<%BANNER%>

Impact of Poverty on Deforestation

Permanent Link: http://ufdc.ufl.edu/UFE0024485/00001

Material Information

Title: Impact of Poverty on Deforestation A Case Study in the Philippines
Physical Description: 1 online resource (73 p.)
Language: english
Creator: Georges, Jessica
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: community, deforestation, forest, income, inequality, managment, poverty
Food and Resource Economics -- Dissertations, Academic -- UF
Genre: Food and Resource Economics thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science IMPACT OF POVERTY ON DEFORESTATION A CASE STUDY IN THE PHILIPPINES By Jessica Georges May 2009 Chair: Carmen Carrio acuten-Flores Major: Food and Resource Economics This study presents an empirical analysis of the effect of poverty and income distribution on deforestation using regional level data from the Philippines. It examines the impact of poverty on deforestation, while accounting for environmental improvement and the fact that poverty and the environment evolve together. In addition, it analyzes whether community based forest management agreements have an impact on the severity of poverty and deforestation. The study estimates two simultaneous equations by using the Two Stage Least Squares estimation method based upon the Environmental Kuznets Curve conceptual framework. The main results are (i) the area of forest hectares given to each community to manage is an important predictor of the amount of deforestation in a region - the larger area a community is allowed to manage the less deforestation occurs; (ii) the increase number of community forest management sites in a region is likely to lead to more deforestation due to fragmentation of the forest; (iii) income inequality does not spur environmental improvement as portrayed in the dominant literature and (iv) indicators related to severity of poverty have a negative impact on deforestation. The results highlight the important role of community forest management in reducing deforestation, but emphasizes that there is an optimal and strategic way that communities should be awarded the area of forest. When awarded properly community forest management can assist in reducing poverty through economic benefits and reduce deforestation through sustainable practices. The evidence from this analysis underscore the role severity of poverty has on deforestation and reveals a way that policy makers can make their community forest management programs more effective.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jessica Georges.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Carrion-Flores, Carmen.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024485:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024485/00001

Material Information

Title: Impact of Poverty on Deforestation A Case Study in the Philippines
Physical Description: 1 online resource (73 p.)
Language: english
Creator: Georges, Jessica
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: community, deforestation, forest, income, inequality, managment, poverty
Food and Resource Economics -- Dissertations, Academic -- UF
Genre: Food and Resource Economics thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science IMPACT OF POVERTY ON DEFORESTATION A CASE STUDY IN THE PHILIPPINES By Jessica Georges May 2009 Chair: Carmen Carrio acuten-Flores Major: Food and Resource Economics This study presents an empirical analysis of the effect of poverty and income distribution on deforestation using regional level data from the Philippines. It examines the impact of poverty on deforestation, while accounting for environmental improvement and the fact that poverty and the environment evolve together. In addition, it analyzes whether community based forest management agreements have an impact on the severity of poverty and deforestation. The study estimates two simultaneous equations by using the Two Stage Least Squares estimation method based upon the Environmental Kuznets Curve conceptual framework. The main results are (i) the area of forest hectares given to each community to manage is an important predictor of the amount of deforestation in a region - the larger area a community is allowed to manage the less deforestation occurs; (ii) the increase number of community forest management sites in a region is likely to lead to more deforestation due to fragmentation of the forest; (iii) income inequality does not spur environmental improvement as portrayed in the dominant literature and (iv) indicators related to severity of poverty have a negative impact on deforestation. The results highlight the important role of community forest management in reducing deforestation, but emphasizes that there is an optimal and strategic way that communities should be awarded the area of forest. When awarded properly community forest management can assist in reducing poverty through economic benefits and reduce deforestation through sustainable practices. The evidence from this analysis underscore the role severity of poverty has on deforestation and reveals a way that policy makers can make their community forest management programs more effective.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jessica Georges.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Carrion-Flores, Carmen.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024485:00001


This item has the following downloads:


Full Text

PAGE 1

1 IMPACT OF POVERTY ON DEFORESTATION A CASE STUDY IN THE PHILIPPINES By JESSICA GEORGES A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MA STER OF SCIENCE UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Jessica Georges

PAGE 3

3 To God and my family

PAGE 4

4 ACKNOWLEDGMENTS First, I would like to thank God for his grace and mercy that endures forever. I would like to acknowledge my family and the Abiding Faith Christian Church, for the constant prayers I would also like to acknowledge my supervisory committee: Flores for the encouragement and assisting me in narrow ing down my thesis topic; Dr. Evans for h is advice and feedback; Dr. Hyden for agreeing to be part of my supervisory committee even though he is retired. I would like to express my gratitude to Mr. Jauregui for helping me understand the statistical software and answering my questions

PAGE 5

5 TABLE O F CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 7 LIST OF FIGURES .............................................................................................................................. 8 ABSTRACT .......................................................................................................................................... 9 CHAPTER 1 INTRODUCTION ....................................................................................................................... 11 Problem Statement ...................................................................................................................... 11 Objectives and Hypotheses ......................................................................................................... 12 Thesis Outline .............................................................................................................................. 14 2 BACKGROUND ......................................................................................................................... 15 General Features of the Philippines ........................................................................................... 15 Deforestation and Poverty in the Philippines ............................................................................ 16 Philippin es Forest History ......................................................................................................... 16 National Forest Policy ................................................................................................................. 18 Philippine Strategy for Sustainable Development (PSSD) ............................................... 18 Master Plan for Forestry Development (MPFD) ............................................................... 18 Forests on a Global Scale ........................................................................................................... 20 3 LITERATU RE REVIEW ........................................................................................................... 23 Environmental Kuznets Curve ................................................................................................... 23 Different Scenarios for the EKC ................................................................................................ 26 Limitations and Criticism of the EKC ....................................................................................... 27 Poverty and the Environment ..................................................................................................... 29 Importance of Forests .................................................................................................................. 30 Climate Change and Deforestation ............................................................................................ 32 Community Management and Forest ......................................................................................... 33 4 ANALYTIC AL FRAMEWORK ............................................................................................... 35 Ordinary Least Squares ............................................................................................................... 35 Two Stage Least Squares ............................................................................................................ 37 Validity of 2SLS .......................................................................................................................... 37 Specification Test ................................................................................................................ 37 Overidentification Restrictions ........................................................................................... 38 Weak Instruments ................................................................................................................ 39

PAGE 6

6 Multicollinearity .................................................................................................................. 40 Heteroskedascity .................................................................................................................. 40 5 EMPIRICAL MODEL ................................................................................................................ 42 Empirical Model Description and Specification ....................................................................... 42 Measurement and Specification Model Variabl es .................................................................... 43 Identification Strategy ................................................................................................................. 49 6 RESULTS .................................................................................................................................... 50 Testing for Endoge neity & Overidentifying Restrictions ......................................................... 50 Testing Instrument Strength ....................................................................................................... 51 Heteroskedascity ......................................................................................................................... 52 Multicollinearity .......................................................................................................................... 53 First -Stage Regression ................................................................................................................ 54 OLS and 2SLS Regression ......................................................................................................... 55 7 CONCLUSIONS AND IMPLICATIONS FOR POLICY ....................................................... 57 Summary ...................................................................................................................................... 57 Conclusions and Implications for Po licy ................................................................................... 57 Further Research and Limitations .............................................................................................. 59 APPENDIX A MAP OF THE PHILIPPINES .................................................................................................... 61 B OFFICIAL POVERTY LINE ..................................................................................................... 62 C FULL RESULTS ......................................................................................................................... 63 D SUMMARY STATISTICS ........................................................................................................ 67 LIST OF REFERENCES ................................................................................................................... 68 BIOGRAPHICAL SKETCH ............................................................................................................. 73

PAGE 7

7 LIST OF TABLES Table page 6 1 Specification test and Overidentification test ...................................................................... 51 6 2 Weak Identification test ........................................................................................................ 52 6 3 Breusch Pagan Heteroskedasc ity test .................................................................................. 52 6 4 Multicollinearity .................................................................................................................... 53 6 5 First -stage Regression and Instrument Significance test .................................................... 54 6 6 OLS and 2SLS Regression ................................................................................................... 56 A 1 Official poverty line (in pesos) ............................................................................................. 62 C1 Full estimation by OLS without cluster ............................................................................... 63 C2 Full estimation by OLS with cluster .................................................................................... 64 C3 Full estimation by 2SLS without cluster ............................................................................. 65 C4 Full estimation by 2SLS with cluster ................................................................................... 66 D 1 Summary statistics ................................................................................................................ 67

PAGE 8

8 LIST OF FIGURES Fig ure page 2 1 Characteristics of the W orlds Forests.. ............................................................................... 21 2 2 Global designated functions of Forests.. .............................................................................. 22 3 1 Environmental Kuznets Curve ............................................................................................. 24 3 2 Different scenarios of EKC .................................................................................................. 27 3 3 Forest contributions ............................................................................................................... 33 4 1 Instrume ntal Variable approach ........................................................................................... 36 5 1 Philippines 2006 Lorenz Curve ............................................................................................ 45 7 1 Philippines Environmental Kuznets Curve .......................................................................... 57 A 1 Map of the Philippines .......................................................................................................... 61

PAGE 9

9 Abstract of Thesis Presented to the Gra duate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science IMPACT OF POVERTY ON DEFORESTATION A CASE STUDY IN THE PHILIPPINES By Jessica Georges May 2009 Chair: Carmen Carri n -Flores Major: Food and Resource Economics This study presents an empirical analysis of the effect of poverty and income distribution on deforestation using regional level data from the Philippines. It examines the impact of poverty on deforestation, w hile accounting for environmental improvement and the fact that poverty and the environment evolve together. In addition, it analyzes whether community based forest management agreements have an impact on the severity of poverty and deforestation. The study estimates two simultaneous equations by using the Two Stage Least Square s estimation method based upon the Environmental Kuznets Curve conceptual framework. The main results are (i) the area of forest hectares given to each community to manage is an impo rtant predictor of the amount of deforestation in a region the larger area a community is allowed to manage the less deforestation occurs; (ii) the increase number of community forest management sites in a region is likely to lead to more deforestation due to fragmentation of the forest; (iii) income inequality does not spur environmental improvement as portrayed in the dominant literature and (iv) indicators related to severity of poverty have a negative impact on deforestation. The results highlight th e important role of community forest management in reducing deforestation, but emphasizes that there is an optimal and strategic way that communities should be awarded the area of forest. W hen awarded properly community forest management can assist

PAGE 10

10 in redu cing poverty through economic benefits and reduce deforestation through sustainable practices. The evidence from this analysis underscore the role severity of poverty has on deforestation and reveals a way that policy makers can make their community forest management programs more effective.

PAGE 11

11 CHAPTER 1 INTRODUCTION The Philippines has been plagued by high inequality, high deforestation and low poverty reductions for the past decades (Asian Development Bank 2007). It is one of the top ten countries with a high deforestation rate and has fallen behind its Asian neighbors in economic development and growth (ESCAP 2006) China and Vietnam are the top two countries in the As ian and Pacific regions, where forest area has increa sed considerably Since the early 2000s, China and Vietnam absolute poverty rates have been lower than the Philippines. Yet the two countries started at higher levels of poverty than the Philippines (FAO 2007; Asian Development Bank 2007). Efforts to reduce poverty cannot be achieve without taking into the account the environment since changes in the ecosystems1 play a role in shaping lives. The primary feature of an ecosystem is populations and environments can interact as a functional unit. When ecosys tems are d amaged they cannot fully provide their services A s a result the damaged ecosystem may not be able to supply the necessary services to people and animals (UNDP 2008). The degradation of the environment tends to affect the poor populations sinc e they are more vulnerable to negative shocks in the ecosystem due to the lack of income or asset s available for them to cope (Nunan et al 2002). Problem Statement Sustainability of the environment and poverty reduction are major concerns of the internati onal community. Over 70 million people live in remote areas of closed tropical forests and another 735 million people live in or near tropical forests. Their well being is strongly 1 Ecosystem is defined as interacting complex sets of plants, animals and microorganism communities together with the nonliving environment.

PAGE 12

12 related to the environment in terms of health, income, security, and decent housing (World Bank 2008). A lack of consensus on the environment and the extent of linkages between deforestation and poverty create challenges for policy makers to plan effectively and provide appropriate regulations. Understanding how to address the poverty and environmental relationship depends on being able to identif y and interpret these linkages properly. Forests play vital ecological roles: such as stabilizing soils, conserving nutrients, and regulating climate Unfortunately, deforestation and poo r land use practices weaken these support services at a rapid rate. Forests also have important environmental components including carbon sequestration which helps the reduction of greenhouse gas emissions (USAID 2008). When forests grow, carbon is removed from the atmosphere and absorbed in wood, leaves and soil. Because forests can absorb and store carbon over an extended period of time, they are considered carbon sinks This carbon remains stored in the forest ecosystem but can be released into the atmosphere when forests are burned. Thu s forest can influence climate change by affecting the amount of carbon dioxide in the atmosphere (Puhe and Ulrich 2001). Despite the forests numerous attributions, deforestation continu es to be five percent a decade and contributes to twenty percent of annual global CO2 emissions the second largest source after the energy sector (World Bank 2008; USAID 2008). Deforestation can cause changes in rainfall and disrupt the trade winds, whic h can create stronger upswellings of hot air. Rainfalls tend to be more powerful and there is an increased likelihood of flash flood because the shrinking forest lessens the landscapes capacity to intercept, retain and transport precipitation (Spray and M oran 2006). Objectives and Hypotheses The primary objective of my research is to present an empirical analysis of the relationship between poverty and deforestation using regional level data from the Philippines. I will be examining the impact of poverty on deforestation, while accounting for environmental

PAGE 13

13 improvement and the fact that poverty and the environment evolve together, in addition to analyzing whether community based forest management agreements have an impact on poverty and deforestation. Thi s will be achieved through the following specific objectives: Estimate the effect of the severity of poverty on deforestation Estimate the effect of income distribution on deforestation Examine the influence of community forest agreements on the incidence of poverty and deforestation Hypothesis 1: Areas with an increased severity of poverty will have higher deforestation. The poor populations overexploit natural resources since they rely heavily on its components to subsist. The more environmental degradat ion that takes places implies less natural resources the poor population has available as input s for their household (Dasgupta, Deichmann, Mesiner, and Wheeler 2005; Bhattacharya 2007 ). The less monetary resource households have make s them more susceptibl e to the food insecurity, natural disasters and disease s This implies that areas with severe depth of poverty will have higher deforestation taking place due to the vulnerability of poorer populations. Hypothesis 2 : Areas with higher income inequality wil l have higher deforestation rates. Greater inequalities will lead to higher deforestation rates because the level of deforestation being done by the non poor population is not counteracted by the level of deforestation of the poor population (Boyce 1994). The wealthy populations have more capacity to overexploit natural resources due to better capital and increased technology. In addition, the wealthy segment of the population may be able to access areas that are off limits to poor populations due to their social standing and political connections. Hypothesis 3: Areas with a higher number of community based forest management agreements will have lower levels of deforestation and less severity of poverty. The Philippines

PAGE 14

14 Department of Natural Resources enter s into a legal agreement with the local community to manage specified amounts of forest land. The local community is able to utilize their hectares of forest by cultivating and selling forest products. This allows the community to earn income without engaging in destructive activities. Thus, the community forest agreements should stop the negative reinforcing cycle of the poverty-environment nexus. Since community are able to earn an economic profit without harming the forest. Therefore, participating in a community forest based agreement should help alleviate poverty due to sustainable forest practices and an increase in economic resources. Thesis Outline This study utilizes panel data from the Philippines to estimate simultaneous equations in which defore station and poverty are treated as endogenous variables that evolve together. This approach allows the poverty -environment al nexus hypothesis to be tested using the Environmental Kuznets Curve (EKC) conceptual framework. The remainder of the thesis is orga nized as follows. Chapter 2 introduces the study site and the state of poverty and deforestation in the Philippines. Chapter 3 presents the literature review with a focus on Environmental Kuznets Curve and the poverty-environment relationship. The latter p art of chapter reviews the ecological and social contributions of forests. Chapter 4 discusses the analytical framework used to estimate the simultaneous equations and presents several methods to test the validity of instrument al variables Chapter 5 provi des the empirical model and measurement of the varia ble used. Chapter 6 presents the findings of the study where descriptive statistics and results are discussed. Chapter 7 presents the conclusions, extension, as well as limitations

PAGE 15

15 CHAPTER 2 BACKGROUND In this chapter, background information on the Philippines is presented. This chapter contains a brief overview of the general features of the Philippines. A synopsis of the Philippines deforestation and poverty situation is outlined. This is f ollowed by a review of the Philippines s forest history and its two major national policies : the Philippine Strategy for Sustainable Development program and the Master Plan for Forestry Development T he specific objectives and programs under the PSSD and MPFD are highlighted The final section of the chapter focuses on forests on a global scale. General Features of the Philippines The Philippine s which is official known as the Republic of the Philippines is an archipelago of 7,107 islands with a total land area of 29 9,764 square km It is a country in Southeast Asia with Manila as its capital city. It is geographically located at 4o23' 21o25' Latitude North and 112oE 127o Longitude East. The island is surrounded by three bodies of water: the Philippine Sea on the east, the South China Sea on the west and the Celebes Sea o n the south (See map in Figure A 1) It has a tropical marine climate with three distinct seasons: wet (June to October), dry (November to February) and hot (March to May). The country is divided into three island groups : Luzon (Regions I V, NCR, CAR) Visayas (Regions VI VIII), and Mindanao (Regions IX XIII, ARMM). It has 17 regions, 81 provinces, 136 cities, and 1,494 municipalities (Republic of the Philippines 2008; Republic of the Philippi nes 2009). In 1898, Philippines declared its independence from Spain. Spain however ceded the Philippines to the United States in the Treaty of Paris along with several other countries. In 1946, the Philippines was able to regained its independence from th e United States The Philippines is the third largest Englis h speaking country in the world (Republic of the Philippines 2009).

PAGE 16

16 Deforestation and Poverty in the Philippines The Philippines is one of the top ten countries with a high deforestation rate. Du ring the period of 20002005, the Philippines s forest decreased at a rate of 2.1 percent per year which is an average loss of 157,000 hectares annually ( FAO 2007 ). Deforestation in the Philippines is reported to have started in 1521, when the Spanish regi me invaded the islands. By 1900, the total forest area had been reduced to 70 percent (Kummer 1996). In 2005, the total forest cover of the Philippines was recorded at 7.168 million hectares or 24 percent of the countrys total land area (FAO 2007). The fo rests play an important role in the lives of the Philippine people since they rely on it for shelter, fuel, food and income (Van Den Top 2003). In 2006, the number of poor people in the Philippines increased by sixteen percent from 2003, this translates in to an increase of 3.8 million people and one out of four families being deemed poor. The cost of achieving the minimum food requirement to satisfy nutritional recommendations rose by twenty three percent from 2003 to 2006. This resulted in 12.2 million peo ple not being able to meet basic food requirements, which is an increase of fourteen percent from 2003. (Philippines 2008). Poverty reduction has not had much success in the Philippines due to the high level of inequality that has continued to worsen. The Gini coefficient in 2006 was over .45, which is among the highest in Southeast Asia (Asian Development Bank 2007). The bottom twenty percent received only four percent of total income and a ccount s for five percent of total expenditure (Philippines 2006). Philippines Forest History The first forestry department was created in June of 1863, under the Spanish regime. Its sole function was to collect data and oversee the proper utilization of the forest. Instead, the government allowed free use of timber and by 1898 when the United States came into rule forest cover has decrease to seventy percent of total land area. The first forest legislation enacted by the

PAGE 17

17 United State s on June 27, 1900 was a means to r egulate cutting and transportation of forest resources A charge was imposed on forest products from public forest for commercial use; while timber cut for personal use remained free but had to be cut under a license. On May 7, 1904, U nited S tates Congress enacted the Forest Act and it remained the basis of all forestry operations until 1975. Rights to forest utilization were granted to a select few individuals and American logging companies started entering the country introducing modern logging techniques. During the American rule, the Philippines became a major exporter of logs and timber and generated substantial revenues. Despite economic gains, reforestation projects had to be implemented in 1916 because of the rapid los s of forest resources; but the reforestation projects were largely unsuccessful and d eforestation continued to be widespread. In 1941, the American rule was d isrupted by the outbreak of WWII and the Japanese imperial forces occupied the Philippines ( Republic of Philippines 2009). The two major forestry policies during the J apanese regime a re: (i) Act No. 13, which prohibited cutting trees near a spring that was used for irrigation and water purposes; (ii) Act No. 42, which limited the maximum amount of hectares that was allowed to be devoted to pastures and reforestation to 2,000 hectares. Aft er the Philippines was able to gain independence from the United States the Bureau of Forestry expanded its responsibilities and created additional divisions in the central office and field. In the first year of independence there were five divisions and forty -four district offices established. The Forestr y Bureau underwent many changes, which included integrating and merging with other departments to better incorporate the need s of the people and the environment. Presently, the Forestry Bureau is now called the Forest Management Bureau (FMB) and is attached to the Department of Environment and Natural Resources (DENR ). The FMB supports the DENR by overseeing all forestry activities (Republic of the Philippines 2009).

PAGE 18

18 National Forest Policy The Philippi ne Constitution of 1987 cl early identified the role of government regarding the environment and natural resources assuming full responsibility for all lands of public domain, water, minerals, coals, petroleum, potential energy, fisheries, forest and etc. T he Philippine government has complete control and supervision of exploration, development and utilization of any natural resource. The Philippines Congress is the responsible authority that determines the specific limits of forest lands and national parks, in addition to instituting logging bans. In order to pursue the provisions of the Constitution, the Philippine Strategy for Sustainable Development (PSSD) and the Master Plan for Forestry Development (MPFD) were implemented. Philippine Strategy for Susta inable Development (PSSD) The PSSD provides the framework for promoting economic growth without engaging in destructive environmental practices. It has the following core strategies and activities: 1 Adoption of the Sustainable Forest Management Act 2 Defining forestland boundaries 3 Conducts environmental impact assessment prior to project implementation 4 Establishes permanent forest estates including an inventory of all remaining natural forest 5 Enhance the administrative capacity of the Department of Environment and Natural Resources though implementation 6 Adoption of Community Based Forest Management as the national strategy to ensure sustainable development 7 Ensures protection of biological resources and overall environmental quality Master Plan for Forestry Dev elopment (MPFD) The Philippine s g overnment with the assistance of Asian Development Bank (ADB) and the Finnish International Development Agency (FINNIDA) formulated a 25 -year Master Plan for Forestry Development (MPFD). This plan was formulated to address the concern of forest

PAGE 19

19 degradation. The MPFD consist of three main umbrella programs that are supported through 15 major programs. The main subprograms are indicated. Man and the Environment a ) People Oriented Forestry : Community Based Forest Management progr ams, Integrated Social Forestry b ) Soil Conservation and Watershed Management c ) Integrated Protected Area System (IPAS) and Biodiversity Conservation: buffer zone management, protected area management and biodiversity conservation d ) Urban Forestry Policy and l egislative reform, public information and integration of forest protection. Forest Management and Products Development Programs a ) Management of Natural Dipterocarp Forests: establishing permanent dipterocarp forest estates and enhancing dipterocarps forest p roductivity b ) Management of Mangrove, Pines and other Natural Forests: management of pines, mangrove and other natural forests c ) Forest Plantation and Tree Farms: reforestation, forest plantation development. d ) Wood Based Industries: provision of new technologie s, development of community based wood processing industries. e ) Non -Wood Forest Based Industries: development of medicinal plant, gum, resin and essential oil industries Institutional Development Programs a ) Policy and Legislation: develop the policy and legal framework and to update the laws and regulations b ) Organization, Human Resources, Infrastructures and Facilities: improve mobility by transforming the department into an extension and development organization c ) Research and Development: increase the quality an d number of staff conducting research and increase funding d ) Education, Training and Extension: to increase the knowledge and skills of field technicians through training.

PAGE 20

20 e ) Monitoring and Evaluation: monitor and evaluate programs by measuring the progress and impacts on clientele and/or intended beneficiaries (Philippines 2003). L ocal people and policymakers have different perceptions of the forest, which often leads to miscommunication. The PSSD and MPFD are important because the frameworks assist policymaker s in protecting the environment, while trying to meet the needs of the Philippine peo ple. Programs that focus on people and the environment such as the Community Based Forest Management Agreements (CBFMA), promote integration of sustainable forest manageme nt practices while allowing people to benefi t economically from the forest services. In essence, the goal of the CBFMA is to lessen deforestation while allowing people to earn income to reduce their poverty status. The income earned will help raise their s tandard of living and the environment is not harmed in the process. The principles in each framework try to ensure that present needs are being met without compromising the ability of future generations to meet their own needs Forests on a Global Scale Fo rests cover approximately thirty percent of Earths land, which is about 3,952 million hectares (FAO 2007) However, forests are not equally distributed around the world; ten countries account for two thirds of total forest area. While deforestation contin ues at an alarmingly rate, activities such as forest planting, landscape restoration and natural expansion of forests have reduced the net loss of forest area (FAO 2006). There are different types of forests worldwide. An estimated fifty three percent of t otal forest areas ar e modified natural forests; these are forests that have had clear indicators of human interactions but have regenerated naturally comprising of native species (Figure 2 1) T hirty six percent of the worlds forest areas are classified a s primary forest s ; which are forest s of native species where there has been no human or ecological disturbance Seven percent are semi -natural forest ; which are forests made

PAGE 21

21 up of native species that have been planted or have had assistance in the natural regeneration process. F orest plantations which are planted forests consisting of introduced species account for four percent of total forest area. Figure 2 1 Characteristics of the Worlds Forest s Source: FAO 2006. Forest worldwide are managed and d esignated for a combination of functions Thirty four percent of forests are primarily used for wood production and non wood forest products ; they provide raw materials and offer a wide arrange of economic benefits such as income from employment in additio n to monetary value from production of goods and services. Eleven percent of total forest areas have been set aside for c onservation of biological diversity ; they help maintain natural habitats and play a role in climate change mitigation. N ine percent of forests around the world are allocated to protect land and water resources; their protective functions range from soil and water conservation to desertification control (Figure 2 2 ). Yet, d eforestation2 is occurring at thirteen million hectares per year w hich is approximately twenty thousand hectares per day. In the period 20002005, net global change in forest area decreased by 7.3 2 Deforestation defined as the loss of forest from logging, shifting cultivation, gr azing and other destructive activities

PAGE 22

22 million hectares per year. Although, forests are heavily exploited, efforts are being made to use and manage them more sustai nably (FAO 2006) Figure 2 2. Global d esignated f unctions of Forest s Source: FAO 2006.

PAGE 23

23 CHAPTER 3 LITERATURE REVIEW The purpose of this chapter is to review prior research pertaining to the poverty environment nexus literature with a focus on the Env ironmental Kuznets Curve. Although, there are several publications tackling similar issues, it remains unclear what impact and relationship environmental degradation and poverty have on each other. The first section presents the Environmental Kuznets Curve (EKC) conceptual framework and the major studies that contributed to the literature. This is followed by the EKCs alternative scenarios and limitations T he next section discusses the poverty and environment relationship. Th e chapter ends with a synopsis of the importance of forest and community interactions. Environmental Kuznets Curve The literature on deforestation and economic development focuses mainly on the Environmental Kuznets Curve (EKC) The EKC is named after Simon Kuznets an economist and statistician who won the 1971 Nobel Prize for Economics. Simon Kuznets proposed that economic inequality increases over tim e while a country is developing; a fter the country reaches a critical average income, inequality in the country begins to lessen (Kuz nets 1955) A graphical representation of Simon Kuznetss theory is the Kuznets curve, from which the EKC is derived. The EKC hypothesis is an inverted-U relationship between environmental degradation1 indicators and income per capita ( Figure 3 1 ). The in tuition behind the EKC is in the first stages of development, as a country moves away from agriculture to industrialization; pollution rises rapidly (Stern 2004) The countrys priority becomes investing in physical capital as a main mechanism of economic growth. People are more interested in jobs and income than clean air 1 Environmental degradation is defined as damage to the natural environment. It can refer to damage to land, water, and air; including loss of biodiversity or loss of natural resources in an area.

PAGE 24

24 and water. The rapid growth results in greater use of natural resources and more emission s of pollutants. T herefore in the initial phases of economic growth, pollutants and environmental degradation increase In later stages of development once a countrys income per capita starts rising, its priorities change and people start to value the environment more. Regulatory structures are put into place to reduce pollutions and the country star ts becoming more environmentally clean. Thus, t he trend of environmental degradation starts reversing and improves as countries per capita income increases (Deacon and Norman 2004). The EKC therefore implies that in order for countries to attain a decent environment they must become rich (Gallet and List 1999). Figure 3 1. Environmental Kuznets Curve The reason the EKC has an inverted -U shape is because t here are three main channels in which income growth impacts environmental quality T hey are the f ollowing: scale effect s composite effect s and technological change. The scale effect is thought to have a negative impact on the environment because as a country increases its output, more natural resources are used in the process. This leads to an increa se in pollution by -product s and waste. The composite effect somewhat offsets the scale effect because environmental degradation will fall as the structure of the economy changes. When the economy changes from rural to urban, the industry mix is

PAGE 25

25 changed to service oriented industries rather than energy intensive reducing pollution The technological change occurs as the obsolete technology and capital are replaced with newer and upgraded technology resulting in cleaner activities. The study of EKC started with Shafik and Bandyopadhyay (1992), they examined the empirical relationship between per capita income, air pollution, deforestation, access to clean water and production of solid wastes. Their findings in most cases were supportive of EKC, but not supportive of a common threshold of per capita income after which a decline in environmental degradation would be observed. Another important study was done by Grossman and Kruger (1995) who estimated the likely effects of increased income on air pollution. In this study, the authors identified a suggested per capita income threshold for countries in regards to airborne sulfur dioxide and smoke pollutants to improved air quality. Panayotou (1993) estimated the EKC using cro ss sectional data and estimated sulfu r dioxide (SO2) nitrogen o xide (NOx), suspended particle m atter (SPM), and deforestation SO2, NOx and SPM were measured on a national level and deforestation was measured as the mean annual rate of deforestation. All the pollutants fit the inverted -U cur ve and the turn ing point for SO2 was $ 3,000 per capita, for NOx it wa s around $5,500, for SPM it was around $ 4,500 and for deforestation it wa s around $ 823 per capita (Stern, Common and Barbier 1996). The Panayotou (1993) study has the lowest turning po int compared to all the major studies but is similar in range to those reported by Grossman and Kruger and Shafik and Bandyopadhyay (Stern 2004) Selden and Song (1994) estimated the EKC using longitudina l data from 22 developed countries and 8 developing countries for four pollutants: sulfur dioxide (SO2), n itrogen oxide (NOx), s uspended particle m at ter (SPM), carbon m onoxide (CO ). Their study showed that the

PAGE 26

26 turning point for emission s were likely to be higher than ambient concentrations (Stern 2004). Al l the variables in their model were significant except CO and their turning points are relatively higher than most studies. Selden and Song (1994) suggest that in countries with low population densities, there is less pressure to adopt strict environmental standards and therefore emissions will be higher (Stern, Common and Barbier 1996). Holtz -Eakins and Selden (1995) use d panel data to estimate the relationship b etween carbon dioxide (CO2) and GDP. In their study, two different models were estimated : a qu adratic and a natural logarithm model Their study suggested that there is a diminishin g marginal propensity to emit CO 2 as countries develop However, t he turning point for the quadratic model is around $35,428 per capita and the natural logarithm model turning point was above $ 8 million. Both of these results exceeded the maximum GDP level of countries in their sample. Torras and Boyce (1998) added to the EKC literature by taking a political economy2 approach. Their argument includes income distributio n and power as important determinant s of environmental degradation. The study shows empirically that more equitable distributions result in improved environmental quality. Different S cenarios for the EKC There are some critics that believe that the EKC h ypothesis will actually be a horizontal line deviating from the maximum conventional EKC ( Figure 3 2) (Dasgupta, Laplante, Wang and Wheeler 2002). This horizontal line occurs because as developed countries start raising their environmental standards, a hi gher cost is imposed on polluters. These environmental costs tend to be fairly higher in the developed counties than the developing countries. As globalization emerges, the polluters are able to transfer their capital and technology and relocate to the low 2 the allocation of scarce resources, not only among competing ends but also among competing individuals, groups, and classes ( Boyce 2001).

PAGE 27

27 regulation countries As d eveloped count ries s ee capital outflow s start to increase the governments start relaxing their regulations (Dinda 2004). G lobalization could promote a race to the bottom in which competition increases investment and jobs ( Dasgu pta, Laplante, Wang and Wheeler 2002 ). However, as the race to the bottom increases, the EKC should flatten and rise toward a higher level of existing pollution (Dinda 2004). Another alternative scenario of the EKC hypothesis is even if pollutants are red uced while income increases, society may be creating new toxics that remain unregulated ( Dasgupta, Laplante, Wang and Wheeler 2002). The traditional pollutants may have had an inverted -U shape but the new toxics replacing them may not (Refer to Figure 3 2) As older pollutants lessen due to regulation, new potential toxics may be increasing and the overall environmental impact is not reduced (Stern 2004). Figure 3 2 Different s cenarios of EKC Limitations and Criticism of the EKC T he EKC hypothesis has limitations that are worth mentioning. First, it appears to explain the inverted -U relationship for air pollutants but is not an acc urate model for all pollutants. Second, there is no universal agreement in the literature on the income level at which

PAGE 28

28 envir onmental degradation starts decreasing. This indicates that the turning point for countries differ and may be unachievable for certain countries. For example if the turning point for a poor country is at an exceedingly high level of per capita income th e environmental benefit of economic gro wth will not be realized. Third studies have found when alternative variables are included in the EKC model, the variables coefficients become i nsignificant and no longer follow an i nverted -U pattern (Stern 2004; Ga llet and List 1999). The empirical results of the EKC literature is under criticism due to the fact that most scholars are skeptical about the quality of data used in the EKC studies (Stern 2004). Most studies have estimated the EKC with cross sectional d ata from different countries and regions because of lack of data availability. This leads to data compatibility issues because of unreliable data collection and the differences in measurements. In addition, the lack of data can lead to sample selection bia s. The data from these cross sectional studies are usually overrepresented by developed countries and highly polluted areas because those are the countries that have the data available and are normally the areas of interest (Shafik 1994). T hus, econometr ic evidence of the EKC model tends to be very inadequate. Researchers pay little attention to the statistical properties of the data being used and assume that if the regression coefficients are individually or jointly significant then the EKC relationship e xists (Millimet, List and Stengos 2003). Despite, the lack of evidence between income and environmental degradation in previous studies, this study uses the Environmental Kuznets Curve framework. T he earlier studies used techniques that may have cause d es timation errors such as using cross sectional data based on reduced form equation making data quality and comparability a serious issue (Stern, Common and Barbier 1996) Since environmental degradation is likely to have feedback effect s on income grow th a more appropriate method to use within the EKC framework is a simultaneous mo del

PAGE 29

29 equation. T hus, the EKC conceptual framework is used in this study because essential ly the EKC allows simultaneous estimation s of feedback effects in the environment. Povert y and the Environment Poverty is a complex and multidimensional concept therefore i t is important when defining poverty, to specify who is classified as poor and the means of measurement that is being used. The Philippines Social Reform and Poverty Alleviation Act define the poor population as: Individuals and families whose income falls below the poverty threshold as defined by the government and/or those that cannot afford in a sustained manner to provide their basic needs of food, health, education housing and other amenities of life ( Philippines Republic Act 8425 1997). Since the early 1970s, it has been agreed that poverty and environmental degradation are linked (United Nations Development Programme and European Commission 1998). The poverty-en vironment al nexus implies that environmental degradation reinforces poverty because the reduction in availability of natural resources make the poor vulnerable to natural disasters; and poverty forces people to degraded their environment through over exploi tation in the quest for subsistence living. According to the vast literature, a frequent indicator used to evaluate poverty and environmental degradation in developing countries is the deforestation rate. The phenomenon of the poor in developing countrie s relying heavily on local natural resources such as forests, pastures and water is well documented (Jodha 2000). The poor have been identified as agents of environmental degradation, because of the dominant view that the poor overexploit natural resource s for their subsistence. The poor are portrayed as trying to meet their short term needs, which prevents them from making investments to replace the natural resources. The environmental degradation cause during overexploit ation further impoverishes them le aving

PAGE 30

30 them more vulnerable (Dasgupta, Deichmann, Mesiner, and Wheeler 2005). The change in availability of natural resources due to the overexploitation of natural resources affects poverty because it implies a shrinking input base for poor households and thus increases the severity of poverty. Environmental degradation reduces the ability of the poor to generate income and usually the poorer households derive a large part of their income from environmental resources (Bhattacharya 2007). The dominant view in the literature is that poverty causes environmental degradation; however, there is some con tradicting empirical evidence. One study done in Kenya for a duration of 60 year s examined the interaction between the Kenyan people and the environment The aut hors concluded that despite being poor the communities manage d to use their resourc e efficiently. The authors noted in their analysis, that Kenya is not a unique experience and the communities method of management for resources is replicated in other plac es (Tiffen, Mortimore and Gichuki 1994). Another study that contradicts the dominant view, is a study done in two northwestern mountainous regions in Nicaragua. It concluded that the non poor populations were the immediate agent s of environmental degradati on since they primarily engaged in environmental damaging practices that depleted the environment (Ravnborg 2003). There is also evidence from a study that used cross sectional data from Latin America that suggested that the non poor and poor populations c ause environmental degradation. The authors concluded that although the poor lack resources to invest in sustainable forest management the non poor lacked the incentives to practice sustainable resource management which result ed in overexploitation of nat ural resources (Swinton, Escobar and Reardon 2003). Importance of Forests For centuries forests have provided not only products and services but an identity for local populations (See Figure 3 3) For indigenous populations the forest is central not only f or their

PAGE 31

31 survival but it also has spiritual and cultural significance. The forest landscapes have symbolic meaning and provide historical associations for the local populations. It also provided protection from natural disasters such as landslides and flooding and acts as a safety net for subsistence livelihoods (United Nations 2000). Forest s more prominent global value is biodiversity and carbon sequestration The Philippines is one of the seventeen mega diverse3 countries in the world. Unfortunately its also considered a biodiversity hotspot4 placing it among the top priority hotspots for global conservation (Conservation International 2009). The forests provide secure habitats for native species of both plants and animals, which provide food and medicine for the local population. When deforestation occurs, game species get eliminated and medicinal plants are scarce. These changes affect plant pathogens and herbivorous insects that vector them creating opportunities for invasive species and exotic pathogens to devastate the forest. These disruptions can cause severe hardship and social disruptions to forest dependent populations. (ESCAP 2006; Laurence and Peres 2006). Forests currently store approximately 800 billion tons of carbon and the Intergovernmenta l Panel on Climate Change (IPCC) suggest ed that up to 87 billion tons of carbon can be sequestered in the worlds forests by 2050 (Sohngen and Mendelsohn 2003) Carbon storage (sequestration) occurs in forests and soils primarily through the natural proces s of photosynthesis (Freer -Smith, Broadmeadow and Lynch 2007). Atmospheric carbon dioxide (CO2) is taken up through tiny openings in leaves and incorporated as carbon into the woody biomass of trees and agricultural crops. Plants absorb the CO2 from the ai r during photosynthesis and ultimately metabolize and store carbon as tissue (biomass) or transfer it to the soil. Changing the way that 3 About 70 percent of the worlds total diversity in flora and fauna are in these countries 4 An area that contains 0.5 percent of the worlds species and flora/fauna unique to a particular geographic location and has lost at least 70% of its primary vegetation

PAGE 32

32 forests are managed can increase the amount of carbon stored by plants and soil and can hold carbon intact for decades or centuries (Lal 2004). The large amounts of carbon forests store in their biomass, gets release back into the atmosphere through deforestation. Climate Change and Deforestation Climate change is emerging as one of the greatest environmental challenges of the twenty first century (FAO 2009) The influence of climate change on weather related natural disaster is acknowledged and supported by the World Meteorological Organization. Different natural disasters affect people and the environment in various ways. The estimated proportion of economic loss in developing countries from natural disasters can be as high as 13 percent of GDP and this estimate is for flooding al one, while developed countries experience a loss of 2 percent of GDP (ESCAP 2006). Atmospheric concentrations of carbon dioxide (CO2) have risen by thirty five percent since the Industrial Revolution. The source of the increase mainly comes from the burning of fossil fuels and deforestation. Impacts of climate change are likely to vary regionally but the aggregated effects will impose a net cost which will increase over time as global temperature rise (IPPC 2007b). Forest s can be used as a source of climate change mitigation through four major ways: reforestation5, sequestration, conservation and su bstitution. Reforestation helps restore degraded areas and counteracts deforestation by creating new carbon sinks Sustainable forest management a ctivities such as increasing rotation length and other appropriate silviculture6 techniques can also enhance t he uptake of carbon dioxide since substantial amounts of carbon is stored in soil ( Bhatti, Lal and Price 2006) Trees remove carbon dioxide from the atmosphere thro ugh photosynthesis and t he carbon is stored in their biomass. 5 Afforestation which is the establishment of new forest is also a source of climate migration 6 B iological and economic theory and practice use to control forest establishment, composition, and growth to satisfy land management objectives

PAGE 33

33 Younger trees sequester carbo n at high rates, while older trees sequester carbon at slower rates allowing the carbon balance to reach a steady state. However, w hen forests are cleared the carbon is released into the atmosphere and acts as a source of greenhouse gases (Freer -Smith, Bro admeadow and Lynch 2007) Conservation protects the carbon that has already been sequestered by using existing forest s through increasing landscape level carbon density (IPPC 2007b ). In addition, s ubstituting off site wood products instead of using fossil fuels can lower carbon emission because wood products store carbon for the duration of their lifetime (IPPC 2007c). Figure 3 3 Forest c ontributions Community Management and Fores t Most forests are legally owned by the government yet are used frequentl y by people from nearby communities. When local populations do not own the forest, benefits derived from its use through unhealthy practices is not a major concern (Mayers and Bass 2004) F orest management requires a shift from a destructive output drive n approach to a focus on the sustainability of the environment. Sustainable forest management is the process of managing forest to achieve desired forest productivity without negative effects on the physical and social

PAGE 34

34 environment ( Philippines 2003). Altho ugh forest products are important, the health and well being of the land should be main objectives. F or the land to meet the needs of the community and still maintain a healthy ecosystem forestry management is best done at the community level. At this lev el, there is a direct vested interest in the overall well being of the community and an increased awareness of the ecosystem beyond the community level. In order for the community to realize the full benefits of a community forest management program, f ores t products and servic es must be used and produced with in the limits of the land tenured for its beneficiaries This is to exclude noncontributing beneficiaries from consuming, thus limiting the free rider7 dilemma (Kimmins 1997) If this mechanism is not p ut into place it i s unlikely others will inves t in maintenan ce and/or protection of the forest. Without excludability the forest resources will be easy to deplete and difficult to protect. Attempts are commonly made by the government and their agencies to regulate forest areas. However, g overnments in developing countries rarely have the personnel or funds to properly enforce regulations to protect their natural resources (Lee, Field and Burch 1990) Allowing communities to assume authority over acres of f orest land is now becoming a viable natural resource management strategy for governments. L ocal communities that live within and/or nearby the forest can greatly determine the outcome of a natural resource management strategy because they can modify the co nsumption of natural resources through local institutions8(Mayers and Bass 2006; Gibson, McKean and Ostrom 2000) 7 A person who consumes without paying or shoulders a less than fair share of cost; consume more than their fair share of a resource. 8 Direct or indirect rules and/or behaviors exhibited by local communities to guide pattern of activities

PAGE 35

35 CHAPTER 4 ANALYTICAL FRAMEWORK O rdinary Least Squares Under the Gauss-Markov theorem the Ordinary Least Square (OLS) is the best linear unbiased estimator (BLUE). This means, OLS estimators are unbiased and have the smallest variance in the class of all linear estimators (Gujarati 2003). However, in or der for OLS to be BLUE the following classical assumptions must be met. I. The regression mode l is linear in parameters y = 0 1x1 2x2 kxk (4 1) II. Random s ampling of n observations {(xi1, xi2,..xik,yi): i = 1, 2,.n} (4 2) III. No perfect collinearly none of the independent variables are constant and there are no exact linear relationships IV Zero c onditional m ean error term has an expected value of zero given any values of the independent variables1. E(u 1,x2,..xk) = 0 (4 3) V Homoskedasticity the variance of the error term, conditional on any explanatory variable is constant2. Var ( u 1.xk2 (4 4) VI Normality the population error ter m is independent of the explana tory variables and 2 E ( u (4 5) Var ( u 1.xk2 (4 5) u ~ Normal (0, 2) (4 6 ) 1 Assumption 1 4 allows OLS to be unbiased. The key assumption for unbiasedness is Assumption 4 2 Assumption 5 plays no role in showing unbiasedness. It implies that OLS has efficiency properties and simplifies the variance calculations.

PAGE 36

36 Wh en an assumption is violated, we have to determine which property no longer holds In the case of estimating an equation involving endogenous variables, the assumption that states that al l explanatory variables must be uncorrelated with the error term is violated; this caused OLS to produce a simultaneous bias. This bias refers to the fact that the expected values of s are not equal to the true s The OLS estimates of the coefficients ar e not centered around the true population values of the parameters being estimated. In addition, the coefficients are inconsistent because the expected values of s do not approach the true even as the sample size increases. One way to help mitigate in consistency inherent in the OLS estimation is to use an alternative estimation method called Two Stage Least Squares (2SLS) The 2SLS is a common technique based on the use of instrumental variables (IV). An instrumental variable denoted as z replaces an endogenous variable and must satisfy two conditions: (i) be uncorrelated with the error term and (ii) be correlated with the endogenous variable in question. An instrument that satisfies these conditions would avoid violati ng assumption IV and therefore provid e consistent estimat ions In Figure 4 1, the instrumental variable is shown as being correlated to the endogenous variable but at the same time having no effect on the dependent variable. Th e use of the instrumental variable approach allows the assump tions to hold and not be violated. Figure 4 1 Instrumental Variable a pproach. This shows z has no effect on the dependent variable except through the endogenous variable. Dependent Variable z Endogenous Variable

PAGE 37

37 Two Stage Least Square s The two stage least squares utilizes a structur al and reduced form equation to estimate a simultaneous model. A structural equation is a standard linear model that is expressed in terms of endogenous and exogenous variables. A reduced form equation is one that expresses an endogenous variable in terms of exogenous variables and the error (Gujarati 2003). Equation 4 7 is known as the structural equation and Equati on 4 8 is the reduced form equation. y 1 0 1y2 2x1 + u1 (4 7 ) y2 0 1z1 + z2 2 (4 8 ) The Two Stage Least Square is a two step procedure of OLS. Stage One : Estimate the reduce d form equation by OLS and obtain its predicted values. Stage Two: Replace the endogeno us v ariable by its predicted value and estimate the equation by OLS (Wooldridge 2000). To estimate a model by using the T wo S tage L east S quares method the equation has to be identified. The two requirements for identification is the order and rank condit ion. The order condition states that the first equation in a two equation simultaneous model is identified if the second equation contains at least one exogenous variable that is e xcluded from the first equation. The rank condition requirement for identifi cation is at least one o f the exogenous variables excluded from the first equation must have a nonzero population coefficient in the second equation (Wooldridge 2006). Validity of 2SLS Specification Test The 2SLS approach uses instrumental variables to rep lace the endogenous variables in simultaneous equations. However, if there is no endogeneity both OLS and 2SLS are consistent .3 3

PAGE 38

38 To test for endogeneity, a structural and reduced form equation is needed. The reduced form is estimated by regressing all the exogenous variables including those in the structural equation and all instrumental variables (Equation 4 9 ). The residual obtain from the reduced form regression, is then added to the structural equation along with the endogenous variable and tested for s ignificance (Equation 4 10). Using the t -statistic, i f the residual coefficient is statistically different from zero, then the variable is endogenous (Wooldridge 2006; Greene 2000). y2 0 1z1 + z2 2 (4 9 ) y 1 0 1y2 2x1 1 2 1 = 0 (4 10) The results show that the residual is not significant, with a p value of .28. However, the Environmental Kuznets Curve conceptual framework allows the s everity variable to be treated as endogenous. The full results from the specification test are presented in Chapter 6 (Table 6 1). Overidentification Restrictions Instrumental variables are used in this study to solve the problem of estimating endogenous variables, thus to avoid violating the assumption of the error term being uncorrelated to any explanatory variable. Inst rumental variables must satisfy two requirements: (i) be uncorrelated with the error and (ii) be correlated with the endogenous variable When the re are more instrumental variables used for an endogenous variable the m odel is said to be overidentified. In an overidentified case, the first requirement can be tested to see if there are any correlations with the structural error. In the case, where there are just enough instruments for an endogenous variable the model is said to be just identified. T he overidentifi cation test cannot be use d to test the exogeneity of an instrument in a just identified equation because it involves a correlation between the instrumental variable and an unobserved error. Testing overidentifying restrictions requires estimating the structural equation by 2SLS and obtaining the residual. After obtaining

PAGE 39

39 the residual, run an auxiliary regression on all exogenous varia bles and obtain the R squared. If nR2 exceeds the commonly used threshold of the chi -squared distribution at the 5 % critical value level, then at least one instrumental variable is not exogenous. If the nR2 is smaller than the critical level at the chi s quare distribution 5 % level than the instruments are exogenous to the structural error and are proper to use (Wooldridge 2006; Wooldridge 2000). Weak I nstruments Instrument exogeneity is not the only criteria necessary for an instrumental variable to be v alid. The 2SLS method w as often estimated without regard to its IV strength and the general rule of thumb was an F -statistic > 10 Now testing instrumental variable strength is becoming common practice and part of the empirical analysis An instrument is c onsidered weak if the correlation between the endogenous variable and the excluded instruments are nonzero but small (Baum, Schaffer and Stillman 2007) The concern is that the instrument may not have enough explanatory power to allow for the inference on Weak Instruments have serious consequ ences because it causes the sampling distribution to be nonnormal thus the point estimates, hypothesis tests and confidence intervals are unreliable (Stock Wright and Yogo 2002). One approach of testing weak instr uments is a method proposed by Stock and Yogo (2005). It is based on the CraggDonald statistic. In the presence of heteroskedascity the Wald F statistic is used instead This is because the CraggDonald statistic is no longer valid due to the violation of the independent and identically distributed assumption. Using the CraggDonald statistic when the errors are heteroskedastic could cause one to interpret the instruments as valid, when in fact the instruments are weak. The Stock Yogo weak test essential ly has t w o components: a maximal relative bias and maximal size. The relative maximal bias is the ratio of the bias of the estimator to the bias of OLS. The maximal size is based on the rejection rate that a

PAGE 40

40 researcher is willing to tolerate if the true reje ction rate is the standard 5 % ( Baum, Schaffer and Stillman 2007) Multicollinearity Multicollinearity is when there is a high correlation between two or more independent variable s As long as there is not perfect correlation, multicollinearity does not vi olate any assumptions (Wooldridge 2006). The problem with multicollinearity is a high degree of collinearity cause s the standard errors to be inflated4. The confidence intervals for coefficients tend to be wide and t -statistics tend to be small. Therefore the coefficients will have to be larger in order to be statistically significant; this makes it harder to reject the null hypothesis When estimating by 2SLS, multicollinearity tends to be higher because the produced by the reduced form equation in the first stage has less variation than the y from the structural equation. In addition, the can be highly correlated with the exogenous variables because the first stage includes all the explanatory variables from the structural equations (Gujarti 2003, Gr eene 2000) Although the problem of multicollinearity is not well defined, a common test to detect for multicollinearity is the variance inflator factor (VIF). A general rule of thumb when detecting for multicollinearity is a variable with a VIF value gre ater than 10. The square root of the VIF shows how much larger the standard error is compared to what it would be if that variable was uncorrelated with the other explanatory variables (UCLA 2008) Heteroskedascity Heteroskedascity is when the variance of the error term is not constant Therefore it violates assumption V which is the homoskedascity assumption that the variance of the error, conditional on the explanatory variables is constant H owever, even if the error term is 4 Large standard errors can be caused by other things besides multicollinear ity

PAGE 41

41 heteroskedastic it will not cause bias in the coefficient estimates. Recall that assumption V is not needed to make OLS unbiased or consistent. Although, heteroskedascity does not cause bias in the OLS estimators, it affects the estimators of the variances making them biased. The OL S standard errors are based on the variances and therefore no longer valid for constructing confidence levels and t -statistics. Heteroskedascity is easily corrected using a statistical software package (Wooldridge 2006; Wooldridge 2000). A common test use d to detect heteroskedascity is the Breusch Pagan test. This test assumes that the errors are normally distributed. First, the model is estimated using OLS and the squared OLS residual is obtained. Then using the residuals, run a regression and obtain the R squared. Using the F -statistic, the p -value is computed and if the p-value is significant, the null hypothesis of homoskedascity is rejected. To test for heteroskedascity when using 2SLS, the Breusch -Pagan test is slightly modified. The 2SLS residuals ar e obtained and a regression of the residuals is regressed on all the exogenous variables However, using the usual F -statistics, a joint significant test is used instead. The null hypothesis of homoskedascity remains the same and is rejected if the exogenous variables are jointly significant. This chapter presents several rigorous econometric tests used to correct ly specify and estimate the empirical model using the Two Stage Least Squares estimation method when OLS is not valid. In this study OLS is not valid since the model involves endogenous variables. Using the instrumental variable approach method allows the all the classical assumptions to hold and not be violated. By using these various econometric test s, the study strives to properly model and mi nimize sta tistical and econometric errors that maybe arise when estimating simultaneous e quations. These routine tests allow correct inferences to be made and the best estimators for the 2SLS method to be obtained.

PAGE 42

42 CHAPTER 5 EMPIRICAL MODEL Empirical Mode l Description and Specification The approach adopted in this study combines the conceptual framework of the Environmental Kuznet Curve with the empirical methodology proposed by Bhattacharya (2007). The empirical model is specified by two simultaneous equa tions. The independent variables were determined by socioeconomic and environmental characteristics. As discussed previously, the estimation model is specified based on the assumption that the environment and poverty evolve together in such a way that a ch ange in poverty will cause a change in the environment. For instance, an increase in the severity of poverty is likely to cause an increase in environmental degradation and environment degradation is likely to impact the severity of poverty. A decrease in the severity of poverty should cause environmental improvement and environmental improvement should decrease the severity of poverty. The following simultaneous equations Equation 5 1 and Equation 5 2 were estimated: D = 1 P + it1 X1 11 + u1 (5 1) P = 2 D + it2 X12 13 + u2 (5 2) where i = represents the 17 regions in the Philippines and t = 2000, 2003, 2006. T he endogenous variables are: D = deforestation and P = severity of poverty; X1 through X11 are predetermined variables. X1 = expenditure; X2 = number of CBFMAs site in each region; X3 = hectares of CBFMAs in each region; X4 = CBFMAs beneficiaries; X5 = area reforested; X6 = population density; X7 = gini; X8 = primary enrollment; X9 = rainfall; X10 = unemployment; X11 = watershed area; the instrumental variable used for severity of poverty is X12 = access to water and X13 = makeshift housing. The summary statistics can be found in Table A 2

PAGE 43

43 Measurement and Specification Model Variables Deforestation : Environm ental degradation will be measured by the fore st cover within forest land during the years 2000, 2003, and 2006 in all 17 region s in the Philippines The data will be obtained directly from the Philippine government bodies: The Forest Management Bureau whi ch is part of the Department of Natural Resources and Environment; National Mapping and Resource Information Authority (NAMRIA) which provides geographic and resource information on land use, forestry, agriculture, water resources, the coastal zone and other natural resources and environmental information Poverty l ine : A traditional way of measuring poverty is using the poverty line. A poverty line is the minimum required income for a household to meet basic food and non food requirements. People who are classified as poor are those whose level of income falls below the poverty line. Once a poverty line has been established, a number of summary statistics can be calculated. These measures are the Head Count I ndex, Poverty G ap Index and the Squared Gap Ind ex. Foster Greer Thorbecke show that these poverty measures can be calculated using the following formula. P= (5 3) In this equation yi is the consumption for the ith household, z is the poverty line, N is the the Head Count Index is calculated This index is the proportion of the population for whom consumption is less than the poverty line Although this measure is easily to calculate and int erpret, this index is not sensitive the Poverty Gap Index is calculated This index is defined as the average difference between poor households consumption and the poverty line. This indicato r shows the minimum cost of eliminating poverty

PAGE 44

44 by using transfers to bring the poor populations consumption up to the poverty line. The drawback with this index is that it does not capture differences in the depth of poverty among the t he Squared Poverty Gap is calculated This index is sensitive to the inequality among the poor. It gives more weight to those who are severely poor (World Bank 2006). This study uses the S quared Poverty G ap as the variable to estimate the severity of pove rty Expenditure : The average expenditure variable is used because expenditure tends to fluctuate less than income. Income can vary, especially when considering workers employed in agriculture and tourism. In addition, informal workers and those who are se lf employed tend to get neglected in income based measures. The average expenditure is used to account for consumption for each region (Asian Development Bank 2005). The expected sign of expenditure is negati ve. T he Philippines is a still a developing coun try and has not reached a threshold where according to the EKC framework it would start reversing its environmental degradation. Thus, an increase in income (expenditure) should cause a decrease in forest cover. Access to w ater : Since poverty is a multidi mensional concept, the use of monetary based measurements alone is not adequate. People without access to clean water and proper sanitation can also be regarded as poor. W ater is essential to everyones wellbeing; however it also needs to be adequately cle an in order to minimize health hazards, including waterborne diseases. Poor water and sanitation is the second leading risk factor associate with the total burden of disease. Decline in environmental quality is likely to affect the health of the poor more severely than the rich, since they have little to no access to clean water or a latrine. Improved sanitary conditions may reduce the disease burden on households and local communities ( Lopez, Mathers, Ezzati, Jamison and Murray 2006). The expected sign of this variable is positive since an increase in the

PAGE 45

45 population gaining access to clean water will reduce the burden on the poor and therefore causing severity of poverty to decrease. Income e quality : The overall inequality in a region is an important measu rement of social welfare. The most commonly used measure of inequality is the Gini coefficient. This index is used to determine the extent to which the distribution of welfare deviates from a perfectly equal distribution. The Gini coefficient is derived fr om the Lorenz Curve and summarizes the total amount of inequality. The Lorenz Curve relates the cumulative proportion of total income received against the cumulative proportions of the population starting with the poorest household (Chotikapanic and Griff iths 1999). The Philippines 2006 Lorenz Curve is g raphically represented in Figure 5 1 : Figure 5 1. Philippines 2006 Lorenz Curve If all individuals have the same income or there is total equality, the income distribution line curve is a straight diago nal line, called the line of equality. If there is any inequal ity in income, then the Lorenz C urve falls below the line of equality. Figure 5 1 shows that the bottom twenty percent only receive four percent of the total income, while the top twenty receive s over fifty percent of total income The Gini coefficient is calculate as area A divided by the sum of areas A and B. Th e coefficient varie s between zero and one, where zero reflects perfect equality

PAGE 46

46 and one indicates perfect inequality The expected sign of this variable is negative since a more unequal distribution will result in less environmental protection and more environmental degradation Housing : The right to housing is guaranteed in the United Nations Universal Declaration of Human Rights, in Ar ticle 25: Everyone has the right to a standard of living adequate for the health and well -being of himself and of his family, including housing Housing provides a sense of personal security and space, in addition to protection from the weather. However i nsecure and /or inadequate shelter threatens well being and reduces the overall quality of life. For housing to be adequate it must provide suitable privacy, space and conform to basic infrastructure standards ( Sachar 1996). Makeshift housing refers to l iving quarters where either the roof or walls are made with salvaged and/or improvised m aterials M ost makeshift housing would not be classified as adequate housing. This is because makeshift housing fails to qualify as decent housing and does not raise th e standard of living for its residents. T he expected sign of this variable is positive. As poverty increases the incidence of makeshift housing should also increase. Rainfall: Rainfall is an important climatic element. Especially since in the Philippines, rain distribution varies from one region to another due to the direction of the winds and location of the mountains. The rainfall averages were obtained for all 17 regions Temperature is not included in the model since there is essentially no difference in the mean annual temperature in the Philippines measured at or near sea level (Philippines 2009). The expected sign of this variable is positive since more rainfall should allow the forest to regain it cover at a faster rate. Areas r eforested: In a majo r effort to stop the alarming rate of deforestation, the Philippines government started to replant forest areas. Therefore, areas reforested by the

PAGE 47

47 government have been included in the model to control for environmental quality improvement. In addition it helps diminish inequality since poor people will not be able to make these environmental improvements due to lack of resources. The expected sign of this variable is positive since restoring areas that been d eforested should increase forest cover. Waters hed: A watershed is an area of land that divide s the catchment area or drainage basin, of one river system or group of river systems from another system or group of s ystems (Columbia Encyclopedia 2008). Having a watershed is one of the most effective ways to address the water resource challenge. It supplies drinking water, provides recreation and most importantly helps sustain life. Food, tourism and manufactured goods depend on healthy and clean watersheds (Environmental Protection Agency 2009) This vari able is used to account for efforts to maintain sustainable water resources and reduce sedimentation. The expected sign of this variable is positive because watersheds help forest provide its services to surrounding area s Community Based Forest Management Agreements: Community Based Forest Management Agreements (CBFMA) are agreements between the Department of Environment and Natural Resources (DENR) and local communities represented by people organizations as forest managers. The DENR gives the CBFMA holde r a specified amount of hectares of forestland where the holder assume responsibility for the protection of the entire forest lands within the CBFMA area against illegal logging and the other unauthorized extraction of forest products. This agreement also included several environmental components such as forest rehabilitation, forest protection, and development of alternative livelihood opportunities not necessarily dependent on forest products. In essence the CBFMA holder has to provide security for their CBFMA area and there are incentives to develop utilize and manage their specific portion of forest land. The number of sites variable is used to account for the presence of a

PAGE 48

48 CBFMA program in the region. The expected sign of the site variable is positive since more agreements imply that more areas have opportunities to be a part of the CBFMA program and manage the forest sustainably. The number of beneficiaries that depend on the CBFMA program have been included to test whether the participants are better off and have less incidence of poverty severity. The expected sign of the beneficiaries variable is positive since more people are participating in the CBFMA program and therefore can benefit economically, reducing their severity of poverty. The hectares of land tenured to the CBFMA program have been included to account for the amount of hectares t hat the holders are responsible for. The expected sign of CBFMA area is positive, because as more area s are devoted to safe management by the local residents, d eforestation should decrease. Socioeconomic variables: The unemployment variable is used to account for all persons who are 15 years old and over as of their last birthday and are reported as: (i ) without work and (ii) currently available for work and seek ing work The expected sign of unemployment is negative, since lack of income may force people to rely heavily on the forest. The crime variable is used to account for the average crime in a region. The expected sign of crime is negative since areas with high deforestation are thought to be areas with high poverty. The assumption being made is areas with high poverty, will have high incidences of crime. The immunization variable is used to account for initial health. The expected sign of immunization is positive. The assumption being made is that more immunizations result in healthier children. If children are healthier then parents will not have to rely on forest products and services as much. The density variable is used to account for population pressure (growth) for each region. The density variable expected sign is negative, because more people in a region may result in more pressure on the forest.

PAGE 49

49 Identification Strategy Povert y severity is identified using two instrument s access to water and incide nce of makeshift housing The access to water variable measures the proportion of the population that is able to access clean water. Being able to have access to clean water reduce the disease burden therefore making households more healthy and productive, thus reducing poverty. The instrument performs well in the first -stage results It has the expected sign (negative) and also is statistically significant. The makeshift housing variable is the proportion of families living in improvised housing People li ving in makeshift shelter are exposed to the crime, weather variations and etc. An increase of makeshift housing assumes that people are not able to afford adequate h ousing therefore the severity of poverty is thought to be increasing. The mhousing variabl e is significant and has the expected sign (positive). The instrumental variables results are presented in Chapter 6 (Table 6 4 ).

PAGE 50

50 CHAPTER 6 RESULTS This chapter presents and discusses the empirical results for the Ordinary Least Squares regression and the Two Stage Least Square regression. In addition several instrumental variable test s such as: endogeneity, overidentification, strength of instruments, and multicollinearity are reported. The study uses the EKC conceptual framework, which allows feedback ef fects of the environment to be estimated by simultaneous equations. The objectives in this study were to estimate the impact of severity of poverty and income distribution on deforestation, in addition to examining the influence of community forest managem ent agreements on deforestation and poverty. The hypothesis used to test these objectives were (i) areas with an increased severity of poverty will have higher deforestation ( ii) areas with higher income inequality will have higher deforestation rates an d (iii) areas with a higher number of community based forest management agreements will have lower levels of deforestation and less severity of poverty. Testing for Endogeneity & Overidentifying Restrictions Results for the endogeneity and overidentifying restrictions test are summarized in Table 6 1. When testing for endogeneity, the residual 2 from the reduced form equation is not significant I t has a p value of .28. The endogeneity test may have fail because there are other important variables that nee d to be included that account for institutional and cultural factors in the poverty-environmental degradation nexus (Duraiappah 1998). However, the poverty environment nexus literature implies that poverty and the environment affect each other (Bhattachary a and Innes 2007; Nelson and Ch omitz 2004; Dasgupta Deichmann, Mesiner and Wheeler 2005). Therefore we will assume that the severity of poverty and deforestation are jointly determined. This assumption does not affect our estimates, since both OLS and 2SL S results are presented and if there is no endogeneity both results consistent.

PAGE 51

51 In Chapter 5, one of requirements for an instrumental variable is it should not be correlated with the structur al error. Testing t his assumption can only be done in ov eridenti fied equations. The overidentified requirement is met since there is more than one IV being used for the endogenous variable Given that the cluster option is used, to test if the IV is correlated to the structural error, the Hansen J test is used Under t he null hypothesis all IVs are uncorrelated with the structural error. The J -statistic behaves like a chi -square random variable with degrees of freedom equal to the number of overidentifying restrictions, which from our model is equal to 1. The 5% critica l value for a chi -square of 1 is 3.84. The J -statistics is .22 with a p -value of .63, therefore we fail to reject the null hypothesis and conclude that the instruments are exogenous. Table 6 1 Specification test and Overi d entification t est Variable Coe fficient Robust Std Error t statistic P value Expenditure 5.490445 2.351354 2.34 0.033** Reforest .6920203 .5196209 1.33 0.202 Site .0008852 .0004861 1.82 0.087 Cbfmarea 2.19e 06 2.72e 07 8.04 0.000*** Beneficiaries 7.85e 06 2.53e 06 3.10 0.0 07*** Density .0001086 .000035 3.10 0.007*** Gini .6523177 1.856659 0.35 0.730 y03 .2724015 .0954605 2.85 0.011** y06 .3177518 .1933737 1.64 0.120 Crime .0290049 .0203832 1.42 0.174 Rainfall .0006074 .0003233 1.88 0.079* Shed 8.91e 07 6. 30e 07 1.41 0.177 Imm .0220391 .009049 2.44 0.027* Unemploy .0494881 .0420528 1.18 0.256 Severity .1380547 .0659229 2.09 0.053* 2 .0829869 .0741377 1.12 0.280 Cons 3.51548 1.535585 2.29 0.036* J Statistic = .22 p -value = .63 significant at 10 %; ** significant at 5 %; *** significant at 1 %. Testing Instrument Strength Result for instrument strength is summarized in Table 6 2. Recall from Chapter 5 that instrumental variables have to be correlate d with th e endogenous variable. However, for proper

PAGE 52

52 inference the instruments must have sufficient explanatory power. Stock and Yogo (2005) developed a table where a researcher could empirically test whether or not the instrumental variables being used were adequate enough to explain their variable in question. If an instrument is weak, the sampling distribution is no longer normal and the test statistics and confidence intervals are not reliable. Based on the Stock and Yogo (2005) when using two excluded in strumental variables to instrument for a single endogenous variable and restrict the bias of the IV estimator at the 10 % level the critical value of the first -stage F -statistic should exceed 19.93. The Kleibergen -Paap rk Wald F statistic is 30.29 which e xceed the critical value threshold; therefore our instruments are not weak and remain valid. Table 6 2 Weak Identification test Kleibergen -Paap rk Wald F statistic: 30.291 Stock Yogo weak ID test critical values: 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maxima l IV size 7.25 Source: Stock Yogo (2005). Reproduced by permission. Heteroskedascity Results for heterosked ascity are summarized in Table 6 3 Heteroskedascity is when the variance of the error term is not constant. When it occurs, standard errors are no longer valid for constructing confidence levels and t -statistics Using the Breusch -Pagan Hall test, we fail to reject the null hypothesis of homoskedascity. The OLS and 2SLS have p -values of .99 and .91 respectively. Heteroskedascity is auto matically corrected for the OLS and 2SLS models estimated with the cluster option Thus, he teroskedascity is not a major concern in this study. Table 6 3 Breusch Pagan Heteroskedascity t est Chi square P value OLS 0.00 .99 2SLS 8.86 .91

PAGE 53

53 Multicolline arity Results for multicollin earity are summarized in Table 6 4 Multicollinearity is a violation of assumption III; which is none of the independent variables can be constant and there can be no exact linear relationships There are several variables that exceed the variance inflator factor threshold of 10: cover, cbfmarea, and density. However, multicollinearity is not considered a problem since all these variables are statistically significant. If the variables lacked significance and exceeded the 10 thr eshold then the model would have a serious multicollinearity issue Even in the presence of extreme multicollinearity none of the Gauss-Markov assumptions are violated (Wooldridge 2006). In addition the mean variance inflator factor for the model is 5 .27 for all variables which is below the rule of thumb threshold. Table 6 4. Multicollinearity Variable VIF SQRT R Squared Tolerance 5 R Squared Cover 10.67 3.27 0.0938 0.9062 Expenditure 7.54 2.75 0.1326 0.8674 Reforest 1.60 1.27 0.6241 0.3759 Sit e 4.98 2.23 0.2007 0.7993 Cbfmarea 10.42 3.23 0.0960 0.9040 Beneficiaries 4.24 2.06 0.2358 0.7642 Density 11.35 3.37 0.0881 0.9119 Gini 3.54 1.88 0.2829 0.7171 Y0 3 3.45 1.86 0.2899 0.7101 Y06 5.66 2.38 0.1766 0.8234 Crime 3.32 1.82 0.30 13 0.6987 Rainfall 2.66 1.63 0.3758 0.6242 Shed 2.10 1.45 0.4763 0.5237 Imm 3.83 1.96 0.2611 0.7389 Unemploy 7.69 2.77 0.1300 0.8700 Severity 6.35 2.52 0.1575 0.8425 Water 3.08 1.75 0.3252 0.6748 Mhousing 2.40 1.55 0.4166 0.5834 Mean VIF = 5.27 5 Tolerance = 1/VIF

PAGE 54

54 First -Stage Regression Results are summarized in Table 6 5. The instrumental variables used in the empirical model were the proportion of the population with access to clean water denoted as water and proportion of families living in makeshif t housing denoted as mhousing. Both instrumental variables are statistically significant and have the correct sign s The water variable is negative because as more people have access to clean water, severity is assumed to be decreas ing The decline in environmental quality is likely to affect the health of the poor more severely than the wealthy segment of the population, since the wealthy are not as vulnerable to environmental changes and can cope (Nunan et al 2002). Therefore, the environmental conditions lessen the poor disease burden. The mhousing variable is positive because as more people live in makeshift housing severity of poverty is assumed to be increasing. People are not able to afford decent housing and reside in inadequate housing structures ma de with improvised materials Table 6 5 First -stage Regression and Instrument S ignificance test Severity Variable Coefficient P value Expenditure 12.97016 0.086 Reforest .2317697 0.928 Site .0026025 0.038 Cbfmarea 8.79e 08 0.936 Beneficiarie s .0000196 0.003 Density .0001888 0.067 Gini 6.773171 0.266 Y03 .0170838 0.974 Y06 .2048727 0.763 Crime .0759954 0.236 Rainfall .0012379 0.208 Shed 3.43e 06 0.206 Imm .0062876 0.865 Unemploy .1751798 0.137 Water .0517464 0.004 Mhousing .6814824 0.037 Cons 11.65057 0.021 Wald F statistic : 30.29 Instruments: water, mhousing

PAGE 55

55 O LS and 2SLS Regression Results of OLS and 2SLS with and without the cluster option are summarized in Table 6 6 The coefficients are the same in the OL S estimations but the level of significance are different. Using OLS with the cluster option makes some variables less significant. The same pattern is exhibited in the 2SLS estimation; the cluster option makes some the variables less significant Regardles s, the variables still maintain their statistical significance In all four estimation methods the severity variable is significant. This gives us evidence that severity of poverty negatively effects forest cover. The 2SLS estimate of severity of poverty on forest cover is 13 % while the OLS estimate is 8 %. The expenditure variable is also statistically significant in all four estimation models. It negatively impacts forest cover which supports the prediction of the Environmental Kuznets Curve in develop ing countries The Gini coefficient is not significant i n any of the model and therefore according to the results income inequality does not seem to impact deforestation. All the Community Forest Management coefficients variables are significant but very s mall. This implies that although they affect forest cover, in order to make a difference in reducing deforestation they need to be occurring at higher levels than they are currently operating. Thus this policy tool is being underutilized or in efficient in some aspect. The density variable is negative and statistically significant but the coefficient is also very small. Th is implies while population density does negatively impact forest cover the current population density in the Philippines is not a major factor in aggravating deforestation. Population density would have to be at higher levels for it to be considered a potential threat to forest cover. The immunization variable implies that the more children are immunized; deforestation will decrease by 2%

PAGE 56

56 Table 6 6 OLS and 2SLS R egression Cover OLS 2SLS OLS w/cluster 2SLS w/cluster Severity .0879076 (0.024) ** .1380547 (0.029)** .0879076 (0.082)* .1380547 (0.038) ** E xpenditure 4.173064 (0.050) 5.490445 (0.031) ** 4.173064 (0.061) 5.490445 (0.025) ** Reforest .7191163 (0.314) .6920203 ( 0.344 ) .7191163 (0.143) .6920203 (0.200) Site .0010997 (0.003) *** .0008852 ( 0.037 ) ** .0010997 (0.050) .0008852 (0.084) Cbfmarea 2.32e 06 (0.000) *** 2.19e 06 ( 0.000 ) *** 2.32e 06 (0.000) *** 2.19e 06 (0.000) *** Beneficiaries 6.70e 06 (0.001) *** 7.85e 06 ( 0.001 ) *** 6.70e 06 (0.006) *** 7.85e 06 (0.005) *** Density .0000972 (0.002) *** .0001086 ( 0.001 ) *** .0000972 (0.009) *** .0001086 (0.005) *** Gini .6291168 ( 0.698 ) .6523177 ( 0.695 ) .62 91168 (0.741) .6523177 (0.722) Y03 .2616552 (0.048) .2724015 ( 0.045 ) ** .2616552 (0.022) ** .2724015 (0.014) ** Y06 .3144232 (0.100) .3177518 (0.105) .3144232 (0.111) .3177518 (0.089) Crime .0244339 (0.179) .0290049 (0.131) .0244339 (0.236) .0290049 (0.182) Rainfall .0005029 (0.074) .0006074 (0.048)** .0005029 (0.099)* .0006074 (0.098) Shed 1.03e 06 (0.167) 8.91e 07 (0.251) 1.03e 06 (0.089)* 8.91e 07 (0.146) Imm .0209371 (0.019) ** .0220391 (0.017)** .0209371 (0.055)* .0220391 (0.051) Unemploy .033532 (0.285) .0494881 ( 0.166 ) .033532 ( 0.413 ) .0494881 (0.286) Cons 2.747172 (0.064) 3.51548 ( 0.039 ) ** 2.747172 ( 0.030 ) ** 3.51548 (0.036) ** Observations R Squared 51 9023 51 51 9023 51 p values in parentheses; significant at 10 %; ** significant at 5 %; *** significant at 1 %.

PAGE 57

57 CHAPTER 7 CONCLUSIONS AND IMPLICATIONS FOR POLICY Summary Efficient targeting of resources to achieve poverty reduction objectives requires information about the poor and their circumstanc es. To be tter underst and the poverty and environmental relationship, c orrectly measuring who the poor are, where they live, and how they respond to programs intended for them is vital Deforestation can generate significant negative externalities ranging from eleva ted risk of erosion, loss of biodiversity, and increased carbon into the atmosphere associated with climate change. The study shows that environmental degradation does negatively impact poverty and measure s should be taken to improve environmental quality. Conclusions and Implications for Policy The Environmental Kuznets Curve predicts developing countries experience more environmental degradation until they have reached a critical income per capita The expenditure variable which has been used instead of income due its stable quality, is statistical significant and negative This implies that the Philippine is to the left of the EKC and still has not reached the expenditure (income ) necessary for environmental degradation to start decreasing (Figure 7 1) Figure 7 1 Philippines Environmental Kuznets Curve

PAGE 58

58 T he severity of poverty variable was statistically significant and negative in all four models. Based on the negative relationship with the cover variable, there is evidence that the severity of poverty has an effect on forest cover. Therefore the evidence supports our first hypothesis that the severity of poverty has an impact on deforestation. As severity of poverty increases the forest cover will decrease and aggravate deforestation. The Gini coeffi cient used to estimate income distribution has a negative sign, which means it would have had a negative impact on deforestation. However, this variable is not significant. Therefore, there seems to be no or little evidence that income distribution has an impact on deforestation. In essence, greater inequality does not seem to influence environmental improvement. The wealthy segment of the population is not investing in environmentally sustainable activities. There is no support for h ypothesis two which sta tes that greater inequality would lead to increased deforestation. The third hypothesis seems to work very well in both in OLS and 2SLS. The number of sites in each region has a statistically significant negative impact on deforestation. The more community based forest agreement that are in the region the more deforestation occurs. However, the area of hectares given to each site has a positive statistically significant effect on deforestation. This could imply that a few larger sites could help decrease de forestation as opposed to awarding several smaller sites. This could provide evidence for policy makers to give larger portions of land for safe management to a larger community and refrain from awarding small parcels of land to small groups. This could i nstead encourage small groups to work with other communities, in order to gain the community a greement. Bigger community involvement and interaction could help slow the rate of deforestation oppose to fragmenting the forest by providing small areas to too many

PAGE 59

59 groups Thus this result gives strong evidence that Community Based Forest Management Agreements has a positive impact on deforestation, but only if awarded properly. However, the beneficiaries variable is statistically negative which means that as the number of people that depend on the area tenured increases forest cover is reduced. This implies that t he more people that depend on an area the faster resource s get depleted. While bigger areas being awarded slows down deforestation, policymakers have to mindful that there is an optimal number of people that should be depending on the area tenured. When too many people are depending on a parcel of forest, even if the forest acres are awarded properly deforestation will continue to occur. There is also evidence that CBFMAs have a statistically significant positive impact on the severity of poverty. The area tenured variable implies that this policy tool in a region is assisting in the reduction of poverty especially amongst the households considered sev erely poor The beneficiaries variable is also statis ti cally positive which means that people who are choosing to participate in this program are benefitting because as beneficiaries of CFBMA s increase the severity of poverty decrease. The site variable is statistical positively related to the severity of poverty variable. This means a s the number of sites increase, poverty increase. This evidence helps support the idea that larger areas should be awarded rather than smaller forest areas. Further Research and Limitations This study could be extended to incorporate linear programming or simulations to estimate the number of optimal sites each region should have. In addition, the optimal number of beneficiaries needs to be identified, so policy makers can est imate how many people should be depend ing on the areas tenured. An in -depth study on the components and techniques that allow the Community Forest Management Agreement to be successful at controlling deforestation should also be explored. Understanding wha t works in context will help replicate the model in other countries or regions that desperately need the knowledge and ideas dispersed.

PAGE 60

60 The limitation of the study is that the data used is at the regional level rather than provincial. Therefore accurate local impacts are not fully realized. In addition, the community forest agreement data is not at a scale where specific community factors can be identified and assessed such as management styles, voting processes, and etc which need to b e accounted for in the model. The data does not tell us whether certain components being implemented in the agreements play an influential role in determining the success of the forest management agreeme nts. Lastly, the 2SLS method used to estimate the simultaneous equation s is consistent, yet i t i s still bias. This means it increases Therefore, t he larger the sample size, the better it is to use 2SLS.

PAGE 61

61 APPENDIX A M AP OF THE PHILIPPINE S Figure A 1 Map of t he Philippines

PAGE 62

62 APPENDIX B OFFICIAL POVERTY LIN E Table A 1 Official p overty l ine (in p esos) 2000 2003 2006 Philippines 11, 458 12,309 15,057 NCR 15,722 16,737 20,566 CAR 13,071 14,033 16,810 Region I 12, 687 13,281 15,956 Region II 11,128 11,417 13,791 Region III 13,760 14,378 17,298 Region IV -A 13,670 14,720 17,761 Region IV B 12,013 12,402 14,800 Region V 11,375 12,379 15,015 Region VI 11,314 12,291 14,405 Region VII 9,659 9,805 13,390 Region VIII 9,530 10,804 13,974 Region IX 9,128 10,407 13,219 Region X 10,509 11,605 14,199 Region XI 10,278 11,399 14,942 Region XII 10,458 11,328 14,225 Region XIII 10,903 11,996 15,249 ARMM 12,999 12,733 15,553

PAGE 63

63 APPENDIX C FULL RESULTS Table C 1 Fu ll e stimation by OLS without cluster Cover Coefficient Std. Err. t P> t Lower Upper 6 Expenditure 4.173064 2.056124 2.03 0.050 8.347218 .0010898 Refor est .7191163 .7035549 1.02 0.314 2.147409 .7091761 Site .0010997 .0003461 3.18 0.003 .0018023 .000397 Cbfmarea 2.32e 06 2.75e 07 8.44 0.000 1.76e 06 2.88e 06 Beneficiarie s 6.70e 06 1.88e 06 3.56 0.001 .0000105 2.88e 06 Density .0000972 .0000285 3.40 0.002 .0001551 .0000392 Gini .6291168 1.610089 0.39 0.698 3.897771 2.639538 Y03 .2616552 .1275555 2.05 0.048 .5206066 .0027038 Y06 .3144232 .1862684 1.69 0.100 .6925681 .0637217 Crime .0244339 .0178125 1.37 0.179 .060 5952 .0117274 Rainfall .0005029 .0002726 1.85 0.074 .0010562 .0000504 Shed 1.03e 06 7.33e 07 1.41 0.167 4.55e 07 2.52e 06 Imm .0209371 .0085275 2.46 0.019 .0036253 .038249 Unemploy .033532 .0308519 1.09 0.285 .0961648 .0291008 Severity .0 879076 .0371942 2.36 0.024 .1634159 .0123992 Cons 2.747172 1.43388 1.92 0.064 .1637596 5.658104 Observation 51 R squared 0.9023 6 95 % confidence interval

PAGE 64

64 Table C 2 Full e stimation by OLS with cluster Cover Coefficient Std. Err. t P> t Lower Upper Expenditure 4.173064 2.073804 2.01 0.061 8.569332 .223204 Reforest .7191163 .4671479 1.54 0.143 1.709426 .2711931 Site .0010997 .0005179 2.12 0.050 .0021977 1.67e 06 Cbfmarea 2.32e 06 2.53e 07 9.1 9 0.000 1.79e 06 2.86e 06 Beneficiar ie s 6.70e 06 2.14e 06 3.13 0.006 .0000112 2.16e 06 Density .0000972 .0000327 2.97 0.009 .0001664 .0000279 Gini .6291168 1.874284 0.34 0.741 4.60242 3.344187 Y03 .2616552 .1036095 2.53 0.022 .4812974 .042013 Y06 .3144232 .1863176 1.69 0.111 .7093989 .0805524 Crime .0244339 .019843 1.23 0.236 .0664992 .0176314 Rainfall .0005029 .0002876 1.75 0.099 .0011125 .0001067 Shed 1.03e 06 5.71e 07 1.81 0.089 1.77e 07 2.24e 06 Imm .0209371 .0 101128 2.07 0.055 .000501 .0423753 Unemploy .033532 .0399186 0.84 0.413 .1181557 .0510917 Severity .0879076 .0474373 1.85 0.082 .1884702 .0126551 Cons 2.747172 1.156105 2.38 0.030 .296339 5.198005 Observations 51 R Squared 9023

PAGE 65

65 Ta ble C 3 Full e stimation by 2SLS without cluster Cover Coef ficient Std. Err. t P> t Lower Upper Severity .1380547 .0606422 2.28 0.029 .261165 .0149444 Expenditure 5.490445 2.445574 2.25 0.031 10.45523 .52 56651 Reforest .6920203 .7220432 0.96 0.344 2.157846 .7738053 Site .0008852 .0004082 2.17 0.037 .001714 .0000565 Cbfmarea 2.19e 06 3.10e 07 7.06 0.000 1.56e 06 2.81e 06 Beneficiar ies 7.85e 06 2.21e 06 3.55 0.001 .0000123 3.36e 06 Densit y .0001086 .0000312 3.48 0.001 .000172 .0000453 Gini .6523177 1.651515 0.39 0.695 4.005071 2.700436 Y03 .2724015 .1312154 2.08 0.045 .5387828 .0060201 Y06 .3177518 .1910698 1.66 0.105 .7056442 .0701406 Crime .0290049 .0187677 1.55 0.131 .0671055 .0090956 Rainfall .0006074 .0002963 2.05 0.048 .0012089 5.85e 06 Shed 8.91e 07 7.63e 07 1.17 0.251 6.59e 07 2.44e 06 Imm .0220391 .0088073 2.50 0.017 .0041593 .0399189 Unemploy .0494881 .035018 1.41 0.166 .1205784 .0216022 Cons 3.51548 1.638422 2.15 0.039 .1893057 6.841654 Observations 51

PAGE 66

66 Table C 4 Full e stimation by 2SLS with cluster Cover Coefficient Std. Err. t P> t Lower Upper Severity .1380547 .0610953 2.26 0.038 .267571 .00853 84 Expenditure 5.490445 2.222425 2.47 0.025 10.20178 .7791148 Reforest .6920203 .5180482 1.34 0.200 1.790233 .4061928 Site .0008852 .0004809 1.84 0.084 .0019047 .0001342 Cbfmarea 2.19e 06 2.91e 07 7.50 0.000 1.57e 06 2.80e 06 Beneficiari es 7.85e 06 2.43e 06 3.23 0.005 .000013 2.69e 06 Density .0001086 .0000331 3.28 0.005 .0001789 .0000384 Gini .6523177 1.798469 0.36 0.722 4.464901 3.160266 Y03 .2724015 .0988196 2.76 0.014 .4818897 .0629132 Y06 .3177518 .1756391 1 .81 0.089 .69009 .0545864 Crime .0290049 .0207871 1.40 0.182 .0730716 .0150617 Rainfall .0006074 .0003459 1.76 0.098 .0013407 .0001259 Shed 8.91e 07 5.83e 07 1.53 0.146 3.45e 07 2.13e 06 Imm .0220391 .0104321 2.11 0.051 .000076 .0441542 U nemploy .0494881 .0448124 1.10 0.286 .1444862 .04551 Cons 3.51548 1.537563 2.29 0.036 .2559912 6.774968 Observations 51 R Squared .8363

PAGE 67

67 APPENDIX D SUMMARY STATISTICS Table D 1 Summary s tatistics Variable Observations Mean Std. Dev. Min Max Year 51 2003 2.473863 2000 2006 Region 51 9 4.947727 1 17 Cover 51 1.254017 .743488 .00026 2.79 Severity 51 3.94902 2.029322 .3 8.6 Expenditure 51 .1131941 .0478784 .007 .23 Reforest 51 .053996 .0691201 0 .31 Site 51 347.1176 214.160 2 0 831 Cbfma Area 51 307969.1 263289.6 0 955283 Beneficiaries 51 42762.82 36616.15 0 204511 Density 51 1244.765 3987.449 74 18165 Gini 51 .4301961 .0431041 .31 .52 Rainfall 51 2814.118 222.4978 2283 3063 Watershed 51 88214.55 72010.76 0 280461 Water 51 76.99608 12.87116 34.1 96.4 Mhousing 51 1.727451 .651484 .7 3.1 Unemployment 51 7.776471 3.213259 2.8 17.8 Crime 51 7.592745 3.887093 1.91 17.12 Immunization 51 79.87843 7.115822 65.8 95 Y03 51 .3333333 .4760952 0 1 Y06 51 .3333333 .4760952 0 1

PAGE 68

68 L IST OF REFERENCES Asian Development Bank. 2000. State of the Environment in Asia and the Pacific New York: United Nations. Asian Development Bank. 2005. Poverty in the Philippines: Income, Assets, and Access Southeast Asia: Asian Devel opment Bank. Asian Development Bank. 2007. Philippines: Critical Development Constraints. Philippines: Asian Development Bank. Baum, C., M. Schaffer, an d S. Stillman. 2007. Enhanced Routines for I nstrumental Variables and GMM E stimation and Testing. Work ing P aper, Boston College. Bhattacharya, H., and R. Innes. 2007. Does Poverty Aggravate Environmental Degradation in Rural India? Working P aper, The Earth Institute Columbia University. Bhatti, J.S., R. Lal, M.J. Apps, and M.A. Price. 2006. Climate Cha nge and Managed Ecosystems. Boca Raton: Taylor and Francis Group. Boyce, J. 1994. Inequality as a C au se of Environmental Degradation. Ecological Economics 11(3): 169 7 8. Chotikapanich D., and W. Griffiths. 1999. Estimating Lorenz Curves Using a Dirichl et Distribution. Working Paper, University of New England Columbia Encyclopedia 2008. Watershed. New York: Columbia University Press Conservation International. 2009. Philippines. Retrieved on March 7, 2009. Available at : http://www.conservation.org/explore/regions/asia -pacific/philippines/Pages/overview.aspx Dasgupta, P., U. Deichmann, C. Mesiner, and D. Wheeler. 2005. Where is the Poverty Envi ronment Nexus: Evidence from Cambodia, Lao PDR and Vietnam. World Development 33 (4):617 38. Dasgupta, S., B. Laplante, H. Wang, and D. Wheeler. 2002. Confronting the E nvironmental Kuznets Curve. Journal of Economic Perspectives 16:147 68. Deacon R. an d C. Norman. 2004. Is the E nvironmental Kuznets Curve an E mp irical Regularity? Working Paper, Department of Economics University of California at Santa Barbara Dinda, S. 2004. Environmental Kuznets Curve Hypothesis: A Survey. Ecological Economics 49: 43155. Duraiappah, A.K. 1998. Poverty and Environmental Degradation: A R eview and Analysis of the Nexus. World Development 26(12): 21691 79.

PAGE 69

69 Environmental Protection Agency. 2009. Watershed. Retrieved on March 9, 2009. Available at : http://www.epa.gov/owow/watershed/whatis.html ESCAP. 2006. State of the Environment in Asia and the Pacific Thailand: United Nations. F AO. 2007. State of the Worlds Forest 2007. Rome : United Nations FAO. 2006. Global Forest Resources Assessment 2005: Progress towards Sustainable Forest Management. Rome: United Nations. FAO 2009 Forestry Retrieved on January 4, 2009. Available at : http://www.fao.org/for estry/home/en/ Forsyth, T., M. Leach and I. Scoones. 1998. Poverty and E nvironment: P riorities for R esearch and P olicy: An Overview S tudy. United Nations Development Pr ogramme and European Commission at I nstitute of Development Studies Sussex. Free r -Smith, P., M. Broadmeadow, and J. Lynch. 2007. Forestry and Climate Change Cambridge: OECD. Gallet, C., and J. List. 1999. The E nvironmental Kuznets Curve: D oes One Size F it A ll? Ecological Economics 31(3): 40923. Gibson, C., M. McKean, and E. Ostrom. 2000. People and Forests: Communities, I nstitutions and G overnance. Cambridge: Massachusetts Institute of Technology. Greene, W. 2000. Econometric Analysis Fourth Edition New Jersey: Prentice Hall. Grossman, G. and A. Kruger. 1995. Econo mic G rowth and the E nvironment Working Pa per, National Bureau o f Economic Research. Gujarti, D. 2003. Basic Econometrics Fourth E dition. Boston: McGraw Hill. Holtz -Eakin, D., and T M. Selden. 1995. Stoking the F ires: CO2 E missions and Economic G rowth. Journal of P ublic Economics 57:85 101. Intergovernmental Panel on Climate Change. 2007b. Mitigation of Climate Change New York: Cambridge University Press. Intergovernmental Panel on Climate Change. 2007c. Impacts, Adaptation and Vulnerability. New York: Cambridge U niversity Press. Jodha, N.S. 2000. Common Property Resources and the Dynamics of Rural Poverty: Field Evi dence from Dry Regions of India. In W. Hyde and G. Amacher, eds. Economics of Forestry and Rural Development: An Empirical Introduction from Asia. M ichigan: University of Michigan Press. Kimmins, H. 1997. Balancing Act: Environmental Issues in Forestry Second Edition. Vancouver: UBC Press.

PAGE 70

70 Kummer, D. 1996. Deforestation in Postwar Philippines. Chicago: University of Chicago Press. Kuznets, S 1955. Economic Growth and Income Inequality. American Economic Review 45(1): 1 28. Lal, R. 2004. Soil Carbon S equestration to M itigate Climate C hange Geoderma 123:122. Laurence, W. and C. Peres. 2006. Emerging Threats to Tropical Forests. Chicago: Universit y of Chicago. Lee, R., D. Field, and W. Burch. 1990. Community and Forestry: Continuities in the Sociology of Natural Resources Boulder: Westview Press. Lopez, A., C. Mathers, M. Ezzati, D. Jamison, and C. Murray. 2006. Global Burden of Disease and Risk F actors Washington D.C.: World Bank and Oxford Press. Mayers, J and S. Bass. 2004. Policy That Works for Forest and People: Real Prospects for Governance and Livelihoods Sterling: Earthscan. Millimet, D. L., J. A. List, and T. Stengos. 2003. The Enviro nmental Kuznets C urve: Real P rogress or M isspecified M odels? Review of Economics and Statistics 85(4):1038 47. Nelson, A ., and K. Chomitz. 2004. The Forest -Hydrology -Poverty Nexus in Central America: An Heuristic Analysis Working Paper, The World Bank Policy Research Nunan, F., U. Grant, G. Bahiigwa, T. Muramira, P. Bajracharya, D. Pritchard, and M.J. Vargas. 2002. Poverty and the Environment: Measuring the Links A Study of Poverty Environment Indicators with Case Studies from Nepal, Nicaragua and U ganda. London : Department of International Development Panayotou, T. 1993. Empirical T ests and P olicy A nalysis of Environmental D egradation at D ifferent S tages of Economic D evelopment Working Paper, Technology and Employment Programme at International Labour Office Puhe, J. and B. Ulrich. 2001. Global Clim ate Change and Human Impacts on F orest Ecosystems: Postglacial Development, Present Situations, and Future Trends in Central Europe. Germany: Springer. Ravnborg, H.M. 2003. Poverty and Environmental Degradat ion in the Nicaraguan Hillsides. World Development 31( 11): 193346. Republic of the Philippines. 2009. History and General Information Retrieved on February 16, 2009. Available at: http://ww w.gov.ph/aboutphil/general.asp Republic of the Philippines, Congress of the Philippines. 1997. Philippines Republic Act 8425: Social Reform and Poverty Alleviation Act. Metro Manila.

PAGE 71

71 Republic of the Philippines, Forest Management Bureau. 2003. Revised Mas ter Plan for Forestry Development Metro Manila. Republic of the Philippines, Forest Management Bureau. 2009. History Retrieved on January 29, 2009. Available at http://forestry.denr.gov.ph/history 1.htm Metro Manila. Republic of the Philippines, National S tatistical Coordination Board. 2000. Family Income and Expenditure Survey Metro Manila. Republic of the Philippines, National S tatistical Coordination Board. 2003. Family Income and Expenditure Survey. Metro Manila. Republic of the Philippines, National S tatistical Coordination Board. 2005. Census Statistics. Retrieved on October 12, 2008. Available at : http://www.gov.ph/aboutphil/general.asp. Metro Manila. Republic of the Philippines, National S tatistical Coordination Board. 2006. Family Income and Expenditure Survey Metro Manila. Republic of the Philippines, National S tatistical Coordination Board. 2008. Official Poverty Statistics: Stateme nt of Secretary Augusto B Santos Metro Manila. Rudel, T. 2005. Tropical Forest: Regional Paths of Destruction and Regeneration in the Late Twenthieth Century. New York: Columbia University Press. Sachar, R. 1996. The Right to Adequate Housing. New York: United Nations. Selden, T. M., and D. Song 1994. Environmental Quality and D evelopment: Is T here a Kuznets Curve for A ir P ollution? Journal of Environmental Economics and Management 27:147 62. Shafik, N. 1994. Economic D evelopment and E nvironment al Q u al ity: An E conometric A nalysis. Oxford Economic Papers : Special Issue on Environmental Economic s 46:75773. Shafik, N. and S. Bandyopadhyay. 1992. Economic G rowth and Environmental Quality: T ime Series and CrossC ountry E vidence. Background P aper for the World Development Report 1992, The World Bank, Washington D.C. Sohngen, B. and R. Mendelsohn 2003. An Optimal Control Model of Forest Carbon Sequestration. American Journal of Agricultural Economics 85: 448 57. Spray, S., and M. Moran. 2006. Tropical Deforestation Lanhan: Rowman and Littlefield Publishers, Inc Stern, D. 2004. The Rise and Fall of the Environmental Kuznets Curve. World Development 32(8 ): 141939.

PAGE 72

72 Stern, D. I., M.S. Common, and E.B. Barbier 1996. Economic Growth and Environmental D egradation: The Environmental Kuznets Curve and Sustainable D evelopment. World Development 24: 1151 60. Stock, J. H., and M. Yogo. 2005. Testing for Weak Instruments in Linear IV Regression. In D.W. Andrews and J. H. Stock eds Identification and Infere nce for Econometric Models: Essays in Honor of Thomas Rothenberg. New York: Cambridge University Press. Stock, J. H., J. H. Wright, and M. Yogo. 2002. A Survey of Weak Instruments and Weak Identification in Generalized Method of Moments. Journal of Business & Economic Statistics 20(4):51829. Swinton, S.M., G. Escobar and T. Reardon. 2003. Poverty and Environment in Latin America: Concepts, Evidence and Policy Implications. World Development 31(11): 18651872. Tiffen, M., M. Mortimore and F. Gichuki 1994. More People, Less Erosion: Environmental Recovery in Kenya. London: John Wiley and Sons. Torras, M., and J.K. Boyce. 1998. Income, Inequality, and P ollution: A R eassessment of the E nvironmental Kuznets Curve. Ecological Economics 25: 147 6 0. UCLA, Academic Technology Services 2008. Adapted from SAS Class Note s. Ret rieved on February 19, 2009. Available at: http://www.ats.ucla.edu/stat/sas/notes/ UNDP. 2008. Poverty & Environmental Indicator s. Cambridge: St. Edmund College. USAID, Environment Department. 2008. Forestry Retrieved on July 14, 2008. Available at : http://www.usaid.gov/our_work/environment/forestry/ Van Den Top G 2003. The Social Dynamics of Deforestation in the Philippines: Actions, Options and Motivations. Denmark: Nordic Institute of Asian Studies. Wooldridge, J. 2000. Introductory Econometrics: A Modern Approach. Cincinnati: South Western College. Wooldridg e, J. 2006. Introductory Econometrics: A Modern Approach, Third Edition Cincinnati: S outh -Western College World Bank. 2008. At Loggerhead: Agricultural Expansion, Poverty Reduction, and Environment in the Tropical Forests Washington D.C.: World Bank. Wo rld Bank. 2008. Measuring Poverty Retrieved on August 7, 2008. Available at : http://go.worldbank.org/0C60K5UK40

PAGE 73

73 BIOGRAPHICAL SKETCH Jessica Georges was born in Miami, Florida. She recei ved her Bachelor of Science in e conomics at the University of Florida in May 2007 with two minors: a nthropology and family and c ommunity s cience. She started the f ood and r esource e conomic s Master of Science program in August 2007 and receive d her degr ee in May 2009 with a minor in political s cience.