Country 'choices' or deforestation paths: a method for global change analysis of human-forest interactions

G.M. Koop, L.A. Tole

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Data used in quantitative studies of global tropical deforestation are typically of poor quality. These studies use either cross-sectional or panel data to measure the contribution of social and land use factors to forest decline world wide. However, there are pitfalls in the use of either type of data. Panel data studies treat each year's observation as a distinct, reliable, data point, when a careful examination of the data reveals this assumption to be implausible. In contrast, cross-sectional studies discard most of the time series information in the data, calculating a single average deforestation rate for each country. In this paper, we argue for a middle road between these two approaches: one that does not treat the time series information as completely reliable but does not disregard it altogether. Using a well-known global forest data set (FAO'sProduction Series Yearbooks ), we argue that the most the data can reliably tell us is whether a country's deforestation rate falls into one of four categories or country 'path choices'. We then use the data categorised in this way in a small empirical investigation of the socio-economic causes of deforestation. This multinomial logit framework allows for the determination of the influence of independent variables on the probability that a country will follow one deforestation path vs. another. Results from the logit analysis of key social and land use indicators chosen for their importance in the literature in driving deforestation suggest that the effect of these variables will differ for countries depending on the particular set of deforestation trajectories in question.
Original languageEnglish
Pages (from-to)133-148
Number of pages16
JournalJournal of Environmental Management
Volume63
Issue number2
DOIs
Publication statusPublished - Oct 2001

Fingerprint

Deforestation
global change
deforestation
panel data
Land use
Time series
logit analysis
time series
land use
Food and Agricultural Organization
method
analysis
trajectory
Trajectories
road
Economics

Keywords

  • tropical deforestation
  • human-forest interactions
  • multinomial logit analysis
  • development studies
  • economics

Cite this

@article{0a93c6c0660148d7ab8c3c02a04b4f11,
title = "Country 'choices' or deforestation paths: a method for global change analysis of human-forest interactions",
abstract = "Data used in quantitative studies of global tropical deforestation are typically of poor quality. These studies use either cross-sectional or panel data to measure the contribution of social and land use factors to forest decline world wide. However, there are pitfalls in the use of either type of data. Panel data studies treat each year's observation as a distinct, reliable, data point, when a careful examination of the data reveals this assumption to be implausible. In contrast, cross-sectional studies discard most of the time series information in the data, calculating a single average deforestation rate for each country. In this paper, we argue for a middle road between these two approaches: one that does not treat the time series information as completely reliable but does not disregard it altogether. Using a well-known global forest data set (FAO'sProduction Series Yearbooks ), we argue that the most the data can reliably tell us is whether a country's deforestation rate falls into one of four categories or country 'path choices'. We then use the data categorised in this way in a small empirical investigation of the socio-economic causes of deforestation. This multinomial logit framework allows for the determination of the influence of independent variables on the probability that a country will follow one deforestation path vs. another. Results from the logit analysis of key social and land use indicators chosen for their importance in the literature in driving deforestation suggest that the effect of these variables will differ for countries depending on the particular set of deforestation trajectories in question.",
keywords = "tropical deforestation, human-forest interactions, multinomial logit analysis, development studies, economics",
author = "G.M. Koop and L.A. Tole",
year = "2001",
month = "10",
doi = "10.1006/jema.2001.0466",
language = "English",
volume = "63",
pages = "133--148",
journal = "Journal of Environmental Management",
issn = "0301-4797",
number = "2",

}

Country 'choices' or deforestation paths: a method for global change analysis of human-forest interactions. / Koop, G.M.; Tole, L.A.

In: Journal of Environmental Management, Vol. 63, No. 2, 10.2001, p. 133-148.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Country 'choices' or deforestation paths: a method for global change analysis of human-forest interactions

AU - Koop, G.M.

AU - Tole, L.A.

PY - 2001/10

Y1 - 2001/10

N2 - Data used in quantitative studies of global tropical deforestation are typically of poor quality. These studies use either cross-sectional or panel data to measure the contribution of social and land use factors to forest decline world wide. However, there are pitfalls in the use of either type of data. Panel data studies treat each year's observation as a distinct, reliable, data point, when a careful examination of the data reveals this assumption to be implausible. In contrast, cross-sectional studies discard most of the time series information in the data, calculating a single average deforestation rate for each country. In this paper, we argue for a middle road between these two approaches: one that does not treat the time series information as completely reliable but does not disregard it altogether. Using a well-known global forest data set (FAO'sProduction Series Yearbooks ), we argue that the most the data can reliably tell us is whether a country's deforestation rate falls into one of four categories or country 'path choices'. We then use the data categorised in this way in a small empirical investigation of the socio-economic causes of deforestation. This multinomial logit framework allows for the determination of the influence of independent variables on the probability that a country will follow one deforestation path vs. another. Results from the logit analysis of key social and land use indicators chosen for their importance in the literature in driving deforestation suggest that the effect of these variables will differ for countries depending on the particular set of deforestation trajectories in question.

AB - Data used in quantitative studies of global tropical deforestation are typically of poor quality. These studies use either cross-sectional or panel data to measure the contribution of social and land use factors to forest decline world wide. However, there are pitfalls in the use of either type of data. Panel data studies treat each year's observation as a distinct, reliable, data point, when a careful examination of the data reveals this assumption to be implausible. In contrast, cross-sectional studies discard most of the time series information in the data, calculating a single average deforestation rate for each country. In this paper, we argue for a middle road between these two approaches: one that does not treat the time series information as completely reliable but does not disregard it altogether. Using a well-known global forest data set (FAO'sProduction Series Yearbooks ), we argue that the most the data can reliably tell us is whether a country's deforestation rate falls into one of four categories or country 'path choices'. We then use the data categorised in this way in a small empirical investigation of the socio-economic causes of deforestation. This multinomial logit framework allows for the determination of the influence of independent variables on the probability that a country will follow one deforestation path vs. another. Results from the logit analysis of key social and land use indicators chosen for their importance in the literature in driving deforestation suggest that the effect of these variables will differ for countries depending on the particular set of deforestation trajectories in question.

KW - tropical deforestation

KW - human-forest interactions

KW - multinomial logit analysis

KW - development studies

KW - economics

U2 - 10.1006/jema.2001.0466

DO - 10.1006/jema.2001.0466

M3 - Article

VL - 63

SP - 133

EP - 148

JO - Journal of Environmental Management

JF - Journal of Environmental Management

SN - 0301-4797

IS - 2

ER -