Project Topics
Get Complete Project Material File(s) Now! »
Data generation process and results’ accuracy with the weight matrix type
The standard Probit model is contrasted to the SAR-probit model as suggested by LeSage et Pace (2009, 2014). Further, we check the sensitivity of the results to various specifications used for the spatial weight structure in the SAR-probit. It finally concludes on the nature of the spatial dependence in various trade-offs, considering the best model.
The table in appendix (A.2.3) displays the results of the standard Probit model that assumes non-spatial relationship among observations alongside with those of the spatial autoregressive model for the trade-off between specialization and diversification. This table shows some stark contrasts between the estimates and marginal effects of both standard Probit model and SAR-probit. It also suggests a significant effect of distance to market as well as distance to the nearest protected area. On the other hand, these effects are not significant for the SAR-Probit. In the latter model, the spatial autocorrelation coefficient, ρ, differs statistically from zero. It is thus clear that estimates and marginal effects of the standard Probit model are biased and inconsistent, allowing non pertinent causations as regards to both distances to the market and the nearest protected area. As result, the decision to choose among specialization and diversification is generated by spatially interdependent processes.
The table in appendix (A.2.4) contrasts the results of various SAR-Probits given the structure of neighbors to illustrate the impact of changing the type of the weight matrix. It shows that varying the type of the matrix does not lead to similar results. Indeed, the Gabriel-relative neighborhood graph (Gabgrahp) reports the absence of spatial patterns. The value of ρGabgraph is close to zero and non significant (ρGabgraph = 0.05), while the KNN and the distance-based matrix report the influence of spatial effects on the households’ likelihood to adopt a livelihoods strategy. Further, differing the number of neighbors yields different results. As the number of neighbors increases, the strength of spatial interaction increases. The spatial weight matrix based on 10 nearest neighbors (10NN) presents higher spatial dependence with ρ = 0.36 compared to 3NN (ρ3NN = 0.23) and 5NN (ρ5NN = 0.29). This analysis validates the sensitivity of results to the weight matrix specification postulated by LeSage et Pace (2009, 2014). Therefore, a good analysis of households livelihoods strategies should involve testing and accounting for spatial effects using spatially explicit econometric models, as well as checking the accuracy of the results with the form of the spatial weight matrix. In the following, we consider the distance-based weight matrix. Indeed, as the households were geo-localized during the field work, the distance-based weight matrix (that yields almost similar results with the 10NN matrix in our study) is better. This matrix allows for the magnitude of interaction among two individuals to be proportional to the inverse euclidian distance among them, while the KNN matrix tends to attribute the same weight to all the k individuals. The quantitative explanatory variables were checked for multi-collinearity, tables (A.2.5) and (A.2.6) in appendix suggest the independence among them.
Spatial dependence
Table 1.2 shows an evidence of spatial autocorrelation among the likelihood of proximal households to choose among livelihood strategies. Indeed, the value of ρ is positive and significant at 1%. This suggests that households tend to mimic the livelihood strategies of their neighbors.
The range of the ρ parameter suggests a difference among the strength of mimicry among the three trade-offs. Indeed, the dependence among closer households’ like-lihood appears to be stronger regarding strategies that value forest rather than choosing or not to choose between forest-based strategies and land-conversion based strategies, or between specialization and diversification.
Coefficient Estimates and Marginal Effects
Table 1.2 presents the sign and the possible variables that drive household heads’ decision in various trade-offs. Tables 1.3, 1.4 and 1.5 present the magnitude or the incremental change resulting from a one-unite change in the independent variables on both own and neighboring likelihood of choosing or not to choosing a strategy in various trade-offs. As a reminder, the independent variables include households’ human, social, natural and financial assets; geographical assets of location and infrastructure; remaining households characteristics and spatial spillover effects.
Specialization vs diversification
Table 1.3 displays the results for the first trade-off where households choose between specialization (either on forest activities, or on cash-crop or small-scale farming) and diversification (mixing cash crop, forest and small-scale farming). One can argue that households specializing in one activity tend to be more income-maximizing oriented, while those tending toward diversification are more risk-coping oriented. Yet, other kinds of characteristics have to be taken into account: some households may have to specialize because there are some barriers to diversification, related to low levels of some assets.
Macro-level empirical studies
A broad existing literature addresses causes of tropical deforestation at national, regional, and global scales using macro-level data in developing countries, considering many type of forests, macroeconomic variables, institutional and policy factors (Kaimowitz et Angelsen, 1998; Pfaff, 1999; Palo, 1999; Bhattarai et Hammig, 2001; Barbier et Burgess, 2001; Geist et Lambin, 2002; Nguyen Van et Azomahou, 2007; Culas, 2007; Damette et Delacote, 2012; Wolfersberger et al., 2015; Tegegne et al., 2016; Combes et al., 2015). Major conclusions from a meta-analysis using results of 150 deforestation models by Kaimowitz et Angelsen (1998) in Brazil, Cameroon, Costa Rica, Indonesia, Mexico, Thailand, Ecuador, the Philippines and Tanzania are that deforestation tends to be greater when economic liberalization and adjustment policy reforms increases; when forested lands are more accessible; when agricultural and timber prices are higher; when rural wages are lower and there are more opportunities for long distance trade. Nguyen Van et Azomahou (2007) use a panel dataset of 59 developing countries over the 1972–1994 period to study the deforestation process. They found no evidence of an Environmental Kuznets Curve and they pointed out political institution failures as factor that can worsen the deforestation process in developing countries. More generally, the evidence supporting the existence of an EKC for deforestation is contrasted (Choumert et al., 2013).
Hosonuma et al. (2012) derive deforestation and degradation drivers using empirical data synthesized from existing reports on national REDD+ readiness activities. They assessed the relative importance as well as the drivers of variability by continent reflecting approximately the period 2000–2010. They used the forest transition model, considering deforestation rate and remaining forest cover in 100 subtropical non-Annex I countries4. They found that, similarly to Asia, the importance of deforestation drivers in Africa varies with different forest transition phases and with different areas. The impact of commercial agriculture on deforestation rises until the late-transition phase and the relative importance of subsistence agriculture remains fairly stable throughout the different phases.
Micro-level empirical studies relating to agent livelihoods decision
There is strong evidence that forests have an important role in insuring livelihood over time and in some cases, contribute to poverty alleviation (Sunderlin et al., 2005). Yet, few studies have investigated the relation between agent livelihoods decision and tropical deforestation at household’s level.
Babigumira et al. (2014) use the CIFOR-PEN dataset, comprising of 7172 households from 24 developing countries to analyze which household and contextual charac-teristics affect land use decision in the developing world. The authors consider the sustainable livelihoods framework and assess the role of various asset types on households’ deforestation. The authors found out that 27% of rural household have cleared forest for agricultural based livelihoods. They also found out that asset poverty does not drive deforestation. Indeed, households with medium to high asset holdings and higher market orientation were more likely to clear forest than the poorest and market-isolated households. They found out that households that cleared forests were closer to the forest and came from villages with higher forest cover.
Relying on a rich panel dataset collected from the Tsimane’ communities in Bolivia, Perge et McKay (2016) analyze the relationship between forest households’ livelihoods strategies, and forest clearing, and the relationship of both to welfare. The authors identify four livelihoods strategies based on households’ reported sources of cash earnings, namely, sale, wage, diversified and subsistence strategy. They find that forest clearing is positively linked to welfare especially for households whose income results from combining agricultural sales and wage activities compared to households adopting other strategies. Households with subsistence strategy are not able to accumulate assets in the long run. As one of the main conclusions, the authors state that households clear only small areas of forest with a positive effect on welfare, enabling accumulation of assets.
Pacheco (2009) define a typology of smallholders that accounts for both livelihoods, farming systems and wealth to analyze smallholders’ deforestation in Uruará and Redenção in the Brazilian Amazon. The author use households survey data from 136 interviews in Uruará and 82 interviews in Redenção area, and find that cattle ranching is associated to greater impact than cash cropping or subsistence agriculture. Contrary to Perge et McKay (2016), a strong correlation between deforestation and the wealth of the farmers is found.
Spatially patterns studies
Spatially explicit econometric studies of drivers of deforestation have taken more importance in the last few years (Ferretti-Gallon and Busch, 2014). According to these studies, most deforestation tends to be located outside the reserve and mountainous area and deforestation occurred primarily within the more accessible Eastern counties and at areas near deforested areas. This illustrates the spread effect of deforestation in the Brazilian Amazon (Mertens et al., 2002). In the same vein, Pfaff et al. (2007) find evidence of spatial spillovers from roads in the Brazilian Amazon’s deforestation. Considering local administrative entity, Amin et al. (2014) found that deforestation activities of neighboring municipalities are correlated with some leakage. As a point of fact, protected areas may shift deforestation to neighboring municipalities.
Using a general spatial two stage least squares model to analyze the determinants of deforestation in 24 Sub-Saharan African countries during the period spanning 1990 to 2004, Boubacar (2012) found that deforestation in one country is positively correlated to deforestation in neighboring countries and that determinants of forest clearing are region specific.
Table of contents :
What drives livelihoods strategies in rural areas? Evidence from the Tridom Conservation Landscape using Spatial Probit Analysis
1.1 Introduction
1.2 Literature Review
1.3 Case study: The Tridom-TCL in the Congo Basin
1.3.1 Data
1.3.2 Livelihoods portfolios
1.4 Theoretical Model specification
1.4.1 Strategic trade-offs in livelihood portfolio selection
1.4.2 A simple microeconomic model of livelihoods portfolios
1.5 Spatial Probit Model
1.5.1 The model
1.5.2 Spatial operator
1.6 What drives livelihoods strategies in the Tridom-TCL?
1.6.1 Descriptive statistics
1.6.2 Spatial dependance and sensitivity
1.6.3 Coefficient Estimates and Marginal Effects
1.7 Discussion and conclusion
2 Households’ livelihoods and deforestation in the Tridom Landscape : A spatial analysis
2.1 Introduction
2.2 Literature Review and contribution
2.2.1 Literature review
2.2.2 Contribution
2.3 Objective and Hypothesis
2.4 A simple microeconomic model of deforestation choices
2.5 Spatial Econometric Procedure
2.5.1 Cross-sectional spatial econometric models
2.5.2 Selection procedure
2.6 Results
2.6.1 Variables and Descriptive statistics
2.6.2 Spatial dependence diagnostic
2.6.3 What are the immediate causes of households’ deforestation in the Tridom-TCL?
2.7 Discussion and conclusion
2.8 Acknowledgements
3 The Value of Endangered Forest Elephants to Local Communities in a Transboundary Conservation Landscape
3.1 Introduction
3.2 Case Study: The Tridom Landscape’s EFEs in the Congo Basin .
3.3 Methodology: Combining Open-Ended and Closed-Ended CV Methods
3.3.1 Overview of CV methods
3.3.2 Survey design
3.3.3 Theoretical model specification
3.3.4 Econometric model specification
3.4 Results
3.4.1 Variable Descriptions and Descriptive Statistics
3.4.2 Econometric results
3.5 Discussion and Conclusion
General conclusion
3.6 General Conclusion
A Appendix of chapter 1
A.1 Figures
A.1.1 Study Area and Location of households surveyed
A.1.2 Absolute Frequencies of Households Strategies
A.1.3 Trade-offs in livelihoods, Deforestation and Per Annum Yiels/ha155
A.2 Tables
A.2.1 Measurement of NTFP in the forest-based strategy
A.2.2 Descriptive statistique among various trade-off
A.2.3 Standard Probit VS SAR-Probit
A.2.4 results’ accuracy with the weight matrix type
A.2.5 Variance Inflation Factor
A.2.6 Correlation matrix among quantitative variables
B Appendix of chapter 2
B.1 Figures
B.1.1 Diagnostic Plots for Regression Analysis
B.1.2 Livelohoods Strategies and Per Annum Yiels/ha
B.2 Tables
B.2.1 Full Spatial Autoregressive Model
C Appendix of chapter 3
C.1 Figures
C.1.1 Bid structure
C.1.2 Aggregate WTP for elephant conservation of the local Tridom population (106* CFA)
C.1.3 Mean willingness to pay per subdivision and ethnic distribution( CFA)
C.2 Tables
C.2.1 Ethnic Representativeness (without protesters)
C.2.2 Spatial Representativeness
D Questionnaire
General bibliography
