Forest Concessions and Deforestation

Abstract

Deforestation is the most critical issue of climate change in Indonesia, as well as in other tropical countries. Reducing deforestation when major forests are managed under concession or permit system is increasingly challenging. The main focus of this chapter is to quantitatively investigate the association between forest permits and deforestation at national level in Indonesia. Hence, this chapter discusses the state of forest management and the forest permit system in Indonesia. Theoretical and empirical literature on the relationship have been extensively reviewed and summarized in the context of Indonesia. Furthermore, the empirical analysis with provincial data for the relation between two major forest permits, the logging permit and the plantation conversion permit, and forest cover loss are implemented. The results are different from the expectation posing a question on the effectiveness of the current forest moratorium policy as well as the forest tariff policy to the attempts in mitigating deforestation. Based on the empirical findings, some possible alternative forest policies in short- and long-term perspectives are discussed.

This chapter contains parts of analytical results in the journal article: Indarto, J., Kaneko, S., and Kawata, K., 2015, Do forest permits cause deforestation in Indonesia?, International Forestry Review, 17(2):165–181.

Appendix

1.1 Material and Methods

1.1.1 Framing the Model

This analysis has been framed in the proximate underlying approach of deforestation (Geist and Lambin 2002). Proximate causes could relate to agricultural expansion, wood extraction, and infrastructure extension. These factors are even sometimes used as proxies of deforestation (Choumert et al. 2013). Underlying causes are typically classified into several factors: economic, demographic, technological, policy/institutional, and cultural factors. In this analysis, forest permits are considered as one of the policy factors undertaken by the government in managing state forests. The government has a strong authority to issue or not, to revoke, to designate location, and to decide how much area will be permitted within production forest. Furthermore, economic, demographic, and technological factors represented by gross domestic regional product per capita (GDRP/capita), population, and oil palm productivity, respectively, are other important factors affecting deforestation. It should be noted that cultural factors and other underlying variables are not incorporated in this analysis.

For economic factors, the immiserization theory postulates that rising economic levels (GDRP/capita) generate off-farm job opportunities that can prompt a shift away from reliance on forests (Rudel and Roper 1997). Conversely, the forest frontier theory suggests that better capital availability generated by economic development in forest regions enables loggers to expand and intensify their logging activities, which may or may not lead to greater deforestation (Rudel and Roper 1997). Generally, increasing per capita income may increase deforestation, but at a certain level, and then the trend is reverted. The existence of the Environmental Kuznets Curve (EKC) for deforestation is confirmed by some studies for cross-country case (Koop and Tole 1999; Barbier and Burgess 2001), Latin America (Cropper and Griffiths 1994; Barbier and Burgess 2001; Bhattarai and Hammig 2001; Culas 2007), and Africa (Bhattarai and Hammig 2001). A different pattern, a U-shaped EKC for deforestation, is found for Asia (Koop and Tole 1999; Bhattarai and Hammig 2001), arguing that it is due to the successful story of reforestation programs in certain tropical countries. Considering reforestation programs promoted through the development of PCP, analysis of this chapter also expects that a U-shaped EKC for deforestation exists in Indonesia.

Demographic factors also have dual effects on deforestation. The Malthusian theory suggests that an increase in population increases pressure on natural resources (Palo 1994). However, the Boserup effect indicates that more population may reduce deforestation through better innovation, technology, and institutions (Bilsborrow and Geores 1994). In the context of Asian countries, certain studies suggest that the population’s effect on deforestation is insignificant (Cropper and Griffiths 1994, Koop and Tole 1999, Culas 2012). This analysis follows the Malthusian theory, expecting that an increase in population increases deforestation. Population of Indonesia increased approximately 1.49 % from 2000 to 2010 and is projected to grow 1.39 % from 2010 to 2015 (BPS 2010).

Technology factors specifically refer to technology in the agriculture sector regarded as one of the main pressures on the forest frontier, whereas technology factors in forest sector is represented by income as stated above. In Indonesia, oil palm development brings a significant pressure on forests. This analysis expects that better agricultural technology, represented by oil palm productivity, is negatively correlated to deforestation. Improving productivity enables oil palm plantations to produce more in the same amount of land (Mahapatra and Kant 2005), resulting in less demand of forest land to be converted and less pressure on forests. On the other side, higher oil palm productivity becomes a strong economic incentive of oil palm expansion. A more available capital generated by an increasing economic, likewise, will induce the development of oil palm plantation, bringing about more pressures on forests.

Since most forests are owned by the government, issuing and/or terminating forest permits is one of the important policies. LP is issued in productive production forest. The holders are allowed to extract timber immediately, but selectively. On the other side, PCP that has been designed as one of the main programs of forest rehabilitation and reforestation should be established in unproductive production or degraded or logged-over forests. In this type of permit, conceptually, the holders can utilize those areas by planting trees first before harvesting timber. In practice, however, immediate clear cutting has been taking place during land preparation stage. Lastly, utilization of production forest is subject to sustainable forest management, expecting that forest permits will not induce deforestation. Therefore, LP is expected to positively and/or insignificantly associated with forest cover loss, while its negative correlation is expected for PCP.

1.1.2 Empirical Model

To estimate the impacts of LP and PCP on forest cover loss, a panel data is developed by using the Forest Cover Change known as Hansen dataset (Hansen et al. 2013a). The main interest of this analysis is the marginal effect of area of valid LP and PCP on forest cover loss. The econometric model is specified as follows:

$$ {\mathrm{DEF}}_{it}=\alpha +{\beta}_1{\mathrm{X}}_{it}+{\beta}_2{\mathrm{LP}}_{it}+{\beta}_3{\mathrm{PCP}}_{it}+ut+{\varepsilon}_{it} $$

where DEF, LP, PCP, X, u, ε, i, and t denote deforestation in ha, LP in ha, PCP in ha, other potential explanatory variables, time or year effect, within-entity errors, province, and year, respectively.

Forest cover loss is used as the proxy of deforestation (DEF). Explanatory variables, including GDRP/capita in thousand Rupiah at 2000 constant price, square of GDRP/capita, population, population density per square kilometer, population growth in annual percentage, oil palm productivity in ton per ha, and timber production in cubic meter, are controlled in this model. Year dummy is expected to capture time effects, for each year, whereas province dummy captures region-specific character in each province. Furthermore, it takes into account roles of sustainable forest management, which is measured by Forest Stewardship Council (FSC) certified area in ha for both permits.

Since deforestation is a complex process, correlation between unobserved components and some explanatory variables is assumed to exist. To this end, the fixed-effect estimator is applied in this analysis (Damette and Delacote 2011). Heterogeneity issue is solved by running robust standard errors.

During analysis processes, those three demographic factors (population density, population growth, and number of population) were separately examined. At the end, since a number of population brought a significant sign, it was chosen to be incorporated in the model. Consequently, results and discussion on demographic factor in this chapter refer to the number of population. It was also decided to exclude oil palm productivity from the analysis for two reasons: its sign was consistently insignificant during the analysis process and excluding this insignificant factor improved the model specification because of the degree of freedom. Explanation of variables included in the discussion is presented in Table 5.8.

Table 5.8 Variables, data sources, and expected signs

1.1.3 Data

The data of this analysis are provincial data covering all provinces in Indonesia. The data of deforestation (forest cover loss), the dependent variable, are based on Global Forest Change or Hansen dataset (Hansen et al. 2013a) and accessed from Global Forest Watch website (Hansen et al. 2013b). In this dataset, forests refer to tree cover, where trees are defined as all vegetation taller than 5 m in height and canopy cover at least 30 % at the Landsat 30 × 30 m pixel scale. By this definition, commercial forestry plantations, as well as primary and secondary forests, are accounted as forests. To be in line with how the government defines forests (MoF 2008), the threshold of 30 % canopy cover was chosen. Threshold of 10 % canopy cover was utilized to check the robustness of estimation results (see Table 5.9). Deforestation is defined as forest cover loss, the disturbance or complete removal of tree cover canopy. It is calculated by subtracting forest extent in the previous year with that in the current year for each province. Positive value of forest cover loss represents the magnitude of deforestation area.

Table 5.9 Estimation results: impacts of LP and PCP on forest cover loss with 10% threshold of canopy cover

Since the government has not designated any LP and PCP in the provinces within Java and Bali islands, this analysis employs 26 out of 33 provinces, excluding the provinces in Java and Bali. This analysis takes into account the dynamics of decentralization in which new provinces were established (Banten from West Java in 2000, Kepulauan Bangka Belitung from Sumatera Selatan in 2000, Gorontalo from Sulawesi Utara in 2000, Papua Barat from Papua in 2001, Kepulauan Riau from Riau in 2002, and Sulawesi Barat from Sulawesi Selatan in 2004). In those cases, the data of the explanatory variables are available 1 year after their establishment years. Therefore, the balanced panel data for this analysis is in the period of 2005–2012. The descriptive summary of the data is given in Tables 5.10 and 5.11 for a more detailed summary.

Table 5.10 Descriptive statistics of main variables, 2005–2012

Table 5.11 Descriptive statistics of panel data, 2005–2012