of this indicator for the 10-year cutting cycle.The treatment in which 50% of basal area is removal, in the 10-year cutting cycle, 5% of VPL ∞ values are above R$1,134.29.In the 43-year cutting cycle, 5% of VPL ∞ values are above R$ 244.36.The same trend is observed in the remaining situations.Figura 2 -Histogramas dos valores de VPL ∞ para os tratamentos, no ciclo de corte de 10 anos.

Silva, S. C. da et al.
Table 6 provides probabilities of VPL ∞ being below zero, indicating economic inviability of the relevant treatment.The 50% treatment shows higher occurrence probabilities of a negative VPL ∞ .The probability ranges from 17.80% in the 10-year cycle to 73.49% in the 43-year cycle, indicating that Cerrado management following this prescription entails high risk.The clear cutting treatment (100% of vegetation removed), on the other hand, entails low risk even in a 24-year cutting cycle (4.83%).
The above check can also be done through analysis of numerical data in Table 7.In the 50% treatment, the expected VPL ∞ value (or mean value) is R$369.46,and its standard deviation is R$ 412.87.Thus, with a little less than one standard deviation a zero VPL ∞ is attained.As for the 100% treatment, 1.8 standard deviations are necessary for a negative VPL ∞ .Additionally, the coeffi cient of variation of the 50% treatment (111.74%)shows higher risk than the coeffi cient of the 100% treatment (55.74%).
The mean VPL ∞ , or expected VPL ∞ value, whose calculation is risk based (Table 7) is invariably higher than the VPL ∞ whose calculation is based on deterministic analysis (Table 4).Take for instance the 80% treatment, with risks being computed, the mean VPL ∞ is R$735.91.In the deterministic analysis, the VPL ∞ is R$ 528.69.
The probability of VPL ∞ being below values calculated by Oliveira (2006) was also assessed without computing risks.Probabilities in Table 8 indicate that in the 100% treatment, 10-year cutting cycle, the VPL ∞ calculated with risks being computed has a 36.77%chance of being lower than the VPL ∞ shown in Table 4 (R$1,086.89),where risks were not computed.

Figure 1 -
Figure 1 -Cash fl ow for Cerrado vegetation management.
), as follows: where: VPL = Net present value; VFL = Net fi nal value; PT = Price of land; i = Annual interest rate; n = Cutting cycle duration, in years.

Figure 2 -
Figure 2 -VPL ∞ value histograms for each treatment, in the 10-year cutting cycle.

Table 1 -
Volumes of removed timber (RET) for each treatment and cutting cycle.

Table 2 -
Costs involved in charcoal production.

Table 3 -
Price of land and charcoal and interest rate.

Table 6 -
Probabilities of VPL ∞ being negative for each treatment and cutting cycle.Tabela 6 -Probabilidades de o VPL ∞ ser negativo para os tratamentos e ciclos de corte.

Table 7 -
A descriptive statistic of VPL ∞ for each treatment, in the 10-year cutting cycle.Tabela 7 -Estatística descritiva do VPL ∞ para os tratamentos, no ciclo de corte de 10 anos.