Accounting for institutional quality in global forest modeling
Introduction
Deforestation accounts for 12% of anthropogenic CO2 emissions (Smith et al., 2014), causes biodiversity loss (Gibson et al., 2011), soil erosion (Smith et al., 2016), ground water stress, and changes in local rainfall patterns (Garcia-Carreras and Parker, 2011). The literature has widely acknowledged the conversion of forest land to agricultural land as the main driver of deforestation (Busch and Ferretti-Gallon, 2017, Gibbs et al., 2010, Mayaux et al., 2013). This logic is typically reflected in land use change models. In the Global Forest Model (G4M global v.4.0), a representative land owner makes a return-maximizing land use decision, based on a comparison of net present values of agricultural and forest land. The biophysical properties and the agricultural suitability of the land are taken into account for the decision (Kindermann et al., 2006, Kindermann et al., 2008).
After Brazil's historic success in curbing deforestation by more than 70% (Tollefson, 2015) through enhanced enforcement and fining of illegal deforestation (Cisneros et al., 2015, Hargrave and Kis-Katos, 2013, Nepstad et al., 2014), more recent literature on deforestation suggests that next to economic and biophysical factors, the quality of political institutions is a key parameter influencing land use change decisions (Bhattarai and Hammig, 2001, Bhattarai, 2004, Buitenzorgy and Mol, 2011, Galinato and Galinato, 2012, Koyuncu and Yilmaz, 2013) .1 Furthermore, examples such as Costa Rica and Colombia, with similar gross national income income trends (World Bank, 2015c), but diverging forest cover trends from 1990 to 2015, suggest that non-economic factors also significantly influence deforestation outcomes. Costa Rica experienced a 7.5% forest cover gain and Colombia a 9% forest cover loss (FAO, 2015a, FAO, 2015b).
By providing intertemporal contracts, institutions help generate regularity in social behavior and can prevent the overuse of common goods (Aoki, 2001). In this article we refer to this capacity with the concept of ‘environmental institutional quality’. It measures the extent to which existing political institutions lead to a sustainable use of common environmental resources. In order to measure environmental institutional quality, this paper builds on the FAO and PROFOR's (2011) forest governance framework that suggests that the quality of political institutions in the forest sector can be measured by three main components, which are (i) regulatory frameworks, (ii) planning and decision making processes, and (iii) the implementation and enforcement of policies. Despite the increasing attention that is paid to the quality of political institutions in deforestation processes in the empirical literature, it is still uncommon to take differences in the quality of institutions into account when modeling forest cover change trends (Benítez et al. (2007) and Wang et al. (2016) represent notable exceptions). Magliocca et al. (2015) make a strong case for using synthesis knowledge to improve process-based land change models. This resesarch project aims at taking this new trend in the empiricial literature into account for future forest cover change simulations of the Global Forest Model. In the Global Forest Model, all factors causing deviations of purely economically motivated land use change decisions are captured by the residual calibration factor. This factor is multiplied by the estimated net present value of forestry, to yield an adjusted net present value of forest land use.
This paper tests the hypothesis whether the residual calibration factor can be reduced by including an additional parameter into the model, which measures the quality of political institutions that are relevant for the sustainable management of environmental resources like forests. Reducing the residual calibration factor would reduce unexplained factors influencing the forest cover change decisions and thus improve the representation of deforestation processes in the model. The hypothesis is first tested through a regression analysis using the residual calibration factor for the 2000 to 2010 period as a dependent variable and an environmental institutional quality indicator as independent variables. In a second step, the indicator is applied to the model, to evaluate, in a third step, whether this can substantially improve the simulation. Finally, a test of the significance of the findings is conducted using forest cover data for 2015. The test indicates that for the countries selected, a better match between the model forecast and observed forest cover change trends can be achieved by accounting for environmental institutional quality.
The remainder of the paper is organized as follows: in section 2 the data and the construction of the environmental institutional quality indicator are presented. Section 3 presents the methodological steps taken to integrate the indicator into the model. Section 4 presents the results, and section 5 discusses the relevance of the findings for other forest cover change models and ecological process models in general.
Section snippets
Environmental institutional quality index
An in-depth review of existing data sources reveals that comprehensive cross country datasets, measuring differences in the quality of institutions affecting the management of natural resources and forests in particular, are unavailable. At the same time, Kishor and Belle (2004) highlight that general governance indicators are unsuitable to measure the performance of institutions charged with the management of natural resources and forests in particular, because general and specific
Methods
The Global Forest Model compares net present agricultural and forestry values for a grid cell (resolution 0.5° × 0.5°) to simulate the land use change decision of a representative land owner, who may be a state or individual, depending on the ownership regime (Gusti and Kindermann, 2011, Kindermann et al., 2006, Kindermann et al., 2008). The subscript i indicates that the values vary for each country. A net present value calculation allows to optimize investment decisions, based on
Regression analysis
Table 2 shows that the regression yields statistically significant results at the 0.1% level (p-value = 0.001), when the model is tested with all control variables (column 1), no control variables (column 2), and each single control variable (column 3–7). Furthermore, Table 3 reports results of a robustness test with a restricted range (0.05–15) of residual calibration factor values and shows that these tests also yield statistically significant results.
Reduction of the residual calibration factor
When using the environmental
Discussion
The above analysis shows that accounting for differences in environmental institutional quality through the in the Global Forest Model, improves the representation of past deforestation trends. While a wider application of this technique could be interesting for other ecological models, data constraints currently considerably limit this potential. This section discusses the findings of our analysis on the context of the broader land use change and environmental modeling literature, as well
Conclusion
Incorporating an indicator on environmental institutional quality into the Global Forest Model significantly reduces the residual calibration factor of the model for the period 2000 to 2010. First tests for the subsequent calibration period (2010–2015) using data from the FAO 2015 Forest Resources Assessment (2015a), show that including the index can contribute to a better forecast of deforestation trends. Future research should concentrate on providing better data on the quality of
Acknowledgments
Funding: This work was financially supported by the German National Member Organization of the International Institute for Applied Systems Analysis (IIASA) and carried out during the Young Scientists Summer Program (YSSP) 2015. It was furthermore supported by the project “Options Market and Risk-Reduction Tools for REDD+” funded by the Norwegian Agency for Development Cooperation (NORAD).
References (78)
Modeling human decisions in coupled human and natural systems: review of agent-based models
Ecol. Model.
(2012)- et al.
Global potential for carbon sequestration: geographical distribution, country risk and policy implications
Ecol. Econ.
(2007) - et al.
Institutions and the environmental Kuznets curve for deforestation: a crosscountry analysis for Latin America, Africa and Asia
World Dev.
(2001) - et al.
2014. Simulating the impacts of reduced rainfall on carbon stocks and net ecosystem exchange in a tropical forest
Environ. Model. Software
(2014) - et al.
Global land-use implications of first and second generation biofuel targets
Energy Pol.
(2011) - et al.
An adaptive agent model for analysing co-evolution of management and policies in a complex rangeland system
Ecol. Model.
(2000) - et al.
Are more complex physiological models of forest ecosystems better choices for plot and regional predictions?
Environ. Model. Software
(2016) - et al.
A geospatial model of forest dynamics with controlled trend surface
Ecol. Model.
(2010) - et al.
Long-term forest management and timely transfer of carbon into wood products help reduce atmospheric carbon
Ecol. Model.
(2009) - et al.
The importance of introducing spatial heterogeneity in bio-economic forest models: insights gleaned from FFSM++
Ecol. Model.
(2015)